Approach To Arabic Linguistics PDF

Approaches to Arabic Linguistics
Studies in Semitic Languages

and Linguistics
Editorial board
T. Muraoka
VOLUME 49
Kees Versteegh
Approaches to Arabic
Linguistics
Presented to Kees Versteegh on the Occasion
of his Sixtieth Birthday
Edited by
Everhard Ditters and Harald Motzki
LEIDEN BOSTON
2007
This book is printed on acid-free paper.

Library of Congress Cataloging-in-Publication Data
A C.I.P. record for this book is available from the Library of Congress.
ISBN 0081-8461
ISBN 978 90 04 16015 6
Copyright 2007 by Koninklijke Brill NV, Leiden, The Netherlands.
Koninklijke Brill NV incorporates the imprints Brill, Hotei Publishing,
IDC Publishers, Martinus Nijhoff Publishers and VSP.
All rights reserved. No part of this publication may be reproduced, translated,
stored in a retrieval system, or transmitted in any form or by any means, electronic,
mechanical, photocopying, recording or otherwise, without prior written permission
from the publisher.
Authorization to photocopy items for internal or personal use is granted by
Koninklijke Brill NV provided that the appropriate fees are paid directly to
The Copyright Clearance Center, 222 Rosewood Drive, Suite 910,
Danvers, MA 01923, USA.
Fees are subject to change.
printed in the netherlands
CONTENTS
Preface ..................................................................................................
xi
Bibliography Kees Versteegh ............................................................
xv
HISTORY
Inside the Speakers Mind: Speakers Awareness as Arbiter of
Usage in Arabic Grammatical Theory .........................................
Ramzi Baalbaki
Pragmatics and Contractual Language in Early Arabic

Grammar and Legal Theory .........................................................
Michael Carter
25
Idmr in the Man of al-Farr: A Grammatical Approach

between Description and Explanation .........................................
Kinga Dvnyi
45
Arabic allad as a Conjunction: An old Problem and a New

Approach .........................................................................................
Werner Diem
67
Les origines de la grammaire arabe, selon la tradition:

description, interprtation, discussion ..........................................
Pierre Larcher
113
Sbawayhis View of the zarf as an mil .........................................

Aryeh Levin
135
Problems in the Medieval Arabic Theory of Sentence Types .........

Yishai Peled
149
viii
contents
Arabic avant la lettre. Divine, Prophetic, and Heroic Arabic .......

Stefan Wild
Inflection and Government in Arabic According to Spanish
Missionary Grammarians from Damascus (XVIIIth Century):
Grammars at the Crossroads of Two Systems .............................
Otto Zwartjes
189
209
LINGUISTICS
The Linguistic Analysis and Rules of Pause in Arabic ...................
Salman H. Al-Ani
247
The Explanation of Homonymy in the Lexicon of Arabic .............

Georges Bohas and Abderrahim Saguer
255
The Periphrastic Bilingual Verb Construction as a Marker of

Intense Language Contact. Evidence from Greek, Portuguese
and Maghribian Arabic .................................................................
Louis Boumans
291
Faula, faila, faala: dispersion et rgularits smantiques dans

les trois schmes simples du verbe arabe .....................................
Joseph Dichy
313
Featuring as a Disambiguation Tool in Arabic Natural

Language Processing .......................................................................
Everhard Ditters
367
Arabic on the Media: Hybridity and Styles .....................................

Mushira Eid
The Use of Morphological Patterns in Arabic Grammars of
Turkic ...............................................................................................
Robert Ermers
Lexical Gaps in Arabic: Evidence from Dictionaries ......................
Jan Hoogland
403
435
455
contents
Masdar Formation ..............................................................................
Joost Kremers
Mthodologie linguistique: organisation de la langue arabe.
Organisation gnrale des langues ...............................................
Andr Roman
ix
475
501
DIALECTS
How to be KOOL in Arabic Writing: Linguistic Observations
from the Side Line ..........................................................................
Gert Borg
Hello, I say, and welcome! Where from, these riding men?
Arabic Popular Poetry and Political Satire: a Study in
Intertextuality from Jordan ...........................................................
Clive Holes
Notes on the Dialects of the Lgt and H amda h of Southern
Sinai .................................................................................................
Rudolf de Jong
Classical and Colloquial Arabic Archaisms .....................................
Alan S. Kaye
527
543
565
595
Do They Speak the Same Language? Language Use in Juba

Local Courts ....................................................................................
Catherine Miller
607
Paradigmatic Stability and Final Laryngeals in Nigerian Arabic:

Why History Repeats itself, without Actually Doing so .............
Jonathan Owens
639
Some Aspects of Diglossia as Reflected in the Vocabulary of

Literary and Colloquial Arabic .....................................................
Judith Rosenhouse
653
contents
Everything you Always Wanted to Know about l, yil to say

in Egyptian Arabic .........................................................................
Manfred Woidich
675
Index ....................................................................................................
701
PREFACE
From when we began to compile a Festschrift for Kees Versteegh on the

occasion of his sixtieth birthday, the first problem we had to tackle was:
Who should we invite to contribute? The broad range of his scholarly
interests, his expertise in different fields, his academic contribution
to them, as well as his network of global contacts made it clear that
the circle of his colleagues and friends is too large to invite all to contribute. So we decided to confine this volume to Kees Versteeghs core
contribution: Arabic linguistics. This decision will surely be regretted
by his colleagues working in related fields of scholarship in which Kees
Versteegh is engaged as well, but we had to make a choice. Moreover,
our time schedule forced us to only select the first set of early contributions brought in. In the end, what does it matter? Someone will have
to prepare another for his sixty-fifth!
Even with this limitation, the present Festschrift has become voluminous, since the eagerness of scholars in the field of Arabic linguistics
to contribute has been overwhelming. It clearly demonstrates: Kees
Versteegh has been widely considered to be one of the most eminent
scholars in this discipline. He also is appreciated as a tirelessly working
editor and co-editor of renowned book series, collective volumes and
encyclopaedias. Moreover, he has won many friends by his kindness
and dependability.
With the list of Kees Versteeghs publications at hand, one realises the
number, diversity, and depth of his scholarly interests, from Hellenistic
elements in Arabic linguistics and other fields of Islamic culture, to
the history of Arabic grammar as a scholarly discipline from classical
to modern times, grammatical and linguistic phenomena described
by scholars of Arabic grammar, changes in written and spoken Arabic
through the ages, as well as early Qurnic exegesis as source for the
beginnings of Arabic grammar. In addition, he ranslated and commented
on Arabic texts, grammatical treatises and other genres including a novel.
His books Greek Elements in Arabic Linguistic Thinking; Pidginization
and Creolization: the Case of Arabic; Arabic Grammar and Qurnic
Exegesis in Early Islam; and The Arabic Language are handbooks. Every
single scholar and student of Arabic linguistics can not bypass them.
xii
preface
Several of his publications have been translated into Arabic, and others
will follow.
At Nijmegen University, where Kees Versteegh is working since
1972, first as lecturer and later as full professor, he organized several
workshops about the History of Arabic Grammar (1984 and 1987)
and the Model of Arabic Grammar in other Languages (1997). With
intense pleasure the participants of these conferences remember these
fruitful academic gatherings occurring in a very informal and convivial
atmosphere. These gatherings resulted in two collective volumes he
edited in cooperation with one of his many colleagues. Thanks to his
co-editorship, international publication projects such as the History of
Language Sciences and the Encyclopaedia of Arabic Language and Linguistics, Arabic linguistics acquired a firm place in the field of linguistics
in general. Generations of Dutch students, translators and interpreters
of Arabic will know his name from the two volumes of the dictionary
Woordenboek Arabisch-Nederlands/Nederlands-Arabisch of which he is
one of the editors.
This Festschrift will be a monument for Kees Versteegh in Arabic
studies, as a homage to his scholarly oeuvre. Several contributions
camd from former students who wrote their Ph.D. theses under his
supervision. As a matter of fact, all contributors studied with him. We
divided the collected articles into three chapters reflecting the foci of
his scholarly oeuvre: history of Arabic grammar, Arabic linguistics, and
Arabic dialectology. We have put the history of Arabic grammar first,
since it is Kees Versteeghs true domain. He wrote his Ph.D. in this field,
dedicated most of his publications to it, and returns to it. He translated
and commented upon al-Zajjjs Kitb al-dh , a theoretical treatise on
Arabic grammar and one of the classical works on this issue. Since the
1980s, his research and publications also have been directed towards
other historical and thematical topics in linguistics, and he certainly
will recognize many of his own ideas in the contributions in all three
sections. His studies on pidginization and creolization in Arabic and his
interest in Arabic dialects express his feeling for the historical development of the Arabic language.
Kees Versteegh is a polyglot. When invited to lecture outside the
Netherlands he enjoys lecturing in the language of is audience, in English, French, German and Arabic, as well as Spanish or Czech. From
the start of his academic careerhe first studied Greek and Latinhe
cherished a fondness for these languages. In his research this became
preface
xiii
apparent in his search for Hellenistic elements in Arabic linguistics.

Together with a colleague of Classical Studies, he also teaches regular
classes on linguistics. His interest inwell, let us sayexotic languages
is exceptional. His eyes sparkle when he tells how he learned Inuktitut,
the language of the Inuit, from an indigenous scholar in Alaska or how
he mastered Hottentot clicks in South Africa.
Many colleagues enjoyed Kees Versteeghs hospitality in his sextons
house in picturesque Batenburg. They will never forget the tasty meals
he prepares himself as the evenings conversation quickly passes from
scholarly issues to personal and private ones, as for example his experiences as director of the Netherlands-Flemish Institute in Cairo or the
birds he suddenly noticed by a Zamalek window. One may be engaged
with Kees in a really basic discussion of Arabic linguistics, only to be
interrupted by his quick sprint outdoors to observe an ornithological
event, afterwards resuming the discussion exactly at the comma or full
stop of his last intervention. Bird watching remains his hobby and on
a walk along the Maas (or any other river), he never forgets his binoculars. Besides linguistics, a guest learns a lot about the birds of the
region, their names in Dutch and other languages, including Arabic.
Guests get free access to his rich private library with its rare books
on Arabic linguistics and even the privilege of borrowing without any
check for solvability.
The study of Arabic has been Arabic-speaking scholars domain for
many centuries. Kees Versteegh has unearthed their understanding of
the language and translated it for Western scholarship. Western scholars
linguistic study of Arabic, according to current standards, started only
at the beginning of the 19th century. So while the western linguistic
approach is much younger than that of those native scholars, Western
Arabic linguistics are on par with even native intellectual efforts, and
Kees Versteegh has contributed to this state of affairs. Had classic scholars of the Arabic language like Sbawayhi or al-Zajjj the change to
witness the linguistic achievements of their modern Western colleagues
like Kees Versteegh, they would have been impressed. They might have
cited the verses of Ibn H amds (Sicily, 447527/10561133):
al adawti tuyrin fish in
lahunna ardu inda l-halli
turajjiu fh durba l-luh ni
al anna afsah ah ajamu

muhammalatu l-wazni l tulamu
fa-tutribun wa-hya l tufhamu
On the tunes of birds that speak true Arabic,

but the most eloquent of them are foreigners
xiv
preface
who use metres that are neglected by al-Khall and unknown
in which they quaver notes of various melodies,
delighting us although they are unintelligible.
This Festschrift would not have seen the light without several helping
hands. Greetje Heemskerk compiled the bibliography of his publications.
Marjolein van der Heul and Ine Smeets produced the raw version of
the index. Elizabeth Bishop copy-edited the English contributions. Joed
Elich of Brill has been willing to publish the book. Ingrid Heijckers
lead us safely through the whole of the production process and beyond.
Renee Otto supervised the publication process. We are very grateful for
their commitment.
Everhard Ditters and Harald Motzki
BIBLIOGRAPHY KEES VERSTEEGH
1977
Greek Elements in Arabic Linguistic Thinking. (= Studies in Semitic
Languages and Linguistics, 7). Leiden: E. J. Brill. Also published as
PhD thesis University of Nijmegen. [Arabic translation by Mahmud
Kankr, 2000: Ansir ynniyya f l-fikr al-luaw al-arab. Amman:
Jamiyyat umml al-mat bi at-tawuniyya].
1978
The Arabic Terminology of Syntactic Position. Arabica 25, 261281.
1979
Die Mission des Kyrillos im Lichte der arabo-byzantinischen Beziehungen. Zeitschrift der deutschen Morgenlndischen Gesellschaft 129,
233262.
Reviews
Ephrem Hunayn Festival Baghdd 47/2/1974. 1974. Baghdad: Mat bt
Majma al-Lua al-Suryniyya, Bibliotheca Orientalis 36, 9697.
Rundgren, Frithiof. 1976. ber den griechischen Einflu auf die arabische
Nationalgrammatik. Uppsala: Acta Universitatis Upsaliensis. N.S. 2(5),
119144. Bibliotheca Orientalis 36, 235236.
1980
Hellenistic Education and the Origin of Arabic Grammar. Progress in
Linguistic Historiography: Papers from the International Conference
on the History of the Language Sciences, Ottawa, 2831 August 1978.
Edited by Konrad Koerner. Amsterdam: J. Benjamins. 333344.
Logique et grammaire au dixime sicle. Histoire, pistmologie, Langage 2. 3952.
Notice bibliographique. Histoire, pistmologie, Langage 2. 6775.
The International Project Onomasticon Arabicum. Bibliotheca Orientalis
37, 291294.
The Origin of the Term qiys in Arabic Grammar. Zeitschrift fr arabische Linguistik 4, 730.
The Stoic Verbal System. Hermes 108, 338357.
xvi
bibliography kees versteegh
Reviews
Ambros, Edith. 1979. Sieben Kapitel des arh Kitb Sbawaih von arRummn in Edition und bersetzung. Wien: Verlag des Verbandes der
wissenschaftlichen Gesellschaften sterreichs. Bibliotheca Orientalis
37, 361362.
Malti Douglas, Fedwa and Genevive Fourcade. 1976. The Treatment
by Computer of Medieval Arabic Biographical Data: An introduction
and guide to the Onomasticum Arabicum. Paris: Centre national de
la recherche scientifique. Bibliotheca Orientalis 37, 362.
Traini, Renato. 1977. Sources biographiques des Zadites (annes 1221200
h.): Letters alif-ha. Paris: Centre national de la recherche scientifique.
Bibliotheca Orientalis 37, 362363.
1981
A Dissenting Grammarian: Qutrub on declension. Historiographia Linguistica 8, 403429. [See also The History of Linguistics in the Near
East. Edited by Kees Versteegh, Konrad Koerner and Hans-Josef
Niederehe. 1983. Amsterdam: J. Benjamins. 167193].
De taalsituatie in de Arabische wereld. De taal van de Islam: Opstellen
over Arabische, Turkse en Afghaanse cultuur. Nijmegen: Nederlandse
Vereniging voor de Studie van het Midden-Oosten en de Islam.
1938.
La conception des temps du verbe chez les grammairiens arabes. Analyses, thorie 3, 4768.
Reviews
Klein-Franke, Felix. 1980. Die klassische Antike in der Tradition des
Islam. Darmstadt: Wissenschaftliche Buchgesellschaft. Bibliotheca
Orientalis 38, 734737.
1982
Progress and Change in the History of Arabic Grammar. Linguistics
in the Netherlands 1982. Edited by Saskia Daalder and Marinel Gerritsen. Amsterdam: North-Holland. 3950.
Structural Change and Pidginization in the History of the Arabic
Language. Papers from the 5th International Conference on Historical Linguistics, Galway, April 610 1981. Edited by Anders Ahlqvist.
Amsterdam: J. Benjamins. 362373.
Vertalingen voor 1ste jaarsstudenten Arabisch (with Gert Borg). Nijmegen:
Katholieke Universiteit Nijmegen.
xvii
Reviews
Dagorn, Ren. 1981. La geste dIsmal daprs lonomastique et la tradition arabes, Genve: Droz; Paris: Champion. Bibliotheca Orientalis
39, 720726.
Rowson, Everett and Seeger Bonebakker. 1980. A Computerized Listing
of Biographical Data from the Yatmat al-Dahr by al-Thalib. Malibu:
Undena Publications. Bibliotheca Orientalis 39, 727729.
1983
A Dissenting Grammarian: Qutrub on declension. The History of Linguistics in the Near East. Konrad Koerner, Hans-Josef Niederehe, and
Kees Versteegh (Eds.). Amsterdam: J. Benjamins. 167193. [See also
Historiographia Linguistica 1981. 8, 403429].
Arabic Grammar and the Corruption of Speech. Arab Language and
Culture. Edited by Ramzi Baalbaki. Beirut: American University of
Beirut. [= al-Abhth 31]. 139160.
Current Bibliography on the History of Arabic Grammar. Zeitschrift
fr arabische Linguistik 10, 8689.
History of Eastern Linguistics in the Soviet Union. Historiographia
Linguistica 10, 289307.
The History of Linguistics in the Near East. (= Studies in the History of
Linguistics, 28). Editor (with Konrad Koerner and Hans-Josef Niederehe). Amsterdam: J. Benjamins.
1984
Arab Grammatical Studies before Sbawayh. Matriaux pour une histoire
des thories linguistiques = Essays toward a history of linguistic theories
= Materialien zu einer Geschichte der sprachwissenschaftlichen Theorien. Edited by Sylvain Auroux, Michel Glatigny, Andr Jolly, Anne
Nicolas and Irne Rosier. Lille: Universit de Lille III. 227238.
Piginigado, kreoligado kaj Esperanto. Hungara Vivo 4, 127129.
Pidginization and Creolization: The case of Arabic. (= Current Issues in
Linguistic Theory, 33). Amsterdam: J. Benjamins.
xviii
Reviews
Bakalla, Muhammad. 1983. Arabic linguistics. London: Mansell. Bibliotheca Orientalis 41, 751754.
Carter, Michael. 1982. Arab Linguistics: An introductory classical text
with translation and notes. Amsterdam: J. Benjamins. Bibliotheca
Nebes, Norbert. 1982. Funktionsanalyse von kna yafalu. Hildesheim:
Olms. Bibliotheca Orientalis 41, 754757.
1985
La Grande tymologie dIbn Jinn. La linguistique fantastique. Edited
by Sylvain Auroux, Jean-Claude Chevalier, Nicole Jacques-Chaquin
and Christiane Marchello-Nizia. Paris: Denol. 4450.
Studies in the History of Arabic Grammar. Papers Presented during a
Workshop held at the University of Nijmegen, April 1619, 1984. Editor
(with Hartmut Bobzin). Wiesbaden: O. Harrassowitz. (= Zeitschrift
fr arabische Linguistik 15).
Survey of Journals. Arab Journal of Language Studies / al-Majalla
al-arabiyya li-d-dirst al-luawiyya 2, 189197.
The Development of Argumentation in Arabic Grammar: The declension of the dual and the plural. Studies in the History of Arabic Grammar. Edited by Hartmut Bobzin and Kees Versteegh. Wiesbaden:
Harrassowitz. 1985. 152173. [= Zeitschrift fr arabische Linguistik
15].
Translation
Salih, Tayyib. Seizoen van de trek naar het noorden. Uit het Arabisch
vertaald en van een nawoord voorzien. Amsterdam: Meulenhoff.
[Dutch translation of Mawsim al-hijra il -aml].
Reviews
Sezgin, Fuat. 1982, 1984. Geschichte des arabischen Schrifttums. VIII.
Lexikographie. IX. Grammatik. Leiden: E. J. Brill. Historiographia
Linguistica 12, 452461.
1986
History of Arabic Grammar. Nijmegen: Instituut voor Talen en Culturen
van het Midden Oosten, Katholieke Universiteit.
xix
Latinitas, Hellnisms, Arabiyya. The History of Linguistics in the Classical Period. Edited by Daniel J. Taylor. (= Historiographia Linguistica
13, 425448). Amsterdam: J. Benjamins, 1987. 251274.].
The Origin of the Romance Languages and the Arabic Dialects. Islo e
arabismo na pennsula ibrica: Actas do XI. congresso da Unio Europeia de Arabistas e Islamlogos, Evora, Faro, Silves, 29 set.6 out. 1982.
Edited by Adel Y. Sidarus. vora: Universidade de vora. 337352.
Word Order in Uzbekistan Arabic and Universal Grammar. On the
Dignity of Man: Oriental and classical studies in honour of Frithiof
Rundgren. Edited by Tryggve Kronholm and Eva Riad. (= Orientalia
Suecana 3334). Uppsala: Almqvist and Wiksell. 443453.
1987
al-Arab f jibl al-Alp. al-Azmina 7, 2630.
Die arabische Sprachwissenschaft. Grundri der arabischen Philologie,
II. Edited by Helmut Gtje, Wiesbaden: L. Reichert. 148176.
Het Arabisch: Norm en realiteit (with Arie Schippers). Muiderberg: D.
Coutinho.
Marginality in the Arab Grammatical Tradition. Papers in the History
of Linguistics: Proceedings of the Third International Conference on
the History of the Language Sciences (ICHoLS III), Princeton, 1923
August 1984. Edited by Hans Aarsleff, Louis G. Kelly and Hans-Josef
Niederehe. Amsterdam: J. Benjamins. 8796.
Nahwiyyna wa-luawiyyna wa-mawqif Dozy iz at-turt an-nahw
al-arab. F l-mujamiyya al-arabiyya al-musira. Edited by Ahmed
El-Ayed and Ibrahim Ben Mrad. Tunis: Dr al-arb al-Islm.
401413.
Reviews
Sublet, Jacqueline. 1985. Cahiers donomastique Arabe, 19821984. Paris:
Centre national de la recherche scientifique. Bibliotheca Orientalis
44, 130133.
1988
De ontwikkeling van de technische woordenschat in Modern Standaard
Arabisch: Naar aanleiding van een recente publicatie. Sharqiyyt 1,
8085.
xx
Geschiedschrijving in de klassieke Arabisch-Islamitische samenleving.

Tussen traditie en wetenschap: Geschiedbeoefening in niet-westerse
culturen. Edited by R.B. van de Weijer, P.G.B. Thissen and R. Schnberger. Nijmegen: Katholieke Universiteit Nijmegen. 133145.
Pourquoi tudier la tradition grammaticale? Le Maroc et la Hollande:
tudes sur lhistoire, la migration, la linguistique et la smiologie de la
culture. Edited by Abdelmajid Kaddouri, Jilali Sab and Abdelmajid
Zeggaf. Rabat: Universit Mohammed V. 207217.
1989
A Sociological View of the Arab Grammatical Tradition: Grammarians
and their professions. Studia linguistica et orientalia memoriae Haim
Blanc dedicate. Edited by Paul Wexler, Alexander Borg and Sasson
Somekh. Wiesbaden: Harrassowitz. 289302.
Early and late Grammarians in the Arab Tradition: The morphonology of the hollow verbs. Zeitschrift fr arabische Linguistik 20,
922.
La tradition arabe. I : Le langage, la religion et la raison. Histoire des
ides linguistiques, I : La naissance des mtalangages en Orient et en
Occident. Edited by Sylvain Auroux. Lige: P. Mardaga. 243259.
The Definition of Philosophy in a Tenth-century Grammarian. Jerusalem Studies in Arabic and Islam 12, 6692.
Vulgair Latijn en Koine-Grieks: De verhouding tussen standaardtaal
en volkstaal. Lampas 22, 7491.
Reviews
Bergter, Annette. 1989. Das Kapitel inna wa-ahawtuh aus dem Manhaj as-slik des Grammatikers Ab H ayyn al-arnt (12561344).
Hildesheim: G. Olms. Historiographia Linguistica 16, 180184.
1990
Are Linguists Ridiculous? Notes on a heavenly discussion between
grammarians in the 11th century. History and Historiography of
Linguistics: Proceedings of the Fourth International Conference on the
History of the Language Sciences (ICHoLS IV), Trier, 2428 August
1987, I: Antiquitity17th Century. Edited by Konrad Koerner and
Hans-Josef Niederehe. Amsterdam: J. Benjamins. 147155.
Borrowing and Influence: Greek grammar as a model. Le langage dans
lantiquit. Edited by Pierre Swiggers and Alphons Wouters. Leuven:
Peeters. 197212.
xxi
Freedom of the Speaker? The term ittis and related notions in Arabic
grammar. Studies in the History of Arabic Grammar, II. Edited by
Michael G. Carter and Kees Versteegh. Amsterdam: J. Benjamins,
1990. 281293 [Arabic translation by Bouchaib Barramou: H urriyyat
al-mutakallim? Mussa lah al-ittis wa-l-mafhm al-murtabiha bihi f
n-nahw al-arab. Fikr wa-naqd 3:24 (1999). 99110].
Grammar and Exegesis: The origins of Kufan grammar and the Tafsr
Muqtil. Der Islam 67, 206242. [Reprint in The Early Islamic Grammatical Tradition. Edited by Ramzi Baalbaki. Aldershot [etc.]: Ashgate/Variorum. 2007. 3773].
Over taal en verandering. Nmegen: Katholieke Universiteit. Inaugural
lecture, University of Nmegen.
Studies in the History of Arabic Grammar, II. Proceedings of the 2nd Symposium on the History of Arabic Grammar, Nijmegen, 27 April1 May,
1987. Editor (with Michael G. Carter). Amsterdam: J. Benjamins.
The Arab Presence in France and Switzerland in the 10th Century.
Arabica 37, 359388.
The Earliest Commentary on the Qurn: Muqtils Tafsr. Makalahmakalah yang disampaikan dalam rangka kunjungan menteri Agama
R.I.H. Munawir Sjadzali, M.A. ke Negeri Belanda (31 Oktober
7 November 1988). Edited by Wim Stokhof and Nico Kaptein. Jakarta:
INIS. 213219. [with Indonesian translation: Tafsir Quran paling
awal: Tafsir Muqatil].
Reviews
Owens, Jonathan. 1988. The Foundations of Grammar: An introduction
to medieval Arabic grammatical theory. Amsterdam: J. Benjamins. Die
Welt des Islams 30, 248250.
1991
Arabic Language Teaching and the Status of Standard Arabic. Proceedings of the Symposium on Differentiation in LSP, Learning and Teaching, Leuven, 710 November 1990. Edited by Serge Verlinde. Leuven:
Instituut voor levende talen. 95102.
Greek Translations of the Qurn in Christian Polemics (9th century
A.D.). Zeitschrift der Deutschen Morgenlndischen Gesellschaft 141,
5268.
Man. 1. In grammar. The Encyclopaedia of Islam: New edition. Edited
by C. E. Bosworth, E. van Donzel, B. Lewis and Ch. Pellat, assisted
by F. Th. Dijkema and S. Nurit. Leiden: E. J. Brill. VI, 346.
xxii
New approaches towards the history of Arabic. Taqaddum al-lisniyyt

fi l-aqtr al-arabiyya [Progrs de la linguistique dans les tats arabes].
Edited by Abdelkader Fassi Fehri. Beirut: Dr al-arb al-Islm.
199216.
The substratum debate in creole linguistics. Diachronica 8, 5980.
Two Conceptions of Irreality in Arabic Grammar: Ibn Him and Ibn
al-jib on the particle law. De la grammaire de larabe aux grammaires des arabes. Edited by Pierre Larcher. Damas: Institut Franais
de Damas. 7792 [= Bulletin dtudes Orientales, 43].
Reviews
Holes, Clive. 1990. Gulf Arabic. London: Routledge. Bibliotheca Orientalis 48, 293296.
Wouters, Alfons. 1988. The Chester Beatty Codex Ac. 1499: A GraecoLatin lexicon on the Pauline Epistles and a Greek grammar. Leuven:
Peeters. Bibliotheca Orientalis 48, 523527.
1992
De val van de Barmakiden volgens Tabar. Elf wijzen van interpreteren:
Essays over het lezen van teksten uit het islamitisch cultuurgebied.
Edited by Ed de Moor. Nijmegen: Instituut voor Talen en Culturen
van het Midden-Oosten. 117126.
Grammar and Rhetoric: Jurjn on the verbs of admiration. Studies in
Semitic Linguistics in Honor of Joshua Blau. Edited by M. Bar-Asher
[et al.]. Jerusalem: Hebrew University of Jerusalem. 113133. [=
Jerusalem Studies in Arabic and Islam 15].
The Debate about Latin and Early Romance. Diachronica 9, 259285.
The History of Linguistics in the Low Countries. (= Studies in the History of the Language Sciences 64). Editor (with Jan Noordegraaf and
Konrad Koerner). Amsterdam: J. Benjamins. [Published in 1988 as
Historiographia Linguistica 15, 1+2].
Reviews
Malina, Renate. 1987. Zum schriftlichen Gebrauch des kairenischen
Dialekts anhand ausgewhlter Texte von Sadaddn Wahba. Berlin:
Schwarz. Bibliotheca Orientalis 49, 526528.
Mitchell, T. F. 1990. Pronouncing Arabic, I. Oxford: Clarendon. Bibliotheca Orientalis 49, 523526.
Sluiter, Ineke. 1990. Ancient Grammar in Context, Amsterdam: VU
University Press. Historiographia Linguistica 19, 401408.
xxiii
1993
Arabic Grammar and Qurnic Exegesis in Early Islam. Leiden: E. J.
Brill.
Esperanto as a First Language: Language acquisition with a restricted
input. Linguistics 31, 539555.
Leveling in the Sudan: From Arabic creole to Arabic dialects. International Journal of the Sociology of Language 99, 6579.
Three is a Crowd: lawyers and linguists on Qurn 4/11. Wolfdietrich
Fischer zum 65. Geburtstag. Edited by Hartmut Bobzin and Otto Jastrow. Wiesbaden: Harrassowitz. 302315 [= Zeitschrift fr arabische
Linguistik 25].
Reviews
Jastrow, Otto. 1990. Der arabische Dialekt der Juden von Aqra und Arbl.
Wiesbaden: O. Harrassowitz. Bibliotheca Orientalis 50, 858862.
Johnstone, Barbara. 1991. Repetition in Arabic Discourse: Paradigms,
syntagms, and the ecology of Language. Amsterdam: J. Benjamins.
1994
Arabic Outside the Arab World. Editor. New Delhi: Bahri Publications.
[= Indian Journal of Applied Linguistics, 20].
The Arabic Linguistic Tradition in a Comparative Perspective. Arab
Journal of the Humanities 12 (no. 46), 282297.
The Notion of Underlying Levels in the Arabic Grammatical Tradition. Historiographia Linguistica 21, 271296. [Arabic translation by
Bouchab Barramou: Mafhm al-mustawayt at-tahtiyya f t-turt
an-nahw al-arab. Recherches Linguistiques 6:2 (2001), 81126].
Reviews
kesson, Joyce. 1990. Ahmad b. Al b. Masd on Arabic morphology =
Marh al-arwh, edited with translation, commentary and introduction. Part 1: The strong verb as-sahh. Leiden: E. J. Brill. Bibliotheca
Broselow, Ellen, Mushira Eid and John McCarthy (Eds.). 1992. Perspectives on Arabic Linguistics, IV. Amsterdam: J. Benjamins. Bibliotheca
Comrie, Bernard and Mushira Eid (Eds.). 1991. Perspectives on Arabic
Linguistics, III. Amsterdam: J. Benjamins. Journal of the American
Oriental Society 114, 107109.
xxiv
Robins, Robert H. 1993. The Byzantine Grammarians: Their place

in history. Berlin: Mouton de Gruyter. Bibliotheca Orientalis 51,
719723.
1995
The Explanation of Linguistic Causes: Az-Zajjjs Theory of grammar,
introduction, translation, and commentary. Amsterdam: J. Benjamins.
(Translation of Ab l-Qsim Abd ar-Rahmn b. Ishq az-Zajjj, Kitb
al-dh f ilal an-nahw).
Arabic Outside the Arab World. (= Indian Journal of Applied Linguistics,
20). (Co-editor). New Delhi: Bahri Publications.
1996
Arab Grammarians: Ab l-Aswad ad-Dual, 3 i-ii; Ab H ayyn, 4
i-5 ii; Ab Ubayd, 5 ii; Ab Ubayda, 6 i; al-Axfa al-Asar, 11 ii;
al-Axfa al-Awsat, 11 ii-12, i; al-Asma, 49 ii-50 i; al-Astarbd, 50
i-51 i; al-Azhar, 59 i-ii; al-Batalys, 70 ii71 i; al-Frb, 288 ii-290
i; al-Fris, 290 i-ii; al-Farr, 291 i-292 i; al-Frzbd, 300 i-ii;
al-Jarm, 326 ii-327 i; al-Jawhar, 330 iii; al-Jurjn, 380 i-381 ii;
Xalaf al-Ahmar, 384 i-ii; Xall, 385 i386 i; Ibn Ab Ishq, 438 i; Ibn
jurrm, 438 i-ii; Ibn al-Anbr (Ab l-Bakr), 438 ii; Ibn al-Anbr
(Ab l-Barakt), 438 ii-439 ii; Ibn Aql, 439 i-440 i; Ibn Durayd, 441
i-ii; Ibn Fris, 442 ii-443 i; Ibn Jinn, 445 ii-447 ii; Ibn al-H jib, 447
i-ii; Ibn Xlawayh, 447 ii-448 i; Ibn Him, 448 i-ii; Ibn Kaysn, 448
ii-449 i; Ibn Mad, 449 ii-450 ii; Ibn Mlik, 450 ii-451 i; Ibn Manzr,
451 i-ii; Ibn Qutayba, 452 i-ii; Ibn as-Sarrj, 452 ii-453 i; Ibn Sdah,
453 ii-454 i; Ibn as-Sikkt, 454 i-ii; Ibn uqayr, 454 ii-455 i; Ibn Usfr,
455 ii-456 i; Ibn Walld, 456 i-ii; Ibn Ya, 456 i-ii; Ixwn as-Saf,
458 i-ii; s ibn Umar at-Taqaf, 461 ii-462 i; al-Kis, al-Mzin, 617
ii- 618 i; al-Mubarrad, 651 ii-657 ii; an-Nahhs, 666 ii- 667 i; al-Ql,
766 i-ii; Qutrub, 767 ii- 768 i; ar-Rummn, 808 i-ii; Sbawayhi, 856
i-859 i; as-Srf, 863 i-864 i; as-Suyt, 895 ii-896 i; at-Talib, 904 i-ii;
Talab, 905 i-ii; Ukbar, 946 i-ii; Yahy ibn Ad, 1028 i-ii; az-Zabd,
1037 i-ii; az-Zajjj, 1038 i-ii; az-Zajjj, 1038 ii-1039 ii; az-Zamaxar,
1039 ii-1040 i; az-Zubayd, 1045 i-ii. Lexicon Grammaticorum. Edited
by Harro Stammerjohann. Tbingen: Niemeyer.
Jahilia. Het woordenboek van De duivelsverzen. Edited by Maarten
Asscher [et al.]. Amsterdam: Contact. 5253.
Linguistic Attitudes and the Origin of Speech in the Arab World.
Understanding Arabic: Essays in contemporary Arabic linguistics in
xxv
honor of El-Said Badawi. Edited by Alaa Elgibali. Cairo: American

University in Cairo Press. 1531.
The Linguistic Introduction to Rzs Tafsr. Studies in Near Eastern
Languages and Literatures. Memorial volume of Karel Petrek. Edited
by Petr Zemnek. Prague: Academy of Sciences of the Czech Republic. 589603.
Western Studies on the History of Arabic Grammar, 19691994. Proceedings of the Colloquium on Arabic Linguistics, Bucharest August 29
September 2. 1994. Edited by Nadia Anghelescu and Andrei Avram.
Bucharest: University of Bucharest. I, 927.
Reviews
Fernndez, Mauro. 1993. Diglossia: A comprehensive bibliography
19601990, Amsterdam: J. Benjamins. Word 47, 288291.
Peled, Yishai. 1992. Conditional Structures in Classical Arabic. Wiesbaden: O. Harrassowitz. Bibliotheca Orientalis 53, 557560.
Wagner, Daniel A. 1993. Literacy, Culture, and Development: Becoming
literate in Morocco. Cambridge: Cambridge University Press. International Journal of the Sociology of Language 119, 129134.
1997
Bedouin and Townspeople: Ibn Khaldns view on the development of
Arabic. The Diversity of Development: Essays in honour of Jan Kleinpenning. Edited by Ton van Naerssen, Marcel Rutten and Annelies
Zoomers. Assen: Van Gorcum. 332340.
Landmarks in Linguistic Thought, III: The Arabic linguistic tradition.
London: Routledge.
Sarf. The Encyclopaedia of Islam: New edition. Edited by C. E. Bosworth,
E. van Donzel, W. P. Heinrichs and G. Lecomte, assisted by P. J. Bearman and S. Nurit. Leiden: E. J. Brill. IX, 5354.
al-Shaybn, Ab Amr Ishk b. Mirr. The Encyclopaedia of Islam: New
edition. Edited by C. E. Bosworth, E. van Donzel, W. P. Heinrichs
and G. Lecomte, assisted by P. J. Bearman and S. Nurit. Leiden:
E. J. Brill. IX, 394395.
The Arabic Language. Edinburgh: Edinburgh University Press. [2nd
revised edition 2001] [Arabic translation by Muhammad a-arqw:
Al-lua al-arabiyya. Cairo: Al-Majlis al-al li-t -ta qfa, 2003].
The Arabic Tradition. The Emergence of Semantics in Four Linguistic
Traditions: Hebrew, Sanskrit, Greek, Arabic. Edited by Wout van
xxvi
Bekkum, Jan Houben, Ineke Sluiter and Kees Versteegh. Amsterdam:

J. Benjamins. 225284.
The Emergence of Semantics in Four Linguistic Traditions: Hebrew, Sanskrit, Greek, Arabic. (with Wout van Bekkum, Jan Houben and Ineke
Sluiter). Amsterdam: J. Benjamins.
Wetenschap in de Islamitische samenleving. In het huis van de Islam:
Geografie, geschiedenis, geloofsleer, cultuur, economie, politiek. Edited
by Henk Driessen. Nijmegen: SUN. 278295.
Reviews
Bohas, Georges (Ed.). 1993. Dveloppements rcents en linguistique
arabe et smitique. Damas: Institut Franais de Damas. Bibliotheca
Dialectologia Arabica: A collection of articles in honour of the sixtieth
birthday of Professor Heikki Palva. Helsinki: Finnish Oriental Society.
1995. Bibliotheca Orientalis 54, 800802.
Durand, Olivier. 1995. Introduzione ai dialetti arabi. Milan: Centro Studi
Camito-Semitici. Bibliotheca Orientalis 54, 799800.
Gille, Christiane. 1995. Das Kapitel al-Mausl (Das Relativum) aus dem
Manhaj as-slik des Grammatikers Ab H ayn al-arnt (12561344).
Hildesheim: G. Olms. Bibliotheca Orientalis 54, 793795.
Khan, Mohammad-Nauman. 1994. Die exegetischen Teile des Kitb alAyn: Zur ltesten philologischen Koranexegese. Berlin: K. Schwarz.
Watson, Janet C. E. 1993. A Syntax of Sann Arabic. Wiesbaden:
O. Harrassowitz. Bibliotheca Orientalis 54, 795799.
1998
Reviews
Endress, Gerhard and Dimitri Gutas. 1994, 1995. A Greek and Arabic
Lexicon: Materials for a dictionary of the mediaeval translations from
Greek into Arabic, fasc. 2 and 3. Leiden: E. J. Brill. Journal of the
American Oriental Society, 118, 108109.
Ingham, Bruce. 1994. Najdi Arabic: Central Arabian. Amsterdam:
J. Benjamins. Bibliotheca Orientalis 55, 294297.
Marais, Jean (Ed.). 1996. Quebecs Aboriginal Languages: History, planning, development. Clevedon: Multilingual Matters. Applied Linguistics
19, 131135.
xxvii
1999
Loanwords from Arabic and the Merger of d /d. Compilation and
Creation in adab and lua: Studies in memory of Naphtali Kinberg
(19481997). Edited by Albert Arazi, Joseph Sadan and David
J. Wasserstein. Winona Lake, IN: Eisenbrauns. 273286. [= Israel
Oriental Studies 19].
The Dutch/ArabicArabic/Dutch Dictionary Project: Aims and
methods (with Jan Hoogland). Revue de la lexicologie 1415,
1326.
Zayd ibn Als Commentary on the Qurn. Arabic Grammar and Linguistics. Edited by Yasir Suleiman. London: Curzon Press. 929.
Reviews
Rizk, Attia. 1992. Ibn a-ajar. M ttafaqa lafzuh wa-xtalafa manhu.
Stuttgart: F. Steiner. Der Islam 76.184186.
2000
Ed de Moor (19362000). Sharqiyyt 12, iii-vi.
Editors Foreword. (With Sylvain Auroux, Konrad Koerner and HansJosef Niederehe). History of the Language Sciences / Geschichte der
Sprachwissenschaften / Histoire des sciences du langage. I. Berlin:
W. de Gruyter. Volume I, HSK 18.1. XXVXXXV.
History of the Language Sciences: An international handbook on the
evolution of the study of language from the beginning to the present =
Geschichte der Sprachwissenschaften: Ein internationales Handbuch zur
Entwicklung der Sprachforschung von den Anfngen bis zur Gegenwart
= Histoire des sciences du langage: Manuel international sur lvolution
de ltude du langage des origines nos jours, I. Editor (with Sylvain
Auroux, Konrad Koerner and Hans-Josef Niederehe). Berlin: W. de
Gruyter. Volume I, HSK 18:1.
Language and logic in the Arabic grammatical tradition. History of the
Language Sciences / Geschichte der Sprachwissenschaften / Histoire des
sciences du langage, I. Edited by Sylvain Auroux, Konrad Koerner,
Hans-Josef Niederehe and Kees Versteegh. Berlin: W. de Gruyter.
Volume I, HSK 18.1. 300306.
Tasrf. The Encyclopaedia of Islam: New edition. Edited by P. J. Bearman,
Th. Banquis, C. E. Bosworth, E. van Donzel and W. P. Heinrichs.
Leiden: E. J. Brill. X. 360361.
xxviii
Reviews
Enk, Gerrit J. van, and Lourens de Vries. The Korowai of Irian Jaya:
Their language in its cultural context. Oxford: Oxford University Press.
Word 51, 134137.
Lallot, Jean. Apollonius Dyscole, De la construction (syntaxe). I. Introduction, texte et traduction. II. Notes et index. Paris: J. Vrin. Word
51, 304308.
2001
Arabic in Europe: From language of science to language of minority.
Lingua e stile 36, 335346.
Arabic in Madagascar. Bulletin of the School of Oriental and African
Studies 64, 177187.
History of the Language Sciences / Geschichte der Sprachwissenschaften /
Histoire des sciences du langage, II. Editor (with Sylvain Auroux,
Konrad Koerner and Hans-Josef Niederehe). Berlin W. de Gruyter.
Volume II, HSK 18:2.
Linguistic Contacts between Arabic and other Languages. Arabica 48,
470508.
Naphtali Kinberg. Studies in the Linguistic Structure of Classical Arabic.
Editor (with Leah Kinberg). Leiden: E. J. Brill.
Reviews
Kropp Dakubu, M. E. Korle meets the sea: A sociolinguistic history of
Accra. Oxford: Oxford University Press. Word 52, 163166.
Owens, Jonathan. 1998. Neighborhood and Ancestry: Variation in the
spoken Arabic of Maiduguri, Nigeria, Amsterdam: J. Benjamins. Journal
of African Languages and Linguistics 22, 101105.
2002
Dead or Alive? The Status of the Standard Language. Bilingualism in
Ancient Society: Language contact and the written word. Edited by
J.N. Adams, Mark Janse and Simon Swain. Oxford: Oxford University
Press. 5274.
al-Zadjdjdj, Ab Ishk Ibrhm b. al-Sar. The Encyclopaedia of Islam:
New edition. Edited by P. J. Bearman, Th. Banquis, C. E. Bosworth,
E. van Donzel and W. P. Heinrichs. Leiden: E. J. Brill. XI, 377378.
al-Zadjdjdj, Abu l-K sim Abd al-Rah mn b. Ish k. The Encyclopaedia of Islam: New edition. Edited by P. J. Bearman, Th. Banquis,
xxix
C. E. Bosworth, E. van Donzel and W. P. Heinrichs. Leiden: E. J.

Brill. XI, 378379.
al-Zamakhshar, Abu l-K sim Mahmd b. Umar. The Encyclopaedia
of Islam: New edition. Edited by P. J. Bearman, Th. Banquis, C. E.
Bosworth, E. van Donzel and W. P. Heinrichs. Leiden: E. J. Brill. XI,
432434.
Translations
Al Jahiz: Traktaat over de lof van de zwarten boven de blanken. De
taal der engelen: 1250 jaar klassiek Arabisch proza. Edited by Arnoud
Vrolijk. Amsterdam: Contact. 135139.
Hunayn ibn Ishaq: De behandeling van staar. De taal der engelen: 1250
jaar klassiek Arabisch proza. Edited by Arnoud Vrolijk. Amsterdam:
Contact. 279282.
Reviews
Abu-Manga, Al-Amin. 1999. Hausa in the Sudan: Process of adaptation
to Arabic. Cologne: R. Kppe. Word 53, 292296.
2003
Arab tradition. International Encyclopedia of Linguistics (2nd ed.). Edited
by William J. Frawley. Oxford: Oxford University Press. 202204.
De Arabische taalkunde in Nederland. Sharqiyyt 15, 2942.
The Arabic component of the Indonesian lexicon. Rintisan kajian
leksikologi dan leksikografi. Edited by Lilie Suratminto and Munawar Holil. Jakarta: Fakultas Ilmu Pengetahuan Budaya Universitas
Indonesia. 216229.
Woordenboek Arabisch-Nederlands. Amsterdam: Bulaaq. Editor (with
Jan Hoogland and Manfred Woidich).
Woordenboek Nederlands-Arabisch. Amsterdam: Bulaaq. Editor (with
Jan Hoogland and Manfred Woidich).
2004
Approaches to Arabic Dialects: A collection of articles presented to Manfred
Woidich on the occasion of his sixtieth birthday. Editor (with Martine
Haak and Rudolf de Jong). Leiden: E. J. Brill.
From Classical Arabic to the Modern Vernaculars. Morphologie: Ein
internationales Handbuch zur Flexion und Wortbildung = Morphology:
an international handbook on inflection and word-formation. Edited
xxx
by Geert Boo, Christian Lehmann, Joachim Mugdan and Stavros

Skopeteas in collaboration with Wolfgang Kesselheim. Berlin: W. de
Gruyter. II, 17401754.
Grammar and Lexicography.. Encyclopaedia of Islam and the Muslim
World. Editor in chief: Richard C. Martin. USA: MacMillan Reference USA. I, 279281.
History of the Arabic Language. Encyclopaedia of Islam and the Muslim
World. Editor in chief: Richard C. Martin. USA: Macmillan Reference USA. 5862.
Meanings of Speech: The category of sentential mood in Arabic grammar. Le voyage et la langue: Mlanges en lhonneur dAnouar Louca et
dAndr Roman. Edited by Joseph Dichy and Hassan Hamz. Damascus: Institut Franais du Proche-Orient. 269287.
One Letter does not make a Word: The doubly weak imperatives in
Arabic. Words, Texts and Concepts Cruising the Mediterranean Sea:
Studies on the sources, contents and influences of Islamic philosophy
and Arabic philosophy and science dedicated to Gerhard Endress on his
sixty-fifth birthday. Edited by R. Arnzen and J. Thielmann. Leuven:
Peeters. 131141.
Phonological Constraints in Arabic Grammatical Theory: The Iltiq
as-skinayn. Romano-Arabica 3, 221235.
Pidginization and Creolization Revisited: The case of Arabic. Approaches
to Arabic Dialects: A collection of articles presented to Manfred Woidich
on the occasion of his sixtieth birthday. Edited by Martine Haak,
Rudolph de Jong and Kees Versteegh. Leiden: E. J. Brill. 343357.
Taalkunde en letterkunde in de kleine letteren. Gij letterdames en letterheren: Nieuwe mogelijkheden voor taalkundig en letterkundig onderzoek
in Nederland. Edited by Joost R. Swanborn. Amsterdam: Koninklke
Nederlandse Akademie van Wetenschappen. 189203.
The Future of Creolistics. Linguistics Today: Facing a greater challenge. Edited by Piet van Sterkenburg. Amsterdam: J. Benjamins.
305319.
Whats Where and Hows What? Interrogatives in Arabic dialects.
Estudios de dialectologa norteafricana y andalus 8, 239251.
2005
Breaking the Rules Without Wanting to: Hypercorrection in Middle
Arabic texts. Investigating Arabic: Current parameters in analysis and
learning. Edited by Alaa Elgibali. Leiden: E. J. Brill. 318.
xxxi
Reviews
Talmon, Rafael. 2003. Eighth-century Iraqi Grammar: A critical exploration of pre-H allian Arabic Linguistics. Winona Lake, IN: Eisenbrauns.
Jerusalem Studies in Arabic and Islam 30, 528535.
2006
A new treatise about the ilal an-nah w : Ibn al-Warrq on inna
wa-axawtuh. Grammar as a Window onto Arabic Humanism: A
collection of articles in honour of Michael G. Carter. Edited by Lutz
Edzard and Janet Watson. Wiesbaden: O. Harrassowitz. 5165.
Arabic Linguistic Tradition. Encyclopedia of Language and Linguistics:
Second edition. Editor in chief: Keith Brown. Oxford: Elsevier. I,
434440.
D d. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
Brill. I, A-Ed. 544545.
Encyclopedia of Arabic Language and Linguistics. Leiden: E. J. Brill.
General editor (with Mushira Eid, Alaa Elgibali, Manfred Woidich
and Andrzej Zaborski). I, A-Ed.
Introduction. (With Mushira Eid, Alaa Elgibal, Manfred Woidich and
Andrzej Zaborski. Eds.). Encyclopedia of Arabic Language and Linguistics. Leiden: E. J. Brill. I, A-Ed. v-x.
History of Arabic language teaching. Handbook for Arabic Language
Professionals in the 21st Century. Edited by Kassem M. Wahba, Zeinab
A. Taha and Liz England. Mahwah, NJ: Lawrence Erlbaum. 312.
History of the Language Sciences / Geschichte der Sprachwissenschaften /
Histoire des sciences du langage. Editor (with Sylvain Auroux, Konrad
Koerner and Hans-Josef Niederehe). Berlin: W. de Gruyter. Volume
III, HSK 18.3.
Leerboek modern geschreven Arabisch. Grammatica, woorden, idioom
en oefeningen. (with Everhard Ditters and Lieke de Jong). Nijmegen:
Radboud Universiteit. 1, 2, 3.
The Study of the non-Western Linguistic Traditions. History of the
Language Sciences / Geschichte der Sprachwissenschaften / Histoire
des sciences du langage. Edited by Sylvain Auroux, E. F. K. Koerner,
Hans-Josef Niederehe and Kees Versteegh. Berlin: W. de Gruyter.
Volume III, 18.3. 27912802.
xxxii
Reviews
Crowley, Terry. 2004. Bislama Reference Grammar. Honolulu. University
of Hawaii Press. Bdragen tot de taal-, land- en volkenkunde 162,
572575.
2007
Encyclopedia of Arabic Language and Linguistics. General editor (with
Mushira Eid, Alaa Elgibali, Manfred Woidich and Andrzej Zaborski).
Leiden: E. J. Brill. II: Eg-Lan.
Encyclopedia of Arabic Language and Linguistics. General editor (with
Mushira Eid, Alaa Elgibali, Manfred Woidich and Andrzej Zaborski).
Leiden: E. J. Brill. III: Lan-Ser.
H araka. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
Brill. II: Eg-Lan. 232236.
Idfa. (With Karin C. Ryding). Encyclopedia of Arabic Language and
Linguistics. Leiden: E. J. Brill. II: Eg-Lan. 294298.
Idmr. (With Michael C. Carter). Encyclopedia of Arabic Language and
Linguistics. Leiden: E. J. Brill. II: Eg-Lan. 300302.
Il. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
Illa. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J. Brill.
II: Eg-Lan. 308311.
Inna wa-axawtuh. Encyclopedia of Arabic Language and Linguistics.
Leiden: E. J. Brill. II: Eg-Lan. 354358.
Interrogative Pronoun. Encyclopedia of Arabic Language and Linguistics.
Leiden: E. J. Brill. II: Eg-Lan. 387389.
Isnd. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
HISTORY
INSIDE THE SPEAKERS MIND:

SPEAKERS AWARENESS AS ARBITER OF USAGE
IN ARAB GRAMMATICAL THEORY
Ramzi Baalbaki
American University, Beirut
1. Introduction
The Arab grammatical theory as represented by the later grammarians is generally characterized by its focus on the formal aspects of the
utterance within well-defined, albeit complex and often controversial
rules of usage. Although it is very difficult to represent linearly the shift
which took place during the evolution of the theory towards formal
aspects at the expense of meaning since there have indeed been some
attempts to restore to meaning its primary position in linguistic analysis, it may be safely argued that this shift started immediately after the
first major grammatical work, namely Sbawayhis (d. 180/796) Kitb.
It is, of course, true that at the levels of morpho-phonology and morpho-syntax Sbawayhis formal considerations were adopted almost in
their entirety by subsequent grammarians. But Sbawayhis method of
probing the relationship, at the syntactico-semantic level, between form
and meaning was continuously eroded by the grammarians attempt
to codify rules, systematize usage, and analyze structure largely on the
basis of formal considerations which govern its constituent elements.
Generally speaking, the continuous shift from Sbawayhis method of
syntactico-semantic analysis culminates in works from the seventh
century A.H. onward, such as Alfiyya commentaries and the extensive
sources (mutawwalt), where pedantic formulae and rigid rules almost
fully replace the vivid and dynamic nature of Sbawayhis analysis which
takes into account both the formal and the semantic aspects of citations
and utterances.1
1
In his Muqaddima (10811084), Ibn Xaldn (d. 808/1406) praises Sbawayhis
Kitb on the grounds that its author did not confine it to the formal rules related to
irb and that it is replete with proverbs and citations from poetry and speech. Those
who study the Kitb are therefore likely to enhance their malaka (natural linguis-
ramzi baalbaki
One of the most significant aspects of Sbawayhis method of grammatical analysis is the pragmatic role he assigns to the speaker (mutakallim), and by extension to the listener (muxtab), as part of the social
interaction which language represents for him. This aspect which lies
at the core of Sbawayhis understanding of language as a form of social
behavior and which embodies his originality, and perhaps the essence
of his value in the history of linguistic ideas, has unfortunately been
severely diminished, if not totally annulled, by later authors. Although
this development is obvious in the works which immediately follow the
Kitbsuch as Mubarrads (d.285/898) Muqtadab and Ibn as-Sarrjs
(d. 316/929) Uslthe shift of focus in linguistic analysis from social
interaction and context of situation to formal considerations becomes
more dramatic in later stages of the history of grammar. Equally unfortunate is that modern scholarship has only recently paid any meaningful
attention to the role which Sbawayhi assigns to the speaker, the listener, and the context in which speech takes place. Carter (2004, 5657)
observes that one of the most striking features of Sbawayhis analysis
is that it concerns itself almost exclusively with language as behavior:
speech is a set of actions, each named according to its intention, e.g.
istifhm asking a question, tatniya making something dual, tanbh
drawing attention to something. He further observes (p. 57) that every
utterance takes place in a context of a speaker . . . and listener and that
this approach places great emphasis on the pragmatic roles of speaker,
listener and context and invites the analyst to propose psychological
explanations of linguistic phenomena.2 Bohas et al. (1990, 38) convincingly argues that, from a typological perspective, grammatical and linguistic systems can be divided into two rough classes: on the one hand,
those which analyze utterances in terms of formal relationships between
their components; on the other hand, those which analyze them in terms
tic ability), although some of them end up mastering grammar as a sina (craft), but
not as a malaka. Contrarily, the books of the later authors (kutub al-mutaaxxirn) are
void of poetry and the speech of the Arabs, and contain nothing but grammatical rules
(al-qawnn an-nahwiyya). Readers of such works, according to Ibn Xaldn, can hardly
be expected to enhance their malaka and can only master the craft. See also Zakariyy
(1986, 23 ff.).
2
See also Carter (2004, 9598) for further discussion of the speakers role and its
significance to Sbawayhis reasoning. It would be particularly interesting to examine
in more detail the effect of the speakers choice (96) and the speakers intention (97) on
utterances in Sbawayhis analysis of speech.
speakers awareness as arbiter of usage
of operations performed by the speaker in order to achieve a specific

effect on the allocutee. Based on this distinction, they maintain that
Sbawayhis approach basically belongs to the latter category, while the
classical grammarians, typically belongs to the former.
It must be stressed that what we shall call the linguistic awareness of
the speaker is, according to Sbawayhi, an essential component of the
competence which is demonstrated in successful interaction with the
listener. By linguistic awareness we refer to what he and a few other
authors perceive as the speakers alertness to the various tools which
language places at his disposal and his ability to use them to decide what
form best expresses the meaning which he intends to convey to the listener. Within this context, the purpose of the present paper is twofold.
First, we shall try to show that although Sbawayhis method, which
assigns a central role to the meaning and to the speakers awareness, was
generally not followed by subsequent authors, some of them have indeed
attempted to restore that central role, albeit from different perspectives.
In a cursory look at the basic elements of the most important of these
attempts, it may be possible to establish that what they have in common
with Sbawayhi is a genuine concern for meaning and for the role of the
speaker. The latter part of the paper looks into how Sbawayhis assumption of the speakers awareness is practically applied in his analysis of
particles whose etymology is essential in determining usage. For this,
we shall examine his analysis of those compound particles which may
be used as single words or split into their constituent elements.
2. Meaning and speakers awareness

The relationship between meaning and the speakers awareness in
Sbawayhis analysis cannot be overemphasized. As a general rule,
whenever Sbawayhi highlights the role of the speaker in the utterance or his competence in producing correct speech, the reader should
expect meaning to be at the center of the authors argument. The Kitb
abounds with examples which support this conclusion, and it is certainly beyond the scope of this paper to investigate the larger number
of formal, semantic and contextual elements which contribute to the
link between meaning and the speakers awareness in Sbawayhis analysis. This notwithstanding, it is important to establish the link between
meaning and one of the most basic concepts in the Kitb, namely taqdr
ramzi baalbaki
(suppletive insertion of elided elements), mainly because, as a grammatical tool, taqdr embraces the formal, semantic and contextual elements that are at the core of correct speech according to Sbawayhi.
Another compelling reason for establishing this link is that it can help
us understand the sharp contrast between Sbawayhis approach and
that of the later grammarians, for whereas they have largely adopted the
formal aspects of his taqdr, they have, more often than not, ignored the
semantic and contextual dimensions of the process of supplying missing
elements to structure. We shall not, however, make any detailed comparison between Sbawayhi and the later grammarians in this respect as
this would require an independent study. Rather, we shall point out two
of the most illuminating principles of the relationship between meaning
and taqdr in Sbawayhis system of grammatical analysis.3
The first principle is that in the process of taqdr, the proposed construction should not contradict the meaning of the original construction, i.e. before the suppletion of the elements which are judged to be
elided. In an earlier study of the harmony which Sbawayhi tries to establish, through taqdr, in several types of constructions (Baalbaki 1979,
714), I discussed his method of breaking up one sentence into two,
both of which share a common feature and thus demonstrate an underlying harmony. For example, the two sentences Zaydan darabtuhu and
a-Abdullhi daraba axhu Zaydan are interpreted, at the level of deep
structure, as *darabtu Zaydan darabtuhu and *a-daraba Abdullhi
daraba axhu Zaydan respectively (Kitb I:81, 102). If a nominal sentence is conjoined to a verbal sentence, as in raaytu Zaydan wa-Amran
kallamtuhu, he intervenes to restore the harmony by supplying a verb to
the nominal sentence, hence the proposed construction *raaytu Zaydan wa-kallamtu Amran kallamtuhu (Kitb I:88). In defending the restoration of a verb to produce a verbal sentence parallel to the first one,
he argues that the introduction of the verb causes no contradiction in
meaning (l yanqud man; Kitb I:8889). Closely related to this argument is his assertion that in utterances which express amr (command)
or nahy (prohibition), the verb may be uniformly elided, as in (idrib)
Zaydan, (l taqrabi) l-asada, and (xalli) t-tarqa (Kitb I:253254).
Sbawayhis discussion of such constructions reveals that the element of
3
Carter (1991, 127128) notes that although Sbawayhi uses the term taqdr exceedingly sparingly, only 24 times in fact (see Troupeau 1976, 167), he does give plenty of
advice on reconstruction without calling it taqdr.
meaning which justifies the elision of the verb is present in the context
of situation since the listener would assume the virtual existence of a
verb of which he is the agent. Consequently, the grammarians intervention to restore this verb would only elucidate, rather than contradict
meaning, based on his understanding of what the speaker has in mind.
The second principle which governs the relationship between taqdr
and meaning represents a further stage in Sbawayhis analysis of constructions. While he remains faithful to the rule that taqdr should not
contradict meaning, he explores the niyya4 (intention) of the speaker to
explain why he may well utter a certain part of the construction although
he has another usage in mind. An example of this is his belief that each
of the two constructions lam tika and l tka has the status (manzila)
of a noun at the level of the speakers niyya, and can thus be interpreted
as lam yakun itynun (Kitb III:2829). The grammatical implication of
this niyya is only fully revealed when the uttered part is virtually replaced
by what the speaker intended to say. Thus, in Farazdaqs line: mamu
lays muslih na aratan * wa-l nibin ill bi-baynin urbuh,
nibin is in the genitive although at the level of the actual utterance it
is conjoined to muslih na, which, being the predicate of laysa, is in the
accusative. According to Sbawayhi, the recurrent use of the preposition bi- with the predicate of laysa (e.g., lays bi-muslih na) reveals the
true intention of the speaker, and hence nibin is in reality conjoined
to a genitive noun which does not feature in speech, but is as valid as
an uttered noun. In other words, the preposition, which the speaker
has in mind, is syntactically valid and operational, and it determines
usage as if it were actually uttered (h att kaannahum qad takallam
bih f l-awwal; Kitb III:29). As far as meaning is concerned, Sbawayhi
asserts that the assumption of the preposition bi- in bi-muslih na does
not alter the meaning5 (l yuayyir al-man) because the preposition is
indeed frequently used with the predicate of laysa. Similarly, Sbawayhi
intervenes in constructions of the type marartu bihi fa-id lahu sawtun
sawta h imrin/surxun surxa t-takl to supply a verb (i.e. yusawwitu,
4
According to Troupeau (1976, 208), the terms naw and niyya occur 13 and 27
times respectively in the Kitb. The concept of intention, however, is often expressed
by much more frequent termsincluding an (136 times), man (891 times), arda
(1361 times), etc. (Troupeau 1976, 150; 102)or by expressions such as kaannahum
ql, tawahham, ixtra, etc.
5
The significance of preserving the meaning of the construction in this line as well
as the meaning of other constructions within the context of Sbawayhis analysis of the
taqdr of an after f is discussed by Baalbaki (2001, 186209, esp. 188).
ramzi baalbaki
yasruxu) before the accusative noun (Kitb I:355f.). The intention of

the speaker is expressed here by the term tawahhama,6 and the actually
uttered words lahu sawtun are said to have the same status as yusawwitu.
Syntactically, the outcome of taqdr is the use of the accusative in sawta
and surxa, whereas at the semantic level, it is implicit that the introduction of yusawwitu and yasruxu does not contradict the meaning because
these verbs share the same root with the nouns they govern.
The study of the relationship between meaning and taqdr in the Kitb
strongly indicates that in his analysis of constructions, Sbawayhi transcends the levels of grammatical correctness and the effect of the operants on case-endings to examine the speakers thinking and the mental
processes involved in the choices he makes. This linguistic awareness on
the part of the speaker becomes a real arbiter of usage and allows him,
for example, to use nibin where niban is expected, or to use sawta
h imrin in spite of the absence of a verb from the utterance. It is this
feature in Sbawayhis analysis that one so much misses in the work of
the later grammarians. In spite of that, some authors were indeed interested in the role of the speaker and the effect of his awareness on the
speech he produces. From this perspective we shall briefly examine the
contribution of two leading figures, Ibn Jinn (d. 392/1002) and Jurjn
(d. 471/1098), and show how their priorities are largely consistent with
those of Sbawayhis. In the light of the speakers awareness and his internal thinking, we shall then examine the analysis proposed by Sbawayhi
for constructions in which the etymology and word-class of certain particles are crucial for the speaker to achieve correct speech and for the
listener to comprehend what is meant.
3. Post-Sbawayhi authors
The first post-Sbawayhi author whose work reflects serious concern for
the speakers awareness is certainly Ibn Jinn. Groomed in the grammatical tradition and himself author of several works which are in full
conformity with the general grammatical theorymost notably Sirr
sinat al-irb, al-Luma f l-Arabiyya, and at-Tasrf al-mulkIbn
6
For the various senses in which the term tawahhum is used in the Kitb, see Baalbaki (1982, 234237). In this particular case of Farazdaqs line, tawahhum refers to the
speakers mental restoration of elided parts in the utterance, resulting in their government of parts actually uttered.
Jinn stands out as a unique scholar whose innovative approach in his

most impressive work, al-Xasis, represents a major step in the direction of determining the theoretical bases of language usage. The traditional phonetic, morphologic and syntactic data becomes in al-Xasis
the subject of study from a methodological and epistemological perspective. Within this framework, the speaker takes a central role in
Ibn Jinns interpretation of linguistic activity. His point of departure
for this is a firm belief in the intuitiveness7 of native speakers of Arabic
as well as in their mental abilities, which are manifest in their production of speech and subtle analysis of a host of linguistic phenomena.
A particularly illuminating example is that in which he tries to examine the linguistic awareness of a Tamm informer of his. When asked
why he would say darabtu axka, and not *darabtu axka, but would
still use axka in daraban axka, the Tamm expresses his astonishment and wittily comments that each of the two expressions has a different perspective (ixtalafat jihat l-kalm; Xasis I:76; cp. 1:250). Ibn
Jinn confidently concludes that the Arab native speakers scrutinize the
syntactic positions of speech elements (adall ay al taammulihim
mawqi al-kalm) and that they knowingly and consciously (an mza
wa-al basra) assign to each element the position and case-ending it
merits. Similarly, he describes vowel mutation which results from the
speakers sense of lightness (istixff ) and heaviness (istitql ) as proof
of keen insight and fineness of perception and observation (li-quwwat
nazarihim wa-lutf istiffihim wa-tasaffuh ihim; I:78).
Ibn Jinns interest in the intuitiveness of Arab native speakers and
in their mental awareness of the various phonetic, morphologic, and
syntactic processes involved in speech is best understood as part of his
attempt to reveal the h ikma which underlies Arabic and which constitutes its intellectual basis. Whether the discussion relates to the reasons
(ilal ), purposes (ard) and intention (qasd ) associated with speech
(I:237, 245), or to the change which forms, etc. undergo due to recurrent usage (II:31), or to the temporal precedence of one part of speech
7
Ibn Jinn often expresses the notion of intuitiveness by derivatives of the root t-b-
(e.g., sun al-br subh nahu f an tabaa n-ns al hd; tahjum bihim tibuhum al
m yantiqna bihi; a-turhu l yuh sin bi-tabihi . . . hd l-qadr; Xasis II:117; III:273,
275 respectively). Cf. also the terms salqiyya and najr (I:76) for intuitiveness, and the
expressions min lutf al-h iss wa-safihi wa-nasat jawhar al-fikr wa-naqihi; quwwat
nafsihi wa-lutf h issihi; I:239; III:75). See also the comments of Suleiman (1999, 6465)
on the intuition of native speakers and the rationality of Arabic within the more general
framework of Ibn Jinn s study of tall (causation).
10
ramzi baalbaki
over others (II:33), or to the onomatopoeic nature of phonemes within

words (II:162, 164), or to the occurrence of two dialects side by side in
speech (III:317), the term h ikma is used as an inalienable trait of the
original wdi (creator) of Arabic and of his logic in deciding on what is
to be used or not and how it should be used. It can thus be argued that
the abstract notion of the wdi finds its practical dimension in the role
of the speaker (or, more generally, the speech community) since Ibn
Jinn ascribes to him an awareness of the linguistic processes that are
the result of the original h ikma. In this respect, Ibn Jinns ideas are very
much in line with Sbawayhis not only because both of them place the
speaker at the center of their linguistic analysis, but also due to their
mutual interest in the communication of meaning as the ultimate aim
of successful speech.
One of the most striking results of Ibn Jinns focus on the speakers
role and of his firm belief that, at all levels of analysis, the linguistic phenomena of Arabic are essentially rational is his discussion of grammatical awmil or operants and their relatedness to meaning. Although he
adopts the traditional division of these operants into two types, lafz (formal; expressed) and manaw (abstract), his interpretation of the awmil
in the light of the speakers role redresses the imbalance between the
two types in the tradition. For example, Jurjn in his al-awmil al-mia
n-nahwiyya (8586; 312), labels ninety-eight of the awmil as lafz and
only two as manaw.8 This imbalance, of course, predates Jurjn, and
it is interesting to note that Ibn Jinns own master, Ab Al al-Fris
(d. 377/987), is reported to have authored a book entitled al-awmil
al-mia (Sezgin 1984, 107). By insisting that all types of amal (rection)
i.e., raf , nasb, jarr and jazm, which cover the three nominal types
(nominative, accusative and genitive) and the three verbal types (indicative, subjunctive and jussive)are in reality ( f l-h aqqa) produced by
the speaker (Xasis I:109110), Ibn Jinn effectively reduces the traditional divide between lafz and manaw operants merely into a didactic
technique that tries to distinguish between amal which is accompanied by an uttered operant and amal which lacks such accompaniment
(I:109). In other words, whether an operant is actually uttered or not,
rection is in all cases the result of the speakers internal thinking and
8
Note also the assertion of some grammarians that no mil may be classified as
manaw unless it cannot possibly be explained as lafz (l yudal il jal al-mil
manawiyyan ill inda taadd
u
r al-laf as-slih ; Suyt, Ham I:159).
11
a reflection of the meaning he intends. Based on this and on the similarly untraditional view that qiys lafz (formal analogical extension)
is not devoid of meaning, Ibn Jinn confidently formulates the general
principle that one cannot but bestow a manaw dimension on what is
lafz, whereas what is manaw may well do without a lafz dimension
(I:111).
It may be useful at this point, at the risk of disrupting the historical
sequence, to bring into the discussion the sixth-century Z hirite scholar,
Ibn Mad (d. 592/1196). Ironically, Ibn Jinns attribution of amal to the
speaker is enthusiastically received by Ibn Mad. He quotes Ibn Jinns
statement fa-l-amal . . . innam huwa li-l-mutakallim nafsihi l li-ay
ayrihi and highlights his use of nafsihi as a corroborative to emphasize
al-mutakallim, followed by the assertion that rection is attributable to
nothing other than the speaker (l li-ay ayrihi; Radd, 77). Ibn Mad
quickly recognizes his odd position as a Z hirite embracing a Mutazilite
view, and thus hastens to resolve the situation by explaining that contrary to the Mutazilites, the doctrine of ahl-al-h aqq (i.e., the Z hirites)
stipulates that case-endings (here, aswt) are in reality produced by
God (innam hiya min fil Allh tal) but metaphorically attributed
to man. Irrespective, however, of this modification and of Ibn Mads
argument that neither the uttered forms of the operants nor their meanings cause rection, his merciless criticism of the grammarians focus on
amal and taqdr rests in part on their disregard for both the speaker and
the intended meaning. In fact, he often refers to the speakers intention
(cf. yanw, 89; yurd, 93) and assesses the relationship between meaning
and taqdr (80, 109). These two aspects of his theory obviously form his
best defense against traditional grammar and firmly place him, albeit
from the different perspective of his Z hirite doctrine, with the few
authors who challenged its shortcomings.
After Ibn Jinn, the most important attempt to reinstate a primary
role to meaning and the speaker who intends it is undeniably that of
Jurjn. Like his predecessor, Jurjn followed in the footsteps of the traditional grammarians in some of his works. In addition to al-Awmil
al-mia n-nahwiyya mentioned earlier, such works include al-Jumal,
most of which is a didactic summary of awmil (chapters 2 to 4), and
a commentary on Friss dh entitled al-muqtasid f arh al-dh .
However, Jurjns two major works in the field of stylistics, primarily
Dalil al-ijz but also Asrr al-bala, represent a major shift from
the traditional syntactical analysis of the grammarians. In his Dalil, he
12
ramzi baalbaki
makes a zealous appeal for placing meaning at the center stage of grammatical analysis and repeatedly argues that nazmwhose narrow sense
corresponds to word order, but which more generally refers to the complex relations among the constituents of a structureis nothing other
than the proper adherence to the discipline of grammar (laysa n-nazm
ill an tada kalmaka l-wad allad yaqtadhi ilm an-nahw, 64). As such,
nazm should aim at what he calls the meanings of grammar (an-nazm
huwa tawaxx man n-nahw, 276, 282, 310, 403404; cf. Asrr, 65).
This means that syntactical rules, which govern the relationships among
the various parts of any utterance, can express the exact meaning
intended by the speaker since speech formation begins in ones mind ( f
n-nafs) and only then are words arranged to formulate the outcome of
this mental process according to a set of syntactical relationships (43f.).
Based on the conviction that the arrangement of meanings (at-tartb f
l-man) is prior to the arrangement of words, and that form is subsidiary to meaning (4445), it follows that any change in syntax is necessarily accompanied by a change in meaning (86f.) The speakers awareness
of the intricacies of syntactic relations hence acts as a virtual arbiter in
his choice of the nazm, which best expresses the intended meaning.
Jurjns theory of nazm is to a large extent a reaction against traditional grammar in which formal aspects acquired greater prominence at
the expense of meaning. Even Sbawayhi, it has been suggested (Baalbaki
1983, 12f.), may have been the target of some of Jurjns critical comments in which he accuses the grammarians of giving too little attention
to meaning. This notwithstanding, Jurjn is surely much closer in spirit
to Sbawayhi than to the later grammarians. Both authors strive to investigate the internal thinking of the speaker and examine its influence on
actual utterances. On a wider scale, Sbawayhi, Ibn Jinn and Jurjn, the
three most original authors in the related fields of nahw (grammar), philology (ilm al-lua), and stylistics (bala) respectively, share the view
that meaning should be the main focus of linguistic analysis. A variety
of conceptssuch as the speakers intuitiveness, competence, intention
and awareness of the tools at his disposalfeature in the works of the
three authors as part of their study of meaning and the mental processes
to which it is related. Unfortunately, however, the three have one more
thing in common, for although their focus on meaning and the speakers
awareness represents the most significant and original aspect of their
contribution, that focus gave way in later writings to an ever-growing
shift towards formal considerations and pedantic formulae which rel-
13
egate meaning and those concepts that are related to it to a much lower
position in their scale of linguistic analysis.
4. The importance of the particles

The importance which Sbawayhi attaches to the speakers competence
and his alertness to the means of successful communication of meaning feature distinctly in his discussion of some particles.9 More specifically, Sbawayhi is interested in proving that the speaker can express the
intended meaning only if he is aware of certain peculiarities of these
particles. The rest of this paper shall deal with etymology of particles
as one such peculiarity, the awareness of which by the speaker, according to Sbawayhi, is crucial for correct speech. It should be noted here
that the later grammarians largely adopt Sbawayhis analysis of particles
and their usage, but this does not at all mean that they have preserved
his method of syntactical analysis or embraced the psychological and
contextual explanations which he emphasizes so much. Rather, they
seem to dwell on the formal side of his analysis, or simply repeat his
views without incorporating them into their own methods of analysis
or developing them in any meaningful way.
Initially, it is essential to introduce the concept of h ikya, which
Sbawayhi applies to those particles which he discusses from the etymological perspective. The original sense of the root h -k-y, to report; to
imitate, is preserved in Sbawayhis use of h ikya for direct speech. For
example, in the construction qla Zaydun Amran xayru n-nsi (III:142),
Zayds words are reported verbatim and therefore qla does not govern
Amrun, which remains in the nominative as in the original utterance.
Derived from this sense is the use of the term to refer to elements that
are intentionally not integrated into the syntactical build of the construction. Commenting on the sentence marartu bi-Zaydin, for example, one
may respond by saying man Zaydin, in the genitive (II:413). This strategy in dialogue can result in the creation of forms that are only used in
9
The term particle is used here in a general sense which includes not only what the
traditional grammarians classify as particles proper (h urf ), but also what they consider
to be verbs or nouns. We shall therefore refer, for example, to amm, interrogative m,
and the halumma as particles although they are traditionally classified as particle, noun,
and verb respectively.
14
ramzi baalbaki
h ikya, as in the use of ayyayni/ayyna or manayni/manna in response

to raaytu rajulayni/rijlan (II:407408). Further removed from the
original sense of the term h ikya is its use to describe proper nouns that
are syntactically whole sentences or phrases and are uttered verbatim
regardless of their grammatical position. These include attested proper
nouns such as taabbata arran (II:269; III:326), baraqa nah ruhu (II:269;
III:326), ba qarnh (II:85; III:207, 326), ibn jal (III:207), and forms
such as xayran minka, driban rajulan and min Zaydin (III:328329),
which are hypothetically proposed as proper nouns and serve as a testing device for the full potential of h ikya. Sbawayhis use of the term
h ikya in connection with proper nouns that are made of more than one
element is most probably what facilitated his generalization of the term
to apply to compound particles, that is, particles which he believes are
etymologically made up of more than one element.
As a preliminary example, we can consider the particle h aytum
which Sbawayhi describes as h ikya. This sense of the term is clearly
distinguished in the Kitb from neighboring concepts, most notably
ism wh id and law. The latter term, law, refers to elements of speech
which are otiose or redundant. The negative particle l, for example,
is considered to be law when it does not signify negation, as in liall
yalama ahlu l-kitbi (Q 57:29), which is interpreted as li-an yalama in
the affirmative (IV:222).10 As for the term ism wh id, Sbawayhi recurrently uses it in connection with the place name H adramawta, which he
cites as an example of two distinct nouns that coalesced into one (e.g.,
the four chapter titles in II:267; III:296, 374, 475). Although at times he
refers to particles which exemplify h ikya as ism wh id (II:417418),
h arf wh id (II:418), or kalima wh ida (III:115), Sbawayhi clearly
demarcates h ikya, as in h aytum, from ism wh id in the sense which
H adramawta embodies. He achieves this by looking into the function of
-m in h aytum and similar particles (see below) and comparing it with
the corresponding element -mawta in H adramawta. The difference lies
in the relationship which each of -m and -mawta has with the preceding
element (III:331). The introduction of -m to h aytu-, he observes, does
not result in the retention of the final damma in h aytu- since h aytam,
with a fath a, is also attested. Moreover, and more importantly, the introduction of -m causes a semantic change since to the adverbial sense of
10
Another example of law is m in mahm (interpreted as m- + -m), and after
conditional in, adverbial id, and pronominal ayy (III:5960).
15
h aytu- is added to the conditional sense which h aytum expresses. Similarly, -m is introduced to an- in amm (amm <an- + -m) to cause
a grammatical change, namely the annulment of the rection of an-.
Because of these semantic and grammatical changes, which its introduction brings about, -m in h aytum and amm is called muayyira.
The difference between -m and -mawta becomes clear semantically and
grammatically. Although Sbawayhi does not explicitly describe the difference, it is implied that h adra- can only have a fath a and not any other
vowel, and that one cannot speak of semantic or grammatical change
caused by -mawta, at least not in the sense which applies to particles.
Furthermore, the term muayyira draws a sharp distinction between
the nature of -m in h ikya and the concept of law which is essentially
antithetical to change.
Based on the above discussion, particles that are classified as h ikya
can be described as typically made up of two constituent elements (e.g.,
h aytum> h aytu- + -m; innam> inna- + -m),11 and the element
which is identified as being introducedeither as a prefix, such as lain laalla, or suffix, such as -m in h aytumis not law. But there is
another dimension to this issue since there are particles, such as exceptive ill, which may seem to be examples of h ikya according to the
above criteria, but which Sbawayhi interprets as non-compound particles. Since Sbawayhi does not use a special term for such particles,
we shall refer to them as non-h ikya or non-compound particles. The
differentiation between the two types is crucial for our understanding
of the role which Sbawayhi assigns for the speaker in using certain particles. Following is an illustrative list of h ikya particles, with the most
essential characteristics which Sbawayhi ascribes to each. These will be
analyzed and then compared to non-h ikya particles.
The h ikya particles12 comprise the following:
11
Only rarely does h ikya involve more than two constituent elements, as in dlika
which Sbawayhi mentions side by side with other h ikya particles that are made up of
two elements (III:332), and laallam in which -m is added to laalla- (IV:221), itself a
compound particle (la- + -alla; III:332).
12
Sbawayhi also uses the term h ikya in connection with personal and demonstrative pronouns (e.g., anta, hd, huli, dka, dlika, etc.; III:332; IV:218), but these are
not relevant to our discussion. Also beyond our scope are particles like lan which may
well be the result of merging two elements (III:5) but which speakers do not normally
recognize as compound particles.
16
ramzi baalbaki
1. innam: h ikya (III:331); h arf wh id (II:418; III:57); laysat -m bilaw (III:57); -m changes inna- by imparting a new meaning to it
(IV:222; cf. II:138); -m changes (i.e., annuls) the rection of inna(IV:222).
2. annam: h ikya (III:331).
3. h aytum: h ikya (III:331); -m does not prevent the final vowel of
h aytu- from being either -u- or -a- (III:331); -m imparts to h aytuthe meaning of conditional (III:331; cf. III:59, 518; IV:222); -m is
muayyira (III:331; cf. III:59), is unlike -mawta in H adramawta
(III:331), and is not law (III:331); h arf wh id (II:418; cf. III:57);
has the status of in, imm (III:59) and ayna (IV:221).
4. imm: h ikya (III:331); -m is attached (madmma) to in- and
may be elided (III:331332; cf. I:266; III:141; IV:222); has the status
of amm in amm anta muntaliqan intalaqtu maaka (III:332);
see also h aytum.
5. amm (in amm anta, as in 4 above); introduction of -m
prevents an- from governing the subjunctive and -m is hence
muayyira (III:331, 332).
6. halumma: h ikya in both H ijz and Tamm dialects (III:332; cf.
I:252; III:529, 534).
7. idm: -m imparts to id- the meaning of conditional (III:56); has
the status of innam and kaannam (III:57); laysat -m bi-law
(III:57); h arf wh id (III:57).
8. lawm and lawl: h ikya (III:333); -m and -l impart a new meaning to law- (III:115; IV:222); -m is muayyira (IV:222223); h arf
wh id (III:115); kalima wh ida (II:180); as excitative (tah dd) particles, they precede only verbs (III:115).
9. lamm (which governs the jussive): -m is muayyira (i.e., it changes
the syntactic properties of lam-; IV:223).
10. kam: h arf wh id (III:116).
11. kad: h ikya (III:332; cf. III:151); ay wh id (III:171).
12. kaayyin: h ikya (III:332; cf. II:171; III:151).
13. kaanna: h ikya (III:332; cf. III:151, 164); ay wh id (II:171).
14. kaannam: h ikya (III:331); h arf wh id (II:418; III:57); laysat -m
bi-law (III:57); -m changes the rection of kaanna- (IV:221; cf.
II:138).
15. laalla: h ikya (III:332).
16. laallam: -m changes the rection of laalla- (IV:221; cf. II:138).
17
17. rubbam: kalima wh ida, like qallam and other similar particles
(abhuhum; III:115); precedes verbs, unlike rubba- (III:116).
18. h abbad: kalima wh ida (II:180).
19. al and am (as interrogative particles): h ikya (III:332).
20. all and hall: h arf wh id (III:5, 115); -l imparts a new meaning to
hal- (IV:222).
21. ill (as a conditional particle): h ikya (III:332).
22. imm (as a conditional particle): h ikya (III:332); see also
h aytum.
23. md and badam: see below.
Sbawayhi spares no effort to demonstrate the coherency of the above
group through cross-references in various and often distant parts of the
Kitb and by using a largely unified terminology to describe the properties of the particles within the group. Furthermore, this coherency is
supported by the characteristics, which its members share particularly
at the level of meaning (nos. 1, 3, 7, 8, 20), syntax (8, 9, 17) and rection (1, 5, 14, 16). This notwithstanding, the particles within this group
considerably vary in their clarity, from the perspective of the speaker,
as to whether they are compound in nature or not. For example, one
can safely assume that idm, kad and laallam are much more easily
recognizable by the speaker as compound particles than, say, halumma
or amm which are far removed from their supposed origin and which
can even be the object of disagreement among grammarians concerning what that origin really is (cf. the case of halumma in Suyt , Ham
II:106107). The most revealing examples, as far as speakers awareness is concerned, are those in which the particle may be either broken down into two elements or used as a single entity. To use the late
grammatical term lamh al-asl13 (lit. recognition of origin), one can say
that the speaker who splits such particles into two elements is aware
of their etymological origin as compound words, unlike the speaker
13
The term lamh al-asl is used by the later grammarians mainly to refer to the
speakers recognition of word class as reflected in usage. One example is that proper
nouns such as H rit and H asan, contrary to the norm, may be prefixed with the definite
article al- since they are originally adjectives. The article is hence called al-allat li-lamh
al-asl (Suyt , Ham I:174175), and the speakers recognition of the adjectival origin
(lamh as-sifa) of such proper nouns justifies its prefixation to them (Ibn Aql, arh
91; Umn, arh I:8586). Due to the obvious similarity between the recognition of an
original word class and the recognition of a particles etymology, the term lamh al-asl is
perfectly applicable to m d, man d, etc. when they are split into two units.
18
ramzi baalbaki
who treats them as single-word particles. Sbawayhi specifically deals

with this issue in a chapter which he designates for md and mand.
Depending on whether d enjoys the independent status of the relative pronoun allad or is integrated with m into a single unit, which
Sbawayhi typically describes as ism wh id (II:417), the speaker respectively uses either the nominative or the accusative in the noun which
follows md. In Labds line al tasalni l-mara m d yuh wilu *
a-nah bun fa-yuqd am dallun wa-btilu, d is given an independent
status equivalent to allad and m by itself acts as interrogative. The corroborative noun nah bun thus acquires the nominative because it modifies m, which has the grammatical function of the subject of a nominal
sentence (i.e., mubtada). Conversely, md could have been treated as
a single interrogative particle which grammatically serves as the direct
object of yuh wilu. In this case, corroborative nah ban would be, like
md which it modifies, in the accusative.
It is clear in Sbawayhis discussion that the speaker has the choice of
using md as a single-word particle or splitting it into its constituent
elements. The mental process which leads up to the speakers decision is
not restricted to the deconstruction of the particle into its elements or
its retention as one unit. It also has to do with the syntactical ramifications of the speakers choice since the case-endings of the noun which
follows the particle have to be consistent with that choice.14 The speakers
competence is thus demonstrable at two levels, namely his awareness
of the nature of md as a compound particle which admits two possibilities of usage, and his confirmation of this awareness by observing
the syntactical implications of each possibility. Furthermore, Sbawayhi
takes into consideration the listeners interpretation of the speakers use
of the particle in order to prove that successful communication also
depends on the listeners awareness of how the speakers choice between
two possibilities of usage affects his own response. The correct response
to m d raayta, for example, would be matun h asanun in the nominative since m itself is mubtada and hence nominative. On the other
hand, md raayta is equivalent to m raayta since the verb governs
14
Note that Sbawayhi does not refer to the fact that, in actual speech, stress may be
an essential factor in differentiating md from m d. As a single word, md would
normally receive stress on its first syllable. Conversely, the separation of the two elements would be indicated by a stronger stress on d than m, perhaps to underline the
likely demonstrative function of d. In short, the difference in meaning between the two
options can be best demonstrated by translating md faalta and m d faalta as What
have you done? and What is this that you have done? respectively.
19
the single-word interrogative particle which precedes it in both cases,

and the answer should thus be in the accusative as in md anzala rabbukum ql xayran (Q 16:30).15
As part of his discussion of md/m d and mand/man d,
Sbawayhi mentions two other particles which the speaker can use as
a single word or as two separate elements. These are kaannam and
h aytum (II:418; cf. also badam II:139). Sbawayhi often examines the
function of the suffixed -m in these and similar particles (see detailed
list above), including annulment of rection (cf. innam) and shift in
the part of speech which the particle precedes. For instance, rubba and
qalla can only precede nounsas one cannot say *rubba yaqlu and
*qalla yaqlubut the suffixation of -m causes the new particle to
precede verbs only (axlashum li-l-fil; III:115). In such cases, correct
speech is contingent on the speakers reanalysis16 of such compound
particles with respect to their constituents as this would ensure proper
distinction between utterances like qallam yaf alu, innam yaf alu,
kaannam yaf alu, etc. and their counterparts with relative m; i.e.,
qalla m yaf alu, inna m yaf alu, kaanna m yaf alu, etc.17 For his
part, the listener is obviously expected to differentiate between the two
types for successful communication to take place.
As we pointed out earlier, there is a group of particles which may
seem to be examples of h ikya but which Sbawayhi interprets as simple
or non-compound (i.e., non-h ikya) particles. In order to highlight the
contrast between these particles and the h ikya particles, Sbawayhi discusses both types side by side in one of his chapters (III:331332). It
is important, however, to determine at which level this contrast exists.
By classifying, for example, the exceptive particle ill with non-h ikya
particles, Sbawayhi does not want to specifically deny that it is historically the result of a merger between in- and -l, in contrast with the
15
Sbawayhi cites the reverse usage, i.e., the accusative after m d and the nominative after md. But although this is grammatically explicable, he asserts that to use the
nominative after m d and the accusative after md is the most appropriate manner
of response (wajh; aqrab il an taxud bihi; II:418419).
16
Cf. the role of reanalysis in the use of tlam as discussed by Anghelescu (2004,
115116).
17
To illustrate this distinction in the case of innam, for example, we can replace it
by inna m in some Qurnic verses where this is syntactically possible. The resulting
constructions are grammatically sound, but they obviously differ in meaning from the
original constructions. Cf. innam/inna m ttuhu al ilmin ind (Q, 28:78); waman yabxal fa-innam/fa-inna m yabxalu an nafsihi (Q, 47, 38); innam/inna m
tuxzawna m kuntum tamalna (Q, 66:7).
20
ramzi baalbaki
conditional particle ill (see no. 21 in the list of h ikya particles). But
irrespective of whether or not he actually differentiates between the two
types of ill at the etymological level, it is obvious that he intends to
establish the contrast between h ikya and non-h ikya particles at the
level of the speakers awareness of their compound versus non-compound nature. In other words, the fact that linguistic analysis may well
prove the compound nature of exceptive ill and similar particles is of
no interest here to Sbawayhi since it has little to do with the speakers
awareness of their etymology. Based on the fact that proper usage and
comprehension of pairs of utterances which have two apparently identical particles is dependent on the distinction between the simple versus
the compound nature of each particle in a pair (e.g., exceptive versus
conditional ill), we can interpret Sbawayhis interest in such pairs as
part of his overriding interest in the competence of the speaker in correctly communicating the intended meaning to his listener. As for why
the particles which he describes as non-h ikya are considerably fewer
in number than the h ikya particles, two reasons may be suggested. The
first is Sbawayhis faith in the speakers ability to break down compound
particles into their componentsalthough some of them are more easily recognizable as compounds than othersand thus arrive at what
the later authors call lamh al-asl. The second reason may have to do
with the grammarians own failure, due to their largely synchronic and
non-comparative approach, to ascertain the compound nature of a large
number of particles (e.g., lam, layta, lta, laysa, kayfa, ayna, etc.).
Within the above confines, the particles which Sbawayhi explicitly
describes as non-h ikya (III:332) are the following:
1. ill: As an exceptive particle, it has the status of difl (oleander)i.e.,
it is a simple or non-h ikya particleand stands in contrast to the
conditional particle ill (see h ikya particles, no. 21).
2. h att: Like ill, it has the status of difl. It should be noted that h att
is the only member of Sbawayhis group of non-h ikya particles
which is apparently not made up of two elements and which has no
counterpart in the h ikya group. We list it here, nevertheless, for the
sake of completeness.
3. amm: As an inceptive particle in constructions like amm Zaydun fa-muntaliqun, it is a non-h ikya particle which has the status
of arw (the like of a thing) and stands in contrast to amm as in
amm anta muntaliqan intalaqtu maaka (see h iyka particles,
no. 5).
21
4. al: When used inceptively, it is a non-h ikya particle which has the
status of words like qaf (back) and rah (quern), in contrast to interrogative al which is a h ikya particle (no. 19 above).
5. am: Like al, it is a non-h ikya particle when used inceptively, in
contrast to interrogative am (no. 19 above).
The distinction between non-h ikya particles and their h ikya counterparts in context is, of course, dependent on the speakers competence
in using the appropriate syntax, stress intonation, etc. But proper communicationas is implied in Sbawayhis textis also to a large extent a
function of the listeners ability to identify each of two identical particles
one of which is simple and the other is compound. This etymological
distinction is to be assumed in the listeners comprehension of sentences
in which these particles appear. Cf., for example:
l tatn fa-tuh additan ill izdadn fka rabatan (III:32) and wa-ill
tafir l wa-tarh amn akun min al-xsirna (Q 11:74; note that the
choice of these two examples is ours, since Sbawayhi does not provide
contrasting examples for ill);
amm Zaydun fa-muntaliqun (III:332), and amm anta muntaliqan
intalaqtu maaka (III:332); and
al innahu dhibun (IV:235) and al rajula imm Zaydun wa-imm
Amrun (I:289).
The fact that the distinction between formally identical particles in
the above sentences on the basis of their simple or compound nature
is accompanied by certain syntactic peculiarities is not without parallels in the grammatical corpus. The mandatory use of f after simple
or non-h ikya amme.g., amm anta fa-muntaliqun versus amm
anta muntaliqan intalaqtu maakais strikingly similar to the mandatory use of lm in constructions like in kna la-slih an (cf. III:104).
Although Sbawayhi does not use a special term for this lm, it acquired
later the name of al-lm al-friqa because its presence indicates that in
is not a negative particle but the lightened form of inna, known as in
al-muxaffafa. In other words, lm contributes to the distinction of two
formally identical particles both of which, unlike the case of amm,
are not compound. In Sbawayhis own words, this lm is mandatory
(alzamah l-lm) in order that the lightened form of inna not be confused with in which has the status of m (li-all taltabis bi-in allat hiya
bi-manzilat m allat tanf bih; II:139).
22
ramzi baalbaki
5. Conclusion
Sbawayhis interest in the speakers awareness of the etymology of particles and its effect on his usage and on the listeners response is yet
another proof of his method of grammatical analysis which aims at
examining the mental operations which the speaker performs and at
determining the formal and semantic effects of these operations. As the
study of md and mand versus m d and man d in the Kitb shows,
successful communication between the speaker and the listener (who in
turn assumes the role of speaker) is measured by their ability to associate each of the two possibilities of usage (i.e., simple versus compound
particle) with the formal and semantic aspects which pertain to it. In
this particular case, what the grammarians refer to as etymology and
is intuitively discerned both by the speaker and the listener represents
the crucial factor upon which successful communication hinges. The
close association between form, meaning, and speakers awarenessas
exemplified in Sbawayhis analysis of h ikya and non-h ikya particleshas certainly been degraded, and at times even totally obliterated by subsequent grammarians. On a wider scale, Sbawayhis model
of linguistic analysis which, like that of Ibn Jinns and Jurjns, largely
rests on exploring the dialectics or interplay, so to speak, between
form and meaning, gave way to an alternative model which is heavily
tipped in favor of formal considerations. Proponents of this model consequently failed to delve, as did Sbawayhi, into the mind of the speaker
in order to pursue the complex processes which result in his choice of
the form that most appropriately expresses the intended meaning and is
expected to have the desired effect and elicit the correct response from
the listener.
6. References
6.1 Primary sources
Ibn Aql, arh = Bah ad-Dn Abdallh Ibn Aql, arh Ibn Aql al Alfiyyat Ibn Mlik.
Ed. by Ramz Munr Baalbak. Beirut: Dr al-Ilm lil-Malyn, 1992.
Ibn Jinn, Xasis = Ab l-Fath Utm
n Ibn Jinn, al-Xasis. Ed. by Muhammad Al anNajjr. Cairo: Dr al-Kutub al-Misriyya, 195256.
Ibn Mad, Radd = Ab l-Abbs Ahmad b. Abd ar-Rahmn Ibn Mad al-Laxm, arRadd al n-nuh t. Ed. by awq D ayf. 3rd ed. Cairo: Dr al-Marif, 1988.
Ibn as-Sarrj, Usl = Ab Bakr Muhammad b. Sahl Ibn as-Sarrj, al-Usl f n-nahw.
Ed. by Abd al-H usayn al-Fatl. Beirut: Muassasat ar-Risla, 1985.
23
Ibn Xaldn, Muqaddima = Waliyy ad-Dn Ab Zayd Abd ar-Rahmn b. Muhammad
Ibn Xaldn, al-Muqaddima. Beirut: Dr al-Kitb al-Lubnn, 1956.
Jurjn, Asrr = Ab Bakr Abd al-Qhir b. Abd ar-Rahmn al-Jurjn, Asrr al-bala.
Ed. By Helmut Ritter. Istanbul: Government Press, 1954.
Jurjn, Awmil = Ab Bakr Abd al-Qhir b. Abd ar-Rahmn al-Jurjn, al-Awmil
al-mia n-nahwiyya f usl ilm al-Arabiyya, bi-arh Xlid al-Azhar. Ed. by al-Badrw
Zahrn. 2nd ed. Cairo: Dr al-Marif, 1988.
Jurjn, Dalil = Ab Bakr Abd al-Qhir b. Abd ar-Rahmn al-Jurjn, Dalil al-ijz.
Ed. by Muhammad Rad Rid. Repr. from the Cairo edition, Beirut: Dr al-Marifa,
1981.
Jurjn, Jumal = Ab Bakr Abd al-Qhir b. Abd ar-Rahmn al-Jurjn, al-Jumal. Ed. by
Al H aydar. Damascus: Dr al-H ikma, 1972.
Mubarrad, Muqtadab = Ab l-Abbs Muhammad b. Yazd al-Mubarrad, al-Muqtadab.
Ed. by Muhammad Abd al-Xliq Udayma. Cairo: Dr at-Tahrr, 196568.
Sbawayhi, Kitb = Ab Bir Amr b. Utm
n Sbawayhi: al-Kitb. Ed. by Abd as-Salm
Muhammad Hrn. Cairo: al-Haya l-Misriyya l-mma, 1977.
Suyt, Ham = Jall ad-Dn Ab l-Fadl Abd ar-Rahmn b. Ab Bakr as-Suyt , Ham
al-hawmi arh jam al-jawmi f ilm al-Arabiyya. Cairo: Matb aat as-Sada, 1327
A.H.
Umn, arh = Ab l-H asan Al b. Muhammad al-Umn, arh al-Umn al
Alfiyyat Ibn Mlik al-musamm Manhaj as-slik il Alfiyyat Ibn Mlik. Ed. by
Muhammad Muhy d-Dn Abd al-H amd. Cairo: Dr al-Kitb al-Arab, 1955.
6.2
Secondary sources
Anghelescu, Nadia. 2004. La langue arabe dans une perspective typologique. Bucharest:
University of Bucharest.
Baalbaki, Ramzi. 1979. Some Aspects of Harmony and Hierarchy in Sbawayhis Grammatical Analysis. Zeitschrift fr arabische Linguistik 2, 722.
. 1982. Tawahhum: An Ambiguous Concept in Early Arabic Grammar. Bulletin of
the School of Oriental and African Studies 45, pt. 2. 22344.
. 1983. The Relation between nahw and bala: A Comparative Study of the Methods of Sbawayhi and Jurjn. Zeitschrift fr arabische Linguistik 11. 723.
. 2001. Bb al-f [ f + Subjunctive] in Arabic Grammatical Sources. Arabica 48.
186209.
Bohas, Georges, J. P. Guillaume and D. E. Kouloughli. 1990. The Arabic Linguistic Tradition. London: Routledge.
Carter, Michael, G. 1991. Elision. Proceedings of the Colloquium on Arabic Grammar,
Budapest 17 September 1991, Kinga Dvnyi and Tams Ivnyi: eds. Budapest: Etvs Lornd University and Csoma de Krs Society. 12133.
. 2004. Sibawayhi. London and New York: I.B. Tauris and Oxford University Press.
Sezgin, Fuat. 1984. Geschichte des arabischen Schrifttums. IX. Grammatik. Leiden: E.J.
Brill.
Suleiman, Yasir. 1999. The Arabic Grammatical Tradition: A Study in tall. Edinburgh:
Edinburgh University Press.
Troupeau, Grard. 1976. Lexique-index du Kitb de Sbawayhi. Paris: Klincksieck.
Zakariyy, Michel. 1986. al-Malaka l-lisniyya f Muqaddimat Ibn Xaldn. Beirut: alMuassasa l-jmiiyya li-d-dirst wa-n-nar wa-t-tawz.
PRAGMATICS AND CONTRACTUAL LANGUAGE IN

EARLY ARABIC GRAMMAR AND LEGAL THEORY
Michael Carter
University of Sydney
It is well known that Arabic grammar and Islamic law enjoy a peculiarly
close relationship. The two sciences are united by a common purpose,
to control linguistic and general behaviour respectively, and they share
a common methodology, namely the inductive derivation of rules from
a linguistic corpus and the deductive application of these universal rules
to particular acts of the Muslim. Where they differ is in their sources.
Grammar relies on the natural, worldly speech of a select range of
human speakers (Bedouin), law on the inspired texts of the Qurn and
the H adt, which are supernatural in origin and holy in status. All legal
systems are linguistic codes of one sort or another, spoken or written,
but the total dependence of Islamic law on a finite body of revealed and
prophetic language is unique. Its modern secular analogue is the type of
law which is derived from a written constitution, and here too, the law
has to be discovered by an essentially linguistic process, whereby there
is often disagreement over the presumed intentions of those (invariably
dead) who framed the document.
The development of a method for interpreting the language of the
Qurn and H adt took several centuries, and at risk of oversimplification it can be said that the two sciences of grammar and law, aided by
imported Aristotelian logic, leap-frogged each other in an evolutionary
series, where the advances of one made further progress possible in the
other. This paper will review the general similarities in the approach
to language in early grammar and law, especially Sbawayhis intuitive
pragmatism (here in the non-technical sense) and his awareness of the
legal implications of grammatical form. There follows a brief account of
some grammatical/legal problems discussed in an intermediate phase
in the 3rd4th/9th10th centuries, and the paper concludes by listing a
number of features of legal methodology which can be linked with ideas
first noted in Sbawayhi, but which only acquired their fully developed
form after the maturing of us l al-fiqh as a discipline.
26
michael carter
Islamic legal hermeneutics proceeds from the axiom that, regardless of its supernatural origin, the language of the holy texts conforms
entirely to the principles of human discourse. Two short quotations, one
from Sbawayhi (d. ca 180/796) and one from a-fi (d. 204/820, and
probably born about the same time as Sbawayhi) will testify that this
notion was well defined at the very birth of the sciences of grammar and
law :
Gods servants were spoken to in their own speech and the Qurn came
down in their language and according to what they mean
ibd[u] llhi kullim bi-kalmihim wa-ja l-Qurnu al luatihim wa-
al m yanna (Kitb Der. I:139/Bl. I:167).
In almost identical wording from a-fi (the original Arabic could not
be checked: this is from Khadduri 1987, 94, and looks very like a quotation from or paraphrase of the Kitb):
God has addressed his book to the Arabs in their tongue in accordance
with the meanings known to them.
The context in a-fis case was the dispute over the possibility of foreign words in the Qurn, while for Sbawayhi it was a syntactical issue
of indefinite expressions such as salmun alayka; although a definite
as-salmu would be expected, the indefinite is an old-established Arab
usage that must be accepted, especially when it appears in the Qurn.
A-fi is regarded as the first legal theorist to give proper weight
to the linguistic aspects of the law, since which time both grammarians and lawyers have shown themselves to be remarkably strict and
uncompromising in subordinating the language of God to the linguistic conventions of ordinary Arabic. The latter, for Sbawayhi, was a dialogue between speaker and listener, both being required to conform to
what are basically ethical criteria to speak well (h asan) and rightly
(mustaqm). For the lawyers there could be no dialogue with God, only
the contemplation of the written record of what he and his Prophet said,
but over time they evolved a system of interpretation in which they
played the role of silent listeners to a speaker of their own tongue, under
the same conditions as natural speech.
For this they constructed an elaborate hermeneutical mechanism
(usl al-fiqh) which, as documented by Ali (2000), exhibits an impressive congruence in many details with the modern branch of linguistics
pragmatics and contractual language in early arabic
27
known as Pragmatics (henceforth with capital P as a school of thought).

The thesis of the present article is that a form of Pragmatics can be discerned in Sbawayhis analysis of speech (kalm) long before the the
usl al-fiqh were codified, and, further, that this approach to language
passed through at least two stages before it grew into the explicit type of
Pragmatics seen in Alis sources.
For the purposes of this article usl al-fiqh will be taken in the broad
sense of the principles of legal argumentation in deriving law from the
texts. As for Pragmatics, here is a recent definition which so closely
reflects the usls position that it might have been written by one of
them, or even by Sbawayhi:
The study of language from the point of view of the users, especially of the
choices they make, the constraints they encounter in using language in a
social interaction, and the effects their use of language has on the other
participants in the conversation [. . .] including aspects of deixis, implicatures, presuppositions, speech acts and discourse structure. [. . .] It has
been characterized as the study of the principles and practice of conversational performancethis including all aspects of language usage, understanding and appropriateness (Crystal 2000, s.v. Pragmatics).
Readers of the Kitb will find all these notions very familiar, and some
will be illustrated below. But first it is necessary to introduce the essential elements of Pragmatics as laid down by Grice.
Grice (1989) treats speech (writing, significantly, does not fit easily
into his model, nor into Sbawayhis, see below) as a cooperative activity
that is both purposeful and rational, in which the participants understand each other by a logically structured process of conversational
implicature. Speech is a quasi-contractual matter, governed by four
maxims (Grice 1989, 26f):
The maxim of Quantity :
1. Make your contribution as informative as is required (for the current purposes of the exchange).
2. Do not make your contribution more informative than is required.
The maxim of Quality :
1. Do not say what you believe to be false.
2. Do not say that for which you lack adequate evidence.
The maxim of Relation:
Be relevant.
28
michael carter
The maxim of Manner:

1. Avoid obscurity of expression.
2. Avoid ambiguity.
3. Be brief (avoid unnecessary prolixity).
4. Be orderly.
All these maxims can be deliberately ignored or flouted, sometimes for
personal reasons, e.g. a desire to mislead or deceive, sometimes for rhetorical purposes, e.g. exaggeration, irony, implying what one is reluctant
to say explicitly, intentional ambiguity etc. There can be and often is a
difference between the words actually spoken (sentence meaning) and
what is really meant (utterance meaning).
In Larcher (1998) an earlier form of Pragmatics is applied to the
category of in, i.e. performative utterances of the type qabiltu hd
n-nikh a I hereby accept this marriage [proposal]. This is the Pragmatics of J. Searle and J. L. Austin of How to do things with words fame (bibliographical details in Larcher), out of which Grices maxims were later
elaborated. However, Grice has been invoked at least twice over the past
few years in articles on Arabic linguistics. Moutaouakil, who quotes the
above maxims in full (1990, 233), deals only with the Pragmatics of the
relatively late scholastic author as-Sakkk (d. 626/1229) and will not
detain us further, except to remark, following Simon (1993, 15) (where
further sources), that it was not until grammar and usl al-fiqh had
themselves achieved systematic perfection that an independent science
of rhetoric could emerge, effectively the last of the Islamic sciences to
appear.
The other is a direct comparison between Grice and Sbawayhi made
by Buburuzan. She interprets the constructions nima l-rajulu abdullhi
What a fine man Abdullh is! and abdullhi nima l-rajulu Abdullh,
what a fine man [he is]! (1993, 424, Kitb Der. I:259f/Bl. I:300f) as
compound expressions where the second part presupposes a question
from the listener. The first part, she says, has violated Grices maxim of
quantity and requires completion in answer to the question who? or
what about him?1
1
It is relevant here to recall a similar analysis of the syntax of the zaydun ja abhu
structure, Zayd, his father came, which Bravmann (1953, 136) explained as deriving from a self-answered question, What about Zayd? His father came; Sbawayhi loc.
cit. actually compares the nima construction to this same type, viz. abdullhi dahaba
axhu Abdullh, his brother has gone. Bravmanns Isolated Natural Subject is not
29
Sbawayhi makes more or less the same statement about the syntax
of marartu bi-rajulayni muslimin wa-kfirin/muslimun wa-kfirun I
passed by two men, one Muslim, one pagan, where the speaker will
choose between the oblique case ( jarr) in adjectival agreement or the
independent (raf ) case as if muslimun wa-kfirun were predicates of
elided subjects, because, as Sbawayhi observes (Der. I:182/Bl. I:214,
and see Der. I:215/Bl. I:252 for another, similar case), the speaker tries
to anticipate the question the listener might pose, either, what sort of
men? or, who were these two men?
Like Grice, Sbawayhi is very concerned with the listeners role in
conversation, and there are many linguistic events in which the listener
influences the speakers choices. In what would be a neither [. . .] nor
[. . .] construction in English, it is qabh . i.e. structurally incorrect, to
say marartu bi-rajulin l frisin I passed by a man who was neither
a knight without completing it with wa-l ujin nor a valiant person or the like, because it is an answer to someone who asked youor
whom you have put in the status of having askedwhether you passed
by a knight or a valiant man (Der. I:313/Bl. I:358, translators italics).
By the same token a listener who answers no to the disjunctive question Is it Zayd who is with you or Bir? when one of them is known
to be there, has broken the communicative contract so gravely that his
answer is classified as muh l [morally] wrong, [semantically] absurd,
i.e. an utterance which is self-contradictory and therefore meaningless
(Der. I:432/Bl. I:483). As if to reinforce the importance of the listener,
Sbawayhi comments that in talking to oneself, e.g. hall af alu why
dont I do this, you are like the listener (Der. I:114/Bl. I:136, kunta
fhi ka-l-muxtabi).
There is even a discussion of what looks like body-language when
Sbawayhi describes how, on seeing the figure of an unknown person,
some sign (ya, the same word as for the verses of the Qurn!) appears
by which you identify him, so you exclaim, Abdullh! Good Lord!
Not only that, the same elliptical exclamation (that is, a predicate without a subject) can be uttered when the sign by which you identify a
person is his voice or perfume, or simply what you hear said about him
(Der. I:240/Bl. I:279).
intrinsically Gricean, but it accords well with the eminently Pragmatic principle stated
by Sbawayhi (Kitb Der. I:346/Bl. I:394, a notion he acquired from his teacher alXall b. Ahmad), that every subject must have a predicate because the listener is expecting it.
30
michael carter
It will be apparent that as well as the speaker and listener, the physical
context is also linguistically relevant. Both Grice and Sbawayhi like to
present their data in the form of utterances set in a described situation.
In Grice it is undoubtedly fictitious and often involves broken-down
cars or sherry parties, in Sbawayhi it probably reflects actual observation, e.g.:
an example of the suppression of the verb which could be expressed in
normal usage is when you see a man who has just returned from a journey
and you say, the best of returns xayra maqdamin, [. . .] where the dependent (nasb) form is as if [the speaker] had [syntactically] constructed it on
the basis of having said may you return, viz. qadimta xayra maqdamin,
and even though he was not heard to say this expression, the arrival of the
other person and the sight of him have the same [linguistic] status as the
speakers saying qadimta (Der. I:114f./Bl. I:136f.).
The similarity with Gricean Pragmatics is unmistakable here: not only

does the speaker engage in a cooperative activity with the listener in a
real context, but that context itself can become an active constituent in
the grammatical form of the utterance (cf. Carter 2002, 7, where the
above item is discussed).
In the same way the vocative particle can be dispensed with when
the listener is standing right in front of the speaker (Der. I:104, 274/
Bl. I:125, 316); the object of a blessing does not have to be indicated
if the intended recipient is obvious from the context (Der. I:131/Bl.
I:157)the speaker however remains free to name the recipient for purposes of emphasis). In a rather violent scenario (Der. I:107/Bl. I:128)
the speaker can dispense with the verb and merely shout the persons
name if he sees someone about to be killed or being abused. This is common in warnings, e.g. al-jidr(a) [mind] the wall!, al-asad(a) [dont
go near] the lion!, and the cry of at-tarq(a), at-trq(a) [get out of] the
way! [Get out of] the way!2
A striking feature of the Kitb is the sheer quantity of commercial and
contractual talk, much of it admittedly trivial, though it does tell us how
preoccupied the Bedouin in the Mirbad were with the price of sheep,
camels and wheat.3 But many items are strictly legal in form and con-
2
These are always printed with dependent (nasb) case endings, however, in the circumstances they are bound to be in pausal form.
3
The Mirbad was not only a market but also a place where H adt scholars came to
check their vocabulary (EI 2, art. Mirbad by C. Pellat), and this may be one of the
reasons why Sbawayhi came to Basra to study tr or H adt. As it happens mirbad
31
tent, involving contracts, debts, sale of goods and property, testimony,

deposits, theft and the activity of fiqh itself. The following list appeared
in Carter 1972, 90ff:
inna f alfi dirhamin la-madrabun (Der. I:99/Bl. I:119); baya l-malat
l ahda wa-l aqda (Der. I:115/Bl. I:137); tabu d-dra h addun minh
kad wa-h addun minh kad (Der. I:137/Bl. I:165: note that the sale of
undefined property was not legal, at least among the H anafs, see Hamilton
1870, 257), lahu ilmun ilmu l-fuqahi (Der. I:151/Bl. I:181); lahu alayya
alfu dirhamin urfan (Der. I:160/Bl. I:190); bitu -a tan wa-dirhaman, qmartuhu dirhaman f dirhamin, bituhu dr diran bi-dirhamin,
etc. (Der. I:165/Bl. I:196); hd dirhamun waznan and others (Der. I:235/
Bl. I:275); inna alfan f darhimika bdun (Der. I:245/Bl. I:285); kam
minkum hidun al fulnin (Der. I:256/Bl. I:297); alayhi aru kalbayni
daynan (Der. I:257/B I:298); al-wadatu ayyuh l-biu (Der. I:284/Bl.
I:326); qadiyyatun wa-l Ab H asanayni = the caliph Al, (Der. I:310/Bl.
I:355); ti l-amra l yaqtau l-lissa (Der. I:402/Bl. I:435); amm juhda
ray (Der. I:418/Bl. I:470). The pre-emptive in a llhu to avoid being
bound by an oath may also be mentioned here: it is called istitn both by
Sbawayhi and in later legal terminology (Der. I:399/Bl. I:448, see also
Carter 1972, 90 n. 2).
Additional legal or commercial material which has come to light since

includes (and is not exhaustive: marginal items have been ignored such
as have you barley, wheat or dates?, though this may well reflect a mercantile context, Der. I:434f/Bl. I:485f):
Placement of merchandise (Der. I:64/Bl. I:76: even in those days it was
better to buy from the top of the pile rather than the bottom!); expressions of time (important for contracts, Der. I:9093, 176/Bl. I:110114,
208); litigation (Der. I:94/Bl. I:114); price variations (Der. I:122, 170/Bl.
I:147, 200); profit, sadaqa and zakh, rendering accounts (Der. I:165f/Bl.
I:196f); weights and measures (Der. I:141f, 183/Bl. I:216, 292f); substitution of goods (Der. I:245/Bl. I:285); giving testimony (Der. I:421/
Bl. I:473); default masc. for mixed genders, i.e. slaves (Der. II:180/Bl.
II:174).
One specifically contractual type of utterance will be discussed here in

some detail, involving the way prices are stated. In a long analysis (over
four chapters, 9295, Der. I:165168/Bl. I:1958) Sbawayhi explores
how deals are struck, beginning significantly with the grammar of
kallamtuhu fhu il fiyya I spoke to him face to face and byatuhu
is mentioned once in the Kitb (Der. II:265/Bl. II:248) but only as an example of the
mif al pattern.
32
michael carter
yadan bi-yadin I traded with him hand in hand (i.e. for cash), where
the dependent (nasb) forms are obligatory, since they are not literal, but
simply mean immediately, on the spot, no matter how physically close
the listener might be. In other words the legal (utterance) meaning is
different from the overt (sentence) meaning: by saying these words in
this form a legal obligation is created regardless of their literal meaning.
This leads to a whole string of commercial expressions, bitu -a tan
wa-dirhaman, qmartuhu dirhaman f dirhamin, bituhu dr diran
bi-dirhamin, bitu l-burra qafzayni bi-dirhamin, axadtu zakta mlihi
dirhaman li-kulli arbana dirhaman, bayyantu lahu h isbahu bban
bban, in all of which the phrase indicating the unit and price (scil. tan
wa-dirhaman one sheep and one dirham) must be stated in its entirety,
otherwise the meaning will not be valid l yasih h u l-man, i.e. legally.
By this Sbawayhi means only in the case of a contractual intent, since
the shortened expressions omitting the price are still meaningful but not
in any contractual sense, e.g. bitu tan tan I hereby sell my [collective] sheep, sheep by sheep, bituhu dr diran I hereby sell him
my house, one cubit, but this would lead the listener to believe that you
were selling your sheep one at a time or that your house was only one
cubit in size, and so on.
Nevertheless, as he observes, the price or the unit are frequently
omitted in ordinary speech, and people will say kna l-burru qafzayni
the wheat was [for sale] at two bushels, omitting the price, or al-burru
bi-sittna the wheat is for sixty [dirhams], omitting the unit of quantity.
They do this, Sbawayhi says, because in the first instance they know
in their hearts ( f sudrihim) that bi-dirhamin is meant and that the
dirham is the standard price unit (allad yusaaru alayhi), so it is as if
they were answering the question, How much you get for a dirham?,
while in the second they and the listener both know what they mean,
as if someone had asked What is the price of a load? and received the
answer The load is [for sale] at sixty [dirhams] (Der. I:166/Bl. I:196).
Sbawayhi advises us to follow the practice of the Arabs in this, though
al-Xall complicates the picture somewhat by pointing out alternative
formulations.
Sbawayhis Pragmatic approach is self-evident here: he puts the
conversation in a real-life setting, which assumes all the Gricean maxims: he distinguishes between utterance and sentence meaning, and he
accounts for the grammatical features of the expressions in terms of the
extralinguistic situation and the intentions of the participants. If a statement such as yajzu an taqla bitu d-dra dirun bi-dirhamin (Der.
33
I:166/Bl. I:197) were taken out of context there would be no way of

knowing whether it was from a legal work, scil. it is [legally] permitted4 to utter [the binding contractual formula] I hereby sell etc., or a
grammatical work, scil. it is allowed [by the rules of grammar] to say, I
hereby sell etc. 5 Only when Sbawayhi goes on to discuss the possibility
of the independent (raf ) case in dirun here (as advocated by al-Xall)
do we learn that the topic is syntax. The goal, however, is to determine
the legal consequences of the syntactic options. Why else would he raise
the matter at all?
In the 3rd/9th century a new weapon enters the methodological
armory, the concept, borrowed from Aristotle, that science has to be
logically structured and its enquiries carried out along systematic, that
is, logical lines. For about a century and a half there was a great deal of
experimentation before the sciences reached the point where they could
be defined and classified, notably in such works as the Mafth al-ulm
of al-Xwrazm, written between 366/976 and 387/997. Two debates
from that period will be mentioned here as evidence of the transitional
stage in the development of grammar and law.
Ibn Walld (d. 332/943) reports a dispute which originated with
Mubarrad (d. 285/898) over the semantic status of commands and prohibitions, whether they were logically complementary, i.e. whether a
command is equivalent to a prohibition from doing the opposite and
a prohibition is equivalent to a positive command to do the opposite.
Ibn Wallds wording (Intisr 42) is: kullu amrin amarta bihi fa-anta f
l-man nhin an xilfihi [. . .] fa-id nahayta amarta bi-xilfihi hence
the positive exclamation h adaraka beware! has the negative meaning,
do not approach. The probable Stoic origins of this controversy (cf. Versteegh 1977, 181) are not the main point of interest here:6 what is important is to observe how the intellectual environment stimulated such
arguments while the usl al-fiqh were still in gestation. Al-Mubarrad
is among the earliest of the grammarians to reveal the direct influence
The old definition of jiz by Bergstrer (1935, 32) is still the most informative:
zulig im moralisch-religisen und zugleich rechtlichen Sinne, und daher rechtsgltig,
rechtswirksam. Replace legal by linguistic to see what the term meant to Sbawayhi:
permissible in a religious-moral and at the same time linguistic sense, hence linguistically valid, linguistically effective.
5
Sbawayhis examples in this section are not unambigously performatives, and they
can be, and have been translated elsewhere as literal statements I have sold etc.
6
The Stoic term pragmata for the things done which are represented by words is the
basis of our linguistic Pragmatics in the sense of doing things with words.
34
michael carter
of Greek ideas before there is any reliable documentation, suggesting

that the ideas circulated informally, perhaps from Christian to Muslim
scholars. Ibn Walld himself is very sceptical, and he flatly asserts that
there is nothing intrinsically positive or negative about commands: only
by inference (istidll) can we determine whether an imperative verb is
ordering us to do something or prohibiting us from doing the opposite.
The language and style of this argument are exactly what we encounter when the usls take up the theme later in the century. In al-Bqilln
(d. 403/1013, and a strong supporter of the notion of complementarity),
the debate is given a thorough airing (Taqrb II:198207), and by this
time both the grammatical and the logical techniques are well advanced,
and the level of argumentation shows that the Aristotelian dialectic had
been fully absorbed. There is a good deal of sophistry from al-Bqillns
opponents, who claim, amongst other things, that imperative and prohibitive verbs have different forms, therefore cannot be complementary
to each other. He matches their sophistry with a reductio ad absurdum,
putting it to them that by their own criteria a positive act would have
to be defined as not refraining from doing the opposite ayru tarki
diddihi and he concludes with the practical argument that the meanings of commands or prohibitions are understood naturally and grasped
immediately from the words themselves, without the need for any kind
of logical inference (while this may not look like Pragmatics, it is really a
question of differing opinions on conversational implicature: the methods of inference were a very contentious issue among usls, see Ali
2000, especially ch. 5). There is thus a clear line of progression from
Sbawayhis casual recognition of amr and nahy as grammatical categories (e.g. Der. I:105/Bl. I:1267), through Ibn Wallds early dialectical
treatment to the fully structured arguments of al-Bqilln and his fellow usls.
In other words, there is now a theory where before there was only data.
Under these circumstances there is a universal tendency for theory to triumph over data, and a very illuminating example is the treatment of the
word kad such and such [an amount]. For Sbawayhi, it was no more
than a dummy numeral (kinya, as he calls it, an allusion to any number,
just as fuln so-and-so is an allusion to any person), and it behaves like
interrogative kam, with a dependent (nasb) complement, kad wa-kad
dirhaman so and so many dirhams (Der. I:256/Bl. I:297).
By the end of the 4th/10th century a more complicated system makes
its appearance, when ar-Rummn (d. 384/994, arh fol. 161r) proposes
that kad by itself has the value of 1120 (because these numbers are
35
regarded as single words) while kad wa-kd stands for compound

numbers from 21 upwards.7 Ar-Rummn was famous for the exaggerated logical rigor of his experiments with language, but he did not go as
far as he could have done. Others did go further, and by the 7th/13th
century we find the fully elaborated system in Ibn Mut (d. 628/1231,
Fusl 23, further references in 2445), exactly mimicking the syntax of
the numerals:
kad wa-kad darhima = from 3 to 10 dirhams.
kad wa-kad dirhaman = from 11 to 99 dirhams.
kad wa-kad dirhamin = from 100 dirhams upwards.8
This exquisitely artificial and almost certainly unattested scheme is connected with the H anafs and (which is not much different in this context) with the Kfans, and it survived at least until the 19th century in
the work of Nsf al-Yzij (d. 1871, in Nr al-Qir, see Fleischer I:568
for references). What it shows is that as the linguistic and legal sciences
evolved they became increasingly abstract, passing beyond the limits of
actual usage. It does not seem likely that Ab H anfa himself explicitly
correlated the syntax of kad with that of the numerals in this manner,
but it fits the reputation of the H anafs for artificial and over-systematic
reasoning, and can be seen as a fine specimen of what can happen when
a legal judgement (scil. how much is meant by kad in such and such a
case?) has to depend more on the methodology than on actual speech.
Classical Arabic was no longer a living language at this time, and the
usls could only consult the rules of the dead language.
In doing so they nevertheless assumed that their one-way communication with the texts was a natural use of language. Here follow some
illustrations of the inherent similarities between Sbawayhis concept of
speech and the lawyers approach to consultations with God.
It has already been said above that Gods speech had to follow the
formal rules of Arabic grammar. Furthermore, as the usls were well
aware, in order to make himself understood, God had to obey the conventions of human communication. For this reason the usls, exactly
like Sbawayhi, strove to account for meaning in terms of the motives
of the speaker (man or God) and the real-life context of the utterance.
A good specimen from Sbawayhi of the dependence of meaning on
7
Through a scribal error the single kad (scil. kad dirhaman) and paired kad (scil.
kad wa-kad dirhaman) are not distinguished in the manuscript.
8
Further references in Carter 2003a, endnote 27 referring to p. 180.
36
michael carter
motive and context is the statement m atn rajulun: it can mean no

man [at all] came to me, not one man came to me[but several], no
[real] man came to me [but a weakling], or no man came to me [but a
woman did] (Der. I:20/Bl. I:27). There are no formal linguistic clues
whatsoever as to which meaning the speaker has in mind: the clues lie
in the extra-linguistic context, in this case, what question might the
speaker be answering, and this is why Sbawayhi spends so much time
offering psychological explanations for the speakers choices. Some centuries later the lawyers had to do the same for God, trying to penetrate
his words to divine his purpose, which they could only do by assuming
that he spoke to his servants in the same way as a rational human being
would under normal circumstances.
Sbawayhi took this for granted. The muxtabna in this next quotation are the same people who are the muxtabna of everyday conversation, but this time it is God speaking to them:
When [God] said your mothers are prohibited to you and so on to the
end of his speech (kalm), those being addressed (al-muxtabna) knew
that this was proscribed for them, and affirmed, but God went on to say
kitba llhi by written decree to add emphasis, tawkdan (Der. I:160/Bl.
I:191, on Q 4/2324).
Striking here is the attempt to explain away the mention of the written
decree: writing is ill-suited to Pragmatic analysis for the simple reason
that the recipient of writing (or in this case the sender!) is usually absent.
There can be no conversation, still less any conversational implicature,
without the presence of both participants. Writing is mentioned only
occasionally in the Kitb, and is completely marginal and secondary.
For the usls, too, it was not part of normal communication: it was a
special case, a kind of act of faith in the future, like farming, lending or
borrowing, and other actions which require a presumption of continuity, istish b al-h l, i.e. that the recipient would still be alive when the
letter was delivered (see Ali 2000, 80). We have to admire the usls
for their commitment to the belief that the spoken words of God and
Muhammad can be directly experienced through their written record:
the paradox was to a large extent resolved by the device of learning the
texts by heart so that, once implanted in the memory, they ceased to be
a document (mush af, kitb) and became a virtual oral event (Qurn,
h adt), neurologically the same as a remembered discourse.
God as a speaker also has the same privilege as humans of presuming
knowledge in his listeners: in Q 3/180 (Der. I:347/Bl. I:395) he says let
37
not those who are miserly with the bounty that has come to them reckon
[being miserly] is better for them . . ., and Sbawayhi explains that God
here omits the word al-buxla being miserly which would be required
as the first direct object of reckon, because the listener, al-muxtab,
will know from the verb yabxalna they are miserly that miserliness
is meant. Anyone who has consulted a tafsr will be familiar with this
method of filling gaps in the Qurn, and the usl sources are no exception in the use of this procedure.
A second point of similarity between Sbawayhi and the usls is the
presumption of sanity. Sbawayhi assumes, without spelling it out, that a
speaker will be mentally capable of formulating an idea and conveying
it successfullythis is implicit in his criterion of mustaqm right, used
for utterances which are fully understood in their intended meaning,
and in the term murd what is intended for the meaning of speech
acts. The lawyers turned this into an overt legal principle. They had to
decide who was allowed to speak, and to eliminate those who were not
legal persons, and therefore had no voice, such as infants and the insane,
and they produced the following conditions of legally valid speech (here
paraphrased from Ali 2000, 42 based on al-mid, d. 631/1233):
1. It must be uttered intentionally.
2. It must be intended for a particular listener.
3. The listener must be rational and understand it.
The first condition presupposes sanity, because only a sane person can
form an intention at all (legal or otherwise), or indeed be a Muslim for
that matter (sanity later formed part of the definition of a Muslim); the
second criterion excludes soliloquy, and the third anchors speech in a
sane society, as well as giving us a hint as to how the usls saw themselves in the dialogue. To recall the notion of leap-frogging introduced
above, it should be noted that these new legal definitions of speech found
their way back into grammar, where the speech of the insane, or of those
talking in their sleep, or even of birds imitating humans, were excluded
for the sole reason that their speech could not be intentional (e.g. airbn d. 977/1570, Nr 10, though it appeared before his time).
God himself comes under the same constraints: in order to communicate with humanity he must speak rationally. His attributes allow for
this: he has an intellect, a will and the power of speech, and his language
is that of the people he is addressing (there is some literature, which cannot be looked at here, on the requirement that God address his prophets
in the language of their own people otherwise the revelation will be in
38
michael carter
vain: you cannot punish sinners for disobeying laws they are unable to
understand). For the usls Gods mode of communication (bayn) is
inevitably a major theme in their treatises: they either begin their works
with a detailed review of the nature of Arabic and of legal semantics,
e.g. the introductory chapters of Ab l-H usayn al-Basr (d. 436/1044),
Mutamad, or else the subject is raised after the epistemological topics
have been covered, e.g. al-Bqilln (d. 403/1013), Taqrb, from I:316.
For Sbawayhi, and later the usls, lexical meaning is arbitrary. Definition by synonyms only leads to infinite regression (Der. II:339/Bl.
II:312), and meaning is nothing more than intention, hence the verb
arda to want and its derivatives are among the commonest terms in
the Kitb (1,362 times, plus 20 in the passive, according to Troupeau
(1976), s.v., and there are also synonyms). It is clear, too, that Sbawayhi
fully aware of the distinction between utterance meaning and sentence
meaning: he refers more than once to man l-kalm and man l-h adt
the [integral] meaning of the utterance i.e. not simply the sum of its
lexical parts, and man itself is almost exclusively used to denote the
meaning of speech acts, not of words, such as the acts of expressing surprise, asking a question, giving an order etc., e.g. the meaning of swearing an oath man l-qasam, the meaning of calling man n-nid, and
even of grammatical categories, the meaning of the dependent form
man n-nasb, the meaning of tanwn, etc.
A significant similarity between Sbawayhi and the lawyers is that they
both define the meanings of the particles (h urf ) in terms of their discourse functions: thus wa- and is used to to bring one thing together
with another and join them without indication of order, and fa- and
[then] is the same except that you leave some scope for one to be after
the other (Der. II:330/Bl. II:304), cf. Ab l-H usayn, Mutamad I:20,
very concisely, wa- is li-l-jam for joining while fais li-l-taqb for
arranging consecutively. Sbawayhis definition of naam yes is interesting: instead of the expected agreement or consent he gives us a
rather legalistic definition: naam indicates promise and belief , ida
wa-tasdq (Der. II:339/Bl. II:312), the former implying some kind of
contractual commitment (yes, I promise do it), the latter indicating
assent to a proposition (yes, I believe what you say), which in our context could mean believing the sellers description of the goods or the
terms of a contract.
Sbawayhi never even asks where meaning originates, but the usls
were obliged to agree on an answer before they could proceed to the
39
derivation of law from the texts. Here is a case where the debates of
the intervening century and a half carried the lawyers a long way from
Sbawayhis agnostic position. They had to reconcile the potentially infinite backward extension of meaning with the historical fact that Arabic
is not the oldest human language. Part of the solution, which will not be
discussed here, was to attribute to Adam the bridging role connecting
the supernatural Arabic he spoke in Heaven with the temporal world he
inhabited after the Fall. He himself did not speak Arabic on earth which
appeared only later, evolving naturally until it reached perfection in the
time of Adams prophetic heir and descendant, Muhammad.9
After Sbawayhi there was considerable discussion of the origins of
language, which has been investigated for the grammarians by Loucel (196364), with the general conclusion that the choice lay between
divine ordination (tawqf) or human convention (tawdu). The lacuna
which Loucel pointed out, that there was no comparable study of the
origins of language in the legal sciences, remains unfilled and will not be
dealt with here. For our purposes it must suffice to note that there was
overall preference for the view that language is in some way conventionally imposed (by wad) but the identity of the imposer, wdi, is left
obscure, perhaps deliberately: it may be God, it may be the first users of
Arabic, it may be all users of Arabic who agree amongst themselves on
the meaning of a word.
The real dispute concerned whether meaning could be imposed independently of a words being used. At one end of the spectrum (Mutazil),
a word does have a meaning before it is used, and at the other (Ibn
Taymiyya), a word cannot have a meaning until or indeed unless it is
used. These issues are well described by Ali (2000): what is important
for this paper is that none of these ideas, including the term wad in
this sense, are found in the Kitb. The concept must have emerged later,
probably under the influence of the Platonic debate over whether words
had meaning by their nature ( physis, cf. Arabic taba, replaced in Islm
by the creating God) or by imposition (thesis, the same as the wad of the
grammarians and lawyers).
From the lawyers point of view it was important to detach meaning
from prehistory: in spite of disagreement about the origins of language,
lexical meaning was taken as given, either a priori or as recorded by lexicographical experts (ahl al-lua), or synchronically by mere usage. It
See Carter (2003b).
40
michael carter
could then be treated as purely conventional, and in this way the lawyers,
after a long debate in which Sbawayhi took no part, eventually returned
to his Pragmatist position. Meaning, regardless of where it comes from,
is what you do with the language, or, put another way, language is only
meaningful for usls when it has legal effects outside the actual utterance. This restriction of meaning to habit and community usage enables
the law to control that communitys behaviour. The result is the same for
both grammarians and lawyers: speakers are obliged to stay within the
habitual codes, whether linguistic or social.
We might say that Sbawayhi took a lawyers view of language and
lawyers a Sbawayhian view. Consider his Pragmatist interpretation of
such verbs as ra etc. to see, consider, regard, be of the view that, of
which he says, even a blind man can use ra to see and say, I regarded
Zayd as the good man raaytu zaydan-i s-slih a (Der. I:13/Bl. I:18).10
The equivalent lawyers position is stated by a-aybn (b. 132/749, d.
189/804, and therefore a contemporary of Sbawayhi): a blind man who
has to feel the goods for sale when making a purchase is in the same
place (mawdi, i.e. legal situation) as a sighted man ( Jmi 81).
The primacy of usage and habit is asserted many times by Sbawayhi,
and there is no better illustration (because one senses a tongue in cheek
here) than his discussion of expressions of praise and blame. They are
not unconstrained, he says, you must follow the speech habits of the
Arabs, so you cannot, for example, praise someone for being a tailor
or a seed-merchant, still less praise a person in terms normally used of
God, e.g. al-h amdu li-zaydin praise be to Zayd!. Sbawayhi, or possibly a commentator, allows himself a pun here by saying that would
be a grave sin, am, playing on tam magnification, the name for
this laudatory construction (Der. I:214f/Bl. I:251). And although it is
correct to use such attested idioms as he is as close to me as where my
waist-cloth is tied huwa minn maqida l-izri, you cannot say he is as
close to me as where the horse is tethered *huwa minn marbita l-farasi
(Der. I:174/Bl. I:206). Ab l-H usayn discusses the interdependence of
meaning and use in a similar way (e.g. Mutamad I:17f, 2228), and it is,
of course the central problem of usl semantics, as Ali (2000) demonstrates in great detail.
10
The example is perhaps deliberately perverse, as the natural reading would be I
saw the good man Zayd, but this is in a chapter on verbs of the heart, so ra must have
the complete sentence Zayd [is] the good man as its direct object.
41
Finally Sbawayhis recognition of the performative and illocutionary

uses of speech can be compared with the far more elaborate and selfconscious expressions of the same ideas in the legal works, written after
the lawyers had had time to absorb Aristotles categories of sentences (a
post-Sbawayhian innovation), which include vocatives, requests, commands and entreaties.
Thus when Sbawayhi defines the vocative noun in y abdallhi as
made dependent (nasb) through the suppression of the verb which [in
this context] is not expressed (Der. I:262/Bl. I:303, nasbun al idmri
l-fili l-matrki ihruhu), he does leave it open for others to supply a
verb such as und I call out to, though his choice of the verb here
is significant, and in fact there are situations where a verb cannot be
restored, e.g. subh na llhi glory be to God, for which there is no verb
available to be suppressed. This is enough (especially when taken with
Sbawayhis other observations on elliptical utterances, see above with
xayra maqdamin) to permit the assumption than he was aware of the
performative aspects of y and other speech elements. In the usl treatment, those who did regard y as representing an elided verb such as I
call are roundly rebuked by Ab l-H usayn (Mutamad, I:20). He firmly
rejects this on the grounds that merely to say the word y and its noun
conveys in itself the information that an act of calling has occurred. Here
he makes explicit what is left implicit in Sbawayhi, as so often happens.
By Ab l-H usayns time it had become a polemical issue, and he accuses
those who would restore a verb in this situation of either oversimplifying for pedagogical purposes (how condescending!) or misinterpreting
what Sbawayhi meant by h arf nid a particle of calling, a performative
element by nature, which needs no verb, as Ab l-H usayn observes.
What this article has tried to show is that Sbawayhi, perhaps by his
own genius, perhaps encouraged and inspired by the community of
pioneering Islamic scholars around him, chose to treat language in a
manner which shows a remarkable affinity with modern Pragmatics.
The nine points which Ali (2000, 3f) regarded as essential pragmatic
insights are taken from usl sources, but equivalents to most of them
can easily be found in the Kitb. Indeed it might have made a tidier version of this paper simply to match them seriatim with material from the
Kitb.
These ideas were, so to speak, embryonic in Sbawayhi, but the marriage of Aristotelian logic and Arab linguistic studies in the 4th/10th
century led to the birth of a much more self-conscious and program-matic
version of the same attitude in legal science, reaching its peak of
42
michael carter
development in the 7th/13th century works which provide the bulk

of the material for Ali (2000). Symbolically we may point to the term
qnn law as an indication of the change of direction, though the term
has never carried much weight in medieval Islm. It would not have
meant much to Sbawayhi, who had half a dozen terms for the correct
use of Arabic, one of them sunna, nor is it prominent in medieval legal
contexts, but perhaps because it was circulating in the 4th/10th century,
e.g. in Frb (d. ca 339/950) Ih s 10, it may have fostered the notion
that a systematic code of behaviour along the lines of Greek ethics was
desirable and possible. As with grammar from that point on, works on
usl al-fiqh display an impressive mastery not only of the large body of
historical Islamic data but also the methods of logical enquiry and dialectical disputation. The two disciplines henceforth continue in parallel, grammar becoming increasingly legalistic to stop the language from
changing, and jurisprudence becoming more and more a grammatical
analysis of the unchangeable texts.
One result is that it was possible to formulate a set of five principles of
communication which, as set out by Ali (2000, 64), present an instructive analogue to Grices four maxims, here briefly paraphrased:
1.
2.
3.
4.
5.
The speakers disposition to make his intention manifest.

The speakers truthfulness.
That what is said should have its due effect.
That what is said should be grasped immediately.
That the existing conventions should be maintained.
With great diligence and subtlety the classical usls applied these and
other principles to the interpretation of Gods speech in search of
Gods law as it has been put. This had not been Sbawayhis goal, but
his exhaustive description of how the Arabic language works between
speaker and listener prefigured the Pragmatics of the usls and provided a basis (filtered through the subsequent grammatical tradition)
for their scholarly exertions.
In a sense there was always an implicit Pragmatics in Arabic grammar
and Islamic law. On the macro-level Islm itself is a covenant with God,
and Islamic society is a kind of social contract in which everything said
between Muslims has a contractual dimension, exactly as in Grices perception of speech. That is one reason why Muslims are enjoined not to lie
to each other, particularly when transmitting the religious knowledge on
which the survival of their faith depends. Sbawayhi could afford to take
truthfulness for granted, since it does not affect linguistic form, but the
43
usls had to stipulate truthfulness as a condition of a valid legal utterance precisely because there was no way to tell from the words themselves whether the speaker was lying, i.e. breaking Grices unenforceable
maxim of Quality, where lying is likewise formally undetectable.
The last point to make is that all this legal matter found in the Kitb
goes to confirm two other aspects of Sbawayhis life and work which tie
him closely to the lawyers. Firstly, from what little we know of his biography we can deduce that he associated not only with revered authorities
on Arabic but also with early legal and religious scholars, and secondly
it is evident from the Kitb that these scholars were the inspiration for
much of his technical vocabulary and methodology. While this does not
solve the problem of the origins of grammar completely, it does make
it more likely that the Kitb is the creation of a single mind, an unprecedented description of Arabic in all its domains, religious, poetic, public and private, in a theoretical framework which drew deeply upon the
principles of the nascent legal system and owes almost nothing to external traditions. This view may not meet with the approval of the dedicatee of the present volume, but it is a tribute to his belief in the right of
dissent that such heresies can be published without fear of legal action.
References
2.1
Primary sources
Ab H usayn, Mutamad = Ab l-H usayn al-Basr, Kitb al-Mutamad f usl al-fiqh.
M. H amdullh with M. Bekir and H . H anaf, eds. Damascus, 1964.
al-Bqilln, Taqrb = Ab Bakr Muhammad b. at -Tayyib al-Bqilln, at-Taqrb wa-lird as-sar. Abd al-H amd b. Al, Ab Zayd, ed. Beirut, 1998.
al-Frb, Ih s = Ab Nasr Muhammad b. Muhammad, Ih s al-ulm. A. Gonzles
Palencia, ed. Madrid, 1932.
Ibn Mut, Fusl = Ab l-H usayn Yahy b. Abd an-Nr, Ibn Mut , al-Fusl al-xamsn.
Mahmd Muhammad at -Tanh, ed. Cairo, 1976.
Ibn Walld, Intisr = Ahmad b. Muhammad b. al-Wald, Ibn al-Walld, Kitb al-Intisr.
Monique Bernards, ed. Changing Traditions. Al-Mubarrads Refutation of Sbawayh
and the Subsequent Reception of the Kitb, Arabic pp. 1212. Leiden, New York, Kln,
1997.
ar-Rummn, arh = Ab l-H asan Al b. s ar-Rummn, arh Kitb Sbawayhi. MS
Feyzulla 1984.
a-aybn, Jmi = Muhammad b. al-H asan a-aybn, al-Jmi as-sar f l-fiqh. Margin of Ab Ysuf, Kitb al-xarj. Blq, 1884.
n Sbawayhi, al-Kitb. (1) Hartwig Derenbourg, ed. Le livre de Sibawaihi. Paris, 18811889. Repr. Hildesheim, 1970. (2) Kitb
Sbawayhi. Blq Press, 18981900. Repr. Baghdad, 1965. (3) Kitb Sbawayhi. Abd
as-Salm Muhammad Hrn, ed. Cairo, 196877 (cross-paginated with the Blq
edition). References to the Kitb are usually in the form Der/Bl.
44
michael carter
as-Srf, arh = Ab Sad as-Srf, arh Kitb Sbawayhi. MS Atif Efendi 2548.
a-irbn, Nr = Muhammad a-irbn al-Xatb, Arab Linguistics, an introductory classical text with translation and notes [Nr al-sajiyya f h all alfz al-jurrmiyya].
Michael G. Carter, ed. Amsterdam. 1981.
2.2 Secondary sources
Ali Mohamed M. Yunis. 2000. Medieval Islamic Pragmatics. Sunni legal theorists models
of textual communication. Richmond.
Bergstrer, G. 1935. G. Bergstrers Grundzge des islamischen Rechts: bearbeitet und
herausgegeben von Joseph Schacht. Berlin, Leipzig.
Bravmann, Meir. 1953. Studies in Arabic and General Syntax. Cairo.
Buburuzan, Rodica. 1993. Exclamation et actes de langage chez Sbawayhi. Revue Roumaine de Linguistique 38, 421437.
Carter, Michael G. 2002. Patterns of reasoning: Sibawayhis analysis of the h l. Proceedings of the 20th Congress of the Union of European Arabists and Islamicists, Part
One, Linguistics, Literature, History [= The Arabist, vol. 2425]. K. Dvnyi, ed. Budapest. 315.
. 2003a. Legal Schools and Grammatical Theory. Arabistikai islamoznanie. Tom 2.
Studi po sluchai 60godishnata na dots. d.f.n. Penka Samsareva, Simeon Evstatiev, ed.
Sofia. 177183.
. 2003b. Talking with and about God, Adam and the Arabic language. Majz,
culture e contatti nellarea del Mediterraneo. It ruoli dell Islam (21st Congress of the
Union of European Arabists and Islamicists, Palermo 2002) [= La Memoria vol. 15].
Antonino Pellitteri, ed. Palermo. 197208.
Crystal, David. 2000. A Dictionary of Linguistics and Phonetics. 4th ed. Oxford.
Fleischer, Heinrich Leberecht. 188588. Kleinere Schriften, gesammelt durchgesehen und
vermehrt von A. Huber, prof. Thorbecke, und F. Bhlau. Leipzig. Repr. Osnabrck.
1968.
Grice, H. Paul. 1989. Logic and Conversation. Studies in the Way of Words, 2240.
Cambridge, Mass. and London.
Khadduri, Majid. 1987. Al-Imm Muh ammad ibn Idrs al-Shfis al-Risla f usl alfiqh. Treatise on the foundations of Islamic jurisprudence. Translated with an introduction, notes and appendices. 2nd ed. Cambridge.
Larcher, P. 1990. lments pragmatiques dans la thorie grammaticale arabe post-classique. Studies in the History of Arabic Grammar II. Kees Versteegh, Michael G. Carter,
eds. Amsterdam. 193214.
. 1998. Une pragmatique avant la pragmatique: medivale, arabe et islamique.
Histoire, Epistmologie, Langage 20, 101116.
Loucel, Henri. 1963, 1964. Les origines du langage daprs les grammairiens arabes.
Arabica 10, 188208, 253291; 11, 5772, 151187.
al-Marnn, Burhn ad-Dn. 1870. The Hedaya or Guide: a commentary on Musulman
laws, trans. by Charles Hamilton, 2nd edition. Standish G. Grady, ed. London (reference is to the reprint 1963).
Moutaouakil, Ahmad. 1990. La notion dactes de langage dans la pense linguistique
arabe ancienne. Studies in the History of Arabic Grammar II. Kees Versteegh, Michael
G. Carter, eds. Amsterdam. 229238.
Simon, Udo. 1993. Mittlelalterliche arabische Sprachbetrachtung zwischen Grammatik
und Rhetorik. Heidelberg.
Troupeau, Grard. 1976. Lexique-index du Kitb de Sbawayhi. Paris.
Versteegh, Kees [C.H.M.]. 1977. Greek Elements in Arabic Linguistic Thinking. Leiden:
E.J. Brill.
ID MR IN THE MAN OF AL-FARR:

A GRAMMATICAL APPROACH BETWEEN DESCRIPTION
AND EXPLANATION
Kinga Dvnyi
Corvinus University, Budapest
1. Introduction
Kees Versteegh stimulated the discussion on the history and development of Arab grammatical thinking in a number of his publications.
In one of his books (Versteegh 1993, 150), he reflects upon my earlier
analysis of al-Farrs linguistic methods in his Man (Dvnyi 1990).
He pointed out the insufficient analysis of idmr and its related terms
in this authors work. In another chapter of the same work (Versteegh
1993, chapter five), he assembled data in an attempt to present the interrelationship between grammarians, readers and commentators who
worked in the 2nd/8th century.
The present contribution would like to pick up these two threads and
examine, on the one hand, in some detail the role of idmr in the Man
l-Qurn of al-Farr and, on the other hand, analyze the role and place
of this grammatical commentary of the Qurn from the point of view of
other grammars (mainly Sbawayhis Kitb) and other exegetical works,
like for example those of al-Axfa and at- Tabar.
It is a well-known fact that al-Farr (d. 207/822) held in great esteem
Sbawayhis (d. 180/796) Kitb, which more than twenty years predated
his composition. Their starting points and approaches were, however,
widely different. Versteegh (1993, 180) has already pointed out that the
interests of the two authors lay elsewhere. To this, we can add that alFarr and Sbawayhi, though working within the framework of practically one grammaror one grammatical idealhad widely different
aims. While al-Farr, in his Man, used his grammatical knowledge
for the analysis of an existing corpus which he described from the point
of view of the listener to this text, Sbawayhi aimed at creating, from
the point of view of the speaker, a comprehensive grammar in which he
used poetical and Qurnic excerpts only by way of illustration.
46
kinga dvnyi
2. An overview of al-Farrs methods
In order to have a brief overview of the methods used by al-Farr, a

few examples from the beginning of the Man will be presented first.
That the corpus al-Farr is working on is the text of the revelation has
special importance. On the one hand, al-Farrlike grammarians and
later rhetoriciansconsiders that irb is a necessary prerequisite for
the understanding of any text and so the text of the Qurn. On the other
hand, since the Qurnic text is usually understood without relying on
the irb endings, there is a strong tendency to analyze different endings at a given place without entailing a change in the meaning. 1 This
method aims at eliminating the problems posed by the different qirt.
Although in most of the cases the grammatical analysis only underlines
and systematizes the interpretation given by the first exegetes of the
Qurnic text, in several cases, however, it is the grammatical analysis
which helps to disclose the meaning of the ya.
2.1
Man I:3 regarding Q 1:2 al-h amdu li-llhi
The task here is to determine the vowel ending of the word al-h amd.
Step (1): The examination of the readings (qirt):
According to al-Farr the readers are in total agreement concerning the
raf ending.2 It is also interesting to note, that al-Farr only mentions
this ending but does not present a grammatical explanation for it.3
Step (2): The elicitation of extra-textual linguistic source:
The Bedouins (ahl al-badw) say three things:
(a) al-h amda li-llh,
(b) al-h amdi li-llh,
(c) al-h amdu lu-llh.
1
A notable exception, where different readings reflect a difference in meaning is e.g.
Q 5:6. For the analysis of this ya, see Dvnyi 198788 and Burton 1988.
2
Makram and Umar (1985, I:5) also list al-h amda and al-h amdi among the readings
of this verse.
3
This is in contrast with al-Axfa (Man I:9 ff.) whose analysis at this place is rather
similar to that of al-Farr, but who also provides a detailed grammatical analysis of the
raf ending.
R in the MAN of al-farr

IDM
47
Step (3): The explanation of the variants:

For (a), i.e. the nasb ending, al-Farr
(i) gives a grammatical rule according to which h amd is a masdar
in place of which a verb could also have been used. This is an
example that fixes the rule;
(ii) supports it with other similar instances in the Qurn, like:
Q 47:4: fa-id laqtumu lladna kafar fa-darba r-riqbiinstead
of which one could say in kalm: fa-drib r-riqba
Q 12:79: mada llhi an naxuda . . .which is the same as:
nadu bi-llhi;
(iii) and props it with Bedouin usage (qawl al-arab) where saqyan
laka may be used instead of saqka llh.
The (b) variant, i.e. al-h amdi li-llh is treated as one word. As such, it is
compared to ibil, where two i vowels follow each other. This is a descriptive explanation making reference to usage. Other parallels are presented as well, among them the (c) variant: al-h amdu lu-llh. Al-Farr
approached this corpus from the point of view of the listener to the text.
And it seems from the second step employed by al-Farr that the listener
could hear some variants. And because al-Farr was interested not only
in the text of the Qurn, but obviously placed it in the context of the
Arabic language as a whole, he analyzed these versions. What is even
more, he proceeded to explain these variants in detail thoughaccording to his knowledgethese variations were not Qurnic readings.
2.2 Man I:78 regarding Q 1:7 [. . . anamta alayhim] ayri
l-madbi [alayhim wa-l d-dllna]
The task here is to determine the vowel ending of the word ayr. There
are neither variant readings nor extra-textual variants.
Step (1): Grammatical analysis:
Al-Farr explains the i in ayri as a nat to alladna, mentioning that it is
definite (marifa) because of the following word (al-madbi).4 He also
4
It is important to note in this respect, that while the analysis of al-Axfa goes along
the same lines (Man I:18) (with the usual difference in terminology), but with one
basic difference, i.e. that he does not define negatively the ending of the word, in other
words he does not say what it is not, rather contents himself with saying what it is. That
is to say, he does not deal with the refutation of grammatically incorrect endings or with
48
kinga dvnyi
categorically refuses to connect ayri to the preceding word, alayhim.

Doing this he seems to argue against the view of those among his contemporaries who carry the surface descriptive analysis or the analysis
based upon proximity (itb) to the extremes.
Step (2): Semantic analysis:
Turning to the semantic side of the explanation, he states that ayri is
connected to wa-l d-dllna which follows it and its meaning is l. He
also adds that if it meant siw, it could not have been followed by l. So
it can be established that al-Farr connects the two types of analysis, the
semantic and the grammatical.
2.3 Man I:1112 regarding Q 2:2 [dlika l-kitbu l rayba fhi]
hudan li-l-muttaqna
The question is whether hudan is in raf or nasb. Though in the case
of this particular word both endings are realized in the same surface
form, the question should be decided both from the point of view of the
semantic interpretation of the structure and for the sake of other similar
structures in the Qurn where even the surface realization is different.
In the case of similar phrases in the Qurn a number of qirt have
been preserved with readings in both raf and nasb, as e.g. Q 31:13 (
lm tilka yt al-kitb al-h akm hudan wa-rah matun/an li-l-muh sinn)
and Q 11:72 (a alidu wa-hd bal ayxun/an).
Step (1): Grammatical statement:
Both endings can be explained in different ways (wajh). If the interpreter
of grammatical structures does not deviate from the intended meaning, he may freely choose between raf and nasb or the different ways.
This freedom is expressed by al-Farr in the use of the 2nd person: id
aradta . . ., wa-in jaalta . . . rafata . . . (Man I:11, 12ff.).
Step (2): The detailed grammatical analysis:
(i) raf
(1) It can be the xabar of dlika l-kitb whichin this caseis analyzed as the mubtada. A paraphrase is given: dlika hudan.
the refutation of grammatically not permissible analyses. This difference will remain
characteristic throughout the two books.

IDM
49
(2) It can be a nat (modifier) of the xabar: l rayba fhi. A Qurnic

parallel is quoted: wa-hd kitbun anzalnhu mubrakun (Q
6:92, 6:155).
(3) It can also be considered the beginning of a new structure which
comes after a complete phrase (al stinf li-tamm m qablahu).5
A Qurnic parallel is quoted: wa-hd bal ayxun (Q 11:72).6
(ii) nasb al l-qat
(1) It is either cut off from al-kitbu, which in this case would be
analyzed as the xabar of dlika.
(2) Or it is cut off from the -hi in l rayba fhi. A paraphrase is given
as: l akka fhi hdiyan.
A ready-made grammatical rule is given as an explanation for both possibilities: li-anna hudan nakira ittasalat bi-marifa qad tamma xabaruh
fa-nasabtah li-anna n-nakira l taknu dallan al l-marifa.
2.4
Man I:1415 regarding Q 2:16 fa-m rabih at tijratuhum
The task is to define the reason why certain structures are permissible in
contrast to other seemingly similar structures that are not.
Step (1): Parallels from kalm al-arab:
hd laylun nimun
Step (2): Parallels from the Qurn:
Q 47:21 fa-id azama l-amru
Step (3): Semantic definition
The permissibility of specific structures greatly depends upon their
communicational value, i.e. they can only be permitted if they can be
understood unambiguously (ulima manhu I:14, 17). In these examples, the reference of the verbs and the adjective is unambiguously not to
5
In connection with this third analysis, attention should be called to the difficulties
of grammatical and semantic analysis arising from the lack of punctuation. No wonder
that writings on qat and istinf developed into a special branch of the Qurnic sciences. On the use of istinf and related terms in early grammar and exegesis in general,
see Versteegh 1993, 132136 and the literature cited there.
6
The reading in nasb can be found at this place in todays printed editions, and this
was al-Farr s reading as well (Man I:12, 2). The raf ending was read by Ibn Masd
and Ubayy among others (see Makram and Umar 1985, III:125).
50
kinga dvnyi
the other word in the phrase which is a common noun (tijra, layl, amr)
but to the people who act behind these words. Similar structures are
not permitted in the case of possible ambiguity, e.g. *qad xasara abduka
is not permitted because of the ambiguity of the word abd (he can both
be trader and the object of trade), as al-Farr puts it, its meaning cannot
be known: fa-l yulamu manhu (Man I:15, 3).
2.5 Man I:16 regarding Q 2:1718 wa-tarakahum f ulumtin . . .
summun bukmun umyun
The task is to explain the endings of summ, bukm and umy.
(i) The raf ending is explained by istinf and the completeness of the
preceding clause in itself (li-anna l-kalm tamma). An additional
reason is that the whole phrase is split between two yas, but this is
not a prerequisite of istinf. It should, however, be mentioned that
istinf is frequently found at the beginning of yas.
(ii) The nasb reading (summan bukman umyan) is explained in two
ways:
(1) according to the meaning (al l-man), i.e. by referring back
to the verb in the preceding clause (tarakahum);
(2) by their being expressions of blame (damm).
A parallel Qurnic passage quoted by al-Farr is Q 9:111112:
wa-man awf bi-ahdihi mina llhi
. . . at-tibna l-bidna l-h midnaexplained as istinf
or the reading in nasb
. . . at-tibna l-bidna l-h midnaexplained as qat
The use of the explanatory terms istinf and qat shows clearly that for
al-Farr the most important issue was to understand the text as a whole.
And one of the first issues to be dealt with was the correct segmentation
of the text. The term qat expresses more clearly the syntactic structure,
whereas the term h l can only refer to the meaning.7
2.6 Man I:17 regarding Q 2:19 yajalna asbiahum . . . h adara
l-mawti
Meaning, however, can also play a decisive role in the explanation of an
irb ending and the determination of the syntactic structure, as it is the
For the difference in al-Farrs usage between hl and qat, see Kinberg 1996, 194.

IDM
51
case, e.g. in Q 2:19 (Man I:17, 1ff.): yajalna asbiahum . . . h adara

l-mawti. The word h adara is in nasb not because the verb yajalna
affects it (*yajalnah h adaran) but by way of specification (tafsr). It is
the same as e.g. in the phrase ataytuka xawfan where fear is not given
but something is given because of it.
3. The explanatory technique of idmr in the Man

The above examples were presented to show the steps followed by
al-Farr in his explanations.
One of the basic explanatory techniques used by al-Farr in the
Man is the term idmr (together with its related forms). In the course
of the explanation of the text of the Qurn, al-Farr is basically confronted with three types of problems:
(i) The meaning of the ya needs clarification;
(ii) Though the meaning is clear, there is a problematical irb-ending;
(iii) The qirt of similar structures at different places in the Qurn
should be harmonized and explained.
It will be seen that the technique of idmr features as a prominent solution in all of these cases. It is outside the scope of the present paper
to examine all the occurrences of idmr and its related terms in the
Man. In the following, however, we shall try to present a typology of
the most significant types of occurrences of this term.
3.1 idmr is supported by another place in the Qurn where the
suppressed element occurs in the text
3.1.1 Man I:13 regarding Q 2:7 xatama llh al qulbihim wa-al
samihim wa-al absrihim iwatun
This ya has a variant reading, iwatan, which entails no difference in the meaning of the ya (wa-manhum wh id ).8 This reading is explained by otherssays al-Farrby the idmr of jaala. This
8
It might be interesting to note that al-Axfa did not mention the existence of different readings here (see al-Axfa, Man I:34).
52
kinga dvnyi
is supported by Q 45:23 where the phrase actually occurs as wa-jaala

al basarihi iwatan. Al-Farr accepts this explanation, i.e. the use of
another Qurnic place to support the explanation by idmr, though he
emphasizes that were it not for the Qurnic parallel, the idmr in this
sentence could not have been accepted. This is because according to
al-Farr, idmr is acceptable only in those utterances (kalm) which
are coherent ( yajtami), i.e. where the beginning refers ( yadullu al) to
the end. The idmr is acceptable (yah sunu) if it is well known (urifa; or
at other places malm).
A kalm example is brought in to illustrate what is meant by the
term yadullu al, and how it operates in structures. The example is as
follows: qad asba fulnun al-mla fa-ban d-dra wa-l-abda wa-lima wa-l-libsa l-h asana. It can be seen from this example, proceeds
al-Farr, that the notion of building (bin) cannot be extended to
slaves and clothes. By quoting this example from kalm, al-Farr appeals
to the listeners linguistic insight and competence in the Qurnic text.
He explains that the first part of the utterance (asba) refers to the
slaves, etc., i.e. qad asba fulnun al-mla fa-ban d-dra wa-[asba]
l-abda wa-l-ima wa-l-libsa l-h asana. Without saying that it should
be repeated, instead he uses the term dalla, yadullu al. Sbawayhi would
rather say explicitly that asba should be repeated before al-abda.
Al-Farr only hints at the idmr of asba.
In the following (Man I:14), Q 56:22 is mentioned as a similar
example: yatfu alayhim wildnun muxalladna bi-akwbin . . . wah rin nin.9 Al-Farr accepts the -in reading on the basis of proximity (itb) but mentions that those who put it in raf (wa-h run nun)
do this on the basis of the meaning, since the cups cannot be co-ordinated with the beautiful companions, and by the idmr of a word like
indahum, fh, maa dlika. Again, as in the previous example from
kalm, al-Farr does not use the term idmr here, but only hints at it.
The limits of idmr are imposed by understandability and coherence,
or rather their lack. This is called by al-Farr qillat al-ijtim. So. e.g. the
idmr of qataltu is not permissible in the following utterance: darabtu
fulnan wa-fulnan [*qataltu]. The reason is that there is no reference
9
Mainly because he considered it the accepted reading. This was the reading of a
great number of readers, among them al-Kis (cf. Makram and Umar 1985, VII:65),
but in todays printed mush af the -un reading can be found.

IDM
53
(dall) to the meaning here, the speakers intention cannot be known by

the listener.
At- Tabar adds (Tafsr I:87) that xatama cannot refer to the eyes
because they are never described by xatama either in the Qurn or in
the kalm.
3.1.2 Man I:35 regarding Q 2:72: wa-id qataltum nafsan
fa-ddratum fh
The jawb of id is present neither in this ya nor several others similar to it in the Qurn. Al-Farr explains the structure by the idmr of
udkur before id, and supports this explanation by similar utterances
in the Qurn where this word is present, as e.g. in Q 8:26 (wa-dkur id
antum qall . . .). Suppression of udkur with id can be inferred from
places like Q 8:26 and Q 7:86. al-Farr adds that if udkur was not
mentioned in Q 7:86, you could still infer that this is what is intended
because it occurs previously.
A parallel case of idmr is also mentioned: Q 7:73 (wa-il Tamda
axhum Slih an) where the known meaning allows the explanation of
this structure by the idmr of arsaln, i.e. arsaln Slih an. This example in itself can be regarded as a sub-case of the acceptance of idmr on
the basis of parallel Qurnic passages, since though the verb arsala is
used in the Qurn, it is mainly the content of the ya which determines
the selection of the suppressed element.10
3.1.3
Man I:166 regarding Q 2:246 wa-m lan all nuqtila
Al-Farr starts his analysis with comparing similar utterances from different parts of the Qurn, some of them with an and some other without it, like e.g.:
Q 14:12: m lan all natawakkala al llh
Q 57:8: wa-m lakum l tuminna bi-llhi wa-r-raslu yadkum litumin bi-rabbikum
Dropping (ilq) an here cannot be considered deficiency (illa) according to al-Farr because of its frequent usage in the arabiyya. The usage
of an relies on the meaning of manaa. According to this explanation,
10
On the meaning of this term in the Man, see Kinberg 1996, 377.
54
kinga dvnyi
the utterance m laka l tusall f jama means: m yamnauka an

tusalliya. The an may be inserted into the m laka phrase because the
latters meaning is identical with manaa. And a last evidence is put
forward from the text of the Qurn itself: Q 7:12: m manaaka all
tasjuda id amartuka runs parallel to Q 15:32: m laka all takna maa
s-sjidna.
idmr is supported basically by kalm
3.2
3.2.1 Man I:32 regarding Q 2:48 wa-ttaq yawman l tajz nafsun

an nafsin ayan
Al-Farr explains this passage by the idmr of fhi after tajz, insisting
on the necessity of distinguishing by reference to the meaning adverbial structures allowing idmr from verbal structures with sifa (prepositional phrase) complement, where idmr is not possible. For example,
anta llad takallamtu fhi should be distinguished from anta llad takallamtu. In doing this he argues with al-Kis and others who understand
too rigidly the formal description and identify the two, formally similar but semantically different types. On the basis of this identification
al-Kis refutes both structures with idmr while others permit both
of them. The same holds true, says al-Farr, for the fhihu alternation
within adverbial structures of time and place but not in verbal prepositional phrases. The phrase tka yawma l-xams is interchangeable
with tka f yawmi l-xams, since the sifa and h here agree in meaning
(muttafaq manhum). But when the meaning differs it is not allowed
to suppress f in place of h and vice versa: yuh ibbuh does not equal
with yuh ibbu fh.
at-Tabar, who quotes extensively (Tafsr I:203, 11ff.) al-Farrs interpretation, adds that the idmr in this ya is possible because it is wellknown (malm), this being the generally accepted explanation for
idmr by the exegetes (ash b at-tawl).
3.2.2
Man I:113, 9, Q 2:185: wa-li-tukmil l-idda
The li- + verb cannot be art11 (cause) of a previous verb because the ww
blocks its impact. Al-Farr discusses in many places that some particles,
11
On the meaning of this term in the Man, see Kinberg 1996, 377.

IDM
55
especially the ww, do not allow us to suppose a grammatical link with

a preceding regent. This phenomenon is explained on the basis of kalm
examples: jituka li-tuh sina ilayya cannot be transformed to jituka
wa-li-tuh sina ilayya only if you mean (jituka) wa-li-tuh sina ilayya
jituka. This second occurrence of the same verbal phrase is, naturally,
suppressed (al-arab tudxiluh f kalmihim al idmr fil badah).
This rule stated, al-Farr cites many Qurnic places where the
same causal structurewa-li and the idmr of a verbal phrasecan
be observed signifying the result of this cause, e.g. wa-kadlika nur
Ibrhma malakta s-samawti wa-l-ardi wa-li-yakna min al-mqinna
(Q 6:75) where the suppressed (mudmar) phrase is a repetition of the
verbal phrase before ww: araynhu.
At-Tabar (Tafsr II:88, 3ff.) quotes two opinions. One is a verbatim
quotation of al-Farrs view without mentioning his name (qla bad
nahwiyy l-Kfa). The other is the view of bad ahl al-arabiyya which
was rejected by al-Farr. This supports our analysis that whenever
al-Farr rejects a seemingly only theoretical possibility of analysis, he
rejects the opinion of a certain group of people without attributing their
view to anybody.
3.2.3 Man I:271 regarding Q 4:46 mina lladna hd yuh arrifna
l-kalim
The meaning is: man yuh arrifna l-kalim. This is supported by kalm
examples where man is suppressed in the beginning of the utterance
(mubtada al-kalm): minn [man] yaqlu or minn [man] l yaqlu.
This is possible according to al-Farr, because min, i.e. the first part
(awwal) of the utterance refers to the meaning of what has been left out,
being its part (bad). Al-Farr considers it important to note that man
cannot be left out (idmr) as a rule, but only in those cases where the
prepositional phrase refers to it. This latter sometimes may be f too: fh
slih na wa-fh dna dlika.
3.3 idmr is supported by the immediately preceding context (the
first part of the utterance or the previous utterance): Man I:141142
regarding Q 2:220: wa-in tuxlithum fa-ixwnukum
With idmr of the rfi, i.e. the nominal subject: hum, or if it is nasb, with
idmr of the second occurrence of the verbal phrase: fa-ixwnakum
tuxlitna. In both cases the suppressed element is given in the first
56
kinga dvnyi
part of the utterance or, sometimes, in a previous utterance, as in

Q 2:239: fa-in xiftum fa-rijlan. This is called ijtim al-kalm. Here,
says al-Farr, it would not be correct to suppose huwa because it is not
an (everlasting) state (dim) but an action. The previous ya (h fi
al s-salawti . . . wa-qm li-llhi . . .) gives a clue for the explanation.
Hence the meaning can be restored as: in xiftum an tusall qiyman
fa-sall rijlan.
At-Tabar (Tafsr II:208) follows al-Farrs argument, and refers to the
same Qurnic passage, etc. but everything is more detailed. It is as if he
worked from a fuller version of the Man.
3.4 idmr is supported primarily by the readings of Abdallh and
Ubayy: Man I:156 regarding Q 2:240: wa-lladna yutawaffawna
minkum wa-yadarna azwjan wasiyyatan
Here al-Farr presents two readings, nasb (wasiyyatan) and raf
(wasiyyatun), and both of them are explained on the basis of idmr. The
first idmr relies on the rule of the absolute object, the suppressed verb
being a jussive of the same root (li-ys . . . wasiyyatan). He refuses, however, to accept the supposition that the previous verb in the yadarna
azwjan phrase could have put wasiyya into nasb. This may have been
held by the supporters of the proximity-theory who always preferred the
explanation on the basis of the neighboring word.
The second, raf, reading (wasiyyatun) may also be explained by
idmr. The source of this idmr is the reading of Ibn Masd or Ubayy:12
fa-matun li-azwjikum or kutiba alaykum al-wasiyyatu li-azwjikum,
both containing raf , the first also a lexical variation.
12
From among the 64 Qurn readers whom al-Farr quotes by name, Ibn Masd
is by far the most frequently quoted, with 411 references. Ubayy is the seventh most
frequently quoted reader, with 92 occurrences. Cf. Dvnyi 1991, 160161.

IDM
57
3.5 The use of major, meaning-triggered grammatical rules in

supposing idmr
3.5.1
h l
3.5.1.1 Man I:24 regarding Q 2:28: kayfa takfirna bi-llhi

wa-kuntum amwtan
The meaning is established as qad kuntum, since the second verb is a
h l to the first one and it refers back to a state prior to the past action
denoted by the first verb.
After the definition of the meaning, al-Farr formulates the following
rule: this type of h l necessitates qad, either overtly (ihr) or in a suppressed way (idmr). If this condition is not fulfilled, i.e. the first verb
does not refer to the past (like e.g. in the case of kda, as) then the
second cannot contain qad in either way.
This rule is reinforced by another type of proof, a parallel place in
the Qurn where qad is being suppressed and where it is generally
understood to be necessary for the correct meaning (Q 12:27: in kna
qamsuhu qudda min duburin fa-kadibat).
At-Tabar (Tafsr I:146, 18ff.) follows al-Farrs arguments but he
adds that the explanation for the suppression of qad lies in the fact that
if the verbal form faala takes the place of h l it is self-evident (malm)
that it requires qad. It is to be emphasized that al-Farr also considers
that only those elements can be suppressed which are well-known to the
speakers.
3.5.1.2 Man I:372 regarding Q 7:4: fa-jah basun baytan aw
hum qilna
The meaning is established: [wa]-hum qilna is a h l, parallel to
baytan. Its grammatical rule is: wa-huwa filun, accordingly the idmr
of wa- is compulsory. The wa- should be present either overtly or in a
suppressed way. All this is supported by an example from kalm where
both ways (ihr and idmr) are correct: ataytan wliyan aw wa-an
mazl or awan mazl.13
13
Az-Zajjj (Man II:317) refutes this explanation without mentioning al-Farr
and says that it is not necessary to suppose wa- in h l in general.
58
kinga dvnyi
3.5.2
The delimitation and rules of idmr in quotation (hikya)
3.5.2.1 Man I:38 regarding Q 2:58 wa-ql h itta tun

Considering h itta tun as a h ikya entails the idmr of the rfi (i.e. the
nominal subject) which is supposed to be m umirtum bihi or simply
hiya. Though not called idmr in this place, it is analyzed as such, as will
be seen below.
It is a quotation (h ikya) which presupposes that it should be correct
(saluh a) if the rfi/nsib or xfid is suppressed. Al-Farr exemplifies
this rule as follows: qultu l ilha ill llh fa-yaql al-qil: qultu kalimatan slih atan. But kalimatan slih atan cannot stand alone without a
nsib.
To decide whether an utterance is quotation or indirect speech is very
important for the Qurnic text where change(s) of speaker and hearer
can occur even within the same ya.
3.5.2.2 Man I:40 regarding Q 11:69: ql salman qla salmun
The word salm is used here in two meanings (al manyayni). In the
first step the two endings (nasb and raf ) are explained. The explanation
of the endings relies on the grammatical rule differentiating indirect
speech from quotation (h ikya). Having fixed this al-Farr adds that
in the case of h ikya the idmr of alaykum should be supposed. In a
third step he demonstrates that this idmr is possible because salm
frequently occurs in kalm alone, i.e. without alaykum.
3.5.2.3 Man I:93 regarding Q 2:154: wa-l taql li-man yuqtalu f
sabli llh amwtun
It is another instance of h ikya where the raf ending is explained by the
idmr of hum.14 The irb ending and the difference between quotation
and indirect speech is demonstrated by substitutional analysis (huwa bimanzilat . . .). He says that the nasb ending is not permitted here because
amwt is an ism and not a qawl, i.e. it is not a nominalized phrase.
He illustrates the possible and impossible usages as follows:
qultu laka xayran, i.e. kalman h asanan.
qultu laka xayrun, is similar to: qultu laka mlun.
14
Az-Zajjj (Man I:229) also accepts this analysis.

IDM
59
In the case of xayr both structures are correct whereas their substitutes
cannot be used in the other structure. Since there are no different readings here it might be supposed that by illustrating these structures in
detail al-Farr may possibly refute analyzing techniques that take into
account only the surface structure.
3.5.2.4 Man I:296, 7 regarding Q 4:171: wa-l taql taltatun
This and similar examples show that it is a general grammatical rule that
necessitates the use of idmr, scil. if there is a marf after the qawl (as
a quotation after the verb qla), there must be a rfi (subject) as well, in
the above example: wa-l taql hum taltatun
General rule: if something is in raf after qla (yaqlu, qawl), there
should be a rfi either overtly or in a suppressed way (idmr). Here alFarr gives a formal explanation of an irb ending.
3.5.3 The id in the relative clause: Man I:157 regarding Q 2:246
ibat lan malikan nuqtil f sabli llhi
The verb may not be nuqtilu (in raf ) referring to the preceding noun
as a sila, because there is no reference back to malikan. However, if one
accepts the yuqtil reading, it can be explained as either sila ( yuqtilu)
or jaz (yuqtil) after an imperative (amr), as in the case of nuqtil.
Al-Farr shows that the form yuqtilu can also be used when there is
no antecedent, with idmr: ibat lan llad yuqtilu. Then he presents a
kalm example: allimn ilman antafiu bihi and allimn llad antafiu
bihi. But if bihi is dropped, then only the jazm is correct, since there is
no reference back (id).
3.5.4 jaz: Man I:178, 4, Q 2:265: fa-in lam yusibh wbilun
fa-tallun
Al-Farr here refers to the grammatical rule of jaz that requires md
forms, thus he supposes the suppression of kna before tallun. The
interesting point of his analysis is, however, when he says: udmirat kna
fa-saluh a l-kalm, because it sheds light on the meaning of idmr: It is
not simply suppression or deletion but the supposition of a suppressed
element which corrects the utterance and makes it fit the grammatical
rules.
60
kinga dvnyi
3.5.5 amm . . . fa-: Man I:228 regarding Q 3:106: fa-amm lladna

swaddat wujhuhum a-kafartum
This is a good example of a well-known grammatical rule (the obligatory use of fa- after amm) which makes the grammarian suppose the
idmr of fa-. At the same time the meaning of the ya also requires the
supposition of a suppressed phrase yuql. After dropping yuql, however, the fais also dropped. In reconstructing the meaning one needs to
reconstruct the utterance as well: fa-amm lladna swaddat wujhuhum
fa-yuql a-kafartum.
In the above five examples idmr is necessitated by some basic grammatical rules considered by al-Farr to be generally accepted and evident to such an extent that he does not even try to explain them with the
exception of the last example.
3.6 The impossibility of idmr: Man I:195197 regarding Q 3:15: qul
a-unabbiukum bi-xayrin min dlikum li-lladna ttaq inda rabbihim
janntun
Jannt is in raf because of the influence of lm. This intervening lm
prevents referring back to the beginning of the utterance (awwal
al-kalm).
He then draws the conclusion: the xfid (here bi-) cannot be suppressed (lam yudmar), therefore one cannot suppose that the bi- (in bixayrin) is suppressed before jannt causing it to be in xafd. This also
means that the irb-ending xafd cannot be used independently. The
same principle is stated in connection with Q 6:96 (Man I:346) where
a h l stands in the way of the effect of a preceding verbal noun (jil):
wa-jilu l-layli sakanan wa--amsa wa-l-qamara h usbnan. Since
a-amsa and al-qamara cannot be in xafd, therefore in al-Farrs view
they take nasb ending according to their meaning.
4. Concluding remarks
4.1
A summary of idmr types
To sum up the method of idmr, it can be established that al-Farr

derives the explanations of specific Qurnic passages from the following main types of sources:

IDM
61
(i) the end of an utterance is related to its beginning, the preceding

context (sometimes on the basis of the different segmentations of
the text), whereas al-Farr (Man I:13, 1415) puts itwainnam yah sunu l-idmru f l-kalmi llad yajtamiu wa-yadullu
awwaluhu al xirihi, i.e. the context is coherent. It is contrasted
with inqataa, where the context is incoherent.
(ii) another place in the Qurn;
(iii) an extra-textual linguistic source: kalm al-arab/badw or sometimes a poem, though it should be pointed out that poems do not
have their central significance they have in the Kitb;
(iv) a grammatical rule often quoted without example or proof. Its main
types are:
(a) only the rule is mentioned;
(b) a rule-like example is given (kam taql . . .);
(c) a grammatically incorrect example is used to support his reasoning (l yajz . . .);
(v) a qira;
(vi) the codex of Abdallh b. Masd or, less frequently, that of Ubayy
b. Kab or the codices of their followers which sometimes contain
non-canonical variations.
4.2
The meaning of idmr
The meaning of idmr is not simply h adf (deletion), but rather the
obligatory supposition of an element. Thus when al-Farr writes that l
budda min idmr kna li-anna l-kalm jaz (Man I:178, 4) it means
just the opposite of obligatory h adf (the obligatory deletion); it means
that it is obligatory to suppose the existence of an element either present
in the utterance or suppressed.
The use of the terms uskitat and ulqiyat also sheds light on the main
characteristic of idmr, i.e. its obligatory nature. For example at Man
I:163 regarding Q 2:246 there are two options to express the same meaning: either with all or only l. The dropping of an, however, cannot be
considered idmr, since both variants are equally correct Arabic (al
wajh al-arabiyya), the structure with only l not being brought back to
the one with an.
62
4.3
kinga dvnyi
Two kinds of traditions in the Man
It can be established that al-Farr relied on two kinds of traditions in

his Man.
I
The grammatical tradition
A) al-Farr and Sbawayhi

Disregarding the terminological differences it can be stated that on the
whole he followed the same grammatical tradition that became exemplified by Sbawayhis Kitb.
a) He takes sides with a Sbawayhi-like grammatical analysis on the basis
of the mil theory, without, however, using the same terminology.15
b) Frequent reference can be found to grammatical rules considered
self-evident and thus in no need of further explanations.
c) We can see the main difference between the Man of al-Farr and
Sbawayhis Kitb not in the method of grammatical analysis but in
their approach. Al-Farr, starting from a complete text, always has
in mind the text as a whole, as a series of utterances, and he proceeds
accordingly. Whereas Sbawayhi cites only specific examples to illustrate the grammatical rules he wishes to present.
B) al-Farr and al-Axfa
The difference between al-Farr and al-Axfa lies in the fact that the
latter basically writes about grammar, while the former deals essentially
with Qurnic exegesis on a grammatical basis. This may account for
al-Axfas lack of interest in eliciting all the known readings at a given
place whereas al-Farr seems keen on mentioning whatever can be
heard (see, e.g. al-h amdu/al-h amda/al-h amdi above), whereas does not
seem to attach great importance to the explanation of structures or endings that he considers self-evident.
15
According to Talmon (2003, 309312) Sbawayhis main concern in syntax is irb
carried out by amal effect. Talmon also postulated that in the Kitb al-h udd, al-Farr
seemed to focus in his syntactic description on sentence-types and the determination
of syntactic relations. We can also experience in the Man that while al-Farr dealt
with irb-endings in a somewhat flexible way, he did not make allowances in the case
of syntactic structures.

IDM
63
C) al-Farrs opinion on the linear interpretation of the text

a) At least half of al-Farrs analysis of idmr deals with the interpretation of different irb endings which he considers fundamental in
clarifying the meaning of the Qurnic text. Therefore he is in constant struggle with those views according to which the short vowel
endings are merely phonetic or morphological phenomena and not
irb endings. The most frequent of these solutions is the choice of
the final vowel on the basis of proximity, itb.16
b) He also introduced some formal descriptive interpretations in which
he might have followed Kfan exegetical methods. He accepts these
explanations if they fit into his concept of the linear descriptive text
analysis. The best examples to demonstrate this are those places
where he insists that certain particles, such as wa-or the prepositions, may block the impact of the previous grammatical structure
(e.g. Q 2:185 as analyzed in 3.2.2 above).
c) al-Farr, however, refuses to accept superficial formal analyses which
run contrary to the meaningful interpretation (e.g. Q 2:48 when he
refuses identification of f in temporal adverbs with f in prepositional
verbs). In such cases he does not say whose opinion, whose analysis is rejected by him. Sometimes he uses the passive (yuql), some
other times it can only be supposed that he is at variance with some
people who only looked upon the text as a linear string of utterances
and derived the explanations quasi mechanically from the preceding
string.
II
The exegetical tradition
Al-Farrs grammatical activities in the field of Qurnic exegesis cannot

be overestimated. It becomes especially evident if we compare his explanations to the relevant passages in at- Tabars Tafsr. It goes without saying that the scope of at -Tabars commentary is much wider than that of
al-Farrs work. But if we limit our comparison to the linguistic exegesis, we find that al-Farrs explanations are most of the time taken over
verbatim without his name being mentioned. One can also suppose that
16
Al-Farr accepts these variant forms in case of different existing sam in the collected corpus, as e.g. in the case of al-h amda/i/u (Q 1:2). If, however, the difference
appears not in the given vowel ending but in the explanation, then he always advocates
the explanation based on iml, as e.g. in Q 1:7 where he does not accept the explanation according to which the ending of ayri would be determined by the preceding
alayhim.
64
kinga dvnyi
at-Tabar had access to a fuller version of the Man, it is, however, also
possible that al-Farr was only one prominent exponent of a common
thinking about grammatical issues in the Qurn and the analyses he
dictated from his memory belonged to a common stock of knowledge, a
long line of grammatical exegetical tradition, with which at-Tabar was
still familiar. It is interesting to note, however, that the two other Man
works, Ab Ubaydas Majz and al-Axfas Man do not seem to have
been incorporated into what became the definitive commentary of the
Qurn for centuries. So it might be concluded that while the grammatical tradition as it was shaped in Basra outshone the Kfan, its trace in
the grammatical analysis of the Qurn is not significant.17
5. References
5.1 Primary sources
Ab Ubayda, Majz = Ab Ubayda Mamar b. al-Muta nn at-Taym, Majz al-Qurn.
Muhammad Fud Sazgn [Fuat Sezgin], ed. 2 vols. Cairo: al-Xnj, n.d.
al-Axfa, Man = Ab l-H asan Sad b. Masada al-Muji al-Axfa al-Awsat , Man
l-Qurn. Fiz Fris, ed. 2 vols. Amman: Dr al-Bar, 1981.
Farr, Man = Ab Zakariyy Yahy b. Ziyd al-Farr, Man l-Qurn. Ahmad
Ysuf Najt and Muhammad Al an-Najjr, eds. 3 vols. Cairo: al-Haya al-Misriyya
l-mma li-l-Kitb, 19802.
n Sbawayhi, al-Kitb. Abd as-Salm
Muhammad Hrn, ed. 5 vols. Beirut: lam al-Kutub, 19661977. n.d.
at-Tabar, Tafsr = Ab Jafar Muhammad b. Jarr at-Tabar, Jmi al-bayn f tafsr
al-Qurn. 30 vols. Cairo: al-Mat baa al-Maymaniyya, n.d.
az-Zajjj, Man = Ab Ishq Ibrhm b. as-Sar az-Zajjj, Man l-Qurn wa-irbuhu.
Abdaljall Abduh alab, ed. 5 vols. Beirut: lam al-Kutub, 1988.
17
An interesting feature of al-Farr s explanatory method is his usageat certain
placesof the 2nd person singular in the case of irb endings and other grammatical
interpretations. The use of 2nd person singular is on the one hand the usual practice
in Sbawayhis Kitb, but Sbawayhi presents his linguistic analysis from the point of
view of the speaker, the producer of different utterances, whereas al-Farr deals with
a concrete text, and what is even more, the text of the revelation. The use of the 2nd
person singular in this case might indicate that al-Farr considered that the sacred text
was the one without the short vowel endings or that it had been revealed according to
what is termed as sabat ah ruf which leaves the reader of the text some freedom in the
vocalic realization. It should, however, be noted that al-Farrsimilarly to other commentators, or in fact Sbawayhi when he deals with the Qurn (e.g. Kitb II:155, 10 ad
Q 5:69)does use the 3rd person singular or the passive when he deals with different
irb endings in the Qurn.

IDM
5.2
65
Secondary sources
Burton, John. 1988. The Qurn and the Islamic Practice of wud. BSOAS 51:1:2158.
Dvnyi, Kinga. 198788. Mujwara: A Crack in the Building of irb. Quaderni di
Studi Arabi 56, 196207.
. 1990. On Farrs linguistic methods in his work Man l-Qurn. Studies in the
History of Arabic Grammar II. Proceedings of the 2nd Symposium on the History of Arabic Grammar, Nijmegen, 27 April1 May 1987, Michael G. Carter and Kees Versteegh,
eds. Amsterdam: J. Benjamins.
. 1991. Al-Farr and al-Kis: References to Grammarians and Qurn Readers
in the Man l-Qurn of al-Farr. The Arabist. Budapest Studies in Arabic 34, 159
176.
Kinberg, Naphtali. 1996. A Lexicon of al-Farrs Terminology in his Qurn Commentary.
Leiden: E.J. Brill.
Makram, Abd al-l Slim and Ahmad Muxtr Umar. 1985. Mujam al-qirt alQurniyya. 8 vols. Kuwait: Jmiat al-Kuwayt.
Talmon, Rafael. 2003. Eighth-century Iraqi Grammar. A Critical Exploration of PreXallian Arabic Linguistics. Winona Lake, Indiana: Eisenbrauns.
Versteegh, C.H.M. 1993. Arabic Grammar and Qurnic Exegesis in Early Islam. Leiden:
E.J. Brill.
ARABIC ALLAD AS A CONJUNCTION:

AN OLD PROBLEM AND A NEW APPROACH
Werner Diem
Cologne
Sunt aliquot quoque res, quarum unam dicere causam

non satis est, verum pluris, unde una tamen sit.
Lucretius: De rerum natura VI 703f.
1. Introduction
This paper is devoted to the Arabic sub-standard phenomenon of the
relative pronoun allad in the function of a conjunction predominantly
meaning that; because. The historical interpretation of this phenomenon, which had been noted by Arab purists for the formula al-h amdu
li-llhi llad Praise be to God that as early as the 9th century C.E.,
has been a topic for Arabists during the last fifty years. Considering the
intensity with which the historical dimension of allad that; because
has been discussed, it may seem superfluous to want to take it up yet
again. This, however, would be a rash conclusion as no communis opinio
concerning the origin of the conjunctional allad has ever been agreed
upon.
The aim of this paper is threefold. First, I shall discuss the theories
as to the diachronic aspects of allad as a conjunction in chronological
order. Then I shall present additional early evidence from documentary sources dating from the 11th12th centuries C.E. and later. After
that, I shall present my own theoretical approach combining important
insights of my predecessors with hitherto neglected aspects. Finally, I
shall deal with the origin of the formula al-h amdu li-llhi llad.
A point that I shall not consider in this paper is the general function of
the H amdalah in formulaic expressions used by todays Arab Muslims.
The important role of this formula, simple and expanded, in everyday
Arabic speech is borne out by numerous examples in Piamenta (1979,
247f., Index) and (1983, 209, Index), among them examples of the
68
werner diem
al-h amdu li-llhi llad type with allad as a conjunction meaning that;
because (Piamenta 1979, 89, 172f., 176).
2. Theoretical approaches
2.1
Meir M. Bravman (1953)
In the chapter The development of the psychological (logical) subjectpredicate relation in his Studies in Arabic and General Syntax (1953),
Meir M. Bravmann, after stating that in constructions of the al-h amdu
li-llhi llad type and similar expressions allad can be interpreted as
that, maintains that the construction is already found in Classical Arabic, for which he adduces a verse by Imra al-Qays. This verse need not
be discussed here as it allows of other interpretations. More important
is Bravmanns general theory about the origin of allad as a conjunction
(1953, 41):
[. . .] we have to assume [. . .] relative clauses which do not link up with
the immediately preceding expression (as al-h amdu li-llhi etc.), but are
parts of an independent new sentence or, more exactly, predicates whose
subjectknown from the preceding sentence or from the situationis
mentally supplied but not linguistically expressed [. . .]. [. . .] the sentence
al-h amdu li-llhi llad lam amut would mean: Thank God. [I am one]
who has not died etc.
Spitaler (1963) rejected Bravmanns theory (see below 2.3), and it was
not discussed by other scholars. I tend toward Spitalers rejection of this
theory as it does not explain satisfactorily the phenomenon in question,
and therefore I shall disregard it.
2.2
Joshua Blau (1961)
In his grammar of Judaeo-Arabic, Joshua Blau treats the conjunctional

allad in 346, which begins with the words allad introducing noun
clauses.1 The paragraph falls into five sections according to the syntactic
status of the allad clauses: (a) attributive clauses, (b) subject clauses,
(c) predicate clauses, (d) object clauses, (e) clauses in which allad has a
causative function. Most of the examples refer to past events, but there
This and the following translations from the Hebrew of Blau are mine.
arabic allad as a conjunction
69
are also some referring to future events with the heads of the clauses
expressing prevention and necessity.
As to the origin of the conjunctional allad, Blau begins section (a)
with the following words (1961, 226):
Relatively usual are attributive clauses (apposition), which represent the
transition from relative clauses,
a statement which is continued in a note with the remark (words in

brackets are my additions):
However, this does not mean that allad introducing noun clauses has
originated from attributive clauses only. It is, for example, possible that
from sentences such as al-h amdu li-llhi llad anan Praise be to God
who helped me allad = k, e- [that; because] originated, marking
object clauses or mipne e [because].
These are quite general statements, which hint to the direction where,
according to Blau, a historical interpretation should be looked for. In his
opinion, as can be grasped from both remarks cited above, the origin of
allad as a conjunction has to be seen (a) in attributive relative clauses
and (b) in the al-h amdu li-llhi llad type.
2.3
Anton Spitaler (1963)
Spitalers approach differs fundamentally from Bravmanns and Blaus

as it is monocausal. He starts from the al-h amdu li-llhi llad type and
assumes that this construction was first re-interpreted (Spitaler does not
use this word) as meaning gottlob, da, after which it was generalized.
The following quotations will illustrate Spitalers approach:
al-h amdu li-llhi llad ist der Anfang eines normalen, von dem Wort Allh
abhngigen Relativsatzes, aber aus seinem eigentlichen Zusammenhang
gelst und sekundr in eine neue Konstruktion bertragen. (1963, 101)
Es ist nun ganz eindeutig, dass das stereotyp wiederkehrende al-h amdu
li-llhi im Lauf der Zeit einer Funktionsschwchung unterlegen ist [. . .].
Dadurch verlor das nachfolgende allad zwangslufig den lebendigen
Zusammenhang mit seinem Beziehungsnomen und der ganze Ausdruck
wurde zu einem starren, wenn auch mit einem ganz bestimmten Affektgehalt geladenen Syntagma, bei dem allad nur mehr als berleitendes
Element, als verbindende Partikel, letztenendes eben als Konjunktion
empfunden wurde. Und nunmehr war es natrlich gleichgltig, welche
syntaktische Form der anschlieende Satz hatte. (1963, 235)
70
werner diem
After presenting some examples semantically corresponding to al-h amdu

li-llhi llad praise be to God that which, however, contain finite forms
of h md I and kr I instead of the nominal al-h amdu li- (1963, 105), and
after rejecting Bravmanns above-mentioned theory (2.2), Spitaler continues with the words:
Mit der Herauslsung aus seinem ursprnglichen legitimen Relativsatzgefge und der bertragung in eine ganz neue Konstruktion hat nun aber
unser Typus seine Entwicklung nicht abgeschlossen; es zeigt sich nmlich,
dass die Formel al-h amdu li-llhi als solche zurcktritt und durch andere
Ausdrcke ersetzt wird, die ebenfalls der usserung des Dankes, der
Freude, frohen berraschung, Befriedigung usw. dienen. [. . .] Die dabei
gegebenen phraseologischen Mglichkeiten sind sehr mannigfaltig [. . .].
Sie werden aber noch dadurch bereichert, dass die Konstruktion nunmehr
[. . .] auch fr die damit kontrastierenden Gefhle des Bedauerns, der Reue
usw. verwendet wird. (1963, 109)
There follow in Spitalers article numerous examples of these latter constructions both from post-classical writings and modern dialects, all of
them referring to events in the past. Additionally, Spitaler presents some
examples of allad that after verbs from other semantic fields, among
them some where the allad-clause, as in some of Blaus examples, refers
to future events expressed by the Arabic imperfect.
While Spitalers approach to the problem is definitely a monocausal
one, he was well aware of the fact that a multicausal approach would
also in principle have been possible, all the more in view of Hebr. aer
(on its origin see now Rubin 2005, 49f.) and Aram. d-, which, besides
their original function as relative particles, developed the meaning
that; because (1963, 106f.). In his final remarks, Spitaler emphatically
defended his own theory on the grounds: (a) that the Arab grammarians
noted the conjunctional allad only for the H amdalah, which therefore
seems to be central, (b) that the al-h amdu li-llhi llad type possesses the
chronological priority within the evidence of the conjunctional allad,
and (c) that it would be implausible to assume two different starting
points of the conjunctional allad converging into one phenomenon
(1963, 111f.).
There is no reference to Blaus Diqduq (1961) in Spitalers article,
probably due to the fact that he was unaware of it when he wrote the
article. In fact, Blau in some respect anticipated Spitalers theory about
the central role of al-h amdu li-llhi llad.

2.4
71
Joshua Blau (1965)
In his Emergence and Linguistic Background of Judaeo-Arabic (1965),

which appeared four years after Diqduq, Blau is essentially of the same
opinion as in Diqduq (1961) about the origin of allad that, and even
the wording reminds one of that used in Diqduq although it is more
explicit:
Since a relative clause may often be mistaken for a substantive clause, the
transition of allad into a conjunction introducing substantive clauses was
easily accomplished. (1965, 109)
After this general statement, Blau gives two examples: wa-kam alimta
min h adti bni Dwrdy jama m faalahu llad abida s-sanama and lillhi l-h amdu llad aslamahu f aydkum, in Blaus translation and likewise, you know the story of B. D., everything that he had done, that he
had served the idols and praise be to God that he has delivered him
into your hands. According to Blau, it is in both cases only the word
order that indicates that allad introduces a substantive clause and not
a relative clause, i.e. that allad is a conjunction and not a relative pronoun. While the second example indeed illustrates this transition, as
allad can be interpreted both as a relative particle and a conjunction
(praise be to God, who delivered him into your hands and praise be to
God that He delivered him into your hands), this, in my opinion, is not
possible with the first example as it admits of the interpretation of allad
as a conjunction only. From a semantic point of view, the allad clause in
this sentence is an apposition to m faalahu, which means that it cannot
be interpreted at the same time as an attribute to Ibn Dwrdy. Thus, this
example does not represent the starting point of the re-interpretation of
allad but allad in its new conjunctional function.
Spitalers theory is not taken into consideration by Blau, but his article
is mentioned in a footnote (1965, 109).
2.5
Joshua Blau (1967)
In contrast with Emergence (1965), where Spitalers article (1963) is

mentioned in a footnote only, having no impact on Blaus theoretical
approach, it is fully considered in the third volume of Blaus monumental Grammar of Christian Arabic, which appeared in 1967. Blau writes
there:
72
werner diem
alladh opens substantive clauses. Beginnings of this construction already
appear in CA2 and it becomes rather frequent in MA. A. Spitaler [. . .]
has brought out in full relief the early and very interesting history of the
type al-h amdu li-llhi-lladh thank God that and its development. The
use of alladh for introducing substantive clauses in ASP, insofar this is
not due to literal translation of Aramaic d-, is mainly from this type and
its developments. This is all the more remarkable since in JA other substantive clauses (especially apposition clauses) introduced by alladh are
quite common. It stands to reason that, in accord with Spitalers findings,
this use of alladh first developed in the type al-h amdu li-llhi. Early ASP
exhibits mainly this stage, whereas in later JA other kinds of alladh introducing substantive clauses have arisen. (1967, 526f.)
Among Blaus examples there is none of the nominal al-h amdu li-llhi
llad type. As to his examples of the corresponding verbal type that contains a reference to God, such as h amida llhi llad or akara llhi llad,
it must be stated that none of them unequivocally represent the new
type with allad as a conjunction because each can also be interpreted
as a relative clause referring to Allh. Of Blaus remaining examples,
many are likewise syntactically ambiguous, while some do represent the
new type. In several cases, Blau himself draws attention to the alternative interpretation as relative clauses. Apart from that, as Blau remarks
himself, some examples might be calques on the Aramaic d-, which, as
has been mentioned above, is both a relative particle and a conjunction
meaning that; because. Thus unequivocal evidence of the new type is
less for Ancient South Palestinian than might be inferred from the number of Blaus examples at first glance.
2.6
Manfred Woidich (1980)
Manfred Woidichs article illi als Konjunktion im Kairenischen (1980)

is an investigation into illi, which goes back to allad, as a conjunction in
modern Cairene Arabic. In his masterly analysis, which shows an interest and a competence in syntax not common for Arabic dialectology,
Woidich gives a thorough picture of illi as a conjunction from a synchronic perspective. Woidichs examples are mostly drawn from written
sources, but he also asked native speakers of what is called Educated
Cairene Arabic. It would be beyond the scope of this paper to comment
2
CA = Classical Arabic, MA = Middle Arabic, ASP = Ancient South Palestinian, JA
= Judaeo-Arabic.
73
extensively on what Woidich found out about illi as a conjunction in

Cairene Arabic. Therefore I shall concentrate on some points that are
also important from a historical perspective.
Woidich begins with the statement that the A (bergeordnete Ausdrcke), i.e. the heads on which the illi clauses as uS (untergeordnete
Stze) depend, are:
uerungen des Dankes, der Freude, der Befriedigung, des Bedauerns usw.
[. . .] Der Sprecher nimmt jeweils emotional Stellung zu einem im untergeordneten Satz (uS) geschilderten Sachverhalt und wertet ihn in diesem
Sinne. [. . .] Wir teilen das gesammelte Material nach der Art der AA ein,
von denen sich zwei Typen unterscheiden lassen. Die einen (Gruppe A)
werten den Sachverhalt des uS direkt, indem sie [. . .] emotional Stellung
nehmen, die anderen (Gruppe B) werten auf eine indirekte Weise, indem
sie angeben, wie eine Person sich oder andere wertet aufgrund des Sachverhalts des uS, oder zu welcher affektisch bestimmten Handlung eine
Person durch diesen veranlat wird. (1980, 225f.)
Examples of type A are il-h amdu li-llh illi ultaha b-nafsak Gott sei
Dank, da du es selbst gesagt hast! and kuwayyis illi wiit fi gbak inta
Gut, da es in deine Tasche gefallen ist! Characteristic of this type is
according to Woidich the absence of resumptive pronouns:
Eine Verknpfung des A mit dem uS durch pronominale Rck- und Verweise kann hier nicht stattfinden, da im A keine Pronomen auftreten.
(1980, 226).
Though this statement is corroborated by the above-cited examples, the

last two examples of Woidichs type A do have a pronominal concatenation, among them ayizni lli dayyat il-bazburt-i bti Es rgert mich,
da ich meinen Pa verloren habe. Woidich considers examples of this
type einen bergang zur Gruppe B (1980, 227). In my opinion, such
examples do not belong to group A but rather to group B, as expressions
like ayiz + object pronoun are a kind of pseudo-verbs, the pronoun
being the logical subject, and there are similar examples in group B.
As for type B, examples are d-ana batnaddim illi gt Ich bereue, da
ich gekommen bin and inta karihni lli bal il-h a Du kannst mich
nicht leiden, weil ich die Wahrheit sage. Concerning the syntactic characteristics of this type B, Woidich says:
Logisches Subjekt des A ist [. . .] eine Person, die im uS als Subjekt auftritt. Dadurch kommt eine Verknpfung der beiden Teilstze durch pronominalen Verweis zustande, der ein Charakteristikum fr diese Gruppe
darstellt. (1980, 227)
74
werner diem
Contrary to this statement, the person in the A (i.e., the head) to whom
the illi clause refers is not always the logical subject, as is, e.g., borne
out by the above-mentioned example inta karihni lli bal il-h a, where
the Person, die im uS als Subjekt auftritt is the (logical) object, not
the (logical) subject of the sentence. Furthermore, there are examples of
type B declared by Woidich to be marginal where there is no pronominal concatenation between the illi clause and an element of the head at
all, e.g. ana mabst illi ma-g I am glad he did not come (1980, 229).
Considering this evidence, it seems to me that the feature of coreferentiality is questionable for Woidichs type B in Cairene Arabic, and it is evident from Blaus, Spitalers and my evidence that it cannot be sustained
in a general diachronic and diatopic perspective. A certain degree of
coreferentiality is per se bound to exist for the simple reason that when
in the heads of sentences emotions are spoken of the causes of these
emotions as expressed in substantival clauses depending on those heads
are in most cases connected with the persons mentioned in the heads of
the sentences as having these emotions.
Important insights in Woidichs article are that in type B the element
of the head to which the illi clause refers must be human (1980, 230),
that the predicates are faktiv, that is, refer to real facts, mostly past
events (1980, 231), and that this kind of illi can always be replaced by
inn- that, with the difference that illi is considered by the informants
as being strker than inn-, that is, more affective (1980, 234).
2.7
Manfred Woidich (1989)
While Woidichs article of 1980 is essentially descriptive, his contribution illi dass, illi weil und zayy illi als ob: zur Reinterpretation von
Relativsatzgefgen im Kairenischen, which was published in 1989,
is diachronically oriented. As announced in the title, Woidich distinguishes between three different kinds of illi: illi that, illi because and
zayy illi as if .
(a) illi that
The type illi that corresponds to type A of Woidich (1980), e.g., il-h amdu
li-llh illi sabitak gottlob, da sie dich verlassen hat and kuwayyis illi
gt gut, da du gekommen bist (1989, 110f.). As for the origin of this
type, Woidich follows Spitaler (1963) in saying that il-h amdu li-llh in
75
sentences like il-h amdu li-llh illi waaik fiyya Lob sei Gott, der dich
mit mir zusammengebracht hat:
unterlag einer Funktionsschwche [. . .]. Als starres Syntagma wurde
diese Formel nicht mehr als analysierbar aufgefat, was zur Folge hatte,
da insbesondere N = allh nicht mehr als Nomen gesehen wurde, dem
ein syndetischer Relativsatz angeschlossen werden konnte. (1989, 111)
While in the assumption of Funktionsschwche Woidich explicitly follows Spitaler, he differs from him in assuming that the al-h amdu li-llhi
llad formula was re-interpreted because there are similar sentences in
Cairene Arabic having the same marked structure of Rhema-Thema,
that is, comment-topic, such as h ilwa di prima ist die! or b ikkalm
da eine Schande sind solche Worte! He sums this up by saying
da die Reinterpretation von illi als Relativpronomen zu illi da ausgelst
wurde durch den Umstand, da die ursprngliche syntaktische Struktur
mit dem eingebetteten Relativsatz nicht der thematisch-kommunikativen
Funktion dieser Stze entsprach. Sie war markiert und wurde durch Reinterpretation der bei diesen Stzen blichen funktionalen Satzstruktur
Prd.Subjekt angeglichen. Dadurch konnte dem illi die Funktion einer
Konjunktion zugeordnet werden, die Subjektsstze einleitet. (1989, 115)
Here we have to ask what Woidich intends to demonstrate from a

general Arabic perspective. Does he want to demonstrate a historical
development valid for Cairene Arabic only, or a general development
in Neo-Arabic the results of which are palpable in Cairene Arabic also?
A reconstruction of the first kind would be flawed by the fact that the
Cairene conjunctional allad cannot be detached from the common
history of Neo-Arabic, and a reconstruction of the second kind by the
fact that it would be problematic to reconstruct a common Neo-Arabic
development on the basis of the specific evidence of one modern dialect,
while disregarding other evidence, older and newer. Since we are concerned here with the second perspective only, which is tantamount to a
general reconstruction of the conjunction allad in Neo-Arabic, we have
to ask what Woidichs theory means for the history of the conjunctional
allad in early Neo-Arabic.
In Neo-Arabic the equivalents of Woidichs above-mentioned two
Cairene examples would probably be something like h ilwa hd and ayb
hd l-kalm. Comparing these sentences with a sentence like al-h amdu
li-llhi llad jita, it seems very improbable that the two types of sentences should have been mentally connected by any speaker of Arabic,
76
werner diem
and thus Woidichs theory seems to me to be farfetched. Nevertheless,

his drawing attention to the Rhema-Thema (or comment-topic) structure, which syntactically corresponds to a predicate-subject structure, is
an important insight, to which I shall come back below when discussing
my own theory (4.3).
Goldenberg comments on Woidichs theory with the words This form
is structurally parallel to the classical Arabic constructions of cleft sentences with m having the same order of constituents (1994, 16/261),
but I doubt the validity of this alleged parallel, as the cleft sentences
mentioned by Goldenberg, apart from their Rhema-Thema structure,
differ fundamentally from the sentences mentioned by Woidich.
(b) illi because
illi because corresponds to type B of Woidich (1980), and examples
are ana farh n illi uftak I am happy because (that) I saw you (1989,
116) or itnaddimit illi gat maya sie bereute es, da sie mit mir gekommen war (1989, 117). As for the meaning of illi in the second example,
it remains to be shown that in Cairene Arabic illi in such sentences can
be replaced by the causal conjunctions an and li-ann-. Interestingly,
no transformations of sentences containing expressions of this type are
among those adduced by Woidich (1980, 235) for the replacement of illi
by the causal conjunctions an and li-ann-.
In view of the semantic difference between the two kinds of illi in
Cairene Arabic as seen by him and also in view of the feature Referenzidentitt (that is, coreferentiality) in the second type, which he
thinks is essential, Woidich assumes for the second type (type B in
Woidich 1980) an origin independent of the al-h amdu li-llhi type (type
A in Woidich 1980). This origin he finds in sentences such as ana h mr
illi dafat il-h isb ich bin ein Esel, der ich die Rechnung bezahlt habe!,
which contains a direct relative clause depending on ana. Since in sentences of this kind there exists a causal connection between the relative
clause and the head, Woidich assumes that illi could be re-interpreted
as a causal conjunction, ich bin ein Esel, da ich die Rechnung bezahlt
habe!, adding that this re-interpretation is also true of alle anderen
Stze dieser Struktur (1989, 118). It goes without saying that if we are
to assume this kind of re-interpretation for early Neo-Arabic in general
(and not only for Cairene Arabic), we would have to assume that allad
had been generalized as a relative particle by then, an assumption which
poses no problem.
77
The fact that Woidich declares his example (13) itnaddimit illi gat
maya sie bereute es, da sie mit mir gekommen war not to belong
to the original type, confirms that, in his opinion, it is in nominal sentences that this type originated (1989, 119). Against this it can be argued
(a) that verbal sentences (such as itnaddamit illi) seem to prevail over
nominal sentences (such as ana h mr illi), and (b) that it seems highly
improbable to me that speakers should have used sentences like ana
h imruni llad addaytu d-dannra I am an ass, (I) who have paid the
dinars, an farh nuni llad najawtu I am glad, (I) who have escaped
or an mutaassifuni llad taaxxartu I am sorry, (I) who was late at all.
For this theory to be accepted, it would be prerequisite to find unambiguous relative clauses of this kind in Classical Arabic. I have checked
more than a thousand items with allad in Tradition, many of them in
dialogues, without finding even one example of such constructions, and
I dare say that it is very improbable that they occurred at all.
To sum up, as long as the syntactic type ana h imr illi dafat il-h isb,
which Woidich presumes to be the starting point of this type, is not
shown to have been a normal construction in pre-Neo-Arabic, I consider sentences of this kind in Neo-Arabic the result of a specific development, which should be explained otherwise, rather than the origin
thereof.
(c) zayy illi as if
In interpreting this third type, Woidich starts with cases such as ir-rgil
firih bna zayy illi laa lya der Mann freute sich ber uns, wie einer, der
einen Fund gemacht hat, in which the syntagma zayy illi wie jemand,
der; like somebody who could be interpreted as meaning als ob; as if .
In cases representing unequivocally the result of this re-interpretation,
only the interpretation of zayy illi as as if is possible, such as ma-kan-i
f tagwub, zayy illi kunt-i bakallim h agar es gab keine Resonanz, es war,
als ob ich mit einem Stein sprche.
As for earlier evidence of this type, the ka-llad mentioned in grammars, which structurally corresponds to zayy illi, has a different function, namely that of a conjunction of comparison with real facts, such as
xudtum ka-llad xd You have plunged as they have plunged Qurn
9:69, already adduced by Reckendorf (1921, 192, 4). For more evidence of this ka-llad, see Blau (1967, 527f.), Hopkins (1984, 238) and
Goldenberg (1994, 27f./276).
78
werner diem
An example of ka-llad that functionally corresponds to zayy illi as a

particle of comparison with unreal or hypothetical facts, as described by
Woidich, was adduced by Spitaler (1962, 108) in a footnote: fa-l yazlu
kadlika ka-llad yuh ibbu an yasxara min sh ibih 3 and he continues to
behave in this manner like one (or as if) intending to mock his friend,
sc. the tomcat (as-sinnawr) who plays with the mouse (al-farah) after
having caught her (al-Jhiz, H ayawn V, 202). Spitalers second example
concerns the well-known other ka-llad type: fa-sallaw ka-llad kn
yaf alna and they prayed as they always had. Another example of
ka-llad as a particle of comparison with hypothetical facts is wa-man
axadahu bi-irfi nafsin lam yubrak lahu fhi ka-llad yakulu wa-l
yabau and who takes it (sc. certain money) with haughtiness of mind
will not be blessed regarding it, like one (or as if) eating without getting
replete (al-Buxr, Sah h , Kitb az-zakh, No 1379), and more evidence
of this kind can be found in Tradition. Since in examples of this kind
allad can still be interpreted as a relative pronoun, they represent kallad in the meaning as if in statu nascendi. Obviously, Spitaler did not
realize that his two examples belong to two fundamentally different types.
The first ka-llad is a conjunction corresponding to kam (accordingly it should perhaps be transcribed as kallad), while the second
ka-llad consists of the particle ka- and the relative pronoun of the 3rd
masc. sing. In Cairene Arabic, this second type was grammaticalized as
zayy illi, with zayy standing for ka-, which is unusual in Cairene Arabic.
Future research should pay attention to the existence of two different
types of ka-llad.
It is, then, evident that ka-llad has to be separated from allad that,
which is the topic of this paper. Therefore, these remarks must suffice,
and I shall not return to this special type.
2.8 Gideon Goldenberg (1994)
In his article allad al-masdariyyah in Arab grammatical tradition
(1994), which was reprinted in 1998, Gideon Goldenberg begins with
general observations, then mentions Spitalers article (1962) and Woidichs second article (1989). Goldenberg describes the aim of his paper
as follows:
3
There follow four further an-clauses dependent on ka-llad yuh ibbu. Spitalers quotation has the variant bi-sh ibih.
79
About the distribution of infinitival allad in Arabic writings in general

I have nothing to add; in the present paper I just wish to adduce some
further evidence for a fuller understanding of the relevant structures as
described and treated in Arabic grammatical literature. Such examination
is important, because syntactical constructions that are considered grammatical by the great masters of Arabic grammar cannot easily be discarded
as inadmissible or non-Classical. (1994, 16f., 262)
Some of the examples that the Arab grammarians commented on are

from the Qurn, while other examples were made up by the grammarians themselves. A survey of the examples in the Concluding remarks
shows that the examples belong to disparate types. In some of them,
allad is combined with a preposition, such as ka-llad as, al llad for
that, bada llad after, in which allad stands for the more usual m. In
other cases, it is again either m, or an, which allad stands for. It might
even be doubted that some of the sentences on which the grammarians
dwelt in length ever occurred in normal speech, such as allad mararta
mamarrun h asanun It is a good passing that you passed, apart from
the fact that the syntactic interpretation of allad in this sentence is far
from being unequivocal.
As far as I can see, Goldenberg makes no attempt to draw historical
conclusions from his examples within the general discourse of conjunctional allad, nor can I do this myself. So, in spite of Goldenbergs thorough and learned approach, the historical dimension of his examples
has yet to be demonstrated.
3. Additional early documentary evidence

In the following, I shall give documentary evidence of allad as a conjunction, which is the result of a wide and systematic reading of premodern Arabic documents. The major part of the evidence comes
from Judaeo-Arabic documents found in the Genizah in al-Fustt (Old
Cairo). Considering that I likewise have read Arabic documents written
by Muslims as well as Christians, the preponderance of the Judaeo-Arabic evidence is noteworthy, being due to the more sub-standard character of Judaeo-Arabic writings as compared with Muslim and Christian
ones. The value of documentary examples lies in that they are chronologically well-defined, which is not necessarily the case with literary
examples, especially when these occur in sub-standard texts which have
come down to us through late manuscripts.
80
werner diem
Meanwhile, many of the Judaeo-Arabic documents which I had read

in their editiones principes have been re-edited by Moshe Gil in his two
comprehensive collections of Genizah documents. Where this is the case
I shall cite both editions, partially in order to give credit to the original
editors and partially because there are occasionally differences between
Gils re-editions and the original versions. While some interesting or
difficult cases of dialectal divergences from the literary language will be
commented upon in footnotes, substandard elements as such are not
systematically marked. Some of my examples have already been cited by
Blau (1961), as became clear to me while preparing this article. I have
retained these examples for the reason that they belong to an exclusively
documentary corpus, whereas Blau draws his examples from both literary and documentary sources, with the majority being literary.
My examples will be presented according to a linguistic typology
which mirrors the historical development of allad as I see it. My division owes much to Woidichs division of his Cairene material into his
types A and B (see above 2.6 and 2.7), even if his division and mine do
not fully overlap from a historical perspective as I assume more subtypes to have existed than he does. Many examples contain vulgarisms,
which I shall not mark systematically.
A. The head of the sentence expresses praise of or gratitude to God
(a) Nominal type
(1) li-llhi l-h amdu llad knati l-qibatu li-xayrin4
Praise be to God that the result (of the affair) was good
Goitein, Arkiyon, No 39r, 21 = Gil, Texts, No 215r, 20 (letter
to al-Fustt, early 11th c. C.E.)
(2) fa-li-llhi l-h amdu wa-l-minnatu llat5 knati l-aqbatu
h amdatan
So to God be praise and gratitude that the result (of the
6
affair) was praiseworthy
Goitein reads [il l-]xayri, Gil li-xayrin. Gil does not mention Goiteins reading.
allat, which seemingly refers to al-minnah, is due to a kind of hypercorrection.
6
Another example with allat that, or, more exactly, li-llat in order that, is lianna qawm mina l-maribati qad sr yusall indan li-llat (!7%%) yutbahum in
Goitein, Kneset, No 4v, 4f. Indeed, Goitein rendered !7%% as the Hebrew final conjunction *.)%. However, Gils re-edition in Texts, No 328, has %% instead of !7%%. So what
we have here is the simple prayer Allhu yutbhum May God reward them!sc. the
Maghrebis for praying in the synagogue of the Jerusalemites in Old Cairo.
5
81
Diem, Geschftsbriefe Wien, No 45r, 10 (letter to a merchant

in Egypt, 12th13th c.s C.E.)
(b) Verbal type
(3) fa-h amidn llha al dlika llad nazara ilayn wa-lam
yumit bin
and we praised God for this (happy outcome),7 namely that
He cared for us and did not have us mocked at8
Goitein, Iggeret r, 16 = Gil, Texts, No 616r, 17 (letter of
al-Mahdyah to al-Fust t, 11th c. C.E.)
(4) wa-h amidtu llha]9 an wa-ax llad masayn10 lah iqn al
xayr wa-fiyah li-Msr11
and we praised God], I and my brother, that we (finally)
reached Egypt on foot well and in good health
Toledano, Teudot, No 2, 1 (letter from Egypt, 1540 C.E.)
B. The head of the sentence does not express praise of or gratitude to God
(a) The head of the sentence contains an abstract noun expressing a
non-personal emotional evaluation of the contents of the allad
clause
(5) wa-m wajadn azan li-qulbin ayra annahu l-waylu
lan nah nu llad in li-hdihi l-masibi wa-narab hdihi
l-akwsa l-murrata
We did not find consolation for our hearts but (all we can
say is) that woe is us that we have lived (long enough to go
through) these disasters und (that we) have to drink these
bitter cups!
Gottheil and Worrell, Fragments, No 27r, 79 = Gil, Texts, No
501r, 79 (letter from Jerusalem to al-Fustt, 1065 C.E.)
7
The details of the happy outcome of the affair in question are imparted immediately before the passage cited.
8
Gil in his translation considers dlika an antecedens of allad in the sense of the
German darber, da. However, the alladclause is an apposition to dlika, which in
its turn refers to details mentioned before. In other words, the sentence is an expansion
of the usual fa-h amidn llha al dlika.
9
The addition is mine. Other additions are possible, but praise of God is the most
probable one.
10
Dialectal form for maayn.The language of the letter is substandard and exhibits features of Moroccan Jewish Arabic.
11
The grapheme renders Masar < Masr < Misr.
82
werner diem
(b) The head of the sentence contains a verb or a participle expressing the emotion of a person caused by the contents of the allad
clause
Joy
(6) wa-an bi-h amdi llhi mutabitun bi-llad ttasaltu ilayhim
I am, thank God, rejoicing that I joined them12 (in
marriage)13
Ashtor, Documentos, No 2r, 11 = Gil, Documents, No 457r,
12 (letter from Jerusalem to Toledo, 1057 C.E.)14
(7) wa-qad sarran llad anfadta lahu rah lahu
It pleased me that you sent him his merchandise
Gottheil and Worrell, Fragments, No 9v, margin, 1113 (letter from al-Fustt to Aden, probably 12th c. C.E.)
(8) fa-qad radtu laka bi-llad uxidtum fa-huli nsun
muhtaimn wa-llhi z-zabbln m rad bi-llad uxidtum
I was content concerning you that you (both)15 were punished <as well as certain other persons were content about
it> for these are decent people. Only the street-sweepers 16
were, by God, not content that you (two) were punished
Diem, Geschftsbriefe Wien, No 10, 8 (letter of a jealous wife
to her husband, 12th c. C.E.)
Wonder
(9) wa-tumma inn ajabu minka llad lam tusb man yaktubu
laka kitb ill daf atan
Furthermore, I am astonished at you that you (allegedly)
found only once somebody writing a letter down for you17
Goitein, Saloniqi, No 1r, 37f. (letter from Saloniki to alFustt, 11th c. C.E.)
12
Sc. the male relatives of the bride.

Ashtor conceives of allad as standing for alladna, translating Estoy contento de
aquellos con que [he emparentado por matrimonio].
14
According to Ashtor, the date is 1053. The difference consists in the reading of the
last letter in the date according to the Jewish era on the margin (ttyd vs. ttyh )
15
Sc. the addressee and his mistress, the plural standing for the dual in vulgar
language.
16
That is, the mob.
17
Sc. a letter to be sent to the writer of the letter, who is the addressees father.
13
83
Anger
(10) wa-an adbnu alayka y-ax katr allad wasal[ta il]
Misra wa-lam tasil il Adana
I am very angry about you, O brother, that you came to
Egypt (or Cairo), whereas you did not come to Aden18
Braslavski, Mishar, r 12 (Letter from Aden to al-Mahdyah,
c. 1149 C.E.)
Reproach19
(11) m baqiya alayn ayun ill llad lam tuarrifn kayfa
knat wasyatu xlika naxuduhu (!) llad lam tuarrifn in
kna wasala laka ayun mina l-kutubi
Nothing remains for us (to say) except that (until now)
you have not informed us as to how your uncles last will
was. What we reproach him (!)20 with is that you did not
inform me as to whether you received any of our letters
Gottheil and Worrell, Fragments, No 9v, 40f. (letter from
al-Fust t to Aden, probably 12th c. C.E.)
Grief
(12) wa-azza alayn dlika katr allad lam yakn ah adun
minn indaka yuwinuka f-m jar alayka f taabika f
m yaxussu amra xlika
and we were very much grieved by that,21 (namely) that
none of us was with you to assist you in that which you
had to endure in your concerns regarding the affair of (the
illness and death of) your uncle
Gottheil and Worrell, Fragments, No 9r, 12f. (letter from
(13) wa-dqa sadrun katr allad lam yakn laka maahu
kitbun yutamminun
and we were very much distressed that he had no letter
of yours with him setting our minds at rest (with regard
to you)
18
The writer intends to say that continuing the travel from Egypt to Aden would have
been easy for the addressee, his brother.
19
Only the second allad in (11) is an example of allad following a verb expressing
reproach. For the first allad, see (17).
20
Scribal error for you.
21
Reference to the illness and death of the addressees uncle, the details of which are
recapitulated by the writer before.
84
werner diem
Gottheil and Worrell, Fragments, No 9r, 28f. (letter from
(14) nulimukum ann bi-xayrin f fiyatin wa-qad azza alayya
katr allad lam takn h dir h att tuxallisa laka rah laka
I inform you (herewith) that I am well (and) in good health,
but (that) I am much grieved that you are not present so that
you might free your luggage (from the authorities)
Amari, Diplomi, 53, 2f. (letter from Tunis to Pisa, 12th
13th c.s. C.E.)
(c) The head of the sentence contains a verb or noun not expressing
an emotion
(15) wa-qad akartu tafaddulahumh arasahum llhullad
qad dakarn f kitbihim bi-s-salmi wa-bi-fili l-jamli f
bb
I am also grateful for their22 kindnessmay God protect
them(consisting in) that in their letter they gave greetings to me and (also consisting) in performing good deeds
to me
Assaf, Meqorot, 51, line 20f. = Gil, Documents, No 298r,
20f. (letter from Jerusalem to al-Qayrawn, 1039 C.E.)
(16) wa-jb lan kitbaka wa-nah nu laysa indan xabarun
h att jbahu r-rajulu l-warrqujazhu llhu ann xayr
allad nalamu
and he brought us your letter while we (still) had no news
of you until that man, the book-seller, brought itmay
God requite him in our stead with good that we (now)
know (how you are)
Gottheil and Worrell, Fragments, No 9v, 368 (letter from
(17) See (11), first allad.
I am the first to admit that 17 examples of allad that are a somewhat

meagre result of a decade-long reading of texts, but this result simply
shows how rare this phenomenon is in texts whose writers intended literary Arabic. Surprisingly scarce is especially evidence of the type A(a)
al-h amdu li-llhi llad.
22
Reference to two addressees, whose letter the writer answers.
85
Asking whether we can conclude from the scarcity of this type in documents, especially letters, that it was likewise marginal in the spoken
language, we must allow for some reservations. The H amdalah occurs
frequently in letters, mostly at the beginning but also in other parts.
Usually the H amdalah is mentioned in the context of news which are
deemed praiseworthy, as, for example, the writers or other persons
good health or a good outcome of a difficult situation. However, this
kind of the H amdalah is usually preceded by the report of the fact to
which it refers, as, e.g., katabtu ilayka . . . wa-an wa-man qibal bixayrin wa-fiyatin wa-l-h amdu li-llhi (al dlika) I am writing you . . .,
while I and my family are in good healthPraise be to God (for this).
This conventional structure prevents the al-h amdu li-llhi llad type
from frequently occurring in letters, which in its turn must also lead
to lower frequency of the re-interpreted al-h amdu li-llhi llad in letters than may have been the case in the spoken language. Documentary
texts where the al-h amdu li-llhi llad type usually does occur are, e.g.,
waqf documents, appointments of high officials and marriage contracts
between persons belonging to the upper class. In these kinds of texts, an
introductory al-h amdu li-llhi llad is expanded into long complicated
passages praising God for bounties related to the topic of the text, but
they are of so elaborated a style that they lack any deviation from the
literary language. To sum up, the rare occurrence of the re-interpreted
al-h amdu li-llhi llad in documents is of no relevance for the question
of its potential frequency in the spoken substandard language.
Scarce is also my evidence of the type B(a), where an abstract noun
expresses a non-personal evaluation of the contents of the allad clause,
especially when compared with the abundant evidence of this type in
Cairene Arabic as collected by Woidich (1980 and 1989). Again it is possible that the scarcity of early evidence of this type in documents is due
to specific circumstances. Expressions of this type are mostly exclamatory and thus have a clear Kundgabefunktion, which makes it more
likely for them to be used in the spoken language than in writing.
4. Historical typology
4.1
Introductory remark
Since the question of whether, and if so, how the different subtypes of
allad that are connected with each other is still open at this stage
86
werner diem
of this investigation (even if the answer to this question is admittedly

anticipated by the arrangement of my own material in the preceding
paragraph), I shall deal with them one by one. In doing so I shall consider whether there is reason to assume that the subtype in question
had an origin of its own independent of the other subtypes or whatever
other origin there might be. The sequence of the subtypes corresponds
to decreasing syntactic and semantic markedness and specificity.
4.2 Types A(a) al-h amdu li-llhi llad Praise be to God that and type
A(b) h amidtu llha llad I praised God that
There seems to be nobody so far denying the validity of Spitalers theory
that in al-h amdu li-llhi llad, to which my type A(a) corresponds, the
relative particle came to be re-interpreted as a conjunction meaning
da; that. In Spitalers view, this development is due to a Funktionsschwchung of allad, whereby das nachfolgende allad zwangslufig
den lebendigen Zusammenhang mit seinem Beziehungsnomen [verlor] (for details see above 2.3), and this opinion is whole-heartedly
shared by Woidich (see above 2.7).
In my opinion, this approach is flawed by the fact that, besides the
nominal type al-h amdu li-llhi llad, attested in the Qurn and elsewhere, there existed a corresponding verbal type, as e.g., in the saying
ascribed to Muhammad, y-maara l-arabi h mad llha llad rafaa
ankumu l-ura O company of the Arabs, praise God, who took the
tithes from you! Ibn H anbal, Musnad, Musnadu l-aarati l-mubairna
bi-l-jannati, No 1566. This verbal type is likewise attested in the form of
the re-interpreted verbal type A(b), along with the re-interpreted nominal type A(a). The existence of the re-interpreted verbal type A(b), whose
verb varies according to the syntactic context, clearly proves that when
the re-interpretation of al-h amdu li-llhi took place its constituents, in
spite of its formulaic character, must have been present in the minds of
the speakers and cannot have been a de-etymologized complex as, e.g.,
add how much? (< *qadr ayyi ayin) in Syrian Arabic or izzayy
how? (< * zayy) in Cairene Arabic. Both the nominal al-h amdu
li-llhi and the verbal h amidtu llha are equally attested as early as in
pre-Islamic poetry,23 which proves that they had existed side by side
from the first.
23
An example of al-h amdu li-llhi occurs in the Muallaqah of Imra al-Qays, verse
124, in Ahlwardts edition, and for examples of the verbal type see Brockelmann (1922,
87
Rather, the re-interpretation took place because the implicit logical structure of many items of the Qurnic al-h amdu li-llhi llad is
causal. A typical example is al-h amdu li-llhi llad najjn mina l-qawmi
z-zlimna Q 23:28, in Arthur J. Arberrys translation: Praise belongs to
God, who has delivered us from the people of the evildoers. This literal
translation of the relative clause is of course correct. However, when
considering the propositional structure of the Arabic sentence we can
easily discern that there is between the relative clause and the head of
the sentence an intrinsic connection consisting in the implication that
God is to be praised because He delivered the believers from the evildoers.24 It should be noted that this causal structure does not automatically exist in all al-h amdu li-llhi llad sentences but, it seems, only or
mainly in those which refer to certain individual bounties of God. In this
respect, it may, e.g., be doubted that the relation in al-h amdu li-llhi llad
xalaqa s-samwti wa-l-arda Praise be to God, who created the heavens and the earth Qurn 6:1 or al-h amdu li-llhi llad wasia samuhu
l-aswta Praise be to God, whose ear comprises all sounds an-Nas,
Sunan, Kitb at-ta lq, No 3406 and Ibn H anbal, Musnad, Bq musnad
al-Ansr, No 23064, is as causal as the first example or causal at all. But
the cases of al-h amdu li-llhi llad in the Qurn and still more those in
Tradition refer mostly to individual deeds of God and thus express also
an inherent causal relation.
A combination of a non-causal relation and a causal relation is, e.g.,
found in the following passage in Ibn Ab d-Dam a-fis (d. 1244
C.E.) Kitb adab al-qad: al-h amdu li-llhi llad ahidati l-uqlu
bi-qidamihi wa-wah dnyatih / wa-waqafat dna idrki jallihi waazamatih || allad btadaa l-maxlqti bi-badi h ikmatih / wa-sawwara
116). Additional verbal examples occur in Labds Dwn No 6, 1 (h amidtu llha wallhu l-h amdu etc.) and No 5, 1 (wa-llhu rabb mjidun mah mdun). For examples of
the related bi-h amdi llhi, see Brockelmann (1922), and an additional example occurs in
the Muallaqah of an-Nbiah, verse 12, likewise in Ahlwardts edition.
24
Such causal function of relative clauses is, it seems, not dealt with by Lehmann
1989 in his otherwise comprehensive morphological and semantic analysis of the relative clause in a great number of languages, including Arabic. Nor is his chapter Vom
Relativpronomen zur Konjunktion (1989, 389393) of much help for the problem of
allad as a conjunction, as he considers any relative pronouns that have lost their inflection to be Konjunktionen if there is no resumptive pronoun in the subordinate clause.
How Lehmann thinks the transition from uninflected relative pronouns to true conjunctions meaning that or because to have been remains unclear, in spite of some
final remarks (1989, 391).
88
werner diem
l-kinti bi-lutfi sanatih etc.25 Praise be to God to whose pre-existence

and uniqueness all minds testify and whose magnificence and power
they are unable to grasp, who made creation with His wonderful and
singular wisdom and designed the existing world with His kind work,
etc. The first allad clause is non-causal as it describes attributes of God,
and thus it is not possible to paraphrase it with a causal clause such as
*Praise be to God because all minds testify to His priority and uniqueness and are unable to grasp His magnificence and power. In contrast
to the first allad clause, the second one is inherently causal, which is
evident from the fact that it can easily be re-formulated as a causal
clause: Praise be to God because He made creation with His wonderful and singular wisdom and designed the existing world with His kind
work, etc. The causal character of this allad clause is due to the fact
that men own their existence to God having created them and thus have
to thank Him for His benevolence. It is also interesting that the content
of the first allad clause is of a more inherent nature than the second in
that it describes inherent, inalienable attributes of God, while the second refers to a deed that God performed of His own will. This hierarchy
of the two allad clauses conforms with a rule of Arabic syntax which
says that attributes follow the noun they refer to in order of decreasing
inherence.
Considering that there is an inherent causal relation in many items
of the Qurnic al-h amdu li-llhi llad between the head of the sentence
and the relative clause, it comes as no surprise that the allad that connects the two parts of the sentence and thus holds the position which a
conjunction explicitly expressing the causal relation would have, should
have been re-interpreted as a causal conjunction. This re-interpretation
was in all probability triggered by parallel constructions in which the
same causal relation is expressed explicitly. These constructions are as
follows:
(a) al-h amdu li-llhi al + noun/pronoun praise be to God for . . . and
h amidtu llha al + noun/pronoun I praised God for . . . In both
constructions, the al phrase indicates the reason why God shall
be praised. They are so frequent, particularly the former, that it is
superfluous to give examples.
25
Adab al-qad I, p. 247.
89
(b) al-h amdu li-llhi (al) an(na) praise be to God (for) that . . . and
h amidtu llha (al) an(na) I praised God (for) that . . . The preposition al can be elided as can any preposition preceding an(na).
Sentences of this kind seem absolutely normal, but I can adduce
only few examples of the verbal type, one from pre-Islamic times
and the other examples later: h amidtu llha an ams Rubayun *
bi-dri l-hni malh yan muqm (Wfir) I praised God that
Rubay got * in the house of disgrace, placed there in a shameful
way (Maqil b. Xuwaylid) Dwn al-Hudalyn I, No 14, 3; istahalla
yah madu rabbahu an l yakna asbahu d h aqqin f l-fayi and
he began to praise his Lord that none who had a right to the booty
had injured him ad-Drim, Sunan, Kitb al-muqaddimah, No 91;
allha ah madu al an jaalan min ulami l-arabyati God I
praise that he made me belong to the scholars of the Arabic language az-Zamaxar, Mufassa l, 2, 2. Another example of the verbal
type with God as the subject of an anna-clause is cited by Blau for
Christian Arabic for kr I: akara llha annahu lam yuxallhi yusu
il abdihi He thanked God that He had not let (or made) him
act unjustly to His servant (19661967, 526). With subordinate
clauses whose subject is coreferential with Allh (second and third
examples), this construction is optional instead of (a), whereas it
is the only possible construction for subordinate clauses whose
subject is not coreferential with Allh (first example). It cannot be
excluded that this construction is more usual in or even restricted to
the verbal type, which is less formulaic than the nominal type. More
examples would be desirable; their being so scarce so far is perhaps
due to my (and my predecessors) not noting them simply because
they are so normal and therefore do not seem to deserve attention.
(c) ah maduhu (al) an(na) I praise Him (for) that . . . Whenever
Allh is replaced by the personal pronoun, only this construction
is possible. This construction is especially frequent as a continuation of the relative clause type in the religious arengas (xutbahs)
of Mamluk documents: al-h amdu li-llhi llad . . . nah maduhu al
annahu . . . This construction would also be the only one possible
for the nominal head lahu l-h amdu but I have no examples.
(d) h amidtu llha id/fa- I praised God because. In this type, the subordinate clause is not dependent on the verb of the head as in the
preceding types but forms a subordinate causal clause. Examples
of each of the two conjunctions are the pre-Islamic verse h amidtu
90
werner diem
ilh bada Urwata id naj * Xirun wa-badu -arri ahwanu min
26
badin (Tawl) I praised my God after (the death of my brother)
Urwah (in battle) because (my son) Xir was (at least) saved, *
and some evil is easier to bear than another (Ab Xir) Dwn
al-Hudalyn III, No 14, 1, and, in Tradition, y-iata h mad llha
fa-qad barraaki llhu O iah, praise God because God has exculpated you! al-Buxr, Sah h , Kitb a-ahdt, No 2467.
These functional and syntactic correspondences are set off in the following table, in which all sentences express the notion of praise being
due to God for the salvation of the speakers, which is directly or indirectly ascribed to Him. Of the types marked with ?, no evidence has
been adduced so far, but they would be normal Arabic from a syntactic
point of view.
(a) Nominal type
Head
Expansion
Head
Expansion
al-h amdu li-llhi llad najjn

al-h amdu li-llhi al najtin
?al-h amdu li-llhi (al) an najjn
Praise be to God who saved us

Praise be to God for our salvation
Praise be to God (for) that He saved
us
?al-h amdu li-llhi (al) an najawn Praise be to God (for) that we were
saved
(b) Verbal type
Head
Expansion
Head
Expansion
h amidtu llha
h amidtu llha
h amidtu llha
llad najjn
al najtin
(al) an najjn
I praised God
I praised God
I praised God
h amidtu llha
(al) an najawn I praised God
h amidtuhu
(al) an najjn
I Praised Him
h amidtu llha
h amidtu llha
id/fa-najjn
id/fa-najawn
I praised God
I praised God
who saved us
for our salvation
(for) that He saved
us
(for) that we were
saved
(for) that He saved
us
because He saved us
because we were
saved
26
bad in the rhyme.
91
It is from this specific constellation that the re-interpretation of allad

in al-h amdu li-llhi llad and h amidtu llha llad as a causal conjunction
meaning (for) that started. Since the relative clause type shared with
the other types both the overall construction and the causal meaning of
the subordinate clause, allad could by analogy be interpreted as a particle expressing this causal relation. It is in sentences such as al-h amdu
li-llhi llad najawn Praise be to God (for) that we were saved and
h amidtu llha llad najawn I praised God (for) that we were saved,
with the subject of the allad no longer being coreferential with Allh of
the head, that the re-interpretation surfaced.
It should be stressed that this re-interpretation of the relative clause
as a causal subordinate clause was only possible because of the three following characteristics of al-h amdu li-llhi llad (and the corresponding
h amidtu llha llad), of which the first is syntactic, the second semantic
and the third morphological.
(a) The relative clause of al-h amdu li-llhi llad is non-restrictive,
which means that it does not serve the purpose of identifying Allh, who
is the object of praise. It is the loose connection typical of non-restrictive relative clauses that permitted the syntactic shift of the subordinate
clause from its status as a relative clause depending on the antecedent
Allh to its new status as a causal subordinate clause which no longer
depends on Allh but on the head of the sentence as a whole.
(b) Verbs such as to praise or to thank have, due to their specific
semantics, an inherent causal complement indicating the grounds for
praise or gratitude. Even if in the case of an isolated H amdalah in a
religious context the reason is not explicitly expressed it is nevertheless there, consisting in Gods general bounties, the knowledge of which
exists in the religious subtext. Thus whenever something positive is
mentioned in the syntactic context of a H amdalah it is in principle subject to being interpreted as being the reason for it. Exactly this happened
in the case of the relative clause of the H amdalah, but it happened only
there because it is only the non-restrictive clause that, due to its structure, has the syntactic potential of such a subordinate clause, while this
is not the case with, e.g., nominal attributes referring to Allh.
(c) The re-interpretation of allad may have been helped by the fact
that at the time when it happened the relative pronouns had probably
already been reduced in the spoken language to allad, which etymologically is the pronoun of the 3rd masc. sing. As a result of this development, the congruence between the antecedent and the relative pronoun
in terms of gender, number and, in the case of the dual, case had
92
werner diem
disappeared, and thus the syntactic connection between the antecedent

and the relative clause had been weakened.
However, behind this development there is yet another factor, which
is of an ontological nature. Of the constructions mentioned above sub (a)
and (b), those where the subject of the subordinate clause is not coreferential with Allh are as a rule not or less possible with other syntactic
objects of to praise, unless the specific reason of praise is mentioned
before or otherwise known to the person addressed. Thus one could say
in English (and correspondingly in German and Arabic and many other
languages), I would like to thank you that you helped me (coreferential subject of the subordinate clause), but usually not I would like to
thank you that we were helped (non-coreferential subject of the subordinate clause), whereas the second, non-coreferential construction is
a normal construction with God as the syntactic object of to thank, to
praise: (I) thank God that we were helped. The difference lies here in
that whenever something (positive) happens it can be attributed by the
believer to God as the one who with His power has caused it. This is why
the H amdalah (nominal or verbal) could be followed from the beginning by non-coreferential causal clauses (al an, an, id) as mentioned
above, and this is also why in al-h amdu li-llhi llad the coreferentiality
of the subject of the relative clause with its antecedent Allh was not an
inherently indispensable syntactic feature, a fact which made it possible
for coreferentiality to be given up in the course of the re-interpretation
of the relative clause.
As for the question of chronological priority of the re-interpreted
nominal and verbal types, Spitaler seems to have assumed the verbal
type to have been secondary (1962, 105). In support of this assumption it
may be argued, as Spitaler has, that there is evidence of the nominal type
from as early as the 9th century C.E. whereas the verbal type is attested
somewhat later, but this does not necessarily mean that the verbal type
had not existed before the earliest written evidence of it. More convincing is the argument that it is only the nominal type al-h amdu li-llhi
llad that is part of the Qurn and thus a formula the Muslims had been
familiar with from the beginning of Islam. Accordingly, we find in Tradition numerous examples of (the correct) al-h amdu li-llhi llad + a
coreferential verb and many examples of al-h amdu li-llhi al + a noun,
but few examples of the verbal type. We could even say that al-h amdu
li-llhi unexpanded or expanded has been and still is part of everyday
Muslim speech, while this is not the case with the verbal type to the same
93
extent. On the other hand, it is the verbal type who seems to exhibit
greater variability, and thus it may have been this type, or perhaps rather
the existence of this type in its various subtypes, that may have triggered
the re-interpretation of the more formulaic nominal type.
Besides the examples of h md, there are examples of the same basic
structures containing synonymous expressions, which shows a first
semantic generalization of the re-interpreted allad. There is my example
(2) fa-li-llhi l-h amdu wa-l-minnatu llat (!) knati l-aqbatu h amdatan,
and among Spitalers examples there is ukuri r-Rah ma r-Rah mn /
allad lam tajid al-malik Qays f hd l-makn danke dem Allbarmherzigen, dass du den Knig Qais nicht an diesem Ort gefunden hast
from the Antar novel, where additionally Allh is replaced by its synonym ar-Rah mn ar-Rah m, reversed there because of the rhyme. Blau
adduces further examples of synonymous verbs (sbh II, kr I, mjd II),
which, however, are syntactically ambiguous (1967).
We have reason to assume that the synonymous roots had already
been in use, besides h md, as alternatives of h md from the first, as we have
nominal examples such as wa-li-llhi l-h amdu wa--ukru To God praise
and thanks are due! (Dietrich, Briefe Hamburg, No 42a, 3 [c. 916/7 C.E.],
and likewise in Tradition fa-laka l-h amdu wa-laka -ukru Ab Dwd,
Sunan, Kitb al-adab, No 4411); wa--ukru li-llhi Thanks be to God!
(Anawati and Jomier, Papyrus chrtien, line 3 [9th c. C.E.]); wa-li-llhi
-ukru katran kam huwa ahluhu wa-mustah iqquhu To God repeated
thanks are due as he is entitled to and worthy of it (Rib, Lettres II,
No 17r, 5 [9th c. C.E.]). Furthermore, there are the Christian formulae
as-subh u li-llhi Praise be to God! and al-majdu li-llhi Glory be to
God!, for which the reader is referred to my Briefe Heidelberg, 24 and 25
respectively. An example of ukr and a relative clause is fa-aqimi -ukra
li-llhi llad snaka bi-dlika Therefore extend in the right way your
thanks to God, who preserved you thereby (Rib, Lettres I, No 5, 22
[9th c. C.E.]). When al-h amdu li-llhi llad and h amidtu llha llad were
re-interpreted, this development comprised the synonyms of h md also.
A final point to be made regards the translation of the re-interpreted
allad as that, da. Asking why allad can be translated with simple
that da, while, according to what I have tried to point out above, its
re-interpretation is based on an implicit causal relation, we should be
aware of a hitherto neglected fact: the syntactic opacity of that, da.
As with Arabic an and anna, prepositions preceding that, da in
English and German can or must be elided, which leads to the basic
94
werner diem
logical relations, as expressed by the prepositions, becoming blurred on

the surface. But the prepositions can also be retained on the surface,
probably more so in German than in English. Thus, instead of the syntactically opaque Lob sei Gott, da wir gerettet worden sind the more
explicit Lob sei Gott dafr, da wir gerettet worden sind is also possible.
4.3
Type B(a) malh allad (it is) good that
Half a dozen examples of this type were adduced by Spitaler (1962, 109f.)
for modern dialects, intermingled however with examples of the farihtu
llad type (for this latter type, see below 4.4). It was Woidich who distinguished between the two groups, of which he characterized the first one
as satzquivalente Ausdrcke (1980, 226). Examples of his type A for
Cairene Arabic are, apart from il-h amdu li-llhi illi Gott sei Dank, da,
e.g., baraka illi ein Segen, da, ya-xsra lli schade, da, and kuwayyis
illi gut, da. Corresponding examples were given by Spitaler for other
modern dialects, whereas he had, according to himself, not come across
examples in literary texts (1962, 110). Earlier evidence of this kind is my
example (5) from the 11th century C.E. containing al-wayl lan nah nu
llad woe is us that.27 Two other examples from that time quoted by
Blau are y-baxtik allad what good fortune for you (fem. sing.) that
and y-h ayf al abk llad what a pity for your (fem. sing.) father
that (1961, 227 sub ). These three early examples contain references to
persons in their heads, but this is also the case with two modern Cairene
examples cited by Woidich for this type A, and therefore this should not
lead us to assume historically different types.
As Woidich (1980) begins his Cairene evidence of this type with two
examples of the re-interpreted il-h amdu li-llh illi it is clear that he considers this type to be somehow the prototype of the whole group. This
conclusion is confirmed by Woidich (1989), where he derives this group
from the al-h amdu li-llhi llad expression, assuming as a first step
Funktionsschwche and as a second step a re-interpretation according to the Satzstruktur Prd.Subjekt as in h ilwa di prima ist die!
or b ikkalm da eine Schande sind solche Worte! As I have remarked
above (2.7), sentences of this kind are not likely to have been connected
by native speakers with al-h amdu li-llhi llad.
27
allad might also be interpreted here as a relative pronoun standing for alladna.
95
In my opinion the solution to this problem should not and need not
be looked for in Cairene examples such as h ilwa di or their early NeoArabic equivalents if any existed. But before going into details, I think it
appropriate to recapitulate some important points:
(a) That the type A(a) al-h amdu li-llhi llad is likely to be the starting
point of the type B(a) malh un allad may be assumed for the simple
fact that it is only al-h amdu li-llhi llad where the transition of the
relative particle allad to a conjunction can be explained, whereas
this is not feasible for malh un allad and the other expressions of
this type.
(b) The sentences of my type B(a) have a comment-topic structure.
(c) When asking for a link between the type A(a) al-h amdu li-llhi llad
and the type B(a) malh un allad, one can rightly assume, as Woidich did, that this link lies in that al-h amdu li-llhi llad had (finally)
been reduced to being a (positive) evaluation of the fact mentioned
in the allad-clause. This reduced meaning of al-h amdu li-llhi llad
would have been equivalent to good that; nice that.
However, there remains the question of the structure of the type B(a)
malh un allad. In this respect, my theory differs from Woidichs. In my
opinion, we should not start from sentences such as Cairene h ilwa di
or b ik-kalm da nor their Neo-Arabic equivalents in order to explain
malh un allad and similar expressions. Asking rather whether there
are other examples of the structure comment + that + topic in Classical Arabic (and Neo-Arabic), we can answer in the affirmative: there is
an absolutely normal type consisting of a noun + an(na) clause, as for
example malmun anna it is known that, h asanun anna it is good
that, barakatun anna it is a blessing that and so on.28 This is why
we can transpose all examples of my type B(a) into normal Arabic by
replacing allad by an(na), with potential small changes due to anna
demanding an accusative. It should be noted that the transformation
is always possible from allad to an(na) but not always from an(na)
to allad. Thus the transformation of malmun an(naka) jita into
?
malm allad jita would be very questionable, and for Cairene Arabic
?
malm illi gt would probably not be acceptable to most speakers. This
28
In modern literary Arabic, the type mina l-malmi anna would be more likely to
correspond.
96
werner diem
is due to the restriction of this allad type to specific heads referring to

emotions.
Starting from sentences evaluating events in an emotional manner
of the type barakatun (h asanun) an(na) + perfect, which were structurally and semantically equivalent to al-h amdu li-llhi llad + perfect,
the next step of the development, as I see it, was the replacement of
an(na) by allad in analogy with al-h amdu li-llhi llad. This development was made possible by two factors: (a) A semantic-syntactic connection between the type al-h amdu li-llhi llad + perfect and the type
barakatun (h asanun) an(na) + perfect was mentally established, due
to Funktionsschwche of the H amdalah. (b) The allad of al-h amdu
li-llhi llad was mentally connected with an(na), because it overlapped
with it in function after it had been re-interpreted as a conjunction.
The generalization of allad led to an opposition between an(na) and
allad, with allad as the marked element being subject to certain restrictions as compared with the unmarked an(na). These restrictions, which
the type B(a) had inherited from the primary type al-h amdu li-llhi llad
Praise be to God who via the secondary type A(a), are as follows:
(a) The structure of the type B(a) is always comment + allad + topic
and cannot be reversed. Thus malh allad jita It is nice that
you came is possible, whereas *?allad jita malh is not. Woidich
remarked this for his type B, with which my type B(b) partially
overlaps, but this restriction is valid of his type A and my type B(a)
also (1980, 235).
(b) The head of the sentence consists of an evaluating expression.
Thus malh allad jita It is nice that you came is possible, whereas
*?malm allad jita It is known that you came is usually not.
(c) The logical subject of the allad-clause is human. Woidich remarked
this for his type B for the syntactic subject of the allad clause, but
the same holds true for this type (1980, 230). Thus a sentence such
as malh llad wasala is possible in the meaning It is nice that he
(sc. a certain person) arrived, whereas the meaning It is nice that it
(sc. a letter or a parcel) arrived is excluded or questionable. What,
however, is attested in both the old and new evidence are allad
clauses with a non-human syntactic subject which however refer
to a logical human subject. Thus, a sentence such as malh allad
wasalan It is nice that it (sc. the letter) reached me or, more freely,
97
It is nice that I received it would probably be possible, though

perhaps not accepted by all speakers of individual dialects.
(d) The subject of the allad clause is almost exclusively pronominal
and as such mostly implied in a finite verb. Thus again a sentence
such as malh allad wasala It is nice that he arrived is possible,
whereas malh allad wasala Ah mad It is nice that Ahmad arrived
is with all probability either not possible at all or at least doubtful in
most dialects.
(e) The allad clause refers to actual, mostly past events. Woidich found
out this feature of Faktivitt for his type B, but this is also true of
his type A and my type B(a) (1980, 231). Thus malh allad jita It is
nice that you came is possible, whereas malh allad taj It would
be nice for you to come is not in most dialects.
These restrictions of allad in type B(a) as compared with an(na) corroborate the assumption of its being the result of a generalization of
the original al-h amdu li-llhi llad Praise be to God who, whereby the
head was replaced with similar expressions while the other features of
the construction were retained:
(a) al-h amdu li-llhi llad najjn has the structure al-h amdu li-llhi +
allad clause, which is not reversible.
(b) The head is evaluating.
(c) The subject of the allad clause, which is Allh, is, in a anthropomorphistic perspective, human.
(d) The subject of the allad clause is pronominal, being implied in the
verb that refers to Allh.
(e) The relative clause refers to past events only due to the semantics of
to praise in the meaning of to thank.
4.4
Type B(b) farihtu allad I was glad that
Turning now to the type farih a allad he was happy that, that is, allad
after verbs expressing emotions, the function of allad as a conjunction
is clearly borne out by cases where it follows a preposition as in (6) and
(8) of my examples which have bi-llad by that.
While Spitaler holds that this type goes directly back to the al-h amdu
li-llhi llad type by way of generalization (1962; see above 2.3), Woidich
98
werner diem
declares it to have had a separate origin in relative clauses where the subject
of the head and the subject of the allad clause are coreferential, as in Cairene
ana h mr illi dafat il-h isb (1989). In my opinion, this interpretation
is flawed in several respects, for which the reader is referred to paragraph 2.7.
Even if I do not find Woidichs historical explanation of this type convincing, it is worthwhile to mention that the sentence which, according
to him, was the starting point shares with al-h amdu li-llhi llad, in spite
of their different structures, the three characteristics mentioned above:
(a) The relative clause is non-restrictive, (b) there is an underlying causal
relation between the head and the relative clause (I am an ass! Why?),
(c) illi goes back to the uninflected allad, to which the relative pronouns
had been reduced.
Before continuing, I think it appropriate and useful to draw attention to the fact that this type B(b) shares with type B(a) the restrictions
(b)(e) mentioned above, that is, all restrictions with the exception of
the one concerning specific heads. As in the case of type B(a), a historical theory must take account of these restrictions and explain why they
exist. The assumption that this type had developed from coreferential
relative clauses as claimed by Woidich would explain the restrictions
(a)(c) but not (d) nor (e), as some emotions can concern both past
and future events. Therefore, it is plausible to assume that this type B(b)
developed from the verbal type A(b) h amidtu llha llad in the same way
as the type B(a) developed from the nominal type A(a) al-h amdu li-llhi
llad. In the first case, features of an impersonal expression were generalized, and in the other case, features of a personal expression.
This development, as I see it, was as follows: The inherited conjunction in Arabic sentences of the type I was glad that he came was in
Classical Arabic, and still is in most dialects, an(na), e.g. farihtu an (or
annahu) ja in the classical language. In a certain sense, praising God
for an agreeable event that has happened is an expression of an emotion. Saying, for example, in German Ich danke Gott, da ich davon
verschont geblieben bin, is equivalent to Ich bin froh, da etc. This
semantic overlap made it possible for the allad of h amidtu llha llad
I praised (thanked) God that to be generalized to verbs expressing
likewise the positive emotions of joy, contentment etc. Subsequently, as
Spitaler assumed, this generalization was extended, by Kontrastanalogie, to verbs expressing negative emotions such as anger, reproach, grief
and so on (1962, 109).
99
As a result of this development, the replacement of the marked allad

in B(b) by the unmarked an(na) is always possible, whereas the replacement of an(na) by allad is only possible where the an(na) clause fulfills
the above-mentioned restrictions.
4.5
Type B(c) alamtuhu llad I informed him that
In Spitalers, Blaus and my material examples remain where allad that

is preceded by verbs not expressing emotions. As I have mentioned
above (2.3), this was a major problem for Spitaler.
In my opinion, the problem is less serious than Spitaler thought it to
be. It is true that the heads of these sentences do not contain expressions
of emotion, but the contents of the allad-clauses are something calling
for positive or negative comments. Take, for example, the two following
examples cited by Spitaler: wa-axbarhu llad qatalahu Antar und sie
teilten ihm mit, dass Antar ihn gettet hatte (1962, 111) and, quoting
Michel Feghali (1928, 313) for Lebanese Arabic, min hayk illi twaffiqti
b-jztik cest cause de cela que tu as t heureuse dans ton mariage!,
with Feghali adding: avec ironie. Supporting evidence for this are also
my example (11) with its first allad and my examples (15)(16). In sentences of this type, allad has the function of expressing a certain emotion or the empathy of the speaker or narrator, the kind of which can be
inferred from the context only. Thus, allad in axbarhu llad qatalahu
Antar indicates a certain emotion on the part of the narrator, which,
depending on the specific context, can either be positive (joy) or negative (grief).
It should be stressed that this kind of affective allad is not a result of
a secondary development but something inherent in allad that from
the beginning. From the moment when in al-h amdu li-llhi llad and
h amidtu llha llad the relative pronoun was re-interpreted as a conjunction it became the marked counterpart of an(na), which through
this development became unmarked. The difference between them lay
in that an(na) continued to express the simple that, while allad as its
marked counterpart expressed a certain emotional involvement of the
speaker.
Should one try to express the specific affective content of allad in
sentences of this type B(c), one could do this, for example, by means
of adverbs. Consequently, the first of the two sentences could be rendered, depending on the context, alternatively with and they informed
100
werner diem
him that Antar had fortunately/regrettably killed him. Strictly speaking,

such explicit translation would be appropriate for any case of the conjunction allad as long as it is restricted to this and the aforementioned
types, translating, e.g., Cairene ana mutaassif illi taxxart with I am so
sorry that I was late.
4.6
Remaining cases
There remain some examples in Spitalers corpus and elsewhere where

the clause following allad (illi) does not refer to past, but to future or
possible events. It is interesting that two of Spitalers three examples are
from Tunisian Arabic, among them ammin elli tsil l-martek w-uldek
sei sicher, dass du zu deiner Frau und deinen Kindern kommen wirst.
Here we have the final point of the development of allad. As HansRudolf Singer pointed out for the dialect of Tunis, illi is used there in
every context (1984, 669), apparently in the same sense and distribution as would be an(na) in Classical Arabic and inn- in many modern
dialects including Cairene Arabic, e.g. andna z-zhar illi lqnhum29 wir
hatten Glck, da wir sie trafen on the one hand (referring to a past
event) and th il-ahid illi ma yfh h atta . . . er gab ihm das Versprechen, da er sie bis [zum Hochzeitstag] nicht anschauen wrde (future
30
event). In this dialect, illi that is unmarked, and its marked emotional counterpart for expressing that is kf (< kayfa how), for which
this function is not an isolated phenomenon in modern dialects nor in
the pre-modern substandard language. This remark about kf typical of
the dialect of Tunis must suffice here as the hitherto unwritten story
of kayfa as a conjunction is too long to be told here in a few words.
The dialect of Tunis is not the only one to have generalized the usage
of allad as a conjunction. The same usage of illi/li that is found in
Maltese Arabic (Schabert 1976, 216 and Aquilina 1987, I 566), which is
no coincidence as La langue maltaise a pour origine un dialecte arabe,
vraisemblablement proche des vieux dialectes citadins de Tunisie (Vanhove 1993, 1). The Jewish dialect of Tripolis in Libya is yet another dialect using lli/li that (Yoda 2005, 278). In addition, similar extreme
examples of illi that are known from other dialects but we lack compre-
29
I have simplified Singers complicated transcription.

See also Woidich (1980, 224) for the same function of illi in the Judaeo-Arabic
dialect of Tunis.
30
101
hensive studies of their exact distribution. For Cairene Arabic, one can
state on the basis of Woidichs two studies that examples as cited above
for Tunisian Arabic would most certainly be declared to be ungrammatical or unusual by native speakers.
As for the early history of Cairene Arabic, one should be cautious
about taking the language of the Jewish-Arabic documents of the Cairo
Genizah from the 11th century C.E. and later as early evidence of Cairene
Arabic, because many of the writers of those documents hailed from
the Maghreb, especially from what is today Tunisia, and there remained
strong bonds between the Jewish traders who had settled in al-Fust t
and elsewhere in the Islamic East and their relatives and partners in the
Maghreb. This means that deviations in those documents in the usage of
allad as a conjunction as compared with todays Cairene Arabic might
be ascribed to the Maghrebine background of their writers. Therefore
extreme examples of the conjunction allad in Judaeo-Arabic documents are to be connected to modern Tunisian Arabic rather than to
modern Cairene Arabic.
4.7
allad vs. an(na)
In describing the development of allad as a conjunction, I have not

dealt in detail with the various functions which it can assume. Some
additional remarks might therefore be helpful.
According to the development of allad as I see it, the relative pronoun
allad in al-h amdu li-llhi llad and h amidtu llha llad was first re-interpreted in the sense of that, which is normally expressed by an(na), and
then, by a gradual generalization, it replaced an(na) in specific syntactic
and semantic contexts. We can express this by saying that allad took
over part of the functions of an(na) step by step, a process that led to
an(na) and allad becoming the unmarked and marked members of an
opposition in many, if not all, dialects.
There is a well-known rule in Arabic that any preposition preceding
an(na) can be deleted. When allad replaced an(na) in specific contexts, which have been described above, allad consequently was felt to
be submitted to the deletion rule as its counterpart allad was. This meant
that where, according to the syntactic context, a preposition was to be
expected before allad in analogy with an(na), this preposition could
be considered to exist in the deep structure of the construction. On the
other hand, since a preposition could be retained before an(na) without
being deleted, allad could likewise be preceded by prepositions, again in
102
werner diem
analogy with an(na). To put it another way, the existence of cases such
as (6) wa-an . . . mutabitun bi-llad ttasaltu ilayhim I am . . . rejoicing
that I joined them, where allad is preceded by the preposition bi- typical of the verb bt VIII on which it depends, proves that the process
of the re-interpretation of allad as a conjunction meaning that had
taken place to its full extent. In the modern dialects, the combination
of illi with prepositions does not seem possible; at least Woidich does
not mention it for Cairene Arabic (1980 and 1989), and I do not have
examples thereof either.
Let us now have a look at some of my examples and see what syntactic status allad has and which prepositions are possibly missing. I
shall insert the prepositions missing in the surface structure but existing
in the depth structure in brackets, and also give a translation of these
prepositions:
(1)
(15)
(13)
li-llhi l-h amdu {al} llad knati l-qibatu li-xayrin

Praise be to God for the result having been good
Remark: The verb is h md I fulnan al to praise s.o. for s.th.
wa-qad akartu tafaddulahum . . . {bi-}llad qad dakarn f
kitbihim bi-s-salmi wa-bi-fili l-jamli f bb
I am also grateful for their kindness . . . of giving greetings to me
and of performing good deeds to me
Remark: The verb is fdl V bi- to be so kind as to do s.th.
wa-dqa sadrun katr {li-}allad lam yakn laka maahu kitbun
yutamminun
and we were very much distressed due to his not having a letter
of yours with him setting our minds at rest
Remark: (li-)llad indicates the cause as (li-)an(na) does.
On the other hand, there do exist many sentences where allad introduces a subject or an object clause so that no preposition can be supplemented in the deep structure:
(7)
wa-qad sarran llad anfadta lahu rah lahu

It pleased me that you sent him his merchandise
Remark: the allad-clause is the subject of the sentence.
There is also the adnominal usage of allad, forming, as is frequent with

an(na), a syntagma that seems to be an apposition to a noun. It would
103
be difficult to elucidate the true syntactic status of allad here. However,

this is not necessary; it suffices to say that the status of allad is here
analogous to the status of an(na), by which it can be replaced.
(9) inn ajabu minka llad lam tusb man yaktubu laka kitb ill
daf atan
I am astonished at you that you (allegedly) found only once somebody writing a letter down for you
5. Summary
In the preceding paragraphs, I have endeavored to sketch a picture of
the development of allad as a conjunction valid of all varieties of NeoArabic as far as they are known. The stages of the development are set
off in the following table:
(a) Original expressions containing relative clauses:
al-h amdu li-llhi llad
h amidtu llha llad
Praise be to God who
I praised God who
(b) Re-interpretation of allad as a causal conjunction on the pattern of
parallel constructions with semantically explicit syntactic means,
this process being due to (a) the relative clause being non-restrictive, (b) verbs such as to praise having an inherent complement
indicating the cause of praise, (c) the relative pronouns having been
reduced to allad, whereby the connection of the relative clause to
the head was weakened:
31
al-h amdu li-llhi llad
h amidtu llha llad
Praise be to God that
I praised God that
(c) Generalization of allad for heads expressing positive emotions, by
(optional) replacement of an(na) with allad, while the syntactic
and semantic restrictions typical of (a)(b) were retained:
malh allad (for malh anna)
farihtu llad (for farihtu anna)
It is nice that
I was glad that
(d) Generalization of allad for heads expressing negative emotions ceteris paribus as in (c):
31
And by analogy semantically related verbs such as kr I, sbh II and mjd II.
104
werner diem
wayl laka llad (for wayl laka

taassaftu llad (for taassaftu
anna)
anna)
Woe is you that
I regretted that
(e) Generalization of allad for heads not expressing emotions, with
allad still having an affective value and retaining its original restrictions as in (a)(d):
alamtuhu llad
I informed him (of the pleasant/regrettable fact) that
(f) Generalization of allad as an unmarked conjunction that without
syntactic or semantic restrictions:
th il-ahid illi ma yfh
He made a pledge that he would not see her (Tunisian Arabic)
Stage (f), which is found in the dialect of Tunis and the dialect of the
Jews of Tripolis (Libya), is the final point of the development of allad as
a conjunction. The development of allad (illi) in Cairene Arabic, which
is the best-known of all Arabic dialects, reached stage (d) only.
6. Appendix: Notes on the origin of al-h amdu li-llhi llad

As demonstrated in paragraph 4.2, there were, besides the relative clause,
other, more explicit means of expanding the H amdalah. This raises the
question, which has not yet been given the attention it deserves, why
in the case of al-h amdu li-llhi a causal relation between a subordinate
clause and the head is expressed by a relative clause if the subject of
the relative clause is coreferential with Allh of the head, and not more
explicitly.
Indeed, in terms of normal Arabic one is in some respect entitled
to say that al-h amdu li-llhi llad expressing causal relations is a special
case, which is, however, so innate in the religious language of Islam that
Muslims are not likely to be aware of this fact. On the other hand, the
very fact that the relative pronoun was re-interpreted, at some time in the
history of Arabic, as a conjunction demonstrates that this relative clause
was something peculiar not wholly in line with the normal syntax of
Arabic. That Western Arabists do not seem to have felt al-h amdu li-llhi
llad to be something special either, must be ascribed to the existence
of corresponding relative clauses in the religious style of the Western
languages. For German, I can say by way of introspection that this kind
of causal relative clauses, apart from the religious language, e.g. Lob
105
sei Gott dem Herrn, der etc., is restricted to a high stylistic level and is
possible for verbs such as danken, preisen, loben only, e.g. Es ist mir
ein Anliegen, an dieser Stelle Herrn N.N. zu danken, der dieses Projekt
hochherzig gefrdert hat. A corresponding formulation in less formal
language is, e.g., Ich mchte Herrn N.N. dafr danken, da er dieses
Projekt untersttzt hat. So the fact that al-h amdu li-llhi llad seems
familiar to us must not prevent us from enquiring as to the reason for
this construction. To this question the following lines will be dedicated.
Considering that al-h amdu li-llhi was in use in pre-Islamic times
already (see above 5.2), we can assume that its expanded form al-h amdu
li-llhi llad is likewise pre-Islamic, although there is no explicit evidence
of it so far, at least none that I am aware of. The basic formula itself is
probably a calque on the Syriac ubh l-alh, as remarked by Theodor
Nldeke and Friedrich Schwally (1909, 112, footnote 1) for the Ftih ah
of the Qurn: entspricht genau syrischem ubh l-alh32 bezw.
teboht l-alh und neutestamentlichem . In this context,
Nldeke and Schwally also mention the so-called Berk in the Old Testament and the Christian liturgy as a parallel to the Syriac and Greek
formulae, without, however, connecting it to the Qurn. Anton Baumstarks article Jdischer und christlicher Gebetstypus im Koran (1922)
is in a certain sense a comment on the short remarks of Nldeke and
Schwally, although he does not refer to them. In the following, I shall
first sum up some important points of Baumstarks article and then add
some deliberations of my own:
(a) In the Old Testament, there is the so-called Berk (blessing) of the
structure brk Yahw Blessed is (or be) Yahwe, to which a nominal attribute or a relative clause can be added, e.g. brk Yahw aer
hiss l etke m miy-yad Misrayim -miy-yad Par Blessed be Yahwe,
who delivered you from the hand of the Egyptians and the hand of
Pharaoh Exodus 18:10.33 A corresponding Qurnic expression is
tabraka in the subtypes tabraka llhu rabbu l-lamna Blessed
be God, the Lord of the whole world Q 7:54 and tabraka llad
bi-yadih l-mulku Blessed be the One in whose hand is the kingdom Q 67:1, as well as the exceptional brika in brika man f n-nri
32
In the original text in Syriac script.

This passage has already been cited by Spitaler (1962, 107) but in another context, where he asks whether this expression might be a parallel to the development of
al-h amdu li-llhi llad. Instead of Spitalers h iss l, read hiss l.
33
106
werner diem
wa-man h awlah Blessed be He Who is in the fire and around it

Q 27:8. Baumstark tends to assume that the Berk was transmitted to Muhammad via Christian formulae in the New Testament
and elsewhere, in which the Jewish Berk lives on, of the type
, a last echo of this development being Benedictus
dominus (Baumstark 1922, 231ff.). A fact not mentioned by Baumstark is that the Qurnic formula is syntactically different both
from the Biblical Berk and the H amdalah, as it lacks a subtype
tabraka llhu llad Blessed be God who.
(b) The type which in the Christian liturgy, especially of the East, has
become the prevailing one is the so-called doxology, which has probably developed from the Hebrew Berk, e.g. o , Gloria tibi
domine, Laus tibi Christe, and so forth, the doxology being [eine]
possessive Form der Prdikation, durch welche Herrlichkeit, Ruhm,
Lob, Ehre oder wie immer man das schillernde griechische
wiedergeben will, als der Gottheit eigen oder gebhrend bezeichnet wird (Baumstark 1922, 234). The dative of the Greek formula
can be expanded by eine partizipiale Apposition oder Anrede, and
the reason for Gods praise can also be given in the form of a
clause. This Christian doxology is likewise found in the Qurn, in
two forms: as subh na + a pronoun or a genitive (which does not
interest us in this context) and as al-h amdu li- (Baumstark 1922,
234ff.). However, as Baumstark remarks:
Der frhchristlichen partizipialen Ergnzung einer Doxologie entspricht
diejenige eines
fast immer vielmehr durch einen Relativsatz [. . .]
(1922, 237)
After dealing with details of subh na and al-h amdu li- in the Qurn and
comparing them to Christian liturgical formulae they are likely to be
derived from, Baumstark finally poses the question of how these formulae may have passed to Muhammad. He supposes that they were brought
to the Arabs by Nestorian missionaries, who translated their Syriac religious texts into Arabic, which he says is typical of the Nestorians:
Die nestorianische Kirche hat aber immer wieder eine auffallende Geneigtheit bekundet, ihre angestammte syrische Kultsprache beim Betreten
neuer Missionsgebiete der Volkssprache derselben zu opfern (1922,
247f.)
So far my summary of Baumstarks article. As we have seen, Baumstark

is aware of the fact that the Qurnic al-h amdu li-llhi is expanded by a
107
relative clause while there are participles and clauses in the Greek
doxologies, but he does not pay further attention to this difference. In
my opinion, the solution to this problem is found in the Syriac context,
through which the doxology was probably transmitted to the Arabs.
Baumstark (1922, 235) mentions himself that the Syriac doxology is
ubh l-, without however further dwelling upon it, while Nldeke and
Schwally (1909), as mentioned above, were of the opinion that al-h amdu
li-llhi goes back to ubh l-alh. It is also interesting in this context
that for Barhebrus, as cited by S. Payne Smith (1879 II, 4026), the
ubh l- formula must have been closely connected to the H amdalah, as
he renders in his dictionary the Syriac ubh l-mrayym l-lmn Praise
be to the Elevated One to all eternity with the Arabic al-h amdu li-l-l
il d-dahri.
In the Syriac New Testament, the simple doxology is part of a sentence
in w-kulleh amm da-h z yab-w ubh l-alh Et omnis plebs ut vidit,
dedit laudem Deo Lk 18:43. Syntactically independent simple doxologies are ubh l-alh Praise be to God (Payne Smith 1903, 563a) and
ubh l-ab w-la-br wa-l-rh qadd Praise be to the Father, the
Son and the Holy Spirit (Payne Smith 1879 II, 4026). As for expanded
doxologies, we find evidence of them in the Breviarium Chaldaicum, e.g.
ubh l-tb da-b-yad h ubbeh gl teboht la-bnayn Praise be to the
Good One, who through His love revealed glory to men (Breviarium
I, kh = 28) or ubh l-mry da-b-yad tuqpeh hg
am la-trn wa-p
raq
l-abdaw Praise be to the Lord, who through His power overthrew
the tyrant and saved His servants (Breviarium III, 333). Examples of
expanded doxologies in colophons of books are ubh l-ab w-la-br
wa-l-rh qadd d-h ayyel wa-dar w-sayya Praise be to the Father,
the Son and the Holy Spirit, who strengthened, assisted and supported
(respectively) and ubh l-ab d-h ayyel w-la-br d-sayya wa-l-rh
d-qud d-aml Praise be to the Father, who strengthened, the Son,
who assisted, and the Spirit of Holiness, who accomplished (Payne
Smith 1879 II, 4026).
We can safely infer from these examples that the expanded Arabic doxology al-h amdu li-llhi llad is nothing but a rendition of the
Syriac ubh l-alh d- in the same way as the simple Arabic doxology
al-h amdu li-llhi is a rendition of the Syriac ubh l-alh. This means
that the Arabic relative pronoun allad corresponds to the Syriac d-,
which in its chief function is also a relative pronoun. It is true that d- is
also a conjunction of wide and vague meaning so that the d-clauses of the
108
werner diem
above-mentioned Syriac doxologies might be translated as causal clauses

as well, e.g., ubh l-ab d-h ayyel w-la-br d-sayya wa-l-rh d-qud
d-aml as Praise be to the Father because He strengthened, the Son
because he assisted, and the Spirit of Holiness because he accomplished.
On the other side, it may be argued that the d-clauses of Syriac doxologies, at least to my knowledge, are all coreferential with the nouns of the
l-phrases and thus likely to have been conceived of as relative clauses
rather than as causal clauses.
Although the relative pronoun of the Qurnic al-h amdu li-llhi llad
is thus accounted for, there remains the question why the Syriac doxology is expanded by d-clauses while the older Greek formulae have participles and clauses. The solution to this problem lies probably in the
difference of the verbal systems. The Greek verbal system, which per se
is more complex than the West Semitic one, is still more so with regard
to the active participle, as there are different Greek participles derived
from the corresponding tenses, whereas the Semitic languages have but
one active participle, whose temporal value depends on various factors.
Therefore, when Greek doxologies with active participial attributes
were translated into Syriac, the problems connected with this disparity
seem to have been avoided by translating the Greek participle, if the
specific context demanded this, with a relative clause containing a finite
verb in the (Syriac) perfect. As the Syriac relative pronoun d- is also a
causal conjunction, Greek expansions of the doxology by means of the
conjunction that; because could likewise be rendered with d-. In
this process of rendering Greek attributive participles and clauses
with Syriac d- clauses, the translators may also have been influenced
by the Berk of the Old Testament with its aer, which is also both a
relative particle and a causal conjunction, and which in the Syriac Old
Testament is rendered with d-. Thus the Hebrew brk Yahw aer hiss l
etke m miy-yad Misrayim -miy-yad Par Exodus 18:10 is rendered
as the Syriac brk- mry d-p
ass kn men d d-Misry w-men deh
d-P
ern, with d- exactly corresponding to aer.
Even if there exists, as I have tried to demonstrate, a historical connection between the Hebrew brk Yahw aer, the Syriac ubh l-alh
d- and the Arabic al-h amdu li-llhi llad, this does not fully explain why
relative clauses were chosen from the beginning and not explicit causal
conjunctions as the Hebrew k, the Syriac mettu l d- or the Arabic al
an or li-an. The reason is probably that the implicit indication of the
cause is more in line with the religious attitude deemed appropriate
109
towards God. If an explicit causal conjunction were used the praise of

God would be limited to a specific deed of His as mentioned in the subordinate clause, that is, God would be praised for this specific deed and
for nothing else. But this would not be adequate as God, according to
Jewish, Christian and Islamic thinking, is entitled to praise in general,
as is clearly borne out by the unexpanded Berk and the unexpanded
doxology. Thus by not explicitly expressing an (implicitly existent)
causal relation the relative clause is much more in line with a general,
unconditioned praise of God than an explicit causal clause would be.
Accordingly, the Arabic H amdalahs, nominal and verbal ones, where
the subordinate clause is introduced by an explicit conjunction (including the re-interpreted allad followed by non-coreferential clauses) do
not imply a general praise of God but are limited to praise of God for
specific facts. In this respect, sentences of this kind belong to the language of everyday life and not the language of religious worship, a fact
which is demonstrated by the contexts in which such sentences occur.
A final remark concerns the Arabic verbal type h amidtu llha llad/
an. There is the Syriac verb abbah to praise, and this verb is likely to
have been used with God as the object. However, the dictionaries do
not contain evidence of a formula such as abbh et l-alh d- I praised
God who (or that) used for the expression of gratitude to God for individual bounties as the Arabic h amidtu llha llad/an is. Furthermore,
the verbal forms in both languages are different. Consequently, the verbal h amidtu llha llad/an is the result of a specific Arabic development
rather than a calque on a Syriac expression. This in turn would suggest
that the doxology had a history of its own in Arabic after it had been
taken over in its formulaic nominal form from Syriac. Now, as Hartmut
Bobzin (2004, 59) has stressed, it can be concluded from certain passages of the Qurn that even in pre-Islamic times Allh had been considered a Schpfer- und Rettergott, a designation of which the notion
of Rettergott (saviour God) is significant in this context. For if Allh
was considered a Rettergott he was especially qualified for the H amdalah as in every day life the H amdalah, nominal or verbal, is mostly
used for the happy outcome of an affair. It seems, therefore, that in the
Arabic H amdalah the Christian doxology might well have melted in a
syncretistic way with pre-existing pagan notions of the role of Allh.
110
werner diem
7. References
7.1 Primary sources

Ab Dwd, Sunan = Ab Dwd Sulaymn b. al-Aat as-Sijistn, Kitb as-Sunan.
See Mawsat al-hadt.
Ahlwardt, Wilhelm see Arazi and Masalha
Amari, Diplomi = Michele Amari, I diplomi arabi del R. Archivio fiorentino. Florence:
Tipografia di Felice le Monnier, 1863.
Anawati and Jomier, Papyrus chrtien = PP. Anawati and Jomier, Un papyrus chrtien
en arabe (gypte, IXe sicle Ap. J.-C.). Mlanges islamologiques 2, 1954, 91102.
Arazi and Masalha, Six Early Arab Poets = Albert Arazi and Salman Masalha, Six Early
Arab Poets. New Edition and Concordance Based on W. Ahlwardts The Divans of the
Six Ancient Arabic Poets. Jerusalem: The Hebrew University, 1999.
Ashtor, Documentos = Eli Ashtor, Documentos espaoles de la Genizah. Sefarad 24,
1964, 4180.
Assaf, Meqorot = Simha Assaf, Meqorot u-meh qarim be-toledot Yisrael. Jerusalem, 1946.
Braslavski, Mishar = Josef Braslavski, Al ham-mishar hay-yehudi ben hay-yam hattikon wa-Hoddu bam-meah ha-12. Zion 7, 19411942, 135139.
Breviarium Chaldaicum = Breviarium iuxta ritum Syrorum orientalium id est Chaldaeorum. Slawwt qnnyt d-khn. Rome, 2002. (Reprint of Eugne Tisserants edition, Rome 1938, itself a reprint of the original edition, Rome 1886.)
al-Buxr, Sah h = Muhammad b. Isml al-Buxr, Kitb al-Jmi as-sah h. See Mawsat
al-hadt.
ad-Drim, Sunan = Abd Allh b. Abd ar-Rahmn ad-Drim, Kitb as-sunan. See
Mawsat al-hadt.
Diem, Briefe Heidelberg = Werner Diem, Arabische Briefe auf Papyrus und Papier aus der
Heidelberger Papyrus-Sammlung. Wiesbaden: O. Harrassowitz, 1991.
Diem, Geschftsbriefe Wien = Werner Diem, Arabische Geschftsbriefe des 10. bis 14.
Jahrhunderts aus der sterreichischen Nationalbibliothek in Wien. Wiesbaden: Harrassowitz, 1995.
Dietrich, Briefe Hamburg = Albert Dietrich, Arabische Briefe aus der Papyrussammlung
der Hamburger Staats- und Universitts-Bibliothek. Hamburg: J. H. Augustin, 1955.
Dwn al-Hudalyn = Kitb arh ar al-Hudalyn sanat Ab Sad al-H asan ibn
al-H usayn as-Sukkar. Ed. by Abd as-Sattr Ahmad Farrj and Mahmd Muhammad
kir. Cairo: Dr al-Urbah, 1965.
Gil, Documents = Moshe Gil, Palestine during the First Muslim Period (6341099). Part I.
Studies. Part II. Cairo Geniza Documents. Part III. Cairo Geniza Documents. Tel Aviv:
Tel Aviv University etc., 1983.
Gil, Texts = Moshe Gil, In the Kingdom of Ishmael. Volume I. Studies in Jewish History
in Islamic Lands in the Early Middle Ages. Volume II. Texts from the Cairo Geniza.
The Jews of Iraq and Persia (nos. 1101). Letters of Jewish Merchants (nos. 102303).
Volume III. Texts from the Cairo Geniza. Letters of Jewish Merchants (nos. 304607).
Volume IV. Texts from the Cairo Geniza. Letters of Jewish Merchants (nos. 608846).
Indexes. Jerusalem: Tel Aviv University etc., 1997.
Goitein, Arkiyon = Shelomo Dov Goitein, Olelot me-arkiyono el Ysuf ibn Awkal.
Tarbiz 38, 1968, 1842.
Goitein, Iggeret = Shelomo Dov Goitein, Iggeret Labrat ben Moe ben Simr dayyan
ha-ir al-Mahdyah. Tarbiz 36, 1967, 5972.
Goitein, Kneset = Shelomo Dov Goitein, Bet hak-kneset we-siyyudo lepi kitbe hagGenizah. Erez Israel 7, 1964, 8197.
Goitein, Saloniqi = Shelomo Dov Goitein, Eduyot qedumot min hag-Genizah al
qehillat Saloniqi. Sefunot, 11, 1970, 1133.
111
Gottheil and Worrell, Fragments = Richard Gottheil and William H. Worrell, Fragments
from the Cairo Genizah in the Freer Collection. New York: Macmillan, 1927.
Ibn Ab d-Dam a-fi, Adab al-qad = ihb ad-Dn Ab Ishq Ibrhm b. Abd
Allh al-Hamdn al-H amaw al-marf bi-bn Ab d-Dam a-fi, Kitb adab
al-qad. Ed. by Muhy Hill as-Sarhn. 2 volumes. Bagdad: Matbaat al-ird, 1984.
Ibn H anbal, Musnad = Ahmad b. H anbal, Kitb al-Musnad. See Mawsat al-hadt.
al-Jhiz, H ayawn = Ab Utm
n Amr b. Bahr al-Jhiz, Kitb al-h ayawn. Ed. by Abd
as-Salm Muhammad Hrn. Cairo: Matbaat Lajnat at-talf wa-t-tarjamah wa-nnar, 1969.
Labd, Dwn = arh Dwn Labd b. Rabah al-mir. Ed. by Ihsn Abbs. Kuwait:
Governmental Press, 1962.
Mawsat al-hadt = Mawsat al-hadt a-arf. al-Kutub at-tisah. CD-Rom. 1st edition.
arikat Saxr li-barmij al-hsib (19911996).
an-Nas, Sunan = Ab Abd ar-Rahmn an-Nas, Kitb as-Sunan. See Mawsat
al-hadt.
Rib, Lettres III = Ysuf Rib, Lettres arabes. Annales islamologiques 14, 1978,
1535; 16, 1980, 129.
Toledano, Teudot = Jacob Moses Toledano, Teudot mik-kitbe yad. Hebrew Union
College Annual 4, 1927, 449467.
az-Zamaxar, Mufassa l = Ab l-Qsim Mahmd b. Umar az-Zamaxar, Kitb
al-Mufassa l f n-nahw. Ed. by J. P. Broch. Christiania etc.: P. T. Mallingius, 1879.
7.2
Secondary sources
Aquilina, Joseph. 1987. Maltese-English Dictionary. Malta.

Baumstark, Anton. 1927. Jdischer und christlicher Gebetstypus im Koran. Der Islam
16, 229248.
Blau, Joshua. 1961. Diqduq ha-arabit ha-yehudit el yeme hab-benayim. Jerusalem:
Magnes Press, The Hebrew University.
. 1965. The Emergence and Linguistic Background of Judaeo-Arabic. A Study of the
Origins of Middle Arabic. Oxford: Oxford University Press.
. 19661967. A Grammar of Christian Arabic Based Mainly on South-Palestinian
Texts from the First Millennium. 13. Louvain: Secrtariat du Corpus SCO.
Bobzin, Hartmut. 2004. Der Koran. Eine Einfhrung. Mnchen: C.H. Beck, 5th edition.
Bravmann, Meir M. 1953. Studies in Arabic and General Syntax. Cairo: Imprimerie de
lInstitut franais darchologie orientale.
Brockelmann, Carl. 1922. Allah und die Gtzen, der Ursprung des islamischen Montheismus. Archiv fr Religionswissenschaft 21, 99121.
Feghali, Michel. 1928. Syntaxe des parlers arabes actuels du Liban. Paris: P. Geuthner.
Goldenberg, Gideon. 1994. allad al-masdariyyah in Arab grammatical tradition.
Zeitschrift fr Arabische Linguistik 28, 735 = Gideon Goldenberg, Studies in Semitic
Linguistics. Selected Writings, 250285. Jerusalem: Magnes Press, The Hebrew University, 1998.
Hopkins, Simon. 1984. Studies in the Grammar of Early Arabic Based upon Papyri Datable to before 300 A.H./912 A.D. Oxford: Oxford University Press.
Lehmann, Christian.1984. Der Relativsatz. Typologie seiner StrukturenTheorie seiner
FunktionenKompendium seiner Grammatik. Tbingen: G. Narr.
Nldeke, Theodor. 1909. Geschichte des Qorns. Bearbeitet von Friedrich Schwally. Erster
Teil. ber den Ursprung des Qorns. Leipzig. Reprint Hildesheim: G. Olms, 1961.
Payne Smith, J. 1903. A Compendious Syriac Dictionary founded upon the Thesaurus
Syriacus of R. Payne Smith, DD. Ed. by J. Payne Smith (Mrs. Margoliouth). Oxford:
Oxford University Press.
112
werner diem
Payne Smith, R. 1879. Thesaurus Syriacus. Oxford: Clarendon.

Piamenta, Moshe. 1979. Islam in Everyday Arabic Speech. Leiden: E.J. Brill.
. 1983. The Muslim Conception of God and Human Welfare as Reflected in Everyday
Arabic Speech. Leiden: E.J. Brill.
Reckendorf, Hermann. 1921. Arabische Syntax. Heidelberg: C. Winter.
Rubin, Aaron D. 2005. Studies in Semitic Grammaticalization. Winona Lake, Indiana:
Eisenbrauns.
Schabert, Peter. 1976. Laut- und Formenlehre des Maltesischen anhand zweier Mundarten. Erlangen.
Singer, Hans-Rudolf. 1984. Grammatik der arabischen Mundart der Medina von Tunis.
BerlinNew York: W. de Gruyter.
Spitaler, Anton. 1962. al-h amdu li-llhi llad und Verwandtes. Ein Beitrag zur mittel- und neuarabischen Syntax. Oriens 15, 1962, 97114 = Anton Spitaler, Philologica. Beitrge zur Arabistik und Semitistik, ed. by Hartmut Bobzin, 230247, 248
(Zustze). Wiesbaden: Harrassowitz, 1998.
Vanhove, Martine. 1993. La langue maltaise. Etudes syntaxiques dun dialecte arabe priphrique. Wiesbaden: Harrassowitz.
Woidich, Manfred. 1980. illi als Konjunktion im Kairenischen. Studien aus Arabistik
und Semitistik. Anton Spitaler zum siebzigsten Geburtstag von seinen Schlern berreicht, ed. by Werner Diem and Stefan Wild, 224238. Wiesbaden: Harrassowitz.
. 1989. illi dass, illi weil und zayy illi als ob: zur Reinterpretation von Relativsatzgefgen im Kairenischen. Mediterranean Language Review 45, 1989, 109128.
Yoda, Sumikazu. 2005. The Arabic Dialect of the Jews of Tripoli (Libya). Grammar, Text
and Glossary. Wiesbaden: Harrassowitz.
LES ORIGINES DE LA GRAMMAIRE ARABE, SELON LA

TRADITION : DESCRIPTION, INTERPRTATION, DISCUSSION1
Pierre Larcher
Universit de Provence (Aix-en-Provence, France)
1. Introduction
Le texte, bien connu, que nous traduisons en annexe, est extrait du
Kitb al-dh f ilal al-nahw daz-Zajjj. Az-Zajjj le Zajjjien est
le surnom sous lequel est connu Ab l-Qsim Abd al-Rahmn b. Ishq.
Ce grammairien dorigine iranienne du IVe/Xe sicle doit son surnom
au fait quil tudia Bagdad auprs du grammairien Ibrhm b. as-Sar
az-Zajjj (m. 311/9234). Il sinstalla ensuite en Syrie, Alep, puis
Damas, avant de se rendre en Palestine, Tibriade, o il mourut en
337/949.
On connat un peu moins dune vingtaine douvrages daz-Zajjj. Une
dizaine environ a t publie. Parmi ceux-ci, deux se distinguent :
le Kitb al-Jumal, qui est un ouvrage de grammaire devant son nom
au fait, non quil traite de phrases, mais quil est constitu de notes
de synthse (jumal) sur les diffrents chapitres de la grammaire. Il a
t publi en 1926 Alger par Mohammed Ben Cheneb et republi
Paris en 1957.
le Kitb al-dh f ilal al-nahw, qui nest pas un ouvrage de grammaire, mais sur la grammaire. Il a t publi par Mzin Mubrak au
Caire en 1959, puis republi Beyrouth en 1973 et 1979.
Vers la fin des annes 60 et le dbut des annes 70, il y eut, avec lexplosion de la linguistique, celle dune sous-discipline : lhistoire de la
linguistique. Le mouvement atteignit mme les arabisants. De par sa
nature mme, le dh dal-Zajjj attira lattention, en particulier celle
1
Ce texte est la version crite de la leon ERASMUS faite au sminaire du Pr. Dr.
Andreas Kaplony, lOrientalisches Seminar de lUniversit de Zrich, le mardi 19 Avril
2005. Que les collgues et tudiants de lOrientalisches Seminar soient remercis pour
leurs remarques et questions, dont a bnfici la version finale.
114
pierre larcher
de Kees Versteegh, qui le traduisit dans le cadre de son MA (1971), et

lutilisa abondamment dans sa thse Greek Elements in Arabic Linguistic
Thinking (1977).
Nanmoins, cest seulement en 1995 que Versteegh publia cette traduction, sous le titre de The Explanation of Linguistic Causes. Az-Zajjjs
Theory of Grammar.
Louvrage est typique du IVe/Xe sicle, en ce que sy rvle partout linfluence de la falsafa (ou philosophie hellnisante), commencer sur la
forme mme de louvrage : celle, dialectique, hrite de lAntiquit grecque, par question et rponse.
Notre texte constitue le chapitre XIV (8990) de louvrage. Il rpond
la question de savoir pourquoi la grammaire a t nomme nahw en
arabe. Notons que cest le mme mot de illa, pluriel ilal, qui apparat
dans le titre du chapitre et dans le titre de louvrage. En revanche, dans
la formulation de la question, au dbut du chapitre, apparat celui de
sabab. Cela peut amener penser que les deux termes sont synonymes.
Et cest srement ce qui dtermine Versteegh traduire ilal par causes
dans le titre de louvrage et illa et sabab dans le texte du chapitre par le
mme mot de reason. Il existe cependant entre sabab et illa la mme
diffrence quen franais entre cause et justification. La cause est objective et relve de lordre logique : A parce que B. La justification en revanche est intersubjective et relve de lordre dialectique : A car/puisque B.
Noublions jamais que le rapport des autres disciplines la falsafa est
dialectique : par les questions quelle pose, la falsafa les oblige rpondre, en se justifiant.
Ce texte est capital, non seulement pour lhistoire de la grammaire
arabe, mais encore celle de la langue arabe. Telle quelle a t comprise
par la tradition arabe, la premire tire en effet son origine de la corruption ( fasd) de la seconde.
2. Description
Le texte nous dit o, quand et comment ce processus a lieu et il nous dit
aussi en quoi il consiste.
O : Basra, cest--dire dans une des villes nouvelles cres la suite
de la conqute islamique, ce qui rpond en mme temps la question du
quand : Basra a t fonde en 16/637.
Comment : par un double processus de sdentarisation des Bdouins
et de mlange des populations arabes avec des populations non-arabes.
les origines de la grammaire arabe
115
Apparaissent dans le texte quelques termes fondamentaux. Dabord celui

de Arab, dans lexpression abn al-Arab (litt. Fils, enfants des Arabes),
oppos tout la fois al-h dira et abn al-Ajam (litt. Fils, enfants des
non-Arabes). Cela veut dire que Arab ne soppose pas seulement ici
Ajam comme Arabes non-Arabes (et, plus particulirement, dans le
contexte local, Persans), mais encore comme Bdouins sdentaires.
Au tmoignage mme du Lisn al-Arab (dsormais LA) de Ibn Manzr,
m. 711/1311 (art. hdr) : al-h adar, al-h adra et al-h dira sont le contraire
de al-bdiya et il sagit des villes, des villages et de la campagne (alh adar wa-l-h adra wa-l-h dira xilf al-bdiya wa-hiya al-mudun wal-qur wa-r-rf ). Al-h dira dsigne donc bien le pays sdentaire par
opposition al-bdiya ou pays bdouin.
Voil pour la sdentarisation. Venons en maintenant au mlange
des populations. Celui-ci nest pas voqu explicitement au travers du
terme habituel de muxlata, mais implicitement au travers de celui de
muwalladna-na. On traduit ordinairement par mtis (< lat. mixticius mlang ). On a voulu voir dans muwallad ltymon de multre
(Kazimirski 18467, art. wld), via lespagnol mulato, mais ce dernier
terme se rapporte peut-tre plus simplement au latin mulus/mula
( mule(t) , cf. en espagnol mme, outre mulo/mula mule(t) , muleto/
muleta jeune mule(t) ), tout le moins a subi une contamination de
cette famille lexicale. Dans la mme veine larticle muwallad de EI2 indique quil sagit dun terme appartenant au vocabulaire des leveurs et
dsignant le produit dun croisement (tawld) entre deux races animales
diffrentes, donc un hybrid, un sang ml et que cest par analogie
que le terme a t tendu aux humains. Mais larticle ne donne aucune
rfrence pour ce sens, que nous ne trouvons pas, par exemple, dans LA
(art. wld). La question se pose donc au linguiste de larticulation de la
dsignation historique du terme et de sa signification. Morphologiquement, muwallad est le participe passif du verbe wallada. Wallada est le
factitif du verbe de base, mais il renvoie lactif ou au passif de celui-ci,
donc walada-yalidu ou wulida-yladu, selon quil est doublement ou
simplement transitif. Le verbe de base walada-yalidu, simplement transitif, signifie engendrer un enfant (et plus particulirement accoucher dun enfant , sil se dit dune femme, ou mettre bas un petit ,
sil se dit dune femelle). Le verbe doublement transitif wallada-h -hu
signifie faire en sorte quune femme accouche ou quune femelle mette
bas , cest--dire laider accoucher ou mettre bas. Muwallida est
un des noms de la sage-femme (qbila). Le verbe simplement transitif
wallada signifie faire natre quelquun ou gnrer quelque chose .
116
pierre larcher
Dans les deux cas, cependant, muwallad signifie, comme seul objet ou
second des deux objets de wallada, engendr . Cest srement par
lide de mis au monde que le terme muwallad a pris, tout la fois
par mtaphore et gnralisation, le sens de tout ce qui est nouveau,
moderne (al-muh dat min kulli ay LA, art. wld). Il peut se dire, soit
de quelquun, soit de quelque chose : il se dit en particulier des potes
modernes (al-muwalladn min a-uar) et des nologismes (summiya al-muwallad min al-kalm muwalladan id istah dathu wa-lam
yakun min kalmihim fm mad ce quil y a de muwallad dans le parler a t ainsi appel, quand on le produit, sans quil ait exist dans le
parler auparavant ). Comme tout ce qui est nouveau, le terme peut sentendre en mauvaise part comme quelque chose de fabriqu, controuv,
apocryphe. Cest sans doute par une extension de ce dernier sens que
trs tt le terme a pris le sens de non purement arabe , pouvant se
dire, l encore, soit de quelquun soit de quelque chose, cf. LA, art. WLD
arabiyya muwallada wa-rajul muwallad id kna arabiyyan ayr mah d
de larabe ou un homme muwallad(a), sil nest pas purement arabe .
Historiquement, le terme sest dit des enfants ns, la suite des conqutes
islamiques, dunions mixtes, gnralement entre des pres arabes et des
mres non arabes. Si donc lon suit le mouvement smantique suggr
par LA, le sens de mtis nest pas mettre au dpart, mais au contraire
larrive dun processus dvolution smantique . . . Notons que ce dernier sens pourrait aussi satteindre par un simple et banal processus de
tadmn, consistant faire entrer dans un mot le sens de toute une
collocation, muwallad tant mis pour muwallad min muxlatat al-Arab
al-Ajam ( issu/produit du mlange des Arabes et des non-Arabes ).2
Maintenant, en quoi consiste prcisment ce processus de corruption de la langue ? Celui-ci est dcrit au travers dune anecdote mettant
en scne Ab l-Aswad ad-Dual et sa fille.
La tradition a vu dans ce personnage du Ier/VIIe sicle le pre de
la grammaire arabe. Pourquoi lui plutt quun autre ? Dabord, parce
quil est dorigine arabe : sa gnalogie complte, telle que donne par les
Tabaqt (21) de Zubayd, m. 379/98990, le rattache aux Kinna, tribu
2
Cest un processus fondamental, tant dans le lexique de larabe classique (e.g. jihd
guerre sainte mis pour jihd f sabl li-llh combat pour Allah , siysa politique , mis pour siysa madaniyya ( gouvernement de la cit ) que dans celui de larabe
moderne (tlib tudiant mis pour tlib al-ilm celui qui cherche le savoir , amn
secrtaire mis pour amn as-sirr dpositaire du secret ).
117
de la rgion de la Mecque. Il serait n vers 606 de notre re et aurait

sjourn chez les Quayr dArabie centrale (nous verrons ultrieurement
limportance de ces notations). Ensuite, ralli Al, il est nomm par ce
dernier cadi puis gouverneur de Basra, en 36/656, mais la disparition de
Al en 41/661 ne lempche pas, comme le montre notre texte (et dautres
sources) dentretenir des relations avec le gouverneur Umayyade Ziyd
b. Abhi. Il serait mort vers 69/688. Enfin, il tait galement pote. Le
personnage est donc le prototype de larabe sdentaris, que son origine
et son parcours dsignent comme un matre et un gardien de la langue.
Il est de Basra, dont la tradition ultrieure fera le sige de lcole dominante de la grammaire arabe, et voir dans un Basrien le pre de la grammaire arabe nest rien dautre quune manire de signaler lanciennet du
travail grammatical dans cette ville et surtout son antriorit par rapport
lcole rivale de Kfa. Il est consensuel socialement et politiquement :
nomades et sdentaires, partisans de Al comme des Umayyades peuvent sen rclamer.
Venons en maintenant lanecdote elle-mme. La fille dit son pre
m aaddu l-h arri. Celui-ci interprte le propos comme une question
sur la chaleur la plus intense ( Quelle est la chaleur la plus intense ? )
et rpond, par suite, ar-ramdau f l-hjira, cest--dire la canicule en
plein midi . Mais la fille rejette cette interprtation lam asalka an hd
je ne tai pas demand cela , ajoutant je me suis tonn de lintensit
de la chaleur (innam taajjabtu min iddati l-h arri). Alors dis, rtorque son pre, m aadda l-h arra ( Quelle chaleur intense !) .
Autrement dit le pre reproche sa fille davoir confondu les structures interrogative (istifhm) et exclamative (taajjub) et davoir employ
lune pour lautre. Cest cela un lah n, autre terme fondamental apparaissant dans le texte : non pas une faute de langage en gnral, mais une
faute contre la flexion dsinentielle, casuelle et modale, en particulier.
Cest cette flexion quon appelle en arabe mme irb. Une telle faute
serait dautant plus grave quelle crerait un quiproquo.
De cette anecdote, il existe plusieurs versions (comme le reconnat
le texte lui-mme) : Versteegh (1997b, 59) cite la version donne par
Srf (m. 368/979), dans son Axbr (d. Krenkow, 19) avec les exemples de m ah sanu as-sami What is the most beautiful thing in the
sky? / m ah sana s-sama How beautiful is the sky! .
A travers cette anecdote, il sagit en fait de mettre au centre de la
arabiyya ou langue des Arabes le irb (mot tymologiquement li
Arab) et de suggrer que cet irb est pertinent, en ce quil distingue des
118
pierre larcher
significations (muwaddih , mubayyin, mufarriq, munabbi . . . li-man

al-lua), pour citer ici quelques-uns des mots que lon relve dans les
sources.
Et cest laggravation de la situation qui amne la constitution de la
grammaire, son nom arabe de nahw tant justifi par le fait que le grammairien se fait lindicateur de la voie suivre (unh hd n-nahw).
Nahw est en effet le masdar du verbe nah -yanh , quon emploie toujours comme circonstanciatif, fig laccusatif (nahwa), de sens vers .
Peu importe si cette tymologie est fantaisiste ou non. Ce qui nous
intresse ici, ce sont les diffrentes interprtations quun linguiste est
susceptible de faire de ce texte pour lhistoire de la langue.
3. Interprtation
3.1
De lhistoire . . .
Au XIXe sicle, la linguistique, ne au tournant du XVIIIe et du XIXe,

sous forme de la grammaire compare (des langues indo-europennes),
devient historique. On ne cherche plus seulement reconstruire en
amont des protolangues (Ursprache). Plus modestement, on cherche
retracer en aval lvolution des langues existantes.
La linguistique historique est une spcialit essentiellement allemande.
On ne stonnera donc pas que ce soient les arabisants allemands qui, les
premiers, se sont intresss lhistoire de larabe. Ils rinterprteront le
fasd al-lua de la tradition arabe, caractrise par les lah n ou fautes de
irb, comme le signe dune volution dun type ancien arabe (en allemand Altarabisch et en anglais Old Arabic) vers un type no-arabe (en
allemand Neuarabisch et en anglais New ou Neo-Arabic). Le type ancien
arabe est videmment caractris par lexistence dune flexion dsinentielle, casuelle et modale, le type no-arabe par la disparition de cette
flexion. Le type ancien arabe est donc plus synthtique et le type noarabe plus analytique. Corollairement, dans le type ancien arabe, lordre
des mots est plus libre, mais, dans le type no-arabe, moins libre. Quand
le type ancien arabe commence se dgrader en type no-arabe, nous
entrons dans le moyen arabe (en allemand Mittelarabisch et en anglais
Middle Arabic). Cest Heinrich Leberecht Fleischer (18011888) qui
projettera, implicitement dans un article de 1847, explicitement dans
un article de 1854, sur larabe cette tripartition clbre en linguistique
historique. Voici ce quil crit en 1854 (23) :
119
Als heilige Sprache des Islam, Organ der Gelehrsamkeit und hhern Wissenschaftlichkeit, Mittelpunkt oder vielmehr ausschliesslicher Gegenstand aller Schulphilologie, steht das Altarabische seinem Abkmmling,
dem Neuarabischen, in der Anschauung des Morgenlandes selbst schroff
gegenber. Nur jenes heisst bei den Gelehrten al-luah, die Sprache, alarabiyyah, das Arabische schlechthin, dieses al-lisn al-mm oder almm, die gemeine Mundart, la lingua volgare.
Il est clair, daprs la description mme quen donne Fleischer, que lancien arabe est larabe classique et le no-arabe larabe dialectal et non
moins clair que si les deux varits coexistent en synchronie, larabe dialectal est explicitement compris comme tant historiquement le descendant (Abkmmling) de larabe classique.
Un peu plus loin (4), il mentionne le moyen arabe. Il lavait dj exactement dcrit (du point de vue de la linguistique historique) en 1847,
propos de la langue dun codex grco-arabe (155), quil compare celle
des Mille et une nuits, auxquelles il avait consacr sa dissertatio en 1836 :
Wie in der Tausend und Einen Nachten sind auch hier einzelne jener
ltern Formen mit der neuern gleichsam noch im Kampfe begriffen ;
willkrlich tritt bald die eine, bald die andere ein . Ltat moyen dune
langue se caractrise en effet par lalternance, en synchronie, dlments
interprtables, en diachronie, comme relevant encore de ltat ancien
(lter) ou dj de ltat moderne (neuer).
Un de ses lves, Ignaz Goldziher (18501921), dans un crit de jeunesse rdig en hongrois et aujourdhui traduit en anglais, compare explicitement la relation entre ancien arabe et no-arabe celle du latin et des
langues romanes, appeles jadis no-latines (Goldziher 1994, 20) :
As French abandoned the case inflection of Latin and developed the
Roman synthesis into analysis, making de lhomme from hominis, so did
the living Arabic of today dissolve the old rajulin into met r-rajul ; as latin
scrip-si developed into French jai crit (. . .), so was Old Arab[ic] aktubu
turned into biddi aktub ou bi-aktub.
Cette conception allemande de lhistoire de larabe se retrouve au

XXe sicle, en particulier chez Johann Fck (18941974) dans son grand
ouvrage Arabya (Fck, 1955[1950]), et, aujourdhui encore, chez Joshua Blau, le grand matre du moyen arabe (e.g. Blau 2002, 16).
Objectivement, cette conception rejoint celle que sen faisait au VIIIe/
XIVe sicle Ibn Xaldn (m. 808/1406) et quil expose deux reprises
dans la Muqaddima : une premire fois dans la section 22 (6757), intitule f lut ahl al-amsr ( Des parlers des habitants des villes ) du
chapitre IV, consacr aux villes et pays , et une seconde aux chapitres
120
pierre larcher
47 et 48 (10731080), consacrs respectivement aux parlers nomades et aux parlers sdentaires de son temps, du chapitre VI, lui-mme
consacr aux sciences . Pour Ibn Xaldn, ce quil appelle langue de
Mudar (luat Mudar, al-lisn al-mudar) est la langue premire (allisn al-awwal) et originelle (al-lisn al-asl) de lArabie et la langue
du Coran et du h adt. Cette langue est parle de manire dautant plus
chtie ( fash a) que ceux qui la parlent sont plus loigns des autres
nations, la manire de parler des Quray tant pour cette raison la plus
chtie et la plus pure des manires de parler arabes (wa-li-hd knat
luat Quray afsah al-lut 3 al-arabiyya wa-asrah uh): si lon retraduit
en termes gographiques la gnalogie, lappellation langue de Mudar
revient dsigner le centre et louest de lArabie comme le domaine de
larabe fash . Ibn Xaldn fait ici la synthse de deux thses, sur lesquelles nous reviendrons ci-dessous : dune part la thse philologique (qui
voit ce domaine comme constitu de deux sous-domaines dits Hedjaz
et Tamm),4 et dautre part la thse thologique (qui identifie, sur la base
de Cor. 14:4, la langue du Coran la luat Quray et, sur une base dogmatique, la luat Quray la lua al-fush ), tout en les croisant avec une
thse philosophique, issue du Kitb al-h urf de Frb (m. 339/950),5
et qui est le corrollaire de la thse liant corruption et mlange .
A linverse, cette langue est dj corrompue ds lpoque prislamique,
l o les Arabes sont en contact avec dautres nations, et se corrompt
encore davantage aprs la conqute islamique et les nouvelles fondations
urbaines, et donc en milieu sdentaire plus encore que nomade, jusqu donner naissance de nouvelles langues. Loriginalit dIbn Xaldn
est en effet de ne pas considrer les dialectes comme de simples formes
dgrades de la langue de Mudar , mais comme des varits autonomes par rapport celle-ci et distinctes delle, en ce quelles ont substitu
la syntaxe base sur la flexion dsinentielle une syntaxe de position
(at-taqdm wa-t-taxr).6
Cest cette expression qui donne, par rcriture, celle de al-lua al-fush .
Sur cette subdivision, cf. Rabin (1951). On comprend pourquoi Ab l-Aswad
d-Dual, natif du Hedjaz, est dit avoir fait un dtour par lArabie centrale . . .
5
Du moins la version de ce texte connue par le Muzhir (I:211212) ou, mieux, le
Iqtirh (20) de Suyt (m. 911/1505), non celle publie par Mahdi en 1969. Sur les deux
versions de ce texte, cf. Langhade (1994, 248258) et Larcher (2006a).
6
Sur Ibn Xaldn et lhistoire de larabe, cf. Versteegh (1997a, 153165) et Larcher
(2006b).
4

3.2
121
la sociolinguistique
En citant lexpression italienne de lingua volgare, certes attire par celle

de al-lisn al-mm(),7 Fleischer montre quil na pas seulement en tte la
linguistique historique, mais aussi le modle italien. Or, dans ce modle,
la langue vulgaire nest videmment pas dsigne en termes diachroniques, cest--dire historiques, comme descendant du latin ; elle lest
au contraire, en termes synchroniques et sociolinguistiques, comme
langue du vulgum pecus, par opposition au latin langue des clercs .8
Mais, l encore, cette conception, qui, pour larabe, sera baptise ultrieurement par les arabisants diglossie,9 rejoint objectivement celle que
se font les auteurs de langue arabe, commencer dailleurs par az-Zajjj
lui-mme. Au chapitre XVII, intitul bb dikr al-fida f taallum alnahw, il demande (95) :
quoi sert dapprendre la grammaire, la plupart des gens parlant naturellement10 sans flexion dsinentielle, quils ne connaissent pas, tout en comprenant les autres et en [se] faisant comprendre deux ( fa-m al-fida f
taallum an-nahw wa-aktar an-ns yatakallamn al sajiyyatihim bi-ayr
irb wa-l marifa minhum bihi fa-yafhamn wa-yufhimn ayrahum
mitl dlika).
7
Fleischer ne donne aucune rfrence pour ces deux expressions. Cest dommage,
car si, lpoque o il crit (milieu du XIXe sicle), lexpression al-arabiyya est, comme
il le note, couramment utilise, par une mtonymie significative, pour dsigner larabe
classique, cest lexpression de al-lua ad-drija qui est utilise pour dsigner larabe dialectal. Lexpression dal-lua al-mmiyya (vs al-lua al-fush) napparatra que vers la
fin du XIXe sicle, du moins comme nom de cette varit, mais ds le Moyen Age, on la
rencontre pour dsigner un vulgarisme au sein de la langue.
8
La comparaison avec la situation italienne ne peut dailleurs tre pousse trop
loin sans aporie. Le domaine arabe na pas connu la rvolution qua connue le domaine
roman (et, mutatis mutandis, lEurope entire), savoir la promotion des langues vulgaires au rang de langues littraires, ce qui fera du latin (et seulement pour un temps,
plus ou moins long selon les pays) le vhicule de la seule culture savante. Ainsi, aprs
Il cantico delle creature (1226) de Saint Franois dAssise (11821226), Dante Alighieri
(12651321), logiquement, crit La Divine Comdie en langue vulgaire, mais traite de
celle-ci en latin (De vulgari eloquentia).
9
Ce terme, venu de la linguistique no-hellnique (1885), a t explicitement introduit en linguistique arabe par William Marais (18741956), dans un article de 1935,
avant que le concept ne soit thoris, partir de larabe et dautres langues, par Charles
A. Ferguson (19211998), dans un article de 1959. Pour le dtail, cf. Larcher (2003).
10
Blanc (1979, 165, n. 20) traduit par spontaneously et Versteegh (1995) par
intuitively .
122
pierre larcher
Il revient sur ce point la fin du chapitre (96) :

Quant aux gens du commun qui parlent larabe sans flexion dsinentielle,
on les comprend. Mais cela est seulement possible pour ce qui est bien
connu et dusage courant, ce dont on a une connaissance familire et est
usit. Mais si, daventure, lun deux se risquait claircir une ambigut,
sans le faire comprendre au moyen de la flexion dsinentielle, il ne le pourrait pas ( fa-amm man takallama min al-mma bi-l-arabiyya bi-ayr
irb fa-yufham anhu fa-innam dlika f al-mutaraf al-mahr wa-lmustamal al-malf bi-d-dirya wa-law iltajaa ah aduhum il al-dh an
man multabis min ayr fahmihi bi-l-irb lam yumkinhu dlika).
Alors que le chapitre XIV concerne le Ier/VIIIe sicle, le chapitre XVII

concerne lpoque dal-Zajjj lui-mme, cest--dire le IVe/Xe sicle. La
situation dcrite dans ce chapitre semble pouvoir tre interprte comme
laboutissement du processus dcrit au chapitre XIV. Au Ier/VIIIe sicle
le type ancien arabe commence se dgrader en type no-arabe. Trois
sicles plus tard, cette dgradation a abouti, non une substitution dun
type lautre, mais, le type ancien arabe subsistant, une coexistence
des deux, chaque type tant caractris non seulement linguistiquement
par la prsence/absence du irb, mais encore socialement et culturellement : dire que le type non flchi est lexpression naturelle de la
plupart des gens revient dire que le type flchi est celle non seulement
dune minorit, mais encore lexpression artificielle de cette minorit
(artificielle, puisque scolairement apprise et non naturellement acquise
et pratique !). Qualifier explicitement la majorit de mma revient non
seulement qualifier implicitement la minorit de xssa , mais encore
semble prfigurer lappellation de al-lua al-mmiyya, aujourdhui utilise pour dsigner larabe dialectal. Enfin reconnatre par deux fois que
labsence de flexion dsinentielle ne nuit en rien la communication
explique que, pour sauver cette flexion (son gagne pain !), le grammairien distingue entre communication quotidienne et communication
savante et cherche, voire fabrique, des structures ambigus, que seule
la flexion, censment pertinente, est mme de dsambiguser (on voit
tout de suite que lexemple propos au chapitre XIV relve exactement
de ce cas de figure et, comme tout arabisant en a fait lexprience, ce petit
jeu perdure, jusqu aujourdhui, dans certains milieux puristes !).
La vrit oblige cependant dire quil subsiste un doute sur cette
interprtation. On doit se rappeler ce que dit al-Jhiz (m. 255/869) dans
le Bayn (I:137). Il noppose nullement mma xssa comme la masse
illettre llite lettre, mais seulement, parmi les gens lettrs, les gens
ordinaires aux happy few. Cest exactement le sens que les spcialistes
123
donnent mma dans la littrature spcialise des ouvrages de lah n almma.11 Il ny a rien l que de logique : lah n prsuppose irb ; par suite,
sil ny a pas irb, il ny a pas lah n ! Si lon projette le sens jhizien de
mma sur le texte daz-Zajjj, il ne sagirait plus alors de deux varits darabe, mais seulement de deux registres dune mme varit : lun
soutenu (avec ralisation du irb) et lautre relch (sans ralisation de ce irb). Jusqu aujourdhui lcole enseigne le irb qui terrorise (irhb) les apprenants, parce que sa ralisation est source de lah n,
do le sage conseil ijzim taslam Supprime la voyelle brve finale, tu
seras prserv de lerreur ! . Un seul lment du texte daz-Zajjj peut
faire pencher la balance en faveur de la premire interprtation, plutt que de la seconde : cest la notation que les locuteurs de larabe sans
flexion dsinentielle nont pas connaissance de celle-ci . La mma, au
sens dal-Jhiz, ne lignore pas ; elle en a seulement une connaissance
imparfaite12 . . .
4. Discussion
On le voit : tout tourne autour du irb, tant dans le texte daz-Zajjj que
dans linterprtation, tout la fois historique et sociolinguistique, qui en
est faite. Pourtant, ds avant le XIXe sicle et lessor de la linguistique historique, une tout autre tendance tait apparue chez certains arabisants.
Ds le XVIIIe sicle, des arabisants, galement allemands dailleurs,
comme Johann Davis Michaelis (17171791) et Johann Gottfried Hasse
(17591806), avaient dvelopp ce que Gruntfest (1991), qui la tudie, appelle A Early theory of Redundancy of Arabic Case Endings .
Notons quils rinterprtaient des ides dj exprimes, au XVIIe
sicle, par lItalien Antonius ab Aquila (Antonio dellAquila), Franciscain
envoy en mission auprs des Chrtiens dAlep et auteur de la premire
11
On lira avec profit larticle Lah n al-mma, de EI2, d Charles Pellat (1914
1992).
12
On notera que dans lunique manuscrit, dat de 617H, qui sert de base ldition du
dh , il y a dans lavant-dernier paragraphe de ce chapitre XVII, une magnifique faute de
irb : lan yumkin ah ad [corrig par lditeur en ah adan] min al-muwalladn iqmat-hu
ill bi-marifat an-nahw ( Nul, parmi les muwalladn, ne pourrait ltablir (la posie),
sauf par la connaissance de la grammaire ). Labsence du alif rvle au minimum quil
nest pas prononc, voire suggre que ah ad a t trait comme le sujet de yumkin et par
suite iqma comme lobjet, ce que peut galement suggrer le masculin yumkin.
124
pierre larcher
grammaire de la langue (arabe) vulgaire (1650).13 Notons galement

que tous les arabisants du temps savaient le latin et le grec : quand on a
souffert dans sa jeunesse sur une version grecque ou latine, on voit tout
de suite la diffrence entre grec et latin classiques et arabe classique. En
grec et en latin, cest la dclinaison qui permet de construire la phrase ;
en arabe, cest la construction de la phrase qui permet de restituer la
dclinaison . . . Les grammairiens arabes le reconnaissent eux-mmes
implicitement, en couplant, dans leur thorie, le irb au amal : le irb
est leffet de l action dun lment dit mil sur un autre dit maml,
ainsi lan, dit nsib, qui dtermine le nasb (subjonctif ) du mudri (inaccompli) dit mansb. Par ailleurs, le fait mme que les grammairiens arabes cherchent des cas o cette flexion serait pertinente suffit prouver
quelle ne lest pas ! On voit tout de suite le caractre artificiel de lexemple
propos par az-Zajjj. Dans beaucoup de langues, structures interrogatives et exclamatives sont confondues, tant distingues par lintonation,
ainsi en franais ou en allemand Quel artiste/Welcher Knstler ? vs Quel
artiste/Welch ein Knstler !
Les exemples donns de l aggravation de la situation , qui font
dfaut chez az-Zajjj lui-mme, mais quon trouve chez dautres auteurs,
ne sont gure plus convaincants. Ainsi le fameux tuwuffiya abn wataraka banna ( notre pre est mort et a laiss des fils ) est si peu probant pour un fasd al-lua interprt en termes historiques que Versteegh
(1997b, 51), qui le cite daprs le Nuzha dIbn al-Anbr (m. 577/1181),
linterprte en termes sociolinguistiques (alors mme que lanecdote
concerne le temps de Ziyd, donc le Ier/VIIIe sicle). En termes historiques, on se serait en effet attendu lemploi des formes no-arabes ab
et bann (respectivement nominatif de la flexion triptote ab// et cas
rgime de la flexion diptote ban/(na) du type ancien arabe), de sorte
quavec la phrase donne en exemple, il ny aurait pas eu de faute !
En termes sociolinguistiques, en revanche, la coexistence de deux types
dont lun a plus de prestige que lautre peut amener le locuteur, dans une
circonstance formelle, substituer lunique forme du type no-arabe,
quil emploie ordinairement, mais pense fautive, lautre forme du type
13
Cf. Fck 1955, 78. Fck considre que la Fabrica overo Dittioniario della lingua volgare arabica et italiana (1636) de Dominicus Germanus de Silesia (15881670) nest pas,
malgr son titre, un dictionnaire, mais une introduction, presque sans valeur, larabe
vulgaire. En tout cas, on voit que ce sont des clercs, Italiens ou lis lItalie, qui, sous
lappellation, valant signature, de langue vulgaire , sont les inventeurs de larabe
dialectal. Cette double qualit, jointe au fait quils taient des hommes de terrain et non
de cabinet, les y prdisposait.
125
ancien arabe, quil nemploie pas et pense correcte : pour viter une faute,
il en commet une autre !
Certes, ces arabisants voyaient dans les dsinences casuelles une
invention des grammairiens arabes, position encore dfendue au XIXe
sicle par Alos Sprenger (18131893). Si personne aujourdhui ne croit
plus cela, en revanche la redondance de la flexion en arabe classique
amne beaucoup darabisants douter que lhistoire de larabe consiste
en une volution du type ancien arabe vers le type no-arabe, caractriss respectivement par la prsence et labsence de cette flexion.
Deux positions se sont fait jour. Lune est trs rpandue chez les arabisants. Lakkadien possdant une flexion casuelle triptote et dautres
langues smitiques exhibant des restes de flexion casuelle, les arabisants
admettent que la flexion casuelle de larabe classique est un trait de haute
antiquit, qui sest maintenu dans le seul registre potique de la langue.
Sous linfluence de la reprsentation diglossique de larabe, rtoprojete
sur lhistoire de la langue, ce registre potique est souvent vu comme
une langue commune (koin), vhiculaire, oppos aux vernaculaires que
sont les anciens dialectes arabes. Cette koin serait galement la langue
du Coran, quelques hedjazismes prs. Malgr les apparences, cette
hypothse arabisante nest pas si diffrente de la thse thologique ! Celle-ci, on la vu, se rsume en une double identification : lune, opre sur
une base scripturaire, de la langue du Coran avec la langue de Quray
et lautre, opre sur une base purement dogmatique, de la langue de
Quray avec al-lua al-fush . Il y a un sicle, Vollers (1906) acceptait la
premire identification, mais refusait la seconde. Il supposait en effet que
le Coran avait dabord t nonc et crit dans le vernaculaire de la Mecque, parler ouest-arabique dpourvu de irb, avant dtre rcrit dans la
langue vhiculaire de la posie, parler est-arabique pourvu de ce irb.
Kahle (1959[1947]) reprit cette hypothse en lattnuant : si le ductus
coranique (rasm) reflte le vernaculaire de la Mecque, les qirt, elles,
refltent la langue vhiculaire de la posie. La plupart des arabisants (cf.
Blachre 1952, 6682) refusent la premire identification et, par suite,
la seconde, mais acceptent la troisime, dcoulant de la suppression
du moyen terme, i.e. langue du Coran = al-lua al-fush . Or, si lon relit
le clbre texte du Sh ib (5253) dIbn Fris (m. 395/1004), dont la
source nest autre que celui attribu al-Farr (m. 207/822) et jadis
exhum par Kahle (1959[1947]), on saperoit que pour concilier vrit
thologique et vrit philologique (i.e. le fait que la langue du Coran,
identifie la luat Quray, exhibe des traits, les fameux hedjazismes ,
qui ne sont pas ceux de la lua al-fush ) ces auteurs imaginaient un
126
pierre larcher
scnario expliquant comment la luat Quray tait devenue la lua

al-fush : comme produit dun processus de koinsation justifi par le
fait que, la Mecque tant un centre de plerinage intertribal, les Quray
avaient ainsi pu slectionner (taxayyur) le meilleur de chaque parler
arabe !14
Mais une autre position, originale et marginale, est en train de gagner
du terrain. La posie archaque pratique la qfiya mutlaqa ( rime absolue ), cest--dire ralise les voyelles brves u, a et i, avec ou sans tanwn,
uniformment comme des voyelles longues , , . Cest dire si le tanwn
ne sert rien !15 La rcitation psalmodie du Coran (tajwd) va plus loin :
elle pratique gnralement la qfiya muqayyada ( rime lie ), cest-dire supprime les voyelles brves u, a et i en finale. Elle supprime galement -un et -in, et ralise -an comme -. Cette prononciation pausale de
-an en - est la seule trace de flexion casuelle qui apparaisse dans le maigre matriel pigraphique conserv16 et cest aussi sa seule manifestation
dans les dialectes modernes (marh aban ou ahlan wa-sahlan prononcs
marh ab et ahl w-sahl). Le cas du suffixe relateur -(V)n quon trouve
dans maint dialecte ancien et moderne est disjoindre, dans la mesure
o sa fonction (marquer la relation mawsf/sifa, celle-ci pouvant tre un
N, un SP ou une phrase) ne le relie en rien au tanwn de larabe classique et amne penser que cest la conservation dun trait archaque (cf.
Ferrando 2000).
Quon doive supprimer les voyelles brves en finale (y compris donc
les voyelles flexionnelles) est videmment un argument en faveur de la
non pertinence de cette flexion sur le plan syntaxique.17 En revanche,
quen dehors de la pause, elle soit ralise, suggre que cette flexion nest
pas syntaxique, mais prosodique. Lide tait dj mise au XIXe sicle
par Johann Gottfried Wetzstein (18151905), dans un article paru en
1868. Les historiens de la grammaire arabe savent aujourdhui, grce
Versteegh (1981[1983]), quil sest trouv au moins un grammairien
arabe, Qutr ub (m. 206/821), pour anticiper cette ide. Nous connaissons sa thorie grce az-Zajjj (ch. VII, 6971, bb al-qawl f al-irb
14
Sur ces deux textes, cf. Larcher 2004 et 2005b.

Comme le suggre en outre la ralisation du long rsultant de la qfiya mutlaqa
comme . . . -an (tanwn at-tarannum) attribue aux Tamm, dans la rcitation potique
(ind). Dans les deux cas, on peut donc avoir larticle et voir et/ou entendre /an.
16
Un exemple dans linscription dUmm al-Jiml (Ve ou VIe sicle ap. JC ?), avec un
mot lu successivement par Littmann en 1929 et 1949 comme iyran et afran. Nous
parlons bien sr ici de la seule flexion marque par des voyelles brves.
17
Cf. lanalyse qui est propose de Cor. 85 :2122 dans Larcher (2005a).
15
127
lim daxala f-kalm pourquoi la flexion sest-elle introduite dans le

discours ? ), qui explique que pour Qutr ub les voyelles de flexion sont
lies lenchanement (wasl) des mots dans la phrase. Des ides, sinon
identiques, du moins comparables sont dfendues aujourdhui par
Owens (1998).
Un bon argument en faveur de la conception prosodique et non syntaxique de la flexion peut tre trouv dans le traitement contradictoire de
Tamd dans le Coran. Contradictoire, dans la mesure o il est trait partout comme un diptote sans tanwn Tamd-u/a, sauf en 11 : 68 ; 25 :38 ;
29 :38 ; 53 :51, o apparat un alif. Alors que H afs an Asim (Coran du
Caire) neutralise ce alif, War an Nfi (Coran du Maghreb) le traite
bien, en ces endroits, comme un triptote avec tanwn et lit Tamdan.
Pourquoi, donc, deux dclinaisons pour un mme mot, alors quaucune
des deux ne sert rien sur le plan syntaxique (en 17 :59, nous avons un
Tamd complment dobjet sans alif) ? On a tt fait dobserver que dans
trois des quatre cas o apparat ce alif, Tamd est coordonn d, galement muni de ce alif (et partout trait lui comme un triptote d-u/
a/i-n). Lapparition du alif peut sexpliquer par un simple et banal ajustement forme/sens : cest la coordination syntaxique de deux mots qui,
smantiquement, vont ensemble, comme noms de peuples reprouvs ,
qui explique quon donne au second le traitement syntaxique du premier. Lexplication ne vaut cependant pas pour le quatrime cas (11 :68),
o Tamd apparat seul : (ka-al-lam yanaw fh) a-l inna Tamda
(H afs)/an (War) kafar rabbahum a-l budal-li-Tamd[a]. Sauf lire
ce alif comme il est crit, cest--dire non comme -an, mais bien comme
-, auquel cas il forme aussitt assonance avec les nombreux - de son
environnement syntaxique immdiat, que nous avons mis en gras . . . On
peut aller plus loin. Le dbut du verset, que nous mettons entre parenthses, va avec ce qui prcde. Le reste sorganise en un paralllisme
marqu par lanaphore de a-l. On voit que si on lit, non budan, do,
par assimilation (idm) du nn au lm budal-li-, mais bud, on na
plus seulement alors une assonance en , mais encore une rime interne
en -d, soit : a-l inna Tamd kafar rabbahum a-l bud li-Tamd
Hol ! Oui, les Thamoud, ils ont t infidles leur seigneur ! Hol !
Arrire aux Thamoud ! .18
18
Tamd est loin dtre un cas unique. On peut galement citer salsil de 76 :4 et
qawrr de 76 :1516. Par ailleurs on a compris que, pour ma part, je ne considre
pas - comme la prononciation pausale de -an, mais -an comme une rinterprtation
ultrieure dun - fondamental . Linscription du Jabal Usays (5289 ap. JC) montre
128
pierre larcher
On ne perdra pas de vue que lhistoire de larabe est toujours dpendante des ides du moment. Ainsi Versteegh (1984), surfant sur la vague
des tudes croles, a propos de rinterprter le fasd al-lua dans les
termes dun processus de pidginisation-crolisation. La conqute islamique a pour effet de mettre en contact des populations arabophones
et des populations non arabophones. On peut donc imaginer, pour les
besoins de la communication, lmergence de langues de contact htrognes, ce quon appelait jadis des sabirs et quon appelle aujourdhui
des pidgins (Fck 1955 [1950], 8, en fait dj lhypothse). Si les enfants
qui naissent dunions mixtes (les muwalladn) la prennent pour langue
maternelle, cette langue de contact devient un crole. Et, enfin, si au
contact de larabe, le crole se rarabise (ou se dcrolise), ce pidgincrole dcrolis peut tre vu comme le point de dpart des dialectes
arabes, sur le modle du Juba-Arabic au Sud Soudan.19 Lhypothse de
Versteegh suscita un vaste dbat et fut gnralement rejete (cf. Holes
1995, 1924). Il me semble nanmoins quon ne devrait pas jeter le bb
avec leau du bain !20 On ne peut exclure a priori que des processus de ce
genre aient jou, localement, un rle. Pour ma part, je me contenterai
de signaler et souligner ici une phrase de la Muqaddima dIbn Xaldn
qui ma fait parler, pour tenter une comparaison loquente, de case de
lOncle Tom sur les bords du Tigre et de lEuphrate .21 Traitant de la
gense des parlers sdentaires dans les diffrentes rgions du monde
musulman (Maghreb, Machrek, Andalousie), il crit propos du second
(10771078) :
De mme, quand les Arabes eurent vaincu les nations de lOrient, Persans et Turcs, et se furent mlangs eux, les langues de ceux-ci se diffusrent parmi eux, par lintermdiaire des laboureurs, des paysans et
des captifs quils prenaient comme intendants, sages-femmes, pres nourriciers et nourrices : leur langue se corrompit du fait de la corruption de
quavant linvention du t marbta les deux ralisations dun mme morphme se traduisent par deux graphies : -h la pause, mais -t en liaison (il reste des traces de cet tat
de choses dans le ductus coranique, avec des exemples de rah mat ou nimat). On voit
dautant moins pourquoi les deux ralisations phoniques du tanwnan de larabe classique auraient donn lieu dans le matriel prclassique une graphie unique (correspondant -) que, dans les textes du moyen arabe, le suffixe relateur -(V)n se traduit dans
la graphie par un n.
19
Ma collgue Catherine Miller, consulte, mindique qu lheure actuelle le JubaArabic est encore loin dtre un dialecte arabe. En outre les choses vont un peu dans tous
les sens, manifestant des tendances contradictoires.
20
Cf. dailleurs, Versteegh lui-mme (2004).
21
Cf. Larcher (2006b).
129
lhabitus, au point de muter en une autre langue (wa-kad al-Mariq

lamm al-Arab alaba al umamihi min Fris wa-t-Turk fa-xlatahum
wa-tadwalat baynahum lutuhum f l-akara wa-l-fallh n wa-s-subiyy
alladna ittaxadhum xawalan wa-dytin wa-adh ran wa-mardi fa-fasudat luatuhum bi-fasd al-malaka h att inqalabat lua uxr).
Il serait videmment intressant de remonter ici aux sources mmes

dIbn Xaldn. Bien sr, la notation va dans le sens de largumentation
dIbn Xaldn (interruption du processus naturel dapprentissage de la
langue de Mudar, par transmission dune gnration lautre dArabes,
qui seule cre l habitus ). Mais mme le plus intgriste des crolistes ne peut nier que la phrase souligne dcrit ce que ce genre de crolistes considre tre la base objective, sur le plan social, dun processus
de crolisation : une population servile, parlant une ou dautres langues
que celle de ses matres. En outre deux, voire trois, des quatre catgories mentionnes renvoient ltymologie mme de crole (< lat. criare
nourrir, lever ), terme qui peut dsigner aussi bien les enfants blancs
que les esclaves noirs et leurs langues (gnralement le parler des Noirs,
mais au moins en un cas, celui de Saint-Barthlemy dans les Antilles,
celui des Blancs).
La linguistique arabe est aujourdhui sous linfluence de la sociolinguistique amricaine variationniste , issue des travaux de William
Labov (n en 1927). Cela a amen au dpassement de la reprsentation
diglossique de larabe. Mais cela a aussi amen, en matire dhistoire de
la langue, renouer avec la vision des plus anciens grammairiens arabes,
notamment Sbawayhi (m. 177/793 ?) et al-Farr. Ceux-ci concevaient
larabe tout la fois comme une langue une et plurielle, lua faite de
lut, cest--dire non pas de varits autonomes, mais de variantes, tribales ou rgionales, bonnes ou mauvaises (cette hirarchie suffit bien sr
pour ne pas en faire des variationnistes avant lheure !). Aujourdhui,
larabe dit classique est gnralement compris, non comme un tat de la
langue, mais comme une langue standardise, cest--dire retenant certaines lut et en liminant dautres. Elle est un aboutissement, non un
point de dpart, et, par suite, les dialectes arabes modernes ne sauraient
en tre des corruptions .
Dans la mesure o beaucoup des traits dcrits par les grammairiens arabes (notamment les plus anciens dentre eux : Sbawayhi et
al-Farr) se retrouvent, identiques ou analogues, dans les parlers arabes daujourdhui, il ny a pas de raison de penser quils ne prolongent
pas les anciens parlers arabes, selon des modalits complexes en fonction des lieux et des temps (cest le travail de la dialectologie historique
130
pierre larcher
que de dbrouiller lcheveau). Simplement, mme si les parlers arabes

daujourdhui relvent uniformment de ce qui est pour la linguistique
historique le type no-arabe, on se gardera bien de croire quils descendent uniformment dun type ancien arabe. Lopposition des deux
types, compte tenu de la non-fonctionnalit de la flexion dsinentielle
en arabe classique mme, apparat trs largement outre et, par suite,
ils ont pu trs bien coexister date ancienne et, non pas en opposition,
mais en continuit . . . Un argument en ce sens peut dailleurs tre trouv
dans la faute releve dans lun des deux plus vieux papyrus dats (22
de lHgire), o lon a Ab Qr pour Ab Qr (cf. Diem 1984, 271). Ab
Qr nest pas un nom arabe, mais larabisation dun nom grec, qui est
Apa Kyros. Pour parvenir Ab Qr, il faut donc bien passer par Ab
Qr, rinterprt comme laccusatif dune flexion triptote Ab//. Cela
nempche pourtant pas le scripteur dutiliser le nominatif Ab Qr
une place rclamant le gnitif ! Le papyrus tant contemporain de la
conqute de lEgypte (celle-ci, commence en 18 de lHgire, sachve en
25 avec loccupation dfinitive dAlexandrie) et, plus prcisment de la
fondation de Fust t (22H), il parat difficile, pour ne pas dire impossible, de voir dans ce fait le signe dune volution dun type lautre, lie
au mtissage des populations dans les centres urbains nouvellement
crs la suite de la conqute ! Et si on a donc bien ici, en mme temps,
rfrence implicite au type ancien arabe et utilisation explicite du type
no-arabe, il vaut mieux y voir la continuation dune situation de langue
plurielle , non encore standardise. Un autre argument dans le mme
sens peut tre trouv dans le Coran mme, avec les incertitudes de la
flexion visible, casuelle ou modale (pour des exemples et une analyse,
cf. Larcher 2005a).
5. Conclusion
Le rcit traditionnel des origines de la grammaire arabe, ne de la corruption de la langue , a lapparence dun rcit historique. Les orientalistes du XIXe sicle, sicle dhistoire et desprit critique, sy sont laiss
prendre, mme si, en 1906, Karl Vollers jeta un fameux pav dans la
mare ! Cent ans aprs, les linguistes arabisants voient dans larabe dit
classique, comme dans toute autre langue classique, une construction.
Ils verront alors ce rcit comme une reconstruction, ayant pour seul but
la lgitimation de cette construction.
131
6. Rfrences
6.1
Sources primaires
al-Frb, Ab Nasr. 1969. Kitb al-h urf (Alfarabis Book of Letters), d. Muhsin Mahdi,
coll. Recherches publies sous la direction de lInstitut de Lettres Orientales de Beyrouth n46, Premire srie : Pense arabe et musulmane. Beyrouth : Dar el-Machreq.
al-Jhiz, Bayn = Ab Utmn Amr b. Bahr al-Jhiz, Kitb al-bayn wa-t-tabyn, d.
Abd as-Salm Muhammad Hrn, 4 parties en 2 vols, Le Caire, 1367/1948.
Ibn Fris, Sh ib = Ab l-H usayn Ahmad Ibn Fris as-Sh ib f fiqh al-lua wa-sunan
al-arab f kalmih, d. Moustafa El-Choumi. Coll. Bibliotheca philologica arabica,
publie sous la direction de R. Blachre et J. Abdel-Nour, vol. 1. Beyrouth : A. Badran
& Co. 1383/1964.
Ibn Manzr, Lisn al-Arab = Muhammad b. Mukarram b. Al b. Ahmad al-Ansr
al-Ifrq al-Misr Jaml ad-Dn Ab l-Fadl Ibn Manzr. Lisn al-Arab al-muh t. Ed.
par Ysuf Xayyt , 4 vols. Beyrouth: Dr Lisn al-Arab. s.d.
Ibn Xaldn, Muqaddima = Wal d-Dn Abd ar-Rahmn b. Muhammad Ibn Xaldn.
al-Muqaddima, t. I du Kitb al-ibar. Beyrouth: Maktabat al-madrasa et Dr al-kitb
al-lubnn, 1967.
Suyt, Iqtirh = Jall ad-Dn Abd ar-Rahmn Ab Bakr as-Suyt , Kitb al-Iqtirh f
ilm usl an-nahw, H aydarbd, 1359 H. [reimp. Alep : Dr al-Marif, s.d.].
Suyt, Muzhir = Abd ar-Rahmn Jall ad-Dn as-Suyt al-Muzhir f ulm al-lua
wa-anwih, d. Muhammad Ahmad Jr al-Mawl, Al Muhammad al-Bajw et
Muhammad Ab l-Fadl Ibrhm, 2 vols. Le Caire : s l-Bb l-H alab. s.d.
az-Zajjj, Ab l-Qsim. 1973. al-dh f ilal an-nahw, d. Mzin Mubrak, 2me edition, Beyrouth, Dr an-Nafis.
. 1957. Al-Jumal. Prcis de grammaire arabe publi avec une introduction et un
index par Mohamed ben Cheneb, coll. Etudes arabes et islamiques, Premire srie :
Manuels. Paris : Klincksieck.
Zubayd, Tabaqt = Ab Bakr Muhammad b. al-H asan az-Zubayd al-Andals, Tabaqt
an-nahwiyyn wa-l-luawiyyn, d. M. Ab l-Fadl Ibrhm, coll. D axir al-Arab 50.
Le Caire, Dr al-Marif, 1984.
6.2 Sources secondaires
Blachre, Rgis. 195219641966. Histoire de la littrature arabe des origines la fin du
XV e sicle de J.C., I, II et III. Paris : Adrien-Maisonneuve.
Blanc, Ham. 1979. Diachronic and Synchronic Ordering in Medieval Arab Grammatical Theory. in Studia Orientalia Memoriae D.H. Baneth Dedicata. Jerusalem: The
Magnes Press, The Hebrew University. 155180.
Blau, Joshua. 2002. A Handbook of Early Middle Arabic, The Max Schloessinger Memorial Studies Monographs 6, Institute of Asian and African Studies, Faculty of Humanities, The Max Schloessinger Memorial Foundation and The Hebrew University,
Jerusalem.
Diem, Werner. 1984. Philologisches zu den arabischen Aphrodito-Papyri. in: Der
Islam 61, 251275.
EI2 = Encyclopdie de lIslam, nouvelle dition, Leiden, Brill, 1960.
Ferrando, Ignacio. 2000. Le morphme de liaison /an/ en arabe andalou : Notes de
dialectologie compare , in : Oriente moderno, 80:1, 2546.
Ferguson, Charles A. 1959. Diglossia. Word, 15:2, 325340.
Fleischer, H. 1847. Ueber einen griechisch-arabischen Codex rescriptus der Leipziger
Universitts-Bibliothek. ZDMG, I, 148160 [repris dans Kleinere Schriften, 1885
1888 t. III, ch. XXII, 378388].
132
pierre larcher
Fleischer, Heinrich. 188588. Ueber arabische Lexicographie und Talibs Fikh alluah. Berichte ber die Verhandlungen der Kniglich Schs. Gesellschaft der Wissenschaften. Philol.-histor. Cl. 114 [repris dans Kleinere Schriften, 18851888, t. III, ch.
IX, 152166].
Fck, Johann. 1955 [1950]. Arabya. Recherches sur lhistoire de la langue et du style
arabe, Paris, Didier [tr. fr. de Arabya. Untersuchungen zur arabischen Sprach- und Stilgeschichte, Abhandlungen der schsischen Akademie der Wissenschaften zu Leipzig,
Philologisch-historische Klasse. Band 45, Heft 1, Berlin, Akademie Verlag, 1950].
. 1955. Die arabischen Studien in Europa bis in den Anfang des 20. Jahrhunderts.
Leipzig: Otto Harrassowitz.
Goldziher, Ignaz. 1994. On the History of Grammar among the Arabs. An Essay in Literary History, translated and edited by K. Dvnyi et T. Ivnyi, Studies in the History of
the Language Sciences, 73, Amsterdam: Benjamins, 1994 [traduction anglaise de A
nyelvtudomny trtnete az araboknl, Nyelvtudomnyi Kzlemnyek 14, 307375,
1878].
Gruntfest, Y. 1991. From the History of Semitic Linguistics in Europe: an Early Theory
of Redundancy of Arabic Case-endings. in: K. Dvnyi et T. Ivnyi, eds. Proceedings
of the Colloquium on Arabic Grammar, The Arabist. Budapest Studies in Arabic 34.
Budapest. 195200.
Holes, Clive. 1995 [2004]. Modern Arabic. Structures, Functions, and Varieties. London:
Longman [Revised Edition, Washington D.C.: Georgetown University Press].
Kahle, Paul. 1947 [1959]. The Cairo Geniza, First Edition: 1947, Second Edition: 1959.
Oxford: Basil Blackwell.
Kazimirski, A. de Biberstein. 18467. Dictionnaire arabe franais, 2 vols. Paris : Thophile Barrois.
Langhade, Jacques. 1994. Du Coran la philosophie. La langue arabe et la formation du
vocabulaire philosophique de Farabi. Prface de Jean Jolivet. Damas : IFEAD.
Larcher, Pierre. 2003. Diglossie arabisante et fush vs mmiyya arabes : essai dhistoire
parallle . Auroux, Sylvain et al. eds. History of Linguistics 1999. Selected Papers from
the Eighth International Conference on the History of the Language Sciences (ICHoLS
VIII), Fontenay-St.Cloud, France, coll. SIHoLS 99. Amsterdam: Benjamins. 4761.
. 2004. Thologie et philologie dans lislam mdival : relecture dun texte clbre
de Ibn Fris (Xe sicle) , dans Le discours sur la langue sous les rgimes autoritaires,
Cahiers de lILSL, n 17. Universit de Lausanne. 101114.
. 2005a. Arabe prislamique, arabe coranique, arabe classique : un continuum ? ,
dans Karl-Heinz Ohlig et Gerd-Rdiger Puin (Hrsg) : Die dunklen Anfnge. Neue
Forschungen zur Entstehung und frhen Geschichte des Islam. Berlin : Verlag Hans
Schiler. 248265.
. 2005b. DIbn Fris al-Farr ou un retour aux sources sur la lua al-fush , Asiatische Studien/Etudes asiatiques, LIX, 3, 797814.
. 2006a. Un texte dal-Frb sur la langue arabe rcrit ? Lutz Edzard and Janet
Watson (eds.). Grammar as a Window onto Arabic Humanism. A Collection of Articles
in honour of Michael G. Carter. Wiesbaden: Harrassowitz. 108129.
. 2006b. Sociolinguistique et histoire de larabe selon la Muqaddima dIbn Xaldn
(VIIIe/XIVe sicle) . Pier Giorgio Borbone, Alessandro Mengozzi e Mauro Tosco
(eds.), Loquentes Linguis Studi linguistici e orientali in onore di F.A. Pennacchietti /
Linguistic and Oriental Studies to Honour F.A. Pennacchietti / Lingvistika kaj orientaj
studoj honore al Fabrizio A. Pennacchietti. Wiesbaden : Harrassowitz. 425435.
Littmann, Enno. 1929. Die vorislamisch-arabische Inschrift aus Umm ij-Jiml , ZS,
VII, 197204.
. 1949. Arabic Inscriptions (Syria : Publications of the Princeton University Archaeological Expedition to Syria in 19045 and 1909. Division IV, Section D). Leiden: E. J.
Brill.
Marais, William. 1930. La diglossie arabe , in LEnseignement publicRevue pdagogique, tome 104 n12 (1930), 401409.
133
Owens, Jonathan. 1998. Case and proto-Arabic. Bulletin of the School of Oriental and
African Studies, I:61/1, 5173 et II:61/2, 215227.
Rabin, Cham. 1951. Ancient West-Arabian. London: Taylors Foreign Press.
Versteegh, C. H. M. 1977. Greek Elements in Arabic Linguistic Thinking. Leiden: E. J.
Brill.
Versteegh, Kees. 1981 [1983]. A dissenting grammarian: Qutr ub on declension. Historiographia Linguistica 8:23, 403429 [repris dans Cornelis H.M. Versteegh, Konrad
Koerner and Hans-J. Niederehe, eds. The History of Linguistics in the Near East, Studies in the History of Linguistics, 28. Amsterdam: Benjamins. 1983.167193].
. 1984. Pidginization and Creolization: The Case of Arabic. Amsterdam Studies in
the Theory and History of Linguistic Science, Series IV-Current Issues in Linguistic
Theory. Amsterdam: Benjamins.
. 1995. The Explanation of Linguistic Causes, Az-Zajjjs Theory of Grammar. Introduction, Translation, Commentary. Studies in the History of the Language Sciences
75. Amsterdam: John Benjamins.
. 1997a. Landmarks in Linguistic Thought III. The Arabic Linguistic Tradition. Coll.
History of Linguistic Thought. London: Routledge.
. 1997b [2001]. The Arabic Language. Edinburgh: University Press [2me d. 2001].
. 2004. Pidginization and Creolization revisited: The Case of Arabic. dans Martine Haak, Rudolf de Jong and Kees Versteegh, eds. Approaches to Arabic Dialects.
A Collection of Articles presented to Manfred Woidich on the Occasion of his Sixtieth
Birthday. Leiden: E. J. Brill. 343357.
Wetzstein, Johann Gottfried. 1868. Sprachliches aus den Zeltlagern der syrischen
Wste. Zeitschrift der Deutschen Morgenlndischen Gesellschaft 22. 69194.
7. Traduction du texte dal-Zajjj

Pourquoi cette sorte de science a t appele nahw
Si lon demande pour quelle raison cette sorte de science a t appele
nahw et sest vue attribuer ce nom, on rpondra : la raison en est lhistoire rapporte au sujet de Ab l-Aswad ad-Dual. Lorsquil entendit
parler les mtis dArabes Basra, il rprouva les fautes de langage quils
commettaient, du fait de leur contact avec la vie sdentaire et les enfants
des non-Arabes. Une fille lui dit un jour : Papa, m aaddu l-h arri ?
[quelle est la chaleur la plus intense ?] La canicule en plein midi,
ma petite fille, lui rpondit-il, ou quelque chose de ce genre, car il y a
divergence dans le rcit. Je ne tai pas demand cela, lui dit-elle, je me
suis seulement tonne de la chaleur intense . Alors dis, reprit-il, m
aadda l-h arra ! [quelle chaleur intense !] . Et dajouter : Nous appartenons Dieu; la langue de nos enfants sest corrompue . Il songea faire
un ouvrage, o il rassemblerait les fondements de larabe, mais Ziyd len
empcha. Nous ne croyons pas, dit-il, que les gens se fient ce livre, ni
quils abandonnent la [bonne] langue et cessent de tirer la puret [linguistique] de la bouche des Arabes . Et, ce, jusqu ce que les fautes de
langage se rpandent, deviennent nombreuses et affreuses. Alors, il lui
ordonna de faire ce quil lui avait interdit. Et Ab l-Aswad fit un livre
134
pierre larcher
contenant la syntaxe de larabe et dit aux gens : suivez cette voie, cest-dire allez dans ce sens ; nahw signifie direction et cest pourquoi la
grammaire a t appele nahw.
On dit quil fut le premier crire dans un ouvrage que le discours
est nom, verbe et particule, dote dune valeur smantique. Interrog
ce sujet, il dclara : je lai emprunt au commandeur des croyants Al b.
Ab Tlib (Dieu tende ses bndictions et son salut sur lui !) .
Il arrive quun nom, un qualificatif ou un surnom lemporte pour une
chose. Celle-ci est alors connue sous ce nom spcifiquement, lexclusion de tout autre objet entrant dans la comprhension de ce nom. Le
fiqh, on le sait, est lintelligence des choses. On dit faqihtu le rcit aussi
bien que je [l] ai compris et un homme faqh ou faqih, cest--dire qui
comprend. Puis le fiqh est devenue la science religieuse, spcialement.
Et quand on dit un homme faqh, on vise seulement lhomme savant en
matire de Loi, mme si toute personne qui comprend une science et y
excelle est un faqh en cette science. Et, de mme, tibb est lhabilet. Cest
de l que lon dit un homme de tibb et tabb, sil est habile. Puis tabb est
devenu insprable de ceux qui sintressent la science des philosophes
ayant pour effet la conservation de la sant et, plus spcialement, permettant de la recouvrer. Les exemples de ce genre de choses abondent.
SBAWAYHIS VIEW OF THE ZARF AS AN MIL

Aryeh Levin
The Hebrew University, Jerusalem
1. Introduction
1.1
The meaning of the term zarf
In Sbawayhis terminology the term zarf (plural zurf ), designates an

expression denoting place or time. The zarf is an accusative as in al-laylata to-night,1 or a combination of an accusative + genitive as in maahu with him,2 xalfa-ka behind you3 and yawma l-jumati Friday,4 or a
combination of a h arf jarr + genitive, as in f-h in it.5
1.2
The syntactic status of the zarf
In given syntactic constructions the zarf is an indispensable part of the

sentence, while in others it is a dispensable part.
When the zarf is an indispensable part it occurs as a predicate in
some types of the nominal sentence, as in the examples abdu llhi fh
(Sb. I:222, 18) and fh abdu llhi (Sb. I:222, 17) Abdallah is in it6
and abdu llahi fh qiman Abdallah is standing in it (Sb. I:222, 15).
A zarf also occurs as the indispensable predicate in sentences beginning with inna and the kind of kna the later grammarians called kna
an-nqisa, as in the examples inna fh zaydan and inna zaydan fh
verily Zayd is in it (Sb. I:222, 20); inna zaydan lafh qiman verily
Zayd is standing in it. (Sb. I:242, 12); m kna fh ah adun xayrun
minka nobody better than you was in it (Sb. I:21, 7) and m kna
1
See Sb. I:176, 1720.

The form maa is conceived of by Sbawayhi and the other grammarians as a noun
taking the accusative (see Sb. I:177, 1415). When following the particle min ma takes
the genitive, as in the example min maihifrom him (ibid.).
3
See Sb. I:170, 1720.
4
See Sb. I:176, 1720.
5
See Sb. I:207, 2021.
6
See Sb. I:222, 1420. Cf. Sb. I:207, 2021.
2
136
aryeh levin
ah adun mitluka fh nobody like you was in it (ibid.). When the zarf
is the indispensable predicate of the sentence it is called mustaqarr,7 lit.,
a place where someone is. As a grammatical term mustaqarr designates an indispensable predicate of the nominal sentence, denoting the
place where the subject is.8 The term mustaqarr is sometimes restricted
by an expression denoting its grammatical quality as an indispensable
part of the sentence, as in mustaqarran taktaf bihi a predicate denoting
the place where the subject is, with which you content yourself [when
intending to express a complete sentence] (Sb I:21, 11), and mustaqarran lizaydin yastan bihi s-suktu a predicate denoting the place where
Zayd is [occurring in the sentences inna fh zaydun and inna zaydan
fh], with which a complete sentence can be satisfied as its complement.9 (Sb. I:222, 2021).
When the zarf is not an indispensable part of the sentence it is called
ayr mustaqarr not a mustaqarr.10 It is said that the zarf in this case is
mulan or law a dispensable zarf which does not operate as an mil
(see below 4).
The form mustaqarr sometimes occurs in combinations referring to a
dispensable part of the sentence. These combinations include restrictive
expressions indicating that the whole combination refers to a dispensable part: in referring to the example fh abdu llhi qimun Abdallah is standing in it (Sb. I:223, 2), where fh is a dispensable part of
the sentence, Sbawayhi says that fh here is a mustaqarr lil-qiym an
expression denoting the place where the act of standing [expressed in
the predicate qimun] takes place.11 Similarly, in referring to the examples fha abdu llahi qiman and abdu llhi fha qiman Abdallah is
standing in it (Sb. I:222,15), Sbawayhi says that qiman is a h lun
mustaqarrun fh [an expression denoting] a h l (= a state) where [the
subject abdu llhi] is. Note that the above combination refers to a part
of the sentence that is a h l and not a zarf.
See Sb, I:21, 717; 222, 1422.

This definition is inferred from Sb. I:222, 1422; See also Sb. I:21, 411; as-Srf
III, 11, 914; as-Srf according to Jahn, 1895, I/2, 73, note 16.
9
Lit. with which silence can be satisfied [after expressing a complete sentence].
10
See Sb. I:21, 1415.
11
See Sb. I:223, 19.
8
sbawayhis view of the zarf as an mil
137
2. The amal in sentences containing a zarf

2.1 A zarf cannot operate as the mil producing the nominative in the
subject
Sbawayhi believes that in nominal sentences a zarf cannot operate
as the mil producing the nominative in the subject (= al-mubtada),
irrespective of whether the zarf occurs as an indispensable predicate
(= mustaqarr) or as a dispensable part of the sentence. Hence, in a sentence like fh abdu llhi qiman (Sb. I:222,15), the mil of the subject
abdu llhi is not the predicate fh, but the abstract mil called al-ibtid.12
In Sbawayhis view, the sense of al-ibtid is the act of putting the noun
in a position where it is unaffected by any word operating as an mil.13
The view that the zarf cannot be the mil of the subject derives from
the notion that in a nominal sentence, a word operating as an mil producing the nominative must be logically identical with the noun affected
by it. For example: in the sentence abdu llhi axka Abdallah is your
brother (Sb. I:6, 11), the subject abdu llhi is logically identical with
the predicate axka, since Abdallah is your brother and your brother
is Abdallah.14 Hence the subject abdu llhi is the mil producing the
nominative in the predicate axka.15 In contrast, in fh abdu llahi
qiman, the predicate fh is not identical with the subject abdu llhi,
since it is an expression denoting the place where the subject is, and
hence fh cannot be the mil producing the nominative in the subject
abdu llhi.16 As a result, this sentence does not include any word that
can operate as an mil producing the nominative in the subject abdu
llhi. Hence, abdu llhi takes the nominative because of the effect of
al-ibtid.17
Sbawayhi contends that examples beginning with inna, such as inna
fha zaydan (Sb. I:222, 20), confirm that fh does not produce the
12
See Sb. I:222, 14223, 18.

See Levin, (forthcoming), mubtada, 4.1.
14
This notion is discussed in detail in Levin, 1979, 199202; Levin, 2002, 359,
15360, 11; Levin, 2006, 110111, 5; Levin, forthcoming, Cahiers linguistiques, 3.2.
15
For this notion see Sb. I:239, 59.
16
See Sb. I:222, 14223, 18.
17
See Sb. I:222, 1419.
13
138
aryeh levin
nominative in the subject, since in the above example inna is the mil
of the subject zaydan, as is shown by its accusative ending.18
Although Sbawayhi believes that the zarf cannot be the mil of
the mubtada, it is inferred that he holds that the zarf can produce the
accusative in nouns occurring as a h l, as in nominal sentences of the
type abdu llhi fh qiman and fh abdu llhi qiman Abdallah is
standing in it (Sb. I:222, 15).19 In his view, in these examples, the mil
producing the accusative in the h l qiman is the zarf fh.20 Sbawayhi
also explicitly says that the zarf is the mil producing the accusative in
words denoting measures of distance occurring as a tamyz 21 (see below
3).
2.2 Arguments confirming that in Sbawayhis view the zarf is the mil
of the hl
Sbawayhi does not explicitly say that the zarf is the mil producing the
accusative in the h l in certain constructions of the nominal sentence.
However, his view in this respect is inferred from some places in the
Kitb text.
2.2.1 In his discussion of sentences beginning with m kna, Sbawayhi
refers to two types:
(i) sentences where the zarf fh occurs as a mustaqarr, i.e., as the
indispensable predicate of the sentence, as in m kna fh ah adun
xayrun minka nobody better than you was in it (Sb. I:21, 7).
(ii) sentences where the zarf fh occurs as a dispensable part of the
sentence (mulan or law), as in m kna ah adun xayran minka
fh (Sb. I:21, 10).
In this discussion Sbawayhi explicitly says that a zarf which is a mustaqarr can operate as an mil.22 He also adds here that since a mustaqarr
18
Sb. I:222, 19223, 1.

Sbawayhi refers to the accusative in these examples both as a h l (Sb. I:223, 1) and
a xabar (Sb. I:222, 140). For this special use of xabar see Levin, 1979, 193196, 2.4.
20
See Sb. I:222, 49. See also Sb. I:167, 1116; Sb. I:218, 616, especially lines
1213. See below 2.22.2.3.2.
21
The later grammarians term tamyz does not occur in the Kitb.
22
See Sb. I:21, 719.
19
139
can operate as an mil, it is preferable to put it in a position where it

precedes the noun affected by it. The more the speaker makes it precede
this noun the better, he says. In contrast, he says that when the zarf is
not a mustaqarr, it is preferable to put it at the end of the sentence, or at
least in a position close to the end.23
2.2.2 In his above discussion, dealing with the possibility that a zarf
can operate as an mil, Sbawayhi does not give any example illustrating
this point. However, it is clear that his statements in this respect refer to
examples such as fh abdu llhi qiman and abdu llhi fh qiman,
discussed elsewhere in the Kitb.24 Since Sbawayhi explicitly says that
in these examples the mil of the mubtada abdu llhi is not the zarf,
but al-ibtid (see above 2.1), it is inferred that the mil producing
the accusative in qiman, which is a h l, is the zarf fh. It should be
stressed that the zarf fh is the only part of the sentence which can be
the mil producing the accusative in qiman, since the mubtada abdu
llahi is logically identical with qiman, and a noun which is logically
identical with another noun can produce in it only the nominative, as in
fh abdu llhi qimun (see above 2.1).
2.2.3 The above inference is supported by the following considerations:
2.2.3.1 In his discussion of the example laka -u tan bidirhamin
tan bidirhamin [The sale of] the sheep became binding on you25
[in a situation where their price is] one dirham a sheep, one dirham a
sheep (Sb. I:167, 13), Sbawayhi says that the accusative tan occurring twice in this example is a h l. He explains how the accusative is
produced in tan as follows: wasra laka -u id nasabta bimanzilati
wajaba -u kam kna fh zaydun qiman bimanzilati staqarra zaydun qiman when you put [tan] in the accusative [the utterance]
laka -u is equivalent to wajaba -u, just as [the utterance] fh
zaydun qiman is equivalent to istaqarra zaydun qiman (Sb. I:167,
13).26 Since Sbawayhi believes that in all verbal sentences the verb is the
23
See Sb. I:21, 919.

See Sb. I:222, 14224,2.
25
The translation here is based on Lanes rendering of the expression wajaba l-bayu
(see Lane VIII, 2922A,1517). According to Sbawayhi laka -u is equivalent to
wajaba -u.
26
See Sb. I:167, 1116. Cf. Sb. I:222, 14223, 1.
24
140
aryeh levin
mil of the h l,27 it is clear that in his view, the mil of tan in wajaba
-u atan bidirhamin tan bidirhamin is the verb wajaba, while the
mil of qiman in istaqarra zaydun qiman is the verb istaqarra. Since
Sbawayhi holds that laka and fh are equivalent to wajaba and istaqarra
which operate as awmil, it is inferred that in his view laka and fh
also operate as awmil in the corresponding examples laka -u atan
bidirhamin tan bidirhamin and fh zaydun qiman respectively. For
similar examples where a combination of li + genitive or li + a relative
clause operates as an mil see Sb. I:223, 18224, 2.
2.2.3.2 In referring to the example fh zaydun qiman fh In it is
Zayd, in it (Sb. I:238, 15) Sbawayhi says: fainnam ntasaba qimun
bistigni zaydin bifh [The h l qiman] takes the accusative because
of the fact that when [the first] fha is added to zaydun, it makes the
sentence complete.28 The significance of this wording is that since the
first fh in the above utterance is the indispensable predicate making
the sentence complete, it is the mil producing the accusative in the h l
qiman in fh zaydun qiman fh.
3. The zarf as an mil of a tamyz denoting a measure of distance

Sbawayhi explicitly says that in the example dr xalfa drika farsaxan
the wandering territory of my tribe is situated in a distance of one
parasang from that of yours (Sb. I:176, 67), the mil producing the
accusative in farsax is the zarf xalfa drika, which is the indispensable
predicate of the sentence.29 He adds that this zarf produces the accusative in farsaxan and in other measures of distance, since it is not logically identical with farsaxan, and since farsaxan is not the sifa of xalfa
drika. He compares the effect of xalfa drika on farsaxan with that of
irna on dirhaman, in the example irna dirhaman.30
27
See Sb. I:15, 1822.

See Sb. I:238, 1418.
29
See Sb. I:176, 612. For xalfa as a zarf see Sb. I:177, 14.
30
See Sb. I:176, 611. For the amal in irna dirhaman see Carter, 1972. See also
Levin, (forthcoming), Cahiers linguistiques, 3.2.
28
141
4. The il of the zarf

According to Sbawayhi, one of the syntactic qualities of the zarf is alil, lit. the abolishment. It is inferred from the text of the Kitb that
this term, when referring to a zarf occurring in certain syntactic constructions, denotes the abolishment of the status of the zarf as an indispensable predicate of the sentence. In this case al-il is opposed to
al-istiqrr the occurrence of the zarf as an indispensable predicate.31
In other syntactic constructions al-il denotes the abolishment of the
status of the zarf both as an indispensable predicate and as an mil. In
these constructions al-il is opposed both to al-istiqrr and al-iml,
i.e., the appliance of the amal of the zarf to the case ending of a certain
noun.32
The following examples illustrate the contrast between al-il on the
one hand, and al-istiqrr and al-iml on the other hand in certain syntactic constructions:
(1) In the example m kna fh ah adun xayrun minka nobody better
than you was in it (Sb. I:21, 7), the zarf fh is the indispensable
predicate of the sentence33 (= al-mustaqarr). In contrast, in m kna
ah adun xayran minka fh (Sb. I:21, 10), the il of the zarf fh
takes place, since the indispensable predicate is xayran minka, and
hence the status of fh as the indispensable predicate is abolished.34
(2) In the examples abdu llahi fha qiman and fh abdu llhi qiman
(Sb. I:222, 15) the zrf fh is the indispensable predicate,35 and it
is also the mil of the h l qiman . In contrast, in fh abdu llhi
qimun (Sb. I:223,2) the il of fh takes place, since qimun is
the predicate, and fh loses its status as an indispensable predicate
and as an mil.36
31
See Sb. I:21, 1415.

The contrast between il and iml is inferred from Sb. I:21, 914, where
Sbawayhi uses the forms milan and yamalni in contrast to al-il and alayta. For
the occurrence of the term iml in the Kitb, also when referring to other awmil, see
Troupeau, 1976, 149, voc. iml.
33
The expression indispensable predicate here is based on Sbawayhis view of xabar
kna (see Levin, 1979, 203205, 2.6).
34
See Sb. I:21, 714.
35
See Sb. I:222, 14223,1.
36
See Sb. I:223, 12.
32
142
aryeh levin
Sbawayhi compares the il of the amal of the zarf with that of the
verbs later called afl al-qulb.37 He says that when the il of the zarf
takes place, it is preferable to pronounce it at the end or close to the end
of the sentence. In contrast, when the zarf is a mustaqarr occurring as an
mil, it is preferable to put it at the beginning of the sentence, like verbs
such as zunnu and ah sibu, which are pronounced at the beginning of
the sentence when they are awmil.38
Sbawayhi illustrates two types of taqdr construction of sentences
containing a zarf mulan:
(i) In referring to the sentence inna bika zaydan maxdun Zayd is
enchanted by you (Sb. I:242, 2), where the il of the zarf bika takes
place, Sbawayhi says that when the speaker expresses this sentence, it
is as if he were saying inna zaydan maxdun. Similarly, when saying
inna fka zaydan la ribun Zayd covets you (Sib. I:242, 5) it is as if the
speaker were saying inna zaydan ribun.39 These taqdr constructions
illustrate the notion that when the above sentences are pronounced it
is as if the zarf bika and fka are not spoken, and hence they cannot
operate as the mil producing the accusative in maxdun and ribun
respectively.
It appears that Sbawayhi holds this view in order to solve a theoretical difficulty arising from one of the main principles of the theory
of amal: in his view, the effect of an mil producing the nominative
or the accusative in the noun is always applied, irrespective of whether
this mil is an indispensable part of the sentence or not.40 This principle
seems to be violated if one assumes that when a zarf such as fh is an
indispensable predicate, as in fh zaydun qiman, it is the mil producing the accusative in the h l qiman, but when fh is a dispensable
part, as in fh zaydun qimun, its amal is abolished. In order to solve
this difficulty Sbawayhi says that when the amal of the zarf is abolished, the zarf does not occur in the taqdr construction. Since according to the grammarians the relevant construction, as far as grammatical
analysis is concerned, is that of the taqdr, Sbawayhi assumes that fh
37
The il of the amal of this category of verbs is discussed in chapter 31 of the
Kitb (= Sb. I:49, 452, 15).
38
See Sb. I:21, 1013.
39
See Sb. I:242, 28. Sbawayhis words expressing this notion are very clear:
. . . kaannaka aradta inna zaydan ribun wainna zaydan maxdun walam tadkur fka
wal bika fauliyat hhun kam uliyat f l-ibtidi (Sb. I:242, 78).
40
See Sb. I:223, 613.
143
does not occur in the taqdr construction,41 and hence it is clear that its
amal cannot be applied.
(ii) Sbawayhi says that when the speaker expresses the sentence
fh abdu llhi qimun he intends it is as if he were saying abdu llhi
qimun fha. This taqdr construction illustrates the view that when
the zarf is not an indispensable predicate and hence is not an mil, the
speaker imagines that it is as if he were pronouncing the zarf at the end
of the sentence, since as regards grammatical theory it is preferable to
pronounce a zarf which is not an mil at the end of the sentence, or at
least in a position close to the end (see above 2.2.1). Sbawayhi says
that when the il of the zarf occurs in an example like fh abdu llhi
qimun the speaker imagines that it is as if he were saying abdu llhi
qimun fh, since in this taqdr construction the zarf which is not an
mil occurs at the end of the sentence.42
5. An interpretation of a difficult passage from the Kitb

(Sb. I:207, 1721)
Sbawayhis discussion in Sb. I:207, 1721 is one of the most difficult
passages in his text. It contains some points relevant to the topic of this
paper.
In referring to the example marartu birajulin maahu saqrun sidan
bihi adan I passed by a man with a hawk who was intending to hunt
with it tomorrow (Sb., Hrn II:52, 67),43 Sbawayhi says:
walam annaka id nasabta f hd l-babi faqulta marartu birajulin maahu
saqrun sidan bihi adan fal-nasbu al h lihi lianna hd laysa bibtidin
wal yubihu fh abdu llhi qimun adan lianna z-zurfa tul h att
yakna l-mutakallimu kaannahu lam yadkurh f hd l-mawdii, faid
sra l-ismu majrran aw milan fhi filun aw mubtadaun lam tulihi
liannahu laysa yarfauhu l-ibtidu, waf z-zurfi id qulta fha axawka
qimni yarfauhu l-ibtidu
41
For taqdr constructions which are shorter than their corresponding literal constructions see Levin, 1997, 146148, 3.3.
42
See Sb. I:222, 22223,6.
43
Derenbourgs edition has idan instead of sidan (see Sb. I:207, 1718). However, Hrns version sidan is supported by Sb. I:206, 8 in Derenbourg. The version
sidan also occurs in all the later grammarians treatises (see, for example, al-Fris, I:
250, 710; as-Srf, VI: 131, 7 (in a quotation from Sbawayhis text).
144
aryeh levin
know that if you put a noun in the accusative in [syntactic constructions
of] this type,44 and you say marartu birajulin maahu saqrun sidan bihi
adan, the accusative ending [occurring in sidan] remains unchanged,
because this [utterance]45 is not [an independent sentence] occurring at
the beginning of the utterance,46 and it does not resemble [the nominal
sentence] fh abdu llhi qimun adan (= Abdallah is standing in it
tomorrow) [where the form qim can take either the nominative or the
accusative],47 because [when expressing sentences such as fh abdu llhi
qimun adan ] the [amal] of the expressions denoting place (= zurf )
is abolished, [and the feeling of] the speaker is that [when expressing his
literal utterance he intends] it is as if he were not pronouncing the zarf
in this place at all.48 When the noun [rajulin] takes the genitive [in the
example marartu birajulin maahu saqrun sidan bihi adan], or when
it is affected by a verb [as in the example raaytu rajulan maahu saqrun
sidan bihi adan]49 or [when it is affected by the mubtada [in an example
such as hd rajulun maahu saqrun sidan bihi adan]50 you do not abolish the amal [of the zarf maa which produces the accusative in sidan],
because the ibtid does not produce the nominative in the noun [rajulin
in the example marartu birajulin maahu saqrun sidan bihi adan], [and
hence this noun is not a mubtada which can be the mil producing the
nominative in the word sid]. [On the other hand, in sentences beginning
with] a zarf,51 when you say fh axawka qimni (= your two brothers
are in it), the ibtid produces the nominative in [the mubtada, which is
axawka]52 [so this mubtada is the mil producing the nominative in the
xabar, which is qimni (Sb. I:207, 1721).53
44
I.e., in syntactic constructions including a relative clause beginning with a zarf,
such as marartu birajulin maahu ksun maxtmun alayhi I passed by a man having
with him a sealed sack (Sb. I:207, 1516). Many examples of this type are discussed in
Chapter 112 of the Kitb (=Sb. I:206,5210,2).
45
I.e., the clause maahu saqrun sidan bihi adan.
46
The term al-ibtid here denotes a position occurring at the beginning of the utterance. For al-ibtid in this sense, see, for example, Sb. II:295, 16; 296, 1115; 297, 36;
362, 1623.
47
See Sb. I:222, 14223, 2. See above 4.
48
I.e., when the speaker says fh abdu llhi qimun he intends it is as if he were saying abdu llhi qimun (for this notion see as-Srf, VI:135, 18136, 5; see above 4).
49
This example does not occur in the Kitb. It has been introduced here according
to al-Friss interpretation (See al-Faris, I:251, 35) in order to explain Sbawayhis
intention.
50
This example does not occur in the Kitb. It is introduced according to al-Friss
interpretation (see al-Faris, I:251, 37), in order to explain Sbawayhis intention.
51
In Sbawayhis manner of expression a combination such as f az-zurf denotes the
sense of in sentences beginning with a zarf . Similarly, the expression f l-fil denotes
in sentences beginning with a verb (see Sb. I:17, 17); f darabain a sentence beginning with daraba (see Sb. I:16, 1820); f knain a sentence beginning with kna
(Sb. I:17, 12).
52
For this notion see Sb. I:222, 14223, 18.
53
Ibid.
145
The following remarks and conclusions are inferred from the above passage. These conclusions are supported by other texts in the Kitb, discussed in this paper.
(1) A zarf can operate as an mil only when it is a mustaqarr, i.e.,
only when it is the indispensable predicate of a certain sentence or of a
certain clause.
(2) The zarf cannot be an mil producing the nominative in the subject or in the predicate. In a nominal sentence such as fh abdu llhi
qimun the mubtada abdu llhi takes the nominative because of the
amal of al-ibtid, and abdu llahi is the mil producing the nominative
in the predicate qimun.
(3) In a sentence such as marartu birajulin maahu saqrun sidan
bihi adan, the noun rajulin cannot be the mil producing the nominative in sid, since rajulin is not affected by the ibtid, and hence it
is not a mubtada. In Sbawayhis view, in a nominal sentence, only a
mubtada can produce the nominative in a noun occurring as a predicate. The noun affected by the mubtada must be logically identical with
it. Since the sentence marartu birajulin and the clause maahu saqrun
sidan bihi adan do not contain a mubtada logically identical with
sid, sid cannot take the nominative. The only word, which can be the
mil of sid in the above utterance is the zarf maa, which can produce
the accusative in sid. Hence it is impossible to abolish the amal of
maa, since if its amal were to be abolished the word sid would remain
without an mil.
(4) But in fh axawka qimni , the mubtada axawka takes the
nominative because of the amal of al-ibtid, and hence it can be the
mil of the predicate qimni. It is also possible to say fh axawka
qimayni in it are your two brothers standing. In this structure
axawka takes the nominative because of the amal of al-ibtid, and
fh, which is a mustaqarr, is the mil producing the accusative in the
h l qimayni.
(5) The form ibtid contained in the expression lianna hd
laysa bibtidin denotes an expression occurring at the beginning of
the utterance. It does not denote here any of the terms it designates
in Sbawayhis terminology of the nominal sentence. The words lianna
hda laysa bibtid express the notion that the clause maahu saqrun
sidan bihi is not an independent sentence occurring at the beginning
of the utterance.
(6) There is another argument which, according to as-Srf, prevents the il of maahu in the clause maahu saqrun sidan bihi adan:
146
aryeh levin
maahu, as-Srf says, includes the antecedent -hu , referring to rajulin

in marartu birajulin maahu saqrun sidan bihi adan. If the il of
maahu were applied, it would have been dropped from the taqdr construction, which in this case would be marartu birajulin saqrun sidun
bihi adan. This taqdr construction cannot exist, since the clause contained in it does not include an antecedent referring to rajulin. Since
in the grammarians view the taqdr construction is the relevant one
as far as grammatical analysis is concerned, it is impossible to apply
here the il of maahu, since this would create an unacceptable taqdr
construction.
6. Conclusions
1
Syntactically, Sbawayhi distinguishes between two kinds of a zarf:

1.1 A zarf which is an indispensable predicate of a nominal sentence, as fh in the example fh abdu llhi qiman In it is
Abdallah standing. This kind of zarf is called mustaqarr [A
zarf] denoting the place where the subject is.
1.2 A zarf which is a dispensable part of the sentence, as fh in the
example abdu llhi qimun fh Abdallah is standing in it.
This kind of zarf is sometimes called mulan or law or ayr
mustaqarr.
2 A zarf which is an indispensable predicate (= mustaqarr) is liable to
operate as the amil producing the accusative in a part of a sentence
occurring as a h l or a tamyz denoting a measure of distance. For
example:
(i) In fh abdu llhi qiman, fh is the mil of the h l
qiman.
(ii) In dr xalfa drika farsaxan the wandering territory of
my tribe is behind that of yours, at a distance of one parasang, the zarf xalfa drika is the mil of the tamyz farsaxan.
3 In contrast, a zarf which is not an indispensable part of the sentence
cannot operate as an mil. In Sbawayhis view, this zarf undergoes
the process of al-il, i.e., the process of the abolishment of its status as an indispensable predicate and as an mil. Hence, this zarf
is called mulan or law [a zarf whose] status as an indispensable
predicate and as an mil is abolished.
147
4 Sbawayhi illustrates two types of taqdr construction of sentences

containing a zarf mulan:
4.1 In referring to examples like inna bika zaydan maxdun Zayd
is enchanted by you (Sb. I:242, 2), where the il of the zarf
bika occurs, Sbawayhi says that when the speaker expresses
this sentence, he intends it is as if he were saying abdu llhi
maxdun. This taqdr construction illustrates the notion that
when pronouncing the above sentence the speaker intends that
it is as if the zarf bika is not pronounced, and hence it cannot
operate as the mil producing the accusative in maxdun.
4.2 When the speaker expresses the sentence fh abdu llhi
qimun he intends it is as if he were saying abdu llhi qimun
fha. This taqdr construction illustrates the view that when the
zarf is not an indispensable predicate and hence is not an mil,
the speaker imagines that it is as if he were pronouncing the
zarf at the end of the sentence, since as regards grammatical
theory it is preferable to pronounce a zarf which is not an mil
at the end of the sentence, or at least in a position close to the
end.
7. References
7.1
Primary sources
al-Fris, Ab Al al-H asan b. Ahmad b. Abd al-affr. (d. 377/987). at-Talqa al
Kitb Sbawayhi. Iwad b. H amad al-Qz, ed. 1410 A.H. = 1990. Cairo.
Sbawayhi. (d. 177/793). Le livre de Sbawaihi. Trait de grammaire arabe. Hartwig
Derenbourg, ed. 18811889. Paris. 2 vols.
Ab Bir Amr b. Utm
n b. Qanbar. (d. 177/793). al-Kitb. Kitb Sbawayhi. Abd
as-Salm Hrn, ed. Cairo, 1977. 5 vols.
as-Srf, Ab Sad. (d. 368/979). arh Kitb Sbawayhi. Ramadn Abd at-Tawwb and
others, eds. Cairo, 19882004. 6 vols.
7.2
Secondary sources
Carter, Michael G. 1972. Twenty Dirhams in the Kitb of Sbawayhi, Bulletin of the
School of Oriental and African Studies. London 35, 485496.
Jahn, G. 1895. Sbawaihis Buch ber die Grammatik, bersetzt und erklrt von G. Jahn.
Vol. I, second pagination. Berlin.
Lane, E.W. 18631893. Arabic-English Lexicon. London (8 volumes).
Levin, Aryeh. 1979. Sbawayhis view of the Syntactical Structure of kna wa-axawtuh,
Jerusalem Studies in Arabic and Islam 1, 185211.
. 1997. The Theory of al-Taqdr and its Terminology. Jerusalem Studies in Arabic
and Islam 21, 142166.
148
aryeh levin
. 2002. An Interpretation of a Difficult Passage from the Kitb. Jerusalem Studies in

Arabic and Islam 27, 356362.
. 2006. An Interpretation of Two Difficult Passages from al-Kitb Referring to the
mil in Elliptical Sentences. Jerusalem Studies in Arabic and Islam 31, 107117.
. (forthcoming). Sbawayhis View of al-mubtada and al-ibtid. Moshe Bar-Ashers
Festschrift.
. (forthcoming). The mil of the xabar in Old Arabic Grammar. In: Cahiers linguistiques de Inalco.
Troupeau, Grard. 1976. Lexique-Index du Kitb de Sbawayhi. Paris: Klincksieck.
PROBLEMS IN THE MEDIEVAL ARABIC THEORY OF

SENTENCE TYPES
Yishai Peled
Tel Aviv University
1. Introduction
In medieval Arab grammatical tradition, the two basic sentence types
jumla filiyya and jumla ismiyya, are normally defined by the first
occurring predicative constituent. A verb followed by its subject signals a jumla filiyya (verbal sentence), e.g. daraba abdu-llhi zaydan
(Abdullh hit Zayd). By contrast, a sentence introduced by a nominatival noun is a jumla ismiyya (nominal sentence), e.g. zaydun rajulun
(Zayd is a man). This binary division corresponds with the grammarians theory of amal (regimen), or, to be more specific, with two basic
types of amal. A jumla filiyya correlates with a verbal mil (operator), whereas a jumla ismiyya corresponds with ibtid, which is considered an abstract mil.1 The ibtid is normally said to consist of a
zero phonological mil and the predicatival relationship between the
mubtada and the xabar, respectively the subject and predicate in this
type of sentence. The basic principle of amal, stipulating that the mil
should precede the maml, applies in both sentence types. In a jumla
filiyya, the verbal mil affects the complements following it; in a jumla
ismiyya, the abstract mil, ibtid, occupies in principle a pre-mubtada
position, from where it assigns the raf case to the mubtada; the latter,
in turn, assigns raf to the xabar (according to Sbawayhis [Kitb I:239]
version). As for cases such as daraba zaydan abdu-llhi and rajulun zaydun, these were presented as cases of taqdm wa-taxr (preposing and
postposing), i.e. as the inverted versions of daraba abdu-llhi zaydan
and zaydun rajulun; in other words, as inverted jumla filiyya and jumla
ismiyya respectively.
For a detailed discussion of this correspondence, see Levin 1985.
150
yishai peled
However, the linkage between the concept of two sentence types on

the one hand, and the theory of amal on the other, turned out to constitute a major problem with regard to the binary division into jumla filiyya
and jumla ismiyya. The grammarians realized that such types as daraba
abdu-llhi zaydan and zaydun rajulun, with their inverted versions, leave
various constructions that do not easily fit into any of the two categories.
The two apparently most problematic cases may be represented by the
two model sentences qimun zaydun (Standing is Zayd), where a participle is followed by a definite noun phrase, and zaydun f d-dr, or f
d-dri (or fh) zaydun (Zayd is in the house/in it), where the predicate
position is occupied either by a definite prepositional phrase or by an
adverbial phrase such as hun, hunka etc. In the latter type, a definite
subject noun may either precede or follow the adverbial/prepositional
predicate; an indefinite subject must obligatorily follow its predicate.
As it were, sentences such as qimun zaydun and f d-dri (or fh)
zaydun may be considered as cases of an inverted jumla ismiyya, pragmatically motivated. And modern linguists, uncommitted to the theory of amal, would probably regard them as such. Yet for many of the
medieval grammarians, they represent, rather, a sentence type in its own
right. To be more precise, f d-dri zaydun is explicitly presented as such
by some of the grammarians; qimun zaydun, by contrast, is often dealt
with in a way that leads one to believe that it was considered by certain
grammarians as representing a third sentence type.2 I would argue that
the controversies that arose over these (and other) constructions point
to what may be viewed as gaps in the medieval theory of sentence types.
In other words, an attempt will be made to show that the theory, based
on a binary conception of jumla filiyya and jumla ismiyya, representing
two types of amal, was highly vulnerable and far from stable. This paper
concentrates on the theoretical problems presented by the above two
constructions. We start with qimun zaydun.
2
To be sure, there were grammarians who analyzed both constructions as an inverted
jumla ismiyya with a fronted xabar; others accepted more than one type of analysis. For
a discussion, cf. Ibn Ab r-Rab, Bast I:583ff.
the medieval arabic theory of sentence types
151
2. qimun zaydun
In his bb al-ibtid, Sbawayhi (Kitb I:239) discusses the option of
mubtada-xabar inversion. His starting point is that the standard preferred (al-h add) structure is for the mubtada to precede the xabar rather
than the reverse, much as the standard word order in the verbal sentence
is for the fil to precede the maf l. When dealing with inversion, he singles out qimun zaydun as a markedly complex case deserving special
attention. He quotes his teacher al-Xall as saying that qimun zaydun is
an ill-formed (qabh ) sentence unless analyzed as the inverted version of
zaydun qimun. As is well known, some of the later grammarians made
the point that such an inversion is quite problematic, since it places the
maml before the mil, and the Kfans saw a further problem in that it
makes the pronoun implicit in the participial form qimun precede its
antecedent (al-idmr qabla d-dikrsee, e.g. Ibn al-Anbr, Insf I:65).
But these problems were easily dismissed by the claim that qimun zaydun represents a secondary ( far) or surface (lafz) structure, whereas in
the basic structure (man, niyya, taqdr) zaydun, the zhir and mil
precedes qimun, the maml, with the implicit pronoun (the mudmar)
referring backas requiredto zaydun (for a detailed discussion, see,
e.g. Ibn al-Anbr, Insf I:6566, 68).
However, the real problem with qimun zaydun was associated with a
differentverbalanalysis of the participle, known to have been advocated by some of Sbawayhis contemporaries. Citing al-Xall, Sbawayhi
points out that:
fa-id lam yurd hd l-man wa-ard an yajalhu filan ka-qawlihi
yaqmu zaydun wa-qma zaydun qabuh a li-annahu smun
If, however, they do not accept this analysis [= inversion], and want to
treat [qimun] as a verb, in analogy to such sentences as yaqmu zaydun
and qma zaydun, this should be rejected, because [qimun] is a noun
(Sbawayhi, Kitb I:239).
Yet Sbawayhi immediately makes it clear that under certain conditions

an active participle, while categorized as a noun, may implement a verbal
function (yajr majr l-fil). This could be accepted (h asuna indahum),
he maintains, if the participle functions as part of an asyndetic relative
clause (sifa) linked to some antecedent (mawsf), or, otherwise, governed
by a preceding operator such as a mubtada. In other words, qimun
zaydun is disallowed with a verbal analysis, much as dribun zaydan
(hitting Zayd) is unacceptable as a complete sentence. However,
152
yishai peled
qimun zaydun, as part of a larger sentence, is acceptable with a verbal

analysis, in analogy to such cases as an dribun zaydan (I am hitting Zayd).3 As we shall see shortly, Sbawayhis approach to the verbal
analysis of qimun zaydun was later established as a firm principle in
medieval Arab grammatical thinking.
Ibn as-Sarrj (Usl I:5960), whose position regarding mubtadaxabar inversion seems to be similar to that of Sbawayhis, readily accepts
muntaliqun zaydun as an inverted version of zaydun muntaliqun. As
for analyzing qimun zaydun as analogous to yaqmu zaydun where
qimun is not preceded by any supporting element (see below), in
other words, construing zaydun as fil to muntaliqunin principle, Ibn
as-Sarrj, much like Sbawayhi, regards such an analysis as misguided
(qabuh a), yet he admits it as jiz (acceptable). What both grammarians seem to accept without reservation is that the noun following the
participle may be analyzed as a kind of fil provided that it is anchored
(yatamidu al, in Ibn as-Sarrjs words) to some preceding constituent. As an illustration of how this condition can be met, Ibn as-Sarrj
adduces such sentences as: marartu bi-rajulin qimin abhu (I passed
by a man whose father was standing), zaydun qimun abhu (Zayd
his father is standing), a-qimun abka (is your father standing?).
This rule would be developed by later grammarians into a general
principle of itimd, designed to specify the conditions under which a
non-verbal predicate may be analyzed as analogous to a verb preceding its subject.4 This principle stipulates that a non-verbal predicatival
constituent, such as an active participle, or an adverbial/prepositional
phrase (see section 3 below) may exercise amal upon the constituent
following it (the subject) only if supported by (yatamidu al) some element such as an interrogative particle, a relative pronoun, or, otherwise,
when the clause as a whole functions as an asyndetic relative clause or as
xabar to a preceding mubtada. In such cases, the first predicatival constituent is perceived as behaving analogously to a verb. And as a verblike constituent it acts as a mil, assigning the raf case to the following
3
Sbawayhi (Kitb I:239) asserts that while the active participle and the verb may be
similar in some respects, one must appreciate the difference between them. Other grammarians (e.g. Ibn al-Anbr, Asrr, 70) pointed out that the active participle is weaker
than the verb, and cannot, therefore, exercise verbal amal, unless supported by some
preceding element (see below).
4
For the concept itimd as it is used in al-Xalls Kitb al-Ayn with reference to other
grammatical structures, see Talmon 1997, 210.
153
nominal constituent.5 For discussion, see, e.g. Ibn al-Anbr, Asrr: 70;
cf. Goldenberg 2002, 199201.
Ibn as-Sarrj is aware, however, of the implications of a verbal analysis of qimun zaydun for the theory of amal. He argues (Usl I:60) that
in qimun abka (Your father is standing), qimun is assigned the
raf case by the ibtid, and abka is assigned the same case by the verb
preceding it. He indicates further that abka fills a xabar position.
In any event, both Sbawayhi and Ibn as-Sarrj reject the use of dribun
bakran amrun ( Amr hits Bakr) as an independent sentence, on the
ground that the active participle, while being analogous to the verb, is
by definition a nominal, and as such cannot be made to function identically to a verb in terms of case assignment. The above examples, where
the participle is linked to a preceding antecedent (mawsf), a mubtada,
or an interrogative particle, are viewed as analogous to the construction dribun bakran when anchored, under the principle of itimd, to
some external constituent (mah ml al ayrihi), such as a mubtada,
thus presenting a well-formed independent sentence (e.g. hd dribun
bakranthis [person] is hitting Bakr) (Ibn as-Sarrj, Usl I:60; and cf.
Sbawayhis position above; Levin 1985, 125126).
Like Sbawayhi and Ibn as-Sarrj, as-Zajjj (Jumal, 3738) was aware
of the theoretical problems raised by sentences consisting of an active
participle followed by a noun phrase. In particular he demonstrated the
implications for the grammatical agreement between the two constituents. To the extent that qimun in qimun zaydun is conceived of as
xabar muqaddam (a fronted xabar), it must be replaced by qimni or
qimna, once zaydun is substituted by a dual or a plural form respectively. But under the alternative analysis cited by as-Zajjj, in which
qimun is assigned a verbal function, the active participle preceding
its subject should invariably take the singular form. In other words, the
proponents of this analysis would have qimun az-zaydni/az-zaydna
rather than qimni z-zaydni and qimna z-zaydna.6
5
Ibn Ab r-Rab (Bast I:585) remarks that some grammarians rejected the idea of
an adverbial/prepositional phrase assigning case. They argued that such phrases were
different in status (manzila) from the adjective. The latter, they argued, is capable of
inflection, and as such is more powerful than the adverbial/prepositional phrase. Therefore, they concluded, the adverbial/prepositional phrase may not be analyzed as a case
assigner even where the principle of itimd is met. I return to this issue later.
6
See, e.g. Ibn Ab r-Rab (Bast I:584), who also indicates that the proponents of
akaln l-bart must, by extension, say qimni z-zaydni and qimna z-zaydna,
154
yishai peled
This rule, as stated by as-Zajjj, highlights the verbal function of the

participle under this particular analysis, relating qimun az-zaydni/azzaydna to qma z-zaydni/z-zaydna. It should be noted that the medieval grammarians treated participles, as well as other types of adjective,
as complex forms incorporating a personal pronoun (for discussion, see
Goldenberg 2002, 195). Ibn Ya (arh I:8788) states clearly that participles and other adjectives are derived from the verb, and that in virtue
of having a verbal meaning (man fil) they must have a fil.7 However, construing muntaliqun zaydun as modeled on yantaliqu zaydun
implies neutralizing the pronominal element in muntaliqun exactly as
it is done in yantaliqu.8 The argument is that the participle, much like a
regular fil, cannot assign the raf case twice (l yarfau filayni).9
Zajjj (Jumal, 38) indicates that in such cases the active participle
introducing the sentence is assigned raf by the ibtid, whereas the constituent following it is assigned the same case by its verb (bi-filihi
apparently referring to the active participle; cf. Ibn as-Sarrjs analysis
above). Zaydun in qimun zaydun, it is argued, replaces the xabar (yasuddu masadd al-xabar); and the participle preceding it, he points out,
is invariably singular li-annahu qad jar majr l-fili l-muqaddami (for
it behaves analogously to a verb preceding [its subject]).10
Observe that, unlike the vast majority of grammarians, Zajjj did not
make the point of linking the verbal analysis of qimun zaydun to an
obligatory application of the principle of itimd. It should, indeed, be
noted that some of the later grammarians held a narrower version of
this principle, restricting the use of a participle in sentence-initial position to cases where the participle is attached to a negative or interroga-
since in this version of the language the verb preceding the subject agrees with it in
number and gender (for a detailed discussion of akaln l-bart, see Levin 1989).
7
The grammarians, however, recognized that the personal pronoun incorporated
in an active participle cannot qualify as fil in the way an implicit personal pronoun
in a verb can. Thus, while allad daraba zaydun (The one who hit is Zayd) is a perfectly grammatical sentence, allad dribun zaydun is not, since, unlike allad daraba,
allad dribun cannot implement the function of a subject clause (see Jurjn, Muqtasid
I:463464).
8
This is, perhaps, why the Kfans, who rejected the analysis of qimun zaydun as an
inverted nominal sentence, could accept it as modeled on a verbal sentence: under the
verbal analysis the pronoun in qimun is disabled so there is no problem of cataphora
(cf. above).
9
For Ibn Ya (arh I:8788), then, a sentence such as zaydun qimun abhu
(Zayd, his father is standing) consists of a mubtada (zayd) and a xabar, the latter analyzed as a complex construction consisting of a fil (qimun) and a fil (abhu).
10
This type of analysis is attributed to Axfa; see, e.g. Ibn Usfr, arh I:341.
155
tive particle: m qimun az-zaydni (the two Zayds are not standing),
a-qimun az-zaydni (Are the two Zayds standing?). According to Ibn
Aql (d. 1367) (arh I:189), qimun, in each of the last two sentences,
functions as a mubtada, whereas az-zaydni is a fil sadda masadd alxabar (a fil substituting for the xabar). Indeed, this is the common
formula employed by those later grammarians who adopted the verbal
analysis of the construction in question (cf. Carter 1981, 189).
At this point one might ask how frequent in classical Arabic are such
sentences as qimun az-zaydni? I looked for this construction in the
Qurnic text, but no example of it was attested. In all the recorded cases,
a singular participle is followed by a singular noun phrase, or, otherwise,
a singular feminine participle by a plural (non-human) noun phrase.
It is interesting to note, however, that all cases display some kind of a
supporting element. In the vast majority of examples, the construction in question functions as predicate to a preceding subject realized
as either a referential nominal (typically, but not necessarily, in a sentence introduced by inna or one of its sisters), or, otherwise, as a nonreferential damr a-an: wa-zann annahum mniatuhum h usnuhum
(They believed that their fortresses would protect themQ. 59:2),
wa-huwa muh arramun alaykum ixrjuhum (You are not allowed to
expel themQ. 2:85). Huwa in the latter example functions as damr
a-an. One example was noted where the supporting element is the
interrogative particle a-: a-ribun anta an lihat (do you loathe my
gods?Q. 19:46).
What is then the effect of the supporting element that makes [zaydun/rajulun] qimun abhu or a-/m qimun abhu an acceptable
verbal construction, as opposed to qimun abhu? Al-Xall (see above)
does not provide an elaborate answer. He argues, however, that a participle cannot easily replace a verb in pre-subject position, because it is an
ism. A verb and a noun, he maintains, may in certain positions implement similar functions, but they must still be differentiated. Nor did later
grammarians elaborate on the function of the supporting element. But
their discussion of the relevant cases might give us a clue. To phrase the
question differently, how does the supporting element impart further
verbal force to the adjectival predicate that enables it to act analogously
to a verb in such cases? If we compare the two constructions qma zaydun and qimun zaydun, we can see that the difference between the
two is that the finite verb, while devoid of a pronominal element, is still
inflected for person, whereas the participle is not. Lacking either a pronominal element or inflection for person, the participle is excluded as a
156
yishai peled
pre-subject verbal predicate in an independent sentence. The function

of the supporting element zaydun in zaydun qimun abhu, is to make
up for the lack of person inflection in qimun, and thus empower the
latter to implement a verbal function.11
Further, as has been indicated, the supportive function may be
implemented not only by a noun but also by an interrogative or a negative particle. These are, in other words, further sources from which the
adjective could derive a verbal force. Already Sbawayhi attributed a
verbal effect to certain interrogative particles (Sbawayhi, Kitb I:39f.,
41f.the latter dealing specifically with a-). Ibn Ab r-Rab (Bast
II:712) points out that the interrogative particle a- requires a verb (alhamza tlibatun bi-l-fil), and that underlying (taqdr) a-zaydun daraba
amran is a-daraba zaydun amran (Did Zayd hit Amr?) (and cf. his
similar attitude to halII:679). Regarding the construction at issue he
maintains that underlying a-qimun zaydun is a-yaqmu qimun
zaydun, and that the former is derived from the latter by suppressing the
redundant yaqmu, for which qimun serves as an exponent (tafsr).
This explains why the interrogative a- qualified in the grammarians
view as a supporting element in sentences consisting of an adjectival
predicate followed by a subject.
However, the verbal analysis of such constructions as (a-)qimun
zaydun raises a difficult problem for the medieval theory of sentence
types. For if a sentence consists of a mubtada followed by a fil, how
should it be categorized in terms of sentence types? The fact that the
fil is presented as replacing the xabar does not make the issue any
simpler. For under the suggested analysis, a verb-acting constituent is
considered, in terms of irb, as a mubtada assigned the raf case by
the ibtid. The conception of the participle in such cases as mubtada is
quite understandable, given that it is a nominatival constituent in sentence-initial position. If one declines the inversion analysis of sentences
such as qimun zaydun, how else can one account for the raf case of
qimun? The main problem with this analysis lies in its stipulating that
qimun as a mubtada assigns raf to a fil occupying a xabar position.
11
That, I believe, is what is intended by al-Xall (Sbawayhi, Kitb I:239) when he
refers to the participle in such cases as [kna] sifatan jar al mawsfin aw jar al
smin qad amila fhi [the participle] is an adjective agreeing with a head or [otherwise] with a noun acting upon it, and further when he says: l yaknu maf lan f
dribin h att yaknu mah mlan al ayrihi fa-taqlu hd dribun zaydan . . . drib
cannot take an object unless it is linked to some other constituent, as for example in
hd dribun zaydanthis [person] is hitting Zayd.
157
Once we correlate qimun zaydun with qma zaydun, rather than with
zaydun qimun, with the consequence that qimun is to remain in
the singular irrespective of the number of the following noun (qimun azzaydni/az-zaydna), one can hardly see how sentences such as qimun
zaydun under a verbal analysis, let alone qimun az-zaydni, may be
viewed other than as cases of jumla filiyya. Indeed, when presenting the
fil, some of the grammarians, like Ibn Ya (arh I:74), indicate that
the position preceding the fil is available for a verb ( fil) or a nominal that
is analogous to a verb (abahuhu, m huwa f man l-fil min al-asm).
In this latter category they normally include the active and passive participles, as well as such adjectives as h asan (sifa muabbaha bi-smi lfilan adjective analogous to the active participle, e.g. Ibn Ya, arh
I:87). It is argued that in a sentence such as zaydun dribun ulmuhu12
(Zayd, his slave is hitting), dribun, much like yadribu, assigns raf to
ulmuhu. One may infer, then, that Ibn Ya would regard a sentence
such as qimun zaydun as a verbal sentence. Yet I have not recorded
any explicit reference to this type of sentence as a jumla filiyya. In any
case, the prevalent analysis of the construction under discussion was, as
already indicated, mubtada+fil sadda masadd al-xabar. It is not surprising, however, that the grammarians adhering to this analysis did not
commit themselves to explicitly categorizing such sentences as either
jumla filiyya or jumla ismiyya.
A remarkable exception is Ibn Him al-Ansr (d. 1360) who, in his
famous book Mun l-Labb, provides an elaborate discussion of Arabic
sentence types. Ibn Hims classification will be discussed in detail in
section 5 below. As we shall see, he defined three sentence types (rather
than two!) by the kind of constituent introducing the sentence. Thus,
a sentence introduced by a nominal element is a jumla ismiyya. And
among his examples of jumla ismiyya we find the sentence qimun azzaydni. Ibn Him was, indeed, aware of the controversy surrounding
this sentence, indicating that it was accepted as a well-formed sentence
by Axfa and the Kfans. As we saw above, qimun az-zaydni was
adduced as an acceptable sentence in Arabic also by Zajjj, but the latter
did not classify it as jumla ismiyya.
12
Note, however, that in Zamaxars and Ibn Yas examples the construction at
issue is itself a xabar following a mubtada. As we have seen, this is consistent with the
principle of itimd.
158
yishai peled
In sum, then, the analysis of qimun zaydun into a mubtada+fil

replacing a xabar appears to reflect a twofold attempt:
(1) to make this kind of sentence conform to the principle of ibtid;
this is motivated by the fact that the first constituent is a nominal
exhibiting a raf case ending, and;
(2) to apply the principle of verbal tadiya, given the verbal properties
of the active participle.
But, to the extent that this analysis holds, does not it follow (at
least from the grammarians viewpoint) that qimun zaydun represents a sentence type in its own right? To my knowledge, no such
proposition has ever been advanced in medieval Arab grammatical
literature.13
3. fh / f d-dri zaydun
Only a small minority of the grammarians suggested that sentences
such as fh zaydun (In it there is Zayd) should be regarded as representing a sentence type in its own right. They designated this type jumla
zarfiyya, but differed on whether this term should or should not cover
also sentences such as zaydun fh/f d-dr (Zayd is in it / in the house).
I return to this later. At this point, let us examine the grammarians conception of this construction, starting with Sbawayhi.
3.1
Sbawayhi
In his bb al-ibtid (chapter 132), Sbawayhi does not develop any discussion of this sentence sub-type. But elsewhere in the Kitb (I:170171;
cf. Levin 1987, 362, and Owens 1989, 224) he argues that in cases such as
huwa xalfaka (He is behind you) it is the subject huwa that assigns the
nasb case to xalfaka. Indeed, this is consistent with his argument (Kitb
13
Badawi (2000, 8f.) claims that the grammarians recognized three types of sentences
namely: filiyya, ismiyya and wasfiyya, introduced, respectively, by a verb, a noun and an
adjective (a participle or otherwise). He emphasizes the use of different terms for the
subject and predicate in each sentence type, indicating that in the jumlat wasf these are
referred to as mubtada and fil sadda masadd al-xabar. However, the term jumlat wasf
has not been attested in the grammarians writings studied for the present work.
159
I:239) that in zaydun muntaliqun, it is the subject that assigns raf to the
predicate.14 However, what about sentences such as fh zaydun, consisting of an adverbial/prepositional phrase followed by a nominatival
noun phrase? Sbawayhi deals with these cases within the framework of
his discussion of sentences such as fh abdu-llhi qiman and abdullhi fh qiman (Abdullh is in it, standingcf. Talmon 1993, 281).
He starts his discussion analyzing qiman as an accusatival xabar to
abdu-llhi. Then he goes on to indicate that abdu-llhi in these cases:
irtafaa bi-l-ibtidi li-anna llad dukira qablahu wa-badahu laysa bihi
wa-innam huwa mawdiun lahu wa-lkinnahu yajr majr l-ismi l-mabniyyi al m qablahu
is assigned the raf case by the ibtid since the constituent occurring
either before or after it [= the adverbial] is not it [= is not identical in reference with it], but rather signals its location. Yet [this adverbial] functions
analogously to a noun built upon the [subject] preceding it(Sbawayhi,
Kitb I:222).
Sbawayhi, as can be seen, points out that fh is non-coreferential with

abdu-llhi, but rather refers to Abdullhs location (mawdi). But it is
precisely this observation that underlies his endeavor to establish, first
of all, the acceptability of fh abdu-llhi/abdu-llhi fh as a complete
independent sentence. To this end he draws an analogy between fh
abdu-llhi and hd abdu-llhi (This is Abdullh), claiming that
in terms of completeness, the former, much like the latter, is a kalm
mustaqm (a correct sentence) that h asuna [badahu] s-suktu (that
may appropriately be followed by silenceSbawayhi, Kitb I:222 and
239240). Similarly, abdu-llhi fh is presented by him as analogous
to abdu-llhi axka (Abdullh is your brother): in both cases the
second constituent is built (mabn) upon the first. As for abdu-llhi,
Sbawayhi states clearly that, whether preceding or following the prepositional phrase, it is assigned the raf case by the ibtid.
But once a predicatival relationship is established between fh and
abdu-llhi, Sbawayhi reanalyzes the sentence assigning fh the function of xabar and abdu-llhi the function of mubtada. The position of
qiman is then demoted to that of h l (see Figure 1 below). He points
out, however, that since fh represents the persons location, fh abdullhi is paraphrasable by istaqarra abdu-llhi. In other words, fh
14
Note, however, that the amal in huwa xalfaka is presented by Sbawayhi as analogous to that in his model construction irna dirhaman (Twenty dirhams).
160
yishai peled
mubtada
xabar
fh
abdu-llhi
qiman
xabar
mubtada
hl
Figure 1
behaves analogously to the verb istaqarra. As we shall see in 3.2, the verb
istaqarra, or otherwise the participle mustaqirrun, have since become
the grammarians most common device for explaining the grammatical
structure of sentences such as zaydun fh / fh zaydun.
A further indication of Sbawayhis consideration of fh as a verb-like
element is his statement (Kitb I:223) that qiman in the above sentence may, alternatively, be replaced by qimun in the nominative. This,
he explains, is the result of abrogating (alayta) fh. In the medieval
grammarians writings, ilg is normally used as a technical term denoting the annulling of amal. It is typically used with reference to potential
awmil, that is, elements that normally exercise amal upon other elements in the sentence (for discussion see, e.g. Peled 1992a, 150152).
One may infer, then, that Sbawayhi considered fh, in virtue of its acting analogously to istaqarra, as an mil assigning nasb to qiman in
fh abdu-llhi qiman. As we shall see later, such adverbials as fh
were considered by some early grammarians as an mil assigning raf
to the following subject in such cases as fh zaydun. This view is typically attributed to the Kfans. Yet, Sbawayhi (Kitb I:223224) then
enters into an extensive discussion designed to exclude the possibility
that fh in sentences such as fh abdu-llhi qimun is the mil assigning raf to abdu-llhi. He draws an analogy between this sentence and
bika abdu-llhi maxdun (Abdullh is fascinated by you). He argues
that an operator assigning case to an optional constituent (qimun in
the former sentence) has the same status (manzila) as an operator acting upon an obligatory constituent (maxdun in the latter).15 Sbawayhi
emphasizes that in both cases (as well as in similar ones adduced by
him) the adjective is built upon the noun, thus establishing a predicatival relationship between the two. The prepositional phrase, by contrast,
is a law, i.e. a constituent that neither assigns nor receives amal. In fh
15
Indeed, in later grammatical writings, the model sentence bika zaydun maxdun
features regularly in the Basran arguments against the Kfan claim that in fh zaydun
it is fh that functions as the mil assigning raf to zaydun (cf. Ibn al-Anbr, Insf
I:5253).
161
abdu-llhi qimun, fh is only designed to specify the location where

the standing is taking place.
Let us now return to the original construction fh abdu-llhi to
which Sbawayhi devotes a separate bb, following his bb al-ibtid. In
chapter 133 of the Kitb, he deals with such cases as fh abdu-llhi,
tamma zaydun and ayna zaydun. Compared to his discussion above,
Sbawayhi here seems to be less specific about the rfi of zaydun:
wa-llad amila fm badahu h att rafaahu huwa llad amila fhi h na16
kna qablahu
The operator assigning raf to the following constituent is the same operator that assigned it the raf case when that constituent was before it [i.e.
before fh](Sbawayhi, Kitb I:239; cf. editors notes as well as the editions of Blq and Hrn).17
Sbawayhi thus asserts that the operator assigning raf to zaydun in

fh zaydun is the same one that assigns the raf case to zayd in zaydun
fh, where/when zayd occurs before it (i.e. before fh). Apparently,
he leaves it to the reader to conclude that it is the ibtid that functions
as mil in both cases. This could lead one to believe that for Sbawayhi,
fh zaydun represents an inverted version of zaydun fh. Yet Sbawayhi
does not make any explicit claim for taqdm wa-taxr in this particular
case. He presents fh zaydun as a case in which fh:
yaqau mawqia l-ismi l-mubtadai wa-yasuddu masaddahu
occupies the position of the mubtada and replaces it(Sbawayhi, Kitb
I:239; quoted also in Kouloughli 2002, 9).
As we saw in section 2, there were among later grammarians those who

analyzed qimun in qimun zaydun as a mubtada followed by a fil
(replacing a xabar). It seems more than likely that they were influenced
by the above Sbawayhian passage.
Regarding ayna zaydun, Sbawayhi points out that ayna is paraphrasable by f ayyi makn (in which place?), and emphasizes that ayna,
as an interrogative, must obligatorily occur sentence-initially. In other
words, fh zaydun is distinct from ayna zaydun only in that in the latter
case the xabar occupies sentence-initial position obligatorily.
16
The words h aytu and h na alternate in this position in two different versions of
the text.
17
Talmon (1993, 283284) confronts the long version cited here with a shorter one
that to me looks rather obscure.
162
yishai peled
It is interesting to note that Sbawayhi does not include in his discussion such cases as fh rajulun (In it there is a man), where the subject,
being indefinite, obligatorily follows the predicate (much like in ayna
zaydun). Indeed, as we will see later, this construction received little
attention from the grammarians, compared to fh zaydun. As we shall
see, while the grammarians never failed to point out that the predicatesubject order in fh rajulun is obligatory, only a small minority of them
regarded this construction as representing a sentence type in its own
right. For the vast majority, fh rajulun, much like fh zaydun, represented an inverted jumla ismiyya. A detailed discussion of constructions
with an obligatorily fronted xabar is provided in section 4 below.
3.2
The istaqarra/mustaqirrun hypothesis
Following Sbawayhi, the medieval grammarians continued to address

themselves to the twofold problem posed by such sentences as zaydun
fh and fh zaydun. This problem, as we have seen, consisted in establishing a predicatival relationship between the nominal and the adverbial/prepositional phrase, and accounting for the mil assigning case to
each. The suggested solution of positing an underlying linking element
such as istaqarra/mustaqirrun gave rise to extensive discussions that
developed into what may be referred to as the istaqarra/mustaqirrun
hypothesis. As we saw in 3.1, the origins of this theory can be easily
traced back to Sbawayhis Kitb.18
Throughout the centuries, the istaqarra/mustaqirrun hypothesis
has become the common strategy used by the medieval grammarians
in order to account for such sentences as zaydun f d-dr (or, for that
matter, zaydun xalfaka, where xalfaka alternates with min xalfikafor
discussion of the status of the zarf, see Levin 1987, 351357).19 As for
the grammatical status of the adverbial following the mubtada, here the
grammarians differed. For Sbawayhi (Kitb I:222), muntaliqun and fh
were equally admissible as xabar for zaydun. In Ibn as-Sarrjs (Usl
I:6263) view, the xabar in zaydun f d-dr is the underlying mustaqir18
As we have seen, however, Sbawayhi used this device in dealing with sentences
such as fh abdu-llhi qiman, to account for the nasb of qiman; he did not employ
it in cases such as fh zaydun or zaydun fh, pointing rather to the ibtid, in both, as
the mil assigning raf to zaydun.
19
For the Kfan theory of xilf (or muxlafa), see Astarbd, arh I:214; Ibn alAnbr, Insf I:245247; Muji, arh , 87, n. 216; and cf. Carter 1973).
163
run which, when coreferring with the mubtada, is optionally omitted

(h adf). The adverbial f d-dr fills the xabars slot, acting as a complement to the omitted genuine xabar (for a similar view, see Ibn Usfr,
arh I:347). Zaydun xalfaka/f d-dr is thus paraphrasable by zaydun
mustaqirrun xalfaka/f d-dr. Ibn as-Sarrj points out that the omitted
element normally conveys some meaning of existence, and is redundant
because it is retrievable from the adverbial. 20 He emphasizes that postulating an underlying constituent in such cases is obligatory, because, in
itself, f d-dr (or xalfaka) does not qualify as a predicate; that is, it does
not predicate any quality of zayd (laysa bi-h adt),21 but only specifies the
location (mawdi) of Zayd.
Jurjn (Muqtasid I:274275) argues that an adverbial, with either
an explicit or implicit preposition, in principle presupposes a verb with
which it is linked to form a syntactical unit (cf. Ibn al-Anbr, Insf
I:246). Consequently, underlying f d-dr, in zaydun f d-dr, is the verbal
clause istaqarra f d-dr. This indeed represents the view of the majority of grammarians for whom positing an underlying (muqaddar) finite
verb (istaqarra) is consistent with the clausal status of f d-dr in sentences such as zaydun f d-dr. Obviously, once established as the (head
of the) xabar, istaqarra/yastaqirru is referred to as the mil, assigning
case to the following adverbial/prepositional phrase.
However, the assumption of an underlying finite verb in cases such
as zaydun f d-dr was not universally accepted. The controversy here is
linked to the fact that not all the grammarians regarded f d-dr in zaydun
f d-dr as having a clausal status. Some grammarians posited a participle rather than a verb as the underlying element linking the adverbial/
prepositional phrase to the preceding mubtada. We have just seen that
Ibn as-Sarrj was one of the proponents of this hypothesis; indeed, he
20
Ibn Usfr (arh I:347348) emphasizes that using an adverbial/prepositional
phrase as a xabar substitute is only admissible when the deleted element is recoverable
from the surface constructionotherwise, the xabar should appear in full. Thus, for
example, zaydun f d-dr is only allowed if it is intended to convey the meaning mustaqirrun f d-dr, because f, signalling a receptacle (wi), is compatible in meaning
with istiqrr (staying). If, however, zaydun f d-dr is intended to convey the meaning of dh ikun f d-dr ([Zayd] is laughing in the house), then the word dh ik must
occur; for, unlike the meaning of staying, that of laughing cannot be recovered from
the preposition f. Cf. Astarbd, arh I:215, for linking elements like h sil and kin
(be); Levin 1987, 360.
21
Ibn as-Sarrjs use of the term h adt in this case is significant, for it signalizes predicate realized whether as fil or as xabar (cf. Goldenberg 1988, 4649).
164
yishai peled
viewed sentences such as zaydun xalfaka as displaying a single-phrasedrather than a clausal xabar (Ibn as-Sarrj, Usl I:63). Astarbd (arh
I:215) cites Ibn as-Sarrj and Ibn Jinn as two grammarians who advocated the participle rather than the verb hypothesis, on the ground that
the participle as a single phrase (mufrad) is compatible with the basic
structure of the xabar. Another proponent of the participle hypothesis
is Muji (arh , 87) who derives such sentences as zaydun ammaka
(Zayd is in front of you) and amrun min al-kirm ( Amr is one of the
honorable) from the underlying (taqdr) structures zaydun mustaqirrun ammaka and amrun kinun min al-kirm respectively.22 Muji
makes it clear that for him an adverbial/prepositional phrase in xabar
position has the status of, and is therefore a substitute for, a participle
(not a clause). Postulating a personal pronoun implicit in the participle,
Muji argues further that this pronoun moves to, and resides in, the
adverbial/prepositional xabar occupying the position of the deleted participle (in Mujis words: wa-afd d-damru llad kna f smi l-fil il
n-nib anhu fa-statara fhi).23 For further discussion of the istaqarra/
mustaqirrun hypothesis, see Jurjn, Muqtasid I:275ff; Ibn al-Anbr,
Insf I:245247; Ibn Usfr, arh I:344, 349351.
We can see that the istaqarra/mustaqirrun hypothesis is used by the
grammarians to fit the construction zaydun fh/fh zaydun into their
theory of amal, and, by implication, to their binary system of sentence
types. Once zaydun in both zaydun fh and fh zaydun was recognized as a mubtada, both constructions could be said to represent a
jumla ismiyya, the latter being an inverted version of the former. The
xabar, when following the mubtada, is presented as either clausal or
phrasal, depending on whether one assumes yastaqirru or mustaqirrun
to be the underlying linking element. In both cases, this element is made
accountable for the case of f d-dr (see Kouloughli 2002, 1316, for
further discussion). However, the istaqarra/mustaqirrun hypothesis was
not universally accepted. And as we shall see, the alternative hypotheses
had substantial implications for the theory of sentence types.
22
Astarbd (arh I:215) claims that the underlying element is obligatorily deleted,
rejecting such sentences as zaydun kinun f d-dr. He indicates that Ibn Jinn did allow
such constructions, but points out that there is no evidence to support this position.
23
This is evidently Mujiis way of claiming a xabar status for the adverbial/prepositional phrase. Astarbd (arh I:216217) points to Fris and his followers as advocating the same hypothesis. But Srf is cited by him as claiming that the pronoun is
deleted as part of the linking constituent.

3.3
165
Ab Al l-Fris
Ab Al l-Fris (d. 987) is considered to be one of the first grammarians who advanced in an explicit way and developed the idea of sentence
types in Arabic (cf. Owens 1988, 3637). He defined each type, and
spelled out the problematic nature of the dichotomy verb+noun ( jumla
filiyya) versus noun+noun ( jumla ismiyya).24 Indeed, he was the first to
present a detailed argument with the conclusion that zaydun f d-dr is
neither a jumla filiyya nor a jumla ismiyya.
Having defined the two basic sentence types in Arabic, Fris turns
to concentrate upon the construction represented by zaydun f d-dr.
Indeed it looks as though Friss definition of the two basic sentence
types is meant as an introduction to his discussion of this particular
construction (Fris, Askariyya, 105109). He starts by indicating that,
although such sentences are composed of a nominal element (i.e. the
two nouns) and a particle (the preposition), they do not have the same
status as inna sentences, where the particle enters into a sentence made
up of two nouns. This is because f d-dr is non-coreferential with zaydun. And since zayd and f d-dr are not identical in reference, they
cannot be analyzed simply as subject and predicate. However, given that
zaydun f d-dr is definitely a well-formed sentence in Arabic, one must
assume some underlying (muqaddar) linking element to account for the
predicatival relationship between its two constituents. As we saw in 3.2,
this linking element must inevitably be either a noun or a verb (a particle does not bear any reference). To the extent that either of these can
be posited, a sentence such as zaydun f d-dr must eventually belong
either to the verb+noun or to the noun+noun type.25
24
Anxious to provide accurate and valid definitions, Fris (Askariyya, 104105)
points further to the option of a particle (h arf) entering into either of the two defined
jumlas, to form a kalm. What the reader is invited to infer is that the resulting construction is an independent grammatical sentence whose basic type (i.e. filiyya or
ismiyya) is unaffected. He exemplifies this by sentences introduced by hal, inna, m,
qad and lam. (As a matter of fact, the same principle had already been stated by Ibn
as-Sarrj, Usl I:43.)
25
Fris (Askariyya, 109) draws a comparison between the case in question and
address (nid) expressions. He argues that y zaydu (O, Zayd!), much like f d-dri
zaydun, consists of nominal elements and a particle, and constitutes an independent
sentence. The difference between the two, he maintains, is that in the case of y zaydu a
verbal element should be assumed, which renders the address expression a sub-type of
a jumla filiyya, whereas in the case of zaydun f d-dr/f d-dri zaydun no such element
can be posited (see below).
166
yishai peled
However, from this point onward, Friss argument continues in a

direction designed to prove that this is not the case. In other words, a
sentence such as zaydun f d-dr, while conforming to the general principle governing the production of well-formed sentences, constitutes an
exception in that it does not fall into any of the two basic categories,
jumla filiyya and jumla ismiyya. In Friss words:
a-l tar anna l-kalma wa-in kna l yaxl mimm dakarn f l-asli faqad sra lahu l-na h ukmun yaxruju bihi an dlika l-asli
Notice that although a sentence must obey the basic principles indicated
by us, in this case there is an [overriding] rule leading the sentence away
from the basic principles (Fris, Askariyya, 105).
What Fris is now trying to prove is that neither a verb nor a noun can
be posited as a linking element establishing a predicatival relationship
between zaydun and f d-dr, a relationship modeled on that obtaining between the subject and predicate of a regular verbal or nominal
sentence. And if it can be proved that neither a verb nor a noun can be
posited as an underlying linking element between zaydun and f d-dr,
the conclusion must be that sentences such as zaydun f d-dr represent
a sentence type in its own right.
He starts (Fris, Askariyya, 105) by adducing the sentence inna f
d-dri zaydan, were the particle inna enters into the sentence f d-dri
zaydun (his choice of this construction rather than zaydun f d-dr is
significant, as will be seen below.) Then he makes the following two
points:
(1) An underlying verbal link cannot be assumed, because a verb would
exclude the use of inna. In other words, while f d-dri zaydun may
be preceded by yastaqirru, inna and yastaqirru are mutually exclusive: inna f d-dri zaydan is a perfectly grammatical sentence in
Arabic, but *inna yastaqirru f d-dri zaydan is disallowed.
(2) A linking noun cannot be posited either, because that would
amount to assumingfalselythat inna exercises its effect (amal)
upon zaydan across the underlying linking noun (Fris, Askariyya,
108).
Having disqualified both noun and verb as possible underlying linking
elements in cases such as zaydun f d-dr and f d-dri zaydun, Fris
(Askariyya, 108) argues further that in such cases the adverbial constituent as such cannot be claimed to implement a verbal function. This,
167
he maintains, is borne out by the fact that the adverbial may not be preceded by a circumstantial phrase; a sentence such as *qiman f d-dri
zaydun is inadmissible, but it would be allowed if f d-dr had a verbal
value (a sentence such as qiman dah ika zaydun is considered as perfectly acceptable by the grammarians.)
All the above boils down to a rejection of the istaqarra/mustaqirrun
hypothesis, and that, in turn, leads Fris to the conclusion that sentences such as zaydun f d-dr/f d-dri zaydun should be considered
neither as jumla filiyya nor as jumla ismiyya; they must be thought of,
rather, as representing a sentence type in its own right. Note, however,
that Fris did not assign the type of sentence under discussion any special designation. The term jumla zarfiyya to which we will be introduced
below was coined in a later period.
But if one is supposed to assume no underlying element linking the
two predicatival constituents in zaydun f d-dr/f d-dri zaydun, what
is then the assigner of raf to zaydun in such cases? As we saw in 3.1,
Sbawayhi, who was not committed to any theory of sentence types, had
no problem presenting the ibtid as raf assigner to zayd in both zaydun
f d-dr and fh zaydun. But for Fris, making a similar claim would
imply classifying fh zaydun as a jumla ismiyya. Regarding the raf
assigner in zaydun f d-dr Fris does not develop any elaborate discussion, apparently because in such cases one would automatically refer to
the ibtid as the raf assigner. However, when it comes to fh zaydun,
the construction on which he focuses his discussion, Fris presents a
clear position as to the rfi of zaydun. Having shown that neither a verb
nor a noun can be posited as a linking element, and having proved, further, that the adverbial itself cannot be claimed to function as a verb,
Fris (Askariyya, 108109) refers the reader to Ab l-H asan [al-Axfa]
(d. 733), explaining that these are the reasons why Ab l-H asan regarded
the adverbial per se as the rfi when preceding a noun functioning as
muh addat anhu (of whom the message is predicated, subject). Notice
that it is not the term fil that is used with reference to that noun, but
rather muh addat anhu, a term that cuts across all sentence types. As
we shall see, however, later grammarians did not refrain from using the
term fil in this particular context.
Obviously, attributing the assigning of raf to an adverbial/prepositional phrase constitutes a serious problem for the theory of amal. Since
the formulation of this theory, the grammarians always insisted that the
function of case assignment is implemented by either a verb or a particle. Various elements, notably active participles and other adjectives
168
yishai peled
were claimed to have a verbal force. But in our case, as we have seen,
Fris argued that f d-dr was not verbal enough to allow a circumstantial phrase to precede it. So one might ask what it is that qualifies
fh as raf assigner. To my knowledge, this point has never been clarified by the grammarians. And it is no wonder that the concept of jumla
zarfiyya, where a predicatival prepositional phrase assigns raf to the
following subject, remained marginal and never became part of mainstream medieval Arab grammatical thinking. Anyhow, for Fris, Axfas
position regarding the raf assigner in fh zaydun constituted further
support for his thesis that this construction represents a sentence type
in its own right.
4. Obligatory fronting of the xabar

4.1
Formal aspects
A remarkable feature of the grammarians (including Friss) discussion

of the adverbial/prepositional xabar is that they base their argument on
such sentences as zaydun f d-dr/f d-dri zaydun, where the definite
subject may either precede or follow the predicate (cf. Kouloughli 2002,
10, n. 7). But for the third sentence type advocated by Fris, sentences
such as ind mlun (I have money) would surely be a better example.
For in this case the subject follows the predicate obligatorily; reversing
the order is disallowed.
Most of the grammarians adduce such sentences by way of illustrating
an obligatorily fronted (taqdm) xabar. However, within the framework
of their grammatical discourse, obligatory fronting of the xabar presents
a major conceptual problem. A xabar, by definition, must follow, not
precede, the mubtada. The very concept of an obligatorily fronted xabar
appears to conflict with two fundamental principles in medieval Arab
grammatical theory : 1. The formal principle stipulating that the mil
precede the mamlsee, e.g. Ibn Ab r-Rab, Bast I:587); 2. The functional principle placing the constituent representing the given information before the one representing what is new for the addressee. To the
extent that the mubtada is identified with the given, and the xabar with
the new, such sentences pose a serious problem.
Obviously, in cases such as f d-dri rajulun, the grammarians could
not present rajulun f d-dr as the asl of f d-dri rajulun, since the former is disallowed as an independent sentence in Arabic. One may argue,
169
therefore, that in f d-dri rajulun, as well as in other cases of obligatory fronting to be discussed below, the very concept of taqdm is not
normally intended in the sense that the above structure is the output of
reversing the order of some basic structure in which the indefinite rajulun precedes f d-dr. Rather, one must assume that taqdm in such cases
is used in the sense of placing in initial position, with no transformation
involved. Ibn Ya (arh I:86) argues that rajulun f d-dr is excluded
(1) because it could be wrongly interpreted as a noun phrase (with f ddr functioning as the attributesifaof rajulun) rather than as a complete sentence (see also Ibn Usfr, arh I:343), and (2) in order to avoid
introducing a declarative (wjib) sentence by an indefinite noun.26
The idea of transformation in cases such as f d-dri rajulun was not,
however, universally excluded. Not surprisingly, it was suggested, albeit
in a rather idiosyncratic manner, by Ibn Jinn, a grammarian noted
for his originality and for frequently advancing dissenting arguments
incompatible with mainstream medieval Arab grammatical thinking.
In his Sirr sinat al-irb, Ibn Jinn (Sirr I:276) uses the concept asl
marfd (a rejected basic construction) in dealing with cases which
he regards as transformed constructions, but whose underlying structure (asl) is inadmissible. The principle of asl marfd is not explicitly
applied by him to rajulun f d-dr. But in his Xasis (I:300) he maintains
that, while mubtada-xabar is the basic word order of a jumla ismiyya,
a certain intervening factor (rid) might impose the reversal of that
order. The occurrence of an indefinite mubtada at the beginning of an
affirmative sentence constitutes in Ibn Jinns view such an rid, a kind
of contingency imposing the movement of the mubtada into the second
position in the sentence (cf. Peled 1992b, 105106).27 This is regarded
26
Astarbd (arh I:232) maintains that the problem of ambiguity between xabar
and sifa is acute, owing to the common occurrence of an adverbial in xabar position
in Arabic. He cites, however, one case where an adverbial xabar follows an indefinite
mubtada, pointing out that it is perfectly acceptable when the sentence is used as an
exclamation (du). Astarbd also remarks that fronting a non-adverbial xabar to an
indefinite mubtada does not eliminate the ambiguity. Thus, if you transform rajulun
qimun into qimun rajulun, rajulun could be analyzed as xabar of qimun or as an
apposition (badal) to it, whereas a fronted adverbial in similar cases is bound to be interpreted as xabar, due to its nasb case, whether explicitly marked (lafzan), or understood
by position (mah allan).
27
Note that an indefinite mubtada introducing a negative or interrogative sentence
is readily accepted by Ibn Jinn. Thus he admits (Xasis I:300) sentences such as hal
ulmun indaka (Is there a boy with you?) and m bistun tahtaka (There is no
carpet under you), claiming that they are communicatively useful, as opposed to sentences such as rajulun indaka (A man is with you). The argument is that one can
170
yishai peled
by him as a corrective procedure (islh al-lafz):28 the second position,

he reminds us, is in principle the xabars position; and since the xabar is
essentially indefinite, the indefinite mubtada now fills the appropriate
slot as far as (in)definiteness is concerned. He emphasizes, however, that
in the underlying theoretical (muqaddar) level the mubtada precedes
the xabar (Ibn Jinn, Xasis I:318).29
There are other cases adduced by the grammarians as examples of
obligatory fronting of the xabar. The first example presented by Ibn
Ab r-Rab (Bast I:587) is the interrogative construction ayna zaydun
(Where is Zayd?). This grammarian strived to demonstrate that such
cases of obligatory xabar fronting do not violate the basic principle that
the mil must precede the maml. Regarding ayna zaydun he indicated that the basic underlying (asl) structure in this case is a-zaydun
f d-dri am f s-sqi am f l-h nti (Is Zayd at home, in the market, or
in the shop?). The word ayna is an economy device designed to replace
both the adverbials and the interrogative particle a-, as well as the particle am. It implements a xabar function in virtue of its being a replacement for the adverbials. At the same time, it is obligatorily fronted as a
substitute for the interrogative particle a-. Another semantic component in ayna is that of specification (tayn), formally represented in the
basic underlying structure by the particle am. The transformation from
the basic to the final surface structure thus proceeds in the following
stages: First, the xabar ( f d-dri am f s-sqi etc.) is moved to a frontal
position immediately following a- (a-f d-dri am f s-sqi . . . zaydun
the interrogative a- is always positioned sentence-initially). Then ayna
is introduced to replace all the constituents preceding zaydun. It is thus
the adverbial semantic component in ayna that warrants its occurrence
in pre-mubtada xabar position, whereas the interrogative component
accounts for the obligatoriness of the movement. Ibn Ab r-Rab indicates that the same applies to other interrogatives such as mat, kayfa,
negate the existence of, or pose a question with regard to, an unknown entity (mankr
l yuraf ), while there is no communicative value in predicating of an unknown entity
affirmatively.
28
Similarly, Astarbd (arh I:232) regards the obligatory fronting of the xabar in
cases such as f d-dri rajulun as a corrective (musah h ih ) procedure designed to handle
the indefiniteness of the mubtada.
29
Ibn Jinn (Xasis I:319320) then refers to sentences displaying an indefinite
mubtada in sentence-initial position. He argues, however, that these are not predicatival
sentences, in the sense that they are meant to express a wish or imprecation rather than
convey information. Another case is explained by him as paraphrasable by a negative
sentence (cf. n. 27 above).
171
man and m: kayfa axka (How is your brother?), man axka (Who
is your brother?), and the like.
Ibn Ab r-Rabs second case of obligatory xabar fronting is the construction f d-dri rajulun. This, however, is dealt by him in pragmatic
rather than in purely formal terms, and will, therefore, be reviewed in the
following sub-section. The third case is exemplified by Ibn Ab r-Rab
(Bast I:588) by the sentence al t-tamrati mitluh zubdan (on the date
there is butter of an equal amount). He indicates that the reverse order
(mitluh zubdan al t-tamrati) is disallowed since -h in mitluh is an
anticipatory pronoun in both lafz (surface) and martaba (underlying)
structures, thus violating the rule of the anticipatory pronoun (al-idmr
qabla d-dikrcf. section 2 above). For further discussion of the relationship between anaphora and the position of the xabar, see Astarbd,
arh I:232233.
Introducing his fourth and final case of obligatory xabar fronting, Ibn
Ab r-Rab (Bast I:588) cites the exceptive sentences m frisun ill
zaydun (No one is a horseman but Zayd) and m f d-dri ill amrun
(No one is in the house but Amr). In these two sentences the subject
nominal occurs sentence-finally and is dominated by the exceptive particle ill. Reversing the order in such cases, Ibn Ab r-Rab points out,
would not violate the rules of Arabic grammar, but result in a sentence
different in meaning from the original one. The sentence m frisun ill
zaydun assigns to Zayd, and only to him, the attribute of horsemanship.
This sentence, however, is neutral as to whether or not Zayd possesses
other qualities as well. But if the order of constituents is reversed so
as to make the subject zaydun precede the predicate, the resulting sentence m zaydun ill frisun unmistakably excludes the possibility that
Zayd possesses any quality beside horsemanship. Similarly, the sentence
innam frisun zaydun is equivalent in meaning to m frisun ill zaydun, whereas innam zaydun frisun is synonymous with m zaydun
ill frisun, which explains why a mubtada-xabar order is inadmissible
in this related case as well.
Ibn Usfr (arh I:353) adds two more cases where the xabar is obligatorily placed sentence-initially: 1. When the mubtada is a nominalized
clause introduced by anna: f ilm annaka qimun (It is known to me
that you are standing); 2. When the xabar is a kam al-xabariyya phrase:
kam dirhamin mluka (How many dirhams you have!).
The first of these two cases is dealt with also by Astarbd (arh
I:233234) who cites Fris as claiming that the adverbial/prepositional
phrase in such cases exercises amal (raf ) upon the following anna
172
yishai peled
clause with no supporting element (cf. 3.3 above; for itimd, see section 2 above). Astarbd (arh I:233) explains that the reason for the
obligatory fronting of the xabar (whether adverbial or not) in such cases
is that if the anna clause were placed sentence-initially, the word nna
could be misread as inna rather than anna. For, between the two particles, it is the former rather than the latter that is associated with the
initial position in the sentence. Astarbd points out further that if the
xabar precedes the anna clause it is bound to be correctly analyzed as
xabar to the following clause as a whole rather than as a fronted constituent governed by anna, because a constituent within the scope of inna/
anna cannot be preposed to either of these particles. Furthermore, once
the adverbial/prepositional phrase is established as the xabar of the following clause, then the particle heading that clause will be easily read as
anna, because a mubtada clause, being a noun clause, cannot be introduced by inna. For further discussion of this issue, see Ibn Ya, arh
VIII:5960.
As we have just seen, f d-dri rajulun was only one item, and not necessarily the first, on the list of constructions presented by the grammarians as examples of obligatory fronting of the xabar. But it was apparently
the most difficult to deal with in purely syntactic terms. For one thing,
like the related f d-dri zaydun, it presented a challenge to the theory
of amal. For another, the indefiniteness of the mubtada could not, in
itself, constitute sufficient grounds for ruling out its occurrence in sentence-initial position (cf. Astarbd, arh I:202207 for a detailed discussion of cases of an indefinite mubtada in sentence-initial position).
The grammarians main formal explanation for the obligatoriness of
predicate-subject order in this case was that an adverbial/prepositional
phrase following an indefinite nominal could be wrongly interpreted
as an attribute rather than a predicate. But as we have just indicated,
sentences with an indefinite mubtada do occur in Arabic. Indeed, as
we will see shortly, the strongest argument against *rajulun f d-dr was
pragmatic rather than syntactic.
4.2
f d-dri rajulunpragmatic aspects
When examining the construction f d-dri rajulun in terms of information structure, most of the grammarians appreciated that sentences of
this kind represent a special case. Indeed they recognized that in these
cases it is the definite adverbial/prepositional phrase, occupying sentence-initial position that represents the given information, whereas the
173
following nominatival indefinite phrase signals the new information,

and not the other way around (see, e.g. Ibn Ya, arh I:8687).30 Ibn
Ya (arh I:86) argues that, judging by the meaning of sentences such
as laka mlun (You have money), it is the definite complement of the
preposition (in this case the pronoun -ka) that represents the muh addat
anhu (what the sentence is about), even though, formally ( f l-lafz), it
is the nominatival noun that implements that function.31 In support of
his claim he indicates (1) that laka mlun is paraphrasable by anta d
mlin, and (2) that an indefinite nominal is inadmissible in the position
of the complement of the preposition: *li-rajulin mlun is disallowed
(lam yakun kalmanis not an [acceptable] sentence).
Another attempt to relate the construction in question to a mubtadaxabar order was made by Ibn Ab r-Rab. This grammarian presents
a number of cases where the xabar is obligatorily fronted (cf. 4.1
above), taking great pains to demonstrate that fronting the xabar in
these cases is wholly justified, if not on purely formal, then on functional/semantic, grounds. Regarding f d-dri rajulun, Ibn Ab r-Rab
states the following:
fa-hd yulzimu t-taqdma wa-l yajzu taxruhu fa-taqla rajulun f
d-dri li-annahu l yubtadau bi-n-nakirati wa-innam jza l-i-btidu
hun bi-n-nakirati li-anna l-maqsda l-ixbru an-i d-dri bi-annah
masknatun wa-laysati n-nakiratu l-maqsdata bi-l-ixbri wa-kna l-aslu
an taqla ad-dru mamratun bi-rajulin tumma ard l-i-xtisra fa-ql
f d-dri rajulun wa-alzam d-dra t-taqdma li-annah l-muxbaru anh
bi-l-h aqqati
in such cases [the xabar] is preposed obligatorily. It may not be postposed to yield rajulun f d-dr, because an indefinite noun may not fill a
mubtada position. In our case, the mubtada32 may be indefinite, because
the intention is to predicate of the house that it is inhabited, rather than
to predicate of the indefinite noun. Underlying [our sentence] is the sentence ad-dru mamratun bi-rajulin. But for the sake of brevity, they say
30
According to Talmon (1993, 285287), the idea is already attested in ninth century
writings where the locative is typically referred to as sifa and the nominatival noun following it as xabar as-sifa.
31
This obviously rests on the assumption that in a sentence containing only one
nominatival noun, it is this noun that should be construed as the muh addat anhu. This
term, while referring literally to a pragmatic function, signals in the grammarians usage,
the subject, irrespective of sentence type; its counterpart h adt signals the predicate (see
Goldenberg 1988, 4649, for discussion).
32
Notice that yubtadau and ibtid are both construed in this case as used as
mubtada or implementing a mubtada function.
174
yishai peled
f d-dri rajulun. They obligatorily prepose the [phrase f] d-dr, because
it is really the house that is predicated of [i.e. the topic] (Ibn Ab r-Rab,
Bast I:587588).
In other words, the sentence f d-dri rajulun is paraphrasable by ad-dru

mamratun bi-rajulin (The house is inhabited by a man), where
ad-dru obviously functions as mubtada. Ibn Ab r-Rab explains that
the meaning intended by f d-dri rajulun is that the house is inhabited;
it is ad-dr of which something is predicated, not the indefinite rajulun.
The version f d-dri rajulun, Ibn Ab r-Rab maintains, is preferred to
ad-dru mamratun bi-rajulin for reasons of economy. The constituent
ad-dr is obligatorily preposed, for it is the one that implements the
function of muxbar anhu, which obviously entails that rajulun functions as xabar. (For the terms muxbar/muxbar anhu, see Goldenberg
1988, 4651).
A different, indeed exceptional, view on this issue is held by Jurjn
(Muqtasid I:308309). Sentences such as ind mlun (I have money)
are dealt with by him within the framework of his discussion of sentences with an indefinite mubtada. For Jurjn, the mubtada in this
particular case, albeit indefinite, implements the same pragmatic function as a definite sentence-initial mubtada. Mlun in the above sentence
implements the function of mubtada
li-ajli h usli l-ixtissi f l-xabari id kullu wh idin l yalamu anna indaka
mlan
because the xabar signals some specification [regarding the mubtada],
for it is not common knowledge that you have money (Jurjn Muqtasid
I:308).
For Jurjn, then, ind in its sentence-initial position makes the same
contribution to the communicative value (ifda) of the sentence as
would a xabar occupying a post-mubtada position. Much like other
grammarians, however, Jurjn argues that in such cases the xabar is
obligatorily fronted because mlun ind would be wrongly interpreted
as a noun phrase, with ind analyzed as a complement (sifa) to mlun
(see 4.1 above).
Most of the grammarians, however, did not follow this line of thought.
Rather, they discerned a discrepancy between the syntactic analysis of
sentences such as f d-dri rajulun into xabar and mubtada, and the
pragmatic functions of muxbar anhu and xabar implemented by f ddr and rajulun respectively. They did, however, emphasize that in terms
175
of definiteness sentences like laka mlun conform to the principle that

the definite constituent, representing the given information, should precede the indefinite noun signalling the new information. In any case,
the grammarians could not accept an analysis of f d-dri rajulun into
a mubtada followed by a xabar, on the ground that if a sentence of this
kind is introduced by inna, the noun phrase rajul automatically takes
the nasb case, which marks it unmistakably as subject. For discussion,
see Kouloughli (2002, 1617), and his references.
5. Ibn Him al-Ansrs tripartite division

5.1
Background
Returning now to the question of the raf assignment in cases such as

fh zaydun, most of the grammarians from Sbawayhi onwards regarded
ibtid as the operator assigning raf to zaydun. This, however, was by no
means universal, as we have seen. Indeed, Ibn al-Anbr (Insf I:5155)
attributes this position to the Basrans while presenting the other (much
less common) position as Kfan. We learn that the Kfans, as well as
the Basran grammarians Mubarrad and Axfa (cf. 3.3 above), regarded
the adverbial/prepositional phrase in the above case as the assigner of
raf to the following noun zaydun.33 Ibn al-Anbr maintains that both
the Basrans and the Kfans resort to Sbawayhi for support for their
respective claims. The Basrans obviously point to chapter 133 in the
Kitb, where, as we have seen, Sbawayhi refers to the ibtid (though
without using the actual term) as the mil assigning raf to zaydun. The
Kfans, for their part, draw upon a number of cases where, according to
Sbawayhi, an adverbial assigns raf to a following noun (Ibn al-Anbr,
Insf I:52). It is upon such cases, Ibn al-Anbr argues, that the Kfans
base their claim that in fh zaydun it is fh that should be regarded as
the rfi of zaydun.
Astarbd (arh I:218) indicates that the analysis of zaydun in f
d-dri zaydun as fil of the adverbial/prepositional phrase was advanced
by the Kfans, as well as by Axfa in one out of two statements he made
33
This position is clearly evidenced in Farrs Man l-Qurn (e.g. I:195196;
III:133). See Talmon (1993, 279) for further details and references.
176
yishai peled
on the subject.34 The argument in both cases is that the adverbial/prepositional phrase has a verbal force (man l-fil), analogously to qim in
qimun zaydun.
In Astarbds view, the Kfans position emanates from their categorical objection to xabar fronting, irrespective of whether the xabar
is a phrase or a clause. This objection, in its turn, is designed to forestall the occurrence of an anticipatory pronoun. To this, however,
Astarbd offers an outright rejection, claiming, as could be expected,
that the anticipatory pronoun occurs only in the surface structure; in
the basic structure the mubtada precedes the xabar with no anticipatory pronoun involved (cf. section 2 above).
As for Axfa, according to Astarbd he did not object to xabar
fronting, and (in his other statement) indeed regarded the ibtid as the
assigner of raf to zaydun in f d-dri zaydun. Axfas position here is
divergent from the one cited above. It rests upon two assumptions: 1.
The verbal force of the adverbial/prepositional phrase is weaker than
that of the adjective. 2. The acceptability of the construction f drihi
zaydun (in his house Zayd [is located]). In this case, the option of analyzing zaydun as fil is excluded on the ground that it would lead to an
unacceptable anticipatory pronoun (since the pronoun, under this analysis, would be cataphoric both in the lafz and man configurationscf.
Astarbd, arh I:202).
The debate, as could only be expected, concludes with the Basrans
having the upper hand. There is evidence to suggest, however, that, as in
many other cases, the position dismissed as Kfan represented a view
that was much more widely accepted than the mainstream medieval
grammarians would have us believe. As we shall see in the next sub-section, it remained viable centuries later, in the writings of one of the most
prominent medieval grammarians, Ibn Him al-Ansr. The relevance
of this issue to our discussion is clear: it is closely related to the question
of whether or not sentences such as fh zaydun represent a sentence
type in its own right.
34
Ibn al-Anbr (Insf I:51) adds Mubarrad to the proponents of this kind of analysis (and cf. Ibn Usfr, arh I:158159).

5.2
177
Ibn Hims categorization and definitions
Coming now to the 14th century grammarian Ibn Him al-Ansr

(d. 1360), we find both a critical discussion of the concept jumla as compared with kalm, and an unambiguous division into three sentence
types. Recall that Friss starting point was that there were two basic
sentence types in Arabic. This was followed by an elaborate argument
designed to prove that zaydun f d-dr represented a sentence type in its
own right. For Ibn Him the tripartite division is an established linguistic fact, and he makes it the starting point of his discussion. He does not,
however, ignore the problems raised by this division, as we shall see.
Ibn Him (Mun, 492) starts by defining the concepts kalm and
jumla. The latter is defined by him as a construction made up of either
fil + fil or, otherwise, mubtada + xabar. Then, however, he distinguishes three types of jumla, ismiyya, filiyya and zarfiyya, introduced
respectively by a noun, a verb and an adverbial (zarf) or a prepositional
phrase ([ jrr wa-] majrr).35 Being aware of the problem arising from
his defining each type by the initially occurring constituent, he remarks
(Mun, 492) that the definitions refer only to predicative constituents
(musnad and musnad ilayhi). Thus, the sentence a-zaydun axka (Is
Zayd your brother?) and the conditional clause in qma zaydun (If
Zayd stands up) are, respectively, ismiyya and filiyya, even though the
first noun in the former and the verb in the latter are each preceded by
a particle.
Once the problem of the particles is settled, Ibn Him appears to
be remarkably uncompromising in applying his principle, that Arabic
sentence types must be defined by the initial predicative constituent.
And this, indeed, leads to some conspicuous peculiarities. The jumla
ismiyya, for instance, is exemplified by him by the following three sentences: zaydun qimun (Zayd is standing), hayhti l-aqq (How far
is the ravine!) and qimun az-zaydni (Standing are the two Zayds).
While the first of the three sentences is straightforward, the other two
are not. The word hayht in the second example is regarded by the
medieval grammarians as ism fil representing, as the term suggests, a
special word category whose members are considered as neither nouns
nor verbs (see, e.g. Astarbd, arh III:165ff. for details). Such asm
35
A fourth type, jumla artiyya, which, as he indicates, was proposed by Zamaxar,
is rejected by Ibn Him on the ground that the conditional clause should be categorized
as jumla filiyya.
178
yishai peled
al-af l, in constructions like hayhti l-aqq, are normally described as

occupying a verb position, with the implication that the following noun
implements the function of fil. However, since hayht as an ism fil is
viewed as a special kind of noun (rather than as a kind of verb), a sentence introduced by it must, according to Ibn Hims rigid principle of
classification, be regarded as a jumla ismiyya rather than as jumla filiyya.
The special problems relating to the third case (qimun az-zaydni)
have already been discussed in section 2 above, and will not be repeated
here. It is, however, noteworthy that for Ibn Him the fact that qimun
az-zaydni is introduced by a participle (viewed by the grammarians as
a nominal element) was sufficient for classifying this sentence as jumla
ismiyya; the fact that qimun is followed by a noun in the dual form did
not require, in his eyes, any further argument or elaboration.
The jumla filiyya is illustrated by Ibn Him by six sentences introduced by a verb. Out of these, five are straightforward: he uses qma,
yaqmu, qum, duriba and zanna to demonstrate that a jumla filiyya
may be introduced by any finite verb form. Specifically, by adducing
the sentence zanantuhu qiman (I believed him to be standing), Ibn
Him makes the point that a cognitive verb may, like any other verb,
introduce a jumla filiyya. As is well known, in medieval Arab grammatical theory, from Sbawayhi onwards, cognitive verbs such as zanna
(zanna wa-axawtuhzanna and sisters) are presented as analogous
to inna and kna (and their respective sisters), in that they enter into
(yadxulna al) sentences composed of a mubtada and xabar, nullifying
in the process the abstract operator ibtid, and assuming in its stead the
function of a formal mil assigning case to both nominal constituents in
the sentence (see, e.g. Sbawayhi, Kitb I:6; Ibn Ya, arh VII:7778).
In any case, the status of zanna as a verb was never disputed. By
contrast, the verbal status of kna was a matter of controversy among
the grammarians. The vast majority of grammarians considered kna
as a semantically deficient (nqis) verb,36 in the sense that it lacks the
semantic component of action.37 As such, its only function is to signal
the time of the nominal sentence into which it enters. Significantly,
36
Unless signalling existence, in which case it is labelled kna at-tmma complete
kna, and treated as an ordinary verb.
37
This had been recognized already by Sibawayhi, although in his account (Sibawayhi, Kitb I:16) he depicted kna as analogous to daraba in terms of transitivity
(tadiya): like daraba it takes two nominal complements, one in the nominative, the
other in the accusative.
179
while acknowledging the predicative relationship between zaydan and

qiman in zanantu zaydan qiman, the grammarians analyze the two
nominal constituents as direct objects (maf l) of zanna. By contrast,
in kna zaydun qiman, zaydun and qiman are normally analyzed in
terms of a jumla ismiyya, namely as mubtada and xabar, or otherwise,
as ism kna and xabar knarespectively.
True to his rigid definitions of the three sentence types, Ibn Him
includes in his examples of jumla filiyya the sentence kna zaydun
qiman (Zayd was standing). And this already seems to run counter
to the mainstream conception of jumla filiyya. Indeed, the grammarians do not normally bring kna into their discussions of sentence types.
But given their special treatment of this verb, it is highly unlikely that
they would have classified kna zaydun qiman as jumla filiyya just by
the verbal morphology of kna.
But the most interesting for our present discussion is Ibn Hims
jumla zarfiyya. This he illustrates by the two sentences a-indaka zaydun
(Is Zayd with you?) and a-f d-dri zaydun (Is Zayd in the house?),
where the first predicatival constituent is an adverbial and a prepositional phrase, respectively. Ibn Him points out that sentences such as
these can qualify as jumla zarfiyya only
id qaddarta zaydan filan bi-z-zarfi wa-l-jrri wa-l-majrri l bi-listiqrri l-mah d
fi wa-l mubtadaan muxbaran anhu bihim
if you assume zayd to be a fil [acted upon] by the adverbial/prepositional phrase, not by a deleted [verb/participle conveying the meaning of]
istiqrr, and [only if] you do not analyze zayd as a mubtada for which
the adverbial/prepositional phrase serves as xabar (Ibn Him al-Ansr,
Mun, 492).
The significance of this passage lies in that it seems to suggest that Ibn
Hims sentence-type definitions were not as rigid as they appeared
to be when we looked at his definitions and illustrations of the jumla
ismiyya and the jumla filiyya. It now turns out that for him, the predicative constituent that comes first in the sentence is not in itself the only
criterion for determining the type of sentence. Rather, for a predicatival constituent to qualify as sentence-type identifier it must act as mil
upon the second predicatival constituent.
Another important point to note is that in both of Ibn Hims examples the adverbial/prepositional phrase is preceded by the interrogative
particle a-. This may be taken to suggest that by the time of Ibn Him
the principle of itimd, which, as we have seen (section 2 above), can be
180
yishai peled
traced back to Sbawayhi, had already been firmly established in medieval Arab grammatical thought. When Ibn Him states that a sentence
can only qualify as a jumla zarfiyya if the second constituent functions
as fil to the first, illustrating this with examples displaying the interrogative particle a- preceding the first predicative constituent, one is
bound to conclude that for him a jumla zarfiyya is a sentence whose first
predicatival constituent is an adverbial/prepositional phrase, where that
phrase acts as a verb, thus assigning raf to the following constituent on
the strength of the principle of itimd.
Ibn Him then goes on to make a critical remark directed at
Zamaxar. He indicates that Zamaxar exemplified jumla zarfiyya by
the phrase f d-dr in zaydun f d-dr (cf. Ibn Ya, arh I:88). This position, he argues, is based:
al anna l-istiqrra l-muqaddara filun l ismun wa-al annahu h udifa
wah dahu wa-ntaqala d-damru il z-zarfi bada an amila fhi
on [the assumption] that [the underlying word conveying] istiqrr is a
verb, not a noun, and that that verb was deleted alone while the pronoun
implicit in it moved to the adverbial phrase, after [the verb] had exercised
amal upon the adverbial (Ibn Him al-Ansr, Mun, 492).
Recall that zaydun f d-dr was described by Fris as representing a

sentence type in its own right on the ground that, in his view, neither a
verb nor a noun could be posited as a linking element between zayd and
f d-dr. Here Zamaxar is quoted as elevating f d-dr to the status of
a clausal xabar (jumla zarfiyya). This is done, so the argument goes, by
positing an underlying verb that is deleted while the pronoun implicit
in it is transferred to the adverbial, the verb having exercised amal upon
that adverbial. In other words, the clausal status of f d-dr stems from
the pronoun it receives from the deleted verb istaqarra. What we see
here is, indeed, another attempt to account for the predicatival relationship between zaydun and f d-dr, two non-coreferential elements, as
well as for the irb of the xabar. However, this attempt is based on the
istaqarra hypothesis (3.2 above), and that is precisely the reason why it
is rejected by Ibn Him. Zamaxars analysis is incompatible with Ibn
Hims conception of jumla zarfiyya. For Ibn Him, once an underlying verb is assumed, the clause should be regarded as a jumla filiyya;
exactly as, when one posits a xabar-mubtada relationship between f
d-dr and zayd, the sentence must be considered as jumla ismiyya. This
will be further clarified below.
In 3.3 we pointed to Friss reference to Axfa, who had attributed
to the adverbial/prepositional phrase the function of operator assigning
181
raf to the nominal following it. He did not, however, refer to that nominal explicitly as fil. The term muh addat anhu which he used, signals
in medieval Arabic grammatical literature the subject of the sentence,
whether a fil or a mubtada. Indeed, the specific grammatical status of
the nominatival constituent, as determined by the mil assigning it the
raf case, never ceased to be a topic of debate among the grammarians.
Yet one thing emerges quite clearly. The analysis of zaydun as fil in
both f d-dri zaydun and qimun zaydun, normally attributed to the
Kfans and Axfa (see section 2 above), always comes up when these
two constructions are discussed. It was never abandoned. However, of
these two constructions, it is only fh/f d-dri zaydun that was considered, albeit by a small number of grammarians, as a sentence type in
its own right. The reason for this should by now be clear. The opening
predicative constituent in each of the three sentence types was regarded
as an operator (mil) assigning case to the following constituent(s): sentence types were unmistakably correlated with amal types. And since
the participle (qimun) could not be viewed as other than a verbal or
a nominal element, it could not be regarded as introducing a sentence
type in its own right. By the same token, a sentence such as zaydun f ddr, introduced as it is by a nominatival noun, could only be defined as
jumla ismiyya. The concept jumla zarfiyya was by and large associated
with cases where a zarf could be claimed to be a mil assigning raf to
the nominal following it. As we will see in the next sub-section, it was
Ibn Him, an eminent proponent of the tripartite division, who also
appreciated and spelled out the problems arising from the actual notion
of sentence types in Arabic, whether two or three.
5.3
Problems
As we have seen throughout, the problems the grammarians encountered in categorizing Arabic sentences stemmed from the fact that their
conception of sentence types was deeply embedded in the theory of
amal. This is manifested also in the way these problems are illustrated
by Ibn Him (Mun, 493497). He offers an illuminating discussion
of ten cases where a sentence can be construed as either a jumla filiyya
or a jumla ismiyya, or, otherwise, raise a controversy among grammarians as to the right categorization. Significantly, no case is cited as an
unambiguous jumla zarfiyya. Since the basic arguments recur throughout his discussion, I will review only four of his examples that, I believe,
well illustrate the problematic aspects of the traditional categorization
of sentence types.
182
yishai peled
Let us start with Ibn Hims fourh example md sanata (What have
you done?). He points out that this sentence may be paraphrased as
either m llad sanatahu, or as ayya ayin sanata. Since llad sanatahu
is a nominalized constituent, the sentence, according to the first paraphrase, must be categorized as a jumla ismiyya. Ibn Him indicates that
the first constituent m is analyzed as a fronted xabar by Axfa, and
as mubtada by Sbawayhi. By contrast, the proponents of the second
paraphrase, ayya ayin sanata, would categorize the same sentence as a
jumla filiyya, analyzing ayya ayin as a fronted direct object. (And see,
further, Ibn Hims discussion of the sentence md sanatahu.)
Ibn Hims sixth example reads qm axawka (Your two brothers
stood up). This sentence is presented by him as acceptable, subject to
specific types of analysis. (To what extent this construction was in actual
use in medieval Arabic is immaterial for the present discussion). First,
the sentence could be categorized as jumla filiyya if (1) the ending - in
qm is interpreted as a dual-marking particle (h arf tatniya), much as
the -t in qmat Hindun is analyzed as a feminine marker (and not as a
pronoun); or alternatively if (2) the ending - is interpreted nominally
and the following axawka is analyzed as apposition (badal) to it. Second, qm axawka may be categorized as jumla ismiyya with a fronted
xabar (with the ending - interpreted nominally and axawka analyzed
as a postposed mubtada). Note that Ibn Him does not mention the
possibility of analyzing qm axawka as a jumla filiyya with axawka
functioning as fil to qm (luat akaln l-bart see above, section
2, n. 6).
The seventh example presented by Ibn Him is nima r-rajulu zaydun
(What a nice man is Zayd). This sentence, he explains, may be viewed
as an inverted jumla ismiyya, with nima r-rajulu functioning as a preposed xabar to zayd. Under an alternative analysis, however, zaydun
could function as xabar to a deleted mubtada. Ibn Him argues that
under the latter analysis, nima r-rajulu zaydun consists of two asyndetically coordinated clauses, the first one (nima r-rajulu) verbal, and the
second nominal.
But perhaps the most interesting is Ibn Hims second example,
where he makes the following statement regarding a-f d-dri zaydun
and a-indaka amrun:
fa-inn in qaddarn l-marfa mubtadaan aw marfan bi-mubtadain
mah d
fin taqdruhu kinun aw mustaqirrun fa-l-jumlatu ismiyyatun dtu
xabarin f l-l wa-dtu filin munin an-i l-xabari f t-tniyati wa-in
qaddarnhu filan bi-staqarra fa-filiyyatun aw bi-z-zarfi fa-zarfiyyatun
183
if we analyze the nominatival constituent as mubtada, or otherwise, as

a nominal assigned the raf case by a deleted mubtada such as kinun or
mustaqirrun, then the sentence should be considered as nominal, with a
xabar under the first analysis, or with a fil replacing the xabar under the
second. If, however, we analyze it [i.e. zaydun or amrun] as fil of [an
underlying] istaqarra, then the sentence is verbal; if [the operator assigning raf to the fil] is the adverbial, then the sentence should be considered as a jumla zarfiyya (Ibn Him al-Ansr, Mun, 494).
Here Ibn Him offers four ways for analyzing sentences such as a-f
d-dri zaydun, correlating each analysis with a different sentence type.
These are the four options as presented in the above passage, in Ibn
Hims order:
1. zaydun could be analyzed as mubtada. This would imply that the
sentence is a jumla ismiyya, with the adverbial/prepositional phrase
implementing the function of (a preposed) xabar.
2. We could posit an underlying mubtada, such as kinun or mustaqirrun, assigning the raf case to zaydun. In this case zaydun would
implement the function of fil replacing (munin an) the xabar. The
sentence under such an analysis would be regarded, according to Ibn
Him, as jumla ismiyya.
3. zaydun could be analyzed as fil assigned the raf case by the underlying verb istaqarra. In this case the sentence would be considered as
jumla filiyya.
4. If, however, we analyze zaydun as a fil receiving its raf case from the
preceding adverbial/prepositional phrase, then the sentence should
be regarded as jumla zarfiyya.
In Ibn Hims view, then, a fil is not necessarily preceded by a finite
verb. But it is only when the operator assigning raf to the fil is a finite
verb (whether explicit or underlying) that the sentence may be conceived of as jumla filiyya. When the raf assigner is a participle (whether
explicit or underlying) or an adverbial/prepositional phrase, the sentence must be conceived of as a jumla ismiyya in the first case, and as a
jumla zarfiyya in the second, even though the actual use of the term fil
suggests that the participle and the adverbial / prepositional phrase in
such cases behave analogously to a verb.
As can be seen, Ibn Him presents the four options without any
attempt to defend his categorization. His analyses are consistent with
his definitions of the three sentence types (see 5.2 above), and manifestly reflect the controversies relating to the constructions in question.
184
yishai peled
The proponents of the first option would presumably regard a-f d-dri
zaydun as the inverted version of a-zaydun f d-dr. The occurrence of
the interrogative a- in this case is irrelevant, as is the case also under
the third analysis, where the sentence is presented as an unmistakable
jumla filiyya. Indeed, positing an underlying verb like istaqarra in order
to account for the raf case of zaydun in sentences of this kind was common practice among the grammarians, as we saw in 3.2. What is really
remarkable in Ibn Hims third analysis is that it leads to the important
conclusion that under a certain analysis a sentence such as a-f d-dri
zaydun could be conceived of as jumla filiyya.
Under the second analysis, zaydun implements the function of fil
following a deleted mubtada, thus occupying a xabar position. The
adverbial/prepositional phrase, under this as well as under the third
analysis (see above), would be analyzed as an adjunct. Obviously, the
second analysis is reminiscent of the analysis of (a-)qimun zaydun
into a mubtada followed by a fil replacing the xabar, as we saw in section 2. Note that, unlike Sbawayhi (Kitb I:239; and cf. 3.1 above), Ibn
Him does not view the prepositional phrase as occupying a mubtada
position. Rather, the mubtada in this case is an underlying participle.
Here, at any rate, a sentence whose subject is labeled fil is categorized
as jumla ismiyya.
As we have already indicated, the fourth analysis is consistent with
Ibn Hims theory of three sentence types, each defined by the predicative constituent introducing the sentence (and acting as mil upon
the second constituent). But we have already seen (3.3 above) that it
remains unclear how an adverbial/prepositional phrase can function
as a verb assigning raf to a following nominal constituent. Following
Sbawayhi, it was often argued that such a phrase may act analogously to
a verb when preceded by a supporting element such as the interrogative particle a- (itimdcf. sections 2, 4.1 above). But does that in itself
warrant categorizing the construction a-f d-dri zaydun as representing a sentence type in its own right? The concept of jumla zarfiyya does
not seem to have been seriously discussed in the writings of the medieval Arab grammarians. Evidently, the vast majority found it difficult to
fit the concept of jumla zarfiyya into their theory of amal. Indeed, this is
manifested even in Ibn Hims position, which does not present a-f ddri zaydun as a straightforward jumla zarfiyya. Rather, it makes it clear that
the actual identification of a sentence as jumla zarfiyya is essentially dependent upon conceiving the adverbial/prepositional phrase as
a mil assigning raf to the following nominal. The other two types,
185
in contrast, could be determined straightforwardly, since both the verb

and the ibtid were established awmil in the medieval theory of amal.
Hence the grammarians adherence to the binary system of jumla filiyya
and jumla ismiyya.38
6. Summary
Since the grammarians theory of sentence types grew out of, and has
always been closely related to, their theory of amal, it is not surprising
that in elaborate discussions of Arabic sentence types, particularly those
of Fris and Ibn Him, problems relating to the categorization of certain constructions were couched in terms of case assignment (amal).
The basic types of jumla filiyya and jumla ismiyya are shown throughout to represent two types of amal: verbal tadiya and ibtid. Sentences
such as qimun zaydun and f d-dri zaydun/rajulun are shown to be
problematical in terms of amal. With regard to qimun zaydun, we have
seen that many grammarians advocated the rather awkward analysis of
mubtada+fil sadda masadd al-xabar. This was designed to deal with
the essentially nominal nature of the participle occurring sentence-initially, as well as with its verb-like behavior in this particular case. Apart
from Ibn Him who regarded this construction as an example of jumla
ismiyya, the proponents of the above analysis did not commit themselves
to any clear-cut categorization of this particular structure. Regarding f
d-dri zaydun/rajulun, the very fact that this construction displays an
adverbial/prepositional predicative constituent in sentence-initial position, gave rise to the hypothesis that it represents a sentence type in its
own right, a jumla zarfiyya. And it comes as no surprise that this was
associated with the hypothesis that in such cases it is the adverbial/prepositional phrase that assigns raf to the nominal constituent following it.
Obviously, this hypothesis and the long established istaqarra/mustaqirrun hypothesis were mutually exclusive. In 3.3 we saw Friss attempt to
refute the istaqarra/mustaqirrun hypothesis, arguing from the theory of
amal. This was his line of defending a tripartite sentence-type system.
38
For a modern study advocating a three-type division, see Kouloughli (2002, 2124)
who argues that sentences such as f d-dri rajulun/zaydun (referred to by him as locative sentences) should be viewd as representing a sentence type in its on right, since
they exhibit a number of syntactic and semantic properties not shared by regular topiccomment sentences.
186
yishai peled
For him, indeed, claiming that fh zaydun represented a sentence type

in its own right was tantamount to presenting fh as the assigner of raf
to zaydun. But such an argument could never be accepted by the majority of grammarians, because it was considered a major violation of a central principle of the theory of amal. The vast majority of grammarians
did find the istaqarra/mustaqirrun hypothesis a convenient tool for fitting such constructions as zaydun fh and fh zaydun into their theory
of amal. The question of sentence type was apparently secondary. Once
they established the status of mubtada to zaydun in both cases, they
could argue that both represent a jumla ismiyya, irrespective of whether
the underlying element linking the adverbial/prepositional phrase to
the mubtada is a verb or a participle. Fh zaydun was thus conceived
of as the inverted version of zaydun fh. To Ibn Him, as we have seen,
this was unacceptable. For, if underlying fh zaydun is the structure
yastaqirru fh zaydun, it follows that zaydun is assigned the raf case
by the underlying verb occupying sentence-initial position. And Ibn
Him, indeed, drew the conclusion following from that assumption,
namely that under the above analysis fh zaydun must be categorized
as jumla filiyya. But this position of Ibn Hims is clearly exceptional
in the medieval grammatical literature. The majority of grammarians,
conforming to the theory of amal, did posit istaqarra as the underlying
case assigner to zaydun, but they never went as far as categorizing fh
zaydun as jumla filiyya. Rather, fh zaydun has always been regarded
an ibtid construction, clearly associated with jumla ismiyya.
7. References
7.1 Primary sources
Astarbd, arh = Rad ad-Dn Muhammad b. al-H asan al-Astarbd, arh Kfiyat
Ibn al-H jib. Emil Bad Yaqb, ed. Beirut: Dr al-Kutub al-Ilmiyya, 1998.
Fris, Askariyya = Ab Al l-Fris, al-Masil al-askariyyah. Muhammad a-ti r
Ahmad Muhammad Ahmad, ed. Cairo: Matb aat al-Madan, 1982.
Farr, Man = Ab Zakariyy Yahy b. Ziyd al-Farr, Man l-Qurn. Ahmad
Ysuf Najt and Muhammad Al an-Najjr, eds. Cairo: ad-Dr al-Misriyya li-t-Talf
wa-t-Tarjama, 19551972.
Ibn Ab r-Rab, Bast = Ubaydallh b. Ahmad b. Ubaydallh Ibn Ab r-Rab, al-Bast f
arh jumal az-Zajjj. Ayyd b. d at-Tabt, ed. Beirut: Dr al-arb al-Islm, 1986.
Ibn al-Anbr, Asrr = Ab l-Barakt Abd ar-Rahmn b. Muhammd b. Ab Sad alAnbr, Kitb Asrr al-Arabiyyah. Muhammad Bahjat al-Bayt r, ed. Damascus:
Matbt al-Majma al-Ilm al-Arab bi-Dimaq, 1957.
Ibn al-Anbr, Insf = Ab l-Barakt Abd ar-Rahmn b. Muhammd b. Ab Sad alAnbr, Kitb al-Insf f masil al-xilf bayn an-nahwiyyn al-basriyyn wa-l-kfiyyn.
Muhammad Muh ad-Dn Abd al-H amd, ed. Beirut: Dr al-Fikr, n.d.
187
Ibn Aql, arh = Bah ad-Dn Abdallh Ibn Aql, arh Ibn Aql al alfiyyat Ibn Mlik.
Muhammad Muhy ad-Dn Abd al-H amd, ed. n.p: Dr Sab, n.d.
Ibn Him, Mun = Jaml ad-Dn Ibn Him al-Ansr, Mun l-labb an kutub alarb. Mzin al-Mubrak and Muhammad Al H amdallh, eds. Beirut: Dr al-Fikr,
1985.
Ibn Jinn, Sirr = Ab l-Fath Utmn Ibn Jinn, Sirr sinat al-irb. H asan Hindw, ed.
Damascus: Dr al-Qalam, 1985.
Ibn Jinn, Xasis = Ab l-Fath Utmn Ibn Jinn, al-Xasis. Muhammad Al an-Najjr,
ed. Cairo: al-Haya l-Misriyya l-mma li-l-Kitb, 19861988.
Ibn as-Sarrj, Usl = Ab Bakr Muhammad b. Sahl Ibn as-Sarrj, Al-Usl f n-nahw.
Abd al-H usayn al-Fatl, ed. Beirut: Muassasat ar-Risla, 1987.
Ibn Usfr, arh = Al b. Mumin b. Muhammad b. Al Ibn Usfr, arh jumal azZajjj. Shib Ab Janh, ed. Mosul: Ihy at-Turt al-Islm, 19801982.
Ibn Ya, arh = Muwaffaq ad-Dn Ya b. Al Ibn Ya, arh al-Mufassa l. Cairo:
Maktabat al-Mutanabb, n.d.
Jurjn, Muqtasid = Abd al-Qhir al-Jurjn, Kitb al-muqtasid f arh al-dh . Kzim
Bahr al-Marjn, ed. Baghdad: Dr ar-Rad li-n-Nar, 1982.
Muji, arh = Ab l-H asan Al b. Faddl al-Muji, arh uyn al-irb. Abd alFatth Salm, ed. Cairo: Dr al-Marif, 1988.
n Sbawayhi, Al-Kitb. Hartwig Derenbourg,
ed. Hildesheim and New York: G. Olms, 1970.
Zajjj, Jumal = Ab l-Qsim Abd ar-Rahmn b. Ishq az-Zajjj, Kitb al-jumal f nnahw. Al Tawfq al-H amad, ed. Beirut: Muassasat ar-Risla, Dr al-Amal, 1988.
7.2
Secondary sources
Badawi, El-Said M. 2000. Ray f man l-irb f fush t-turt: H lat al-jumla lismiyya. Diversity in Language: Contrastive Studies in Arabic and English Theoretical
and Applied Linguistics, Z. M. Ibrahim, S. T. Aydelott and N. Kassabgy, eds., 120.
Cairo: The American University in Cairo Press.
Carter, Michael G. 1973. Sarf et H ilf: Contribution lhistoire de la grammaire arabe.
Arabica 203, 292304.
. (ed.). 1981. Arab Linguistics: An introductory classical text with translation and
notes (irbns Nr al-sajiyyah). Amsterdam: John Benjamins.
Goldenberg, Gideon. 1988. Subject and Predicate in Arab Grammatical Tradition.
Zeitschrift der Deutschen Morgenlndischen Gesellschaft 1381, 3973.
. 2002. Two Types of Phrase Adjectivization. Sprich doch mit deinen Knechten
aramisch, wir verstehen es!: 60 Beitrge zur Semitistik, Festschrift fr Otto Jastrow
zum 60. Geburtstag, W. Arnold and H. Bobzin, eds. 193208. Wiesbaden: Harrassowitz.
Kouloughli, D. E. 2002. On locative sentences in Arabic. Zeitschrift fr arabische Linguistik 41, 726.
Levin, Aryeh. 1985. The distinction between nominal and verbal sentences according
to the Arab grammarians. Zeitschrift fr arabische Linguistik 15, 118127.
. 1987. The views of the Arab grammarians on the classification and syntactic function of prepositions. Jerusalem Studies in Arabic and Islam 10, 342367.
. 1989. What is meant by akaln l-bartu? Jerusalem Studies in Arabic and
Islam 12, 4065.
Owens, Jonathan. 1988. The Foundations of Grammar: An Introduction to Medieval Arabic Grammatical Theory. Amsterdam: John Benjamins.
. 1989. The Syntactic Basis of Arabic Word Classification. Arabica 36, 211234.
Peled, Yishai. 1992a. Amal and Ibtid in Medieval Arabic Grammatical Tradition.
Abr-Nahrain 30, 146171.
. 1992b. Cataphora and Taqdr in medieval Arabic grammatical theory. Jerusalem
Studies in Arabic and Islam 15, 94112.
188
yishai peled
Talmon, Rafael. 1993. Two Early non-Sbawaihian Views of amal in Kernel-Sentences. Zeitschrift fr arabische Linguistik 25, 278288.
. 1997. Arabic Grammar in its Formative Age: Kitb al-Ayn and its attribution to
Xall b. Ah mad. Leiden: E. J. Brill.
ARABIC AVANT LA LETTRE. DIVINE, PROPHETIC, AND

HEROIC ARABIC
Stefan Wild
Bonn
1. Introduction: the Qurn

The history of the Arabic language is indelibly marked by the fact that
the Qurn has made Arabic a prophetic language with its own holy
book and a worldwide appeal. Arabic had been a mantic language in
pre-Islamic times. When the poet recited verses or when the soothsayer
uttered his sayings, reciter and listener were sure that behind his voice
there was another voice. This voice really speaking was that of a higher
power.
With the Qurn, the mantic voice behind the voice of the Prophet
Muhammad became in the believers ear the voice of the one and only
God. Muslim dogma and the consensus of the unlearned considered
the Qurn to be the direct, undiluted Arabic word of God. The status of
Classical Arabic, the standardization of Arabic including the development of Modern Standard Arabic, the diglossia Standard Arabic versus
Arabic dialects, the nature of the Arabic vocabulary, Arabic orthography,
Arabic style and vocabularyall are unthinkable without the Qurnic
fact. The history of the Arabic language down to our times cannot be
written without constant reference to the Qurn.
Conversely, the Qurn is deeply marked by its Arabness. The Qurn
is the first literary document in Arabic. It is also the first Arabic document to mention the Arabic language. In contradistinction to the attitude of the Jewish Bible and the Christian New Testament toward their
own linguistic forms, the language of the Qurn is an important topic
of Qurnic self-reflection. While neither the Bible nor the New Testament refer to their Hebrew (or Aramaic) and Greek literary forms, the
Arabic character of the Qurn is in its self-view a cornerstone of its
divine quality.
The adjective arab in the Qurn always refers to the language, never
to a tribe or a social class. The term as an ethnic designation is also
190
stefan wild
very rare in Jhiliyya poetry.1 In the Qurn, arab refers always to the
language of the holy text. It occurs eleven times in the Qurn, and only
in Suras traditionally dated to the Middle and late Meccan period. In
six passages, the adjective arab is a qualification of the word Qurn,
a word meaning primarily recitation, reading aloud . . . (Sura 12:12;
20:113; 39: 2728; 41:14; 42:7; 43:14). The conjunction lisn arab
Arabic tongue occurs three times in the Qurn (Sura 16:103; 26:195,
46:12), and is used to describe the language of the Qurn. The conjunction h ukm arab (Sura 13:37) an Arabic judgment also refers to the
holy text, and in 41:44, in which the possibility of a Qurn ajam a
non-Arabic Qurn is dismissed, the adjective arab again refers to the
holy text. The Arabic quality of Qurnic revelation could scarcely be
more solidly established.
On the other hand, the Qurn seems indifferent to the linguistic
shape of preceding revelations. In the Qurn, the only language used is
Arabic. Arabic is also the only language mentioned by name. The Qurn
does not specify in which languages Nh, Ibrhm, Isml, Ms, s or
other prophets and messengers spoke to their peoples or in which languages their holy books might have been.
God speaks Arabic to Adam and his wife, Satan whispers in Arabic
(Sura 20:120), the angels and the jinn speak Arabic (Sura 72:115),
Moses addresses the Pharaoh, Joseph addresses the Egyptian ministers
wife in Arabic, Jesus speaks Arabic from his cradle, D l-Qarnayn and
the People of the Cavethey all use Arabic. Every single soul is made to
speak Arabic at the Day of Judgment, animals like the ant (Sura 27:18) or
the hoopoe (Sura 27:22), even inanimate entities like Hell (Sura 50:30)
speak Arabic. Everybody and everything that speaks in the Qurn must
necessarily speak Arabic, because Arabic is the only language used
throughout the Qurn.
But the intention of the text is in no way to convey that all mankind
throughout history shared and will share the same language. I do not
know of any exegete who concluded from the Qurnic accounts that
the language used between Ms and the Egyptian Pharaoh was Arabic
or that the language used between the Egyptian notables wife and her
lady-friends must have been Arabic, or that the Messiah spoke in Arabicjust because the Qurn reproduces their words in Arabic. It is a
different matter for Adam and Isml (see below).
Cf. Agha and Khalidi: Poetry and Identity 70.
divine, prophetic, and heroic arabic
191
On the other hand, the Qurn does mention the existence of languages other than Arabic and emphasizes that the plurality and variety
of human languages is a sign of divine grace. The divine creation of different languages is as important as the creation of heaven and earth:
wa-min ytih xalqu s-samawti wa-l-ardi wa-xtilfu alsinatikum
wa-alwnikum nna f dlika la-ytin li-l-limn
And of His signs is the creation of heavens and earth and the variety of
your tongues and hues. Surely in this are signs for people who know (Sura
30:22).
Even as God created man and woman, heaven and earth, he created
different colors (alwn, or: kinds of human skin) and different human
languages (cf. the enumeration in Sura 30:2025). The existence of different languages is one of Gods signs for those who know. Unless the
word ya is here taken to mean a portent of warning, such a view seems
difficult to reconcile with the myth of the Babylonian tower (Gen 11:
19), according to which the origin of a multitude of human languages
is divine punishment. Sura 30:22 may even be an inter-textual stab at the
narrative of the Babylonian Tower. There were, however, extra-Qurnic
traditions that preserved the motif that the difference between human
languages was due to an act of divine wrath (p. 196).
A second important element in the Qurnic linguistic outlook flows
forth from this esteem of other languages. In the course of history, God
sent each messenger (rasl) to a specific people (qawm), and this messenger brought the divine message to that people in its language.
And We have sent no messenger save with the tongue of his people that he
might make all clear to them
wa-m arsaln min raslin ill bi-lisni qawmih li-yubayyina lahum
(Sura 14:4).
The primary raison dtre of the Qurn is that the Prophet Muhammads
message was in Arabic. Whereas the other prophets and messengers had
been sent earlier with a message in the languages of their peoples, who
did not speak Arabic, Muhammad is sent to the Arabs. The Arabic language vouchsafes the understanding of the Arabic-speaking audience.
For a great part of the Qurn, the fact that this revelation was in Arabic
was the most important difference between Muslim revelation and all
previous revelation.
These Qurnic elements influenced a theological-linguistic scenario
that gave rise to a particular image of linguistic history. The claim that
192
stefan wild
Arabic as the language of the Qurn had become a divine language,

led to the construction of a religious pre-history of Arabic. There were
three competing strands of knowledge that were interwoven to form a
comprehensive imaginaire of the history of Arabic and by implication of
human languages in general:
1. the philosophical question of the origin of language in general;
2. exegesis of the Qurn, including the Qisas al-anbiy literature, i.e.,
reports on earlier prophets, which depended heavily on Biblical and
Aggadic material; and
3. historical traditions about pre-Islamic history, including the genealogies of the Arab tribes.
One of the main issues was the origin of Arabic, an issue often framed as
the question Who was the first human being to speak Arabic? MuslimArab scholarship tried sometimes to disentangle, sometimes to combine and often to harmonize the heterogeneous strands of philosophical
speculation, exegetical H adt, and genealogical traditions. In many cases,
however, contradictory reports were just left standing side by side. An
extensive and useful overview of much of the Arabic material was given
by as-Suyt in his Muzhir.2 The following remarks will outline the main
lines of this colorful and often contradictory linguistic yarn.
2. Adams Arabic and Nhs Syriac

The basic divide between Arab philosophers and theologians concerned
with the origin of language was whether it rested on human convention
(isti lh , tawdu) or on a divine act of revelation (ilhm, wahy, tawqf ).3
The Platonic controversy whether human language was what it was physei or thesei, by nature or by imposition, here took a new form. A further
problem was whether, theologically speaking, Arabic had a special linguistic status that set it apart from all other languages.
2
Andrzej Czapkiewicz, The Views of the Medieval Arab Philologists on Language and
its Origin in the Light of Al-Suyutis Al-Muzhir, Krakow 1989. Czapkieviczs translations
are sometimes hard to understand.
3
For the difference between tawqf and ilhm, cf. J. van Ess, Theologie und Gesellschaft IV 325.
193
An important part of tradition seems to be in general agreement that

Adam, who is in Islam a prophet, was the first human being to speak Arabic. The revelationists based their view mainly on the early exegesis of
Qurn 2:31 wa-allama dama l-asma kullah and He (God) taught
Adam all names. Weiss explains: According to at-Tabar, the majority
of early exegetes, including the noted Companion and tafsr-authority
Ibn Abbs, interpreted this verse as meaning that God taught Adam the
Arabic names of all existing things. This meant that Adams language
was revealed by divine teaching. It also clearly implied that God taught
Adam language in its entirety.4 According to Ibn Jinn (d. 392/1002),
most Mutazilites, on the other hand, taught that all languages including
Arabic rested on convention, not on revelation.5
Even those who accepted the revelationist theory could argue that
Sura 2:31 meant that Adam was taught the names of all things created in
all languages, in Arabic, of course, but also in Persian, Syriac, Hebrew,
Greek, and all other languages as Ibn Jinn explains. Adam and his offspring spoke all these languages. But in the course of time, Adams offspring spread over the world and each of them stuck to one of these
languages until this language became his main language and other
languages eluded him because of his lack of contact with them (qad
fussira hd bi-an qla inna llha subh nah allama dama asma
jami l-maxlqti bi-jami l-lugti l-arabiyyati wa-l-frisiyyati wa-lsuryniyyati wa-l-ibrniyyati wa-r-rmiyyati wa-gayri dlika min siri
l-lugti fa-kna damu wa-waladuh yatakallamna bih tumma inna
waladah tafarraq f d-duny wa-aliqa kullun minhum bi-lugatin min
tilka l-lugti fa-galabat alayhi wa-dmahalla anhu m siwh li-budi
ahdihim bih, Ibn Jinn, Xasis I 41). This model safeguarded the priority of Arabic; Arabic as the language of the divine word must be prior
to all other languages, whether their origin was revealed or due to convention. But according to some, all other languages were also revealed
to Adam by God. Others with a different timetable thought that different linguistic communities emerged only after the Flood. The idea that
Adam had known all human languages found a parallel in the belief of
some Shiites in the 9th century that the Prophet Muhammad had known
Weiss, Muslim Discussions 37.

aktaru ahli n-nazari al anna asla l-lugti innama huwa tawdu wa-sti lh l
wahy wa-tawqf (Al-Xasis I 4041).
5
194
stefan wild
all languages.6 The plurality of human languages is sometimes seen as

due to a benevolent divine intervention: He (God) shaped every nation
conforming to a language in which He made them speak and which
He made easy for them.7 God had made His Arabic revelation easy
for the Arabs (e.g. Sura 54:17); in a similar way, earlier revelations had
been made in other languages in order to make them easy for their
listeners.
The most widely accepted historical model that explained the existence of languages other than Arabic was built on the supposition that
Arabic had been at a certain time the universal language. The idea of
a universal common language can also be found in Gen 11:1, where,
however, this language does not have a name. For most Arabic scholars,
Arabic was the earliest existing language.8 But this universality had at
some point in history come to an end, and decadence, corruption, and
confusion ( fasd, tah rf, tabalbul) had set in. This statement was hard
to reconcile with the Qurnic assertion that God himself had created
the variety of human languages (Sura 30:22). The reason for the development of other languages was nevertheless frequently seen either in
a divine act of punishing Adam or mankind, or in a general confusion
(balbala), which mixed up what had been one common human language in a gradual process of corruption (tah rf ). Tah rf and tabalbul
were sometimes expressly linked to a punitive act of God, sometimes
they seem to be seen more as a general tendency of history. According
to Ibn Askirs chronicle, Arabic was Adams language in Paradise until
he disobeyed God. Then God deprived him of Arabic and he started
speaking Syriac, evidently considered a lesser language or a corrupted
form of Arabic. When Adam repented, however, God gave the Arabic
language back to him.9
A second less revelationist tradition was the following:
The first language with which Adam came down from Paradise was Arabic. When the contact with Arabic became far and distant, Arabic was
corrupted (h urrifa) and became Syriac (suryniyya) which is a Nisba
to ard sr or Suryna. This is the land of al-Jazra, where Nh and his
Van Ess, Theologie und Gesellschaft IV 324.

az-Zubayd, Tabaqt an-nahwiyyn (ed. Muhammad Ab l-Fadl Ibrhm), Cairo
1954/1373, 1 jabala kulla ummatin mina l-umami al lugatin antaqahum bih wayassara lahum.
8
lugatu l-arabi asbaqu l-lugti wujdan, as-Suyt , Muzhir I 28, 4.
9
as-Suyt, Muzhir I 30, 10.
7
195
people lived before the flood. Syriac resembled Arabic, but it was corrupted (muh arraf ), and Syriac was the language of all people in Nhs ark
except for one man, whose name was Jurhum, and his language was still
the original Arabic. When they left the ark, ram b. Sm married one of
his (Jurhums) daughters. From them the Arabic language came down on
his offspring Aws Ab d, Ubayl, Jir Ab Tamd, and Jdis. d was
given the name of Jurhum, because he was their maternal forefather. And
Syriac stayed with the offspring of Arfaxad b. Sm until it reached Yajub
b. Qaht n. He was in Yemen. There the Ban Isml settled and the Ban
Qaht n learned from them the Arabic language.10
It is interesting to see that the term tah rf, which is the usual word
denoting the falsification of the revealed scriptures by Jews and Christians, is here used to explain a fact of imagined linguistic history : the
corruption and subsequent loss of Arabic. According to this account,
Nhs language was not Arabic, but Syriac. Arabic had survived only
with Jurhum and his tribe.
Abdalmalik b. H abb (d. 238/852) developed a stemma of prophetic languages. According to him, these prophetic languages are Arabic, Syriac,
and Hebrew. All the sons of Israel spoke Hebrew, the first to speak it
was Ishq. Syriac was the language of five prophets: Idrs, Nh, Ibrhm,
Lt and Ynus. And twelve of the Prophets spoke Arabic: Adam, Sht,
Hd, Slih, Isml, uayb, al-Xidr, the three in surat Ysn i.e. the three
nameless messengers who were sent as the Companions of the City
(ash b al-qarya, sura 36:13ff.), Xlid b. Sinn al-Abs, the legendary
forerunner of the Prophet Muhammad, and the Prophet Muhammad
himself.11
Others added more and more details. According to al-Azraq, Nh
had eighty men with their families on his ark. When the ark came to a
halt on Mount Jd, Nh founded a village called Tamnn (eighty).
The next morning, they found that their tongues/languages had been
confused and that there were now eighty languages, one of which was
Arabic. They did not understand each other any more. 12 This account
offers no explanation for the emergence of these eighty languages, but
as-Suyt , Muzhir I 30,-6 according to Abdalmalik b. H abb.

Claude Gilliot and Pierre Larcher Language and Style of the Qurn. EQ 3, 109
135, especially The mythical narratives on the superiority of Arabic 118ff.
12
fa-btan qaryatan wa-sammh Tamnin fa-asbah dta yawmin wa-qad tabalbalat alsinatuhum al tamnn lughatan ih dh l-arabiyyatu (al-Azraq, Axbr Makka,
/ed. F. Wstenfeld) vol. 1, 20.
10
11
196
stefan wild
is based on the idea that Nh and his followers originally did not speak
Arabic.
In other accounts, Ibrhm is said to have spoken two languages,
Syriac and Hebrew. When he fled from the persecution by Namrd in
Kta in the land of Babylon, his language was Syriac (lisnuh suryn).
Namrd wanted to catch him and gave order that anybody speaking
Syriac should be arrested and brought before him. But when Ibrhm
crossed the Euphrates from H arrn, God changed his language and
Ibrhm miraculously started speaking Hebrew (ibrn). He escaped
because his persecutors did not know this language.13
A further version closer to the narrative of the Babylonian Tower is
reported by at- Tabar: Namrd has a high building (sarh ) made, until it
reaches the sky. God destroys the building and on that day the languages
of people become confused from fright. They then speak in seventythree languages. This is why the place was called Bbil. The language of
mankind before that was Syriac.14 A different kind of etiological explanation is linked to Yarub (see below section 4). The etymological connection between balbala and the name Bbil is often invoked and finds
its way into Arabic lexicography.15 This connection between the Arabic
word for confusion (balbala) and the name of Bbil is prefigured in the
Bible, in which the name of the city of Babel is linked to the Hebrew
verb balal, also meaning to confuse (Gen 11:7). The number of existing languages is given in different ways. According to al-Masd, there
were seventy-two languages divided under Nhs sons: the descendants
of Sm spoke nineteen languages, the descendants of H m seventeen,
and those of Yfit thirty six.16
As far as I can see, nobody in pre-modern times ever claimed that the
prophet Ibrhm spoke Arabicalthough his unequalled importance
for the link between earlier monotheistic religions and Islam, especially
through his role in building the Kaba, would have made such a claim
attractive. Abraham is called a Muslim in the Qurn (h anf muslim
13
Ibn Sad, Tabaqt ed. Eugen Mittwoch, Leiden 1905, I:1 21, 14; a similar tradition
in Ibn Mutarrif at -Taraf, Qisas al-anbiy no. 124.
14
fa-tabalbalat alsunu n-nsi min yawmaidin mina l-fazai fa-takallam bi-taltati
wa-sabna lisnan fa-li-dlika summiyat Bbil wa-innam kna lisnu n-nsi qabla
dlika s-suryniyyata., At-Tabar, Trx I 322.
15
Lisn al-Arab s.v. bll: summiyat ardu Bbil li-anna llha h na arda an yuxlifa
bayna alsinati ban dama baatha r-rh an fa-h aarahum min kulli ufuqin il Bbil fabalbala llhu bih alsinatahum tumma farraqahum tilka r-rh u fi l-bild.
16
al-Masd, Murj ad-dahab, ed. Barbier de Meynard I 78.
197
sura 3:67)but never an Arab. This claim was put forward only by zealous Arab nationalists. When in 1999 Pope John Paul II announced his
plan to visit Ur in Iraq, members of the ruling Bat -party criticized that
the Pope in his announcement had not spoken of Ibrhm as an Arab.
They argued: Ibrhm was born in the land of the Arabs and he lived
in Iraq, in the Sumerian city of Ur. When he was chased out of his place
of birth because of his monotheist creed, Ibrhm started his combative
journey through the lands of the Arabs, H arrn, Palestine, Egypt, and
Mecca . . . Combative was a favorite attribute of praise in the Bat-party.
The Bats do not expressly identify Arabic as Ibrhms languageeven
though the reader is forced to conclude that the language that the combative Arab Ibrhm spoke must have been Arabic.17 Such a linguistic
myth could be elaborated. There is frequently a tendency in modern
popular Arab discourse to call Arabic what is elsewhere called Semitic,
in order to extol the importance of Arabic. The Canaanite tribes preceding the Israelite settlers are then called Arab Canaanite tribes, and the
variant of Akkadian to be found in the Ebla texts is called Arabic.18 For
the claim that the Aramaic of the Nabatean inscriptions is really Arabic, see below.
3. Ismls Arabic
Next to Adam, the most important figure with whom the introduction of Arabic is firmly connected is Ibrhms son Isml. In one of the
most important foundational Muslim narratives, God orders Ibrhm
to migrate to Mecca with young Isml and the latters mother Hagar.
Ibrhm builds the Kaba together with Isml. There they meet members of the (Arabic-speaking) Jurhum tribe. Isml grows up with their
children, learns how to shoot (the bow) and to speak in their language,
and he takes a Jurhum-wife.19 Another report says, without reference to
where and how, Ismail learned Arabic: Sarah gave Hajar to Ibrhm, he
slept with her and she bore him Isml, who was Ibrhms eldest son.
His name used to be Iml which was later arabicized (wa-kna smuh
17
Amatzia Baram, Der moderne Irak, die Baath-Partei und der Antisemitismus in
Jahrbuch fr Antisemitismusforschung 12 (2003) 99119 p. 114.
18
Welt des Islams 21 /1982, 240f.
19
Ibn Qutayba, al-Marif, ed. Tarwat Uka, Cairo 1960, 34.
198
stefan wild
mwyl fa-uriba).20 This Arabic tradition closely follows the Biblical

account (Gen 21:20)even the bow is already thereand preserves
in Iml a dialectical Hebrew form of the classical Hebrew Yiml.
Abraham had given his son a Hebrew name, because he spoke Hebrew.
The link between Isml and the Arabic language became part of the
awil-literature. The Prophet had said: all Arabs descend from the offspring of Isml b. Ibrhm. The question who was the first to speak
Arabic could, therefore, be answered and the answer is contained in a
short didactic dialogue in which an Ab Jafar Quray b. Uqba b. Bar
asks and a Muhammad b. Al answers:
Q: Who was the first to speak Arabic?
A: The first to speak Arabic was Isml b. Ibrhm, when he was thirteen
years old.
Q: And what did people speak before that?
A: Hebrew.
Q: And what was Gods language that was sent down to his messengers
and servants?
A: Hebrew.21
Here, Adam is not even mentioned as the first recipient of Arabic. And
Gods language of pre-Arabic revelation is simply Hebrew. In awilcollections, the particular that Ishmael was the first to write in Arabic is
not too frequent.22 See below p. 200.
The report that Isml picked up Arabic from Arabic-speaking tribes
was widely disseminated, but there was a competing tradition saying
that Isml received the Arabic language by revelation on the day he was
born. The three other sons of Ibrhm stuck to their fathers (Hebrew)
language.23 According to a H adt, the Prophet Muhammad recited (the
verse) an Arabic Qurn for people who know (Sura 41:3); then he said:
Isml received this Arabic language by a great revelation (ulhima
Ismlu hd l-lisna l-arabiyya ilhman). This tradition contradicted
the view that Isml had learnt Arabic from the Jurhum tribe. In the
latter case, the tribe must have logically preceded Isml in speaking
Arabic. Another H adt mentions Ismls age: when he learned Arabic,
he was 14 years old.24
20
Ibn Sad, Tabaqt I/1 23, 9.

Ibn Sad, Tabaqt I/1 24, 16. Tottoli p. 82. Cf. Sibt Ibn al-Jawz, Mirt 310; Ibn
Askir, Trkh II 331.
22
See the comment of ibl, Mah sin al-wasil f marifat al-awil 143.
23
24
as-Suyt, Muzhir I 22, 13.
21
199
Ibn Sad transmits a report that contradicts the idea that Isml was
born with Arabic and traces Arabic only to Ismls offspring. According to this tradition, Isml never spoke Arabic, because in his filial
piety he did not deem it permissible to act differently from his father.
According to this report, the first of his offspring to speak Arabic were
the Ban Rala bint Yajub b. Yarub b. Ldan b. Jurhum b. mir b. Saba
b. Yaqtn b. bir b. likh b. Arafxaad b. Sm b. Nh.25 Nevertheless,
the fact that Isml forgot his fathers Hebrew is expressly stated in a
tradition traced to Muhammad b. Salm:26 The first one to speak Arabic
and to forget his fathers language was Isml. Muslim Arab tradition,
therefore, agrees that Ibrhms language was not Arabic, whereas there
is disagreement on whether it was Isml who was the first to speak
Arabic and, if so, when and how Isml learned it. For the tribal aspects
of such traditions, cf. below under 4. Yarubs Arabic.
The most frequently quoted account, successful as a canonical H adt,
was, however, that Isml learned Arabic from the Jurhum tribe in his
youth (wa-shabba l-ghulmu wa-taallama l-arabiyyata minhum, i.e.,
Jurhum, s. Buxr, Sah h , Anbiy 21). He marries twice, in both cases
a wife from the Jurhum tribe. This again meant, of course, that Isml
was, strictly speaking, not the first one to use Arabic, because he had
to learn it from somebody else. In some reports, the gift of the Arabic
language is mentioned next to other privileges of Isml: the traditionists report that Isml was the first one to speak Arabic, the first one to
build the h aram after his father Ibrhm, and the first one to install the
rites of pilgrimage. He was also the first to ride full-grown horses, which
before were wild and could not be ridden. Some say: Isml was the first
whose tongue God opened to speak Arabic. And when he grew up, God
gave him the Arabic bow. This report implies that God revealed Arabic
to Isml and that Isml did not have to learn it from the Jurhum.27
This report became part of Adab literature. At- Talib mentions it in his
Latif al-Marif. The first person to speak Arabic was Isml, peace be
upon him; all the Arabs came subsequently from his progeny, except for
three tribes, those of Auz, H adramawt and Taqf. He was the first to
25
26
27

as-Suyt , Muzhir I 32, -2.
al-Yaqb, Trx I 22.
200
stefan wild
ride horses, these mounts being originally wild horses which had never
been ridden before.28
There was also some concern about the fact that the long time that had
elapsed between the period in which Isml had started speaking Arabic
and the time of the Prophet Muhammad made it difficult to accept that
the Arabic language should not have changed during this interval. The
Prophet gave the answer: Ismls language had been obfuscated (darusat). But Jibrl came and made me retain the language and I retained it,29
i.e. the Qurnic message saved Arabic.
In other versions, it is not only the Arabic language whose origin is
attributed to Adam or Isml, but also the Arabic script, which is in
this case imagined as written on clay tablets: The first who installed
the Arabic script, the Syriac script, and all other scripts was Adam, 300
years before his death. He wrote them in clay and baked them. When the
earth was hit by the flood, each people received its script and used it to
write. Isml b. Ibrhm received the Arabic writing.30
Some of these conflicting mythological reports are woven together
and transformed into scholarship by modern Wahhabi scholars. Taking
such accounts as factually historical, Muhammad Musta f al-Azam is
led in a recently published book31 to a re-writing of linguistic history
on a grand scale. He asks himself the question What language did the
Nabateans speak? And the answer is:
Growing up in Makkah from his earliest childhood Isml, eldest son of
Ibrhm, was raised among the Jurhum tribe and married within them
twice. This tribe spoke Arabic, and so, undoubtedly, must have Isml.
The Jurhum Arabic probably lacked the sophistication and polish of the
Quraishi Arabic, preceding it as it did almost by two thousand years.
Ibn Ushta records a statement from Ibn Abbs, that the first person to
initiate set rules for the Arabic grammar and alphabet was none other
than Isml. Eventually, Allah commissioned Isml as a messenger and
prophet, to call his people for the worship of the one true God Allah, to
28
In: C.E. Bosworth (ed.), The Book of Curious and Entertaining Information. The
Latif al-marif of al-Thalib, Edinburgh 1968, p. 40, cf. Lammens, La cit arabe de
Taif la veille de lHgire, Beirut 1922, 5768.
29
as-Suyt , Muzhir I 35, 3.
30
as-Suyuti, Itqn IV 143 quoting Ab Bakr M. b. Abdallh b. Muhammad b. Uta
al-Isbahn (d. 360) and his Kitb al-mash if after Kab al-Ahbr: cf. as-Suyt, Bugya
59; as-Safad, Wf III 347; and Nldeke GdQ II 53 and GdQ III 1 fn. 2.
31
Muhammad Musta f al-Azam, The History of the Quranic Text. From Revelation
to Compilation. A Comparative Study with the Old and New Testaments, UK Islamic
Academy, Leicester 2003.
201
establish prayers and pay alms to the poor. Since Allah sends every messenger in the language of his own people, Isml must have preached in
Arabic. Genesis credits Isml with twelve sons, among them Nebajoth/
Nabat; born and nurtured in these Arabian surroundings they must have
adopted Arabic as their mother tongue. These sons may have preserved
their fathers message by using the prevailing Arabic script; certainly, they
would not have resorted to whatever script was then current in Palestine
(Ibrhms homeland), since two generations had already lived in Arabia.
When Nabat subsequently migrated northwards he must have taken the
Arabic language and alphabet with him. It was his descendants who established the Nabatean Kingdom (600 B.C.E.105 C.E.).
Al-Azam dates Isml and his early Arabic at around 1400 B.C.E. and
concludes: The Nabatean language and script were . . . a form of Arabic
(121).
Such a sentence disregards that the Nabateans spoken language was
indeed Arabic, while the texts they wrote were in Aramaic. The thesis
that Nabatean is a form of Arabic comes close to a linguistic panorama,
which sees in all languages usually called Semitic a form of Arabic.
The Arabic language and script, in their primitive forms, gave birth to
the Nabatean and most probably predated the Syriac (121) is but one
example of such a view.
4. Yarubs Arabic
A further source in tracing the first Arabic speaker was tribal history.
The Arabs, in their own self-view, were not only a linguistic community,
but also marked by common ancestors. Descent was a primary symbol
of a tribal community, and the self-esteem and prestige of individuals
and communities were linked to the purity of descent. The efforts of the
Arab genealogists to establish a link between living Arab tribal groups
and past forefathers were, of course, highly tenuous. The skeptical observation of Ibn H azm (d. 456/1064) that on the face of the earth there is
no one whose descent from them is verifiable did not prevent the emergence of the most speculative lineages.32 By and large, the Arab tribes
claimed descent from one of two ancestors, either a North Arabic origin
32
Ibn H azm, Jamharat ansb al-arab, ed. Lvy-Provencal, Cairo 1948, 8, quoted by
EI 2nd ed. I 546.
202
stefan wild
connected to the North Arab Adnn (also to Maadd and Nizr, later
Qays) or a South Arabic origin connected to Qaht n (also called Yamanis, later also Kalb). Some tribes such as d, Tamd, ram, Jurhum,
Tasm and Jads were believed to have disappeared before Islam. Hostility between the offspring of Qaht n and those of Adnn emerged after
the advent of Islam in the form of the cleavage between the Ansr in
Medina and the Quray. The fact that the Prophet Muhammad belonged
to the Quray brought enormous prestige to the Adnn, i.e., the North
Arabs. The narrative that linked Isml to Arabic was counterbalanced
by the argument that Isml had learned Arabic from the Jurhum, a
South Arabic tribe. While al-Masd accepted that Isml had been
given Arabic by God, he did not deny that Yarub b. Qaht n, the ancestor of the Yamanis, was the first to speak Arabic. Which of the three
Arab group, the extinct Arabs, the Nizris, or the Yamanis was the first
to speak Arabic? The Nizri and Yamani groups vigorously upheld their
own claims to this honour. The Nizris held that Isml was given the
language by God, while the Yamanis contended that Isml had learned
the language from a Yamani tribe living in Mecca.33 This conflict of
claims masked a deeper social and political conflict between the two
groups. Al-Masd, while granting that Yarub Ibn Qaht n, the ancestor
of the Yamanis, was the first to speak Arabic (Murj sec. 71), believes
that Ishmael too was granted this honour by God, independently of his
association with the Yamanis.34 The tribal family trees set up by Arab
genealogists always have an agenda. When names are inserted or omitted, insertion and omission usually serve a purpose. When in the list
of Qaht ns ancestors there are two new names, those of Yajub and
Yarub, inserted between Saba and Qaht n (Yoqtan) in a family tree
that is otherwise based on Gen 10, 132, this happens because these
two names fulfill two important functions. Yarub symbolizes and personalizes the change from Syriac to pure Arabic, while Yajub/Yaman
gives his name to the land Yaman. The appearance and etiological function of persons with such telling names is a common feature of Arab
genealogy.35
A tradition quoted by Yqt and traced to ad-Dnawars Mujlasa
links Bbil with Yarub and Arabic as the heavenly language: When God
33
Tarif Khalidi, Islamic Historiography. The histories of Masudi, Albany 1975, 116;
conflicting claims are set forth in Tanbh 7983 and Murj 99699.
34
Tarif Khalidi, Islamic Historiography ibid.
35
Manfred Kropp, Geschichte der reinen Araber II 379f.
203
assembled mankind in Babylon, He sent them an Eastern, a Western, a

Southern, and a Northern wind. And He assembled them in Babylon
and they assembled to see why they had been called. Suddenly, a voice
called out: He who has the West on his right, the East to his left, and
faces Meccahis is the language of the people of heaven. Then Yarub b.
Qaht n stood up, and the voice said: Yarub b. Qaht n b. Hd, you are
the one. And he was the first to speak Arabic.36
According to Ibn Durayd, the pure Arabs were seven tribes d,
Tamd, Imlq, Tasm, Jads, Umayyim und Jsim. Most of them have
vanished except for some remnants dispersed in the sub-tribes. And
Yarub, whose name was Muhazzim b. Qaht n, received his byname
because he was the first whose tongue switched from Syriac to Arabic.
This is the meaning of al-Jawhars statement in as-Sih h : the first to
speak Arabic was Yarub b. Qaht n. 37 Here, the eponymon Yarub is
linked to the root rb, another example for the favorite method of the
genealogists of connecting proper names with imagined history, etymology with linguistic history.
In this mixture of traditions, Yarub b. Qaht n is finally also located in
Babylon and celebrates his virtues in the style of a poetic self-praise.
The majority of historians among them, the author of Tawrx alumam (al-Isfahn), and the author of al-Marif (Ibn Qutayba) tell us
that Yarub b. Qaht n was the first to speak in clear Arabic, the first to
bear the crown in Yaman, the first who was greeted by his sons with the
formula May you avoid the curse and your day be happy!
Ibn Hishm, the author of the Book of Crowns, which he wrote on
the kings of H imyar reports: It was Yarub who went with the Arabs
to Yaman and settled there. Therefore Yaman was called after him,
because Yarubs name was Yaman. Al-Bayhaq says: Yarub was the
first to speak in clear Arabic and left the confused way of talking used
by the Arabs. He and his offspring were successful and God bestowed
on them the land of Yaman as heritage. To them belong the kings of
Tubba, who conquered the regions of the earth. On the strength of this
privilege they are ranked before Adnn, while the Adnn can boast the
prophethood of Muhammad. . . . At the time, in which the languages
had already been confused, the Ban H m had come to the highlands of
36
37
Yqt, Mujam al-buldn ed. F. Wstenfeld, I 447, 19.

as-Suyt , Muzhir I 31, -3.
204
stefan wild
Yaman. . . . And Yarub was the most high-minded youth in Babylon and
spoke clear Arabic, while the others spoke in confused language. 38 And
Yarub recited the following verse:
I am the youth favored by the richest gift, the happy one, well known for
his virtue. I am the son of Qaht n, the influential and rich, I spoke in
Arabic, while the people were in (linguistic) confusion. (I spoke) in the
clearest unambiguous language and in the perfect language of the kingdoms after me.
an l-gulmu d n-nasbi l-ajzali / al-aymanu l-marfu bi-t-tajammuli /
an bnu Qaht na l-hammmi l-aqyali / arabtu wa-l-ummatu f tabalbuli /
bi-l-mantiqi l-abyani gayri mukili / wa-mantiqi l-amlki badi l-kamali.
This is pure poetic self-praise in the rajaz meter as we know it from the
earliest Arabic poetry. According to this report, Yarub b. Qaht n was
the first to speak clear Arabic and turned away from the confused version of Arabic (namatu l-arabiyyati l-mubalbalati). He let his sons greet
him with the greeting abayta l-lana (may you avoid the curse!) and
with good morning! This Arabic is not so much seen as a prophetic
language but as a heroic language spoken by a tribal noble forefather.
The hero of this poem, Yarub, praises himself for speaking pure Arabic,
a royal language of noble kings, while lesser Arab mortals used mixed
and confused languages.
In a further verse, Yarub predicts the coming of the Prophet
Muhammad:
Muh ammadu l-hd n-nabiyyu l-mursalu / li-llhi darru l-mjidi l-mustaqbili
Muhammad, the guide, the God-sent Prophet/ how praiseworthy is the
blessed one who is coming.
Such a tradition combined tribal history with revelation. And to round

things out, the report claims that the same Yarub was the first to recite
Arabic poetry and to put it into meters. He invented the poetic genres
and composed praise-poetry, self-praise, and love poetry.39 Earlier traditions see the Ban Hshim and the Ban Yarub in fierce competition. When the rival of the Prophet Muhammad, Musaylima the Liar,
who traced his descent to the Ban Yarub, heard of Muhammads death,
he hoped to outstrip the Prophet and is said to have recited the verse:
38
39
Kropp, I Arab. text 9ff.

Kropp 11, transl. 149.
205
Passed away the prophet of the Ban Hshim / and rose up the prophet
of the Ban Yarub!40
5. The Prophet Muh ammads Arabic

As we have seen, after the advent of Islam, the unique character of Arabic was extrapolated by theologians, philosophers, and genealogists back
to times immemorial. In their view, after the revelation of the Arabic
Qurn, Arabic became even more of a special language. Arabic itself,
not only its Qurnic form, was something sacred, superhuman. Yet, in a
famous description, a-fi (d. 204/820) explained the special character of Arabic mainly by its most important hermeneutical challenge, i.e.,
its vast and ramified vocabulary. According to him, this was the primary
reason for the privileged status of Arabic:
We do not know that any man except for a prophet can claim to know the
Arabic language completely. However, nothing Arabic escapes the Arabs
collectively, so that there would be nobody among the Arabs who knows
it. The knowledge of Arabic among the Arabs resembles the knowledge
of the traditions (sunan) among the scholars of religious Law. We do not
know anybody who could claim to know all traditions, so that not a single
one would escape him. But when the knowledge of traditions of all scholars is collected, the whole of tradition is found. When the knowledge of
each of these scholars is divided, something of the traditions will escape
each of them. But what escapes him will be found with someone else.41
The Prophet, who as an individual can claim to know Arabic completely,

is, of course, Muhammad. His knowledge of Arabic is superhuman. But
the Arabs collectively also know the whole language.
The Prophets own mastering of Arabic was related to the Qurnic
revelation, as well. When a man admires the Prophets rhetorical talent and says, What a good speaker you are: We have not seen anybody
speaking better Arabic than you! the Prophet answers: This is my right.
For the Qurn was revealed to me in clear Arabic.42 And the Prophet
could boast: I am the best Arabic speaker (an afsah u l-arab43 or an
40
Ibn Katr, Bidya vi, 341, quoted according to M. J. Kister, Musaylima in EQ 3
(2003) 462.
41
A-fi, Risla, ed. Ahmad Muhammad kir, Cairo 1940, 42, 8ff.
42
as-Suyt , Muzhir I 35, 6.
43
Muzhir I 209, 2.
206
stefan wild
arabukum).44 It is not always clear whether such statements refer to

Qurnic revelation or to the Prophets non-Qurnic words. In the same
vein, the Prophet was immune against linguistic mistakes (lah n) in his
speech. He was reported to have said: I am from the Quray and I grew
up among the Ban Sadhow should I commit a linguistic error? (an
min Quray wa-naatu f ban Sad wa-ann fiyya l-lah n).45 The Ban
Sad b. Bakr b. Hawzin were the clan of Muhammads wet-nurse. In this
case, the Prophets claim to linguistic excellence was definitely rooted in
his tribal background and not in divine grace. Among the further signs
of the Prophets unprecedented eloquence was that he was said to have
coined expressions that allegedly had never been used in Arabic before
him and that later became part of an elevated Arabic style, such as mta
h atfa anfih, he died a natural death.46
In some traditions, the Prophet became a second Adam. In a H adt,
he said: My community appeared to me in water and clay. And I was
taught all names even as Adam was taught all names.47
Arab grammarians connect the linguistic quality of the Arabic of the
Quray with pre-Islamic times and with the role the Quray played at
the Kaba in Mecca. A privileged position vis--vis the Arabic language
was also conferred on the companions of the Prophetthey also spoke
nothing but the purest Arabic. The proof was that they never coined
a new word (m alimnhumu sta lah al xtiri lugatin aw ih dthi
lafzatin lam tataqaddamhum).48 This was most probably connected to
the fact that the companions were the guardians of the Prophets word
and deeds. If their Arabic had been doubtful, it could have influenced
the correctness of tradition. In a comparable manner, the ethical, philosophical, and religious letters and pronouncements, ascribed under the
title Nahj al-balga (The Method of Eloquence) to Al b. Ab Tlib,
were considered by many Shiites in essence and style as second only to
the Qurn.49
44
Ibn Sad, Tabaqt I 113.

Ab t- Tayyib al-Lugaw, Martib an-nahwiyyn, ed. Muhammad Ab l-Fadl
Ibrhm, Cairo n.d. <1955>, Maktabat Nahdat Misr 6,1.
46
Muzhir I 209, 5.
47
Muzhir I 35, 4.
48
Muzhir I 10, 9.
49
Nahj al-Balgawa-huwa majm m xtrahu -arf Ab l-H asan Muh ammad
ar-Rd b. al-H asan al-Msaw min kalm Amr al-muminn Ab l-H asan Al b. Ab
T lib, ed. Subh Slih, Beirut 1982; Moojan Momen, An Introduction to Shii Islam, The
History and Doctrine of Twelver Shiism, New Haven/London 1983, 25.
45
207
6. The fate of Arabic after the Islamic conquest

The Arabic language celebrated its greatest triumph with the vast expansion of the Islamic conquest and the concomitant expansion of the Arabic language. But most Arab grammarians and lexicographers did not
see it that way. In an almost incredible volte-face, they saw the fate of the
Arabic language sealed and its purity doomed to corruption precisely
at the moment when the Muslim faith and the Qurn won over the
world:
wa-lam tazal il-arabu tantiqu al sajiyyatih f sadri islmih wa-md
jhiliyyatih h att azhara llhu l-islma al siri l-adyni fa-daxala
n-nsu fhi afwjan wa-aqbal ilayhi arslan wa-jtamaat fhi l-alsinatu
l-mutafarriqatu wa-l-lugtu l-muxtalifatu fa-fa l-fasdu f l-lugati l-arabiyyati.50
The Arabs did not cease to speak Arabic according to its disposition in
the beginning of the era of Islam and in the pre-Islamic past until God let
Islam prevail over all other religions. Then people entered Islam in masses
and turned to it in flocks. In Islam all separate languages and different
tongues came together and therefore corruption ( fasd) spread in the
Arabic language.
This rather anticlimactic statement marks the beginning of Arabic as

the language of an Islamic world civilization. For many an Arab grammarian and lexicographer, this period was at the same time the starting
point of its decadence and corruption.
7. References
Agha, Saleh Said and Tarif Khalidi. 2002/03. Poetry and Identity in the Umayyad Age.
Al-Abhth 5051, 55119.
al-Azam, Muhammad Musta f. 2003. The History of the Quranic Text. From Revelation
to Compilation. A Comparative Study with the Old and New Testaments. Leicester: UK
Islamic Academy.
al-Azraq. 1858. Die Chroniken der Stadt Mekka (F. Wstenfeld, ed.), vol. 1 Axbr Makka.
Leipzig.
Czapkiewicz, Andrzej. 1988. The Views of the Medieval Arab Philologists on Language
and its Origin in the Light of Al-Suyutis Al-Muzhir. Krakow: Universitas Jagellonica
Acta Scientiarum Litterarumque CMIX.
50
az-Zubayd, Tabaqt an-nahwiyyn wa-l-lugawiyyn (ed. Muhammad Ab l-Fadl
Ibrhm), Cairo 1954, 1.
208
stefan wild
EQ 20012006 = Jane McAuliffe, ed. Encyclopaedia of the Qurn, vol. 15. Leiden:
E. J. Brill
Fischer, A. and A.K. Irvine. 1978. Kaht n. EI 2nd ed., 4: 447449.
Gilliot, Claude and Pierre Larcher. 2003. Language and Style of the Qurn. EQ 3,
109135, especially The mythical narratives on the superiority of Arabic, 118ff.
Goldziher, Ignaz. 1873. Beitrge zur Geschichte der Sprachgelehrsamkeit bei den
Arabern. Sitzungsber. d. Kaiserl. Ak. d.W. Wien, Phil-Hist.Kl. 73, 51152.
Ibn H azm. 1948. Jamharat ansb al-arab. Ed. . Levy-Provencal, Cairo.
Ibn Mutarrif at -Taraf, Ab Abdallh M. b. A. al-Kinn. 2003. Qisas al-anbiy (The
Stories of the Prophets). Roberto Tottoli, ed. Berlin: Islamkundliche Untersuchungen
253.
al-Jumah Muhammad b. Sallm 1916. Tabaqt a-uar. Joseph Hell, ed. Leiden.
Khalidi, Tarif. 1975. Islamic Historiography. The histories of Masudi. Albano.
Kropp, Manfred. 1975. Die Geschichte der reinen Araber vom Stamme Qaht n aus dem
Kitb Nashwat at-Tarab f trkh jhiliyyat al-arab des Ibn Sad al-Maghrib, hrsg. u.
bersetzt von Manfred Kropp, Bd. I: Einleitung und Text, Bd. II: bersetzung und
Anmerkungen, Diss. Heidelberg.
Loucel, Henri. 1963. Lorigine du langage daprs les grammairiens arabes. Arabica 10 I
188208; II 253281; 11 (1964) III 5772; IV 151187.
Rubin, Uri. 1990. H anfiyya and Kaba. An inquiry into the Arabian pre-Islamic background of dn Ibrhm. JSAI 13, 85112.
as-Suyt , Abdarrahmn Jalladdn. n.d. al-Muzhir f ulm al-luga. Muhammad Ahmad
Jd al-Mawl Bek, Muhammad Ab l-Fadl Ibrhm, Al Muhammad al-Bjw, eds.
Cairo n.d., 2nd ed., Dr Ihy al-Kutub al-arabiyya, vol. 12.
Tottoli, Roberto, ed. 2003. The Stories of the Prophets by Ibn Mutarrif al-Tarafi, edited
with an introduction. Berlin, 253.
. 2002. Biblical Prophets in the Qurn and Muslim Literature. Richmond.
Weiss, Bernard G. 1974. Medieval Muslim Discussions on the Origin of Language.
ZDMG 124, 3341.
INFLECTION AND GOVERNMENT IN ARABIC ACCORDING

TO SPANISH MISSIONARY GRAMMARIANS FROM
DAMASCUS (XVIIITH CENTURY): GRAMMARS AT THE
CROSSROADS OF TWO SYSTEMS?1
Otto Zwartjes
University of Amsterdam, NIAS
1. Introduction
Grammars of Hebrew written in Europe in the Renaissance by Christians could benefit from the Hebrew grammatical tradition. Johannes
Reuchlin (14551522) quotes in his De rudimentis hebraicis (1506)
Priscian and Rabbi David (i.e. Qimhi) as well (Law 2002, 247248). In
grammars of Sanskrit written in Europe we see also that the framework
of grammatical description has been derived from the Indian
grammatical tradition. The German Jesuit Heinrich Roth (16201668),
as Hauschild (1988, 1314) observes, uses with perfect familiarity the
technical terms of Indian grammar [. . .] Roth stands entirely within
the Indian grammatical tradition, and probably he used the practical
grammar of Anubhti Svarpcrya, called the Srasvata Vykarana,
which was in general circulation in Hindustn, Bihr and Benares. [. . .].
Another candidate, though a less likely one, would be the grammar
called Mugdhabodha, which was composed in the second half of the
13th century by Vopadeva, but the usage of which was more common
1
This article is an elaborated version of paper delivered at the IIIrd International
Conference on Missionary Linguistics, Hong Kong and Macau, 12th15th March 2005.
The organization of the conference and participation in Hong Kong has been made possible by financial support of the Norwegian Research Council (Norges Forskingsrd)
and the Language Centre of the Hong Kong University of Science and Technology. I
have to acknowledge Kees Versteegh who inspired me to do research in the field of the
History of Linguistics and particularly Michael Carter for his valuable corrections and
suggestions. Thanks to my colleague Maria Cndida Barros, I came across the reference
to the grammar of Lucas Caballero. Research has been made possible by the Radboud
University (TCMO) where my 2 research on Pedro de Alcal started. I continued this
research topic at the University of Oslo, supported by the NFR-project OsProMil (Oslo
Project on Missionary Linguistics). I am grateful to the Rogge Library (Strngns) for
the reproduction of the MS. I gladly acknowledge Pierre Winkler for his translations
from Latin.
210
otto zwartjes
in Bengal . . . The work of Roth was a masterpiece, which does not differ
considerably from current grammars, which similarly depend on the
linguistic achievements of Indias own grammatical tradition (Hauschild
1988, 1314). In the grammars of Japanese written by the Portuguese
Jesuit Joo Rodrigues (15611634), particularly in his description of
particles and verbal endings, we can also find information concerning
the study of tenifa or tenivofa (the study of particles and verbal
endings) from contemporary Japanese scholars (Maruyama 2004, 155).
As has been demonstrated by Gregory James (2007), some missionaries
describing the Tamil language, such as Bartholomus Ziegenbalg (1682
1719), were familiar with some ancient Tamil works on grammar.
These missionaries worked under favourable conditions compared to
their colleagues who described languages, particularly Amerindian and
Austronesian, without any written tradition or an adequate indigenous
grammatical framework they could rely on.
The use of Arabic grammatical terms in the first grammar written in
Europe of vernacular Arabic, the Arte para ligeramente saber la lengua
arabiga (1505) of Pedro de Alcal (Order of St. Jerome) has been the
subject of an article written by William Cowan (1981). In de Alcals
grammar, some technical terms were incorporated in the descriptive
framework, including terms such as damir, temiz, masdar, amr and
xucla. In this article, Spanish grammars of Arabicvernacular and classicalwritten by Franciscans in Damascus and completed, copied, or
printed in Spain in the 18th century occupy our attention, particularly
the grammars of Francisco Caballero and Juan de la Encarnacin (18th
century) and Francisco Caes (17301795). Of the first we have an
unpublished manuscript, which has escaped the attention of researchers until today,2 and of the second a printed work has been conserved,
2
Bibliographical information concerning Francisco Caes can be found in Schnurrer (1811, 79, no. 113) BICRES III, and in Monroe (1970) we find some historical background. However Caes is not mentioned by Fck (1955), Dannenfeldt (1955), and
Killean (1984) and particularly the grammar of Caballero has been neglected by all.
After a century-long period of silence and total neglect, an important monograph on
Bernardino Gonzlez appeared recently, together with a facsimile edition of the dictionary (Intrprete arbico) and his grammar (Eptome) (Lourido Daz 2005), not long after
this paper had been delivered in Hong Kong. When the proofs were almost ready for
publication, I received a copy of this monograph, courtesy of Emilio Ridruejo. Lourido
Daz (2005, I, 2122) had traced seven manuscripts of the dictionary, and six copies
of Eptome de la gramtica rabe made by Bernardino Gonzlez pupils, probably for
their own use. One copy was completed by Blas Francisco de Salamanca in 1704, the
second by Lucas Caballero and Juan de la Encarnacin between 1709 and 1710, the
so-called Tingstadius copy. Two copies were compiled in 1719, one from the El Escorial
inflection and government in arabic
211
studied sporadically by dialectologists describing the urban colloquial

Arabic of Damascus (Lentin 1997), but which also escaped the attention of researchers working in the field of the history of linguistics. In
the Damascene Franciscan tradition, the authors used a great number
of Arabic grammatical terms, adapted in a hispanicised form, which
substituted, accompanied or sometimes superseded Latin terminology :
examples are: arabicacin (irb), moziones (h arakt) or the partculas nasbantes, chazmantes and charrantes. Obviously, these Franciscan
authors were informed by the Maronite Christians, but in the prologue
of the grammar of Francisco Caes, we read that Spanish Franciscans
not only were familiar with the work of Pedro de Alcal, but also that
they were inspired by north-European grammars and dictionaries of
Arabic, such as the famous grammar written in Latin by Thomas van
Erpen (= Erpenius 15851624) and lexicographical work of Jacob Golius
and the other from the University of Valencia and two further anonymous and undated
manuscripts from the Real Academia de la Historia. Lourido Daz states that all these
works were calques of that of Bernardino Gonzlez (Lourido Daz 2005, I, 13). All the
Latin grammars analysed in this article are also listed in two footnotes by Lourido Daz
(2005, I, 130 and 135) but very little importance is given to the influence of these on the
grammatical tradition of the Franciscans linked to Damascus. After having consulted
all these Latin grammars, we have come to the conclusion that the Spanish grammars
of Arabic completed by Franciscans in Damascus or in Spain were heavily inspired by
the Latin grammars, and in some cases they are Spanish calques, or translations of the
Latin examples. This is particularly evident in the grammatical examples and the use of
literally almost the same orientalising terminology, inherited from the Arabic tradition.
Thanks to the evidence of Lourido Daz study, a direct link between the Franciscans in
Damascus and Spain and the Holy Congregation of the Propaganda Fide and the San
Pietro di Montorio can now be confirmed. Juan de la Encarnacin learned Arabic from
his teacher Lucas Caballero, a pupil of Bernardino Gonzlez. The latter, in his turn, was
a pupil of the Italian Fray Bonaventura da Molazzana, who taught at the San Pietro di
Montorio and who arrived in Seville in 1693. It is known that the grammars and dictionaries used at the San Pietro di Montorio were those of Dominicus Germanus and
Philip Guadagnoli, among others, and it is thus probable that Bernardino Gonzlez had
direct access to the Italian grammatical tradition. It is also important for the purposes
of this article to know that the work of Bernardino Gonzlez was also obligatory in
the curriculum for Spanish and Portuguese missionaries (Lourido Daz 2005, I, 34).
Morevoer, the Portuguese Arabist de Sousa was born in Damascus, so all these grammars are thus linked and use common sources. Germanus of Silesia was educated in the
Holy Land, and was later an Arabic instructor in the El Escorial Monastery in Spain.
Although Lourido Daz monograph is without any doubt extremely important for all
those interested in the bio-bibliographical data related to Bernardino Gonzlez and his
successors, little importance is given in it to the influence of Latin sources and almost
nothing is said about possible Arabic sources, the significance of these works from the
perspective of the history of linguistics or the history of Arabic. Are these grammars to
be considered as key creative productions on the part of Spanish missionaries, or are
they nothing more than a chain in a long tradition? In the future we hope to give an
answer to this question.
212
otto zwartjes
(15961667). Other sources mentioned by Caes are the grammars of

Felipe Guadaoli,3 Antonio de Aguila,4 Agapito de Valle Flammarum,5
and Francisco Gonzalez,6 and the Franciscan grammarian also informs
us that he used a considerable number of manuscripts, without specifying which sources these were. In Renaissance grammars, we find usually two subdisciplines in syntax. In the Spanish tradition, syntax, often
called construccin, can be subdivided in construccin de rgimen and
construccin de concordancia and Francisco Caes does not form an
exception. In this article we concentrate on the first and we particularly
attempt to analyze how the concept of mil (often translated as governor) has been incorporated into this model. The subject of concord
(agreement) has been analysed in Zwartjes (2007). The use of Arabic
grammatical terms in these grammars will be analyzed, and we will concentrate on the following questions related to morphosyntax:
Which Arabic grammatical terms are used in these Spanish grammars
particularly in the sections dedicated to inflection and agreement?
What did they mean and why did they use them?
Which sources could they have used?
2. The grammar and dictionary of Arabic of Pedro de Alcal

As we all know, Arabic speaking communities are diglossic. Pedro de
Alcals dictionary and grammar of Granadan Arabic is obviously a
description of colloquial Arabic. His purpose was to teach the ordinary
people (los populares) and not the language of the wise (alfaques).
The aims to compose his dictionary are also slightly different if we
compare them to other missionary dictionaries in the New World,
where missionary composed dictionaries for their own use and for
the novices from the Old World. As we can read in the prologue to his
dictionary, Pedro de Alcal wrote his dictionary not only for the Old
Christians who wanted to learn Arabic, but also for the New Christians.7
Schnurrer (1811, 47, no. 72).

Antonius Ab Aquila (Schnurrer 1811, 50, no. 78).
5
Schnurrer (1811, 59, no. 85), or Agapito Valle Flemmarum (da Val di Fiemme).
6
I have not identified this author yet, but this could possibly be Bernardino Gonzlez,
as Lourido Daz suggests (2005).
7
Ca assi como los aljamiados (o cristianos viejos) pueden por esta obra saber el arauia,
viniendo del romance al arauia: assi los arauigos (o nueuos cristianos), sabiendo leer la
4
213
As has been demonstrated in recent research, mainly from Federico

Corriente (1988), it is true that the language described is predominantly
colloquial,8 but at the same time it is also obvious that there is some
interference between the colloquial and the classical registers. The main
objective of all missionary grammars is practical: the teaching of a
certain language. Nevertheless, missionary works can be predominantly
didactic, showing many paradigms with few explanations and linguistic
theory, while in other works the didactic-pedagogical approach is
much more theoretical. Pedro de Alcal also states that his grammar
is more a practical introduction, not a learned theoretical work. His
approach reveals itself to be universalist, since he observes that there are
definitions and explanations in one language in respect to the expression
of its concepts in its own terms, these are the same in all other languages
regarding the expression of their own concepts. Thus for the same reason
that this name Pedro is a proper noun in Latin, it is also such in Arabic
(Alcal 1505[1883], 2). This observation is characteristic of Renaissance
grammars in general and can be found in many other sources from this
period. As Vivian Law observed:
Questions as what is a proper noun?, what is a verb? how many word
classes are there? what are properties of the conjunction? are as close
to universal as any you are likely to find in a medieval grammar. Such
concerns apply equally well to any European language; indeed, they had
already been transferred from Greek to Latin. There is no inherent reason
why they should not also be asked about Old Irish or Old Icelandic: one
can find proper nouns (for instance) just as easily there as in Latin. (Law
2002, 191)
Obviously, Alcal did not find it necessary to give his pupils definitions
of the parts of speech, since they are the same for all languages. So one
would wonder why the author decided to include Arabic grammatical
terminology, which seems to be in contradiction with his own universalistic approach. A possible explanation is that he did this only for
scientific reasons. He might have introduced them with the purpose to
have a more adequate or sophisticated framework to fit in phenomena
letra castellana: tomando primero el arauia, ligeramente pueden venir en conocimiento

del aljamia). (Pedro de Alcal 1505, prologue, ii v.) See also Cowan (1981, 358). [As
the aljamiados, or old Christians can learn Arabic through this work, coming from
Romance to Arabic, so the Arabs (or new Christians), having mastery of the Castilian
alphabeth, taking first the Arabic, easily can have knowledge of the aljamia].
8
His purpose was hazer vocabulista de la habla comun y usada de la gente deste.
(ibid.) to compose a dictionary of the common speech and used by the people.
214
otto zwartjes
he was not so familiar with. As has been demonstrated by Cowen, this

is only in a few cases a satisfactory explanation. Sometimes Arabic
terms are only used as equivalents or translations of Latin terms. Clear
examples are the names for the cases in the nominal paradigm. Pedro de
Alcal recognizes when dealing with declensions that the Arabic noun
has only one declension, since all Arabic nouns are invariable.9 However,
admitting that there is only one declension, he states that Arabic nouns
have six cases. We reproduce here the table according to Cowan (1981,
359):
TABLE 1
THE SIX CASES ACCORDING TO PEDRO DE ALCAL
Alcal
Translation
Classical term
mubted
subject
mudf
genitive
maxrr
dative
maful10
object
munde
vocative
zarf
adverb
nominatiuo
mubtada
genitiuo
mudf
datiuo
majrr
acusatiuo
maf l
vocatiuo
mund
ablatiuo
zarf
As William Cowan observed:10

they are not true cases in either the Arabic or the Latin sense of being an
inflection added to a noun, a fact that Alcal was quite aware of, but are
regarded by him in the same way that modern theoreticians of case grammar regard the syntactic positions in an utterance. Such relations are not
expressed by inflections, since colloquial Arabic has none, but through
abitudines or conocimientos. The markers of the equivalents of these Latin
cases are in fact combinations of prepositions and particles and the definite article (lil, maal, lal, etc.) (Cowan 1981, 359).
9
La declinacion de todos los nombres arauigos es vna solamente. Porque todos los
nombres arauigos son inuariables (Alcal 1505, capitulo nono).
10
In the original text, a small hamza in superscript is placed on the first vowel u.
215
The information given by William Cowan, however, is not complete,

since translations from his dictionary are not included here. Although
we do not find all the terms in his dictionary, we do find some of them:
mubted is not only translated by Pedro de Alcal as nominative, but
also as principio de oracion (Corriente 1988, 12), which is not without
importance (see below). zm mudf is rendered as posesivo nonbre.11
The term abitud seems to be inherited from other grammars than
the Latin and Castilian grammars of Antonio de Nebrija (c. 14441522).
The term habitudo is used by Ferdinandus Nepos in his Materies (completed between 1469 and 1485) and Juan de Pastrana Compendium
grammatica (1462).12
Item per hanc regulam primo iuvenes component per unum casum tantum
sic dicendo: la tierra terra; de la tierra terre, etc.; uel per duos, sic: la tierra
del rey; uel per tres et quatuor et amplius, sic: o leyente la lection a los scolares en el general de las escuelas componitur o legens lectionem scolaribus
in generali scolarum, dando cuilibet casui propriam habitudinem, interrogando cuius casus, numeri et cuius declinationis hoc principio. (Codoer
2000, 90).
However, it is not so clear what Pedro de Alcals exactly means with

the term abitud (pl. abitudines), since he uses it as a synonym of the
definite article (Nota que porque ay algunas abitudines en cada vno delos
casos que en alguna manera parescen preposiciones, porque se preponen
alos casos, avn que en verdad no lo sean, mas son articulos) (Alcal
1505[1971], 26). In another section, Alcal uses the term as a synonym
of preposicion, since according to his observations, the maxrr case
(see below) has the four abitudines la, lal, li, lil, whereas the term
abitud is not used for ba, bal, bi, bil, fa, etc. which are described in
the paragraph on the zarf case (Alcal 1505[1971], 27). Here they are
just called prepositions and not abitudines and there is no explanation
11
Other terms not mentioned by Cowen, are jezme, translated as consonante (letra
mazjun [sic] (Corriente (1988, 34), and mdde, translated as acento (Corriente 1988,
189).
12
With this rule, the novices first build constituents with only one case [= head of
the NP], saying la tierra terra, de la tierra, terre, etc., or with two cases [= head
+ complement of the NP], as la tierra del rey, or with four and so on, as: reading the
lesson in general for the scholars of the schools, assigning the appropriate habitudo
[= grammatical form] to whatever case [= grammatical function]. The term habitudo
casualis is also used by Nepos in relation with government: Haec enim regula maxima
est in construendo [. . .] quia talem casum regit dictio qualis fuerit habitudo casualis.
(ibid.).
216
otto zwartjes
why he used these different terms. In other sections, Alcal uses abitud
as synonym of conocimientos (see also Zwartjes 1993, and 1994).
In Arabic grammatical theory a nominal sentence can be divided into
topic for which mubtada (lit. what is begun with) is normally used,
and comment (xabar), or predicate (Owens 1988, 32), or according to
Sbawayhis terminology also called mabny al-l-mubtada (what is built
upon what is begun with) (Owens 1990, 45). Mudf is the word Pedro
de Alcal uses for the genitive, which in Arabic tradition means literally
what is added, i.e. possessed (Owens 1988, 34; 1990, 104). Majrr from
the same root as jarr (see below), means pulling, or governing the -i
inflected form. Pedro de Alcal follows in his sections about the prepositions the Latin system and tries to apply Arabic terms to them, without
realizing that in the dialect he describes, case-endings are not used, and
without realizing that in classical Arabic nominal declension, there are
three inflectional vowels, the -u, the -a, whereas for the verbal inflection
the three vowels -u, -a and (zero ending) can be distinguished. Pedro
de Alcal did not take the Arabic inflectional endings as starting point,
but the Latin prepositions in alphabetic order: prepositions + accusative, prepositions + ablative, etc. and at the same time he translated the
names of these Latin terminology into Arabic:
Capitulo XXXII. De las preposiciones.
Hallamos en el Arauia todas las preposiciones que en la gramatica [latina],
y ayuntadas a essos mesmos casos, que son maful y darf (que son acusatiuo
y ablatiuo), y son las del acusativo las siguientes:
A
A
ad
al
apud
aynd
circa
carib
circa
qued
ante
acbal
longe
bad
ante
cudim
[. . .]
[. . .]
Las preposiciones del darf (que es ablativo) son las siguientes, conuiene
saber:
Con con con
Ba bal bi
con
bil
en
fa
en
fal
en
en
fi
fil . . .
(Alcal 1505[1883], 26).
Chapter XXXII. About the prepositions.

We find in the Arabic language all the same prepositions as in Latin grammar, and they are combined with the same cases, which are mafl and
zarf (which are accusative and ablative), and those which can be combined
with the accusative are the following:
A.
A
ad
al
apud
aynd
circa
carib
circa
qued
ante
acbal
longe
bad
ante
cudim
[. . .]
[. . .]
217
The prepositions of the darf (which is ablative) are the following:

con con
ba bal
con
bi
con
bil
en
fa
en
fal
en
fi
en
fil . . .
As we can see, Pedro de Alcal did not separate the prepositions from
the definite article (al).
Mafl is the term used for object. Sbawayhi distinguishes five subtypes: mafl bihi direct object, mafl fhi locative object, mafl maahu
accompaniment object, mafl lahu reason object, and mafl minhu
object from it. (Owens 1990, 160). For the locative object, instead of
mafl fhi the term zarf is also used (Owens 1990, 51,141151), which
is the term Alcal uses here for the ablative. Mund is the direct translation of vocative. According to Cowan (1981, 360), Alcal was apparently trying to make unfamiliar material intelligible to his audience, but
at the same time to avoid a direct equation with the Latin categories. In
fact, the first might be true, but we must be aware that in his paradigm of
the case-system, we do find an equation with Latin cases, and we never
find any traces of the four traditional Arabic inflectional endings, -zero,
-u, -a, and i and never the original Arabic names for these inflectional
endings are introduced here.
Other technical grammatical terms in Alcals grammar are: amir
(conocimiento;13 cl. Ar. damr conjunct pronoun), temiz (cl. Ar. tamyz
accusative of specification),14 amr (imperatiuo; cl. Ar. amr imperative), xucla seal; cl. Ar. akl(a) orthographic sign:15
Es otrosi de notar, que los Arauigos non tienen letras vocales como los Latinos, mas tienen en lugar dellas ciertas seales, que ellos dizen xclas, con
las quales y con todos los caracteres suso dichos leen y escriuen lo necessario
(Alcal 1505[1883], 4).
It has to be observed that the Arabs do not have the letters for the vowels
as the Latins, but instead of them, they have certain signs, which they call
xuclas, and with all the above-mentioned characters they read and write
anything which is necessary.
13
Cf. Zwartjes (1994).

Tem is also translated as conocimiento in his dictionary (Corriente 1988, 197).
15
In his dictionary Alcal translates the word alma as (signo por) seal (Corriente
1988, 140). The seal is also used in his grammar for the article (seal de demostracin),
so this term could be the technical grammatical term, but as happens often in these
grammars, it is not always possible to distinguish between language and metalanguage.
14
218
otto zwartjes
The origins of the mnemonic vowel terms with and without tanwn
(the nunated forms) which expresses indefiniteness in Arabic are
unknown to me and this could be a local teaching method, not recorded
elsewhere:
TABLE 2 THE VOWELS AND THE NUNATED FORMS
ACCORDING TO PEDRO DE ALCAL (1505[1883], 4)16
minib
minib
minib
minenbn
mininbn
minunbn
minb
fath a
kasra
damma
fath atn
kasratn
dammatn
sukn16
a
i
u
an
in
un
The remaining Arabic terms analyzed by Cowan are alif cequin (alif skin
silent alif ), and in his dictionary we find iarab (oracin de gramtica;
clas. Ar. irb (the inflectional endings , a, i, u (see below)17 and harf
(letra; Cl. Ar. h arf), which are not analyzed at all in the grammatical
treatise.
Summarizing the preceding paragraphs, we can conclude that in most
cases Pedro de Alcal could easily use Latin terminology, such as the
names for the cases or the imperative. The use of an exo-grammatical
term damr for the conjunct pronoun does make sense since traditional
grammar did not have precise equivalents from contemporary sources
yet. Probably, Pedro de Alcal understood very well that the conjunct
pronoun in Arabic can be used differently from the Spanish pronouns;
they can also be affixed to prepositions and nouns, for instance and
that explains probably the reason why he used the Arabic term. In the
remaining cases, Arabic terms are used for mnemonic or pedagogicaldidactical reasons. It is questionable if these terms made his teaching
16
The terms fath a, kasra and damma are not found in his grammar which gives us
the impression that Pedro de Alcal did not know them.
17
Again, we find more information in the dictionary, neglected by Cowan: iarb is
translated as declinacion de palabras (Corriente 1988, 134), which is not unimportant
because the author avoids the term noun here, since irb is used for nouns and verbs
as well.
219
method more attractive or more easily accessible to his pupils. 18 Seen

through our modern eyes, the grammar of Pedro de Alcal was a real
novelty, a pioneer work. His transcription system of the Arabic alphabet
is the first in Europe, although his knowledge of classical Arabic was evidently limited; there are quite a few inconsistencies while using his own
transliteration system. It is also obvious that Pedro de Alcal did not
have any knowledge of Arabic grammatical theory and his Greco-Latin
approach is particularly visible in the lacking of insights in the analysis
of derivations. As had been observed by Fck [Pedro de Alcal] erkennt
nicht die Bedeutung des Wurzelbegriffs, so dass dem Leser der arabische
Formenbau in der Nominal- wie in der Verbalbildung undurchsichtig
bleibt (1955, 33). Nevertheless, as has been stressed by several scholars
at the conferences on missionary linguistics,19 it is easy to point at the
shortcomings of these grammars and dictionaries measured by the standards of our own time, but when studied and analyzed in its own historical and cultural context, there is no doubt that there is still a wealth
of material to be studied. As observed by Smith Stark certain preconceptions about pre-modern descriptive work have resulted in its neglect
among those in the Western tradition (2005, 4). Monographs and studies on the description of Arabic by Spanish missionaries are still non-existent. Positive evaluations are scarce when the grammars are concerned,
whereas the majority agrees that the lexicographical work of Pedro de
Alcal and others have been of great value. Dannenfeldts observation,
according to whom Both of these [Alcals] works are based on solid
18
Missionaries usually emphasised that the language they were learning was easy
to learn, although others label the language under description as difficult. They tried to
use the most transparent and less obscure paradigms and explanations. The reason to
re-write existing grammars was almost always because predecessors were too obscure.
Probably, the grammar of Pedro de Alcal could be perfectly understandable without
the use of Arabic grammatical terminology. The same could be said of the use of Hebrew
posodical-grammatical terms by Oyanguren de Santa Ins in his grammar of Tagalog
(1742, 208209), such as milehal (stress on the penultimate syllable, instead of the usual
ultimate syllable, in connection with stress assignment), athnach (semicolon or pause)
and metheg (one type of the several secondary accents, avoiding the loss of vowels in
pronunciation, or a sign, pointing a vowel, which usually would be reduced to schwa
but which is to be fully pronounced in this particular place). The terms atnach and
meteg are both so-called cantillation marks in the Hebrew Bible from Masoretic times.
Did the pupils of Oyanguren de Santa Ins know Hebrew, or is this pure pedantry or
snobbism?
19
The International Conference on Missionary Linguistics took place in Oslo (2003),
So Paulo (2004), Hong Kong and Macau (2005), Valladolid (2006), Mrida (Yucatn,
2007) and the sixth will be organized in vora, Portugal.
220
otto zwartjes
philological methods and are evidences of humanist learning in Spain

(1955, 33),when compared to Fcks (1955, 33)is without any doubt
an exception. Although Pedro de Alcals grammar served as the model
for the Gramtica arbigo-castellana (still unpublished) composed by
the Hiernoymite Patricio Jos de la Torre (17601819),20 and although
we have some evidence the Maronite Joseph Simon Assemani of Syrian origin (16871768)21 quoted from his grammar, Alcals dictionary
was particularly widely known and used throughout Europe. Johannes
Gabriel Sparvenfeldius possessed probably Alcals catechism, which has
been appended to his grammar and his dictionary already in 1706 but
in that period it had become already a rare book (Schnurrer 1811, 16).22
If the original work was not longer available, scholars used the original
editions or the re-edition from 1776 from Patricio Jos de la Torre with
transcriptions into Arabic characters, which was the base for the Supplment of the Dutch Orientalist Reinhart Dozy (18201883) (Monroe
1970, 38). One of the main purposes was to publish the work of Pedro
de Alcal in Arabic script, something that was completely unnecessary
according to Pedro de Alcal. Although we find an entire page with the
Arabic alphabet in his grammar, an observation below it tells us all
characters can be substituted by Latin or Castilian letters:
Estos son los caracteres y nonbres de las letras arauigas, las quales todas se
pueden suplir con nuestras letras Latinas o castellanas, de manera, que para
la comun algarauia no ay necessidad de las saber ni conocer todas, mas
solamente quarto conuiene saber kha, dil , te, ay, cuyos sones no tenemos en
nuestro ABC latino. (Alcal 1505[1883], 34)
These are the characters and names of the Arabic letters, which all can
be substituted by our Latin or Castilian letters, so that there is no need
to learn or to know them all for the common speech, but only four [are
necessary], namely the kha, dil , t e, ay, whose sounds we do not have in
our Latin ABC.
We have seen that for didactic reasons, Pedro de Alcal used these
mnemonic words, which are not derived from canonical grammatical
works of the Arabic tradition. Other grammarians found a different
20
Schnurrer (1811, 88, no. 128). BICRES III, no. 84.

Schnurrer (1811, 16) observes that Assemani quotes from his grammar ex grammatical recitat, but also adds that he actually used material from Alcals dictionary
(non Grammaticae sunt, sed Vocabularii, ibid.).
22
Ioan. Gabr. Sparvenfeld. 1706. Catalogus Centuriae Librorum Rarissimorum Manuscript.& partim Impressorum, Arabicorum, Persicorum, Turcicorum, Graecorum, Latinorum, &c. Upsala: John Henry Werner.
21
221
solution. One of the most remarkable solutions can be found in the

Fabrica Arabica (1640) of Dominicus Germanus of Silesia (15881670).
In this case, we have translations of the Latin names of the cases into
Arabic. In the following table we see these literal:
TABLE 3 THE CASES ACCORDING TO DOMINICUS
GERMANUS FABRICA ARABIC (1640)23

al-mutasamm
al-mutawallid
al-mustat
al-mutak
al-mund
al-mustaqti
Nominativus
Genitivus
Dativus
Accusativus
Vocativus
Ablativus23
3. The grammars of Lucas Caballero, Juan de la Encarnacin, and

Francisco Caes: Grammars at the crossroads of two systems?
3.1
Introduction
After the foundation of a great number of missions in the East, the

importance of the Arabic language for preaching the Christian faith
continued to increase. Paul V in a papal bull dated 1610 had commanded
the various religious orders to teach Oriental languages in their colleges.
In the early 16th century, Arabic was taught in Seville at the Colegio
trilinge.24 The Franciscans decided to found colleges in Salamanca, Alcal,
Paris and Toulouse for the teaching of Arabic, Greek and Hebrew (Monroe
1970, 26). According to Monroe, Bernardino Gonzlez (c. 16651735)
composed an Arabic dictionary in Seville, which was completed by
Franciscans in Jerusalem in 1709 (Monroe ibid.), an unpublished work.
Jos de Len began to compile a new dictionary of Arabic and Bernardino
Gonzlez was sent to Damascus in order to complete his work.25 As Monroe
23
From the verbs samm to denominate, walada to give birth, at to give, ak to
complain, nad to call and qatta to cut off, to disjoin, the tenth derived form means to
deduct, which seems to be an approximate translation of auferre.
24
In this short account, there will be no space to summarize the study of languages
during the Middle Ages in Muslim Spain.
25
Franciscans had already arrived in 1233 in Damascus. Propagating the Christian
faith was not permitted by Sultan Malik al-Araf, but they took care of Europeans who
settled there. After several cases of martyrdom and imprisonment, the Cadi of the city
222
otto zwartjes
observed, the work was the most complete ever to have been composed
in Spanish after that of Pedro de Alcal. [. . .] The work of Gonzlez and
his collaborators was lost until Asn Palacios (1901) came across a copy
containing additions up to the year 1727 made by the friars of Damascus
(Monroe 1970, 27). One of the successors of Bernardino Gonzlez was
Francisco Caes who settled in Damascus at the Spanish Franciscan
College in 1757 (Monroe 1970, 28). Caes grammar of Arabic has been
printed in two different editions, a first in 1775 (Madrid, Don Antonio
Perez de Soto, and a second in 1776 (Schnurrer 1811, no. 113, BICRES
959 and 971), entitled Gramatica arabigo-espaola, vulgar y literal. Con
un diccionario arbigo-espaol, en que se ponen las voces mas usuales
para una conversacin familiar, con el texto de la Doctrina Cristiana en
el idioma arabigo.
I came across another copy from the same Franciscan tradition, composed by Lucas Caballero and Juan de la Encarnacin as we can read in
the colophon, which escaped the attention of scholars who have worked
in this field. The manuscript has been identified by Magnus and Aare
Mrner in his Spanien i svenska arkiv. The title of this manuscript is Compendio de los rudimentos y gramtica rabe en que se da notizia de la lengua verncula y Vulgar y algunas reglas de la literal Iustamente, 1709, and
in the colophon 1710 (another author, Juan de la Encarnacin, finished
the text San Diego, Seville). The work is based on Bernardino Gonzlez
as we can read in the title, and Lucas Caballero, lector actual Arabo en
el Colegio de Damasco composed (recopilado) this manuscript, which
has been donated by Johan Adam Tingstadius (17481827),26 bishop of
Strngns, Sweden, from 1803, to the Rogge library, which has belonged
administratively to the Royal Library of Stockholm since 1968.
As the titles of the grammars of Caes and Caballero demonstrate, the
language under description is not only classical Arabic, but the urban
dialect of Damascus. Apart from Alcals grammar of colloquial Arabic of Granada, European scholars usually did not pay much attention
to lower registers, so the linguistic works of these Franciscans work-
granted the privilege of being able to open a public chapel and in 1668 the Franciscans established themselves in a Maronite church, which they left in 1719 when they
acquired a new church in the Christian Quarter of Bb Tma. The foundation of the
college where Arabic was taught dates from this period.
26
Tingstadius was a professor in Oriental languages at Uppsala. He published, for
instance in 1770, a treatise entitled Dissertatio philologica de natura et indole linguarum
orientalium communi (Uppsala: Johan Edman) and in 1794 his Dictiones arabicae ex
carmine Tograi, hebraismum biblicum illustrantes. Uppsala: Johan Edman.
223
ing in Damascus are of great importance. However, they were not the
only grammarians who described non-Classical registers. Antonio ab
Aquilas grammar published in 1650 is not only a grammar of classical
Arabic (ad grammaticae doctrinalis intelligentiam) but also colloquial
Arabic (ad vulgaris dumtaxat idiomatic), probably the reason why he
called the grammar Arabicae linguae novae et methodicae institutiones.
Dominicus Germanus (Germanus of Silesia; 15881670)27 composed a
dictionary in 1636 with the title Fabrica overo Dittionario della lingua
volgare arabica et italiana, copioso di voci e locutioni, con osservare la
frase delluna e dellaltra lingua (Roma. Nella stampa della Sac. Congr. De
Propag. Fide) followed three years later by his Fabrica linguae Arabicae
cum interpretatione latina et italica, accommodata ad usum linguae uulgaris et scripturalis (Roma. Typis Sac. Congreg. De Prop. Fid.).
In 1800 a work has been completed by Franciscus de Dombay (1758
1810) with the title Grammatica linguae Mauro-Arabicae juxta vernaculi
idiomatis usum, accessit vocabularium Latino-Arabicum (Vindobonae:
apud Camesina)28 but, according to Schnurrer, this title is misleading;
although this grammar describes the common speech of the people in
the Maghreb (Arabicus sermo in Mauritania quo vulgus uti solet), he
observes that all words are good Arabic (non sunt vulgari idiomati propria, sed omnia bene Arabica) (Schnurrer 1811, 95).29
27
Germanicus was a teacher of Arabic at the mission school St. Peter in Montorio,
Rome. He assisted with the preparation of the Arab Bible, he published dictionaries,
and commentaries on the Qurn. He was teacher and translator at the court of Philip
IV of Spain.
28
I have not been able to consult this grammar yet.
29
Although this is not the aim of this paper, I wish to show just a few elements from
these sources that are important records of colloquial Damascene Arabic from the
beginning of 1700. Particularly the word lists are full of colloquialisms, but also the
grammar of Lucas Caballero has many colloquial elements, to mention a few: omission of vowels: muqatla instead of muqtala, the use of the -u- vowel as a prefix for the
imperfect tense (64), b- future suffix (p. 24), which is colloquial (in Egypt it is the present
tense). However, Caballero is not always consistent, we find both faaltu as faaltum (2
person pl.m.), the use of - instead of -na for the second person feminine singular in
taf al (= Haztu fem.) (72). It is remarkable that sometimes we find even hybrid forms,
such as antum faaltu. However, we find also classical elements, such as the use of the
feminine plural in the verbal paradigms, which is not used in colloquial urban speech
today. It is also significant that the order of the persons singular in the verbal paradigms
is not the traditional one 73v72r. Caes has 3 (masc. Sing, 3 fem. Sing, 3. plur. 2 masc.
Sing. 2 fem. Sing. 2 plur. 1 sing 1 plur. Instead of 3,3,2,2,1 (sing.), 3, 3, 2, 2, 1. I am grateful to Manfred Woidich for his comments on this footnote.
224
otto zwartjes
Grammars of Classical Greek were available throughout Europe, but

Spanish priests were engaged in the writing of vernacular Greek too, as
we can read in the prologue of Francisco Caes (1775, s.n.).30
In the following paragraphs we will treat the use of Arabic terminology in both grammars separately.
3.2
Semantic and inflectional vowels and the concept of amal
The questions to be answered now are: which Arabic grammatical

terms are used, particularly in the sections dedicated to inflection and
government, what did they mean and why did Caballero and Caes use
them?
Lets start with the vowel system, according to the paradigm of Caballero are:
TABLE 4A
bu
ba
bi
THE VOWEL SYSTEM ACCORDING

TO CABALLERO31
Damma31
Fatha
Kesra
O que inclina a V
A que inclina a E
E que inclina a I
Unlike Pedro de Alcal, Caballero gives also in an appendix of this

Chapter the Arabic technical terms for these vowels when they are used
as inflectional endings, accompanied with translations into Spanish:
Los Gramaticos a estas mociones dan otros nonbres conbiene a saber que
significan, eleuacion, ereccion, y contraccion: al-raf, al-nasb, wa-l-jarr y
estas mociones duplicadas llaman tanun (tanwn) esto es nunacion que es
30
It must be emphasized that missionary sources, often written in Spanish, are in
many cases the only existent sources which can give us information of vernaculars once
spoken in early ages of languages of which we only have more detailed information concerning the literary or classical register. Priests understood very well that in China the
teaching of Mandarin was not so useful in regions were other dialects were spoken.
Grammars of modern Greek circulated since 1638, the Grammatica lingu grc
vulgaris was printed by Simon Portius. The first Spanish grammar of modern Greek
was composed by Pedro Fuentes, as we can read in the following quotation from the
same prologue: Por lo tocante la lengua griega ha impreso su gramatica vulgar Fr.
Pedro Fuentes observante, que residi en el Seminario de Nicosa en Chipre, y ahora est
imprimiendo la gramatica literal.
31
In their tables of the vowel system, both authors also give the names of the vowels
in Arabic script, not reproduced here.
225
lo mismo que decir addicion de un sonido de nun o n y biene a sonar on, an,
en vg. Racholon, Racholan, Racholen, Homo, Hominem, Hominis [. . .] pero
esto ueras mas claro en el tratado del nonbre (77 r.).
The grammarians give other names to these motions [h arakt], namely
the names which signify elevation, raising, and contraction, ar-raf ,
an-nasb, wa-l-jarr, and these motions, when duplicated, are called tanwn,
which is nunation which is the same as adding the sound of a nn, or -n,
which approximately sounds as on, an, en vg. Racholon, Racholan, Racholen, Man (nom.), man, (acc.), man (gen.), but this you will see more clearly
in the chapter about the noun.
Caes table resembles that of Caballero, although there are some

differences. The order of the vowels is different, Spanish translations of
the Arabic vowel terms are given, and instead of de verb inclinar we
find declinar:
TABLE 4B
ba
bi
bu
THE INFLECTIONAL VOWELS ACCORDING TO CAES

(1775, f. 7)
Fatha, Apercion
Kesra, Fraccion
Domma, Colleccion
A que declina e, y a clara. Ba, Be.

E que declina en e, i claro.
O que declina en o, y u. como Bo, Bu.
Also here, we find almost the same observation as above, which

demonstrates that the works are closely related to each other and that
they probably derived from a common source (or sources):
Los gramaticos las tres mociones dichas les dan otros nombres, es saber:
al Fatha le llaman Nasbo, esto es ereccion, al Kesra Charro, contraccion,
y al Domma Rafo, elevacion. Suelen tambien duplicar las dichas mociones
de esta suerte ( ) ( ), y entonces les dan el nombre tanuin, esto es, nunacion,
nun vocal, que viene sonar an, en, on. (Caes 1775, ff. 89).

The grammarians give other names to these three above-mentioned

motions, namely: they call the Fatha Nasbo which means erection, they
call Kesra Charro which means contraction, and Domma Rafo elevation.
They are used to dublicate the above-mentioned motions in this way: ( )
( ), and they give them the name tanwn which is nunation, or vocal nn
which sounds rougly like: an, en, on.
In Chapter 4, both authors deal with the cinco signos:

(1) Secun (quietud), which has according to both the synonym chezm
(caballero), or chiasmo (Caes), (jazm) which Caballero translates
as anputacion and Caes as corte;
226
(2)
(3)
(4)
(5)
otto zwartjes
Tex did (duplicacion, corroboracion);
Maddo (extension produccion);
Vasalo (union, conjuncion); and
Hamza (punzamiento).
A chapter dealing with how to read Arabic without vowel signs was
appended by Caballero, not present in the grammar of Caes. Although
there are differences, both grammars are from the same tradition and
both authors (re-)formulated probably the lost version of Bernardino
Gonzlez, or quoted directly from other sources, such as Agapito Valle
Flemmarum, who almost has the same definitions as Caess, although
the vocalizations of the Arabic terms and the order is slightly different:
TABLE 4C
bu
ba
bi
THE INFLECTIONAL VOWELS ACCORDING TO

AGAPITO VALLE (1687, f. 79V.) 32
Dzhamma, collectio
Fathha, aperitio
Kesra, fractio
o declinans ad u, & u claru

a declinans ad e, & a claru
e declinans ad i, & i claru
The Tratado III deals with the noun. Here we find sometimes some
parallels with the grammar of Pedro de Alcal, particularly since
the term seal has been used in both sources, or notificacin for the
article33 and the abitudines of Pedro de Alcal resemble much the
seales del nombre. Since colloquial Arabic nouns are not inflected
according to cases, we find in the grammars of Alcal, Caballero and
Caes equivalents of the Greco-Latin cases for didactic reasons:
Los arabes aunque en la lengua vulgar reconocen tres numeros en el nombre,
es saber: singular, dual, y plural no conocen distinction de casos. Y asi el
nombre en qualquier caso termina con una misma voz. (Caes 1775, 59).
The Arabs, although they recognize in the colloquial speech three numbers in the noun, being, singular, dual and plural, they do not know the
distinction in cases. And thats why the noun ends with the same sound
in whichever case.
32
The vocalization of Martelottus is slightly different: Dzhammon, Phathhon and
Kafron.
33
Pedro de Alcal uses seal de demostracin. See Zwartjes (1992).
227
Elsewhere in his grammar, i.e. after the Numeros cardinales and

Numeros ordinales we find a comparable description:
Los Arabos en la lengua Vernacular o Vulgar no tienen terminacion de casos
en el nombre assi como los Castellanos (16v.).
In the vernacular or vulgar speech, the Arabs do not have the ending of
cases in the noun as the Castilians have.
The case system according to Caballero is as follows:

TABLE 5
THE INFLECTIONAL VOWELS ACCORDING TO

CABALLERO (1709, 14r.)
Nominativo Vocatiuo
raf un
Rajulun
Vir, o vir
nasbun
Rajulan
jarr un
Rajulin
Virum o Virum
Viri
Viro
A viro
(this column in Arabic script)
Acusativo Vocatiuo
Genitiuo
Datiuo
Ablatiuo
What did these terms mean in the Arabic grammatical tradition? Before
Sbawayhi, no distinctions have been made between the vowels which
are used in classical Arabic for the declensional endings, and the other
vowels, for instance: there was no disctinction between both vowels i
in the genitive al-kitb-i. An important novelty of Sbawayhi is that
he distinguished the first i that is non-declensional, from the final i
which is declensional (Versteegh 1997, 19).
When we analyze Arabic terminology in our 18th century grammars,
the sections about the particles are even more interesting. Caes and
Caballero use both a metalanguage inherited from the Arabic tradition
that had already been developed by Sbawayhi:
These endings follow eight courses: accusative (nasb), genitive ( jarr),
nominative (raf ), apocopate ( jazm), a-vowel ( fath ), i-vowel (kasr),
u-vowel (damm), zero-vowel (waqf ). (Translation by Versteegh 1997, 36).
The main distinction is whether a certain vowel is declensional or not.

The Arabic terms used by Caballero have all to be related to the Arabic
term amal that generally is translated as governance, or dependence/
dependency, which resembles the 20th century theory of government
and binding. As explained by Owens:
228
otto zwartjes
The governor is said to govern the governed in some case or mode form
(irb):
For the nouns these forms are: u nominative (raf ); a accusative (nasb); i
genitive ( jarr);
For the verb only the imperfect verb shows mode inflection: u indicative
(raf ); a subjunctive (nasb); jussive ( jazm).(Owens 1988, 39).
Carter demonstrated in several publications (1991, 1993, 1994)dealing

with different grammatical theoriesthat the translation of technical
terms are in many cases inexact, problematic, anachronistic and can lead
to misconceptions.34 Carter argued that the basic meaning of the concept
of amal is different, stressing that the interpretation of Wei: amal
und regere: Es ist nicht recht verstndlich, wie man hier schwanken
kann. amala f kann ja doch gar nicht heien Gouverneur sein ber.
[. . .]. amala f ist ttig sein, arbeiten. The translation governor is
according to Wei an example of wishful thinking (Wunsch der Vater
des Gedankes), because the concept of governance was already widely
used in the Greco-Latin framework. A translation that brings us closer
to the original meaning an etwas arbeiten, auf etwas einwirken (Wei
1910, 384).
Particles (h urf )35 can also be defined as governors, since they can
govern cases. The subclasses of some particles can be defined in terms
of dependency, i.e. which nominal or verbal ending they govern. If a
particle (h arf ) governs a genitive, such particles are called h urf jarr
(Owens 1988, 10), for instance bi (by means of ), or min (from) etc.
In both the Compendio delos Rudimentos y Gramatica Araba of Lucas
Cauallero, as the Gramatica arabigoespaola, vulgar, y literal of Francisco Caes we find hispanicised forms of these subclasses of particles.36
At the beginning of the chapter dealing with the parts of speech, both
Caballero and Caes give us the tripartite division of the parts of speech,
which is an Arabic division, according to Caes (Pero los arabes las
34
The same happened when Romans translated Greek grammar. An illustrative
example is the term casus accusativus which is a wrong translation of the Greek term
ptosis aitiatike. It is not the anklagender Fall, but das von der Handlung Betroffene,
dasjenige, dem etwas gechieht. (L. Lersch: Die Sprachphilosophie der Alten, Bonn 1838
1841. vol. 2, 186, quoted in Carter 1993, 131).
35
H arf does not only mean particle, since it has in fact much more meanings, such
as edge, letter, sound, word. See for a detailed overview of the most important meanings
the first Appendix of Owens (1990, 245248).
36
We have not been able to consult a dictionary, which is particularly devoted to the
particles, the Diccionario de partculas rabes [s.a.], composed by Mariano Rizzi y Franceschi (18th century; BICRES III, no. 71).
229
[partes de la oracion] reducen a tres, que son: verbo, nombre, y particula

(h arf ). As a matter of fact, the authors follow this tripartite division, but
in the second chapter of the 4th treatise, which deals with the particles,
the author prefers to follow the Latin model:
Capitulo II. De las particulas separadas.
Muchos son los generos, que hay de estas particulas, y para evitar confusion,
las dividiremos al modo de los latinos, en adverbios, conjunciones, preposiciones, y interjeciones. (71).
Chapter II. About the separate [= not suffixed] articles.
The subcategories of these particles are many, and in order to avoid confusion, we divide them, according to the Latin manner, into adverbs, conjunctions, prepositions and interjections.
In the pages dealing with classical Arabic, a separate chapter following

the final chapter in syntax, Francisco Caes again prefers the Arabic
model. In the following quotation he deals with them after the chapter
on syntax describing classical Arabic:
Aqui es preciso advertir, que asi como el nombre se declina por la variacion
de las terminaciones, que tiene en los casos, como adelante se ver; igualmente el verbo en el modsareo se conjuga de tres maneras: por variarse
su terminacion en las personas. Esta variacion proviene, de que los arabes
anteponen al verbo en el modsareo dos generos de particulas, que se llaman
nasbantes y chazmantes. Se llaman nasbantes del verbo nasaba puso,
plant, fix, porque en fuerza de ellas el domma de la tercera radical del
verbo se pierde, y convertido en la mocion fatha, la fixan, y plantan sobre la
dicha radical, sin que padezca mutacion. Las chazmantes se llaman asi del
verbo jazama sec,37 cort, porque estas particulas le cortan la mocion la
tercera radical; y poniendole el signo secun, la dexan quiescente, liquida.
(Caes 1775, 109110).
Here, it is necessary to observe, that as the noun declines through the variation of its endings which they have in the cases, as we shall see below, so
on the same manner the verb in the mudri38 conjugates in three ways: by
varying the endings according to persons. This variation results from the
fact that the Arabs put before the verb in the mudri two classes of parti-
37
Probably seccion, since secar means to dry.

Usually translated as imperfect, literally the resembling verb (Baalbaki 2004,
XIII, 23), because they resemble the nouns, since both share the same declensional
vowels (irb) -u (subject and indicative mood; and -a which is direct object and
subjunctive mood). The nouns do not have zero endings (apocopate or jussive mood
in the verbal system), whereas the nouns have the -i ending (genitive), which is not
present in the verbal paradigm.
38
230
otto zwartjes
cles, which they call nasbantes and chazmantes [= which co-occur with
the nasb (a ending) and those which co-occur with the jazm ( ending)].
They are called nasbantes from the verb nasaba, to put up, to plant, to fix,
because through their force, the domma [damma] of the third radical of
the verb is cut off and converted into the fath a [a ending/motion]; they
fix this radical and settles down on it, while it does not suffer any change.
The chasmantes [those which co-occur with the zero-ending] are called so
from the verb jazama, to cut off, to truncate, because these particles cut off
the vowel [motion] from the third radical, and put on this the sign sukn
[motionless, vowelless] and leave them silent or liquid.
In Chapter IV, dealing with the noun, Caes gives us three classes of
particles, charrantes, chazmantes y nasbantes:
Particulas charrantes son unas preposiciones, que antepuestas al nombre le
colocan en el caso charro, genitivo (139) [. . .] Aqui se debe advertir, como
en arabe lo mismo es de decir particula nasbante, que en latin preposicion
de acusativo, y asi antecediendo al nombre le colocan en dicho caso, [. . .]
Estas particulas manera de los verbos, admiten afixs, y rigen los nombres,
colocando el sujeto en nasbo, acusativo, y el predicado en rafeo, nominativo, de suerte, que se viene hacer una permutacion del nominativo con
el acusativo . . . (143)
the particles which are charrantes are some prepositions, which placed
before the noun, put it in the jarr [i] ending, or genitive. Here it must be
observed, that as in Arabic it is the same to call them particular nasbante,
as in Latin, prepositions which combine with the accusative, and as such
they are placed before the noun, they put it in this so-called case. [. . .]
These particles, when combined with verbs, allow affixes, and govern the
nouns, placing the subject in nasb, or accusative and the predicate in raf ,
or nominative, so that there is a permutation from the nominative to the
accusative.
Caballero uses the same classification and terminology, although his

hispanicised forms are slightly different: particulas charrantes, nassuantes, and chesmantes. Although in Caes description, only classical
Arabic is commented on, Caballero also dedicates a paragraph on
the particulas Chesmantes, ojo, Vulgares (particles which correspond
with the optative Utinam and other Latin forms, such as quare, quia,
quoniam, etc.).
If we compare the use of Arabic terminology in the works of Caballero and Caes with those used by Erpenius, we can conclude that the
latter did not use Arabic terms so extensively. Worthy of mention are for
instance: Fatha, damma, kesra in the first book De elementis . . .. We find
a Latinized form of the term jazm: post gjezman constanter manent
231
(1620, 20), also used as the verb Gjezmare, gjezmant, or in the passive
form gjezmatur (47), verbum hamzatum (70). The term motiones is not
used as a translation of h arakt but it for the change a noun undergoes
if the feminine ending is added to the masculine form.39 Nunnatio is
used as well (141), but the Arabic terminology for inflectional endings
as used by the Spanish missionaries is not recorded.
3.3
Possible sources
The use of non-Western metalanguage in itself has its own tradition. Not
only Pedro de Alcal used non-Western terminology, but in Northern
Europe we see also that Hebrew grammars used Hebrew terminology
in an adapted form.40 Which sources could the Franciscans have used?
In the prologue of the grammar, Caes informs us that he completed
an eclectic grammar in agreement with the taste of everyone (que
sea del gusto de todos), using the most useful aspects (he procurado
aprovecharme de lo bueno que en ellas he visto ibid.) of earlier sources
and adding material from his own 16 years long experience:
He procurado con el mayor cuidado y desvelo leer, y releer para el ajuste de
esta, las gramaticas de Fr. Pedro de Alcal, de Fr. Felipe Guadaoli, de Toms
Erpenio, de Fr. Antonio de Aguila, de Fr. Agapito de Valle flammarum, de Fr.
Francisco Gonzalez, & c. Asimismo me he valido de un considerable numero
de manu-scritos, que me han franqueado gustosos algunos aficionados la
lengua arabe. Finalmente he aprovechado lo que me ense el estudio, y la
experiencia por espacio de diez y seis aos, que estube predicando, y confesando en arabe en las misiones del Asia (ibid.).
I have taken the greatest care to do my best to read and read over again the
grammars of Fr. Pedro de Alcal, Fr. Philip Guadagnoli, Thomas Erpenius,
Fr. Antonio de Aguila, Fr. Agapito de Valle Flammarum, and Fr. Francisco Gonzlez, and others. Likewise I have used a considerable amount
of manuscripts, which some aficionados of the Arabic language have
passed to me. Finally I have benefited from what study and experience
39
Motio est nominis Masculini in Foemininum converse; sitque additione terminationis foemininae (Chapter V).
40
Johannes Reuchlin, who published his Hebrew grammar one year after the publication of Alcals, introduces the verb dagessare: quando he uel aleph repellunt nun passiue significationis, dagessatur prima ut . . . (1974 [1506] Liber III, 590). See also Geiger
(1871, 129). The presence of Oriental elements in Western grammars is ofcourse not
only present when loans are used. Translations and mistranslations produced sometimes terms which are not longer recognized as from oriental origin. An example is the
history and development of the concept of radix (root).
232
otto zwartjes
have taught me over the period of sixteen years that I spent predicating
and confessing in Arabic in the missions of Asia.
To start with the first grammar mentioned, Pedro de Alcals Arte, we

can conclude immediately that this grammar has not been the source
for the terms used for the inflectional endings. Thomas Erpenius has
been apparently used, but we observed that Erpenius reduced the exotic
grammatical terms in his grammars,41 although he maintained them in
his Latin translation of Arabic treatises written by Ibn jurrm and
al-Jurjn, which could have been also the direct sources of Caes. Ibn
jurrm was born in Morocco in 12731274 and died in Fez in 1323.
He is the author of a grammatical compendium entitled Muqaddima
al-jurrmya Mabdi ilm al-Arab where he exposes the inflectional
system of Arabic, called irb. This treatise on syntax has been widely
used until the present day and it is one of the later works downstream the
long tradition starting with Sbawayhi. The Muqaddima was known in
Europe since the 16th century.42 This work has been printed for the first
time in Europe in 1592 (Medici, Rome). A translation by Peter Kirsten
(15771640), into Latin appeared in 1610 (Breslae, 1610), followed by
a translation of Erpenius (Leidae, 1617). In 1631, another translation
appeared by Thomas Obicini (15851632).43 The Kitb al-awmil al-mia
n-nahwiyya (Book of the Hundred Regentia) of Abd al-Qhir al-Jurjn
(died in ca. 1080) has been translated by Erpenius in 1617 and published
together with the Muqaddima: Grammatica Arabica dicta gjarvmia &
libellus centum regentium cum versione Latina & comentarijs (Leidae,
1617). As we shall demonstrate below, scholars in Rome, such as Philip
Guadagnoli, knew this work. Erpenius used in his translations of these
works loans from Arabic in his specified grammatical terminology. To
mention an example:
41
In this article, we quote from the Rudimenta (1620). For a more complete analysis,
the Grammatica Arabica, quinque libris methodice explicata a Thoma Erpenio (Leidae,
1613) and the Grammatica Arabica (Leidae, 1636) have to be taken into account.
42
See his article Ibn jurrm (Encylopedia of Islam. New Edition, 3, 697), and Ben
Cheneb (1927, 381382).
43
Thomas Obicini (15851632) was abbot of the Franciscan convent at Aleppo
between 1613 and 1619 and in 1621 he returned to Rome where he founded the college
at the St Peter Convent of Montorio where arabic was taught for the missionaries who
were being prepared to spread the faith in the East. He was responsible for the supervision of the type designs of Oriental types at the Propaganda Press.
233
partes autem eijus sunt Rafa, Nasba, Chafda and Gjezma, e quibus convenit
Nominibus Rafa, nasba and Chafda, non autem Gjezma: verbis vero Rafa,
nasba, and Gjezma: non autem Chafda (f. 11)
However, as has been demonstrated by Fck (1955, 68), Erpenius

replaced in his own grammars, when possible, Arabic terminology by
Latin equivalents, and his Rudimenta could not have been either the
source of inspiration of Caballero and Caes either. Erpenius maintained
the Arabic terminology of the vowels:
TABLE 6
Nomen
THE VOWEL SYSTEM ACCORDING TO THOMAS

ERPENIUS (1636, f. 16).4445
Figura Potestas
Phatha44
Ba
Dhamma
Bu
Kesre
Bi
Nunc a purum & clarum ut in amabam, nunc cum e mixtum, id est Graecum, ut multi id nunc pronuntiant45
Nunc u purum & clarum, nunc cum o mixtum, id est o obscurum
i simplex
Erpenius does not maintain the Arabic terms of the subclasses of the
particles, as he rendered them in a Latinized form in his translation of Ibn
jurrm, but obviously he attempted to fit them into the Latin model.46
If we compare the translation into Latin of Erpenius with Obicinis, we
can conclude that the latter also used the Arabic terms for inflectional
endings.47 Obicini firstly gives the Arabic term, written in the Arabic
script, then a translation is given followed by a description or paraphrase
with the purpose to explain the Arabic terms: ar-raf as elevatio, the
definition of an-nasb is accusativus, quasi patiens positum sub agente
(without translation), al-xaf is rendered as depressio, & amplectitur
44
In the left column the terms are also written in Arabic script, and in the second
column the Arabic letter b is given, together with its appropriate vocalisation.
45
Sometimes a pure and clear a as in amaba, other times mixed with an e i.e. as in
Greek as many pronounce it now, etc.
46
For instance, we find definitions such as: de syntaxi Particularum: Praepositiones
omnes tum separatae regunt genitivum f baitin . . ., instead of the terms nasbantes, etc.
Nevertheless, in his Rudimenta we still can find verbs as gjezmare (1620, f. 47).
47
Rafaa, & Nasba, & Chafda, & Gezma. At nominibus ex ijs (conueniunt). Rafaa, &
Nasba, & Chafda, non autem Gezma, Verbis ver ex eisdem, Rafaa, & Nasba, & Gezma,
non autem autem Chafdha (Obicini 1631, f. 3). Agapito Valle Flemmarum has almost
the same definition (1687, 194).
234
otto zwartjes
genitivum, dativum, & ablativum, quasi obliquus, and finally, al-jazm

as abscisio, idest casus privatio, quasi dictione cadentium scindat. This
means that Obicini decided to maintain Arabic terminology, and when
explained appropriately, the learner could take advantage of these Arabic
terms which were developed for describing the Arabic language.
And now lets come back to the concept of mil f. In Troupeau (1962),
the most important translations in this period is listed: In Kirstens and
Obicinis translation we have agens, whereas the first also uses efficiens
and the latter uses also regens. Erpenius uses both regens as operans.
As Michael Carter observed, we see until today that the basic meaning
has been neglected by many scholars:
Anderseits kann man fast an eine Verschwrung glauben, die Grundbedeutung des Terminus amal auch heute noch zu unterdrcken, um den
unbegrndeten Mythos zu verewigen, der arabische Begriff vom strukturellen Verhltnis zwischen Satzteilen sei identisch mit dem lateinischen,
d.h. eine Art Rektion. Da die Grundmetapher der lateinischen Rektion
eine durchaus hierarchische, senkrechte Einordnung der Satzteile voraussetzt, die arabische amal Operation, aber im Gegenteil eine waagerechte
Beziehung schildert, scheint die moderne Sprachwissenschaft nicht anerkennen zu wollen. (Carter 1993, 133).
Carter observed that Wei was an exception and demonstrated that

the translation by Kirsten unfortunately did not have any impact on
later translations or interpretations: Wie eine Stimme in der Wste
verbleibt noch der Aufsatz von J. Weiss. Die wrtliche und systemtreue
bersetzung operans von Kirsten hat sich leider nicht durchgesetzt
(Carter 1993, 134).
In the first place we have to add that Franciscus Martelottus and Philip
Guadagnoli (15961656), both neglected by Troupeau, also deserve our
attention in this context, since we find in their grammars of Arabic a
very precise analysis and translation of the concept. To start with the
latter: the concept of al-irb is translated as arabicatio and amal as
efficientia or operatio, and al-awmil as regentia:
Ratione autem, quo vnuquodque in suo vel Casu vel Modo & quacumque
dispositione locatur, quam Constructionem latini dicunt, Arabes interdum
al-irb, Arabicationem, proprius autem amalun efficientiam, seu operationem appellant. Nomen autem illud, seu Verbum, seu Particula, eiusmodi
operationem exercens, scilicet cuius vi aliquid in tali vel tali dispositione
locatur, dicitur, milun operans, regens. Locatum ver ex eius dicitur
mamlun operatum, rectum. Exinde dicuntur, al-awmilu regentia. Quibus omnibus notis, nihil superest Grammatico. Collegit autem quidam,
235
cognomine Giargianius, in libello Regentia omnia, quem propterea nominavit De Centum Regentibus. Regentia enim, vel sunt lafziyyatun explicita, vel
manawiyyatun implicita.48 (Guadagnoli 1642, f. 248).
However, the system according to which a constituent has to be placed
in the appropriate case or mode and on which convenient position has
to be placed, this system which the speakers of Latin call Constructio
[= syntax], is system is called al-irb by the Arabs which means arabicization, but they call this more appropriately amalun efficiens [an act
which produces a certain effect], or operation [an act caused by force of
an operator]. When a Noun, a Verb or a Particle produces such effects in
this manner, and when by force of this effect something has to be placed in
a certain disposition, we call this milun [producing a certain effect upon
something, to govern]. The element which has been effected is called
mamlun [the governed]. Consequently, al-awmilu are called regentia
[governers]. Although all this is well-known, nothing has been transmitted by The Grammarian,49 but someone with the name Giargianius has
collected in a booklet all the regentia, which can be subdivided in two
subclasses, the lafziyya (expressed) and the manwiyya (abstract).
As we see, Guadagnolis description is not only accurate and complete,

since the original Arabic is given and the literarly meaning had been
maintained, but he also mentions his source, which is the hundred
regentia written by al-Jurjn (Giargianius).50
Do we find traces of these translations of amal other grammars written in Latin of this period? The answer is positive. In chronological
order we shall summarize what other grammarians from Rome taught
us, starting with one of the earlier grammars that appeared in Rome
after the Medici translation of the work of Ibn jurrm, the Institutiones
of Franciscus Martelottus.
Martelottus not only mentions Erpenius in his prologue (1620, 38),51
but he also deals with methodology. Should the Arabic model with
48
Operators can be expressed (lafziyy), and abstract (manaw). The first class are
the particles or verbs or nouns that are either actually uttered or elided but understood,
while the latter are abstract causes that do not involve uttered or restored linguistic
elelements. (Baalbaki 2004, XV, 2358). This means that elided elements can produce
effects.
49
We could not identify this Grammarian.
50
Also in other paragraphs we see direct translations in the work of Guadagnoli,
which remain close to the Arabic original, such as ignoratum, for majhl, usually
erroneaously translated as passive, opposed to cognitum for the active (marf )
(Guadagnoli, 1642, f. 255).
51
Scripsit autem eleganter admodum de dictionibus hisce, quemadmodum & de litteris eruditissimus Orientalium linguarum in Leidensi Academia professor Thomas Erpenius uniusquam leuasset quoq. nobis huiis secundi libri labores partem (1620, 38).
236
otto zwartjes
its technical terms be maintained, or should they be abandoned and

replaced by Latin terminology? Martelottus decided to follow the traditional method when dealing with the single word classes (Priora
vero hic loci propri de singulis dictionibus singillatim ordine debito,
ac recta methodo explicanda sunt (1620, 37), but we can infer from his
preface, he obviously follows the Arabic model, and he does this explicitly.52 However, Martelottus does not leave the Arabic terminology out
of his volume:
Ubi omisso Arabicorum grammaticorum ordine Latinis admodum dissono,
in rebus pluribus nobiscum, quoad methodum conuenisse comperimus.
Caeterum quamvis praedicta Arabum methodo deflectentes, nostro nos
ordine Latinis magis consono procedamus, omnino grammaticalibus vocibus ubique utemur, eorumque ordines, ac procedendi rationes, divissiones,
ac series enucleabimus, ut faciliro cunctis ad eosdem perlegendos authores
pateat aditus. (Martelottus 1620, 38).
Whenever the difference between Arabic grammars and the Latin way of
construction was left out of consideration, we discovered that, with respect to the system of the language, Arab had many similarities with ours.
After all, to what extent we ourselves may differ from this language system
of the Arabs and are differing more and more as well from the Latin way
of construction, we should still use in our research the Arab terminology
and developing our knowledge we will explain the systems, the methods
of construction and concatenation, in order to make it easier for us all to
read the same authors.
Martelottuss methodology is in our eyes extremely modern. It tries to

bridge the gap between exo- and endo-grammaticalization. An eclectic
approach, combining the best elements of both traditions and their
corresponding technical terms is the best way to understand the Arabic
language. Martelottus also dedicates an entire chapter to inflection (De
irb, seu Inflexione), translated as Arabificatio, or Arabicatio (1620,
98). The concept of mil is translated again as operans and the author
quotes directly from the Arabic grammatical tradition:
irb apud Grmaticos tayru awkhir al-kalm il al-ikhtilf alawmil: variatio ultimorum, seu extremitatum dictionum, ob diversitatem
operantium (1620, 98).
52
In his grammar we find Arabic terminology extensively, together with the Latinized
form, for instance the traditional Arabic classification of the consonants: Chalchiiton,
lahuiiaton, sciagiariiaton, asliiaton, natiiaton, dhalchiiaton, sciaphahiiaton, liniiaton
(1620, 35).
237
Variation of the final parts or extremities of the words, caused by the diversity of the operators (by the different effects produced by the operators).
At the end of the 17th century, Agapito Valle Flemmarum treats the
particles in detail in his chapter entitled De syntaxi Particularum,
and particularly describes the effect they have as operators on the
inflectional system, translating al-h urf al-mila as:
1.
2.
3.
4.
5.
particlae operantes in nomen simplex Giarrantes,

et in nomen simplex nasbantes,
et in summam,
et in verbum nasbantes,
et operantes in verbum Giezmantes (1687, 268).
Another Franciscan, Antonio ab Aquila, also used Arabic terminology

for the particles, although we do not find the same definitions. In his
classification of the subclasses, we find a Latin terminology, such as
particulae coniunctae and particulae separatae, with subclasses, such
as particula interrogationis, iurandi, status in loco, ad determinationem,
optativi, conformandi, negationis, conditionalis (1650, Caput XXV,
388 ff.). However, he also uses Arabic terms, such as particula Gizemeh
(f. 403).
And finally, the question rises wether we find traces of the concept
of mil in the two Spanish grammars. The answer is again positive, as
we shall demonstrate. Lucas Caballero gives rules for classical Arabic
(gramtica literal) where particles have a certain effects:
De las particulas charrantes
Despues de las particulas que hemos puesto en el Arabo corresponden a
nuestras quatro ultimas partes de la oracion, de las quales particulas, la
mayor parte son literales; siguiendo el intento de dar algunas reglas de la
grammatica literal, quiero explicar algunos efectos que tienen las particulas puestas o otras que a ellas siguen . . . (Cabellero 1709, f. 7r.; emphasis is
mine)
After the particles we have put in Arabic, four final parts of speech correspond with ours, whose major part are literary utterances: following our
intention to give some rules of literary grammar, I want to explain certain effects these particles have when placed, or other which follow after
them.
In the grammar of Caes, we find exactly the same term, hispanicised

as operacion:
238
otto zwartjes
. . . quando se les une la particula m quedan absolutamente privadas de su
operacion, y regimen, de suerte que pierden la fuerza que tenian de colocar
el sujeto en el caso nasbo y se queda en el rafeo, nominativo, como ciertamente que escrivir Pedro, innam sayaktubu Butrus (Caes 1775, 144;
emphasis is mine).
. . . when combined with the particle m they remain absolutely deprived of
their operation and government so that they loose their force which they
had before to put the nasb (a inflection, or direct object) and it remains
in the raf (u ending, or subject) or nominative, as in certainly to write to
Peter innam sayaktubu Batrus.
To sum up, we have found the following authors as possible sources of

the two Franciscan grammarians of Damascene Arabic: Ibn jurrm
and al-Jurjn (through translations of the Medici edition, Erpenius,
Obicini and maybe also Kirsten) as the main Eastern sources. Western
sources mentioned by name are Erpenius, Golius (mentioned by Caes),
Agapito Valle and Guadagnoli. It has to be observed that missionary
grammarians in Rome were familiar with some of the most important
Arabic authors. The name of Al-Xall b. Ahmad al Farhid (died in 780)
who codified and established a system of 15 meters has been mentioned
by Guadagnoli. Since the grammars of Caballero and Caes do not have
a final chapter on prosody, they did not deel the need to use this source.
Another prominent pioneer, who established the foundations of Arabic
grammar is of course Sbawayhi, whose work became known in the
West through translations of Jahn. It is surprising that we find already in
1620 his name in the grammar of Martelottus, which is evidence for the
fact that missionary-grammarians in Rome knew who he was. Dealing
with the verba ternaria, Martelottus explains that the forma masdari is
irregular, and in that context he mentions Sbawayhi:
In verbis ternarijs, ut iam diximus, forma masdari irregularis est, omnes
autem quidem Author Arabicus, nomine Sibauai ad 32. reuocauit, videlicet
(1620, 213).
In the ternary verbs, as we said before, the masdar-form is irregular;
indeed, the Arabic author with the name Sbawayhi reduced all these to
32.
The same author, Martelottus, also mentions another Arabic source in

his chapter dealing with the Constructio Particularum where he divides
the particulas operantes in five subclasses, in agreement with an Arabic
work called Lucerna, or Al-Misbh , the particulae operantes in nomen
simplex giarrtes, particulae operantes in nomen simplex nesbantes,
in verbum nasbantes, in verbum gezmantes, and the original text in
239
Arabic script is given in the same table as well. Although Martelottus

does not give the name of the author, we think this work is probably
the treatise written by al-Muta rriz, entitled Al-Misbh f ilm an-nahw.
Al-Mutarriz (11441213) compiled this treatise which became a textbook in the madrasas of the East. The Misbh itself wa sbased on three
small grammatical monographs of al-Jurjn (Lichtenstdter 1936, 847
and Sellheim (EI [New edition] 7, 773).
4. Conclusion
Summarizing, we can conclude that Pedro de Alcals use of Arabic
terminology seems to be unsystematic and the reasons why he used
them remain unclear, particularly when he uses the Arabic names for
the cases. They do not reflect the Arabic inflectional endings and his
model was obviously Latin grammar. The use of the term damr is an
exception, since the suffixed pronouns can be used in a different wayattached to verbs, nouns and particlescompared to Latin. Pedro de
Alcals mnemonic terms of the vowels stand alone, and we do not find
any use of them in other works and in his dictionary we do not find the
terms fath a, kasra, and damma. Although we find in Alcals dictionary
the translation obrar for the Arabic root amala there are no traces of
Arabic theory concerning operators or government, related to this
term.
The earliest translations of al-Jurjn and Ibn jurrm are without any
doubt an important milestone for the development of the study of nonWestern grammatical theory in the West and probably for some of them
a real new discovery, which can serve as an enrichment of the Western system, as Martelottus postulates. Direct influence of these works
can be found in the grammars of Martelottus, Ab Aquila, Agapito and
Guadagnoli, and the Franciscans in Damascus continued this tradition.
Although Caballero and Caes were not the pioneers themselves, they
were probably the first grammarians who introduced extensively Arabic
terminology in the Spanish metalanguage, as we have demonstrated.
In Northern Europe, however, we see a different approach. Erpenius did not aim at orientalising Western grammatical terminology,
except for the terms for the vowels and some other terms. The concept
of mil has followed its own course in grammatical theory and in the
20th century it was absorbed by anachronistic terms such as government and dependency, as Carter demonstrated. Nevertheless, we have
240
otto zwartjes
demonstrated that the Franciscans were totally aware of the right connotations of mil although we have to admit that sometimes they use
gobierno or regimen as a synonym for efficiens or operans. Guadagnoli and Martelottus gave us without any doubt the most detailed
analysis, and probably Caballero and Caes have been inspired by their
works. It was surprising that not only works of Ibn jurrm, al-Jurjn
and al-Muta rriz are mentioned by name by some of the grammarians
working in Rome, but even Sbawayhi is mentioned by name in this
relatively early period (1620).
Missionary linguistics in Rome, particularly the achievements of
scholars and teachers who published grammars in the seventeenth
century at the Polyglot Press of the Sacra Congregatio de Propaganda
Fide deserve to be studied more in detail in future. Particularly those
authors who tried to combine exo- and endo-grammatical terminology and approaches have been innovative. How the learners of Arabic
appreciated this bridging approach is another question. Many scholars
preferred in their teaching curriculum the more Latin-based grammar
of Erpenius and his work was without any doubt a great success during many centuries. However, I agree with Martelottus that there is no
reason to postulate that the Arabs differ from our system. We can also
say, we ourselves may differ from this Arabic language system. While
using their own terminology, which has been developed for their own
linguistic phenomena, we will make progress in the understanding of
not only the language but of the linguistic model as well.
5. References
5.1
Primary sources
Alcal, Pedro de. [Before 5 Feb. 1505a]. Arte para ligeramente saber la lengua arauiga.
Salamanca: Juan Varela.
. [Before 5 Feb. 1505b]. Vocabulista arauigo en letra castellana. Salamanca: Juan
Varela.
. [ca. 1506a]. Arte para ligera mente saber la lengua arauiga emendada y aadida y
Segunda mente imprimida. Salamanca: Juan Varela.
. [ca. 1506b]. Vocabulista arauigo en letra castellana. Salamanca: Juan Varela. [Biblioteca Nacional, R 31638.] Re-edition: Paul de Lagarde. 1883. Petri Hispani de lingua arabica libri duo. Gottingae: In aedibus Dieterichianis Arnoldi Hoyer. Reprint:
Osnabrck: Otto Zeller, 1971.
Aquila, Antonio Ab. 1650. Arabicae linguae novae, et methodicae institutiones. Non ad
vulgaris dumtaxat Idiomatis; sed etiam ad Grammaticae doctrinalis intelligentiam, per
Annotationes in Capitum Appendicibus suffixas, accomodatae. Authore F. Antonio ab
Aquila Or. Min. Sancti Francisci strict. Obser. Theologo, atque in Collegio Sancti Petri
241
Martis Aurei Sacra Congregatione de propaganda Fide Arabicae linguae deputato

lectore. Opus tum omnibus Arabicae linguae studiosis, tum potissumum Apostolicis
Viris, per Asiam & Africam Fidem propagaturis, utile & necessarium. Romae: Typis
Sac. Cong. De Prop. Fide.
Caballero, Lucas. [Cauallero] 1709. Compendio de los Rudimentos y Gramatica Araba en
que se da suficiente notizia de la lengua Vernacula o Vulgar y algunas Reglas de la literal
Iustamente P.M.R.F. Bernardino Gonzalez hijo de la Pro de la Concepzion en Espaa
Lector Jubilado en Arabo y Misionero Apostolico del Oriente y recoplada por el Re.do P.
Fr. Mo Apostolico hijo de la Proa de los Angeles Lector actual Arabo en el colegio de
Damasco. Ms. Rogge Library, Handskriftssamlingen, Sweden.
Caes, Francisco. 1775. Gramatica arabigoespaola, vulgar, y literal. Con un diccionario
arabigo-espaol, en que se ponen las voces mas usuales para una conversacion familiar,
con el Texto de la Doctrina Cristiana en el idioma arabigo. Madrid: En la Imprenta de
Don Antonio Perez de Soto.
Dombay, Franciscus de. 1800. Grammatica linguae Mauro-Arabicae juxta vernaculi idiomatic usum, accessit vocabularium Latino-Arabicum. Vindobonae: apud Camesina.
Erpenius, Thomas. 1617. Grammatica Arabica dicta Gjarumia, & Libellus centum regentium cum versione latina & commentariis Thomae Erpenii. Leidae: Ex typographia
Erpeniana Linguarum Orientalium.
. 1620. Rudimenta lingv arabicae. Accedunt ejusdem Praxis Grammatica; & Consilium de studio Arabico feliciter instituendo. Leidae: Ex Typographia Auctoris.
Germanus [of Silesia], Dominicus. 1639. Fabrica linguae Arabicae cum interpretatione
Latina & italica, accomodata ad usum linguae vulgaris, & scripturalis. Romae: Typis
Sac. Congreg. De Prop. Fide.
. 1640. Fabrica Arabica copiosioribus impensis atque vberiori structura. Romae.
Gonzlez, Bernardino 2005 [c. 1705]. Intrprete Arbico. Edicin y estudio preliminar
por Ramn Lourido Daz. 2 vols. Madrid: Real Academia de la Historia/Ministerio de
Asuntos Exteriores y de Cooperacin.
. 2005 [c. 1705]. Eptome de la gramtica arbiga. Edicin y estudio preliminar por
Ramn Lourido Daz. 2 vols. Madrid: Real Academia de la Historia / Ministerio de
Asuntos Exteriores y de Cooperacin.
Guadagnoli, Philippi. 1642. Breves Arabicae linguae Institutiones. Romae: Ex Typographia
Sac. Congregationis de Propaganda Fide.
Martelottus (= Martelotti), Franciscus. 1620. Institutiones Linguae Arabicae Tribus Libris
distributae. In quibus uberrime quaecumque ad litteras, Dictiones & Orationem attinent, explicantur. Authore P. Francisco Martelotto Martinensi, Sacerdote, Theologo,
Clericorum Regularium Minorum. Romae: Excudebat Stepahsnus Paulinus.
Nepos, Ferdinandus. 2000.[c. 14681485]. Materies. (Ms. See Codoer).
Obicini, Franciscus. 1631. Grammatica Arabica Agrumia appellate. Cum versione Latina
ac dilucida expositione. Adm. R.P.F. Thomae Obicini Noniensis, Diae. Nouariau, ord.
Minorum Theologi, ac linguarum Orientalium in Collegio S. Petri Montis Aurei, de
mandato Sacrae Congregationis Fidei propaganda Magistri. Romae: Typis Sac. Congregationis de Propaganda Fide.
Oyanguren de Santa Ins, Melchor. 1738. Arte de la lengua japona. Dividido en quatro
libros segun el Arte de Nebrixa, con algunas voces proprias de la escritura, y otras de
los lenguages de Ximo, y del Cami, y con algunas perifrasis, y figuras. Mxico: Joseph
Bernardo de Hogal.
. 1742. Tagalysmo elucidado, y reducido (en lo possible) la Latinidad de Nebrija, con
su Syntaxis, Tropos, Prosodia, Passiones, &c. y con la allusion, que en su uso, y composicion tiene con el Dialecto Chinico Mandarin, con las Lenguas Hebrea, y Griega. Mxico:
D. Francisco Xavier Snchez, en la Calle de San Francisco.
Reuchlin, Johannes. 1974[1506]. De rudimentis hebraicis libri III. Pforzheim. Reprint:
Hildesheim New York: Georg Olms Verlag.
242
otto zwartjes
Rodrigues, Ioo. (Rodriguez) 16041608. Arte da Lingoa de Iapam composta pello Padre
Portugues da Copanhia de Iesv diuidida em tres Livros. Com Licena do Ordinario, e Svperiores. Nagasaqui: no Collegio de Iapo da Companhia de Iesv Anno.
1604 (-1608 colofon.)
Rodrigves, Ioam. 1620. Arte Breve da Lingoa Iapoa tirada da Arte Grande da mesma
lingoa, pera os que comecam a aprender os primeiros principios della. Pello Padre Ioam
Rodrigvez da Companhia de Iesv Portugues do Bispado de Lamego. Diuidida em tres
Livros. Com Licenca do Ordinario, & Superiores. Amacao: no Collegio da Madre de
Deos da Companhia de Iesv.
Roth, Heinrich. 1988[ca. 1662]. Grammatica Linguae Sanscretanae Brachmanum Indiae
Orientalis. Facsimile edition: Arnulf Camps & Jean-Claude Muller: The Sanskrit
Grammar and Manuscripts of Father Heinrich Roth S.J. (16201668). Facsimile edition
of Biblioteca Nazionale, Rome, Mss.Or. 171 and 172. Leiden: E. J. Brill.
Valle Flemmarum, Agapiti . 1687. Flores Grammaticales arabici idiomatis collecti ex
optimis quibusque Grammaticis nec non pluribus Arabum monumentis, ad quam maximam fieri potuit brevitatem atque ordinem revocati: Studio, & Labore Fr. Agapiti
Valle Flemmarum Ordinis Minorum S. Francisci reformatorum Provinciae Tridentinae,
in Semionario Patavino lectoris. Opus omnibus Arabicae Linguae Studiosis perutile, &
necessarium. Patavii: Ex Typographia Seminarii Opera Augustini Candiani.
Ziegenbalg, Bartholomus. 1985 [1716]. Grammatica Damulica. Qu Per Varia Paradigmata, Regulas & Necessarium Vocabulorum Apparatum, Viam Brevissimam Monstrat, Qua Lingua Damulica Seu Malabarica, Qu Inter Indos Orientales in Usu Est, &
Hucusque In Europa Incognita Fuit, Facile Disci Possit: In Usum Eorum Qui Hoc Tempore Gentes Illas Ab Idololatria Ad Cultum Veri Dei, Salutemque ternam Evangelio
Christi Perducere Cupiunt . . . Hal Saxonum: Litteris & Impensis Orphanotrophei.
Ed. Burchard Brentjes & Karl Gallus, Halle: Martin-Luther-Universitt Halle-Wittenberg, 1985.
5.2
Secondary sources
Baalbaki, Ramzi. 2004. Grammarians and Grammatical Theory in the Medieval Arabic
Tradition. Aldershot: Ashgate Variorum.
Ben Cheneb, Mohammed 1927. Ibn djurrm. Enzyklopdie des Islm. 2.381382.
Leiden Leipzig: Brill Harrassowitz.
BICRES (see Niederehe).
Carter, Michael G. 1991. Review Owens 1988. Journal of the American Oriental Society,
111.2: 395397.
. 1993. Probleme bei der bersetzung von Fachsprache am Beispiel des Arabischen. In Frank, Armin Paul, Kurt-Jrgen Maa, Fritz Paul, und Horst Turk, eds.
bersetzen, Verstehen, Brcken bauen. Geisteswissenschaftliches und literarisches
bersetzen im internationalen Kulturaustausch. Berlin: Erich Schmidt., 1.130141.
. 1994. Review Owens 1990. Journal of the American Oriental Society, 114.3: 472
475.
Codoer, Carmen. 2000. Gramticas Latinas de transicin. Juan de Pastrana, Fernando
Nepote. Introduccin y edicin crtica. Salamanca: Ediciones Universidad de Salamanca.
Corriente, Federico. 1988. El lxico rabe andalus segn P. de Alcal. (Ordenado por
races, corregido, anotado y fonmicamente interpretado. Madrid: Universidad Complutense de Madrid.
Cowan, William. 1981. Arabic grammatical terminology in Pedro de Alcal. Historiographia Linguistica, 8.2/3:357363.
Dannenfeldt, Karl H. 1955. The Renaissance Humanists and the Knowledge of Arabic.
Studies in the Renaissance, 2, 96117.
243
Escavy, R., M. Hernndez Terrs, A. Roldn, eds. 1994. Actas del Congreso Internacional de Historiografa Lingstica. Nebrija V-Centenario 14921992. Murcia, 14 abril
1992. Murcia: El Taller, vol. III: Nebrija y otros temas de historiografa lingstica.
Frank, Armin Paul, Kurt-Jrgen Maa, Fritz Paul, und Horst Turk, eds. 1993. bersetzen,
Verstehen, Brcken bauen. Geisteswissenschaftliches und literarisches bersetzen im
internationalen Kulturaustausch. Berlin: Erich Schmidt.
Fck, Johann. 1955. Die arabischen Studien in Europa bis in den Anfang des 20. Jahrhunderts. Leipzig: Otto Harrassowitz.
Geiger, Ludwig. 1871. Johann Reuchlin. Sein Leben und seine Werke. Leipzig: Duncker
and Humblot.
Hauschild, Richard. 1988. Notes on the Content of the Three Manuscripts of Heinrich
Roth. Arnulf Camps and Jean-Claude Muller: The Sanskrit Grammar and Manuscripts of Father Heinrich Roth S.J. (16201668). Facsimile edition of Biblioteca Nazionale, Rome, Mss.Or. 171 and 172. Leiden: E.J. Brill.
James, Gregory. 2007. The terminology of declension in Early Missionary Grammars
of Tamil. In Zwartjes, Otto, Gregory James and Emilio Ridruejo, eds., Missionary
Linguistics III/Lingstica misionera III. Morphology and Syntax. Selected papers from
the IIIrd and IVth International Conferences on Missionary Linguistics, Hong Kong/
Macau/Valladolid, 167190.
Juynboll, W.M.C. 1931. Zeventiende-eeuwsche beoefenaars van het Arabisch in Nederland. Utrecht: Kemink en Zoon.
Kerkhof, Maxim, Hugo de Schepper, Otto Zwartjes, eds. 1993 Espaa: Ruptura 1492?.
Amsterdam: Rodopi.
Killean, Carolyn G. 1984. The development of Western Grammars of Arabic. Journal
of Near Eastern Studies, 43.3: 223230.
Law, Vivian. 2002. The History of Linguistics in Europe. From Plato to 1600. Cambridge:
Cambridge University Press.
Lentin, Jerome. 1997. Recherches sur lhistoire de la langue arabe au Proche-Orient
lpoque moderne. Thse de Doctorat dtat, Universit de Paris III.
Lichtenstdter, Ilse. Al-Mutarriz. Enzyklopdie des Islm, 3:847. Also New edition:
R. Sellheim, 7:773.
Lourido Daz, Ramn. 2005. Estudio preliminar. See: Gonzlez (primary sources).
Maruyama, Toru. 2004. Linguistic Studies by Portuguese Jesuits in Sixteenth and Seventeenth Century Japan. In Zwartjes, Otto and Even Hovdhaugen, eds. Missionary
Linguistics [I]/Lingstica misionera [I]. Selected papers from the First International
Conference on Missionary Linguistics, Oslo, 1316 March 2003. Amsterdam: John Benjamins. 141160.
Monroe, James. T. 1970. Islam and the Arabs in Spanish Scholarship (Sixteenth century to
the present). Leiden: Brill.
Mrner, A.R. and Magnus Mrner. 2001. Spanien i svenska arkiv. Skrifter utgivna av
Riksarkivet, 16. Stockholm: Riksarkiv, 6061.
Niederehe, Hans-Josef. 2005. Bibliografa cronolgica de la lingstica, la gramtica y la
lexicografa del espaol (BICRES III). Desde el ao 1701 hasta el ao 1800. Amsterdam: John Benjamins.
Owens, Jonathan. 1988. The Foundations of Grammar. An introduction to Medieval Arabic Grammatical Theory. Amsterdam: John Benjamins.
. 1990. Early Arabic Grammatical Theory. Heterogeneity and Standardization.
Amsterdam: John Benjamins.
Schnurrer, Christianus Fridericus de. 1811. Bibliotheca Arabica. Halae ad Salam: I.C.
Hendelii.
Smith-Stark, Thomas. 2005. Phonological description in New Spain. In Zwartjes, Otto
and Cristina Altman, eds.. Missionary Linguistics II/Lingstica misionera II. Orthography and Phonology. Selected Papers from the Second International Conference on
Missionary Linguistics, So Paulo, 1013 March 2004. Amsterdam: John Benjamins,
164.
244
otto zwartjes
Troupeau, G. 1962. Trois traductions latines de la Muqaddima dIbn urrm. tudes

dOrientalisme ddies la mmoire de Lvi-Provenal. Paris: Maisonneuve et Larose,
1.359365.
Versteegh, Kees. 1997. Landmarks in Linguistic Thought III. The Arabic Linguistic Tradition. London: Routledge.
Wei, Josef. 1910. Die Arabische Nationalgrammatik und die Lateiner. Zeitschrift der
deutschen Morgenlndischen Gesellschaft, 64:349390.
Zwartjes, Otto. 1993. El artculo en las gramticas pioneras de Nebrija y Alcal y las
gramticas grecolatinas. In Kerkhof, Maxim, Hugo de Schepper, Otto Zwartjes, eds.
Espaa: Ruptura 1492?. Amsterdam: Rodopi, 261286.
. 1994. Tradicin e innovacin en las gramticas pioneras de Antonio de Nebrija
y Pedro de Alcal: la categora gramatical del pronombre. In Escavy, R., M. Hernndez Terrs, A. Roldn, eds. 1994. Actas del Congreso Internacional de Historiografa
Lingstica. Nebrija V-Centenario 14921992. Murcia, 14 abril 1992. Murcia: El
Taller, 3:651665.
. 2007. Agreement asymmetry in Arabic according to Spanish missionary grammarians from Damascus (XVIIIth century). In Zwartjes, James, Ridruejo, eds., forthcoming. Missionary Linguistics III/Lingstica misionera III. Morphology and Syntax.
Selected papers from the IIIrd and IVth International Conferences on Missionary Linguistics, Hong Kong/ Macau/Valladolid, 273303.
, and Even Hovdhaugen, eds. 2004. Missionary Linguistics [I]/Lingstica misionera
[I]. Selected papers from the First International Conference on Missionary Linguistics,
Oslo, 1316 March 2003. Amsterdam: John Benjamins.
, and Cristina Altman, eds. 2005. Missionary Linguistics II/Lingstica misionera II.
Orthography and Phonology. Selected Papers from the Second International Conference
on Missionary Linguistics, So Paulo, 1013 March 2004. Amsterdam: John Benjamins.
, Gregory James and Emilio Ridruejo, eds. 2007. Missionary Linguistics III/Lingstica misionera III. Morphology and Syntax. Selected papers from the IIIrd and IVth
International Conferences on Missionary Linguistics, Hong Kong/ Macau 1215 March
2005. Valladolid, 811 March 2006. Amsterdam: J. Benjamins.
LINGUISTICS
THE LINGUISTIC ANALYSIS AND RULES OF

PAUSE IN ARABIC
Salman H. Al-Ani
Indiana University, Bloomington
1. Introduction
The Arabic term waqf pause is a grammatical concept. It is a verbal
noun derived from the verb waqafa that could mean to stop, to come
to a standstill or to pause. This verb is used both in transitive and
intransitive form. The term waqf is from the transitive verb waqafa.
Another verbal noun wuqf standing still is derived from the intransitive
verb waqafa (Farrj 2001, 1314). The term waqf is more commonly used
and has a significant point of reference in both linguistic and religious
connotations that mean to hold something at a pause or stop. This term
waqf and its plural form awqf mean an endowment in Islamic law
which signifies the dedication of property or land that cannot be sold.
The chain of speech utterances may be divided into spoken group
events. This is usually regulated by the phonological, syntactic and
semantic rules of the language. Normally words in the chain of the
utterances of speech in prepausal forms either end with vowels or consonants. In Arabic vowels at the end of a speech event in prejunctional
state are a sign of continuity and consonants signal waqf (H assn 1973,
270271). When speaking or reading aloud with the prepausal mode
almost all short vowels at the end of words, phrases and sentences will
be dropped. This linguistic phenomenon, in Arabic, is referred to as
pause forms.
Oral reading and recitation have been highly emphasized in the
Arab-Islamic tradition. Poets in Arabia in the pre-Islamic period composed their odes to be recited and the members of the poets tribe in
turn memorized and recited these poems whenever the opportunity
availed itself for them to do so. When the Qurn was first revealed it
emphasized recitation. The very first verse states recite or read aloud,
iqra. The tradition of oral reading, in general, is engrained in the hearts
and minds of most Muslims and Arabs.
The Arabic sources that dealt with this topic were focusing on the
rules of waqf in Classical Arabic with special emphasis on the rules of
248
salman h. alani
recitation in the Qurn. This aspect of waqf will not be covered in this
paper. The emphasis is on the phenomenon of waqf with the explanation of the basic linguistic features of waqf of Modern Standard Arabic.
Previous research on waqf, in western languages, is rather limited. There
are four studies that made a basic contribution to waqf or pause. The following is brief account of these studies.
1.1
F.T. Mitchell (1990)
Mitchell wrote that pause . . . relate(s) first and foremost to words that
occur in apocopated form before pause, that is predominantly, final in
the phrase and sentence. He added that: In discourse, it should not
be expected that pause will always neatly correlate with grammatical
phrasal divisions nor that the most appropriate divisions are always
observed by speakers or readers, but the principle should always be
followed that, wherever a pause is made, the preceding word should be
pronounced in its pausal form.
He postulated the following rules: (i) a final short vowel, for example
those of case, tense and mood is omitted; (ii) the sign of the indefinite,
i.e., in nunation (-n) is omitted together with the preceding vowel, in
the nominative (-un) and genitive (-in) cases of nouns and adjectives;
(iii) accusative (-an) may be replaced by a long fath a (); (iv) the feminine singular and unit ending t marbtah is replaced by /-h/ in all
three cases (Mitchell 1990, 99100).
This brief account of waqf by Mitchell covers the basic rules of pause
of al-arabiyyah. It is intended to aid the learner of Arabic to pronounce
and read aloud correctly. He used a short transliterated Arabic passage
read aloud by what he called . . . a speaker trained in the high Classical tradition to illustrate the pause form rules. The comments and the
pause rules almost mirror the traditional statements of the Arab grammarians on waqf (Mitchell 1990, 100101).
1.2
Clive Holes (2004)
Holes defined waqf as: pause is defined (rather vaguely) as an audible

break in delivery (Holes 2004, 63). The focus of his analysis is on the
spoken Modern Standard Arabic. The illustrative examples he used were
taken from Syrian radio broadcast and television news and speeches of
a former Egyptian leader, Gamal Abdul Nasser. He (Holes) said that
linguistic analysis and rules of pause in arabic
249
pauses depend on what is said in the spoken Modern Standard Arabic,

the audience and the speakers intentions.
The conditions where prepause occur are: (i) final short vowels that
appear in prejunctural position are omitted; (ii) the feminine gender
morpheme t marbtah in certain nouns and adjectives in prepause,
together with the /-u, -i, and -a/ are deleted. He added that /-ah/ is pronounced as /-a/; (iii) The inflectional suffixes /-un/ nominative, /-an/
accusative and /-in/ genitive were discussed individually. The /-un/ is
omitted in prepause. The nature of the accusative /-an/ depends on its
grammatical status. When it is used as an adverbial marker as in kullyan
completely the majority of speakers retain it, however few realize it as
/-/. When /-an/ is used as marker of an object of nouns or an adjective
that agrees with one, it is sometimes retained and sometimes omitted.
Holes analysis of the inflectional suffix /-in/ is very interesting. He made
a distinction of /-in/ when it is used as a marker of a noun in the genitive
case and /-in/ as a marker of an attributive adjective in agreement with
a genitive head noun. The suffix /-in/ usually is omitted, however, Holes
gave some examples where it is retained (Holes 2004, 6368).
1.3
Muh ammad Farrj (2004)
The focus of Farrjs long Arabic article is on the traditional grammatical

concept of waqf. The emphasis of this article was on the rules of waqf in
Classical Arabic Grammar and its application for the recitation of the
Qurn. He stated that there are two basic features of waqf. These are
wajh manner and mah all place. The purpose of mah alli l-waqf place of
pause is to regulate the speech event and leads to improving its division
where juncture or pause may occur. The wajh, plural awjuh, provides
the speaker or the reader with the guide for waqf so the speech will
appear harmonized in its utterances. He added that waqf has a wazfa
task which serves for clarity of pronunciation and meaning that help in
understanding of what is being read or spoken (Farrj 2004, 1223).
1.4
Tamm H assn (1973)
H assn stated that waqf through its various means by its nature is a
mifsal, separator of speech where it is possible that the chain of speech
may broken into spoken groups. Every one of these, when its meaning
is complete, is considered a speech event. However, if the speech is not
250
salman h. alani
complete then the speech event may consist of more than one event
(H assn 1973, 270).
2. The Arabic linguistic analysis of and rules for waqf

2.1
The waqf status of the t marbtah
One of the most common modes of waqf is applied to words that end with
a t marbtah. This t marbtah is considered to be a morpheme that
primarily marks the feminine endings of nouns and adjectives. However
there are some feminine nouns and adjectives that do not always end
in t marbtah. In Arabic script some feminine nouns and adjectives
may be written with t mabstah. This is the regular t that appears at
the end of words. This is especially true in the script of the Qurn. In
fact we find sometimes the same word written with t marbtah and in
other contexts it is written with a t mabstah. The word rah ma mercy
for example is written in either t marbtah or t mabstah.
There are nouns and adjectives that end in a t marbtah that have a
masculine meaning. Proper names such as H amzah, Talh ah, and several
others are written with t marbtah. The rules of waqf apply to them as
they apply to the feminine nouns and adjectives. Therefore it is not the
gender of the word but rather the form of the word that determines the
application of the waqf rules.
The t marbtah in nouns and adjectives that appears in prepause
form is deleted and replaced by /-h/. This deletion, of course, includes
the vowels: /-i/, -/u/ and /-a/ in the definite and /-in/, /-un/ and /-an/
in the indefinite. The deletion takes place in all three cases: nominative,
accusative and genitive. With prejunctural, on the other hand, the t
marbtah and the vowels both in the definite and indefinite are retained.
This is illustrated with the word madrasah school as in the prejunctural
forms below:
Indefinite
Definite
Nominative
Genitive
Accusative
madrasa-tun
al-madrasa-tu
madrasa-tin
al-madrasa-ti
madrasa-tan
al-madrasa-ta
251
The case endings in the indefinite /-tun, -tin, -tan/ and the definite /-tu,
-ti, -ta/ all will be deleted in the prepausal forms. The word madrasah
school and the word al-madrasah the school both in definite and
indefinite forms will have the same endings in the prepausal forms. The
reason is that the /-h/, placed between slashes, is sometimes weakened to
the point that one really cannot even hear it. This is especially the case of
Modern Standard Arabic, read aloud or spoken by radio and television
broadcasters. I have examined and analyzed the speech segments of
several announcers and observed both the dropping off of the /-h/ and
the retaining of it. In careful delivered speeches especially of religious
nature the /-h/ is almost always retained.
2.2
The deletion of final short vowels
The final short vowels, in waqf, are deleted. This deals with al-h arakti
l-irbiyyah, the vowel marks that signal case endings, tense and mood.
The following rules apply to nouns and adjectives derived from strong
verb roots:
(a) The case endings in the definite nouns are indicated by /-u/ in the
nominative case, /-i/ in the genitive case, and /-a/ in the accusative
case. All of these short vowels that mark the case endings are omitted in waqf. The following sentences illustrate the prejunctural and
prepausal forms:
Prejunctural forms
Prepausal forms
Meaning
ja-l-walad+u
mina-l-walad+i
raaytu-l-walad+a
ja-l-walad
mina-l-walad
raaytu-l-walad
the boy came

from the boy
I saw the boy
(b) When the indefinite nouns and adjectives that end in nunation,
the sound /-n/ pronounced but not written. The case endings in the
indefinite nouns and adjectives are indicated by /-un/ in the nominative case, /-in/ in the genitive case and /-an/ in the accusative
case. Both nominative and genitive endings are omitted in prepausal forms. However the accusative case maker is changed into an
alif /-/ that requires a special treatment. The word walad boy in
the following sentences illustrates both prejunctural and prepausal
forms.
252
salman h. alani
Prejunctural forms
Prepausal forms
Meaning
ja walad+un
min walad+in
raaytu walad+an
ja walad
min walad
raaytu walad+aa
a boy came
from a boy
I saw a boy
(c) The indefinite accusative case ending is /-an/. Also certain types of
adverbs end in /-an/. This ending is replaced by /-/ in prepausal
forms as illustrated in the table above. In the case of adverbs some
speakers retain the nunation while others replaced it with /-/.
However, in prepausal and words in isolation may be retained but
oftentimes it changes to /-/ following the normal rules of waqf. The
word id an meaning if is written orthographically either with nn
/n/ or with the alif tanwn. In prepausal form pronounced as /-/.
(d) Tense and mood in verb-vowel endings are omitted.
2.3
Prejunctural forms
Prepausal forms
Meaning
katab+a
yaktub+u
lan yaktub+a
katab
yaktub
lan yktub
he wrote
he writes
he will not write
Words that end in long vowels
There are two main grammatical categories of words that end in long
vowels. These are called in traditional Arabic grammar as the al-manqs
defective as in the word al-wd the valley and al-maqsr shortened
as in the word al-fat the youth.
1. The rules of waqf on words that end in the long vowel /-/ such as
al-muh m the lawyer which is derived from a finally weak verb h am
and its imperfect yah m to defend ends in /y/. Nouns that are derived
from finally weak verbs like h am of the pattern of al-muh m always end
/-/. This category of words should not be confused with words ending in
y-n-nisbah as in words like lubnniyyun Lebanese or even with words
like zabyun deer both of these words are written in Arabic script with
/y/, however the source of this /y/ is not a final radical as the nouns
253
derived from finally weak verbs like h am and its imperfect yah m to
defend which ends in /y/ as a radical (Farrj, 2001, 7172).
The vowel endings of al-manqs defective are determined by the case
endings and whether they are definite or indefinite. When words of the
manqs defective are in the accusative case and indefinite they end in
/-/ in the waqf as qbaltu muh my I met a lawyer after the omitting
of the nunation. When they are definite they end in /-/ as in qbaltu
l-muh m I met the lawyer. When the manqs words are in genitive or
nominative case they end in the omitting of the final vowel /-/ as in filwdin the valley and hd wd this is a valley. However when they are
definite the words of the manqs end in long vowel /-/ as in fil-wd in
the valley and hd l- wd this is the valley.
2. The rules of waqf of words that end in the long vowel /-/ such as
al-fat the youth and which are primarily derived from finally weak
verbs are called in traditional Arabic grammar al-maqsr the shortened.
The waqf on these words is always end in an alif /-/ in all three case
endings and regardless how orthographically they are written with alif
maqsrah or regular alif. What matters here is the pronunciation and
not the script.
3. Concluding Remark
It is worthwhile to mention that the rules of waqf are not always adhered
to by readers and speakers of Modern Standard Arabic. The Arabic
language for sometime has been going through processes of change.
Some readers and speakers are not using the al-h arakta-l-irbiyyah
case endings. The often heard statement that states sakkin taslam use
sukn and you will be safe reflects the state of affairs of the on going
change of the Arabic language. Therefore the rules of waqf outlined
above when considered should be used as guidelines not to be applied
in a rigorous and strict manner.
254
salman h. alani
4. References
4.1 Primary source

Ibn Ya. n.d. arh al-mufassa l. Cairo: Maktabat al-Mutanabb.
Farrj, Muhammad. 2001. Al-waqf wa-wazifuhu inda n-nahwiyyn wa-l-qurr. Annals
of Arts and Social sciences, Monograph 159 Volume 21. Kuwait: Kuwait University.
H assn, Tammm. 1973. Al-lua al-arabiyya mabnh wa-manh. Cairo: al-Haya
al-mma lil-Kitb.
Holes, Clive. 2004. Modern Arabic Structures, Functions and Varieties. Revised edition.
Georgetown, Washington, D.C.: Georgetown University press.
Mitchell, T.F. 1990. Pronouncing Arabic. Oxford: Clarendon Press.
THE EXPLANATION OF HOMONYMY1 IN THE LEXICON

OF ARABIC
Georges Bohas and Abderrahim Saguer
ENS Lettres et Sciences Humaines, Lyon
1. Introduction: the theoretical framework and earlier analyses

This article is a follow-up to our previous paper entitled Sur un point de
vue heuristique concernant lhomonymie dans le lexique de larabe.2 By
adopting a heuristic viewpoint, we take into account the fact that we have
not yet finished exploring all the matrices of Arabic, and thus proceed
by means of successive evaluations and provisional hypotheses.3 Certain
points remain to be clarified; nevertheless, the level of explanation we
can achieve has become clearer, as well as the explanatory methods that
we are able to offer within the framework of the Theory of Matrices and
Etymons (TME). Although there are already numerous publications on
this subject, it is worth underlining that, within the TME framework,
the lexicon is organized on three levels:4
1 / The matrix: a non linearly-ordered combination of a pair of phonetic feature vectors linked to a notional invariant; for example,
{[labial], [coronal]} to strike a blow.5
1
It is worth defining homonymy in opposition to polysemy. Polysemy is a word
which brings together several meanings between which users can recognize a link
(Nyckees, 1998: 194); the meanings are different but related. Homonymy is distinct
from polysemy in that, in the case of homonymy, it seems impossible to re-establish a
plausible semantic relationship (Nyckees, 1998: 194) between the different meanings,
for example: flies certain insects and flies the opening at the front of a pair of trousers
or to sound to make a noise and to sound to measure the depth of waterdifferent non
related meanings.
2
Bohas and Saguer (2006).
3
That is: in a not-rigorously demonstrated manner but justified by reasons of internal coherence (see the website www.memo.fr Einstein, Albert); and accepting that you
cannot explain everything.
4
See Bohas (1997, 2000), Dat (2002).
5
This is a property of the language that was proved both formally and semantically
by Bohas and Darfouf (1993), developed in Bohas (1997), which consists in the fact that
256
georges bohas and abderrahim saguer
2 / The etymon: a non linearly-ordered bi-consonantal base made up

of two phonemes taken from a given matrix exhibiting both the features of this matrix and its notional invariant; for example, {b,t} to
strike a blow with a sharp object.
3 / The radical: an etymon that has developed by diffusion of the final
consonant or by incrementation, or that results from the blending of
two etymons; the radical includes at least one vowel and vectors the
notional invariant; for example: /bvtar/, to cut, to cut the tail (Bohas
2000, 9). The radical is the domain in which diverse morphological
and Ablaut processes take place (Guerssel and Lowenstamm 1993,
Segeral 1995).
So far, ten matrices have been accounted for; most of these have already
been subjected to in-depth studies (see footnote 14).
Matrix 1
{[labial],6 [coronal]}
Notional invariant: to strike a blow
Matrix 2
{[labial], [-voiced] }7
[+continuant]8
Notional invariant: movement of air
Matrix 3
{[labial], [pharyngeal]}9
Notional invariant: (a) tightening
Matrix 4
{[coronal], [pharyngeal]}
[-dorsal]10
[-voiced]
a binary combination {a, b} is realized in the order a+b and in the order b+a while keeping the same notional invariant.
6
[labial] characterizes sounds produced with a constriction of the lips. For matrices
1, 2, 3, 6 we integrate on-going research which shows that the feature [labial] should not
be restricted by [-sonorant] (see Mansouri, 2006).
7
[voiced] Sounds produced with vibration of the vocal cords are said to be voiced
([+voiced]), whereas other sounds are said to be non-voiced ([-voiced]), see Dell (1973:
56).
8
[continuant] Sounds with the feature [+continuant] are produced without interrupting the flow of air through the oral cavity, those with the feature [-continuant] are
produced with total interruption of the flow of air at the oral cavity, see Halle (1991:
208).
9
[pharyngeal] characterizes segments that the Arabic tradition calls gutturals, that
is: , h, , h , x, and q. For the problems posed by the characterization of this class, see
Kenstowicz (1994: 456ff).
10
[dorsal] characterizes sounds produced with a constriction created with the back
of the tongue between the soft palate and the uvula (velar and uvular consonants; rear
vowels).
explanation of homonymy in the lexicon of arabic
257
Notional invariant: low voice, muffled, hoarse noise

Matrix 5
{[coronal], [dorsal]}
Notional invariant: to strike a blow11
Matrix 6
{[labial], [dorsal]}
Notional invariant: curvature
Matrix 7
{[dorsal], [pharyngeal]}
Notional invariant: the cries of animals
Matrix 8
{[+sonorant12], [+continuant]}
[+lateral13]
Notional invariant: the tongue
Matrix 9
{[+nasal], [+continuant]}
Notional invariant: the nose
Matrix 10
{[+nasal], [coronal]}
Notional invariant: traction14
The data on which we have based our study are to be found in the
Kazimirski, and have been checked in the Qms and/or the Lisn.
When they are based on another source, this is mentioned.
2. Explanatory methods
In the paper quoted above, we demonstrated that the homonymy of a
radical may be attributed to three causes:
A. the fact that it is the result of blending: it manifests the meanings of
both the etymons that are its source.
11
See Diab (2005) who brings a modification to the formulation of the notional
invariant of this matrix.
12
[sonorant] Sounds with the feature [+sonorant] are produced with a constriction
which does not influence the capacity of the vocal cords to vibrate spontaneously. Those
with the feature [-sonorant] have a constriction which reduces the global flow of air and
makes voicing more difficult. Thus the natural state for sonorants is [+voiced] and for
non sonorants (termed obstruents) is [-voiced], see Kenstowicz (1994: 36).
13
[lateral] A [+lateral] sound is produced by making a constriction with the central
part of the tongue while lowering one or both lateral edges so that air escapes around the
side(s) of the mouth, see Kenstowicz (1994: 35).
14
For matrices 1 to 6, see Bohas (2000), Dat (2002), for an in-depth study of matrix 6,
see Serhane (2003), Bohas and Serhane (2003), for matrix 7, see Bohas and Dat (2005),
for matrices 8 and 9, see Bohas (to be published) and for matrix 10, see Saguer (2003).
258
B. the fact that its etymon is the realization of several matrices: it manifests the meanings of these matrices.
C. the fact that two etymonial analyses are possible, such as [nX]Y and
n[XY].
Below, we illustrate each case with an example from Bohas and Saguer
(2006).
A. Homonymy resulting from blending
Let us consider the verb araza, which attests two meanings (hereafter
senses):
S1 to prick something with a needle, to drive in, to plunge (a sharp
instrument), to plunge a tail into the ground to lay eggs (of
locusts);
S2 to give but very little milk (of pregnant camel).
The same semantic load is found in rizun;
S1 that drives in, plunges a sharp instrument, a goad into something;
that plunges a tail into the ground to lay eggs (of locusts);
S2 that gives but little milk (camel).
Since it is not possible to establish a plausible semantic relationship
between the two senses, this is an obvious case of homonymy. And yet
we observe the existence of the following words:
arra15 F. III : to be found in small quantities (of milk of a female);
irrun
: a small quantity, generally, such as a small quantity of
milk in a female
murrun
: that has little milk in the udders (camel)
Phonetically, the etymonial analysis can only be { ,r}, since their radical
has no other consonant,16 and it is obvious that they attest sense S2.
Moreover, the precise meaning of the verb razza is: to plunge a tail into
the ground to lay eggs (of locusts); to stick, drive in and fix firmly one
object into another or into the ground. Thus it clearly attests sense S1
and is analysed as the etymon {r,z}.17 Therefore, the explanation is that
araza includes senses S1 and S2 because it results from blending of the
15
16
17
We use boldface for the segments that make up the etymon.

We call these non ambiguous radicals.
Non ambiguous radical.
259
two etymons {,r} and {r,z}. The way this blending occurs is represented
in model A:18
A
Cj
Ci
Ci
Si
Ck
Sj
Cj
Si
More explicitly :
r
lack of milk
Si
Ci
+
Ck19
SJ
x20
rz
to drive a sharp object into
Sj
araz
1920
Henceforth, in similar cases, we shall talk of explanation through

blending.
B. Homonymy through the realization of several matrices
1) An easy case
Let us take the verb mata"a, which means:
S1 to strike somebody with a stick
S2 to tighten, stretch out a rope
The m is [labial], the t is [coronal]: the etymon {m,t} can thus be a realization of matrix 1:
{[labial], [coronal]}
And for this reason assumes the sense S1 to strike somebody with a
stick.
But the m is also [nasal] and the etymon {m,t} can also be a realization
of matrix 10:
18
See Bohas (1997: 175f and 2000: 49).

The obligatory contour principle (OCP) explains the fusion of the two Ci into a
single segment. See McCarthy (1986) for the definition and various applied examples
referring to Semitic languages of this principle which forbids adjacent identical elements
at the same level. Since, OCP has given rise to a multitude of studies the list of which
would be superfluous here.
20
We use x to indicate blending.
19
260
{[+nasal], [coronal]}
Notional invariant: traction
And for this reason assumes the sense S2 to tighten, stretch out a
rope.
2) A more complex case
The verb natara attests the following meanings:
S1 to disperse
F. I
: to scatter, disperse, disseminate
F. II
: to scatter a lot, in large quantities: intensive of F. I
F. V
: to be scattered, dispersed, to disperse
F. VI
: to be scattered, dispersed, to disseminate, to spread
S2 actions concerning the nose
F. I natura : to blow ones nose
F. II
: to blow ones nose
F. II
: to draw up water through the nostrils
F. VIII
: to blow ones nose
F. VIII
: to draw up water etc. through the nostrils and to
expel it through the nostrils
S3 to pull, tear off/out
F. I
: to remove, to take the clothes off the body of somebody, to strip
S4 to strike a blow with a sharp object21
F. IV
: to pierce somebody with a sharp instrument and to
make the blood flow
The first hypothesis we can formulate is that natara develops the etymon: {n,t} and that for this reason, it is a realization of matrix 9:22
{[+nasal] [+continuant]}
Notional invariant: the nose
The phonetic substance of this matrix comprises, on the one hand, the
two nasals, m and n, and, on the other, the various fricatives.
21
We will propose no explanation for this sense; as we said in the introduction, we
have not yet explored all the matrices of Arabic, the notional invariant sharp is without
doubt important but as yet we know nothing of it.
22
See Bohas (to be published) for a detailed study of this matrix.
261
The ramifications of the notional invariant are as follows:

1. the nose
1.1. the organ itself and what affects it
1.2. the specification of parts (the top, the sides)
1.3. to be sharp, > protruding, > to precede
2.1. specifications of the organ (big, small . . .)
2.2. animal or human presenting these specifications
3. to lift the nose: movement of pride or contempt
4. the nose and air: to breathe in; to breathe out; to perceive odours,
to smell
5. the influence of the nose on the voice: nasal sound; similar animal
cries (buzzing-grunting)
6. various liquids (mucus, phlegm) which pass through the nose
We may note that, in this organization, senses S2 to blow ones nose,
to draw up water through the nostrils, fit into both 6. and 4.
But, since n is [+nasal] and t [coronal], {n,t} can also be a realization of
matrix 10:
{[+nasal], [+coronal]}
Notional invariant: traction
Moreover, other realizations of the {n,t} etymon can be found in:
natala [nt]l
: to remove the garment or breastplate
nanaa [n]n : to remove (ones clothes)
Sense S3 to pull, tear off/out of F. I: to remove, take the clothes off the
body of somebody, to strip is thus incorporated into this matrix.
We can therefore understand why the {n,t} etymon is homonymic: if we
take into account the t the feature [coronal], then we see that it realizes
matrix 10 traction, while if we take into account the feature [+continuant], then we see it realizes matrix 9 the nose.
In this case we shall speak of ambiguity originating from the fact that
an etymon is the realization of several matrices.
C. Homonymy due to the possibility of two etymonial analyses
Let us pursue the analysis of natara. We have seen that it also attests the
sense to disperse, to scatter. It is true that we can establish both a phonetic and a semantic relation with:
tarra [tr]r
: to disperse, disseminate
the etymon of which can only be {t,r},23 and with:
23
No ambiguity.
262
tra t[w]r
: to be lifted and scatter in the air
tartaratun [tr]tr
: dispersion, dissemination.
farata f[rt]
: to be dispersed, disseminated (of a tribe)
This leads us to analyse natara by positing {t,r} as etymon and the n as
an initial crement. Therefore, the homonymy stems from the fact that in
A and B the form is analysed as [nt]r, which may have two matrix links,
and in C as n[tr]. We shall say in the latter case that homonymy arises
because several etymonial analyses are possible.
3. The two levels of explanation

In Bohas and Saguer (2006) we studied cases concerning radicals
containing an n. In this paper, we shall extend the study to include the
analysis of radicals containing an l. According to previous studies,24 l may
have the status of initial crement/prefix25 or of matrix segment. Earlier,
Hurwitz (1913, 5560) had already recognized this prefix status:
The preformatives are thus seen to possess a fairly definite, though remote,
relationship to each other. The sibilants and gutturals are to be traced to
causative stems; the dental t and liquid n to reflexive stems; the liquids m,
l, and i are to be connected etymologically with the reflexive n, and the
preformative y may be considered to be a denominative stem.
When it has a matrix function, l may form an etymon with the second
radical: [lx]y, or the third radical: l[x]y.26 In other words, a radical [lvxy]
may present all its possibilities, as can be seen in the table below:
Etymon:
{x,y}
{l,x}
{l,y}
Analysis:
l[xy]
[lx]y
l[x]y
Radical:
24
<lvxy>
The most recent are those of Saguer (2000, 2002, 2002b).

We use prefix when l has a semantic-grammatical value and initial crement when
it does not.
26
This expresses the fact that in this form the etymon is [ly] and x is an inset
crement.
25

3.1
263
The explanation through attribution to identified matrices
The verb labaxa demonstrates eight senses:

S1 to be fleshy (of the body)
S2 to beat, strike somebody
S3 to kill somebody
S4 to take a thing from somebody, from the hand of somebody
S5 to get something out of somebody using trickery
S6 F. III: to slap somebody in the face
S7 F. V: to perfume oneself with musk
S8 to say something foolish, insulting to somebody
It should be noted that several senses are polysemically related; in other
words, a plausible semantic relation can be established between them,
as follows:
S2, S6 and S3 are manifestations of the notional invariant to
strike a blow; to kill is related to this concept through the
cause>consequence relationship and S6 specifies the mode of
action.
S4 and S5 come under the notional invariant: to pull, bring something to oneself.
Thus five homonymic senses remain:
A S1 to be fleshy (of the body)
B S2+S6+S3 to strike a blow
S2 to beat, strike somebody
S6 F. III: to slap somebody in the face
S3 to kill somebody
C S4+S5 to pull, bring something to oneself
S4 to take a thing from somebody, from the hand of somebody
S5 to get something out of somebody using trickery
D S7 F.V: to perfume oneself with musk
E S8 to say something foolish, insulting to somebody
We are going to attribute each of these homonymic senses to a source
matrix.
Take sense A S1labaxa: to be fleshy (of the body).
264
It is easily noted that this form is semantically linked to:

rabxun r[bx]
: fat, thick and with a soft and loose body
habayyaxun h[bx]
: young chubby
: fat, thick, hard
anbaxun [bx]
muxabxabat [xb] xb : handsome and plump (camels)
This leads us to deduce that labaxa should be analysed as a form
incorporating the etymon {b,x} with initial incrementation of an l
which has the role of a prefix denoting the sense we have called static,
following Joon (1923, 95). This is defined as the appropriation by the
subject of the quality or the state X. The Hebrew examples cited by Joon
are: kbed he is heavy, qton: he is small (1923, 95).
This etymon {b,x} is a realization of the matrix 6 {[labial], [dorsal]}
which has curvature as its notional invariant. One of the rst manifestations of curvature in the convex form is precisely the concept of
fatness:27 fat, plump, robust, that which realizes a swollen form: , as
in:
bajja -F. VII
bjilun
fajia
hijaffun
: to be plump, to have rounded flanks (of animals fattened by pasture)

: fat, replete
: to have a fat belly
: who has a fat belly, paunch
For sense A, labaxa to be fleshy (of the body) is analysed as l[bx] and is
thus a realization of the matrix {[labial], [dorsal]} which has the notional
invariant curvature.
Let us consider sense B S2+S6+S3 to strike a blow; S2 to beat,
strike somebody; S6 F. III: to slap somebody in the face; S3 to
kill somebody. It is easy to establish a link between the form and other
manifestations of the etymon {l,b} such as:
wabala w[bl] : to strike somebody with a stick
labana [lb]n : to strike somebody violently, to batter somebody to
death with a stick
lah aba l[h ]b : to strike somebody with a sabre
and for the implied sense:
habila h[bl] : to lose ones son through death
27
See Bohas (2000: 108).
265
This enables us to establish that labaxa includes the etymon {l,b}, itself a
realization of the matrix {[labial], [coronal]} which has to strike a blow
as its notional invariant.
However, note the existence of the following set:
laxxa [lx]x
laxaba [lx]b
laxama [lx]m
: to slap somebody in the face

: to slap somebody in the face
: to strike somebody in the face
This enables us to extract the etymon {l,x} connected, once again, to

the sense to strike a blow. The segment l is analysed as [coronal} and x
as [dorsal], so that this etymon is a realization of the matrix {[coronal],
[dorsal]}, which leads us to the conclusion that labaxa in the B sense
results from blending between two etymons {l,b} and {l,x}, a B type
blending (Bohas 2000, 50).
For sense C. = S4.+S5. to pull, bring something to oneself : S4. to
take a thing from somebody, from the hand of somebody; S5. to get
something out of somebody using trickery we shall justify an etymonial
analysis l[b]x, that is, an etymon {l,x} with an inset crement b. Indeed, a
phono-semantic relation can be established with:
In the order xl:
xalaja
xal (hly)
xalxala
saxala
xalasa
: to pull, attract to oneself

: to pull out
: to completely strip a bone of its flesh
: to take, remove, ravish through trickery
: to take away, remove ravish in an instant, unexpectedly, to pull out
F. III : to pull something from somebody; to seize; to grab
something
F. V : to remove, take away
F. VI : to reciprocally pull out, and to pull something each
from his own side
F.VIII : to pull something towards oneself in a hurry (the same
meaning as the first item of the list with the addition
of celerity).
In the order lx :
malaxa
: to forcibly pull something towards oneself seizing it
with ones teeth or hands
F. V. : to pull out, to burst e.g. the eye of a prey (birds of
prey)
F.VIII : to pull, extricate, pull out (a tooth, an eye); to pull out
of the scabbard (a sabre, etc.)
266
Therefore the etymon {l,x} carries two homonymic senses: to strike a

blow, as a realization of the matrix {[coronal], [dorsal]} and to bring
something to oneself as a realization of another matrix, which remains
to be studied and, we suspect, includes the features:
{[+approximant28], [+continuant]}
[coronal]
with the notional invariant: to bring something to oneself
Besides the words above, this matrix manifests itself in:
Etymons with l
salla
salaba
salata
salaxa
salaba
sahala
halaba
halata
halada
lah
lahaba
lahata
laxasa F. II
halaa
halaba
halata
halama
alaha F. II
28
: to pull, extract gently one object from another

: to snatch something with force from somebody
: to extract, to pull
: to skin, to remove the skin from a sheep; to remove
ones clothes and shirt
: to pull, to extract the marrow from bones
: to peel, to strip bark, skin
: to pull out hair, bristles, horsehair
: to peel something by removing the skin, bark
: to pull, extract, pull out something from its place
: to remove the inner bark of a tree, wood
: to remove the bark from a piece of wood, to strip it of
its bark
: to remove the bark from a piece of wood, to peel it
: to remove, to extract the purest part
: to remove the flesh from the skin of a dead or bled
white animal
: to take the milk, to milk
: to pull out, to remove in flakes
: to remove tinea from the skin
: to strip, to remove the clothes from somebody
This composition, [+approximant], seems too complex to characterize the class r, l.

[coronal]
This is due to the fact that, according to Yeou and Maeda (1994), the pharyngeals and
uvulars of Arabic are also characterized by the feature [+approximant]. Indeed, the continuation of our studies will enable a discussion of this point. If the gutturals do figure
in this matrix, and assuming that it may then be formulated simply as {[+approximant],
[+continuant]}, this would constitute proof that the gutturals and r, l are indeed members of the same class: [+approximant].

Etymons with r
haraba
harasa
harada
saraqa
sar
xaraa
xarata
xarafa
'araza
'ar'ara
'araqa
'arama
'ariy
qa'ara
267
: to strip somebody, to pillage (a caravan, a tribe)

: to steal something; in fields, pastures
: to take, by milking, all the milk from a camel
: to steal
: to remove, separate, distance something from somebody
: to attract a camel to oneself with a hooked stick
: to strip of bark and make even
: to pick a piece of fruit from a tree
: to pull with force
: to move the cork of a bottle to uncork it; to remove,
take out the cork; to pull, to pull out, to burst an eye
: to strip a bone of the flesh that was on it by eating it
: to eat the flesh that sticks to the bone
: to be naked, stripped of ones clothes (cause> consequence relation)
: to pull out with the root, from top to bottom and to
cause to fall
This abundance of data, the notional domain of which has still to be

organized, is nevertheless sufficient to justify the existence of this
matrix:
[+approximant], [+continuant]
[coronal]
to bring something to oneself
This data also justifies the analysis that posits that in labaxa sense C =
S4+S5 to pull, bring something to oneself : S4 to take a thing from
somebody, from the hand of somebody; S5 to get something out of
somebody using trickery is a manifestation of this matrix.
As to labaxa, D S7 F.V to perfume oneself with musk it is analysed as
l[bx] in which the etymon {b,x} is found, as in:
baxara [bx]r : to perfume somebody or something with incense
baxira
: to smell bad
: who has unpleasant, fetid breath
abxaru
baxxa [bx]x : to snore during sleep
This etymon is a realization of matrix 2.
{[labial], [-voiced]}
[+continuant]
268
which combines the labials b and f with voiceless fricatives.

The ramifications of the notional invariant of the conceptual field
are:
movement of air: wind, breath
expulsion of air in man or animal
29
consequences (various odours)
Below we provide some realizations of the matrix with f:
{ ft}
nafata
{ fh}
fahh a
fahfah a
fah -F. II
fha/fawah a/
lafaha
nafaha
{ fx}
faxxa
fxa/fawaxa/
nafaxa
: to blow (on something)

: to hiss (snake); to hiss during sleep
: to be hoarse
: to season food
: to spread ones perfume; to smell good or bad
: to blow (of warm wind)
: to spread ones odour; to blow (of cold wind)
: to snore and hiss talking of somebody who sleeps;
to spread (of an aroma)
: to spread (of an odour); to hiss (wind); to release
wind (of a man)
: to blow with the mouth; to break wind
Finally, sense: E S8 to say something foolish, insulting to somebody.

Note that labaxa commences with an l [+lateral] and nishes with a x
which is a [+continuant] segment. The analysis is thus l[b]x, of the
etymon {l,x}, which is also present in:
: to be very talkative and say a lot of rubbish
laxiya
xatila
: to talk a lot and to say only trivialities
In this case, this etymon is a realization of matrix 8, exactly like:
{l, }
la
: to speak, in general to say futile things, to utter vain,
flippant or careless words, to make mistakes in talking, to make an error
29
This sign indicates that a semantic relationship exists, here: cause>consequence.

laiya
F. X
Matrix 8
269
: to make an error talking, to be mistaken

: to gather locutions, idioms, or to pay attention to
words and locutions (lut), particularly of nomadic
Arabs, i.e. to draw from them knowledge of Arabic
words
{[+sonorant], [+continuant]}
[+lateral]
Notional invariant: tongue
As we shall often be returning to the organization of the notional field of

this matrix, it is worthwhile examining it here:30
The ramifications of the notional invariant include:
0. The tongue and its characteristics
1. The tongue and the physical actions that are its peculiarities
1.1. to make an action of the tongue
1.2. to seize, to pull with the tongue
1.3. to lick
1.3.1. consequence (1): to moisten and stick
1.3.2. consequence (2): to be smooth, polished
1.4. to savour, to taste
2. The tongue as an instrument of language: to speak, to speak in various
ways, to be talkative, to malign, to wrestle in words with somebody, to
hurt with malevolent words; to speak with authority > to order.
3. The tip of the tongue: pointed, to be sharp, to become a point and
thus to prick. Here we have a relationship of the type part/whole:
only the tip of the tongue is taken into consideration. Note a possible
interference with 2. he who has a pointed tongue is likely to utter
hurtful remarks. We link to 3. the cases in which the pointed characteristic is explicitly given as cause.
4. Tongue of fire: to blaze, to burn, to scorch.
Consequently, all the senses expressed by this radical are made explicit
in the complex diagram below which constitutes its lexicogenetic tree,
and traces its phono-semantic composition, thus accounting for the
homonymy :
30
This matrix is studied in detail in Bohas (2006).
270
Matrix level
M6
M2
M1
M8
M5
{[labial],
{[labial],
{[coronal],
{[+lateral],
[+continuant]}
[coronal]}
[dorsal]}
[+continuant]}
air> odour
to strike a blow to strike a blow the tongue
{[labial],
[dorsal]}
curvature
Etymoniall evel
{b,x}1
l[bx]
{b,x}2
{b,l}
{l,x}2
{l,x}1
[lb]x
l[b]x
Radicall evel
labax
Once this tree is exhaustively constructed, the analysis is finished. In the

present case, we have managed to link each sense to a matrix. This is not
always possible.
3.2
The explanation through attachment to etymons
In other cases, the analysis only enables the etymons to be identified

without reaching the matrix level. Consider the verb laata which has
the senses:
S1 to hurt, to strike, to injure, to harm somebody (either by shooting an arrow at him, or by looking at him with the bad eye
S2 to print on an animal, on the neck, a mark with hot iron, to
mark it
S3 to hurry, to hasten
S4 to delay paying a debt, to put off its payment
S5 to go to pasture (livestock)
Each sense is homonymic in relation to the others, but, in each case,
it is possible to establish a phono-semantic relation with other words,
enabling us to identify etymons, as we shall demonstrate.
For S1 to hurt, to strike, to injure, to harm somebody (either by shooting an arrow at him or by looking at him with the bad eye. Taking into
account:
271
laata l[]t
lahat l[h]t
lta l[w]t
: to touch, attain somebody, e.g. with an arrow

: to strike, to attain somebody with an arrow
: to strike, to attain somebody with an arrow or by looking at him with the bad eye
we may establish that the three verbs all demonstrate the phonetic constant lt and very similar meanings (indeed for laata and lta, the meaning is identical), enabling the identification of an etymon {l,t}.
For S2, compare laata to print on an animal, on the neck, a mark with
hot iron, to mark it with the elements of the following paradigm:
'alata
: to mark (a camel) on the neck with a transversal
mark
'alama
: to mark, to distinguish by a mark, by some sign
'alaba
: to mark something, either by making incisions or by
pressure
ra'ala F. II : to incise the ear of a beast to mark it
This enables us to establish the presence of a phonetic constant l and
a semantic constant to mark, thus identifying the etymon {l,' }. Let us
add the form:
laxafa
: to print a large cautery, to make a large burn on somebody
in which lx constitutes either a variant31 of l or a clue to identifying a
matrix.
For S3, it is worth comparing la'atato hurry, to hasten with:
'at F. II [' t]w : to hurry somebody, to tell him to hurry
'abata
' [b]t : to throw a horse into the race so as to make it
sweat
: to go, to advance rapidly and by lengthening the
hata'a h[t' ]
neck (of camel)
'atawwad ['t]d : rapid, hurried and tiring (voyage)
This enables us to identify the phonetic constant ,t and the semantic
constant to go rapidly, which leads us to deduce that laata should be
analysed as an etymon {',t} with the incrementation prefix l bearing the
reflexive meaning.32 Here again, note the variant:
xta
: to pass rapidly
31
32
We explain this notion the following pages.

Which corroborates Hurwitz earlier quoted in subsection 3.
272
We may relate S4 laata: to delay paying a debt, to put off its payment
to:
tla t[w]l F. II : to allow ones debtor an extension
tal [tl]w
: to wait, to be waiting, to defer
talla [tl]l
: to allow an extension, to give respite to ones
debtor
latat [lt]t
: refusal to pay or to recognize what one owes to
somebody
The phono-semantic constant presented by the above data enables us to
identify the etymon {l,t}.
Note that S3 and S4 are two contradictory senses: to go rapidly and
to delay. This enantiosemy33 can be explained by the fact that the word
is analysed as a blending of two etymons with opposing senses: lt to
delay and x t to go rapidly (D type blending: (Khatef, 2003 and 2004)).
This enantiosemy is homonymic, since the two senses have nothing in
common. The enantiosemy, which is usually presented as a quirk of
Arabic is, in fact, trivially predicted by TME (for more examples, see
Bahri 2003).
Finally, for laata S5 to go to pasture (livestock) a semantic link may
be established with:
: to graze in this or that place; to go to graze freely;
ra'
to put to graze, to take to graze
: to leave to graze freely
rata'a F. IV
'ra
: to go away, to move away (of, amongst others, a
horse that goes off grazing here and there)
: to graze freely
raba'a
These forms are obviously related to the etymon {r,' } linked to the sense
to go to graze (freely). It seems plausible that the etymon l from laata
S5 to go to pasture (livestock) is an allophone of this etymon, r and
l being from the same class [+approximant] [coronal]. The definition
of the allophones of etymons is set out in Bohas and Dat (2007):
33
We use enantiosemy (i.e. reverse semantics) for words that mean something and its
opposite, such as big and small, to rent which means both to take temporary possession
for the payment of a fee and to give temporary possession for the receipt of a fee.
273
Phonetic evolution may provoke a modification of the signifier and bring

about the appearance of what we will call etymon allophones, which are
phonetic variants of matrix etymons (from the set of a given matrix).
Most frequently, the etymon allophones bring into play acoustic factors, which explains the confusion of segments in the communicational
process. We posit that if:
[b] / {a , _} and [c] / {a , _}
where {a,b} is a matrix etymon and [b] and [c] are phonemes with
one or several shared phonetic features other than the vector of features
demanded by the matrix combination, and that they correspond to two
lexical items that are conceptually related, linkable (not necessarily identical), then [b] and [c] are the free variants of the phoneme (belonging to
the paradigm defined by the vector of features) that enters into the composition of the matrix etymon.
The allophonic forms, which enlarge the logical number of etymons belonging to a binary matrix of features, characterize the etymons, enlarged or
not, the articulation of which is weakened34 or loosened in verbal communication; these allophones are, over time, recovered and incorporated
into the lexicon of the language. These are free phonetic variantshistorical and/or dialectalof successful (widespread) innovations that co-exist
with the source-forms, such as the elements in the following list (from the
lexicon of Hebrew).
s s / /
ntas
: to demolish, to knock over, to knock down, to

pull out
ntas
: to break, to destroy
sdh Niph. (hapax) : to be desolated, ravaged
dad Pi.
: to break up lumps of soil, to harrow, to level a
piece of land
dad
: exercise violence, to desolate, to wreck, to
destroy, to devastate
In all these examples, the existence of the minimal pair engenders no
major lexical opposition, the semic difference often resulting from the
translation.
This argumentation enables us to consider laata S5 to go to pasture

(livestock) as a realization of the etymon {l,' } allophone of {r,' } given
the (obvious) semantic relation and the phonetic relation we have
established, but, in our analysis of this word, we have yet to trace it to a
clearly identified matrix.
34
The natural tendency of the speaker is to limit effort of his speech and to avoid
sharp shifts in the use of speech organs. (Lipinski, 1997, 186).
274
laata, which realizes the etymons: {',t}, {l,t} and {l,' }, is therefore
typical of the second level of explanation, namely identification of the
etymons. When we go on to study specific cases in the fourth part, we
will move from one of the two levels of explanation to the other. In some
instances we can link the etymon to a matrix; in others we may only link
the radical of a word to an etymon, while in others, we can provide no
analysis for the simple reason that, as things stand, identification of the
matrices and etymons is not yet complete. The objective we pursue is to
fully organize the lexicon into matrices. Is this a chimera, as some do
not fear to say? In our eyes, the best answer is provided by Darwin:
[. . .] it is those who know little, and not those who know much, who so
positively assert that this or that problem will never be solved by science.35
4. Case study
Within the framework defined above, we shall now analyse some
words.
4.1
lah ana
S1 to become fond of somebody, to fall in love with somebody

S2 to speak Arabic badly
S3 to speak a particular slang with somebody so as not to be understood by others
S4 to understand, to hear a word, an expression (link speak/understand)
Let us start by studying sense S1 to become fond of somebody. A
phonetic and semantic relation can be easily established with:
h anna [h n]n
: to groan, to make a groan of tenderness (of
certain animals e.g. a female camel showing
tenderness to her young)
nh a n[w]h
: to coo, to groan (of pigeons)
h anh ana [h n]h n
: to have and to show tenderness, compassion,
with emotion and worry
35
Charles Darwin: The Descent of Man and Selection in Relation to Sex, 4th paragraph, quoted in Quiniou (2006); for the full text visit www.gutenberg.org.
275
h an [h n]w
: to have great tenderness for somebody (also

used of a mother who through love of her
children, does not want to re-marry)
h aniba [h n]b and F. II : to feel a sentiment of pity, of compassion for
somebody
sah ana s[h n] F. III
: to treat somebody with goodness
The above leads us to deduce that, for this sense, lah ana should be
analysed as a form incorporating the etymon [h ,n] through initial
incrementation with an l which acts as a prefix marking the middle
voice.
In the semantic organization of matrices, the starting point is a concrete sense. In this we agree with Hurwitz (1913, 72):
It must also be borne in mind that primitive ideas are generally concrete,
and that an abstract idea is secondary, in that it is often based on some
objective aspect involved in the expression of the abstract idea, as when
anger is denoted by a reddening of the face, displeasure, by a falling of the
countenance etc.
The example under study is particularly illuminating. The word h anna

which has the abstract sense to be moved, to have compassion for
somebody, to feel great tenderness for somebody has the matrix
sense to groan, to make a groan of tenderness (of a female camel).
Exactly like anna to groan which comes from the same matrix:
Matrix 4
{[coronal] , [pharyngeal]}
[-dorsal]
[-voiced]
Notional invariant: low voice, muffled, hoarse noise
From the physical groan, as a noise, we pass to why we groan, to what
the groan expresses. In h anna, the two senses are maintained, whereas,
in lah ana, only the abstract sense appears.
We have thus identified the matrix of which lah ana to become fond
of somebody is a realization. Let us now examine senses S2 and S3
which are clearly linked. S2 to speak Arabic badly, S3 to speak a
particular slang with somebody so as not to be understood by others.
For these senses, the analysis [lh ]n is used, in other words, the etymon is
{l,h}, which is itself a realization of Matrix 8 {[+lateral], [+continuant]}
linked to the notional invariant tongue; furthermore, we have already
seen that to speak is one of the developments of the tongue notional
invariant:
276
2. The tongue as an instrument of language (cf. 3.1 above): to speak,

to speak in various ways, to be talkative, to malign, to wrestle in words
with somebody, to hurt with malevolent words; to speak with authority > to order.36
The same sense is found in:
laxxa [lx]x
: to be unintelligible, to speak (especially Arabic) in an unintelligible way
laxlaxniyy [lx]lx
: who has difficulty speaking or only speaks
lecherous language
laxiya [lx]y
: to be very talkative and say a lot of rubbish
la [l]w
: to say futile things, to utter vain, flippant or
careless words
S4 to understand, to hear a word, an expression that can be found
in lah ina with to be intelligent, seems to accept the same analysis. The
causal relation we assume, i.e. speak > understand > be intelligent is,
in fact, explicit in Syriac; in the latter, mll means both who speaks
and who is intelligent or in Greek in which logiks covers two series of
meanings: I. which concerns speech and II. which concerns reason.37
Therefore the lexicogenetic tree of the word may be constructed:
M8
M4
{[+lateral],
[+continuant}
tongue
{[consonantal],
[pharyngeal]}
muffled voice . . .
{l,h}
{h,n}
[lh ]n
l[h n]
lah an
36
37
See for the organization of the conceptual field above at the end of subsection 3.1.
See Bailly (1950).

4.2
277
lasaa
Reorganizing the Kazimirski, the following senses can be said to be

included:
S1 to head into the interior of the land
S2 to prick somebody, to bite (of a scorpion or snake)
S3 to hurt somebody with malevolent words, with features of
satire
S4 to malign somebody
S5 F. IV to sow enmity between men
These five senses can be reduced to two:
A to make ones way into the land
B in relation with tongue, as we shall show.
As far as the first (A.) sense (to head into the interior of the land) is
concerned, this form is semantically linked to:
nasaa n[s]
: to head into the interior of the land, country
tasaa [ts]38 x [s] : to travel, to cross a country, to head into the interior
of the land
asaa [s]
: to be remote, to be situated at a great distance
This fact leads us to deduce that lasaa should be analysed as a form
incorporating the etymon {s,' } through initial incrementation of a
prefix l marking the reflexive meaning, in the same way as the n at the
beginning of the form nasaa.
For the other (B) senses:
S2 to prick somebody, to bite (of a scorpion or snake)
S3 to hurt somebody with malevolent words, with features of
satire
S4 to malign somebody
S5 F. IV to sow enmity between men
The etymon {l,s} that the radical develops is a realization of Matrix 8
organized around the notional field tongue as was earlier discussed at
the end of subsection 3.1.
Of these four senses,
S2 to prick somebody, to bite (of a scorpion or snake) is linked to S3.,
like:
lasaba
: to prick somebody (said properly of a snake)
38
Tassa: F. II: to head into the interior of the land, country. The radical thus results
from the blending of the two etymons.
278
S3 to hurt somebody with malevolent words, with features of satire is

linked to S2 or S3, as is:
ladaga
: to hurt somebody with ones tongue, i.e. with a biting
remark
S4 to malign somebody is linked to S2, like:
lasana
: to give blows with the tongue, i.e. to malign somebody,
to tear him apart
S5 F. IV to sow enmity between men is in a cause > consequence
relation with S2 and S3. We can therefore show the lexicogenetic tree
which, for {l,s}, goes back to the matrix, and which, for the etymon {s,' }
remains at the etymonial level:
M8
M?
{[+lateral],
[+continuant]}
tongue
{l,s}
{s,' }
[ls]
l[s]
to head into the land
lasa
4.3
lasafa and lasifa
This entry covers the following senses:

S1 to fit by placing one next to the other and one over the other
(e.g. stones in constructing a wall, in building)
S2 lasifa to be dried and stuck to the bones (of skin on a very
scrawny body)
S3 to wind a strap of sinew around the base of an arrow
S4 to shine, to gleam
These four senses do seem to be in homonymic relation, as it is difficult
to see how one could draw some sort of plausible semantic relation
between them.
279
For S1 to fit by placing one next to the other and one over the other
(e.g. stones in constructing a wall, in building), comparison with the
words below leads us to identify a common etymon {s,f } to arrange, to
organize:
saffa [sf ]f
: to arrange in order
xasafa x[sf ]
: to fit and join solidly
sannafa s[n]f
: to compose, to make (a work, a book)
For S2, we recognize matrix 8 as in item:
1.3.1. consequence (1): to moisten and stick, as in:
lassa [ls]s F. VIII : to attach oneself, and to stick strongly
lasiqa [ls]q
: to be stuck to the bones
lasiqa [ls]q
: to be stuck
lasaa [ls]
: to be dried and stuck to the bones (of skin or a
very scrawny body)
For S3 to wind a strap of sinew around the base of an arrow,
consider:
: to wind a solid strap or a flattened sinew around
rasafa r[sf ]
the tip of an arrow to make firm the iron that has
been fitted
'afasa [ fs]
: to wind a ifs around the mouth of a bottle
'asaba [sb]
: to bandage, to wind a headband, bandage around;
to put a dressing on (the head, a member)
Examining the above, we may identify the etymon {s,f }, which is itself a
realization of Matrix 6, f [labial] and s [dorsal],39 of which the notional
invariant is curvature; and of which to surround and to wind around
are consequences.40
Finally, lasafa in sense S4 to shine, to gleam can be brought into
relation with:
walafa w[lf ]
: to shine time after time with repeated flashes that
come in uninterrupted succession (of lightening)
jafala
: to shine
39
40
Emphatics are characterized by the features [dorsal], [pharyngeal], [coronal].

See Bohas (2000: 115117).
280
The above enables us to identify the etymon lf to shine. As to the matrix

of which it is apparently a realization, it should contain the feature
[+approximant], as in:
balaqa F. VIII : to shine, to gleam
baraqa
: to shine, to gleam, to be shiny
lamah a
: to shine
lamaa
: to shine
ramah a
: to shine (of lightening)
In any case, as in walafa and lasafa, the composition {[+approximant],
[labial]}41 connected to the notional invariant to shine can be observed.
This is a complex matrix that has yet to be studied. Suffice it to say that,
in the present state of our knowledge, the etymon has been positively
identified, while the matrix has not. We can trace the lexicogenetic tree
thus:
M8
M?
M6
{[+lateral],
{[+approximant] ?}
[+continuant]}
tongue
shininess
{[labial], [dorsal]}
curvature
{l,s}
{l,f }
{s,f }1
{s,f }2
[ls] f
l[s] f
l[sf ]
l[sf ]
to organize
lasaf
4.4
lasaba and lasiba
S1 to prick somebody (of snake)

S2 to give somebody a lash with a whip
S3 lasiba to lick; to attach and to stick to something
41
The same remark as in footnote 28 is appropriate here. If gutturals appear to be a
part of this matrix, then they join the class of approximants, otherwise, [approximant]
will have to be restricted by the addition of [coronal].
281
Take S2 to start with. This form is related both semantically and

phonetically to:
saba a [sb] : to whip somebody with a whip until he bleeds
This fact leads us to deduce that lasaba should be analysed as a form
incorporating the etymon {s,b} through initial incrementation of the l,
which has no semantic value. The etymon {s,b} comes from Matrix 1
{[labial], [coronal]}, the notional invariant of which is to strike a blow,
with the specification, in this case, of the means with a whip, and other
specifications in:
rabasa
: to strike with the hand
safaa
: to strike, to give a blow (especially of birds, when, in
fighting, they give each other vigorous blows with
their wings)
sfa
: to strike somebody with a sabre
nasama
: to strike the ground with a foot
The two other senses come from the matrix {[+lateral], [+continuant]}
tongue, the first enters the heading:
3. The tip of the tongue: pointed, to be sharp, to become a point, and
thus to prick. As in:
lasaa
: to prick somebody, to bite (of scorpion or snake)
lasana
: to form a point, to give a pointed form, (e.g. to a shoe,
etc.)
: to prick (of scorpion)
salama
: to prick somebody, to make a bite (of a snake)
F. IV passive : uslima: to be pricked by a snake
And the second the heading:
1.3. to lick and 1.3.1 (see the end of subsection 3.1 above)
as in:
lassa
: to lick (a vase, a frying pan, etc.)
lasada
: to lick (a vase)
ladasa
: to lick
: to lick
lahasa
Drawing up the lexicogenetic tree is therefore an easy matter:
282

M8
M1
{[+lateral],
[+continuant]}
tongue
{l,s}
{s,b}
[ls]b
l[sb]
to strike a blow
lasaba
4.5
lah afa
Whereas the analysis of lasaba presents no particular difficulty, that

of lah afa is formidably complex, which is why we have left it till last.
Indeed, this verb attests several senses, as follows:
S1 F. I: to lick something
F. IV: to pressure somebody, to trouble, to ask with insistence
S2 F. I: to envelop somebody in a sheet, a blanket
F. III: to help, to assist somebody42
F. IV: to clothe and envelop somebody in a garment
F. V: to envelop oneself in a sheet, in a cloth, in a piece of material
F. VIII: to envelop oneself in a piece of material
S3 F. I P.: to suffer losses in ones belongings, flocks, etc. 43
F. IV: to do something bad to somebody, to cause him some
harm
S4 F. IV: to pull out (e.g. somebodys nail)
S5 F. IV: to burn, to have something consumed in fire44
Let us start with S1. The radical includes l and the fricative h , and can be
a realization of Matrix 8: {[+sonorant] , [+continuant]}
[+lateral]
42
43
44
This sense is linked to this block, see below.

P. indicates the passive.
Attested in Ass al-bala.
283
Notional invariant: tongue

to lick coming under the heading 1.3.45
The following forms develop the etymons {l,h} and {l,' }, that are
analogous realizations of this matrix:
lahika [lh ]k
: to lick
lahisa [lh ]s
: to lick
: to lick
la'iqa [l]q
la' [l]w F. V : to lick
The meaning expressed by F. IV that we have placed under sense S1, to
pressure somebody, to trouble, to ask with insistence, seems to correlate
well with the idea of licking, exactly as in French in which coller, to
stick can mean46 to impose ones presence on somebody and collant
sticky: that which one cannot rid oneself of. Thus, we have the following
chain: lick >stick >insist, the sense that we find in other manifestations
of the etymon {l,h}:
lahh a [lh ]h
: to insist on something to somebody, to persist in
asking him (for) something
hafala h [ f ]l F. V : to insist
halata [h l]t
: to insist, to pressure, to scold somebody, and to
swear
mahala m[h l] F. V : to insist and to pressure somebody
For S.2, the emergence of the sense to envelop undoubtedly comes from
the blending of the two etymons: lf x h f, which can be clearly seen in:
laffa
: to envelop, to twist, to surround with something
h affa
: to surround somebody with something, to envelop with something
Is it possible to go further and identify the matrix from which these
etymons stem? Without doubt, if we take into consideration the
following data:
laffa
: (also means:) to gather, to collect from all sides
daffa
: to gather, to bring together47
A hypothesis defended by Cantineau (1951), amongst others, is that in
protosemitic there was a lateralized emphatic that we shall write dl. In
Bohas and Janah (2000) it was argued that this dl had split in Arabic into
45
46
47
See above subsection 3.1.

See the Petit Robert
See also: mudf d[y] f : surrounded, attacked, encircled on all sides.
284
two phonemes: d and l. A word bearing one meaning had thus given
birth to two words with the same sense (modulo a few nuances). Thus
we would have:
dlaffa to bring together
daffa
laffa
to gather, to bring together to gather, to collect from all sides
This means that the l of la a is not a true l, but lexically a dl and that,
as such, it has the feature [pharyngeal] of the emphatics. The pair:
laffa
: to envelop, to twist, to surround with something
h affa
: to surround somebody with something, to envelop
with something
is in fact a dlaffa/h affa pair and the etymons dlf/h f are realizations of
Matrix 3:
Matrix 3 {[ labial] , [ pharyngeal]}
Notional invariant: (a) tightening
The relation between to envelop, to surround and to tighten is merely
one of cause to consequence.
The sense expressed by F. III: to help, to assist somebody seems to be
accounted for in terms of metaphor: to help is to surround somebody
with assistance, protection or affection. As an argument in favour of this
relation, we have the verb h affa that explicitly shows this meaning, since,
in the Kazimirski, after to surround comes the sense to be constantly
around somebody, and to be attentive to serve or protect him.
S3 constitutes a single semantic block:
a : F. I P.
: to suffer losses in ones belongings, flocks, etc.
b : F. IV
: to do something bad to somebody, to cause him some
harm
Let us bring lah afa into relation with the forms which manifest the same
properties:
falla F. IV
: to lose ones flocks
falaa
: to lose, to reduce to nothing
faliya
: to be cut, separated from the rest of the body
lafaa
: to peel, to skin
wafala
: to peel something by removing the bark
We introduce an etymon {l,f }; l is [coronal] and f [labial]. This etymon
can thus be a realization of Matrix 1:
285
Loss, harm comes under the heading B.3,48 global consequence, as in:
h afata
: to destroy, to lose
talifa
: to perish
As for S4 F. IV: to pull out (e.g. somebodys nail) it is a realization of
a Matrix under study:
{[+approximant], [+continuant]}
[coronal]
with the notional invariant: to bring something to oneself.
This manifests itself in the words quoted earlier in section 3.1. As f and
h are both continuants, it seems reasonable to consider the radical as a
blending of the two etymons that both realize this matrix: lh x lf, B type
blending.49
There remains sense S5 F. IV: to burn, to have something consumed in
fire. In this meaning, we can relate lahafa to:
lafaha
: to burn, to cause harm through its intensity (of fire, or
a very warm wind)
fayh
: heat caused by a star
sahafa
: to burn, to have something consumed in fire
The above relation reveals the etymon { f,h}; the semantic relation with
to blow remains to be established in order to link it to Matrix 2.
As for the residual cases
F. II
: to let the bottom of ones garment scrape the ground,
to wear it very long so that it trails, by extension: to
walk proudly
F. IV
: to come to the foot of a mountain and lih f: the foot of
a mountain
At this stage in our research, all that can be noted is that the latter is
perhaps to be related to h ffatun: edge, margin, extremity, although we
cannot establish this with certainty. All these comparisons are of limited
interest, but it is worth remembering, as we said at the start, that certain
points are still unclear.
48
49
See Bohas (2000: 79).

See Bohas (2000: 50).
286
Consequently, we shall limit ourselves to those senses that are clear, in

order to construct the lexicogenetic tree:
M8
M1
{[+lat], [+cont]} {[labial], [coronal]}

tongue
strike a blow
M3
M7
{[phar], [lab] }
tightening
{[+approx], [+cont]}
to bring to oneself
{l,h}
{l,f }
{h,f }
{l,h} x {l,f }
[lh ] f
l[h ] f
l[h f ]
[l,h ] x [l,f ]
lah af
5. Conclusion
Our method is thus distinct from that of the partisans of the
triconsonantal root. For them, it is sucient to identify the three
consonants in order to consider the analysis complete, even if
semantic incongruities and incompatibilities are evident, and even
if this identi cation provides no explanation of phono-semantic
links between words, such as homonymy and enantiosemy. What
does reassure them, however, is that they can pride themselves
on having reached a state of certainty.... that the root of maktab =
ktb is a certainty! Yet why should the root of istadaytu be dw rather
than dy, and, if it is dw, what phonetic motivation is there to be
found in istafaltu to justify the passage from w to y in istadaytu?
Indeed, this certainty is not as de nitive as they would have us believe.
With our approach based on argumentative reasoning, we might make
mistakes: one word might, perhaps, be matched up with another on
the basis of such and such a property without us having noticed this
relation. Given the explanatory results of our approach, which adopts
a heuristic point of view, this is a risk we assume.
The reader will have noticed that all our research has been carried
out taking the lexicon of Arabic as a synchronic whole. As we have often
repeated, in the study of the lexicon, it is vain to go back to a previous
287
biliteral stage in order to diachronically derive a triliteral stage. In other

words, the old debatebiliteral or triliteralis immaterial. The binary
composites (the matrices and the etymons) and the ternary composites
are there before our eyes: you only have to open a dictionary to find
them. Every radical is, according to the level of explanation, binary or
ternary, as the explanation of homonymy demonstrates all too well.
Admittedly, no one is obliged to overstep the conception, invented
by the Arab grammarians, of the tri- or quadri-consonantal root. However, to restrain oneself to this is to forego the opportunity to provide
explanations for the phono-semantic phenomena contained in the lexicon, phenomena that everyone may observe. To work with a theory that
posits that the Sun rotates around the Earth enables one to explain a
certain number of observable phenomena;50 but to work with a theory
that posits that the Earth rotates around the Sun enables one to explain
a greater range of phenomena, and fits in better with other knowledge
we have of the movement of the stars.
6. References
Ass al-bala = Ab l-Qsim Mahmd b. Umar az-Zamaxar, Ass al-bala,
Abdarrahm Mahmd, ed. Bayrt: Dr at-ti ba wa-n-nar.
Bahri, A. 2003. Lnantiosmie en arabe, Doctoral thesis. University Paris 8.
Bailly, A. 1950. Dictionnaire grec franais. Paris: Hachette.
Bohas, Georges. 1997. Matrices, tymons, racines, lments dune thorie lexicologique du
vocabulaire arabe. Paris: Peeters.
. 2000. Matrices et tymons, dveloppements de la thorie. Lausanne: Editions du
Zbre.
. 2006. De la motivation corporelle de certains signes de la langue arabe et de ses
implications. Cahiers de linguistique analogique, 3, 1141.
, and N. Darfouf. 1993. Contribution la rorganisation du lexique de larabe, les
tymons non-ordonns, Linguistica Communicatio, 5/12, 55103.
, and A. Janah. 2000. Le statut du dd dans le lexique de larabe et ses implications.
Langues et Littratures du Monde Arabe, 1, 1328.
50
We find in The Legend of Alexander: For the Sun is the servant of The Lord, which
interrupts its course neither day nor night. This idea that the stars are in the service of
God seems to date back at least to Bardesane (born in 154): Neither the Sun, nor the
Moon, nor the other beings which are superior to us in any thing have received power
over themselves, they are on the contrary subject to a law and, consequently, they do
what they have been ordered and never anything else. The Sun never says I shall not
rise at the given hour, nor the Moon I will no longer have phases, I will neither wax nor
wane, . . . All these creatures are servants and remain subject to a law: they are instruments of the wisdom of The Lord, Who is infallible. (See Teixidor 1992).
288
, and M. Dat. 2005. La matrice acoustique {[dorsal], [pharyngal]} en arabe classique et en hbreu biblique, premire esquisse. Mlanges de lUniversit Saint-Joseph,
LVIII, 125143.
, and M. Dat. 2007. Une thorie de lorganisation du lexique des langues smitiques :
matrices et tymons, Lyon: ENS ditions.
, and A.R. Saguer. 2006. Sur un point de vue heuristique concernant lhomonymie
dans le lexique de larabe. In Edzard, L. & J. Watson (eds), Grammar as a Window
onto Arabic Humanism. A Collection of Articles in Honour of Michael G. Carter, Wiesbaden, Harrassowitz Verlag, 130154.
, and R. Serhane. 2003. Consquences de la dcomposition du phonme en traits.
In: Angoujard, J.-P. and S. Wauguier-Gravelines (Eds.): Phonologie. Champs et perspectives. Lyon: ENS ditions, 131-155.
Cantineau, Jean. 1951. Le consonantisme du smitique. Semitica, IV, 7994.
Dat, M. 2002. Matrices et tymons. Mimophonie lexicale en hbreu biblique. Doctoral
thesis, Lyon: Ecole Normale Suprieure Lettres et Sciences Humaines.
Dell, F. 1973. Les rgles et les sons: introduction la phonologie gnrative. Paris: Hermann.
Diab, S. 2005. La matrice {[coronal], [dorsal]}, Les tymons impliquant le jm, Masters 2
Research Paper. Lyon: cole normale suprieure lettres et sciences humaines.
Guerssel, M. and J. Lowenstamm. 1993. The Derivational Morphology of the Classical
Arabic Verbal System. ms. UQAM and University Paris VII.
Halle, M. 1991. Phonological Features. In W. Bright (ed.): Oxford International Encyclopedia of Linguistics. Oxford: Oxford University Press, 207212.
Hurwitz, S. 1913 [1966]. Root-Determinatives in Semitic Speech, a Contribution to Semitic
Philology. New York: Columbia University Press.
Joon, P. 1923. Grammaire de lhbreu biblique. Rome: Institut biblique pontifical.
Kazimirski, A. de Biberstein. 1860. Dictionnaire arabe franais, Paris: Maisonneuve et
Cie.
Kenstowicz, M. 1994. Phonology in Generative Grammar. Oxford UK : Blackwell.
Khatef, L. 2003. Statut de la troisime radicale en arabe: le croisement des tymons, Doctoral thesis. University Paris VIII.
. 2004. Le croisement des tymons: organisation formelle et smantique. Langues
et Littratures du Monde Arabe, 119138.
Lisn = Jaml ad-Dn Ab l-Fadl Muhammad b. Mukarram b. Al b. Ahmad b. Ab
l-Qsim b. H abqa Ibn Manzr, Lisn al-Arab, s.d. Abd Allh Al al-Kabr, Muhammad
Ahmad H asab Allh, Him Muhammad a-dil, eds. Cairo: Dr al-Marif.
Lipinski, E. 1997. Semitic Languages. Outline of a Comparative Grammar. Leuven:
Peeters.
Mansouri, W. 2006. La place du trait [sonorant] dans les matrices de larabe. Masters 2
Research Paper, Lyon, cole normale suprieure lettres et sciences humaines.
McCarthy, J.J. 1986. OCP Effects: Gemination and Antigemination. Linguistic Inquiry,
17,2. 207263.
Nyckees, V. 1998. La smantique, Paris: Belin.
Qms = Majd ad-Dn Muhammad b. Yaqb al-Fayrzbd, Al-Qms al-Muh t.
Bayrt: Muassasat ar-Risla.
Quiniou, Y. 2006. La mort scientifique de Dieu. Le nouvel observateur. Hors-srie,
3841.
Le Petit Robert, dictionnaire alphabtique et analogique de la langue franaise. 1967
[1993]. Paris: Dictionnaires Le Robert.
Saguer, A.R. 2000. Lincrmentation des prfixes dans le lexique de larabe. Le cas du n.
Actes du colloque Journes de linguistique arabe et smitique, Langues et littratures du
monde arabe, 1, 5782.
. 2002a. Lincrmentation des prfixes dans le lexique de larabe. Le cas du m.
Langues et littratures du monde arabe, 3, 2957.
289
. 2002b. Z hirat al-isbq f l-judr al-arabiyya. Agadir: publications de lUniversit

Ibn Zuhr.
. 2003. La matrice {[+nasal), [coronal)}, traction en arabe. Premire esquisse.
Langues et littratures du monde arabe, 4, 138183.
Segeral, Ph. 1995. Une thorie gnralise de lapophonie, Doctoral thesis. University
Paris 7.
Serhane, R. 2003. Etude de la matrice {[labial], [dorsal]} en arabe, Doctoral thesis. University Paris 8.
Teixidor, J. 1992. Bardesane dEdesse, la premire philosophie syriaque. Paris: Les ditions
du Cerf.
Yeou, M. and S. Maeda. 1994. Pharyngales et uvulaires arabes sont des approximantes: caractrisation acoustique. 20me Journes dtudes sur la Parole, Trgastel,
409414.
THE PERIPHRASTIC BILINGUAL VERB CONSTRUCTION AS A

MARKER OF INTENSE LANGUAGE CONTACT.
EVIDENCE FROM GREEK, PORTUGUESE AND
MAGHRIBIAN ARABIC
Louis Boumans
University of Nijmegen
1. Introduction
I would like to start this paper with some terminological notes. I use
the dichotomy community language versus superimposed language
to refer to the typical unequal social-economic status of the bilingual
speakers languages. These terms refer to local as well as global power
relations and their sociolinguistic consequences. For instance, whether
in Portugal, Brazil or the United States, Portuguese speaking people
learn English in order to gain access to valuable information and upward
social mobility, i.e. education, media and employment.
I will use the term socially dominant for the language the individual
speaker is most exposed to in her daily life. This could be either the
community or the superimposed language, depending on the local situation. Thus English is more likely to be socially dominant for a particular
Portuguese/English bilingual member of the Portuguese community in
the US, while Portuguese will be socially dominant for most bilinguals
living in Portugal.
The terms matrix language and embedded language are grammatical notions referring solely to local syntactic units of analysis in bilingual
speech. The higher order constituent is the matrix in which lower order
constituents are embedded. In mixed sentences higher and lower order
constituents are in different languages. In most instances the community
language functions as the matrix language and superimposed language
elements are embedded. However, the reverse occurs in a minority of
cases so the terms should not be confused.1
1
Since the early 1990s, Carol Myers-Scotton has been the most influential promoter of
the insertion approach to code-switching and the terms matrix and embedded language.
292
louis boumans
The present article deals with verbs from a superimposed language

that function as embedded elements in community language discourse.
The community language may or may not be socially dominant, and I
argue that this makes a difference for the way in which foreign verbs are
embedded.
2. The integration of foreign verbs

There are three ways in which foreign verbs are integrated in the matrix
language, two of which are common. One is the complete morphological integration. Some basic form of the foreign verb, typically the verb
stem or the infinitive, is treated as the verb stem of the receiving matrix
language, and verbal categories of the latter are expressed by matrix
language morphology. Gloss (1) is a Moroccan Arabic/French example
showing the French verb stem montr- (from montrer [m ~tre] to show)
with an Arabic prefix and suffixes.2
(1) wa
adi y-montr.i-w-l-ek . . .
Q
FUT 3-show-PL-to-2SG
Are they going to show you . . .? MA3/French (Wernitz 1993, 308)
The insertion of foreign verb stems without overt morphological integration in matrix languages lacking verbal morphology, such as various
Austronesian languages (Van Staden 1999) can be considered as a subcategory of the morphological integration strategy, even if the integration is not overtly expressed by ML morphemes.
I concur with the fundamentals of her original Matrix Language Frame (MLF) model
(Myers-Scotton 1993), except for the definition of the matrix language. In my view, all
syntactic constituents function as a matrix for lower-order constituents, whereas in the
MLF model only the Complementizer Phrase functions as a matrix. I refer to earlier
work for more details on this approach to code-switching (Boumans 1998, Boumans
and Caubet 2000, Boumans 2002).
2
The vowel i is not a proper suffix. Embedded French verbs are modelled on a class
of Arabic verbs ending in a vowel. This vowel is subject to a/i ablaut. Cf. Caubet (1993),
Boumans (1998), and Boumans and Caubet (2000).
3
The following abbreviations are being used: in the main text: MA Moroccan Arabic;
in the glosses to numbered examples: 1,2,3 first, second, third person; ACC accusative;
AGR agreement; ART article; AUX auxiliary; DEF definite article; FUT future tense;
IMPF imperfective; INF infinitive; M masculine; NEG negation; PASTPART past participle; PL plural; PROGRPART progressive participle; PRT preterit; REL relative clause
marker; SG singular.
the periphrastic bilingual verb construction
293
The other common strategy is the periphrastic construction. This

strategy is equally common in the worlds languages and can also be
illustrated with Moroccan Arabic, this time in contact with Dutch:
(2) ba
y-dir
li-h
aanvall-en
want
3-do
to-3SG
attack-INF
It [the bird] wanted to attack him. MADutch (Jamal, 16, Utrecht 2000)
In his book Bilingual Speech (2000), Muysken presents a typology of

verb integration with special attention for the periphrastic type. In the
periphrastic construction, he argues, the foreign verb can be a nominalization, an infinitive or an adjunction. In the first case, the nominalized
verb is the complement of the helping verb, in the case of an infinitive the helping verb must be analysed as an auxiliary. In the adjunction
analysis, the foreign verb and the helping verb form a kind of verbal
compound.
Which of these three analyses is most appropriate does not only depend
on the language pair involved. As I discussed elsewhere (Boumans 1998;
2000), strategies and grammatical constructions differ among individuals belonging to the same bilingual community. Compare examples (2)
and (3). In Jamals utterance in (2), the patient participant of the Dutch
transitive verb aanvallen to attack is expressed as the MA indirect object
li-h to-him. The Patient role is expressed as an indirect object, because
the Dutch verb occupies the Direct Object position of the MA transitive
verb dar to do. This is in fact the common construction in the Dutch
Moroccan community. Now in Samirs example in (3), the patient of the
transitive verb controleren to supervise surfaces as the direct object suffix -hom them. The construction in (2) can be described as a helping
verb plus a verbal noun object, while the one in example (3) is in fact
very similar to the monolingual Dutch auxiliary construction. Compare
(3) with its Dutch translation in (4). Note that the pronominal direct
object of the main verb cliticizes to the finite auxiliary, even if an adjunct
constituent (morgen tomorrow) comes between the two verbs.
(3) kun adi
y-dir-hom
controler-en?
who
FUT
3do-3PL
supervise-INF
Who will supervise them? MA/Dutch (Samir 20, Nijmegen 1991)
(4) wie
gaa-t
hun
(morgen)
controler-en
who
go-2SG 3PL
tomorrow
supervise-INF
Who will supervise them (tomorrow)? Dutch translation of (3)
294
louis boumans
The variation between speakers, and even between utterances of the

same speaker, can be explained by the varying levels of competence in
both languages, and by the effect of grammaticalization and conventionalization (Backus 1996).
The third, less common strategy for the incorporation of foreign
verbs deserves mentioning here for the sake of completeness. This strategy consists in inserting inflected verb forms rather than verb stems.
The inflected foreign forms are mapped onto the ML paradigm and
express ML verbal categories. A famous case in point is Mednyj Aleut,
also known as Copper Island Aleut (Thomason and Kaufman 1988,
2338). In this variety of Aleut, Russian inflectional patterns in finite
verbs replaced Aleut ones while most other grammatical subsystems
remained intact. Crucially, the Russian verbal inflections express Aleut
tense and aspect categories. Similarly, Igla (1991) reports on a dialect of
Romani nowadays spoken in a suburb of Athens in Greece. The speakers
of this dialect of Romani moved in from Turkey in the 1920s but no longer speak any Turkish. However there are still approximately 30 verbs of
Turkish origin that continue to be inflected with Turkish suffixes while
following the Romani verbal paradigm, cf. Table 1. That is, the Turkish
inflectional suffixes express Romani inflectional categories.
TABLE 1 THE PRESENT TENSE OF TURKISH VERBS AND OF VERBS
OF TURKISH AND NON-TURKISH ORIGIN IN AJIA VAVARA ROMANI
(FROM IGLA 1991, 52)
present
sg 1
sg 2
sg 3
pl 1
pl 2
pl 3
Turkish
Ajia Vavara Romani
to write
yaz-ar-m
yaz-ar-sn
yaz-ar
yaz-ar-z
yaz-ar-snz
yaz-ar-(lar)
to write
yaz-ar-um
yaz-ar-sun
yaz-ar
yaz-ar-us
yaz-ar-sunus
yaz-ar-(lar)
to bring
an-av
an-es
an-el
an-as
an-en
an-en
(5) i
thagarni kurta-du len
e
rom-en
ART queen
save-PRT REL ART gypsy-ACC.PL
(..) the queen saved (them) the gypsies. Romani/Turkish (Igla 1991, 53)
This latter strategy is actually very rare in language contact situations.

However, it may be more common in the case of two closely related
languages or varieties. For instance Malkiel (1986) draws attention to
the Spanish conjugation pattern of a number of Portuguese verbs. Also
295
non-tensed verb forms such as participles, infinitives and imperatives

are quite readily interchangeable between languages, cf. (6).
(6) Le estaba
poniendo
atencin qu estaba
3.IO was.3SG put.PROGRPART attention what was.SG
recorded
record.PASTPAR
He was paying attention to what was recorded. Spanish/English (Pfaff
1979, 300)
3. Explanations for the distribution of the two common strategies

Why do speakers opt for one strategy rather than the other? The two
common strategies in particular, morphological integration and periphrastic constructions, allow us to compare the languages and linguistic
situations involved.
3.1
Characteristics of the matrix language
When one makes a list of all language contact situations and the attested
strategies for the integration of foreign verbs, it becomes clear that
genetic or areal typological factors play an important role. In the IndoIranian and Turkic languages, for instance, the periphrastic construction appears in virtually all language contact situations. In connection
with this, Muysken (2000, 194) speaks of a large linguistic area, in
this case, stretching from Sri Lanka to Greece. On the other hand, most
western European languages seem to prefer the morphological integration strategy. A plausible explanation for the areal bias is that speakers
who are accustomed to a certain strategy of incorporating foreign verbs
will reuse this strategy in new contact situations. Some examples of this
kind of bilingual knowledge will be discussed below. Still, the two MA
examples cited in (1) and (2) show that the matrix language is not the
only factor deciding which incorporation strategy speakers will use.
3.2
Characteristics of the embedded language
One way to explain the difference between MA/French and MA/Dutch

code-switching is to look for differences in the superimposed embedded languages. Heath (1989) argues that the phonological shape of the
French verbs facilitates their incorporation in Arabic. Bilingual speakers
associate the stressed final vowel /e/ of most French infinitives and other
296
louis boumans
verb forms with the final /i/ of the prefix conjugation (imperfective) of
a class of MA verbs.
Dutch infinitives typically end in an unstressed suffix -en that is pronounced as a schwa. It is not obvious whether this makes them phonologically less similar to the MA finite verb in /i/ than French infinitives.
Firstly, MA phonology does not allow for the schwa in open syllables.
This may lead Moroccan listeners to ignore word-final schwas in Dutch,
and interpret the infinitives as consonant-final. Alternatively, however,
Moroccans may interpret the Dutch final schwa as a full front or back
vowel. Both tendencies can be observed in the speech of Moroccan
learners of Dutch.
Moreover, the vocalic ending in French verbs cannot be decisive,
since Spanish, Italian and English verbs are morphologically integrated
in North African varieties of Arabic in the same way as French verbs.
Many of these Romance and English infinitives end in a consonant.
Arabic/Spanish language contact is still common in the (formerly)
Spanish occupied northern parts of Morocco (cf. Heath 1989, Herrero Muoz-Cobo 1996). Cohen (1912) notes interesting observations
on the Jewish dialect of Algiers, where the Spanish (or Lingua Franca)
infinitive ending -ar is even extended to embedded French verbs. E.g.
refuzarit from French refuser [rfyze] in (8). Numerous verbs of Italian
and English origin are found in Maltese, another variety of Maghribian
Arabic (Aquilina 1965 [1987], Camilleri 1994, Mifsud, 1995).4 The stem
extension -jain Maltese verbs of English origin, as in (10), is a reflection of the Sicilian ~ Italian infinitive suffix -are.
(7)
frin-ar-t
~ frin-ar.i-t
brake- INF-1SG ~ brake- INF.STEM EXTENSION -1SG (Sp. frenar)
I braked Tetouan Arabic/Spanish (Heath 1989, 184)
(8)
refuz-ar-it
refuse-SPANISH INF-1SG
(Fr. refuser)
I refused Jewish Arabic of Algiers/French (Cohen 1912, 432)
(9)
ti-ppartiipa-w
3-participate-PL (It. partecipare)
you (pl) participate Maltese/Italian~ Sicilian (Camilleri 1994, 437)
4
As a matter of fact, the type of verb integration illustrated in (1) does not occur in
the Middle Eastern varieties of Arabic. Instead the periphrastic construction is more
common in that region.
297
(10) ni-bbli.ja-ha
1-bleach.STEM EXTENSION-3SG
I bleach it Maltese/English (Camilleri 1994, 443)
Thus, verb stems from various embedded languages and with diverse
phonological characteristics can be morphologically integrated into
Moroccan or Maghribian Arabic, and phonology does not seem to be an
explanation for the periphrastic construction in the case of MA/Dutch.
A second way in which the embedded language might influence the
selection of the verb integration strategy is the use of periphrastic verb
constructions in the embedded language itself. Following this line of
thought, the bilingual MA/Dutch periphrastic construction could be
inspired by periphrastic constructions with doen to do in standard or
non-standard Dutch.5 Again, the comparison with the other language
pairs speaks against this hypothesis. Periphrastic constructions with a
do verb are much more common in English than they are in Dutch,
while their abundance in French and Spanish may be similar to that in
Dutch.6 Therefore, the occurrence of do constructions in monolingual
Dutch likewise does not explain why the periphrasis strategy is chosen
to insert Dutch verbs in MA matrix clauses.
3.3
Characteristics of the sociolinguistic setting
In the case of the North African language contacts, neither the host language nor the embedded language is the sole factor determining the way
in which the loan verbs are integrated in Arabic. For this reason I conjecture that the sociolinguistic setting in the Netherlands, where MA is
a minority language, is responsible for the periphrastic construction in
MA/Dutch. As a more general hypothesis, I suggest situations in which
language contact is more intense, like migration to an urban industrialised society, favour the use of the periphrastic construction.7
5
Jacomine Nortier and Roeland van Hout made this suggestion when we discussed
this paper at the SS14 workshop on borrowing in Gent, April 4, 2002, and on earlier
occasions.
6
With the exception of certain infrequent constructions, do-periphrasis in Dutch is
associated with non-standard regional varieties or child language. Cf. Nuijtens (1962,
15457), Giesbers (1984), and Cornips (1994).
7
One could argue that the morphological integration of French verbs is found both
in North Africa and in the North African diaspora in Francophone Europe and Canada. In this particular case, however, code-switching strategies that were already established in the homeland have simply been maintained in the diaspora.
298
louis boumans
In order to shed some light on this issue and to test the intensity of
contact hypothesis, the following two sections survey verb integration
in two additional language pairs for which both integration strategies
are attested: Greek/English and Portuguese/English. In both cases the
influence of English has been described for immigrant communities in
Anglophone countries as well as in the Greek and Portuguese speaking
homelands.
4. Greek/English
4.1
Greece
The morphological integration of Italian verbs consists of adding Greek

suffixes to the Italian infinitive in -are.8
(11)
It. posare to pose
It. schizzare to sketch

Mainland Greek/Italian (Van Dijk-Wittop Koning 1963)
(12)
fundro
It. fondare to anchor
barkro
It. barcare to board
kompletro It. completare to fill up [cargo]
Mainland Greek/Italian (Hartley 2001)
According to Swanson the mildly productive verbal suffix -aro entered

the Greek language in late Byzantine times (1958, 40). The Italian infinitive marker has been generalised to verbs of French and English origin
as well:
(13)
v Fr. dbuter to make ones debut
Fr. lancer to launch

Mainland Greek/French (Contossopoulos 1978, 42)
(14)
flertaro
to flirt
stoparo
to stop (a machine)
sutaro
to make a shot (soccer, basketball)
Mainland Greek/English (Swanson 1958)
8
The authors on loanwords and codeswitching in Greek make use of different writing and spelling conventions. The spelling of the source publication is retained in the
examples cited here.

(15)
299

/parkaro/
to park

/manadzaro/ to manage

/triparo/
to trip
Mainland Greek/English (Apostolou Panara 1991, 50)
Interestingly, the recycling of a foreign verb marker like the Italian

-are to mark verbs from a third language is yet another example of how
bilingual experience influences the way in which speakers treat words
from foreign languages. Note the close parallel with -ar in the Jewish
Arabic dialect of Algiers (8), and the -ja stem extension after English
verb stems in Maltese (10).9
Most sources on loan verbs in Modern Greek mention only this
-ar- extension with complete morphological integration of the foreign
word. However, Apostolou Panara, while asserting that verbs all adapt,
i.e. are morphologically integrated, mentions that in some cases, along
with the single lexeme we attest a periphrasis with the Greek verb /kano/
(to do, to make) preceding a noun or a gerund (1991, 50), cf. the examples in (16). A few verbs occur exclusively as part of the periphrastic
construction (17), whereas the periphrasis is judged ungrammatical in
most cases (18).
(16)
/flertaro/
besides /kano flert/
to flirt
/manadzaro/ besides /kano manadzing/ to manage
/stokaro/
besides /kano stok/
to stock
(17)
/kano kambing/ but not */kambaro/ to camp

/kano serfing/ but not */serfaro/
to surf
(18)
/triparo/ but not */kano trip/ to trip

/parkaro/ but not */kano park/ to park
Mainland Greek/English (Apostolou Panara, 1991, 51)
It is possible that a closer study of the English vocabulary in Modern

Greek would reveal regularities that govern the use of periphrasis.
Periphrasis with kno could be associated with particular types of events,
for instance habitual or durative activities. The periphrastic construction may be more common with recent English loans, or in the speech
9
In Dutch a similar situation obtains with the French infinitive marker -er [er] (Treffers-Daller 1994), which is sometimes used for English loanwords as well, e.g. formatter-en to format.
300
louis boumans
of speakers who are more exposed to English. kano kambing may be a

calque after the French expression faire du camping (cf. kno ski in (24)
below, possibly after French faire du ski). The available data do not allow
for such conclusions at the moment. Still, I think it is safe to conclude
from the literature that in Greece the periphrastic construction occurs
as a minor pattern with some verbs only, while morphological integration is the rule.
4.2
Cyprus
Cyprus has been under British rule for more than eighty years, during
which English was the language of administration. First, from 1878 on
Britain administered the island in agreement with the Ottoman Empire.
Then Cyprus was annexed by Britain when the Ottoman Empire enters
into World War I on the side of Germany, and subsequently the island
became a British Crown colony under the British rule. In 1960 Cyprus
gained independence, and (Modern Standard) Greek became the language of administration. The English language remained influential
through the tourism industry and the large international community
on the island.
There are a number of studies dealing specifically with the influence
of English on Cypriot Greek. Papapavlou (1997) cites a list of English
loan words found in written sources and tape-recorded speech. His list
includes 23 verbs in the -aro conjugation. He makes no mention of the
periphrastic construction.
(19)
flrdro
to flirt
rejistrro to register
riskro
to take a risk
arcro
to change (money)
to check (inspect)
akhro
Cypriot Greek/English (Papapavlou 1997)
Similarly, Evripidou, also on the basis of written sources and recorded

speech, lists 19 English-origin Cypriot Greek verbs, all of them integrated in the -aro class (2001). Goutsos discusses Greek/English codeswitching among members of a middle-class Cypriot family in Limassol
(2001). This study differs from the other ones, as Goutsos does not list
English loan words but rather focuses on the discourse functions of
language choice in the conversations. He does not comment on integrated verbs of the -aro conjugation. He does cite three instances of the
301
periphrastic construction. As in Apostolou Panaras data, there are two

types of construction, one in which the English verb is represented by
the infinitive (or verb stem), see (20) and (21), the other with the gerund (22).
(20)
a kmno shower
I will shower
Cypriot Greek/English (Goutsos 2001, 203)
(21)
i jti tan kmnis wash

no because when you wash
(22)
. . . llun tus pu na rti i jaj su en kmni swimming

I was saying to them, when your grandma comes, she will swim.
The periphrastic construction is not a pervasive phenomenon in the

speech of Goutsos informants, nor is code-switching in general (Goutsos 2001, 216). The examples also show the absence of an established
convention on how to insert the English verb into the periphrastic construction. Still it is noticeable that the do-construction is recorded on
Cyprus and not in mainland Greece. Goutsos comments: Compared
to SMG [Standard Modern Greek], CG [Cypriot Greek] draws more on
English resources for borrowing and creates mixed compound forms
with de-lexicalized verbs (2001, 204).
Although Apostolou Panara shows that the periphrastic construction does occur in mainland Greek, Goutsoss comment confirms the
impression that it is a marginal phenomenon. If it is indeed true that
the periphrasis is more common in Cyprus, it is attractive in the light
of the present discussion to explain this difference as resulting from the
greater impact of English on the language of the former crown colony.
Moreover, Goutsoss examples stem from a really bilingual setting,
whereas the loanwords collected by the other authors stem mostly from
written sources that do not presuppose a bilingual readership.
4.3
Mainland Greek in the Diaspora
The oldest study on American Greek mentions just one English-origin

verb: mouvaro to move (Lontos 192526). However, this particular
form probably goes back to an Italian form muevere, as Hartley (2001)
suggests in his paper on loan words in Greek nautical terminology.
302
louis boumans
A much more elaborated study of American Greek is provided by Seaman. He mentions the same form muvro (1972, 165), but it becomes
clear that the periphrastic construction with kno to do, make is the
productive way to incorporate English verbs in his data: In verbal loans
from English, /jno/ seems to be the auxiliary for passives and /kno/ for
actives (1972, 166). His examples of the passive construction with jno
to be, become are ambiguous, as they might also be analysed as cases
of a copula plus predicate, and there are only a handful of examples. On
the other hand, Seaman cites 47 examples of the periphrastic construction with kno. In 33 cases the embedded English element is unambiguously a verb; 11 cases may involve either a verb or a noun, e.g. control,
welcome. Some examples are reproduced below (see also Muysken 2000,
212).10
(23)
kno, cover up
I cover up
knis, brush ta ndja su You (sg) brush your teeth
kane, punch
s/he punched [a meal ticket]
American Greek (Seaman 1972, 16768)
Tamis found the same construction in Australia (1986, cited in Muysken

2000, 212). The construction is also known from Canada (24) and Brussels (25). Compared with Seamans study, the information on the latter
diaspora communities is rather scarce. It is not clear whether in these
cases the periphrastic construction is really the productive strategy for
incorporating English or French verbs. The Canadian examples cited
in (24), involving computer terminology and sports, remind of the
use of the periphrastic construction in Greece described by Apostolou
Panara.
(24)
kno ski
kno save [computer]
kno print [computer]
kno jogging
Montreal Greek/English (Muysken 2000, 213, based on Hatzidaki p.c.)
(25)
kno dmnager
to move house
Greek/French (Muysken 2000, 213, based on Hatzidaki p.c.)
10
In Table 7.7 on p. 212 of Muyskens book the words jno and kno have been
reversed.

4.4
303
Cypriot Greek in London
There are a number of publications (and many unpublished theses) on the

language of the Greek Cypriot community in London. Gardner-Chloros
(1992) mentions the occurrence of morphologically integrated verbs
like tekaro to check, which is not surprising as these words are found
in Cyprus (and Greece) as well. In addition, however, she points out the
periphrasis with kno as an innovative construction, occurring with English adjectives, nouns, participles . . . (Gardner-Chloros 1992, 127).
(26)
kamno use
to use
kamno respect
to respect
kamno developed to develop
kamno spelling
to spell
Cypriot Greek in London (Gardner-Chloros, 1992, 127)
Zarpetea compares the language use of three generations of London

Cypriots (1995). She notices the use of both integrated loan verbs and
the periphrastic construction in the first as well as the second-generation
immigrants. The periphrastic construction still occurs in the speech of
the third generation, young children who use only little Greek:
(27)
wash cups, do you want me to do the plates now mum?

I have washed the cups, . . .
Cypriot Greek/English (Zarpetea 1995, 581)
As a final note on morphologically integrated English verbs in immigrant Greek, I would like to draw the attention to the fact that there is
much overlap between the lemmas mentioned in the various sources.
This is particularly striking in Zarpeteas paper. Out of the seven verbs
she cites, five also occur in other sources, cf. Table 2.
TABLE 2
OVERLAP BETWEEN LOAN VERBS IN ZARPETEA (1995)

AND OTHER SOURCES
Zarpetea 1995, 578
other sources
to cancel
Gardner-Chloros 1992, 126, Papapavlou 1997,

232, Evripidou 2001, 24
Papapavlou 1997, 240, Evripidou 2001, 24
Papapavlou 1997, 240
Apostolou Panara 1991, 50, Papapavlou 1997,
236
Lontos 192526, 309, Seaman 1972, 167, Hartley
2001

to check
to charge
to park
to move
to book
to packet
304
louis boumans
There is no such overlap in the examples cited for the periphrastic construction with kmno (mainland Greek kno).11 This indicates that the
morphologically integrated verbs found in London may well have been
coined in the more monolingual setting in Cyprus. As noted above,
to move is probably not even of English origin. This is a further indication that in the Cypriot diaspora, the periphrastic construction replaces morphological integration as the most productive strategy
for the incorporation of new English verbs. The chances are that the
same development has taken place in the American Greek community
studied by Seaman (1972).
5. Portuguese/English
I have been able to trace surprisingly few studies on English loan words
in Portuguese, whether European or Brazilian. Compared with Greek,
there are also few studies on Portuguese speaking communities in
Anglophone countries. The picture that arises from the available data is
approximately the same as for Greek.
5.1
Portugal
The website of the Instituto Superior Politcnico de Viseu in Portugual

hosts a short list of English loanwords in Portuguese, compiled by two
students and their English teacher (Queiroz, Rodrigues and McKenny
1999, accessed June 2002). The authors are not explicit about their data
sources, except that visitors of the web page are invited to contribute
observations. As it concerns a Portuguese project, however, I assume
that most of their sources are likewise of Portuguese, rather than Brazilian, origin.
Their list of English loan words contains four verbs integrated in
the Portuguese conjugation class ending in -ar (28), in addition to the
11
Apostolou-Panara, Goutsos, Gardner-Chloros and Zarpatea cite 14 different English verbs in the periphrastic construction. kmno wash is shared between Goutsos and
Zarpatea. kmno use is cited in the papers by Gardner-Chloros and Zarpetea. But these
two papers must be based on (partly) the same data. I infer this from the fact that the
example
very busy I know that you have (are) very busy is cited in both
(Gardner-Chloros 1992, 128; Zarpatea 1995, 578).
305
denominal form stressar (from stress) and the de-adjectival compactar

(to compress from compact, computer terminology).
(28)
draftar to draft
driblar to dribble
linchar to lynch
snifar to sniff [drugs]
European Portuguese/English (Queiroz et al. 1999)
In addition this list contains six deverbal English nouns denoting activities in the domains of sports and information and communication technology: jogging, kickboxing, surf; scan, upgrade and zap. The web site
does not indicate how these words are used in Portuguese. Nicholas
Hurst (p.c. 18 Apr 2002) cites examples of periphrastic constructions
involving the Portuguese verb fazer to do (29). The examples show that
there is much variation: as in Greek contexts, the English verb is used in
either the citation form (infinitive) or the gerund. Moreover, the indefinite article um may or may not be present.
(29)
5.2
fazer zapping
fazer um scanning
fazer surf
fazer um upgrade
European Portuguese/English (p.c. Nicholas Hurst 18 Apr 2002)
Brazil
Kennedy (1971) and Harmon (1994) quote a number of English verbs

occurring in Brazilian dictionaries and other written sources. All are
integrated into the so-called first conjugation class ending in -ar:
(30)
chutar
to shoot [ball game]
drenar
to drain
ranquear to rank
treinar
to train
Brazilian Portuguese/English (Kennedy, 1971)
Harmon does not mention the construction with fazer. Kennedy cites
just one example of a somewhat enigmatic periphrastic construction:
fazer o footing, translated in English as take a stroll.
306
5.3
louis boumans
American Portuguese
There is a classical study of the language of Portuguese immigrants

in the US by Pap (1949). He gives many examples of morphologically
integrated English verbs. These are added to the most frequent en most
productive Portuguese conjugation class in -ar, with the necessary phonological adaptations. In some cases the English verb stem is extended
with an -e- such that the infinitive ends in -ear.
(31)
raid-ear
to ride
canec-ar to connect
damp-ar to dump
s(e)leir-ar to slide
chinj-ar
to change
raiv-ar
to drive
North American Portuguese/English (Pap 1949, 95100)
While morphological integration of English verbs is the most common

pattern, Pap also lists a large number of instances of a periphrastic construction with fazer. The nominalized English verb stem occupies the
direct object position. In contrast with the contemporary European
examples cited by Hurst, the foreign verb stem is preceded by the Portuguese masculine definite article o.
(32)
fazer o boda
to bother
fazer o spoil
to spoil
fazer o save
to save
fazer o find out to find out
fazer o give up to give up
North American Portuguese/English (Pap 1949, 105106)
6. Discussion
The data from North African varieties of Arabic in contact with European languages suggest that diaspora communities are more inclined
to use the periphrastic alternative for incorporating foreign verbs. The
studies on Greek and Portuguese in contact with English do not unambiguously confirm this. Morphological integration as well as periphrastic constructions are reported for all contact situations, whether in the
Greek and Portuguese speaking countries or in the respective immigrant communities in Anglophone countries.
307
Broadly speaking, however, the literature on Greek and Portuguese

does point in the same direction. Integration in the productive conjugation classes, Greek -ro and Portuguese -ar, is predominant in the
studies on English loanwords in Greece, Cyprus, Portugal and Brazil.
Do-constructions, on the other hand, feature prominently in the studies on migrant communities.
A caveat is in order: studies on the influence of English as a foreign
language are traditionally oriented toward loan words, i.e. words in
isolation, in written sources. These studies tend not to identify foreign
verbs in the periphrastic construction as verbs at all, which can easily
lead to the suggestion that embedded verbs are either absent or consistently integrated into the morphology of the matrix language. For obvious reasons, studies on bilingualism in migrant communities focus on
oral communication. Thanks to the recognition of code-switching as
an interesting linguistic phenomenon, the latter studies also pay more
attention to the higher degrees of linguistic organisation like clauses and
utterances. Hence the emphasis on one verb strategy or the other may be
to some extent an artefact of the different research traditions.
Still, at least the Greek case supports my hypothesis, in particular
the studies by Apostolou Panara on Greece and Seaman on the Greek
community in the US. Both authors discuss the different types of verb
integration, and designate one as the default strategy: morphological
integration in Greece and periphrasis in the US. The studies on the London Cypriots suggest that periphrasis is the most productive strategy
there as well.
With respect to the sociolinguistic factors in verb integration, I conclude that, all other things being equal, morphological integration is
the norm in homeland settings where the matrix language is socially
dominant, while the periphrastic construction is found in migrant communities where the embedded language is dominant. The migrant communities concerned here are actually undergoing language shift. Stated
in more general terms, the periphrastic construction characterises situations of intense contact with the language of the embedded foreign verb.
As mentioned above, other factors may override the sociolinguistic
ones, as in the case of matrix languages that opt for the same integration
strategy in no matter what sociolinguistic setting.
308
6.1
louis boumans
Automatization of superimposed language practices
The next question is why speakers more often revert to periphrasis

in situations of more intense contact. I can think of the following
explanations:
Firstly, automatization of superimposed language practices: In general the degree of integration of foreign words is inversely proportional
to the speakers knowledge of the embedded language. With increasing
practise and competence, the linguistic habits characteristic of that foreign language become entrenched in the speakers brain, and can more
easily be reproduced. This even results in the disintegration of formerly
integrated foreign words, a process known as denativization (Haugen
1950). In Morocco, for instance, French-Moroccan Arabic bilinguals
now prefer to reborrow the same French terms in shapes closer to the
French prototype (Heath 1986, 114). Denativization has often been
described with respect to phonology of loan words from English, a language whose influence is expanding (cf. Khubchandani 1968, Hasselmo
1969, De Reuse 1994, Van Ness 1994). I would like to stress this point,
because it is a commonly held belief that loanwords become, on the
contrary, more integrated over time (Nortier and Schatz 1988, Heath
1989).
The replacement of morpho-phonological integration by periphrasis
as the productive strategy for the integration of foreign verbs fits into
the denativization process, since the periphrastic construction entails
little or no phonological or morphological adaptation to the matrix language. This type of change could be taking place in for instance Greece
or Portugal, where exposure to English is still increasing.
Denativization refers to a diachronic change in situations where language contact increases over time. However, the underlying mechanism
of automatization of superimposed language practices also explains the
differences between immigrant and homeland bilingualism.
6.2
Loss of community language practices
Secondly, attrition or loss of community language practices: the morphological integration of foreign verb requires the application of matrix
language morphological procedures. These procedures are typically the
same as those used for deriving verbs from nouns or adjectives. The
loss or erosion of community language morphology, a common effect
of decreased use of the language and incipient shift (Andersen 1982,
309
El Aissati 1997), cuts off morphological integration as a strategy for

embedding foreign verbs.
6.3
Collocational complements
Thirdly, triggering by collocational complements of the embedded verbs:

Many examples of periphrasis with foreign verbs involve a collocation
with a foreign direct object noun or other complement, cf. wash and
cups in Greek/English example (27) above and diep, gesprek and voeren
in the Arabic/Dutch example below.
(33)
ma
ne-qder
eh n-dir
NEG 1-can NEG er 1-do
diep-e gesprekk-en
voer-en
deep-AGR conversation-PL conduct-INF
I cant er carry out deep conversations.
MA/Dutch, Jamal, 20, Eindhoven 1991 (Boumans 1998, 245)
The periphrasis leaves the original embedded language collocation

intact. Therefore the collocation might trigger this construction. The
usage of foreign collocations presupposes a higher level of competence
in that language and hence more intense exposure to it.
7. Conclusions
In nearly all cases, foreign verbs are combined with matrix language
inflection in either of two ways: they are morphologically integrated and
inflected with ML affixes (if any), or they are embedded in a periphrastic
construction with an inflected ML verb. Some MLs, such as Magribian
Arabic, Portuguese and Greek, allow for both solutions. A comparison
of Magribian, Portuguese and Greek bilingual communities suggests
that the periphrastic integration of foreign verbs is favored in situations
of more intense contact with the superimposed language.
8. References
Andersen, R. 1982. Determining the Linguistic Attributes of Language Attrition. In
R. Lambert and B. Freed, eds. The Loss of Language Skills. Rowley, MA: Newbury
House, 83118.
Apostolou Panara, A. 1991. English Loanwords in Modern Greek: an Overview. Terminologie et Traduction, 1.91, 4560.
310
louis boumans
Aquilina, J. 1965 (1987). Teach Yourself Maltese. Valetta: Hodder and Stoughton.
Backus, A. 1996. Two in One: Bilingual Speech of Turkish Immigrants in the Netherlands.
Tilburg: Tilburg University Press.
. 2001. Synchronic Register Variation in Monolingual and Bilingual Turkish (project
description).
Boumans, L. 1998. The Syntax of Code-switching. Analysing Moroccan Arabic/Dutch
Conversations. Tilburg: Tilburg University Press.
. 2000. Periphrastic Verb Constructions in Moroccan Arabic/Dutch Code-switching. In A. Fenyvesi and K. Sandor, eds. Language Contact and the Verbal Complex of
Dutch and Hungarian. Working Papers of the 1st Bilingual Language Use Theme Meeting. Szeged: University of Szeged, Teacher Training College, 6784.
. 2002. Repetition Phenomena in Insertional Codes-witching. In A. Rouchdy, ed.
Language Contact and Language Conflict Phenomena in Arabic. London: Curzon,
279316.
, and Caubet, D. 2000. Modelling Intrasentential Code-switching: a Comparative
study of Algerian Arabic/French in Algeria and Moroccan Arabic/Dutch in the Netherlands. In J. Owens, ed. Arabic as a Minority Language. Berlin: Mouton de Gruyter,
pp. 11380.
Camilleri, A. 1994. Language contact between Maltese and English: code-switching
and cross-linguistic influence. In D. Caubet and M. Vanhove, eds. Actes des premires
journes internationales de dialectologie arabe de Paris. Paris: INALCO, 43149.
Cohen, M. 1912. Le parler arabe des juifs dAlger. Paris: H. Champion.
Contossopoulos, N.G. 1978. Linfluence du franais sur le grec: emprunts lexicaux et
calques phrasologiques. Athnes: s.n.
Cornips, L. 1994. De hardnekkige vooroordelen over de regionale doen+infinitiefconstructie. Forum der Letteren, 35.4, 28294.
De Reuse, W. 1994. English Loanwords in the Native Languages of the Chukotka Peninsula. Anthropological Linguistics, 36.1, 5668.
El Aissati, A. 1997. Language Loss among Native Speakers of Moroccan Arabic in the
Netherlands. Tilburg: Tilburg University Press.
Evripidou, D. 2001. Lexical Borrowing: A Study of English Loanwords in the Greek Cypriot
Dialect. Lancaster: Lancaster University.
Gardner-Chloros, P. 1992. The Sociolinguistics of the Greek Cypriot Community in
London. Plurilinguismes, 4, 11236.
Giesbers, H. 1984. Doe jij even lief spelen? Notities over het perifrastisch doen. Mededelingen NCDN, 19, 5776.
Goutsos, D. 2001. A Discourse-analytic Approach to the Use of English in Cypriot
Greek Conversations. International Journal of Applied Linguistics, 11:2, 194223.
Harmon, R.M. 1994. Aspectos lingsticos dos emprstimos em portugus. Hispania,
77:3, 463469.
Hartley, A.H. 2001. Loanwords in Modern Nautical Greek. (2002).
Hasselmo, N. 1969. On Diversity in American Swedish. Svenska Landsml och Svenskt
Folkliv, 92, 5372.
Haugen, E. 1950. The Analysis of Linguistic Borrowing. Language, 26:2, 21031.
Heath, J. 1986. Hasta la Mujerra! and other Instances of Playful Language Mixing in
Morocco. Mediterranean Language Review, 2, 1136.
. 1989. From Code-Switching to Borrowing. A Case Study of Moroccan Arabic. London: Kegan Paul International.
Herrero Muoz-Cobo, B. 1996. El rabe marroqu: aproximacin sociolingstica. Almera: Universidad de Almera.
Igla, B. 1991. On the Treatment of Foreign Verbs in Romani. In P. Bakker and M. Cortiade, eds. In the Margin of Romani. Gypsy Languages in Contact. Amsterdam: Inst.
ATW, Univ. Amsterdam, 505.
311
Kennedy, J.H. 1971. The Influence of English on the Vocabulary of Brazilian Portuguese. Hispania 54:2, 327.
Khubchandani, L.M. 1968. The Gender of English Loanwords in Sindhi. In B. Krishnamurti, ed. Studies in Indian Linguistics. (Professor M.B. Emeneau sastipurti volume).
Poona: Deccan College, 18088.
Lontos, S.S. 192526. American Greek. American Speech, I, 30710.
Malkiel, Y. 1986. A Spanish Conjugational Model Superimposed on Portuguese. Mediterranean Language Review 2, 5166.
Mifsud, M. 1995. Loan Verbs in Maltese. A Descriptive and Comparative Grammar.
Leiden: E.J. Brill.
Muysken, P. 2000. Bilingual Speech. A typology of Code-Mixing. Cambridge: Cambridge
University Press.
Myers-Scotton, C. 1993. Duelling Languages. Grammatical Structure in Codeswitching.
Oxford: Clarendon Press.
Nortier, J. and Schatz, H. 1988. Van nwoordwisseling naar ontlening, een vergelijkend onderzoek. Forum der Letteren, 29, 16178.
Nuijtens, E.T.G. 1962. De tweetalige menseen taalsociologisch onderzoek naar het
gebruik van dialect en cultuurtaal in Borne. Assen: Van Gorcum.
Pap, L. 1949. Portuguese-American Speech. New York: Kings Crown Press.
Papapavlou, A. 1997. The Influence of English and its Dominance in Cyprus: Reality or
unfounded fears? Journal of Mediterranean Studies, 7:2, 21849.
Queiroz, A., Rodrigues, J. and McKenny, J. 1999. Emprstimos da lngua inglesa para a
portuguesa. English loan words and phrases in Portuguese. Escola Superior de Educao de Viseu. 2002.
Seaman, P.D. 1972. Modern Greek and American English in Contact. The Hague:
Mouton.
Swanson, D. 1958. English Loanwords in Modern Greek. Word, 14, 2646.
Tamis, A. 1986. The State of the Modern Greek Language as Spoken in Victoria. Melbourne, University of Melbourne.
Thomason, S.G. and Kaufman, T. 1988. Language Contact, Creolization and Genetic Linguistics. Berkeley: Univ. of California Press.
Treffers-Daller, J. 1994. Mixing Two Languages. French-Dutch contact in a comparative
perspective. Berlin: Mouton de Gruyter.
Van Dijk-Wittop Koning, A.M. 1963. De continuteit van het Grieks: onderzoek naar de
herkomst van de woordvoorraad van het Nieuwgrieks. Zwolle: W.E.J. Tjeenk Willink.
Van Ness, S. 1994. Die Dimensionen lexicalischer Entlehnungen im Pennsylvaniendeutschen von Ohio (USA). Zeitschrift fr Dialektologie und Linguistik, 61.3,
27997.
Van Staden, M. 1999. Where does Malay end and Tidore begin? In R.A.C. Dam, ed.
Perspectives on the Birds Head of Irian Jaya, Indonesia. Proceedings of the conference
Leiden, 1317 October 1997. Amsterdam: Rodopi, 691716.
Wernitz, C.J. 1993. Bedingungen und Voraussetzungen fr Sprachwechsel. Eine Untersuchung zum Sprachwechsel bei bilingualen Marokkanern in Frankreich. Frankfurt: Peter
Lang.
Zarpetea, P. 1995. Code-switching and Lexical Borrowing (loanwords) in the Speech
of Three Generations of Greek Cypriots in London (Harringey). Studies in Greek
Linguistics. Proceedings of the 16th Annual Meeting of the Department of Linguistics.
Thessaloniki: Aristotle University, 57687.
FAULA, FAILA, FAALA :

DISPERSION ET RGULARITS SMANTIQUES
DANS LES TROIS SCHMES SIMPLES DU VERBE ARABE
Joseph Dichy
Universit Lumire-Lyon 2
Pour Kees Versteegh

Aut cuspis sic vita fluit, dum stare videtur.
1. Introduction1
Les schmes du verbe augment (mazd) en arabe sont traditionnellement associs des valeurs smantiques de base, telles que litration,
lintensivit, la causativit et la factitivit, la rflexivit, etc. Si les listes dexceptions qui accompagnent ces descriptions indiquent que le
problme, malgr des avances significatives, est loin dtre rsolu de
manire satisfaisante, la question des valeurs smantiques associables
aux trois schmes simples (mujarrad) ou non-augments faula, faila et
faala2 semble offrir encore plus de rsistance lanalyse.
La description de ces schmes dans les grammaires arabisantes est,
lorsquelle existe, bien plus sommaire que pour les verbes augments.3
1
Je reprends ici, sous une forme largement revue, le chapitre 3 de mon cours de prparation lagrgation darabe du Centre national denseignement distance (CNED),
2002 et 2003. Par souci dexplicitation, jai prsent quelque peu en dtail le cadre conceptuel ncessaire au traitement de cette difficile question, et conserv des dfinitions ou
des indications qui pourront paratre videntes, mais ne le sont pas ncessairement.
Ce travail doit beaucoup au cadre thorique et aux travaux de Jean-Pierre Descls,
ainsi quaux sminaires que nous avons anims, et continuons danimer ensemble,
depuis 2001, sur les valeurs associes en contexte aux formes verbales en arabe et en
franais. Je remercie galement les membres de son quipe de recherche, notamment
Brahim Djioua, pour de nombreuses et fcondes discussions.
2
La convention qui prvaut dans la tradition linguistique arabe dsigne ce que nous
appelons les consonnes radicales 1, 2 et 3 (de la racine smitique) respectivement par
f, et l.
3
Ainsi pour les grammaires en langues occidentales : Neyreneuf et Al-Hakkak (1996,
3542) signalent des nuances de sens . . . pour chaque forme drive, mais non pour
les verbes simples (id., 2829). De mme, les descriptions consacres par Blachre et
314
joseph dichy
Il suffit dobserver quils demeurent dsigns dans la tradition arabisante occidentale comme la Forme I. Cette dsignation relve dune
ide gnrale qui rpond, dfaut de dcrire les donnes linguistiques,
une cohrence interne. La Forme I est appele verbe primitif par A.
Sylvestre de Sacy, par rfrence au verbe nu mujarrad des grammaires
arabes (1831, I : 123). A cette forme primitive, Sylvestre de Sacy associe
les grandes divisions smantiques gnrales des verbes (id. 121122) sur
lesquelles il fondera ensuite son analyse des formes drives (id. 129143).
Mais il nidentifie pas de valeurs smantiques associes en propre aux
schmes faula, faila et faala, dans lesquelles il ne voit quune forme
primitive ou nue (mujarrad) partir de laquelle les formes drives ou
augmentes (mazda) sont construites.
Si la tradition arabisante a pu ainsi rapporter ces trois schmes une
seule forme, cest sans doute aussi pour une autre raison : ces derniers
apparaissent en effet comme instables, tant du point de vue formel, o lon
observe des modifications de la voyelle de la 2e radicale (section 2 ci-dessous), que du point de vue smantique, o les notions trs gnrales de
transitivit, de verbes daction ou dtat par lesquelles on a essay de les
caractriser rencontrent un nombre lev de contre-exemples. Cest ainsi
que les trois pages dindications essentiellement formelles consacres
cette question par la grammaire da-artn sont prcdes de cet avertissement : On ne peut tablir [la forme des verbes relevant de] ces schmes
(awzn) quen ayant recours des ouvrages lexicographiques (1910, 11).
Faut-il donc renoncer ? Il y a de solides raisons de penser que non. Plusieurs essais de mise en ordre et danalyse ont t tents. La synthse la
plus rcente de ces tentatives est, ma connaissance, celle de Pierre Larcher (2003, chap. II), qui comporte en outre des propositions originales. Signalons galement les analyses incontournables dAndr Roman
(1983, II : 886900). Dautres auteurs, au premier rang desquels Marcel
Cohen, Paul Joon, Jean Cantineau et Henri Fleisch ont t utiliss. En
outre, un nombre considrable dobservations et danalyses a t effectu
dans les textes des sciences linguistiques arabes mdivales. Je ferai largement usage de ces travaux, sans toutefois les discuter en dtail, pour
Godefroy-Demombynes (1952, 4970) aux formes drives contrastent avec labsence
danalyse de la Forme I. Belot, (1922, 15), Badawi, Carter et Gully (2004, 60) ne signalent pour la forme primitive ou de base (al-fil al-mujarrad), de manire trs prudente, que la transitivit ou son absence en fonction de la voyelle mdiane de faala,
faila et faula. On trouve en revanche des propositions de description du sens dans
Caspari-Uricoecha (1881, 3233), Wright (189698 I : 30), Brockelmann (1948, 35) ou
Boormans (1967, 10), ainsi, naturellement, que dans la tradition linguistique arabe (voir
la synthse de Nr ad-Dn 2002, 177186).
faula, faila, faala
315
viter dallonger indment cette tude. A lenttant problme pos par la

dispersion smantique des verbes de schme simple en arabe jai cherch
apporter une solution fonde sur une dfinition aussi rigoureuse que
possible des termes utiliss, ainsi que sur une conception, me semble-t-il,
renouvele de la relation entre sens et forme dans les schmes (au sens
que prend ce terme dans le domaine smitique) en arabe (Dichy 2003).
Les propositions qui suivent sappuient en outre sur un dpouillement
systmatique des donnes lexicales dans le roman de Tawfq al-H akm,
Yawmiyyt nib f l-aryf (abrg en Yawm.) dune part, et dans plusieurs dictionnaires arabes de lautre (notamment Hans Wehr-Cowan
1979 et al-Mujam al-Wst). Jai fait largement usage de la base de donnes lexicale DIINAR.1 (Dictionnaire informatis de larabe, version
1 Dichy, Braham, Ghazeli et Hassoun 2002 ; Dichy et Hassoun 2005 ;
http ://www.elda.org), du concordancier informatis construit par
R. Abbs (2004), ainsi que de la liste indexe de 20.000 verbes ralise pour les besoins du Bescherelle des Verbes arabes (Ammar et Dichy
1999). Les rfrences ces dernires sources ne sont pas donnes, la
diffrence des textes arabes mdivaux, des travaux arabisants de lpoque moderne ou, au besoin, des dictionnaires ; Yawm est indiqu avec
le numro de la page suivi de la ligne lorsque le sens doit tre interprt
en fonction du contexte.
2. Les schmes du verbe non augment, aspects formels

2.1
Lalternance vocalique de la voyelle de la 2e radicale
Les trois schmes de base du verbe simple ont t identifis dans

les sciences mdivales arabes du langage partir de la voyelle de la
deuxime radicale ou voyelle mdiane du verbe au paradigme du md
(ou suffix),4 respectivement u, i et a. Les trois familles morphologiques correspondantes sont numres sous forme dexemples (amtila)
4
Les termes daccompli et inaccompli, dachev et dinachev ou de perfectif et imperfectif, pourtant reus, comportent un problme, celui de dsigner la forme morphologique par un terme dnotant lune des valeurs aspectuo-temporelles que cette forme
peut prendre en contexte. Do le choix de dsigner ces paradigmes verbaux par leur
nom arabe ou par le trait morphologique qui les caractrise. Ainsi, le md (traditionnellement accompli), dans lequel le morphme de personne prcde la base du verbe,
correspond ici au suffix ; le mudri (traditionnellement inaccompli), dans lequel ce
morphme prcde la base, est dsign du terme de prfix (cf. Moscati, d. 1964, 1312 :
prfix-conjugation vs suffix-conjugation).
316
joseph dichy
par Sbawayhi (m. vers 180/796) ds lintroduction du Kitb (I : 12), et

reprises au cours de ce trait au moyen des conventions faala, faila et
faula (dsignes comme des formes construites, abniya) avec des indications morphologiques, syntaxiques et smantiques (Kitb IV : 567,
notamment 38). Je laisse de ct ici la question, qui a t pose, de savoir
si la forme sujet non exprim (dite passive) fuila ne constituerait pas
un schme part entire.5
Selon la reconstruction du proto-systme de la langue arabe, telle
quelle a t propose en diachronie (et en tout cas dans lorganisation
qui sous-tend le systme de la langue), la forme simple du verbe comporte trois schmes, reprsents dans le tableau ci-dessous (voir notamment Cantineau 1950, 77 ; Fleisch 1957 et 1968, 115119 ; Roman 1983,
II : 894) :
Md. ( suffix)
Mudri'( prfix)
Schme simple 1
faula
yaf ulu
Schme simple 2
faila
yaf alu
Schme simple 3
faala
yaf ilu
FIGURE 1
LES SCHMES POSTULS DU VERBE SIMPLE DANS

LE PROTO-SYSTME DE LARABE
Selon Moscati, d. (1964 : 122), qui se situe dans la perspective des tudes
smitiques compares, lantiquit de ce schma vocalique trois termes en arabe est confirm par quelques-unes des manifestations les plus
anciennes du groupe smitique du Nord-Ouest, cest--dire, par lamorite,
lougaritique, et les gloses de Tell Amarna . . .6 Dans cette reconstruction,
comme le souligne Roman (loc. cit.), les schmes sont toujours identifia-
5
Voir sur ce point Cantineau (1950) ; Roman (1983 II : 897900 ; 2005, 33), qui voit
dans la premire voyelle la marque de la diathse (subjective ou objective) ; Fleisch
(1957 ; 1968, 246 ; 1979) ; Larcher (1996 ; 2003, 2628). Nr ad-Dn (2002, 185186)
signale des traits tardifs, dans lesquels la forme fuila est considre comme un quatrime principe (asl), i.e., un quatrime schme du verbe simple, et non une forme de
la conjugaison (comme dans , m. en 643/1245, arh al-Mulk : 30).
6
The antiquity of this threefold vocalic scheme in Arabic is confirmed by some of
the oldest manifestations in North-West Semitic, i.e. Amorite, Ugaritic, and the Tell
Amarna glosses. In the prefix-conjugation the variation in the second vowell is at least
partly paralleled : u or i corresponding to a, and a to i, while u generally remains (Moscati, d. 1964, 122).
faula, faila, faala
317
bles en fonction de la voyelle de la deuxime consonne radicale (en gras

dans la transcription), et ce, dans chacun des deux grands paradigmes
de la conjugaison. Dans ltat de langue arabe tel quil nous est parvenu,
ds les poques classiques, les trois schmes simples ci-dessus ne sont
plus directement identifiables, comme le rappelle le tableau suivant :
Schme simple 1
Schme simple 2
Md( suffix)
Mudri' (prfix)
faula
yaf ulu
faila
yaf ilu
Sous-catgorie formelle
yaf ilu
Schme simple 3
Sous-catgorie formelle 1
Sous-catgorie formelle 2
yaf alu
faala
yaf ulu
yaf alu
FIGURE 2 LES SCHMES DU VERBE SIMPLE ET LEURS SOUS-CATGORIES FORMELLES, TELS QUILS NOUS SONT PARVENUS
Cette situation continue de permettre lidentification au md (suffix) de trois schmes distingus au moyen de la voyelle qui affecte la
2e radicale, mais non au mudri (prfix). Lorsque lon traite daspects
formels, comme dans la conjugaison ou dans un dictionnaire, il est
donc ncessaire de caractriser le verbe par lalternance vocalique de la
voyelle mdiane aux deux paradigmes du suffix et du prfix, exemple :
daraba/yadribu frapper, dalternance vocalique a/i. (Jcrirai ici, selon
une convention reue : daraba i.)
2.2 Les zones de stabilit ou dinstabilit formelles de lalternance
vocalique
Comme le montre la figure 2, le schme simple 1 (alternance u/u) est
formellement stable. Dans les schmes faila et faala en revanche, on
trouve des sous-catgories formelles dans lesquelles la voyelle de la 2e
radicale varie. Ces variations sont souvent corrles la prsence, au
sein de la racine, dune cause de transformation (illa) :7 deuxime et
Sur le statut pistmologique de cette notion dans la tradition linguistique arabe
318
joseph dichy
troisime radicales identiques, consonne radicale /w/ ou /y/, ou encore

radicale // en premire, deuxime ou troisime position. Les relations
entre ces causes de transformation et les variations observes dans les
schmes du verbe simple ont t, sinon expliques, du moins largement
inventories dans les sciences linguistiques arabes mdivales (voir
rsum dans a-artn, 1910 : 1014 ; Qabbwa, 1998 : 8594).
2.2.1
La sous-catgorie formelle faila/yaf ilu du schme simple 2
La sous-catgorie formelle du schme simple 2 ( faila/yaf ilu) est corrle dans la plupart des cas une racine de premire radicale w ou y. Elle
ne concerne quun petit nombre de verbes. Trois dentre eux, repris de
Sbawayhi (Kitb IV : 3839) et rgulirement cits dans les traits linguistiques arabes mdivaux ou par les grammaires, relvent de racines
sans cause de transformation : h asiba i ou a croire, estimer ; baisa i ou
a tre misrable ; naima i ou a tre doux, fin. Ils peuvent tre considrs comme ngligeables, car ils supportent galement une ralisation a
de la voyelle mdiane du mudri prfix.8
2.2.2
La sous-catgorie formelle faala/yaf alu du schme simple 3
Une seule modification de la voyelle mdiane est suffisamment rgulire. Elle est largement signale dans les ouvrages arabes mdivaux
(cf. par exemple, al-Zajjj, Jumal, 396397). Il sagit de la sous-catgorie formelle 2 du schme simple 3 ci-dessus ( faala/yaf alu) : elle se
produit pour simplifier si la 2e ou la 3e radicale du verbe est lune
des six consonnes darrire (h urf al-h alq), i.e. appartient lensemble
{, h, h, , , x}. Cette rgle sapplique selon une gradation, de manire
hautement frquente pour les deux consonnes les plus antrieures (les
consonnes glottales et h), et nettement moins frquente pour les deux
mdivale, et notamment chez Ibn Jinn (m. en 393/1002), voir Guillaume 1984. La
liste et les combinaisons des causes de transformation (il peut y en voir plus dune dans
une racine donne) ont t prsents sous une forme modlise dans Dichy (1993) et
Ammar et Dichy (1999, 3134).
8
Voir par exemple Ibn Xlawayh, m. en 370/981, Laysa f kalm al-arab Point ne
se trouve dans le langage des Arabes . . . (sous-entendu hormis, lexception de), chap.
10 ; az-Zajjj, m. en 337/949, Jumal : 398 (plus cohrent que le prcdent). Sur le statut
de la classe h asiba/yah sibu, notamment chez Ibn Jinn, m. en 392/1002, cf. Guillaume
(1984, 427433).
faula, faila, faala
319
consonnes les moins antrieures ( et x) (cf. Ibn Ya, arh al-Mufassa l

VII : 153154, repris in Ammar et Dichy 1999 : 24).
2.2.3
La sous-catgorie formelle faala/yaf ulu du schme simple 3
La sous-catgorie formelle 1 du schme simple 3 ( faala/yaf ulu) ne

parat pas, quant elle, pourvoir tre lobjet de corrlations systmatiques entre la voyelle de la 2e radicale et des traits phontiques affectant
les consonnes du verbe. On trouve ainsi un nombre non ngligeable de
verbes dont le prfix (mudri) admet les deux voyelles mdianes i et
u, ainsi : azafa/yazifu et yazufu (an) se dtourner (de) ; qafala/yaqfilu
et yaqfulu revenir (de voyage), rentrer, etc. La prsence, dans de tels
exemples, des deux ralisations met en dfaut toute corrlation entre les
traits phontiques des consonnes radicales et la voyelle mdiane dans la
langue arabe telle quelle nous est parvenue, avec ses normes stabilises
par les travaux des lexicographes et lautorit des grammairiens.
3. Aspects smantiques, concepts de base

Considrons maintenant les valeurs smantiques associables aux schmes simples. Les deux notions qui ont t traditionnellement utilises
constituent un cadre conceptuel dont on verra quil ne peut suffire. Il
sagit du trait syntaxique de transitivit ( 3.1.1) et de la distinction entre
verbes daction et verbes dtat ( 3.1.2). Ces notions devront tre reprises dans un cadre conceptuel largi (sections 3.2 3.4). Les descriptions
proposes ci-dessous sappuieront sur une srie dexemples, choisis pour
illustrer la plupart des types de valeurs smantiques qui caractrisent
ces schmes.
3.1 Les deux principales notions traditionnellement utilises pour
dcrire le sens des schmes simples
3.1.1 Le trait syntaxique de transitivit/intransitivit
La premire distinction que lon rencontre dans les grammaires arabisantes comme dans les sciences mdivales arabes du langage est de nature
syntaxique. En fonction de la voyelle a, u ou i affectant la deuxime
consonne radicale, ces verbes sont dits :
320
joseph dichy
pour le schme faala, transitifs, bien que comportant un nombre

considrable dintransitifs,
transitifs et intransitifs pour le schme faila, et
toujours intransitifs pour les verbes en faula.9
Les traits syntaxiques de transitivit ou dintransitivit sont cruciaux
dans la thorie de la rection qui prvaut dans la tradition linguistique
arabe mdivale. Leur incidence sur le sens des verbes nest toutefois
pas directe : elle en est plutt une consquence quune cause. Une autre
question que la thorie de la rection laisse dans lombre est celle de la
transitivit indirecte, dans laquelle le complment du verbe est un syntagme prpositionnel. La transitivit, envisage sous ses deux formes,
directe et indirecte sera naturellement incluse dans les valeurs smantiques associes aux verbes des schmes simples, particulirement en ce
qui concerne le schme faala.
3.1.2
La partition traditionnelle entre verbes daction et verbes dtat
Les arabisants et les smitisants occidentaux ont gnralement fond

leurs analyses des verbes des schmes simples sur une partition gnrale en verbes daction, ou actifs, et verbes dtat ou statifs, les seconds
tant ncessairement intransitifs, et les premiers, transitifs ou intransitifs selon les cas. Dans son principe, le schme faala aurait ainsi correspondu des verbes daction (transitifs ou intransitifs), le schme faula
des verbes statifs toujours intransitifs exprimant un tat permanent
ou une qualit inhrente naturelle, et le schme faila des verbes gnralement intransitifs indiquant soit un tat ou une condition temporaire, soit une qualit purement accidentelle dune personne ou dune
chose (Wright 189698 I : 3638). Dans cette description, le schme
faila pose un problme, qui a t trait en recourant la notion de verbe
moyen dE. Benveniste (section 3.4 ci-dessous).
Si la notion de verbe dtat, une fois redfinie, rsiste lanalyse, celle
de verbe daction ne laisse pas, dans lusage qui en a t fait, dapparatre
comme indiffrencie. Que faire, par exemple de verbes comme zanna
u croire, saqata u tomber, halaka i prir, wajaba i devoir, amala u
englober ou ra a voir, qui relvent tous du schme faala, rput cor9
Cf. Sbawayhi (Kitb IV : 38), ainsi par exemple que az-Zamaxar (Mufassa l, 277
279). Lintransitivit des verbes dtat est rappele notamment par Wright (189698,
vol. I, 3638) ou Fleisch (1957 ; 1979). Pour lhbreu, voir par ex. Gesenius-KautzschCowley (1910, 4143) ou Joon (1923, 4041).
faula, faila, faala
321
respondre aux verbes daction ? Et que dire des paires akala u manger,
de schme faala vs. ariba a boire, de schme faila, ou ra a voir, vs.
samia a entendre ? La rponse consistant poser une catgorie intermdiaire ou moyenne (Marcel Cohen 1929 ; Joon 1930 ; Fleisch 1957),
qui concerne au premier chef le schme faila tout en incluant des verbes
en faala, ne rsout que trs partiellement le problme. Les processus de
glissement smantique entranant le passage dun verbe daction un
verbe dtat, sils ont donn lieu quelques fines observations,10 napportent pas non plus de solution de porte gnrale. Mais le principal problme, comme la section 4.3 lillustrera, est que lapparente transparence
de la notion de verbe daction aura masqu la complexit des donnes
et la forte dispersion smantique du schme faala. On verra dans la
suite de ce travail quil nest de fait pas ncessaire de recourir aux verbes
daction pour dcrire les verbes en faala, malgr lintrt que prsente
par ailleurs cette notion.
Passons maintenant la dfinition des proprits fondamentales
ncessaires lanalyse des schmes simples : lagentivit (section 3.2), les
notions de procs, dvnement, de processus et dtat (sect. 3.3), celle de
verbe moyen ou de diathse interne (sect. 3.4).
3.2 La proprit dagentivit
Lagentivit se dfinit comme la participation de lagent au procs dnot
par le verbe, le critre de cette participation tant le contrle (Lyons
1978/90, 3.4 ; Descls 1994) de lagent sur ce procs. Il y a en arabe trois
degrs dagentivit :11 lagentivit pleine dans laquelle lagent contrle
effectivement le procs ; partielle, dans laquelle il nexerce quun contrle imparfait, et la non agentivit, dans laquelle il ne jouit daucun contrle (lagent existe, mais il ne concide nullement avec le sujet grammatical). Ces trois degrs sopposent eux-mmes lagentivit neutralise,
qui correspond des procs dans lesquels il ny a aucun agent, ce qui se
10
Cf. par exemple pour lhbreu, Joon 1923, 41, pour larabe, Larcher 2003,
2728.
11
Cf. Roman (1990, 4243 ; 1999/2005, B-2.3.2). La dfinition ci-dessus de lagentivit est en partie diffrente. Cf. galement lanalyse de Fleisch (1968, 116117), qui pose
une bipartition du verbe arabe : 1 le verbe agent (le sujet tant considr en tant
quagent) ; le verbe de qualit (le sujet tant simplement le qualifi), et distingue, dans la
premire catgorie lagent pur et simple de lagent intress.
322
joseph dichy
prsente typiquement dans les verbes valeur dtat caractristique (

4.1.1 ci-dessous). Exemples (verbes de schme simple) :12
(a) verbes dagentivit neutralise (valeur smantique tre x) ou nonagentifs (valeur : devenir x) :
schme faula : karuma u, tre, devenir karm noble-et-gnreux ;
arufa u devenir, puis tre arf , i.e. noble, de premier rang social ;
saura u devenir, tre sar, petit ou : tre trs jeune ; latufa u tre,
devenir latf, subtil, fin, gracieux ;
schme faila : baliha a devenir ou tre ablah, faible desprit, stupide ; salia a devenir ou tre chauve (asla) ; karia a tre ou devenir
rid, craquel (notamment : peau) ;
schme faala : xadaba i devenir, tre vert (plante, herbage, lieu . . .) ;
xaffa i devenir, tre xaff, lger (antonyme de schme faula : taqula
u devenir, tre taql, lourd) ;
(b) verbes non-agentifs (le sujet grammatical nest pas lagent du procs
dont il est le sige) :
schme faila : dasiqa a se remplir ( ras bords), dborder (bassin . . .) ; marida a tomber malade ; qarima a avoir faim de viande
(par effet de privation) ;
schme faala : sakana u, se calmer (mer, vent, bruit . . .) ; waqaa a
tomber et galement : se trouver (ville, montagne, lieu . . .) ; halaka i
prir ;
(c) verbes partiellement agentifs :
schme faila : adiba a tre, se mettre en colre ; h azina a tre ou
devenir triste, prouver de la tristesse ; wamiqa i chrir ; aiqa a
tre ou tomber passionnment amoureux ; samia a entendre (sens
de loue) ;
schme faala : ra a voir (vue) ; amma u sentir (odorat) ;
(d) verbes pleinement agentifs :
schme faila : ariba a boire ; rakiba a chevaucher, monter (pour se
dplacer) ; h amida a louer (louange),
12
Certains des exemples donns dans ce travail existent aussi avec des formes correspondant dautres schmes du verbe simple, quelquefois avec le mme sens. Ces autres
formes ne sont pas reprises ici, malgr lintrt quet reprsent la discussion.
faula, faila, faala
323
schme faala : daraba i frapper ; jalasa i sasseoir ; da u revenir ;

dahaba a partir, aller ; tabaxa u (ou a) faire cuire ou bouillir, cuisiner ; talaa a monter ; akala u manger ; qla u dire.
La sous-catgorie (a) inclut des verbes des trois schmes et les sous-catgories (b) (d), des verbes des schmes faila et faala. Ces exemples, qui
anticipent par leur classement sur la suite de lexpos, montrent que la
proprit dagentivit, malgr limportance qui est la sienne dans la description smantique des schmes simples, ne peut, elle seule, constituer le principe qui en permettrait le classement.
3.2.1
Agent, patient, sujet grammatical
Il est essentiel, pour bien comprendre la notion dagentivit, de distinguer les termes dagent (et corollairement, de patient) du sujet grammatical du verbe. Les formes verbales de larabe incluent en effet toujours
leur pronom sujet, qui correspond aux morphmes de personne suffixs
au md et prfixs au mudri, et concide avec le sujet grammatical du
verbe. Ainsi : jat l-amra correspond mot--mot : elle est venue la
princesse (Ammar et Dichy 1999, 1214). Cette structure est occulte
par lemploi indiffrenci, dans la tradition linguistique arabe, de fil
pour lagent comme pour le sujet grammatical.
Lagent ( fil) est ce qui effectue le procs dcrit par le verbe. Dans les
verbes dcrivant une action, lagent est lauteur de celle-ci. Dans les verbes de perception ou de sentiment, par exemple, il est le sige de celui-ci.
Dans les verbes dcrivant un tat, le procs, comme on vient de le voir,
ne rfre pas un agent : le sujet grammatical inclus dans la forme verbale
est le sige de cet tat, quil sagisse dune personne comme dans raufa u
tre, devenir ou se montrer doux, compatissant ou dune chose, comme
dans daula u, devenir/tre troit ou petit (quantit). Dans les verbes
dagentivit partielle ou entire, lagent concide avec le sujet grammatical. Ce dernier, la voix dite passive, concide avec le patient (qui subit
le procs dcrit par le verbe), exemple : suiltu jai t questionn. Dans
les formes non-agentives, le sujet grammatical subit galement le procs ou en est le sige, exemples :
saqatat al l-lard elle est tombe par terre (quil sagisse dune personne ou dune chose) ;
darasa r-rasmu la trace (du campement) sest efface : lagent est ici le
vent du dsert. (Ce thme est classique dans la posie arabe ancienne ;
324
joseph dichy
cf. la fin du 6e vers de la Muallaqa grande ode du pote emblmatique de lant-islam Imru al-Qays : Wa-hal al rasmin drisin min
muawwal, Sur traces effaces, qui sen irait pleurant ?)
3.3 Les proprits dvnement, de processus et dtat ( partir de
Descls 1994)
Le terme gnral de procs (auquel correspond en anglais celui de
situation)13 traduit le fait que le sens du verbe est inscrit dans une dure,
cest--dire, dans un intervalle de temps. Cette dure peut tre perue de
trois manires, correspondant aux notions aspectuelles dtat, dvnement ou de processus. La dfinition de ces termes premiers est emprunte
directement J.-P. Descls.14 premire vue, ces dfinitions pourraient
apparatre comme dtournes : les trois termes aspectuels dvnement,
de processus ou dtat, qui dcrivent des phnomnes affectant des verbes en en contexte sont rutiliss, dans ce travail comme dans ceux dont
la publication suivra, pour dsigner dans la smantique des schmes de
larabe des primitives incluses dans le sens grammatical associ aux schmes simples et certains schmes augments. Mais y regarder de plus
prs, sagissant de valeurs smantiques primitives, il ny a rien dtonnant
ce quelles puissent galement tre observes dans les proprits lexicales de familles morphologiques, telles que les schmes simples et les
schmes augments. La mise en lumire de ces observations constitue
une premire contribution thorique du travail prsent ici la dfinition de ces valeurs primitives. Leur confrontation aux donnes de larabe
ma en outre conduit en proposer des extensions, particulirement en
ce qui concerne les sous-catgories de ltat ( 3.3.3).
13
Le terme de procs ne se traduit pas directement en anglais. Comrie (1985, 5),

Lyons (1978/1990, 3.4) ont recours au terme de situation pour recouvrir les tats dune
part, les vnements et les processus de lautre.
14
Mais voir aussi Comrie (1976, 1985), Lyons (1978/1990), ou, diffremment, la synthse de Maingueneau (1994). La thorie des invariants cognitifs du langage et des primitives smantico-cognitives de J.-P. Descls est sous-jacente aux analyses prsentes
ici, notamment : lide que les primitives sont des oprateurs abstraits dtermins par
des proprits formelles, lhypothse selon laquelle elles sont ancres sur la perception
et laction, ainsi quun nombre important de points qui dcoulent de ces conceptions
(Descls 1990 ; 1994 ; Descls et Guentchva paratre). Le lecteur familier de ces travaux notera enfin que le terme de schme est employ ici la manire des travaux
arabisants et smitisants. On distinguera donc cet usage de celui des schmes smantico-cognitifs (Descls, 1990 ; Abraham, 1995 ; Descls et al., 1998).
faula, faila, faala

3.3.1
325
vnement
Laspect vnement [. . .] est lexpression dune occurrence qui apparat

sur un arrire-fond stable [. . . et] tablit une coupure entre lavant vnementiel (pas encore) et laprs vnementiel (ne plus). La dure de
lvnement est donc borne gauche par un avant et droite par un
aprs. Cet aprs peut concider avec un tat engendr par lvnement,
ou tat rsultatif (Descls 1994, 7375).15 Les exemples ci-dessous (verbe
concern en gras) illustrent quelques proprits fondamentales de laspect vnement :
m kidtu hd s-sabh a arifu finjna l-qahwa al maktab h att
waradat ira tilifuniyya . . . (Yawm., 123), peine, ce matin-l,
avais-je sirot le caf [pos] sur mon bureau, quun tlgramme tlphon survint . . . (mot--mot, lorsque apparut). La premire proposition de cette phrase (m kidtu . . .arifu peine . . . avais-je sirot)
constitue larrire-plan sur lequel se dtache lvnement h att waradat lorsque apparut, vnement qui dtermine au sein de la situation
dcrite, un avant et un aprs.
wasala dayfun mundu sa notre invit est arriv depuis une heure :
de lvnement est arriv rsulte un tat, dont la dure englobe le
moment de lnonciation. Le locuteur rapporte lvnement de larrive de linvit, et exprime par les mmes mots ltat dans lequel son
interlocuteur et lui se trouvent, et qui correspond laprs du procs :
leur invit est l depuis une heure. On dira, en simplifiant, que lvnement engendre ltat qui rsulte de lui.
ariqa r-rajulu lhomme sest noy (ou se noya) : lvnement dcrit
par le procs engendre ltat rsultatif noy, exprim en arabe par
ladjectif arq, associ ce verbe.
3.3.2
Processus
Laspect processus exprime un changement saisi dans son volution

interne. Tout processus exprime ncessairement un changement initial
(. . .) qui indique le dbut du processus. Le processus peut tre dcrit
15
Je fais dans ce travail un usage gnrique (ou hypronymique) du terme dtat rsultatif. Ce dernier inclut ici ltat rsultant, qui est par ailleurs trs utile pour lanalyse des
valeurs aspectuo-temporelles en contexte, par exemple pour la description de ltat-prsent, concomitant lacte dnonciation (Descls, 1994, 73). Ce choix, limit au prsent
travail, est d au fait que les valeurs smantiques associes aux familles morphologiques
constitues par les schmes ne sont pas sensibles cette distinction contextuelle.
326
joseph dichy
comme une succession dtats. Il peut, selon le cas, sorienter ou non vers
un terme explicite, qui correspond alors au dernier tat du processus ou
tat rsultatif (Descls 1994, 7580). Exemples :
Dans la phrase ci-dessus, le syntagme m kidtu arifu finjna l-qahwa
peine avais-je sirot le caf exprime un processus dont le terme est
spcifi : un moment donn, qui correspond ltat final du processus, le breuvage est entirement bu. La phrase dit mot--mot : . . . sirot
la tasse de caf ; la valeur smantique associe cette expression dans
le lexique de larabe implique que le contenu de la tasse a t bu en
entier.16
Dans arabati -amsu le soleil sest couch ou hal turdu an
tujanninan ? tu veux me rendre fou (folle) ? cest galement dans
la valeur lexicale du verbe que le terme du processus (le coucher du
soleil, la folie) est inscrit.
Le sens de amiya a perdre la vue comporte un terme, qui correspond dans le lexique ladjectif am aveugle. Ce verbe dcrit, selon
le contexte, soit un vnement, soit un processus.
3.3.3 tat, types dtats et catgories drives ou corrles
Laspect tat exprime la stabilit de la situation rfrentielle reprsente,
cest--dire quil rfre une absence de mouvement et de changement.
Ltat dcrit un procs qui nest born ni gauche (en son commencement) ni droite (le terme nest pas spcifi), et ne fait rfrence aucun
agent (Descls 1994, 7173).17 Il y a plusieurs types dtats. Deux dentre
eux, ltat caractristique ( 3.3.3.a) et ltat rsultatif ou tat acquis
( 3.3.3.b) intressent directement les schmes du verbe simple. Le second,
qui est produit par un changement dtat, correspond un glissement
smantique troitement corrl lusage des verbes dtat en discours.
Lanalyse ci-dessous dbouche sur ladjonction, la valeur smantique
de ltat caractristique, de celle de lacquisition dtat. Cette dernire
16
Ainsi, le dictionnaire glose lexpression raifa l-ina, mot--mot siroter un rcipient comme en aspirer le contenu jusquau bout (al-Mujam al-Wast : racine /r--f/) ;
de mme, ariba l-ina boire un rcipient correspond ariba kulla m fhi en boire
tout le contenu (Ilys et Nsf 1995, 164).
17
Comparer avec la dfinition de Maingueneau (1994, 64) : Les tats . . . nont ni
dbut, ni fin, ni milieu, ils ne supposent ni agent ni changement (ex. Luc est paresseux).
la diffrence de Comrie (1976, 1985), Lyons (1978/1990), Descls (1994) et du prsent
travail, cet auteur noppose pas tat, vnement et processus.
faula, faila, faala
327
notion relve de celles, plus gnrales, dtat rsultatif et de changement

dtat. Lacquisition dtat constitue en effet la principale valeur smantique corrle ltat caractristique ( 3.3.4).
3.3.3.a La notion dtat caractristique
Les donnes de larabe mont conduit proposer une sous-catgorie de
la notion gnrale dtat, celle dtat caractristique (Dichy 2002/2003).
Ce dernier est exprim par un verbe associ dans le lexique un adjectif,
traditionnellement appel adjectif assimil ou analogue [au participe]
(sifa muabbaha ci-aprs 4.1.2). Ltat caractristique par opposition ltat non caractristique , dcrit une proprit ou un attribut
de lobjet, pos, selon le point de vue, comme dominant, intrinsque,
inhrent ou inscrit dans la nature de celui-ci. Cet tat peut, comme on
le verra au paragraphe suivant, tre acquis lissue dun vnement ou
dun processus.
Le terme dtat caractristique, dont le contenu sera confront aux donnes du schme faula la section 4.1, est prfrable lemploi habituel
des grammaires arabisantes qui oppose, dans les verbes du schme simple, qualit permanente ou durable qualit temporaire (par exemple,
Wright 189698, I : 3638). La dure relative dun tat tout comme
celle dun processus ou dun vnement correspond en effet lune
des valeurs aspectuo-temporelles attribues une forme verbale donne
en fonction du contexte, et ne doit pas tre confondue avec une valeur
smantique morpho-lexicale, associe comme telle un schme donn.
Ainsi, h azina a tre ou devenir triste (de schme faila, adjectif correspondant : h azn) est traditionnellement cit comme exemple de
qualit temporaire ou transitoire.18 Il rfre toutefois une qualit
durable dans les noncs suivants : fa-h azina il xiri umrihi Et il
fut triste jusqu la fin de ses jours ; inna lladna man . . . l xawfa
alayhim wa-l hum yah zann Ceux qui ont eu foi . . . aucune crainte
18
Ce verbe est cit par Caspari-Uricoecha pour illustrer un tat transitoire et passager ou bien une proprit ou une qualit qui naffecte lobjet du verbe que pendant peu
de temps (1881, 32) ; par Wright, comme illustrant a temporary state or condition,
or a merely accidental quality in persons or things (189698 I : 30) ; par Brockelmann
(1948, 35) comme exemple de verbes exprimant toujours une qualit ou une situation
accidentelle ou temporaire (Die Form faila steht durchweg fr zufllige, vorbergehende . . . Eigenschaften und Zustnde) ; ou Boormans (1967, 10), comme dcrivant un
tat passager.
328
joseph dichy
[ avoir] pour eux, il ne connatront [ternellement] aucune tristesse

mot--mot : et aucunement ne sattristeront (Coran 2 : 62). H azn sera
donc mieux dcrit comme un adjectif dtat non caractristique.
Inversement, le verbe xalua u mriter le nom de dprav, tre un
dprav (adjectif correspondant : xal), qui relve du schme faula,
rput renvoyer des verbes de qualit permanente ou durable
est glos dans le dictionnaire par lexpression taraka l-h ay wa-rakiba hawhu, abandonner [toute] honte et se livrer ses passions
(al-Mujam al-Wast : racine /x-l-/). La qualit de xal dprav
(notoire) nest donc pas permanente. Xal peut en revanche tre
dcrit indpendamment toute rfrence une dure comme un
adjectif dtat caractristique.
Exemples de ces deux sous-catgories de la notion dtat :
verbes associs un adjectif dtat caractristique :
tla u tre tawl long ; hayifa a tre ahyaf , avoir la taille lance ;
awira a tre awar, borgne ; baxula u tre baxl, avaricieux (au
sens de tre avare par nature)19
verbes associs un adjectif dtat non caractristique :
farih a a tre farih , joyeux (lun des contraires de h azina ci-dessus) ;
yaisa a tre yis dsespr ; atia a tre atn ayant soif , avoir
soif ; marida a tre mard malade. . .
3.3.3.b Verbes dtat vs verbes de changement dtat produisant un
tat rsultatif
Pour H. Fleisch, cependant :
Le verbe de qualit nest pas un statif. Il signifie acqurir une qualit, autrement dit : devenir tel (daprs la qualit en question) : karuma = devenir
19
Les proprits de lagentivit ne relvent pas dinvariants linguistiques, mais de la
reprsentation du procs vhicule par le vocabulaire de chaque culture. Ainsi, dans la
culture arabe mdivale, on peut, soit (a) se montrer avare envers quelquun dans une
situation donne, ou (b) tre avare par nature ( f t-tab) : cf. par exemple la prface du
Kitb al-Buxal Livre des avares dal-Jhiz, m. en 255/868 ; semblablement lavarice est,
lpoque de Molire considre comme pouvant tre inscrite dans le caractre de celui
qui, comme Harpagon, est un avaricieux. Le premier cas (a) est exprim au moyen du
verbe pleinement agentif baxila a al se montrer avare envers qqun, et le second (b),
par le verbe dagentivit neutralise baxula u tre un avaricieux (lavare est le sige du
vice inscrit dans sa nature).
faula, faila, faala
329
karm (gnreux) ; ou bien, comme consquence de lacquisition : avoir

une qualit, soit : tre tel, cest un rsultatif, alors karuma = tre karm
(gnreux) (1968, 117).
Fleisch suit vraisemblablement ici la dfinition que donne al-Mubarrid

de la valeur smantique des verbes en faula :
innam huwa li-l-h l allat yantaqil ilayh l-fil . . . nahwu karuma wazarufa wa-arufa elle (cette classe de verbes) dsigne uniquement ltat
dans lequel entre le sujet [du verbe] (mot--mot : ltat vers lequel se
dplace le sujet), exemples : karuma, zarufa et arufa (Muqtadab I : 209
traduction des exemples : devenir et donc tre noble-et-gnreux, devenir, puis tre raffin et devenir, puis tre prminent, illustre, noble).
Ces deux descriptions font de ltat caractristique une valeur seconde

par rapport la valeur devenir qqch. Or Sbawayhi dcrit les verbes de
schme faula comme dnotant les caractres qui peuvent tre dans les
choses (al-xisl allat taknu f l-ay Kitb IV : 28). Ces termes ont
t par la suite largement repris dans les sciences linguistiques arabes
mdivales (cf. par ex. az-Zamaxar, Mufassa l : 279). Dans cette conception, la valeur fondamentale des verbes de qualit est stative, alors que
chez al-Mubarrid et Fleisch, elle correspond au changement dtat. Mais
faut-il vraiment opposer ces conceptions ? Est-il possible de dcider de
la valeur smantique qui serait premire par rapport lautre ?
Considrons les exemples suivants :
H asuna u tre h asan, beau, qaduma u tre qadm, ancien ou hayifa
a tre ahyaf , qui a la taille lance dsignent la base un tat caractristique.
amasa i ou u devenir/tre ensoleill (jour) ; asaqa i devenir/tre
noir, obscur (nuit) rfrent fondamentalement un changement
dtat. Il sagit de verbes signifiant devenir qqch, dcrivant un processus qui aboutit un tat rsultatif non caractristique.
adiba a se mettre/tre en colre ou ja u avoir faim/tre affam (=
ayant faim) correspondent fondamentalement des verbes de changement dtat, dans lesquels le nouvel tat (non caractristique) qui
affecte le sujet grammatical rsulte du procs. Les adjectifs correspondants, adib, adbn en colre et ji, jawn ayant faim sont des
tats rsultatifs non caractristiques.
Le verbe awira a devenir/tre borgne dsigne selon le contexte
soit un changement dtat (devenir qqch) dbouchant sur ltat caractristique tre awar, borgne, soit cet tat caractristique lui-mme
(qui peut tre rsultatif ou non).
330
joseph dichy
Un verbe dtat caractristique correspondant fondamentalement

un tat, comme h asuna u tre beau ou hayifa a avoir la taille lance
ci-dessus peut aussi tre amen par le contexte dsigner le rsultat
dun vnement ou dun processus, exemple : hayifa l-na l-fat le
jeune homme est maintenant ahyaf , a maintenant une taille lance
(lnonc prsuppose : maintenant quil a grandi).
Ces exemples montrent qu la question pose (ce qui, de la valeur
smantique stative ou du changement dtat serait premier par rapport
lautre), il ny a pas de rponse gnrale, qui demeurerait valable dans
tous les cas. La rponse doit au contraire tre propose pour chaque
verbe, en fonction de son sens lexical (et des connaissances encyclopdiques associes celui-ci). Cest ainsi que :
amasa i ou u devenir/tre ensoleill ou ja u avoir faim dcrivent
dabord un changement dtat (devenir qqch), et en second lieu, ltat
non caractristique correspondant.
Le sens des verbes dtat caractristique qaduma u tre ancien ou
jayida a avoir le cou long (signe de beaut) est peu compatible avec
un contexte dans lequel ces verbes prendraient la valeur devenir tel
(bien quun tel contexte demeure, comme on vient de lindiquer, possible).
Awira a, qui dcrit un tat caractristique, est compatible tant avec la
valeur stative tre borgne quavec celle du changement dtat (devenir tel), sans quil soit possible de dcider de la valeur qui driverait
de lautre.
3.3.4. Lvnement de lacquisition dtat, cas particulier du changement
dtat et de ltat rsultatif
Mais il convient, pour comprendre la prsence dans ces verbes des deux
valeurs stative et dacquisition dtat, dinterroger plus avant la notion
dtat rsultatif. Ce dernier est engendr par un procs (vnement ou
processus) correspondant un changement dtat (Descls 1994, 71 et
suiv. ; Descls et Guentcheva paratre, chap. II). Dans les donnes suivantes, cet tat met en jeu des notions dans lesquelles les proprits de
ltat, de lvnement et du processus se trouvent combines :
(a) tat rsultatif (caractristique ou non) engendr par un vnement :
Le verbe amura u galement ralis amira a est glos dans les
dictionnaires comme devenir mir, notamment par succession (al
faula, faila, faala
331
Mujam al-Wast, al-Munjid, Hans Wehr-Cowan 1979). Ce verbe

dcrit, dans cette acception, lvnement dun changement dtat, et
le nom-adjectif associ amr, qui dsigne ltat caractristique mir,
prince, celui qui gouverne un mirat, correspond un tat rsultatif.
Mais il peut galement, comme nous le verrons au paragraphe suivant, dsigner ltat caractristique correspondant, indpendamment
dun vnement ou dun processus qui laurait prcd.
H azina a tre ou devenir triste, sattrister dsigne ltat non caractristique h azn triste ou lentre dans cet tat. Lvnement que constitue cette entre peut tre caus par un autre vnement, par exemple
larrive dune mauvaise nouvelle : lamm samia xabara axhi, h azina
adda l-h uzni lorsquil entendit les nouvelles de son frre, il tomba
dans une profonde tristesse (mot--mot : il sattrista dune intense
tristesse). Ladjectif h azn, dcrit ci-dessus ( 3.3.3.a) comme un
adjectif dtat non caractristique peut donc en outre correspondre
un tat rsultatif, h azina relevant, dans ce dernier cas, des verbes de
changement dtat.
(b) tat rsultatif (caractristique ou non) correspondant au dernier tat
dun processus :
Comme rappel ci-dessus, un processus correspond une succession
dtats. Ltat rsultatif peut donc concider avec le dernier tat du processus qui le produit. Exemples :
azuma anuhu ayan fa-ayan sa situation sociale est devenue peu
peu (ou devint peu peu) de grande importance (azm) ;
sawida a devenir ou tre aswad noir ; saura u devenir ou tre
(socialement ou quantitativement) sar petit ;
xadira z-zaru la rcolte a verdi ou verdit (Munjid, racine /x-d-r/ ;
Ilys et Nsf 1995, 113).
Ces donnes montrent que la notion dtat rsultatif est loin daller de
soi. Elles appellent deux remarques gnrales :
3.3.4.a Valeur aspectuo-temporelle du suffix (md) et vnement
engendrant un tat exprim ou non par un adjectif assimil
On connat en franais la valeur daccompli du prsent que peut prendre le pass compos et qui indique dans une nonciation qui inclut
la fois le pass compos et le prsent de lnonciateur le rsultat
prsent dun procs accompli antrieurement (Maingueneau 1994,
332
joseph dichy
6768). Descls (1994, 73) dcrit ce rsultat comme un tat-prsent

concomitant lacte dnonciation. Comparons par exemple :
(1) Alors tu te dpches ? Je finis !
(2) Alors tu te dpches ? Jai fini ! (Leeman-Bouix 1994, 6162).
En (1), linterlocuteur est en train dachever quelque chose, le verbe
conjugu au prsent a donc valeur dinaccompli ; en (2), le pass compos exprime un processus la fois accompli et achev (voir, pour cette
distinction, Descls 1994, 7680).
En arabe, la conjugaison au md (suffix) comporte galement la proprit dengendrer un tat rsultatif, exemples :
saimtu taklifa l-h ayti . . . je suis las des fardeaux [dont nous charge]
la vie (Muallaqa grande ode du pote prislamique Zuhayr b. Ab
Sulm, 46e vers) : la forme conjugue au suffix (mot--mot : jai t
ou je fus las . . .) engendre ltat rsultatif tre las.
La traduction franaise : je suis las est contrainte dexprimer cet tatprsent (au sens de Descls 1994, 73), mais elle ne peut rendre par une
forme verbale lvnement qui lengendre. En arabe le md (suffix) de
saimtu englobe ces deux interprtations (cf. lexemple de wasala au
3.3.1) ;
atitu jai soif : la forme au suffix (mot-a-mot : jeus ou jai eu soif )
engendre ltat avoir soif (en traduction : jai soif mme commentaire que ci-dessus), qui pourrait tre glos comme : la soif sest maintenant installe en moi (ou sinstalla dsormais en moi).
Il est essentiel, pour comprendre le sens des schmes associs la notion
dtat, dobserver que ces exemples illustrent la prsence en arabe de deux
catgories diffrentes de verbes, selon que ltat rsultatif est exprim ou
non par un adjectif associ au verbe par la drivation morpho-lexicale.
Ltat rsultatif engendr par saimtu nest pas lexicalis : le cas est donc
comparable mutatis mutandis lusage du pass-simple franais.
Pour atitu, ltat qui rsulte du procs correspond une entre lexicale
(i.e. un adjectif assimil). Ainsi :
faula, faila, faala
333
En reprenant les exemples (a) du 3.3.4, qui dcrivent des vnements :

Lnonc amura -iru (le pote est devenu/devint le prince
[rgnant] ou le pote rgne/rgna) peut tre glos comme : la position de prince est maintenant (ou fut dsormais) occupe par le pote.
Le sens du verbe amura u est donc bien devenir, puis tre mir, et
non uniquement devenir mir, comme dans les dictionnaires cits
plus haut.
lamm samia xabara axhi, h azina adda l-h uzni lorsquil entendit
les nouvelles de son frre, il tomba dans une profonde tristesse correspond : lorsquil entendit . . . ltat de tristesse sinstalla en lui.
Les exemples donns en (b), qui portent sur des processus, mritent un
mot dexplication : lachvement du processus parvenu son dernier tat
constitue lui-mme un vnement (Descls, 1994, 72, 7780) :
Dans xadira z-zaru la rcolte a verdi ou verdit cet vnement correspond au moment o la rcolte parvient au degr de mrissement
attendu (ltat de verdeur).
Dans azuma anuhu ayan fa-ayan sa situation sociale est devenue/devint peu peu de grande importance, cet vnement est celui
de lachvement du processus devenir peu peu, qui atteint alors son
dernier tat (celui de grande importance).
Dans tous ces cas, (1) il y a un avant et un aprs de lvnement du
changement dtat et de linstallation du nouvel tat, et (2) cet tat est
dsign dans le lexique par un adjectif assimil (sifa muabbaha). Lvnement pourrait de ce fait tre glos comme atteindre ou acqurir ltat
x, dans lequel x correspond un adjectif ou un nom-adjectif : atn
qui a soif ; amr mir ; h azn triste ; axdar vert ; azm de grande
importance.
3.3.4.b Les relations asymtriques tre x o devenir x et devenir x
tre x
La prsence dans le lexique dun adjectif assimil permet de mieux saisir le lien entre tre x (valeur stative) et devenir x (valeur dacquisition
dtat) qui stablit entre les verbes associs un adjectif dtat caractristique ou non caractristique dune part, et les verbes de changement
dtat de lautre. Ainsi :
334
joseph dichy
qadumat al-yawma dawlatuhum leur dynastie est devenue aujourdhui

ancienne peut tre glos comme : leur dynastie est devenue tant
ancienne (ou devint tant ancienne). Il sagit dun exemple de verbe
dtat caractristique (qaduma u tre ancien) dans lequel on observe
un glissement smantique en contexte de la valeur stative (tre qadm
ancien) celle dacquisition dtat (tre devenu tant qadm). Cest la
rencontre au sein de lnonc du sens lexical tre ancien et de la valeur
aspectuo-temporelle dachvement associe au md (suffix) ellemme renforce ici par le marqueur temporel al-yawma aujourdhui,
qui permet le glissement smantique tre x o devenir x observ.
adiba r-rajulu lhomme est (ou fut) en colre (adib) correspond
: ltat tre en colre sest accompli en lui. Cet exemple est celui dun
verbe dacquisition dtat non caractristique, dans lequel on rencontre un passage de la valeur de changement et dacquisition dtat
(devenir en colre) la valeur dtat rsultatif acquis (tre en colre),
la seconde tant engendre par la premire. Il peut tre glos comme :
tre devenu tant adib (en colre). Cest galement la rencontre en
contexte du sens lexical tre en colre et de la valeur dachvement
que prend le md (suffix) qui permet lengendrement de ltat acquis
tre x par la valeur de changement et dacquisition dtat devenir x.
Il y a donc deux cas gnraux de modification smantique corrls aux
verbes dtat. Les verbes dont la valeur de base est celle de ltat caractristique connaissent un glissement smantique tre x o devenir x
(glissement de ltat au changement et lacquisition dtat). Dans les
verbes exprimant fondamentalement un changement correspondant
lacquisition dun tat (caractristique ou non), ltat acquis est engendr
par lvnement qui concide avec lachvement du processus dacquisition, ce qui peut tre reprsent schmatiquement comme : devenir x
tre x (dans lequel le symbole se lit engendre).
3.3.5 Les deux valeurs smantiques associes la notion dtat : ltat
caractristique et lacquisition dtat
Lanalyse ci-dessus conduit donc distinguer, pour les verbes associs
un adjectif dtat caractristique, dune part ltat caractristique (tre x)
et de lautre lacquisition dtat (devenir x), qui correspond un changement dtat. Il leur est associ, respectivement, des adjectifs dtat caractristique et des adjectifs dtat acquis.
faula, faila, faala
335
Lexamen des donnes relatives aux verbes associs un adjectif dtat

non caractristique permet en outre dobserver que ces verbes relvent
tous du changement dtat : ils appartiennent de ce fait la catgorie des
verbes dacquisition dtat. A condition de tenir compte des glissements
smantiques dcrits plus haut (tre x o devenir x et devenir x
tre x), lanalyse de Fleisch et la conception dal-Mubarrid cites plus
haut ( 3.3.3.b), qui ne sapplique que partiellement aux verbes associs
un adjectif dtat caractristique, sapplique en revanche pleinement
aux verbes associs un adjectif dtat non caractristique, que la tradition arabisante appelait tat temporaire. On en a examin ci-dessus
plusieurs exemples.
3.4 Benveniste revisit : la notion de verbe moyen (ou diathse interne) ;
le mdio-passif
Lautre grande distinction sur la dfinition de laquelle il convient de
revenir est celle de verbe moyen. On appelle verbe moyen selon la dfinition dE. Benveniste, un verbe qui dcrit un procs dont le sujet est
le sige :20
Ici, le sujet est le lieu du procs, mme si ce procs, comme cest le cas
pour lat. fuor [jouir, avoir profit de] ou sanskr. manyate [prouver une
agitation mentale], demande un objet ; le sujet est centre en mme temps
quacteur du procs ; il accomplit quelque chose qui saccomplit en lui, natre, dormir, gsir, imaginer, crotre, etc. Il est bien intrieur au procs . . .
(1950/1966, 172).
3.4.1
Ambigut de cette dfinition ; les verbes mdio-passifs
La formulation ci-dessus pose un double problme :

(a) Dans les verbes dtat aussi on pourrait dire que le sujet est le lieu du
procs, mme si lon voit bien quil ne peut en tre lacteur. Les verbes
dtat, dans lesquels lagentivit est neutralise, nont pas de diathse, et
ne constituent en aucun cas une sous-catgorie du verbe moyen. Cette
20
A propos de la notion de verbe moyen en arabe classique, outre M. Cohen (1929),
Joon (1930), Cantineau (1950) et Fleisch (1957 ; 1968, 116117), comparer la description ci-dessus avec P. Larcher (1995, 2956 ; 2003, 2226) ou, pour larabe dialectal gyptien, de C. Audebert (2002).
336
joseph dichy
restriction, qui est implicite dans la dfinition ci-dessus, se dduit galement de cette remarque :
Il ne sagit donc nullement de faire concider la diffrence de lactif au
moyen avec celle des verbes daction et des verbes dtat (Benveniste
1950/1966, 172).
(b) Si le sujet est acteur du procs dont il est le sige, sil accomplit
quelque chose, cela suppose quil jouit dun certain degr dagentivit.
La proprit dagentivit est incluse de mme dans la notion de diathse
laquelle Benveniste a recours, notamment lorsquil propose en conclusion de larticle cit, de substituer diathse interne (par opposition la
diathse externe) au terme de verbe moyen (par opposition au verbe
actif ou externe). Or le passage cit plus haut donne comme exemples
de ce type de verbes natre, gsir et crotre, qui sont non agentifs.
Pour traiter cette difficult, il convient demprunter au mme article
(mais non sur ce point lanalyse) dE. Benveniste la notion de verbe
mdio-passif, pour dcrire les verbes non agentifs hors verbes dtat
bien entendu comme nam u crotre, grandir, waqaa a f se trouver,
tre situ ou dans, mad passer (temps), mta u mourir, etc.21
3.4.2 Diathse interne vs diathse externe et proprits de transitivit/
intransitivit
Le verbe moyen soppose au verbe actif ou externe :
De cette confrontation [entre deux types de verbes] se dgage assez clairement le principe dune distinction proprement linguistique, portant sur
la relation entre le sujet et le procs. Dans lactif, les verbes dnotent un
procs qui saccomplit partir du sujet et hors de lui.22 Dans le moyen, qui
est la diathse dfinir par opposition, le verbe indique un procs dont le
sujet est le sige ; le sujet est intrieur au procs (Benveniste, 1950/1966,
172). Lactif est une production dacte, rvlant plus clairement encore la
position extrieure du sujet relativement au procs ; et le moyen servira
dfinir le sujet comme intrieur au procs (op. cit. 173).
21
Je fais mienne ici une suggestion de Jean-Pierre Descls, communication orale.

Il est intressant de noter titre de comparaison que Sbawayhi considrait que
la transitivit fait entrer un verbe dans le chapitre des actions visibles ou audibles ( f
bb al-aml allat tur wa-tusma Kitb IV : 6). Cette description est reprise dans le
commentaire dal-Srf (en note de la mme page) et dans Ibn Ya, arh al-Mufassa l
VII : 157.
22
faula, faila, faala
337
Le verbe actif ou externe est par dfinition un verbe transitif :

Ainsi se constituent partir du moyen des actifs [. . .] qui se caractrisent
toujours par ceci que le sujet, pos hors du procs, le commande dsormais comme acteur, et que le procs, au lieu davoir le sujet pour sige, doit
prendre un objet pour fin (ibid.).
Rappelons que la transitivit doit sentendre, en ce qui concerne les donnes traites ici, comme directe ou indirecte (i.e. complment prpositionnel). Exemples de verbes de diathse externe :
h akama ayan matriser qqch. ; aml al fulnin ayan dicter
qqch. qqun ; dafaa a pousser, repousser ou payer ; kataba u
crire ; qla u dire . . .
Les verbes moyens ou de diathse interne, dans la dfinition de Benveniste, peuvent tre aussi bien intransitifs que transitifs directs ou indirects. Exemples de verbes du schme simple :
verbes moyens intransitifs :
baqiya a, rester, demeurer lorsque le sujet est humain ; farih a a se
rjouir, prouver de la joie ;
verbes moyens transitifs directs :
malla (dalternance vocalique i/a) tre ennuy par, prouver de lennui [ cause] de ; xa i craindre ;
verbes moyens transitifs indirects :
aara u bi-, sentir, ressentir ; raiba a f dsirer qqch, et son contraire
raiba a an dsirer viter ou spargner qqch.
4. Valeurs smantiques de base des schmes simples

Le cadre conceptuel tant pos, considrons maintenant les valeurs
smantiques associes aux trois schmes faula, faala et faila.
4.1
4.1.1
Les deux valeurs smantiques du schme faula

Les faula valeur dtat caractristique
Les verbes de ce schme sont traditionnellement dcrits comme exprimant un tat ou une qualit durable ou constante, par opposition aux
338
joseph dichy
verbes en faila, qui dnoteraient un tat temporaire ou instable (cf.

par exemple az-Zamaxar, Mufassa l : 230). A cette rfrence la dure,
jai propos de substituer la notion dtat caractristique, selon la dfinition ci-dessus ( 3.3.3.a). Ces verbes dcrivent effet une qualit ou un
tat pos comme constitutif ou caractristique du sujet grammatical.
Wright dsignait ce dernier comme correspondant une qualit inhrente naturelle (a naturally inherent quality 189698, I : 3638),
et Sbawayhi, dj cit, comme les caractres susceptibles daffecter les
choses, mot--mot qui peuvent tre dans les choses (al-xisl allat
taknu f l-ay Kitb IV : 28).
Dans les faula valeur dtat caractristique, le procs tat est, comme
on la vu, dagentivit neutralise. Ces verbes sont, comme il est attendu,
intransitifs.23 Exemples (hors glissement smantique de type tre x o
devenir x) :
qabuh a u tre qabh laid (valeur physique ou morale) ; h asuba u
tre h asb noble, de haute naissance ; h aluma u tre h alm clment ;
sabuh a u tre sabh brillant, radieux (visage) ou beau, joli (garon,
personne) ; katufa u tre katf pais, dense.
Faula est le schme simple le plus stable dun point de vue formel. Nous
avons vu (figures 1 et 2) que la voyelle de sa deuxime consonne radicale
demeurait inchange au paradigme du prfix (mudri). Ce schme est
aussi le plus stable smantiquement. Sa valeur de base, ltat caractristique,
est toutefois lobjet dun glissement de sens qui lui est troitement corrl, le changement et lacquisition dtat ( 3.3.3.b et 3.3.4 ci-dessus).
4.1.2
Les faula valeur dacquisition dtat caractristique
Ce glissement smantique, trs frquent, est illustr par lexemple :

ayra anna hdihi l-jama . . . d. a'ufa h udruh l-alaniyy . . . mundu
an . . . Mais la prsence publique de ce rassemblement . . . sest affaiblie (mot--mot, est faible) . . . depuis que . . . (Majallat an-Nahr, 2
juillet 2006, 13, col. 1).
23
Cf. Sbawayhi : Il nexiste pas dans le langage des Arabes de verbe transitif en
faula (laysa f l-kalm faultuhu mutaaddiyan Kitb IV : 38).
faula, faila, faala
339
Des exemples similaires ont t vus aux 3.3.4. a et b. Le contexte indique clairement un dbut du procs (depuis que . . .) : le verbe daufa u
(adjectif dtat caractristique : daf faible) ne rfre donc pas un tat.
La forme conjugue au md exprime ici un vnement, qui concide
avec lachvement du processus dcrit par le verbe (laffaiblissement) :
ce processus a atteint son dernier tat, ce qui dtermine un avant et
un aprs. Cet aprs correspond un tat rsultatif que lon peut dcrire
comme un tat acquis : cette prsence publique est devenue faible ou
est dsormais faible.
On observe, travers les usages illustrs par cet exemple et ceux qui
ont t analyss plus haut, que la valeur smantique tre x incluse dans
celle de ltat caractristique se trouve trs souvent ralise en contexte
comme un changement dtat (devenir x). Ce changement a, son tour,
la proprit dengendrer un tat rsultatif. Un tel phnomne nest nullement propre larabe. On peut comparer, au moins partiellement, avec
les exemples suivants, qui sont emprunts au franais :
Ah ! Maintenant, tu es beau ! (dit une maman son fils, qui revenait en
tranant les pieds de chez le coiffeur).
Lucie est dsormais notre anctre.
Dans les deux cas, les prdicats tre beau et tre notre anctre paraissent
incompatibles avec un contexte qui leur assignerait un commencement
ou un terme. Or les marqueurs temporels maintenant et dsormais
prsents dans ces exemples assignent aux procs un dbut, ce qui a pour
effet doprer une translation des prdicats verbaux ci-dessus de la catgorie de ltat vers celle du changement et de lacquisition dtat. Cest
comme si lon disait : tu es devenu beau (maintenant que tu as les cheveux coups) et Lucie est devenue lanctre de lhumanit (depuis sa
dcouverte et les recherches qui ont suivi).
Le dictionnaire de Hans Wehr-Cowan (1979) traite le plus souvent
mais pas systmatiquement le glissement smantique de ltat caractristique vers lacquisition dtat comme relevant des proprits lexicales
des verbes concerns, exemples :
(a) glissement smantique trait comme inclus dans le sens du verbe :
daufa u to be or become weak ; katura u to be . . . numerous suivi
de : to increase, augment ; saura u to be or become small (1er sens
de ce verbe) ;
340
joseph dichy
(b) glissement smantique non pris en compte dans le sens du verbe :

h asuna u to be handsome, beautiful ; qabuh a u to be ugly ; zarufu
u to be charming (. . .) witty ; saura u (2e sens de ce verbe) : to be
young ; qaduma u to be old, ancient.
Si ce dictionnaire semble hsiter, cest que la question de savoir si ce
glissement smantique est assez frquent pour tre inclus dans le sens
lexical du verbe nest pas toujours facile trancher. Les traductions proposes dans Wehr-Cowan proviennent vraisemblablement des usages
attests dans les fiches des auteurs. Ces donnes montrent en tout tat de
cause que, pour un verbe donn du schme faula, les deux cas prsents
plus haut ( 3.3.4.b) se rencontrent. Voyons de quelle manire :
4.1.2a Le glissement smantique tre x o devenir x
Dans les verbes en faula dont le sens est fondamentalement celui de
ltat caractristique, comme dans h asuba u tre h asb noble, de haute
naissance, un contexte dacquisition de cet tat est presque toujours possible. Cette valeur doit alors tre considre comme smantiquement
seconde, quelle soit inscrite dans le lexique ou quelle soit limite un
nonc donn. Le glissement smantique observ correspond alors au
schma : tre x o devenir x, exemples :
h asuba l-na jruna notre voisin maintenant est devenu un noble
(nonc ironique) ;
karuma r-rajulu lhomme sest montr noble-et-gnreux.
Dans ces deux noncs, la prsence dun adjectif dtat caractristique
associ par les structures morpho-lexicales de la langue aux verbes en
faula suggre une valeur attributive : notre voisin a acquis le titre de
h asb, i.e. de noble de haute naissance (do lironie) ; lhomme a mrit
le nom de karm, noble-et-gnreux.
4.1.2.b Lengendrement de ltat acquis par lvnement ou le
processus du changement dtat (devenir x tre x)
Dans les faula dont le sens de base est celui de lacquisition dun tat,
comme dans badua u devenir incomparable, sans pareil (al-Mujam
al-Wast et al-Munjid, racine /b-d-/), le contexte associe au verbe un
tat acquis, qui est engendr, soit au terme du processus de changement
dtat (ltat acquis correspond alors au dernier tat du processus), soit
par un vnement. Exemple :
faula, faila, faala
341
badua ibn sna bi-ilmihi xayran wa-arran Avicenne fut incomparable par sa science, en bien comme en mal. Du vivant du philosophemdecin, ltat acquis et pris une valeur dtat-prsent dont la dure
inclut le moment de lnonciation (Descls, 1994, 73), la traduction
tant alors Avicenne sest rvl . . . ou se rvle . . .
Comme dans les exemples du paragraphe prcdent, le verbe en faula
peut tre interprt comme prenant une valeur attributive dont le prdicat est ladjectif dtat caractristique correspondant : Avicenne mrita
le nom de bad, incomparable et admirable. Ces exemples ralisent le
schma : devenir x tre x (dans lequel se lit engendre).
4.1.3
Schme faula et adjectifs assimils et/ou participes
Si lon carte le glissement smantique tre x o devenir x, le sens

des verbes valeur dtat caractristique exclut les participes actif (ism
al-fil) et passif (ism al-maf l). Les valeurs aspectuo-temporelles de
base des participes sont en fonction du contexte celles de lachvement (qui peut engendrer un tat rsultatif ) ou du progressif (dont le
terme peut, ou non, tre spcifi).24 Ces valeurs ne sont pas compatibles
avec celles de ltat caractristique, dont la dure nest pas borne et qui
ne comporte aucune progression. En outre, le participe passif qui ne
peut, quant lui, tre driv que dun verbe transitif, est donc en quelque sorte doublement exclu.
Les adjectifs assimils ou analogues [au participe] (sifa muabbaha)
qui sont associs aux verbes en faula ainsi quaux verbes dtat caractristique ou non relevant dautres schmes compensent en quelque sorte
labsence de participes. Comparer par exemple les phrases nominales :
hum kuram ils sont gnreux ; dans cette phrase, karm, plur.
kuram, est ladjectif assimil associ au verbe dtat caractristique
karuma u ;
(a) hum tlibn ils sont en train de demander/ils ont demand
ou (b) hum matlbn ils sont demands ; tlib et matlb sont les
24
Pour les deux valeurs de base des participes (dachvement et progressive), comparer avec Roman (1990, 3940). Pour une analyse dveloppe de la valeur smantique
des participes en contexte et notamment du progressif terme spcifi ou non, voir
Dichy (2002/2003, 5.2.2). En ce qui concerne la relation morphologique entre certains
schmes et ladjectif assimil ou le participe actif, comparer la discussion mene ici avec
Larcher (2003, 2526).
342
joseph dichy
participes actif et passif de talaba u demander, qui ne relve pas des

verbes dtat.
Cette division entre dune part les verbes dcrivant un tat et dots dun
adjectif assimil et de lautre les verbes auxquels sont associs des participes, rencontre toutefois un problme. Au chapitre consacr ladjectif assimil (sifa muabbaha), az-Zamaxar (Mufassa l : 230) associe
celui-ci la valeur smantique dimmutabilit man t-tubt (qui renvoie
ce qui est tbit immuable, tabli en quelque sorte hors nonciation du
temps). Il oppose cette dernire la valeur doccurrence dans le temps
man l-h udt, valeur quil associe aux participes. Or il nillustre pas
ces derniers par des verbes dcrivant des vnements ou des processus (comme je lai fait ci-dessus avec talaba u demander), mais par des
verbes associs un adjectif dtat caractristique. Le premier exemple
donn est celui de h sin, qui correspond un participe actif de h asuna u
tre beau (adjectif dtat caractristique h asan). Az-Zamaxar souligne
la valeur doccurrence dans le temps man h udt de la forme h sin en
insrant celle-ci dans les phrases :
huwa h sin al-na ou adan, mot--mot : il est en train de devenir/il
est devenu beau maintenant, ou il le deviendra demain.
Les exemples suivants sont : krim, participe actif de karuma u tre
noble et gnreux, til, participe actif de tla u tre ou devenir long
et diq, participe actif du verbe dtat caractristique dqa i tre ou
devenir troit (adjectif dtat caractristique dayyiq), pour lequel
az-Zamaxar cite le verset : wa-diqun bihi sadruka, alors que tu
es angoiss par cela, mot--mot, ta poitrine tant oppresse (rendue
troite) par cela (Coran 11 : 12).
Ibn Ya commente, dans ce verset, le choix du participe actif diq au
lieu de ladjectif dayyiq en indiquant que le premier rfre une troitesse (i.e. une oppression) [correspondant ] un accident [se produisant] dans le prsent et non un [caractre] immuable (dq rid f l-h l
ayr tbit arh al-Mufassa l VI : 83).
Si les verbes associs des adjectifs dtat caractristique peuvent tre
compatibles avec le sens des participes actifs, cest en raison :
(a) soit du glissement de sens qui permet ces verbes, lorsque leur sens
premier est celui de verbes dtat, de prendre celui de lacquisition dtat
(glissement tre x o devenir x),
faula, faila, faala
343
(b) soit du fait que ces verbes rfrent, fondamentalement, lacquisition

dun tat caractristique, cest--dire, un processus ou un vnement
engendrant un tat rsultatif (schma devenir x tre x).
Les quatre exemples cits par az-Zamaxar relvent de ces deux cas.
Si h asuna tre beau relve de la premire catgorie, on a vu qual-Mubarrid, cit au 3.3.3.b, considrait karuma comme ayant une valeur
dacquisition dtat, et que le dictionnaire de Hans Wehr-Cowan oscillait
entre la valeur de base attribuer chaque verbe. La valeur attributive
se distribue, comme on la vu, sur les verbes relevant tant de (a) que de
(b). Il apparat clairement, en consquence, que lon ne peut rduire les
valeurs smantiques du schme faula au seul sens de verbe dtat : il sy
ajoute lacquisition dtat.
4.1.4 Deux consquences de lagentivit neutralise ou de la non
agentivit des faula
Lagentivit neutralise des verbes valeur dtat caractristique, et la
non agentivit des verbes dnotant lacquisition dun tat caractristique entranent galement limpossibilit pour ceux-ci de limpratif :
ce dernier nest possible que si un certain degr de contrle exerc par
lagent sur le procs est envisageable (cest pourquoi les verbes pouvoir
et devoir nont pas dimpratif en franais). En outre, lagentivit neutralise est incompatible avec la transitivit. Il sensuit que la conjugaison
des verbes correspondants exclut toutes les formes du passif (ce dernier
ntant possible quavec des verbes transitifs, dans le cadre de contraintes
smantiques prcises).25
Le schme faula est relativement peu frquent dans les textes, notamment
en arabe littraire moderne. On lui prfre sans doute les constructions
adjectivales correspondantes. Cela ne signifie nullement que ce schme
soit sorti de lusage ou en voie de disparition, comme le montre lexemple
de daufa ci-dessus, emprunt la presse actuelle. On trouve galement
assez souvent dans la prose contemporaine des emplois de construction
impersonnelle, la faveur, pour ainsi dire, de lagentivit neutralise :
l yah sunu an . . ., il ne convient pas que . . . [Yawm., 79, 18] ; yajduru,
il convient, suivi de an ou dune forme infinitive . . .
25
Les critres qui permettent, interdisent ou restreignent lapparition de limpratif

ou du passif en arabe ont t prsents dans Ammar et Dichy (1999, 1920).
344
4.2
joseph dichy
Les valeurs smantiques de base du schme faila
Le deuxime schme du verbe simple est dalternance vocalique i/a.

Comme on la vu au 2.2.1, une variante en i/i porte sur un petit nombre
de verbes : sur 15 ou 16 verbes inventoris par les traits de morphologie
arabe mdivale, huit sont attests en arabe moderne par le dictionnaire
de Hans Wehr (cf. liste dans Ammar et Dichy 1999, 24). Cette variante,
qui affecte essentiellement des verbes de 1re consonne radicale w ou y est
purement formelle, et nentrane pas de diffrence de valeur smantique
avec le schme faila dont elle relve. Exemples :
warita i hriter ; waliya i suivre ou succder .
On rencontre dans les verbes du schme faila tous les degrs dagentivit (pleine ou partielle) ou labsence de cette proprit (non agentivit,
agentivit neutralise), ainsi que les diffrentes formes de transitivit ou
de son contraire. Do la partition des sens associables ce schme qui
a t observe par les auteurs qui se sont attachs le caractriser.26 La
valeur la plus gnrale est celle de verbes moyens ou de diathse interne
(au sens revisit prsent plus haut), ce schme comportant en outre un
nombre limit de verbes associs un adjectif dtat caractristique.
Il y a quatre grandes catgories smantiques de verbes en faila, distingues principalement au moyen des proprits dagentivit et de transitivit.
4.2.1 Les faila associs un adjectif dtat caractristique (verbes dtat
ou dacquisition dtat)
Un certain nombre de verbes dalternance vocalique i/a (schme faila)
relve des verbes associs un adjectif dtat caractristique. Ces derniers
incluent tous les traits prsents plus haut propos du schme faula, et
comportent comme eux les deux grandes valeurs smantiques troitement corrles tre x et devenir x. Les verbes en faila relevant de cette
sous-catgorie sont intransitifs, dagentivit neutralise (valeur tre x)
ou non agentifs (mdio-passifs, de valeur devenir x), et ne peuvent avoir
de participe actif que dans les conditions, rarement ralises, prsentes
plus haut ( 4.1.3). Ils dcrivent pour la plupart un tat caractristique
qui correspond :
26
Outre M. Cohen (1929) et Joon (1930), voir notamment : Fleisch (1979, 220
suiv.) ; Ammar et Dichy (1999, 2324).
faula, faila, faala
345
soit une particularit physique ou psychique (dfaut ou qualit)27

affectant un sujet humain ou un animal, domestique ou appartenant
un troupeau,
soit une particularit, souvent mais pas ncessairement connote
ngativement, affectant une entit concrte ou abstraite.
Ladjectif assimil est le plus souvent du schme af al, fm. fal.
Une srie limite de verbes de cette sous-catgorie relve de racines 2e
radicale w ou y. Jen ai dnombr 71, dont 18 de 2e consonne y dans Ilys
et Nsf (1995). Ces verbes ont pour particularit de ne pas appliquer
les rgles de transformation phonologiques habituelles.28 Exemples :
(a) Verbes adjectif dtat caractristique dcrivant une qualit ou un
dfaut le plus souvent physique (humain ou animal) :
jayida yajyadu avoir le cou long (signe de beaut) adj. ajyad, fm.
jayd ; awiza yawazu tre ou tomber dans le besoin (adj. awaz) ;
layisa yalyasu tre ou se montrer courageux, tenir sa position (dans
un combat) adj. alyas ; rawia yawau tre ou se montrer admirable (humain) pour son intelligence, sa beaut, son courage, etc. (adj.
arwa) ; awira yawaru tre ou devenir borgne (adj. awar) ; qawida
a avoir le dos et le cou allongs (cheval . . .).
(b) Verbes adjectif dtat caractristique dcrivant une particularit
affectant une entit non anime concrte ou abstraite :
awisa yawasu tre ou devenir difficile, incomprhensible, abscons
(affaire, explication) adj. awas et aws ; awija yawaju tre ou
devenir tordu, incurv, tortueux (adj. awaj, fm. awj) ; ayisa
yayasu tre ou devenir doux, souple (adj. ayas) ; qawira yaqwaru
tre vaste (maison) adj. aqwar.
Un nombre galement limit, mais plus important, de verbes en faila
associs un adjectif dtat caractristique relve de racines sans transformation, exemples :
27
Voir notamment : Sbawayhi, Kitb IV : 17 ; az-Zamaxar, Mufassa l : 278. La synthse ci-dessus provient de mon propre examen des donnes.
28
Rappelons que dans xfa, yaxfu craindre, dalternance vocalique i/a, la 2e
consonne radicale w est lobjet de transformations en fonction de la voyelle qui la suit
et la prcde immdiatement, la diffrence des verbes de 2e radicale w ou y dont on
trouve ici des exemples.
346
joseph dichy
(c) Verbes dcrivant une qualit ou un dfaut chez un humain ou un

animal :
baliha a devenir faible desprit ou tre stupide, idiot (adj. ablah, fm.
balh) ; ramida a devenir, puis tre chassieux (il) ou (mot--mot)
devenir ayant, puis avoir lil chassieux (humain) adj. armad ;
darida a perdre ses dents, devenir ou tre dent (adj. adrad) ;
sawifa a (mot--mot) devenir ayant, puis avoir beaucoup de laine,
tre laineux (ovin ou caprin) adj. aswaf.
(d) Verbes dcrivant une particularit affectant une entit concrte ou
abstraite :
xadira a verdir, puis tre vert (adj. axdar vert) ; sawida a noircir, puis tre noir (adj. aswad noir) ; xaliqa a suser, puis tre us
(vtement) adj. axlaq ; sawisa a se gter, tre rong par des vers
(viande, nourriture) adj. aswas.
(e) Verbes de cette sous-catgorie dont ladjectif nest pas de schme
af al :
yasira yaysaru devenir puis tre facile (adj. yasir et yasr).
La valeur smantique de base des verbes prsents en (c), (d) et (e) est
celle de lacquisition dtat, la diffrence des verbes des sous-catgories
(a) et (b). Dans leur sens de base, les premiers sont donc non agentifs, et
correspondent des verbes mdio-passifs.
4.2.2
Autres verbes mdio-passifs et non agentifs en faila
Dautres verbes du mme schme sont non agentifs et intransitifs. Ils

correspondent, non des verbes diathse interne ou moyens, comme
dans les deux paragraphes suivants, mais des verbes mdio-passifs.
Exemples :
verbes de sujet grammatical non-humain :
rasila a tre longue et pendre (chevelure) ; rawiya a tre irrigu,
arros (terre) ; xadiba a verdir (arbre, jardin, terre) ;
verbes de sujet grammatical humain :
ariqa a se noyer (au sens propre), mourir noy adj. assimil
arq ; marida a, tre ou tomber malade, adjectif assimil : mard.
faula, faila, faala
347
Tous les verbes de cette catgorie ne sont pas associs, dans les dictionnaires, un adjectif assimil. Dans le cas des deux derniers exemples
ci-dessus, il est noter que le participe actif est attest, mais avec une
diffrence de sens du plus grand intrt pour la distinction entre les
valeurs smantiques du participe actif et celles de ladjectif assimil (cidessus, 4.1.3) :
Le Munjid al-lua l-arabiyya l-musira (2000) donne lexemple de
anqaztu riqan jai sauv un noy (racine /-r-q/), et Roman : l
takul hd t-tama fa-innaka mridun in akaltahu Ne mange pas
de cette nourriture, car tu seras malade si tu en manges (1999, 71,
daprs le Lisn al-arab, XIIIe s., racine /m-r-d/).
Comparons ces emplois. Les deux adjectifs assimils arq et mard
correspondent des tats rsultatifs engendrs par lvnement ou concidant avec la dernire tape du processus dcrit par le verbe (Descls
1994) : le processus se noyer a pour dernire tape ( !) tre mort noy,
dcrit par ladjectif arq, de mme que tomber malade entrane ltat
dcrit par ladj. mard. Par contraste, les participes actifs correspondants riq et mrid ont une valeur de progressif terme spcifi (Dichy
2002/2003, 34, 36, 60) :
le procs de noyade na pas atteint son terme dans lexemple du Munjid (le sauvetage a pu avoir lieu) ;
le terme (ou lachvement) du procs tomber malade est projet dans
le futur dans lexemple du Lisn cit par A. Roman. Ce dictionnaire
glose, significativement mridun la suite de la phrase ci-dessus par
ay tamradu cest--dire : tu seras malade.
Il est significatif ce quatteste notamment lexemple du Munjid al-lua
l-arabiyya l-musira que ces drivations restent vivantes en arabe
contemporain, mme si leur frquence est faible.
4.2.3 Les verbes de schme faila diathse interne (ou moyens) et
dagentivit partielle
Un nombre beaucoup plus important de verbes du schme faila correspond des verbes diathse interne, ou verbes moyens : le procs affecte
le sujet, que le verbe soit transitif ou non (ainsi : rire de qqun ou rire).
Ces verbes sont transitifs indirects ou intransitifs et dagentivit partielle
348
joseph dichy
(lagent, ntant pas autonome par rapport au procs, ne conserve quun

contrle limit sur celui-ci). Ils peuvent soit admettre un participe actif,
mais non un adjectif assimil, soit tre associs dans le lexique un
adjectif assimil, et nadmettre de participe actif que dans les constructions vues au paragraphe prcdent et en 4.1.3. Exemples :
(a) Verbes admettant un participe actif, mais non un adjectif assimil :
verbes intransitifs :
baqiya a rester, demeurer qq part (agent humain) [Yawm., 12, 12] ;
nma a dormir ; dah ika a rire (en emploi absolu, diffrent de : rire
de . . ., se moquer de . . .) ;
verbes transitifs directs ou indirects :
dah ika a min ou al rire de, se moquer de qqun ; nasiya a oublier
qqch ; samia a entendre qqch ; saima a sennuyer de . . . (qqch)
[Yawm., 50, 10] ; h asiba a anna . . . penser, croire, estimer que . . . ;
alima a anna . . . savoir que ou apprendre que . . . alima a bi-, avoir
connaissance de . . . .
(b) Verbes associs dans le lexique un adjectif assimil, mais nadmettant de participe actif que dans les constructions lies loccurrence dun
procs (h udt) 4.1.3 :
verbes intransitifs :
atia a avoir soif (adj. atn) ; taiba a prouver de la fatigue, tre
fatigu (adj. taib) ; xafira a prouver un fort sentiment de timidit,
tre trs timide, notamment pour une jeune fille (adj. xafir);29 salima
a tre sain et sauf , chapper ( un mal ou un danger) adj. salm,
mais galement slim ;30
verbes transitifs directs ou indirects :
h afiya a bi- accueillir trs chaleureusement qqun (adj. h afiyy) ; adiba
a al tre, se mettre en colre contre qqun (adj. adib).
Dans les exemples que lon vient de voir, la catgorie morphologique
de ladjectif assimil ne correspond pas un tat caractristique, mais
29
Il existe un second adjectif assimil, mixfar grande timide : associ ce dernier,
xafirat tre une grande timide relve des verbes valeur dtat caractristique, dans
lesquels lagentivit est neutralise.
30
Dans des cas tels que slim ladjectif assimil est dit, dans la grammaire arabe traditionnelle, prendre la forme (sa) du participe actif. Autres exemples : yis dsespr,
bis misrable, mhir adroit, habile (artisan) . . .
faula, faila, faala
349
plutt ltat rsultatif produit pas le procs (vnement ou processus) :

le verbe salima produit ainsi salm sain et sauf , mais galement slim
min . . . sauf de . . . (mal ou danger). Dans un verbe exprimant un processus, comme dans adiba se mettre en colre, ladjectif adib dcrit
le dernier tat de ce processus ; si adiba se trouve, dans un contexte
donn, dcrire un vnement, ladjectif dcrit ltat rsultatif engendr
par cet vnement.
Ces verbes permettent de prciser la notion dagentivit partielle. Dans
les exemples ci-dessus, le contrle limit du procs par lagent peut surtout tre glos ngativement :31
refuser de rester quelque part, voire dtre sain et sauf dun danger
(et agir pour sortir de ce lieu ou de cette situation) ; combattre ou surmonter un sentiment ou une motion (rire, ennui, timidit, chaleur
humaine ou colre envers quelquun), une sensation (sommeil, fait
dentendre, soif , fatigue), un mouvement de la pense ou processus cognitif (croire, oublier, savoir), etc.
On notera enfin que la relation entre le verbe et, soit un participe actif,
soit un adjectif assimil, est lexicalise (et non compositionnelle, comme
elle ltait sans doute aux plus anciennes poques de la langue arabe) : on
doit recourir aux connaissances lexicales pour la construire, et non aux
seules connaissances grammaticales ou une dduction partir du sens
du verbe.
4.2.4 Les verbes du schme faila diathse interne, transitifs et
dagentivit entire
Lautre grande partie des verbes du schme faila est constitue de verbes
galement moyens (ou diathse interne), dagentivit entire et de rgime
syntaxique transitif direct ou indirect. Dans cette configuration, le procs comporte un agent autonome contrlant effectivement (du fait de
lagentivit entire) lvnement ou le processus dcrit par le verbe et un
patient (du fait de la transitivit). Il englobe en outre dune manire ou
dune autre lagent ou se droule au moins partiellement en lui (diathse
31
Cf. Roman (1990, 4243), qui dcrit lagentivit partielle comme une agentivit
ractive.
350
joseph dichy
interne). Ces verbes ont un participe actif, et nadmettent pas dadjectif

assimil. Exemples :
ariba a, boire ; radiya a bi- tre satisfait de . . . ; lah iqa, suivre, poursuivre qqun, lah iqa bi-, suivre qqun au sens de le rejoindre [Yawm.,
66, 21] ; rakiba a, monter, chevaucher ; watiqa i bi-, faire confiance
(qqun).
4.3
Les valeurs smantiques de faala et leur dispersion
Le troisime schme de la forme simple du verbe est faala. Il comporte

comme on la vu une alternance vocalique de base ( faala i) et deux
variantes rsultant de divers conditionnements phontiques ( faala u et
faala a).
On pourrait poser que ce schme comprend majoritairement des verbes relevant de la catgorie des verbes actifs ou externes qui soppose
structurellement celle des verbes moyens, au sens donn ces termes
dans larticle de Benveniste cit plus haut (section 3.4). Ce trait serait
alors ce qui distingue faala des schmes faila et faula.
Or il existe un nombre trs lev de verbes ne rpondant pas cette
dfinition de principe. Nous avons vu que ce schme connaissait une
importante dispersion sur le plan formel : lordonnance systmatique du
tableau des schmes de base du verbe simple (fig. 1) est principalement
rompue dans le tableau des schmes du verbe simple et de leurs sous-catgories (fig. 2) par les trois alternances vocaliques possibles du schme
faala. Les exemples ci-dessous montrent en outre que cette dispersion
se vrifie galement du point de vue smantique.
Considrons dabord les catgories manant des proprits dagentivit
et de transitivit :
Verbes pleinement agentifs et transitifs (directs ou indirects) :
amana u bi- avoir foi en ou dans ; qla u dire ; saala a interroger,
questionner ; sh a i bi- crier vers, en direction de, contre qqun, le
hler ; da u il revenir .
Verbes pleinement agentifs et transitifs indirects, auxquels est associ un
adjectif exprimant un tat rsultatif, ou tat acquis :
latafa u bi- se montrer gentil, prvenant envers qqun (adjectif assimil, exprimant un tat rsultatif : latf cf. Coran 42 : 19) ; daxala u
f entrer dans un clan ou une tribu, se placer sous sa protection (adj.
exprimant un tat rsultatif : daxl) ;
faula, faila, faala
351
Verbes pleinement agentifs et intransitifs :

makata u rester, demeurer (sujet humain) ; jalasa i sasseoir ; haraa
a se prcipiter (tte baisse) ; sakata u se taire (aprs avoir parl) ;
samata u se taire, garder le silence.
Verbes dagentivit partielle transitifs (directs ou indirects) :
ra a voir ; fza bi- gagner, obtenir qqch ; aara u bi- sentir, ressentir qqch.
Verbes dagentivit partielle intransitifs :
zahara a apparatre (intentionnellement) lorsque lagent est humain32
[Yawm., 32, 22] ; warada i, apparatre intentionnellement, arriver
(avec un agent humain).
Verbes non agentifs intransitifs :
mad i passer, se passer (laps de temps, vnements ou processus
dont on prend en compte la dure) [Yawm., 32, 16] ; bad u paratre
(avec un sujet non humain) [Yawm., 32, 17] ; h na i se produire accidentellement [Yawm., 61, 15] ; araba u se coucher (soleil) ; mta u
mourir.
Verbes non agentifs transitifs (directs ou indirects) :
waqa a al tomber sur [Yawm., 32, 17] ; h aw i contenir, inclure,
comprendre.
Verbes dagentivit neutralise ou non-agentifs et intransitifs, associs
un adjectif assimil dcrivant un tat caractristique (verbes dacquisition dtat) :
dqa i tre ou devenir troit (adjectif dtat caractristique : dayyiq) ;
jda u tre ou devenir excellent (adj. jayyid) ; xaffa i tre ou devenir
lger (adj. xaff ) ; adda u tre ou devenir intense, dur ou violent
(adj. add) ; samala u devenir ou tre rigide ou sec (adj. smil et
saml).
Reprenons cette analyse en classant les verbes de schme faala selon les
grandes catgories prsentes plus haut. On a :
32
Lagentivit est susceptible de varier selon que le sujet grammatical rfre un
humain (qil) agissant intentionnellement ou non, ou un non humain (ayr qil).
Avec un sujet grammatical non humain, ou humain, mais dpourvu, dans le contexte,
dintentionnalit, zahara ou warada sont des verbes non agentifs. Un autre exemple (du
schme faila cette fois) est dans Yawm. (16, 15) : fa zafira n-nawmu bi-jufn, le sommeil triompha de mes paupires. Dans cette phrase, le verbe zafira perd le trait dagentivit quil aurait avec un sujet grammatical humain, et devient non agentif (le sommeil ne
disposant daucun contrle sur le procs, dont il nest dailleurs pas lagent). En revanche,
waqaa tomber demeure non agentif, que le sujet soit humain [Yawm. 16, 12] ou non
humain [Yawm. 16, 20 et 32, 17].
352
joseph dichy
(a) Verbes valeur dtat caractristique ou dacquisition dtat caractristique repris en partie des exemples ci-dessus :
dqa i tre ou devenir troit (adjectif dtat caractristique : dayyiq) ;
lna i tre ou devenir souple (sens physique), ou tendre (sens psychologique affectant un sujet humain) (adj. layyin) ; jda u tre
ou devenir excellent (adj. jayyid) ; sda u tre ou devenir le chef
(adj. sayyid) ; xaffa i tre ou devenir lger (adj. xaff ) ; qalla i tre
ou devenir peu nombreux (adj. qall) ; adda i et u tre ou devenir
intense, dur ou violent (adj. add) ; samala u devenir ou tre rigide
ou sec (adj. smil et saml) ; fasada u et i se corrompre, pourrir,
i.e. pour une viande, de la nourriture ou les murs dune personne,
etc. : devenir, puis tre corrompu (adj. fsid, fasd).
(b) Verbes diathse interne ou moyens repris des exemples ci-dessus :
Verbes associs dans le lexique un adjectif assimil exprimant
un tat rsultatif :
latafa u bi- se montrer gentil, prvenant envers qqun (adj. latf ) ;
daxala u f entrer dans un clan ou une tribu, se placer sous sa
protection (adj. daxl) ;
Verbes sans adjectif assimil :
amana u bi- avoir foi en ou dans ; makata u rester, demeurer
(agent humain) ; jalasa i sasseoir ; haraa a se prcipiter ; ra a
voir ; da u il revenir ; fza bi-, gagner, obtenir qqch ; zahara
a apparatre, se manifester intentionnellement (agent humain)
[Yawm., 32, 22] ; warada i apparatre intentionnellement, arriver (agent humain).
(c) Verbes mdio-passifs repris des exemples ci-dessus :
mad i passer, se passer (laps de temps, vnement ou processus
dont on prend en compte la dure) [Yawm., 32, 16] ; bad u paratre (avec un sujet non humain) [Yawm., 32, 17] ; h na i se produire
accidentellement (vnement) [Yawm., 61, 15] ; araba u se coucher (soleil) ; mta u mourir ; saqata a tomber, chuter ; h aw i
contenir, inclure, comprendre.
(d) Verbes diathse externe (transitifs) :
qatala u tuer ; daraba i frapper ; h asaba i compter (des objets) ;
madda u tendre, tendre ; sah ara a charmer, ensorceler ; sajana u
emprisonner ; xadaa a tromper, duper, leurrer ; fasala i sparer,
diviser ; qataa a couper ; fatah a a ouvrir ; ba i vendre ; qla u
dire ; saala a interroger, questionner ; talaba u min an . . . de-
faula, faila, faala
353
mander qqun de . . . ; sh a i bi- crier vers, en direction de qqun, le

hler [Yawm., 61, 11].33
4.3.1 Les corrlats formels de la prsence de verbes adjectif dtat
caractristique dans le schme faala
Les verbes adjectif dtat caractristique (tat caractristique ou acquis)
du schme faala, dont on aurait attendu la prsence parmi les faula ,
sont associs des corrlats formels. Ils relvent :
soit des verbes concaves (ajwaf, i.e. de 2e consonne radicale w ou y),
pour lesquelles la ralisation de lalternance vocalique u/u ne semble
atteste que dans le cas de tla/yatlu adj. tawl (bien que lon
puisse galement voir dans cette forme un verbe en a/u) ;
soit des verbes redoubls (mudaaf, de 2e et 3e consonnes radicales
identiques), pour lesquels il nexiste quun tout petit nombre de verbes en faula, qui sont signals comme des exceptions par Sbawayhi.34
Deux verbes redoubls en u/u sont attests en arabe contemporain
(Hans Wehr-Cowan 1979) : arra (1e pers. arurtu) u se monter ou tre
mauvais, malfaisant (adj. arr) et labba (1e pers. labubtu) u se montrer ou tre intelligent (adj. labb autre alternance vocalique : i/a).
Ces deux verbes sont en outre donns par le dictionnaire dal-Fayrz
bd, dans la glose de h abba (1e pers. h abubtu) u il devenir un
proche [qui reoit des marques damiti] de qqun (adj. h abb), comme
tant, avec ce verbe, les seuls membres de cette classe formelle.35
33
Les verbes de parole du schme simple sont en arabe des verbes diathse externe.
Le passage une diathse interne se fait au moyen de schmes augments incluant ce que
Roman (1990 ; 1999/2005) appelle le morphme-cho t voir Ammar et Dichy (1999,
2728) ; Dichy (2002/2003, 4.2). Exemples : sala a interroger [schme simple, verbe
diathse externe] vs sala an interroger, questionner (qqun) sur, puis tasala an sinterroger sur [schme augment incluant le morphme-cho, verbe diathse interne] ;
qla u dire [schme simple, verbe de diathse externe] vs qawwala attribuer (des paroles) qqun, puis taqawwala prtendre, allguer (au profit, dans lintrt, en faveur de
soi-mme) [schme augment incluant le morphme-cho, verbe de diathse interne].
34
Sbawayhi indique quil ny a presque pas de verbes en faula de 2e et troisime radicale identiques (ou redoubls, mina t-tadf Kitb IV : 3637). Ibn Xlawayh (Laysa . . . :
27), signale labsence de verbes redoubls en faula, lexception des deux cas de labba u
(dj cit par Sbawayhi, loc. cit.) et azza u avoir peu de lait (dit dune chvre).
35
Malgr lindication cite, on trouve dans le dictionnaire dal-Fayrz bd trois
autres verbes : azza (3e pers. fm. sing. azuzat) u voir (pour une chvre) son lait diminuer, avoir peu de lait (adj. azz), ainsi que damma (1e pers. damumtu) u devenir ou
tre trs laid, repoussant (adj. damm), et fakka (1e pers. fakuktu) u, se montrer ou tre
354
joseph dichy
Toutefois, il ne sagit pas dune rgle oprant dans les deux sens : tous les
verbes concaves ou redoubls de schme faala ne relvent pas, il sen
faut de beaucoup des verbes adjectif dtat caractristique. Les exemples de samala u devenir ou tre rigide ou sec et de fasada i et u voir
4.3, exemples (a) montrent quil existe galement un certain nombre de faala adjectif dtat caractristique qui ne relvent daucune des
deux catgories formelles que lon vient dindiquer. Leur sens de base est
toutefois celui de lacquisition dun tat caractristique plutt que celui
de verbes dtat.
4.3.2 Un glissement smantique partiellement responsable de la
prsence de verbes diathse interne et de verbes mdio-passifs
dans le schme faala
Les verbes diathse externe sont agentifs et transitifs : le procs verbal affecte un objet externe lagent. Un glissement smantique trs
rpandu et observable dans plusieurs langues peut toutefois modifier la
diathse de ces verbes, qui devient alors interne. Deux cas se prsentent,
selon que la valeur smantique rsultant de ce glissement est agentive ou
non. Dans le premier, le sens produit est celui dun verbe moyen ou a
diathse interne : le verbe est agentif et le procs affecte lagent. Dans le
second, le sens produit par le glissement est mdio-passif : le procs, qui
est non agentif, affecte le sujet grammatical du verbe (lagent existe, mais
il nest pas dsign). Comparons les donnes de plusieurs langues :
Verbes agentifs diathse interne, partir de verbes diathse
externe :
En franais, le sens de base du verbe transitif plonger est celui dun
procs diathse externe dans lequel un agent plonge un objet
dans un liquide, ex. : Pierre plongea sa chemise dans leau de la lessive. Ce verbe a pour pendant une forme pronominale exprimant
une diathse interne, se plonger, dont le sens est mtaphorique,
ex. : se plonger dans un livre (mais non *dans la piscine). En arabe,
les sens de plonger et se plonger sont associs au verbe atasa i.
Joon donne un exemple similaire en hbreu : taval tremper et se
tremper (1923, 95). La diffrence est dans la prsence en franais
dune marque morphosyntaxique, celle des verbes pronominaux.
stupide ou faible, ces deux derniers verbes pouvant galement tre dalternance i/a. Cf.
aussi Ibn Xlawayh, Laysa . . . : 73 (note de lauteur).
faula, faila, faala
355
Observons toutefois quen franais, de manire en partie similaire

taval en hbreu et atasa en arabe, il existe, outre les deux sens
que lon vient de voir, une forme intransitive moyenne plonger, de
sens physique, sans marque morphosyntaxique associe.36 Tous ces
exemples ont en commun dtre lexicaliss (la relation forme-sens
ne rsulte par dun phnomne de drivation compositionnel, mais
est inscrite dans le lexique).
Le verbe diathse externe daraba i frapper, battre en anglais,
to hit, to beat est cit dans tous les ouvrages, en langue arabe
ou en langues occidentales, pour illustrer sens du schme faala,
lexemple mille fois reproduit tant : daraba zaydun amran Zayd
a frapp Amr. Il comporte toutefois un sens dans lequel le procs
exprim par le verbe ne rencontre pas dobjet externe lagent, mais
affecte ce dernier (le verbe est donc diathse interne), comme
dans lexpression fige daraba f l-ardi parcourir la terre (ou une
contre, une rgion . . .). Un glissement smantique analogue se
produit en franais dans les expressions battre les chemins (sens
propre), ou battre la campagne (sens figur).
Verbes mdio-passifs (et donc non agentifs), partir de verbes diathse externe :
Dans axada qalbuhu yadribu f sadrihi son cur se mit battre
dans sa poitrine, le verbe daraba prend une valeur non agentive ; le
procs affecte le sujet grammatical. Il est significatif que le mme
glissement smantique est observable en franais avec battre et en
anglais avec to beat dans les expressions correspondantes.
Le verbe diathse externe aqqa u fendre, briser (en anglais to
break) prend galement le sens de clore (bouton de fleur . . .),
cest--dire, fendre sa gaine, ou, pour une dent pousser. Ces
deux sens sont associs, respectivement, aux formes infinitives
(masdar) aqq et uqq. Comme dans tous les exemples ci-dessus,
le sens mdio-passif ou diathse interne drive du sens transitif
et diathse externe : il sagit dun phnomne de polysmie, inscrit
dans le lexique.
36
Toutefois, la diffrence des exemples emprunts larabe et lhbreu, le sens
physique de plonger intransitif et moyen cit ici comporte en franais un ajout
smantique (ide de plongeon), les donnes attaches ce verbe tant relativement
complexes. On a, par exemple, soit : plonger dans la piscine de son jardin, soit plonger
les mains dans leau froide du bain, mais on ne peut plonger (= faire un plongeon) dans
son bain . . .
356
joseph dichy
4.3.3 Les faala diathse interne ou mdio-passifs ne correspondant ni

un corrlat formel ni un glissement smantique
Il existe par ailleurs au sein du schme faala des verbes dont le sens est
ou en tout cas parat tre originellement celui de procs diathse
interne ou mdio-passifs, et qui ne relvent pas formellement des verbes
concaves ou redoubls. Exemples :
faala diathse interne :
dahaba a partir, aller, sen aller (agent humain) ; ma i marcher ;
daxala u entrer (agent humain) ; akala u manger (alors que ariba
a boire est, comme attendu, de schme faila) ; lamasa a toucher et
ra a voir (mais samia a entendre relve de faila) . . .
faala mdio-passifs :
zahaqa a prir, svanouir, disparatre ; waqaa a tomber ou se trouver, tre situ (pour un lieu, une ville, etc.) ; dahaba a sen aller (sujet
grammatical non humain) ; fasada u et i se corrompre, pourrir (adj.
fsid, fasd).
4.3.4
Dispersion et rgularit dans le schme faala
Les exemples ci-dessus montrent quon trouve parmi les verbes en faala
toutes les catgories et proprits smantiques associes aux deux autres
schmes. la dispersion formelle prsente au dbut de ce travail semble
donc rpondre une dispersion smantique quivalente, dont on constate
lexamen quelle ne semble pas pouvoir tre rduite des corrlats formels ou des glissements smantiques.
Une grande rgularit smantique peut toutefois tre observe : les
verbes diathse externe apparaissent exclusivement dans le schme
faala.
Cette zone de stabilit est, selon toute vraisemblance, rendue possible
par lorientation particulire du glissement smantique qui permet un
verbe diathse externe de devenir un verbe diathse interne ou un
verbe mdio-passif lorsque le procs verbal revient sur lagent ou sur
le sujet grammatical, ou, selon la formule de Benveniste, quand le sujet
devient intrieur au procs (1950/1966), exemples :
wadaa a poser, dposer est un verbe diathse externe : lagent
effectue laction dimprimer un objet le mouvement correspondant.
Le mme verbe prend le sens diathse interne de accoucher, qui
peut tre soit intransitif, soit transitif (Coran 3 : 36 et 46 : 15). Lagent
devient, dans ce cas, le lieu du procs. De mme, dans h amala i, le
faula, faila, faala
357
sens diathse externe porter (un objet . . .) devient, avec une diathse interne : porter [un enfant], devenir ou tre enceinte.
Le verbe wasala i joindre, unir une chose une autre (avec la prp.
bi-) est, dans son sens premier, diathse externe ; il est transitif un
double objet (de manire, respectivement, directe et indirecte). Au
sens de tre uni quelquun par un lien de parent, damiti . . . et de
frquenter assidment quelquun, lagent est en mme temps lun des
deux objets du procs : le verbe est alors diathse interne et transitif
( un seul objet).
Ces exemples, qui sajoutent ceux qui ont t prsents au 4.3.2,
illustrent un processus qui opre dans un sens, le sujet grammatical ou
lagent devenant le lieu du procs (comme dans wadaa et h amala) ou
lobjet de celui-ci (comme dans wasala). On peut en revanche formuler
lhypothse que le glissement smantique de la diathse interne vers la
diathse externe nest pas possible, compte-tenu des donnes de larabe :
une fois le sujet impliqu dans le procs, le verbe conserve cette proprit smantique. Le passage de la diathse interne (exemple : jalasa i
sasseoir) une diathse externe sopre en construisant un autre verbe,
de mme racine, mais de schme diffrent, ainsi : ajlasa faire asseoir
(schme IV, af ala, de valeur causative-factitive).
Le passage dun verbe dtat caractristique un verbe diathse
interne se produit galement en changeant de schme, ainsi :
latufa u tre ou devenir gentil ou subtil est de schme faula ; latafa
u traiter quelquun avec gentillesse, de schme faala, est diathse
interne, la gentillesse affectant lagent.
Ces donnes dobservation permettent donc de formuler une hypothse
de porte descriptive.
5. Structure de la relation entre sens et forme

dans les schmes simples
La dispersion de la relation entre sens et forme dans les schmes du
verbe simple nest donc pas irrductible. Les raisons de cette dispersion ne sont pas, me semble-t-il, chercher dans la dtrioration dun
schma originel ou dun proto-schma qui entrerait en cohrence avec
celui des relations formelles entre les deux paradigmes de la conjugaison
objet de la figure 1. Sil parat raisonnable de partir dun proto-schma
358
joseph dichy
prsentant sur le mode hypothtique la relation entre sens et forme des

schmes du verbe simple dans son principe, il est essentiel de ne pas sarrter cette hypothse. Celle-ci est fonde sur lide que la relation sens
forme dans les schmes serait essentiellement, voire originellement,
de nature bijective. Or je montrerai plus loin que ce nest gure le cas.
5.1 Le proto-schma, reconsidrer, des relation entre sens et forme
dans les schmes simples
Le principe de ce proto-schma associerait, de manire exclusive, la
valeur smantique des verbes adjectif dtat caractristique au schme
faula, celle des verbes diathse interne ou mdio-passifs faila, et
celle des verbes diathse externe faala :
Schme
Sens grammatical
faula
Verbes adjectif dtat caractristique (tat ou

acquisition dtat)
faila
Verbes diathse interne ou mdio-passifs
faala
Verbes diathse externe
FIGURE 3 LES RELATIONS ENTRE SENS ET FORME DANS LE

VERBE SIMPLE, CONSIDRES DE MANIRE HYPOTHTIQUE
DANS LEUR PRINCIPE
Toutes les relations ci-dessus sont bijectives (ce qui est reprsent par
une flche deux ttes ). Il est peu vraisemblable quun tel systme ait
jamais exist en ltat, car, en sus dinluctables transformations morphophonologiques, des glissements smantiques dune catgorie lautre et
divers effets de figement se sont ncessairement fait jour ds les toutes
premires poques de la langue. Un tel glissement est dj perceptible,
dans la figure 3, dans le fait que les verbes adjectif dtat caractristique peuvent se raliser comme des verbes dacquisition dtat, qui sont
mdio-passifs. Or les verbes en faila peuvent galement correspondre
des mdio-passifs. Cest donc plus une reconstruction hypothtique du principe qui gouverne ces relations qu une reconstitution diachronique que ce schma nous convie. Ce principe est reconsidrer
de deux points de vue : il masque larbre conceptuel des notions qui le
sous-tendent ( 5.2) ; il est mis en dfaut par les grandes lignes de glissements smantiques ou de modifications formelles observables dans les
schmes simples ( 5.3).
faula, faila, faala

5.2
359
Larbre conceptuel des notions incluses dans cette structure
Conceptuellement, les grandes notions qui nous ont permis dtudier

les sens grammaticaux des schmes simples ne sont pas mettre sur
le mme plan : comme on la vu plus haut, les verbes diathse interne
(ou moyens) et mdio-passifs dun ct, et les verbes diathse externe
de lautre sont dans une relation dopposition. Or, les uns et les autres
peuvent correspondre un vnement ou un processus, mais non
un tat opposition fondamentale que masquait la distinction couramment rpandue entre verbes daction et verbes dtat. Lensemble
constitu par les verbes valeur dtat caractristique soppose dune part
aux verbes mdio-passifs qui incluent les verbes dacquisition dtat
caractristique ou non , et aux verbes et diathse interne ou externe
de lautre.
Lagentivit enfin opre elle mme une partition entre les verbes
valeur dtat caractristique, dans lesquels cette proprit est neutralise, et les verbes mdio-passifs, qui sont non-agentifs. Par contraste,
les verbes diathse interne ou externe sont, selon le cas, entirement
ou partiellement agentifs. Ces oppositions sont rsumes dans larbre
conceptuel suivant :
notion aspectuelle dtat
(agentivit neutralise)
Verbes valeur
dtat caractristique
FIGURE 4
notions aspectuelles dvnement ou de processus

non agentivit
Verbes
mdio-passifs
( y compris
dacquisition dtat)
agentivit entire ou
partielle
diathse
interne
diathse
externe
Verbes
diathse
interne
Verbes
diathse
externe
LES VALEURS SMANTIQUES DE BASE DES VERBES

DE SCHME SIMPLE, ARBRE CONCEPTUEL
Conventions : Les proprits smantiques sont en caractres gras ; les

catgories smantiques des verbes du schme simple sont en italiques.
On notera quen raison de phnomnes de polysmie, un mme verbe
peut appartenir des catgories smantiques diffrentes : les proprits
ci-dessus peuvent donc correspondre soit des catgories lexicales, soit
des occurrences en contexte.
360
joseph dichy
5.3 Glissements smantiques luvre dans la relation sens-forme dans

les schmes simples
Si lon part de laspect formel des schmes, on observe que :
(a) tous les verbes en faula correspondent un verbe adjectif dtat
caractristique (valeurs dtat caractristique ou dacquisition de
celui-ci) ;
(b) presque tous les verbes en faila relvent de lune ou lautre des
valeurs du verbe diathse interne ou du mdio-passif. Ce schme
inclut par ailleurs un sous-ensemble, rduit en nombre, de verbes
adjectif dtat caractristique ;
(c) les verbes en faala, cependant, nappartiennent que partiellement
aux verbes diathse externe.
Si lon part au contraire du sens grammatical, on observe que :
(d) tous les verbes diathse externe sont inclus dans le schme faala ;
(e) les verbes adjectif dtat caractristique sont en contrepartie reprsents dans les trois schmes ;
(f) les verbes moyens et mdio-passifs sont rpartis entre les schmes
faila et faala (dans le schme faula, les verbes dacquisition dtat
sont mdio-passifs).
Les deux relations (a) et (d) sont de la forme tous les . . . sont . . . ; la relation (b) est du type presque tous les . . . sont . . .. Ce qui prcde peut tre
reprsent par la figure suivante :
Schme
faula
faila
faala
FIGURE 5
Sens grammatical
Verbes adjectif dtat
caractristique
(tat ou acquisition dtat)
Verbes diathse interne ou
mdio-passifs
Verbes diathse externe
LES RELATIONS ENTRE SENS ET FORME DANS
LE VERBE SIMPLE
faula, faila, faala
361
Conventions :
Les flches dont le trait est doubl (
) indiquent une relation mettant en jeu la totalit des lments de lensemble considr. La
convention se lit : pour tout schme ou sens grammatical situ au dbut
de la flche, le sens ou le schme est celui qui est indiqu par la pointe de
la flche (cf. section 4.1).
La flche interrompue de trait doubl (
) indique une
relation mettant en jeu la quasi totalit des lments de lensemble
considr, avec toutefois un ensemble dexceptions clairement identifiable. La flche il ny en a quune se lit : pour presque tout schme
situ au dbut de la flche, le sens grammatical est celui qui est indiqu
par la pointe de la flche (voir section 4.2).
Les flches en pointills (
) indiquent une relation qui ne
concerne quun nombre rduit et ferm de verbes, dans linventaire
duquel apparaissent des contraintes morphologiques, comme par
exemple dans les verbes dtat caractristique du schme faala
( 4.2.1 et 4.3.1).
La flche en trait plein (
) signale une relation entre sens
grammatical et schme qui concerne un nombre important et en tout
tat de cause ouvert de verbes. Ces verbes sont produits par un processus identifiable de glissement smantique ( 4.3.2 et 4.3.3).
6. Le dcrochage partiel entre sens et forme :

une relation non bijective
Le dcrochage partiel entre sens et forme illustr ci-dessus est d au jeu
de deux types de contraintes : formelles (incluses dans les flches transversales en pointill) et smantiques (flche transversale en trait plein).
Pour comprendre la structure gnrale laquelle on est confront, il faut
tenir compte du fait que la relation entre les schmes et leur sens grammatical nobit par une relation bijective. On rencontre en effet, soit des
relations schme sens grammatical, comme dans faula et faila, soit
une relation schme sens grammatical, comme dans faala, mais non
la relation bijective ( deux sens) schme sens grammatical, comme le
laisserait croire une conception nave du langage.
Un tel dcrochage est frquent dans les langues : on le rencontre,
typiquement, dans la relation entre les actes de langage (Searle) et leurs
362
joseph dichy
ralisations dans une langue donne. En reconnaissance, il est galement trs frquent dans les valeurs aspectuo-temporelles associes en
contexte une forme verbale : une mme forme supportant plusieurs
interprtations, cest la mise en rapport de celle-ci avec divers indices
contextuels qui permet au processus de comprhension doprer (Descls et al. 1998 ; Descls et Guentchva, paratre). Le reprage dindices
est particulirement important dans le processus de lecture en arabe,
lcriture courante tant dpourvue des signes diacritiques secondaires
dits de vocalisation.
Dans la constitution du lexique arabe, la relation entre les sens grammaticaux associs un schme donn et les sens lexicaux qui sadjoignent
un nom ou un verbe est soumise leffet du principe de figement
lexical (PFL), qui soumet le sens grammatical des schmes, ds le
niveau du mot, des carts smantiques (Dichy 2003, 204208). Dans le
verbe simple, des glissements smantiques tels que ceux qui sinstaurent
entre les valeurs de verbe dtat (tre x) et dacquisition dtat (devenir
x) ou celui qui permet certains verbes diathse externe de prendre une valeur de diathse interne, soumettent les relations entre le sens
grammatical des schmes et leur forme de fortes pressions. Le protoschma de ces relations (figure 3) persiste toutefois dans le schma des
relations observables (figure 5), mais avec des connexions entre sens et
forme qui, ntant pas bijectives, sinscrivent dans la structure gnrale
des rapports entre le sens et la forme que lon observe dans le processus
de comprhension du langage humain.
7. Rfrences bibliographiques
7.1 Sources primaires
Ibn Xlawayh, al-H usayn b. Ahmad (m. en 370/981). Laysa f kalm al-arab, d. Ahmad
A. At t r. Beyrouth : Dr al-Ilm li-l-malyn, 1979.
Ibn Ya (m. en 643/1245). arh al-Mulk f t-tasrf, d. Faxr ad-Dn Qabbwa. Alep :
Al-Maktaba l-arabiyya, 1973.
. arh al-Mufassa l. Le Caire : Maktabat al-Mutanabb / Beyrouth : lam al-kutub,
10 tomes en 2 vols.
al-Mubarrid (ou al-Mubarrad), Ab l-Abbs (m. en 285/898). Al-Muqtadab, d.
Muhammad A. Udayma. Le Caire : Al-Ahrm, 1399/1978, 4 vols.
Sbawayhi (m. v. 180/796). Al-Kitb, d. Abd as-Salm M. Hrn. Le Caire : Al-Haya
l-misriyya l-mma li-l-kitb, 1977, 5 vols.
Yawm. = Tawfq al-H akm. 1937. Yawmiyyt nib f l-aryf. Le Caire : Dr Misr li-t tiba.
az-Zajjj, Ab l-Qsim (m. v. 340/952). Al-Jumal f nahw, d. Al T. al-H amad. Beyrouth : Muassasat ar-risla, 1988 (4e d.).
az-Zamaxar, Ab l-Qsim. Al-Mufassa l f ilm al-arabiyya. Beyrouth : Dr al-Jl, s.d.
faula, faila, faala

7.2
363
Sources secondaires
Abbs, Ramzi. 2004. La conception et la ralisation dun concordancier lectronique pour

larabe. Thse de doctorat en sciences de linformation, Lyon, ENSSIB/INSA.
Abraham, Maryvonne. 1995. Reprsentations smantico-cognitives des verbes (chantillon des verbes de mouvement) : problmes et mthode. Thse dinformatique, Paris :
EHESS.
Ammar, Sam et Joseph Dichy. 1999. Les verbes arabes. Paris : Hatier (coll. Bescherelle).
Audebert, Claude. 2002. Verbes actifs et moyens dans le parler du Caire : une suite.
Annales islamologiques 36, Le Caire, I.F.A.O.
Badawi, Elsaid, Michael G. Carter, and Adrian Gully. 2004. Modern Written Arabic. A
comprehensive Grammar. London and New York : Routledge.
Belot, J. B. 1922. Cours pratique de langue arabe. Beyrouth : Imprimerie catholique.
Benveniste, mile. 1950/1966. Actif et moyen dans le verbe, in : Problmes de linguistique gnrale. Paris, Gallimard, t. 1 (1966), 1950, repris in 1966, ch. XIV, 168175.
Blachre, Rgis et Maurice Godefroy-Demombynes.1952. Grammaire de larabe classique. Paris : Maisonneuve (3e d.).
Boormans, Maurice. 1967. Grammaire darabe littral. Feuilles de travail. Rome : Institut
pontifical dtudes arabes (document mimographi).
Brockelmann, Carl. 1948. Arabische Grammatik. Leipzig : O. Harrasowitz.
Cantineau, Jean. 1950. La notion de schme et son altration dans diverses langues
smitiques. Semitica III. 7383.
Caspari, C.P. 1881. Grammaire arabe. Trad. E. Uricoecha. Paris : Maisonneuve.
Cohen, Marcel. 1929. Verbes dponents internes (ou verbes adhrents) en smitique.
Mmoires de la Socit de Linguistique de Paris XXIII : 4, 225248.
Comrie, Bernard. 1976. Aspect. An Introduction to the study of verbal aspect and related
problems. Cambridge University Press.
. 1985. Tense. Cambridge University Press.
Descls, Jean-Pierre. 1990. Langages applicatifs, langues naturelles et cognition. Paris :
Herms.
. 1994. Quelques concepts relatifs au temps et laspect pour lanalyse des textes.
Studia kognitywne 1, SOW, Varsovie. 5788.
, Valrie Flageul, Christiane Kekenbosch, Jean-Marc Meunier et Jean-Franois
Richard. 1998. Smantique cognitive de laction : 1. Contexte thorique. 2. tude
exprimentale de la catgorisation des verbes daction. Langages 132 (Cognition, catgorisation, langage). 2847 ; 4868.
et Zlatka Guentchva. paratre. Aspectualit, temporalit : une approche cognitive
et formelle partir des langues. Document mimographi.
Dichy, Joseph. 1993. Knowledge-system simulation and the computer-aided learning
of Arabic verb-form synthesis and analysis. Processing Arabic Report, 6/7. T.C.M.O.,
Universit de Nimgue. 6784 ; 9395.
. 2002/2003. Structure de la drivation lexicale en arabe : sens et forme des verbes et
des drivs nominaux les plus immdiats, Cours de prparation au CAPES darabe,
session 2003, question de linguistique. Paris : C.N.E.D.
. 2003. Sens des schmes et sens des racines en arabe : le principe de figement lexical
(PFL) et ses effets sur le lexique dune langue smitique. In Sylvianne Rmi-Giraud
et Louis Panier, dirs. La polysmie ou lempire des sens. Lexique, discours, reprsentations. Presses universitaires de Lyon. 189211 (www.concours-arabe.paris4.sorbonne.
fr/cours/dichy.doc)
et Mohamed Hassoun. 2005. The DIINAR.1 ! Arabic Lexical Resource, an
outline of contents and methodology. The ELRA Newsletter, vol. 10, 2, April-June 2005.
510.
, Abdelfattah Braham, Salem Ghazali et Mohamed Hassoun. 2002. La Base de
connaissances linguistique DIINAR.1, (DIctionnaire INformatis de lARabe
version 1). In A. Braham, d. Actes du colloque international sur Le Traitement automatique de larabe. Tunis : Universit de La Manouba. 4556.
364
joseph dichy
Fleisch, Henri. 1957. tudes sur le verbe arabe. Mlanges Louis Massignon. Institut
Franais de Damas. 153181.
. 1968. Larabe classique. Esquisse dune structure linguistique. Beyrouth : Dr alMachreq.
.1979. Trait de philologie arabe vol. 2. Beyrouth : Dr al-Machreq.
alyn, Musta f. 1912. Jmi ad-durs al-arabiyya. 19e d. revue par Muhammad A.
an-Ndir. Beyrouth, al-Maktaba l-Asriyya, 1994, 3 vols. en un.
Guillaume, Jean-Patrick (1984), Quelques aspects de la thorie morpho-phonologique
dIbn Jinn. propos des verbes glide mdian. In G. Bohas et J.-P. Guillaume, ds.
tude des thories des grammairiens arabes. I. Morphologie et phonologie. Institut franais de Damas. 338490.
Ilys, Jzf et Jirjis Nsf. Mujam ayn al-fil. Beyrouth : Dr al-Ilm li-l-malyn, 1995.
Joon, Paul. 1923. Grammaire de lhbreu biblique. Rome : Institut pontifical, rd.
1965.
. 1930. Smantique des verbes statifs de la forme qatila (qatel) en arabe, hbreu et
aramen. Mlanges de lUniversit Saint Joseph, XV : 1, 332.
Larcher Pierre. 1995. O il est montr quen arabe classique la racine na pas de sens et
quil ny a pas de sens driver delle. Arabica XLII. 291314.
. 1996. Drivation lexicale et relation au passif en arabe classique. Journal asiatique
284, 2. 265290.
. 2003. Le systme verbal de larabe classique. Aix-en-Provence : Publications de
lUniversit de Provence.
Leeman-Bouix, Danielle. 1994. Grammaire du verbe franais, des formes au sens. Paris :
Nathan.
Lyons, John. 1978/1990. Smantique linguistique, trad. fran. de la 3e d., revue et corrige de Semantics II, Cambridge University Press (1978). Paris : Larousse, 1990.
Maingueneau, Dominique. 1994. Lnonciation en linguistique franaise. Paris : Hachette.
Moscati, Sabatino, d. 1964. An Introduction to the Comparative Grammar of the Semitic
Languages. Phonology and Morphology. Wiesbaden : O. Harrassowitz (2e d., 1969).
Al-Mujam al-Wast. 1973. Ibrhm Ans, Abdalhalm Muntasir, Atiyya as-Sawlih,
Muhammad X. Ahmad, eds. Le Caire : Dr al-Marif, 2e d. (1e d. 1960).
Al-Munjid f l-lua l-arabiyya l-musira. 2000. Subhi H amw, d. Beyrouth : Dr alMariq.
Neyreneuf Michel et Ghalib Al-Hakkak. 1996. Grammaire active de larabe. Paris : Livre
de Poche.
Nr ad-Dn, Ism. 2002. Abniyat al-fil f fiyat Ibn al-H jib. Beyrouth : Dr al-Fikr
al-lubnn (1e d. 1982).
Qabbwa, Faxr ad-Dn. 1998. Tasrf al-asm wa-l-af l. Beyrouth : Maktabat al-Macrif,
3e d.
Roman, Andr. 1983. tude de la phonologie et de la morphologie de la koin arabe. Marseille : Jeanne Laffitte, 2 vols.
. 1990. Grammaire de larabe. Paris : P.U.F. (coll. Que sais-je ?).
. 1999/2005. La cration lexicale en arabe, ressources et limites de la nomination dans
une langue humaine naturelle. Presses Universitaires de Lyon, 1999. 2e d. revue et
augmente, Presses Universitaires de Lyon et Universit de Kaslik, 2005.
a-artn, Rad. 1912. Mabdi al-arabiyya. Beyrouth : Imprimerie catholique (1e d.).
Sylvestre de Sacy, Antoine. 1831. Grammaire arabe lusage des lves de lcole spciale
des langues orientales vivantes. Paris : Imprimerie royale (2e d.). Rimpr. photomcanique. Paris : Institut du Monde arabe, 2 vols. s.d.
Versteegh, Kees. 1997/2003. The Arabic Language. New York : Columbia University Press,
1997. Trad. arabe : Al-lua al-arabiyya, trxuh wa-mustawaytuh wa-tatruh,
trad. Muhammad a-arqw. Le Caire : Al-Majlis al-al li-t -taqfa, 2003.
. 2004. Meanings of speech. The category of sentential mood in Arabic grammar.
In Joseph Dichy et Hassan Hamz, ds. Le voyage et la langue, Mlanges en lhon-
faula, faila, faala
365
neur dAnouar Louca et Andr Roman. Damas : Institut Franais du Proche-Orient,

269287.
Wehr, Hans, John M. Cowan (ed). 1979. A Dictionary of Modern Written Arabic (ArabicEnglish), edited by J. M. Cowan, 4th edition. Wiesbaden : O. Harrassowtiz.
Wright, W. 189698. A Grammar of the Arabic Language. Rimpr. Beyrouth : Librairie
du Liban, 1974.
FEATURING AS A DISAMBIGUATION
TOOL IN ARABIC NLP
Everhard Ditters
University of Nijmegen
1. Introduction
Featuring is the affixation of, general as well as language dependent,
second-level labels (such as gender1 or animate) to first-level nonterminals (such as noun, verb or particle) in a formal description of
the syntax and semantics of a specific natural language. Such a formal
grammar may serve as input for the automated processing of the natural
language (NLP) concerned: Modern Standard Arabic.
About 98% of Modern Standard Arabic texts, whether in printed or
electronic form, are represented in a non-vocalized shorthand form.
The analysis of this kind of data, automated or otherwise, has to cope
with an exponential combinatorial ambiguity unless:
one considers the units of linguistic description beyond word level
such as constituent, sentence, paragraph or even text-level;
one combines this description with an adequate, coherent, consistent,
and as exhaustive as possible, featuring system for analysis purposes.
Therefore, our objective is: the design of a featuring system as a tool
for the disambiguation2 of undesired analysis results in the automated
processing of Arabic text data.3
1
Feature names and feature values are represented in the text in Italics. We use two
types of features: inherent and inherited features. The former represents the intrinsic
semantic value of the entry concerned. The latter follows as modifier the entry it modifies according to language dependent concord and agreement rules.
2
We use the term disambiguation for attempts to obtain a single and most probable
syntactically and semantically correct analysis of input data in an automated processing
environment.
3
Language phenomena such as homonymy, polysemy and even antinomy should
also be controlled by means of adequate featuring.
368
everhard ditters
A comprehensive inventory of general or universal linguistic features

is not (yet) available for Modern Standard Arabic. This is mainly due to
the complexity of the description of semantic features. Usually, linguistic
theories and descriptive language models tend to become language
independent. However, we preferwhile exploiting current, language
independent linguistic theories and descriptive modelsto describe the
features that are specific for the language under consideration.
If the linguistic model chosen for the description may have some
impact on the organization of the featuring system, it remains possible to
use heuristics and to refine the formal description any time we consider
it worthwhile in order to avoid undesired ambiguities. So we are free to
insert or to combine features such as agreement and concord,4 in order
to account for regularities, relationships and dependencies occurring
between different constituents, or between different elements within a
constituent.
Considering Modern Standard Arabic language description, we
distinguish three layers of featuring: a morphological, a syntactic and a
semantic layer. They describe the form, the function and the meaning
of elements used in a, agreed upon by convention and expandable by
language evolution, system of information interchange, tailored to the
needs of users of any natural language system. This tri-partition roughly
coincides with traditional classifications into orthography, phonology,
morphology, syntax, stylistics, discourse, and even semantics.5
A description of the Arabic phonemes for analysis purposes does not
require a differentiation in terms of their production characteristics. A
listing of phonemes (26 consonants, three glides [semi-consonants or
semi-vowels],6 and three vowels [in a short and long variety]), together
with a number of graphemic alternatives, constitutes the basis for our
phonemic description of Modern Standard Arabic.
4
We distinguish between concord and agreement and reserve the latter to describe
regular patterns in relationship between an explicit agent and its predicate in a verbal sentence (Sv) involving gender (Ditters 1992, 169, n. 13; Kihm 2006, 1415; cf. also:
Bahloul 2006, 4348). We use the term concord for feature-value sharing within the
noun phrase (NP) between the head and its modifiers, as well as in the nominal sentence
(Sn) for the marked relation between the topic and comment, involving, if applicable,
definiteness, gender, number, and case.
5
We are definitely not trying to introduce a formalized dynamic description of the
Arabic concept of the world. We rather follow a static semantic approach using finite
enumeration of pertinent (static) semantic features. See also subsection 4 below.
6
For coherency within our description, we adopt Sbawayhis inventory of 29 Arabic
consonants (Hrn 1982, 4, 431).
featuring as a disambiguation tool in arabic nlp
369
The introduction of a feature such as verbtype may well account for

alternative realizations of the finite forms of so-called weak verbs. A
feature like complement structure7 controls the number (argument-1
[arg-1], [arg-2], [arg-3] etc.) as well as the format (noun phrase (NP),
prepositional phrase (PP), complement clause (CCL) etc.) of possible
complements of a lexical verbal entry. By means of predicate-subject
and predicate-object matching features we are in a position to considerably refine our formal description and to distinguish, for example,
between ordinary and metaphorical language use. However, for the disambiguation of analysis results our main task at this moment remains: to
draft a general pattern for nominal and verbal featuring; and to compose
a basic set of (semantic) features and feature-clusters embedded in an
adequate description of verbal complement structure.
A remark about the linguistic model chosen for the description may
be useful. We still opt for a Phrase Structure Grammar (psg) model
based on immediate constituency (ic), but enlarged with a second level
of description to account for prevailing relationships and dependencies
within a Modern Standard Arabic sentence.8 We even foresee the
introduction of a third level of description, grouping together pertinent
semantic features with information about their characteristic form of
realization at syntax level. Inspired by Saads (1982) syntactic-semantic
study of the Classical Arabic verb in Case Grammar (cg) terms, we
will use a number of Fillmores (1968) cg ideas as well as later
developments for the sub-categorization and patterning of verbs in
Modern Standard Arabic.9
Finally some words about the processing environment. We use the
agfl-system,10 a parser generator for research and applications in
7
To be elaborated upon later at syntax level when speaking about: tri-, bi-, mono-,
and intransitive verbs or the reduction to a one-argument realization used to uniquely
emphasize the semantic load of the verbal entry ; and at the semantic level while speaking
about: semantic feature hierarchy; the minimal and the maximal realization of complement structure; compulsory and optional complement structure realization; case roles;
and verb sub-categorization.
8
For other applications such as information retrieval (IR) or text-summarizing
(TS), we use a Dependency Grammar (DG) approach since it appears to express more
adequately semantic properties in terms of nodes and relations between nodes (Ditters
and Koster 2004).
9
Only recently we decided to include al-Saffr (1979) in our research for his semantico-syntactic features in Case-Grammar terms.
10
For more information about the AGFL formalism, now in its 2.4th Windows version, see Koster (1991) and www.cs.ru.nl/agfl. For more information about the formal
description of Arabic, see Ditters 1992; 2003a; 2003b; and forthcoming.
370
everhard ditters
nlp. The acronym refers to Affix Grammars over Finite Lattices.11 The
parsers generated by the system are top-down recursive backup parsers
(Koster 1974, Meijer 1986), based on non-deterministic concepts and
the unification-principle (Nederhoff 1993). The agfl-formalism is
part of the family of two-level grammars: a context-free grammar is
augmented with set-valued features for expressing agreement between
syntactic categories. The formalism is, in principle, suited for describing
morphological and syntactic structure, as well as finite semantics in
terms of animate, concrete, human, volition and many others.
We discuss feature-sets at morphologic ( 2), syntactic ( 3), and
semantic level ( 4). We end with a conclusion ( 5) and a list of
references used ( 6).12
2. Morphology
Strictly speaking, morphology relates to the formal description of the
individual Arabic parts of speech: verb, noun and particle, and, if needed
or useful, of their differentiation into a further sub-categorization.13 In
a broader sense, morphology also comprises feature names and values
such as: aspect, case, definiteness, derivation, gender, number, person,
tense, voice, and many other features used at syntax level. Morphology,
finally, has to account for punctuation marks and other textual noise,
normally more conveniently stored in the lexicon module.
Literary Arabic has been described as a predominantly root and
pattern language type.14 Moreover, vowel-pattern variation combined
11
Here, the term affix, a variable with a finite set of values, has to be taken in its formal and not in its linguistic sense.
12
Within the morphological, syntactic and semantic sections, we use the traditional
Arabic language parts-of-speech (POS) differentiation into noun, verb and particle as
headings for the subsections. In these subsections we only discuss the, for us, relevant
features.
13
Cf. Sbawayhis tri-partition (Hrn 1982, 1:12) and the subdivision of POS into 7
classes by alSq (1977, 214).
14
Cohen (1970, 49 ff.) has been one of the first to describe, for automated Arabic language processing, a frame of mostly three, sometimes four and rarely five, consonants or
semi-consonants filled with a combination of vowels (including the absence of a vowel
at a certain slot) expressing semantic differentiations to the global meaning of the consonantal root combination. Elements of a small subset of the phoneme inventory are used
to produce other derivates of the base frame, whether of the category verb, noun or
adjective, with their own specific variation on the global meaning of the consonantal
root combination.
371
with the insertion of a finite set of auxiliary phonemes triggers a more

or less regular, more or less predictable change in the semantic load of
the base root meaning.
The Arabic phoneme inventory has extensively been described elsewhere.15 As far as the consonants are concerned, the description comprises a cluster of features referring to their way and place of production,
as well as to their marking for velarization and degree of voicing. As far
as the vowels are concerned, they have been described by their place and
the openness of that place, set out against the position of the tongue in
their production process.
For analysis purposes, we have the Arabic grapheme inventory with
variant scriptures for alternatives occurring in initial, median, end or
independent position. In order to account for fully, partially or nonvocalized Arabic input data, the listing of the Arabic vowel graphemes
will include an empty vowel realization.
2.1
Verbs
Assigning morphological features and feature values to verbs, we

account for: the realization of finite verb occurrences on the basis of
number and place of the root consonants (radical, r1,16 r2, r3, and less
frequent: r4 and r5); the realization of the vowel (a, i, u) of the second
consonant (C2) of the base (Stem I) verb realization in perfect and mood
aspect (vowperf, vowimperf ); and the realization of prefixed or infixed
vowels (vow or sukn).
Morphological features, attached to elements of a finite verb realization
at syntactic level (see also 3.1 below), affect values concerning: aspect,
tense,17 voice, person, gender and number. For the description of a finite
15
Cf. Fleisch mainly used three, sometimes four, features to define the elements of
the consonantal system, which ends up in a 28 u 4 matrix (1961, 1:5665) or a 16 u 5
matrix (1968, 19); Versteegh (2001, 20) used a 9 u 6 matrix; Saad (1982, 6) used a 7 u
9 matrix for the consonants and a 4 u 2 matrix for the vowels. See also the phoneme
featuring of Bohas and Saguer (2007, 255 ff.) in this volume.
16
A listing of the Arabic phonemic (consonant and vowel) system is given in the lexical module(s). For analysis purposes, a single description of the r1 suffices. For generation purposes, features describing occurrence incompatibilities between an r1, r2 and r3
(r4, r5) should be provided for.
17
As Badawi et al. (2004, 362 ff.) do, we distinguish between a perfect and a nonperfect (mood) value of the feature aspect of the verb. In combination with modal and
temporal verbs and/or adverbials a complete range of temporal and aspectual differentiations (tenses) can be described (see on aspect and tense also Eisele 2006, 195201;
Bahloul 2006, 506; and Reese 2006, 5053).
372
everhard ditters
verb form of the so-called weak verbs with alternative realizations in its
conjugation, a variable verbtype accounts for alternative verb realizations.
To account for the differentiation in base and derived stems we use the
feature derivation. With that we can list, in the form of a context-free
metagrammar, the following feature names or non-terminal affixes
(in capital letters) with their finite-set of values or terminal affixes (in
lower case), connected to a finite verb form:18
ASPECT
MOOD
TENSE
VOICE
PERSON
GENDER
NUMBER
VERBTYPE
::
::
::
::
::
::
::
::
DERIVATION
DERIVED
RADICALS
R1
R2
R3
R4
R5
VOW
VOWPAT
VOWIMPERF
VOWPERF
::
::
::
::
::
::
::
::
::
::
::
::
perfect; MOOD.
indicative; subjunctive; jussive; imperative.19
temporal; modal.
active; passive.
first; second; third.
feminine; masculine.
singular; dual; plural.
r1r2r3; r2=r3; r1=w; r1=y; r2=a; r2=w;
r2=y; r3=w; r3=y; r1=w, r3=y; r2=w, r3=y;
r1=hamza; r2=hamza; r3=hamza.
base; DERIVED.
ii ; iii ; iv ; v; vi; vii; viii; ix; x.
three; four; five.
alphabet.
R1.
R1.
R1.
R1.
vowel; sukn.
VOW.
VOW.
VOW.
18
We will use the following conventions for the formal description of features and
the finite-set of values at the second level of description which closely follow the AGFL
convention:
feature names are written in upper case;
feature values are written in lower case;
the rewrite symbol is a double colon ::;
a single left-hand entry is rewritten in one ore more feature names
and/or feature values at the right-hand side;
alternative realizations at right-hand side are separated by a semicolon;
options in the right-hand side are separated by a vertical bar;
a rule will be closed by a period ..
19
The archaic energetic-1 and energetic-2 finite verb forms are not accounted for in
our formal description of Modern Standard Arabic.

2.2
373
Nouns
Features, at the morphological level assigned to elements of the category

noun, concern variables such as: person, gender and number.20 For the
description of nominal plural building, it is worthwhile to differentiate
between an external and internal formation. Besides person, gender and
number, the feature case is a pertinent part of the declension system,
in Modern Standard Arabic closely connected with the type of
declension.
For the description of a differentiation in base and derived stems for
deverbal nouns (active and passive participles, infinitives, nouns of
time, place, doing an action once, or referring to instruments used in an
occurrence), we use the feature derivation. By means of morphological
base rules, we describe the subset denominatives of the category noun,
such as: individuality nouns, multitude nouns, vessel nouns, relation
nouns, abstract quality nouns and diminutives.21
The feature variable determination with values as definite, indefinite
or morphologically neutral is another inherent or inherited (at syntax
level) characteristic of elements of the category noun. Elements of
the sub-classes demonstratives, personal pronouns, proper nouns, and
relative pronouns are inherently definite. Elements of the sub-classes
indefinite and interrogative pronouns definitely possess an inherent
indefinite value.
As far as the definite article is concerned, we follow Wrights interpretation of the Arabic grammatical tradition in describing this article
on the basis of its deictic and not of its generic value as a sub-class of the
demonstratives.22 Furthermore, elements of the sub-classes adjectives,
common nouns, elatives, numerals, and quantifiers may, or may not,
receive a definite or an indefinite value depending on their function
and occurrence in phrases at syntax level. So we list the following
morphological feature rules for a noun:
CASE
DECLENSION
:: nominative; NOMINATIVE.23
:: invariable; diptote; triptote.
20
Since identical looking feature values are attached to distinct non-terminal names
(different category labels), we can reuse these feature names without any risk for undesired ambiguities.
21
Cf. Wright 1974, 1:109177.
22
Cf. Wright 1974, 1:264270. See also Fleisch 1961, 1:339347.
23
The AGFL formalism allows for the use of logical markers such as +, -, and others
in combination with feature names and values.
374
everhard ditters
DENOMINATIVE
DERIVATION
DERIVED
DETERMINATION
DEVERBAL
GENDER
NOMINATIVE
NUMBER
PERSON
PLURAL
2.3
:: individuality; multitude; vessel; relation; quality;

diminutive.
:: base; DERIVED.
:: ii; iii; iv; v; vi; vii; viii; ix; x.
:: definite; indefinite.
:: actpart; paspart; infinitive; time;
place; once; instrument.
:: feminine; masculine.
:: genitive; accusative.
:: singular; dual; PLURAL.
:: first; second; third.
:: external; internal.
Particles
According to the traditional description24 of particles we may distinguish

between: adverbs, conjunctions, interjections, and prepositions.25 A formal
description should include:
PARTICLE TYPE
2.3.1
:: ADVERB; CONJUNCTION;
INTERJECTION ; PREPOSITION.
Adverbs
A for us still useful sub-categorization of adverbs, at morphological level,

is that in invariable, as far as case marking is concerned, and bound and
free forms, as far as the orthographic representation is concerned.
ADVERB
2.3.2
:: bound; free; invariable.
Conjunctions
As far as the coordinating particles are concerned, a distinction should

be made in cumulative and selective particles because of the importance
of number value in concord and agreement phenomena at syntax level.
We then divide the selective particles into: alternative, consecutive,
explicative, exclusive, inclusive, restrictive, and successive elements.26 This
sub-categorization will be extensively used at the next higher level of
description.
24
Cf. Wright, 1974, 1:282296.

CF. Badawi et al. 2004, 174219; Cantarino, 19745, 2:253 ff.; El-Ayoubi et al. 2003,
1:2, 275460.
26
Cf. Ditters 1992, 222228.
25
375
For the conditional particles we prepare their application at syntax

level with a differentiation as to the nature of the condition in: possible,
real and unreal.
CONJUNCTION
COORDINATING
SELECTIVE
CONDITIONAL
2.3.3
:: COORDINATING; CONDITIONAL.
:: cumulative; SELECTIVE.
:: alternative; consecutive; exclusive; explicative;
inclusive; restrictive; successive.
:: possible; real; unreal.
Interjections
Elements of this sub-section of particles, rather close to discourse and

formalized language use, can, as far the morphological level is concerned,
be divided into vocatives and exclamations.27 However, we also have to
decide whether or not proverbs and frozen expressions should be
lodged among the interjections and labeled as adverbs or adverbials at
syntax level. In any case we also like to mention here the category of
formulaic greetings.28
INTERJECTION
2.3.4
:: vocative; exclamation; f-expressions; proverbs;

greetings.
Prepositions
The standard differentiation of prepositions at this level is into a primary

and a secondary group. The secondary group consists of an open set of
noun-derived entries, marked for their function by means of a definite
accusative case value and, at syntax level, engaged in a construct state-like
link with the prepositional complement, itself marked with genitive case
value (NPgen). The primary group comprises a finite-set of non-derived
entries, some of which are, in their orthographic representation, directly
bound to the prepositional complement while others are unbound.29
PREPOSITION
PRIMARY
:: PRIMARY; secondary.
:: bound; unbound.
27
A still poorly described domain of frozen or set expressions like: greetings, insults,
proverbs and similar insertions should be included here. Cf. Bergman 2007, 136137.
28
Cf. Elzeiny 2007, 202207.
29
Cf. El-Ayoubi et al., 2003, I:2:574592.
376
everhard ditters
3. Syntax
At syntax level, elements of the earlier mentioned word categories

and sub-classes are assembled into constituents, labeled: verb phrase
(vp), noun phrase (np), and particle phrase, sub-categorized according
to the element realizing the compulsory head function.30 These phrases
realize specific compulsory and optional functions within the two main
units of linguistic description at sentence level: the nominal (sn) and the
verbal (sv) sentence.31 It may be clear, that the combinatorial behavior
of elements within a constituent, and that of constituents at sentence
level, comprising mutual relationships and dependencies, is controlled
by features and feature-condition rules.
In Modern Standard Arabic text data one distinguishes within the
nominal sentence (sn) the compulsory topic and comment functions
(Figure 1a). In a verbal sentence (sv) we are dealing with an compulsory
predicate function and several optional complements (Figure 1b).
Optional sentence adverbials can occur in both sentence types.
A np with a nominative case value for the head (npnom) always32
realizes the topic function.33 The comment function is realized by: a np34
with nominative case value for the head (npnom); an adjp with nominative
case value for the head (adjpnom); a vp; a pp; or an adverb phrase (advp).
A vp always fills in a sv the predicate slot. Sentence adverbials (sadv) are
realized by: a np with accusative case value of the head (npacc); a pp; an
advp; a cp; or a ccl.35
30
As mentioned in 1, a particle phrase could be subdivided into a prepositional
phrase (PP), an adverbial phrase (ADVP), an interjectional phrase (IP), and (to remain
consistent) a conjunctive phrase (CP), or a complement clause (CCL).
31
Distinct concord and agreement phenomena mark a productive syntactic (and
semantic) distinction between a nominal and a verbal sentence type in MSA. Discourse
sensitive emphasis on the topic agent should be maintained, side by side with the possibility of emphasizing the action performance of, sometimes, the same agent in typical
VSO-oriented approaches. Cf. Ditters 2001, 3137.
32
The optional occurrence of a particle, like inna with an emphasizing semantic load
and governing its complement by an accusative case value, only represents an alternative
realization within the base structure.
33
An alternative topic realization is a CCL (complement clause), introduced by the
emphatic particle inna, governing the head of the following NP in the accusative case.
34
A sub-class of the nouns is constituted by different subsets of adjectives. We should
add the adjective phrase (ADJP) with a nominalized adjective in head position as possible alternative for a head or modifier function in the sentence.
35
For detailed structural descriptions of phrases in MSA, see Ditters 1992, chapters
III and IV, and for a formal description of sentence structures see Ditters forthcoming.
377
In Figure 1a are listed the function slots within the sn, as well as the
categories able to fill these slots. One may easily see how many different
realizations of a nominal sentence are accounted for in this diagram
where alternatives are represented within square brackets and optional
elements within parentheses.
FIGURE 1A
SLOTS AND FILLERS WITHIN THE SN

Sn
Slots
TOPIC
COMMENT
(Sadv)36
Fillers
[NPnom]37
[NPacc]38
[ADJPnom]
[ADVP]
[CCL]
[NPnom]
[PCCL]
[PP]
[VP]
[ADVP]
[CCL]39
[NPacc]
[PP]
Figure
2
In Figure 1b are listed the function slots within the sv, as well as the
categories able to fill these slots. One may easily see how many different
realizations of a verbal sentence are accounted for in this diagram
By means of variables as predicate-subject match (psmatch) and topiccomment match (tcmatch) we describe regular patterns in concord and
agreement between the elements involved in a sv or sn, checking at the
same time the matching of a verb-argument-1 relation as well as the
compatibility of the elements involved in a topic-comment occurrence. In
the same way we control the occurrence of other argument realizations
of a verbal entry by means of a predicate-object match (pomatch).
36
The AGFL-formalism allows for free sequence variation at any level of description (this
means for us: at function and category level) of entries within the formal description and
processing. Therefore we do not need to list all mathematically possible realizations.
37
An alternative realization in the form of a CCL filler in a topic slot, as recorded from
Classical Arabic data (Qurn 2:184): an tasm (topic) xayrun la-kum (comment),
is, for its poor frequency in modern text data, not accounted for in our formal description of the Sn.
38
We include this alternative realization to account for an absolute negator-head
combination in topic position (see below 3.2).
39
In our formal description of Modern Standard Arabic we also include under the
variable name CCL the protasis (the condition posed) in conditional and hypothetical
sentences.
378
everhard ditters
FIGURE 1B
SLOTS AND FILLERS WITHIN THE SV

Sv
Slots
PREDICATE
(Sadv)
Fillers
VP
[ADVP]
[CCL]
[NPacc]
[PP]
Figure 3
3.1
Verb Phrase
At an early stage in the Arabic grammatical tradition, regular patterns

in syntactic behavior and the meaning of the elements involved in
verbal constructions brought grammarians to a general description
of the verbal complement structure40 and the grouping together of
verbs.41 Briefly summarizing this tradition we can say, that a finite-verb
realization (marked for perfect aspect, indicative, imperative mood or
their alternatives)42 functions as the compulsory head of a vp. In a
minimum vp configuration, an implicit agent provides values for gender,
number and person of the verbal head. Particles may precede the head
realizing optional pre-modifying (prem) functions such as a modal,
aspectual or temporal modifier, or carrying a negation value (neg). In
a maximum vp configuration, an explicit agent, objects (compl) as
well as other optional adverbial (adv pom) or peripheral (phr pom)
modifiers may follow the head. The general structure of a vp is presented
in Figure 2, where alternatives are included in square brackets and
optional elements in parentheses:
40
Within the grammatical tradition grammarians spoke about: mutaaddin (transitive) and ayr mutaaddin (intransitive), further differentiated into and complemented
with: maf l bi-hi (direct object), maf l mutlaq (absolute object), maf l f-hi (object of
time or place), maf l la-hu or li-ajli-hi (object of cause or reason), maf l maa-hu (concomitative object), hl (circumstantial object), tamyz (object of specification). Cf. Sbawayhis al-Kitb. Hrn, 1:3454, 297310, 367384; see also Fleisch 1968, 177185.
41
We mention briefly: incomplete verbs (al-af l an-nqisa), the verbs of hope (af l
ar-ra), the verbs of beginning (af l a-ur), the verbs of the heart (af l al-qulb),
the verbs of praise and blame (af l al-madh wa-d -d amm), the verbs of approximation (af l al-muqraba), and the verbs of esteem (af l at-tafdl). Cf. a.o.: Ayoub 1980;
Cuvalay 1994, 1996; see also Wright, 2:4752.
42
The negation of the perfect aspect by means of the particle lam followed by a finite
verb form in the jussive or the negation particle lan governing a finite verb form in the
subjunctive.

FIGURE 2
379
GENERAL STRUCTURE OF THE VP43

VP
Slots
(PREM)
(NEG)
Fillers
[ASP] [NEG PART]

[MOD]
[TMP]
HEAD
(COMPL)
VERB
[ADJPacc]
[CCL]
[NPnom]
[NPacc]
[PP]
[VP]
(ADV POM) (PHR POM)

ADVP
[NPacc]
[PP]
[ADJPacc]
[NPacc]
[PCCL]
[PP]
[VP]
An active or passive voice value of the verbal head, in combination with

the variable transitivity44 (with values as intransitive or transitive), affects
the realization of a verbal complement structure, not only in the number
of compulsory or optional arguments, but also in the form of these
arguments as fixed in the lexically given DNA-structure of the verbal
entry concerned. With the term form we here refer to the argument
realization by means of a np with nominative case (npnom) for an
explicit agent, a np with accusative case (npacc) for direct object(s) or its
equivalent,45 and any subject or object attribute,46 a complement clause
(ccl), a co-referential vp for verbal extension,47 a pp as prepositional
object or its equivalent.48
Each finite verb occurrence in MSA, intransitive and transitive alike,
has in its complement structure a slot for an implicit agent or argument1 realization in the form of a suffixed pronoun (in past tense) or a
discontinuous pre- and suffixed pronoun combination (in present tense).
43
Ditters 1992, 286 ff.

We prefer the term transitivity, with the values: intransitive, transitive, bi-transitive
and tri-transitive to the term valency or valence since the latter is less apt to determine
non-optional core meanings.
45
A complement clause (CCL) introduced by the particle an or anna.
46
In de form of an NPacc or an indefinite adjective phrase marked for accusative case
value (ADJPindef,acc).
47
With the variable name verbal-extension we describe, in our formal grammar, the
occurrence of co-referential verb clusters, sometimes called auxiliaries, such as: incomplete verbs (kna or laysa, to be(come) or not to be); inceptive verbs (badaa, to begin
and the like); continuity verbs (istamarra, to continue and the like); anticipation verbs
(kda, to be on the point and the like). Cf. Haak 2006, Badawi et al. 2004, 422 ff.
48
A prepositional complement clause (PCCL) introduced by a preposition followed
by the particle an or anna.
44
380
everhard ditters
Combined with an explicit agent in the form of a npnom, the finite verb
realization immobilizes into a 3rd person singular realization liable to
vary in gender value.
For monotransitive verbs with a direct object in the form of a npacc, in
arg-2 position, this argument will assume the arg-1 position in case of
a passive voice value of the verbal head. The same holds for bitransitive
or tritransitive verbs: the first npacc direct object will assume the arg-1
position in case of passive voice realization.
We may summarize the verbal complement structure in Modern
Standard Arabic in two tables. The first matches the features form,
function, and transitivity against the number of possible completive arguments. Together with the semantic load of the verbal entry concerned,
we have the actors for the predicate-subject (psmatch) and the predicate
object matching (pomatch).
TABLE 1
VERBAL COMPLEMENT STRUCTURE (COMPLETIVE)

arg-1
arg-2
function
agent
v-extension
transitivity intrans/trans intrans/trans
form
[NPnom]
[VPi]
arg-3
arg-4
arg-5
object-1
object- 2
monotrans
bitrans
[NPacc]
[ADJPindef,acc]
[PPprep]
[NPacc]
[CCLan/anna/inna]
[PPprep]
[PCCLprep, an/anna]
arg-6
object-3 indirect object

tritrans
intrans/trans
[NPacc]
[PPprep]
[ADJPindef,acc]
The following table (Table 2) presents, in a general way, slots and

possible filler for (mainly circumstantial) elements occurring within
the complement structure of a lexically given verbal entry. As far as the
featuring is concerned, it is gradually being introduced in the form of
subscripts to the phrasal heads at syntax level.
TABLE 2
VERBAL COMPLEMENT STRUCTURE (CIRCUMSTANTIAL)

arg-7
arg-8
arg-9
arg-10
arg-11
function situation absolute source

goal
time
form
[NPacc]
[NPacc] [PPmin/an] [PPprep] [NPacc,time]
[CCLwa]
[PPprep,time]
arg-12
arg-13
arg-14
place
instrument result
[NPacc,place] [PPprep]
[NPacc]
[PPprep,place]
[PPprep]
For the description of predicate-subject and predicate-object relations

in MSA (psmatch and pomatch), we have so far exploited the Arabic
grammatical tradition. Coming to Modern Standard Arabic, we like to
381
adopt, in the next section ( 3.1), Saads work (1982, 2026) in a Case
Grammar perspective for Classical Arabic.
Some of the features linked with a lexical verbal entry are listed
below:
AGREEMENT
[PSMATCH
[POMATCH
TRANSITIVITY
MONOTRANS
BITRANS
TRITRANS
3.2
::
::
::
::
+human; human.
]49
]
intransitive; MONOTRANS; BITRANS;
TRITRANS.
:: acc; prep.
:: acc2; accprep; prep2.
:: acc3; acc2prep; accprep2; prep3.
Noun Phrase
We distinguish within the np one compulsory and several optional

functions. Only a noun (or a nominalized adjective) can realize the
compulsory head function. Among the optional functions we mention:
negator (neg), a mutually exclusive predeterminer (predet) and postdeterminer (postdet), postmodifier (pom), and, for some head realizations, a complement (compl).50 In the following general structure
of the np (Figure 3), optional functions are included in parentheses and
mutually exclusive functions in square brackets. 51
Relationships and dependencies within the np are specified and
controlled by features and feature-condition rules. In a neg-head
combination it is the negative particle l that imposes the unique
occurrence of an indefinite common noun in head-position provided
with a definite version of the accusative case value. Personal pronouns and
proper nouns, interrogative and indefinite pronouns (of the subset-type:
man and m) in head-position, are inherently marked as definite and
indefinite respectively, thus they may only occur with an optional pom.
49
These rules will be discussed later.

For example: a masdar in head function with a subjective complement in the slot
for postdeterminer (NPgen) and an objective complement in the slot for COMPL (NPacc),
both realizations in the form of a suffixed pronoun thus realizing a postdet and a
compl function: hubb--h = love-mine-her my love for her.
51
In Figure 3 we have included neg among the reciprocally exclusive function realizations. However, here we underline the incompatibility of a predet and a postdet
in a np. Cfr. the examples in Cantarino 1974, I:114; 1975, II:220222), Badawi et al.,
2004, 464466. El-Ayoubi et al. (20012003, I:12) do not mention a neg; neither in the
Vorfeld der Nominalgruppe, nor in the Nachfeld der Nominalgruppe.
50
382
everhard ditters
FIGURE 3
GENERAL STRUCTURE OF THE NP52

NP
Slots
([NEG])
Fillers [NEG PARTl]
([PREDET]) HEAD
[ART]
NOUN
([POSTDET])
[NPgen]
(POM) (COMPL)
[NP]
[ADJP]
[PP]
[REL CL]
[PP]
[NPacc]
Only (de)verbal nouns (those marked with acc, or acc2 for transitivity) as
well as some nominalized adjectives in head-position may occur with a
compl.53 Other relationships and dependencies concerning definiteness,
person, gender, number, and case, will be discussed below.
3.2.1
Definiteness
Elements of the noun subsets proper nouns and personal pronouns have
the value definite for definiteness with consequences for post-modifying
elements at constituent and sentence level. Elements of the noun-subsets
indefinite pronouns and interrogative pronouns have the value indefinite.
Common nouns, (de)verbal nouns, and nominalized adjectives receive
their value for definiteness by the occurrence of a predet or a postdet
modifier.
3.2.2
Person
Only elements of the noun subset personal pronouns vary in their value
for person ranging from first, second to third, while varying in number
and case feature values. All elements of other noun subsets bear the fixed
value third for the feature person.
3.2.3
Gender
Occurrences of the subsets proper nouns, personal pronouns, and common

nouns possess a lexically given value for the feature gender, whether this
52

For a more detailed discussion about adjectives in the context of the ellipsis of the
head of an NP and of nominalized adjectives see Ditters (forthcoming).
53
383
is marked morphologically by a default null-value, by one of the feminine markers or not. Elements of the subsets indefinite and interrogative
pronouns bear the default gender value masculine, but may, in retrospective and supported by the unification-principle of the agflformalism, receive an inherited feminine value provided by the context
in which they occur beyond sentence level. At syntax level we may have
to deal with the interference of semantic feature values such as animate,
concrete, human, volition etc., here mainly related to as: concord and
agreement phenomena.
3.2.4
Number
The singular and plural realization of elements of the subsets proper

nouns, personal pronouns, and common nouns are given lexically. A
dual number value realization is normally the result of the application
of a grammar-rule except for the personal pronouns. Only infinitives or
masdars always have a singular number value. Apparent evidence to the
contrary is caused by the exploitation of masdars for the creation of an
abstract lexicon.
3.2.5
Case
Case is a syntax dependent application feature. Its values (nom, gen, and
acc) are related to the function the element concerned has at phrasal or
higher levels of linguistic description. A null-value for case, any other
ambiguity concerning diptotic np-head occurrences, and gender issues
for a first person realization have to be disambiguated beyond phrasal
and sentence levels of description.
3.2.6
Trash
Earlier we mentioned feature-clusters. In some of them nps are involved

(concord, agreement, ps-match, po-match, and tc-match). In this section
we discuss the occurrence of a noun in the head-position of a phrase at
syntax level. Feature values of such a noun should be capable to match
with elements occurring within the same constituent or with feature
values of heads of other constituents within a given context.
Summarizing some conclusions at this point, we list the following
formal rules connected with a lexical noun entry at syntax level:
384
everhard ditters
[AGREEMENT
[CONCORD
[PSMATCH
[POMATCH
[TCMATCH
CASE
DEFINITENESS
HEADREALIZATION
3.3
::
::
::
::
::
::
::
::
+human; human].54
full; partial.]
]
]
]
accusative; genitive; nominative; null.
definite; indefinite.
demonstrative; personal; proper; relative;
indefinite; interrogative; adjective; common;
elative; numeral; quantifier; verbal.
Particle Phrase
As far as the occurrence of particles at syntax level is concerned: earlier we

distinguished between an adverb phrase (advp), a prepositional phrase
(pp), an interjectional phrase (ip), and a conjunctive phrase (cp).
3.3.1
Adverb Phrase
At the morphological level, we sub-categorized the adverbs into: invariable, as far as case marking is concerned, and bound or free forms, as far as
the orthographic representation is concerned. One feature of the bound
forms involves the temporal aspect of a verb realization itself marked by
a non-perfect or mood tense (sa, directly linked to a finite verb form,
then suggests a future temporal aspect).55 The interrogative a affects
the whole of a succeeding verbal or nominal sentence. The affirmative
la mainly occurs, in Modern Standard Arabic, as a premodifier to the
particle qad in an advp. At syntax level, the bound and free forms are
further differentiated into, among others, time, degree and manner
adverbials, attached to verbs and adjectives as well. Finally, we have to
distinguish between adverbs as head of an advp, on the one hand, and
other realizations of an adverbial function at phrasal or sentence level
such as pp, a npacc or a pccl at the other.
Formally, an adverb phrase (advp) can be described as a, at syntax
level occurring, constituent that, in its minimal configuration, has an
adverb in the compulsory head function. Optional functions in an advp
can be labeled as a pre- or postmodifier of the head. A premodifier may,
for example, figure as a discontinuous negative particle to adverbials as
54
These rules will be discussed later.

The so-called TMA (tense, mood and aspect) auxiliary particles have been discussed
by Kinberg (2001). See also Eisele 2006.
55
385
qattu never or faqat no more,56 or in the form of a preposition with

an head as in min qablu previously. An optional postmodifier can be
realized by a PP.57
FIGURE 4
GENERAL STRUCTURE OF THE ADVP 58

ADVP
Slots
(PREM)
HEAD
(POM)
Fillers
[PART]
[PREP]
[ADV]
[NPacc]
[PART]
For the adverb we suggest the following combination of feature names

with corresponding values:59
ADVERB
ARGUMENTATION
CONCLUSIVE
CONFIRMATION
DISCOURSE
FOCUS
FREQUENCE
LOCAL
MEASURE
MODALITY
TEMPORAL
SEQUENCE
SUFFIXED
3.3.2
:: ARGUMENTATION; CONFIRMATION;
DISCOURSE ; FOCUS; FREQUENCE;
interrogative; LOCAL; MEASURE;
MODALITY; SEQUENCE; TEMPORAL.
:: evaluative; comparative; CONCLUSIVE.
:: positive; negative.
:: certain; affirmative; limitative.
:: greetings; wishes; emphasis.
:: specifying; inclusive; exclusive; cumulative.
:: number; repetition; iteration; coinciding.
:: position; dimension; direction.
:: degree; quantity.
:: expression; noun; adjective; nisbe.
:: past; neutral; future; SUFFIXED; relative.
:: coinciding; preceding; succeeding.
:: idin; dka; pronoun.
Prepositional Phrase
Within a prepositional phrase we distinguish two compulsory functions: a prepositional complement (pcompl) and a linker (plink) that
connects that complement to the next higher level of linguistic description. The complement function can be realized by: an adverb phrase
56
Cf. El-Ayoubi et al. 2003, I:2, 275460, and 406. Badawi et al. 2004, 161174.
In the discussion about the NP we already mentioned the use of an accusative NP
(NPacc) for adverbial purposes.
58
59
Cf. El-Ayoubi et al. 2004, I:2, 284 ff., and Wright 1974, I:282 ff.
57
386
everhard ditters
(advp); a complement clause (ccl) introduced by the particle an or

anna (cclan|anna); a noun phrase with genitive case value (npgen); or
a verb phrase with a head marked for subjunctive mood (vpsubj). The
prepositional linker function is usually realized by an element of the
particle-subset of prepositions. However, there are also expanded constructions in which a preposition, followed by an infinitive or abstract
noun, realizes the header function,60 and prepositional clusters in which
an infinitive, marked with an indefinite accusative case value (npacc) and
followed by a preposition realizes the linker function.61
FIGURE 5
GENERAL STRUCTURE OF THE PP62

PP
Slots
HEADER
Fillers
[PREP]
[EXPAND-P]
[P-CLUSTER]
COMPL
[ADVP]
[CCLan|anna]
[NPgen]
[VPsubj]63
A good basis for a discussion about semantic features in the next section
certainly is the sub-categorization of prepositions into: time, place, and
ideal (or manner) features (Wright, 1974: II:129). With corpus based
evidence, forwarded by Cantarino (1974, II: 262 ff.) and El-Ayoubi
et al. (2004, I, 2:466 ff.) we are able to further differentiate this subcategorization of prepositions into features values:64
PREP
IDEAL
COINCIDING
HYPOTHETIC
PLACE
TIME
60
:: IDEAL; PLACE; TIME.

:: adversative; benefactive; causal; COINCIDING;
comparative; content; discourse; explicative; final;
HYPOTHETIC; instrumental; measure; modal;
partitive; possessive; specification; substitution.
:: comitative; simultaneous.
:: concessive; conditional.
:: destination; direction; local; partitive; position.
:: direction; partitive; temporal.
Cf. El-Ayoubi et al. 2004, I:2,574.

Cf. El-Ayoubi et al. 2004, I:2,584 ff.
62
63
Not to confuse with the inflectional phrase as in the Government and Binding
approach (cf. Fassi Fehri 1993, 16 ff.).
64
See also Badawi et al. 2004, 167 ff.
61
387
3.3.3 Interjectional phrase or clauses

Earlier ( 2.3.3) we differentiated the interjectional phrase (ip) into:
vocatives; exclamations;65 frozen expressions, proverbs and greetings. Two
of them, vocatives and exclamations, lend themselves easily for a general
structure description (see Figure 6).
Other optional ips ( f-expressions, proverbs, greetings, and insults) may
provisionally receive, in our formal description of Modern Standard
Arabic, the label of a phrasal or sentence adverbial since the syntactic
structure of such an insertion can easily be described by means of the
basic general syntactic patterns we already suggested. However, the
semantic value of their occurrence in a given context has still to be
evaluated and described.66
FIGURE 6
GENERAL STRUCTURE OF THE IP

IP
Slots
(PREM)
I-HEAD
(I-COMPL)
(TAIL)
Fillers
[PART]
[INTERJ]
[NPacc]
[Sv]
[NPnom]
[NPacc]
[PPil|bi|al]
[PPli]
[Sadv]
[PART]
A few comments may make this scheme more transparent. The line
of slots tells us about the compulsory function of an ip head (i-head)
with an optional tail in case of a discontinuous realization like w-h,
be it for Modern Standard Arabic a bit archaic, as in wzaydh Woe
upon Zayd! (Badawi et al., 2004:37). It further mentions the optional
occurrence of a premodifier (prem) or a complement (i-compl) of
the i-head. In the filler-section alternative realizations are listed, also
comprising a possible sv-sadv combination as man yaqtul yuqtal Who
kills will die killed. For this reason we included the term clause in the
heading of this subsection.
65
66
Cf. Firanescu 2007, 7981.

Cf. Bergman 2007, II:136137.
388
3.3.4
everhard ditters
Conjunctive phrase (cp) or Clause (ccl)
Following the Wright tradition, at the morphological level, we made a

distinction between a separable and an inseparable occurrence of conjunctions, as well as between connective and conditional conjunctive particles (Wright, 1974, I:290294). At syntax level, however, we also prefer
to distinguish between coordinating and subordinating conjunctions.
3.3.4.1 Coordination
The compulsory head-function in the cp is realized by an element of
the subset coordinators of the conjunctions. An example of an optional
premodifier would be the narrative connector wa preceeding the
adversative coordinator lkin but (Badawi et al., 2004:555). The compulsory complement function may be realized by a variety of alternatives (Figure 7).
FIGURE 7
GENERAL STRUCTURE OF THE CP

CP
Slots
(PREM)
C-HEAD67
COMPL
Fillers
[PART]
CONJ
[NP]
[VP]
[ADJP]
[ADVP]
[Sn]
[Sv]
3.3.4.2 Subordination
The compulsory head function in a complement clause is realized by
an element of the subset subordinators, which may contain a null value,
for example, in caseof the occurrence of a purposive li followed by a
verb with a subjunctive mood value. In our formal grammar of Modern
Standard Arabic, the compulsory complement function may vary
between a sn and a sv.
67
The conjunctive head may be empty is case of asyndetic coordination (Badawi
et al. 2004, 539 ff.

FIGURE 8
389
GENERAL STRUCTURE OF THE CCL

CCL
Slots
(PREM)
HEAD
COMPL
Fillers
[PREP]68
[PART]
SUB
[Sn]
[Sv]
The variables and features we like to retain for the conjunctive phrase or
clause are:
CONJUNCTION
COORDINATING
SELECTIVE
SUBORDINATING
SINGLE-CLAUSE
DOUBLE-CLAUSE
CONDITIONAL
:: COORDINATING ; SUBORDINATING.
:: cumulative; SELECTIVE.
:: adversative; alternative; consecutive;
disjunctive; exclusive; explicative; inclusive;
restrictive; successive.
:: circumstantial; consequential; SINGLECLAUSE; DOUBLE-CLAUSE.
:: comparative; complemental; purposive;
locative; temporal.
:: CONDITIONAL; temporal.
:: possible; real; unreal.
4. Semantics
We discussed so far feature variables with distinct values at the morphological and syntactic level of description, variables such as: aspect, tense,
definiteness, gender, number, person, and case. Moreover, with featureclusters such as concord and agreement (involving: human, person,
gender, and number), we are able to describe regularities between a
predicate and an explicit agent (ps-match), between a predicate and
possibly realized complements (po-match) in a sv, as well as between
topic and comment realizations (tc-match) in a sn.
As far as our semantic level of description is concerned, we earlier
insisted to remain, in this paper, within the limits of static semantics as
used in almost all other applications of NLP.69 Table 3 summarizes our
68
In that case we can speak of a prepositional complement clause (PCCL).

We refer to the general domains of machine translation and information retrieval,
still using the basics on static semantics of Yngve (1966). See also Koster (2004).
69
390
everhard ditters
TABLE 3
Levels
morphology
syntax
semantics0
semantics1
semantics 2
Features
morphological
morpho-syntactic
syntactical
syntactico-semantic
semantical0
static-semantical
semantical1
dynamic-semantical
semantical2
etc.
working space ending at level-zero (semantics0) of static semantics. The

description of features at the levels listed in the squared box remains to
be done.
Thus, a comprehensive assignment of semantic features to all lexical
entries, which can occur as head element of one of the different phrasal
constituents, does not (yet) exist for automated analysis purposes of
Modern Standard Arabic. That may not be necessary when a heuristic
feature assignment for a ps-match, a po-match or a tc-match could
already meet our disambiguation objectives.
The more or less detailed sub-categorization of features for headers
in an advp and a pp ( 3.3.12) is a good example of our hypothesis.
This feature sub-categorization supplies such an advp with a specific
semantic load in the context of its occurrence. For the header of a pp, this
sub-categorization presupposes a matching with some corresponding
feature values of the head realization of the prepositional complement
for a grammatically and semantically correct analysis result. Therefore:
if the head-element of a prepositional complement shares the maximum
number of pertinent features with the prepositional header, alternative
analysis results should and could be discarded.70 If a noun like bustn
possesses a feature hinting at an open-air outdoors location we could
easily match the local feature values of the header and the prepositional
70
The fine-tuning of this matching occurs in the cyclic process of testing the description on new text data. Moreover, this matching produces optimal results when unification provides contextual information. A sequence as min tahtihi remains ambiguous as
long as the anaphor hi is not disambiguated in context.
391
complement and interpret the sequence f l-bustni in the garden as an

attributive or adverbial adjunct with local value. As a matter of fact, we also
have to provide individual lexical elements of verb and noun categories
with a minimum set of pertinent feature names and feature values.
4.1
Verbs
In our formal description of Modern Standard Arabic for automated

analysis purposes, individual lexical verbal elements have to provide
information about:
compulsory and optional completives and circumstantials within the
complement structure of a given verbal entry (transitivity);
the combination of a verbal entry with elements of other word
categories to denote a particular meaning of such a verbal entry
(collocation);71
minimum qualifications for possible implied or explicit agents and
completives (in terms of case-roles and syntactic realization).
4.1.1
Transitivity
Within the framework of the diinar-project72 verbal entries have been

compiled according to the number and the syntactic realization of their
arguments. In this way the verbal lexicon is first split up in intransitive and
transitive verbs. The transitive verbs are then further divided into: mono-,
bi- or tri-transitive verbs. Finally, these verbal entries can be divided
into sub-categories governing a direct object(s) (npacc or ccl), and/or a
prepositional object (pp or pccl). A further differentiation concerns the
lexical realization of the preposition (ppli, ppmin, ppan etc.).
What is still missing in this approach is the semantic dimension as,
for instance, Levin (1993) did for English.73 She compiled 49 verb classes
with sub-categories. The members of each sub-class share the general
meaning of the class as well as a number of syntactic characteristics. So,
the members of the sub-class 10.2 of banish verbs (Levin 1993, 123):74
71
Cf. Hoogland 1993; El-Gemei 2006, 434439.

Information about the DIINAR lexical database can be found at: http://catalog.
elra.info. Information about DIINAR-MBC can be found at: http://sites.unin-lyon2.fr/
langues_promodiinar/accueil.htm.
73
See also Levin & Rappaport Hovav 1996.
74
These members are: banish, deport, evacuate, expel, extradite, recall, and remove.
72
392
everhard ditters
relate to the removal of an entity, typically a person from a location. The
location argument is expressed in a prepositional phrase headed by the
preposition from. [. . . .] Unlike the remove verbs, these verbs allow to
phrases as well as from phrases, though not simultaneously.
The application of such an approach for the description of verbal entries

in Modern Standard Arabic is given in Table 4 where square brackets
point to mutually exclusive arguments and parentheses to optional
occurrences:75
TABLE 4
arg-4
manner
arg-3.2
destination
arg-3.1
source
arg-2
object
arg-1
agent
(PPbi)
(PPbi)
(PPbi)
(PPbi)
(PPbi)
(PPbi)
(PPbi)
(PPbi)
(PPbi)
(PPbi)
(PPbi)
(PPbi)
(PPbi)
(PPbi)
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPan|min,loc|pos76)]
[(PPan|min,loc|pos)]
(NPnom,+human)77
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
" #
$
% $
% $#
&' (
) * (
+) ,-
. /
0 ) (
12
) 12
3 4'
+ 56
7 8 9
(PPbi)
(PPbi)
(PPbi)
(PPbi)
(PPbi)
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
[(PPil|li,loc|pos)]
NPacc,+human
NPacc,+human
NPacc,+human
NPacc,+human
NPacc,+human
NPacc,+human
NPacc,+human
NPacc,+human78
NPacc,+human
NPacc,+human
NPacc,+human
NPacc,+human
NPacc,+human
NPacc,+human|
PPbi,+human
NPacc,+human
NPacc,+human
NPacc,+human
NPacc,+human
NPacc,+human
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
(NPnom,+human)
8 :
; 4<
: =
&: =
+>=
75
entry
A vertical bar | here denotes alternatives in subscript variables and features.

The feature alternatives loc and pos comprise a physical location as well as a social
function.
77
An optional explicit agent has the same feature values as the implied agent
78
The object of this entry is typically realized by the collective nouns army, troops
or alternatives.
76

4.1.2
393
Collocation
At this point we question the way of featuring specific verbal head-advp,

verbal head-np, and verbal head-pp combinations resulting in a, for the
combination, specific semantic value. The problem is that the matching
of the individual feature values of the elements involved does not necessarily point to the semantic value of the combination. A cheap solution
will be to list this kind of combinations as they are in the lexicon.
4.1.3
Case-roles and realization
According to Abdul-Raof (2006, 343), the use of case-roles may

adequately describe deep structure relations between noun phrases and
the verb. If this quote is equivalent to: feature matching between the
heads of a np and a vp (occurring as comment in a sn), or between the
heads of a vp and nps (occurring as dependent or independent nps, and
realizing a completive and/or a circumstantial function in a sv), we like to
explore the potentiality of Case Grammar for our purposes.
Winograd (1983, 311 ff.; 623624) describes the evolution of a caserole based description of semantic relations within syntactic structures,
and lists some 65 case roles. For Arabic, we refer to Abdul-Raof (2006)
and LeTourneau (2006) on Case Roles and Case Theory respectively, but
we will exploit Saads monograph (1982) for further featuring in a CG
framework.
With 12 roles Saad easily represented our 14 columns in Table 1
and 2 in case grammar (CG) terms, as is shown in Table 5a and 5b
below.79 In the columns 5 (range) and 12 (comitative), we recognize the
circumstantials: al-maf l al-mutlaq (cognate object) and al-maf l
maahu (accompanying object):
TABLE 5A
CASE ROLES IN MSA
arguments
case role
agent
instrument
patient/
object80
target
range
source
abbreviation
PO
Ta
79
The row abbreviation has been inserted to introduce names of variables within
a representation of the complement structure of verbal complement structure in case
frame terms.
80
We prefer to differentiate between a +human object (= patient) and a human
object.
394
everhard ditters
TABLE 5A (cont.)
arguments
verb class
agentives
agentives
agentives;
verbs of a
mental,
physiological,
emotional state
or process
agentives;
verbs of a
mental,
physiological,
emotional state
or process
features
+animate81
-animate
form
NPnom
NPacc|PP
|PCL82
NPnom|NPacc
NPnom|NPacc
|PCL
TABLE 5B
case role
abbreviation
verb class
10
goal
G
verbs of change,
experience,
transformation,
action in motion
or time
place
P
time
Ti
manner
M
local;
position
time
manner;
degree
PP
NPacc|PP
NPacc|PP
PP
6
verbs of
change,
transformation,
experience,
action in
motion
or time
NPacc
AJDPnom|
NPnom|PP
CASE ROLES IN MSA
features
form
11
12
beneficiary comitative
B
C
PP
NPacc|PP
With a paragraph on co-referential case roles, Abdul-Raof (2006,

345) gives a hint for the fine-tuning of verbal sub-categorization in
Modern Standard Arabic. He lists five cases wherein the role of Agent
is co-referential with the role of Object (= Saads patient), Source, Goal,
Experiencer (= Saads patient), and Beneficiary respectively.83
The verbal entries of Table 4 share an optional explicit subject84
(Agent) in the form of a np, marked for a syntactic nominative case
value and a semantic +human value (npnom,+human). Moreover, they share
81
Saads (1982, p. 20) suggestion for an alternative realization form (animate) to
describe astronomical bodies and semi-autonomous machines, gets a more elegant
foundation in Dichy (2005).
82
A vertical bar | here denotes alternatives at form level.
83
The co-referentiability between an Agent and more than a single other case role
remains to be studied.
84
An implicit subject of a finite verb form is always present or is neutralized and
overruled by an explicit agent.
395
an compulsory direct object (Patient) in the form of a np, syntactically

marked for accusative case value and semantically as +human
(npacc,+human). Above that, they share two optional, mutually exclusive
local circumstantials, expressing the origin (Source) and the destination
(Goal) of the action being or been performed (ppan|min,loc|pos or ppil,loc|pos).
Sometimes there is even mention of an optional circumstantial in the
form of a ppbi, marking the manner (Manner) in which the action is
performed (bi lquwa by force). Finally, one could witness the occurrence
of an optional circumstantial in the form of an advp or pp marking the
moment (Time) of the action performed (advptime or pptime).
Such a description can be represented by means of a frame with a
line for the lexical entry concerned, abbreviations for the cases with
realization forms between curled brackets ({}), while parentheses (()) and
square brackets ([]) denote optional occurrence and mutual exclusion
respectively. A vertical bar | separates alternative realization forms.
For our example of the sub-class of banish verbs, this results in the
following case frame for the members of this sub-category:
[__ A({NPnom,+human}), P{NPacc,+human}, [(S{PPan|min,loc|pos})], [(G{PPil,loc|pos})],
(M{PPbi}), (I{PPbi}), (Ti{ADVBtime|PPtime})]
During the analysis process of Arabic text data, a lexical verbal entry will
be recognized as far as its consonantal root and stem is concerned, while
the information about verbtype, derivation, aspect, tense, voice, person,
gender, and number are being stored. In the meantime, information
about the case frame of the verbal entry will be loaded from the lexicon
and matched against the string(s) following the finite verb form.
The next step in our description seems to comprise the drafting of
case frames for the Arabic equivalents of members of the other subclasses of the class of remove verbs. This predicts a cyclic process of the
drafting and testing of case frames for the Arabic equivalents of the
members of all the other 48 classes Levin (1993) listed for English.
We like to retain the following case roles in the description of the
complement structure of verbal entries for agent and object matching
with different kinds of completives and circumstantials: agent (a),
instrument (i), patient (po), target (ta), range (r), source (s), goal (g),
place (p), time (ti), manner (m), beneficiary (b), and comitative (c).
We like to retain the idea of pairing case roles by co-referencing, as
Abdul-Raof (2006, 345) gives with: Ai=Oi,85 Ai=Si, Ai=Gi, Ai=Ei, and
85
Abdul-Roaf only lists 8 case roles and does not employ the term Patient. Therefore
396
everhard ditters
Ai=Bi. It certainly may refine the description of elements of different

verb classes and the complementstructure of verbal entries.
ACTORS
MONO
BENEFICIARY
PAIRS
[TRIADES
4.2
:: MONO; PAIRS; TRIADES.

:: agent; instrument; object; patient; target;
range; source; goal: place; time;
manner; BENEFICIARY; comitative.
:: client; recipient.
:: A=PO; A=S; A=G; A=E; A=B.
:: ].
Nouns
At the end of the introduction we stated that the formalism for automated
processing is suited for describing morphological and syntactic structure, as well as finite semantics in terms of animate, concrete,
human, volition and many other. At the morphological and syntactic
level we already discussed a number of features attached to elements
of the category noun. Besides that, the differentiation of relevant
prepositional features in time, place, and ideal presupposes a matching
with corresponding features attached to the head of a prepositional
complement, in casu a noun. The question is: how to make an inventory
of the minimum number of required semantic noun features, and how
to set up the organization of this inventory.
A frequently used approach consists in the drafting of a conceptual
semantic tree. Hartmann (1974, 169) used Moravcsiks (1970) conceptual
tree starting with the node countable for her Transformational Generative Grammar of Arabic. Muhtaseb (1988, 65 ff.) developed (far too)
detailed conceptual trees describing the subject and action hierarchy for
his Natural Language Understanding System (NALUS). More recently,
Dichy (2005) proposed a conceptual tree of three layers starting with
concrete, using two other basic features: animate and human. This
tree generates the lexical categories: person, animal, perceptible
entity, and idea or non-physical experience. By a transfer of some
features, characteristic for +animate and +human entities, to animate
and human entities by assignment of specific values as intention,
+volition, or +motion, Dichy also generates the derived categories:
it is not yet clear how to harmonize his co-referential pairs Agent = Object (Ai=POi) and
Agent = Experiencer (Ai=Ei) with the set of 12 case roles of Saad.
397
pseudo-animated perceptible entity, collectivity, and pseudoanimated perceptible entity.

We like to retain the following base set of semantic features (BASEF)
for elements of the category noun: concrete, animate, and human. For
the moment, we also like to incorporate Dichys sub-features: intention,
motion, and volition. Moreover, for elements of the category noun
we have to include features required in the matching with features of
elements of the categories verb and particle. In 4.1 we saw that those
features are required for predicate-subject (psmatch) and predicate-object
(pomatch) matching. In this subsection they are required in the topiccomment (tcmatch) matching. In the following section these features
are needed in a preposition-complement ( pcmatch) matching.
In conclusion, the human value of a noun plays a role in agreement86
and concord87 phenomena involving, in the Sv, the elements of the predicate-explicit agent ( psmatch) matching or the elements of the predicatepatient/object ( pomatch)88 matching, and, in the sn, the elements of the
topic-comment (tcmatch)89 matching.
Below we list our features rules for elements of the category noun,
easily expandable with other variables and values:
BASEF
SUBF
4.3
:: concrete; animate; human; place; time.

:: intention; motion; volition.
Particles
Concerning the vocatives we would like to discriminate at the semantic

level between occurrences related to +human and +animate, and human
entities. We prefer to further differentiate the exclamations into subcategories expressing: agreement, commands, dissent, enthusiasm, sorrow,
surprise, presentations, and quantitatives.
For our purposes it is meaningful to divide the group of adverbs by
means of a variable advtype into elements expressing: manner, place,
86
If an explicit agent (arg-1) is marked for +human, there is full gender matching
between predicate and subject, otherwise (if human), there is partial gender matching,
i.e. the verbal entry matches gender values with a subject marked for singular or dual
number, but it receives feminine gender value with plural subjects.
87
A NPnom,+human in the topic position of a Sn triggers full gender and number matching
with elements, liable for such a matching, in comment position.
88
In the case of a NPacc patient or object of a finite verb form marked for passive
voice.
89
In the case of a ADJPnom, NPnom or VP occurrence in comment position.
398
everhard ditters
time, quantity, affirmation, negation, or interrogation.90 Another variable

may function as a basket for elements expressing: restriction, doubt,
wish, or exclamation.91
The semantic feature exclamation also may occur as one of the values
of the variable inttype of the interjections. Other values are: attention,
confirmation, support, and sociality. Values for the variable preptype and
conjtype can be borrowed from their differentiation at syntax level. The
preptype features play a role in the preposition-complement ( pcmatch)
matching.
INTYPE
VOCATIVE
EXCLAMATION
ADVTYPE
DECLARATIVE
QUES
CONJTYPE
PREPTYPE
:: VOCATIVE; EXCLAMATION; f-expressions;

proverbs; greetings; insults.
:: +human ; +animate,-human.
:: agreement; commands; dissent; enthusiasm;
presentations; quantitative;
sorrow; surprise; warning; wishes.
:: DECLARATIVE; QUES; manner; place; quantity;
time.
:: affirmation; negation.
:: doubt; exclamation; interrogation; restriction;
wish.
:: cumulative.
:: instrument; measure; place; time.
5. Conclusion
We discussed variables and values of features of importance for the disambiguation of analysis results from the automated processing of contemporary Arabic text data. We considered three levels of description.
Morphological feature variables and values serve as input for the syntax
level. Feature variables and values from the semantic level help to refine
the description of language structure at syntax level to arrive at a single,
most probable, syntactically as well as semantically correct, analysis result.
At syntax level we presented structural descriptions of sentence types
and basic sentence phrasal constituents in Modern Standard Arabic.
The general complement structure including optional and compulsory
90
We are here dealing with the interrogative adverbs hal and a marking Yes-No
questions.
91
A number of refinements to this description has been effectuated on the basis of
El-Ayoubi et al. 2003, 275460.
399
completives and circumstantials of a verbal entry were summarized in

the form of tables.
At the semantic level we suggested a description of Arabic verbal
entries on the basis of the number and the syntactic realization of their
arguments as well as a correspondence in meaning la Levin (1993) for
English. Arabic equivalents of banish verbs served as an example. Then
we started to translate the accumulated information into Case Grammar
terms using Saads (1983) twelve cases for the description of the role of
arguments within the possible complement structure of a verbal entry.
We started to complete the description with information about type of
verb classes, realization form and final semantic features values of the
complements. This exercise resulted in the presentation of a case frame
for the Arabic banish verbs.
In the next parts of this semantic section we discussed a base set of
semantic features for elements of the noun and particle categories to be
used together with case frames for Arabic verbal entries (sv), as well as
for topics complemented by comment realizations (sn). A draft for predicate-explicit agent, predicate-object, topic-comment, and prepositioncomplement matching has been presented.
What remains to be done is:
to test the structural descriptions of sentence types and phrasal
constituents against new contemporary Arabic text data;
to apply Levins (1993) concept for verb categorization to the Arabic
verbal lexicon;
to test the results from this application against new contemporary
Arabic text data;
to translate the corrected results from that application into a
refinement of the minimum set of semantic features for elements
of the noun and verb categories;
to fill in the gaps and to expand the slots in figure 9.
400
everhard ditters
FIGURE 9
concrete +
animated +
human +
adult +
male +
determinated +
Ahmad
man
Fayruz
woman
6. References
Abdul-Raof, Hussein. 2006. Case Roles. In Versteegh, Kees et al., eds. EALL, vol. 1,
343347.
Ayoub, Georgine. 1980. Af lu l-qulb en arabe standard: lments pour une analyse.
In Bohas, Georges, ed. tudes Arabes: Analyses Thories. 1980, 1, 154.
Badawi, Elsaid, Michael Carter and Adrian Gully. 2004. Modern Written Arabic: A
Comprehensive Grammar. London and New York: Routledge.
Bahloul, Maher. 2006. Agreement. In Versteegh, Kees et al., eds. EALL, vol. 1, 4348.
. 2006. Copula. In Versteegh, Kees et al., eds. EALL, vol. 1, 506511.
Bergman, Elizabeth. 2007. Frozen Expressions. In Versteegh, Kees et al., eds. EALL,
vol. 2, 136137.
Bohas, Georges and Abderrahim Saguer. 2007. The Explanation of Homonymy in the
Lexicon of Arabic. In Ditters, Everhard and Harald Motzki, eds. Approaches to Arabic
Linguistics Presented to Kees Versteegh on the Occasion of his Sixtieth Birthday. Leiden:
E.J. Brill, 255289.
Cohen, David. 1970. Essai dune analyse automatique de larabe. In Cohen, David, ed.
tudes de linguistique smitique et arabe. The Hague: Mouton, 4978.
Cuvalay, Martine. 1994. Auxiliary Verbs in Arabic. In Elisabeth Engberg-Pedersen,
Lisbeth Falster Jakobsen, and Lone Schack Rasmussen, eds. Function and Expression
in Functional Grammar. Berlin: Mouton de Gruyter, 265283.
Cuvalay-Haak, Martine. 1996. The Arabic Verb: A Functional Grammar Approach to
Verbal Expression in Classical and Modern Arabic. PhD-thesis. Amsterdam: University
of Amsterdam.
Dichy, Joseph. 2005. Spcificateurs engendrs par les traits [anim], [humain],
[concret] et structures darguments en arabe et en franais. In Bjoint, Henri and
Franois Maniez, eds. De la msure dans les termes. Hommage Philippe Thoiron.
Travaux CRTT, Lyon: Presses Universitaires de Lyon, 151181.
Ditters, Everhard. 1992. A Formal Approach to Arabic Syntax: the Noun Phrase and the
Verb Phrase. PhD thesis Nijmegen University. Nijmegen: Luxor.
. 2001. A Formal Grammar for the Description of Sentence Structure in Modern
Standard Arabic. In Kaufman, Morgan, ed. Workshop Proceedings: Arabic Language
Processing: Status and Prospects, CRNSUniversit des Sciences Sociales Toulouse,
France, 3137.
. 2003a. Non-coinciding Phrasal Heads. In Chu, Hsing-Wei, Jos Ferrer, Tran
Nguyen, and Yuongquan Yu, eds. Proceedings of the Joint International Conference
401
on Computer, Communication and Control Technologies (CCCT03) and the 9th

International Conference on Information Systems Analysis and Synthesis (ISAS03).
Orlando, Florida, USA: International Institute of Informatics and Systemics, 5156.
. 2003b. An AGFL for the Description of Non-coinciding Phrasal Heads. In Callaos,
Nagib, William Lesso, Belkis Sanchez, and Elizabeth Hansen, eds. Proceedings of the
7th World Multiconference on Systemics, Cybernetics and Informatics. Orlando, USA:
International Institute of Informatics and Systemics, 107112.
. 2006. Computational Linguistics. In Versteegh, Kees et al., eds. EALL, vol. 1,
455465.
and Jan Hoogland. 2006. Corpus Linguistics. In Versteegh, Kees et al., eds. EALL,
vol. 1, 195201.
and Cornelis Koster. 2004. Transducing Arabic Phrases into Head-Modifier (HM)
Pairs for Arabic Information Retrieval. In Nikkhou, Maktab, ed. Proceedings of the
NEMLAR 2004 International Conference on Arabic Language Resources and Tools,
Cairo, Egypt: Magnet Creative Communications, 148154.
Eisele, John C. 2006. Aspect. In Versteegh, Kees et al., eds. EALL, vol. 1, 455465.
El-Gemei, Dalal Mahmoud. 2006. Collocation. In Versteegh, Kees et al., eds. EALL,
vol. 1, 434439.
Elzeiny, Nagwa. 2007. Greetings. In Versteegh, Kees et al., eds. EALL, vol. 2, 202207.
Fassi Fehri, Abdelkader. 1993. Issues in the Structure of Arabic Clauses and Words. Dordrecht : Kluwer Academic Publishers.
Fillmore, Charles. 1968. The Case for Case. In Bach, Emmon and Robert Harms, eds.
Universals in Linguistic Theory. New York: Holt, Rinehart and Winston Inc. 188.
Firanescu, Daniela. 2007. Exclamations. In Versteegh, Kees et al., eds. EALL, vol. 2,
7981.
Fleisch, Henri. 1961. Trait de philologie arabe, vol. 1: Prliminaires, phontique, morphologie nominale. Imprimerie Catholique: Beyrouth, 5665.
. 1968. Larabe classique : Esquisse dune structure linguistique. Recherches publis
sous la direction de lInstitut de Lettres Orientales de Beyrouth, deuxime srie:
Langue et Littrature Arabes, volume 5. Beyrouth: Dar El-Machreq.
Haak, Martine. 2006. Auxiliary. In Versteegh, Kees et al., eds. EALL, vol. 1, pp.
216221.
Hartmann, Regina. 1974. Untersuchungen zur Syntax der Arabischen Schriftsprache.
Wiesbaden: Harrassowitz.
Hoogland, Jan. 1993. Collocation in Arabic (MSA) and the treatment of collocations
in Arabic dictionaries. In Dvnyi, Kinga, Tams Ivnyi, and Avihai Shivtiel, eds.
Proceedings of the Colloquium on Arabic Lexicology and Lexicography. Budapest: Etvs
Lornd University and Csoma de Krs Society, 7593.
Kinberg, Naphtali. 2001. Studies in the Linguistic Structure of Classical Arabic. Kinberg,
Leah and Kees Versteegh, eds. Leiden: E.J. Brill.
Khim, Alain. 2006. Adjective Phrase. In Versteegh, Kees et al., eds. EALL, vol. 1, 14.
Koster, Cornelis. 1974. A Technique for Parsing Ambiguous Grammars. In Siefkes, D.,
ed. GI-4 Jahrestagung, Lecture Notes in Computer Science 26 (1975). Springer, 233246.
. 1991. Affix Grammars for Natural Languages. In H. Alblas and B. Melichar, eds.
Springer LNCS 545. Heidelberg, 469484.
. 2004. Head-Modifier Frames for Information Retrieval. In Proceedings CICLing2004. Springer LNCS 2945, 420432.
LeTourneau, Mark S. 2006. Case Theory. In Versteegh, Kees et al., eds. EALL, vol. 1,
347353.
Levin, Beth. 1993. English Verb Classes and Alternations: A Preliminary Investigation.
Chicago, Ill.: University of Chicago Press.
Levin, Beth and Malka Rappaport Hovav. 1996. Unaccusativity at the Syntax-Lexical
Semantics Interface. Linguistic Inquiry Monograph 26. Cambridge, MA: MIT Press.
Meijer, Hans.1986. Programmar: A Translator Generator. PhD-thesis, Nijmegen: University of Nijmegen.
402
everhard ditters
Moravcsik, Julius. 1970. Subcategorization and Abstract Terms. Foundations of

Language 1970 (6), 473487.
Nederhoff, Mark-Jan. 1993. A New Top-down Parsing Algorithm for Left-recursive
DCGs. In Programming Language Implementation and Logic Programming, 5th
International Symposium, Lecture Notes in Computer Science. Tallinn, Estonia:
Springer, Volume 715, 108122.
Saad, George Nehmeh. 1981. Transivity, Causation and Passivization: A semantic-syntactic
study of the verb in classical Arabic. Library of Arabic Linguistics. Monograph No. 4.
London: Keagan Paul International.
Al-Saffr, Abdul Emr Dhhir. 1979. A Semantico-Syntactic Study of Classical Arabic in
Case-Grammar Terms. PhD SOAS. London: University of London.
alSq, Fdil. 1977. Aqsmu l-kalmi l-arab min h aytu -akl wa-l-wazfa. al-Qhira:
Maktabatu l-Xnj.
Sbawayhi. al-Kitb. Hrn = Ab Bir Amr b. Utmn Sbawayhi. al-Kitb. 5 vols.
Abd as-Salm Muhammad Hrn, ed. 1977. Cairo: al-Haya al-Misriyya al-mma
li-l-Kitb.
Versteegh, Kees. 2001. The Arabic Language. 2nd revised ed. Edinburgh: Edinburgh
University Press.
, Mushira Eid, Alaa Elgibali, Manfred Woidich and Andrzej Zaborski, eds. 2006.
Encyclopedia of Arabic Language and Linguistics, Volume I, A-Ed. Leiden: E.J. Brill.
, eds. 2007. Encyclopedia of Arabic Language and Linguistics, Volume II, Eg-Lan.
Leiden: E.J. Brill.
Winograd, Terry. 1983. Language as a Cognitive Process. Volume I: Syntax. Reading:
Addison-Wesley.
Yngve, Victor. 1966. A framework for syntactic translation. In David G. Hays, ed. Readings in Automatic Language Processing. New York: American Elsevier, 189198.
ARABIC ON THE MEDIA: HYBRIDITY AND STYLES*

Mushira Eid
University of Utah
1. Introduction
The purpose of media programming is to inform and to entertain. Some
programs can be classified as informative, others as entertainment, and
others as both. The language used on these programs is expected to
reflect these discourse functions. In a diglossic situation, such as that
found in the Arabic speech communities, the choice involves two (at
times more) relatively distinct varieties, or linguistic codes, of a language: the high (standard variety) for information and the low (local
dialect) for entertainment. In Arabic these correspond to fush and
mmiyya, respectively. Programs classified as being both (information
and entertainment) raise a question as to what form of Arabic is, or
should be, used. These programs, like the broadcasting media in general, represent hybrid contexts that mix, and at times merge, the public
and the private, the formal and the informal. They serve as bridges, or
in-between spaces, where linguistic differences and cultural identities
are negotiated in the production of a program or a performance.
The relationship between language and the media is particularly interesting in the Arabic context due to the diglossic nature of its speech communities. Two major forms of Arabic characterize these communities:
one is the spoken everyday language of communication (mmiyya vernacular/colloquial/dialect), which is also written in certain contexts,1
the other a literary variety which is the medium of written communica-
* An earlier version of this paper was presented at the Center for the Advanced Study
of Language (CASL), University of Maryland, College Park, June 1, 2006.
1
Personal letters, for example, may be written in dialectal Arabic. In Egypt, certain literary genres are also written in Egyptian Arabic including plays, poetry, and a
few biographies (Awad 1965; El Assal 20022003). There is also one novel written in
Egyptian (Musharrafa n.d.). The practice of including some form of dialect in literature
mixed with fush is a relatively common practice, particularly when it involves dialogue
(Cachia 1990, 5975, chapter four The use of colloquial in modern Arabic literature;
Eid 2002).
404
mushira eid
tion and is also spoken in some formal contexts ( fush lit: eloquent,
literary/standard). The two varieties differ in linguistic form, manner
of acquisition, function or use, and social meaning/value. The definition
of diglossia as it was first presented by Ferguson (1959) is given in (1).
1. Diglossia is a relatively stable language situation in which, in
addition to the primary dialects of the language (which may include
a standard or regional standards), there is a very divergent, highly
codified (often grammatically more complex) superposed variety,
the vehicle of a large and respected body of written literature, either of
an earlier period or in another speech community, which is learned
largely by formal education and is used for most written and formal
spoken purposes but is not used by any sector of the community for
ordinary conversation (Ferguson 1959, Word 15:336).
The definition in (1) presupposes a relationship based on separation,
both linguistic and cultural, of the two varieties fush and dialect. It also
reflects the official discourse and predominant attitudes toward them.
Because of its being the language of divinity and heritage in Islam, the
fush is considered the language of high culture and prestige whereas
the dialects are local, highly divergent, and often not mutually intelligible, at least at the outset. Fush is also considered a unifying force, the
pan-Arab national language, in the words of Shawqi Daif (2001) luat
ub al-umma jamcan the language of all the peoples of the [Arab]
nation, whereas the dialects represent the daily language of a single people . . . understood only by its people hence divisive. (Quoted in Boussofara-Omar 2006, 108; see also Haeri 2003 for further discussion.)
Social, political, educational, and technological changes in the 20th
century have brought about change in the relationship between the two
varieties, particularly in their contexts of use, their market value, and
their social meaning. Accessibility to fush increased, for example, due
to changes in the educational system. In Egypt the availability of free
public education through high school and higher education increased
the number of people who can use both varieties. Contexts in which
fush is used increased as well, the broadcasting media as a spoken context for fush being one.2 As a result, the media serves as a liberating
force: by providing contexts that are not restricted to one variety, the
2
The media are often criticized for allowing the dialects to creep into domains originally reserved for fush . Such intrusions will eventually dismantle the ties that bond
the peoples of the umma [Arab nation] (Shawqi Daif 2001, cited in Boussofara-Omar
(2006) op. cit.). The broadcasting media, however, are not a domain originally reserved
arabic on the media: hybridity and styles
405
media allows users of Arabic access to many forms of Arabic, dialectal

and fush . This position contradicts the predominant discourse, as represented, for example, in the quotes from Shawqi Daif, that argues for
the separation of the two in order to maintain the purity of the fush
against intrusions from dialect (but see footnote 2).
The media, particularly the broadcasting media, has a unique role to
play in negotiating the relationship between the two varieties. Because
of its dual role, to inform and entertain, the media serves as a bridge
between the public and the private, the local and the global. It represents different forms of discourse: formal (speeches, lectures, news
broadcasts) and informal (conversations, story-telling, joking) and creates in-between spaces that serve as excellent sites for the negotiation of
identities. It does so by bringing public content into the privacy of the
home and taking private content to the public view to audiences that are
local and, when aired over satellite channels, global as well. For the Arabic broadcasting media there is in addition a pan-Arab audience that
needs to be addressed. Some channels focus on that pan-Arab audience,
e.g. al-Jazeera. Others, e.g. the Egyptian Satellite Channel (ESC) and the
Lebanese Broadcasting Channel (LBC), focus more on local audiences.
The varieties of Arabic heard over these channels naturally reflect the
intended audience.
In this paper I start from the position that the broadcasting media
(henceforth just media) represent hybrid contexts, which for Arabic
means contexts that allow both fush and dialect. I focus on the interview as a genre whose purpose is to inform and to entertain and whose
context is appropriate for either or both varieties of Arabic.3 I examine
two interviews aired over the Egyptian Satellite Channel. I show that the
language that emerges in these interviews is a mix, identifiable at times
as fush and at times as mmiyya (here Egyptian) but more often as an
in-between variety, a hybrid form of Arabic. Three major styles are
identified, corresponding roughly to Labovs reading, careful, and casual
styles. The styles vary depending on degree of formality and interaction,
topics, and participants (their perceptions of themselves, projections of
for fush . The statement must therefore be applicable to Arabic in the written media,
e.g. newspapers.
3
Eid (2004) shows that interviews and panel discussions from al-Jazeera are at times
conducted completely in fush , but they may include dialectal features depending on
speaker, topic, and degree of interaction. The programs analyzed, however, were limited
to different forms of political discourse.
406
mushira eid
identities, and perceived purpose of communication). For Arabic, they

also involve the alternating use of features of fush and dialect. I discuss
characteristics of each style in relation to the nature and source of the
hybridity. In concluding the article I briefly discuss hybridity in relation to code-switching and argue that they are different constructs.
1.1
Correspondences between fush and Egyptian Arabic
Table 1 illustrates four types of correspondences between fush and Egyptian Arabic (henceforth, Egyptian) to be used in the subsequent analysis of speech styles: phonological, syntactic, morphological, and lexical.
TABLE 1
SOME CORRESPONDENCES BETWEEN FUSH AND

EGYPTIAN ARABIC
Phonological
q~
t ~ t ~ s
t ~ s
d ~ d ~ z
d ~ z
d ~ d ~ z
d ~ z
diphthong~long vowel
ay ~
aw ~
Syntactic
Relative markers
Demonstratives
Negatives
Complementizer
Morphological
Passive (u-i/u-a ~ it-)
Verbs: Stem IV
I
af al
faal
Fush
Egyptian
Gloss
qalb
alb
heart
aktar
tawra
axad
kadlik
du
hr
du
lm
aktar
sawra
axad
kazlik
duhr
zulm
bayt
xawf
bt
xf
house
fear
alladi, allati, . . .
hda, hdihi, . . .
laysa/l/lam/lan
an
illi
da, di, . . .
mi, ma-
in/
who, whom . . .
this, . . .
not
that
kutiba
yuktab
ajab
it-katab
y-it-kitib
N/A; rare
agab
was written
to be written
became
to please, like
anf
lianna
indi
kalima
kitb
fannn
manaxr
an
andi
kilma
kitb
fannn
aksar
*tawra
axaz
*kadlik
zuhr
*dulm
asbah
more, most
revolution
he took
also, likewise
noon
injustice
Lexical
Phonologically-related
Unspecified
nose
because
I have
word
book
artist
407
On the phonological level, three major correspondences between fush

and Egyptian are illustrated: (1) uvular stop /q/ vs. glottal stop / /; (2)
the interdental fricatives / t d z/, which at times correspond to the stops
/ t d d/, at other times to the sibilants / s z z /; and (3) the diphthongs
/ay aw/ corresponding to the long vowels / /.4 Correspondences in
(1) and (3) apply across the board and therefore need no comment: if
a fush lexical item has features in (1) or (3), the corresponding Egyptian item has the corresponding feature in (1) or (3).5 The interdental
fricatives series, however, require some comment. The fush fricatives
correspond in some words to the stops (aktar ~ aktar more), in others
to the sibilants (ulm ~ zulm injustice). At times both options are possible. For aktar there is a third alternative aksar, but not for du
lm since
*dulm is not an actual word. This distribution is problematic for the
classification of lexical items, a problem to be addressed briefly in discussing hybridity.
On the syntactic level, four features are chosen for analysis: relative
clause markers, demonstratives, negatives, and complementizers. They
all serve as functional heads of their respective clauses with clearly divergent forms in the two varieties. Except for complementizers, they have
also been studied extensively in the literature of Arabic code-switching
(e.g. Bassiouney 2006, Boussofara-Omar 2006b, and Eid 1982, 1988.)
On the morphological level, two verb forms have been selected for
this study: passives and Stem IV verbs. Fush and Egyptian passive verbs
are formed by different morphological processes. In fush the passive is
marked derivationally by the vowel melody /u-i/ in the first and second syllable, respectively, (e.g. kataba wrote ~ kutiba was written). In
Egyptian it is marked inflectionally by the prefix it- (e.g. katab wrote ~
itkatab was written). Stem IV verbs, formed on the pattern /a-CCVC/
like aqbal approach or agab like, are rare in Egyptian. They are sometimes replaced by verbs from different roots, e.g. gih to come instead of
aqbal, but more often they are replaced by Stem I (CVCVC) of the same
verb, e.g. agab instead of agab.
4
Following the transcription requirements for this volume, long vowels are represented as single segments / /. For reasons that have to do with syllable weight and
identification of minimal word, to be explained below, these should be understood as
sequences of two identical vowels /aa ee ii uu oo/.
5
The opposite, however, is not true. Not all words with sibilants, for example, in
Egyptian have a corresponding interdental fricative in fush ; some words may have a
sibilant in both, e.g., fush : sam sky and Egyptian sama.
408
mushira eid
Finally, lexical differences between the two varieties come from different sources. The examples in Table 1 illustrate two of such areas.
Some differences are the result of word usage, the association of a meaning with two different lexical items, one considered fush usage and the
other Egyptian as, for example, anf and manaxr nose. Other differences are phonologically-related as, for example, the difference between
indi and andi I have. Because fush and Egyptian are varieties of the
same language, it is expected that an overwhelming majority of lexical items would be shared. Table 1 provides examples of words that are
identical in both varieties, e.g. kitb book and fannn artist. I refer to
such words as ambiguous or unspecified to mean non-distinct, or not
specified, as to language variety. I use the term both for words that
include features from both varieties, i.e. hybrid forms.
1.2
Hybridity
A hybrid is something that is mixed, and hybridity is simply the mixture. The term has its origin in biology.6 In linguistics it refers to a word
parts of which come from two different languages. The term has become
central to major theoretical discussions among the discourses of race,
post-colonialism, identity, multiculturalism, and globalization. In theories of cultural studies, it is understood as recognition of two identities
or a mixture of identities but also as a refutation of assimilation into
a dominant culture. Bakhtin, for example, defines hybridity [hybridization] as a mixture of two social languages within the limits of a single
utterance, an encounter, within the arena of an utterance, between two
different linguistic consciousnesses, separated from one another by an
epoch, by social differentiation, or by some other factor (1981, 358). The
term has also been used in relation to mixed language varieties that
result from code-switching among bilinguals (Hinnenkamp 2003).
In Arabic hybrid or intermediate forms, as they are sometimes called,
include features from both varieties fush and dialect and, therefore,
they cannot be clearly identified as belonging to one or the other.7
Table 2 illustrates Arabic hybrid forms with examples selected from the
6
It comes from the Latin: hybrida, a term used to classify the offspring of a tame sow
and a wild boar.
7
See Schulz (1981, 8789) for a discussion of hybridity versus mistakes.
409
interviews. In the following section, I also illustrate this with sample

texts from the interviews.
Here and in all other examples italics indicates Egyptian Arabic,
bold indicates fush , and unmarked typescript indicates material that
is unspecified as to Arabic variety, hence ambiguous or shared material.
The first example in Table 2 is a hybrid, because it includes the Egyptian present tense marker, the prefix ba-, attached to a fush verb form
marked by the vowel melody /a-a/ of the Stem. The second example
includes the fush person marker prefix /yu-/, marked by the /u/ vowel
rather than the Egyptian /i/, and the fush mood marker, the final /-u/,
but it also includes the sibilant /s/ of Egyptian, instead of the fush /t/.
TABLE 2
Hybrid
Fush
1. bataammala-taammal
2. yu-massil-u yu-mattil-u
3.a. salt ta
talta
3.b. talsa
3.c.
? salsa
HYBRID FORMS
Egyptian
Impermissible Gloss
hybrids
b-[]a-tammil
*bataammil I contemplate
yi-massil, *yimattil *yi-massil-u he represents
talta
*talsa/salta three
*talta/talta
? salsa
The variation in correspondences between fush /t/ and Egyptian stops

and sibilants creates lexical possibilities for hybridity and ambiguity. The
last example in Table 2, with the different pronunciations of the word for
three, illustrates this point. The hybrid forms (3ab) allow the mixture,
or co-occurrence, of fush /t/ only with sibilants (e.g. talsa). Combinations with Egyptian stops are impermissible hybrids (*talta). This
suggests that the sibilants themselves can be analyzed as in-between
spaces or ambiguous segments: at times they correspond to fush /s/ as
in the many shared words with sibilants (e.g. sitta six), at other times
they correspond to fush /t/ as in the example (2) in Table 2. Moreover, at times they correspond to Egyptian /t/ as in the last examples in
(3). The impermissibility of hybrids involving either the sibilants or the
interdental fricatives with the stops, as in /*talsa/ or /*ta lta/, suggests
that sibilants and interdental fricatives may be considered members of
the same category or group, since they can substitute for each other.
This distribution also raises a question as to the identification of sibilant
pronunciation, particularly in forms where the only Egyptian correspondence to the interdental fricatives is the sibilant. Are the examples
410
mushira eid
(3ab) in Table 2 hybrids, as I classified them, or are they simply alternative pronunciations for the interdental fricative? It is an open question
with arguments for both alternatives.
The modus operandi adopted in earlier research contextualized the
decision (Eid 1982, 1988). In lexical items where speakers have all three
choices (t, t, and s), the sibilant pronunciation is counted as fush since
the speaker has an alternative, namely /t/ pronunciation, should she or
he have opted for Egyptian. Likewise, in lexical items where Egyptian /t/
pronunciation is not available, sibilant pronunciation of fush interdental fricatives counts as Egyptian for similar reasons.8
The above discussion of hybrid forms illustrates the concept through
lexical items. In this article I consider hybridity to be a global feature of
text or discourse. It applies to forms of language that look more like a
collage, created out of two or more linguistic codes where the borders
between language varieties are blurred or no longer distinct.
2. Styles
Speech styles are forms of verbal self-expression and interaction. A style
is based on a speakers choice of linguistic form, discourse strategies,
and modes of interaction. Styles vary depending on degree of formality,
medium (spoken or written), and context or situation (conversation in
a coffee shop, for example, versus a lecture, speech, or an academic presentation). For Arabic, styles may also vary depending on language variety as well as the degree and nature of the mix between fush and dialect
features.
Three styles are identified in each interview : reading, formal, and
casual styles. While they correspond to Labovs classification of speech
styles, they are not identical to it. The difference lies in the formal and
casual styles, where for Labov casual is conversational and emerges in
relaxed situations when the participants become more comfortable with
each other and forget the context of the interview.9 For Arabic the styles
8
Alternatively, sibilant pronunciation may be considered ambiguous, or shared by
both varieties. The high frequency of sibilant pronunciation in fush contexts, e.g. reading, speeches, and recitation of literary texts, supports this view. This pronunciation is
not permitted in recitations of the Quran, however.
9
Labov has a fourth style D: word lists, which is relevant as an elicitation technique
but not applicable to natural conversations as represented in these programs.
411
also correspond to the relative distribution of linguistic features from

fush and mmiyya. In this section I illustrate the three styles and provide a characterization of each based on a quantitative analysis of fush
and Egyptian features in them, as described in Table 1.
Eight segments are selected to illustrate the three styles, four segments
from each of the two interview programs. The interviewees in these programs are Yehia Haqqi and Anis Mansour. Both are represented by three
segments illustrating formal and casual styles. The reading style is represented by two other speakers, one from each interview.
2.1
Styles illustrated
The reading style can be easily identified. It occurs when an individual

is seen reading from a script, as in the Haqqi interview, or heard narrating, for example through the use of voice-over, as in the Mansour
interview (see section 3). Segment 1 and 2 from the Mansour and Haqqi
programs, respectively, illustrate the reading style.
Segment 1 with the voice-over is pre-recorded in the studio. No evidence can be found of any Egyptian features, with the exception of the
substitution of Egyptian /g/ for fush /j/, as in the word az-zawgat-u
for az-zawjat-u the wife. This substitution has become so common in
Egypt that it is considered an accepted fush pronunciation in almost all
readings except those of the Koran and other religious texts. Otherwise,
segment 1 can be considered pure fush , or as pure as it can get. Naturally, this ignores the issue of lexical items and particles, or function
words, that are ambiguous or unspecified for variety: galla majesty,
sah fa journalism, f in, and, in if. Case markers are pronounced, for
example, and vowel patterns are those of fush , e.g. al- for the definite
article not il- as in as-sah fa.
SEGMENT 1: San Mansour (interviewer)
as-sahfa hiya z-zawgat-u at-tniya fi hayt-i l-ktib-i l-kabr ans
mansrin gz-a ha t-ta'br. bada"at 'alqat-u ans mansr bi shibat-i
l-galla as-sahfa fi 'm-i alf-in wa tisimiat-in wa sabat-in wa arban.
wa knat al-bidya fi gardat-i l-ass.
Journalism is the second wife in the life of the distinguished writer Anis
Mansour, if such an expression is permitted. Anis Mansours relationship with her royal highness [journalism] began in 1947. The beginning
was in the newspaper al-ass.
412
mushira eid
Segment 2 read by Fouad Duwara in the Haqqi program has a few more
ambiguous lexical items; more importantly, however, is the Egyptian
pronunciation of the years 1962 and 1970. This is common in readings,
for example, of the news and on-the-air commentaries as is the case of
this reader, who is also reading on-the-air. Otherwise, fush is maintained throughout in vowel patterns, e.g. h wal not Egyptian h wil, and
other pronunciation differences.
SEGMENT 2: Narrator (Fouad Duwara in Haqqi interview)
wa fi ibrl 'm alf w tusumiyya itnn wi sittn 'uyyin-a yehya haqqi ra"s-an
li-tahrr-i magallat al-magalla wa zalla yatawalla mas"liyataha hatta
December alf w tusumiyya w sabn wa hwala tawl tilka s-sanawt an
yuhfiz-a li l-magalla ala i'raha alladi t-taxadathu li nafsih mundu
inih wa huwa sigil it-taqfat-i r-rafa.
And in April 1962 Yehya Haqqi was appointed editor-in-chief of alMagalla journal. He continued to be responsible for it until December
1970. He tried during all those years to preserve for the journal the
motto it had adopted for itself since it was established, and that is [to
be] a record of high culture.
Segments 35 are samples from Haqqis speech. The first, segment 3, has
all the characteristics of a reading style, although Haqqi was not reading. The segment has all the features characteristic of fush including
case markers, vowel patterns, etc. and no features of Egyptian. It also
includes only a few words (6 total) unspecified for variety ( fann art, sir
struggle, maa with, zaman time, bir transient, and xlid immortal).
SEGMENT 3: Haqqi (on art)
al-fann 'ind mawqif-un drmiyy. "aql-u "anna l-fann sir maa z-zaman,
"annahu yurd-u "an yatasayyada l-bir li yuqayyidah, sir'-un maa
l-mawt. wa ma'a dlik hnam [indam] taqif "amma lawha ta'ur
"anna hda l-bir hda l-xlid allad xalladtahu innam huwa bir.
Art for me [is] a dramatic moment. I say that art [is] a struggle with
time, that it wants to capture the transient to preserve it, a struggle with
death. And despite this when [when] you stand in front of a painting
you feel that this transient, this immortal that you immortalized, is only
transient.
Segment 4 differs from segment 3 in that it has some evidence of Egyptian Arabic features from syntax, morphology, lexicon, and pronunciation patterns: the relative marker illi, the tense/aspect prefix ba-, the
413
negative main malh, the word wayyit, and the pattern in baladna,
not baladina the expected fush pronunciation.
The segment also includes a relatively large number of words unspecified for variety (27), almost equal to those specified for fush (26), and
a minimal (6) specified for Egyptian. This segment also includes the
hybrid form /bataammal/, reflecting and perhaps also constructing the
Egyptian and day-to-day context of the discourse. 10
SEGMENT 4. Haqqi (his hobby)
min asman il-hadya allati niltuha hdihi s-sra wa hiya min rasm
akbar musawwir s-sn f dlika l-waqt. il-huwya l-wahda illi xaragt
bha huwyat gami wayyit isyn ma-lh- ayy qma kbra lkin li kull
asya makna fi qalbi. wi bataammal kayfa axtr ihd hdihi l-usa. wa
hun aknu fi hla min al-istihz qallan wa urdu an atamid ala ay
fa axtr asan min agr-i misr min agr il-fallahn fi baladna.
One of the most precious gifts that I received is this picture and it is a
painting of one the best painters of China at that time. The only hobby
that I came out with is the hobby of collecting some canes that dont have
any value but each cane has a place in my heart. And I contemplate how
to choose one of these canes. And here I am being a little sarcastic and I
want to depend on something so I choose a cane from the trees of Egypt
from the trees of the peasants in our country.
Finally, segment 5 of the Haqqi sample includes more instances of lexical items marked for Egyptian (14) by comparison to previous segments.
The number is still the lowest by comparison to that of lexical items
marked for the predominant fush (36 instances) and those unspecified for variety (25 instances). Segment 5 is interesting from another
perspective: it is the only segment of the three from Haqqi that actually
has one or more sentences predominantly in Egyptian as, for example,
the last sentence. Finally, this segment also illustrates variation in the
use of fush interdental fricatives versus Egyptian sibilants through the
alternation muwaddaf ~ muwazza f employee. I have counted the form
with the sibilant /z/ as Egyptian since the alternative with the stop pronunciation (*muwaddaf ) does not exist.
10
The hybrid form bataammal in this segment and yumassilu in segment 5 are
excluded from the count in each case for the obvious reason: each has features of both
varieties.
414
mushira eid
SEGMENT 5: Haqqi (on government employees)

fi hdihi l-fatra istatana an nungiz bada l-aml. rubbam l yahiqq-u
l an adkurha lkin l bas min an aql . . . fa knat tagriba arba giddan xussan fi l-muwazzafn bi n-nisba li hdihi l-wazrt at-taqfiyya.
anta bayna namatayn min al-muwazzafn: muwazza f idri lkin sifr fi
l-fann wa muwazzaf fannn lkin sifr fi l-idra. timsiku min hina yigri min
hina wa il xirih. fa kna l-amal asr giddan. al-husl ala muwazza f
yumassilu xayra at -t arafayn an yakna fannnan wa fi l-waqt nafsu
yistah mil innu yuud ala maktab wi yira waraa wi yiraf yiktib gawb
kwayyis.
During this period we were able to accomplish some goals. Maybe I
dont have the right to mention them but its ok that I say . . . so it was a
very strange experience especially in the employees of these ministries
of culture. You are between two types of employees: an employee [who
is] administrator but zero in art and an employee [who is] an artist but
zero in administration. You catch him from here, he runs from there and
so on. So work was very difficult: the finding of an employee who is good
at both, to be an artist and at the same time tolerates to sit at the desk and
read a paper and know how to write a letter well.
The speech samples from Haqqi demonstrate that in this speech style
features from Egyptian are consistently much lower in frequency than
those from fush as well as those unspecified for variety. One sample
(segment 3) has no features of Egyptian and includes almost all the
characteristics of a reading style.
What, one might ask, would explain such differences? Topic is a very
likely explanation. In segment 4 with the most fush , Haqqi speaks
about art and what it means to him: a topic that lends itself to fush but
more importantly, a topic he has written and lectured about throughout his long career.11 In segments 5 and 6 the topic shifts to relatively
more personal topics and situations: his hobbies and his experiences as
a government official, respectively. The style shifts accordingly. The last
segment includes the most dialectal features, perhaps because it is also
the one most related to everyday events and behaviors as a result of his
comments on government employees. This is not to say that it is necessary that Egyptian be used in this case. Other segments in the program
show Haqqi not resorting to Egyptian under similar circumstances. In
fact, these two segments are the only ones in his speech sample that
include any features of Egyptian.
11
In an earlier segment Haqqi talks about the short story and the essay, stressing that
the purpose of the former is to entertain the reader (imt giving pleasure to) and the
latter is to inform (ilm) the reader.
415
Haqqis style can be said to conform to the discourse on diglossia by

relating Arabic variety to topic: fush for the abstract, informative, and
literary, but Egyptian for everyday situations. Haqqis speech sample
shows that fush can be used successfully and appropriately in descriptions and narrations about everyday situations, although it would be
odd to use it in actual day-to-day interactions.
A different style is represented in segments 68 from the Anis Mansour
interview. Segment 6 consists of a question from the interviewer, Sanaa
Mansour, followed by A. Mansours answer. The question addresses
the issue of opposition to a mainstream position: What happens when
one goes against the tide by taking an unpopular or oppositional stand.
Mansours answer includes more instances of lexical items unspecified
for variety (28) than instances of either Egyptian (14) or fush (15). This
segment also shows an almost equal distribution of Egyptian and fush
lexical items.
SEGMENT 6: A. Mansour (Swimming against the tide)
Question 1. SM:
bi-ylu l yaksab katran man yasbah didd it-tayyr. hadritak kisibt
il-kitr walla dayman bi-tisbah didd it-tayyr.
They say: He does not gain much who swims against the tide. Have you
[sir] gained much or were you always swimming against the tide?
Answer 1. AM:
da yatawaqqaf ala anni tayyr. ahynan min al-maslaha l-ma aw
il-wata niyya annaki tasbahi didd it-tayyr iza kn it-tayyr inhill aw
fasd, tafha, adam axz il-ay bi giddiyya. hun yusbih min al-wgib
ala l-ktib innu yasbah didd it-tayyr, innu yimil tayyr mudi liannu
wagbu innu yuslih aw yartafi bi n-ns. fa da yatawaqqaf ala anhi tayyr
illi fh xf aw mahazr innu yusbah didd-u.
This depends on which tide. Sometimes it is for the public or national welfare that you swim against the tide, if the tide immorality or corruption,
triviality, not taking matters seriously. Here it becomes the duty of the
writer that he swim against the tide, that he create a counter tide because
it is his duty that he reform or uplift the people up. So this depends on
which tide is fearful to swim against.
In segment 7, which is a follow-up on the question above, there is a clear

shift in style for both interviewer and interviewee. Sanaa Mansours
question here is an excellent example of what I call collage. It consists
of 9 words of which 4 are Egyptian, 3 are unspecified for variety, and 2
416
mushira eid
are specified for fush .12 She receives a one-word answer in Egyptian,
which she follows up with a question in clearly marked Egyptian syntax,
indicated by the WH-question word h what, and unmarked lexical items: kn and in-natga. Her question generates another collaged
response from A. Mansour, a response that is predominantly Egyptian
(16), minimally fush (2), with unmarked items in between (7).
SEGMENT 7: Mansour (Swimming against the tide, continued)
Q2. SM: gatlak fatart katra fi h aytak innak sabaht didd it-tayyr.
Did you have many times in your life that you swam against the
tide?
A2AM: kitr
A lot.
Q3SM: kn in-natga h
What was the result?
A3AM: wala h ga. w-adni id uddmik ahu. w-ixtalaft kasran w
ittafat iktr lkin madm inti muqtania aw amna f daawtik
aw fi bi wighit nazarik mayhimmik h ga aw mayhimmin
h ga.
Nothing, and here I am sitting in front of you. I have disagreed
often and agreed often but as long as you are convinced and honest in your mission or in your point of view, you dont care for
anything or I dont care for anything.
The last segment 8 from Mansour is also predominantly Egyptian (35

instances), minimally fush (4), with unmarked items (20) in between.
Had it not been for the last sentence which includes a sentence, or part
thereof, in fush , this whole segment could have been considered Egyptian. The unmarked lexical items, although a majority, have little impact
on the reader in terms of identification of variety. The syntax, phonology,
and lexicon of Egyptian predominate, and the identification as Egyptian
goes unchallenged.
SEGMENT 8. Anis Mansour (More about Kamel El Shenawi, ad h kan
zarf How nice he was!)
giddan yani maraf aullik nukatuh. kan yil ala Sawla H igzi allh
yirh amha innaha min adabha -idd ablima tiftah id-durg tixabbat alh
il-awwil. masalan kan andi arabiyya zuayyara kida baby Ford w-ana mi
12
The fush in her first question (segment 6) is the result of a quoted proverbial
statement.
417
saww kwayyis yani fa kn yil h in il-arabiyyt il-karro timi gamb

minni timil voo [li d-darag di] yani fa kull adt Kamel El Shenawi
marah wi xiffit damm w malib wi tayr fi k-kalm lkinnu abb w axx w
umm w nima s-sadq wa laysa lahu nazr fi h ayatna s-sahafiyya.
Very much [meaning] I dont know. I tell you his jokes. He used to say about
Salwa Higazi, God have mercy on her, that because of her politeness she
would knock on the drawer first before opening it up. For example, I used to
have a small car, a Baby Ford, and Im not a good driver [meaning] so he
used to say that karro carts would go beside me and sound voom. [laughter] to this extent [meaning]. So all his get-togethers were fun, humor, tricks,
and play with words but he was a father, a brother, a mother and the best of
friends and there is no one like him in our journalistic life.
Unlike the two previous segments, fush here comes at the end as a
sequence, a sentence or part thereof, and is not intermingled with the
earlier narrative. It serves as a conclusion to Mansours narrative on
Kamel El Shenawi and his recollection of his sense of humor. The conclusion, that there is no one like him, is what Mansour asserts and wants
the audience to remember. The code-switch, or style shift, to fush helps
him accomplish this goal. Mansour relies on his knowledge of code differences to set the conclusion apart from previous narration, thereby
signaling the conclusion and highlighting its content. The switch, or
shift, is therefore meaningful at this particular moment and within the
context of the text. This, one might argue, represents the basic difference between hybridity and code-switching. Hybridity creates a speech
sample that resembles a collage of elements from one code or another.
In that sense its a global feature of a text. Code-switching as a conversational and stylistic strategy is meaningful in a specific context within a
text; in this sense its local within a text.
The three segments 6 through 8 from Mansour, like those from Haqqi,
show an increasing shift towards Egyptian, which may also be related
to topic. The shift in topic here is not as strong as it was in the Haqqi
samples. The difference between segment 6 and 7, for example, is not
so much in the topic as it is in locus. Segment 6 addresses the issue of
swimming against the tide on a general, more conceptual level whereas
the question in 7 shifts the locus to the person, to Mansour himself and
his experiences in swimming against the tide. The shift to the personal
tends to be accompanied by an increasing shift towards Egyptian. Likewise, the last segment (8) is located in the personal: reminiscing about
a friend and sharing jokes he used to tell with the audience to demonstrate his sense of humor. Both are discourse functions most appropriately conducted in Egyptian.
418
mushira eid
On the basis of these samples, two conversational styles are identified:

one is fush -based, the other Egyptian-based. The two styles incorporate
features from both varieties and include a significantly large number of
lexical items that are ambiguous or unspecified for language variety.13
The texts produced are often like a collage in which the borders separating one variety from the other are often blurred making the product look more like a hybrid text or discourse rather than one where the
boundaries are maintained and the switch from one to the other variety
is clear. The speech samples from Mansour fall into the first category,
those from Haqqi into the second.
The analysis in this section identified three styles and illustrates some
of their linguistic and discourse features. In the next section I present
a frequency-based quantitative analysis of speech samples from Haqqi
and Mansour to identify overall characteristics of the two styles they
represent.
2.2 Quantitative Analysis
The quantitative analysis is based on speech samples of approximately
the first 10-minutes from each speaker, Haqqi and Mansour. The results
are presented in Table 3.
The overall totals at the bottom of the table show clear differences
between the speakers styles. Haqqis is clearly fush -based (80%) and
Mansours Egyptian-based (74%). Although the two speakers have clear
preferences for different varieties of Arabic, they prefer one to the other
by about the same degree: a ratio for Haqqi of 80% fush to 20% Egyptian and for Mansour 74% Egyptian to 26% fush .
Table 3 also provides a breakdown of the distribution of fush and
Egyptian features according to the selected phonological, syntactic, and
morphological features. The figures from Haqqis speech show a bias
towards fush in all areas, but particularly in the syntactic and morphological features with 96% and 100%, respectively. The highest percentage
of Egyptian features in his speech sample comes from phonology (28%).
Of the four phonological variables used in the analysis, the alternation
between fush interdental fricatives and Egyptian stops and sibilants is
the most significant; it is responsible for 85% of the Egyptian phono-
13
Since Schmidt (1974) very few studies, if any, have addressed the role of the lexicon
in Arabic mixed varieties and problems in the identification of lexical items by variety.

TABLE 3
STYLESHAQQI AND MANSOUR

Fush
Phonological
q~
t ~ t ~ s
d / d ~ d / d ~ z/z
diphthong ~ long vowel
TOTALS
Syntactic
Relative markers
Demonstratives
Negatives
Complementizer
TOTALS
Morphological
Passive (u-i, u-a ~ it-)
Stem IV: a-fal pattern
TOTALS
OVERALL TOTALS
419
Haqqi
Egyptian
Mansour
Fush
Egyptian
70
4
29
20
123
72%
4
20
21
3
48
28%
20
3
2
2
27
19%
61
16
30
10
117
81%
9
12
6
20
47
96%
1
0
1
0
2
4%
0
4
10
5
19
25%
10
15
20
13
58
75%
13
13
26
100%
0
0
0
0%
9
7
16
52%
6
9
15
48%
196
80%
50
20%
62
26%
181
74%
logical features in Haqqis speech. In addition there is a strong difference in the distribution of the voiced and voiceless interdental fricatives
by comparison to their corresponding Egyptian pronunciation (stops
and sibilants). Compare the 29 instances of the fush voiced fricatives to
the 21 instances of Egyptian pronunciation and the 4 instances of fush
voiceless fricatives to the 20 instances of their corresponding Egyptian
pronunciation.
A very different picture emerges from the analysis of Mansours style.
While Egyptian is clearly the dominant variety for him, based on the
distribution of features in all three components (phonology, syntax, and
morphology), the contribution from fush is not as minimal as was the
distribution of features from Egyptian, the less dominant variety, in the
Haqqi sample. This is most striking in the morphology, for example,
where the distribution of fush and Egyptian features in Mansours
sample is almost evenly split (52% to 48%, respectively); in Haqqis
sample morphological features of the less dominant (Egyptian) variety
420
mushira eid
FIGURE 1
STYLESPERCENTAGES OF MIX HAQQI

MANSOUR
100
80
60
40
20
0
Egyptian
Fush
Egyptian
Fush
Pronology
28
72
81
19
Syntax
96
75
25
Morphology
100
48
52
Haqqi
Mansour
are nonexistent. Likewise, the distribution of syntactic features in Mansours style shows relatively more contribution form the less dominant
( fush ) with 25%; in Haqqis the less dominant (Egyptian) contributes
only 4% of the syntactic features. The phonology in Mansours style is
consistently Egyptian-dominant at 81%, stronger than the phonology of
Haqqis dominant variety at 72%. The strongest contribution from the
non-dominant fush in Mansours sample comes from the /q ~ / alternation, with fush /q/ representing 25% of all instances of this variable.
In Haqqis style the Egyptian sibilant pronunciation of the interdental
fricative accounted for 85% of all Egyptian pronunciation features. A
graphic representation of these distributions is provided in Figure 1.
The focus of the discussion so far has been on linguistic features (of
phonology, morphology, and syntax) that are different in the two varieties, hence serve to distinguish one from the other allowing speakers to
identify a word, a speech segment, or part thereof, as belonging to one
or the other variety. There is naturally a significant amount of overlap
in the three components. This overlap, or shared features, is typically
ignored in discussions of Arabic mixed varieties since similarity can be
assumed in the analysis of same language varieties.
The lexicon is an area perhaps ignored most in analyses of mixed Arabic varieties, partly because of assumed similarity and partly because
421
of difficulties speakers encounter in categorizing lexical items according to variety. Schmidt, for example, argues that speakers of Arabic are
often unable to classify lexical items as belonging to one or the other
variety, fush or Egyptian, and that the difficulty may be explained on
the basis of topic or semantic fields (1974, 6076). When speakers were
asked to provide fush and Egyptian equivalents in four semantic fields
(body parts, foods, arts, and politics), they had more difficulty with the
last two than they did with the first two. Results of this type can have
more than one interpretation. They can be taken to show a complex,
integrated, interconnected system whereby some areas of the lexicon
such as arts and politics are shared and others distinctevidence for
complementarity typical of subparts of a whole. They can also be interpreted as pointing to areas of deficiency, or incompleteness, in one or
the other language system and used to argue for the superiority of one
or the other variety. Regardless of interpretation, these subsystems of
Arabic are interconnected in ways that are yet to be discovered and satisfactorily explained.
To get a sense of the distribution of lexical items by variety and the
impact the lexicon may have on identification of a base variety, I have
analyzed the lexical items in the 10-minute speech samples of Haqqi
and Mansour based on the presence or absence of clearly marked features of either variety. The absence of such features makes the lexical
item unspecified for variety, or ambiguous. As explained earlier, hybrid
words would have features from both.14 Table 4 provides the overall distribution of lexical items from the 10-minutes speech samples of Haqqi
and Mansour according to the three categories: fush , Egyptian, and
unspecified.
The category unspecified refers to words (content morphemes) and
particles (grammatical morphemes)15 that cannot be distinguished as to
variety and can therefore be considered as ambiguous or shared items.
The results support the idea of a dominant variety underlying diglossicbased stylistic variation. Haqqis lexicon is predominantly fush (74%)
with a smaller 23% of his lexical items unspecified, leaving a minimal 3%
marked as Egyptian. Mansours lexicon is unexpectedly only 45% marked
as Egyptian; but he has an almost equal amount (41%) unspecified and
14
The analysis of hybrid words in the speech samples of Haqqi and Mansour has not
been completed.
15
I adopt McCarthys and Princes (1990) definition of a minimal word in Arabic as
consisting of at least two moras.
422
mushira eid
TABLE 4
THE LEXICONPERCENTAGE OF MIX

Haqqi
Egyptian
Fush
Unspecified
TOTAL
25 (3%)
683 (74%)
214 (23%)
922 (100%)
Mansour
391
119
354
867
(45%)
(14%)
(41%)
(100%)
smaller but sizeable percentage (14%) marked as fush . The major difference between the two styles is the proportion of the lexicon allotted to the predominant variety by comparison to the unspecified lexical
items and to those marked for the non-dominant variety. In Mansours
style the discrepancy between the dominant and non-dominant variety
is not as marked, partly because the contribution from the unspecified
component of the lexicon is stronger than it is in Haqqis style (41% to
23%, respectively). And while the proportion of lexical items from the
dominant and non-dominant is higher in both styles, the difference is
much larger in Haqqis sample than it is in Mansours, 74% to 3% in the
former and 45% to 14% in the latter. A graphic representation of these
results is provided in Figure 2.
The graphic view in Figure 2 suggests a relatively more balanced
distribution in Mansours style than in Haqqis. The graph representing Haqqis style looks as though one component is suppressed (Egyptian), another is allowed some presence (the unspecified), but the third
is asserted as an overpowering presence ( fush ). In Mansours there is
not one single category that is overpowering, but there is one ( fush )
less represented than the othersan unequal participant perhaps.
The graphic view also suggests that the linguistic boundaries are more
clearly defined in Haqqis style than they are in Mansours. For Haqqi
the domain is no doubt fush ; Egyptian may creep in only slightly, if
necessary. For Mansour, the domain is more inclusive.
The ambiguous, or unspecified, lexicon serves to mediate the difference between the two varieties. It creates a shared, or an in-between
space, consistent with the two distinct codes. In doing so, it contributes
to this linguistic collage by allowing smooth transitions from one code
to the other, thereby blurring the distinction between the two. Hybrid
forms, I would add, serve a similar purpose: they can be heard and
interpreted as one, the other, or both varieties. The result is a style that
does not sound too colloquial (dialectal) or too literate ( fush )a
balancing act that allows each speaker to accommodate the situation

FIGURE 2
423
THE LEXICONPERCENTAGE OF MIX
80
60
40
20
0
Haqqi
Mansour
Egyptian
45
Fusha
74
14
Both
23
41
and create personas and identities that are sufficiently separate yet similar enough to be still viewed as one.
The question remains as to why these two speakers, both highly educated, intellectuals and writers, would choose styles so different from
each other in what appears to be very similar context: one conforming to a great extent to the predominant (official) discourse of separation ( fush for public, formal contexts) and the other nonconforming
through its extensive use of Egyptian. The answer to this question
requires a more detailed analysis of the interviews themselves including contexts and topics of conversation as well as speakers purpose in
communication.
3. The Interviews
An interview is a conversation or discussion between two participants
with the purpose of gathering information by the interviewer about an
individual, an institution, a topic, or some other issue. Since the interviewee agrees to participate, s/he is expected by the interviewer to be
forthcoming and cooperative in this conversation. The interviewer,
having initiated the event, is likely to have his/her agenda as well. Both
have something to gain from participation in the event, thus a motive
for a successful performance.
An interview may be private or public and formal or informal,
depending on the purpose of the interview. Participants are expected to
424
mushira eid
follow certain rules, usually set ahead of time, regarding procedure and
at times topic as well. An interview on radio or television is in addition a
public performance intended to create the semblance of a conversation
whose purpose is to convey information to the audiences and to entertain them as well. The ultimate success or failure of such an interview
depends on how the performance is staged and how participants interact.
In this section I describe the two interviews selected for this study
in terms of three components that define an interview: set-up, information, and interaction. I relate these aspects of the interviews to the
styles identified above and to speakers linguistic choices and purpose in
communication.
3.1
The interviews compared
The two interviews are similar in set-up and type of information solicited and conveyed. They differ primarily in forms of interaction and
interviewees purpose in conversation, more specifically in the identities, or persona, they negotiate and attempt to construct through the
interview. The interviewees, Yehya Haqqi and Anis Mansour, are both
famous men of letters, very much involved in the literary and cultural
scenes of their time. Yehya Haqqi (19051992) is a literary critic, essayist and short story writer. He had tremendous influence on several generations of modern Arab writers and is considered one of the fathers
of modern Arab culture. Many of his works have been translated into
several foreign languages and are being taught in many academic institutions, especially his masterpiece Qindl Umm Hashim The Lantern
of Umm Hashim. Anis Mansour is a journalist, an essayist, and a story
writer. He is known for his daily column in al-Ahram mawqif Positions/Opinions and has written numerous books including his famous
h awla l-lam fi 200 yawm Around the world in 200 days. Both interviews are incomplete, with the early part missing from each. The amount
of recorded time available from them, however, is comparable: approximately the last 3035 minutes of each program.
3.1.1
Set-up
Both interviews take place on-location outside the studio in the homes
of the interviewees. In the Mansour program, the interview is conducted
outdoors by the swimming pool with the interviewer and interviewee
seated across a small table. In the Haqqi program, the interview takes
425
place inside his home. Some scenes are shot in his study, others in his
living room, around the dining room table, or in the entry hall.
3.1.2
Format
The Mansour interview adopts a question-answer format and is hosted

by Sanaa Mansour, an experienced interviewer and host of numerous programs on the ESC including Good Morning, Egypt.16 In the
Mansour program she performs the interviewer role, asking the questions and directing the flow of conversation through topic introduction,
follow-up, commentary, and interruption where necessary. From this
perspective, the format of this interview is the conventional interviewer-interviewee. But it actually is far from being so, as will be shown
below.
In the Haqqi program there is no interviewer on the screen. The
questions are not aired, only responses and comments. The program
also includes other participants: Haqqis daughter Nuha, his French wife,
and two of his colleagues, Fouad Duwara and Mohammad Rumeish; the
former is a critic and the latter a writer, as per information projected on
the screen. Both are friends and colleagues who had worked with Haqqi
during his term as editor of al-Magalla.
Despite the difference in format (one-to-one vs. multiple participation), one participant in each program performs the role of commentator and topic introducer. In the Mansour program this role is performed
by the interviewer, Sanaa Mansour, and in the Haqqi program by his
colleague, Fouad Duwara. In performing these roles they read short segments, thus illustrating the non-conversational reading style.
3.1.3
Textuality
Both programs rely on the visual, for entertainment and for reinforcement of information. Images related to the topics under discussion are
often projected on the screen during the conversations. In the Haqqi
program, for example, close ups of paintings appear as Haqqi talks
about gifts he received during his visit to China as director of an Egyptian film festival and about the significance of animals in Chinese art as
illustrated in the paintings. Likewise, pictures of the movie stars Fatin
16
To my knowledge, interviewer and interviewee are not related despite the same
last name.
426
mushira eid
Hamama and Omar Sharif are projected on the screen as Haqqi talks
about Fatin Hamamas popularity among Muslims in China. Pictures of
Yehya Haqqi himself, sometimes a younger Haqqi, also appear on the
screen showing him standing by a car in Paris, for example, as Rumeish
narrates a story about Haqqis trip to Paris to undergo a surgery.
In the Mansour interview, pictures of his books are projected on
the screen when a specific book is mentioned as happened with h awla
l-lam fi 200 yawm (Around the World in 200 Days); likewise images
of newspaper articles or quotes appear on the screen when mentioned
in the conversation. Each segment begins with scenes illustrating the
topic to be discussed with the interviewers voice-over introducing the
segment. To introduce the last segment on the intellectuals and the cafs
they frequented, for example, scenes of Cairo streets with various shops
were shown during the introduction. Later in this segment, scenes of
the Brazilian coffee shop, A. Mansours favorite, were shown when the
interviewer mentioned it in her questions. Pictures of Kamel El Shenawi,
Taha Hussein, Tawfiq al-Hakim, and Abbas al-Aqqad also appear at different points during the program as their names are mentioned.
This interweaving of materials from different text typesvisual, musical, writtenmakes both interviews multi-textual. It serves to create a
collage bringing into the main text of the interview other sub-texts and
sub-effects that contribute to the voice or voices represented in each
interview and ultimately influence the style speakers choose to adopt
when speaking in their voice and the voices of others.
3.1.4
Topics
Both interviews can be described as reminiscences, reflections on the

lives and times of these two professionals. Consequently the topics in
both interviews revolve around the personal and the professional.
In the 35 minutes available from the Mansour interview the discussion is divided into four distinct segments. The first segment is personal
dealing with his views on such matters as having children and the negative stance he has taken towards women. The second revolves around
his life experiences as a journalist/writer, including reflections on his
relationship with leading intellectuals of his time (Kamel El Shenawi,
Taha Hussein, Abbas al-Aqqad, and Tawfiq al-Hakim). The third is
focused on his hectic (perhaps antagonistic) relationship with Abbas
al-Aqqad, and the fourth on cafs frequented by various intellectuals
including himself, a phenomenon the interviewer refers to as taqfat
a-ri (Street Culture).
427
Likewise, in the 32 minutes available from the Haqqi program, the

topics are focused on three areas: Haqqis artistic interests and his views
on literature and art (which for him transcend the literary and include
fine arts and music), his work as editor in chief of al-Magalla and as
government official in the Ministry of Culture, and his daily routines
and personal life as father and husband. These areas and the specific
topics they include are not as clearly bounded in the Haqqi program as
they are in the Mansour interview. Segments dealing with the personal
and private are interspersed with those dealing with the professional
and public. The sequencing in the first part of the program, for example,
illustrates this point. It starts with Haqqi talking about the short story
and the essay, the former being an art form very much like a painting.
Fouad Duawar follows commenting on Haqqis contribution to the performing and fine arts. Haqqi follows to talk about what art means to
him. The topic then shifts to the personal through Nuha, Haqqis daughter, who talks about him as a speaker and a writer. Haqqi follows to talk
about subject matter and description in his work with comments on the
richness of the Arabic language, meaning fush . A switch to the personal follows again from his daughter and then his wife, and so on. The
segments flow one into the other and are not typically introduced with a
reading segment, for example, as is the case in the Mansour program.
This difference in topic organization and shift can be attributed to
production decisions on program set-up and format. But it has implications for the interviews and the speech styles chosen by the speakers. It
may, for example, have reduced, even eliminated, the on-screen interaction among the participants allowing, for example, Haqqi and Duwara
to maintain their predominant fush style.17 The one-to-one set up in the
Mansour program allows for interaction and forms of discourse, such as
interruptions and corrections, that generate spontaneous speech, which
in turn triggers the more casual, conversational style which for Arabic
is dialect-based.
3.1.5
Voice
A major difference between the two interviews involves the issue of

voice: who speaks about what and in whose voice. Despite the difference in set up noted above, the two interviews are multi-voiced, but in
17
Duwara is included here although his speech style, other than reading, is not discussed in this article.
428
mushira eid
different ways. We hear Haqqi and Mansour speak in their voice, both
physically and metaphorically. They speak primarily in the I, not the
we, since neither is there to represent a group or organizational identity.
Mansour tends to represent others more so than Haqqi. When he speaks
in the name of the young generation of writers of his time, for example,
he speaks in the we. Haqqi brings in other fush voices by remembering
lines of classical Arabic poetry, for example, and reciting them for the
audience on the screen.
The one-to-one set up in the Mansour interview makes him the only
speaker, the interviewer being the other. Through these two speakers, we
hear other voices as well. The interviewer Sanaa Mansour speaks in her
own voice when, for example, she expresses an opinion, makes a comment, or asks a question. But she also brings into the interview Mansours voice from the past by quoting his public pronouncements and
published views, at times challenging and questioning them, at other
times allowing him to elaborate and narrate. In so doing, she merges
the past and the present thereby adding to the multi-textual collage of
this program a time-based dimension. Although Anis Mansour speaks
in his voice, he brings in the voices of others through his narration. In
the story-telling segments of the interview, for example, he quotes and
reports the speech of others; we hear them but only through Mansours
voice and his perspective.
Through the multi-person set up in the Haqqi program, we hear and
see four other people, in addition to Haqqi, speak about his life and his
accomplishments. Haqqis comments, however, are limited to Haqqi the
professional: his views on the arts and his track record as an editor and
government official. This public persona is reinforced by his two colleagues, who provide their perspectives on Haqqi as a public figure, a
colleague, and a friend in the profession. Projections of Haqqis private
persona are left to his wife and daughter. As a result, the program as a
whole maintains the boundaries between the two personas: they remain
mostly separate as they come in different voices. Perhaps because of this
separation, segments dealing with the personal/private in this interview
are interspersed with those dealing with the professional/public allowing
the two to merge into one unified whole through program production.
3.2
Projections of Identities
Although the identities projected in these interviews fall into the general categories of the personal and the professional, the persona created
429
by the main figure in each of these programs are very different. This difference is communicated through the speakers voice, i.e., the language
that expresses the persona. The style differentiation described above can
be viewed as representations of speakers voices. In this section I discuss
in relatively more detail the persona being created in the interviews,
relating it to topic, speakers voices, and the interview as a whole.
The Haqqi interview maintains the boundaries between the personal and the professional and, perhaps as a result, between fush and
Egyptian. The set-up and production of the program helps maintain
the boundaries by assigning participants in the program different roles
in the construction of Haqqis overall Persona. Haqqi himself addresses
only the professional, either as his own choice or as that of the producer;
his fush -based speech style reflects this choice. Others interviewed in
the program speak to the personal as, for example, his daughter Nuha,
or to both the personal and professional as does his colleague Meleish;
their styles vary accordingly and appear to be closer to Mansours than
they are to Haqqis.
The focus on the professional with Haqqi allows him to express
his views on literature, art, and literary criticism and to reflect on his
accomplishments in the public sphere as an intellectual, a government
official, and editor of a literary journal. He expresses his views on the
short story, for example, as an art form whose primary purpose is to
entertain, or give pleasure to, the reader (imt al-qri) as opposed to
the essay whose purpose is to inform (ilm). The voice we hear is Haqqis
throughout: his views on art, broadly defined to include literature, fine
arts, and performance arts, recollections of his three years administrative experiences as Director of the Arts Department in the Ministry of
Culture, and his role as editor of al-Magalla. His voice is complemented,
and at times reinforced, by other participants. Haqqis reflections on his
role as editor of al-Magalla (that he made lots of friends through this
job) are picked up in both Rumeish and Duwaras comments. They talk,
for example, about Haqqi as being dedicated to his work and supportive
of other writers. Rumeish recalls his first meeting with Haqqi to submit
for publication his first short story, while Duwara comments on Haqqis
dedication as a journal editor and his travels to solicit manuscripts for
the journal.
On the personal, or private, domain Haqqi briefly mentions his
hobby of collecting canes from all over the world. He describes, and
shows the audience, a few of his favorite canes and explains how his
choice is often determined by his mood. His conclusion to this segment
430
mushira eid
relates the practice of using canes to the public domain, explaining how
it used to be a practice among ministers and politicians but is now a
thing of the past. All other aspects of his private persona are left for
others to project. His daugther, Nuha, speaks of Haqqi as a father and a
professional, merging the two identities a little. She gives the audience a
glimpse of Haqqi the writer and the public speaker through her eyes as
his daugther. She describes his eloquence as a speaker and the ease with
which words come out of his mouth when he speaks; but apparently
they dont come as easy when he writes. Nuha describes how more difficult and time-consuming writing is for him, comparing the writing process to a complicated child birth (wilda mutaassira). She projects him
as a kind and nurturing father, describing their promenades in Maadi,
where he would urge her to observe the beauty in nature and to listen
to the birds, and their walks in Paris, where they would frequent art
museums. She contrasts his perspective of himself versus her perspective of him. While Haqqi says he is old at 80, she sees him as a young
man (b) still, more knowledgeable than anyone in her generation. A
more detailed description of his private life is left to his French wife to
construct for the audience. She speaks in French with a voice over in
Arabic, thus adding one more voice to the many voices in the program.
His wife describes their life together, their walks in Maadi and in Paris,
their visits to museums, and his eating habits (eats very little). As these
life events are narrated, we see the two of them at times sitting at the
dinner table, other times in their living room or in his study, and at
times joined by their daughter. His colleague Rumeish provides a combined personal-professional account, for example, in his narration of
how Haqqi cancelled plans for surgery in Paris to return via Libya to be
with the Egyptian people during the 1967 war.
In the Haqqi program, then, the boundaries between personal
and professional identities are maintained as a result of participants
(assigned) roles in this overall construction of Haqqis identities. This
allows Haqqi to speak in his fush -based style and maintain the public
image he has created for himself throughout his career, a supporter of
the fush and its heritage. From this perspective the Mansour interview
is very different.
The Mansour program, unlike Haqqis, does not separate the professional and the personal. The boundaries between them are fuzzy, or
blurred, making the two identities appear merged into one entity, a collage of the personal and the professional created primarily by Mansour
himself with the help of the interviewer. Since this program is set up
431
with Anis Mansour as the only speaker, it becomes his responsibility to

create this hybrid persona, the merger of the personal and the professional. This Mansour does through a speech style in which the boundaries between the two varieties of Arabic are no longer maintained.
The interviewer Sanaa Mansour brings up for discussion positions
Anis Mansour has taken in public on certain issues. While these topics all belong to the public and therefore professional domain, almost
all are handled on a personal level as narrative rather than as intellectual debates on a more abstract level. The segment discussed earlier on
swimming against the tide is one example. Most of the questions are
handled on a concrete, personal experience level allowing Mansour to
engage in the narration of events in his voice but often reporting the
voices of other participants in past events.
In the Mansour interview, the personal is also public. In the first
segment, for example, the interviewer brings up Mansours views on
children, marriage, and women. Most, if not all, these views come as
quotations from, or paraphrases of, Mansours published articles, books,
or other public pronouncements. In the discussion regarding his negative views on women, the conversation at some point centers around
his wife. Despite his opposition to marriage and his negative views on
women, Mansour is in fact married and his wife is a public figure. The
interviewer wonders if his wife is then different from all other women?
Likewise in the discussion related to children, Mansour explains that
although he is opposed to having children (and gives his reasons based
partly on the professional), he does love children: all children in his
family have pictures with Mansour carrying them on his shoulder. Here
the professional justifies the personal and vice versa. The justification he
strongly voices in concluding this segment is based on the professional
and is expressed with a switch to clearly marked fush syntax and morphology of the embedded that-clause: yihimmini "an "akna mufakkiran aw ktiban aw "adban giddan Its very important for me that I
be a thinker or a writer or a man of letters. This is the public persona
he projects of himself, but always together with the personal.
Differences in the interview set-up and the specific topics covered in
the interviews are partly responsible for the overall persona created of
the main figures. The bulk of the responsibility however lies with the
individuals being interviewed. The persona projected by Mansour is
clearly multi-voiced. He plays many roles and speaks in many voices. He
plays his own person as a writer, an intellectual, and a family person. But
he also plays the roles of others in his narratives: Salwa Hegazi, Kamel
432
mushira eid
Elshenawi, El-Aqqad, etc. In his narratives he assumes their roles by

quoting them, for example, directly or indirectly, by telling their jokes
to illustrate their sense of humor, and by reporting on their accomplishments and modes of interaction as well. Throughout the interview his
voice is concrete, personalized, with a clear sense of humor. As a result,
he engages in different forms of discourse, e.g. telling jokes, narration,
and argumentation/persuasion, and takes on different personas, e.g. the
story-teller, the writer, the friend. He projects himself as a person who
values a sense of humor, friendship, nurturing, and being there for others; these are the values he likes in others and praises them for having
such values. Haqqi, on the other hand, plays one role: the professional,
hence public, role of the intellectual, writer-artist, and administrator.
He speaks in one voice, that of this public persona. This single voice is
clearly reflected in his fush -based choice and the discourse of explication and narration. The personal he leaves to others, to his daughter and
his wife who complement this professional persona with the personal,
Haqqi the man: father, partner, and friend.
On some level, then, the two interviews turn out to be similar in topic,
voice, textuality, and overall informal setting. But they are actually very
different in terms of the roles played by the major figures. The overall
purpose in both programs is the creation of a public and private persona for each interviewee. In so doing, the programs inform the audiences about these individuals and entertain them as well. Entertainment
comes partly through the collage effect created during the interviewa
collage that extends to the language itself as reflected in the diversity of
speech styles and the many voices, images, and persona presented in the
programs, but not necessarily created by the two major figures. While
Haqqi plays one role and speaks in one voice, for example, the program
itself is multi-voiced. Likewise, while the Mansour program creates
boundaries separating its segments and topics, the speakers manage
to cross these boundaries merging the personal and the professional
through their conversational interaction and stylistic choices.
4. Conclusion: Hybrid people and hybrid contexts produce

hybrid language
The analysis presented in this article suggests that the media provides
a hybrid context whose purpose is to inform and entertain. Using the
433
interview as a genre that lends itself to this type of hybridity, I have

argued that the language used in this context is also a hybrid produced
by hybrid speakers, the degree of hybridity (mixture) being dependent
on a number of variables including purpose in communication, projections of identity, and the extent to which a speaker is comfortable
with this hybridity. The language variety produced in such contexts is
like a collage of features drawn from three linguistic spaces in Arabic:
one unspecified for variety, one specified for fush , and one specified
for Egyptian. This form of the language, like hybrids in general, can be
viewed as being neither fush nor Egyptian, since it does not conform to
either, or as both, since it includes features of both. I leave this question
open, for who is to say what, for example, a culturally-hybrid, bilingual
person would be? Is an Arab-American, for example, neither Arab nor
American, or is she/he both Arab and American? In concluding this
article I return instead to elaborate on the view of hybridity as a global
feature of text and code-switching as a local feature (meaningful on a
local level).
It is hard, if not impossible, to predict when or why a person would
switch from one language, or variety of a language, to another. A quick
glance at the sample texts included in section 2 provides enough evidence to support this point. The discussion of these texts also shows that
sometimes the best one can do is identify areas (content, topics, criteria)
where people tend to move from one code to another more extensively
than other times. The end result is a performance where language variety
matches program goal as set by the satellite channel, speakers purpose,
topic, and intended audiences. At other times, however, it was possible
to identify a reason, stylistic or otherwise, for the movement form one
variety to another (e.g. Mansours segment 8).
I have used the terms hybridity and code-switching to capture this
difference. Others have used code-mixing instead of hybridity leaving
the latter to be inclusive of both (Hinnenkamp 2003). Either way two
levels or types of analysis are acknowledged: local analysis that focuses
on the specific motives for alternations, providing a rationale of local relevance, and a global analysis that focuses on motives that can be found
in categories such as identity and group membership, providing a rationale of global relevance to the speakers involved. I have used hybridity in relation to global analysis to capture this cultural connection and
relevance to speaker. I hope to show in future research that for Arabic, and possibly other alternations across codes of the same language,
434
mushira eid
the product may be better analyzed in terms of hybridity understood,

as suggested in this article, as a collage with fuzzy boundaries across
subspaces.
5. References
Awad, Luws. 1965. Mudakkirt tlib bita (Memoirs of a Student on a Study Abroad
Mission). Cairo: Muassasat Rz al-Ysuf.
Bakhtin, M.M. 1981. The Dialogic Imagination. Texas: University of Texas Press.
Bassiouney, Reem. 2006. Functions of Code Switching in Egypt. Evidence from monologues. Leiden and Boston: Brill Academic Publishers.
Boussofara-Omar, Naima. 2006a. Diglossia. In Versteegh, Kees et al., eds. Encyclopedia
of Arabic Language and Linguistics, vol. 1, 629637.
. 2006b. Neither Third Language nor Middle Varieties but Arabic Diglossic Switching. Zeitschrift fr Arabische Linguistik/Journal of Arabic Linguistics 45:5580.
. 2004. Diglossia as Zones of Contact in the Media. Al-Arabiyya 37:101130.
Cachia, Pierre. 1990. An Overview of Modern Arabic Literature. Edinburgh: Edinburgh
University Press.
Eid, Mushira. 1982. The Non-Randomness of Diglossic Variation. Glossa. 16:1.
5484.
. 1988. Principles for Switching Between Standard and Egyptian Arabic. Al-Arabiyya 21, 5180.
. 1992. Directionality in Arabic-English Code-Switching. In Aleya Rouchdy, ed.,
The Arabic Language in America: A Sociolinguistic Study of a Growing Bilingual Community. Detroit: Wayne State University Press, 5071.
. 2002. Language is a Choice: Variation in Egyptian Womens Written Discourse.
In Aleya Rouchdy, ed., Language Contact and Language Conflict in Arabic. New York
and London: Routledge Curzon, 203232.
. 2004. Media Performances as Discourse Events. Arabic Media and Public Appearance Forum, Center for the Advanced Study of Language (CASL), University of Maryland, College Park, June 810, 2004.
Haeri, Niloofar. 2003. Sacred Language. Ordinary People. New York: Palgrave Macmillan.
Hinnenkamp, Volker. 2003. Mixed Language Varieties of Migrant Adolescents and
the Discourse of Hybridity. Journal of Multilingual and Multicultural Development
24:1&2.
Mejdell, Gunvor. 2006. Code-Switching. In Versteegh, Kees et al., eds. Encyclopedia of
Arabic Language and Linguistics, vol. 1, 414421.
Muarrafa, Musta f. n.d. Qantara allad kafar. Cairo: Muassasat T ibat al-alwn
al-muttahida.
Schmidt, Richard. 1974. Sociostylistic Variation in Spoken Egyptian Arabic: A re-examination of the concept of diglossia. Ph.D. dissertation, Brown University.
Schulz, David. 1981. Diglossia and Variation in Formal Spoken Arabic in Egypt. Ph.D.
dissertation, University of Wisconsin-Madison.
Versteegh, Kees et al., eds. 2006. Encyclopedia of Arabic Language and Linguistics, vol. 1.
Leiden and Boston: Brill Academic Publishers.
THE USE OF MORPHOLOGICAL PATTERNS IN

ARABIC GRAMMARS OF TURKIC
Robert Ermers1
s-Hertogenbosch
1. Introduction
The use of meaningful (mor)phonological patterns2 (wazn, bunyat, pl.
awzn, abniyat) is very common in Arabic linguistic thinking. In this
contribution we examine the way these patterns were used in Arabic
grammars of Turkic, especially in the works of Ab H ayyn al-Andalus
(d. 745/1345) and Mahmd al-Kar (11th century).
2. Patterns in the Arabic linguistic tradition

In the Arabic grammatical tradition, each word is analyzed in terms of a
root that contains a certain number of base radicals. Most roots contain
three radicals, but there are also roots with two-, four or even five. For example the verb karuma he was kind is built upon the three radicals k-r-m.
In lexicographical works k-r-m is often placed in the same three-radical
cluster as, say, r-k-m and m-k-r. A much later development is the alphabetical order in which k-r-m follows, e.g., k-r-, which starts with Ibn Fris
(d. 390/1000) (cf. Wild 1965, 66). Within a given entry the radical patterns are further ordered according to the vowels they contain.
In order to make the vowels stand out more, the Arab scholars use the
paradigmatical root /f--l/ (cf. Wright 1986), which in our example karuma yields faula. In this case faula is considered the wazn. In the lexicographical sequence the pattern faula follows faala. For Arabic words
this system based on abstract patterns makes much sense. Verbs that
have the same pattern usually possess similar qualities, e.g. transitivity ;
verbs of the type faula are among other things usually intransitive and
1
2
[email protected], P.O. Box 2176, 5202 CD, s-Hertogenbosch, Netherlands.

For the sake of brevity here referred to as morphological patterns.
436
robert ermers
most of them describe a state of mind or a quality (cf. e.g. Wright 1986,
I, 30, see also Versteegh 1992). Verbs with the same verbal pattern have
similar predictable morphological derivations, such as the formation of
the verbal noun masdar. For verbs of the type faula one possible form
of the masdar is falat e.g. karmat kindness. In regard to nouns of
the same pattern the forming of the plural is predictable in the same
way, the pattern filat may have the plural form fawil, e.g. qidatpl.
qawid base. Frb (d. 350/961) writes in his Dwn al-Adab that plural forms, derived from a pattern which usually yields such plurals, are
not included as entries (I:87). Also more or less predictable is the fact
that the pattern falat may refer to abstract nouns which are not usually
pluralized (similar examples in Irtif I 7397). In this respect the grammarians also indicate augmented radicals in Arabic patterns, e.g. af ala,
iftaala, istaf ala, in which -, -t-, and -stare augmented radicals which
add a specific meaning to the root.
Another obvious motive for collecting words in the same pattern, is
that this is handy when writing poetry. Words with the same pattern can
easily be used in rhyme schemes (cf. also Wild 1965, 66).
The use of morphological patterns thus appears to have two basic
objectives. The first is to determine which consonants in a given word are
basic and which are augmented, or more precisely: to determine which
ones are the basic radicals in a given word. This is for instance important
in establishing a words etymology. In the second place the pattern is
used for illustrating paradigmatic patterns, such as verbal conjugation,
the building of regular plural forms and some types of declension, in
which consonants (e.g. /w/ in /-uwna/ [m pl] and alif /"/ in /-a"t/ t [f
pl] are assigned special meanings (cf. Versteegh 1985). The scope of this
article is limited to the first objective.
For Turkic the general advantage of bringing together words with
similar patterns is evident too, but the morph(ono)logical arguments
do not apply. In Turkic languages meaningful elements usually have the
form of suffixes to a given stem, never infixes or prefixes, e.g. kas cut!,
kas-d he cut (3sg pt), kas-i-d they cut together, and qul slave
qul-lar slaves, qul-juq little slave, qul-juq-lar little slaves, etc. From a
given vowel sequence or consonant pattern no predictions as to plural
and verbal conjugations can be made and the same is true for words that
bear certain suffixes.3 In Arabic grammars of Turkic languages, some
authors make reference to morphological patterns too.
3
If we disregard the consonant assimilation and vowel harmony, that is.
morphological patterns in arabic grammars of turkic 437

3. Two Arabic treatises on Turkic
In the Arabic linguistic tradition the meaningful patterns are the basis
for the arrangement of lexicographical dictionaries. One important
work in this respect is Frbs (d. 350/961) lexicographical work Dwn
al-Adab (cf. Wild 1965, Haywood 1965). The pattern is also used in
Arabic grammars of Turkic. Especially, the Dwn Lut at-Turk (henceforth Dwn) compiled by the 11th century scholar Mahmd al-Kar
is set up in the same way as Frbs work (cf. Ermers 1999a, 19).
On top of the division in patterns, Kar, like Frb, uses a primary categorization of eight chapters (each of which is named Kitb
book), the final two (i.e. 7 and 8) being additions to Frbs division.
This superdivision reflects a primary interest in a morphological and
perhaps an orthographical arrangement. One item that is reflected in
the headings is the position of the glides, the socalled weak consonants,
and other non-regular consonants, such as hamza ().
The division of Dwn is as follows:
1. Kitb al-Hamz (29160): words with initial hamza,
2. Kitb as-Slim (160406): words having all sound consonants,
3. Kitb al-Mudaf (406445): words containing a geminate consonant
or two identical consonants,
4. Kitb al-Mitl (445493): words having an initial weak consonant,
i.e. y or w,
5. Kitb D awt at-talta (493535): words having a medial weak consonant, i.e. y, w or alif,
6. Kitb D awt al-arbaa (535599): words having a final weak consonant,
7. Kitb al-unna (599622): words containing [] or [n],
8. Kitb al-Jam bayn as-skinayn (622638): words containing clusters
of consonantal sounds that do not exist in Arabic, e.g. sirtl- climb
(635), td- turn down (633:15).
In this last chapter of his book Kar must have had in mind the cluster
/-ydt-/ in /tiydtiy/ he turned down, which includes the suffix -t for 3sg
past tense, which is the way Turkic verbs are given throughout Dwn.
Each chapter is further subdivided in separate sections on nouns and
verbs respectively. The more than 6700 entries are further collected in
minor paragraphs, often under a heading that contains a given morphological pattern (cf. Dankoff and Kelly 19825, Auezova 2005). Kar in
438
robert ermers
total lists some 109 main patterns, some of which contain subdivisions,
in total 143 patterns (see scheme 4 in appendix). The number of patterns
in Frbs Dwn al-Adab is much higher, about 1677.
Another scholar who deserves mention in this respect is Ab H ayyn
al-Andalus (d. 1345), who in his Kitb al-Idrk li-Lisn al-Atrk lists a
large number of meaningful morphological patterns according to which
in his view Turkic words are being construed. He mentions 91 nominal
patterns and 44 verbal ones (Idrk 101:10104:16), a total of 135. One
reason which makes it interesting to have a close look at Ab H ayyns
views on Turkic is that they can easily be compared to those on Arabic
through his oeuvre on Arabic grammar and Quranic exegesis. One of
his most prominent works on grammar is Irtif ad-D arab min Lisn
al-Arab, whose setup bears a strong resemblance with Idrk. In Irtif
(I:2072) there is a similar, but much more lengthy and elaborate section on morphological patterns.
4. The application of morphological patterns

In his classification of verbal and nominal roots (asl) in Idrk, Ab
H ayyn differentiates between theoretically possible patterns (qisma
aqliyya) and those that are actually used (masm).4
In Idrk (101:13) Ab H ayyn writes that for biradical nouns there
are 12 theoretically possible forms, all of which are used. These are the
following: (1) fa like san you, (2) fi like kim who, (3) fu like yuz face,
(4) fau, (5) faa, (6) fai, (7) fuu, (8) fii, (9) fua, (10) fia, (11) fiu, (12)
fui. Obviously these can be divided into four groups of three sequences
of vowel patterns each, while after some reshuffling a more transparent
sequence results:
CVC
CaCV
CuCV
CiCV
SCHEME 1
(1) fa
(5) faa
(9) fua
(10) fia
(2) fi
(6) fai
(12) fui
(8) fii
(3) fu
(4) fau
(7) fuu
(11) fiu
PATTERNS OF BI-RADICAL NOUNS IN IDRK
4
Wild (1965, 37) mentions a similar distinction in Kitb al-Ayn between mustamal
used and muhmal not used, litt. neglected.

It is difficult to tell whether Ab H ayyn has a preference for a certain
sequence of the vowels. The regular sequence, though is a, u, i, since in
Arabic grammar a is considered lighter than u and i. The first three
items are listed in the sequence a-i-u, but Ab H ayyn refrains from
holding on to this from the second set onwards.
For nouns5 the following patterns are given:
CVC
(1) fal,
back
(3) ful,
kurt tree
(2) fil,
sirt back
(in Arabic
CVCVC V
analysis CVCVCVC)
(7) faal,
(9) faal,
(8) faal
(16) fuul,
(17) fual
(18) fial
CVCVC
(19) faal,
(21) faul,
(20) fail
(22) fual,
(23) fuul
(25) fial,
(26) fiul,
(24) fiil
(in Arabic
CVCC V
analysis: CVCCVC)
(4) fal,
(6) fal,
(5) fal
(13) ful,
(14) ful,
(15) ful
(10) fil,
(12) fil,
(11) fil
"ard
SCHEME 2
PATTERNS OF TRI-RADICAL NOUNS IN IDRK
In scheme 1 and 2 Ab H ayyn describes nouns, the verbs being dealt
with at the end of the section (103:13). Note that in the second listing (cf.
scheme 2) Ab H ayyn is more consequent in holding to the sequence
a-i-u, which seems to evolve, than in the first. Further, it seems that
some of the patterns are incomplete, for example, the nominal pattern
fu-: there is no form fuil* which completes the set (16, 17). Likewise,
there is only one nominal triradical pattern that starts with fi . . . (18),
where one would expect three, the other two being fiul* and fial*. In
the third place, there is no noun that is formed according the pattern
fuil* (22,23). In sum, only four patterns appear to be lacking from this
overview. Nevertheless, it is easy to invent more patterns which would
fit in this triradical scheme, e.g. fual*, fual* and fual*.
Between parentheses the original sequence in Idrk.
440
robert ermers
For verbs Ab H ayyn gives the following patterns:

Number of radicals
fa-
fu-
fi-
Uni-radical
CV
(2) fa
(3) fu
(1) fi
(5) fa
(8) faa
(13) fai
(4) fu
(12) fua
(7) fuu,
(11) fui
(6) fi
(10) fia
(9) fii
(14) fal
(17) faal
(19) fail
(18) faul
(24) fala
(27) fali
(32) faala
(16) ful
(22) fual
(20) fuul
(15) fil
(23) fial
(21) fiil
(26) fula
(28) fuli
(31) fuila
(25) fila
(35) falal
(34) falil
(33) falul
(39) falula
(40) falala
(41) faalla
(42) failla
(37) fulal
(36) fulul
(38) filil
Bi-radical
CVC
CVCV
Tri-radical
CVCC
CVCVC
CVCCV
CVCVCV
Tetra-radical
CVCCVC
CVCCVCV
CVCVCCV
Penta-radical
CVCVCCVC
CVCCVCCV
CVCCVCVC
(43) fulila
(44) fualla
(45) fuulla
(47) faallal
(48) falalla
(50) falulal
SCHEME 3
(29) fiala
(30) fiila
(46) fiilla
(49) fililla
PATTERNS OF TURKIC VERBS IN IDRK
Note that there are no long vowels in the verbs.
5. Primary and augmented radicals

The rules are, at least so it seems, not applied systematically. In the first
place augmented radicals are not reflected in the pattern and conversely,
radicals are presented in such a way that one thinks that they are considered augments. Ab H ayyn dedicates a large section to augmented
radicals (Idrk 111:17116:13), in which he incidentally investigates
the etymology of some compounds (Idrk 103:12), but the augmenta-

tion is not reflected in patterns in the way this is regularly done in the
Arabic linguistic tradition. The nouns and verbs he cites as belonging
to a given pattern are not analysed further in terms of basic radicals.
For example, "usurmaq fart could be analysed as fuul-maq*, since it is
an obvious derivative of the verbal stem "usur-. The same holds for the
noun ta arjuq sack, for which the pattern faal-juq* could be posited
(Idrk 103:2,5). Instead the former is attributed the pattern fuullal, and
the latter faallul, the regular way of dealing with hexa- and heptaradical
words, but neither of which gives any clues as to the status the radicals
of the respective suffixes possess.
Ab H ayyns formulates his goal in arranging words according to
patterns as follows: It is necessary to study the structures of each of
them, so that the primary radical (al-h arf al-asl) may be distinguished
from the augmented (zid). Only then can the primary radical be compared with the primary radical and the augmented radical with the
augmented radical, (Idrk 104:15, translation Ermers 1999b). In our
introduction we already briefly touched upon the importance of being
able to differentiate between original radicals and augmented ones, e.g.
in af ala, iftaala, istafala, mafalat. The augmented radicals are called
h urf al-man particles of meaning (cf. Versteegh 1995, 120). One
would expect the Turkic augmented radicals in a similar way to be indicated in the pattern, e.g. faal-dir* causative, faal-il* passive, faal-in*
reflexive form.
Kar takes a similar position in his Dwn. Much in the same way
Ab H ayyn does in Idrk he discusses in a separate section the augmented radicals for nouns and verbs (Dwn 1316). In the nouns he
deals with the glides and hamza only, e.g. tar bag, "adir stallion,
along with -n, as in bazn /baza"n/ hammer, in which the added
meaning of the consonant alif /"/ is not obvious. The augments in the
verbs are dealt with in more detail. Kar mentions as many as eleven
augmented consonants, i.e. 1. alif, 2. t, 3. r, 4. s, 5. , 6. q, 7. k, 8. l, 9. n, 10.
l (lm-alif), 11. y, all of which are added because of [certain] meanings ( fa-kull wh idin minh tuzd li-man, Dwn 14:2). Here we
suffice with a few examples, e.g. t in its causative meaning for the transitivity of the verb (li-t-tadiyat al-fil, 14:5) in "ari-t-t he dried, which
basic form is (asluhu) "ar-d it dried up. Another example is s, which,
according to Kar, expresses the meaning of wishing to carry out
that action/verb (man at-tamman li-iqmat dlik al-fil, 14:10 ), e.g.
suv "ij-a-sa-d he wished to drink ("ij-) water (suv). The -s- is also used
when the action of the verb is not actual (wa-lam yakun wuq dlika
442
robert ermers
al-fil minhu h aqqatan), e.g. "ul mank kul-um-sin-d ([uriya annahu]

yadh aku ilayya) [it seemed as if] he smiled at me (in which kul- means
laugh and kul-um-sin smile).
Even though patterns play a crucial role in Dwn, only very occasionally the augmented radicals are made visible, to the extent that even
the ones dealt with in this section of the work are never referred to.
The most obvious example of a suffix that is visible, is -d, the 3sg past
tense suffix that Kar adds to verbs, e.g. faal-d (Dwn 305). The
use of this form is most likely a calque from the Arabic grammatical
tradition, where it is the form that is most unmarked (like karuma, lit.
he was kind), and one could therefore argue that /-diy/ is not really
considered an augmentin any case it is not mentioned in his listing
(Dwn 135).6 A more interesting example is the pattern fiinl (see no.
13 in scheme 4, in appendix, Dwn 82), which contains nouns like, e.g.
itind what is being pushed, and aqind (suv) running (water). Based
on this patternif we first disregard a in aq-f- stands for , and -- for
t/q, the stems are reconstructed as it- push and aq- run (of water)
respectively, both of which are common in Turkic, with a suffix -nd. But
in regard to the augments itself more can be said.
pattern
Turkic 1
Turkic 2
i
i
a
t
q
i
i
i
n
n
n
l
d
d
i
i
i
y
y
y
All examples Kar supplies in this subsection end in -ind, which in

spite of the categorization as triradical, suggests that in his view /-l-/,
which then on a metalevel stands for -d-, is part of a triradical root and
that only -n- and - are augments. If /-l-/ is nevertheless considered a
literal part of the augment, he still has to explain under which phonological circumstances /-nl-/ is realised as -nd. In Arabic grammar, the
phonological changes in consonants are typically explained by means of
qalb, the exchange of one consonant for another, based on a set of rules,
rather than insertion of meaningful consonants. In any case, a pattern of
the type find* would have been a more logical choice.
The patterns falal /fal1al2/ and fanal (Dwn 611) occur in one heading under a section of tetraradical nouns in the book of Nasal Words.
This heading introduces nouns like tauj /tankuj/ something which
6
Perhaps for this reason Ab H ayyn does not include -d in his references to Turkic verbs; he applies the patterns to the stem only, e.g. fu for "ub kiss!, fal for tart for
weigh!.

rises in the air (EDT7 [520] te), and qujr /qujnka"r/ ram. In
these nouns , which in both words is represented by /-nk-/, counts for
two consonants, e.g.:8
pattern 1
pattern 2
Turkic
f
f
t
a
a
a

k
l1
n
n
a
a
u
l2
l
j
Other arguments for this reasoning are in the first place the fact that
according to the morphological pattern the augment is /-n-//-k-/
being perhaps more basiceven though Kar does not specify to
which one of both patterns the nouns in fact belong. In the second place
nouns like taik /tankik/ air and saak /sank1ak2/ cup and suuk /
sunk1uk2/ bone (Dwn 604) are considered to have a doubled consonantif we take into consideration the directly preceding heading.
On a phonological level this suggests that /-k-/ is considered a separate
consonant, whereas from a morphological point of view it is probably
regarded as more basic than /-n-/. All this points to the assumption that
n and k are regarded as separate consonants.
In spite of this, nouns that are quite comparable to those above, like
taut (Dwn 603) /tankut/ name of a Turkic tribe and suqur /sunkqur/ falcon, kaa /kanka/ advice are assigned a triradical pattern of
the type faal, faul, fail.
pattern
Turkic
f
t
a
a

nk
u
u
l
t
In those instances /--/ corresponds with the cluster /-nk-/; this nevertheless appears to be a more general morphology for words with .
In Dwn there are a few instances in which the problem is the other
way around: augments are made explicit when there is no apparent need
for it. The preceding case of fanal may represent such an instance, since
there is no apparent augmented meaning. A speculation along these
lines is that, perhaps Kar in fact was in doubt as to which pattern
to use here, so he wrote down both. Other examples are the pattern
faland that is applied to such nouns as udrund selected and afdind
7
EDT, here and henceforth, refers to: Etymological Dictionary of Pre-ThirteenthCentury Turkish (Clauson 1972).
8
Clauson in EDT apparently supposes that, in view of the number of radicals, /-n-/
alone must stand for , whilein his viewa superfluous /-k-/ (for g) is added, which
results in the erroneous form teg.
444
robert ermers
collected (Dwn 84). If we for both nouns take /-l-/ in fal-and as the
final consonant of the stem, which seems to be the case, it follows that
the paradigmal suffix or augment must have the form -nd. The projected stems then must be udr-* and afd-* respectively.9 Here we have
a similar problem as in regard to the pattern fiinl, which we discussed
earlier. If instead Kar considered /-l-/ as the first consonant of the
suffix, one would have expected him to propose the pattern fa-dand*,
because otherwise the occurrence of -d- and -r- on the position of /-l-/
is not accounted for.
Kars discussion of the few internal augments in nouns he gives
is interesting too. Here the augments are placed between the radicals of
the stem, e.g. fawal for yuwlij lambswool (Dwn 456), fayal for nouns
like qaymaj lambswool (Dwn 522), both of which are considered
triradical. In these cases /w/ and /y/ respectively again, are presented
as augments, quite comparable to /y/ in fuayl, a regular Arabic pattern
for diminutives, e.g. fulays small coin (< fils). But the meaning of the
inserted glides here is quite unclear. If the comparison to fuayl would
hold, this suggests that these would be etymologically related to forms
like yulij* and qamaj*, which could not be found. In view of the fact
that the augment in these instances does not seem to carry a particular meaning, one could argue that the pattern faalal, or perhaps faalil
would have served as well, and the question remains why Kar did
not apply them here instead of proposing the insertion of augments.
There seems to have been a general degree of confusion as to the
applicability of morphological patterns in other works as well. Talmon
(1997, 172) in this respect notes that an authoritative source like al-Xall
(d. 791/175?) in his Kitb al-Ayn is not very consequent in the use of
patterns either: muatft, which according to one view takes the pattern (wazn) of mufalt, and according to another mufalat. A similar
example, still according to Talmon, is andawat, which belongs to two
patterns: fanalat and faallwat. Some consonants, especially hamza, alif,
ww, y and nn are sometimes referred to as basic (asl) and others
as non-asl. In Irtif we came across a similar discrepancy, sanbitat a
period of time is assigned the pattern falatat, but Ab H ayyn acknowledges that others apply fanalat instead: it is said that its pattern is . . .
Only in the first case this agrees with EDT [70] dr-, whereas for the second Clauson suggests evdin- pluck [7].

(wa-qla waznuh), obviously depending on how /-n-/ is interpreted. In
another instance here too the arbitrariness in assigning the patterns is
obvious (Irtif I:33). For example, firsan is attributed the pattern filan
(Irtif I:33), even though /-n/ cannot be considered an augmented consonant. In this way, the use of the pattern does not give any clues for
determining which consonant is augmented. In Frbs Dwn al-Adab
similar examples are found, e.g. anbajn be great which follows the
pattern afaln (Dwn al-Adab I:280). In this example the form with
-n obviously is not a derivative from based on a form (anbaj), without
the suffix. Based on this brief survey we may conclude that Kars
approach is not very exceptional.
6. Other lacunae in the system

Earlier we mentioned that Kar uses some 109 headings in which
he mentions patterns, sometimes more than one. The total numbers of
patterns he refers to is 146. Not surprisingly, in Dwn the same pattern
recurs several times through the work. Within the different chapters
the same patterns almost inevitably recur. To mention only a few, af l
(see no. 4 in scheme 4 in the appendix) is the same as 14; 2 is identical
with 29 and 51; 5 with 32 and 60, more examples of similar patterns in
scheme 5 in the appendix).
Let us examine two occurrences of the same pattern. The pattern afl
(see no. 4) refers to "arqr mountain sheep (Dwn 71), but it also refers
to a noun like "ar fishing hook (see no. 14, Dwn 83).
In book I, Kitb al-Hamz, there is no apparent reason for giving any
pattern with initial hamza /-/. The chapter heading itself suggests that
the initial hamza /-/ is basic and thus should be represented by /f-/. The
only reason for quoting hamza could be that it is augmented, i.e. when
it is not a basic consonant. Yet seven (i.e. the items 3, 4, 11, 12, 14, 17
and 23see appendix scheme 4) out of the 24 morphological patterns
begin with -, e.g. (3) af al, (4) afl, uf l, if l, (11) afl, etc. where
one would have expected f- instead. Let us study two examples in more
detail:
A. "arqr 1. Book I, words with initial hamza, 2. nouns 3. vocalization
of middle radical (i.e. q), 4. Section on af l (and other patterns)
with a vowel on the middle radical, 5. R.
446
robert ermers
B. "ar 1. Book I, words with initial hamza, 2. tetraradical nouns, 3.

pattern af l, 4. a double sequence of the same radical within one
root (i.e. ).
B is explicitly placed under the heading of tetraradical words; it then
follows that A is considered triradical. Apart from being explicitly
quoted in the pattern, the initial hamza in A is also the only reasonable
candidate for being considered an augment, which is in contradiction
with the main chapter heading. It is difficult to speculate on the reasoning Kar follows here that would account for these two instances of
analysis.
There is also a small number of words that have a double entry in the
work in different places. One is kuk light shadow (EDT 753). Apparently Kar has interpreted the pattern of this word in different ways:
A. 1. Book II regular words, 2. Nouns, 3. Words with added glide
between the second () and third radical (l), 4. The pattern fal and
the like, 5. (nouns with) K (Dwn 225:8; Auezova 2005, no. 2606)
B. 1. Book V words with a middle weak consonant, 2. [Nouns], 3. Sections on difficult (?) words, 4. The pattern fal and the like (871),
5. (nouns with) K (Dwn 521:12; Auezova 2005, no. 5500)
If we apply the patterns fal /faa"la"/ (1) and fal /faa"luw/ (2) to
kuk /kuiyka"/, in which /f-/ stands for /k/, /--/ for /--/, etc., the following scheme emerges:10
pattern 1
pattern 2
Turkic
f
f
k
a
a
u
a
a
i
''
''
y
l
l
k
a
u
a
''
w
''
The application of either pattern suggests that -y- is not basic, but
inserted as an augment between -- and -k-. The goal here is of course
obvious: lengthening of /i/, but -y- is still not accounted for as a meaningful augment. B is placed in the main division, the Book of words with
a glide as a middle radical out of these.11 If -y- is not considered a meaningful augment, it must be an instance of prosodic lenghtening. This
same Book starts, as expected, with nouns like th th /ta"h/, but they
10
We will discuss below the fact that the vowels of the schemes do not correspond at
all with those in the Turkic, and that /-"-/ is apparently realized as y, and /-w/ as ".
11
If we correctly interpret dawt at-taltat, cf. Dwn al-Adab I 76,80.

are to our surprise considered biradical (t uniyya). In other words:
the alif is, in spite of the heading, not considered a basic radical, but
apparently a mere instrument for indicating lengthening. This suspicion
is reinforced by Kars following remark elsewhere the lenghtening
consonants may be elided when they are pronounced quickly; in that
case one says tah tah (wa-h urf al-ln suqita min inda surat an-nutq
bih, fa-yuqlu tah tah, Dwn 439:12). Thus the status of the glide still
remains, at least in these words, uncertain.
Finally there is the status of the vowels within the pattern. Ab H ayyn
is both in Idrk and Irtif quite accurate in his selection of matching
verbs and nouns to a given pattern, and the same holds for Frb in
his Dwn al-Adab. Whenever Kar presents a given pattern in a
heading, he nevertheless lists almost everywhere words which follow
an entirely different vowel pattern. To give but a few examples: fal is
applied to yij tree which in fact follows fil, yipr musk in fact fil,
yulr rein in fact ful (see item 60 in the scheme below; Dwn 456:3).
These instances are often covered with subtitles like f h araktihi literally
in the vowels (also 84:11) after the given pattern. In this way, patterns
containing /a"/ (indicating ) are also often applied to words with /
uw/ or /iy/, e.g. fal for yant response, which in fact has the pattern
fal (Dwn 456:3). It seems that Kar throughout his work uses the
patterns for indicating the mere position of the vowels and glides rather
than as a precise pattern.
In some instances Kars presentation is in contradiction with the
regular practice. When Kar introduces the pattern faal (cf. Scheme
4, no. 76), explicitly described as with the first and second consonant
vocalized (muh arraka al-f wa-l-ayn), and further specified as from
the weak words which contain a w (dawt al-ww) (Dwn 507:6).
However, w is not mentioned at all; Kar exclusively lists nouns of the
type (CC), such as qb /qa"b/ pot, tz /ta"z/ bald (Dwn 509:6). It is
unclear how Kar applies this pattern here, since alif ("), which must
be the second or middle consonant, cannot carry a vowel; it can only be
preceded by /a/.
7. The glides
The status of the glides is an important issue in morphological patterns. The patterns were a great aid to the Arab grammarians for determining the status of the glides. In Arabic grammatical theory the glides, i.e.
448
robert ermers
/"/ ( alif), /w/ and /y/, are considered consonants. For example, bbun
door is derived from a projected form /ba1wa2bun/* of which the underlying pattern is /fa1a2lun/, by replacement of /w/ by //, which results
in /b1aa2bun/*, and a subsequent elision of /a2/: /ba1bun/. Even though
glides are used as instruments for indicating lengthening of the preceding vowels, they are not considered vowels themselves. In this way, alif
(/"/), whose only function is expressing lengthening of /a/, is a consonant as well. The glides, like a number of other consonants (e.g. /"/,
/n/, /t/, etc.), can be inserted or prefixed as augments. For example, the
verbal form qtala he battled is, according to the Arabic grammatical
tradition, derived from qatala he killed by inserting an alif /"/ according to the pattern /fa"ala/, rather than by lengthening /a/ (cf. Bohas
1982, 168).
This does not mean that the concept of lengthening does not exist. In
a practical sense the concept of long vowel does exist in Arabic grammatical theory, but in the analysis a sharp distinction is made between
meaningful augments on the one hand, i.e. the insertion of phonemes,
and prosodic lengthening, i.e. lengthening for non-phonemical reasons
(ib) on the other. Or, in other words, the analysis of a given word is
based on the question whether a given long vowel is the result of either
ziydat (meaningful addition, or insertion) or of ib (prosodic lengthening). If we think further along the lines of Arabic linguistic reasoning,
there is hardly a plausible reason to give for reflecting prosodic lengthening in orthography. This is especially true for lexicography. Prosodic
lengthening typically occurs when the word is put in a context, not in
isolation and it is made explicit in poetry, not in regular prose.
The question of how to interpret lengthened vowels is also an issue in
our source material on Turkic languages. Ab H ayyn starts his expos
on this problem with the uniradical nouns s water, which he writes as
/suw/, and y /ya"/ bow, j /jiy/, moist (?). His almost casual remark
that the semiconsonants (h urf al-liyn wa-l-madd) [w], ["] and [y] form
no part of the root (asl, 101, 12), but rather arise from lengthening
of the vowels (nawi an ib al-h arakt) now gains importance. In
the case of s, y and j, Ab H ayyn says, not f /fuw/ or y /fa"/ is
intended, but rather fu+ and fa+, respectively (in which we use + for
prosodic lengthening). In this respect qis (qisa") short too must be
ranged under the pattern fia+. In spite of all this, the author still produces the patterns fal for words like barj all, faal for taraz scales
(see the patterns 418 in the listing given above). In these patterns the
glides, which, according to the author, here serve to indicate prosodic

lengthening, might as well be mistakenly regarded as a meaningful part
of the pattern. It seems that Ab H ayyn clearly was in doubt as to how
to reflect the morphological patterns.
The same ambivalence towards long vowels and their morphological
classification is found in Kars Dwn. Kar says that in general it
is possible to delete the glides (Dwn, 516:6) as in Arabic, in the way
zin decorate! is formed by eliding /y/ from the form /ziyn/*. The reason
he says so, no doubt is the fact that in Uygur script vowels are indicated
by means of the glides, without the possibility, as in the Arabic script, to
distinguish between long and short vowels (see Ermers 1999a, 130).
In regard of the previous one is inclined to think that vowel length is
unlikely to be considered a phoneme in Kars lexicon. This is not the
case. Kar takes either the long form or the short as an entry. In some
instances he gives the other as a dialectal exception to the general rule.
In other words, he deals with the short and long forms in separate paradigmal forms. A nice illustration is the pair "aj- open (Dwn 92:4) and
"aj- be hungry (Dwn 95:17). Each of them has a separate entry in the
chapter on the biradical verbswhere // counts as the first consonant
but only "aj- be hungry is dealt with in the subdivision (lit.) Insufficient
(al-manqs). A similar example is "z, which has a number of meanings,
e.g. fat; deaf; self; valley; heart (Dwn 35:1), "z fat (Dwn 30:6) and
the verbs "z- cut (Dwn 93:1), and "z- pass first (Dwn 96:11), which
is also listed under the subheading al-manqs. In the case of "it/"t push/
bless (Dwn 95:10), which occur under the same pattern, lengthening
may have been used to indicate a different pronunciation altogether, e.g.
t- (cf. EDT 36). Even though of one and the same entry more than one
version does occur throughout the work, from the separate treatment it
is evident that for Kar length is a phoneme. One could even make
the assumption that the fact that length can be made explicit in Arabic
script gave Kar the opportunity to arrange his lexicon in a different
way than he could have in the Uygur script.
8. Summary
In this article we made a brief inventory of the way Arabic grammarians applied morphological patterns to Turkic words. In Arabic linguistic thinking, the patterns are a convenient instrument for indicating
which consonants in a word are basic and which are not. Non-basic
450
robert ermers
consonants are usually attributed a special meaning, such as, e.g., causativity or diminutive, which is added to the root. Furthermore, based
on the pattern words that contain glides, such as bbun can easily be
reconstructed in terms of regular patterns (*/bawabun/). An additional
motivation for lexicographical interests appears to have been bringing
together words with the same or a similar pattern. This comes in handy
for writing poetry, since words with the same pattern can be used in the
same rhyme scheme. Surprisingly, the two sources we examined, Diwn
Lut at-Turk (11th century) and Kitb al-Idrk li-Lisn al-Atrk (14th
century) hardly use Turkic meaningful consonants and suffixes in combination with the patterns, even though the authors are clearly aware of
them. This appears to be a general feature of lexicographical works: the
patterns seem to be applied more or less arbitrarily. In Dwn Kar
goes one step further, in that he roughly indicates the position of the
vowels in a given pattern by means of /a/, while at the same time in those
instances he deals with words with u, and i, .
9. Appendix
9.1 An overview of the morphological patterns in Dwn Lut at-Turk
Scheme 4 below contains the patterns mentioned in the chapter and section headings in Dwn. Sometimes more than one morphological pattern is mentioned (e.g. in 1, 2, 4, 9, 28, etc.). The roman numerals refer
to the number of radicals mentioned in the closest preceding heading.

Book I, Kitb al-Hamz (29160): words with initial hamza
Nouns (Dwn 33)
III (1) fal, ful, fil, (2) faal, faul, fail, (3) afal, (4) afl, ufl,
ifl, (5) fal, fal, (6) fal, fal, (7) fal, (8) fuul, (9)
fainl, (10) fal, ful, fil, (11) afl, (12) af aln,
(13) fiinl
IV (14) af l, (15) fal, (16) faall, (17) aflil, (18) fall,
(19) faland, (20) fuln,
V (21) faalal, (22) faaln, (23) afalal, (24) fulil, (25) falil
(Dwn 91)
Verbs
(26) faal (Dwn 97), (27) falal-d
Book II, Kitb as-Slim (160406): words having all sound consonants
Nouns (Dwn 172)
(28) fal, ful, fil, (29) faal, faul, fail, (30) fil, (31) fl,
(32) fal, fal, fal, (33) fal, fal, fal, (34) faln, fuln,
filn, (35) fal, (36) faln, faln, (37) faanl, (38)
fulul, (39) falal, fall, fulal,
V? (40) falil, V? (41) faalal, faall, (42) fulul, (43) fulund,
(44) faalal, (45) faalln,
VI (46) falall, (47) faalalal
Verbs (Dwn 305)
III (48) faal-d
IV (49) falal-d
V (50) faalal-d
Book III, Kitb al-Mudaf (406445): words containing a geminate or two identical
consonants
Nouns
III (51) faal
IV (52) fall
V (53) faalal
Verbs
III (54) faal-d (Dwn 415)
IV (55) falan-d
Book IV, Kitb al-Mitl (445493): words with an initial weak consonant, y or w
Nouns
III? (56) fal, ful, fil (Dwn 447), (57) faal, faul, fail, (58)
falal, (59) fawal, (60) fal, (61) fal, (62) fal, (63) faln,
(64) faln, (65) faanl
IV (66) falal, fall, (67) falil, (68) faall
V (69) faalal, (70) faaln, (71) falalin (Dwn 468)
Verbs
(72) faal-d (Dwn 473)
IV (73) falal-d
V (74) faalal-d (Dwn 491)
Book V, Kitb D awt at-Talta (493535): words with a medial weak consonant, i.e. y,
w or alif
Nouns
(75) fal, ful, fil (Dwn 494), (76) faal, (77) fil, (78) fal,
(79)
fal, (80) fal, (81) faln, (82) fayal, (83) falil
V (84) faalal (Dwn 523)
Verbs
(85) faal-d (Dwn 526), (86) falal-d (Dwn 529)
Book VI, Kitb D awt al-Arbaa (535599): words having a final weak consonant
Nouns
(87) faal, faul, fail (Dwn 540), (88) fil, (89) fal, (90) fal,
(91) fall
VI (92) falal, (93) faalll (Dwn 552)
Verbs
III (94) faal-d (Dwn 565)
IV (95) falal-d, (96) fal-d, fal-d, fal-d
V (97) faal-d, V (98) falal-d (Dwn 597)
452
robert ermers
Book VII, Kitb al-unna (599622): words containing [] or [n]

Nouns
(99) faal, faul, fail (Dwn 602) (100) fal; (101) fal,
(102) fal
IV (103) falal, fanal, (104) falal,
IV (105) faanll (Dwn 613)
Verbs
(106) faal-d (Dwn 615)
IV (107) falal-d, (108) fall-d, (109) fal-d (Dwn 619)
Book VIII, Kitb al-Jam bayn as-skinayn (622638): words containing clusters of
consonantal sounds, does not refer to morphological patterns.
SCHEME 4
MORPHOLOGICAL PATTERNS EXPLICITLY

MENTIONED IN DWN
The patterns mentioned above recur in different places throughout the

work. The scheme below indicates which patterns are repeated:
1
2
3
4
5
6
7
8
9
10
11
12
13
18
16
17
14
19
15
20
SCHEME 5
pattern
afl
faal
fal
fal
falil
fal
faalal
faalal-d
faaln
faal-d
faall
fil
fal, ful, fil
fal
faln
faland
falal
fall
falald
fulul
occurs in items no. in scheme 4

4, 14
2, 29, 51, 57, 76, 87, 99
5, 32, 60, 78, 89, 100
35, 96 ww
40, 67, 83
10, 80, 102
21, 41, 44, 53, 69, 84
50, 74
22, 70
48, 54, 72, 85, 94, 106
41, 68
30, 77, 88
1, 28, 56, 75
6, 33, 61, 79, 90, 101
34, 63, 81
19 (noun), 55 (verb)
35, 58, 66, 103
39, 52, 66, 91, 108 (verb)
27, 49, 73, 86, 95, 107
38, 42
DOUBLE OCCURRENCES OF MORPHOLOGICAL

PATTERNS IN IN DWN

10. References
10.1 Primary sources
al-Andalus, Ab H ayyn, Muhammad b. Ysuf. Irtif ad-D arab min Lisn al-Arab. (3
vol.) Musta f Ahmad an-Namms, ed. 198419871989. (I: Cairo: Matbaat an-Nar
ad-D ahab; II/III: Cairo: Mat baa al-Madan).
Kitb al-Idrk li-Lisn al-Atrk. Amet Caferolu, ed. 1931. Istanbul: Evkaf
Matbaas.
al-Frb, Ab Ibrhm Ishq b. Ibrhm. Dwn al-Adab. Ahmad Muxtr Umar, ed.
197478 (4 vol.) Cairo: al-Haya al-mma.
al-Kar, Mahmd b. al-H usayn b. Muhammad. Diwn Lut at-Turk. Facsimile edition of the MS; Ankara: Turkish Ministry of Culture. 1990.
al-Xall = Ab Abd ar-Rahmn al-Xall b. Ahmad al-Farhid. Kitb al-Ayn. al-Maxzm
and as-Smarr, ed. Beirut: al-Alam.
Auezova, Zifa. M. 2005. Diwan Lughat at-Turk. Translated (into Russian), with introduction by Z.-A. Auezova; indices by R. Ermers. Almaty: Daik Press.
Bohas, Georges. 1982. Contribution ltude de la mthode des grammairiens arabes en
morphologie et en phonologie daprs grammairiens tardifs. Thse Universit de Paris
III (1979).
Clauson, Gerard. 1972. EDT = An Etymological Dictionary of Pre-Thirteenth-Century
Turkish. Oxford.
Dankoff, Robert and Kelly, James. 19825. Compendium of the Turkic Dialects. Translation of Mah mud al-Kars Dwn Lut at-Turk. 3 Vol.: I (1982); II (1984); III
(1985) indices to Vol. I and II. Harvard: University Press.
Ermers, Robert. 1999a. Arabic Grammars of Turkic. The Arabic Linguistic Model Applied
to Foreign Languages. Leiden: E.J. Brill.
. 1999b. Translation of Ab H ayyn al-Andaluss Kitb al-Idrk li-Lisn al-Atrk.
Leiden: E.J. Brill.
Haywood, John 1965. Arabic Lexicography. Leiden: E.J. Brill.
Owens, Jonathan. 1988. The Foundations of Grammar: An introduction to medieval Arabic grammatical theory. Amsterdam: Benjamins
Talmon, Rafael. 1997. Arabic Grammar in its Formative Age. Kitb al-Ayn and its attribution to H all b. Ah mad. Leiden, New York, Kln: Brill.
Versteegh, Kees. 1985. The development of argumentation in Arabic grammar: The
declension of the dual and the plural. Studies in the History of Arabic Grammar, ed. by
Hartmut Bobzin and Kees Versteegh, 152173. Wiesbaden: O. Harrassowitz.
. 1992. Grammar and rhetoric: Jurjn on the verbs of admiration. Studies in Semitic
Linguistics in honor of Joshua Blau, M. Bar-Asher et al., eds. Jerusalem: Hebrew University of Jerusalem. 113133.
. 1995. The explanation of Linguistic Causes. Az-Zajjjs theory of grammar. Introduction, translation, commentary. Amsterdam: Benjamins.
Wild, Stefan. 1965. Das Kitb al-Ain und die arabische Lexikographie. Wiesbaden: Harrassowitz.
Wright, William. 1986. A Grammar of the Arabic Language. Translated from the German
of Caspari. Cambridge: University Press. (2 vols) (First edited in 1859/1862).
LEXICAL GAPS IN ARABIC LEXICOGRAPHY WITH

EVIDENCE FROM ARABIC DICTIONARIES
Jan Hoogland
Nijmegen University
1. Introduction
Between 1997 and 2003 we compiled with a dedicated group of specialists
at the Arabic department of the University of Nijmegen a twin set of
Arabic-Dutch and Dutch-Arabic dictionaries. Kees Versteegh, Manfred
Woidich, and myself were the responsible editors.1
It may be clear, that the compilation of these dictionaries not only
resulted in their publication, but also in the gathering of a lot of valuable
experience leading to some striking conclusions. One of these was the
observation that, in the Dutch-Arabic volume, many source (Dutch)
language units could not be paired with an equivalent in Arabic, since
they appeared to be non-existent in Arabic. As a matter of fact, almost
25% of all Dutch entries (words or expressions) could only be translated
by means of paraphrases. Paraphrased descriptions are, in most cases,
explanations and not direct equivalents of the source language word or
expression.
From this observation the question did arise: do these untranslatable
units represent lexical gaps in Arabic? In other words: do they represent
concepts without any lexicographically acceptable one-to-one translation
in Arabic?
It goes without saying that all dictionary compilers also benefit from
the work of others. Also the authors of the Nijmegen dictionary referred
to a large number of existing monolingual, bilingual and multilingual
dictionaries in order to single out adequate equivalents for the Dutch
entries selected to occur in the dictionary. Too often we concluded
that the other dictionaries also contained paraphrased descriptions or
definitions instead of one-to-one translations.
1
The whole process of compiling has been described on the project site (www.let.
ru.nl/wba).
456
jan hoogland
Upon this observation we came to the following hypothesis: when

a certain entry is translated into Arabic by means of a paraphrased
description or definition while all, or almost all, bilingual dictionaries,
included in this enquiry, translate the same entry in the form of a
description or definition, it is justified to conclude, that the entry
concerned does not have a one-to-one equivalent in Arabic, and that we,
therefore, are dealing with a lexical gap in Arabic. In order to verify this
hypothesis, a large sample of entries from the Dutch-Arabic dictionary,
with one-to-one source equivalents in various other dictionaries, was
gathered to compare their translations into Arabic.
The source language units can be divided in simple words (lexical units),
compound words and expressions. Compounds in Dutch are simply
written together as one word and consequently have to be entered as
independent lemmas in the dictionary.
This article limits a comparison of translations for simple words
from various bilingual dictionaries, since it turned out difficult to find
unambiguous translations in English, French or German for Dutch
compounds. The lack of those translations as a starting point for a
comparison of Arabic translations would result in data that would be
difficult to compare with each other.
2. Definition of description
As stated in the introduction, descriptions are, in most cases, not direct
equivalents of the word or expression in the source language, but
explanations, i.e. expressions of more than one word which provide
the dictionary user with an explanation in the target language. Since a
description is a combination that is not lexicalized, it is not entered in
the database as an expression in the target language.
Descriptions represent a unidirectional translation relation, i.e. they
are not included in the reversion process to produce the reverse part
of the dictionary. Assuming the description describes a concept that
represents a lexical gap in the target language, it is obvious this concept
could not be entered in the reverse part of the dictionary as an entry.
First of all, a sample of 25 Dutch simple words translated with
descriptions was gathered. These words were chosen randomly from the
underlying database. These words can be found in Table 1 below.
lexical gaps in arabic: evidence from dictionaries
457
In order to be able to look up these words in English-Arabic, FrenchArabic and German-Arabic dictionaries it was necessary to obtain
translations of the Dutch words from the sample in these languages.
The table below shows these 25 Dutch words with translations in
three western languages. Some Dutch words which were candidates
to be included in the sample, turned out to be untranslatable in one
or more western languages. The absence of an obvious translational
equivalent in one of these languages of course would cause problems
in using that language as source language for the comparison of Arabic
translations. For this reason, some randomly selected Dutch words,
translated with descriptions in the Nijmegen Dutch-Arabic dictionary
were not included in the sample.
TABLE 1
ENGLISH, FRENCH AND GERMAN TRANSLATIONS

OF THE SAMPLE
Dutch
English
French
bestek
francofiel
frankeren
freak
fresco
gletsjer
glibberen
cutlery
francophile
to frank
freak
fresco
glacier
slither
couvert
francophile
affranchir
fana
fresque
glacier
glisser
gniffelen
snigger
log
guest
logement
lommerd
loods
loops
lotgenoot
ouderwets
lodging (house)
pawnshop
pilot
in heat/season
partner (in
misfortune /
adversity)
old-fashioned
ouvreuse
ouwel
usherette
wafer
German
Besteck
frankophil
frankieren
Freak
Fresco
Gletscher
glitschen,
schlittern, rutschen
rire sous cape, rire schmunzeln, in
tout bas, rire dans sich hinein lcheln
sa barbe
hte
Gast, Logiergast,
Logierbesuch
auberge
Gasthaus
mont-de-pit
Leihaus, Leihamt
pilote, lamaneur Lotse
en chaleur, en rut brnstig
compagnon,
Leidensgenosse,
compagne
Schicksalsgenosse
dinfortune
vieux jeu, dmod, altmodisch
pass de mode,
dsuet, surann,
archaque, prim
ouvreuse
Platzanweiserin
pain ( m. ) azyme, Oblate
dautel, hostie non
consacre
458
jan hoogland
TABLE 1 (CONT.)
Dutch
English
French
German
ribbel
rif
riposteren
roe
rog
trampoline
transcriberen
traumatisch
rib, cord
reef
riposte
rod
ray
trampoline
transcribe
traumatic
cte
rcif
riposter
verge
raie
trampoline
transcrire
traumatique
Rippe
Riff
ripostieren
Rute
Rochen
Trampoline
transkribieren
traumatisch
If the assumption that the concepts behind these words represent

lexical gaps in Arabic is correct, at least a part of these words have to
be translated with descriptions in other dictionaries too, if they were
included in these dictionaries at all. So in order to investigate this, the
translations of these words were checked in a number of dictionaries.
For this comparison the following dictionaries were consulted:
Al Mawrid English-Arabic
Oxford, Doniach English-Arabic
El Mounged English-Arabic
Al-Kamel Al-Kabir French-Arabic
Al Manhal French-Arabic
Schregle German-Arabic
Van Mol Dutch-Arabic
However, before presenting the results of this comparison of the seven

dictionaries mentioned, it seems useful to present the Arabic translations
from the Nijmegen dictionaries here first. Obviously, these should not
be part of the comparison, since the lack of translational equivalents for
these words in the Nijmegen Dutch-Arabic dictionary were the starting
point for the comparison. Inclusion of these translations would distort
the results of the comparison.
Tables 3a and 3b below contain the Arabic translations of the Dutch
words from the seven dictionaries introduced above.

TABLE 2
459
ARABIC TRANSLATIONS FROM THE NIJMEGEN

DUTCH-ARABIC DICTIONARY
DU_LU
Eng_LU
bestek
francofiel
frankeren
freak
fresco
gletsjer
glibberen
gniffelen
log
logement
lommerd
loods
loops
lotgenoot
ouderwets
ouvreuse
ouwel
ribbel
rif
riposteren
roe
rog
trampoline
cutlery
francophile
to frank
freak
fresco
glacier
slither
snigger
guest
lodging (house)
pawn shop
pilot (nautic)
in heat/season
partner (in
misfortune/adversity
old-fashioned
usherette
wafer
rib, cord
reef
riposte
rod
ray
trampoline
transcriberen
traumatisch
transcribe
traumatic
DA_Nijmegen
I@ #
?

@ A' BC@ DEF ,H

J LK M1 : J 6N .K O
!
P "2 P Q @
RS O> T
! UV W
>K K5 X 1Y 5Z
K _ 1 `
(3[
\

])
3[

4 a b@ &!
A
h g
Id
c ! e 5 = f g

a) eL Q
(BL ij[
kLl m
n 6
P op[ q b 1 +K O!
W 1
!, Jrs: ( W 1
!
Aa@, m @, 1 t
4b $(J[ E )
1 j5 6 ha1W
va
q!
{w 1C
(xn j@ y z C ])
JrLS !
| )
}~
1
HWK 1
w- >" k
- V
5 q! J (
Jc 9B " ) -
} -4
j q! '
K O"
qjn: W 1
} J K5n !
# + >=
1 #
&> Jc
7
1
Jc LK : = Jc ! 5 " AS T !
460
jan hoogland
TABLE 3A
ARABIC TRANSLATIONS FROM 3 OF THE

7 OTHER DICTIONARIES
DU_LU
Eng_LU
bestek
cutlery
francofiel
francophile
frankeren
to frank
freak
freak
fresco
fresco
gletsjer
glacier
glibberen
slither
gniffelen
snigger
log
logement
guest
lodging
(house)
lommerd
pawn shop
EA_Mawrid
EA_Oxford
EA_Mounged
:qjE>>>>>( o H4>>a !

J>"@
q>>!
qjE>>>>>>( A>>'B!@ qjE(
o
IE>>W@
3@nZ I@#
H
qjE>>>>>E
>b1j@
@# JLM1: . >b >! + .O!

DLM1: n!
JLM1: J6N
>>>>M16
(nX#)
"2 A>>>5a 1X# q>! &>>:'# (>sW#) +(-#)
(J!B{ @#)
>>>>>a1
}@ >>>>>>a1>"
}# >>>>>s:a 1X#
n" H>>(;6! a1
> A> &>>{
.s_' UV>>W
J
&>>>{ 1>>>a
>>>6 -=
?1j
%{)\1Y5 J!>(1! ->
->X v>( >o!
(7 @# }-\ &>>>>{ }"
+d9 Rd_
6:X
1>>>=
s$ 1>>=
s$ 1>>>=
s$
&>>>>> A4>>>>>a ->On! &>>{ A>>w
A4>>= A>>w
A4>>>>a JsO>>E
ew =1>>Z
ew
fg) hpa :s: fg hg #1> I-) hOQ
(Id! e5=
!T! h>>p ->>>>>>T:"
(4>T
( mn:") +a4<
: q!3!:# ]) JW@1:! J61 ij" ]) JW@1:! J61

;! 1!
([ ij" 1X!@ ([
1X!@
.T! :qb1 I>=b- +O! Ib- + qb1!
qjb1>>>>> eL>>>>>>> }E
+"! 19)
@#
(qb+"! 1>9
q>b-
461
TABLE 3A (CONT.)
DU_LU
Eng_LU
EA_Mawrid
EA_Oxford
EA_Mounged
loods
pilot
(nautic)
]) q: W1! ] Jrs: W1!

(61! @# n9 r# @# rLC +!
(B! n9) -['
loops
in heat/
season
lotgenoot
partner (in
misfortune/
adversity)
Lb :
=# r' \
J }L
>W1!
Jrs:>>>>>>
+>! >>6
@# >>>>>nL
->>>d' n#
n9
>>6 .>- m@
:
(J>>>>>dE)
ouderwets
oldfashioned
ouvreuse
usherette
ouwel
wafer
ribbel
rif
rib, cord
reef
riposteren
riposte
roe
rod
rog
ray
trampoline
trampoline
(J[) i9@
] i[:
w1 AL' w12 q! (e!)
:2 x( +" JL~ (-#) 9
-
6
W1! :+L @- W1 T6
{! -r vb' rL
( yz! ])
q! -@! J99- }"1 }W1
H 6
|
~1 ]
( )
J!{ 6- PQ
Pp! ()
:1O[ L J>a1V>> W
- (
(Js=>X1!)
C ( 19
n' Xa 3a P>a1>>( >> J:2.
->>p &>>>{
J>a[
'
.a 5'
qjn:>>>>>> h 1 Dn:W
1>>>>O[
h=-
J5n! w) Dd!*
}>>>>>>d h"w 1Y [
qjd!1>>>>>>> [' L{ 4:a
(w[
q!4 L{ &:'

.b8! Ja12)
@# aw +" (Ja1=
|[) x()
W1 >>>>>T6
rL>>>>> @vb>'
}W1
}"1
|
m6 - (
J: C (
3a P>a1>>>( ->p &>>{
J>a[
>5'
.>>>>>a>T
qjn:>>>>>W
1>>>O>"
[ w) Dd!*
4:a h"w 1Y
(w[ [' L{
462
jan hoogland
TABLE 3A (CONT.)
DU_LU
Eng_LU
transcriberen
transcribe
traumatisch
traumatic
EA_Mawrid
EA_Oxford
EA_Mounged
q! W q! +ra
qY@ 3fW#
+ra }E ) 1
@# !
J"T !
(Ja
>1>X >>Q-
+= >>>M
J4! J">TE
q>! @#
&>>>> ka1>>W
Ja{ J"E
+ }K@ M
J"T C
Ja{
>Q- >1>X
TABLE 3B
ARABIC TRANSLATIONS FROM THE REMAINING 4

OTHER DICTIONARIES
DU_LU
Eng_LU
bestek
francofiel
frankeren
freak
fresco
gletsjer
glibberen
FA_Kamil FA_Manhal GA_Schregle DA_Mol
q!) oC Hw

J!
qj
JEW@ J!
JE>>W@
JE>>>@
q@ D(@
qj(@
J@
( J26@
francophile
M1: .
.O! J>>>>>6N Aa
DLM1: n!
>>>>>M1:
J>s>>>>>>M1:
to frank
H>>(- P6 %{ U
PQ@
U>> J(- - q!) (- &>>>{ P"2
(AO!
J(- &>>{
V
freak
P! R>O>T!
.5>T!
fresco
@# 1Y>>>>5Z
1>a>>>>5Z ko %{ =
->X v>(- v>(- ->X
16
-> o!
-$
Jso! Ja-$
) H-) L$ 1
glacier
' .*!
J!
] }a ] J(@ [d9
JL 3[\ (JL[ A2rC
A2rC ]@
1` (JL[
J! L$
slither
w A4< Aw w A4< Aw
A,4 4
cutlery
J!
JE>>W@
qj(@
1>>=
s$
463
TABLE 3B (CONT.)
DU_LU
gniffelen
log
logement
lommerd
loods
loops
lotgenoot
ouderwets
ouvreuse
ouwel
ribbel
rif
Eng_LU
snigger
>>>6 h>>Q hg vZ

1>(
Js:
eLQ '>! e>sQ
3 v"
esQ
eLQ
vsa
lodging
34>= mn6 34>>=
v! mn6 34<
(house)
}
pawn shop
qb1>> is" 7( hr" eL ij"
7!) Ib1 hr"

qb1
+"! eL
o:"@ qbJ@ !
(1:
pilot
a }~- W1! W1! }~>W1!
>W1!
(nautic)
rs:(
+L
in heat/
.2) +o, .a
season
(: .- (s)
s
partner (in
>>>6 ha1>>W
>n(a As6- qY19
misfortune
15 >b
/ adversity)
>>a1>>>W
>>>" >>>>6
oldq! 4 9 w1> q! w1> &>>{
AL'
fashioned
AL' w12
va
va 1T!
aw +"
A>>>s' hT!)
-
w1> (J9 L"
usherette
:{! JrL! !{) * J
yz! ] !{ yzC

:L@ j( @# -p W1
-p W- (v($ q!#
{!
wafer
1j6 4 }~19
m>9
JLg JL" (ia>>>>>l)
rib, cord
PQ 2W PQ
= w@1
@# .=$)
(vX q! w-~
reef
eL- J: 1) J: J>E # .>W
J>dW
( :1 2W 19 [
(1O[
( %{ Ja1
C
guest
464
jan hoogland
TABLE 3B (CONT.)
DU_LU
riposteren
roe
rog
trampoline
transcriberen
traumatisch
Eng_LU
- J{1>>l - @$
H>9 H>_b &>{ - (J{@)

>p! H>_" J:2 Jr"
(eL~ w-[C
.jp9)
rod
(" a
'
ray
L h=-@
qjn:>>>>>W
1 Dn:W s 1>>O>"
?x(# R$)
(. . . Ja1
K
4:!
trampoline
kn!
transcribe - 1X - 1X
d!

traumatic >Q- >1>X
>1X >>>d! >"
>"
>>Qriposte
>5'
>>>M
>>Q-
So, in order to verify if the concepts behind these Dutch words do

represent lexical gaps in Arabic, it is useful to compare and to evaluate
the translations presented in these other dictionaries that provided the
same units with a description, if the word in question is included in the
other dictionary at all.
For all translations obtained from the 7 dictionaries mentioned above,
information was entered if the dictionary in question has included the
Dutch word or its English / French / German equivalent as an entry, and
if so, if this translation is a description. For this an evaluation system was
devised. All translations are evaluated with a code from the following
code system:
0=
1=
2=
3=
4=
5=
entry not included

entry is included, translation is an equivalent (not a description)
entry is included, translation is description
entry is included, translation is a neologism
entry is included, translation is a hyperonym
entry is included, translation is inaccurate or translation belongs to
a different or more specific meaning
465
6 = entry is included, translation consists of a neologism and a

description
Some remarks need to be made in relation to these values. For the sake
of clarity examples for each of these 6 categories will be presented. These
examples are taken from Table 3 a and 3b above.
Value 0 is obvious. If a word is not included in a dictionary, the value for
this word is 0. For example: glibberen (to slither) is not included as an
entry in D-A Van Mol, nor in G-A Schregle.
All other values (16) imply that the word occurs as an entry in the
source language of the dictionary heading the column.
Value 1 implies the translation is a translation equivalent. For example:
glibberen / to slither is included and its translation seems to be a real
equivalent in E-A Oxford (ew 1), E-A Mounged (ew A4< Aw),
and F-A Kamil (A,4 4 w A4< Aw).
Value 2 refers to a description. For example: glibberen / to slither
is included, and its translation is a description in E-A Mawrid (&
JL ).
This applies, first of all, to the real explanations which can be found,
for example, in Oxford English-Arabic at bestek / cutlery ( ox H4a !
IW@ A'B!@ DEF( q!) or logement / lodging (ij" ]) JW@1:! J61
q 1X!@ ([ ). However, one might as well define as descriptions
those combinations of two words that are not frequent. If we look at
the translations for francofiel / francophile for example, we find the
translation M1: . in both F-A Al Kamel and F-A Al Manhal. This is
a combination of words that is not lexicalized, and therefore should not
be entered in the database as an expression.
Value 3 refers to a neologism. For example: lommerd / pawnshop is
translated with a neologism in F-A Kamel (qb1 ij"). Some dictionaries
offer the user made up words in case of a lexical gap in Arabic. In Al
Manhal French-Arabic these neologisms are marked with an asterisk
466
jan hoogland
and an explanation comes with the neologism. See the examples of value
6 below.
Value 4 represents a hyperonym translation, i.e. the word in the target
language is actually the translation of a hyperonym of the source language
word. For example: transcriberen / to transcribe is translated as +=
M }K@ in several dictionaries. These translations do not express the
fact that to transcribe means decoding from one system and encoding
in another system. Therefore these translations express a more general
meaning and not the specific meaning of the source language word.
Value 5 as an evaluation means that the translation can be inaccurate or
wrong, or that it refers to a different, more specific meaning of the source
language word. For example: rif / reef is included, but its translation is
inaccurate in G-A Schregle (J# .W).
Value 6, finally, is a combination of 2 and 3, since it consists of a
neologism, with a description (mostly between brackets) added to it.
For example: gletsjer / glacier is included, but its translation consists
of a neologism and a description in F-A Manhal ([d9 H-
JL[ A2rC ] J(@) and F-A Kamil (3[\ ] }a ' .*! )
JL[ A2rC ]@ JL).
Before presenting all evaluations in a table, it is useful to introduce
the concept translation profile. This profile is a compilation of al the
evaluations, indicating the number of dictionaries in which a translation
was found, followed by an enumeration of all the evaluations. This profile
gives a quick indication of the number and types of translations found
in the various dictionaries.
For example 3:223 means a word was found in three dictionaries, and
in two of these dictionaries the word was translated with a description
(code 2), and in one with a neologism (code 3).
Below is a table containing columns with the Dutch words, the English
translations and 7 columns for all 7 sample dictionaries, and a final
evaluating column containing the translation profile of the concept.
3. Interpretation of Table 4
467
The translation profiles of the words should indicate if these words

represent concepts that can be considered lexical gaps in Arabic. The
higher the number of dictionaries in which these words were included,
and the higher the number of description translations or neologism
translations, the higher is the probability that the concept represents a
lexical gap in Arabic. So, profiles containing value 2 (description) and
value 6 (neologism + description) are the clearest indications.
Seven Dutch words have been included in all 7 dictionaries. These
words are: bestek / cutlery, gletsjer / glacier, logement / lodging,
loods / pilot, ouderwets / old-fashioned, rif / reef , and transcriberen
/ transcribe. A more thorough look at the translation profiles of these
words is useful in order to determine if the coding system seems reliable.
1. The word bestek / cutlery has the following profile 7: 2222222. It
seems justified to conclude this word/concept indeed represents a
lexical gap in Arabic since there is unanimity among all dictionaries
in this, and probably no other dictionary or other reference work
will mention an Arabic word for this concept. Obviously, we do not
have to go very deep into Arab or Islamic culture to conclude tat this
lexical gap can be explained by the fact that in the Arab world people
were and are still used to eat with their hands.
2. The word ouderwets / old-fashioned has the same profile:
7: 2222222. So, obviously this concept also represents a lexical gap
in Arabic.
3. The word loods / pilot has the following profile: 7: 1111222. The
Dutch word loods, as a person, is monoseme and can only mean
a pilot in shipping, and certainly not airplane pilot or any other
type of guide. The translation W1! is correct but requires a specific
context in order to distinguish this specific meaning from other more
general meanings of the word W1!.
4. The word transcriberen / transcribe has the following profile:
7: 2222444. As indicated above, there are presented description
translations and hyperonym translations.
5. The word rif / reef has the following profile 7: 2255666. Three
dictionaries have the word J: as a neologism, in all cases followed
or preceded by a description. However, this word was not found with
this meaning in the text corpus or on the internet.
6. The word gletsjer / glacier has the following profile: 7: 1116666.
468
jan hoogland
TABLE 4
Dutch_LU
Engl_Tr
bestek
francofiel
frankeren
freak
fresco
gletsjer
glibberen
gniffelen
loge
logement
cutlery
francophile
to frank
freak
fresco
glacier
slither
snigger
guest
lodging
(house)
pawnshop
pilot
(nautic)
in heat/
season
partner (in
misfortune
/adversity)
oldfashioned
usherette
wafer
rib, cord
reef
riposte
rod
ray
trampoline
transcribe
lommerd
loods
loops
lotgenoot
ouderwets
ouvreuse
ouwel
ribbel
rif
riposteren
roe
rog
trampoline
transcriberen
traumatisch traumatic
EVALUATION OF THE TRANSLATIONS IN

7 DICTIONARIES
E_A_ E_A_ E_A_ F_A_ F_A_ G_A_ D_A_ Transl.
Mawr Oxf Moun Kamel Manh Schreg Mol Profile
2
2
2
2
2
6
2
2
4
2
2
2
2
0
2
1
1
2
0
2
2
2
2
0
2
1
1
2
0
2
2
2
2
4
2
6
1
2
4
1
2
2
2
4
2
6
1
2
4
1
2
2
2
0
2
6
0
2
2
4
2
0
0
0
0
1
0
0
4
1
7: 2222222
6: 222222
6: 222222
3: 244
6: 222222
7: 1116666
5: 11112
6: 222222
5: 24444
7: 2224445
6
2
2
2
6
2
3
1
2
1
2
1
0
1
6: 222366
7: 1111222
5: 22222
3: 224
7: 2222222
2
2
4
6
2
4
1
2
2
2
2
2
2
1
1
6
2
2
0
2
2
1
1
6
6
5
4
6
2
0
1
3
6
2
2
6
2
1
1
0
0
4
0
5
0
4
0
3
0
0
0
5
0
4
0
0
5: 22266
6: 222245
4: 2244
7: 2255666
5: 22222
6: 111444
5: 11111
5: 23366
7: 2222444
6:111111
The equivalent L$ 1` which is mentioned in three dictionaries

seems to have become a lexicalized expression during the last years.
As for the neologisms presented by the other four dictionaries, the
fact that four dictionary compilers have added descriptions to these
neologisms illustrates that they are not completely at ease about the
understandability of the neologisms they present.
7. The word logement / lodging has the following profile 7: 2224445.
This profile seems to justify the conclusion that this word is a
469
problem word. Indeed, it can easily be translated with a hyperonym,

since it is a certain type of accommodation for staying overnight.
The Dutch Van Dale dictionary defines a logement as a place to stay,
a home where one can obtain a temporary stay (and food) against
payment (presently less in rank or appreciation than a hotel). Neither
is lodging the same as a hotel. So translations with mn6 or 34< should
be evaluated as hyperonym translations.
Another way of looking at the profiles is the following:
There are nine Dutch words that were translated with a description in
all dictionaries in which they were included:
bestek / cutlery
ouderwets / old fashioned
francofiel/ francophile
frankeren/ to frank
fresco/ fresco
gniffelen / to snigger
loops / in heat
ouvreuse / usherette
riposteren / riposte
7: 2222222
7: 2222222
6: 222222
6: 222222
6: 222222
6: 222222
5: 22222
5: 22266
5: 22266
These are the words of which the translation profile consists of code 2
and 6 only. It seems reasonable to sustain that words found in four or
more dictionaries (out of seven) and being translated with descriptions
(possibly in combination with a neologism) in all those dictionaries
indeed do represent a lexical gap in Arabic.
So the concepts expressed by the words bestek / cutlery, ouderwets
/ old-fashioned, francofiel / francophile, frankeren / to frank, fresco
/ fresco, gniffelen / to snigger, loops / in heat, ouvreuse / usherette
and riposteren / to riposte can be considered lexical gaps in Arabic.
There are five Dutch words that were translated with a description in
most dictionaries in which they were included:
lommerd / pawn shop
lotgenoot / partner in misfortune
ouwel / wafer
rif / reef
6: 222366
3: 224
6: 222245
7: 2255666
470
jan hoogland
trampoline / trampoline
5: 23366
Since most of the translations are descriptions, and the others should
be qualified as inadequate translations (neologisms, hyperonyms or
inaccurate), we may assume that these concepts also represent lexical
gaps in Arabic.
As a matter of fact, this system for encoding types of translations and
combining them in a translation profile seems to be a useful method for
comparing dictionaries and identifying lexical gaps in Arabic.
When we take a closer look at the Dutch words meeting with lexical
gaps in Arabic, it turns out, not surprisingly, that they can be categorized
as follows:
the source unit expresses a typical Dutch or non-Arabic concept
causing a lexical gap in Arabic: fresco / fresco, ouvreuse /
usherette, transcriberen / to transcribe;
the source unit expresses a technical concept, too rarely used
inside the Arab world in order to enter a dictionary and to receive
a general translation, thus this concept represents a lexical gap in
Arabic: frankeren / to frank, trampoline / trampoline.
the source unit is related to a non-Islamic religious concept,
for example: ouwel / wafer, coming from Catholicism within
Christianity.
4. Conclusion
4.1
Conclusion concerning the identification of lexical gaps in Arabic
By comparing the Arabic translations of twenty-five Dutch words,

we concluded that nine were translated with descriptions in all the
dictionaries in which they were included. This justifies the conclusion
that these words can be considered to represent a lexical gap in Arabic.
Five words were translated with descriptions in most of the dictionaries
in which they were included, which also indicates they should be
considered to represent lexical gaps in Arabic.
Therefore, the hypothesis formulated at the beginning of this
contribution may be considered to be confirmed by this conclusion.

4.2
471
Conclusions concerning lexicography in general
As a result of the comparison described above, some conclusions

concerning the lexicography of Arabic can be drawn as well.
The first conclusion concerns the use of Arabic neologisms in
dictionaries. During the stage of comparing the descriptions in various
dictionaries, it was noticed that in some dictionaries newly coined words,
which should be considered neologisms in Arabic, were presented as
translations.
However, when neologisms are introduced in a bilingual dictionary as
a translation equivalent for an unknown foreign language word, there is
no context in the users mother tongue. As a matter of fact the dictionary
user will only run into such a neologism by looking up a foreign word
in a foreign language-Arabic dictionary. When the user is confronted
with a neologism without any further explanation, he/she will probably
remain in confusion about the meaning of the foreign language word
as well as to the meaning of the Arabic neologism. Some dictionaries
provide neologisms with an additional explanation or description.
This will help the user both to understand the meanings of the foreign
language word and the neologism in his own language.
If the same dictionary is used by foreigners who use the dictionary
to produce Arabic, and these foreigners use the neologisms presented,
there is again the risk of the Arab receiver of the message who will not
understand the meaning of the neologism.
To conclude, the following question presents itself: why do dictionary
compilers decide to invent new words as translation equivalents for
foreign words / concepts? Is this related to a certain way of language
purism in order to prevent the use of loan words? Do they realize that
the dictionary users still will be left in confusion? The result of this policy
will be the occurrence of ghost words, i.e. words that do only occur in
dictionaries and not in authentic texts.
A second conclusion concerns the occurrence of inaccurate translations in the dictionaries. As indicated in Table 4, some inaccurate
translations have been met during the process of comparing
the translations. Inaccurate translations are represented by value 5 in
the table. Fortunately, the total number of translations, qualified as
inaccurate, is rather limited, since value 5 occurred only three times on
a total of 142 translations in Table 4.
No dictionary is free of inaccuracies, and no one knows this better
than the present author who was involved in compiling the Dutch-
472
jan hoogland
Arabic dictionaries. Some descriptions in the Nijmegen Dutch-Arabic

dictionary could have been replaced with real translational equivalents.
As stated before, some Dutch words in the sample turned out to be
translatable with equivalents in at least some of the dictionaries. This
implies first of all that the concepts behind them certainly do not represent
lexical gaps in Arabic. And secondly, these words should not have been
translated with descriptions in the Nijmegen Dutch-Arabic dictionary.
This, in a certain way, can also be considered a series of inaccuracies
in the said dictionary. Would the authors involved in translating these
specific words have consulted all the dictionaries available, this might
have led to the insertion of a translational equivalent in stead of a
description. Not only would this have enriched the Dutch-Arabic part,
but also the reverse part, since translational equivalent relations are
bidirectional and consequently would have been entered in the ArabicDutch part as well.
Some dictionary makers rely heavily on their colleagues or
predecessors. Some similarities between certain pairs of dictionaries
in the comparison above, however, are striking. If we, for instance,
look at the translations for logement / lodging, lommerd / pawn
shop, loods / pilot, loops / in heat, ouvreuse / usherette, ouwel /
wafer, riposteren / to riposte, in the dictionaries E-A Oxford and E-A
Mounged, we find a number of literally identical translations. This is so
obvious, that one may conclude, that the compiler of the most recent one
of the two dictionaries (Mounged English-Arabic) depended heavily on
his predecessor (Oxford English-Arabic), without mentioning this in
the introduction of his dictionary, and even without entering the Oxford
dictionary in the list of sources consulted. The possibility that both of
the dictionaries were based on a common third ancestor requires further
research.
5. References
Balabakk, Munr. 1981. Al Mawrid: A modern English-Arabic dictionary. Beirut: Dr
al-Ilm li-l-Malyn.
Doniach, Nakdimon. 1972. The Oxford English-Arabic Dictionary of Current Usage.
Oxford: Oxford University Press.
Hoogland, Jan, Kees Versteegh, and Manfred Woidich, eds. 2003. Woordenboek ArabischNederlands Nederlands-Arabisch. 2 Vols. Amsterdam: Bulaaq.
Idriss, Souheil. 1996. Al Manhal Dictionnaire Franais-Arabe. Beirut: Dar Al Adab.
Mol, Mark van. 2001. Leerwoordenboek Nederlands-Arabisch. Amsterdam: Bulaaq.
473
Reda, Youssof M. 1996. Al-Kamel Al-Kabir plus, dictionnaire du franais classique et

contemporain, franais-arabe. Beirut: Librairie du Liban.
Schregle, Gtz. 1977. Deutsch-Arabisches Wrterbuch. Beirut: Librairie du Liban.
Theodory, Constantin. 1996. Al-Mounged English-Arabic. Beirut: Librairie Orientale.
MASDAR FORMATION
Joost Kremers
University of Cologne
1. Introduction
Arabic morphology includes a nominal form, traditionally called
masdar, whose meaning and form is related to that of a verb. The form
relation, although clearly visible, is complex, the meaning relation is
straightforward: the masdar is a complex event nominal, in terms of
Grimshaw (1990), and names the action expressed by the verb, retaining
the verbs argument and event structure.1
The masdar has been analyzed among others by Fassi Fehri (1993)
and Kremers (2003), who both base themselves on Abneys (1987) analysis of English gerunds. These analyses are purely syntactic, however.
Neither discusses how the morphology of these forms functions, the
tacit assumption being that the masdar form is derived by some postlexical process that does not play a role in syntax.
Ackema and Neeleman (2004) discuss so-called mixed categories,
such as English gerunds, in the framework of their theory on word formation and morphological structure. They argue that the peculiar syntactic properties of such structures follow from the way in which they
are formed morphologically.
Arabic masdars share the peculiar syntactic properties of other mixed
categories, but Ackema and Neelemans analysis does not provide any
direct insight into the reason why this should be so. Their analysis crucially depends on the concatenative nature of morphology in the languages that they discuss, and since masdar formation in Arabic uses
non-concatenative morphology, the analysis at first sight does not carry
1
The same nominal forms can also have non-event meanings, making them result
or simplex event nominals, in Grimshaws terms. These lack the argument and event
structure of the corresponding verb, and have no systematic meaning relation to it. See
Fassi Fehri (1993) and Kremers (2003) for some discussion. Traditionally, these nouns
are not called masdars.
476
joost kremers
over straightforwardly. As I show in this paper, however, nothing needs

to be changed about Ackema and Neelemans analysis if we adopt a prosodic model for Arabic morphology, along the lines of McCarthy and
Prince (1990b, 1996).
2. Abneys analysis of English gerunds

In his influential PhD dissertation, Abney (1987) presents an analysis of
the English gerund that tries to account for the fact that the gerund can
be used in at least three different constructions:
(1) a. Johns singing of the Marseillaise
b. Johns singing the Marseillaise
c. John singing the Marseillaise
In all three constructions in (1), the gerund singing is an event nominal,

in the sense that it retains the argument and event structure of the underlying verb to sing (Grimshaw 1990). In other words, all three phrases
describe an event, and the subject and object of the event are obligatorily
expressed.2 Abney observes (as others have done before him) that there
is a decreasing degree of nominality in the structures (1ac): (1a), the
so-called Ing-of construction, is the most nominal, in that it expresses
both the subject and the object with nominal means: the Saxon genitive s and the dummy preposition/case marker of, respectively. (1b),
the Poss-ing construction, is more nominal, in that the object is marked
with accusative case, rather than with of. Yet, the subject is still marked
with the Saxon genitive. Lastly, (1c), the so-called Acc-ing construction,
marks the object with accusative, and the subject as well. This is therefore the most verbal of the three constructions, in that there is no nominal marking present anymore.
Crucially, the fourth logical possibility, the subject being marked with
accusative and the object with of, does not occur:
(2) *John singing of the Marseillaise
2
If one or both of the arguments are not expressed, e.g. in Johns singing, the gerund is
no longer a complex event nominal, but rather a simplex event or result nominal, which,
as Grimshaw (1990) shows, have markedly different properties.
masdar formation
477
This fact can be accounted for if we assume, as Abney does, that gerunds
start out as V projections, and change into an N projection somewhere
along the way. Until the point where the change takes place, arguments
can be licensed through case, but after the change, only nominal licensing mechanisms (s and of ) are available. Once the change has taken
place, it cannot be undone, which accounts for the impossibility of (2).
The analysis is supported by the observation that (1bc) allow adverbs
but no adjectives, while (1a), the purely nominal gerund type, allows
adjectives but no adverbs:
(3) a. Johns constant/*constantly singing of the Marseillaise
b. Johns *constant/constantly singing the Marseillaise
c. John *constant/constantly singing the Marseillaise
The analysis that Abney proposes assumes that there is an affix -ing,
which attaches to a verbal category, changing it into its corresponding
nominal category. It can attach at three levels: at V, creating an N, at VP,
creating an NP, and at IP, creating its corresponding nominal projection DP. Crucially, -ing is not a head, i.e., it does not project a syntactic
phrase of its own. It just attaches to a projection, changing its category.
The tree structures that Abney proposes are the following:
(4) Ing-of:
DP
Johns
D
D
NP
N
-ing
PP
V
sing
of the Marsaillaise
478
joost kremers
(5) Poss-ing:
DP
Johns
D
D
NP
-ing
VP
PP
the Marsaillaise
sing
(6) Acc-ing:
DP
-ing
IP
I
John
I
VP
V
sing
DP
the Marsaillaise
There is of course one unattractive aspect to these structures: it is not

clear how the verb stem sing and the suffix -ing combine to form the
gerund singing. In fact, it is not at all clear what it really means for an
affix to attach to an XP.3
3. Ackema and Neelemans treatment of mixed categories

Ackema and Neeleman (2004) retain the idea that what is different about
the different gerund structures is the level at which the nominalizing
3
Abney notices this problem himself. In order to solve it, he argues for an elaboration
of X theory, which basically makes a distinction between below X0 and above X0 syntax
which is not altogether convincing.
masdar formation
479
affix attaches in syntax. However, they place this idea in the context of a
much broader theory on word formation, which enables them to show
what exactly it means for an affix to attach to different levels of projection. Before we look at how they deal with gerunds, it is necessary to
discuss some of the aspects of their theory.
3.1
Short outline of the theory
Ackema and Neeleman argue for a view of the language faculty that is
inspired by Jackendoff (1997, 2002). In this view, the language faculty
contains three generative systems, one for syntax, one for semantics and
one for phonology. The output of these generative systems are linked to
each other by mapping rules.
In this model of the language faculty, a lexical item is not just a conglomerate of the semantic, syntactic and phonological properties of a
word. Rather, the three types of features are essentially separate, functioning in separate components of the language faculty, linked through
mapping rules. A word such as tree has the syntactic representation
N[+count,sg], which is used in the syntactic module, and it has the
phonological representation /ti:/, which features in the phonological
module. If we then abbreviate the semantic concept as tree, we can
represent the lexical item tree as in (7), where the double arrows indicate two-way mapping relations:
(7) tree N[+count,sg] /ti:/
Affixes function the same way: they have, apart from a semantic representation, which does not concern us here, a morphosyntactic one,
which Ackema and Neeleman represent in small caps, as affix, and a
morphophonological one, which they represent with slashes, as /affix/.
The point is that what is traditionally seen as a single affix actually consists of three separate elements, linked by mapping principles.
There are different types of mapping principles. Ackema and Neeleman argue that there are (at least) three general mapping principles, of
which two concern us here.4 The first is Linear Correspondence:
4
The third mapping principle is Quantitative Correspondence which states that no
element in the morphosyntax is spelled out more than once.
480
joost kremers
(8) Linear Correspondence

If X is structurally external to Y, then F(X) is linearly external to F(Y).5
Linear Correspondence states that a morphosyntactic representation

such as [[root sf1] sf2], in which sf1 is first attached to root, and sf2
then to the complex [root sf1], is mapped onto a morphophonological
representation in which the order of the two suffixes is retained, i.e., /
rootsf1sf2/. It can in principle not be mapped onto a structure in which
both suffixes have switched places: */rootsf2sf1/.
The second general mapping principle is Input Correspondence:
(9) Input Correspondence
If an affix selects (a category headed by) X, then F(affix) takes F(X) as its
host.
Input Correspondence states that if an affix in morphosyntax attaches

to an element X of a specific category, or to a projection of X, then the
morphophonological form to which affix is mapped must attach to
the morphophonological element associated with X. In other words, the
phonological form of a nominal affix cannot attach to a phonological
form that is associated with a syntactic V head; it must attach to something that is associated with an N head.
Note that both Input and Linear Correspondence only apply when
F(affix) has an overt phonological form. When F(affix) is empty (i.e.
//), they do not apply (or apply vacuously). Since both principles regulate the distribution of phonological material, it stands to reason that
they do not apply when there is no phonological material to distribute.
Apart from these general mapping principles, there are also lexical
mapping principles, as in (7). Note that not only words, but also affixes
can be described with lexical mapping rules. For example, the English
agentive suffix er is normally mapped onto the phonological form //:
to write writer. This fact is recorded in the lexical entry for the affix:
5
The notation F(X) refers to the phonological structure onto which the syntactic
structure X is mapped. It is equivalent to the slash notation /x/ that Ackema and Neeleman use, but in my opinion less confusing. I define it as in (i):
(i) a. D(X): the subtree that has X as root
b. F(X): the phonological material onto which D(X) is mapped.
masdar formation
481
(10) agentive noun er[N,sg] //
There are, however, specific cases in which er receives an idiosyncratic

mapping. One example is the case of the verb to type, of which the agentive noun is typist, not *typer. In other words, we can say that there is an
idiosyncratic mapping rule of the form in (11):
(11) [[type] er] /taip/ /ist/
This rule says that a syntactic structure of the form [[type] er] is
mapped onto a phonological structure /taip/ /ist/. Note that mentioning
the phonological form of the stem in the rule is in fact redundant: there
is no reason to assume that the lexical entry for type cannot provide its
phonological form in (11). Therefore, I adopt a slightly different notation for such idiosyncratic mapping rules. Instead of (11), I will write
the following:
(12) er/type /ist/
(12) expresses that er, when attached to type, is mapped onto /-ist/. 6
Affixes are well-known to have selectional restrictions. For example, the
English agentive suffix -er must attach to a word (i.e., it cannot attach
to a phrase), and moreover, this word must be a verb. The point that
Ackema and Neeleman make is that these selectional restrictions are in
fact of non-uniform nature, and must therefore lie in different modules
of the grammar.
By representing a suffix such as -er as is done in (10), it becomes possible to express this point: Ackema and Neeleman state that the syntactic
affix er has the requirement that it must attach to something of category V, while the phonological affix // has the requirement that it must
attach to something that is a (phonological) word.
Note that I use a hyphen here in the affix /-ist /. In Ackema and Neeleman's original
formulation, this is not necessary, because the mapping rule explicitly mentions the stem
and with that, the position of the affix with respect to the stem. In my reformulation,
some way is needed to indicate that the affix is actually a suffix. Note also that although
I use a double arrow in (14), this is not meant to indicate that this is a one-to-one mapping; the phonological form /-ist / does not always signal an agentive noun, cf. words
such as communist, guitarist. Here, Ackema and Neelemans formulation may seem to
have an advantage, since their equivalence is one-to one. As explained in footnote 30,
however, this is not the case in every kind of idiosyncratic mapping rule that we need.
482
joost kremers
This analysis means that er does not require that it be attached to a

head; instead, it can attach to any projection of category V.7 In this way,
Ackema and Neeleman account for the existence of agentive nouns such
as in (13):
(13) spring-in-t-veldje
jump-in-the-field.dim
little madcap
The phrase spring-in-t-veld is a VP, containing an intransitive verb and a

locative PP. In order to derive its meaning as an agentive noun, Ackema
and Neeleman assume that the Dutch er suffix is present in the syntactic
structure, and that it is idiosyncratically mapped onto //:8
(14) [[spring [in t veld]] er] /spri/ /int/ /vlt/ //
That is, there is a mapping rule of the form:
(15) er/VP //
In (14), the suffix er attaches to the VP [spring [in t veld]]; being a

syntactic suffix, er does not require a V head, it only requires a structure
of category V. However, mapping er onto // would violate Input Correspondence, which requires that F(er) take F(V) as its host, even when
it attaches to VP. As mentioned above, Ackema and Neeleman argue
that the alternative mapping onto // voids Input Correspondence.
Therefore, (14) is possible.
3.2
English gerunds
Having discussed the essentials of Ackema and Neelemans theory, we

can now look at their analysis of English gerunds. First, Ackema and
Neeleman assume that the actual nominalizing affix in gerunds is not
-ing. The reason for this is that -ing does not just derive nominal categories, it also derives the (verbal) present participle, which indicates that it
is probably best categorized as a non-finite verb suffix. Therefore, there
must be another affix in the structure that is responsible for the nomi-
7
That is not to say that the attachment is completely free. Some requirements do
apply. Since er saturates the external argument of the verb it attaches to, it cannot attach
to a VP of which the external theta role has already been assigned.
8
I have ignored the diminutive suffix -je here. Although it is attached to the noun
veld phonologically, morphosyntactically, it attaches to the entire structure. (It is not the
field that is small, but the madcap.) See Ackema and Neeleman (2004) for details.
masdar formation
483
nalization. Because no such overt affix exists, Ackema and Neeleman

assume that its phonological form is null.9
As in Abneys analysis, the nominalizing affix is attached to different levels of projection. Note that if the affix had an overt phonological
form, the syntactic structure could not be mapped onto a phonological
structure without violating either Input Correspondence or Linear Correspondence. Suppose, for example, that the affix attaches to VP, creating a Poss-ing construction:10
(16) Poss-ing:
DP
John
D
D
NP
VP
AFFIX
Adv
constantly
VP
V-ing
singing
DP
the Marsaillaise
Note that the nominalizing affix is assumed to be a suffix in (16). It

attaches to a projection of category V. If its corresponding F(affix) were
overt, Input Correspondence would require it to attach to the /word/ that
the head of this projection is mapped onto, i.e., singing. That, however,
would violate Linear Correspondence: in the syntactic structure affix
is external to the VP, and therefore Linear Correspondence requires its
phonological form to be external to F(VP). However, in a phonological
realization such as /Johns/ /singing/ /affix/ /the Marseillaise/, this would
not be the case: the Marseillaise here follows the /affix/, meaning that the
/affix/ is internal to F(VP).11
The same conclusion is reached by Yoon (1996).

Ackema and Neeleman do not specify to which levels of projection the affix
attaches in English, they only do this for Dutch nominal infinitives. I simply assume that
English behaves the same. Note, by the way, that Ackema and Neeleman only discuss
Ing-of and Poss-ing, not Acc-ing. I assume the reason for this is that Acc-ing is more
likely a participial structure than a nominal one, as argued also by Reuland (1983).
11
Note that this situation would obtain if we were to assume that -ing is the nominalizing affix.
10
484
joost kremers
A realization with the /affix/ after the object DP would comply with
Linear Correspondence, but would violate Input Correspondence,
because the /affix/ can no longer attach to /singing/.12 However, on the
assumption that the /affix/ has no overt form, neither Input Correspondence nor Linear Correspondence applies (or they apply vacuously),
and hence no mapping principle is violated.13
4. Arabic masdars
4.1
The data
Let us now turn to the Arabic masdar. First, as observed by Fassi Fehri
(1993) and Kremers (2003), Arabic masdars are typical mixed categories. They can assign (overt) accusative to their objects, as in (17a), but
the object can also be licensed through a preposition, as in (17b):
(17) a. aqlaqa-n
-ntiqd-u
-r-rajul-i
annoyed-me criticizing-nom
the-man-gen
the mans criticizing the project annoyed me
b. aqlaqa-n
-ntiqd-u
-r-rajul-i
annoyed-me criticizing-nom the-man-gen
the mans criticizing of the project annoyed me
-l-mar-a
the-project-acc
li
-l-mar-i
to the-project-gen
I will refer to constructions of the type in (17a) as masdar + acc, and to

constructions of the type in (17b) as masdar+li. Note that in both cases,
the subject of the masdar is marked with genitive case, making the structures equivalent to the Poss-ing and Ing-of constructions, respectively.
Furthermore, (17a) allows an adverbial expression, while (17b) allows
an adjective:
12
Note, by the way, that it is really immaterial whether we assume that the affix
adjoins to the left or the right of the VP. We would run into the same problems with the
adverbial constantly.
13
The analysis makes a strong prediction: if the nominalizing suffix is overt, a deverbal noun cannot assign accusative to its object if this object follows the verb. Ackema
and Neeleman show facts from Norwegian and Quechua that suggest that this prediction is borne out.
masdar formation
485
(18) a. aqlaqa-n
-ntiqd-u
-r-rajul-i
annoyed-me
criticizing-nom the-man-gen
bi-stimrrin
-l-mar-a
with persistence the-project-acc
the mans persistently criticizing the project annoyed me
b. aqlaqa-n
-ntiqd-u
-r-rajul-i
-l-mustamirr-u
annoyed-me
the-persistent
li -l-mar-i
to the-project-gen
the mans persistent criticizing of the project annoyed me
In contrast, masdar + acc cannot be modified by an adjective:14

(19) *aqlaqa-n
-ntiqd-u
-r-rajul-i
-l-mustamirr-u
annoyed-me
the-persistent
-l-mar-a
the-project-acc
the mans constant criticizing the project annoyed me
Fassi Fehri (1993) analyzes masdars much along the lines of Abney
(1987). He argues that there is an (abstract) masdar affix that can attach
at the V level or at the VP level, turning the verbal projection into an N
or an NP. In Kremers (2003), I present a somewhat different analysis.
Instead of having an affix attach to a verbal projection at different levels,
I argue that during the derivation, a nominal functional category can
take a verbal category as its complement. This analysis is based on the
idea that there are certain parallels in the nominal and the verbal projection lines. Essentially, I argue that at any point in the verbal projection
line, a verbal functional head can be replaced by its nominal counterpart, thus deriving the mixed nature of the construction.
However, as remarked above, neither of these approaches takes into
account the morphology of masdar forms. Within Ackema and Neelemans framework, however, we can do better: it becomes possible to see
how masdar formation works given the syntactic analysis. Before we go
into that question, I will first discuss the morphology of masdars.
14
Somewhat unexpectedly, however, masdar+li can be modified by adverbial
phrases:
(i) aqlaqa-n -ntiqd-u
-r-rajul-i
bi -istimrrin
li -l-mar-i
annoyed-me criticizing-nom the-man-gen with the-persistence tothe-project-gen
the mans constantly criticizing of the project annoyed me
As suggested in Kremers (2003), this may be due to the fact that the adverbial in (i) is
a PP.
486
4.2
joost kremers
Masdar formation
Arabic lexical items generally have a root and a stem. The root is a
sequence of (usually) three consonants,15 and is not in itself pronounceable. From a root, stems can be derived by applying prosodic morphemes
(McCarthy and Prince 1996). A stem thus formed is a pronounceable
form, and is the basis for further conjugation and derivation, by means
of pre- and suffixes or by additional prosodic morphological processes
(McCarthy and Prince 1990b).
The original analysis holds that stems are derived by applying CVtemplates to a root. For example, the perfective stem of the base verb is
formed by applying the template CVCVCV to a root (McCarthy 1981).
More recently, however, it has been argued that this approach cannot
account for certain facts. In particular, it allows templates that do not
occur (McCarthy and Prince 1990a, 1990b). Therefore, a newer approach
is developed, in which templates are expressed in prosodic terms.16 The
perfective stem of the base verb is then expressed as a prosodic template
of the form , i.e., a structure of two short syllables.17
It is important to note that a root consists of just three consonants,
and unlike much derivational morphology in Arabic does not have any
prosodic specification. That is, the root does not specify in which positions in the syllable the consonants appear. For example, the initial consonant can appear in the onset (20a) or in the coda (20b):
(20) root: /ktb/
a. ka.ta.b to write
b. ak.ta.b to dictate
15
There are some quadriliteral roots, and there are categories that can be considered
biliteral roots. These will not be discussed here.
16
Prosodic here refers to the prosodic hierarchy (Selkirk 1980, Truckenbrodt in
press), i.e. the hierarchy of prosodic constituents that constitute a phrase, i.e. Utterance >
intonational phrase > phonological phrase > prosodic word > foot > syllable > mora. For
the present discussion, only syllables () and mor () are relevant. A syllable consists
of an onset (the initial consonants) and a rhyme (the vowel and any final consonants).
The onset is extramoraic, the rhyme contains either one mora or two, for light and heavy
syllables respectively. A light syllable is represented as , i.e. a syllable of a single mora,
while a heavy syllable is represented as . In Arabic, the first mora of a syllable is always
associated with a vowel. The second mora can be associated with a consonant or with the
same vowel as the first mora, when this vowel spreads (i.e. when it is long).
17
The third syllable in a form such as kataba he wrote is not part of the template.
Stems in Arabic always have a final extrametrical consonant (McCarthy and Prince
1990a), and it is usually the third root consonant which fills this position. This consonant normally resyllabifies when an ending is added, in this case the -a of the third
person masculine singular.
masdar formation
487
The same goes for the second consonant, although it rarely happens in
the verbal system that it appears in a coda. In the nominal system, this
is not uncommon, however:
(21) root: /f l/
a. fa.a.l to do
b. fi.l action, act
Note that (21) shows that a root does not even specify whether two consonants appear in the same syllable or not. This is solely specified by
the prosodic template of the particular stem formed from the root in
question.
Table 1 lists the perfective stem and masdar forms of triliteral verbs.18
TABLE 1
ARABIC MASDARS
Stem
perfective stem
masdar
I
II
III
IV
V
VI
VII
VIII
IX
X
XI
XII
XIII
XIV
XV
KaTaB
KaTTaB
KTaB
aKTaB
taKaTTaB
taKTaB
inKaTaB
iKtaTaB
iKTaBB
istaKTaB
iKTTaB
iKTawTaB
iKTawwaB
iKTanBaB
iKTanBay
(irregular)
taKTB
muKTaBa(t)
iKTB
taKaTTuB
taKTuB
inKiTB
iKtiTB
iKTiBB
istiKTB
iKTTB
iKTiwTB
iKTiwwB
iKTinBB
iKTinBy
18
As discussed by Wright (1981), Classical Arabic had alternative masdar forms for
most verb stems. Nonetheless, the forms in Table 1 are the standard forms, which is why
I confine myself to them here.
488
joost kremers
In this table, the capitalized consonants KTB belong to the root, any
other non-capitalized consonants are stem affixes.
Looking at Table 1, one quickly notices that one particular vowel pattern dominates the masdar forms: /i/, although on some occasions
(when the corresponding verb stem has a long vowel, such as in III and
XI), the /i/ is also long, yielding //.
Some more analysis is possible, though. McCarthy and Prince (1990a)
argue that the last vowel in a verbal form is always short in finite forms
and often long in non-finite forms (masdars, but also passive participles of stem I, and some deverbal instrumental nouns). They therefore
analyze vowel quantity of the final vowel as a template suffix indicating (non-)finiteness. With that modification, the masdar is no longer
marked by /i/ or //, but simply by /ia/, with vowel length determined by other factors. We can therefore say that the Arabic masdar
contains two morphemes: a nominalizer with the form /ia/, and a nonfinite suffix -.
This template suffix, - for finite forms and - for non-finite forms,
combines with a base stem template, which is monosyllabic.19 So the
stem I template for finite forms exists of the stem base - plus the finite
suffix -. The stem II template is formed from the stem base - plus
the finite suffix. In addition to these morphemes, stems can also have a
prefix, such as the tain stems V and VI, n- in stem VII and st- in stem
X, etc.20
We can now analyze a form such as /infil/, the masdar of stem VII
of the root /fl/, as containing four distinct morphemes: the consonantal
root, the stem VII marker, the non-finite marker and the nominalizer:
(22) root: /f l/
stem VII: (n)
nominalizer: /ia/
non-finite: -
Putting these together requires that we analyze phonological structure

as consisting of several layers, or autosegmental tiers (Leben 1973). The
basic tier is the segmental tier, which contains the segments (phonemes)
19
Although it may contain a detransitivizing prefix /n-/ or /t-/, as in stems VII
and VIII, which consists of an extrametrical consonant that resyllabifies into a coda
position.
20
The stem VIII infix -t- is analyzed as a prefix as well, after which a metathesis rule
swaps the initial consonant and the prefix.
masdar formation
489
of the utterance. For Arabic, we need to assume an additional syllabic

tier. The root morpheme in (22) maps onto the segmental tier, as does
the nominalizer. The non-finiteness morpheme maps onto the syllabic
tier. The stem VII marker is mixed: it maps onto the syllabic tier, but at
the same time fills a slot in the segmental tier. When all four morphemes
of (22) are put together, we obtain the following representation:
(23) ()
a
()
The top row in (23) represents the syllabic tier: the first two syllables
are given by the stem VII marker. The first syllable is extrametrical (it
is later syllabified by the insertion of an epenthetic /i/), and its coda
position is filled by the /n/. The third syllable is the non-finiteness suffix, which is heavy (i.e. has two mor), as indicated. The final syllable
is again extrametrical, and is added by default, since every Arabic stem
ends in an extrametrical syllable.
The bottom row represents the segmental tier. Crucial is of course
the question how the various slots in both tiers are associated with each
other. The initial /n/ is straight-forward: the stem VII marker specifies that it is associated with the first (extrametrical) syllable. Just as
straightforward is the final root consonant /l/. Because a stem must end
in an extrametrical syllable, the third consonant must always take this
position.
The remaining segments, /f/ and // of the root, and /a/ and /i/ of
the nominalizer, are associated through the principle of Left-to-Right
Association (Leben 1973). The root consonants fill the coda positions of
the second and third syllable, and the vowels of the nominalizer fill the
peaks. Because the third syllable has an additional mora, and because
there is no segmental material anymore to fill it, the vowel /a/ spreads to
the second mora, which results in a long vowel.
Most of the masdar forms can, mutatis mutandis, be analyzed in this
manner: for the masdar forms of stems IV and VIIXV, all that needs to
change in (23) is the stem template.21 As is clear from Table 1, however,
21
Note that the masdar of stem IX /iKTaBB/ is formed on the base of the underlying
form /iKTaBaB/. The gemination of the third root consonant is the result of deletion of
the /a/, a common process in this context.
490
joost kremers
things are different for the remaining forms: they appear to be quite
idiosyncratic. In spite of the apparent complexity, however, we can show
that they all make use of the same stem template that the corresponding
verb forms use. Where they differ from other masdars is the form of the
nominalizer and the non-finite morpheme, which is sometimes absent.
Let us look at the various forms. First, masdars V and VI have an
idiosyncratic nominalizer, that takes the form /au/ rather than /ia/.
They also lack the non-finite morpheme , since the forms are not
*/taKaTTB/ and */taKTB/. Here, we can either say that they have an
idiosyncratic non-finite marker -, or that the lack of a second syllable
in the template causes not just the stem affix but the entire verbal stem
template to be used as a basis for the masdar. Either way, we obtain the
same result:
(24) stem V/VI:
-
t
nominalizer: /au/
I will assume that the nominalizer does not associate with the prefixed
syllable in the stem template. Therefore, the /a/ of the nominalizer associates with the second (heavy) syllable of the stem template, and the /u/
with the third. Presumably, the /a/ of the first, prefixed, syllable is a copy
of the /a/ of the second syllable.22
Stem II masdars can be dealt with in a similar way. On the face of it,
the stem II masdar does not seem to be formed on the template of the
stem II verb, as it contains a prefix /ta/ which the verbal template lacks,
and its vowel pattern is /ai/, not /ia/, as the default nominalizer specifies. Furthermore, it does not show gemination of the second consonant,
which seems typical for the verbal template.
However, McCarthy and Prince (1990b) observe that Arabic phonology does not distinguish between CVV and CVC syllables: all that
counts is their prosodic status as heavy syllables. The templates of stems
II and III (and likewise stems V and VI) are therefore identical: their
first syllable is , without any specification how the heaviness of the
22
One fact that supports this assumption is that Classical Arabic had an alternative
masdar form for stem V verbs, namely /tiKiTTB/. On the assumption that the vowel
in the prefix is a copy of the vowel in the next syllable, this form is completely regular:
it contains the masdar morpheme /ia/ and the non-finite suffix , as indicated by the
fact that the final vowel is long.
masdar formation
491
syllable is obtained. An additional gemination rule operates on stem II

and V verbs, but not on stem III and VI.
Suppose now that we say that the gemination rule does not apply in
stem II masdars (unlike stem V masdars), and that the nominalizer of
stem II is idiosyncratic. We then obtain the following:
(25) stem II:
nominalizer: /tai/
non-finite: -
Here, the stem II template is not specific to the masdar, it is the same
template that forms finite verb forms of stem II. In order to see how
these morphemes yield the masdar form /taktb/, let us see how they are
associated with the template. First, we add the non-finite suffix to the
stem base, and associate the nominalizer with the resulting template:
(26)
The first syllable here has two mor because this is specified in the stem
II template, and the second syllable has two mor because it is the nonfinite suffix. The vowel /i/ cannot be associated with the second mora of
the first syllable because Arabic does not allow two vowels in a single
syllable.23 When the root is intercalated into the template, the second
mora of the first syllable can be associated with the first root consonant.
The second root consonant can be associated with the onset of the second syllable, and, as usual, the last consonant is associated with a (newly
created) extrametrical syllable. At the same time, the /i/ associated with
the first mora spreads to the second, creating a long vowel:
(27)
t a k
()
23
Long vowels and diphthongs are either the result of lengthening or because the second mora is filled by a semi-consonant, not because two vowels end up in one syllable.
492
joost kremers
In this way, we can derive the stem II masdar by just assuming an idiosyncratic nominalizer, in the same way that we have analyzed stem V
and VI masdars.24
The stem III masdar is quite idiosyncratic as well. Its form is
/muKTaBa(t)/, where the final /-a(t)/ is the regular feminine ending.
This masdar is a so-called masdar mmy or m-masdar. The m-masdar is
an alternative masdar formation found in Classical Arabic, in which the
feminine form of the passive participle is used as masdar. This masdar
formation has mostly disappeared, but in stem III, it is the dominant
one.25 I will not go into the details of participle formation in Arabic
(see McCarthy 1981 for some discussion), but like masdar formation,
it is templatic, with an /m-/ prefix that marks the participial form and a
vowel pattern that indicates voice.
Summarizing, we can say that there is a regular nominalizer /ia/,
which applies in the majority of cases. This affix is combined with a nonfinite suffix . Stems II, V, and VI have an idiosyncratic nominalizer, and stem III has an idiosyncratic masdar formation. Stems V and
VI lack the non-finite suffix, and instead get a default as second syllable. Lastly, stem I masdars are all idiosyncratic. In the next section, I
will discuss the syntactic and phonological processes that underlie the
masdar formation in more detail, and discuss the mapping rules needed
to account for them.
4.3
The masdar affix
As we have seen, the syntactic properties of masdars are essentially

identical to those of gerunds. Therefore, the starting point for the analysis should be that the syntactic structure of masdars is the same as well:
there is a syntactic masdar affix that attaches to V or to a projection of
V.
If masdar attaches to V, we obtain the equivalent of the English Ing-of
construction: masdar+li. Mapping of such a structure to phonology
24
Note that the /ta/ element in the masdar morpheme is not a prefix, contrary to the
/t/ element in stem V and VI forms. If it were, a form /taKaTTB/ would result. Instead,
it must be part of the nominalizer.
25
Classical Arabic had three other stem III masdars, /KiTB/, /KTB/ and /KiTTB/,
but none of these were as common as the m-masdar.
masdar formation
493
is unproblematic in any event, as we have seen above, because V and

masdar are sisters in the tree. Input and Linear Correspondence are
easily satisfied.
However, Linear Correspondence is only apparently satisfied. As we
will see, there is no way in which masdar in a masdar+acc construction
can adhere to Linear Correspondence. We must therefore assume that
masdar is not subject to it at all, so that even in masdar+li constructions, adherence to Linear Correspondence is only apparent.
(28) shows the tree of a masdar + acc construction. masdar attaches
to VP, which yields a structure in which the object can be assigned accusative case:
(28)
DP
D
NP
MASDAR
VP
Subj
V
V
Obj
However, on the assumption that masdar is subject to Linear Correspondence, the mapping to phonology would violate at least one mapping principle. The reasoning is identical to the one discussed in section
3.2 for the English gerund constructions. masdar attaches to the VP, so
Input Correspondence requires that /masdar/ attaches to /v/. Doing so
would violate Linear Correspondence, however: masdar is external to
the VP, but having /masdar/ attach to /v/ leaves it internal to it, as it then
occupies a position between /subject/ and /v/.
The analysis that made this configuration unproblematic for English,
saying that the nominalizing /affix/ is phonologically null, is not available for Arabic. As we have seen above, the Arabic masdar formation
uses an overt nominalizing morpheme. Another possible solution that
easily comes to mind is to adjust the structural relations in the tree in
such a way that masdar and its intended host V are adjacent, so that /
masdar/ can attach to /v/ without violating Linear Correspondence.
Such structural rearrangement obviously implies movement. As
argued in many works (e.g., Ritter 1991, Kremers 2003), in possessive
constructions in Arabic and Hebrew (of which (28) is one, because the
494
joost kremers
subject is assigned genitive case), N moves to D. If we apply this movement to (28), the V head26 and masdar would end up adjacent.
Such an analysis might seem attractive at first sight, because the order
in masdar constructions is always NSO. Therefore, the V head must
move to a higher position: the tree in (28) could not derive the correct
order. Moving V to a position adjacent to mas d
ar in (28) would enable
Linear Correspondence to be adhered to while at the same time deriving the correct word order.
There are some problems with this analysis, however. First, there is no
real consensus that N-to-D movement really takes place: Borer (1999)
argues against it, for example. Furthermore, even if one assumes that it
can take place, N does not always move to D in the Arabic noun phrase.
If there is no genitive-marked DP in the noun phrase, N-to-D movement
presumably does not take place. Therefore, it cannot be guaranteed that
/masdar/ and /v/ will always, under any conceivable construction, end
up adjacent.
The root of these problems, however, is the tacit assumption that
masdar is a prefix.27 The phonological shape of the masdar affix, however, suggests that this assumption is incorrect: as discussed above, much
of Arabic morphology consists of prosodic templates that map onto a
prosodic tier in phonology (see, e.g. McCarthy 1981). The masdar affix
is such a prosodic template, as we have seen in section 4.2. It is neither a
prefix nor a suffix, and hence does not appear before or after the verbal
root. Rather, one would say it appears simultaneous with it.
At first sight, this seems a problematic conclusion. masdar is attached
to a syntactic structure, and syntactic structures are at some point linearized. The common assumption is that linearization is total: it applies
to every terminal element in the tree. That is, the linear structure that
results from a syntactic tree contains all the terminal elements in that
tree, and for any pair of distinct elements x,y, a linear order is defined,
either x>y or x<y (Kayne 1994). Linear Correspondence is basically a
formalization of this assumption.
26
Of course, there is no N head in (28), so we would have to assume that V moves
to D.
27
Note that the entire solution would be impossible if /masdar/ were a suffix: the
object would then always end up between /v/ and /masdar/, no matter where V moves
to.
masdar formation
495
However, as mentioned above, Ackema and Neeleman assume that

Linear and Input Correspondence do not apply when a syntactic element
maps onto // in phonology. That is, the mapping principles are sensitive to the phonological form of the elements they apply to. It should be
no surprise, then, that given the prosodic nature of the masdar affix, it
is also not subject to Linear Correspondence.
More precisely, the masdar morpheme maps onto the prosodic tier,
and therefore does not (and in fact cannot) require linearization with
respect to the material on the segmental tier. However, it does require
association with material on the segmental tier. This is obtained through
Input Correspondence: because masdar attaches to the VP, F(masdar)
must attach to F(V), which is the root morpheme. Since the root morpheme consists of segmental material, the required association is
obtained.
Having said that, let us look at how masdar formation takes place
in detail. In section 4.2, I have argued, following McCarthy and Prince
(1990b), that the morphosyntactic process of masdar formation involves
two morphemes: a nominalizer, which I will indicate with noml, and
a non-finite suffix n-fin. Above in this section, on the other hand, I
have assumed a masdar affix that must be able to attach to a structure
in phrasal syntax, in order to derive the Arabic equivalent of Poss-ing
constructions. The question we must ask, then, is what is the relation
between masdar on the one hand, and noml and n-fin on the other.
Obviously, noml and n-fin in some way make up masdar. I propose
the following structure:
(29)
NOML
NOML
N-FIN
This is a morphological structure, which is then inserted into the syntactic structure at the position of the masdar affix:
496
joost kremers
(30)
DP
NP
VP
masdar
noml
noml
Subj
n-fin
V
Obj
V
V
vii
root
I assume that the V head is composed of the root and the verb stem
marker, here stem VII. The masdar affix, as indicated, is composed of
the nominalizer and the non-finiteness suffix. When the tree in (30) is
mapped onto phonology, Input Correspondence will make sure that
F(masdar) is properly associated with F(V). The formation of the
masdar form /infil/ will then proceed as described above.
Note that whether a syntactic element is subject to Linear Correspondence is not a function of its morphological form alone. F(V) itself consists of autosegmental morphemes, but cannot be exempt from Linear
Correspondence: the entire masdar form consist of autosegmental morphemes, and at least one of them must be subject to Linear Correspondence, otherwise the form could not be linearized with respect to the
other terminal elements in the structure. The natural assumption is that
the root, which is not a syntactic affix, unlike all the other morphemes,
is this element. Note that the root is a phonological affix, because it cannot form a stem of its own. Syntactically, however, it is not, because it
does not require adjunction to a structure of a specific category. This
syntactic difference between the root and the other elements is presumably the result of a semantic difference: the root is a lexical item (in the
traditional sense of the word) and as such member of an open semantic
class. All the other morphemes are functional or derivational, and part
of closed semantic classes. It therefore seems reasonable to assume that
members of open classes must always be mapped onto the segmental tier
in some way, and are thus always subject to Linear Correspondence. The
masdar formation
497
result of this assumption is that the masdar form is linearized according

to the position that the head V takes in the syntactic structure.28
Stem V/VI and stem II masdars differ very little from regular masdars.
All that is required is to assume the following idiosyncratic mapping
rules:
(31) stem V:
stem II:
noml/V
n-fin/V
noml/II

/ua/

/tai/
That is, noml in the context of a stem V template is mapped onto /ua/,
and n-fin onto -.29 Furthermore, when attached to a stem II template,
noml is mapped onto /tai/ rather than the default /ia/.
The system can also account for the stem III masdar. Recall that this
masdar is a so-called m-masdar, i.e., it is not formed with some genuine
masdar affix. Rather, a more or less suppletive strategy is used, which
employs the feminine passive participle. Let us say that there is an idiosyncratic mapping rule of the following form:30
28
Note that this means that V must move, because masdar constructions have the
order NSO. At first sight, it might be tempting to argue that the NSO order is derived by
spelling out the masdar in the position of masdar, but this would not work for masdars
that license their objects with the preposition li: the analysis states that in such masdars
the masdar affix attaches to V rather than VP, which would predict a surface order of
SNO for such constructions. The actual surface structure is NSO, however, the same as
for masdars that assign accusative.
29
The stem V rules obviously also apply to stem VI, and, although not discussed
here, to the second quadriliteral stem as well. The rule may in fact refer not to the stem
templates but to some other property: as McCarthy and Prince (1990b) discuss, these
three verb stems share properties with each other that are not found in other verb stems,
indicating that they form a class of their own. Presumably, the mapping rules refer to
this class.
30
I mentioned in footnote 6 that Ackema and Neelemans formulation of idiosyncratic mapping rules cannot be read as a one-to-one mapping in all cases. (32) is one
such case. Ackema and Neelemans rules differ in that they mention the phonological
form of the host as well as of the affix. But doing that in (32) would still not establish
that a form consisting of F(prt.pass.f)+F(III) is equivalent to a stem III masdar, as it
could (obviously) also be a feminine passive participle of a stem III verb. Note that this
is indicative of a general asymmetry between syntax and phonology : a syntactic structure is always mapped onto one particular phonological structure, but a phonological
structure may have more than one syntactic equivalent. In other words: phonological
structures can be ambiguous.
498
(32) [noml noml n-fin]/III
joost kremers
F(prt.pass.f)
I will not go into the formation of the participle here. What is relevant
is the fact that an idiosyncratic mapping rule of the form in can exists.
Rather than specifying the phonological material that the syntactic
structure under consideration is mapped onto, the rule specifies a different (morpho)syntactic form whose phonological mapping must be
applied.
5. Conclusions
Ackema and Neelemans (2004) theory on word formation can give a
straightforward account of so-called mixed categories in languages
where morphology is concatenative. Non-concatenative morphology,
such as that of Arabic masdars, at first sight does not yield to an analysis
in terms of Ackema and Neeleman. If, however, we adopt the common
analysis of non-concatenative morphology in terms of autosegmental
tiers, we find a natural way to exclude the masdar morpheme from Linear Correspondence, which is the greatest obstacle to the application of
Ackema and Neelemans analysis to masdars: Linear Correspondence
only applies within an autosegmental tier in phonology.
At the same time, Input Correspondence still applies to the masdar
morpheme, accounting for the fact that it takes the verb as its host, and
not some random root in the rest of the structure.
6. References
Abney, Steven. 1987. The English Noun Phrase in its Sentential Aspect. PhD thesis, MIT:
MIT Working Papers in Linguistics.
Ackema, Peter and Ad Neeleman. 2004. Beyond Morphology: Interface conditions on
word formation. Oxford: Oxford University Press.
Borer, Hagit. 1999. Deconstructing the construct. In Kyle Johnson and Ian Roberts,
eds. Beyond Principles and Parameters. Dordrecht: Kluwer Academic Publishers,
4390.
Eid, Mushira and John McCarthy, eds. 1990. Perspectives on Arabic Linguistics II.
Amsterdam: John Benjamins.
Fassi Fehri, Abdelkader. 1993. Issues in the Structure of Arabic Clauses and Words. Dordrecht: Kluwer Academic Publishers.
Grimshaw, Jane.1990. Argument Structure. Cambridge, Mass: The MIT Press.
Jackendoff, Ray. 1997. The Architecture of the Language Faculty. Cambridge, Mass: The
MIT Press.
masdar formation
499
. 2002. Foundations of Language: Brain, meaning, grammar, evolution. Oxford:

Oxford University Press.
Kayne, Richard. 1994. The Antisymmetry of Syntax. Cambridge, Mass: The MIT Press.
Kremers, Joost. 2003. The Noun Phrase in Arabic: A minimalist approach, PhD thesis,
University of Nijmegen. LOT Dissertation Series 79.
Leben, William. 1973. Suprasegmental Phonology. PhD thesis, MIT.
McCarthy, John. 1981. A prosodic theory of nonconcatenative morphology. Linguistic
Inquiry 12(3). 373418.
McCarthy, John and Alan Prince. 1990a. Foot and word in prosodic morphology: the
Arabic broken plural. Natural Language and Linguistic Theory 8. 209283.
. 1990b. Prosodic morphology and templatic morphology. In Eid and McCarthy,
eds. Perspectives on Arabic Linguistics II Amsterdam: John Benjamins. 154.
. 1996. Prosodic morphology 1986. Ms. University of Massachusetts, Rutgers University.
Reuland, Eric. 1983. Governing ing. Linguistic Inquiry 14(1). 101136.
Ritter, Elizabeth. 1991. Two functional categories in noun phrases: Evidence from
Modern Hebrew. In Susan Rothstein, ed. Perspectives on phrase structure: Heads and
licensing. San Diego: Academic Press. 3762.
Selkirk, Elizabeth. 1980. The role of prosodic categories in English word stress. Linguistic Inquiry 11(3). 563605.
Truckenbrodt, Hubert. In press. The syntax-phonology interface. In Paul de Lacy, ed.
Cambridge Handbook of Phonology. Cambridge: Cambridge University Press.
Wright, William. 1981. A Grammar of the Arabic Language. Beirut: Librairie du Liban.
Yoon, James. 1996. Nominal gerund phrases in English as phrasal zero derivations.
Linguistics 34. 329356.
MTHODOLOGIE LINGUISTIQUE :
ORGANISATION DE LA LANGUE ARABE
ORGANISATION GNRALE DES LANGUES
Andr Roman
Lyon
1. Introduction
Archimde (~ 287~ 212), ayant tabli la thorie du levier, aurait
demand pour soulever le monde un point dappui .
Lon peut voir dans le levier le symbole des mthodologies. Dans le
point dappui , le symbole de leur premier point dapplication.1
Un exemple admirable, clbre, de cette double dmarche est donn
par Ren Descartes dans son Discours de la mthode (1819 ; 29 ; 32) :
Je crus que jaurais assez des quatre [prceptes] suivants [. . .]
Le premier tait de ne recevoir jamais aucune chose pour vraie, que
je ne la connusse videmment tre telle [. . .]
Le second, de diviser chacune des difficults que jexaminerais, en
autant de parcelles quil se pourrait, et quil serait requis pour les mieux
rsoudre.
Le troisime, de conduire par ordre mes penses, en commenant
par les objets les plus simples et les plus aiss connatre, pour monter peu peu, comme par degrs, jusques la connaissance des plus
composs ; et supposant mme de lordre entre ceux qui ne se prcdent
point naturellement les uns les autres.
Et le dernier, de faire partout des dnombrements si entiers, et des
revues si gnrales, que je fusse assur de ne rien omettre.
Ces longues chanes de raisons, toutes simples et faciles, dont les gomtres ont coutume de se servir, pour parvenir leurs plus difficiles dmonstrations, mavaient donn occasion de mimaginer que toutes les choses,
qui peuvent tomber sous la connaissance des hommes, sentre-suivent en
mme faon, et que, pourvu seulement quon sabstienne den recevoir
aucune pour vraie qui ne le soit, et quon garde toujours lordre quil faut
Le symbole est la peinture dune mtaphore.
502
andr roman
pour les dduire les unes des autres, il ny en peut avoir de si loignes
auxquelles enfin on ne parvienne, ni de si caches quon ne dcouvre. Et
je ne fus pas beaucoup en peine de chercher par lesquelles il tait besoin
de commencer : car je savais dj que ctait par les plus simples et les plus
aises connatre.
Au contraire [des sceptiques] tout mon dessein ne tendait qu massurer,
et rejeter la terre mouvante et le sable, pour trouver le roc ou largile.
Et remarquant que cette vrit : je pense, donc je suis, tait si ferme et si
assure, que toutes les plus extravagantes suppositions des sceptiques
ntaient pas capables de lbranler, je jugeai que je pouvais la recevoir, sans
scrupule, pour le premier principe de la philosophie que je cherchais.
Ce problme du premier point de lapplication dune mthodologie,

Ferdinand de Saussure lvoque dans une question, Unde exoriar ,
Do commencer ? (2002, 281). Sa question nest pas, semble-t-il, une
citation. La langue latine, dans laquelle il la formule, est la langue de la
premire voie qui sest ouverte lEurope, la voie romaine.2 La langue
latine est la langue de la culture europenne.
1.1 Une langue et lhomme
Voici le linguiste devant une langue et devant lhomme. Il peut commencer son tude par lhomme car lhomme est, dans le monde, le seul
animal qui parle. La capacit propre lhomme, lhomme seul, qui fait
de lui un homo loquens, un h ayawn ntiq,3 est, selon toute vraisemblance, reflte par les langues que lhomme parle.4
Si cela est,
ou bien la capacit de parole de lhomme, cette capacit constitutive
de lhomme, tant reconnue, la constitution gnrale de la langue,
qui est une actualisation de cette capacit, pourra tre reconnue dans
cette capacit ;
ou bien la constitution gnrale de la langue tant reconnue, la constitution de lhomme, qui en est le crateur, pourra tre saisie dans la
langue.
Voir R. Brague, 1992.

Al-Maydn cite dans son Majma al-Amtl 2 : 291, n3958, ladage : m l-insnu
law l l-lisnu ill sratun mumatta latun aw bahmatun muhmalat .
4
Il ne sagit pas ici des capacits physiologiques de lhomme, ses capacits auditives,
articulatoires, car celles-ci ne portent que sur les sons que lhomme doit entendre, doit
produire comme il parle, et non pas sur la grammaire de la langue.
3
mthodologie linguistique : organisation de la langue
503
Faut-il commencer par lhomme ? Faut-il commencer par la langue ?

Traditionnellement les savants qui se sont intresss aux langues ont
tudi les langues et non pas lhomme. Ou, exceptionnellement, quand
ils se sont intresss lhomme, cest pour affirmer que la langue a t
donne lhomme par Dieu, ou, comme Noam Chomsky, pour avancer,
axiomatiquement, que la langue est inne dans lhomme.
1.2
Mthodologie
Lexemple de Descartes incite rechercher dabord une mthodologie.

La premire esquisse de la mthodologie requise peut se rsumer dans
les deux oprations suivantes :
classer ;
dfinir.
Classer , cest--dire constituer en un ensemble des entits runies sur
quelque trait commun, reconnu, un trait propre, prsent ici, absent l,
qui les diffrencie, les oppose aux autres entits. Les plantes . Parmi
les plantes, les arbres . Parmi les arbres, le palmier . . . Les fruits
des arbres . Les fruits rouges , la fraise , la framboise , la groseille . . .5
Dfinir , cest--dire tablir, dans un ensemble donn par un classement, une certaine identit de chacun de ses lments. La plante est
un vgtal . L arbre est une plante dune certaine hauteur . . . Classer est une opration humaine, immdiate, constante. Dfinir est
tout autant une opration humaine, immdiate, constante.
La casuistique est une manifestation, universelle, clatante, de mise
en uvre, voulue objective, de ces deux oprations.6 Le jugement de
valeur , si commun, est une autre mise en uvre, subjective, de ces
5
Les langues ngro-africaines ont un systme de classes nominales qui repose
sur une rpartition des tres et des objets et, postrieurement sans doute, des abstractions en un certain nombre de catgories , in Les langues du monde, 2, 740741.
6
On a soutenu avec vraisemblance que la tradition grammaticale arabe avait
emprunt la casuistique le cadre gnral de sa dmarche (Carter 1972, particulirement). Versteegh (1980, 1114 ; 1993, 3336) a replac cette hypothse dans une perspective plus assure.
504
andr roman
deux oprations. Classer et dfinir apparaissent comme des oprations binaires, simples.7
Tous les chercheurs ont class. Exemples de classement des grammairiens, les parties du discours : le verbe ; l adverbe . . . ou, dans
le discours, la phrase . . . Et ils ont dfini.8 Comme tous les hommes,
toujours.
Mais, la diffrence de lhomme naf, le chercheur, doit dfinir en
attribuant chaque lment dun ensemble donn une identit tablie
de telle sorte que cet lment puisse tre, constamment, reconnu par
une diffrence raisonne, irrductible, persistante, qui loppose rgulirement ce qui nest pas lui.
Et le chercheur, la diffrence de lhomme naf, doit rviser ses classements au fur et mesure du progrs de ses dfinitions. Lidentit du
dernier classement auquel il aboutira ne pourra tre que la mme identit des lments quil a identifis, qui le composent.
Ce faisant, le chercheur doit inscrire lobjet de sa recherche dans le
temps, car rien sur terre na dexistence hors du temps. Il lui faut donc ne
pas mconnatre lhistoire de lobjet de sa recherche.
2. Sbawayhi
Les dfinitions donnes par Sbawayhi (al-Kitb, I : 12) des trois parties
du discours sont brves : 9
[La particule] est [dans la langue] pour un sens qui nexiste ni par le nom,
ni par le verbe. Le nom, cest homme, cheval.10 Le verbe, ce sont les
7
Schuler (1990, 252) a pu prsenter la dmarche gnrale de la plaidoirie dune
cause, de sa premire position sa dernire position dprcatoire , sur un algorithme exactement binaire.
8
Voir infra les dfinitions des parties du discours donnes par Sbawayhi dans son
Kitb. De mme al-Frb, lun des plus minents et des plus clbres philosophes
musulmans [. . .] surnomm le second matre, le premier tant Aristote [. . .] , mort
Damas, en 339/950, dans son Ih s al-ulm, p. 5. Le nom h add, singulier de h udd,
signifie communment limite et manire . Il a, chez Sbawayhi, o il est un terme
de mthodologie, gard son sens de manire (Troupeau, 1976, s.v. h add). Cest, semble-t-il, dans le Kitb al-h udd du grammairien et philosophe mutazilite ar-Rummn,
mort en 394/994, quil a pris, suivant sa pente, le sens de dfinition . Semblablement, le
terme orismos, dfinition , employ par Aristote, est rapprocher de horos, bord .
9
Voir sur la dfinition des parties du discours Versteegh (1995, 2242).
10
Certains manuscrits du Kitb, la somme de la tradition grammaticale arabe
naissante, livre imparfaitement dit, ajoutent mur h it).
505
schmes drivs de lexpression phonique des vnements lis aux noms et

qui sont construits pour ce qui est pass, ce qui sera, ce qui est encore. 11
fa l-kalimu : ismun wa-filun wa-harfun ja li-manan laysa bi-smin wa-l
fil fa-l-ismu : rajulun wa-farasun. wa-amm l-filu fa-amti latun uxidat
min lafzi ahdti l-asmi wa-buniyat li-m mad wa-li-m yaknu walam yaqa wa-m huwa kinun lam yanqati.
La seule partie du discours recevoir dans ce court paragraphe une dfinition est le verbe. Le verbe se conjugue. Sa conjugaison offrait une prise
Sbawayhi. Mais, remarquablement, ce que Sbawayhi dit ici du verbe,
il ne le dduit pas du verbe. Lexpression du temps quil prte au verbe,
est lexpression du droulement du temps quil constate dans le monde.
Du nom, il sest born donner deux exemples. De fait, dans la langue
arabe de son temps, le nom nest pas construit. Il nest saisissable que par
sa relation ce quil nomme. Il faut donc, pour le reconnatre, regarder
dans le monde lentit quil nomme.12
Au demeurant, les grammairiens arabes, avant mme Sbawayhi,
avaient class, faussement, le relatif h aytu, l o , parmi les noms.
Linclusion dans lensemble des noms dun lment qui nest pas un nom,
rendait impossible la dfinition du nom. Il fallait sortir h aytu de lensemble des noms, recomposer le premier ensemble, encore mal assur,
des noms.
De la particule, qui ne lui apparat pas construite non plus, Sbawayhi
relve quelle assure, complmentairement, les tches smantiques, rfrentielles, que le verbe et le nom nassurent pas.13 Ces tches sont les
tches de relation et de localisation, que lexprience immdiate montre
comme tant ncessaires.
Ce texte, qui sera exactement repris, perptuellement, apparat
comme une image du monde : ce qui est , le nom ; ce qui est dans
11
La constatation que les modus nexistent que dans ou par les res.
Un autre grand grammairien, Ibn Fris, mort Rayy en 395/1004, a, dans son livre,
as-Sh ibi f fiqh al-lua wa-sunan al-arab f kalmi-h, reproch Sbawayhi, p. 85,
davoir, pour le nom, donn non pas une dfinition mais des exemples (tamtl).
Lui-mme a retenu, comme tant plausible, la dfinition suivante : Le nom est ce qui est
implant sur le nomm comme on le mentionne et qui lui reste attach. , dukira l an
badi ahli l-arabiyyati anna l-isma m kna mustaqirran al l-musamm waqta dikri-ka
iyy-hu wa-lziman la-hu wa-hd qarb . Mais cette dfinition nest rien dautre que la
reconnaissance du fait que le nom est rapport immdiatement au nomm, cest--dire
sans le truchement du systme de la langue.
13
En fait plusieurs particules sont analysables ; exemple la ngation, /law/, du mode
rel, faite de la ngation /l/ et du morphme /w/ (< /u/) du mode rel ; voir A. Roman
1998.
12
506
andr roman
le temps , le verbe ; avec, entre lun et lautre, le lien , indispensable,

verbalis par la particule qui les situe, qui est diffrente, par son rle,
du verbe et du nom .
Ici, la tradition grammaticale arabe a choisi, semble-t-il, dobserver dans la langue le monde, non pas lhomme donc et non plus, tout
dabord, la langue. Or la langue ne peut reflter limmense complexit
du monde.14 Cette porte du monde qui ouvrirait sur la langue est une
fausse porte.15
Quant la quatrime unit fondamentale de la langue que composeraient ensemble le nom, le verbe, la particule, cette unit qui serait leur
finalit dans la langue, cette unit, la phrase , les grammairiens non
arabes en ont prsum lexistence. Ils ont chou la dfinir. 16 Les grammairiens arabes, eux, ne se sont pas attachs cette unit incertaine.
3. Discussion
Sil faut commencer ltude des langues smitiques par un ensemble
taill dans ces langues, pourquoi ne pas commencer par un ensemble,
caractristique de ces langues, lensemble des suites ordonnes de trois
consonnes sur lesquelles les units de nomination de ces langues apparaissent construites, le plus souvent.
14
Les systmes matriels qui composent le monde, dune part, sont ouverts, la
diffrences des systmes linguistiques ; dautre part, la combinatoire qui structure les
systme linguistiques est binaire voir infra , tandis que les combinatoires multiples
qui structurent les systmes matriels sont n-aires, avec n suprieur 2 .
15
La proposition de Thom, rapporte dans Petitot, Entrevue avec Ren Thom, est,
senzaltro, plus pittoresque. J.P. : Votre hypothse est [. . .] que les actions archtypales
comme capturer, prendre, couper, lier etc. sont devenues par ritualisation les matrices
de toutes les structures syntaxiques R.T. : Oui. Elles ont captur les structures plus
complexes. La meilleure preuve cest quil nexiste pas de verbes de valence suprieure
quatre. Cest la rgle des phrases de Gibbs . Ce passage ici repris de P.M. Lavorel 1980 :
I, 475).
16
G. Mounin, 1960, a recens, semble-t-il, toutes les dfinitions de la phrase. Aucune
nest opratoire. G. Kleiber, 2003, qui rfute le nouveau dcoupage du discours en clauses
et en priodes, concde qu on a tout gagner tenir compte de la dimension mmorielle introduite par Berrendonner, qui tait totalement absente des analyses traditionnelles et dont la nouveaut consiste montrer que les relations de discours narticulent
pas deux segments textuels, mais un segment textuel et une information en mmoire
discursive . Mais comment voir une pice de la langue dans cette dimension mmorielle ou dans la priode dfinie hypothtiquement comme une unit ergonomique
(A. Berrendonner, confrence, Lyon, 23.10.2003) ?
507
Ainsi, pour reprendre lexemple de toutes les grammaires, la mme

squence, ktb, se retrouve dans : /katbun/, crire (nominatif) ;
/kitbatun/, temps dcriture, criture (dun manuscrit) (nominatif) ; /aktubu/, Jcris (mode rel) ; /ktibu-n/, crivant (nominatif) ;17 /maktabun/, objet dcriture, bureau (nominatif) .
Cette squence de trois consonnes, la tradition grammaticale arabe
la nomme, tardivement, asl, souche .18 Mais ce nom prometteur restera lettre morte faute dune hypothse sur le rle de la squence quil
nomme, faute de lhypothse dun plan gnral, 19 faute de la prise en
compte du temps.
Aprs la tradition arabe, la tradition orientaliste a vu dans cette
squence la racine de ces formes toutefois sans prendre en considration ni son rle, ni lorganisation de la langue quelle implique. En consquence elle posera la question du nombre des consonnes composant ces
squences non pas dans le cadre dune organisation globale reconstruite
mais uniquement en diachronie, la recherche donc dun nombre primitif et non pas dun nombre dtermin systmatiquement ou, sinon,
rfrentiellement : elle cherchera tablir si les racines primitives ont
compt deux consonnes ou trois consonnes.
Ainsi la tradition orientaliste a, dans son tude de la racine ,
pris le temps en compte mais non pas lhypothse dune systmatique
gnrale.
Les deux traditions, arabe et non arabe, accepteront donc des racines
de deux consonnes, de trois consonnes, de quatre consonnes et mme,
parfois, de cinq consonnes. Ce faisant, elles auront empch la recherche dune organisation de la nomination. En effet, une telle irrgularit
des signifiants de racines autrement comparables exclut la rgularit
essentielle tout plan densemble.
17
Sens secondaires : secrtaire, crivain .

Dans le Kitb, asl dsigne le principe (dune loi de la langue), ltat primitif dun
h arf (i.e. dun son de la langue) hors de tout conditionnement ou encore dun schme
qui est donc non anomal. Dans le Sh ib (29), le asl englobe ce qui traite de la constitution de la langue (mawd al-lua), de sa materia prima (awwaliyya), de sa construction (mana) puis des rgles suivies par les Arabes dans leurs communications (rusm
al-Arab f muxtabti-h), des ressources multiformes de lloquence (iftinn) qui sont
leur disposition sur les deux modes rel (tah qq) et figur (majz).
19
Ibn Fris est original par sa proposition, qui semble tre sans autre exemple dans la
tradition grammaticale arabe, dune saisie totalisante de la langue. La langue est, dans le
Sh ib, prsente, dentre de jeu, dune part, dans son plan gnral et, dautre part, dans
sa double relation un Dieu crateur et la crature humaine que ce Dieu a dou dune
parole qui est la matrice de sa pense.
18
508
andr roman
Dans toutes les langues du monde, les racines apparaissent comme les
squences de phonmes ou, sinon, de syllabes,20 qui chanent les units
de nomination des entits et des expriences que lhomme a inventes
dans le monde : /katb/, /kitbat/, /aktubu/, /ktib/, /maktab/ . . .
Les racines qui regroupent les entits, les expriences, reconnues par
lhomme comme des units parentes, sont la premire nomination de
ces entits, de ces expriences.
Les racines sont leur premier tablissement dans les langues.
Le premier rle des racines est de leur donner dans les langues une
forme et une dimension telles que les langues puissent les manier.
Le nombre des lments qui composent les racines doit rpondre
cette exigence de maniabilit.
Les racines, condition de leur maniabilit, ne compteront que quelques lments. Cest l une premire rduction, trs forte, impose par
les langues la nomination du monde.
Une deuxime rduction ncessaire est ralise par labstraction qui
aboutit donner, en effaant leurs diffrences, des entits diverses un
mme nom commun , des expriences diverses un mme verbe
commun .
Les racines, porteuses de ces sens abstraits, cest l leur deuxime
rle, devront compter, chacune, assez dlments pour satisfaire aux
besoins de nomination des entits et des expriences que lhomme veut
nommer.
Dans les langues smitiques, les racines des noms communs et des
verbes communs comptent, rgulirement, trois consonnes parce
que la combinatoire de trois consonnes est la premire combinatoire
mme de produire en nombre suffisant les arrangements qui seront
leurs signifiants.21
20
Les langues smitiques sont les seules avoir construit leurs systmes de nomination sur des racines de consonnes. Les langues tons les ont construits sur des racines
de voyelles. Les autres langues, sur des racines de syllabes. Les systmes de nomination
des langues smitiques et des langues tons vont se rorganisant sur des racines de syllabes.
21
Manifestement, les racines produites par la combinatoire de deux seules consonnes
sont en nombre insuffisant. Les racines produites par la combinatoire de trois consonnes
sont, par contre, en nombre plus que suffisant. Ce sont donc ces racines que la langue
arabe a utilises, rgulirement, dans son systme de nomination. Jamais elle na utilis
dans son systme de nomination de racines de deux consonnes. Les racines de quatre consonnes, secondaires, C1C :2C3C4, ont t inventes dans les formes du verbe
racines de trois consonnes et modalit ditration ; le signifiant de cette modalit tait
la longueur de la deuxime consonne, C :2 ; cette consonne longue a t rinterprte
509
Ce nombre trois , suffisant, est un nombre trs bas. Il est donc satisfaisant.22
La tradition arabe et la tradition non arabe ne poseront pas non plus la
question du statut des autres consonnes qui flanquent ces racines. La
reconnaissance par elles de //, dans /aktub-u/, comme tant la premire
personne du verbe ne touche que le signifi de cette unit de nomination. De mme lidentification par elles du tanwn /n/ comme un article indfini dans les autres formes cites. Quant lidentification de
/t/ dans /kitb-a-t-u-n/ comme un suffixe , de /m/ dans /maktab-u-n/,
comme un prfixe , elle est encore plus imprcise : ces deux consonnes, /t/ et /m/, sont, pour elles, des zawid, des lments ajouts , des
formans dans des wazn, des schmes , filat pour /kitbat/, maf al, pour
/maktab/. Mais ces wazn, ces schmes, ne sont que les mimes des formes
quils prtendent analyser.23
4. laboration en schmes
4.1
Schme 1
Il est incontestable :
que les racines de la langue arabe sont, gnralement, des racines de
trois consonnes ;
que les units de nomination construites sur ces racines reoivent,
gnralement, dans les phrases, une voyelle dsinentielle, brve, qui
est soit le signifiant dun mode, dans les seuls verbes, soit le signifiant
comme une consonne double ; cette consonne double a, par dissimilation dabord, donn
naissance deux consonnes diffrentes ; exemple : /faqqaa/ > /farqaa/, craquer qqc. .
Naissance dun nouveau paradigme. Quant aux racines de cinq consonnes, ce sont des
chimres. Voir Roman, 2005, le chapitre Une brve histoire de la langue arabe .
22
Il est remarquable quune morphologie de larabe toute construite sur des racines
de deux seules consonnes est possible. Mais elle ne suffirait pas aux besoins de la nomination. Au demeurant la nomination smitique, par racines de consonnes, a succd
une nomination par racines de syllabes, dj nombreuses ; voir A. Roman, op. cit., loc. cit.
23
En ralit, ces wazn, ces schmes sont produits en trompe-lil par le sous-systme syllabique de la langue. En effet, ses deux seules syllabes, CV et CVC, produisent mcaniquement des squences rgulires ; mais chacune de ces squences nest le
plus souvent que la figure dun sens global, la figure dune unit de nomination construite sur une racine syllabique, cest--dire sur un radical .
510
andr roman
dun cas dans les autres units de nomination (la voyelle /u/ dans
chacun des exemples donns) ;
que jamais une consonne na un signifi modal ou casuel.
Ces constatations suggrent un premier schma de la langue arabe :
SCHMA I
DE LA LANGUE ARABE24
CCC V
4.2
Les schmes II et III
Hypothtiquement une langue peut tre dfinie comme linterconnexion de deux systmes interdpendants : un systme de nomination
produisant des units de nomination, et un systme de communication,
sa syntaxe, dans le cadre duquel ses units de nomination entrent en
relation pour dire une exprience.
Dans le premier schma prsent comme lesquisse dune organisation gnrale probable, CCC reprsente, videmment, le systme de
nomination ; V , le systme de communication :
SCHMA II

CCC
systeme de nomination
DE LA LANGUE ARABE
V
systeme de communication
La raison de cette rpartition des tches entre les consonnes, seules

pouvoir entrer dans les racines de la langue, et les voyelles, seules
signifier les cas et les modes, se dcouvre dans la structuration du systme syllabique de la langue : en arabe classique, tout arrangement de
consonnes et de voyelles qui, dans le cadre dune syllabe, ne serait ni
CV , ni CVC , constitue une syllabe a-systmatique produite par
une contrainte phontique ou par une pause syntaxique ; un tel systme syllabique, , dtermine, dans le fonctionnement de la langue
la disjonction du sous-ensemble des phonmes consonnes, {C}, et du
sous-ensemble des phonmes voyelles, {V} ; en effet deux squences de
syllabes CV et CVC ne peuvent composer les signifiants de deux
units de nomination qui, elles-mmes, constituant une paire lgitime,
sopposeraient, en bonne rgle, par lopposition simple dune consonne
24
est pour racine ; C , pour consonne ; V , pour voyelle .
511
et dune voyelle ; ce que montre sommairement le colonage ci-dessous

de deux squences diffrentes :
CV C.CV
CV.C VC
dont lopposition {C vs V} est non pas simple mais double.25

Cette disjonction, ds lors que les consonnes et les voyelles peuvent tre
utilises indpendamment les unes des autres, permet effectivement
lattribution systmatique de tches diffrentes aux consonnes et aux
voyelles.
Cest ainsi que sest trouve constitue la structure fondamentale de la
langue arabe, des langues smitiques.26
Cette raison syllabique trouve, la rgularit de son effet implique
la rgularit du plan gnral qui a t suppos.
SCHMA III
= {CV, CVC}
DE LA LANGUE ARABE
{C}
Systme
de nomination
{V} =
Systme
de communication
5. Systme de nomination
5.1
Le temps
Dans lensemble des units du systme de nomination, quel lment

choisir qui serait susceptible de lordonner ? Le temps, sans doute.
En effet, le temps est essentiel aux expriences. En consquence il est
essentiel aux langues. Cest grce sa saisie et sa mmoire du temps,
25
En franais, diffremment, les paires { aorte vs porte }, { pote vs porte }
opposent une voyelle une consonne ; en franais, consonnes et voyelles peuvent occuper, dans les formes, les mmes positions.
26
Invitablement, les consonnes et les voyelles, disjointes par le sous-systme syllabique, sont conjointes dans les syllabes mais les racines slectionnent dans les syllabes
les consonnes qui sont radicales.
512
andr roman
que lhomme a pu crer ses langues. Heureusement, le temps peut tre

saisi, navement, comme une donne floue sans doute mais linaire et
simple.
Lespace, lautre coordonne du monde, est toujours ralis dans des
entits multidimensionnelles, souvent irrgulires, que leur complexit
a mises, demble, hors de porte des ressources dont les langues disposent mme si telle ou telle forme, telle ou telle silhouette, peut tre abstraite en un type : les dimensions trop nombreuses de lespace dbordent
les possibilits des langues.27
Les langues se sont donc constitues, avec la materia prima de la voix,
dans une premire classification, labore sur lopposition rfrentielle :
[+ temps]
vs
[ temps]
Elles ont, en consquence, produit, ncessairement, leurs units de

nomination,
soit comme des units inscrites dans un droulement apparent du
temps, dont le temps est lune des composantes, leurs modus exemple : /ktib/, crivant ;
soit, contre la ralit du monde, comme des units trangres au
temps, leurs res exemple : /ayx/, cheikh .
5.2
Le temps et la racine
Selon toute vraisemblance, le systme de nomination se sera construit,

sur les deux fondements reconnus : le temps et la racine.
En arabe, aucune unit de nomination, res ou modus, ne peut commencer par deux consonnes implosives. Le patron syllabique impose
leur sparation par une voyelle :
27
Le nom commun na pas t dtach de lespace. Il tait concret. Ou, plutt, son
caractre concret, ainsi dsign, par hypallage, est apparu quand lhomme a invent des
entits abstraites, des entits dtaches de lespace et par l-mme du temps, partir
donc, chaque fois, dun modus. Mais ce modus rifi, ce nouveau nom, reconnu comme
le nom dune entit noccupant aucun espace, ce nouveau venu, dsign comme un nom
abstrait , par la mme hypallage, na pas t signifi comme tel par un nouveau morphme : la langue, pour le discriminer na invent aucun morphme dnotant la prsence
ou labsence de lespace dans les entits nommes par ces noms. Aussi la premire langue
aura-t-elle t vocale, le geste tant une criture sur un espace suggr, sur une donne,
donc, non prise en compte par la systmatique de la langue.

*/CC . . . /
>
513
/CVC . . . /28
La premire voyelle des res communes construites sur trois consonnes

radicales, cette premire voyelle que le systme syllabique impose, semble avoir t le signifiant de lanimit.
Exemples :
/kalb/, chien
vs
/milh /, sel
Nulle voyelle entre les deuximes et troisimes consonnes radicales. Le

genre et le nombre, qui sont les autres dterminants propres aux res,
sont, systmatiquement suffixs.
Exemples :
/kalb-a-t/, chienne
vs
/kalb--t/, chiennes
La premire voyelle des modus communs galement construits sur des

racines de trois consonnes, cette mme voyelle impose, tait, est reste irrgulirement, le signifiant de la diathse, subjective ou objective.
Paradigmatiquement, les modus ne pouvaient se distinguer des res que
par la prsence dune voyelle entre les deuxime et troisime consonnes
radicales. Cest effectivement cette voyelle qui les a distingues des res.
Exemple :
/kalab/, rage
vs
/kalb/, chien
Cette voyelle, la h arakat al-ayn de la tradition grammaticale arabe, est,

dans /kalab/, le signifiant de la prsence du temps. Elle est le premier
signifiant du temps. Un temps non prcis.
/kalab/ est, dans la tradition grammaticale arabe, un masdar.
Le terme masdar, source, origine , a t choisi parce que certains
grammairiens, les grammairiens dal-Basra, les premiers semble-t-il, ont
fait de cette forme, hors plan, la forme mre du verbe.29
28
Ibn Fris, as-Sh ib, p. 3839 : inna l arab-a [. . .] l t-ajmau bayna skin-ay-ni
wa l t-abtadiu bi skin-i-n .
29
Voir Ibn al-Anbr, Al-Insf f masili l-xilf bayna n-nahwiyyna l-basriyyna wa
l-kfiyyn vol. I, chapitre XXVIII ; al-Frb, op. cit., p. 6. ; Fleisch, 1979 : 149 ; 1995.
514
andr roman
La tradition na donc pas dfini le masdar en lui-mme.
Il apparat ici comme un modus infinitif.
Lopposition du modus infinitif, /kalab/, la res anime, /kalb/, est un

exemple de la relation biunivoque, gnrale, , qui seffectue, dans le
systme de nomination de chaque langue, comme une opposition, vs ,
entre deux paradigmes ; exemple :
2 1
/katab + t a/
Tu as crit
vs
vs
1 2
/t a + ktub(u)/
Tu cris
vs
vs
/katab + t i/
Tu as crit (femme)
ou entre deux schmes ; exemple :

/katab + t a/
Tu as crit (homme)
Dans le systme de nomination, cette relation biunivoque reste une,

car, tant elle-mme opposition , elle ne peut gnrer, par opposition,
aucune autre relation, diffrente, qui serait univoque.
5.3
Des racines dune seule consonne
Les racines de trois consonnes ne sont pas les seules racines des langues
smitiques. En effet, ct des units de nomination commune, qui
sont des images diffrencies du monde, et sont donc trs nombreuses,
il existe dautres units de nomination non plus commune mais gnrale
ou banale qui sont des images indiffrencies du monde, peu nombreuses donc.30
Le grand nombre des units de nomination commune a impos leur
construction sur des racines de trois consonnes.
Le petit nombre des units de nomination gnrale ou banale a permis leur construction sur des racines dune seule consonne.
Les res banales sont :
la premire et la deuxime personne, remployes, tour tour, dans
chacune de leurs occurrences ; exemples : //, je , de racine , la
30
Banal , qui se met la disposition de tout le monde , Littr.
515
racine de la premire personne ; /ka/, te, de toi , de racine k, la

racine de la deuxime personne ;
les res montres, remployes semblablement, tour tour ; exemple : /
t/, celle-ci , de racine t, la racine des res dostension ;
les res reprsentes, remployes semblablement, tour tour ; exemple : /hu/, lui, le, de lui , de proto-racine *c, la racine des res de
reprsentation.31
Les res gnrales sont :
les res vagues, exemple : /m/, quoi ? , de racine m, la racine de la
res gnrale ; t, racine du temps gnral, homophone de la racine t
des res dostension ;32 n, racine du lieu gnral.33
Les modus de racines monoconsonantiques sont :
le modus /f/, avatar de lancienne racine, p, du modus gnral ;34
le modus /s/, galement ralis //, faire , avatars de lancienne
racine *c ;35
le modus /yy/, tre , de proto-racine *c ;36
le modus d assertion , de proto-racine *m ;37
31
La proto-consonne */c/, occlusive medio-palatale sourde, se transformera en perdant, par lnition, son occlusion : */c/ > *// > // > /s-/ > /h/ > /-/ ; ou, sa sourdit : */c/ >
*/ / > /y-/ /-yy/. Seules les consonnes qui sont affectes dun tiret sont encore employes
dans la langue arabe historique en tant que pices de son systme. La branche sourde est
termine par locclusive glottale // qui a remplac, contre-courant, la constrictive /h/
dont elle est issue. En effet // est moins diffrente que /h/ des voyelles devant lesquelles,
gnralement, elle se trouve. Quant /yy/, cest la mme consonne vocalique /y/, mais
elle a t allonge pour assurer sa dfense contre les conditionnements des voyelles car
elle est toujours, en fin de forme, en position intervocalique ; /y/, diffremment, est toujours en dbut de forme.
32
Lhomophonie de certaines racines monoconsonantiques est annule par leurs distributions diffrentes.
33
la diffrence de /m/, dans /m-aktab/, de /t/, dans /kitba-t/ (chacun de ces deux
morphmes appartient la forme dans laquelle il se trouve), /n/, est une autre unit
de nomination en fonction dexpansion annective de /maktab/. Unit de nomination
non spcifie, /n/ est la tte du paradigme des expansions annectives, spcifies, de la
langue arabe ; /maktab-u-n/ signifie, littralement, bureau dun lieu indtermin >
un bureau ; do lidentification traditionnelle de /n/ comme un article indfini ;
do son nom traditionnel, tanwn, cest--dire nounation , prsence de /n/ .
34
Exemple dans /kayfa/, Comment ? . Ce modus gnral, concurrenc par les deux
modus tre et faire tait sans emploi ; il a t cr par le jeu des oppositions binaires
constitutif de la langue. Il nexiste que dans /kayfa/.
35
Exemples dans les verbes de paradigmes / + af ala/ et /(i)s + taf ala/ : / + ahadta/,
Tu as fait que qqn tmoigne ; /(i)s + tahadta/, Tu as fait que toi-mme sois tmoin .
36
Exemple : /m + aktab i + yy/, bureaucratique ; littralement : tre bureau .
37
Exemple : */allh - u + m/ > /allh + u/, Allh.
516
andr roman
le modus d invocation , de proto-racine ;38

le modus d exclamation , de proto-racine .39
La combinatoire des consonnes dans le cadre des racines de trois consonnes est double par une combinatoire de racines.
Exemples :
/ - a - ktub u/, Jcris , construit sur les deux racines et k-t-b ;
/kitbat/, temps dcriture , construit sur les deux racines k-t-b
et t ;40
/m a ktab/, objet dcriture, bureau , construit sur les deux racines m et k-t-b ;
/m a ktab i yy/, bureaucratique , construit sur les trois racines
m, k-t-b et *c > /yy/.
Ainsi le systme de nomination exploitait deux combinatoires :
une combinatoire de consonnes radicales ;
une combinatoire de racines associant :
soit une racine monoconsonantique une autre racine monoconsonantique ;
soit une ou deux ou trois racines monoconsonantiques une racine
triconsonantique.
5.4
Units non construites sur des racines
Face aux units de nomination construites sur des racines, la langue

a produit des units non construites sur des racines. Ces pices de la
machinerie de la langue, ses modalits , sont dans le systme de
nomination les partenaires des racines. Elles jouent par rapport elles
un rle de dterminants .
la diffrence des racines, les modalits sont les images linguistiques
dentits du monde qui peuvent tre apprhendes, navement, non pas
comme des entits communes mais comme des entits universelles, le
temps, la vie . . . 41
38
Exemple : /a + ayx-u/, Cheikh !

Exemple : */allh - a + / > /allh + a/, Allh !
40
Exemples dans le verset II :280 : /fa nazirat-u-n il maysarat-i-n/, Alors attendre
le temps [quil faudra] jusqu ce que [le dbiteur] soit dans une aisance [suffisante] .
41
Navement, une entit peut tre perue comme universelle si elle est apparat toujours prsente. Sa prsence constante implique sa prsence partout. Luniversalit ainsi
39
517
Les modalits de temps sont employes particulirement dans les

conjugaisons. La modalit de vie a t employe comme une modalit
d animit , mle, femelle . . . 42
6. Les res, les modus et le temps

Les langues dtachent les res du temps.
Elles rattachent les modus au temps, or les res nexistent pas hors du
temps,
et les modus nexistent que dans ou par les res.
Les langues ne peuvent donc garder les res hors du temps et elles ne
peuvent davantage garder les modus hors des res. Il faut donc que, dans
le discours, elles accouplent les res et les modus.
Le premier couple {res modus} construit le noyau de chaque phrase
structure.43 Ses deux composantes sont ainsi relies, lune lautre, ipso
facto, par une relation intrinsque, biunivoque, galitaire : .
cette relation biunivoque, galitaire, de dclaration du lien, indissociable, entre deux units de nomination,44 sopposent, ventuellement,
deux relations univoques, les deux seules relations mme dentrer en
opposition avec elle :
lune galitaire, la coordination , + ;
lautre non galitaire, la subordination , .
entendue lie le temps et lespace, indissociablement. Et elle nimplique avec lespace et

le temps que la vie, qui nat du temps, qui est avec le mouvement dans lespace lautre
mesure du temps. Le temps et la vie . . . sont signifis par des modalits ; lespace nest
jamais dnot que par des noms .
42
Les modalits aspectuelles spcifient diffremment non pas la relation des modus
au temps mais, par hypallage, le droulement intrinsque des modus.
43
Les modalits aspectuelles spcifient diffremment non pas la relation des modus
au temps mais, par hypallage, le droulement intrinsque des modus.
44
Les phrases non structures sont les onomatopes, units unaires trangres
lorganisation de la langue. Les langues animales sont unaires.
518
andr roman
SCHMA IV
DES RELATIONS BINAIRES
[BIUNIVOQUE]
[UNIVOQUE]
[HIRARCHISE]
[NON HIRARCHISE]
DCLARATION

SUBORDINATION COORDINATION

+
Les trois relations ainsi dfinies, ds lors quelles sont les seules relations
possibles, sont communes toutes les langues.
Ds lors quelles sont communes toutes les langues, le systme quelles constituent est universel.
Ainsi le systme de communication sest tabli pour toutes les phrases
de toutes les langues du monde sur le plan suivant :
PLAN UNIVERSEL DE LA PHRASE
{x (+ e . . . ) }
{y
(+ e . . . ) }
noyau
+
extensions
extensions
{e (+... )}
{e (+... )}
+
extensions
{x (+ e . . . ) }
.
.
.
.
.
.
{y
.
.
.
(+ e . . . ) }
.
.
.
Dans ce plan, les units de nomination x et y sont, in prsentia

ou in absentia, les deux lments structurellement ncessaires du noyau
de la phrase, ses lments fondamentaux. La relation biunivoque, ,
qui les solidarise, les constitue en une structure unique : un duo de
deux voix .
Dans ce plan, x , y ..., e , e . . . sont des units de nomination ventuellement appeles par le locuteur, selon son besoin ; ce sont
des extensions, des complments sans aucune ncessit structurelle.
Un tel systme est simple.
519
Sa simplicit tient au trs petit nombre des oppositions en jeu, deux

seulement.
Sa simplicit tient aussi au caractre abstrait des trois relations nes
de ces oppositions.
Commun toutes les langues, ce systme est diffrent dans chaque
langue, superficiellement, par les morphmes spcificateurs propres
son systme de communication.
Le recours ces morphmes est rendu ncessaire par le caractre abstrait des trois relations en jeu. Et cest ce mme caractre abstrait qui le
rend possible.
Les morphmes spcificateurs propres la relation galitaire, biunivoque, de dclaration , sont principalement :
La diathse subjective ou objective ;
Le mode rel ou potentiel ou irrel ;45
Les modaux affirmatif, ngatif, interrogatif, impratif . . .
Les morphmes propres aux relations univoques sont :
les morphmes dfinisseurs de la relation univoque, galitaire, + ,
qui sont les coordonnants ;
les morphmes dfinisseurs de la relation univoque, non galitaire,
, qui sont les subordonnants , la modalit exceptive.
Ainsi les morphmes qui spcifient la relation biunivoque et ceux qui
spcifient les relation univoques sont complmentaires.
7. Lorganisation de la langue arabe

Ont t constates :
dans le systme de nomination,
la prsence dune structure radicale dans les units flchies ; labsence de structure radicale dans les units amorphes ;
la prsence du temps dans les modus ; labsence du temps dans les
res ;46
45
La tradition grammaticale arabe na pas reconnu les modalits de mode.

Cette opposition, fondamentale, {res vs modus} nest plus vivante dans les langues
historiques.
46
520
andr roman
la relation, biunivoque, dopposition, entre les paradigmes euxmmes et les schmes qui les composent ;
dans le systme de communication,
les relations biunivoque, de dclaration ; univoques, de coordination,
de subordination.
Ces quatre oppositions, binaires, retraces dabord, dans la langue arabe
primitive rvlent un plan de la langue arabe exactement binaire.47
ARBRE DE LORGANISATION GNRALE PRIMITIVE DE LA LANGUE
ARABE
[SIGNIFIANTS]
[SIGNIFIS]
[NOMINATION]
[COMMUNICATION]
[TEMPS]

+
[RACINES]
phonmes modus aspects res
animit
syllabes
8. La structuration binaire de la langue

La reconstruction prsente de la systmatique de la langue arabe ne
saurait surprendre. Cest cette mme reconstruction quaboutirait une
tude qui commencerait, non plus par lobservation de la langue, mais
par lobservation de lhomme.
En effet, la parole de lhomme est convenue, ncessairement : il ne
saurait parler au hasard.
Si la convention qui rgle sa parole tait brute, cest--dire amorphe,
chaque entit, chaque exprience du monde, serait dite, immdiatement,
absolument, par une onomatope.48
47
La relation biunivoque, , est d opposition dans le systme de nomination ; elle
est de dclaration dans le systme de communication.
48
Ouille ! , sil faut un exemple, ne peut que crier une douleur de linstant.
Ouille ! ne peut crier le souvenir dune douleur passe, la crainte dune douleur future.
521
Les limites, limprcision, dune telle langue sont manifestes.

Une telle langue ne peut tre la langue de lhomo sapiens, faber.
La convention qui rgle sa parole nest donc pas amorphe mais
structure.
La premire structuration qui existe est la structuration binaire ; au
de, cest--dire hors structuration, le son profr est unaire ; il est
ralis comme une onomatope ; au-del de la structuration binaire la
structuration n-aire (avec n suprieur 2) tait et reste inaccessible
lhomme sans autre recours que son corps. Lhomme a forcment structur ses langues binairement, ds lors que la structuration binaire est la
premire de toutes les structurations possibles et quelle est la seule quil
puisse matriser.
9. Conclusion
La systmatique de ses langues a t impose lhomme par sa capacit
de combinatoire binaire, son langage naturel, sa premire langue commune, la premire langue de ses conventions.49
Le plan primitif de lorganisation gnrale des langues humaines
naturelles, ne de la capacit binaire de lhomme, ne peut tre diffrent
du plan gnral de la langue arabe.
49
Les langues se sont constitues, par pas de deux , chaque signifiant propos
par le systme des sons trouvant, ventuellement, son signifi . Exemples de pas
de deux : C1V1C2C3 est le schma des res ; C1V1C2V2C3 est le schma des modus ;
C1uC2C2R3 est objectif ; C1aC2V2C3 est subjectif ; V2 = /u/, si le sujet nest que le lieu
du modus ; V2 = /a/, si le sujet produit le modus par son action ; V2 = /i/, si le sujet
produit le modus par sa raction ; si le sujet produit le modus par son action, la transitivit ainsi ralise est soit afficiente, alors V1 = /a/ ; soit dficiente, alors V1 = // . . . Jeux
dopposition qui produisent les diffrences ncessaires mais aussi jeux analogiques qui
produisent les paradigmes ncessaires et, parmi eux particulirement, les jeux iconiques
qui mlent les analogies abstraites et les analogies concrtes ; exemple : laspect achev est
signifi par lordre {C - CCC} des racines ; laspect inachev par lordre inverse, {CCC
- C}.
522
andr roman
ARBRE DE LORGANISATION GNRALE DE LANGUES
CAPACIT BINAIRE
[SIGNIFIANTS]
[SIGNIFIS]
[NOMINATION]
[COMMUNICATION]
[TIME]

+
[RACINES]
phonmes modus tenses
syllabes
res
animit
aspects
Lorganisation primordiale retrouve par la dmarche qui commence dans

ltude de la langue et par la dmarche qui commence dans lobservation
de lhomo loquens est lorganisation binaire, simple et puissante.
La capacit de combinatoire binaire est le seuil de lhominisation.
10. Rfrences
Brague, Rmi. 1992. Europe, la voie romaine. Paris : Criterion.
Carter, Mike G. 1972. Les origines de la grammaire arabe, Revue des tudes islamiques,
40, 6997.
Descartes, Ren. (15961650). 1987. Discours de la mthode, texte et commentaire par
tienne Gilson. Paris : Vrin.
al-Frb (m. 339/950). 1350/1931. Ih s al-ulm, dition tablie et prface par Utmn
Muhammad Amn. Le Caire.
Fleisch, Henri. 1961. Trait de philologie arabe, Vol. 1 : Prliminaires, phontique,
morphologie nominale; 1979, vol. 2 : Pronoms, morphologie verbale, particules.
Beyrouth : Dar El-Machreq.
Ibn al-Anbr (513/1119577/1181). s.d. Al-Insf f masil al-xilf bayna n-nahwiyyna
l-basriyyna wa-l-kfiyyn (publi avec le Kitb al-Intisf min al-Insf de son diteur
Muhammad Muhy d-Dn Abd al-H amd), 2 vols.
Ibn Fris (m. 395/1004). 1382/1963. as-Sh ibi f fiqh al-lua wa-sunan al-arab f kalmih, d. Musta f uwaym. Beyrouth: Badrn.
Kleiber, Georges. 2003. Faut-il dire adieu la phrase ?. LInformation grammaticale,
98. 1722.
Lavorel, Pierre Marie. 1980. Aspects de la performance linguistique Contribution
neurolinguistique et psycholinguistique lanalyse des systmes langagiers, Thse dtat.
Lyon : Universit lyon II.
523
al-Maydn Ahmad (m. 518/1124). s.d. Majma al-amtl, d. Muhammad Muhy

d-Dn Abd al-H amd (2 vols.). Damas-Beyrouth : Manrt Dr an-Nasr.
Meillet, Antoine et Cohen, Marcel. d. 1952. Les langues du monde, nouvelle dition,
CNRS, 2 vols. Paris : Atlas.
Mounin, Georges. 1960. Dfinitions rcentes du langage. Diogne 31, 99112.
Roman, Andr. 1998. Les particules dcomposes ou la reconnaissance des composantes des morphmes de ngation de la langue arabe . In D. Leeman et A. Boone,
eds. Du percevoir au dire. Hommage Andr Joly, Paris : LHarmattan, 8796.
. 2005. La cration lexicale. 2e d. Lyon-Kaslik : Presses Universitaires de LyonPresses Universitaires de Kaslik.
Saussure, Ferdinand de (18571913). 2002. crits de linguistique gnrale, texte tabli
par S. Bouquet et R. Engler. Paris : Gallimard.
Schuler, Bernard. 1990. Rhetorica. Rhetorica, 8, 229254, University of California
Press.
Sbawayhi (m. circa 180/786). 13851397/19661977. Kitb, 5 vols. Abd as-Salm
Muhammad Hrn, ed. Cairo.
Troupeau, Grard. 1976. Lexique-Index du Kitb de Sbawayhi. Paris : Klincksieck.
Versteegh, Kees. 1977. Greek Elements in Arabic Linguistic Thinking. Leiden : E.J. Brill.
. 1980. The Origin of the term qiys in the Arabic grammar. Zeitschrift fr arabische
Linguistik, 480, 730.
. 1993. Arabic Grammar and Qurnic Exegesis in Early Islam. Leiden : E. J. Brill.
. 1995. The Explanation of Linguistic Causes : Az-Zajjjs theory of grammar Amsterdam : John Benjamins.
DIALECTS
HOW TO BE KOOL IN ARABIC WRITING:

LINGUISTIC OBSERVATIONS FROM THE SIDE LINE
Gert Borg
Nijmegen University
1. Introduction
Students of Arabiccertainly those of a former generationhave often
been led astray by a fatal fallacy: the Arabic language they study, in its
grammar and syntax, shines out as a neat structure of blissful regularity. They learn by heart a set of rules with a high rate of predictability,
that govern the language that was used to convey Arab thought and creativity for 15 or 16 centuries. One might say that just like the Orient
is the dream of the Orientalists, Arabic seems to be the dream of the
Arabists.
But most Arabists are confronted with the harsh reality of the linguistic situation in the Arab world during their first visit: a reality called
diglossia. It is bitter in a way to have to learn a second Arabic language
in order to be able to survive in everyday situations.
The actual situation of the foreign learner in the Arab world is not very
helpful either: if he or she gives the impression of being able to speak
some Arabic, the addressee may well decide to speak a higher variety
of the language out of politeness and respect, or start rattling in vernacular assuming that for a foreigner to speak MSA would imply even
more strongly that he or she has an easy command of lower language
levels. As a result the linguistic confrontation between native speaker
and foreign language learner in the Arab world is hardly ever a natural
process.
Nevertheless the gifted and dedicated learner of Arabic can achieve
astonishing results in mastering this complicated linguistic situation.
Diglossia of course has some consequences for the teaching process
of the Arabic language; fundamental choices have to be made. Do we
teach Modern Standard Arabic, one of the modern dialects or some
kind of mix between the two? Must this mix be presented synchronically or should it be taught diachronically, in this case leaving open the
question with which variety of Arabic to begin.
528
gert borg
The choices and options become aggravated by growing pressures to

make university programs more time efficient and productive. Andas
if this were not enoughtraditional academic ambitions aim at a wider
study of history of the Arab world, of culture and of religion as well,
apart from the obvious purpose, that students of Arabic Studies are
expected to know their way around in Arabic and in the present day
Arab world.
Although we usually consume Arabic in written form, we soon come
to realize, that the linguistic reality of the Arab world is by no means
as static as the written material suggests: among Arabs new needs and
concepts emerge, new answers and solutions to the diglossic situation
are implicitly or explicitly formulated and this momentumshaped by
the mobility and flexibility of actual peopleis unfolding in a typically
linguistic way: anarchy.
It is obvious, that something static as a language course, a grammar or
even recently collected audiovisual material cannot possibly follow the
volatile dynamics of this process.
2. Ih n, sawt jl bi-h luh

In this contribution I will discuss a single phenomenon that represents
the present stage (August 2005) of this flexibility and anarchy in written
Arabic in Egypt: the monthly Ih n, sawt jl bi-h luh is a glossy magazine in its first year. It is printed and distributed in Cairo where it sells
for LE 5 (= around 0.75). The names of the main editors are d arf
al-Jazr, Jhn Mahmd and arf al-Alf. The number of contributing
editors (muh arrirn) is 13.
It is not easy to identify the intended readership, but on the basis of
the various contributions and the goods advertised in it (mobile phones,
cars, beer, traveling and leisure) one might assume roughly that the
readers would be between 18 to 30 year olds, well to do and with a wide
range of interests from pop music, sex and vintage cars to (anti-American) politics, recent Egyptian history and backgrounds of hot news
items in Egypt and abroad.
The general character of this magazine can be qualified as:
critical
easy to read
cultural
open minded
fairly independent
humorous and satirical
how to be kool in arabic writing
529
The political sympathy of the editors lies apparently with the Kifya
popular movement in Egypt.
The most significant feature of this magazine is its use of two language levels in various ways: contributions in MSA, others in Egyptian
colloquial (mmiyya) and some with a mixed use of language.
I conducted a basic round of questioning about the opinion of the
reading public and the following image emerged: it is immensely popular, also in circles that are not intended, like 14 year old school girls and
their mothers, who appreciate the easy reading of this magazine. Some
readers however regret the use of colloquial whichin their opinion
hampers children in learning real Arabic, but they enjoy the reading
nonetheless.
In a lecture at the Netherlands-Flemish Institute the Egyptian linguist Madha Doss emphasized, that the editors of this magazine use
Egyptian colloquial, not because they do not know how to write MSA/
fush they actually seem to publish in MSA elsewherebut that they
choose to use colloquial according to the character of this particular
magazine.
To illustrate the use of modern colloquial in Egypt and to give an
impression of the purposes for which it is used I selected one issue of
this magazine to be discussed here: nr. 8, published in August 2005.
2.1
Fragment 11
Let us first have a look at a contribution that is purely in MSA. It is the

beginning of a contribution entitled: ahr al-irhb al-aswad2 . . . black
month of terror, about the terrorist attacks in London and the murder
of the Egyptian Ambassador in Iraq:
(yawm al-irhb al-aswad) kna hd huwa l-unwn ar-ras li-jardat
al-ahrm as-sdira sabh yawm al-juma [blank] yliy wa-knat
al-jarda bi-hd al-unwn turu il l-ah dt al-irhbiyya llat waqaat f
l-yawm as-sbiq, h aytu knat iddat infijrt qad waqaat f qalb al-sima
al-bartniyya london mdiya bi-h ayt al-aart wa-kna tanzm
al-qida f bild al-rfidn qad alana an qatl as-safr al-misr f l-irq
hb arf . . . ,
1
By the nature of this contribution many quotes are from texts in which it is not
always clear whether the Arabic should be understood as MSA or mmiyya. Therefore
the transliteration may sometimes be inconsistent.
2
Cf. pp. 1011. All loose page numbers refer to this specific issue of Ih n.
530
gert borg
(a summary of al-Ahrms report of the terrorist attacks in London and
the killing of the Egyptian Ambassador in Iraq).
Up to this point the choice of language and register may have been dictated by the source that is quoted, al-Ahrm, but after this the article
continues:
ill anna al-irhbiyyn lam yaktaf bi-hd al-yawm li-l-qiym
bi-hajamtihimi llat stamarrat tiwl ahr yliy mustahdifah (sic: h)
amkin mutafarriqa min al-lam. Fa-qad istayqaza al-misriyyn sabh
yawm as-sabt 23 yliy amalan f l-istimt bi-ajzt at-tawra qabla an
yufja bi-anna bosla at-tafjrt al-irhbya qadi ttajahat il madna
arm a-ayx al-misriyya . . . (the author turns to the recent terrorist attacks
in Egypt).
In his own words the editor continues the language and register as set
by the caption, referring to the terrorist attack in Sharm el-Sheikh. The
remaining part of this article is more of the same, a language register
that I would qualify as just above the level of average Media Arabic. The
level even rises a little towards classical in a context in which the editor
compares these erring groups of Muslims ( firaq islmiyya dlla) with
the Xawrij, quoting extensively from H adt and Qurn.
The choice of language in this contribution is quite predictable:
commentary on news item o Media Arabic or higher
religious/historic context o Classical Arabic
2.2
Fragment 2
The Ih n-editor al-Alf conducts an interview with an elderly man who
is fishing from the kbr ag-gmia. The photographer he has with him
is named Huseyn.
ajba h usayn f taraddud la, mi hanistann li-h adde m samaka titla.
ibtasam ar-rajul wa-ml li-yaftah antatah wa-axraj minh samaka
ka-annah muidda xiss san li-t-taswr wa-abakah f xutt f as-sinnra
wa-rh yahuzzuh h att tazhar ka-annah sh iya wa-btilab amm
al-kmr
(Huseyn refuses to wait to shoot his photographs because of the failing
light:) No, were not going to wait till a fish pops up. The man smiles
and turns to his bag, producing a fish as if prepared for the photo, hooks
it to the rod again and starts shaking it till it looks like its alive, playing
for the camera
531
saaltuh d nh h y ustd h asan?. anm. anm?. anm . . bil-mm. amm?. anm. anm . . . ah . . .
(the interviewer asking about the kind of fish: a very realistic conversation
indeed).
2.3
Fragment 3
A truly remarkable article is wijha nazar,3 point of view, a contribution physically surrounded on its page by a quotation from the Qurn:
rabban l tuxidn in nasn aw axtan . . . (Q. 2/286). The title of
this contribution is y rabb . . . anta rif (O Lord, You are aware . . . )
qla llh: wa-id saalaka ibd ann fa-inn qarbun ujbu dawata
d-di id dan (= Q. 2/186) y rabb anta qarb minn, lh ana bad
annak? y rabb anta rif kulle h ga, wa-rif (q)adde h ana muhtglak
wa-rif (q)adde h gahl wa-duf byixalln abad annak wa-amil h gt
tzaallak aw mamal al-h gt ell turdk aw amilh bass mi bi-t-tar(q)
a ell turdk. Ana rif ana (q)adde h wih i mak, wa-rif enna l-h gt
wi-niam ell anta addthl akbar min ayye h ga a(q)dir amilh
alanak, lh ba(q) l-whid byistashal ennuh yamil (q)alle h ga wahuwa fkir ennuh kedah xals amal kull al-matlb? . . . lh al-duny dalma
kedah quddm al-wh id wa-lh al-h ayt saba? Ana mi rif wa-mah ad
rif lh kulle h ga btih sa l f l-duny btih sa l kedah lh . . . I feel guilty about
all my wrongdoings in the light of your benevolence, o Lord, and I dont
understand the world anymore . . .
This text is very personal in tone, it is a monologue with God, almost

like a prayer and it is in mmiyya throughout, except verbal forms like
turdk.
What we have here is a religiously inspired text, in the context of
Qurnic Arabic, but written for a purpose that isapparentlybest
served in colloquial.
2.4
Fragment 4
H ukkm l yahbna amrk, an article about world leaders who defy

the US and its politics,4 is also mainly in MSA and Media Arabic. Typical
markers for this second type of Arabic are for instance passages like:
3
4
Cf. p. 18.
Cf. pp. 2526.
532
gert borg
tumma dahab5 Castro il qatar li-tazz al-alqt bayna l-baladayn
xssa tan wa-anna l-alqti llat tarbut qatar bi-l-wilyt al-muttah ida
watqa. tumma dahab il sriy wa-wasafa ziyratahu li-dimaq bi-annah
muhimma wa-annahu yaqif il jnib sriy . . . . . . .etc.
. . . then Castro travelled to Qatar to strengthen the relations between the
two nations, especially because Qatars relations with the US are solid.
Then he went to Syria and described his visit to Damascus as important
and (stated) that he stood next to Syria . . .
The type of language corresponds neatly to the subject of this contribution.

2.5
Fragment 5
Announced as a grain of seriousness (h abba jadd) we find another

contribution in MSA on the daring subject of the role of sex in matrimonial relations: m hiya h udd al-alqa l-h amma bayna l-azwj? 6 (what
are the boundaries of the intimate relationship between married partners?). With captions like an-nazar il awrat az-zawja (looking at your
naked wife), h urriyyat al-wad al-jasad atn al-jim (freedom of body
positions during intercourse), it does not shy away from daring subjects.
The whole article is written in MSA except for some anonymous reactions by boys (ray a-abb) and girls (amm l-fatayt fa-tah addatna
ka-t-tl). These reactions follow separate patterns:
in the boys section the introductory remarks are in MSA, which
gradually turns into mmiyya. Finally the boys are cited in full
mmiyya:
wa-al l-rami mimm qad yazunn al-bad min annahum l yatahaddatn
f tilka l-mawd . . . . ad-dukr arab an rayihim bi-bagh a7 wa-kaannan gn lahum a-ag-garh . qla ahaduhum: ih n makbtn8
wa-madtn min (!) al-mugtama . . . . although some expected them not
to talk about such subjects, the men expressed their view with impudence
as if we were touching a sore point. One of them said: we are oppressed
by society . . .
5
The use of the verb dahaba is extremely rare in Media Arabic: compare the Nijmegen Arabic corpus by Jan Hoogland.
6
Cf. pp. 2829.
7
A typical masdar for mmiyya, not found in classical and MSA dictionaries, but in
Badawi-Hinds.
8
A typical mmiyya word.
533
the girls section is less predictable:

wa-ih d l-fatayt qlat f tamallul: wa-h-yifd bi-h marifat ray f
al-awd ag-ginsya . . . whats the use of knowing my opinion about sexual positions!
wa-law yiz tifham nafsiyyat al-bant mila izzy f l-mawd dah f flm
woman on top fa-huwa yah k qissa t fatt rmnsiyya tazawwajat bi-bb
wasm (if you want to know how womans psychology works in this matter,
than look at the film woman on top; it tells the romantic story of a girl
that married a handsome boy . . . )
Then the story of the film is being told in MSA.
2.6
Fragment 6
A constant mix between MSA and colloquial is a contribution by Yumn

Bassiouni, called al-istina bi-mumatti ln kmbrs min ajl az-ziyrt
ar-rasmiyya,9 calling in extra representatives at official visits for help,
a hilarious story about the hiring of extra patients to accommodate
the official opening of a hospital wing for children on the 4th floor of
the Ab R Academic Hospital. The first line of this story is indicative for its mixed character: wa-alan uakkid annahu mi mugarrad
kalm, hunka mitln h adat muakka ran yuakkidn al mawd
al-istina . . . to make sure that its not only a rumour, here you have
two examples that occurred recently, confirming the topic of calling
in . . . .
From a linguistic point of view the use of mmiyya in this article is
interesting: a remarkable colloquial factor is the use of mmiyya conjunctions instead of MSA ones: apparently this contributes to the fluent
style of writing:
. . . h att az-zar (!) byitl alan yith att f makn ziyra rasmiyya tnya
(for li . . . )
. . . wa-lkin inn al-mawd yawsal (for lkinna)
. . . wa-alan uakkid . . . (for li . . .)
. . . - zayye m binf f t-tilifizyn(for kam)
. . . kullu deh wa-l-umr m zlat tabd tabiyya li-l-ya . . . (for hd
wa-)
Cf. p. 31.
534
gert borg
The mmiyya can also be used to indicate direct speech without using
quotation marks. Doing this makes the scene lively and realistic, because
we can hardly do anything else than picturing the usual hanger around
in such buildings, all the more so because typically the floor that is said
to be still closed is the floor that was meant to be inaugurated from the
beginning:
. . . f d-dr ar-rbi ell lissa muftatah nuh . . . on the fourth floor that
they didnt open yet.
2.7
Fragment 7
Ih n apparently did some investigative research into the behavior of

female singers on the various MTV-like channels and the impact it has on
the audience in an article with the title man tufaddil min al-muanniyt?,
Which of these singers do you like most?.10 Some youths give their
commentary in which the choice of language seems significant for
emphasizing the point of view. A conservative 23 year old boy says:
masxara al-an ntija an s ahwl al-mujtama wa-laysat al-sabab fh
wa-l-ns mudnibn h aytu annahum hum yuajjin tilka al-an . . . the
cause of those ridiculous songs is the deterioration of society, not the other
way around; people have to blame themselves because they are encouraging these songs . . .
A 25 year old girl sounds a bit more committed:

al-fdiy klb ba isff wa-axjal min muhadat qanawt al-an amm
wld video-clips are just pulp; I feel embarrassed to watch them when
my father is around
This 24 year old girl is very dismissive:

hirt hd az-zamn asbah na yulinna an anfusihinna f t-tilifizyn f
akl fdiy klb the whores of this age have come to express themselves
through the video-clip
And this 27 year old girl sounds critical:

law nrakkaz al urbitn mumkin namil h gt original bass al-Arab
(ih n yan) dyn if we concentrate on our Arab identity we could come
up with something original, but the Arabsthat means weare lost
10
Cf. pp. 3233.
535
A 36 year old male:

bs samr law m amalit kedah mh ad hayiufh if Busi Samir wouldnt
act that way, nobody would notice her
The background of the speakers shifts from:
conservative and intellectual to

sympathetic but dismissive
conservative and denouncing
conservative and nationalistic
realistic
and with these attitudes the language variety shifts correspondingly from:
MSA to
popular (mmiyya)
MSA
street wise (mmiyya)
popular (mmiyya)
2.8
Fragment 8
Walid Irfas contribution is a review of Paulo Coelhos book The Zahir,

which was popular at the time. The language level chosen for this article
is close to Classical Arabic.
2.9
Fragment 9
A special page is reserved for reviews of DVDs. These are all in MSA. The
writer, Marwn Qadr, also comments on the Broadcast and Television
Festival in July 2005.11 On this subject his criticism is harsh: wa-ka-da
kull al-qimn alayhi (sc. al-mahrajn) faal al-mahrajn faalan daran
min h aytu t-tanzm wa-l-iftith wa-l-xitm . . . as usual with all organizers the festival failed completely in organization and during opening and
closing sessions. Only once Qadr deviates slightly from MSA: muzam
ad-duyf wa-l-mutarikn lays . . . , most guests and participants were
(plural) not . . . , but this is far from abnormal in Media Arabic. When
11
Cf. pp. 3637.
536
gert borg
the author voices his indignation about the unavoidable balad/fellh

dancer performing at these occasions he shifts to mmiyya:
l in anh f ayyi htifl h att wa-law kna tahr ibn al-jrn wa-llad
yablu min al-umr 7 sanawt. Lh .. lh al-raqsa l-fallh ? H arm alk
zihiqn gaddad, huwa int mi btitfarrag al ayye stirdt aw h aflt
xitm aw iftith f ayye h itta barra masr. Kulluh km wa-l-wufd aw
bi-man asah h ruas al-wufd km tn . . . there is no avoiding her
at any festive occasion, even if it be the neighbors son circumcision whos
only seven years. Why . . . why this peasant dancer? Shame on you, were
fed up with it, find something new(?), its because you havent seen any
festivities of openings or closures anywhere outside Egypt. All that is one
thing, but these delegations or more specifically the delegation leaders are
another!.
2.10
Fragment 10
In an article with the title bsbr axdar m yisw12 (a green passport

isnt worth anything) Yumn Bassiouni complains about the first class
treatment visitors to Egypt receive from the authorities if they carry foreign passports. As witnesses to this practice she cites a few personal stories, parts of which are in mmiyya. One account is by Mohamed Sami
(22) who traveled in Egypt in the winter season with a relative of his
who carries an American passport:13
ya ibn xlat f amrk wa-ndiran m yat il misr . . . . . wa-kunn
rjina min xarga14 bi-l-layl mutaaxxir(!) jiddan . . . wa-awqafan addbit wa-kna kik(!) fn li-anna ibn xlat kn lbis rt f izz al-bard!
wa-awdaha muhammad anna ad-dbit akka f annahum mutatn
h ga muxaddara muxalliyhum mi h ssn bi-nafsihim wa-lbsn kedah,
xssa tan wa-anna lukna qarbih f l-arab mi salma wa-sammam addbit annahum yatl a al al-qism. al-h ga l-wah da ell anqadatn
annan gibn li-d-dbit gawz safar ibn xlat al-amrk wa-hadadnhu
annan hanatakh f s-sifra l-amrkya an al-bahdala ell h asalit/hasalat
la-n wa-bi-t-tl mumalat ad-dbit itayyarat tamman wa-tadar kitr
alan m namil makil! . . . yan ibn xlat lamm kn byistamal
al-bsbr al-masr maa dalika d-dbit , m la r s mumla wa-awwal
m azhar bsbruh al-agnab kn lahu al-ihtirm a-add wa-l-adab f
l-mumala
my cousin lives in the US and rarely comes to Egypt . . . we were returning from Kharga very late at night . . . and a policeman stopped us and
12
13
14
Cf. p. 41.
Passages in colloquial are in italics.
Written is xurga which seems improbable.
537
suspected us because my cousin was wearing shorts in this extreme cold!

Mohamed made it clear that the policeman was wondering whether they
had taken some drug that left them not feeling themselves while they were
dressed that way, especially because the accent of his relative in Arabic was
not ok. The policeman decided that they had to go to the police post. The
only thing that saved us was that we handed the policeman my cousins
foreign passport and we threatened him that we would complain about
him at the American Embassy about the illtreatment that we encountered.
All of a sudden the policemans behavior changed completely and he apologized deeply so that we wouldnt cause him any problems! So when my
cousin used his Egyptian passport (in dealing) with that policeman, he
encountered maltreatment and as soon as he waved his foreign passport
he was treated with deep respect and decency.
In this passage the story changes rapidly from account to direct speech
and the language changes accordingly.
2.11
Fragment 11
In the first lines of the contribution tanfsah15 (reassurance?) we find a

remarkable blend of MSA and mmiyya:
Kun Anta. Ql iraf nafsak dah ahamm wa-asab wa-aktar al-ns
l yarifn!! Amm d h ga arba kulle m inn abuss f bitqat arif ana
mn wa-ahdat at-taxarrug wa-ahdat ag-g wa-al-bsbr . . . Be yourself. They say: Know yourself. That is the most important and the most
difficult thing, while most people dont!! That is something strange: every
time I look at my ID I know who I am . . . , my degree, my army card and
my passport . . .
But then the mmya takes over almost completely:

Huwa deh ana wa-l deh elli itfarad alayy enn aknuh . . .
Amil ell inta yizuh . . . yiz arabiyya bi-rmt bitru mayyah (unzur
namdaj 2)16 yiz tilab mazk! Yll nirh ri muh ammad al bukra
mumkin yigibak gtr wall darabkah . . . yiz tiqr!?wa-llhi fikra mi
batt laindin maktabt fh kull al-mawdt ell titxayyilh, mamak
fuls?! inzil sq17 al-ezbekya fa (!) al-ataba . . .
That is me and not the one who imposed on me that it is me . . . Do whatever you want . . . You want a car with remote control that sprays water (see
15
16
17
Cf. p. 42.
Referring to a photograph.
The text reads sr, but that seems improbable.
538
gert borg
example18 2). You want to play music!19 Lets go to Muhammed Ali Street
tomorrow. Maybe youll like a guitar or a darabukka . . . you want to read
sure, not a bad ideawe have bookshops with all subjects that you can
imagine, you dont have money?! Go to al-Ezbekiyah and Ataba.
In this fragment MSA is used for the general introductory remarks; the
more subjective fictitious monologue is in colloquial.
2.12
Fragments 1216
A few remaining articles about (muscle) cars, sporting exercises, quality food, gadgets and the unavoidable horoscope are all completely in
MSA.20
2.13
Fragments 17 and 18
In a magazine that is so obviously playing with the possibilities that language levels offer it is only natural that the reader would find contributions about language itself. In this issue of Ih n we find two, both under
the common and significant title Arbk slng.21 We find two lists of
say and dont say items (qul and l taqul), here represented as
opposites:
Say
matar
silsila
gurnl
bartamn
fingn
Dont say22
natar
sinsila
gurnn
batramn23
fingl
Say
dafar
muhandisn
bsbr
isbniy
dalwaq/t
Dont say
dfa/ir (?)
muhandizn
bzbr
izbniy
dawaq/t
And under the heading qms Ih n (Ih ns dictionary) we find a quasi-classical explanation of the word sakalnsun (sic):
kalima taxdum jam al-ard al hasb al-man l-murd: yumkin
istixdmuh bi-man kullih f t-tamm kulluh f s-sakalns, wa-yumkin al-ira li-l-by frnd aw al-grl frnd al annahum as-sakalns.
18
Example refers to a photograph accompanying the text.

Mark the calque for azaf.
20
Cf. pp. 4450.
21
Cf. p. 6.
22
These lists seem to be provocative to me; please go ahead and say gurnn!
23
Badawi-Hinds mentions both but apparently prefers batramn, a glass or plastic
jar with a lid, a jam jar.
19
539
kam yumkin al-qawl bi-anna axs (m f t-tarwa?) bi-annahu f

s-sakalns. s-sakalns tustaxdam f l-qahw li-tasmiyat al-kktlt:
qahwa al yzabd bi-l-fawkih . . . lx. f l-matim, as-sakalns tur
il sndwitt al-fl bi-t -tamya bi-l-bayd aw sandwitt as-samak bi-ggambar bi-s-sabt (?) bi-t-tahna wa-llat turaf aydan bi-sandwitt
al-fiygr. sakalns tamsr li-kalima aksalns . . .
an all purpose word depending on the intended meaning: it can be
used as a generic term: its all sakalns. It can refer to your boy friend
or girl-friend: they are sakalns. One can also say that someone . . . (?) is
in sakalns. sakalns is being used in coffee shops to indicate cocktails:
coffee on tea, cream on fruit etc. In restaurants sakalns refers to bean
sandwiches with ta mya and egg or fish sandwiches with shrimps with ?
with ta hna that are also known as viagra sandwiches. sakalns is an egyptianisation of the word excellence.
A wonderful play of words and styles.
3. When is mmiyya being used in this magazine?

The use of mmiyya seems neither inconsistent nor indiscriminate in
this magazine; I will try to clarify the preference of the contributors for
using mmiyya in its various contexts:
1. mmiyya is used for representing direct speech and to add a realistic
flavor
la, mi hanistann li-h adde m samaka titla (fragment 2)
also for fictitious direct speech: yiz tiqr!?wa-llhi fikra mi
batt la(fragment 11)
2. for adressing the reader
zayy m binf f t-tilifizyn (fragment 6)
3. for contexts with humorous, hilarious or ironical purposes
f d-dr ar-rbi ell lissa muftatah nuh (fragment 6)
h att z-zar byitl alan yith att f makn ziyra rasmya tnya
(fragment 6)
4. for intimate and personal expression
fragment 3: the prayer
5. for clarifying the social status or political point of view of the speaking individual
the interviews in fragment 7
540
gert borg
6. for emotive contexts, like anger or indignation

the way indignation that is expressed about the baladi dancer in
fragment 9
fragment 5: ih n makbtn wa-madtn min al-mugtama
4. Orthography
It is obvious that for an Arabic periodical in most Arabic speaking
countries to use written mmiyya is like walking a minefield: it will be
frowned upon by the conservative cultural and intellectual elite and the
periodical might very well be unable to penetrate a wider Arab market
although in the case of Egyptian colloquial this risk is limited. It opens
however a thrilling space of new and unexplored opportunities. It has to
be said though that this experiment faces some practical problems, one
of which is orthography.
For Classical and Modern Standard Arabic the orthography is fairly
straightforward. Even in written Media Arabic the orthographical problems are mainly confined to transliteration of loan words and foreign
names. But for written colloquial the situation is completely different:
the Arabic graphemes do obviously not cover the mmiyya stock of phonemes and no consistent convention has been developed yet. If this trend
of writing colloquial will persist, it would mean, that new orthographical
standards will emerge. It will be hard to continue the standard of oneto-one equivalents as in written CA and MSA. For written colloquial
the orthography of the future might well be less stable and predictable.
In a few instances we can already get an impression of some of the
difficulties: do we spell as we speak or spell as the conventional spelling
of MSA tells us to spell: for example the (ayn) is clearly disappearing
from the urban Cairene dialect. As a consequence Ih n writes byarif
for the MSA yarifna. To do this on the cover even seems to be meant
as an elegant provocation, because the same phenomenon ( for ) cannot be witnessed in the remaining articles except for the article that the
cover announces: istift ih n: 95% min a-abb byarif (with final
alif ) man bs samr . . . inquiry by Ih n: 95% of the kids know who
Busi Samir is (fragment 7).
In other issues of Ih n, not quoted here, I noticed a tendency to
represent stressed vowels by lengthening them, using matres lectionis,
which may be confusing for the unaccustomed eye.
541
Furthermore we find inconsistencies for instance in prepositions with

the first person suffix: liy, alayy, but biyyah.
The conjunctions wa and fa are increasingly represented and
therefore considered as separate words, not only in advertisement, but
also within running contexts like these: kull an-ns knit rfa maslaht
aktar minn: mm wa bb, al-mudarrisn, . . . everyone knew better
than me what was in my interest: ma and pa, the teachers, . . . (from
fragment 11); inzil sq al-ezbekya fa al-ataba . . . go to the Ezbekiya
market or Ataba (fragment 11).
5. Conclusions for teaching

For the present and coming generations of students of Arabic it might
be important to become acquainted with the notions and fashions of
their own generation in the Arab world. For them to learn about differences and similarities will probably contribute to motivation and
mutual understanding.
These notions and fashions are for a large part expressed in language.
The present state of the Arabic language and its day-to day developments cannot however be covered by course books and vocabularies
because these cannot possibly keep up with the tempo of modern urban
life, the media or the internet.
The pace of changes in the Arab world as reflected in the quick development of the written Arabic language confronts us, teachers of the
Arabic language and culture, with the question whether we can still
convey the concept of the written Arabic as a continuum; or should we
actually split the current language acquisition curricula into the categories classical and modern? The implications of options like these are
far reaching.
There is at least one practical consideration: teachers of modern Arabic can hardly do without regularly being in touch with the Arab world
in order to adequately follow the linguistic developments in the Arab
world. This can be achieved either by following the (satellite) media
orbetter stillby regular visits to Arab countries to witness the fast
developments firsthand.
542
gert borg
6. References
Hinds-Badawi: Hinds, Martin and El-Said Badawi. 1986. A Dictionary of Egyptian Arabic: Arabic English. Beirut: Librairie du Liban.
Ih n, sawt jl bi-h luh. 2005. August issue.
Nijmegen Arabic Corpus: a collection of MSA texts from the written media compiled by
Everhard Ditters and Jan Hoogland.
HELLO, I SAY, AND WELCOME! WHERE FROM,

THESE RIDING MEN?
ARABIC POPULAR POETRY AND POLITICAL SATIRE:
A STUDY IN INTERTEXTUALITY FROM JORDAN1
Clive Holes
University of Oxford
1. Introduction
In Arab literary studies, popular poetry, that is, poetry composed in a
non-standard form of the language, remains a relatively unexplored reservoir of creative activity. There has been a tendency for native and western critics alike to ignore it, or at best pigeonhole it as folklore, devoid
of literary value, and written in a debased form of the language. One
of the few great Arab writers to stand out against the prevailing opinion
was the 14th century historian, sociologist and polymath Ibn Xaldn,
a writer who in this, as in so much else, was sui generis. His words are
worth quoting, since he puts his finger squarely on the main reasons for
the prejudice against popular poetry (ignorance) which applies with as
much force today as it did six centuries ago:
Most contemporary scholars, philologists in particular, disapprove of
these types (of poems) when they hear them, and refuse to consider them
poetry when they are recited. They believe that their (literary) taste recoils
from them, because they are (linguistically) incorrect and lack vowel endings. This, however, is merely the result of the loss of the habit (of using
vowel endings) in the dialect of the (Arabs). If these (philologists) possessed the same (speech) habit, taste and natural (feeling) would prove
to them that these poems are eloquent, provided that their (own) natural dispositions and point of view are not distorted. Vowel endings have
1
The present paper is based on data gathered in the course of a field-based investigation into the practice of Bedouin poetry in Jordan and Sinai, involving the collection of
poetry composed over the last 50 years, and particularly over the last 20, direct from the
poets themselves. The fieldwork was carried out by Dr Said Salman Abu Athera, a Bedouin originally from Beersheva but now resident in Jordan, and the translation, glossing
and annotating of the poems was done by me. Our book Y Kundlzza Rys! Politics
and Popular Poetry In the Contemporary Arab World is to be published by Brill in 2007.
544
clive holes
nothing to do with eloquence. Eloquence is the conformity of speech
to what one wants to express and the requirements of a given situation,
regardless of whether the u-ending indicates the subject and the a-ending
the object, or vice versa. (Ibn Xaldn, tr. Rosenthal 1958, 3:4145).
With the beginnings of western interest in Arabic literature, from the

mid-19th century until the early decades of the 20th, there was a good
deal of academic effort devoted to collecting examples of Arabic popular
poetry, but with a mainly philological rather than ethnographic or literary inspiration (e.g. Wallin 1851; 1852; Socin 19001; Landberg 1901;
Meissner 1903, though Musil 1928 is an exception). In recent years there
has been a fresh burst of interest, particularly in the contemporary Bedouin poetic tradition (Sowayan 1985; Abu-Lughod 1986; Ingham 1986;
Bailey 1991; and Kurpershoeks monumental five-volume study of the
Bedouin poetry of Najd, 19932005). These later studies have focused
on the practice and social functions of poetry, and on its ethnographic
value as both the product and record of Bedouin societies and modes
of thought that are now rapidly passing into history. There is evidence
to suggest, however, that the tradition of popular poetry in todays Bedouin society is far from dead; on the contrary, it seems to be transforming itself into a voice of socio-political commentary and criticism which
transcends purely local tribal concerns and addresses regional and even
international issues. During the 20th century, Arabic popular poetry
was first deployed by urban poets, in the absence of any other public
forum for protest, as a potent political weapon, the two prime examples
being Mahmd Bayram at-Tniss barbs against Egypts British colonists during the early part of the century (Booth 1990), which led to his
expulsion from Egypt, and his heir, Ahmad Fud Nigms coruscating
attacks on Anwar Sadat and the policy of economic and cultural infith
pursued by the Egyptian government during the 1970s (Abdul-Malek
1990), a campaign which landed him in prison on several occasions.
Popular Bedouin poetry, on the other hand, has until relatively recently
been little concerned with politics and society outside those of the tribal
dra; only with post-Second World War political independence, and
a perceived encroachment of the state on the concerns of their communities have Bedouin poets gone down the same path as their urban
counterparts and turned their poetry into an instrument for expressing
communal popular protest.
But what is meant, in the early 21st century, by the term Bedouin in
the phrase Bedouin popular poetry, since virtually no one now leads the
arabic popular poetry and political satire
545
Bedouin life of nomadism? As a statement that today refers to membership in a genealogy-based social structure, rather than to a way of life,
ih na badu, or more frequently ih na arab, is still a proud boast. Individuals whose families may have been settled for many generations continue
to use this phrase to identify themselves, however far their life-style has
deviated from the Bedouin stereotype of yesteryear. In countries like
Jordan, Iraq and Saudi Arabia, the concentric circles of family ties, clan
membership and tribal affiliation into which self-proclaimed Bedouin
are born continue to exercise a strong pull on their imagination and provide the backdrop against which their social relationships are conducted
and political attitudes forged, even in large cities.2
Contemporary Bedouin poetry is composed in a stylised form of
the poets spoken dialect, but this does not mean that it is a purely oral
form, or that the poet has had any formal training. Poets come in all
shapes and sizes, and not from any particular class or background: he
(or she, as there are some skilled female poets) can be a tribal elder, a
supermarket owner, a smuggler, or a doorman;3 what counts is a gift for
composing, honed through years of listening to other poets. The oldest
generation of poets now in their 70s and 80s are almost without exception illiterate, and use a vocabulary and forms of poetic diction now
difficult for younger generations to understand.4 Nowadays, however,
many of those who would describe themselves as Bedouin poets, like
the one discussed here, are literate in Modern Standard Arabic, but, for
a variety of audience- and topic-related reasons, some of which I will
discuss below, choose to compose in dialect, and use a vocabulary closer
to that of everyday speech. Written versions of their poems not infrequently appear in newspapers and are published in locally and cheaply
produced dwns.
2
In Egypt, in the mouths of the settled farmers and town-dwellers of the Nile valley,
arab is a term of opprobrium, signifying backwardness and stupidity; but for the Sinai
Bedouin, resentful of the power over their lives which the settled population of Egypt
now exercises, these farmers and town-dwellers are contemptuously referred to as ban
firn the sons of the Pharaoh, an insult that alludes to their supposed slave origin and
habit of subservience.
3
Poetry was collected from poets fitting all these descriptions in the course of the
present project.
4
A good example is the poetry of the illiterate Unz Ab Slim at-Turbn of southern Sinai, considered by many the premier Bedouin poet of the region, who spent many
years in Egyptian prisons for smuggling offences, and died in his 80s in 2000.
546
clive holes
2. Ghassn Surr a-baylt ( Ab Surr)
One of the Jordanian poets from whom we have collected a large number
of unpublished poems, though many of them have circulated by word
of mouth and on cassette, is Ghassn Surr a-baylt (Ab Surr).
Ghassn was born in az-Zarq, north Jordan, in 1954, the son of the
paramount sheikh of the al-Uwayst section (ara) of the Ban H asan,
a north Jordanian tribe. He now lives in a large house in al-Mafraq. He
describes his social background with the tongue-in-cheek epithet ladu:
an abbreviation of l filh wa l badu neither cultivator nor nomad, a
term used to describe people historically of Bedouin origin who are now
fully settled. Ghassn was the eldest son in his family, with seven brothers. When his father died, he was around 14 years old. His father left him
virtually nothing (300 JD), having sold off or squandered almost all the
family wealth. As a result, Ghassn was forced to leave school immediately and find work to support his siblings and mother. At that time, in
the late 1960s, the Iraqi army still had units in Jordan following the 1967
war with Israel. This presented opportunities for smuggling. At that
time the Iraqi army was using Russian weapons and ammunition, but
the Bedouin in Iraq still carried British weapons from the days of British influence before the revolution of 1958, just as the Bedouin of Jordan
still did. When Iraqi units rotated home, the young Ghassn would sell
them British weapons and ammunition, which they smuggled across the
border and sold on at a profit to Iraqi Bedouinthere were no border
checks on troop movements. This ammunition and weapons smuggling
subsequently broadened into a more extensive illicit tradecigarettes
and other goodsand extended also to Syria, which was a much nearer
border. By the early 80s, as a young man, Ghassn had begun to travel
regularly to West Germany and the countries of the eastern bloc, from
where he brought back cars and other goods to sell in Jordan. Eventually, the profits from this enabled him to buy land and start businesses,
including supermarkets. He is now, at around 50 years of age, a relatively wealthy man. That is the bare bones of Ghassns journey from
childhood rags to adult riches. He is at pains to point out that none of
what he owns and has achieved was through family influence, but rather
through thirty years of his own efforts and honesty with business partners: what he has achieved has been, as he put it, bi s-sadga wa l bi
l-garba (through friendship and not through family ties).
547
Ghassn is a complex character. Despite his lack of formal education he has read quite widely, and is in the habit of quoting Plato in
his conversation. It became obvious from talking to him that he has a
thorough knowledge of the history of the Levant. He also, despite (or
perhaps because of the means by which he acquired) his wealth, has a
well-developed social conscience. This is strongly reflected in his poetry,
most of which was composed during the 1980s when he was struggling
to make a living and continually coming up against economic and social
barriers. His constant poetic targets are: the economic inequalities of the
Jordan of this period; the incompetence, hypocrisy, cupidity, corruption
and arbitrary powers of government ministers and officials; the nepotism and snobbery endemic in Jordan at all levels; and perhaps most
emphatically, the loss of the sense of honor and dignity which, in his
view, no longer underpins the fabric of Jordanian society in the way it
once did. All of these he had to battle against to achieve what he has.
The satirical poem presented here, as well as many others, could not
be published in Jordan, and the poet ran into serious problems with
the intelligence services at the time when his poetry first began to circulate informally some twenty years ago. How did, and does, Ghassns
poetry circulate? Most often a finished poem or poems would be recited
in a gathering in the poets majlis, and recorded there on a cassette. This
would be copied and distributed to local shops and sold for a very low
price. Bus drivers would often play such cassettes to their passengers. The
fax machine, when it became common in the 1980s, was another means
of circulation of the poem in written form. The latest method is via text
messaging on mobile phones. Although Ghassn still composes poetry,
he has turned away from the provocative and often personal attacks
of the 1980s, and stays behind what he describes as al-xutt al-h amr
(the red lines) that must not be crossedin particular anything that
could be construed as critical of the Hashemite royal family. His mission
remains, however, to speak up for the little people of Jordanian society,
in particular the Bedouin, whose voice and concerns can rarely be heard
above what he describes in another poem (al-Jarid the Newspapers)
as the sycophantic blather of the government-controlled media.
Ghassn commented thus on the poem presented below:
A famous Jordanian song, popular in the 1960s, was y marh ab, y hal,
minn ar-rakb, minn? The song praised the heroes of our nation and
encouraged manliness (rugla), gallantry (ahma), self-respect (karma)
and freedom (h urriyya). But peoples ideas have changed, as have their
dealings with each other. People have become so preoccupied by earning
548
clive holes
a crust and making ends meet that they have no time to think about anything else, and if they do they get distracted and lost.
3. Rashd al-Kln in Abdh Mss musical setting

The words of the original song (Ghawnmah 2002, 1501) are by Rashd
al-Kln. A famous Jordanian gypsy singer and rebec player of the time,
Abdh Ms (19271977), made them famous by putting them to
music. Ms was well known for his popular nationalistic songs praising the army, national heroes, and the Hashemite dynasty, as well as
sentimental ditties for religious occasions. Ghassn takes many of the
original lines of Abdh Mss song and, while preserving the original
meter and rhyme scheme, alters the words to provide a series of poetic
snapshots of the state of Jordan in the early 1980s. The effect of inserting such material into the structure of the song, whilst at the same time
preserving much of the original bombastic phraseology, is pungently
satirical and bathetic. The impact of Ghassns poem on a local audience
depends precisely on the fact that Mss song from an earlier, more
confident period of Jordanian history was so well known to ordinary
Jordanians from its frequent airing in the local media.
Rashd al-Klns original words are as follows: 5
1
y marh ab, y hal, mnn

ar-rakb, min wn? 6
agbal aln d-dih y zna
gblah
h inn dr7 al-id, tallba li
d-dn
Hello, I say, and welcome! Now

where from, these riding men?
The mornings come upon us,
see how beautifully, again!
Against the foe8 were ruthless;
our debts we neer disclaim,
5
Each verse of the poem is divided into two hemistiches, all the opening hemistiches as a group, and all the closing hemistiches as a group, being metrically identical,
and each group having a different rhyme, -n (or -n) for the opening hemistiches and
-lah for the closing ones. This is a common arrangement. Unfortunately it is almost
impossible to imitate in an English translation, so I have rendered the poem into English
rhyming couplets, with fourteen syllables to each hemistich, and tried to be as faithful as
possible to the meaning and tone of the original.
6
A standard Bedouin greeting.
7
dir pl dr ruthless, remorseless.
8
Lit aggression < CLA ad.
549
wa l-gr m yigbalah illa

r-rad xlah
xd il-marik lin min ymn
salfn9
wa l-hm di llan wa r-rh
fadw lah
For only he accepts abuse whose

menfolk have no shame.
The fray we enter swaggering,
our heads we hold up high,
The Hashemites provide our
shade10for them we live or
die!
wa t -t r krin11 lina, y t rn Revenge is second nature; for
each wrong we take back two,
bi t rn
y sbin h aggan l budd m From any who usurps our
rights, we wrest them back
nnlah
anew!
nizh af ala ll ba wa xn al- We march against oppressors
who break their word and creed,
ahd wa d-dn
wa nds ala ll ta wi nigzh We stamp upon the tyrants
head, and punish his misdeeds!
b af lah
So salute our homeland brave
h ayyhum nim l-watan,
hearts! So salute them, Husseins
h ayyhum gund ih sn
pride!
Theyre the ones who wear
rab il-kaff l-h umur wa
red head-cloths, head-ropes
l-ugul mayylah
slanting to the side!
O wife who grieves in widows
y mgana12 bi n-nay13 kaff
weeds, weep not, nor solace
dum il-n
seek,
Weep no more tears of sorrow,
kaff dum il-as a l-xadd
no tears trickling down your
sayylah
cheek.
Our gallant lads have come to
g k n-nim, laf,14 wi
tbair y zn
you, theyre here, be of good
cheer!
9
salf pl salfn brave, swaggering. Cf. Musil 1928, 561 salf a strong gust of wind, Classical Arabic salif vainglorious, boastful.
10
I.e. protect us.
11
kr pl. kiyr custom, habit. Cf. Bailey 1991, 451.
12
mgana woman wearing a guna = womans black head covering reaching down to
her behind.
13
nay sadness, dejection. < Classical Arabic n-w-. Cf. na bi l-h iml to groan under
a burden.
14
lafa (yilfi) come, arrive.
550
clive holes
igbn fg il-higin, w usd
xayylah
add l-bandig zurug
rah saffn15
u mudarratin uhub nrn

gattlah
10
11
madfi mwallafa, tirm ala

l-gln16
saly17 l-qanbil raad u burg
alah
wa nsr gu ww s-sam
mitlaf a bi l-n18
tixw 20 ala l-mitad lil l-mt
ayylah
12
tixw ala l-mitad mixlbh

bi h addn
m dg tam as-salma min
nisat glah
13
14
15
16
17
18
19
20
21
22
aswad sabh al-id ymin

alhum n
u ubr yim da f il-galb an
h lah
min kull dabbbtin ummt
ganzrn
tuhdur hadr il-bah ar, a
l-gm22 sayylah
Like eagles riding camels, like

fierce lions, those cavaliers!
They wield machine guns,
double-clipped, the metal
glinting blue,
And armoured cars, their colour
grey, death-dealing bullets
spew!
Their guns co-ordinated, and
they let fly to each side,
A thunderous volley of shellfire,
it lights up the countryside.
And vultures19 hover in the
sky, wrapped round by cotton
clouds,
They plunge down on the
enemy, and bring him his death
shroud!
They plunge down on the
enemy, their claws as sharp as
nails,
And none who dares to
challenge them has lived to tell
the tale.21
Black morn for the aggressor,
the day him ill befell!
By dust his weak heart blinded,
his fate he could not tell!
There comes from every battle
tank, twin caterpillar-tracked,
A roar deep as the oceans,
gainst the foe they launch
attacks!
saff cartridge magazine or clip of a firearm.

g l pl. gln side, direction.
saly blazing volley (of shots, cannon fire).
Sc. m clouds. The change to n is a poetic licence to maintain the rhyme.
I.e. fighter pilots.
xawa to plunge, dive.
Lit: None tasted safety that went in their direction.
Normal Bedouin term for enemy.

15
16
u fhd bn il-h ufar tuhgum

ala rigln
gannsa li l-id w inmr
gawwlah
gu nd ih sn
rabi il-kaff l-h umur wi
l-ugul mayylah
551
And darting tween the battle

lines23 two-legged cheetahs
growl,
And hunt down the aggressor,
just like panthers on the prowl.
pride!
Theyre the ones who wear
red head-cloths, head-ropes
slanting to the side!
4. Text comparison
In his version, Ghassn takes the sentimental nationalism of Mss
refrain:
h ayyhum gu nd ih sn
rab il-kaff l-h umur wa l-ugul
mayylah

pride!
Theyre the ones who wear red
head-cloths, head-ropes slanting
to the side!
(line 6 of the original, repeated in line 16)

and transforms it into a variable refrain of his own in which he in turn
ridicules Jordans political leaders:
h ayyhum iyx il-arab il
takstuh24 bi antnn
rab it-taks l-h umur wa
d-daym bi iglah
So salute our Arab leaders, on their

cars two aerials fixed!
Theyre the ones who drive red
limos, head-ropes full of despots
tricks!
(line 5 in Ghassns version).

23
Lit: ditches, i.e. defensive trenches.

taks (in construct phrase and when suffixed takst-) pl. taks car, limousine (as
well as taxi).
24
552
clive holes
its well-paid but incompetent officials:

u h ayy wizrat il-il il alh agatn
bi d-dn!
rab it-tawbig h agar u amwl
sayylah
And salute the High Price Ministry

thats saddled us with debts!
Theyre the ones in stone-built
condos, with their liquid cash
assets!
(line 13 in Ghassns version)

returning more wistfully at the end to Mss subject, the army:
u ym id-dgt wn rh in-nim Where did they go, now times are
wn?
hard, those brave heart lads with
pride?
rabi il-kaff l-h umur wi l-ugul
The ones who wear red head-cloths,
mayylah . . .?
head-ropes slanting to the side . . .?
(line 19 in Ghassns version)
The confident vainglory of Abdh Mss opening lines, in which he
brags about the pride, dignity and valour of King Husseins Bedouin
troops, becomes in Ghassns poem an abject confessional put in the
mouths of ordinary Jordanians who have learnt to accept incompetence
and oppression and to keep their mouths shut and look out for themselves. Compare the original:
il-rakb, min wn?
agbal aln d-dih y znat
gblah!
h inn dr al-id, tallba li d-dn
wa l-gr m yigbalah illa r-rad
xlah
xd il-marik lin min ymn
salfn
wa l-hm dillan wa r-rh
fadw lah
Hello, I say, and welcome! Now

where from, these riding men?
The mornings come upon us, see
how beautifully, again!
Against the foe were ruthless; our
debts we neer disclaim,
For only he accepts abuse whose
menfolk have no shame.
The fray we enter swaggering, our
heads we hold up high,
The Hashemites provide our shade
for them we live or die!

wa t- t r krin lina, y t rn bi
t rn
y sbin h aggan l budd m
nnlah
553
Revenge is second nature; for each

wrong we take back two,
From any who usurps our rights,
we wrest them back anew!
(lines 14)
with Ghassns:
il-bal min wn?
agbal aln l-il y maba
igblah! 25
h inn gh id-duwal w il
nartad bi l-bn26
wa l-kull rd l-gr h att
-aham xlah27
wa l-amr tanfdah lin u min
drn xann
wa l-gir sr ar-rabb wa r-rh
fadw lah
wa s-samt krin lin wa
a l-h agg m nin
wa l-h agg min il-mursil28
wallah m nnlah
Hello, I say, and welcome! But now

wheres this scourge come from?
Price rises came upon us, and
theyve hit us like a bomb!
We act like world-class donkeys,
with our cruel fate content,
Each one of us accepts abuse, even
those of high descent.
We do as we are ordered to; its
not our place to ask,
Money rules our lives; our souls are
mortgaged to that task.
Keeping quiets second nature; we
dont stick up for peoples rights,
Our rights we beg from minions,
then give back without a fight.
(lines 14)
The symbolic widow of lines 78 of the original, weeping for a husband
killed in battle, and comforted by the arrival of the army of the nim
(gallant lads) of the army:
25
< y m ab igblah lit: how awful its coming!

= w illi nartadi bi l-bn those who are content with adversity.
27
Ironically echoing the sentiment of the popular Bedouin saying dawwir li wildik xl find a maternal uncle for your son, i.e. find a brave man and marry his sister
because the bravery will run in the family. Nowadays, the poet is saying, the once noble,
independent Bedouin are ordered about by moral pygmies, and money has become the
yardstick of a persons worth.
28
mursil servant, messenger-boy.
26
554
clive holes
g k n-nim, laf,
wi tbair y zn
igbn fg il-higin, w usd
xayylah
Our gallant lads have come to you,

theyre here, be of good cheer!
Like eagles riding camels, like fierce
lions, those cavaliers!
is told in Ghassns version (line 7) to cry even more, as there is no longer any comforting presence to reassure her:
rh n-nim u madaw,
l tafrah y zn
b syf il-h arb u m dall
xayylah
Our gallant lads have gone for

good, theres nothing left to cheer,
They sold their swords off long ago;
their cavalry disappeared.
Instead of the exultant heroism of:

madfi mwallafa, tirm
ala l-gln
saly l-qanbil raad u burg
alah
Their guns are at the ready, and

they let fly to each side,
A thunderous volley of shellfire,
it lights up the countryside.
(line 10 of the original)

the widow, and all other ordinary Jordanians can expect no protection at
all; on the contrary, they have become the targets of (economic) attack:
awmrah mh add a rah
yirm ala l-gln
saly l-awmir raad u
alt gattlah
Hes got his orders ready, and he lets

fly to each side,
A thunderous volley of memos,
death-dealing stuff inside!
(line 9 of Ghassns version)

the orders in this case being those of the Prime-Minister Mudar Badrn
to raise the price of the basic necessities of life beyond the point where
they can afford them.
The original song concludes (lines 815) with a stirring picture of the
derring-do of the Jordanian army and air force; the corresponding lines
in Ghassns poem depict a sleazy world of fawning senators selling off
the national patrimony, accepting bribes, lining their own pockets, and
hanging around night-clubs and brothels (Ghassns lines 1117):

il-wisk lah wi l-z lah
wa li h sbah ha-l-girn
wa -aab b h amdullah
m yiba nxlah29
fawwad ard l-arab
li s-sr luh ahrn
li l m yinirif ammah
u l yinirif xlah
u h ayy wizrat il-il il
alh agatn bi d-dn!
rab it-tawbig h agar
u amwl sayylah
din yxk30 u sin31
wn il-harba wn?
u h litn b hal-balad
wallah mi h lah
wisiln dr il-h add min
xn wa l-n32
ill ala abwb id-dant33
inmr gawwlah
l yiurrak kubr i-anab
wa l-fard maa mitn
bi d-dabdaba34 yibd u bi l-madh
yinh maglah
iyh tt lah gu ww z-zarf
h ukkmn girn
u min agal hal-girn yanh ar
lak iylah
555
Hes got his whisky, young roast

lamb, hes got our money too,
The people feed on wheat-chaff
and just thank the Lord they do.
He gave away our tribal lands to
folk just two months here,
To people with no lineage, no
pedigree thats clear.
And salute the High Price Ministry
thats saddled us with debts!
Theyre the ones in stone-built
condos, with their liquid cash assets!
Were lost, my brother, wandering . . . oh where is our escape?
By God, the nations state is dire:
no order here, no shape.
To blame? Our sheikhs an
senators; the mess were in is foul:
They spend their time in sleazy
dives, like panthers on the prowl.
Dont be fooled by big moustaches,
guns with double cartridge clips,
What they write begins with
weasel words and ends with
flattering quips.
Our rulers slip them money, in
brown envelopes its brought,
Theyd kill their kids to get one, slit
their throats without a thought.
Ghassn then rubs it in by concluding his message with a variation of

the opening lines:
29
nxla poor quality wheat left after sieving.

A so-called bi-polar address form, also common in Eastern Arabian dialects. See
Yassin 1977.
31
s to wander, go aimlessly from one place to another. sin u din were wandering
lost. Cf Classical Arabic sa to scatter (people, things).
32
I.e. a member of the maglis il-ayn the Senate, all appointed by the King.
33
I.e. brothels and drinking dens.
34
dabda b to talk meaninglessly or ambiguously, utter weasel words.
30
556
clive holes
y marh ab y hal! imnn

il-il min wn?
agbal aln l-bal, y maba
igblah!
SoHello, I say, and welcome! But

now wheres this price rise from?
A scourge is come upon us, and its
hit us like a bomb!
(line 18)
and a prediction of even worse to come from an incompetent government:
win xilsat il-arkt, yallah
is-salmah
itxalln il-wh id min id-dill
h ltah h lah
But when the battles over, and the

situations calm,
You lot will mess things up again;
youll do us yet more harm!
(line 20)
Ghassns poem is four lines longer than the original, and the formal
correspondences of the original lines to those of Ghassns version (i.e.
where Ghassns lines mimic the diction of the originals) are as follows:
Rashd al-Klns lines
14
5
68
9
10
1114
15
1
6, 16
correspond to Ghassns lines

14
57
8
9
1014
15
1617
18 (reprise of 1)
19
20
This is Ghassns complete (1982) text:

1
y marh ab, y hal, mnn ilbal min wn?
Hello, I say, and welcome! But

now wheres this scourge come
from?
557
agbal aln l-il y maba

igblah!
h inn gh id-duwal w il
nartad bi l-bn
wa l-kull rd l-gr h att
-aham xlah
wa l-amr tanfdah lin u min
drn xann
wa l-gir sr ar-rabb wa r-rh
fadw lah
wa s-samt krin lin wa a
l-h agg m nin
wa l-h agg min il-mursil wallah
m nnlah
h ayyhum iyx il-arab il
takstuh bi antnn
rab it-taks l-h umur wa
d-daym bi iglah
Price rises came upon us, and

theyve hit us like a bomb!
We act like world-class donkeys,
with our cruel fate content,
Each one of us accepts abuse,
even those of high descent.
We do as we are ordered to; its
not our place to ask,
Money rules our lives; our souls
are mortgaged to that task.
Keeping quiets second nature; we
dont stick up for peoples rights,
Our rights we beg from minions,
then give back without a fight.
So salute our Arab leaders, on
their cars two aerials fixed!
Theyre the ones who drive red
limos, head-ropes full of despots
tricks!
y mgana bi n-nay zd dum O wife who grieves in widows
il-n
weeds, weep on, the futures
bleak,
zd dum il-as a l-xadd
Weep yet more tears of sorrow,
sayylah
let them trickle down your cheek.
rh n-nim u madaw, l
Our gallant lads have gone for
tafrah y zn
good, theres nothing left to
cheer,
b syf il-h arb u m dall
They sold their swords off long
xayylah
ago; their horsemen disappeared.
kull nadil h kim bihum h ukm Now any numb-skull scoundrel
il-and i-n
treats them like his abject tools,
Makes porters out of highborn
u xall wgh il-balad bi s-sg
attlah
men who once this country
ruled.35
35
The reference is to Mudar Badrn, the Jordanian Prime-Minister of the time. There
is also a whiff here of a sense of lse-majest: Jordanians of pure Bedouin descent regard
the Badrn family, which has its roots in urban Syria, with disdain.
558
clive holes
9 awmrah mh add a rah yirm

ala l-gln
saly l-awmir raad u alt
gattlah
10 saww tnakit il-kz36 bi h dd
dnrn
wa a h sb g in-ns titkt ar
amwlah
11 il-wisk lah wi l-z lah wa li
h sbah hal-girn
wa -aab b h amdullah m
yiba nxlah
12 fawwad ard l-arab li s-sr luh
ahrn
li l m yinirif ammah u l
yinirif xlah
13 u h ayy wizrat il-il il
alh agatn bi d-dn!
Hes got his orders ready, and he

lets fly to each side,
A thunderous volley of memos,
death-dealing stuff inside!
A jerry-can of kerosene put up to
two JD,37
To make his stacks of cash pile
up, the people go hungry.
Hes got his whisky, young roast
lamb, hes got our money too,
The people feed on wheat-chaff
and just thank the Lord they do.
He gave away our tribal lands to
folk just two months here,
To people with no lineage, no
pedigree thats clear.
And salute the High Price
Ministry thats saddled us with
debts!
rab it-tawbig h agar u amwl Theyre the ones in stone-built
sayylah
condos, with their liquid cash
assets!
14 di n yxk u sin wn ilWere lost, my brother,
harba wn?
wandering . . . oh where is our
escape?
u h litn b hal-balad wallah
By God, the nations state is dire:
mi h lah
no order here, no shape.
15 wisiln dr il-h add min xn To blame? Our sheikhs and
wa l-n
senators; the mess were in is foul:
ill ala abwb id-dant inmr They spend their time in sleazy
gawwlah
dives, like panthers on the prowl.
16 l yiurrak kubr i-anab wa
Dont be fooled by big
l-fard maa mitn
moustaches, guns with double
cartridge clips,
bi d-dabdaba yibd u bi l-madh What they write begins with
yinh maglah
weasel words and ends with
flattering quips.
36
37
A twenty-litre jerry can of kerosene, used for cooking on primus stoves.

I.e. (in 1982) very expensive.
559
17 iyh tt lah gu ww z-zarf

h ukkmn girn
u min agal hal-girn yanh ar
lak iylah
Our rulers slip them money, in

brown envelopes its brought,
Theyd kill their kids to get
one, slit their throats without a
thought.
18 y marh ab y hal! imnn
SoHello, I say, and welcome! But
il-il min wn?
now wheres this price rise from?
agbal aln l-bal, y maba A scourge is come upon us, and
igblah!
its hit us like a bomb!
19 u ym id-dgt wn rh
Where did they go, now times
in-nim wn?
are hard, those brave heart lads
with pride?
rabi il-kaff l-h umur wi l-ugul The ones who wear red headmayylah,
cloths, head-ropes slanting to the
side . . . ?
20 win xilsat il-arkt, yallah isBut when the battles over, and
salmah
the situations calm,
itxalln il-wh id min id-dill
You lot will mess things up again;
h ltah h lah
youll do us yet more harm!
Thus where the original song was a jingoistic, bombastic romp, in
Ghassns hands it is transformed into a mordant diatribe against the
Badrns governments economic mismanagement, its cavalier treatment
of ordinary Jordanians, and the corruption and cupidity of its officials.
It is precisely the parodying of the diction of the original that gives the
poem its punch as a piece of political satire.
The relationship of language level, poetic form and subject matter is
worth pondering. Historically, Bedouin popular poetry was concerned
solely with the local tribal milieu: with vaunting the merits of ones tribe
over those of a neighbour, with praising heroic warriors and generous
shaikhs, with ridiculing enemies, against a backdrop of descriptions
of desert flora and fauna and the movement of the tribes in constant
search for water and good pasture. In this, as has often been observed,
Bedouin popular poetry is the direct descendant of the old pre-Islamic
poetry, and is similarly restrictive in matters of meter and rhyme. For
all these topoi, the only possible linguistic vehicle was the vernacular,
both because it was the only form of Arabic the poets and their audience knew, but also because of the organic nexus between locale, subject matter and language. But now that the traditional subject matter for
the exercise of poetic talent has all but disappeared, this ancient poetic
560
clive holes
tradition has shown its protean nature: it continues to be used, even by

Bedouin poets who are literate, to serve many of the same functions,
though now on a much broader social and political canvas. If one wants
to emoteto praise, to damn, to ridicule, to gloat, to cajole, to bewail, to
lovehow much more effective is the vernacular because of its infinitely
greater allusiveness, colour, concreteness and idiomaticity for the ordinary person, compared with the sanitised and institutionalised blandness of the modern standard language.
As I write this article, I have to hand the latest in a long-running
series of poems written since 2003 by a south Jordanian Bedouin poet,
Muh ammad al-Fanti l al-H ajy on the American-British invasion of
Iraq, some of which have been published in local newspapers and magazines. The poems recycle standard Bedouin poetic tropes, such as the
journey (similar to the Classical rah l), whereby a rider, after first having
had the noble qualities of his camel38 described by the poet, is sent with
a message (the poem) to kinsmen encamped far away; or the ascent
whereby the poet climbs to the top of a nearby hill (rigm), there to wrestle with the cauldron of his emotions (higs, hgs, hawgs) and finally
be seized by poetic inspiration to write his poem39 . . . except that in these
poems, it is leading international politicians who are thus depicted! For
example, the poem Fadas wa Fads40 opens with George Bush making
an ascent in his home state of Texas to ponder on his woes during the
2004 Iraq insurgency:41
araft rigmin f marb takss
rigmin tawl min al-baar
r mns
Ah climbed atop a Texas peak, out

west in our proud nation,
A peak so hah no folks live there, to
fahnd some isolation.
38
Nowadays, it is quite common for the poet, without a trace of irony, to describe
the qualities of a modern steeda Toyota Landcruiser, for instancein terms similar
to the ways a fine riding camel would once have been described: its strong chassis, its
smooth gear-change, its fine performance, etc. See, e.g., Kurpershoek 1999:110114 for
examples from Najd. We recorded similar descriptions of cars in Bedouin poetry from
Sinai.
39
Kurpershoek 1999:3134.
40
These pet names are often given by the Jordanian Bedouin to dogs (rather like
Rex and Rover in English). In the poem they are insultingly applied to the pet rulers
of unnamed Gulf States who, in the poets view, treacherously provided bases for the
Americans at the time of their invasion of Iraq.
41
And speaking in a Texas accent, which I have tried to imitate in my translation!

u darabt min hamm xms
bi l-asds
u h asst f rs kam targ ngs
u gadt wagt bn higsin u hgs
wi l-galb tgad bih kam nr
fns
561
Ah brooded deep upon mah woes,

Ah felt at mah wits end, when
Ah heard a bell toll in mah head
what did that bell portend, then?
My worries crowded in on me, my
state it was concernin,
Mah heart aflame as if it were a
lanterns oil-wick burnin!
Another character who has had poems put in his mouth by this poet
is Saddam Hussein, in prison in Iraq, but stillpoeticallyshaking a
defiant fist a the Americans. And to George Galloway, founder and sole
representative of the UKs Respect political party, he has dedicated a
typical piece of Bedouin madh , praising Galloways opposition to Tony
Blair and what the poet sees as his dogged, unpopular but heroic support for the Iraqi people.
Popular poetic commentary on contemporary events such as has
been exemplified in this paper is nothing new: its just that the focus
has broadened. Before the birth of the nation state in the Arab World,
and even in the period since, Bedouin tribal poetry was one of the main
means by which partisan communal sentiments could be articulated. A
good example of this was a dispute between the H uwt t, a Jordanian
tribe, and the Ban Atiyya of Saudi Arabia, which rumbled on into the
late 1980s in the form of an extended poetic debate of claim and counterclaim. The dispute centred on Tubayq, an area in the far southeast of
Jordan which the Jordanian government ceded to the Saudis in 1964
in return for a stretch of Red Sea coastline which would enable Jordan
to improve its naval facilities away from the prying eyes of the Israelis.
Initially, this exchange of land and redrawing of the border caused no
problems, and the H uwt t on the Jordanian side continued to drive
their animals to seasonal pastures on the Saudi side, exactly as they had
always done. They incidentally benefited from the fact that many goods
were cheaper in Saudi Arabia, and border controls were lax or non-existent. A number of events, however, changed all this. During the Jordanian civil war of 1970, Russian machine guns became easily available, and
enterprising H uwt t began a lucrative smuggling operation into Saudi
Arabia. Drugs were also smuggled in. This led to a clampdown by the
Saudis, and the blacklisting of many H uwt t. Then, after the shocking siege of the grand mosque in Mecca by Islamic fundamentalists in
1979, the Saudis put even stricter border controls in place. One measure
562
clive holes
was the digging of a ditch, 3m wide by 3m deep along the border with
Jordan in Tubayq. No camel or car could cross it except at designated
control points situated far apart. H uwt t who had been used to driving
their animals 10 miles to pasture were now faced with driving them 100
miles to find the nearest border post and 100 miles back again. Customs
controls became much stricter. A system of registration documents
was introduced for the family members who moved with the migrating
flocks. There were cases of H uwt t mothers who had given birth while
in Saudi Arabia being detained there because, when they tried to return
to Jordan, the number of family members did not correspond with
the number on the registration document. There were intimate body
searches of all females at the border checkpoints. This was bad enough
for the socially conservative Bedouin, but it also closed off the last
avenue for smuggling. On the other hand, the Jordanian borders
remained open to Saudis without let or hindrance, a fact that caused
huge resentment. This unequal treatment, coupled with what was perceived by the Jordanian Bedouin as the central governments lack of
economic help to the people of the south, was one of the factors that
precipitated rioting in Maan and Al-Jafr in 1989.
A well-known H uwt poet, Barrk Dish Ab Tyih, wrote several
emotional poems about the dispute that the Ban Ati yya, on the Saudi
side of the border, did not like. Ban Ati yya poets replied with poems
which were recorded and passed back to the H uwt t. The initial skirmish might have ended there. However, it was followed by a long and
insulting poem from a young H uwt poet, Nad Tmn Ab Tyih. The
poetic tit-for-tat then escalated, with ten poems by Ban Atiyya poets in
reply, releasing much pent-up personal vituperation. This caused an outraged reaction among the H uwt t, who had always regarded the Ban
Atiyya as inferior to themselves. Seventy years ago, the result would
undoubtedly have been a tribal war; on this occasion, the dispute was
finally defused in 1990 by mediation, and the signing of an agreement
that neither side would write any more poems on the subject of Tubayq.
The key point here is that poetry was, and is, regarded by the Bedouin
as a suitable vehicle for airing important issues of the moment, rather
than a letter to the provincial governor, still less lobbying a remote and
seemingly uncaring central government.
563
5. References
Abdel-Malek, Kamal. 1990. A Study of the Vernacular Poetry of Ah mad Fud Nigm.
Leiden: E.J. Brill.
Abu Athera, Said Salman and Holes, Clive. Forthcoming 2007. Y Kundlzza Rys!
Politics and Popular Poetry in the Contemporary Arab World. Leiden: E.J. Brill.
Abu-Lughod, Lila. 1986. Veiled Sentiments: Honor and Poetry in a Bedouin Society.
Berkeley: University of California Press.
Bailey, Clinton. 1991. Bedouin Poetry from Sinai and the Negev. Oxford: OUP.
Booth, Marilyn. 1990. Bayram al-Tunisis Egypt: Social Criticism and Narrative Strategies. Exeter: Ithaca.
Ghawnmah, M. Abdh Ms: Ridan wa-Mubdian, Amman, Dr al-Kind, 2002.
Ibn Khaldn, tr. Rosenthal, Franz. 1958. The Muqaddimah. An Introduction to History.
London: Routledge and Kegan Paul.
Ingham, Bruce. 1986. Bedouin of Northern Arabia: Traditions of the l-D hafr. London:
Kegan Paul International.
Kurpershoek, P. Marcel. 1994. Oral Poetry and Narratives from Central Arabia I: The
Poetry of al-Dindn. Leiden: E.J. Brill.
. 1995. Oral Poetry and Narratives from Central Arabia II: The Story of a Desert
Knight. The Legend of lwh and other Utayba heroes. Leiden: E.J. Brill.
. 1999. Oral Poetry and Narratives from Central Arabia III: Bedouin Poets of the
Dawsir Tribe. Between Nomadism and Settlement in Southern Najd. Leiden: E.J.
Brill.
. 2002. Oral Poetry and Narratives from Central Arabia IV : A Saudi Tribal History.
Leiden: E.J. Brill.
. 2005. Oral Poetry and Narratives from Central Arabia V : Voices from the Desert.
Glossary, Indices, and List of Recordings. Leiden: E.J. Brill.
Landberg, Carlo le Comte de. 1901. tudes sur les dialectes de larabie mridionale.
Leiden: E.J. Brill.
Meissner, Bruno. 1903. Neuarabische Gedichte aus dem Iraq. Mitteilungen des Seminars fr orientalische Sprachen zu Berlin VI. 57125.
Musil, Alois. 1928. The Manners and Customs of the Rwala Bedouin. New York: American Geographical Society.
Socin, Albert. 19001. Diwan aus Centralarabien. Leipzig: Teubner.
Sowayan, Saad Abdallah. 1985. Nabati Poetry: The Oral Poetry of Arabia. Berkeley: University of California.
Wallin, Georg August. 1851. Probe aus einer Anthologie neuarabischer Gesnge in der
Wste gesammelt. Zeitschrift der Deutschen Morgenlndischen Gesellschaft 5. 123.
. 1852. Probe aus einer Anthologie neuarabischer Gesnge in der Wste gesammelt. Zeitschrift der Deutschen Morgenlndischen Gesellschaft 6. 190218, 369378.
Yassin, M.A. 1977. Bi-polar terms in Kuwaiti Arabic. Bulletin of the School of Oriental
and African Studies 40. 297330.
NOTES ON THE DIALECTS OF THE LGT AND H AMD AH

OF SOUTHERN SINAI
Rudolf de Jong
ACLC, University of Amsterdam
1. Introduction
The subject of this contribution1 is the dialect of the Lgt (or Ulayqt),
a bedouin tribe who live in the western central part of southern Sinai
(see map below).2 In addition, notes on the dialect of the H amda h have
been included. Some texts recorded among the Lgtwith additional
information in footnotesmay serve for further illustration.
The H amda h are only few, and are often regarded as a clan (or family) of the Lgt, although various sources claim that they were present
in Sinai before the Lgt.3 We shall see that there are some notable differences between these two varieties of speech.
1
With great pleasure I dedicate this contribution to Kees Versteegh. My dedication
is with deep respect for his stature in our field of Arabic Studies, with gratitude for the
inspiring thoughts he has shared with our community and with fond memories of the
(too few) occasions I had the honor to work with him.
2
The material used for this article was collected in the framework of my own research
into the bedouin dialects of southern Sinai. This project is funded and supported by The
Netherlands Organization for Scientific Research (abbreviated in Dutch as NWO) and
the Amsterdam Center for Linguistics and Communication (ACLC) at the University
of Amsterdam. I am sincerely grateful to both organizations, as I am thankful also to
Manfred Woidich for his advice and support during my researches. I am no less grateful
to my desert guideand now dear friendId Abuw Silm (al-Atra at-Turbniy), who
has been my ally in this project and without whose contributions this research would
not have been possible.
3
Murray (1935, 291), for instance, writes that the H amdah [. . .] are now only about
twelve tents strong, and live mostly near Serabit el-Khadim and Bir Nasb, where they are
reckoned as part of the Aleiqat. For the location of Sarbt al-Xdim, see Google Earth
at (appr.) 29 00 05 N33 28 01E.
At -Tayyib (1997, 480481) mentions that the H amdah joined the Lgt under the
Lgiy x at that time (A.D. 1914) Mdaxxal Slmn of the Lgiy clan named Awld
Silmiy. Von Oppenheim (1943, 162, 164) writes that the H amdah
were angegliedert to
the Olkt (in his transcription).
On the webpage http://members.nova.org/~lroeder/alegat.htm (authored by Larry
Roeder), under History of Hamada, it is reported that the H amdah
[. . .] probably
[joined the Alegat] shortly after the capture of Nuweiba. We do know that from then
until the 1880s this was a sub-tribe (or section) of the Alegat. Then in 1880 internal
566
rudolf de jong
Reported to have arrived in Sinai at some time in the fourteenth

century,4 the Lgt today number around 1,000,5 and are found predominantly in an area named ar-Ramlah, near where Jabal Sarbt al-Xdim6
stands, situated just south-west of the Th escarpment, and north of
Wdiy Frn and in and to the east of the town of Abuw Znmah.7 Also
farther north, to the south of Rs Sadr and in Wdiy arandal (not far
north of H ammm Farawn) families of the Lgt are said to reside, as
well as farther south, near the town of at-Tr.8
Approximate distribution of bedouin tribes in southern Sinai 9
politics forced a split when the Hamada requested permission to live under the protection of the Muzeina. By 1935, they had become integrated enough in Muzeina affairs to
be considered an integral part of that tribe.
4
See Bailey 1985, 48. For more information on subdivisions of this tribe, their xs,
history, territories, etc., see at -Tayyib 1997, Part 2, 475489.
5
Such numbers are of course approximations.
6
Sarbt al-Xdim, some 40 kilometres east of Abuw Znmah (on maps usually
spelled as Abu Zinima), is famous as the site of turquoise mines operated since early
pharaonic times, and the temple of Hathor, which is the only pharaonic temple built
outside Egypt proper. See also the webpage about the Sinai at http://www.arcl.ed.ac.uk/
arcx/remot esense/sinai/.
7
Also in Wdiy Isla. The Lgt are reported to form an alliance in Sinai with the
H amdah
and Mznah, see At -Tayyib (1993, 705) and Bailey (1991, 5). At- Tayyib (1993,
706) reports that their territory stretches from ar-Ramlah to Wdiy arandal. One of my
own Lgiy informants mentioned these areas too, but said their drah stretches up until
the area named ar-Rynih, north of Rs Sadr. For a map locating their (there transcribed
as Algt) drah, see Bailey (1991, 4).
8
Wdiy s Sahaw, mntigit Mbajjmah, Br anNasb and Wdiy Lihyn were mentioned
to me by a Lgiy informant as parts of the H amdiy
drah.
9
Abbreviations: Ah = Ah aywt, Tr = Tarbn, H w = H wtt, Db = Dbr (see remark
below), Ty = Tayhah, Lg = Lgt, Bd = Badrah, Jr = Jarjrah (see remark below),
dialects of the lgt and h amd ah

1.1
567
General
In the course of the ongoing research into the dialects of southern Sinai,
a picture has been emerging in which dialects of southern Sinai appear
to constituteto an as yet undefined degreea homogeneous group.
This group is typologically related to the dialects of group II in the north
of Sinai: that of the Agylah, andeven more clearly sothat of the
Samnah.10
As a group, the dialects in the south stand separate from the dialects of group I, or the Negev-type of dialect (to which the dialects of
the Ah aywt and Tarbn belong, and also those of the Tayhah and
Jarjrah).11 The southern group also shows important differences with
the dialect of the Mznah, who live in the eastern part of southern
Sinai and its southern tip.12 We shall see that of the two dialects treated
here, H amdi y appears to have the most in common with this southern
Sinaitic group, while Lgiy occupies a position typologically somewhat
nearer to group I.13
2. Phonology
2.1
Consonants
As bedouin dialects spoken in southern Sinai, the phoneme inventories

of Lgiy and H amdi y do not hold many surprises. As far as consonantal phonemes are concerned, all three interdentals /t/ , /d/ and /d/ (the
latter as a reflex for both *d and *d) are present. A voiced and unaffricated /g/ reflects Classical Arabic *q and a voiced affricate /j/ reflects *j
H m = H amda h, Sw = Sawlh ah, Gr = Garrah, AS = Awld Sad, Mz = Mznah, Jb =

Jbliyyah, BW = Baniy Wsil.
I have met Jarjrah in the area (named Malbad, see Google Earth 29 29 29.50 N.,
33 05 36.50 E.) as marked on this map. Murray (1935, 263) also refers to the Jarajira as a family of the Aleiqat. A tribe also indicated on this map are the Dbr, whose
presence was mentioned to me by sources in the field. They are said to be related to the
Masd, who live in and around the village of Jilbnah in the northwest of Sinai.
10
As described in De Jong (2000, chapter 2).
11
Being based on research in progress, this conclusion is tentative.
12
For a preliminary report on research into the dialect of the Mznah, see De Jong,
The Dialect of the Mznih of Southern Sinai (to appear in Estudios de Dialectologa
Norteafricana y Andalus (EDNA) 8 in honor of Peter Behnstedt).
13
As described in Blanc (1970) and De Jong (2000). Murray (1935, 263 ff.), writes
that the Lgt [. . .] are said to talk like the Terabin.
568
rudolf de jong
(while the fricative allophone [] appears to be as current as in e.g. the

dialect of the Mznah in the southeast of Sinai).
There are also exceptions to the interdental reflexes of Classical Arabic interdentals: e.g. (z for *d) bizr seed and aza ~ iza if (compare
Classical Arabic *id). In the loan masalan for instance, s reflects *t.
Although z as a phoneme is rare (and marginal as such) in Sinai bedouin dialectsthe current reflex for both *d and *d being the interdental emphatic d
it is regular in lexemes (loaned from Modern Standard
Arabic or Cairene Arabic) like zabbat, yzabbit arrange; do properly,
bz, ybz go bad (of food) (from the Turkish root boz-mak).
Also, like in many southern Sinaitic dialects (though not all), a minimal pair such as btk your (m. sg) housebtk your (f. sg.) house yields
the two (unaffricated) phonemes (heavily velarized) /k/ and (otherwise neutral) /k/.14
Another feature of Lgiy and H amdi y is its regular progressive
assimilation of initial h of suffixes to preceding voiceless consonants,
e.g. fattte # (< fatt + ha)15 he shredded it (f. sg.) (said of bread, when
one prepares fattah), maslah tta (< maslah t + ha) her affairs, arssa
(< ars + ha) her groom, dabah h a he slaughtered her, msikka having taken hold of her and a very current (in both varieties, as well as
in many other dialects of Sinai) reciprocal assimilation of + h in e.g.
mih h a (< mi + ha) with her.
2.2
Vowels
As for vowels, phonemic opposition of short vowels /i/ and /u/ is as limited as it is in other bedouin dialects of Sinai, but the minimal pair Xidr
male given namexudr green (c. pl.) will isolate these vowels as separate phonemes in Lgiy and H amdi y. Finding a minimal pair to isolate
/a/ as a phoneme is not a problem in Sinai; for Lgiy a pair like xa he
enteredxu enter! works like in any other Sinai bedouin dialect.
Long vowel phonemes are //, // and //, and additional (unconditionally monophthongized diphthongs *ay and *aw) // and //. Result-
14
The opposition is about as widely used in Lgiy as it is in other dialects of southern
Sinai. In H amdi y, however, the suffix -kiy is also used instead of -k for the 2nd p. f. sg.,
while V(C)uk (CC-uk) is regular for the 2nd. p. m. sg. in both H amdiy
and Lgiy.
15
The spelling here with triple t is for the sake of morphological transparency. The
pronunciation is, however, not noticeably different from doubled t (IPA [(t)]). See also
fn 90.
569
ing monophthongs and tend to be realized lower when they are

preceded by some of the back spirants (notably x, and h and ), or
by certain (primary or secondary) emphatics: da f guest, and also xat
thread. ar must, and I have also heard Farn (Wdiy) Frn and less
(but still) lowering in nuss n two halves and masta rnih ruler; other
than, h l (in which // is near, or just above IPA [ :]) year, but also quite
low // in rdin get (f. pl.) water. When h preceedes, however, there is
no noticeable lowering, as in hl cardamon and also IPA [o] in Abuw
lHl Blue Hole.16
In some cases a diphthong will remain unchanged, as in e.g. (loan
from English) dayf dive, lay (~ l) to me, aw or, law if (loan from
Modern Standard Arabic), or when Systemzwang is operative, as in
mawjd present, in the imperative form awuw watch out (m. pl.) and
also in the f. sg. form h awly (m. sg. ah awal) cross-eyed, where retention of the diphthong prevents a (near) homophonic clash with h l my
year.
2.2.1 Raising the short vowel a
Short a is raised in a variety of positions preceding primary stress. In
the cases cited below, such raising is concluded to be optional, since it
does not occur always. Raising of a tends to be towards [u] when it precedes C, or when followed by w, and towards [i] in otherwise neutral
environments.
Lgiy ilh (~ alh) on him (notice the short base,17 but in H amdi y
more regularly ilh), dibht I slaughtered and in sandhi also i- mb
on the side.
Short a preceding stressed : gibyil tribes, iml north and ibn
hungry, kisln lazy. Short a preceding stressed : kitr much, bid far,
digg flour, ijn dough. Although such raising is predominant, forms
with unraised a also appear, e.g. targah method, kabr large, and even
more regularly so when h , precedes a, as in h add iron, ars groom,
xarf autumn.
16
Abuw lHl, lit. The Sphinx, but here as a folk etymology for The Blue Hole, which
is a popular dive site about 12 km north of D ahab.
17
This must have developed in analogy to such variation as luh (~ much less frequent)
lh to him. Comparable forms (alh and alh) were also recorded in the dialect of the
Tarbn of the north and that of the Rmlt (respectively), see De Jong (2000, 181).
570
rudolf de jong
Short a preceding stressed : unb (~ anb) south, isSudiyyih

Saudi Arabia, gud camel that has not yet cut its canine teeth,18 lugh
impregnated (of a she-camel) and uby (~ aby) my father. Absence
of such raising was also recorded, as in ar emperor fish.19 Such raising
appears to be less regular when h precedes, as in h amlih animal led to a
party to be slaughtered for those present,20 ars bride, azz pole. The
gahawah-vowel (see below) a preceding in an open syllable remains
unaffected, e.g. maxarm pierced, maarf known, mah att placed.
Preceding stressed : iluk on you, fidt I sacrificed and middt I
stretched, xit I entered, istinnt I waited. Raising towards [u] preceding w: in muwwl poems, ruwynah well-watered, suwwt I did.
In some cases comparable raising has led to morphological restructuring.21 Examples are irbt I drank, nist I forgot, ligt I found. Notice
that in such cases underlying |a| no longer has a surface form a.22
Not only when preceding primary stress, but also when preceding
secondary stress (and following primary stress), a may be raised in
certain positions. Examples are (Lgiy): nwikal he was eaten, ttifag
he agreed, jjimal the camel and lmutar the rain. Raising of a in
these previous four examples is optional, but raising of a has become
steady (i.e. it has become morphologically restructured, although it has
remained underlying |a|) in open syllables of imperfect forms such as
(underlying |a| underlined) ynbisit he rejoices, yttifig he agrees.23
2.2.2
Final ah and *- ( )
The fem. morpheme tends to be raised, and not only in pause. The height
of raising is from [e] to [i] (e.g. sayyrih # small, madrasih # school,
mirfih # acquaintance, h jih # thing, zibdih # butter) provided such
18
See gad, Bailey (2004, 457) (glossary).

See (spelled shaour) webpage http://www.saudi-fisheries.comArabic/fish_
product_2.htm
20
In other dialects h amlah is used in the meaning of clan, see Blanc (1970, 114).
21
a no longer appears at the surface as a, but as i. This i, however, is not dropped
in open unstressed syllables (e.g. iribt, not ribt for I drank, the latter of which can
be heard e.g. in the dialect of alAr, see De Jong (2000, 515)), and is therefore to be
regarded as still underlying |a|, see also following fn.
22
In Negev dialect, on the other hand, the |a| reappears as a in closed syllables, e.g.
arbit she drank, cf. Blanc (1970, 134).
23
In closed syllables however, underlying |a| does indeed reappear as a, e.g. yinbstu w
they rejoice, yittfgin they (f.) agree. Notice that in the dialect of alAr, this vowel i is
also underlyingly |i|, which may be concluded from its elision in a form like yniwkil it
is eaten and (not ynwikil), see De Jong (2000, 521).
19
571
raising is not inhibited by phonetic factors, such as preceding emphatics (e.g. h tah #, ajarah # tree, H amda h # name of the tribe/clan) or
(some) X (e.g. srh ah # taking out to graze, ladah sting, bite, but notice
raising in tisih # nine.
Extreme raising of final *- () may be heard in the f. sg. *fal pattern
for colors and bodily deficiencies, e.g. arjy limping, h awly cross-eyed,
aby dark-colored, (a gahawah-form) ah aby light-colored. When
velarized consonants precede, raising remains absent (and often a glottal catch will be audible): bd white, h amr red, safr yellow, zarg
blue, black, xadr green and also tarm having a gap in ones teeth.
Other cases of raising of - () include: salt i evening prayer, ti
summer, btti her house, ilf i diy this viper (although lf iy was also
recorded), ngatti h h i we cut it (f. sg.) to pieces and also ji he came.
Although the final glottal catch presumably originated as a pausal feature, it may often be heard in (especially lento forms in) sandhi as well,
keeping such endings separate from forms ending in t marbtah.
However, when a directly precedes in open syllable, and/or velarization is effective, such raising tends to remain absent, e.g. sma sky,
lada the lunch, d wa he returned home before sunset, ma he
went, fda he sacrificed, but also biyffa he sees her, (a gahawahform) taxarizha24 you stitch it (f. sg.).
3. Stress and phonotactics

As for ordering of rules, the rules for elision stress and anaptyxis are
executed in that same order, i.e. like in other bedouin dialects of Sinai.
3.1
Stress
A sequence CaCaC is usually stressed CaCC in Lgiy, while H amdi y

much more regularly stresses CCaC. The article is often stressed in
both dialects, as in lCaCaC (e.g. ljimal the camel, lmutar the rain,
llah am the meat, aar the hair), but is in a limited number of cases
unstressed in H amdi y alCCaC, e.g. alfrah the wedding ceremony,
ilnam the goats, ilbsal the onions. Comparable stress patterns occur
24
Raising of final in the pronominal suffix -h in neutral environments appears to
occur at random; I have not been able to discover a pattern.
572
rudolf de jong
in the perfect forms of verbal measures VII (or n-1) and VIII (or 1-t):
they tend to be stressed inCCac and iCtCaC in H amdi y, but nCaCaC and CtaCaC in Lgiy, while both varieties stress ynCiCiC and
yCtiCiC. Examples are (Lgiy) nkabas, ynkibis be jinxed, stawa,
ystiwiy become ripe/cooked (H amdi y) inbsat, ynbisit rejoice and
ittfag, yttifig agree.25
3.2
The gahawah-syndrome
As has already become apparent in several examples cited above, the

gahawah-syndrome has left its traces in Lgiy and H amdi y as well.
Some additional examples are ghawah coffee, h awal cross-eyed,
th alib she milks, yazlin they (f.) spin. Notice that when suffixed, the
fem. suffix -ah in gahawah-forms becomes -at, irrespective of whether
the suffix is vowel-initial or consonant-initial, e.g. gahawat my coffee,
gahawathin their (f.) coffee.26
3.3
Stress in gahawah-forms
In verb forms, the gahawah-vowel a has a tendency to behave like

an ordinary anaptyctic, in the sense that it remains unstressed, where
it would otherwise receive stress, (gahawah-vowels underlined) e.g.
th ar tuw you (m. pl.) plough and ya zlin they (f. pl.) spin;27 compare
these to forms like (anaptyctic high vowel underlined) tgu duw you
(m. pl.) sit and contrast with stress in e.g. saltuw you (m. pl.) asked.
3.4 Elision and anaptyxis
Generally, elision of high short vowels i and u (except those that are
underlying |a|) takes place like elsewhere in Sinai.28
25
Notice that group II (i.e. the dialects of the Samnah and Agylah) dialects in the
north have stress patterns inCCaC, yinCCiC. On the intermediate position between
groups I and II in the north of the stress-type inCCaC, ynCiCiC, see De Jong (2004,
159).
26
See remarks below.
27
Comparable forms with vowel-initial suffixes were heard in group I, see De Jong
(2000, 105).
28
See De Jong (2000, chapters IIII, 2.4).
573
4. Morphology
As bedouin dialects do, Lgiy and H amdi y too differentiate between
the masculine and feminine in plural forms of verbs and nouns.
4.1
4.1.1
Nominal morphology
Pronominals
The personal pronominal for the 3rd person masc. pl. in Lgiy and
H amdi y is hum m a, while h in Lgiy and huwwa in H amdi y tend to
be used for the 3rd person masc. sg.29 The suffixed form of the 3rd p.
masc. sg. is -u(h) in both dialects. The 3rd p. m. pl. pron. suffix is usually
-huw in Lgiy, but -hum in H amdi y. Both dialects have hinna and -hin
for the f. pl.
The 2nd person masc. pl. is often intum in H amdi y, but intuw in
Lgiy, the latter of which is also heard in southern Sinai dialects farther
east. Similarly, the suffixed pronominal directly elicited in H amdi y was
-kum, while I only heard -kuw in Lgiy, the latter of which is also heard
in dialects farther to the east. Both dialects have intin and -kin for the
f. pl. The 2nd p. sg. suffixes -k and -k, which are typical for many southern Sinai dialects,30 are also present in H amdi y and Lgiy.
Like almost all dialects of Sinai, Lgiy and H amdi y have stressed 1st
p. c. sg. suffixes -n (obj.) and - (poss.).31
4.1.2
The article
The vowel of the article tends to vary. The article is il- or al-, which
appears to be mainly dependant on the vowel in the first syllable of the
following nominal. If this syllable contains (surface) a, the article tends
to be al, assamm the poison, assaddah the dam, assa fr the yellow,
attamr the dates, alamaliyyah the affair, almayyah the water, alh all
the small cattle and also aldra h the corn.
29
Some central and eastern dialects of southern Sinai have huwwa for the m. pl.,
while h is used for the masc. sg.
30
See De Jong (2004, 163164).
31
See Blanc (1970, 130131), De Jong (2000, 675, map 38) and ibid. (2004, 163, with
remark *** on 164165).
574
rudolf de jong
In other cases, the article il- is much more likely to be used (though
al- occurs in such positions as well), e.g. ilwild the boys, issin the goatskin (used for churning butter), iddims the stone, ilkibdah the liver,
iddinya the world.
In stressed positions (i.e. in Lgiy), the article is almost invariably
al- (or assimilated allomorphs with initial a-), e.g. ssa h an the bowl,
ljimal the camel, ssalag the hunting dog, lh atab the firewood and
dd
a h a the morning, but ti (~ mati) the winter.
Apart from assimilating to sunletters, l of the article will also often
assimilate to j, e.g. jjimal the camel, ajjawlig the carpets, ijjihhl the
youngsters.
4.1.3
T in construct state
The fem. morpheme T (-ah) becomes -at, when aC directly precedes,

e.g.: ragabatha her neck, ragabat my neck, and (gahawah-form) gahawatha her coffee,32 gahawat my coffee and also sanatn two years.
Examples in sandhi are: h arakat stn something like two hours,
maslah at alh urmah the affair of the woman.
Otherwise, T becomes -it, whose high vowel is dropped in open syllables, e.g.: (in closed syllables) jurrithe her traces, rh tuk your smell,
ngitkin your (f. pl.) she-camel, ilbtk your (f. sg.) pack/can, wijitha
her pain (i.e. the pain she causes) and arabiyythum their (m. pl.) car.
In sandhi: wayyit rizz a bit of rice, rih lit ats a diving trip, aklit ixd r a
vegetable dish
Instances with dropped i (often anaptycticshere underlinedare
inserted to eliminate a resulting cluster) ilibt my pack/can, arabytuk33
your car, and in sandhi u l t almayyah of (gen. exponent) the water. rht
ilbirdign the smell of the orange, giblt iliLgt the tribe of the Lgt.
4.2 Verbal morphology
A notable feature of verbal morphology is the vowel harmony in preformatives of the imperfect forms, which is common in many Sinai dia-
32
In this respect, Lgiy differs from dialects of the Negev, where one will hear (also)
gahawat(+), but gahawithin, see Blanc (1970, 142). Other dialects in Sinai show yet
other ways of treatment of T in gahawah-forms, see De Jong (2000, various chapters,
3.1.10.3). Treatment of T in Lgiy is much like what was described for Smniy (i.e.
the dialect of the Samnah) of group II in the north, see De Jong (2000, 279280).
33
Notice that the form is not arabiyytk! Contrast with the form arabiyythum listed
above.
575
lects, but in our dialects discussed here the first p. c. singular is included
in this rule (so that in measure 1 of regular verbs the 1st p. c. sg becomes
homophonic with the m. sg. imperative form), e.g. iktib I write, udr ub
I hit and arab I drink, but also ugm I rise, il I carry, anm I sleep
and jiy I come.34
Another characteristic of bedouin dialects in the south of Sinai
(including Mzniy), though not of all, is the apocopation of 2nd p. masc.
sg. forms of tertiae infirmae verbs. This occurs in all measures (indicated in Roman numbers here), e.g. in Lgiy: (I) tim you go, (I) talg
you find, (II) tsaww you make/do, (III) tlg you find, (IV) tit you give,
(V) taa you have dinner, (VI) talg you meet (with),35 (VIII) titir
you buy and also (with shortened base vowel) tij (~ tjiy) you come.
In a similar fashion, masc. imperatives of tertiae y verbs are usually
apocopated, as in the examples ijr run, ans forget, saww do/make!, it
give! etc. Also when suffixes are appended, the apocopated forms are
used, e.g. ansuh forget him!, sawwha make it (f. sg.)!, ituh-yyh give it
to him! and itwarrh-iyyh you show her to him.36
5. Typological positioning of Lgiy and H amdi y inside Sinai

Lgiy may not very convincingly be part of the group I or Negev dialecttype (as illustrated below in a comparison with the group I dialect of
the Tarbn around Rs Sadr, who are the northern neighbours of the
Lgt), but it is still nearer to group I than H amdi y.37 In this respect the
34
Forms like atlub and aktib (like in dialects of northern Sinai, see De Jong (2000,
3.2.1.2 of chapters 1V)) were also recorded, but the forms with vowel harmony in
the 1st p. c. sg. turn out to be much more regular than I had previously noticed (contrast
De Jong (2004, 166), where only atlub and aktib are listed).
35
Another feature typical of many southern Sinai dialects is the reduction of initial
tt- (> t-, sometimes erroneously referred to as a haplological drop of ta-) in ta- initial
measures (V and VI).
36
The origin of such apocopation may lie in a reasoning by analogy (extrapolation
leading to paradigmatic levelling within the verb system): if forms like e.g. imperfect
tgudiy and tkitbiy (or imperative gudiy and kitbiy) are fem. forms for the 2nd person singular, and the forms tugud and tiktib are used for the 2nd p. sg. masculine, then
dropping the -iy ending from a 2nd p. f. sg. forms like e.g. titiy will yield the form for
the 2nd p. masc. sg.: tit.
37
Largely illustrating Murrays remark (1935, 263 f, see fn. 13 above). The dialect of
the Tarbn of Rs Sadr (not very different from other varieties of Turbniy) is taken
here as representative for group I (or Negev-type) dialects. For the latter type, see Blanc
(1970) and De Jong (2000, chapter 1).
576
rudolf de jong
following dialectal featuresfor the greater part in recapitulationmay

be considered (this list is not exhaustive):3839
Lgiy and H amdiy*
Turbniy
Phonology
phoneme */k/
allophones of /j/
diphthongs *ay, *aw
/k/ and /k/38

[d] ~ often []
// and //
raising of a preceding
(extreme) raising *- ()
velariz. in pl. of katr
middt
ti, lai
ktar (no velarization)
only /k/
[d], rarely []
phon. conditioned ay
and aw
maddt
tiy, laiy
ktr (velarized)
Stress
CaCC (Lg)
CCaC (~ few CaCC)
(H m)
lCaCaC (Lg)
alCCaC ~ lCaCaC
(H m)
Parts of Speech
Article and relative
al-/-il & alliy / illiy
pronoun
Demonstratives: singular
m. (h)da, f. (h)diy
coll., plural dill (ih) (~ hdil)(Lg)
dill (ih) (~ hdl
)
(H m)
Negated pronoun 3. sg.
mh, mh (~ mha)
Pronoun suffix 3. m. sg
-uh
3. m. pl.
-huw (Lg), -hum
(H m)
2 m. sg.
C-uk
2 f. sg.
V(C)-kiy, CC-ik39
Suffixed prepositions
muh
ilh (Lg), ilh (H m)
fh, f uk
Verb impf. 1st p. c. sg.
iktib, udrub,
arab,

ugm, uxu, etc.
Apocopated impf. 2 m. sg. tim, talg, tlg
tit, taadd, etc.
38
CaCC
lCaCaC
only al- & alliy

m. hda,
f. hdi y
hdal
/
hd
l

mh, mh (~ mha)
-ah /-ih
-hum
C-ak
invariable -kiy
mah
alh
fah, fak
aktib, adrub,
arab,

agm, axu, etc.
timiy, talga, tlgiy,
titiy, taadda, etc.
But this phonemic opposition is not as widely used in Lg.

This is not entirely certain; my H amdiy
informant produced btk your (f. sg.)
house, but also naxal tkiy your (f. sg.) date palm.
39

Verb to come
perf. 3.m. (sg.)
impf. 3.m.
Suffixed prepositions:
Vowel impf. to eat, to take
Vowel perf. 3rd p. f. pl. impf.
-m in 2nd m. pl. perfect
Interrogatives
when?
who?
Adverb
here
Selected lexicon
mother
baking sheet
i (pl.) um ~uw
(sg.) ja (pl.) jaw
yiy, yuw
muh
ilh (Lg)
yjiy, yjuw
mah
alh
ykul, yxud (H m)

ykil, yxid (Lg)
katabin, glin,
yarabin
katabtuw (Lg),
katabtum40~-tuw
(H m)
ykil, yxid
kataban, gl an
yaraban
katabtuw
imtn
mn
mat (~ wagt)
min
nih(-niy)
hniy
um m
z
am m
sj
577
* Where not specified, the listed form is heard in both dialects.40
6. Conclusion
The comparison drawn above shows a number of differences between
Lgiy/H amdi y and Turb niy. Notwithstanding the characteristics
shared only by H amdi y and Turb niy, however, Lgiy still appears to
be typologically nearer to Turb niy than H amdi y.
7. References
at-Tayyib, Muhammad Sulaymn. 1997. Mawsat al-qabil al-arabiyya, Part 2, Cairo:
Dr al-Fikr al-Arab.
Bailey, Clinton. 1974. Bedouin Weddings in Sinai and the Negev. Folklore Research
Center Studies, vol. 4, Jerusalem Magness Press, 105132.
. 1985. Dating the Arrival of the Bedouin Tribes in Sinai and the Negev. Journal of
the Economic and Social History of the Orient 28, 2049.
. 2004. A Culture of Desert Survival. Bedouin Proverbs from Sinai and the Negev. New
Haven: Yale University Press.
40
Verb forms with final -m were only recorded during direct elicitation. Such forms
also occur in group II of the north, see De Jong (2000, 298299).
578
rudolf de jong
Behnstedt, Peter. 1979. Die nordmittelgyptischen bukara-Dialekte. Zeitschrift fr arabische Linguistik 3, 6295.
Blanc, Haim. 1970. The Arabic Dialect of the Negev Bedouins. Israel Academy of Sciences and Humanities, Proceedings, vol. 4, 112150.
de Jong, Rudolf. 2000. A Grammar of the Bedouin Dialects of the Northern Sinai Littoral. Bridging the Linguistic Gap between the Eastern and Western Arab World. Leiden:
E. J. Brill.
. 2004, Characteristics of Bedouin Dialects in Southern Sinai. Approaches to Arabic
Dialects. A Collection of articles presented to Manfred Woidich on the occasion of his
sixtieth birthday, ed. by Martine Haak, Rudolf de Jong, Kees Versteegh, Leiden: E. J.
Brill, 151175.
Hava, J.G. 1982. Al-Farid ad-durriyyah, Arabic-English Dictionary (fifth edition). Beirut: Dar el-Mashreq.
Lavie, Smadar. 1990. The Poetics of Military Occupation. Berkeley: University of California Press.
Marx, Emanuel. 1999. Oases in South Sinai. Human Ecology 27, 341357. New York:
Kluwer Academic.
Murray, George W. 1935. Sons of Ishmaela Study of the Egyptian Bedouin. London:
Routledge.
Oppenheim, Max Freiherr von. 1943. Die Beduinen, Band II: Die Beduinenstmme in
Palstina, Transjordanien, Sinai, H edjz. Leipzig: O. Harrassowitz.
Stewart, Frank Henderson. 1990. Texts in Sinai Bedouin Law, part 2 (Texts are in Arabic). Wiesbaden: O. Harrassowitz.
Tckholm, Vivi. 1974. Students Flora of Egypt. 2nd ed., Beirut: Cairo University (Cooperative Printing Company).
8. Texts
The speaker is a member of the Lgt, who was 38 years old at the time
of recording. He was born in Sarbt alXdim (about 40 km east of
Abuw Znmah) in the area called ar-Ramlah, where he had lived until
his 29th. He has had no school training. (S) = Speaker of Lgiy, (R) =
interviewer.
8.1
Hunting rabbits
8.1.1 Transcript
1. (R) hatkallimni fi sayd ilarnib? 2. (S) a:ywah. alarnib dilleh41 . . . fi
nnahr bitnm. fi nnahr . . . bitnm m btarta.42 3. ibtarta ar h
billl . . . ugb almaarib . . . ibtusrub. iza nymah fi h ajr, fi ajarah . . .
41
dill (-ih) is the c. pl. demonstrative pronoun for near deixis. Notice the absence of
velarization.
42
The single negation m + verb form is regular.
579
ibtusrub h fi lll bitdawwr alma. 4. idda ll43 h srbeh, iw srh ah . . .

rr45 b iwayyeh . . .
l assu bih . . . . la:44 lfajir. alfajir xals giddm aam la dd
u
5. bitdawwir lhe . . . ajarah . . . bitdawwir lhe jibl, bitdawwir lh-ayyi
h jeh, iw bitgtir fha bitnm. 6. iygl luk46 alarnab nymeh. alarnab
dd
a h a, w issu bih . . . tulgutha . . . iw h nymeh. 7. ibyalguw jurritte,
yalguw jurritta ind ijar da ind ijar da ind ijar da. 8. iw
biygtiruw mi jurritte . . . lam biyrawwh uw ind . . . manmha.
8.1.2
Translation
1. (R) Are you going talk to me about hunting rabbits? 2. (S) Yes. These
rabbits . . . they sleep during the day. They sleep during the day and they
dont graze. 3. They only graze at night . . . after sunset. They run out,
if they sleep in the rocks, into the bushes . . . They run out at night to
look for food. 4. They keep running . . . grazing till the morning, (all
the time) until daybreak. Daybreak, (after that its) over, just before the
sun rises . . . 5. It looks for a bush . . . it looks for a mountain, it looks for
anything, and goes (in) there to sleep. 6. Hell tell you the rabbits are
asleep. The rabbits in the forenoon, which is the morning . . . you seize
them . . . while they are asleep. 7. They find their tracks, they find its tracks
near this bush, near that bush. 8. And they follow its tracks . . . until they
come to . . . its sleeping place.
8.2.1
Transcript
9. fh . . . ns ibyulgutha . . . b duh. iw fh ns biyannin alhe. biyannin

alha yaniy bitkn imh47 ash . . . igsayyreh. 10. ibrn kdiy . . . hda
ym ffa48 nymah fi ajarah, iyannin alha w yuxbutha b ilash . . . 11.
iw linnha49 tbbah fhe.[. . .] iw fh ns . . . induh salg. 12. ssalag hda
43
dd
a l: assimilation of initial t to follwing d in tdall.

Such prosodic lengthening of the vowel is often used to indicate the long duration
of time.
45
dd
rr, see fn 43.
u
46
The suffixed preposition l: (sg.) 3.m. luh, 3.f. lha, 2.m. luk, 2.f. lkiy (~ seldom lik),
1.c. lay ~ l and (pl.) 3.m. lhum , 3.f. lhin, 2.m. lukum, 2.f. lkin, 1.c. lna.
47
The suffixed preposition m(i): (sg.) muh, mih h a, muk, mik (~ frequent mikiy, i.e.
like in group I), m and pl. mih h um , mih h in, mikum , mikin, mina.
48
ffa: fha one of many examples of assimilation of h to a preceding voiceless
consonant.
49
The particle linn or lann is often used to present a sudden or unexpected turn in a
narration, see Blanc (1970, 145).
44
580
rudolf de jong
. . . biynaggizhe . . . biynaggizh-ssalag, 13. w imarabbh . . . ind ssalag i

mna . . . biyda lliy . . . ssalag yizatti . . . lamma yulgutha. 14. m ykilha
lamma yjbha la sh buh. iygm sh buh dbh h e. hda larnab. (R)
aywah . . . 15. (S) iw fh ns biyannin alha tann bi lash biyirmha.
16. iw fh ns ibyulgutha. imh abtuh, imh bantuh muh
h jah lm ffa fi ajarah, ramha alha w lagatha. hda larnab.
8.2.2
Translation
9. There are people who seize her with the hand, and there are people
who take aim at it. They take aim at it, that is, theyll have a small stick
with them. 10. Of two spans of the hand (in length) . . . this (person)
when he sees it asleep in a bush, he creaps up on it and hits it with the
stick. . . . 11. And lo it (i.e. the stick) hits it (i.e. the rabbit). And there
are people who have a hunting dog. 12. This hunting dog will make it
jump around. The hunting dog makes it jump around. 13. And he has
taught it . . . with the hunting dog . . . the hunting dog keeps following it
closely . . . until it grabs it. 14. He wont eat it until it has brought it to its
master. His master then slaughters it, this rabbit. 15. And there are people who take good aim at it with the stick, which he throws. And there
are people who seize it. 16. Hell have his cloak with him, (or) his jacket
with him, or something, (and) when he sees it in the bush, he throws it
at it and seizes it. This is (what they do with) the rabbit.
8.3.1
Transcript
17. (R) iw fh bardu

h . . . biysawwuw faxx? (S) biysawwuw. alfaxx da
grayyib, alfaxx da grayyib byistagduw50 alh nns mi zamn. 18.
igrayyib byistagduw a lfaxx. yibnh,51 w iyh uttu w fh . . . iyh uttu w fh
giirit52 birdagneh, iyh uttu w fh ayyi h jah lfaxx. 19. iw h btjiy a
53
rhtuh . . . hda byibnh ar ugb almaarib. ym lanam tidw
iy, w
alh all dwa xals . . . m f h all, ibyibnh. 20. m byibnh fi nnahr.
50
A measure X (or ista-1) verb of the root q-d-y: istagda, yistagdiy (ala) take up as a
new habit by following an example.
51
yibnh lit. they build it, here they set it (i.e. the trap).
52
giirit: a bukara-form of girit.
53
da wa, yidwiy
go home in the evening, see Stewart (1990, 214) (glossary). In

Turbniy this verb is measure IV, e.g. albant midiwyt
the girls are going home (just

before sunset). The root d-w-y
is probably related to CA d-w- light; using the last day
light to go home.
581
iw tniy h jeh byibnh fi lgd . . . wagt assiyyl ym assiyyl, fh barm iw

h l, 21. ibyibnh kamn fi irg54 assiyyleh. fh barm biyth h bitrawwi
diy55 fi lgd, fi ahar assabah w ihr issittah w ihr dillah lgd. 22. fh
barm biyth fi ssiyyl, talgha56 tamalliy fi ssiyyl alarnib bitrawwih . . .
8.3.2
Translation
17. (R) And are there . . . (people) who make a trap? (S) Yes, they do.
This trap is (a) recent (thing), this trap is something recent, which
people have imitated, not people of the old days. People have started
copying the (method of hunting with a) trap. They set it, and they place
in it . . . they place the peel of an orange in it, they put anything in the
trap. 19. And it (the rabbit) is attracted to its smell. This (person) sets it
only after sunset. When the small cattle come back home. And when it
has come back home . . . thats it . . . there is no small cattle (left roaming
around), (then) they set it. 20. They dont set it during the daytime. And
besides this they set it in summer . . . the time when the acacias bloom,
when the acacias have fruit57 and stuff. 21. They also set it next to the
acacia tree. There is fruit that falls down. It comes normally in summer,
the month of July, the months of June and those of summer. 22. There
is fruit that falls from the acacias, (so) youll find rabbits going to the
acacias all the time.
8.3.3
Transcript
23. ibyibnuw lha fa . . . alfaxx f-assiyylah diy, iw kamn fi ssiyyl d k,
fi ttalah58 diy byibnuw lhin. 24. iw yulugtha f- alfaxx . . . yulgutha
min ragabatta min . . . min rjilha min h jeh. 25. assu bih ym yjiy l
alfaxx iw lannha malgtah. 26. iw fh ns ibyugud . . . igrayyib lha . . . fi
lll . . . yasma xabit alfaxx. 27. ym ytulguh byasmauh (R) aywah . . . (S)
54
fi irg, litt. in the root (of) has been grammaticalized as a preposition meaning
next to (presumably next to some standing object), also e.g. fi irg alh tah next to the
wall.
55
Assimilation of bitrawwih + diy.
56
talgha: apocopated 2nd p. m. sg. imperfect of the verb ligiy, yalga find, followed by
suffix -ha.
57
Lane (1863, part 1, 195): [. . .] the fruit of the talh [or acacia gummifera, which is
of the trees called idh] [. . .]
58
talah (with t!) is a water course between two mountain peaks, and can be used as
a pass between mountains.
582
rudolf de jong
ibyarf alfaxx sdha . . . ibylgutuh . . . fi lll 28. iw fh ns ibyibnh . . . iw

biygtir ibld bideh . . . zayy xamsah kluh aarah kluh ssu bh ibyjiy luh.
29. ym yjiy luh linnuh, linnuh59 lgitha. hda larnab. 30. amma lazl
hda, m kull anns ibtugnus luh, alizln dillah m biyth in fi l . . . fi
lwtiy, ar fi jjbl, fi jjbl albideh . . . 31. (R) ar tatla fg . . . (S) ar tugnus
lhin fi lijbl albideh . . . alazl. 32. w alazl law arwh uk60 . . . ibyunguz min induh. 33 law arwah rh tuk . . . yaniy jt a rrh w arwah , xals
biygtir min induk. hda lazl. 34. alliy h lbadan w alh jt dillah
kulluh. hda . . . hda wn ilinsn ibylugtuh.
8.3.4
Translation
23. They set a trap for it in (under) this acacia, and also in those (other)
acacias, in this watercourse they set (traps) for them. 24. and they catch
it in the trap. . . . they catch it by its neck . . . by its leg . . . by something. 25.
In the morning, when he comes to the trap, there it is, caught. 26. And
there are people who sit (and wait) . . . near it . . . at night . . . hell hear the
snapping of the trap. 27. When it releases hell hear it . . . (R) Yes . . . (S)
and he knows that the trap caught it . . . (and) he seizes it . . . at night 28.
and there are people who set it . . . and they go away to a far place61 . . . like
five kilometres, ten kilometres (and) in the morning he comes (back) to
it. 29. When he comes to it, lo, lo, it has caught it. That is the rabbits.
30. As far as these gazelles are concerned, not all people hunt them.
These gazelles dont come down to the lower areas, (they stay) only in
the mountains . . . 31. (R) You have to go up . . . (S) You have to hunt them
in the far mountains . . . Gazelles. 32. And gazelles, when they smell
you, they jump away from where they are. 33. When they smell your
scent. . . . that is, if you come down with the wind and he smells (you),
thats it, he gets away from you. These are the gazelles, 34. which are the
ibex and all these things. This . . . this is where a person catches them.
59
60
61
See fn. 49.

The (superscript) anaptyctic vowel is voiceless.
bld lit. land, here place.

8.4
583
Snakes
8.4.1 Transcript
62
35. (R) fh bardu
h, rif fi ddd? (S) iddd allafiy? (R) aywah . . . (S)
allafiy dilleh . . . hda l . . . ayyi insn ibyuktilhe. 36. hdiy ygl luk
almzyeh, in kaltuk bidduk,63 ir kn64 daktr walla bidduk, aza
f-albarr kamn m h waluk daktr ir kn insn h wiy. 37. fh ns
f-larab . . . biykn h wiy. [. . .] fh hwytuh xabit, iw fh ihwytuh lah s,
alliy hda biyrudd assamm. 38. hda ddd, alliy h lagrab . . . assa fr diym m dl . . . iw fh lfiy . . . alliy h . . . h tti bn. hda la . . . hda lam kaltuk
r kn insn h wiy walla daktr. 39. fh ns katr mtuw minha. imn
alamaliyyah diy. (R) mi ssamm . . . (S) mi ssamm, hda ssamm . . . 40.
(R) tab w ilh wiy biysawwiy h? (S) alh wiy biyrudd issamm . . . biyrudd
assamm . . . 41. (R) kf biysawwiy? (S) alh wiy byaxabtuk xabit . . .
yaxabtuk . . . talat xabtt . . . ib rguh . . . (R) ib rguh . . . (S) hda biyruddd
assamm. lamma trawwh addaktr, aw mumkin kamn m . . . m trawwh
addaktr ihwytuh jiyydeh, bitkaffiy xals. 42. hda b innisbah la . . . la
lagarab65 w alfiy w atti bn . . . w alh jt dillah.
8.4.2
Translation
35. (R) Are there also . . . do you know about creepy crawlies? (S) Vipers?
(R) Yes. . . . (S) Yes, these vipers . . . these . . . anyone will kill them. 36. This
one will say to you the advantage, if it bites you, you need, by necessity
a doctor or you need, if it is in the desert, and there is also no doctor
around (you), a hwiy person. 37. There are people among the bedouins . . . who are hwiy. Hitting is part of his hwiy-activity, and licking is
part of his hwiy-activity, which brings back (i.e. out of the body) the
poison. 38. This is the creepy crawly, which is the scorpion, this yellow
one with a tail . . . and there is the viper . . . which is the snake. This one,
when it bites you, you need a hwiy or a doctor. 39. There are many
people who died from it, from this business. (R) from the poison . . . (S)
62
dd (n.u. -ih, pl. ddn) is used for any crawling animal, like snakes, scorpions, and
lizzards.
63
bidd, rather than widd (as in group I), is used to express want or need. Here it us
used to express necessity from the perspective of the speaker, like English should, as in
he should see a doctor, see De Jong (2000, 239).
64
The context calls for words like ayr kn. In the recording I hear what has been
transcribed above, but perhaps it should read (strongly reduced) er kn.
65
ag araba bukara form for agrab.
584
rudolf de jong
From the poison, this poison. 40. (R) Okay, and what does the hwiy do?
(S) The hwiy brings back the poison, he brings back the poison . . . 41.
(R) how does he do it? (S) The hwiy gives you a good slapping. He slaps
you three times . . . with his saliva. . . . (R) With his saliva. (S) This brings
back the poison until you get to the doctor, and its also possible that
you dont (have to) go to the doctor. If his hwiy-work is good, thats
enough, thats it. 42. This in reference to the scorpion and the viper and
the snake . . . and these things.
8.5
8.5.1
Falling in love
Transcript
43. (S) lwalad . . . lwalad biygm . . . aza srih . . . biyf albint . . .

44. (R) bifha wn? (S) biyfha fi masrah h a w h srh ah b anamhe. 45.
jabatuh . . . iw h l . . . biygm . . . biyxarrf um m uh, h m biyxarrf ubh,
biyxarrf um m uh. 46. ummu
h bitxarrf ubh. bitgl lwalad . . . ryid bint
ifln . . . imn larab iliflniyyah. . . . 47. biygm ubh . . . biyrawwih l ub
ha . . . iw . . . biyrawwih l ubha w biyxarrfuh. 48. w um m uh . . . bitrawwih
l um m albint, iw bitxarrifhe. gl abuw lbint mh maxatbeh . . . iw binjawwizkuw . . . hda fh kalm tniy. 49. gl abuw lbint maxatbeh66 . . .
hda fh kalm. gl abuw lbint kamn yizha walad ammha . . . hda
fh kalm tniy. 50. iw fh ns biyffa . . . min tanabh.67 tanabh alliy
induh biyf bint ibtijbuh . . . biyxarrf um m uh. 51. biygl ana yz ilbint
lifln diy. 52. bitrawwih l um m ha w bitxarrifha w tasalha w itgl h
lbint maxatbah walla bint ass ibnt walla kidiy walla kidiy . . . 53.
bitxarrifha. baadn ubh . . . biyrawwih l arrijjl . . . abuw lbint . . . iw biyxarrfuh. 54. biyxarrfuh gl h biyjawwzuk xals albint mh maxatbah,
iygl luk xals. 55. iza lbint isayyrih biygl luk xalluk ugub sanah.
xalluk ugub xamis t-uhur, xalluk ugub sanatn . . . 56. biygm byadbah
alhe . . . biyjb tuh gl luh xalluk ugub sanah, sanatn . . . albint
isayyrih. 57. biygm biyjb tuh w yadbah h a. dabah h -ath algasalah.
58. min ym ath lgasalah xals irif hdiy h rumtuh, ib sinnt All h w
rasl-ath . . . gsalatuh . . .
66
maxatbah, gahawah-form of maxtbah.

tinb, tanaba member of the same encampment, see Stewart (1990, 273274,
glossary).
67
585
8.5.2 Translation
43. (S) The boy . . . the boy then goes . . . when he is out herding . . . he sees
the girl . . . 44. (R) Where does he see her? (S) He sees her in her herding place where she is herding her goats. 45. If he likes her. . . . and in
case. . . . he then goes . . . and speaks to his mother. He doesnt speak to his
father, he speaks to his mother. 46. His mother speaks to his father. She
says The boy wants (to marry) so-and-sos daughter . . . from so-and-sos
family . . . 47. His father then goes to her father . . . and . . . he goes to her
father and speaks to him. 48. And his mother . . . goes to the girls mother,
and speaks to her. If the father of the girl says she is not engaged . . . and
we will marry her into your family . . . (then) theres more talk. 49. If the
father of the girl says Shes engaged . . . (then) there is a talk. If the girls
father says Her cousin also wants (to marry) her, (then) theres more
talk. 50. And there are people who see her, of their neighbours . . . one
of his neighbours that he has sees a girl that he likes . . . 51. hell (then)
speak to his mother. Hell say I want (to marry) so-and-sos daughter.
52. She goes to her mother and speaks to her and asks her and she says
the girl is engaged or she is still an unbetrothed virgin or thus or thus...
53. After that his father speaks to her . . . he goes to the man . . . the girls
father, and he speaks to him. 54. He speaks to him. If he says that hell
marry the girl you, she is not engaged, hell say to you okay. 55. If the
girl is young, he says to you wait (until) after one year, wait (until)
after five months, wait (until) after two years. 56. He then slaughters
for her. He gets his sheep, if he says to him wait (until) after one year,
two years . . . the girl is (still too) young. 57. He then brings his sheep
and slaughters it. When he has slaughtered it, then they have betrothed
her.68 58. From the moment they have given him his twig, thats it, he
knows that this (girl) is (going to be) his wife, in the tradition of God
and His Prophet69 they have given him . . . his twig.
68
ath gasalah, lit. they have given him a twig. The twig is given to the groom in
betrothal ceremonies, who is then mxid gasalatha holding her twig, i.e. she has been
betrothed to him, see Bailey (1974).
69
The phrase b sinnt Al l h w rasluh in the tradition of God and His Prophet may
often be heard in additionused almost as an excuse in pre-emptionto the description of traditions, of which the islamic origin is doubtful (or non-existent).
586
8.6.1
rudolf de jong
Transcript
59. iw h l . . . gm xallha w ugub sanatn . . . zabbat70 lh atab, iw jb

idda byih , iw jb ibyt aar. 60. iw azam anns, iw gl lhuw lfarah
ind . . in All h jjimah jjyah jjimah lliy drha, iw banna lbyt . . . ym
alxims . . . assu bih . 61. iw sawwa. . . . atta h wiy. iw juw anns d kull
wh id ib h amltuh.71 kull wh id ib tuh . . . kull wh id ib h amltuh
w h rumtuh muh. 62. ym alfrah ind ifln [linna] imrawwh n
nah da ruh, kull wh id ib h amltuh. iw rawwah uw, iw ugub salt
alasir kamn . . . dabih . 63. dabah iw saww-laa. iw fh ns bitsawwalfitr . . alfarh 72 . . . arrjl w alih ra yyim biysawwin. 64. iw gassam . . . kurr
rijjl 73 firh ah iw h iggtuh lh amatuh fha . . . gassam alhuw . . . kullhuw.
65. iw ugm a74 bat iam baadn . . . ugub salt i . . . h awalha . . .
gm at iddah h iyyih.75 66. fh ns bitsawwiy dah h iyyih, iw fh ns bissawwiy marbah.76 67. almarbah lliy h arrafh iy.77 w iddah h iyyah-lliy
h . . . alxumus. iddah h iyyah kull wh id biygl min induh . . .
8.6.2
Translation
59. And then the situation is . . . he leaves her be,78 and after two
years . . . he arranges the firewood, and brings animals for slaughter, and
brings tents. 60 And he invites people, and says to them the wedding
party is at my house, God willing, coming Friday, Friday next week
and he has set up the tents on Thursday . . . in the morning . . . 61. And
70
Although z as a phoneme is rare (and marginal as such) in Sinai bedouin dialects
(the current reflex for both *d and *d is interdental emphatic d), it is regular in lexemes
(loaned from MSA or Cairene Arabic) like zabbat , yzabbit arrange; do properly (Stewart (1990, 285286, glossary) transcribes z), n(i)zm system and z(u)rf circumstances,
see De Jong (2000, 60).
71
Although h amlah is listed by Blanc (1970, 114) in the meaning of clan, in dialects
of other tribes it means animal led to a festive occasion to be slaughtered as a present
for the host.
72
farh , sg. firh ah large flat (and thin) round bread baked on the sj.
73
kurr rijjl: assimilation of kull rijjl every man.
74
ugm a: assimilated ugb + ma.
75
dah iyyah a type of bedouin dance, see Bailey (1991, 436, glossary). During
dah iyyah, men stand in a line, while women dance in a line opposite to them.
76
In a marbah older men stand in a square facing each other.
77
It was later explained to me that the rafh iy is a dance unlike marbah: during
rafh iy, in which younger people take part, the boys will sing their rhyme, while the
girls dance in front of them (one at a time, often while holding a stick), which is much
more in line with the meaning associated with the root r-f-h , greet (new spouses) in
Hava (1982).
78
The verbal perfect forms used by the speaker have here been translated into imperfect forms.
587
he has prepared the food (to be served before the wedding ceremony),
and the people have come, everyone with his slaughter animal, everyone
with his sheep. Everyone has his slaughter animal and his wife with him.
62. On the day of the wedding ceremony with so-and-so we go over
(to him) and bring him over. Everyone (goes) with his slaughter animal and goes, and also after afternoon prayer . . . (there is) slaughtering.
63. He slaughters and prepares dinner. And there are people who make
flat (unleavened) bread . . . farh . . . the men and women make them.
64. And he distributes . . . (to) every man a flat bread and his portion in
which there is his meat. . . . he distributes it to them . . . all of them. 65.
And after the sun has set, after that . . . after the evening prayer . . . round
about it (i.e. that time) . . . the dancing starts. 66. There are people who
do a dah iyyah, and there are people who do marbah. 67. The marbah
is the same as the rafh iy. And the dah iyyya is . . . the clapping of hands.
(during) The dah iyya everyone says his own lines (of poetry) . . .
8.7.1 Transcript
68. hda xams t-infr aarah . . . xamistar nafr biysawwuw, w
arraggsah h urmah wih dih . . . ibturgus alhuw. hda b innisbah l
addah h iyyeh. 69. w almarbah hdiy . . . m . . . biyrawbuw fh ar anns .
. . ijjihhl79 . . . alliy hinna lbant . . . ibyrugsin . . . w ilwild. 70. ijjihhl hdil
biysawwin . . . almarbah. 71. assmir da biysawwh ar aiyyb. anns
alkibir hdil bysumruw. ibysumruw a jl . . . ismuh smir. 72. amma
ddah h iyyah hdiy . . . biysawwha . . . kitr . . . fi Sna nih biysawwha
wjdah ddah iyyah. 73. ym alfar[h awt], iw ym azzawyir, iw ym
ald . . . biysawwuw ddah h iyyah. hda h . . .
8.7.2 Translation
68. This, five, ten . . . fifteen persons do this, and the dancer is one
woman . . . who dances before them. This as far as the dahiyyah is concerned. 69. And this marbah . . . only young people take part in the
marbah . . . which is the girls . . . who dance . . . and the boys. 70. These
youngsters take part . . . in the marbah. 71. In the smir only older people take part. These old people take part in the smir. They do a smir
on the fringe (of the main festivities) . . . it is called smir. 72. As far as
this dahiyyah is concerned . . . they do it . . . a lot . . . Here in Sinai they do
it a lot . . . the dahiyyah. 73. When there are weddings, and when there
79
jhil lit. ignorant is commonly used in Sinai for young or child.
588
rudolf de jong
are visits (to sheikhs tombs), and when there is the feast . . . they do a
dahiyyah. Thats it . . .
8.8.1
Transcript
74. (R) w alars . . . w al . . . (S) ilurs fi lbt (R) fi lbt . . . 75. (S)
ywah . . . alurs . . . itsawwiy zzaffih humma , iw jbha b ilarabiyyt iw
waddha . . . albt. 76. alurs hadiy btugud saba t-iyym, iw h fi lbt.
77. m btatla minnuh wala biyhiddh bt alfrah . . . mabniy . . . 78. sabi
t-iyym mn ajjimeh . . . l ajjimeh. 79. iw h gdih80 lars . . . batnuh . . . iw
irssa indaha. l(a:) jjimah jjyih. 80. ajjimah jjyih ssawwiy sib,
dabah luh h tn . . . 81. kamn h idi r alliy yiz yah da r,81 w alliy
mh yiz mh yiz . . . 82. iw ugub kidiy diy[t] waddha druh.
aza druh . . . hlih . . . waddha-yyha. mh hleh druh, iw h l . . . 83.
bitrawwih ind ummh
a. ibtasrah ib anamha. 84. lam biysawwiy druh,
iw h l . . . iw hdiy h rumtuh induh. 85. hda b innisbah l ijjawzeh zayy
kidiy.
8.8.2
Translation
74. (R) And the bride and the groom. (S) And the bride is at home (R)
At home . . . 75. (S) Yes . . . the bride, they organise a wedding procession.
And they have brought her with cars and brought her over to the house
(tent). 76. The bride spends seven days (while she is) in the tent. 77.
She does not go out of it (i.e. the tent), nor do they take it down the
wedding tent . . . stands (lit. is built) 78. For seven days from Friday till
Friday. 79. And she sits inside it, the bride . . . while her groom is with
her. (all the time) until the next Friday. 80. The next Friday you celebrate
the weeks feast, he slaughters one or two sheep (for himself) . . . 81. And
also whoever wants to attend is present, and whoever does not want,
does not want . . . 82. And after that he then takes her to his house. If
his house is . . . ready, he takes her to it. If it is not ready his house, 83.
then the situation is. . . . she goes to her mother. She takes her cattle out
to graze. . . . 84. When he prepares his house, then the situation is . . . and
this is then his wife with him. 85. This is with reference to the wedding,
like this.
80
in gdah is quite high; when in neutral environmentsin the CCiC pattern
of the active participle tends to be realized near IPA [:].
81
yah dar
he is present, notice the absence of the gahawah-syndrome here.
589
8.9.1 Transcript
86. (R) bardu
h alars ibturud? (S) ha? (R) fh ns biygluw inn alars
ibturud . . . (S) ibturud. iza mh rydtuh . . . ibtuurud innuh. 87. itgl
mn yztuh tgtir, tudxul ibyt bid . . . an um m ha w ubha tgtir
bad . . . 88. itgl mn yztuh. biyjbha luh tniy w biy . . . asil min
ym mh yiztuh turud . . . marra h marrtn taltih xals lzim
iytallighe . . . xals mh yiztuh. 89. hdiy lars. hda mitjaww . . .
mxidha asib . . . hda lzim iytalligha . . . 90. yiztuh, ibtugud imn
awwil l turud wala h jeh. 91. min ym yizt lwalad da xals. hda
lars . . . biysawwiy btuh, aza ind um m uh, aza ind arbnuh, aza ind
nasbtuh . . . taba art. 92. irt ind awwil ma jawwazh. aza gluw
la lzim tuskun indina . . . xals, biysawwiy btuh induhuw.
8.9.2
Translation
86. Does the bride flee as well?82 (S) What? (R) There are people who
say that the bride flees . . . (S) She flees. If she doesnt want him . . . she
flees from him. 87. She says I dont want him, she goes away and
enters houses83 far away . . . from her mother and her father, she goes far
away . . . 88. She says I dont want him. They bring her back to hem again
and he . . . because when she does not want him she flees . . . Once, twice,
three times, and thats it, he has to divorce her . . . Thats it, she doesnt
want him. 89. This is the bride. This one (man) married . . . has taken
her by force . . . this (person) must divorce her . . . 90. If she wants him,
shell sit with him from the start and wont flee or anything. 91. When
she wants the boy, thats it. This groom . . . sets up his house, be it with his
mother, be it with his clan, be it with his kin (of fathers side) . . . according to the stipulations. 92. Conditions (that were agreed on) from the
moment they married him (to her). If they say No, she has to live with
us . . . then thats it, he sets up his house with them.
82
With some tribes it was customary for the bride to flee from her groom, see Murray
(1935, 182183).
83
The verb daxal, yudxul here actually means to enter a house to seek refuge (i.e. as
a daxl).
590
rudolf de jong
8.10
Agriculture
8.10.1
Transcript
93. (R) bardu

h . . . anns ibyazrauw batt x? (S) ibyazrauw batt x . . . 94.
wagt lmatar fh ns ibyazrauw batt x . . . -lmatar. (R) kf biysawwiy? (S) biysawwiy saddeh . . . imsawwayih min zamn . . . 95. ibyah aritta
-jjimal . . . ym lmatar biytx . . . iw biynaggitha . . . 96. (R) b ilijml, b
iljamal, fh fard? (S) aywah byah aritta . . . hda l algamih . . . 97. (R) iw
fh fard walla bitgluw h? (S) aywah, farid . . . farid . . . (R) iw alha bg?
(S) aywah, marbtah f-ljimal . . . iw msikka . . . arrijjl mn-wara. 98.
hda . . ibyah arit l algamih . . . hda . . . h ittih l algamih , iw xalla h ittih
l al . . . l albatt x. 99. albatt x da . . . nggatuh b ilfs. 100. bizruh fi jambuh, iw yah afir ingr a tl b ilfs iw h biynaggit. 101. albatt x hda
indma tili diy byatla -lmatar. min ym assaddih ruwynih byatla.
ibyunur . . . 102. biytaabbak fi baadu
h biysr batt x wjid . . . bass
anniswn m tjh wala h add . . . 103. min ym ibybudruh biygtruw
annuh [ba] a jl. ibynkibis 104. in nkabas bz84 albatt x. hda . . . l
albatt x. 105. w algamh hda . . . ibyatla ym rabbna biyrd. 106. algamh
hda lliy binsawwiy minnuh l . . . addagg w alh l da, hda h. w attibin
. . . alliy h biyjbh l ajjml . . .
8.10.2 Translation
93. (R) Do people also grow watermellons? (S) They grow watermellons . . . 94. At times of rain, there are people who grow watermellons . . . on rain (water). (R) How does he do (it)? (S) He makes a dam . . . It
was made long time ago . . . 95. He ploughs it with the (plough drawn
by a) camel . . . When the rain falls . . . and he sows it. . . . 96. (R) With
camels, with the camel, is there a plough? (S) Yes, he ploughs it . . . this
wheat . . . 97. (R) And is there a plough, or what do you call it? (S) Yes, a
plough . . . (R) And with a (large trumpet-shaped) funnel on it? (S) Yes,
tied to the camel . . . and holding it . . . is the man at the back end. 98. This
(man) ploughs for (sowing) the wheat, a piece (of land) for the wheat,
and he leaves a piece (of land) for the . . . for the watermellons. 99. These
watermellons, he sows them with the hoe. 100. He keeps his seed by his
side, and he digs holes with the hoe while he sows.85 101. These water-
84
85
bz, ybz go bad is a loan from Egyptian; therefore z.

Lit. let drip, here in the meaning of dropping one or two seeds in a hole at a time.
591
mellons, when it comes up normally, it grows on rain (water). When the

dam86 is well-watered it comes up and spreads out 102. It becomes intertwined, and the watermellons become plentiful. But the women dont
come there . . . not one. 103. From the moment he sows it, they go away
from it (and stay) on the side. (if not) It is jinxed. 104. If it is jinxed the
watermellons will go bad. This . . . is with reference to the watermellons.
105. And this wheat . . . comes up when our Lord wills it. 106. This wheat
is what we make flour from and this situation, thats how it is. And the
straw . . . which they take (as fodder) to the camels . . .
8.11.1
Transcript
107. (R) bardu

h indukuw rh iy? (S) ha? (R) walla bitgluw , rh iy walla
rh h? (S) arrh h . . . [. . .] arrh h lliy h btath an algamih . . . hda-smha87
rh h h ajr . . . 108. (R) aywah. w alh urmah alliy btath an? (S) aywah
lh urmah btath an, iw fh irjl ibyath anuw. 109. (R) aywah, biysawwuw
? (S) biysawwuw fh ns ibyath anuw . . . aldra h. aldra h hdiy biysawwuw minha . . . biysawwuw minha . . . libbih. 110. (R) ismuh ? ildra h?
(S) aldra h88 hdiy biysawwuw minha libbih. allibbah hdiy biysawwha
mn aldra h . . . ibyath anha a rh h.
8.11.2
Translation
107. (R) Do you also have hand mills? (S) What? (R) Or what do you
say? rh iy or rh h? (S) hand mill . . . The hand mill which grinds the
wheat . . . this is called the stone hand mill . . . 108. (R) Yes. And it is the
woman who does the grinding? (S) Yes, the woman does the grinding,
and there are men who grind (as well). 109. (R) Yes, what do they do?
(S) They do, there are people who grind . . . the sorghum. This sorghum
from which they make . . . they make libbah89 from it. 110. (R) What is
it called? ildra h? (S) From this corn they make the libbah. This libbah
they make it form the corn . . . they grind it on a hand mill.
86
Interestingly, saddah here means the area covered by water, rather than the barrier blocking the water or dam itself.
87
ismha (with Cairene type of stress) instead of simha.
88
In group I dialects the form is more typically dra () (or with velarization marked in
the interdental, dr a()) and with article ddr a , see De Jong (2000, 82). For a remark on d
as a reflex for *d, see De Jong (2000, 332333).
89
Stewart (1990, 245, glossary) lists libbih as a small round of bread. A libbah, often
in other dialects referred to as gurs, is what men typically prepare themselves when they
are travelling, and women are not present. It is baked in glowing embers in clean sand.
592
8.12.1
rudolf de jong
Transcript
111. an l makn wala h jah byath an . . . w jjimal kamn, aza induh
jiml walla . . . jamaln walla burn, 112. aza m f h urmah fdy ih...
int ibtath an . . . biddi90 . . . biddi. biddi . . . yaniy aldra h . . . itxallha
mi dagg . . . itxallha majrih jari. 113. hda l alburn . . . l azzamil . . .
itain.91 hda kull ym . . . lzim tujuru lhin . . . imn aldra h diy. 114.
algamih hda . . . hda byath aninnuh duriy a rh h . . . dagg . . . dagg hda
zuwwdeh. 115. itdibb almazwad . . . alh l hda . . . addagg imn algamih .
aldra h hdiy . . . l alburn biyjirha. 116. [kull ym] hda lh l ulitta
btath an algamih , w ibtath an aldra h, iw kulluh. wi . . . w attibin hda . . . l
ijjml a tl. m biyath anh walla h jih bass ibyidirsh daris . . . ibydirsuw lgamih minnuh . . . iw biywaddh a lj[l] . . . 117. (R) iw byidirsh b
h? fh lh ah kidiy walla . . . (S) aywah . . . (R) ismuh h? (S) fh ns ibydirsuh bi l . . . b jjimal . . . biysawwiy luh madrs, ibydirsuh. 118. iw fh ns
halh n ibyidirsh b lmakan. (R) rif kilmit ilhjal?92 (S) alhjal . . . (R)
bitgl yaniy alhjal? (S) aywah . . . [. . .] alhjal hda biysawwuw minnuh,
ibydirsuw minnuh b jjimal. iw ar insn rif iysawwh . . .
8.12.2 Translation
111. Because there were no machines or anything, he grinds . . . and also
the camels, if he (a man) has a camel or . . . two camels or (more) camels. 112. If there is not a woman free (to do it) for them . . . you grind . . .,
you grind coarsely, you grind coarsely. You grind coarsely . . . that is, the
sorghum . . . you do not turn it into flour . . . but you turn it into coarsely
ground corn. 113. This is for the camels . . . for the camels . . . to feed them
(at dinner time). This (you do) every day . . . you have to coarsely grind
(some) of this corn for them (i.e. the camels). 114. This wheat . . . this they
(f.) grind normally on the hand mill . . . flour . . . flour, this is (for) provisions. 115. You fill the sack . . . (in) this case . . . with flour from wheat.
This corn is for the camels, they grind it coarsely . . . 116. [Every day] This
90
ddi, assimilated dd < td in tdi; da, ydi grind corn coarsely, see Stewart (1990,
211, glossary).
91
tain: an apocopated 2nd. p. m. sg. form (ta) of measure 2 verb aa, yaiy
feed dinner, with initial h- of the suffix -hin (camels are usually referred to in the f. pl.)
assimilated to the preceding voiceless consonant. The spelling here with triple is for
reasons of morphological transparency. The pronunciation is, however, not audibly different from doubled (IPA [( :)]).
92
hjal: a threshing board with sharp stones in its underside, on which the man
stands while it is being pulled by a camel.
593
is how it goes (lit. its situation), you grind the wheat, and you grind the
corn, and everything. And this straw . . . goes (lit. is) straight to the camels.
They dont grind it or anything, they only thresh it properly. They thresh
the wheat from it . . . and they put it aside . . . 117. (R) And with what do
they thresh? Is there like a (wooden) board, or . . . (S) Yes . . . (R) What is
it callled? (S) There are people who thresh it with . . . with camels. . . . hell
make (for himself) a threshing floor. 118. And there are people nowadays who thresh it with machines. (R) Do you know the word hjal? (S)
the hjal. . . . (R) Do you call it a hjal? (S) Yes . . . [. . .] This hjal (people)
do with it, they thresh with it with camels. And one has to know (how)
to do it . . .
CLASSICAL AND COLLOQUIAL ARABIC ARCHAISMS

Alan S. Kaye
California State University, Fullerton*
1. Introduction
This essay1 argues that modern spoken Arabic dialects sometimes
retain very archaic Semitic features. In fact, they may even preserve
Proto-Semitic forms that have been lost in Classical Arabicanother
indication that Classical Arabic is not to be regarded as their ancestral
proto-language (see Kaye 1976 for fuller explication of this idea). Rather,
there were other Arabic dialects spoken alongside Classical Arabic all
throughout history that served as the ancestral inputting ones to the
contemporary picture of Arabic dialects.
The orthographic representation of Hebrew l no as <l>2 may best be
interpreted as evidence for the reconstruction of Proto-Semitic *la no.
No matter what vowel preceded the glottal stop, I do not think any Semitist can deny that the Proto-Semitic word ended in a glottal stop.3 Final
* Shortly after Alans untimely death we received the proofs of his article. In deep
respect we insert his paper in this volume as it was at his passing on in appreciation for
what he contributed to our common field of interest (the editors).
1
This was originally presented to the joint session of the 216th meeting of the American Oriental Society and the 34th annual North American Conference on Afroasiatic
Linguistics on March 17th, 2006, in Seattle. I am thankful for the stimulating discussion
by the audience participants. I also wish to express my gratitude to Vit Bubenik, Gideon
Goldenberg, Wolfhart Heinrichs, Bob Hoberman, Jonathan Owens, Adrian Mcelaru,
Gary A. Rendsburg, Judith Rosenhouse, Aaron D. Rubin, Avi Shivtiel, Laurence J. (Tawfiq) Surfas, Rainer Voigt, and Bill Young for useful comments on a preliminary version.
2
According to Koehler and Baumgartner (1958, 466) and (1998, 466), this spelling
occurs 466 times in the Old Testament, while the plene spelling <lw> occurs only 35
times. Gary A. Rendsburg [p. c.] notes that the plene spelling occurs especially in Jeremiah, and that there are more instances of the interrogative and presentative /hall/
spelled plene than there are defectiva.
3
Let me comment on the notion of a pho(a)ethematic glottal stop, which occurs
in some Arabic dialects in final position; e.g., /mia/ he went. These are clearly innovations within Arabic (noted by the medieval Arab grammarians as well) and are paralleled by similar occurrences in Neo-Aramaic dialects (thanks to Bob Hoberman,
p. c.). Neo-Aramaic /la/h/ is paralleled by Israeli Hebrew /lo/ and, in many ways, by the
English voiceless unreleased bilabial stop in yep and nope.
596
alan s. kaye
graphemic aleph in Biblical Hebrew4 /q r / he read presents evidence

for an earlier */q r / or perhaps an even earlier */q ra/ < */qara/
(cf. /q r they read), which is equivalent to Classical Arabic /qaraa/.5
Similarly, the Hebrew form <l> points to an earlier *l < earlier */la/6
= colloquial Arabic la, la, or laa (Hinds and Badawi 1986, 775). The
Biblical Hebrew variant spellings <lh> (Koehler and Baumgartner 1998,
474) and <lw> (Koehler and Baumgartner 1998, 475) prove that the
glottal stop was no longer pronounced. Still, the ubiquitous spelling with
the glottal stop must be accounted for. Moroccan Arabic (so too Algerian Arabic) even has lla no and lla7 no way!; absolutely not! (Sobelman and Harrell 1963, 130 for the former alternant) demonstrating that
4
Hebrew refers to Biblical Hebrew unless otherwise noted.

The problem of stress is not a relevant consideration here and its discussion is thus
left unmarked. Let me take up the matter of the Hebrew final graphemic glottal stop.
Could this merely be an orthographic convention to distinguish two common words:
no, not and to him, originally spelled <lh> (= Classical Arabic <lh>), making it homographic to to her, and later spelled <lw>? The answer is clearly in the negative, since the
Proto-Semitic form contained a glottal stop.
6
Cf. Proto-Semitic ra head > Classical Arabic ras colloquial Arabic rs =
Hebrew r and Proto-Semitic *kas = colloquial Arabic ks = Hebrew ks (spelled with
aleph), although Gideon Goldenberg remarks that Anton Spitaler thought the glottal
stop in the last example was a hypercorrectiona theory that I find farfetched. Bob
Hoberman [p. c.] also notes the Hebrew verbs ymar he will say and ykal he will eat,
etc., that are spelled with aleph but where its phonetic realization as a glottal stop disappeared. Let me hasten to add that when it comes to a glottal stop, it is important to point
out that Hebrew /yir/ he will see retains it, whereas Classical Arabic /yar/ he will
see does not. Cf. Israeli Hebrew /lo/ absolutely not!. Although it is conceivable that the
glottal stop might go back to Proto-Hebrew times, it is much more likely that the latter
form is evidence of history repeating itself. Although the final glottal stop in this word
may originally be due to sound symbolism (cf. English yep and nope that are marked for
informal register, corresponding to more formal yes and no), the fact remains that it is
Proto-Semitic. Proto-Hebrew glottal stop in final position disappeared only to surface
again in Israeli Hebrew (see below) as a result of a linguistic cycle to give the root structure more than a mono-consonantal baseas a kind of Systemzwang. This is parallel to
Modern English ask < Old English aks, which is also once again the form (aks) in several
modern English dialects.
Concerning the Israeli Hebrew development, Judith Rosenhouse [p. c.] notes that it
seems to be a recent phenomenon and appears to be restricted to young adults, especially females, and seems to indicate abruptness or vehemence, particularly as an answer
to a yes-no question in order to stop someones nagging. When I suggested that this
might be a result of colloquial Arabic influence on Israeli Hebrew, she replied that this
was not possible, since it was used by native speakers of Hebrew who do not have direct
contact with Arabic speakers. Still, I pose the following question in rebuttal: does one
have to have direct contact to have influence? Bob Hoberman [p. c.] notes that the final
glottal stop in Israeli Hebrew is older than Rosenhouse thinks and it is not restricted to
just young adults and females, and is not mainly confined to nagging. Arabic full
investigation of this is, in my view, a desideratum.
7
The form with the geminated l actually gives a tri-consonantal body to this root: ll.
5
classical and colloquial arabic archaisms
597
Arabic dialects do not like to tolerate a mono-consonantal root. The first

cited Moroccan form at least points to a bi-consonantal or geminated
root with double /l/, although such roots are also not characteristic of
Semitic languages in that C1C1C2 roots are unknown.
Before continuing the discussion of the importance of the aleph in
the Hebrew word for no, not, it is necessary to point out that Classical
Arabic, although usually very conservative in matters of phonology, is
sometimes innovative. I will now discuss two significant innovations.
First, many contemporary spoken Arabic dialects preserve the ProtoSemitic imperfect vowel preformative (the so-called taltalah) */i/ rather
than */a/ = Classical Arabic /a/. Colloquial Arabic /yifqid/ corresponds
do Akkadian /ipqid/, Aramaic /nepqud/, Hebrew /yipqod/, Geez /
yfqad/, and Classical Arabic /yafqid/ (Proto-Semitic fqd = to seek).
Most Semitic languages and most modern Arabic dialects have /i/ or
a vowel clearly originating from it, as opposed to Classical Arabic /a/.
Although Moscati et al. (1964) postulate the vowel */a/ as Proto-Semitic,
they also remark (1964, 143): Some scholars, however, regard the vowel
i of the prefix as primary, alongside a. I believe there is much more evidence in favor of Proto-Semitic */i/ than */a/.
Second, the Classical Arabic voiced alveo-palatal affricate ( jm) is
clearly secondary deriving from Proto-Semitic */g/. Although I reconstruct a voiced lamino-palatal fricative *// for Proto-Colloquial Arabic,
the entire story of the development in Arabic dialects is not germane to
this investigation (see Kaye 1972 for the details).
Looking at the cognate Semitic languages, Ugaritic l (variant spelling l,8 syllabically spelled la-a indicating l) does not support the colloquial Arabic archaic forms with glottal stop, but matches up exactly
with Akkadian l and Classical Arabic l. Aramaic, Old South Arabian
and Geez lend further support to the earlier glottal stop with their graphemic alephs, although the Geez all is problematic from a comparative Semitic point of view. (Hebrew does have a negative al for negative
imperatives with cognates in Ugaritic, Phoenician, Biblical Aramaic,
Geez, and possibly Akkadian ul ).9 Thus, any notion that colloquial Arabic la() might be secondary is, I believe, an erroneous perspective due
8
The Ugaritic spelling with a glottal stop is erroneous, although Koehler and Baumgartner (1998, 466) cite it. Gideon Goldenberg [p. c.] informs me that the spelling with
aleph was originally postulated by G.R. Driver; however, others thought that this was a
verb meaning to be weakened or gleaming.
9
See Koehler and Baumgartner (1998, 45).
598
alan s. kaye
to the graphemic evidence from the ancient Semitic languages. Rather,

Classical Arabic l (which might represent /la/ since the grapheme for
the hamza is latesee further below) should be viewed as the later development < Proto-Semitic */la/. When the final glottal stop disappeared in
this Classical Arabic word, the previous vowel was lengthened as a result
of compensatory lengthening. It should be noted that this form without
the final glottal stop breaks with root bi-consonantalism or tri-consonantalism in favor of mono-consonantalisma very rare, if not impossible, situation in the classical and modern Semitic languages, especially
for words (prepositions and other function words, such as b- in or lto; for, are mono-consonantal, although the latter mono-consonantal
form is certainly related to the bi-consonantal root l in Hebrew and the
tri-consonantal ly in Classical Arabic (Koehler and Baumgartner 1998,
48).10 Moreover, it is not impossible that <l> in the very early stages of
Classical Arabic could have represented /la/, but that the hamza was not
written on top of the alif.11 This spelling represents the scripta defectiva,
much in the way the hamza of Modern Standard Arabic is often omitted in /uktub/ write! (masc. sg.) as it should be in /waktub/ and write!
(masc. sg.). This hamzatu lwasl should, strictly speaking, be written in
word-initial position, at least in some varieties of Modern Standard
Arabic.12 I can only speculate that the Akkadian graphemes might also
represent an earlier pronunciation in a pre-Akkadian dialect or in early
Akkadian with a final glottal stop before the general loss (in East Semitic
as a whole) of the pharyngeals and laryngeals.
Since one is confronted with the fact that Proto-Hebrew *-a > (as
in /q r / he read discussed above), one may be forced to hypothesize
that the Proto-Semitic word for no has a long , thus yielding a long
via the Canaanite vowel shift. It is important to point out that *qara
he read (previously discussed) does not yield qar or the like; thus the
inputting original forms had to be different because of the very different
10
There are other prepositions in the Semitic languages which clearly have bi-consonantal roots, such as Hebrew m- from < *min (Koehler and Baumgartner 1998, 535)
usually assimilated to the former form, and Hebrew al on < ly.
11
Jonathan Owens notes [p. c.] that the hamza was a post-hoc addition to Arabic
orthography, which is why the orthographic rules relating to it are so complicated.
12
Gideon Goldenberg [p. c.] comments that in early Modern Standard Arabic (e.g.,
Blq Press, Cairo), fully vocalized Modern Standard Arabic was printed with the hamza
sign on the alif only when the glottal stop is phonemic, such as in urdu I want. He
maintains: The rule to write the hamza where the glottal stop is a positional alternant
was adopted by the Egyptian Ministry of Education, then became rather popular.
599
outputs of the development. Bob Hoberman [p. c.] suggests that /a/ >
// was only in closed syllables where the // was the syllable coda. The
relative chronologies of the various sound changes are also important
factors to consider affecting vowel quantity.
2 Biblical Hebrew13 and Comparative Semitic Evidence for Proto-Hebrew 14
*// in Final Position Proving Proto-Hebrew */l/ < Proto-Semitic */la/
no
1. /b /15
enter; come
Classical Arabic /ba/, Akkadian /bu/, Ugaritic */bw/,
Phoenician /b/, Old South Arabian /bh/ (Koehler and
Baumgartner 111)
2. /b r / create
Old South Arabian /br/ build (Koehler and Baumgartner
14647)
Classical Arabic /baraa/ is not in Koehler and Baumgartner (1998).
Wehr (1974, 49) lists it (create).
3. /gb/ ditch
= Akkadian /gubbu/ < */gubu/ (Koehler and Baumgartner 163)
4. dr
/der n/ abhorrence
= Classical Arabic /daraa/ repel (evil) (Koehler and
Baumgartner 217)
5. /d/ grass
Akkadian /diu/ spring, Biblical Aramaic /dit/, Syriac
/t/ (according to Carl Brockelmann via Rainer Voigt
[p. c.], /te/ ~ /ta/), Classical Arabic /adiya/ be moist,
Old South Arabian /d/ season of grass; spring (Koehler
and Baumgartner 220)
13
The Biblical Hebrew data in Section 2.0 of this essay have been taken from Koehler
and Baumgartner (1998), an updated and revised version of Koehler and Baumgartner
(1958). These 40 roots offer the evidence from other Semitic languages of roots ending
in aleph, corresponding to Proto-Semitic */la/ no, not.
14
It makes little difference whether this happened in Proto-Hebrew or a pre-Hebrew
dialect.
15
It is important to point out that the aleph is pronounced in various conjugated
forms of tertiae-aleph roots; e.g., /b / they entered, /y b/ they will enter.
600
alan s. kaye
6. /h/
7.
8.
9.
10.
11.
12.
13.
he
Classical Arabic /huwa/ but Bedouin Arabic /h/. Cf.
the // in Geez /wt/ (Koehler and Baumgartner
226). /h/
she works the same as /h/. Cf. Geez /yt/. Koehler and Baumgartner (1998) do not cite Bedouin Arabic
*/h/, but it probably exists (Koehler and Baumgartner 226). Koehler and Baumgartner (226) reconstruct
Proto-Ethiopian Semitic */hatu/ he and */hati/
she. Gideon Goldenberg [p. c.] notes that Qumranic
Hebrew /ha/ he and /ha/ she parallel the Classical Arabic forms with /-a/.
hb
/nhb / hide oneself
Classical Arabic /xabaa/ hide (Koehler and Baumgartner 270)
/h g /
shame; confusion
Classical Arabic /xajaa/ look ashamed (Koehler and
Baumgartner 275)
/h t /
miss the mark
Classical Arabic /xati a/ make a mistake; sin Wehr
(1974, 245)
Koehler and Baumgartner (1998) cite Classical Arabic
/xati ya/. Ugaritic /xt /, Biblical Aramaic /xt /, Old
South Arabian /xt/, Geez xt (Koehler and Baumgartner 288)
fall ill
/h l /
Classical Arabic /xalaa/, Old South Arabian /hl/ fall
ill (Koehler and Baumgartner 298). Gary A. Rendsburg [p. c.] notes that this root is a by-form of /hly/
become weak (Koehler and Baumgartner 300) =
Middle Hebrew and Jewish Aramaic /hl/.
/hm (h)/ sweat; butter
Ugaritic /xmt/; Jewish Aramaic /xm ()/ (Koehler and Baumgartner 308)
/y re/
fear
Ugaritic /yr/ (Koehler and Baumgartner 399); Avi
Shivtiel [p. c.] adds Classical Arabic waraa to repel,
listed in Lane Vol. 8 (1893, 2933).
/k l /
restrain; keep from
14. /kisse/
15. /ks/
16. /l bo/
17. /y s /
18. /l b/
19. /m le/
20. /m s /
601
Classical Arabic /kalaa/, Egyptian Aramaic, Jewish Aramaic, Syriac, Geez kl (Koehler and Baumgartner
436)
seat
Akkadian /kuss/, Ugaritic /ks/, Phoenician /ks/,
Aramaic /k rse/, Syriac /kursy/ with plene written
/u/, Classical Arabic /kurs/ < */kursi/ (Koehler and
Baumgartner 446)
headdress of the moongod at the time of the full moon;
full moon
Akkadian /kusu/, Phoenician /ks/, Syriac /ks/ ~
/kas/ (Koehler and Baumgartner 446)
East of Jordan (Gary A. Rendsburg [p. c.] translates
entrance; on the way to.)
Assyrian /labu/, Egyptian /rbw/ (Koehler and Baumgartner 470)
go out
Phoenician, Ugaritic /ys/, Old South Arabian /wz/,
Geez ws ~ wd (Koehler and Baumgartner 393)
lion
Akkadian */labu/ (Koehler and Baumgartner 472 state
that they follow Benno Landsberger), Ugaritic /lbt/, Old
South Arabian /lb(t)/. Avi Shivtiel [p. c.] adds Classical
Arabic /labwa/ lioness, not in Koehler and Baumgartner,
but listed in Wehr (1974, 857). Vit Bubenik [p. c.] alerts
me to the alternant /labua/ lioness (Madina 1973, 591).
The Proto-Semitic root is lb. The form with the semivowel is secondary.
be full
Classical Arabic /malaa/ fill; /malia/ be full, Old Aramaic, Biblical Aramaic, Ugaritic /ml/, Akkadian /mal/,
(Koehler and Baumgartner 523); Geez /mala/, not in
Koehler and Baumgartner, added courtesy of Aaron D.
Rubin, [p. c.].
meet; find
Classical Arabic /ant/ < /amt/, Jewish Aramaic /ms /,
Biblical Aramaic /mt /, Syriac /mt/ ~ /mtay/, Old
South Arabian /mz/, Geez ms (Koehler and Baumgartner 553)
602
alan s. kaye
21. /m r /
feed
Classical Arabic /maria/ agree with (of food), Ugaritic /
mr/ (Koehler and Baumgartner 563)
22. /mass / Masu
= /masu/ name of a North Arabian tribe (Koehler and
Baumgartner 570)
23. /n /
half done
Classical Arabic /na/ be raw, uncooked (especially of
meat) (Wehr 1974, 1014) (Koehler and Baumgartner 570)
24. nb
denominative of /n / prophet
Classical Arabic /nuba/ prophecy, /tanabbaa/ (Form
V) to prophecy (Wehr 1974, 937) (Koehler and Baumgartner 586)
25. nd
drive (cattle); detach, remove from
Geez nd drive cattle
26. /n /
lift up
Classical Arabic /naaa/, Phoenician, Moabite, Ugaritic
/n/, Old South Arabian /n/, Geez ns (Koehler and Baumgartner 638) Biblical Hebrew /nm/
clouds; damp; fog, Classical Arabic /na/ hovering
clouds (Koehler and Baumgartner 638)
27. /n /
lay claim; lend
Middle Hebrew ny, Classical Arabic /nasaa/ grant
credit (Wehr 1974, 959) (Koehler and Baumgartner
638)
28. /s l (h)/ participle /msull m/ be paid pointing to sl
Classical Arabic /salaa/ pay promptly (Koehler and
Baumgartner 658)
29. /pl/
miracle
Classical Arabic /fal/ good omen (Koehler and Baumgartner 759)
30. /pr/
zebra16
Classical Araic /fara/ wild ass; onager (Wehr 1974, 701).
The Akkadian cognate is purmu. (Koehler and Baumgartner 775). Vit Bubenik [p. c.] states that Akkadian
par mule is closer phonologically than is purmu.
16
Although Koehler and Baumgartner (1998, 775) translate zebra, this is erroneous
for onager (thanks to Wolfhart Heinrichs and Gary A. Rendsburg for this correction).

31. /s b /
32. /s me/
33. qy
34. qn
35. /q r /
36. /r p /
37. / ne/
603
army
Akkadian /sabu/, Old South Arabian /db/, Geez /
sb/. Loanword in Egyptian /dbi/ (Koehler and Baumgartner 790)
be thirsty
Classical Arabic /zamia/ (Wehr 1974, 583)
(Koehler and Baumgartner /zamiya/ seem to be erroneous.)
Ugaritic /zm/, Akkadian /samu/, Old South Arabian
/zm/, Geez sm (Koehler and Baumgartner 806)
/q (h)/, impf. /y q/ vomit
Classical Arabic /qa/, Geez qy vomit, Akkadian
/qu/ [sic] for /qu/ excrete (Egyptian q vomit)
(Koehler and Baumgartner 83637) (also Koehler and
Baumgartner 1958, 83637)
/qinne/ be envious of
Classical Arabic /qanaa/ become intensely red, /qni/
blood-red, deep-red (Wehr 1974, 791) (Koehler and
Baumgartner 84243)
read
Classical Arabic /qaraa/ read, Ugaritic /qr/ read
Koehler and Baumgartner (851) relate Classical Arabic
read with /qr/ II encounter, befall and not with /qr/
I call, recite > read (Koehler and Baumgartner 849).
In my opinion, this view is erroneous. Their theory is
also expressed in Koehler and Baumgartner (1958,
849). They further note that Biblical Hebrew qry = /
q r (h)/ encounter (Koehler and Baumgartner 1958,
853) = Classical Arabic /qaraa/ to go about (not in
Wehr 1974) and Classical Arabic /qar/ receive as guest
(Wehr /qar, yaqr/ receive hospitably; entertain (1974,
761)
heal
Classical Arabic /rafaa/ mend; darn, Old South Arabian /rf/ repair, Geez rf stitch together (Koehler
and Baumgartner 903)
hate
Classical Arabic /anaa/ hate (Wehr 1974, 487)
604
alan s. kaye
(Koehler and Baumgartner /aniya/ for Classical Arabic

appears to be erroneous). Ugaritic /n/, Old South Arabian /n/ (Koehler and Baumgartner 925)
38. /b / Sheba
Classical Arabic /saba/, Sheba Akkadian /sabe/ mankind (?), Old South Arabian /sb/ Sheba (Koehler and
Baumgartner 939)
39. / w/ a bad thing
Classical Arabic /sa/ be evil; bad (Koehler and Baumgartner 951)
40. /t /
room
Akkadian /tu/ room; guard chamber (Koehler and
Baumgartner 1015)
3. Conclusion
The basic thesis of this work is to affirm that Classical Arabic should
stop being referred to as the preserver of all, or nearly all, original ProtoSemitic phenomena. This paper suggests that certain Classical Arabic
forms are indeed secondary, as e.g., /l/ no discussed above. Consider
also Classical Arabic qalb heart < * lbb heart. The latter root is also fully
preserved in Classical Arabic lubb heart = Hebrew leb and Akkadian
libbu < Proto-Semitic *libbu (= Proto Afro-Asiatic */lib-/ ~ */lub-/ (Orel
and Stolbova 1995, 362). Moreover, Egyptian ib supports Proto-Semitic
*libbu, while the vocalism in Classical Arabic lubb can be explained as
the result of regressive labial assimilation. Thus, Classical Arabic q- in
qalb is a remnant of a prefix of some kind or root determinative17 used
in an old Arabic dialect or an earlier Semitic language antedating Classical Arabic (cf. Classical Arabic qadima or qadama to gnaw; compress
the lips (Wehr 1974, 544 < damma bring together further suggestive of
q- as a root determinative of some kind). Additional comparisons along
the lines of the present investigation of Classical Arabic and colloquial
Arabic dialects with other Semitic languages will undoubtedly further
17
Harsusi /helbb/ and Mehri /hewbb/ have /h-/ before the root /lb/ heart, which is
apparently lost in Soqotri /elbeb/ and Sheri /b/. Mokilko (East Chadic) /ulbo/ heart
looks as though it displays */q/ > //, as occurs in many eastern Arabic dialects.
605
demonstrate that spoken Arabic dialects occasionally preserve more

original forms than does Classical Arabic.
4. References
Hinds, Martin and El-Said Badawi. 1986. Dictionary of Egyptian Arabic: Arabic-English.
Beirut: Librairie du Liban.
Kaye, Alan S. 1972. Arabic /iim/: Arabic Synchronic and Diachronic Study. Linguistics
79, pp. 3172.
. 1976. Chadian and Sudanese Colloquial Arabic in the Light of Comparative Arabic
Dialectology. The Hague: Mouton and Co.
Koehler, Ludwig and Walter Baumgartner. 1958. Lexicon in Veteris Testamenti Libros.
Leiden: E.J. Brill. (Abbreviated Koehler and Baumgartner 1958.)
. 1998. A Bilingual Dictionary of the Hebrew and Aramaic Old Testament. Leiden:
Brill. (Abbreviated Koehler and Baumgartner 1998.)
Lane, Edward William. 1893. An Arabic-English Lexicon. London: Williams and Norgate. (Reprinted 1968, Beirut: Librairie du Liban.)
Madina, Maan Z. 1973. Arabic-English Dictionary of the Modern Literary Language. New
York: Pocket Books.
Moscati, Sabatino, Anton Spitaler, Edward Ullendorff, Wolfram von Soden, eds. 1964.
An Introduction to the Comparative Grammar of the Semitic Languages (Porta Linguarum Orientalism). Wiesbaden: Otto Harrassowitz.
Orel, Vladimir E. and Olga V. Stolbova. 1995. Hamito-Semitic Etymological Dictionary.
Leiden: E.J. Brill.
Sobelman, Harvey and Richard S. Harrell, eds. 1963. Arabic Dictionary of Moroccan
Arabic. Washington, D.C.: Georgetown University Press.
Wehr, Hans. 1974. A Dictionary of Modern Written Arabic. 3rd ed. by J. Milton Cowan.
Ithaca, N.Y.: Spoken Language Services.
DO THEY SPEAK THE SAME LANGUAGE? LANGUAGE USE

IN JUBA LOCAL COURTS
Catherine Miller
IREMAM-CNRS of Aix en Provence, France
1. Introduction
1.1 Language Description and Language Uses: the unavoidable
discrepency
K. Versteeghs book (1984) on Pidginization and Creolization in Arabic
was one of the first attempts to link two previously separated domains,
Arabic studies and Pidgin/Creole studies. The controversy that followed
publication of his book contributed to diffusion of information about the
main contemporary Arabic based Pidgin-Creole varieties of the South
Sudanese basin, namely Ki-Nubi and Juba Arabic (and former Turku).
No serious Arabicist can now ignore the three above mentioned names,
and a number of publications have been devoted to the description of
these varieties. To recall but a few: B. Heine (1982), X. Luffin (2005),
U. Mahmud (1979), C. Miller (1984, 1992), J. Owens (1991, 1997),
M. Tosco (1993, 1995), R. Watson (1984), I. Wellens (2005), E. Yokwe
(1995).
Most of these studies provide a systemic description of Ki-Nubi or
Juba Arabic, and compare them with other Pidgin-Creole languages,
with other Arabic vernaculars, or with local African languages in order
to establish their specificity and autonomy and to postulate hypotheses about their genesis and development. Although still considered as
the orphan of the orphans (Owens 2001), marginalized in both the
Creole Studies and the Arabic studies, Juba Arabic and Ki-Nubi start
to be rather well described, even if many more need to be done. One of
the impacts of all these studies is the acknowledgement that Ki-Nubi
and Juba-Arabic are indeed autonomous and specific varieties that cannot be confused with any other Arabic vernaculars. They are identified
through their specific name; as it is well known, to name something is
to provide it with a specific identification.
An interesting development of this linguistic achievement is the fact
that, since Ki-Nubi and Juba Arabic have been identified as specific
608
catherine miller
languages, the local speakers (Nubi people from Uganda or Kenya and
all Sudanese Southerners) are expected to speak what the linguists have
described as Juba Arabic and Ki-Nubi. This is what is happening for
example in the case of language tests applied to asylum seekers in a
number of European countries. If somebody claims to be a Southern
Sudanese but speaks an Arabic variety close to Northern Sudanese colloquial varieties instead of the Juba Arabic described by the linguists,
he cannot be a true Southerner. Asked to provide some counter-expertises, I realized that the way of speaking of these doubtful Southerners
was sometimes rather similar to some cases I recorded long ago in some
local courts of Juba, with well-attested Southerners!
In many contexts, most Sudanese Southerners constantly shift from
a more Creole level of Juba Arabic (or basilectal level) to a level more
influenced by Northern Sudanese Colloquial Arabic (mesolectal level).
The coexistence of different levels of Juba Arabic was first highlighted by
U. Mahmud (1979), who applied the Creolistic concepts of continuum
and basilectal and mesolectal varieties to describe the variation found
in the verbal system. I dedicated a number of papers to emphasize the
degree of variation and diversity included under the generic label of
Juba Arabic (Miller 1984, 1987, 1989).
In the continuum approach, each pole (basilectal/mesolectal) is characterized by a number of specific features (isoglosses), while the intermediate levels of the continuum are characterized by different degrees
of occurrence and melting of the various features. It appears, however,
that it is extremely difficult to draw an implicational scale that will be
characterized by a regular acquisition of dialectal features along the
continuum scale. In fact, each speaker tends to have its own way of mixing the various features, some focusing more on phonological features,
others on lexicon or morphological features.
I myself, insisted on the fact that the evolution of Juba Arabic was far
from linear, and was not automatically leading to a process of decreolization. Different trends of change and restructuration were simultaneously
recorded in a city like Juba and different influences were operating on
the daily language; one leading towards a rapprochement to Khartoum
Arabic, the other toward what I have called a process of vernacularization (Miller 1987).
This diversity and variability of Juba-Arabic raise the following questions: where are the boundaries of Juba Arabic? On which criteria can
we decide that a person speaks or does not speak Juba Arabic? Are the
language use in juba local courts
609
speakers conscious that they are mixing different features and variables
in their speech? Do the speakers and auditors have the same perception
than an outsider linguist?
1.2
Language Uses in Juba Local Courts
In order to illustrate the diversity prevailing under the generic label of

Juba-Arabic, I choose to present complete samples of interaction, instead
of short extracts given as examples. These samples were recorded in
1981 and 1984 in two local Courts of Juba, the capital-city of Southern
Sudan. Local courts were of particular interest because very different
kinds of people were coming to present and defend their cases. Taking
into account that more than twenty years have passed since the mid
1980s, and that almost nothing is known about the present linguistic
reality of Juba, such samples could be considered as a kind of linguistic
testimony of what speaking Juba Arabic meant in the 1980s.
The term local court and its Arabic equivalent makma ahliyya ~
makama ahliya was applied to Southern Sudanese A and B Courts at the
levels of the village, the district or the city. Various languages were used
in these local Courts.
Outside Juba, the A Court was the villages Court headed by the
villages chief, assisted by two or three elders. The language of communication was, most of the time, the local African vernacular (Bari, Kakwa,
Zande, Moro, etc.). The B Court was the district Court headed by the
paramount Chief assisted by the village chiefs. It dealt with matters not
solved in the A Court. In multilingual districts of Equatoria, such as Yei
B Court, the languages spoken were local vernaculars (Kakwa, Moro,
Madi, Avokaya), Juba Arabic, Bangala (the local variety of Lingala) and
a few English.
In Juba, the only A court was the Garawiyya, located in the center
of Juba, near the main market, and dealing with low-level criminal cases
(robbery, neighbors quarrels, etc.) involving any person living in Juba,
whatever his ethnic affiliation and period of stay. The court was headed
by an appointed local judge (Ramadan, 49 y. old) and two assistants. The
main language of communication was Arabic (Juba Arabic up to Northern Sudanese Arabic (NSA). A few speakers used an African vernacular
and were assisted by a translator. The summary of the case was always
pronounced in Arabic but written in English.
The Kator B court was the Bari local court (the Bari being the main
local group of the Juba-Rejaf area), located in Kator district. It was
610
catherine miller
dealing with personal status cases involving at least one Bari participant.
The President of Kator B court was a Bari (Stephen, a retired policeman), assisted by two or three Bari elders. The language of communication was mainly Juba Arabic, Bari and a few English.
The localization and specialization of each court influenced the
language use. In Juba, the Garawiyya court was far more exposed to
the influence of NSA than the Kator B court. In the Garawiyya, only 4
speakers, not resident in Juba, spoke in their native vernaculars, while
35 speakers spoke only in Arabic (with 16 speaking a more mesolectal
variety and five speaking NSA). In Kator, 10 speakers spoke mainly in
Bari against 24 in Arabic (with none speaking NSA and 14 more or less
mesolectal):
FIGURE 1
Courts
LANGUAGE USES IN KATOR AND GARAWIYYA

LOCAL COURTS
Speakers* Non-AR Vernaculars Juba Arabic
Kator B
Garawiyya
34
39
Total
73
10 Bari
3 Bari
1 Dinka
14
NSA
English
24
30
54
* Number of speakers does not include the Judges
The above figure classifies each speaker according to the use of one
dominant language (English, NSA, JA and Non Arabic Vernaculars). In
fact, the delimitation between each language was not that clear-cut and
there were many cases of language mixing, as will be evident in the Corpus presented in 5. Appendices, the texts: 5.1, 5.2, and 5.3.
2. Presentation and Brief Analyse of the Corpus

The samples presented in the Appendix are taken from three cases, considered to be representative of three types of speaker. The participants
involved in Case 1, belong to the important Nilotic Dinka-speaking
group, a non-local Juba ethnic group. Not well-educated, they speak
Juba-Arabic as a second language. In Case 2, the participants belong to
a small ethnic group from Western Southern Sudan (Bay from Bahr al
Ghazal Region), are educated, work in the police or the army, and speak
a level of Arabic very much influenced by NSA and WSA (Western
611
Sudanic Arabic). In Case 3, speakers are all Bari Juba citizens and speak
Bari or a vernacularized Juba Arabic, very much influenced by Bari
features.
In both courts, the procedure was rather similar. The case starts with
an oath (on the Bible, the Qurn, or a spear according to the participants religion),see (s.9) in Case 1 and (s.14) in Case 2. Then the plaintiff tells his/her story; after some questions, the judge summarizes the
story and asks the defendant to present his/her own version. Additional
questions follow; sometimes, additional testimonies are required. After
listening to all parties, the Court deliberates and announces its verdict.
While most cases took between 30 to 40 minutes in Juba, they could
take hours in the villages. Cases tended to be shorter in the Garawiyya
than in Kator.
Speakers are presented as follow: J.= Judge S. = Courts Secretary. T. =
Translator, P. = Plaintiff, D. = Defendant and W. = Witness. Elements
in italic are considered to be average basilectal Juba Arabic. Elements
in bold are considered to be more mesolectal. Underlined elements
are borrowings from the local languages, mainly Bari. Plain words are
English borrowings. Many elements are common to both basilectal and
mesolectal Juba Arabic. They have been put in both italic and bold, but
the distinction between more basilectal or more mesolectal remains
very fluid. The transcription of the Bari sentences have been done in
1987 by a Bari student, who followed the standard Bari written system
used in Equatoria.
Mesolectal features taken into consideration include a) phonological features (consonants , x, , t, d, s, , h, long vowels and geminates),
b) morphological features (object suffix pronouns, definite article, plural and gender affixes, TMA and person verbal affixes) c) lexical items.
It must be noted that the insertion of mesolectal features lead often to
mix forms, which are neither Northern Sudanese Arabic (NSA), nor
Creole Juba-Arabic. African vernacular features taken into consideration include:
a) phonological features (p, , , , , ", , j strong palatalization and
affrication);
b) lexical items and idiomatic expressions. Syntactic features will not be
discussed here.
The brief analysis that follows intends to higlight the individual level as
well as main examples of mixing with focus on mesolectal insertion. But
612
catherine miller
readers are invited to go to Appendix 1, 2 and 3 in order to evaluate the

degree of mixing within each individual speech.
2.1 Text 1, The Dinka Case, Garawiyya Juba, 1981. (Transcript in
Appendix 1)
Summary: A Dinka woman coming from a village accuses a Dinka man
living in Juba to have rob her a sac of sorghum when she came down
from the ferry. She speaks only Dinka and her speech is translated by
another Dinka from Juba. The case started the day before but was postponed in order to look for an additional witness who could prove the
identity of the defendant. The woman affirms that he is called Maluk
while the defendant claims to be called Deng. At the beginning of the
second audience, the secretary summarizes the facts (s.1) and then the
Judge is looking for the witness (s.2 to s.9) who finally arrives and supports the plaintiff against the defendant that he knows since 1975 (s.10
to s.20); the later is finally recognized culprit by the Court (s.21).
This case illustrates a type of language use rather common in the
Garawiyya Court, that I will define as a formal basilectal non vernacularized type of Juba-Arabic. This type of speech is very common among non
L1 speakers of JA (rural people, older generation) in formal situations
and lead to many mix forms. Although it contains most of JA basilectal
features (such as verbal invariable forms), it is characterized by numerous instances of phonological variation (s/; k/x; j/z; a/, etc.) and by
the frequent dropping of the verbs final epenthetic vowel (kelem vs. JA
kelemu speak, gal vs. JA gale say, baaref vs. JA bi-arifu know, etc.). A
number of lexical items are taken from NSA like der want (NSA dyir/
JA aoju), bet house (NSA bt/JA jua), particle aw or (JA wela), etc.
Some old terms inherited from the Ottoman military language remain
like dosoman troubles, fight (cf. s.8) and appear to be part of the court
vocabulary. One notes the irregular insertion of a number of mesolectal
morphological features borrowed from NSA such as definite article al(l/el) the, verbal suffixes and prefixes as well as affix pronouns. In Case
1, all speakers speak this formal non vernacularized JA, with various
degree of mesolectal insertions.
The two Dinka participants (the translator and the witness) use independent personal pronouns and invariable verbal stems like Translator
(T.) in sentence (s.3) nina m dr mi we dont want to go or sentence
613
(s.5) ya na klem le ede I talked to him like this. The pronunciation shows the influence of the Dinka Mother Tongue, particularly concerning the realization of the phonemes // /s/ and /z/ realized as [] (a
frequent pronunciation among Nilotic speakers) like in en what (JA
senu /NSA en); ol somebody (NSA zl/ JA jol), etc. Vernacular influence is also noticeable in the realization of the vowels, with a tendency
to realize /e/ like an open [] and /o/ like an open [ ]: k (s.3), le (s.5),
bls (s.12), barf (s.13), nugasr (s.12), b and ka (s.16), etc.
The Dinka witness (W) realizes a number of inflected verbal forms
and affix pronouns:
(s.12)
nugasr (JA ana bigeseru/NSA agaar)to limit, shorten, (s.20) axk

(JA aku btak) your brother, gutta lek (JA ana gale le-ita) I told you,
negabel (JA. ana bigabalu/NSA agbil) I meet, nesalem (JA ana biwodi
salam to/NSA asallam) I great, neddu (JA ana wodi leo/NSA nadd) I
gave him). It may be noted that 1st sg. imperfective form is n- + verb
(similar to WSA) and not a- + verb (= NSA).
The secretary (S) has a JA pronunciation: (s.1) asrin twenty (NSA

arn), asan because (NSA an), jl person (NSA zl) but izebu he
brings (JA jibu/ NSA ijib).
The Judge displays many instances of phonological variation: z/j (s.2)
win zl el-brefu jl da (with dominance of [j] cf. (s.4); x/k (s.6) axu
brother, (s.15) kemsa five, (s.17) inta bikutu you put; /s (s.4) aan/
asan because, (s.17) musu isnt it.
He keeps many invariable verbal forms: cf. (s.4) nina il we take, (s.8)
nina balasu we stop, (s.21) inta bigul sehi you say right, etc. and many
independent personal and possessive pronouns: (s.9) el-mahkma nadi
eta the Court calls you; (s.15) inta baarfo aglaK bitao you know his
behavior? (s.21) weled btak suker winu where is your small boy?
The mesolectal variables include the 2nd sg. personal pronoun inta
you (JA ita), pl. deictic del this (JA de), definite article l (JA de) and
a number of conjugated verbal forms: (s.9) twori you show, (s.21)
kizibta you lied, sarkta you stole, etc. The conjugation of verbs leads
to some cases of malapropism such as (s.9) nasaltak we ask you with
the coexistence of n- 1st pers. imperfective and -t 1st or 2nd pers. perfective. Another mix form is the verb gul in (s.4) bas iji wri gl he
just come to tell that (JA bas bija wori gale/ NSA iji igl or iji iworri
innu). The insertion of mesolectal features increases at the end of the
case (s.22), when the Judge is enouncing his verdict and is using a number of technical terms such as tazwir falsification, jiza faults/penalty.
614
catherine miller
Note however that the Judge always alternates basilectal and mesolectal
realizations like the alternation sarkta/ita sereG in (s.22).
A number of mix JA/NSA realizations recorded in Case 1, were almost
permanent in the Judges speech of the Garawiyya and appear also in
Case 2. They seem to indicate that, indeed, this type of mix level was
considered as the appropriate level in the Garawiyya Court.
2.2 Texte 2, The Bay Case, Garawiyya Juba, 1981. (Transcript in
Appendix 2)
Summary: a young Bay police-lady (Hawa) from Bahr al Ghazal accuses
two Bay guys to have tried to attack her the night before. She recalls
that she was going back to her home when the two young men started
to insult her and beat her. Fortunately she was able to call for help and
a soldier came to rescue her and, after many events, managed to bring
the two boys to the police office. The two defendants contest the ladys
version and claim to be victims of a confusion, i.e. the soldier took them
instead of the real culprit. The lady and the two boys agree that they
never met before, in spite of the fact that they all belong to the same
small tribe (Bay). The two Judges argue between them concerning the
importance of this common ethnic background. The soldier is called as
a witness and makes a very long testimony, which supports the policewomans story. After many discussions and controversies, the two boys
are recognized culprits and have to pay a fine of 36 Sudanese pounds.
This case illustrates a more mesolectal level, mastered by the persons
who have been educated in Arabic and particularly those coming from
the region of Bahr al-Ghazal. However, each speaker has his specificity,
and while the young lady is the most influenced by colloquial features,
the men alternate between basilectal and mesolectal features and create many mix forms. A number of lexemes appear to be shared by all
the participants such as der (JA auju) want, bet (JA jua) house, gul (JA
gale) say, fat (JA futu) pass, darab (JA dugu) heat, suf/uf (JA ainu) see,
le-raet-ma (JA lakadi) until, etc.. At the level of the verbal system, one
notes many occurrences of participial forms such as jay coming, gaad
staying, el working and the frequent use of TMA auxiliaries/verbal
particles like kan, bikun, gam/gum, gaad, lisa, biga/baga, ja, aoz.
Hawa (s.15 and s.17), the young policewoman speaks a level very close
to NSA or rather to Western Sudanese Arabic (WSA). She pronounces
many long vowels, as well as velar fricative /x/ and sibilant // but she
realizes few pharyngeal and emphatic consonants: i.e. (s.15) saa watch
615
(NSA saa), saba morning NSA ab), talta I came (NSA talat) but
end of (s.17) sa'a fekka the watch dropped. She conjugates almost all
verbal forms: cf. (s.15) jit, saketta, zitta, hum zdu and (s.17) gumta,
wageftu, gulta, tetfahim, netfahim, maet, etc. Object pronouns and
possessive pronouns are systematically suffixed: (s.15) garas-ni fi-sulb-i
he pinched me in my buttocks; (s.17) dagga-ni, axwn-i, masak-ni,
xam-ak, bedugg-k, etc. She uses a number of TMA auxiliaries, like
gam and ja to mark inchoative:
(s.15) ana gumta zitta katwa btay
(s.17) da tawli masakn ja garasni
waled de ja gam darabni bunya
I accelerated my walk
He immediately sized me and
start to beat me
The boys started to beat me
strongly.
Hawa masters the morphological rules of NSA and shows very few
occurrences of malapropism or mix form. Gender and plural agreement
however remains irregular: (s.15) ana jay I came-masc., ana mya
I went-fem.; (s.17) arabiya aba yagif the car (f.) refused (masc.) to
stop (masc.), awld der iyamal mayi mukel boys (pl.) want (sg.) hemakes with me problems/ the boys want to make me problems.
Defendant 1 (John Gabriel, s.28, s.30, s.34) starts answering both Judges
with mesolectal features:
(s.7)
J2 sgol btk ? Your work
D1: m el/ tleb Not

working/student
(s. 24) samtu I heard-it

(s.28) ana kunt jay min hay kmryal li-l-bT I was coming from district
Commercial to the house.
Then he alternates between few inflected verbal forms and many

invariable verbal forms as well as between affixed and independent
pronouns:
(s.30): ma blis gaadn isugg zl de / uman ft giddm/ ana ja wara/ nama
ana wosol mbil/
/he and police stay-pl they-drive person this/ they pass (sg.) in front/ I
came (inv.) behind/ when I arrive (inv.) Mobil/
He and the police officer were driving this person, they passed in front,
I was coming behind when I arrived at the Mobil Station.
(s. 38) bass darab bitoman ana ja katal /only fight of-them I came (inv.) kill
(inv.)/
I just came to stop their fight.
616
catherine miller
Defendant 2 (William Peter, s.45, s.47, s.51) realizes many inflected verbal forms and affix pronouns:
(s.52)
gabodni zamn gale nemi-l-nukta niuf/ min wisolta unak an

nera mukila bta dn kif
/they-size-me (3rd perf.) time say I-go (1st imp.) the post I-see (1st
imp)/from I-arrived (1st perf.) there for I-see (1st imp) problem of
John how/
They captured me when I went to the police station to see/since I
arrived there to assess the problem of John . . .
Like the Dinka witness in Case 1, William uses the WSA n- 1st sg. imperfect pers. pronoun instead of NSA a-. Note also the use of verb gale say
to introduce an embedded sentence (zamn gale time that).
The witness (W. the soldier) starts with a formal mesolectal level:
(s.75) ana kunta aal fi-l-jawazt, ana sken fi Muluk I was working in the passports, I lived in Muluk. But very quickly he alternates
between basilectal and mesolectal features: ana jit ja wosolu fi medresa
/ana ft/ fialan fatet.. I arrive near the school/ I passed/ of course I
passed; baadin ana jt ana gul when I arrived I say, ana gulta tayib
ya axwna matakum sakal bit fi tarka zede I said good oh my brothers, dont quarrel a girl in this way (note JA negative imperative marker
matakum dont!).
He ends up speaking almost only Juba-Arabic, including at the phonological level ( >s, x > k). Unlike the speakers of Case 1, his level of
Juba Arabic is more grammaticalized. It includes many verbal TMA
markers as well as embedded sentences:
(s.75bis) d tawli gum amol musakl fo mbili li raat ma aoz kaser mrat bta
arabiya zatu
He immediatly started making problems in the Mobil station until
he was almost breaking the mirror of the car
ana bija ft fi jawazt ya neselem kelem kabr li jamaa fi taakir el ana
akr line zaman
I was going to the passeport (office) in order to bring the news to
the people about the delay that I delayed before.
Compared to Case 1, Judge 1 (Ramadan) shows the same level of phonological variation (s/, x/k, z/j) but uses more lexical and morphological mesolectal features (see s.25, s.37, s.52, s.74, s.79). It may be noted
that in (s.25) when Judge1 is summarizing the testimony of the police-
617
woman, he makes more gender agreement than she did and employs the
2nd and 3rd fem. personal pronouns inti and iya:
(s. 25) el-bint de gl iya maya The girl said (masc.) that she was going
(fem.);
inti jya min wn you (fem.) coming (fem.) from where?;
saa btao fakkat waga watch (fem.) of-him drop (fem.) fall (mas.).
But he also mixes with more basilectal JA and consistently uses the
invariable form gul say:
(s. 25) intakum gul inti ermuta sakeT you-pl say you (fem.) prostitute only
(s. 52) nker u gl woket dak kalam inti gul ma hasal klu-klu
deny and say time-that word you (f.) say not happen all-all/
He denied and said (that) at that time what you say never happened.
As in Case 1, the conjugation of verbs and auxiliaries leads to cases of

malapropism or mix forms.:
(s.52)
asa nafarn de badi sumit kalamu /now person-Dual this after I-listened word-his/
now after I listened the word of these two persons (sumit vs. JA asuma
and NSA samit)
fi tarka gmo daribtu bi daraba adid /in way they-stand you-beat
with beat strong/ On the way you started to beat her strongly (gmo
vs. NSA gumtu).
Compared to Judge 1, Judge 2 (see s.66, s.70, s.76, s.78), sticks to a more
basilectal pronunciation ( > s, x > k, etc) and grammar:
(s. 66) musu barau rtan bta bay keda wonos /Neg alone language of Bay
like-this spoken/
Isnt it a specific Bay language which is spoken? (note the impersonal
structure rendered by shift of stress to the final syllable of wonosu).
Case 2 highlights the coexistence of different levels (types?) of (Juba)

Arabic and different degrees of mixing. The level of the police-lady and
the intermediate level of Judge 1 and D2 could be an indication that the
mesolectal level is indeed the prestigious urban formal level that people
speak or try to speak in formal context like the Court. This hypothesis is
however not sustained by the data from Kator B Court, which indicates
another type of urban use.
618
2.3
catherine miller
Texte 3, Kator B Court, Judge Stephen, 1984. (Transcript in
Appendix 3)
Summary: This case starts in a noisy atmosphere and the Judge Stephen
asks for silence from the audience (s.1). The secretary and the policeofficer try then to stop a person who wants to talk (s.2 to s.5). The case
concerns the non-payment of a dowry. The plaintiff Gabriel (from s.10
to s.30) accuses the defendant (Santino) to have married his sister during the first civil war and to have never paid the dowry. The sister had
died and Santino did not take care of the four children, and one died in
an accident. Gabriel now takes care of the children and is asking for the
payment of the dowry. From (s.31) to (s.33) there is another interruption from a woman. In (s.35) the defendant explains why he was in the
incapacity of paying the dowry. A number of relatives come to give their
testimony (parts not reproduced in the Appendix). At the end, the family of the defendant agrees to pay the remaining of the dowry and the
Judge Stephen (s.36) asks everybody to reconcile.
The plaintiff is speaking in Juba Arabic while the Defendant is speaking in Bari. The Judge Stephen, as well as the police and the secretary
shift between Juba-Arabic, Bari and a few English (see s.30 for an example of code mixing Bari-English).
This Case illustrates a type of language use very common among
the Bari population of Juba and surrounding areas, with a deep interpenetration between Bari and Juba-Arabic, which certainly helped the
vernacularization of JA.
When the Judge Stephen and the plaintiff are speaking in JA, one
notes the phonological influence of the Bari languages with realizations:
f>p
pi in (NSA fii); pogu on (NSA fgu),
z>j
julumin spoiled (NSA mazlmn)
> aat banaat girls (JA banat/NSA bant), maat die (JA mutu/
NSA mt)
e>"
gdiyat case (JA gediya/NSA gadya) (/"/ is noted in Bari
script and by Bari speakers).
A number of words are common to both JA and Bari like mali dowry
from Arabic maal money, or kurju cultivate from Bari kuruju. Some
expressions are idiomatic translations from Bari expressions such as
(s.38) tusu bujak ~tufu buzaK spit saliva, i.e. give benediction.
619
When the defendant is speaking in Bari, one notes the presence of a

number of Arabic (sometimes from Ottoman military Arabic) words like
temporal markers kan was/before, badin then/at that time, asa now,
discourse markers taban of course kalas OK, numerals (all numbers
are in Arabic) and words linked to administration or political events
such as hawadis civil war, hukum judgement, korbaat whip.
Inter-penetration between the two languages goes over phonological
and lexical transfers. It includes also syntactic similarities like the use of
verb say (gale in JA, adi in Bari) to introduce embedded clauses (Miller
2001).
(s.7)
taban ana sufu zede gal kan uwo ro ptisu mal . . . .

/naturally I saw like-this say if him go look dowry . . ./
I realized that if he was going to look for the dowry
(s. 36) kirut adi monye sarji adi ukum nagwon kju monye a ukumbe adi asa
man dek pitn kwe
Then her father sent a message that the judgement passed before was
that now I want the remainder of my money.
Note that almost none of the mesolectal features, so common in the

Garawiyya, were recorded in this case. I noticed that the Judge Stephen
never tried to imitate a mesolectal level, even if some of the young participants were talking in mesolectal JA. If he had to alternate and adapt
to the speaker, it was between JA and Bari and not between JA and
mesolectal. Stephens way of speaking is still a rather formal and juridical way of speaking and cannot be associated with an informal level of
JA. However his speech (particularly s.37), as well as that of Plaintiff
Gabriel, was rather similar to the level of JA broadcasted by the radio of
the Sudan Council of Churches and is rather similar to what linguists
will consider as Juba Arabic (see Watson (1984) in particular).
3. Conclusion
The three cases illustrate different levels/types of Juba-Arabic. Of
particular interest was the attitude of the two Judges, Ramadan in
the Garawiyya and Stephen in Kator. Ramadan realized a number of
mesolectal features and used some Arabic Juridical terms. However, he
was adaptating his speech level to his interlocutors (cf. compared degree
of mesolectal insertion between Case 1 and Case 2) and he never tried
to assert his Authority through the exclusive use of a high linguistic
620
catherine miller
norm. He also never commented about the language use of the other
participants. As for Stephen, he always spoke a basilectal vernacularized
Juba Arabic and was playing between Bari and Juba Arabic. He often
made some comments about the Southern identity (ena fi januub here
in the South, kalam bta januub words/languages of the South, arabi
bitana bta januub our Southern Arabic) and once criticized a young
man talking like a Northerner.
The language use of the Judges indicate that while the Garawiyya
Court was influenced by its surrounding Arabized Malekiyya neighborhood, the Kator B Court was influenced by the Bari surrounding, even
if most speakers prefer to speak Arabic rather than Bari. Two types of
urban models were present here. On the one hand, the Northern Sudanese Arabized urban model symbolized by the merchant community
(both Arabs and non Arabs living near-by), which still had an influence
upon part of the Southern population (old settlers as well as newcomers).
On the other hand, the East African urban model (Nairobi, Kampala)
brought back by the returnees and supported by the local Churches and
the political Southern activists. In 1984, at the breaking of the second
civil war, these two urban models were still coexisting together with
more rural traditional ways of life. Language diversity, as well as religious
and ethnic diversity, were considered natural components of the city.
How far did somebody like Hawa, the young Bay policewoman,
and Stephen, the former policeman or Gabriel consider that they were
speaking the same language (Juba-Arabic?), or that they were speaking
two different languages? This was a question that I did not ask at the
time of recording and that I cant answer. I later worked with some Bari
informants in Khartoum. They could very easily reproduce Creole/basilectal Juba Arabic features, when asked to do so. For those who knew
NSA, they were perfectly able to distinguish between the two systems.
They had therefore a clear consciousness of what JA was, as a distinctive
linguistic system. I noticed, however, that when I asked some of them
(students trained in linguistics) to transcribe some of the tapes recorded
in Juba local Courts, they tended to systematically transcribe them in a
basilectal Juba Arabic phonology, without reproducing the mesolectal
variations. It was as if, for them, people from Juba were speaking one
language. I did not discuss with them the reasons of their attitudes and
I dont know if they were not aware of these variations.
The recording of natural corpora help to better grasp the natural
diversity. But it makes the linguistic analyses more complex and renders
the concept of autonomous linguistic system rather problematic. As it
621
is also the case with Standard and Colloquial Arabic, it is possible to

describe the two poles of the continuum as two different systems. But
it is hardly possible to decide where the boundary is between these two
languages within the continuum.
4. References
Heine, Bernd. 1982. The Nubi Language of Kiberia. An Arabic Creole. Berlin: Dietrich
Reimer.
Luffin, Xavier. 2005. Un crole arabe: le kinubi de Mombasa, Kenya. Muenchen: Lincom.
Mahmud, Ushari. 1979. Variation and Change in the Aspectual System of Juba Arabic.
Ph.D. thesis, Georgetown University.
Miller, Catherine. 1984. tude socio-linguistique du dveloppement de larabe au Sud
Soudan. Thesis, University of Sorbonne Nouvelle-Paris III.
. 1987. De la campagne la ville. volution fonctionnelle de larabe vhiculaire
en Equatoria (Sud Soudan). Bulletin du Centre dEtude des Plurilinguismes (Nice) 9.
126.
. 1989. Kelem kalam bitak: langues et tribunaux urbains en Equatoria. Matriaux
Arabes et Sudarabiques (Paris) 2. 2358.
. 1993. Restructuration morpho syntaxique en Juba-Arabic et Ki-Nubi: propos du
dbat universaux/superstrat/substrat dans les tudes croles. MAS-GELLAS Nouvelle
Srie 5. 137174.
. 2001. Grammaticalisation du verbe dire et subordination en Juba Arabic. Leons
dAfrique. Filiation, rupture et reconstitution des langues: un hommage G. Manessy,
ed. by Robert Nicola, 455482. Leuven: Peeters.
Owens, Jonathan. 1991. Nubi, Genetic Linguistics and Language Classification.
Anthropological Linguistics 33, 130.
. 1997. Arabic-based Pidgins and Creole. Contact Languages. A Wider Perspective, ed. by Sarah G. Thomason, 125172. Amsterdam-Philadelphia: John Benjamins
Publishing.
. 2001. Creole Arabic: the Orphan of all Orphans. Anthropological Studies 43,
3. 348378.
Tosco, Mauro and Jonathan Owens. 1993. Turku: A Descriptive and Comparative
Study. SUGIA 14. 177268.
Tosco, Mauro. 1995. A Pidgin Verbal System: The case of Juba Arabic. Anthropological
Linguistics 37, 4. 423459.
Versteegh, Kees. 1984. Pidginization and Creolization: the Case of Arabic. AmsterdamPhiladelphia: John Benjamins.
Watson, Richard. 1984. Juba Arabic for Beginners. Juba: Summer Institute of Linguistics.
Wellens, Ineke. 2005. The Nubi Language of Uganda. An Arabic Creole in Africa. Leiden:
E.J. Brill.
Yokwe, Elisai M. 1985. The Diversity of Juba-Arabic. Studies in African Linguistics,
Supplement 9. 323328.
622
catherine miller
5. Appendices
5.1
The Dinkas case: Garawiyya Juba, 1981
(1) S. ombre gedya fi mantr tultu mya tnn u asrn / tabn ombre azim
asan mra de ibu ns/ jl el ga-ref asn zdu bta jl de / gal yani
de wa gl mlk/
Yesterday the case in the register (was recorded as) three hundred
twenty two. Of course it has been reported so as the woman could bring
people, somebody who knows him [i.e. the defendant] and could testify. Because the man said he is called Deng and she said he is called
Maluk.
(2) J. win sud/ zl el-brefu jl da / inta
Where is the witness? The one who knows this person? You?
(3) T. ama ol igl en ya wktu / gl ns tmon fi k k / lkan mai
lomon / oman gl zl de armi / mumkin fi jerma tni nina m dr mi
/ umon kfu b / m dr msi/
Listen she says what Wakotu (??) /she says that their people are in
Konyo Konyo/ if she goes to them / they say this person is a thief / maybe
there are other crimes / we dont want to go / they are afraid of him, they
dont want to come.
(4) J. Nna bas l haj-e-en/ nna ndin jl /wo biji an ije-wri isim ta jl de
bs / ms kalm tni/ bas iji wri gl ya mahkama na brefu jol de / sim
t filn / asn wo indi tnn isim ena / jl de biwri yat el-numro biyref
fg / bes kalm tni m f/
We take him for what? / we called the person / (so that) he comes to
show the name of the person only / Nothing else / He just comes and
tells Court I know this person / his name is so / because he has two
names here / The person tells which number (name) he knows / there
is nothing else.
(5) T. Ya na klem le ede b"t nagfa / gl l/ lkn intkum dr kalam/
klita ol ir wodna ma bl fi k k
Indeed I talked to him like this but he refused / he says to me / if you want
words / lets somebody go take us with the police in Konyo Konyo.
(6) J. La la / tyeb hwa indu xu / jl bita f / rajel bita wn ?
No, no / OK does she have a brother / does she have somebody / where
is her husband?
(7) T. ay f fi bra
Yes he is outside.

(8) J.
623
Tyeb / kan uwo ra ma axu bitao / ma rajel bitao / uwa birh wri nas
del / yajeb nas del / ma fi kalam dosoman / keli nasma blis lokan fi blis
Good / she should go with her brother / her husband / she goes to tell
this people / she brings this people / there will be no troubles / lets listen
to the police if there is a policeman.
[interruption] The policeman refuses to go with the woman and says:
/ you can not approve a sick man / you cannot approve it a all / because
we blis ma negder/. ./
(8bis J.) its up to us / lakin de ya ma besma min kalm tay aja zede / de fikra ana
kelem ta sakeT maales / kan keda kwayis / izakan u ma jebu nas de / jol
/ nas el beld el ja wri isem / nina balasu-l-gedya /
Its up to us / but this one does not listen to me / this was a simple
idea / no matter / if so ok / if she does not bring the people / the person / people of the village who come to show the name / we will end
the case.
[the case is suspended, the woman goes outside and comes back with a
witness]
(9) J.
inta bahalef harba wala bk / bibl / asma el-mahkma ndi eta aan
haja / nasaltek aja / twori kalam mosbt inta baaref / eta ma tddisu
/ mara de gl aslu inta baarfu jl de / jl de gabel keda besuf ?
You swear on the spear or on the Book? / the Bible? / Listen the Court
call you for something / we ask you something / you tell the right thing
you know / you dont hide / this woman say that you know this person /
this person did you see her before?
(10)
W. jol de / ana baaref yisim bita / isem t maluk / maluk aywen

This person / I know his name / his name is Maluk / Maluk Aywen
(11) J. inta arfu da mitin
Since when do you know him?
(12) W. Ya zl taban ana bigul neya /asa jina asa al ma ksel ydi / fa wd
de ma al u bls masa alna / u wd de taban u barefu zol de taban /
lakin u mubleK / badelma ana baref zol de / ana ma mumken nugasr
gediya bita l / interruption /
O naturally I will say the truth / actually I work as a cleaner / and

the boy came to me and the police came to me / and this boy of course
he knows this person / but he is the plaintiff / since I know him (not
clear if it is the plaintiff or the defendant) I cannot shorten(?) his case
if
624
catherine miller
(13) J.
tegriban jo kem sanawt eta barfo?

Approximately, for how many years do you know him?
(14) W. aywa s / min kema u abain
Yes true / since 1975.
(15) J. inta mudda de / min kemsa u sabain li haddi nahar de / jl de min de
inta brfo uwo / hl inta baarfo aglaK bitao / u jl kwayis wala kan
indu jerima bta nas biil haja bta nas
You, during this period / since 1975 until now / this person since you
know him / do you know his behavior? / Is he a good person or someone
who has crime of people stealing things of people?
(16) T.
s kalm ya ayed el-reys / inta tabyan ta-aref zol wa inta yikun

kunta gyad fi bet wd / inta ma botaaref tani jol de / bref b u binum
wn / inta ainu fi ka / teselem ydu u bref taban / u lekin inta m
ref baaml enu aw baaml enu
You are right Mr. the President / Of course you know a person if you
are staying with him / you dont know else / this person you just know
where he sleeps / you see him on the way / you great him you I know
him of course / but you dont know what he is exactly doing.
(17) J.
m batl / Deng Mayen inta bikutu isomu bitak musu Deng Aywen /
lakin dk bigul isom btk Malk Aywen / indak sual le / asal /
Not bad / Deng Mayen didnt you put your name as Deng Aywen?
/ But this one says that your name is Maluk Aywen / Do you have a
question for him? / Ask!
(18) D. uwa d / ana indu sual wa gidamu / uw akyuT / ana nam seta yom
/ uwa ma adana akel wala mya / de lujal jr / [laughter in the room]
/ mara de dr fgri ana bi bi b-l-awanta
This one / I have one question for him / Is he a brother? / I slept six days
/ he did not give me food nor water / Hes a ?! / This woman wants to
make me poor by her tricks!
[The witness and the defendant start to argue between themselves in Dinka]
(19) J. mutarjem winu where is the translator?
(20) W. nerja le kalam de / uw gal enu / izakan ana axk ita addni futur aw
ita adni ay / u gal keda / lekin ya assi ana gutta lek jebel / l-zol de
ana baaref u-binum wn / wa baamol enu fi-l-yom u baamol !enu ana ma
baaref /negabel fi ska nesalem yeda / fa neddu l -akel/ m indi akwn
rabto kulu
Shall I return to you the words? / What did he say? / if I am your brother
you give me breakfast or tea / He said like this / but I told you before /
625
this person I know where does he sleep / and what does he do during the
day I dont know / I meet him on the way, I great him / and I would give
him food? / Didnt I have brothers that I all raised?
(21) J. to W. izakan ita kan kazb / inta begul sehi lama kan l-mahkma sala
/ lokan gal ana aarfa kalam btau uwa jol batl / lakin inta betkelem ez-zn / l-haG / inta gul ma baaref el aja uwo baamol /
bes inta ligo fi seka bes / de mus kwayis? / tb ya mara / weled btak
suker winu
If you were a liar / you would have say true when the Court ask you / if
you had say I know hos words he is a bad man / but you speak the right /
the truth / you say you dont know what is he doing / you just meet him
on the way / isnt it fine? / Ok woman / where is your small boy?
(22) J. to D.
delwokti maakama ligak bi galata galatatn / alla inta kizibta

/ inta gulta isma bitaki Deng / wa hini akwana btk igulu l de
ms isema btak / da tazwir / isem btak Maluk Aywen / de barau
jiza / jiza numro itnin / sarkta dura bta mara de / inta sarkta
dura li"ana inta gr isma bitak / lokan ita ma serG inta ma mumkin bitager isma btk /. . . [the story continues again and again..] /
fa keda /nina binsuf nas / daiman mujrimin ketr / nama gabod
bikr isem /. . ./ an keda kalam de nihna gul inta muznb alat-tnin / fa kalam el-mahkama / inta biyemsi sita suur fi sijin /
izakan inta ma jibtu talta u taltin gine bta mara de / w jiza
btk an inta xin l-aman / inta bmsoku talata suur / kulu juma
s-sijin btak tesa sar / izakan ma dafa grs / kan dafa grus inta
tla bara min sijin
Now the Court find you with two faults / today you lied / you said
your name is Deng / and your brothers say no its not your name / this
is falsification / your name is Maluk Aywen / this alone is a fault / the
fault number two / you stole the sorghum of this woman / you stole
the sorghum because you changed your name / if you did not steal you
would not change your name / [the story continues] and like this / we
see people / many criminals always / when they are taken they change
their names / because of this we say that you are culprit for two things
/ and the decision of the Court / you go six months in jail / if you dont
bring the 33 pounds of this woman / and your fault that you are a perjurer / you take three months / the total of the jail is nine months / if you
dont pay the money / if you pay you will got out from the jail.
626
5.2
catherine miller
The Bays case
/. . ./ indicates a cut.
/. . ./
(4) J2. smu mnu
Your name?
(5) J2. abla ?
Tribe
(6) J2. sken wn
Living where?
(7) J2. sgol btk
Your work?
(8) J2.
(9) J2.
(10) J2.
(11) J2.
(12) J2.
(13) J2.
(14) J1.
D1. John Gabriel

D1. Bay
D1. Hay is-sinema
D1. ma el/tleb
Im not working/
student
mrk
D1. tnn u irn
Your age?
Twenty two
mutm numro tnin smu mnu
D2. welyam pitr
abla
D2. bay
sgol btk
D2. ma el/tleb
student / ah yr ed
D2. omr/ sabaater sna
yur ed by ingliz/ eta mus asa tleb wela kf
Your age in English / arent you student or what?
ya askri inta bitalfa / ah an iglu l-hG / kut yidek na / tabn
mesya mu kda / gl wl al-azm / kitb el-mukaddes / agl el-hG
/ kl-el-haG / wel y gr el-hG / kda wor el-mdu li-l-mahkma
/ nafern da mal lk nu
soldier you take an oath / ah in order to say the truth / put your hand
here / of course Christian isnt it ? / say Lord the Great / Sacred book
/ I say the truth/ all the truth / nothing else than the truth / so tell the
matter to the Court / what did these two-persons make to you?
(15) P. wellay ombreh hawli saa saba fi-l-misa ana jay bi tark bta-l-mdrasa tijriyya de / ana jt / f askri mi giddmi lbes rasmi /
baadin el-awld bard kaman fi giddmi / ana jt talta bi jambum
/ el-be henk gl le ya axi salm m f / el bi-jay bardu kaman gl le
salm m fi / ana saktta / aslu m gutta hja / ana mya / baadin
ana gumta zitta katwa btay / hum bard zdu katwa waray / el-be
henk ja garasn fisulbi / badin el-bi jay ja daggan
By God, yesterday / around seven oclock in the evening I was coming

by the road of the Commercial School / I came / There is a soldier walking before me, wearing his official clothes / then the boys also are before
me / I came near them / this one told me of my brother, no greetings? /
the other also said no greetings? / I kept silent / I did not say any thing /
627
I was walking / then I accelerated my walk / they also accelerated their

walk / this one started to pinch my buttocks / and then the other started
to beat me . . .
(16) J1. b-l-yedd wela bi kur
With the hand or with foot?
(17)
P. bi-yddu garasn fi sulb bi-ydd / baadin el-bi-jay ja daggan
keda bi-ketfa / ana gumta wagftu / gulta mlkum ya axwni / gl le
mlkum kf / ma titfahim ma'an / ana gult nitfahim ma'akum
agl makum en / gl le inta askut / inta m ermta u bass / ana
gulta lem ukrn / ana met / el-bi jay da tawli masakn ja garasni
/gulta le inta mlek asa / gl li axlaK xamak asa ana beduggk
/. . ./ tawli gumta nadt askri / gutta ya askari taal haj f el-jamaa del al-ajamn min / tawli hu / waled de ja gam darabni
bunya / askari jire ja / ja gl intum mlkum ya awld /. . ./ gulta hum
ajyomni / an keda ana korokta lek / askri da gl xls xls de
mawdua bast /"arahkum / woled de lisa der amol maaya mukel
/. . ./ sukna maan l-merkez tawli /. . ./ askari da / f arabiya tli
keda / askri tawli wgef arabiya / arabiya aba yagif / mai wgef
musfa keda /. . ./ lamm- jere mi li-l-arabiya henka / kullum
itnn tawli jmu foK / de gm bdrd fK / gaadn nesru / baadin de gm bdrbni bi wara kamn /. . ./ sa' fekka min yddi / ma
ref kn wa ya lu / wela kan wg wn / na ma rfu /. . ./ ya
gumna jna lahadd-l-nkta / sbbo le / inta en / nahl dnek wa
inta ermta u bass / inta mi riya skeT /. . . . /
With his hands he pinched my buttocks with his hand / then this one
slapped me / I stopped / I said whats the matter my brother / he said
what the matter how? / let you understand us ! / I said I understand you
I tell you what? / He said to me you shut up / you are just a pimp / I told
him thank you / I went / this one sized me directly and pinched me / I
told him whats the matter / he said close your mouth now I will beat you
/. . ./ I immediately called the soldier / I said soldier come and see this
people who are attacking me / immediately he / the boy started beating
me badly / the soldier run /asked what the matter boys / I said they
attacked me / this is why I shout after you / the soldier said ok ok this
is a simple matter / lets go ! / the boy still wanted to make me problems
/. . ./ we brought them to the post directly / / the soldier / there was a
car coming like this / the soldier stopped the car immediately / the car
refused to stop / it went stopping at some distance / when he run after
the car there / both of them immediately attacked me again / this one
beated me again / we were fighting / then this one hit me from behind /
my watch came off from my arm / I dont know if he took it / or if it fall
628
catherine miller
somewhere / I dont know /. . ./ so we came up to the police-station / they

insulted me / what are you / Curse on your religion you are just a pimp
/ you go in the street for nothing . . .
/. . ./
(24) J1.
(25) J1.
intum samt kalm da

D1. samtu
Did you hear these words?
I heard
el-bint de gl iya maya w intakum ta-azmtlu skeT / gl dik ya
ermta / u de gerres fislbu hene / w-inta bardu daggest bi ketf
/ u badin gl mlkum / intakum gul inti ermuta skeT / inti jya
min wn / fi tarka gmo daribtu bi daraba adid / wa gam korrk
l-askri ja / w bardu baad el-asker / kamn intu gwum darbtu
baad / asn askri mi an ijib / yemsi kelem lel-l-sowwaK el-kan
waggofu / fat giddm / enta azemtolu darobtu li-raet ma saa btao
fakkat waga / wa wahed halaK kamn / da amn / entu gabelt fi
ska / inta /. . ./
The girl said she was walking and you attacked her for nothing / this
one said pimp / and that one pinched her buttock here / and you also
slapped her / and then she said whats the matter / and you said you are
just a pimp / from where are you coming / on the way you start to beat
her badly / and she cried the soldier came / and also after the soldier /
you start to beat each other / during the time the soldier went to bring /
went to talk to the driver who stopped / he went in front / you attacked
her and beated her until her watch came off / and a hear-ring also / this
is sure / you met in the way / you /. . . ./
(28) D1.
el-bit-de / ana kunt jay min hay kmryal li-l-bT / fi wad gidmi
/ ana ma baarf gbel / ma bit de wara / towones m bit de k /
umon gam fi akla / an na zeyde / ana askate umon /
The girl / I was coming from the Commercial District / there is one
in front of me / I dont know him before / with the girl behind / he discusses with the girl / they start quarreling / because Im like this / I make
them silent.
(29) J1.
(30) D1.
da awli s kem
This about what time?
awli s sbaa / fi jay bls / fi jay / bit krk le bls / fialn el-blis
j / woddit len mukla klu / ana mi le-beT / yalla ez-zl eldaraba-l-bit de / ma blis gaadn isugg zl de li-nukta / uman
ft giddm / ana ja wara / nama ana wosol mbil / uman lisa / jl
el-dagga bit u blis / ana ja ligm lsa fi mukla / dl fi mukla unk
/ min kede nama blis ainu na / el-bit gl aywa na gibel hink / ana
629
gl filn ana gibel hink / akala hsel ana f / ya al-amal min ene
ana ma rif / blis masako jl de u jl de ft / ya ja bas masakan
whed be-rejlan /. . ./
Around seven oclock / a policeman came / the girl called after the police
/ of course the police came / and brought us the all problem / I was going
home / Then the person who hit the girl / he and the police they were
driving this person to the police-station / they went in front / I was coming behind / when I arrived to the Mobil station / they were still, the one
who beat the girl and the police / I found them still quarrelling / from
this when the police saw me the girl said yes I was there before / I said
of course I was there before / the quarrel happened I was there / then I
dont know what happened / the police sized the person and the person
went / then one took me by the legs /. . ./
/. . ./
(34) D1.
away / badaln talta / na / blis u bt da / sillna fi nkta / amdulla

na lig waed min az ma mbil enk / baaref ns bT / na gl
kwayis / kan keda ami klem ns bt / gl na fi nkta / il-ana fi
nkta / minnak axuy de / nama nsma kede / r fi nkta enk / nama
anu axuya ja / gl ya tni f / kutt tawli jowa / u badin sab / bit
de gam j len / gl saa woddur u en / u filan ana ma uftu saa
klu klu /
Yes / we stayed three / me, the police and the girl / they took us to the
police station / Amdullah I found one I wanted in the Mobil station /
he knows the house / I said good / if its so go and tell the people of the
house that I am in the police-station / they took me in the station / from
there my brother / when he heard like this / he went to the post / when
he saw my brother coming / he said the other came / they put him inside
immediately / and then in the morning / the girl came and told us that
her watch was lost and what / and of course I did not see her watch at
all
(35) J1.
tyib / lama kn wled de /el-biyamml makil maa bt de / inta

ma mait / inta ajiju / inta kan barak wela ma wald da /
Good / when there was this boy / who was doing the trouble with the
girl / you did not go / you support / you were alone or with this boy?
(36) D1.
na bara
I was alone
630
(37) J1.
catherine miller
inta bara / tyeb badin hinay de / lama blis gabdu weled dk
mi bi / u gamaa kal da / inta ajijtu maa el-bint-da / blis
lamma j / hl giddm el-b lis eta amoltu makil maa bint da
You were alone / good and then / when the police captured that boy
and went with him / and the group quarrel / you supported the girl / the
police when he came / did you in front of the police make quarrel with
this girl?
(38) D1.
ana ma awmol makel / bass darab bitman na ja katal

I did not make quarrels / I stopped their fight
/. . ./
(45) D2.
el-zl de ombare ma biill nihna / ana gaad fi-l-bet / ya rasel wdd

da min el-mbl / wodd de maa unk / gl ya ns el-bet John gabad
/ ya ma kan nemi yisf li-l nkta
The person of yesterday did not take us / I was at home / and he sent
the boy from the Mobil station / the boy went there / he says people of
the house John has been captured/ so I had to go and see at the police
station.
/. . ./
(50) J1.
(51) D2.
nta ma fs / u gabodk mittn

You were not present / where did they capture you?
gabodni zamn gale nemi-l-nukta niuf / min wisolta unak an
nera mukila bta djn kif / bass waed askri gl ya lzem bikn
wahed minhum / bas ja fi hene / tni ma barja bt / u fialan jabni
fi-l-arsa / ya sab bit de maa unk / djn gal ya bit bta ombare
enay jabni fi arsa unk / /
They captured me when I went to the police station to see / when I

arrived there in order to assess the problem of John / one soldier said he
must be one of them / he came here / I didnt come back to the house /
and of course they took me in jail / and in the morning the girl came /
John said this is the girl from yesterday who brought me in jail /. . ./
(52) J1.
ya bit / asa nafarn de badi sumit kalmu / da numro wahed de

nker / u gl woket dak kalam inti gul ma hasal klu-klu / gl uwa
mi liga inta kunta akal maa wahed tni / w uwo mi aan iyazizkum / iyazizkum bas inta maa zl dak / el-wokT inta ndit askri
ja / w-el-askri gam sgo / sG el-waled maaki / inta mi u bardu
waratum kaman / kddam giddm henk intum wageftu kaman bitkelem kalmt keda / lama u masa wosolu lekum henk / bls gl xls kan
izan keda ta'ali inta kaman maa el-jamaa del / gam xala dk u alu uwa
631
girl / these two persons after I heard their words / number one this
one denied / and said at that time, the story you tell did not happen at
all / he was going and met you quarreling with somebody else / and he
came to support you / support you against this person / when you called
the soldier came / and the soldier took the boy with you / you go and he
is behing you also / in front in front there you stopped you talked / when
he arrived near you / the police say ok if it is like this come you also with
the people / he let that one and sized him
/. . ./
(66) J2.
/. . ./ keda ita indu rtn / musu barau rtan bta bay keda wonos
Do you have a language / isnt it a specific Bay language which is
spoken?
(67) D1. ana borton ndogo lakin ma fi rtan bay

I speak Ndogo but there is no language Bay
(68) J2. kulu ndgo intum trtn ndgo
You all speak Ndogo?
(69) P.
ana borton ay / norton kalam enu lakin
Yes I speak / but what shall I say?
(70) J2. de bigl uman beroton enay de rutn / de bigul l / ma boroton ndgo
/ beroton bay / nina azin rfu uman sei bay aw /. . ./ :
This one says that they speak their tribal language / this one says no /
he doesnt speak Ndogo / he speaks Bay / we want to know if there are
really Bay!
(71) J1.
tayyb / izakan bay / el-fakka bikun fi enu / lokn ma bay bikun fi

enu / nesma el-jarma bass
Ok / if Bay / the difference would be in what? / If not Bay what would

happen?/ we listen to the crime only!
(72) J1.
hata kan abla tni bara ma tam / jol el-arfu jol amel ma ajam /
biwori gul ya de / kan gabil wela ma bay biwori / bas el muhim inta
jeb el-blis el-kn ahall al-mawdu /. . ./
Even if it is another tribe it doesnt matter / the person who knows the
person who attacked him / he shows that it is him / if from the Bay or
not he shows / the important you bring the policeman who solved the
matter!
/. . ./
632
(74) J1.
catherine miller
asma / al-sakya da ixtarrat gl inta shid / lama kan hsel beinakum
maa nafarn del maskal / eta keda gul s el eta bitaarefi bi zabti / eta
jt kif / eta legt maamal enu keda / gul len / le-l-mahkma
Listen the lady (?) decided that you are the witness / when the problems happened between you and this two persons / say what you know
exactly / how did you come/ what did you find? / tell us / the Court!
(75) W.
taban siytu / ana kunta saal fi kart / fi-l-jawazt / mal el-jensiyat / badin ana sken fi muluk / lama ana jit ja wosolu fi medresa
Komeriy l / ana ft / taban al-bit de askri ana ma baaref / lbis
maleki / ana ma biyaref askari aw ma askri / faaln ftt / giddm
swiya / baad-ma fatt / al-bit de gum bikor bi-isma-l-blis / gal ya
blis ya blis tal agaod / fialan f gann btatna / bta al-blis / ay jol
bikrek bi-isim el-bls ma mafrd tafgo / lajem ita-ajri suf fi senu fi
senu / tekusu / fialan ana jay / ana jay legtum del itnn klu biajem
el-bit de / baadin ana jt / ana gul ya jamaa mlkum fi sen / gal
el-bit de sottemuni u ft keda / ana gulta tayib ya axwna matakum
sakal bit fi tarka / bi tarika zede / keda tosf el-jisim btna w-el jisim
bitkum wa jisim el-bit / taajem el-bit fi tarka bi-l-ll zede m sh /
wa lisa bikroko / lsa bikroko / fa ana asala-l-bit / hasala enu maak
/ gul nas del ya fat keda / ns del ajemu fi tarik / agru sulbu bitao /
u nama kelem umon setemu / fialan uman setemu /. . ./ bad-ma ana
rajo enka / ana raja / ana ja legitum lisa bisakal ala bit / nama ana
raja min mahal taban ana ft keda / baad-ma ana asma korokoro
waray b-ism el-blis / ana jire tawli ja wara w ana ja legitum l-itnin
del kullu / uman yau / fa filan kelemtum liw / ana gul ya akwnna
nina janbiyin ma mafrd neskel maa badna / hajt zede ma kwayis
/ el-muskila de best tkelwokT // el bit de je gum kelem gl ya askri
il tawasolna fi nukta //
Of course Sir / I was working in the cards / in the passports / the place
of nationalities / then I am living in Muluk / when I arrived at the Commercial school / I passed / of course I dont know that the girl is a soldier
/ she wore normal clothes / I dont know if she is a soldier or not / of
course I passed / a little bit in front / after I passed / the girl started to call
by the name of the police / she said police police come in / of course
in our laws / of the police / anyone who call after the police you cant
let him / you must run see what is happening / you search / of course I
came / I came and found them both attacking the girl / I came / I said
people whats the matter / the girl said they insulted me and went / I said,
good my brothers dont quarrel a girl in this way / by this way / you see
my body and your body and the body of the girl / you attack the girl by
night like this its not correct / and they still shout / they still shout / and
633
I asked the girl / what happened with you / she said this people passed
like this / this people attacked her on the way / they pinched her buttock
/ and when she spoke they insulted her / of course they insulted her /. . ./
when I came back / I came back / I found that they still quarrel on the
girl / when I came back from the place I went there / after I listened the
cries behind by the name of the police / I run directly back and found
them both / there they are / and I spoke to them / I told them brother
we are southerners no need to quarrel between us / things like this are
not good / the problem is small until now /. . ./ the girl spoke and said
soldier you must bring us to the police office /. . . ./
(75bis) W.
ma f maal hukuma keda gerb fogo /. ./ fi arabiya gaad j /. . ./ ana

gum jire bi-wara arabiya de bi-ztu /. . ./ min ana jire henak / ana
asma baga el-bit de biga bikore / uman tni keda bidugu el-bit biwara /. . ./ filan ana jere kede kede kede / ana ja gabodtu da / xalas
amdullah / fi fi mujra waga fi jua mujra /. . ./ negedem kede kede
kede likaat nina wosolu mobli /. . ./ d tawli gum amol musakl
fo mbili li-raat-ma aoz kaser mrat bta arabiya zatu / aoz amol
muskil fi jua arabiya ztu / baadu sd el-arabiya uf kede m f /
da gum maragu bara / ana bardu gum maragu /.../ tni arabiya
ja / rakebna fog le-raat wosoluna fi nokta / baad-ma wosoluna
fi nokta / ana gul jol da jere kals lakin axu da biyaref mahal de /
d nker gul m biyaref da / d maal btao barao u d barao / u ma
biyaref klu klu / ana gul kf / lajem de aku btao li"anu uman hader
al-muskel giddmi ana /. . ./ el-hamdulilay lam fatna balK /
mowdu negla askri de masi mustafa / ana biga ft fi-jawazt ya
neselem kelem kabr li jamaa fi taakir el ana akr lina zaman / ana
msi wodi kabar li-hum enay / lama fi rojyu bitay / el-hamdullay d
kamn gum wasal / ana gal ya de bs ya de bs / uman l-itnn el
ajam ala bit ya de bs /
There is no governmental place nearby /. . ./ a car came / I run after the

car /. . ./ when I run there / I heard the girl shouting / they were again
beating her from behind /. . ./ of course I run like this like this . . ./ I
captured this one / ok thanks to God / there is a sewer canal / he fall in
the sewer /. . ./ we accompany like this like this until we arrived at the
Mobil station /. . ./ this one started to make problem in the station until
he was near to break the mirror of the car / he almost make problem
inside the car / then the owner of the car see its not possible / this one
went out / I also went out /. . ./ another car came / we went in until we
arrived at the police station / after we arrived in the station / I said this
one run away but his brother must know the place / he denied he said
he doesnt know him / this one has his own place and this one also / and
634
catherine miller
he doesnt know him at all / I said how? / It must be his brother because
they attended the troubles in front of me /. . ./ Thanks to God when we
opened the case / the matter of the transport of the soldier to the hospital / I went to the passports to bring the news to the people there about
my delay before / I was going to bring the news to them there / when I
was coming back / Praise God this one also arrived / I said this is him
this is him / the two who attacked the girl, this is him /. . ./
/ story continues . . . ./
(76) J2.
Tayib intakum ref nefsu kalatnin

Good do you recognize that you are faulty?
(77) D1.
la la la kalm ke ana wosol dak ana baray / kalam de kadb / ana ma

jay maa jol / ana jay baray
No no this words I arrived there I was alone / these words are lie / I
didnt come with somebody / I came alone
(78) J2.
ya de suhud kelem kalam bta / keda wonosu ma suhud / asalo /

ind"k sual yeslo
But the witness spoke / so discuss with the witness / ask / do you have
a question for him?
[Discussion between the defendant and the soldier /. . ./]
(79) J1.
intum kalatanin / mahkama gul "annu intu aemtu bit da fi-t-terk

bidun ay munasaba / u atemtu u gul ermuta / u into garestu-ha
li-raet-ma sa"a bitwdda / tamenu xemsa u arbain gine / w indu
wahed fardeK halla wodur / tomon sabaa gine u nus / de klu bi
kuss maskel btakm maao fi-t-teriK /. . . continue /. . ./
You are faulty / the Court says that you attacked this girl on the way
without any reason / and you insulted her and said pimp / and you pinch
her until her watch got lost / its price was 45 pounds / and she had earrings which got lost / their price seven pounds and half / all because of
your quarrel with her on the way/. . . ./
5.3
Kator B Court 1984, Record 3/ 1984
Judge Stephen to the public

(1) J.
. . ./ numro tnin koreraK / izakan nasi bikore bi-ataku ktir ma bikli

makama ymsi / izakan nas biasma kalam de bmsi bi-r / u gdiyat
bemsi bi sra / gowm / mumkin nihina baamolu isrin gediya pi yom
/ izakan ma pi pogu ijaat / ajaat bi sikil de /. . ./ arjuk sabab wa intum
635
nas mutaimin mutamain / el ja asan asma makama / intum ma bija

ena sakiT / eta jet asan testepiT fi makama / asan kul wa baadin
bukura bisala kalam pi bet / aw mara kaman banaat kan ja asma
asan ma baamalu kalaT pi biyutum / sukran / istamir gediya tni /
Number one, shoutings / if people shout laugh a lot it does not let the
Court going / if people listen the words go quietly / and the case goes
quickly / fast / we can do twenty cases in one day / if there is no troubles /
things like this /. . ./ I praise young people and you the defendants / those
who came to listen the Court / you didnt come here for nothing / you
came to benefit in the Court / because each one later tomorrow solve
his problem at home / or women, also girls they come to listen in order
not to make mistake in their house / thank you / continue next case!
(2) S.
(3) J.
(4)
(5) J.
(6) J.
(7) J.
(8) J.
(9) J.
(10) P.
Taban Sabastian ../ Alexi Lado

S. to the J. lisa ma ja
Not yet arrived!
to a man nearby) kede belay atla awal haja / ihina fi gdiya / wa mata
lakbat / ita yau ga-asma kalam ini / ita ja bikelem / istena lakad bad
gdiya / msi
Please go away from here first thing / we are in a Case / dont interfere
/ you have heard the words here / you will speak / wait until after the
case / go!
to policeman kli jol de yeji hina / korbaaT btak win ? /
Lets this person going here / where is your whip?
Police
yau le Wani
Its with Wany
ilu min Wany / u de jatu barfu urkali ? / masi tow gene-gene bita enay /
Take it from Wani / and this one does he know the Courts police? /
go and sit there!
mublek minu / Mogga lo munu
D. ana
Who is the Plantiff? / Whos Mogga?
I
a isim bitak
D. Gabriel Morbe
Ah, your name?
Barinit wen
D. lobunuk
Bari from where?
u eta majlum knyo u knyo.
From what are you victim?
de taban kan pi ukti keda / kan juju ukti fi jaman hawadis / badin ana
ruwa nadi ml / / gal mal maap / taban jaman de hawadis / taban ana
sufu zede gal kan uwo ro ptisu mal / mumkin boro ligo harib / taban pi
jaman de pi aja de /
Of course there was my sister / he married my sister during the civil

war / then I went to ask for the dowry / he said there is no dowry / of
course it was during the civil war / I realized that if he was going to look
636
catherine miller
for the dowry he would maybe find the war / of course at that time there
were such things
(11) J.
(12) J.
(13) P.
hawdis pi sanatu km

Troubles of which years?
badin ada ta awadis mitin
Then the end of the troubles when?
(9) P.
tamanya u sitin
Sixty eight
tnen u sabain / badin kida pi itnin sebeyin / taban ina kaman nadi mal
/ gal mal maap / ila nina amul aja de / dgit mrisa / anina aju wodi
le uwo / gali kasara ita jibu / ana ma indu haja tani / badin uwo amulu
karama / badin dgig mrisa de nina wodi le aja de / le uwo / ma amul
lena haja tani / badi swiya yaba ja mat /anina amulu karama ta yaba
wkit maat / wa asa ukti de / ukti de mat / wokan ma wdi le ana ml
/ bes teletin jine eli kan wdi le nina
Seventy-two / then in seventy-two / we asked again for the dowry / he

said that there is no dowry / only we do this thing / beer flour / we want
to bring it to him / he said you bring it for nothing / I have nothing else /
then he made a ceremony / then this beer-flour we brought it to him / he
did not do anything else for us / later on our father died / we made a ceremony for our father when he died / and now my sister / my sister died /
and he did not give us the dowry / only 30 pounds that he gave us.
(14) J. mt mitin
When did (s)he died?
(15) P. mat aja de fi tnin u tamanin / saar idaser
(S)he died in seventy-two / in November
(16) J. indu iyal
Does (s)he have children?
P. indi iyl
He has children
(17) J. kam
How many?
(18) P. fi tegriban arba / waiT kan ge dowru ma rokuba / uwo ma wdi baal
pogo / ja waga fi nr / gam ja mat
Approximately four / one was walking on his knees / he did not take
care / he fall in the fire / he died
(19) J. pi talta
P. pi talta ay
There is three?
Yes there is three.
(20) J. talta de fi bant fogo
P. fi wae bit u badin tnin yl / tnin woled
Among the three
There is a girl and two
there are girls?
boys/two boy.
(21) J. eli mutu de woled
The one who died was a boy?
(22) P. la / bineya / wa asa itnin iyal de uwo ma biwdi baal fogo / asa itnin
de ge geni le ana
No / a girl / and now the two boys he does not take care of them / now
these two are staying with me.

(23) J.
(24) P.
(25) J.
(26) P.
(27) J.
(28) P.
(29) J.
(30) P.
637
ge gene le eta / kalam kwes / aah /

They stay with you, good, aah?
wa min de taban ina julumin / bineya tanina ma mijawaz / bes ina
akudu teletin jine fakat / u bineya kaman maat / iyal kaman ana lisa
yau ge akilu
And from there of course we have been spoliated / our daughter is
not married / we just took 30 pounds only / and the girl died / and the
children I am still feeding them.
itakum yau ge raba
You (pl.) are the ones who raise them
nina yau ge raba / yau ana majlum ma kalam de
We are the one who raises them / I am spoliated by this story
ita masi leo pi bet / le bet btomon gale malu ita keli iyal de kede . . .
Did you went to him in the house / to their house to ask why you left
the children like this
walay insala ana masi aslu ma bijibu kabar
By god, even if I go he will never inform me
ma ja kurju
He did not cultivate (for you)?
bikurju lakin oslu ma bowodi le iyal de
He cultivates, but he never gives to the children
/. . . . . ./ Interruption with another case

(31)
a woman : nan kan a tu i diit na kora na torobo na a ilu utu ti mrri

mrri yi ko wate
I went at the time of the distribution of the furniture and then he

fought with me and his wifes
(32) J.
ina kju kdyo kiyang baligga / kede kiyang sina / maybe nan kan a
mistake / ado yie gwodam nyo Wani ? ma tini wuni ilo korobat / ti
ko yege o kunen ko doggu kito kadi / bonggwat ban le /
She had reported the case earlier / wait first of all like this / I might have
been mistaken / why do you stand still Wani ? you go receive that whip /
let him carry these, carry these furniture / are there no clothes?
(33) Woman: bonggwat ma a kekeren
Clothes! He torn them
(34) J. Santino makme ko
Santino there it is
(35) D. Walayi / a hal diri bijab i yemba nio nagwon nan yembi kiyasir nanyit
na i diit na hawadis / a de i inu diit ni kan ukum a itinasar jine
badin kayayu a saba jine / badin kjulu anan medde sona anan di min
taab kulu anan kan lunggi luaser baba uba a ko doggi male kwe kunu
uba nyu / kirut ni a kine mali kirut taban jur kaman ktir bayn a nan
/ kalas utu adi kalas lo utu taban bubul yemba makune mali lepeng aje dippa salet / loutu gwe a komonit ti poki ibang / nyena diri i
gwe i dida ko ina uro / taban ko yaba ko atu a nan bubul tindu o
638
catherine miller
nagwon kata / wu hal kajelu ni a baba lo twane / badin ne a monye lo
twane / kirut adi monye sarji adi ukum nagwon kju monye a ukumbe
adi asa man dek pitn kwe ti gurut kune ko na utu nan dek sitin
jine /
By God / really during the time of my marriage when I married his sister during the civil war / by then the Judgement here was twelve pounds
and ours was seven pounds / At that time, on seeing this, and in order to
avoid these troubles, I decided to call my fathers brother from Juba so
that he could go to collect my dowry from Juba there / Then from there
this dowry, of course the village is large, not me alone / ok, people said
ok, this man can marry, here is the dowry, he has prepared a cooking
place / He has become an in-law, let him be in the house / at this point
I started to stay with this woman / of course, if her father comes, I can
give him what ever possible / and at that time my father died / then
her father died too / then her father (i.e. an uncle) sent a message that
the judgement passed before was that now I want the remainder of my
money of this woman I want sixty pound /. . . . ./
[The case continues in Bari and JA with many different participants. At
the end the Defendant and his family agree to pay the remaining of the
dowry]
Conclusion from Judge Stephen:
(37) S. dgiga / dgiga / ah bagi mal el kede eta bi silu lau pi bet le nsibat inak
/ ah / u wdium kabar gal nina jain / asan bisilu iyl de kli ruwa tusu
bujak / uman kaman biptisu kruP kwes kida / u jama biji yesrub / wa
nas umu yani makasutin / nas klu / kulu jene beji intum bestenu / jebu
gdiya tani
Minute / minute / ah the rest of the dowry you take it to them at home
to your in-law there / and give them news that we are coming / so that
they take the children for spitting saliva (benediction) / they also look
for a good sheep like this / and the people come to drink / and the mothers people are happy / the maternal uncles people / every child you wait
for (??) / bring another case!.
PARADIGMATIC STABILITY AND FINAL LARYNGEALS IN

NIGERIAN ARABIC, OR WHY HISTORY REPEATS ITSELF,
WITHOUT ACTUALLY DOING SO
Jonathan Owens
CASL, University of Maryland
1. Introduction
Arabic presents a rich, if underutilized, domain for examining processes
of linguistic change. At one extreme are the challenges presented by the
rapid and dramatic restructuring witnessed in the Creole Arabic varieties of the southern Sudan and East Africa (Versteegh 1984, 2004). At the
other are the mundane, day-to-day, decade to decade, century to century processes which move slowly through the different varieties, sometimes leading to marked changes, sometimes not. In this paper I would
like to document one such change, which has happened not once, but a
number of times in the history of Arabic, and relate the phenomenon to
a more global interpretation of Arabic language history.
At issue is the conceptualization of Arabic language history. The
reigning paradigm today, indeed one established some 150 years ago,
is that a classical language, or Old Arabic, by various processes of
simplification passed into the modern dialects, or Neo Arabic. One of a
number of problems with this model is that it leads one to collapse what
are often internally differentiated developments into a common mold,
which are then assumed to substantiate the Old Arabic/Neo Arabic
dichotomy. Why this militates against a properly nuanced reading of
Arabic language history is that what may lie behind linguistic changes
are a combination of diverse local developments and fundamental
linguistic principles. There have indeed been many changes in the many
varieties of Arabic spoken throughout Asia, Africa and in former times,
Spain. The fact of change in one variety, however, does not justify an
historical model based on the Old/New split.
This point will be illustrated on the basis of verbs ending in a glottal
stop, or in some cases, a final voiceless glottal fricative /h/. Classical
Arabic, of course, has a glottal stop, whereas nearly all modern dialects
do not.
640
jonathan owens
Thus, in Classical Arabic forms such as bada begin and ja come

appear in most dialects as bada and j. For purposes of this paper I
will assume that there has been a change of , whereby verbs such
as bada fell together with verbs like ban he built.1 In contemporary
dialects these will typically belong to the same class of weak-final verbs,
as the following sampling indicates:
(1) Algerian bd-t, bn-t I began, I built, Nigerian bad-t/ban-t,
Eastern Libyan bid-t/bin-t, Mardin (Turkey) baday-t/banay-t, etc.2
A related development is attested in contemporary Western Sudanic

Arabic comprising NE Nigeria, northern Cameroon, Chad and the
western part of the Sudan (see Owens 2006, chapter 5). In the rest of
the paper I will describe the development in the western Sudanic area
on the basis of a relatively detailed corpus-based analysis of Arabic from
Maiduguri in NE Nigeria, and then address the question of its significance for the history of Arabic.
2. Nigerian Arabic3
Nigerian Arabic is unremarkable vis vis other Arabic dialects in its
verbal structure. Like all other varieties of Arabic it has two basic verb
forms, perfect and imperfect. The perfect stem is marked by suffixes
indicating person, number and gender, in the imperfect these being
indicated by suffixes and prefixes. Both perfect and imperfect stems
have two conjugations based on whether the stem vowel is high or low.
In (2) strong verbs are illustrated, with the verbs katab write and libis
1
The proviso for purposes . . . leaves open the possibility that the modern glottal
(stop)-less dialects go back to original glottal-less dialects in Old Arabic (see Rabin
1951). Certainly the glottal-stop-less varieties go back to pre-diasporic Arabic, as
attested in their widespread distribution across the Arabic-speaking world, and their
attestation in Old Arabic sources. If there are proto-forms of Arabic without the glottal
stop, the thrust of the paper would need to be reorientated. A few Yemeni dialects have
a glottal stop (Behnstedt 1985, 43).
Even in classical times, there was a variety without the glottal stop that a glottal stopless variant which was prominent enough that in the Koranic reading tradition (qirt)
recitations are fully allowed without the phoneme (Ibn Mujahid ).
2
The value of the suffix, -t, -t, -ayt is an issue independent of the status of the glottal stop.
3
Research support for this work was provided by the German Research Council
(DFG).
paradigmatic stability and final laryngeals
641
wear, which belong to opposite conjugations in both the perfect and

imperfect:
(2) Low stem vowels: perfect /a/ katab; imperfect /a/ bi-lbas he dresses
High stem vowels: perfect /i/ libis; imperfect /u/ bu-ktub he writes4
By and large the assignment of verb stems in imperfect and perfect to

the two classes is lexically determined, and as seen in the example, one
and the same verb stem can, indeed as a rule is, assigned to a different
class in the perfect and imperfect tenses.
Final weak verbs in Nigerian Arabic also follow the same classificatory
schema, as can be seen in the following examples in (3):
(3) Low stem vowels: bana he built, bi-lga he finds
High stem vowels: ligi he found, bi-bni he builds
2.1
Final laryngeals //, /h/ in Nigerian Arabic, some data
So far as the basic paradigmatic facts go, Nigerian Arabic is fundamentally identical to other Arabic dialects, and broadly similar to classical
Arabic. One phonological change involving two sounds has, however,
complicated the distribution of lexical forms.
The two sounds are h and . Historically speaking, NA h has two sources,
OA (or pre-diasporic), /h/ as in ahar month < OA *ahar, and OA *h
as in hilim dream, < h ilm. NA // derives from //, as in irif he knew,
< OA arifa. The change of OA /h / and // to /h/ and // is one common
to Arabic in Chad, Nigeria and parts of the western Sudan. It happens,
however, that /h/ and // themselves are weak sounds, and liable to
variation of different kinds. The range of variants includes the following:
They may be kept: biarif he knows, ahar month, whid one and ahamar
red.5
They may change to the semivowel /y/ next to an /i/: biyarif, wyid.
Lastly, they may be deleted altogether: biarif, shr, wid.
As in many dialects, there is no phonemic contrast between short high front and
back vowels. In this dialect, however, /i/ and /u/ must be lexically specified, as they are
unpredictable. As far as verbs go, given a lexical stem specification, the pre-formative
vowel is usually determined by vowel harmony rules: if the stem vowel is /u/, the prefix
vowel is /u/. Otherwise it is /i/.
5
Via the so-called gahawa-complex, whereby an /a/ is inserted in the sequence of
guttural C + C, in this case < *ahmar < ah mar.
642
jonathan owens
The variation spoken of here is at the individual level, determined by

contextual and socio-linguistic factors (see Owens 2006, 242 for statistical treatment); one and the same speaker may in one place for month
give ahar, and in another r.
The question I will investigate in this paper is what happens to
verbs like simi hear, ga#a 6 cut and karah hate, verbs with final
/h. There are two reasons for concentrating on the laryngeals in this
position. First it is only in this position that a change in the laryngeals
is nearly categorical, as will be explained.7 Secondly, it is here that the
phonological change has direct morphological consequences. If they
are deleted in final position, what effects, if any, the deletion has on the
morphological structure of the language need to be specified. Given a
form such as tismao you.M.PL hear, if the final glottal stop is deleted,
the form tismao would automatically arise. Tisma-o, with final V-V is
not a paradigm otherwise attested in the dialect, however. Alternatively,
it could collapse with the already existent weak final paradigm, tilga you
get, tilg-o they get (see (3) above), which would yield tism-o. Which
alternative emerges is described in this section.
The data for addressing this question is of two types. One is textual
data, which will be summarized in 2.2 below. First I present the results
of a test in which 8 Arabs from Maiduguri were presented orally with a
series of test sentences in which one word was left out. The respondents
had to complete the sentence with the correct form. All of the answers
involved verbs with a final // or /h/, the purpose of the test being to see
in which contexts the final laryngeals were kept or deleted. A sample
question was the following, involving the verb simi hear, the first
sentence requiring a perfect verb, the second an imperfect:
(4) hu simi an-ndim haw ana kula .
he heard the man and I also
hi tisma l-kalm haw hinna kula .
she hears the sentence/matter and they F also
The /#/ is emphatic and implosive.

It is my impression that the deletion in this position is statistically more common
than in word initial or medial position, though I have not checked this.
7
643
The most obvious response for the first was heard him, the Arabic for this
being either sim-t-a heard-I-him if the final // is deleted, or simi-t-a
if it is kept, while for the second either b-isma-ann-a 3-hear-FPL-it
(if kept) or bism-ann-a (if deleted). Note that in the imperfect a final
V-initial suffix (-an FPL in the second example) replaces the final stem
vowel if the final laryngeal is deleted.
The respondents were all under 30 and most are educated. It was found
that those older than 30 had trouble concentrating on the questions, so
unfortunately age differential could not be taken into account.
A total of 16 sentences were asked. The anticipated responses were
distributed among different morphological classes of verbs, both basic
and derived verbs for example, and among different inflectional contexts,
though only subject suffixes were tested for.
Questions were asked for both perfect and imperfect verbs, as
illustrated in (4) above. Two types of suffixes can be distinguished here,
those that begin with a C (e.g. 1SG -t) and those that begin with a V
(e.g. FPL -an). In the 16 test sentences, the answers divided into forms
involving the following suffixes:
(5) Perfect
verbs with subject person suffix -t
3MPL -o
3FSG -at
Imperfect
3 MPL -u/o9
total
N = 58
N = 3
N = 1
N = 5 (3FPL -an, N = 2)
N = 16
There were 128 total responses (16 8). In all but 2 cases the responses
conformed to the anticipated answers. The two deviant answers are
ignored here, so in all there are 126 responses. The results are summarized in (6). In (7) three typical responses are given.
(6) Stems appearing with:
8
This stands for any subject suffix that begins with a -t, including for example 1 sg,
sim-t I heard, 2 FSG sim-ti you F heard and 2 MPL man-tu you MPL prevented.
9
In the imperfect the MPL suffix is -u after a high stem vowel, bim-u they go from
the stem bimi, and -o after a low stem vowel, bilg-o they find, from the stem bilga,.
644
jonathan owens
(6)
Stems appearing with

(a)
(b)
(c)
/h =
, stem vowel kept
/h kept
(= weak verbs)
(mixed stems)
(= strong verbs)
Perfect
-t
36
-o
11
-at
2
Imperfect
-u/o
14
-an
8
Total
71
before C-initial suffixes:
36
before V-initial suffixes:
35
(7)
(a)
/h =
0
1
0
2
12
6
1
8
2
25
8
53
51
(b)
mixed stem
(c)
/h kept
Perfect
-t
ga#-t I cut
ga#a-t
Cf. weak verb ma-t I went
cf. strong verb katab-t
-o
sim-o they heard
daba-o they killed sim-o they heard
cf. ma-o they went
Imperfect
-u/o
bism-o they hear
bisma-o
bisam-o
cf. bilg-o they get
The verbs in column (a) behave like weak-final verbs, examples of which
are given in brackets in (7). The verbs in column (c) behave like strong
verbs, verbs with 3 consonantal roots. The mixed stem in column (b)
has attributes both of stems with deleted final laryngeal and of those
with the laryngeal maintained. Like the former the laryngeal is deleted;
like the latter the final stem vowel is maintained.
There is a fundamental contrast defined by the variable C- or V-initial
suffix.
Final /h disappear almost categorically before a C-initial subject suffix

(36 cases where /h = (column a), 2 where they are kept (column c)).
The verbs are then conjugated like weak-final verbs, that is, verbs with a
CVCV stem, e.g. ga#-t I cut < *ga#a-t , cf. ma-t I walked. Otherwise
the maintenance of /h is slightly dominant statistically, though only in
one instance is the dominance overwhelming, namely in the form b-at
she sold with the 3 FSG suffix -at. Whether the conditioning factor here
645
is the suffix itself or the verb stem (CC-, which I suspect is the case) is
not apparent in the data since only one -at frame was used.
As with C-initial suffixes, usually before a V-initial suffix if /h are
not used, the stem is shifted to the weak-final class, e.g. ga#-o they
cut, budb-o they M slaughter. In two cases, however the stem vowel
was kept: daba-o they m slaughtered, bitba-o they M follow. Also, in
three cases where /h are not used, stress was irregularly shifted to the
penultimate syllable, bukr-u they M hate.
None of the respondents categorically used or disposed of the final
/h, even in a discrete sub-class of forms (e.g. perfect stems, excepting
the C-initial conditioning factor). Table 1 summarizes the global scores
for individuals.
TABLE 1
INDIVIDUAL RESPONSES
Respondent
/h kept
/h dropped
4
12
2
14
9
7
2
14
9
6
13
3
9
7
4
11
For 6 of the respondents there is textual material against which their test
scores can be compared. These texts include standard interviews as well
as less formal situations. Of the six, one has no tokens of final /h in his
text, and one has only 1. The total scores from the texts, classified into
morphological context, are given in Table 2.
TABLE 2
TEXT SCORES OF SIX INDIVIDUALS
Perfect
/h

t/n
0
10
-o
6
7
Imperfect
at
5
5
u/o
10
5
an
3
0
0
6
AP
Total
V,
0
2
24
35
The contexts are the following; in the perfect: suffix -t/n 1 SG, 2, 1PL,
MPL -o, FSG -at, in imperfect MPL -u/o, FPL -an, 3 MSG with object
suffix, AP with plural suffix. An example of each (taken from texts) is as
follows, where relevant giving one example with the final laryngeal kept,
one with it absent.
dabt I slaughtered, daboh-o they slaughtered, dab-a they M
slaughtered it F (these 2 tokens from same speaker); wagaat she fell;
simata she heard him, bism-o they M hear; tugo you M fall; bimba-an
646
jonathan owens
they F are sold; basem I hear him; tbn (AP, active participle)
following-MPL (<tbi). Note that the AP and object-suffix contexts
were not used in the test frames.
In two cases the sample in the textual material is not very representative.
The FSG -at suffix is represented only on two lexemes, one speaker with
five tokens of wagaat she fell, another with five of sim-at she heard
(etc.). Also, the FPL -an occurs only in one speakers text.
In most respects the textual and test material agree: in texts h/ are
categorically dropped before the subject suffixes beginning with -t/n,
and they are similarly nearly categorically absent in the test. In both sets
of data the stems shift to the weak-final class. Similarly, in both there is a
greater degree of variability before the MPL suffix (-u/o) in both perfect
and imperfect verbs, the ratios of kept/dropped tokens being roughly
the same. In Table 3 the percentages are calculated by dividing the total
tokens of stems where /h are kept by the total of laryngeal stems with a
MPL suffix. The token count is given before the percentage.
TABLE 3
test
text
TEXT COUNT: /H KEPT BEFORE MPL

SUFFIX/DROPPED, % KEPT
kept
perfect
dropped
kept
imperfect
dropped
11
7
12
6
48
55
14
10
25
5
64
66
The texts also indicate that before object suffixes there is a categorical
shift to the weak-final paradigm, regardless of whether or not the suffix
begins with a V or C; thus daba-hin he slaughtered them F, budb he
slaughters it M, rather than dabah-hin/budbaha.108
There is also a fair degree of agreement between the test scores of
individual speakers and their treatment of /h in natural speech. Speaker
4, for instance has the lowest percentage retention of /h in the test, and is
also lowest in the text count, while speaker 6 has the highest percentage
retention in both. Only speaker 7 has a lower percentage retention in the
text than in the test.
10
After strong verbs object suffixes are suffixed directly to the final -C, with no other
changes occurring, e.g. katab-a he wrote-it, katab-hin he wrote-them. After verbs ending in a final -V the final vowel lengthens before a suffix, ligi-hin lig-hin he found
them. Before object suffixes (which were not tested in the frames like (4)) the laryngealfinal verbs shift to the weak-final class.

TABLE 4
647
TOTAL RETENTION/DELETION OF FINAL /H

IN TEXTS OF 5 SPEAKERS
Speaker
/h kept
deleted
0
1
4
11
9
2
6
11
5
8
For the present sample of speakers it can then be said that verbs with
final /h belong to a mixed paradigm: before C-initial subject suffixes
they belong to the paradigm of weak-final verbs; before V-initial subject
suffixes they are treated variably, in cases even by the same speaker,
sometimes as CVCVC stems, sometimes as CVCV, weak- final stems.
2.2
Larger sample
Looking at the larger sample of texts, as summarized in Owens (1998),

the extent of deletion in the pre-vocalic context was indicized over three
sample groups, Maiduguri interviews (N = 58), Maiduguri group conversations (N = 61) and village interviews (N = 52). Aspects of the corpus can be briefly summarized here. In the following, an index of 0%
means that the laryngeal is never deleted, of 100% that it always is. Table
5 groups all speakers together, while Table 6 divides them according to
place and sex. The following indices pertain only to the context V /h-V,
i.e. V-initial suffixes.
TABLE 5
Table 5a: Place
Village
Maiduguri
TOTAL SAMPLE
Table 5b: Sex
71%
79%
Male
Female
TABLE 6
Maid
Villages
Table 5c: Age
79%
71%
<32
3249
>49
83%
83%
61%
PLACE X SEX
Male
Female
81%
74%
75%
54%
More so than the data examined thus far, the overall scores point to a loss
of final laryngeals in pre-vocalic position, the loss being most pronounced
648
jonathan owens
among younger, male residents of Maiduguri, with Maiduguri well ahead

of the village overall among both males and females. The text scores
show a higher degree of deletion overall than do the test scores described
above. This is interesting, if the test scores indicate a higher degree of
monitoring (the reading list effect in the classical, early sociolinguistic
studies), it would indicate that the laryngeals are somehow there to a
higher degree than the test scores would show. The lower female scores
would deserve comment, if adequate comparative data were available,
which it is not. In western studies, females have consistently been shown
to lead changes. However, even if western sociolinguistics attempts to
draw far-reaching conclusions from such tendencies (Chambers 1995,
139), the present data cautions against overgeneralizing, to say the least.
Looking at the data overall, a change can be said to have occurred
in the apportionment of final verbal /h verbs: before C-initial suffixes
they merge with weak final verbs whereas before V-initial suffixes
they variably merge with weak-final verbs. Factors speak both for and
against the variable merger going to completion. In favor thereof, urban
dwellers, especially younger ones, have the highest degree of merger.
Rural dwellers, however, have a lower degree, and at this point in the
history of Arabic in Nigeria, they constitute the overwhelming majority.
Moreover, there are structural factors favoring retention: in explicit tests
the final laryngeal was retained more often than in spontaneous speech.
Moreover, in other positions, V-/h-V sequences, such as word-internally
(ahar month), are generally maintained, so at this point at least, the
high degree of variability is most marked at the right morphological edge.
Moving to the next section, the current situation will be represented as
having a split paradigm, with the possibility, that in the future laryngealfinal verbs will merge completely with weak-final ones.
Before proceeding to the final section, it is relevant to note that
an identical treatment of laryngeal-final verbs is found in Abbeche
Arabic in eastern Chad (Roth-Laly 1979, 810). This shows that the
split of laryngeal-final verbs into two morpho-phonological classes is
old enough to be a pan-western Sudanic Arabic trait. Again, however,
without more detailed study of the situation in eastern Chad it is not
possible to say more than that. If the loss of laryngeals is more advanced
in Chad than in Nigeria, one would predict that they would also be lost
eventually in Nigeria. If, on the other hand, the same complementary
treatment of final laryngeals is found as in Nigeria (governed by form of
suffix), one would rather see the split as stabilized.
649
3. Morphological Regularity, paradigmatic stability

Before continuing, it is relevant to examine the manner in which the
laryngeal-final verbs have split into two classes. As seen in (7), the laryngeal-final verbs either end in a laryngeal (column c in (7)) or in a vowel
(column a). They thus either remain in the class of C- final verbs or they
shift to the class of weak-final verbs:
(8) strong verb (C-final)
ga#aan they F cut
cf. katab-an they F write
weak verb (V-final)

ga#ti you FSG cut
cf. ma-ti you FSG went
What happens only rarely is that a new morphological class is created,

one with properties of both the strong and weak verbs. In (6) there are
two such tokens, attesting to the fact that such forms do in fact occur.11
They are given in column b; both tokens occur before V-initial suffixes.
A form like daba-o they M killed derives from dabah-o with deletion
of the final laryngeal. Rather than switch to another stem class, which
is what usually happens when the final laryngeal is deleted (dab-o), the
final stem vowel has been maintained. In essence, a new morphological
class has been created. This class, however, has not become widespread
or established in the speech community.
Looking at the development structurally, the deletion of the final
laryngeal leads to the splitting of /h-final verb stems into two preexisting morphological classes, the distribution of these being broadly
defined by morpho-phonological context. What did not happen is that
the loss of the final laryngeal in one context led to the development of a
new morphological class based on the stem CVCV-suffix. Morphological
stability takes precedence over forms, which would be derived by the
logic of phonological rules.
The WSA developments would ostensibly appear to support the idea
of linguistic history repeating itself: final laryngeals in Arabic verbs
tend to drift into weak-final ones. This repeats a similar process, which
occurred in pre-diasporic Arabic, as illustrated in (1) in section one
(bada-t/bad-t). Under one interpretation of Arabic language history,
one could this add drift to the catalogue of features characteristic of New
Arabic (allowing for the cautionary note 1 above).
11
In the texts from speakers from the present sample no mixed forms occur. In other
texts, however, they are attested.
650
jonathan owens
However, the data allows for a more nuanced description. The change
described for WSA can be broken down into two parts. On the one hand,
what may be termed a principle of paradigmatic stability, one aspect of
morphological stability, can be invoked:
(9) Principle of paradigmatic stability: do not create new paradigms
In its categoricality this statement is clearly too strong, and one can imagine adding many conditions to it, but it serves present purposes. Both
the Old Arabic and the WSA data obey the principle: the loss of a final
laryngeal in both cases did not lead to the creation of a new paradigm.
Rather, the laryngeal-less forms simply collapsed, or in the case of WSA,
are still in the process of collapsing, into already-existing paradigms.
The other part is the phonological change that creates the condition
for the collapse into pre-existing paradigms. This change is one of
happenstance. In WSA, /h/ became /h/ and // became //. This type of
change is not unique among varieties of Arabic. // has moved to // in the
Tihama, /q/ appears as // in Cairene, Damascene, as well as elsewhere.
Maltese presents a complicated picture of its own. and // merged in
//, which in turn was lost, leading, as in WSA, to the merger of *//-final
verbs with weak finals.129
(10) sm-ayt
tf-ayna
I heard
we threw (< df)
Why the changes occurred in WSA is, frankly, not clear at this point, as
is the question, why in some dialects and/or h continue on to . One
12
In Maltese the final -a of the suffix is conditioned by the historical pharyngeal, and
hence contrasts with, say, bn-eyt I built, where the suffix goes back to the diphthong
*ay-t.
In fact, the historical phonology of Maltese remains to be worked out in detail. Not
least is the problem of a not inconsiderable dialect variability, with its potential importance for historical reconstruction (cf. Owens 2006, chapter 7 on imla in Maltese).
Mifsud (1995, 308-9) explains the final /y/ in sm-ayt etc. as a change of the historical
pharyngeal trace to /y/. This analysis is interesting in and of itself, but probably deserves
an article of its own. Briefly, while Mifsuds analysis still maintains the paradigmatic
stability principle, it is on an a priori basis more complicated than the treatment offered
here. Mifsud notes that in general verbs with historical final voiced pharyngeals merge
with weak-final verbs (e.g. nitfa we throw, like ninsa we forget). The current analysis sees the merger as having occurred throughout all inflectional paradigm members,
allowing for the underlying phonetic conditioning of [a] due to the pharyngeal trace.
Mifsuds analysis would split the paradigm of voiced pharyngeal-final verbs. In the process this creates an otherwise unattested CCay-C stem, where the /y/ represents C3 of the
root, and hence would contrast with the split paradigms-analysis offered above for WSA,
since in the present analysis laryngeal-final verbs split into two existent paradigms.
651
can, of course, appeal to sub-stratal influence. This is quite plausible in

the case of WSA. However, why should the shift have occurred in WSA,
but not in another notable Sprachinsel where Arabic is a remarkable
minority language, Uzbekistan?13 In short, the phonological shift to /h/
and // and the further loss of these sounds, is history in its truest sense,
contingent on particular developments in particular places, in particular times. This contingency is in evidence in WSA in the very variability
of the process. It is unfortunate that longer term trend studies cannot
be followed through on to trace the further development of this phenomenon. At what point, if ever, does the change go to completion, and
why?
In any case, once these sounds are given up, the paradigmatic stability
principle takes over. This obtains in all varieties of Arabic that have (1)
above, in WSA, as described here, and in Maltese as well, as suggested
around the discussion of (10) above.
In short then, the sequence of developments is as follows:
(11) loss of final laryngeal
(12) obeisance of the paradigmatic stability principle (see (9) above)
In other words, given the loss of the laryngeal, collapsing into the weakfinal paradigm follows automatically. In the current framework, only
(11) is history proper. (12) is suggested to be the instantiation of a general linguistic principle in Arabic verbal morphology. In a sense, so long
as Arabic exists, (12) will be operative in the way described here. Its
reappearance at different points in the history of Arabic, however, is not
compatible with a conceptualization of Arabic as having changed from
and Old type to a New type. To the contrary, because the same structural forces are at work in pre-diasporic times as are at work in 2006, no
change has occurred.
4. Appendix
Verbs used in test frames (given in the form of a possible answer): dbaho
they slaughtered; simt I heard; wjt I faced; karaho they hate; ga#t
I cut; garti you FSG stopped; mantu you MPL prevented; binfaan
they F are useful; budbaho they M slaughter; bisaman they F hear;
13
Moreover, in WSA emphatic sounds are fully maintained, but they are lost in
Uzbekistan Arabic.
652
jonathan owens
tuwjuhu you MPL face; bikrahu they M hate; bitbao they M follow;
bat she sold; bibu they M sell; ga#ao they M cut.
5. References
Behnstedt, Peter. 1985. Die nordjemenitischen Dialekte. Teil 1: Atlas. Wiesbaden:
Reichert.
Chambers, Jack. 1995. Sociolinguistics Theory. Oxford: Blackwell.
Ibn Mujhid, Ahmad b. Ms. 1972. As-Saba f l-qirt. Ahmad awqi D ayf, ed. Cairo:
Dr al-Marif, 1972.
Mifsud, Manwel. 1995. Loan Verbs in Maltese. Leiden: E.J. Brill.
Owens, Jonathan. 1998. Neighborhood and Ancestry: Variation in the Spoken Arabic of
Maiduguri, Nigeria. Amsterdam: Benjamins.
. 2006. A Linguistic History of Arabic. Oxford: Oxford University Press.
Rabin, Chaim. 1951. Ancient West Arabian. London: Taylors Foreign Press.
Roth-Laly, Arlette. 1979. Esquisse grammaticale du parler arabe dAbbeche. Paris:
Geuthner.
Versteegh, Kees. 1984. Pidginization and Creolization: the Case of Arabic. In Amsterdam:
Benjamins.
. 2004. Pidginization and Creolization Revisited: The Case of Arabic. In Haak,
Martine, Rudolf de Jong, and Kees Versteegh, eds. Approaches to Arabic Dialects:
A collection of articles presented to Manfred Woidich on the occasion of his sixtieth
birthday. Leiden: Brill, 343358.
SOME ASPECTS OF DIGLOSSIA AS REFLECTED IN THE

VOCABULARY OF LITERARY AND COLLOQUIAL ARABIC
Judith Rosenhouse
Swantech Ltd., Haifa, Israel
1. Introduction
1.1
Preliminary remarks
This paper focuses on diglossic differences between literary and colloquial Arabic mainly from the morpho-phonological and lexico-semantic
aspects.
The lexical part of the language is probably the area that most clearly
distinguishes between these varieties of Arabic. Our survey deals with
phonology, morphology and lexicon/semantics.
After the introduction, the second part deals with phonological
features that distinguish literary Arabic from colloquial Arabic (e.g.,
consonants such as /q, j/ and /t/, the vowel of word-final imla, and the
emphatics). The main differences between the two varieties of Arabic
reflect the well-known typical segmental features of colloquial Arabic.
Next, morphological or morpho-phonological comparison between the
phonological system of cognate lexical items in colloquial Arabic (as
spoken in Israel) and literary Arabic is undertaken.
Literary Arabic uses a considerable number of lexemes in their
foreign forms often adapted to Arabic morpho-phonology ; in such cases
colloquial and literary Arabic lexemes are often cognate. We look briefly
into the lexical areas where these phenomena occur.
These features can be described within language interference or
language contact as terms of one linguistic approach, or as code mixture
according to another. From this point of view, these features do not differ
much from phenomena in other Arabic dialects (Boucherit, 2000, Heath
1989, Owens, 1998, Taine-Cheich, 2000) or languages (e.g., Ritchie and
Bhatia, 2004).
654
1.2
judith rosenhouse
Introductory remarks
The centuries-long development of modern literary Arabic, elsewhere

also Modern Standard Arabic, and colloquial Arabic, the two faces
of the Arabic language, and the differences between literary Arabic
and colloquial Arabic, have been the topic of much linguistic debate
especially since Fergusons (1959) well-known paper on diglossia (e.g.,
Blau, 1977, Blanc, 1960, Bousaffara-Omar, 2005, Diem, 1974, Fernandez,
1993, Fck, 1950, Haeri, 2003, Holes, 1995, Kaye, 1994, 2002, Mejdell,
2002, Meiseles, 1975, 1980, Monteil, 1960, Stetkevych, 1970, Versteegh,
1997). As this area is also of much interest for Professor Kees Versteegh
I dedicate this paper to him.
Literary and colloquial Arabic reveal differences on all linguistic
levels, including phonology, morphology and syntax, in addition to
vocabulary, as described in Ferguson (1959). Our recent work on
trilingual dictionaries (Rosenhouse, 2001, 2004) has given us a genuine
and tangible feel for this situation from the lexical aspect of Arabic
diglossia in Israel, and the present paper is partly based on items in
those volumes.
Some random examples of such differences are shows in Table 1
below. We take into account the relevant literature, and rely on a lexical
search of several Arabic newspapers from which we gathered examples
for the studied points.
TABLE 1
EXAMPLES OF LITERARY AND COLLOQUIAL ARABIC

NON-COMPATIBLE LEXEMES
Literary Arabic
Colloquial Arabic
Gloss
dumya
ihd
ihtada
zujj
kura
zawraq
lube
tarh
libes el-kundara
qazz
tbe
flka
Puppet
Abortion
To put shoes on
Glass
Ball
Boat
The situation of the Arabic language has recently been described as

follows in Ayyoub: In societies where a substantial segment of the
population is literate, and two variants of language usage exist, the gap
between the literary/written and spoken Arabic tends to diminish. With
the gap disappearing, a continuum emerges, offering the speaker a rich
array of possibilities (2002, see also Kaye 2002). Such a mixed language
diglossia as reflected in the vocabulary of arabic
655
can be examined at present from oral material, or from written texts

dating from the 7th century to the present (e.g., Blau 1977,1 and see
Chapter 1 in Mazraani 1997, for a review of relevant literature). For
Arabic, this mixture is known also as middle Arabic (al-lua al-wust,
Badawi 1973).2
The importance of the lexical aspect in the distinction between literary
and colloquial Arabic is partly due to the sheer number of the lexical
items, which naturally far exceed the number of grammatical structures
and phonological categories.
Our research hypothesis and therefore starting points are based on
two assumptions: 1. More literary Arabic lexemes (and more loanwords
from various languages) are used in colloquial Arabic, than colloquial
Arabic lexemes are used in literary Arabic (cf., e.g., Al-Falay 1996,
chapter four). This apparently is because literary Arabic often uses loan
translation (calque) or new terminology coined by official authorities
such as academies of Arabic language in Egypt and elsewhere (e.g.,
Shraybom-Shivtiel 2005) instead of loanwords. 2. When foreign lexemes
are used in literary Arabic or colloquial Arabic, they usually adapt to the
Arabic system phonologically, and also morphologically when necessary
and possible.
1.3
Literary and colloquial Arabic cognate words
Literary and colloquial Arabic cognate words vary mainly in phonological and/or morphological features and not semantically (Cadora 1979).
Most of the colloquial Arabic examples referred to below reflect colloquial Arabic as spoken in Israel. The picture we get from this country may differ somewhat from that in other countries. In fact, literary
Arabic is not uniform either, and literary Arabic lexemes vary in different
1
Several views of middle Arabic do not consider it a separate variety or a variant of
the language, for it lacks a grammatical coherence: we simultaneously find correct forms
of the literary Arabic and deviations from these forms. We even find hybrid forms that
are neither literary nor dialectal (Ayyoub, 2002. Cf. e.g., Kaye, 2002, Mejdell, 2002).
2
Diglossia is also related to bilingualism, since two language systems are dealt with
(cf. Fasold, 1984, Rosenhouse and Goral, 2004). Thus, when speakers use colloquial
Arabic lexemes in a literary Arabic context, or vice versa, they may do it either for specific functional (stylistic) purposes or as an unintentional process of code switching.
Although we do not deal with code switching in this paper, I believe that the use of literary Arabic words in a colloquial Arabic context is not to be considered code switching
or code mixing if the literary Arabic word is already integrated in colloquial Arabic and
used in it as the only accepted lexeme for that specific notion.
656
judith rosenhouse
Arabic-speaking countries, in particular in the fields of technical

and scientific vocabulary (Holes 1995, Badawi 1997). For simplicity,
we refer to literary and colloquial Arabic as distinct and independent
systems.
The following sections of the paper focus on a comparison of
contemporary literary and colloquial Arabic vocabulary as reflecting
Arabic diglossia, and not on their diachronic development. After a short
description of phonological and morphological features of literary and
colloquial Arabic vocabulary ( 2), three aspects relating to semantics
are examined:
(1) the use of foreign lexemes in literary and colloquial Arabic ( 3.1);
(2) the use of colloquial Arabic words in literary Arabic and literary
Arabic words in colloquial Arabic ( 3.2);
(3) semantic differences between cognate literary Arabic / colloquial
Arabic words ( 3.3). Section 4 concludes this essay.
2. Phonological and morphological differences between literary

and colloquial Arabic
The brief description of phonological and morphological differences
between literary and colloquial Arabic lexical pairs in this section is
intended to demonstrate processes that distinguish literary from colloquial Arabic. These processes appear in the oral articulation of literary
and colloquial Arabic and mark them clearly. Ferguson (1959) considered such lexical literary / colloquial Arabic pairs as an important feature of diglossia. For a detailed phonological and morphological analysis
of colloquial / literary Arabic lexical items in the discourse of speakers
from three countries (Egypt, Iraq and Libya) see Mazraani (1997).
2.1
Phonology
Phonological differences between literary and colloquial Arabic in Israel

and elsewhere are among the most studied and noted features in the
literature of Arabic diglossia (e.g., Altoma 1969, Blanc 1960, Meiseles
1980, Fischer and Jastrow 1980, Holes 1995, Mazraani 1997, Rosenhouse
1984, Versteegh 1997). Usually, speakers of a colloquial Arabic dialect
use the same consonants for literary Arabic speech as for colloquial
Arabic. The emphatics /s, d, t, / and the somewhat language-specific
657
pharyngeals /, h /, as well as plosives, fricatives, lateral and semivowels are pronounced in the same manner in both (spoken) literary and
in colloquial Arabic (at least in Israel), and therefore are not discussed
here. That is, most speech sounds are shared by the phonological systems of literary and colloquial Arabic. In Israel as in other dialects (see
Kaye and Rosenhouse 1997), the phonological differences between literary and colloquial Arabic refer mainly to uvular /q/, the three interdentals /t, d, z/, and the allophones / - j/ for Semitic *g. The phoneme /k/
has a palatalized variant // used in certain rural and Bedouin dialects
in Israel, but its use seems to be somewhat dwindling (at least in Israel)
due to the effects of prestigious urban colloquial Arabic (in Israel) dialects and the spread of literary Arabic due to school education. Another
case is that of the emphatic /d/ which merges with // in certain (rural)
dialects. Sometimes such dialect speakers do not merge these phonemes
in literary Arabic. The vowel system of many colloquial Arabic dialects
(in Israel) includes two vowels that do not exist in literary Arabic: /e, o/,
and the parallel long monophthongs /, / which usually reflect the literary Arabic diphthongs /ai, au/ respectively. When speaking in literary
Arabic, the traditional literary Arabic articulation of these diphthongs,
as diphthongs, is usually retained in Israel.
Let us discuss now the problematic phonemes of the system as they
occur in Israel. Mainly Muslim, Christian, and Druze speakers of rural
dialects pronounce the phoneme q as uvular. In urban dialects this
phoneme is articulated as glottal // and in Bedouin dialects and some
rural dialects its pronunciation varies, often under certain conditions,
between /g/ and /j/. However, Bedouins and urban speakers do
articulate the phoneme /q/ as uvular when reading the Qurn or when
certain literary Arabic lexemes are used in words such as al-qurn the
Koran, qadiyye problem, issue, qawmiyye nationalism, or dimuqrtiyye
democracy3 (cf. Mazraani 1997, Haeri 2003). A similar process refers to
the other phonemes under discussion here: the colloquial Arabic dialects
vary in their articulation between the urban /d-d-z/ for /z/ and /s, t/ for /
t/, and the respective traditional literary Arabic articulation /d/ and /d/
which simultaneously characterizes rural and Bedouin dialects. The four
basic emphatics /s, d, t, z/ (especially at word-final position) are often
3
In non-phonetic dictionaries the transcription usually hides such dialect differences. Also in our dictionary only one letter transcribes q in colloquial Arabic.
658
judith rosenhouse
pronounced as non-emphatic phonemes in some colloquial Arabic

dialects (in Israel), mainly urban ones, as in /s/ for /s/ in colloquial Arabic
/sand/ for literary Arabic/sundq/ box (see Rosenhouse 2002). These
phonemes are, however, transcribed in colloquial Arabic dictionaries
(in Israel) without consideration of the option of this non-emphatic
articulation. Colloquial Arabic (in Israel) has, on the other hand, a
few emphatic consonants that do not exist in literary Arabic, and are
usually phonetic variants of non-emphatic phonemes. The phenomenon
is marginal in rural and urban colloquial Arabic in Israel and is more
prevalent in Bedouin dialects (in Israel and elsewhere). But it may form
phonemic distinctions in certain words such as /b$ba/ pope vs. /b$%ba/
her door, or /a&&a/ God vs. /al-la/ he said to her.
The vowel /a/ of the feminine ending in noun and adjective forms,
which in modern literary Arabic is pronounced /a/, may be /a, e/ or
/i/ in colloquial Arabic (in Israel), with differences depending on
dialect features and phonetic factors. The phonetic condition of urban
and rural dialects in this region is that if the consonant preceding
the suffix is pharyngeal, laryngeal, emphatic, or uvular, the vowel
remains /a/; cf. /sa/ hour, clock; otherwise it bends upward (imla
of classical Arabic), as in e.g. /sane, sini/ year. The /a/ is also a feature
characterizing (colloquial Arabic) Bedouin dialects. Thus this suffix
uses the vowel /a/ in formal (literary Arabic) and Bedouin speech, but
also in (non-Bedouin) colloquial Arabic when preceded by a back or
guttural consonant. The difference in the use of /e/ or /i/ in this ending is
less easy to define, because although it is sometimes apparently dialectdependent (see examples in, e.g., Geva-Kleinberger 2004) it often reflects
free variation and is sometime also a more central /i/. Consequently,
four contexts of articulation of the t marbta as /a/ (rather than /e/
or /i/) can exist: (1) formal, literary Arabic; (2) a Bedouin dialect; (3)
phonetically-conditioned urban/rural colloquial Arabic (in Israel); or
(4) a mixed/switched utterance using literary Arabic pronunciation
within a colloquial Arabic structure (in Israel).
The features briefly described here for colloquial Arabic (in Israel) are
similar in principle to those in many other colloquial Arabic dialects,
including Mesopotamian q'ltu and g'l't dialects (Blanc 1964, Jastrow
1978).
Unlike consonants, the vowel system differs in colloquial Arabic
from literary Arabic almost in all the Arabic dialects. This is expressed
by expansion of the phonetic vowel system on the one hand and the
merging of several vowels to a smaller set of phonemes on the other
659
hand (the latter is conspicuous in North African dialects). The wider the
differences between the traditional literary Arabic phonological system
and that of daily colloquial Arabic, the more literary Arabic articulation
errors may be found in the speech of such colloquial Arabic speakers.
This also is a typical feature of the code-mixing phenomenon. Examples
abound in any recorded colloquial or literary Arabic text (e.g., Blanc
1960), and even presidential speeches (see Mazraani 1997).
2.2
Morphology
Morphological differences between literary and colloquial Arabic (in

Israel) exist in various categories. The pronominal system exhibits wellknown differences in the free and bound pronouns in the nominal and
verbal systems. See examples in Table 2.
TABLE 2 EXAMPLES OF MORPHOLOGICAL DIFFERENCES
BETWEEN LITERARY AND COLLOQUIAL ARABIC
Literary Arabic
Colloquial Arabic
Gloss
nah nu
antum
druki
drukunna
katabtu
taktubna
nih na, ih na

intu(m)
drek
drku/drkun/drken
katabt
(b)tukutbi/(b)tikitbi
we
you, pl. m.
your sg. f. home
your pl. f. home
I wrote
you sg. f. write/will write
The suffixed object bound pronouns also differ: literary Arabic talabtuka
vs. colloquial Arabic (in Israel) talabtak I requested you, m. sg., literary
Arabic fatah ahu vs. colloquial Arabic (in Israel) fatah o he opened it, etc.
In the nominal system we note literary Arabic af al vs. ifal in several
local dialects of colloquial Arabic (in Israel), e.g., literary Arabic aswad
colloquial Arabic (in Israel) iswad black in the colors groups. Another
example is the female form of some adjectives, which in literary Arabic
are of the fal pattern whereas in colloquial Arabic they are regular,
i.e., with the t marbta, e.g. literary Arabic sakrn colloquial Arabic
(in Israel) sakrne drunk. Differences between literary and colloquial
Arabic (in Israel) also occur in plural patterns4 as the examples in
Table 3 show.

4
We refer to those lexemes where the singular is identical or at least cognate in both
literary and colloquial Arabic.
660
judith rosenhouse
TABLE 3 EXAMPLES OF DIFFERENCES BETWEEN PLURAL FORMS

IN LITERARY AND COLLOQUIAL ARABIC (IN ISRAEL)
Literary Arabic
Colloquial Arabic (Israel)
Gloss
br
ahur, uhr
numr
nuqs
bah h rna
byra
(u)hr
nmura
nawqes
bah h ra
wells
months
tigers
defects, faults
seamen
Such differences are, of course, in addition to basically phonetic

differences of cognates, existing in singular as well as in plural forms
see Table 4.
TABLE 4
PHONETIC DIFFERENCES BETWEEN COGNATE LITERARY

AND COLLOQUIAL ARABIC LEXEMES
Literary Arabic
Colloquial Arabic (Israel)
Gloss
ism
far
mintaqa
ibar
asnn
daqiq
astida
usum
fr
mantiqa
ubar
snn
daqyeq
astze
name
mouse
region
needles, injections
teeth
minutes
gentlemen, professors
In the verb system also certain verb patterns vary between literary and
colloquial Arabic (in Israel). See regular verb patterns such as literary
Arabic qadaracolloquial Arabic (in Israel) idir was able, literary
Arabic arafacolloquial Arabic (in Israel) iref knew, and their
respective non-past patterns: literary Arabic yaqdirucolloquial Arabic
(in Israel) yidar will be/is able, literary Arabic yarifucolloquial
Arabic (in Israel) yiraf will know/knows. Differences are also found in
verbs of C1//, e.g., literary Arabic yaxudu- colloquial Arabic (in Israel)
yxud/yxud/yxed will take/he takes/, C2w/y/ pattern, e.g., literary
Arabic xiftucolloquial Arabic (in Israel) xuft I was afraid, C3w/y
pattern, e.g., literary Arabic baqiyacolloquial Arabic (in Israel) baqa/
biqi he remained, literary Arabic qaraacolloquial Arabic (in Israel)
qara, qiri he read, and many more.
Differences of these kinds are well known from the literature, and in
a dictionary presenting both literary Arabic and colloquial Arabic (in
Israel) (such as Rosenhouse 2001, 2004) one at once sees the abundance
661
of these forms. However, the main goal of a dictionary study is word

meanings, and to this issue we now turn.
3. Semantic considerations of the differences between literary Arabic

and colloquial Arabic lexemes
3.1
The use of foreign lexemes
Like other languages, Arabic has absorbed many lexical items from
foreign languages from its classical past to the present. Since the second half of the 20th century the major source of foreign loanwords in
Arabic has been the American English language, which has become the
modern lingua franca of the world. Before that, French, Italian, Spanish,
Turkish, Persian, Latin, Greek and others made lasting impacts on Arabic vocabulary (Rosenhouse, to appear).
Literary Arabic was revived in the 19th century through the diverse
efforts of writers, journalists, scientists and linguists. These efforts led
to the establishment of academies for the Arabic language in Damascus
(1919), Cairo (1934), Baghdad (1947) and Amman (1976). Coining new
terminology for modern notions is one of the declared tasks of these
academies, as it is of many language academies elsewhere. The linguistic
methods applied include borrowing terms and integrating them into
the language phonologically and morphologically; extending meanings
of existing (Arabic) words; analogical creations based on existing
roots; translation of foreign words into Arabic (calque); and blending
(Stetkevych 1970, also quoted in Suleiman 2005, Shraybom-Shivtiel
2005). Such Arabicizing of foreign terminology goes on in other Arabicspeaking countries too (e.g., Al-Qahtani 2002, for Saudi Arabia, and
Kharbush 2002, for Jordan see Badawi 1997). This activity has added
thousands of new words to the literary Arabic vocabulary, although
there is no full agreement between the various language institutes
(whatever their titles) in Arab countries and although many of these new
words remain buried in dictionaries without enjoying general public
use (Badawi, 1997). Whatever the distribution of lexical innovations
in literary Arabic vocabulary, this process clearly contributes to the
decrease in the number of foreign terms in literary Arabic, particularly
in the semantic fields of modern technology and science.
Unlike literary Arabic, colloquial Arabic is not subject to any official
language policy and it develops spontaneously, albeit under the effect of
internal and external circumstances. Accordingly, foreign terminology
662
judith rosenhouse
flows freely into colloquial Arabic and is integrated in keeping with

the needs of its speakers (see Rosenbaum 2000/2). As noted, this
terminology now comes mainly from English, but residues of the effects
of other languages from earlier periods can still be found.
In Israel the situation is somewhat different from Egypt (as Rosenbaum
ibid. describes it), since in Israel Hebrew is the dominant language
(Rosenhouse forthcoming). Since the establishment of the state of Israel
in 1948 native speakers of Arabic in Israel have been in contact with
Hebrew officially and spontaneously. Since they use Hebrew daily, the
latter has a strong impact on their Arabic mother tongue. The fact that
they study Hebrew at school from grade three to the end of high school
reinforces their Hebrew proficiency. Hebrew has been the major source
of borrowing of foreign lexemes for colloquial Arabic in Israel for the
last five decades or so. English lexemes (including American-English)
also penetrate colloquial Arabic directly, or through Hebrew that often
serves as mediator (see, e.g., Abdeen 2004, Amara and Spolsky 1986,
Koplewitz 1989). A simple dictionary search soon reveals many cases
where literary Arabic uses Arabic words (often modern innovations)
while colloquial Arabic (in Israel) uses foreign words for the same items
(see Table 5). Literary and colloquial Arabic also share identical foreign
loanwords. Some of these words in colloquial Arabic (in Israel) may be
transferred directly from the foreign language, or indirectly via literary
Arabic, in particular when they refer to higher and modern lexical
domains such as politics, religion or other cultural areas, e.g., ittisl
communication, often by phone, ijr:t procedures (in court), asm
muddda antigens, isti niyy artificial, ihtiyt reserves, ih s iyyt
statistics, awz safar passport (see more examples in Table 6). The
best sources for such examples are newspapers and weekly magazines,
which usually use the literary Arabic variety nearest to middle Arabic.
On the other hand, a large share of lexemes of basic daily colloquial
Arabic (in Israel) vocabulary refers to proper names of products and
place names, which are not usually translated into Arabic, e.g., jeans,
coca cola, jeep.5
5
Usually, such words are not translated into any other language, though they may be
adapted to them phonologically and morphologically, as described above.
663
TABLE 5 EXAMPLES OF ARABIC LEXEMES IN LITERARY ARABIC

AND FOREIGN LEXEMES IN COLLOQUIAL ARABIC
(FOUND IN ROSENHOUSE 2001)
Literary Arabic words
Foreign colloquial
Arabic words
Translation
art an-nr
atra
amat al-itil
tabtba
ajala ihtiytiyya
miqwad
muh arrik
mujhir
funduq
fyz
sandw
bgiyye
rakt
spr
strin
matr
mikroskb
otl
(electric) fuse
sandwich
plug (in a car)
racket (in tennis game)
spare (tiyre in a car)
steering (in a car)
motor
microscope
hotel
TABLE 6 EXAMPLES OF IDENTICAL FOREIGN LEXEMES IN

LITERARY AND COLLOQUIAL ARABIC (FROM ROSENHOUSE 2001,
AND AL-AHRM, 4/3/2006)
Literary Arabic
Colloquial Arabic
Gloss
al-aksada
influenza at-tuyr
al-interbank ad-dlriyy
arf istitl ar-ray
al-aksade
Id.
Id.
Id.
aristuqrtiyya
bank
bra
brtn
btagz
daktr
ad-diktafn
dmuqrtiyy
elektrniyya
mashariyya
anarl
hmglbn
istd
ihz al-mbil
Id.
Id.
Id.
Id.
Id.
Id.
Id.
Id.
elektrniyya
fza
Id.
Id.
istd
belefn6
oxidation
bird influenza
The Dollar Interbank
archives of public
opinion polls
aristocracy
bank
beer
protein
gas stove, oven
doctor
dictaphone
democratic
electronic
vase
general
hemoglobin
stadium
mobile phone
Based on the Colloquial Arabic term for this device.
664
judith rosenhouse
TABLE 6 (CONT.)
Literary Arabic
Colloquial Arabic
Gloss
malyn
malynr
mawqi gugl
milyr
mda
miknkiyy
msq klsk
tmtk
al-tizm
ar-rtr
sln br
sekretra
sigr
simfnia
sinryu
slkn
sirmik
as-sundq al-aswad
amb
slaidt
fniyya
klta
talifn (also htif)
zarfa
Id.
Id.
Id.
Id.
mda
mekanki
msq klskiyye
tmtki
Id.
Id.
Id.
Id.
Id.
Id.
Id.
silikon
Id.
es-sandq el-iswad
Id.
Id.
Id.
uklta
talafn
zarfe
million
millionaire
Google site
milliard/billion
fashion
mechanic
classical music
automatic
autism
Rotary (club)
saloon bar
secretary
cigar
symphony
scenario
silicon
ceramics
the black box in a vehicle
shampoo
slides
Chauvinism
chocolate
telephone
giraffe
3.2 Literary Arabic words in colloquial Arabic and colloquial Arabic

words in literary Arabic
Diglossia has been often described in rather idealistic terms, as comprising two varieties, literary Arabic and colloquial Arabic. However,
more and more voices (in the 20th century) describe Arabic on a continuum of various degrees and types of intermingling elements of these
two poles or varieties (see Ayyoub 2002, Badawi 1973, Blanc 1960, Diem
1974, Ferguson 1996, Fischer and Jastrow 1980, Meiseles 1975, 1980,
Mitchell 1986, Kaye 1994, Holes 1995, Versteegh 1997, etc.).
Since literary Arabic is the prestigious variety of Arabic, while colloquial Arabic is hardly deemed (by its native speakers, even today) to
merit any official attention or reference, colloquial Arabic speakers will
naturally tend to borrow from literary Arabic more than literary Arabic
665
(writers or speakers) would tend to borrow from colloquial Arabic.7 This

prestige factor is also well known in language communities that borrow
from other languages (e.g. Owens 1988, Thomason and Kaufman
1988/1991, Thomason 2001, Weinreich 1967).
Moreover, since communicating in literary Arabic is obviously more
difficult for colloquial Arabic speakers than communicating in colloquial
Arabic, both lexical and grammatical elements from colloquial Arabic
slip into these speakers literary Arabic utterances by code-mixing and
codeswitching processes (cf. Haeri 2003, Meiseles 1975, 1980).
The varieties developed in Arabic (typical of the 20th century) are
entitled al-lua al-wust middle Arabic (Meiseles 1975, 1980, Holes
1995, Kaye 2002, Versteegh 1997), and educated spoken Arabic
(Badawi 1973, El-Hassan 1978, Mitchell 1986), among others. It has
been observed that the modern version of Middle Arabic establishes
bridges between literary Arabic and dialects (Ayyoub 2002). Moreover,
the deliberate use of dialect within the written text has resulted not in
the dialect as a distant variant but as a level of literary Arabic (Ayyoub
2002; see also Rosenbaum 2000/2, 2004, and Somekh 1991, 1993).
A major field of differences between literary and colloquial Arabic
vocabularies concerns the semantic fields of the items. literary Arabic
has been used for high scholarly fields since classical times, and still
today it is the carrier of modern scholarly terminology in the realms
of religion, culture, science and technology, namely areas not typically
considered part of the colloquial Arabic world (see examples in section
3.1). However, with the integration of these modern topics and concepts
in the speakers daily life, their lexical elements (from literary Arabic)
penetrate the daily discourse, via the mediating stage of code mixing
until they integrate in colloquial Arabic and become an inseparable part
of its vocabulary.8
An important part of colloquial Arabic oral communication and
vocabulary is often described as dealing with emotions and personal
needs (also in Holes 1995, Versteegh 1997). This is complemented by the
fact that personal topics are not normally discussed in literary Arabic.
7
That colloquial Arabic borrows and uses literary Arabic items more than literary
Arabic borrows from colloquial Arabic is readily seen in any dictionary.
8
For more examples of such lexemes see Diem 1974, 4647, section 9.54, and Abd
Al-ls introduction (1971, 7) to his dictionary of Colloquial Egyptian Expressions
Whose Origins are Arabic (i.e., cognates).
666
judith rosenhouse
Although literary Arabic certainly has ways of emotional expression, as

written literary Arabic prose and poetry show, the general attitude to
emotion in literary Arabic, even in modern writers, is still generally not
as warm as their attitude to the flowing and rich colloquial Arabic style.
The difference between the serious, preaching style of literary Arabic
and the spontaneous colloquial Arabic speech has been demonstrated
and explained, for example, through the simultaneous use of literary and
colloquial Arabic in three Arab leaders speeches (Mazraani 1997). Arab
writers often complain that literary Arabic vocabulary and structure
need to be facilitated and made more natural, supple and personal,
like colloquial Arabic. For example, Tawfiq al-Hakim and Mahmud
Taymur write in the beginning of the 20th century about their struggle
to make their works natural by using colloquial Arabic words and
expressions (see for example Diem 1974, 117 ff.),9 and Mikhail Nuayma
(among others) puts in quotation marks the colloquial Arabic words
he uses in his early stories (e.g., sanatuha al-jadda, sat dahab), thus
marking their difference from the main story style. As a result of this
distinction, however, many Arab writers end up using colloquial Arabic
in their worksin drama, in novels and short stories, and always in
caricaturesto make their expressions more realistic as well as lively. In
Egypt this spontaneous development has reached the level of a writing
system (Rosenbaum 2000/ 2002, 2004).
Literary Arabic has developed and is now much more fluent and
easy than it was at the turn of 19th century, when the revival movement
was gaining power (Haeri 2003). Still, the general attitude to colloquial
Arabic and literary Arabic has remained largely unchanged.
3.3 Cognate literary/colloquial Arabic words and semantic differences
between them
The basic and largest part of the vocabulary in Arabic is shared by literary and colloquial Arabic, and includes pronouns, kinship terms, body
parts, social structure, animals, geographical objects, basic activities
(verbs), etc. These lexemes are cognate items, namely the same words
(expressed in the same consonantal roots), used in different forms
9
From another angle, Frayha (1955) describes his sons frustration and problems in
answering a simple question in literary Arabic. Frayha (1959) later suggested a method
for accomplishing the simplified language (although later on he stopped publicizing this
position).
667
(inflections, derivations) in literary and colloquial Arabic. The main differences between literary and colloquial Arabic cognates seem to be not
semantic but phonological or morphological (see section 2 above), and
lexemes where the semantic notion is not shared by literary and colloquial Arabic are relatively few. Cadora (1979) analyzes the lexicon in the
Syro-Lebanese region, to which colloquial Arabic (in Israel) belongs,
and we therefore refer to it here. Cadora realizes that Swadeshs list
(1952), on which his work is based, includes words whose meanings are
so general that cultural innovation or discontinuation does not affect
them greatly (26), which reduces their validity. Still, he finds (27) in the
Syro-Lebanese varieties checked against this vocabulary list, that 151
(75.5%) of the 200 lexical items have non-contrastive compatibility (i.e.,
more or less, lexical similarity) This rate is even higher (165 or 82.5%) if
the dialect of Dr Ez-Zr is removed from the analysis. The comparison
of literary Arabic and Syro-Lebanese colloquial Arabic varieties reveals
an even higher compatibility, which on the average reaches 91% (29).
Comparing the studied Syro-Lebanese dialects (except Dr Ez-Zr)
among themselves, Cadora finds 96% non-contrastive compatibility
between each pair of dialects (32). The differences between these dialects
and more distant ones (Casablanca, Cairo, Jidda, Baghdad) are larger, of
course, but lets keep to our region.
Cadora prepared another list, based on Swadeshs list (1952), and an
unpublished list by Ferguson and Said (1958), with his modifications of
them. This new list finally had 101 contrastive compatible items. From
the results of this work, on the whole, he finds that Arabic tends to use
cognate words in literary and colloquial Arabic, while different words
(rather than cognates) occur in cases where the meanings differ between
literary Arabic and colloquial Arabic.
Differences between literary and colloquial Arabic cognates may
arise when the literary Arabic items have several semantic fields, not
all of which are used in modern colloquial Arabic lexical items. The
colloquial Arabic items may refer to one of the several semantic fields of
the literary Arabic item. Such cases may be due to the fact that modern
literary Arabic draws on a huge dialectal inventory from the past, which
is not known in all the colloquial Arabic dialect regions. Among the
most famous examples, is f, Literary Arabic (rarely used) to polish vs.
colloquial Arabic he saw (it is probably famous also because it appears
in Ferguson 1959). Other examples more specific to the region of
colloquial Arabic in Israel are ujra, in literary Arabic wages, colloquial
Arabic taxes (wages in colloquial Arabic in Israel is more often ajar,
668
judith rosenhouse
ma or rteb, which are also used in literary Arabic); muallim in literary

Arabic means a teacher or instructor, but in colloquial Arabic teacher,
as well as a boss in a workshop, an office etc.10 Kabs(a) is in literary
Arabic pressure, attack, colloquial Arabic heavy (rain) (parallel to the
English expression its raining cats and dogs); fatina (literary Arabic)
/ fiten (colloquial Arabic)literary Arabic be clever, smart, colloquial
Arabic remember; qalbliterary Arabic heart, colloquial Arabic ones
internal part of the body, used also for the heart and entrails.
4. Conclusion
This chapter has focused on literary and colloquial Arabic vocabulary.
An important feature of colloquial Arabic vocabulary is that it is more
affected by literary Arabic lexemes than literary Arabic is by colloquial
Arabic lexemes (as expected). The reasons for this have been studied
elsewhere and we did not analyze them here. Suffice it to say that this
picture is largely due to the prestigious status of literary Arabic compared to colloquial Arabic (see Rosenhouse forthcoming).
We have also seen that phonological and morphological features mark
many auditory / articulatory differences between literary and colloquial
Arabic. Our semantic comparisons between literary and colloquial
Arabic lexemes have shown both diverging and converging trends of
development in each of these varieties.
A relatively small part of the sum total of Arabic lexical items is
composed of foreign loanwords that penetrate Arabic due to the
cultural trends of modernization / Westernization. In this section of
the vocabulary we also note that colloquial Arabic uses more foreign
loanwords (in Israel sometimes borrowed via Hebrew) than literary
Arabic does. This state is apparently due to the fact that in many cases
literary Arabic uses translations (calque) and newly coined terms rather
than the foreign ones used in colloquial Arabic (see Abu-Haidar 1992,
Shraybom-Shivtiel 2005).
We did not find many colloquial Arabic items in Israeli newspapers.
Those found occurred mainly in advertisements and caricatures
where foreign loanwords relating to politics, shopping and economics,
10
Wehr (1971) gives also master of trade as one of the meanings of this word.
669
sports and entertainment, etc., are used, as in many other language

communities in the world. On many hoardings (especially in Arabicspeaking townships or villages in Israel), however, colloquial Arabic (of
Israel) vocabulary and structures play an important role, even if they
are not the norm. A considerable measure of linguistic influence on the
general public comes from mass media such as radio and TV, where
advertisements are orally and visually published. These recurring ads
affect the public subconsciously so that this source of mixture between
literary and colloquial Arabic, which was on the increase all through
the 20th century, has been going on up to now. Abu-Haidar (1992, 104)
sums up the effect of modern standard Arabic on dialect convergence
(in Baghdad) by saying that certain lexemes (in their MSA forms)
are now part of Muslim Baghdadi and Christian Baghdadi speakers
everyday experiences (105). She adds: salient features of Muslim
Baghdadi and Christian Baghdadi have for a long time constituted
closely defined boundaries separating the two dialects from each other
and from Modern Standard Arabic. At the present time, however,
these boundaries are shifting as some salient features become obsolete and more unassimilated Modern Standard Arabic forms are diffused
into both dialects (p. 105). In her conclusion Ayyoub (2002) states:
From its journey across the 20th century, the Arabic language carries two
remarkable traits: never in the past have the written and spoken languages
been intricately intertwined; never before have the centrifugal forces of
standardization affected the dialects as much as now . . .
A separate question is, however,

what will become of the Arabic language in the future? If the past century
has sensibly modified the modalities of functions between literary and
dialectal Arabic, present for more than one and a half millennia, multiple
paths remain possible (Ayyoub 2002).
Literary Arabic is a source of fascination (for researchers and native

speakers alike, if we may add) because of its relation to both the secular
and the sacred (Haeri 2003). Moreover, economic constraints and
globalization alike favor literary Arabic, as a common language for a vast
geographic area, more advantageous than multiple languages (Ayyoub
2002). Ayyoub (2002) poses additional questions about the future of
literary Arabic and colloquial Arabic in the 21st century: Will a form
of Middle Arabic be the future? Will the dialects become the national
languages of tomorrow?
670
judith rosenhouse
We see that the diglossic makeup of Arabic is changing and evolving

all the time, including the structure and vocabulary of this language.
In spite of the generally unfavorable attitude to code mixing/switching
between some foreign language and colloquial Arabic, these processes are
apparently inevitable under the modern global village circumstances.
These foreign lexical elements penetrate colloquial Arabic, the receiving
language (or matrix language, according to Myers-Scotton 1993), and
usually become an integral part of it. A similar process takes place in
literary Arabic, although apparently at a smaller (slower?) rate. Since
this process existed also in the past (see, for example, Al-Falay, 1996,
chapter four) without lessening the vitality of the Arabic language,
we assume that the similar process of modern times will not lead to
an entirely different future development. In spite of the huge number
of lexical items (including borrowed and innovated words) and their
semantic fields in a language like Arabic, lexemes are an open class and
may be added, exchanged, renewed or discarded according to need.
When thinking of structural changes of Arabic due to code mixing or
code switching phenomena, we do not consider the lexical developments
of literary and colloquial Arabic as described here to be the straw that
breaks the camels back. Moreover, the features and examples in this
chapter do not differ much, in principle, from those found in other
Arabic dialects (Boucherit 2000, Heath 1989, Owens 1998, Taine-Cheich
2000). We therefore assume that the Arabic language will continue its
basic path sweeping along all the side trimmings noted in the changes
of its vocabulary. We also support the view that the relation between
the studied linguistic varieties in Arabic (Arabic dialects and literary
Arabic) is similar in many respects to the relations between different
languages in bilingual situations (Fasold 1984).
The above analysis may be seen within the framework of phenomena
of language interference or language contact according to one linguistic
approach, in addition to bilingual code mixing according to another.
We hope the examination of the intertwining vocabulary of literary
and colloquial Arabic will contribute to the attempts to improve
our understanding of the Arabic language. Since the paths that the
Arabic language will take do not have to be identical to the paths of
the past, further attention to the issues raised here should prove to be
interesting.
671
5. References
Abd al-l, Abd al-Munim Sayyid. 1971. Mujam al-alfz al-mmiyya dt al-usl
al-arabiyya. Cairo: Maktabat Nahda al-Misriyya.
Abu-Haidar, Farida. 1992. Shifting Boundaries: The effect of modern standard Arabic
on dialect convergence in Baghdad. In Broselow et al., Perspectives on Arabic
Linguistics IV, 91106.
al-Falay, Ibrh$m Salh. 1996. Izdiwjiyyat at-lua: An-nazariyya wa-t-tatbiq. Ar-Riyd
King Sad University.
Al-Hassan, S.A. 1978. Educated spoken Arabic in Egypt and the Levant: A critical
review of diglossia and related concepts. Archivum Linguisticum, 8(2):112132.
Altoma, Salah J. 1969. The Problem of Diglossia in Arabic: A comparative study of classical
and Iraqi Arabic. Cambridge: Harvard University Press.
Arnold, Werner, and Hartmut Bobzin, eds. 2002. Sprich doch mit deinen Knechten
Aramisch, wir verstehen es! 60 Beitrge zur Semitistik Festschrift fr Otto Jastrow zum
60. Geburtstag. Wiesbaden: O. Harrassowitz.
Ayyoub, Georgine. 2002. An Odyssey of Words: Evolution of the Arabic language in the
20th century. Al-Jadid Magazine 8:40 (Summer).
Badaw$, Sa$d Musta f. 1973. Mustawayt al-lua al-arabiyya al-musira f misr. Cairo:
Dr al-Marif.
Badawi, Mohamed. 1997. Probleme des Fachwortschatzes im Arabischen dargestellt
insbesondere an der Terminologie der Teleinformatik. Hildesheim: Georg Olms Verlag.
Bhatia, Tej. K., and William. C. Ritchie, eds. 2004. The Handbook of Bilingualism. New
York: Blackwell.
Blanc, Haim. 1960. Stylistic Variation in Spoken Arabic: A sample of interdialectal
educated conversation. In C.A. Ferguson, ed., Contributions to Arabic Linguistics,
81156.
. 1964. Communal Dialects in Baghdad, Cambridge, Mass.: Harvard University
Press.
Blau, Joshua. 1977. The Beginnings of the Arabic Diglossia. A study of the origins of
Neoarabic. Afroasiatic Linguistics 4:428.
Boucherit, Aziza. 2000. Reflexions sur le contact de langues partir du cas dAlger. In
Manwel Mifsud, ed., Proceedings of the Third International Conference AIDA, Malta,
8388.
Bousaffara-Omar, Naima. 2005. Diglossia. In Kees Versteegh et al., eds., Encyclopedia
of Arabic Language and Linguistics, vol. 1, 62937.
Broselow, Ellen, Mushira Eid, and John McCarthy, eds. 1992. Perspectives on Arabic
Linguistics IV, Papers from the Fourth Annual Symposium on Arabic Linguistics.
Amsterdam/Philadelphia: J. Benjamins.
Cadora, Frederic. 1979. Interdialectal Lexical Compatibility in Arabic: An analytical study
of the lexical relationships among the major Syro-Lebanese varieties. Leiden: E.J. Brill.
Diem, Werner. 1974. Hochsprache und Dialekt im Arabischen: Untersuchungen zur
heutigen arabischen Zweisprachigkeit. Wiesbaden: F. Steiner.
Elgibali, Alaa, ed. 1996. Understanding Arabic: Essays in Contemporary Arabic Linguistics
in Honour of El-Said Badawi. Cairo: American University of Cairo Press.
Fasold, Ralph, W. 1984. The Sociolinguistics of Society, Cambridge: Blackwell.
Ferguson, Charles Albert. 1959. Diglossia. Word, 15: 325340.
. 1996. Epilogue: Diglossia revisited. In Alaa Elgibali, ed., Understanding Arabic: Essays
in Contemporary Arabic Linguistics in Honour of El-Said Badawi, 4967.
. (ed.). 1960. Contributions to Arabic Linguistics. Cambridge, Mass.: Harvard
University Press.
, and Majid Said. 1958. Lexical variants in Arabic dialects (unpublished).
672
judith rosenhouse
Fernandez, Mauro. 1993. Diglossia: A Comprehensive Bibliography, 19601990 and

Supplements. Amsterdam: J. Benjamins.
Fischer, Wolfdietrich, and Otto Jastrow. 1980. Handbuch der arabischen Dialekte.
Wiesbaden: O. Harrassowitz.
Furayha, An$s. 1955. Nawa arabiyya muyassara. Beirut: Dr a%-%aqfa.
. 1959. Tabst qawid al-lua al-arabiyya al usl jadda: iqtir wa-namdij.
Beirut: American University of Beirut.
Fck, Johann. 1950. Arabiyya: Untersuchungen zur arabischen Sprach- und Stilgeschichte.
Berlin: Akademie-Verlag.
Geva-Kleinberger, Aharon. 2004. Die arabische Dialekte der Stadt Haifa in der ersten
Hlfte des 20. Jahrhunderts. Wiesbaden: O. Harrassowitz.
Haeri, Niloofar. 2003. Sacred Language, Ordinary People: Dilemmas of culture and politics
in Egypt. New York: Palgrave Macmillan.
Hary, Benjamin. 1996. The importance of the language continuum in Arabic
multiglossia. In Alaa Elgibali, ed., Understanding Arabic: Essays in Contemporary
Arabic Linguistics in Honor of El-Said Badawi, 6990.
Heath, Jeffrey. 1989. From Code Switching to Borrowing: Foreign and diglossic mixing in
Moroccan Arabic. London: Kegan Paul International.
Hetzron, Robert, ed. 1997.The Semitic Languages. London: Routledge.
Holes, Clive. 1995. Modern Arabic: Structures, functions and varieties. London: Longman.
Jastrow, Otto. 1978. Die Mesopotamisch-arabischen qeltu Dialekte. Vol. 1: Phonologie und
Morphologie. Wiesbaden: O. Harrassowitz.
Kaye, Alan S. 1994. Formal vs. Informal Arabic: Diglossia, triglossia, tetraglossia, etc.,
polyglossia- multiglossia viewed as a continuum. Zeitschrift fr Arabische Linguistik,
27:4766.
. 2002. Diglossia. The state of the art for the new millennium. In W. Arnold and
H. Bobzin eds., Sprich doch mit deinen Knechten Aramisch, 37988.
, and Rosenhouse, Judith. 1997. Arabic dialects and Maltese. In R. Hetzron, ed.,
The Semitic languages, 263311.
Koplewitz, Emanuel. 1989. The use and integration of Hebrew lexemes in Israeli spoken
Arabic. Multilingual Matters 71.181195.
Levin, Arieh. 1995. Grammar of Spoken Arabic in Jerusalem. Jerusalem: Magnes Press
(in Hebrew).
, and Friedmann, Yohanan, eds. 2004. Studies in Honor of Moshe Piamenta.
Jerusalem Studies in Arabic and Islam 29, Jerusalem: Magnes Press.
Mazraani, Nathalie. 1997. Aspects of Language Variation in Arabic Political Speech
Making. Richmond, UK: Curzon.
Mejdell, Gunvor. 2002. Features of lua wustmixed discourse in spoken Arabic of
Egypt. In A. Youssi et al., eds. Aspects of the Dialects of Arabic Today, 31728.
Meiseles, Gustav. 1975. Oral Literary Arabic: Its main characteristics in speech and
reading, Ph.D. thesis, The Hebrew University in Jerusalem (in Hebrew).
. 1980. Educated spoken Arabic and the Arabic language continuum. Archivum
Linguisticum 11:118143.
Mifsud, Manwel (ed.). 2000. Proceedings of the Third International Conference AIDA,
Malta, 29 March2 April 1998. Paris: Association International de Dialectologie
Arabe.
Mitchell, Terrence F. 1986. What is educated spoken Arabic? International Journal of
the Sociology of Language 61, 732.
Monteil, Vincent. 1960. Larabe moderne. Paris: C. Klincksieck.
Myers-Scotton, C. 1993. Duelling Languages: Grammatical structures in code-switching.
Oxford: Clarendon Press.
Nuayma, M$x$l. 1961.6 Kna y m kna: Short Stories, Beirut.
673
Owens, Jonathan. 1998. Neighbourhood and Ancestry: Variation in the spoken Arabic of
Maiduguri, Nigeria. Amsterdam/Philadelphia: John Benjamins.
al-Qahtani, S. 2002. Arabization and the Theory of Language Planning: An Applied Study
of Terminology Arabization in Saudi Arabia. Beirut: Center for Studies of the Arabic
Union.
Ritchie, W.C., and Tej K. Bhatia,. 2004. Social and psychological factors in language
mixing. In T.K. Bhatia and W.C. Ritchie, eds. The Handbook of Bilingualism, 336352.
Rosenbaum, Gabriel M. 2000/2. Do you parler Arabi? Mixing colloquial Arabic
and European languages in Egyptian literature. Materiaux Arabes et Sudarabiques
(Nouvelle Serie), 10:1147.
. 2004. Egyptian Arabic as a written language. In A. Levin and Y. Friedmann, eds.
Studies in Honor of Moshe Piamenta, 281340.
Rosenhouse, Judith. 1984. The Arabic Bedouin Dialects: General problems and a close
analysis of North Israel Bedouin Dialects. Wiesbaden: O. Harrassowitz.
. 2001. Trilingual Practical Dictionary: Hebrewliterary Arabic-colloquial Arabic.
Rosh Ha-ayin: Prolog.
. 2002. Phonetic trends of colloquial Arabic dialects in Israel. In W. Arnold and H.
Bobzin, eds., Sprich doch mit deinen Knechten Aramisch, 599611.
. 2004. Trilingual Practical Dictionary: Literary Arabic-Hebrewcolloquial Arabic ,
Rosh Ha-ayin: Prolog.
. Forthcoming. The English language and its impact on Arabic in Israel. In. J.
Rosenhouse and R. Kowner, eds., Globally Speaking.
, and Mira Goral. 2004. Bilingualism in the Middle East and North Africa: A focus
on the Arabic-speaking world. In T.K. Bhatia and W.C. Ritchie, eds. The Handbook
of Bilingualism, 835868.
, and Rotem Kowner, eds. Forthcoming. Globally Speaking: Motives for Borrowing
English Loanwords in World Languages. Clevedon, UK: Multilingual Matters.
Shraybom-Shivtiel, Shlomit. 2005. The Revival of the Arabic Language as a Mission of the
National Ideology in Egypt. Jerusalem: Magnes Press (in Hebrew).
Somekh, Sasson. 1991. Genre and Language in Modern Arabic Literature. Wiesbaden:
O. Harrassowitz.
. 1993. Colloqialized fush $ in modern Arabic prose fiction. Jerusalem Studies in
Arabic and Islam, 16:17694.
Stetkevych, J. 1970. The Modern Literary Arabic Language: Lexical and stylistic developments. Chicago: University of Chicago Press.
Suleiman, Yaser. 2006. Arabiyya. In Kees Versteegh et al., eds. Encyclopedia of Arabic
Language and Linguistics, vol. 1, 173178.
Swadesh, Morris. 1952. Lexico-statistics dating of pre-historic ethnic contacts with
special reference to North American Indians and Eskimos. Proceedings of the
American Philosophical Society, XCVI:452463.
Taine-Cheich, Catherine. 2000. Reflexions sur le statut des emprunts dans les langues
mixtes partir du cas mauritanien. In M. Mifsud, ed. Proceedings of the Third
International Conference AIDA, 10712.
Thomason, Sara Grey. 2001. Language Contact: An Introduction. Washington D.C.:
Georgetown University Press.
, and Terrence Kaufman. 19881/19912. Language Contact, Creolization and Genetic
Linguistics. Berkeley: University of California Press.
Versteegh, Kees. 1984. Pidginization and Creolization: The Case of Arabic. AmsterdamPhiladelphia: J. Benjamins.
. 1997. The Arabic Language. Edinburgh: Edinburgh University Press.
, et al., eds. 2006. Encyclopedia of Arabic Language and Linguistics. Vol. 1. Leiden:
E.J. Brill.
674
judith rosenhouse
Wehr, Hans; J. Milton Cowan, ed. 1971. A Dictionary of Modern Written Arabic.
Wiesbaden: O. Harrassowitz.
Weinreich, Uriel. 1967. Languages in Contact: Problems and findings. The Hague:
Mouton.
Xarb&, A. 2002. (arakat at-tarb f l-urdunn. Amman: Ministry of Education.
Youssi, Abderrahim, Fauzia Benjelloun, Mohamed Dahbi, and Zakia Iraqui-Sinaceur,
eds. 2000. Aspects of the Dialects of Arabic Today: Proceedings of the 4th Conference
of the International Arabic Dialectology Association (AIDA), Marrakesh, April 14.
Rabat: Ampatril, 317328.
EVERYTHING YOU ALWAYS WANTED TO KNOW ABOUT

L, YIL TO SAY IN EGYPTIAN ARABIC*
Manfred Woidich
University of Amsterdam
1. Introduction
The idea of contributing to the Festschrift for our esteemed friend and
colleague Kees Versteegh with an article on the verb l, yil (gl, yigl)
to say in Egyptian Arabic dialects and its various idiomatic uses, came
to me when I was reading a draft of his lemma Serial Verbs, which he
had written for the Encyclopedia of Arabic Language and Linguistics.1 In
this lemma, he reconsiders my view of items like (01) as an originally
paratactic but now grammaticalized construction in order to introduce
a pseudo-complement,2 and prefers to regard it as serialization, that is,
as a serial verb construction, albeit not without hesitation.
(01)
huwwa raddi alk

allak
?
huwwa radd al-k
l-l-ak
he
answered to-you said-to-you what
What did he answer you? LAB 118,8
Unfortunately, I do not share this view and I shall reaffirm my position

in section 4 below. Since this pseudo-complementation is not the
only case for l, yil to appear in contexts and functions that involve
grammaticalization, it seemed appropriate to give an overview of several
other cases where l, yil clearly has lost its original lexical meaning
by a process known as bleaching and acquired new meanings and
developed into a function word. After making some brief remarks on
the syntactic behaviour of l, yil as a lexical verb in section 2, several
steps and ramifications in the history of l, yil will be described in
* I should like to thank Rudolf de Jong for going through an earlier draft of this article and giving me some valuable hints. Needless to say, any remaining errors are mine.
1
To appear as Versteegh 2007.
2
Woidich 2002, in particular pp. 183184.
676
manfred woidich
section 3 based on data collected over the last 40 years mostly from
written sources but also from some recordings of rural dialects. The
starting point of these developments apparently is the use of l, yil by
the speaker to introduce reported speech3 on discourse level4 within a
pragmatic strategy, namely introducing a direct or an indirect quote, be
it the speakers own words or the speech of somebody else, in order to
show rejection and non-acceptance of an utterance, to give reasons for
acting in a certain way, to explain intent, to make a comparison between
two things or for other reasons. The final sectionsection 5deals
with a lexical aspect and gives some examples of the use of l, yil with
vocatives and in delocutive derivations.
2. On propositional level: direct vs. indirect speech

Here I will briefly describe a syntactical point, that is, the introduction
of direct and indirect speech. Direct or quoted speech follows l, yil
as an asyndetic sentence without a complementizer:
(02) ana mu ayllik hti )a)a
Did I not say to you, bring an ashtray? RUH 112,-1;
(03) alitlaha tali uudi maya
She said to her, sit down with me! FWQ 107,-7
(04) ns kitr bau ylu f-nafsuhum w ana mli
Many asked themselves, what business is this of mine? BAHN 122,9
A sentence reporting indirectly what was said, however, may either

be introduced by the complementizer inn (05, 06) or be connected
asyndetically (07, 08):
(05) iddakatra lu inn abak tabna wayya
The doctors said that your nerves are a little bit off HAM 95,7f
(06) l innu aynm andina llla diyyat wi bass
He said that he would sleep with us only that night LAB 16,8
(07) **a kkallim wi l aybt ba++a
Daddy called and said that he would spend the night out of the house
LAY 113,3
Following Gldemann et al. (2002, viii), I use the term -reported speech as a generic term for both direct and indirect speech.
4
In many languages, the verb -say constitutes a source of various pathways in grammaticalization that can lead to distinct types of function words, see Heine et al. (1993),
and Heine et al. (2002, 261 ff ).
l, yil to say in egyptian arabic
677
(08) dil kn l ayxud agza nnaha+da w ayfu aa f-Alma

dil has said, he would take the day off and they would have a look at
the apartment in Alm'a WAZ 374,12
3. l, yil (gl, yigl) on discourse level

Reported speech introduced by l, yil (gl, yigl) serves in various ways
and on different levels for text constitution and as a discourse device.
In narration it is commonly used for telling a story (3.1) by means of
reporting it as a dialogue. In a similar way, reasons and intentions may
be presented to the hearer in the form of reported speech (3.2), that is,
in the form of the speakers own words. Whereas the last two uses do
not exhibit any syntactic differences in comparison to the normal one
in a simple proposition (see 2), there are phrases formed with l, yil
that deviate in certain respects from the normal usage and suggest that
the items concerned areor at least have started to begrammaticalized.
So, ulti serves as a turn-giving device asking for a positive reaction
(3.3), l marks a short-cut (3.4), l serves as a marker of incredulity
(3.5), tiul as a comparison marker (3.6), baullak as a turntaking device (3.7), biylu as a reference to hearsay and general
knowledge (3.8).
3.1
Narrative device: story-telling by means of reported speech
A very common way of telling a story is to present it as a sequence of direct

utterances, that is, as a constructed dialogue in which each utterance is
introduced by quotative l, yil (gl, yigl). The phrases are juxtaposed
asyndetically and are not connected by means of wi and. Speakers make
extensive use of this means, as can be seen in any collection of texts of
Arabic dialects recorded in the field. For the audience, a story structured
step by step in this way is more insightful and easier to follow than a
purely descriptive narration, since the narrator gives the story a structure
that conveys liveliness and persuasion and involves the hearer. 5 A short
passage from a Bri text may illustrate this:
5
Involvement of the hearer is considered one of the most important pragmatic strategies in Georgakopoulu et al. (2004, 136f).
678
manfred woidich
(09)
glat isma&&a alk y-w+ad, isma&&a alk mlak. gultilha-na

anaalam wu nji an nafsi anijlib min ilba+bax an nissabba,
rut wagat min fg ilmanma. galatli kaamt rijli ya wldi. gultilha
ana gultluk ijilbi wa+y? wu baadn galatli +aa-txallni nmt
kamn wu rijli maksra, gultilha lyi li. galatli jaddak Abda&&a
lGum mt huwwa w rijla maksra. gultilha min ? galatli lamma
knaw aytiu fi a&amt, gultilha k ytiu fi a&amt galatli
l nta mi rif ilikkwa? gultilha l ma-ufth wala xabir ja.
galatli badri, )irg issuww diyya . . .
She said: May God keep you safe, my boy, may God protect you!
Whats wrong with you? I said to her: I was dreaming and [in my
dream] I jumped involuntarily from the footbridge to have a swim.
So I fell from the sleeping roof. She said to me: You sprained my
ankle, my boy! I said to her: Did I tell you to jump after me?
And then she said to me: You are going to let me even die from
a broken leg. I said to her: It was not my fault. She said to me:
Your grandfather Abda**a lGum died from a broken leg. I said to
her: How come? She said to me: When they were pulling on the
colossoi.6 I said to her: How come they pulled on the colossoi? She
said: Why? Dont you know the story? I said to her: I did not see
it, nor do I know anything. She said to me: In former times, this
tourists way . . . (Woidich 1980, 236,2731)
This use of directly reported (quoted) speech introduced by l, yil

(gl, yigl) is characteristic of the narrative style and gives a dialogue
structure to the whole narrative or parts of it. The narrative becomes
like a theatre play delivered by the speaker, and indirect speech and
complicated subordinate structures are avoided. There is no sign of
grammaticalization, and l, yil (gl, yigl) is subject to inflection
following the logic of the story and remains in its position in front of
the quoted speech.
3.2
Argumentation by quoted speech
In a similar way, comments on a statement may be given by the speaker

to the hearer in the form of direct speech, mainly as the speakers own
words, presenting the argumentation as an inner monologue. Such selfquotations are introduced by ult I said and follow a statement. Speakers
report to the hearer the reasons for decisions or conclusions they made,
6
ia&amt *nm (literally the idols) is the Bri word for the Colossoi of Memnon
on the West Bank of Luxor.
679
or the reasons they had for acting in a particular way. 7 This comment
explains either why what is described in this statement occurred (10,
11) or what the intention behind it was or is (13, 14, 15). In other cases
it tells the hearer what follows for the speaker from what is described
in the statement (12). Again, presenting the argumentation as internal
dialogue makes it more insightful to the hearer and easier to follow and
involves him directly.8 Using ult in this way is very common in standard
Egyptian Arabic, in both its spoken and its written form. If an intention
is involved, the verb is in the y-imperfect.
(10) ruti +af a idayya )awwli mdarya wii, ulti la-ykn alh afrt ismu
(11)
(12)
(13)
(14)
(15)
i/rab
I immediately lifted my hands up to protect my face. I said, it may have
got an evil spirit on it, I am afraid, whose name is i/rab [hit!] LAB
214,4 (reason)
ana grt ulti yimkin miawwi ga kida walla kida min wa+a-mmi
I hurried up, I said, perhaps he has somehow saved something behind
the back of my mother HAM 28,1 (reason)
laetkum )ayyibn ulti ya bitt uudi
I found you to be good people, I said, stay here, girl! RUH 30,5
(consequence)
mi gayyili nm ult atmaa
I could not sleep, so I said, let me go for a walk SIB 89,3 (intention)
il0ayya btizaf kamn ala irka. ulti nsibha abli ma ti+a
The water is creeping towards the company, too. I said, let us leave
before it is flooded FWQ 50,1 (intention)
lammt uwayyit xaab kasri ala ittitn awli wi ult awalliha addaffa bha
I collected a little bit of broken wood and some corn cobs and said, let
me set fire to it to warm me up MHR 7,6 (intention)
(15) shows that a syndetic connection with wi and is also possible. (16,
17) provide evidence that this discourse strategy is not confined to a
quotation of the speaker himself using the 1st person but may be applied
in the same way in narratives in the 3rd person as well. While we cannot
see in (16) whether direct or indirect speech is involved, the use of a 1st
pl. verbal form in (17) signals direct speech:
7
For similar functions of self-quotation in German (i.e. reporting on decisions, reasoning, intention), see Golato 2002. Further development of items meaning ;saying
i.e. original quotation markers to a grammatical markers expressing causality oris
widespread in the languages of the world, see e.g. Lord 1993, 177 for se say in Twi and
for the Turkish gerundium diye saying, see Kissling 1960, 191; for Bengali bole having
said see Chisarik et al. 2003.
8
See Georgakopoulu et al. 2004, 136f.
680
manfred woidich
bai a kkanaba lwa++aniyya ma-lta ayyuha ga, zaar a kkursiyyn

iluddamiyyn iltahum ya mawlya zayyi ma xalatini, l tiba an)a
lli wa+a
He looked at the rear seats, but did not find anything, he glanced at the
front seats, but found them empty, he said, it must be the boot LAB
59,4 (conclusion)
(17) fa xad ba/tih wu , a+abiyya mi lminim gl nifu (ilmi Ysif yikn
iann
He set out and what? [he took] a car from the mines, he said, let us have
a look at ilmi Ysif, perhaps hes gone mad (Ba<ariyya: Manda)
(intention)
(16)
3.3
ulti what did you say? as a discourse routine
ulti ,literally What did you say?immediately follows a statement

conveying a suggestion or presenting an astonishing fact to the hearer
(21). ulti is connected to this statement by the intonational contour.
At first glance, the perfect ult makes it look like a request to repeat
something that has been said, pretending that a comment has been
overheard. But this is not the case since ulti is closely connected to
the statement with no pause and there is no time for the hearer to show
any reaction to this statement. Rather, it functions as a turn-giving
discourse marker, which asks for a positive comment on the statement
that is presented hereby as surprisingly good news to the hearer. For
that reason, it should be translated as How about that! which should
be understood as Arent you surprised and am I not to be praised?
Examples:
(18) tikkaffil inta bi maawr il-iyl wi ddurs w ana astannh, ulti ?
You are put in charge of the errands for the children and lessons, and I
wait for him, (so) what do you say now?! BTR 11,3
(19) wala yhimmak ya (sn xui ala lbutk xud illi yigibak wi kullu bi ttas)
tidfa tultumt gin dilwati wi lbi bi ttas) ilmirayya ala irn ah+i
xamas t-alf gin, ulti ?
Dont mention it, isn, go into the boutique, take what you like and
pay for everything in instalments; you pay 300 pounds now and the rest
in easy instalments, 5000 pounds over 20 months, so how about it?!
RAS 13,5
(20) tir ilwat illi yigibak, fi nnr, fi //alma. illi tfu, ulti ?
You go whenever you like, with daylight, in the darkness, just as you see
it (whenever it is convenient for you), isnt this a good suggestion? SIK
106,10
(21) ana tnzilti an ai fi kkaka li x Ima, ultu baa?
I gave up my rights to the cake in favor of x Ima, what do you say
now? MAX 93,4
681
Given this meaning, we would expect a continuous form of the verb

such as til or bitl , rather than of the perfect ulti . But replacing
the perfect here with an imperfect, ulti would mean losing its special
effect, which originates in a violation of the discourse record principle:9
the speaker acts as if his statement had already been commented on,
thereby constructing a reality that contradicts the speaker/hearer record
of the discussion so far and attracts the attention of the hearer and makes
it more relevant to him. At the same time, the rhetorical intonation of
ulti suggests that the hearer is assumed to consider this statement
as positive and to react accordingly. Apparently, this discourse routine10
did not yet lead to grammaticalization and to the forming of a particle,
since ulti still follows the rules of concord, as can be seen in (21). (22)
shows that giving the turn may be done not only by the speaker but also
by a third participant in the discourse:
(22) : ana mumkin anm uddm ilbb ilxargi laddi ma lmukila titall,
bass tiwidni nni ma-f addi yxui min innaya ttanya li lbt: ulti
ya Ar? A: ittafana.
: I can sleep in front of the outer door until the problem is solved.
You only have to promise me that nobody will enter the house from the
other side.: What did you say [what do you mean], Ar?A:
Agreed! MUW 134,4.
Again, the use of perfect by in (22) makes no sense and ulti has to be
seen as a turn-giving discourse marker.

3.4 l say! as a discourse marker
l say!the masculine imperative of l, yilmay be used in two
different ways. Within a text it introduces what a longer explanation or
description boils down to and serves as a short-cut in order to finish a
topic and to switch to another one. It corresponds to in short or to cut
a long story short in English, as in:11
(23) )iliti ba+a wi lwd iabi huwwa lli tabas bidli, a&&h yir/a annu, l
amadti +abbina lli )liti b gildi
9
See Schwenter et al. 2005 for a case of discourse record manipulation regarding
English too.
10
This is why I prefer to call it a routine, i.e. a discourse technique to reach a certain
goal, and not a marker, which in my view would include some grammaticalization as is
the case with l say!; see 3.5.
11
There are other markers with a similar function such as ur ikkalm, ilad,
nahaytu.
682
manfred woidich
I was found not guilty and the apprentice was jailed instead of me, may
God bless him, in short, I praised God that I had come away safely DAR
71,10
(24) wi msik sikkna yiz yiui bha ba)ni. girt minnu. )r wa+ya. l inns
ah
With a knife in his hand, he wanted to rip my belly open. I ran away,
he went after me. To cut a long story short, people held him back DAR
93,8
There is no evidence for female li being used in this way, but apparently
the plural lu serves the same purpose when several persons are
addressed. So l in this case should be considered not as a marker but
as a routine:
(25) inns yadb simit fh markib mal, risyit ala baladhum wi dl gary
an yitiru. lu awwil markib itbit fi sat
The people had only just heard that a salt ship had moored in their town
when they came running in order to buy. In short, the first shipful was
sold within hours MAL 36,8
(26) aadit ah+i w ah+i w ah+. lu iblit
She stayed a month and a month and [another] month. In short, she got
pregnant MAL 39,33
Another use of l say! derives from a discourse routine, that is, selfcorrection in the form of a request to the hearer to correct a piece of
informationvery often numerical informationadduced by the
speaker. As the speaker seeks confirmation from the hearer, he suggests
that he is not sure about his estimate and that this information is not
100% reliable. Within a sentence l may thus indicate an approximation,
and correspond to perhaps; like, for example like the English: the whole
affair lasted, say, 10 minutes, as can be seen from (27). l underwent
grammaticalization and is frozen as a particle, since neither the feminine
nor the plural form are possible here and it is moved into the sentence,
standing for example in front of the direct object (29) or an adverbial
expression (30). More often than not, it combines with ygi about as a
reinforcement of this approximative meaning (28, 29, 30):
(27) ilamaliyya atxudlaha talat t-iyym l a+baa
The operation will take three days, perhaps four YUN 72,9
(28) km ya Nabawiyya km?l ygi irn gin aw akta+
How many, Nabawiyya, how many?Perhaps about twenty pounds or
more RUH 137,7
(29) iggamustn ayilibulna kulli ym l ygi mt klu laban
The two buffalos will give us about 100 litres of milk a day HAM
90,4
683
(30) wi lamma ifyu aadti ala fariti l ygi sa, itnn

And when it had finished, I sat down on my bed for an hour or two
RUH 7,7
3.5
Discourse marker for rejection and non-acceptance l and l
The entry l in Hinds et al. (1986, 722b) gives l as a modal of

incredulity with just imagine! fancy that! can you credit it! as
translation equivalents. And ll nonsense! baloney! humph!,
which introduces and closes a word or sentence as an exclamation of
jeering disbelief.12 I shall try to give a more precise description of its use
and at the same time sketch the historical development.
As the examples below show, an l introduces or follows the reported
utterance made by somebody else and urges the hearer to interpret this
utterance as rejected and not acceptable to the speaker. It functions thus
as a discourse marker and is fully grammaticalized as it is no longer
inflected (31, 33, 35). As becomes clear from the examples, it is indirect
speech here, not direct speech.
(31) Munra: bal li a/ritik ana mi ataal andi add, ana mi xaddma.
Bahga: mi xaddma! samn! l ma-h xaddma
Munra: I am telling you, I shall not work with anybody, I am not a maid.
Bahga: Not a maid! Did you hear! Not a maid, incredible! NAS 81,4
(32) ilminayyil ibn i))abbxa bn Ummi Anwa+, l gayy yux)ub Gamalt
This damned son of the cook, Umm Anwars son, incredible! He is
coming to ask for the hand of Gamalt! FWQ 90,12
(33) lat ilumm bitaadni wi btiddni darsi fi linniyya.ilinniyya?l ana
mi fa/ya l-binti wi an kida hiyya bitr bt zimilha da an ummu
hiyya-lli anna alha
It happened that the mother let me sit down and gave me a lesson in
affection.Affection?As if I would not be free for my daughter and
this is why she goes to the home of her colleague because his mother
always shows affection towards her SIG 131,3
In both (31) and (33), the context leaves no other possibility than that
it was a female speaker who made the incriminated utterance, even
though l remains masculine. l may follow the reported speech (34)
and may even occur in both positions at the same time (35). The latter is
particularly common with single words, as in (36):
12
For a brief discussion, see Woidich 1995, 265.
684
manfred woidich
(34) wi biylu ma-bnitaali l

And they say, can you believe it, [that] we do not work! WAR 20,5
(35) daxalit tialli l bitalli l )ayyib xallha talli
She went in to pray, what she calls pray, well, let her pray then! NAS
130,12
(36) dukham! sama biylu ala abbahathum ? dukham! l dukham l!
Those (people)! Did you hear what they said about her parents? Those!
Imagine, those! GIL 142,3
Grammaticalization is further corroborated by the fact that l acquires

increased mobility and may be moved into the sentence itself (37, 38)
and occur within subordinated sentences (39):
(37) dl fi Mari l biyiksabu bi lmt alfi gn
In Cairo they earncan you believe!a hundred thousand pounds!
SMB 93,5
(38) glu w xawat fi lBaw)i, illi lbalad ilmidna di. yiglu imu+uk1n. w
2 , wala xadu wala amilu,
gl xadu ax mirsm dk, dahab. bassi da kalm
2
hayugudu yibitu lxawat?
They said that foreigners came to ilBaw>i, which is the main town. They
say Americans. They came, allegedly, they took [with them] a golden
statuette in the shape of a cock. This is babble: they neither took anything
nor did anything. Would foreigners go around digging up things?
(Ba<ariyya: Manda)
(39) lit imbri innak l ma-bitru ilmadrasa
She said yesterday that youincredibly enough!do not go to school
TAY 38,11
In all cases from (31) through (39), the rejection of real speechthat is,
utterances that were really madehappens by reporting it, introduced
by l and pronounced with the appropriate intonation. In first instance,
it is this typical intonation of indignation and annoyance that makes it
clear to the hearer what the speaker means: disapproval and rejection.
Reporting it alone would not be sufficient for this purpose; intonation
must be an integral part of the construction. The semantic content of
this intonation materializes in grammaticalized l. This means that
disapproval and rejection now belong to the semantic content of l,
which in this way is recruited as a pragmatic marker for disapproval
and rejection.13 This makes it possible that the development goes further
13
Indeed, for at least some speakers/writers l seems to be no longer associated with
the verb ql in this meaning. In MRR and BAHN, for example, we find it written with
Alif Mdda as 3 whereas l as a verb preserves the original orthography and is written
with qf 39 throughout.
685
and l no longer needs to refer back to a real utterance or to an earlier

part of the discourse (see 31, 35, 36), but rather to a general saying as
in (40). l thus loses its function as a discourse marker and undergoes
a development into something like a sentence adverb, but its semantic
content remains unchanged. As both (40) and (41) show, the sentence
introduced by l was not uttered earlier:
(40) yiba ++gil lbis ya w akitta w alha ban)u w lfif ala rabtu kufiyya
f wi byirtii mi-lbardi wi llabwa m+tu walla bintu maya gambu mal)i
wala hiyya sala f-bardi walla f-talg. l irriggla yita))u wi nniswn
yitarru
So the man wears a collar, a jacket, and over it a coat, and wraps his
neck in a woollen scarf, and still he is shivering from cold, and his wife,
the bitch, or his daughter, walks naked beside him not bothered by cold
or snow. Unbelievably, the men cover themselves and the women strip
themselves DAR 163,3
(41) ilfustn )ili dayya alha, l batahli albisu!
The dress was too tight for her, incredible, she sent it to me to wear!
FWQ 92,5 (the person who sent the dress does not show up personally)
A further extension we notice when l introduces a sentence starting

with yani, an, or biylu. Just as in the examples above, l no longer
introduces a quote or reported speech, but expresses disapproval of a
fact or, more properly, the inference of the earlier statement. More often
than not, a sense of irony is involved. So, in (42) the first statement by A
invites the inference that A, as a matter of course, does not wet himself,
an inference that is rejected by B, ironically by using l together with
yani, which itself could convey a similar meaning.14 l serves here as
a reinforcement of this inference and adds the disapproving semantic
element. In the same vein, in (43)which is a quote from Bayram
at-Tniss isSayyid wi m+tu f-Barsthe pretentious behavior of the
umda described in the first part of the sentence and its inference, is
rejected and ridiculed by introducing this inference in the second part
of the sentence with l.
(42) A: ti+af ya ammi Wad iligli da biyuxxi ala nafsuB: l yani inta-lli
ma-bituxxi?
A: Do you know, Uncle Wa<d, this calf wets itself!B: Baloney,
would you pretend not to pee?! RUH 40,4
14
For a possible role of Sudan Arabic yani in the grammaticalization of Beja miyaad
3fait de dire by language contact, see Vanhove 2004, 149.
686
manfred woidich
(43) (after having listened to a short text in unintelligible newspaper talk)

yim ilumda yubrum anabu wi yhizzi dmu l yani fhim
The mayor twists his moustache and shakes his head, thus pretending
that he had understood SMB 84,10.15
l may be deleted in this case, since yani alone already conveys a

similar meaning and connects the two utterances in much the same
way, albeit without the sense of irony, cf. ya sitt ina wakln fl yani
mbarimn Madam, we have eaten fool, that means, we are full RAS
15,3 and i))ulumba mumkin titaal ala a+abiyyitn, yani mumkin
taman a+abiyyt yimlu banzn fi nafs ilwat The pump can serve two
cars, that means, eight cars can refuel at the same time NUS 35,3.
In (44) below, it is not so clear if rejection of the inference and irony,
which can clearly be understood from (42) and (43), are present and we
may even note here a further step in the semantic development towards
the sense of as if , just as in (45):
(44) nazln yisallimu laslia di l-ns fi mar, a))n ilmanga foha a)a l yani
humma tugg+ fakha
They went to deliver these weapons to people in Cairo, they had put
mangos on it as a cover, thus pretending that (~ as if) they were fruit
merchants BAHN 116,11
(45) lustz Ibdu baa txall masalan yidhin awaln buu min igginb kirm
abya/ l yani +awi
let, for example, Ustz Ibdu smear white cream on the side of his
mouth, as if it were foam ABM 206,-5
Similar semantic extensions may be stated for l an and l biyl, in

both cases l could be deleted:
(46) wi lingilz gum l an yimu mar
And the English came in order towhat they calledprotect Egypt
MAL 79,28
(47) ana ftakartu sak+n wi biyxa++af l biyl inn ilad+a )ilitlu wi huwwa
mi gamb sr ikkinsa
I considered him drunk and raving, when hebaloneysaid that the
Virgin appeared to him when he was walking besides the wall of the
church RUH 114,5
(48) da ddinya hayga wi lukma alba ddinya hnk l biylu lau ia+t
hink biyiftau
The world is upset and the government is turning everything round
there, they saybaloneythat they found traffic signs there that open
the road RUH 108,4
15
See SMB 85,4 for a similar passage.
687
This grammaticalization of l apparently started when it was used as

an exclamation with a strong tone of disapproval, introducing reported
speech16 uttered by somebody, but not approved by the speaker. The
following step must have been the switch from report to the inference
of the utterance, and finally from there to independent, generally
known facts, not specifically uttered previously. The connection with
a real utterance made by somebody was thus lost. This semantic
developmentnamely semantic bleaching and reinterpretationis
accompanied by a morphological reduction of l, that is, the inflection
is lost, and a syntactic flexibility appears: it may occur within sentences
and in subordinate clauses, and may be repeated producing a kind of
circumfix around a rejected phrase (35, 36).
The discourse marker l has l as an equivalent, which is nothing
other than the common rhetorical question formed with , a type of
discourse strategy that abounds in narrative contexts in Egyptian and
other dialects of Arabic.17 An utterance like (49) biyl alki di bitibbini,
l inti bitibb He says about you: she loves me. Nonsense, you and
loving him! FWQ 90,6 can be transformed into a narrative question,
as in (49) biyl alki di bitibbini, l , inti bitibb with the same
disapproval, disbelief and incredulity as the exclamation. Again, there
is lack of concord (5053), that is, l no longer refers to any referent.
Grammaticalization as a discourse marker is further evidenced by the
possibility of incorporation of l into the sentence, and, conversely,
by the fact that l may be dropped from sentences (5153) below
without disturbing the syntactical structure:
(50) xadit minni baiyyit ilirn illi fa/ln kamlit iah+, l atixzin bilibf
wi mallabt wi abn wi makaront
She took the rest of the few remaining piasters for the month, allegedly
she would store corned beef, tins, soap and pastas GIL 120,4
16
All my examples show indirect speech, not direct quotes, as far as this can be seen
from the reference of the pronouns. Direct quote as the original structure from which
the further development started cannot be excluded and is even probable. The lack of
examples could be seen as proof of early grammaticalization with loss of the starting
point structure. This view is corroborated by the fact that no inflected examples of l in
this sense could be found.
17
Any sentence may be split up into a kind of cleft sentence with a first part ending
with what? and the rest of the sentence as answer to this question: m wid minhum
, wxid blu mi lamaliyya, m zanna , ala aibna m aibna )aban , badi ma
allu ana +yi dort il0ayya, xa+ag ala ba++a One of them then what? He realized what
was going on, he hemmed in what? our friend, our friend of course what? After he had
told him I go to the loo, he headed out the door [Cairo: recorded text]. See Woidich
2006, 50.
688
manfred woidich
(51) knu biylu nni arb ilistinzf di l ma-kani laha lazma

They used to say that this war of attritionjust baloneywas not
necessary BAHN 78,3, = knu biylu nni arb ilistinzf di ma-kani
laha lazma
(52) ittit binti mafa dikit alayya w ayyilitni an)a l ala ma txu
ittawalt
A worthless squirt of a girl fooled me and let me carry a bagcan you
credit it?while she went to the loo SIG 102,5
(53) knu lbuada amln l tugg+ manga
Theyfar be they from youacted asunbelievable!mango traders
BAHN 82,6, = knu lbuada amln tugg+ manga
3.6
Comparative tiul
A rhetorical question formed by an introducing til You could say

followed by the interrogative suffix - as in tiul or provided with a
full negational ma- . . . - as in ma-tul, and a subordinated sentence as
reported speech, follows a proposition which describes an unexpected
and surprising fact as in (54, 55).18 Reported speech here communicates
a reason for this fact, but this reason is not the true one and in fact
contradicts reality. You could only compare it to the true reason. For
tiul ~ ma-tul, therefore, a pathway from Couldnt you say . . .?! =
You could say . . .! to As if it were > like can reasonably be assumed.19
On the discourse level, sentences with comparative tiul convey the
astonishment and excitement of the speaker.
(54) sabni wala saal anni, tiul illa kunti bawakkilu )b mi tasiyya w lamit
+s?
He left me and did not inform about me any longer, could you say that
[= as if] I had fed him with bricks, not with broth and meat of the head?
DAR 90,15.
(55) badi wayya asma dawa w kalm tiul il/a malyna ns
After a while I heard noise and talking, as if the room was full of people
GIZ 56,8
(56) labsa am ama+ min r ikmm wi riglha iryna ma-tuli ladaitni a+aba
She was wearing a red sleeveless shirt and her legs were naked, [I was
shocked] as if a scorpion had stung me MRR 80,8
(57) nazln al iddni iddni ma-tul innu huwwa nir ilwaf
They showered him with Give me, give me!, as if he were the supervisor
of the endowment NAZ 62,9
18
See Woidich 1989, 124 and 2006, 281.

For similar pathways of grammaticalization of say to like, as if in other languages, see Heine et al. 2003, 268.
19
689
tiul may be followed by inn introducing indirect speech (57). (56)

differs in some way from (54, 55) since the comparison expressed by
means of reported speech does not refer directly to the fact, but refers to
its consequence, which is not expressed itself [hence put into brackets
in (56)]: the unexpected seductive appearance of the lady causes a shock
to the speaker, which may be compared to the sting of a scorpion. We
notice here a shift of reference from a fact to the inferences thereof
similar to what occurred with l (cf. 3.5), which suggests that tiul
got more independence in its use and that it is on its way to become a
function word with the notion as if.
The question is how far tiul has developed in this way. It still may vary
and be inflected according to the person spoken to, that is, we can find it
as fem. tiul or pl. tiul followed by a complete sentence:
(58) mah dilwati aklitha wi labri bar/ak, ana mi fhim anhi banna aw fil
illi xtara ikkaki bi akli da? tiul illa byimilu xal)it mna
But nowadays, it takes you to the grave when you eat it. I do not
understand: which mason or bricklayer invented this sort of cake? You
could say they mix mortar! SMM 55,5
(59) wi lmg )li nzil tiul milya arr zara
And the waves went up and down, you could say like a ladies dark-blue
silk cloak DAR 141,6
(60) min sit ma simit ilxaba+ im wi hiyya nazla la)mi w buka tul ibnaha
Since she heard this awful news, she has been constantly slapping her
face and crying, as if it were her (own) son RUH 152,7
On the other hand, there are examples in the singular when we would
expect the plural, as in (61), a sentence that addresses a general public
(cf. ma-taxzun pl. Do not blame me! three lines further in the same
opening speech):
(61) w ah layym ammla tigri, tiul alla zalna minna!
And the days fly past, as ifalasthey were angry with us MAL 1,3
Whereas the facts with regard to inflection remain inconclusive,

other features suggest that tiul is on its way to be grammaticalized.
In (54, 58) above, both parts apparently still have separate intonational
contours as indicated by the authors by means of a question mark. But,
on the other hand, we see in (62, 63) below that tiul can be followed by
one argument only, not a complete sentence, which suggests that tiul
has been incorporated into the sentence and that it is used just as the
preposition zayy like. Indeed, the whole sequence has to be read here
with a single intonational contour.
690
manfred woidich
(62) amml yilsa fi llisn tiuli a+aba

It stings the tongue, as if it were a scorpion TAY 169,8
(63) w inta mkaa+ wi manfx a lxir tiuli ras wiz+a
And what are you, grave looking and utterly pompous, as if you were a
prime minister RAS 82,920
Moreover, tiul can leave its original head position and be moved within
the sentence, as in:
(64) lat makn azla zayyi ma huwwa, faritu tuli lissa mafra dilwati
I found azlas place unchanged, his bed was as if it had been made just
now LAB 49,2
(65) wala assi b-ayyuha ga r inn ila+abiyya wafa w humma tuli nizil
alhum sahm a&&h guwwa lutumbl
Nor did he feel anything but that the car had stopped and that they were
as if Gods arrow had fallen on them in the car (i.e. they had suddenly
fallen still) LAB 91,7
The original position in (64) would be tiuli faritu lissa mafra . . . and
in (65) tiul nizil alhum . . . . This strongly suggests that tiul in these
cases is grammaticalized and serves as a function word to express an
unreal comparison.
til you could say without a interrogative suffix - is rarely found in
Egypt, for example in (66) )amaha lazz tili hiyya malban It tastes
nice, youd say it is Turkish delight ARA 58,-1 and in texts from
Dakhla-Oasis (alMiyya): atgardit til di agr It rolled down as if
it were a stone. These last two examples coincide with the widespread
use of til in other Arabic dialects in the form of tegel and apparently
grammaticalized as a preposition like or something similar; see for the
Da%na tigel, tegel tu dirais Landberg (1942) 2542; for the Rwala tesma
eb nejrahom te4el dammm It is as if one could hear the angry voice
of their mortars, like to the sound of a great drum Musil (1982, 84,3)
and passim; for Souchne bitfu tik&innu cabal er sieht aus wie ein Berg
Behnstedt (1994, 353), originally When you see him, youd say it is a
mountain; for Syrian nomads see Cantineau (1937, 196); for Tunisia
see Marais (1959, 3310f.); for Morocco Colin (1993, 1624).
20
An example from Bri dialect in Upper Egypt: ilfurusa +a00a tgli ikle The
mare gallops like a young stallion.

3.7
691
Turn-taking device:21 baullak
Another phrase commonly used as a discourse marker by a speaker

in order take the lead in the conversation and to start a new topic, is
baullak Ill tell you something! Listen! see (66, 67) or simply bal . . .
I say . . . as in (68) without addressing the addressee directly. It attracts
the attention of the hearer by inviting the inference that he is telling him
something important. It is often used to prevent a longer discussion and
to give directions and instructions.
(66) : aywa ana ib ilfa/li alkum, adi nnu lolya ana ma-kani f addi
minkum gih hina, wi fi lxir absinni ba++a mi ayzn tidaxxalni ana
aywa baullak ! ana mi f/i li lhartala btatak di! lissa yiz abu) i/ati
w au)) kamirti . . .
: Yes, it is due to me, I mean, without me, none of you would have come
here, and, in the end, you shut me out and do not let me come in here any
longer!Yes, listen! I am not free for this nonsense of yours now! I want
to fix my lighting and set down my cameras . . . MUW 40,2.
(67) N: la ya **a, la! Fatiyya lzim tikammil talimha. kifya lli ga+li ana w
5a uxti min adit ilbt wi lirmn min ittalm.B: baullik ya Nagba,
ya binti. kalm kitr mi yiz asma w ana xal qa++a+t . . .
No, Daddy, no! Fat<iyya must finish her education. Enough what
happened to me and my sister @a from sitting at home and being
excluded from education!B: Ill tell you something, Nagba, my
daughter! I do not want to hear much fuss and I have made my final
decision . . .! HUD 41,8
(68) S: la, wala ga hti lahwa.H: ma-h f-di. itfa/dal ilahwa ya B!M:
mi yiz ya sitt, a&&h ilani an ahwitik, kida ya sitti Szi, inti ayzni
ar fhaS: bal ma-tsibha hina, u00l fn ilu)a Sayyid?
S: No, nothing, bring the coffee!H: I have it in my hand. Here is the
coffee, please!M: I do not want, Mam, I can do without your coffee!
Is it that, Mrs Suzy, you want me to die?S: Id say, leave it here, where
is Master Sayyid? ABM 193,27
At the same time the use of baullak creates kind of intimacy between
the speaker and the addressee, preventing the directions from coming
over too rude.
21
As to the exclamation of strong assertion bitl fha no question about it! you bet!
you said! adduced in Hinds et al. 1986, 722b, I could not find any example in the literature except the ones given in this entry: idirti truddi ala lmudr?bitl fha w atamtu
kamn Did you dare to answer the boss back?You bet I did! and I cursed him too!
692
3.8
manfred woidich
Reference to hearsay information and general knowledge
biylu they say introduces what is known from hearsay or is general

knowledge and from a non-specific source functioning as an evidential
marker, see Heine et al. (2003, 265). As usual, the 3rd pl is used in this
case, see Woidich (2006, 357):
(69) simiti agt kitr biylu kulli ga hatiba ala lbi)a
I have heard a lot. They say everything will be on a personal identity
card RUH 112,1
(70) inta lzim )allat illi biylu nnak miggawwizha
You must have divorced the one which, they say, you had married
FWQ 120,10
(71) biylu kamn kn skin fha wada w bintaha
They say, too, that a woman and her daughter were living in it RUH
18,10
Often a proverb is quoted in this way:

(72) wi l baa listigl da, da atta biylu lagala min ii)n
And why this haste, they even say, haste is from the Devil WAZ 377,15
4. l, yil introducing pseudo-complements22

In his article on serial verbs in EALL, Kees Versteegh mentions a use
of l, yil to say in Egyptian Arabic, which he considers as a complementizer derived from a serial verb construction [SVC], which in his
view provides a syntactic slot for an object. Quoting an example taken from
Woidich (2002, 183) he compares this use of l, yil to similar developments in Creole languages such as Krio Creole English and Ki-Nubi.
(73) huwwa raddi alk allak ?
What did he answer you? LAB 118,8
Earlier, Catherine Miller brought the development of gl as a

complementizer in Juba Arabic to our attention, Miller (1998, 2000, 2001)
where gl is grammaticalized and used as a complementizer after verbs
of saying and cognition.23 Let me first cite some additional examples:
22
See Woidich 2002, 181ff, Woidich 2006, 401. Phrases of the same type, but used
to express the direction or a goal, are formed with +, as in kuntu btuxrugu tru fn?
where did you go out to? FWQ 62,6, see Woidich, loc.cit.
23
Hopper et al. (2003, 194196) discuss similar developments of say in Akkadian,
see as well Hopper et al. (2003, 1315) for West African languages. For a similar use

(74)
(75)
(76)
(77)
(78)
(79)
693
ilmaxzin )aban atruddi tl ianfi r mawgd

The stores of course will answer and say, this kind is not available
RAQ 38,8
ana kutti baanni bal ?
What, was I busy singing? [film]
inta skit ya (asann, akkallim ana!atikkallimi tli ? waffari kkalm!
You are silent, asann? Then I shall speak!What are you going to
say? Keep your words to yourself ! ABM 144,6
an kida dayman ana zabayni aulluhum iwa ya-bni tu))i ga f-gbak
This is why I always advise my customers and say, take care, my boy,
not to put anything in your pocket YUN 27,3
rafa/u allu ina malna
They refused and said, not our business! NUS 7,6
xayya+ni ammti alli taxdi bni ammik
My aunts made me choose, they said to me: you should take your
cousin RUH 59,3;
The differences between the use of l, yil in our case and gle in Juba
or se in Krio Creole English are indeed significant: the first mentioned
introduces quotations and noun phrasesmostly the question pronoun
what?and is fully inflected, whereas the other two serve as a
complementizer for subordinated sentences of various kinds and are
no longer inflected. Egyptian Arabic l, yil remains connected with
locutive and speech-related verbs, and its use is not extended to other
types of verbs as happens in many West African languages (Lord 1993,
176 ff) and in Juba-Arabic (Miller 2001). There is no trace of l in
Egyptian Arabic functioning as complementizer that, a fact earlier
stated for Arabic dialects in general in Versteegh (1984, 101).
Can we analyse l, yil constructions as an SVC, then? For a verbal
sequence to be recognized as an SVC it has to fulfil some conditions
formulated in Newmeyer (2004, 2f) and Kroeger (2004, 226256). Some
of these are met by our l, yil constructions:24
they are two verbs within the same clause, neither of which is an
auxiliary;
they belong to a single intonation contour25 and refer to a single event;
of the Turkish gerund diye saying see Kissling (1960, 190). For Bedja, see Vanhove
2004.
24
See in particular Kroeger (2004, 229).
25
For a case with no single intonation contour but two contours, see gaadit turgu
wu tanni, glit abbi a+ balad ilLayy, ina&&a m ni yiriddi alayy She started to
694
manfred woidich
they share at least one semantic argument, which is the agent in their
case;
they contain only one grammatical subject. 26
At first glance, this makes the l, yil constructions good candidates for
SVC, were it not for some other diagnostic features that exclude them
from the SVC. SVC do not contain any overt markers of subordination
or coordination. This is not true for l, yil constructions where wi
and may separate the two verbs, suggesting that what we have here is a
coordinate construction:
(80) raddi azla w allu mafhm ya ammi
azla answered and said to him, understood, Uncle! LAB 138,6 (cf.
74)
(81) wi anna w l mawawl
And he was singing Mawwls MAL 45,8 (cf. 75)
(82) af/al ana abar)am w al atta ladwiya kamn biyaddarha
I shall keep muttering: even the medicine they export as well GIL 58,8
(cf. 84 below)
(83) ultilu kulli da, raddi w alli inta-zzayy awlak? mi kuwayyis?
I told him all this, he answered and said to me: how are you? Not
good? ULA 66,2 (cf. 75)
Tense, aspect, modality, negation, etc. are normally expressed only once in
true SVC, and only occasionally are both verbs marked for these. l,
yil constructions, however, usually mark both verbs in the same way,
that is, both verbs appear in the same tense or aspect (see the examples
above), with the exception of bi- and a-prefixes: bi- may be and a- is
always omitted on yil; see (74, 76) above. For bi-, see (84):27
(84) ilbitti bitbar)am til ?
What is the girl mumbling? SIG 126,11
dance and to sing: my darling went to the country ilLayy, he will never return to me, I
hope (Ba<ariyya: MandBa).
26
This excludes, by the way, causative xalla clauses from the serial verb constructions: xalltu ma/a I made him sign contains two agents/subjects.
27
Exceptional is katabu ylu ? What do they write and say? MRR 101,4 which
does not follow the above rules. It contains a sequence of a perfect and an imperfect
instead of perfect in both verbs, as in katabu lu ? What did they write and say? In
both cases, the question word what? asks for the opinion of the writers. Since the
act of writing happened in the past and the opinion expressed by this writing stays present till the moment of asking, an imperfect yilu seems justified for this moment. The
semantic notion here overrules the syntactic one; see Woidich (2003, 131).
695
This is usual when two imperfects of this type are coordinated, see
Woidich (2006, 282). Even different verbal modals with more or less the
same semantic content are possible, as (85) shows, when the verbs are
coordinated by wi and:
(85) umm& h illi inta id tixa++af wi amml tilu da?
What is it then what you are raving about all the time? ABM 39,1
SVC verbs should not contain two overt NPs that refer to the same
argument; see Kroeger (2004, 230). In l, yil constructions this is
possible and both verbs may keep their original semantic and syntactical
structure, cf. (73) and (77) above and:
(86) inta bitwawi ilbinti bitullaha ?
What are you whispering to the girl? MRR 198,11
(87) amml tikallim nafsak til ?
What are you talking to yourself ? LIB 64,12
Finally, as Newmeyer (2004, 3) reports, in a coordinate structure moving

an NP of one of the verbs from its original position should be impossible
(coordinate structure constraint), but it is possible in an SVC. In the
l, yil construction, noun extraction by forming a cleft sentence is
not possible: * lli raddi alk alhlak? or * lli akkallim alu? Strangely
enough, (85) with its coordinated verbs offers such a cleft sentence
containing a moved NP, though it should be not possible in this case
according to the coordinate structure constraint adduced by Newmeyer
loc. cit.
In my view, these are enough reasons not to consider l, yil constructions as SVC. Rather, I think that they developed from paratactic
entailments to intransitive verbs with a semantic component of saying
such as28 6anna to sing, yir to blame, ayya) to cry, bar)am to grumble,
daa to curse, dalla to give a nick-name, ikkallim to talk, kidib to lie,
na)a to pronounce, naa to give advice, radd to answer, rafa/ to refuse,
arrax to shout, wawi to whisper, xa++af to talk nonsense, which do
not have a syntactic slot to express the content of the performative act
and cannot quote directly. By addingbe it asyndetically or by means of
wia sentence with the locutionary verb l, yil that disposes of that
slot, the problem is solved. This coordinative structure describes one
28
The list of verbs with pseudo-complement as registered in my database. 13 of the
28 registered cases include the question l .
696
manfred woidich
fact and shows cohesion as to intonation, but not the syntactic cohesion
normally displayed in SVC. Following Seuren (1991, 196), I prefer to
interpret these l, yil constructions as pseudo-complements, especially
since l, yil is not the only type of these complements. Many verbs with
the semantic content of moving into a direction do not have a syntactic
slot for the direction of the movement. Here a pseudo-complement
with +, yir to go to fills the gap by introducing this direction, as in
mii + fn? Where did he go to?; see Woidich (2003, 181ff ) for more
details. Similar behavior is exhibited by xad, yxud to take and l, yil
to take away, which may be combined with wadda, yiwaddi to bring to
to form utterances such as anlu nwaddh fn AWL 44,6 Where shall
we bring him to?, imbri bi lll alha waddha lbt Yesterday evening
he took it home WAZ 369,-3f.
5. ql in delocutive derivations29
l, yil is combined with set phrases such as vocatives or idiomatic
expressions, thus enlarging the lexicon of the language. This is common
in Cairo Arabic. Here, it will suffice to give some examples:
(88) a))i dlu bn wi+ku w l ya fakk
It hung its tail between its legs and cleared off FAG 119,5
(89) aadt a/rab f l-yit lamma l ya bass
I kept beating him till he could take no more Hinds et al.
(1986, 74b)
(90) ittit dn ala txallha tl ai b raabti
An awful beating, which makes her give in RUH 63,2
(91) kunti bal ya ar/ inai w iblani
I said, O Earth split open and swallow me! = I wanted to vanish into
thin air LAB 218,4
Of a different type is l with sound-related interjections or ideophones

(or expressives, as they are sometimes called) to form descriptive
compounds. Only two examples have come to my attention, but there
are bound to be more:
(92) l wi
(93) l g2y2
29
He vomited Hinds et al. (1986, 947b)

He gave up, resigned30 Azz 129
Plank (2002, 465 and 478).

For gy ~ gy help! see Hinds et al. 1986, 146a; for the Arabic etymology, see
Behnstedt 1997, 36.
30
697
We quite often find this type of gl, yigl with sound-related interjections
as a direct object in Upper Egyptian Arabic, for example in Bri.31 They
mostly describe a sudden event connected with a sound, and occur
commonly in narrative style, not necessarily as intransitive expressions
as described in Plank (2002, 468).
+ gyil ka+a+w bha
He rushed at her
itgl dibb fi lar/
She toppled over
ma++a wida gl dradib
All of a sudden, it went knack
gl daradyy
He crashed down
gult ijlibb
I jumped up and ran away
gl, igl ))
to break wind
glat )irri minnh
It buzzed away
glat inn
It bubbled up (boiling water)
t ittit ul)a kida w gl hub, a+ glib ilgull wu lba wu adda min
r iml
A flat piece came, he made hub and threw down the jars and the bowl
and went off without a burden (Ba<ariyya: Manda)
(103) wu badn ilga)ir gl )), gallih ma tistanna lamma allabiyya tinif
And then the train made >>, he said to it, wait untill the gallabiyya
has become dry (Ba<ariyya: Manda)
(94
(95)
(96)
(97)
(98)
(99)
(100)
(101)
(102)
6. References
6.1
General References
Azz, Araf. 2005. al-Kinyt al-mmiyya al-miriyya. al-Qhira: al-Hadara

Publishing.
Behnstedt, Peter. 1994. Der arabische Dialekt von Soukhne (Syrien). Teil 2 und 3.
Wiesbaden: Harrassowitz Verlag.
1997. Koptisch oder Arabisch? Wiener Zeitschrift fr die Kunde des Morgenlandes
87, 3139.
Cantineau, Jean. 1937. Etudes sur quelques parlers de nomades arabes dOrient.
Annales de lInstitut dtudes Orientales III,119237.
Chisarik, Erika and Wim van der Wurff. 2003. From Say to Because: Grammaticalisation
and Reanalysis. Paper read at the Conference on Comparative Diachronic Syntax,
University of Leiden Centre for Linguistics (ULCL), 2930 August 2003. http://www.
let.leidenuniv.nl/ulcl/events/compdiachr/programme.htm
Cohen, David, Marie-Claude Simeone-Senelle, Martine Vanhove. 2002. The
grammaticalization of say and do: An areal phenomenon in East Africa. In
Gldemann, Tom and Manfred von Roncador, eds. Reported discourse: A meeting
ground for different linguistic domains. Amsterdam: J. Benjamins, 227251.
31
For the very elaborate use of these expressions in Ethio-Semitic, which goes much
further than what we know from these Upper Egyptian Arabic dialects, see Cohen et al.
(2002, 227 and 238 ff). Cf. for the Da7na Landberg (1909, 1268) gl hubs se mit
courir.
698
manfred woidich
Colin, Georges S. 1993. Le dictionnaire COLIN darabe Dialectal Marocain. Sous la

direction de Zakia Iraqui Sinaceur. Volume 6. Rabat: Editions al Manahil.
Fischer, Wolfdietrich und Otto Jastrow. 1980. Handbuch der arabischen Dialekte. Porta
Linguarum Orientalium, Neue Serie 16. Wiesbaden: O. Harrassowitz.
Georgakopoulu, Alexandra and Dionysis Goutsos. 2004. Discourse Analysis. An
Introduction. 2nd print. Edinburgh: Edinburgh University Press.
Golato, Andrea. 2002. Self-quotation in German: Reporting on past decisions. In
Gldemann, Tom and Manfred von Roncador, eds.. Reported discourse: A meeting
Gldemann, Tom and Manfred von Roncador, eds. 2002. Reported discourse: A meeting
ground for different linguistic domains. Amsterdam: J. Benjamins.
Heine, Bernd and Tania Kuteva. 2002. World Lexicon of Grammaticalization. Cambridge:
Hinds, Martin and El Said Badawi. 1986. A Dictionary of Egyptian Arabic. Beirut:
Librairie du Liban.
Hopper, Paul J. and Elizabeth Closs Traugott. 2003. Grammaticalization. Second edition.
Cambridge: Cambridge University Press.
Kissling, Hans-Joachim. 1960. Osmanisch-Trkische Grammatik. Wiesbaden: Harrassowitz.
Klamer, Marian A.F. 2002. Report constructions in Kembera (Austronesian). In
Gldemann, Tom and Manfred von Roncador, eds.. Reported discourse: A meeting
Kroeger, Paul R. 2004. Analyzing Syntax. A Lexical-Functional Approach. Cambridge:
Landberg, Carlo de. 1909. tudes sur les dialectes de larabie mridionale. Deuxime
volume. Da%nah. Deuxime partie. Leiden: E. J. Brill.
1942. Glossaire Datnois. Troisime volume ( w) publi par K.V. Zettersten.
Leiden: E.J. Brill.
Lord, Carol. 1993. Historical Change in Serial Verb Constructions. Amsterdam: J.
Benjamins.
Marais, William and Abderrahmn Guga.1959. Textes Arabes de Takrona. II Glossaire.
Sixime Tome. Paris: P. Geuthner.
Miller, Catherine. 2000. The Grammaticalization of the Verb to say in Juba Arabic.
Proceedings of the Third International Conference AIDA, Malta 29 March2 April 1998,
ed. by Manwel Mifsud, 213218. Malta: Association Internationale de Dialectologie
Arabe.
2001. Grammaticalisation du verbe gale dire et subordination en Juba Arabic.
Leons dAfrique: Filiations, ruptures et reconstitution de langues: Un hommage
Gabriel Manessy, ed. Robert Nicolai, 455482. Louvain: Peeters.
Musil, Alois. 1928. The Manners and Customs of the Rwala Bedouins. New York:
American Geographical Society.
Newmeyer, Frederick J. 2004. Some thoughts on the serial verb construction. Paper
read at: Atelier du 9 dcembre 2004. CRLAO, EHESS, Paris. La notion de construction
verbale en srie est-elle opratoire? Fdration TUL (internet: http://www.typologie.
cnrs.fr/fr/ gabarits/TUL%20Newmeyer.pdf).
Plank, Frans. 2005. Delocutive verbs, crosslinguistically. Linguistic Typology 9,
459491.
Schwenter, Scott and Richard Waltereit. 2005. Presupposition accommodation
and language change: From additivity to speech-act marking. Paper read at From
Ideational to Interpersonal: Perspectives from Grammaticalization. Leuven, 1012
February 2005. Abstract: http://wwwling.arts.kuleuven.ac.be/fitigra/schwenter.pdf
Seuren, Pieter. 1991. The definition of serial verbs. Development and Structure of Creole
Languages: Essays in honor of Derek Bickerton, eds. Francis Byrne and Thom Huebner.
Amsterdam: J. Benjamins, 193205.
699
Vanhove, Martine. 2004. Dire et finalit en bedja: Un cas de grammaticalisation.

Journal of African Languages and Linguistics 25,2, 133153.
Versteegh, Kees. 1984. Pidginization and Creolization: The Case of Arabic. (Current
Issues in Linguistic Theory, Volume 33). Amsterdam: J. Benjamins.
2007. Serial verbs. In Kees Versteegh et al., eds., Encyclopedia of Arabic Language
and Linguistics, vol. 3. Leiden: Brill.
Woidich, Manfred. 1980. XIV. Text aus il-Birt. In W. Fischer and O. Jastrow, Handbuch
der arabischen Dialekte, 235242.
1989. illi da, illi weil und zayy illi als ob: Zur Reinterpretation von
Relativsatzgefgen im Kairenischen. Mediterranean Language Review 4/5, 109128.
1995. Some cases of Grammaticalization in the Egyptian Arabic Dialects.
Proceedings of the 2nd International Conference of Lassociation internationale pour
la dialectologie Arabe, Trinity Hall, University of Cambridge, 1014 September 1995,
259268. Cambridge: Association Internationale de Dialectologie Arabe.
2002. Verbalphrasen mit asyndetischem Perfekt im gyptisch-Arabischen.
Estudios de dialectologa norteafricana y andalus 6, 121192.
2006. Das Kairenisch-Arabische. Eine Grammatik. Porta Linguarum Orientalium,
Neue Serie 22. Wiesbaden: Harrassowitz Verlag.
6.2
Literary Sources
ABM = Amn Bakr, Xamas masraiyyt kumidya. al-Qhira 1986.

ARA = Lu>f al-Xl 1988. il-arnib. Kairo.
AWL Al Slim. 1986. awladna f Landan. al-Qhira.
BAHN = Smi< Farag. 1999. Bnhftrsa. Riwya bi l-mmiyya l-miriyya. al-Qhira.
BTR = Al Slim. 1991. ilBitrl )ili fi betna. al-Qhira.
DAR = usayn afq al-MiDr. 1929. il(aggi Darw w Umm Ismal. al-Qhira.
FAG = A<mad Fud Nigm. 1993. Muzakkirt ilFagmi. al-Qhira.
FWQ = Numn r. 1958. inNs illi f. al-Qhira: Dr an-Nadm.
GIL = Numn r. 1972. igGl iggidd. al-Qhira.
HAM = Bahg Isml. 1994. iln ilamra. al-Qhira.
HUD = Fat<iyya lAssl. 2002. (u/n al-umr. as-sra -tiyya. al-Qhira: al-haya
al-miDriyya al-mma li l-kitb.
LAB = Ysuf al-Qad. 1994. Laban ilafr. Riwyt al-Hill 545. al-Qhira.
LAY = Mu<ammad Inn. 1993. Laylat a-ahab. al-Qhira.
LIB = Mu<ammad Kaml Mu<ammad. 1994. Libit itttalib. al-Qhira.
MAL = Motie Ibrahim Hassan. 1971. in-Ns wil-malik. Copenhagen.
MAX = Ma<md Taymr. 1979. ilMaxba raqam tala))ar. An annotated phonemic
transcription by Stig. T. Rasmussen. Kopenhagen: Akademisk Forlag.
MRR = Rad Rud. 1975. Masra Rad Rud. al-Qhira.
MUW = Wald Ysif. 2000. ilMuw)in mahri. 2000. al-masra< al-arab 142. al-Qhira.
NAZ = Fat<i RaJwn. 1973. Nir Waqf. al-Qhira.
NUS = al-mmiyya al-miriyya f nu wa muwart. Nederlands-Vlaams Instituut
te Caro w.d.
RAQ = Ysuf f. 1997. Rqiat qi) mm. al-Qhira.
RAS = Nihd Gd. 1989. a rRaf. al-Qhira.
RUH = Kaf Abd al-Munim. 2005. Min (alwit irr. Riwya bi l-mmiyya. al-Qhira:
Madnat 6 Uktbar: Sanbil li n-nar wa t-tawz (a>->aba a%-%niya).
SIB = Sad ad-Dn Wahba. 1966. isSibinsa. Masra<iyyt Sad ad-Dn Wahba Nr. 149.
al-Qhira.
SIG = Fat<iyya al-Assl. 1993. Sign innis. al-Qhira.
SIK = Sad ad-Dn Wahba. 1967. Sikkit issalma. Masra<iyyt Sad ad-Dn Wahba
Nr. 184. al-Qhira.
700
manfred woidich
SMB = Bayram at-Tnis. 1925. isSayyid wi m+tu f Bars. al-Qhira.

SMM = Bayram at-Tnis. 1925. isSayyid wi m+tu f Mar. al-Qhira.
TAY = Mu<ammad Taymr. 1922. al-Masra al-mir. Muallaft Muammad Taymr,
al-juz a7-7li7. al-Qhira 1341 (1922).
ULA = Mu<ammad NDir. 2000. la awwil. al-Qhira: al-Majlis al-al li %-%aqfa.
WAZ = Sad ad-Dn Wahba. 1980. ilwazr l ittallga. al-Qhira.
XAMA = A<mad ams ad-Dn al-ajjj. 1988. ilXamasn. al-Qhira: al-masra<
al-arab.
YUN = Mu<ammad Inni. 1993. (alwit Ynis wi masraiyyt uxr. al-Qhira.
INDEX
, 571, 571 n. 24
ending, 544
Abbeche Arabic, 648
Abbs, 315, 363
Abd al-l, 665 n. 8, 671
Abd al-Fatth Salm, 187
Abd al-H usayn al-Fatl, 22, 187
Abd al-Qhir al-Jurjn, 232
Abd as-Salm Muhammad Hrn, 23,
43, 64
Abda**a lGum, 678
Abdallh, 56, 61
Abdalmalik b. H abb, 195, 195 n. 10
Abdeen, 662
Abdel-Malek, Kamal, 563
Abderrahmn Guga, 698
Abdh, Ms, 548, 563
Abdul-Malek, 544
Abdul-Raof, 393395, 395 n. 85, 400
abitud, 215
abitudines, 214215, 226
ablatiuo, 214, 216, 227
ablative, 216217
ablativus, 221
Ablaut, 256
abn al-Ajam, 115
abn al-Arab, 115
Abney, 475477, 478 n. 3, 483, 485, 498
abniya, 316
Abraham, 324 n. 14, 363
Ab Abdallh M. b. A. al-Kinn, 208
Ab Al al-Fris, 10, 165
Ab l-Aswad ad-Dual, 116, 120, n. 4,
133
Abu Athera, Said Salman, 543 n. 1, 563
Ab Bakr M. b. Abdallh b. Muhammad
b. Uta al-Isbahn, 200 n. 30
Ab Dwd, 93, 110
Ab H anfa, 35
Ab l-H asan [al-Axfa], 167
Ab H usayn, 43
Ab l-H usayn al-Basr, 38, 4041
Ab Jafar Quray b. Uqba b. Bar, 198
Ab l-Qsim Abd al-Rahmn b. Ishq,
113
Ab Qr, 130
Ab t-Tayyib al-Lughaw, 206 n. 45
Ab R Academic Hospital, 533

Ab Ubayda, 64
Ab Xir, 90
Ab Ysuf, 43
Abu Zinima, 566 n. 6
Abu-Haidar, 668669, 671
Abu-Lughod, Lila, 544, 563
Abuw lHl, 569 n. 16
Abuw Znmah, 566, 566 n. 6, 578
academy(ies) of, 655, 661
Baghdad, 661
Cairo, 661
Damascus, 661
Jordan, 661
Acc-ing, 478, 483, n. 10
construction, 476
accompli, 315 n. 4
du prsent, 331
accusatif, 118, 130
accusative, 78, 10, 1819, 19 n. 15, 95,
135, 135 n. 2, 138, 138 n. 19, 139140,
142, 144147, 216217, 227228, 230
ending, 138, 144
+genitive, 135
accusativus, 221
acento, 215 n. 11
achvement, 334, 341, 341 n. 24, 347
du processus, 333334, 339
Ackema and Neeleman, 475476, 478
479, 480 n. 5, 481, 481 n. 6, 482, 482
n. 8, 483, 483 n. 10, 484485, 495, 497
n. 30, 498
acquis, 326, 334335, 339341, 350
caractristique, 326327, 329331,
334, 337341, 344, 348, 348 n. 29, 351
non, 327, 330331
constitutif, 338
dernier, 326, 331, 333, 339340, 349
final, 326
instable, 338
passager, 327 n. 18
permanent, 320
rsultatif, 325, 325 n. 15, 326328,
330332, 339, 341, 343, 349350,
352
non-caractristique, 329
temporaire, 335, 338
702
index
transitoire, 327 n. 18
acquisition
dtat, 326327, 330, 334, 338340,
342343, 346, 358, 360
acteur, 335337
lactif, 115
activities
durative, 299
habitual, 299
actors, 396
acusatiuo, 214, 216
acusativo, 227
d, 195, 202203
Adab literature, 199
Adam, 39, 190, 192195, 197198, 200, 206
adaptation(s)
morphological, 308
phonological, 306, 308
add, 86
Aden, 8283, 83 n. 18, 84
adjectif(s), 325329, 331332, 335, 342,
346347, 349350
analogue(s), 327, 341
assimil, 327, 331333, 341342,
345352
dtat
acquis, 334
caractristique, 328, 333335,
339342, 344345, 351354, 358,
360
non-caractristique, 328, 331, 333, 335
adjectival agreement, 29
adjective(s), 17 n. 13, 49, 153 n. 5, 154,
156, 156 n. 11, 157, 158 n. 13, 160,
167, 176, 189, 248249, 250251, 308,
370 n. 14, 373, 376 n. 34, 382 n. 53,
477, 484485
arab, 190
feminine, 250
indefinite, 251
nominalized, 382
adjunct
adverbial, 391
attributive, 391
adjunction, 293
Adnn, 201, 203
adverb(s), 214, 229, 252, 374375, 384,
385, 397, 477, 577
interrogative, 398 n. 90
sentence, 685
temporal-, 63
adverbial, 150, 153 n. 5, 159160,
162163, 167, 169 n. 26, 170, 172,
180, 183, 371 n. 17, 375, 384, 387
constituent, 166
degree, 384
expression, 682
manner, 384
phrasal, 387
phrase, 150, 152, 153 n. 5, 159,
162163, 163 n. 20, 164, 164 n. 23,
167168, 171172, 177180
prepositional, 175176, 179180,
183186
semantic component, 170
sentence, 376378, 387
structures, 54
time, 384
af l
al-madh wa-d-damm, 378 n. 4
an-nqisa, 378 n. 41
al-muqraba, 378 n. 41
al-qulb, 142, 378 n. 41
ar-raj, 378 n. 41
a-ur, 378 n. 41
at-tafdl, 378 n. 41
affirmatif, 519
affirmation, 398
affix, 370 n. 11, 477480, 481 n. 6,
482483, 483 n. 10, 484, 484 n. 12,
485, 492, 497 n. 30
internal, 483
masdar, 485, 492, 494497, 497 n. 28
nominal, 480
nominalizing, 479, 482483, 483
n. 11, 493
phonological, 481, 496
prosodic, 495
stem, 490
syntactic, 481, 496
Affix Grammars over Finite Lattices,
369, 369 n. 10, 370, 372 n. 18, 373
n. 23, 377 n. 36, 383
affixation, 367
affixes, 230, 479481
non-, 372
stem, 488
terminal, 372
Africa, 621, 639
East, 639, 697
African, 607, 609, 611, 621
East, 620
West, 692 n. 23, 693
affricate, 597
affrication, 611
Afroasiatic, 595
Agapito Valle Flemmarum, 212, 226,
231, 232, 233234, 234 n. 48, 237, 239
Agylah
dialect(s), 567, 572 n. 25
index
agent, 321, 321 n. 11, 322323, 326, 326
n. 17, 343, 348349, 351 n. 32,
354357, 376 n. 31, 380, 392394,
394 n. 83, 395, 396 n. 85
explicit, 391
humain, 348, 351352, 356
implied, 391
agentives, 394
agentivit, 321, 321 n. 11, 323, 328 n. 19,
336, 344, 350351, 359
entire, 323, 349, 359
neutralise, 321322, 335, 338,
343344, 348 n. 29, 351, 359
non, 321, 343344, 359
partielle, 321, 323, 344, 347, 349, 349
n. 31, 351, 359
pleine, 321, 344
Aggadic material, 192
Agha & Khalidi, 190 n. 1
Agha, Saleh Said & Tarif Khalidi, 207
agreement, 368, 370, 374, 376 n. 31, 368
n. 4, 377, 381, 383384, 389, 397
Aguila, Antonio de, 212, 231232
gyptisch-Arabisch, 699
Ahaywt, 566 n. 9
dialects, 567
ahl al-arabiyya, 55
ahl al-lua, 39
Ahlwardt, Wilhelm, 21, 86 n. 23, 87
n. 23, 110
Ahmad Fud Nigm, 544, 563, 699
Ahmad Muhammad Shkir, 205 n. 41
Ahmad Muxtr Umar, 65
Ahmad ams ad-Dn al-H ajjj, 700
Ahmad Ysuf Najt, 64
Ahmad, 364
al-Ahrm, 530
id, 59
air, 256, 256 n. 8, 257 nn. 1213, 261,
262, 268, 270
ajam, 115116
Akkadian, 597604
early, 598
pre-, 598
lakkadien, 125
al llad, 79
al, 17, 21
Al-Ani, Salman H., 247
Albany, 202 n. 33
Alcal, Pedrode, 209 n. 1, 210214, 214
n. 9, 215217, 217 n. 15, 218220, 220
n. 22, 221222, 224226, 231 n. 41,
239, 241
Alegat, 565 n. 3, 566 n. 7
Aleiqat, 565 n. 3, 567 n. 9
703
Alep, 113, 123

aleph(s), 595, 596 n. 6, 597, 597 n. 8,
599 nn. 13, 15
Aleppo, 232 n. 44
Aleut, 294
Alexandrie, 130
Al-Falay, Ibrahim S., 655, 670671
al-Alf, 530
Alfiyya commentaries, 3
alayta, 141 n. 32, 160
Alger, 113
Algerian, 640
Arabic, 596
Al-Hassan, 671
Al b. Ab Tlib, 134, 206
Al H aydar, 23
Al Slim, 699
Al Tawfq al-H amad, 187
Al, 117
Ali, 2627, 34, 3637, 3942
Al, caliph, 31
Ali, Mohamed M. Yunis, 44
alif cequin, 218
alif, 123 n. 12, 127, 598
mdda, 684 n. 13
8skin, 218
all, 17
al-la, 658
a&&a, 658
allad, 6772, 7681, 81 n. 8, 82 n. 13,
83 n. 19, 8485, 8889, 91, 93, 103
clause, 68, 7071, 82, 85, 8889,
9597, 102
conjunction, 75, 100101
conjunctional, 70, 75, 79
infinitival, 79
Allh, 69, 72, 89, 9192, 97, 109, 200
allemande, 118119
allemands, 118, 123
allophone(s), 273
etymon, 272273, 568, 576, 657
alternance(s)
vocalique(s), 315, 317, 337, 344, 345
n. 28, 350, 353
Altoma, Salah J., 656, 671
am, 17, 21
amal, 1011, 124, 140 n. 30, 141142,
142 n. 37, 143146, 149152,
152 n. 3, 153, 159 n. 14, 160, 164,
166167, 171172, 180181, 184186,
224, 228, 234235
jarr, 10
jazm, 10
nasb, 10
raf , 10
704
index
Amara and Spolsky, 662

Amari, M., 84, 110
American
English, 661
British, 560
Americans, 560 n. 40, 561, 684
Amerindian, 210
al-mid, 37
mil, 10 n. 8, 124, 135147, 149,
151152, 160, 160 n. 15, 161162,
162 n. 18, 163, 168, 170, 175,
178179, 181, 184, 212, 234, 237, 240
theory, 62
Amn Bakr, 699
amir, 217
$mmiyya, 403, 405, 411, 529, 529 n. 1,
531532, 532 nn. 78, 533537,
539540
amm, 13 n. 9, 1517, 21
Ammar, Sam, 315, 318 n. 7, 319, 323,
343 n. 25, 344, 344 n. 26, 353 n. 33,
363
amorite, 316
amr, 6, 34, 210, 217
Amsterdam, 565, 565 n. 2
an(na) clause, 24, 95, 99, 171
an-, 7 n. 5, 15, 79
8-clauses, 78 n. 3
analogical creations, 661
analysis
automated, 390, 391
etymonial, 258, 261262, 265
anaphora, 171
anaptyctic(s), 572, 574, 582 n. 60
anaptyxis, 571572
Anawati & Jomier, 93, 110
Andalousie, 128
al-Andalus, 78, 435, 438441, 444,
447449, 453
Andersen, R., 308309
angels, 190
Anghelescu, Nadia, 19 n. 16, 23
animal, 345, 346, 396
animate, 367, 370, 383, 394, 396397
non-, 394, 394 n. 81
animit, 520, 522
AnBs, Ibrhm, 364
annam, 16
Ansr, 202
antecedens, 81 n. 8
antecedent, 59, 9192, 151, 153
Antilles, les, 129
antinomy, 367 n. 3
Antiquit grecque, 114
Anubhti Svarpcrya, 209
apercion, 225
apocopate, 227, 230 n. 39, 248, 575576,
581 n. 56, 592 n. 91
Apostolou Panara, A., 299, 301303,
307, 309
apposition, 69, 71, 102, 169 n. 26, 182
clauses, 72
approximant, 266, 266 n. 28, 267, 272,
280, 280 n. 41, 285286
aprs
vnementiel, 325
Aquila, Antonius ab (Antonio
dellAquila), 123, 212 n. 4, 223, 237,
239, 241
Aquilina, Joseph, 100, 111, 296, 310
Arab, 26, 41, 236, 527, 534, 540, 543,
543 n. 1, 551, 557
countries, 541
genealogists, 201202
genealogy, 202
grammarians, 70, 79, 206 207
grammatical theory, 3
grammatical thinking, 45
linguistic studies, 41
literary studies, 543
medieval grammarians, 595 n. 3
Muslims, 67
nationalists, 197
native speakers, 9
North, 201
philosophers, 192
poetry, 3, 4 n. 1
theologians, 192
tribal groups, 201
world, 527528, 541, 543 n. 1, 561,
563
writers, 543
arabe, 35, 114, 116121, 121 n. 8
122123, 131, 133
dAbbeche, 652
ancien, 118119, 122, 124125, 130,
313, 315317, 319, 321, 321 n. 10,
323, 325, 327, 332, 339, 343 n. 25,
349, 353 n. 33, 354355, 355 n. 36,
357, 362, 501, 505, 508, 509 n. 21,
509512, 515 n. 33, 519, 520521, 621
dialectes, 125
classique, 116 n. 2, 119, 121 n. 7,
124125, 126, 128 n. 18, 130, 335
n. 20, 510
contemporain, 347, 353
crole, 621
dialect, 128
dialectal, 119, 121 n. 7, 122, 124 n. 13
gyptien, 335
index
fash , 120
histoire de, 118-119, 120 n. 6, 125
marocain, 698
moderne, 116 n. 2
moyen, 118119, 128 n. 18
no-, 118119, 122, 124125, 130
sdentaris, 117
vulgaire, 124, 124 n. 13
Arabes, 115117, 120, 128129, 133
langue des, 117
populations, 114
populations, non-, 114116, 133
arabes, no-, 124
arab, 189
Arabia, 201, 563
Central, 563
Northern, 563
Arabian, 201
Ancient West, 652
Old South, 597, 599603
Arabic, 9, 9 n. 7, 26, 35, 38, 61, 6768,
70, 72, 75, 79, 87, 87 n. 24, 88, 90,
9293, 95, 98, 101, 104, 106109, 114,
119, 157, 165166, 168169 n. 26
171172, 177, 181, 189211, 211 n. 2,
212, 213 n. 7, 214, 216218, 221, 222
n. 26, 223, 223 n. 28, 224, 224 n. 32,
225230, 232, 232 n. 44, 233, 233
n. 45, 234235, 235 n. 51, 236,
238240, 247248, 250, 252253, 255,
260 n. 21, 266 n. 28, 272, 283, 292,
292 n. 2, 295296, 296 n. 4, 297, 306,
309
dAbbeche, 648
Adams-, 192
Algerian, 596
alphabet, 200201, 219220
Cairene, 7375, 85, 9495, 100102,
104, 568, 586 n. 70
of Cairo, 696
Christian, 71, 89
classical, 13, 68, 72, 72 n. 2, 7677,
95, 98, 100, 155, 189, 210, 216219,
222224, 227, 229230, 237, 247,
368 n. 5, 369, 369 n. 10, 370, 370
n. 12, 377 n. 37, 381, 393, 395, 396,
399, 403, 404, 405, 405 n. 2, 410,
415, 418, 420, 421, 421 n. 15, 427,
431, 433, 435, 437439, 449,
455456, 458, 465468, 470,
471472, 475, 486, 486 nn. 1617,
487 n. 18, 489, 490 n. 22, 491492,
492 n. 25, 493, 493 n. 25, 527, 529,
529 n. 1, 530531, 535, 537,
540541, 544, 549 n. 9, 555 n. 31,
705
559, 565 n. 1, 567568, 578, 595,
595 n. 3, 596, 596 nn. 56,
597598, 598 n. 11, 599604, 604
n. 17, 604605, 607, 609610, 614,
618, 619621, 639, 640, 641, 640
n. 1, 643, 648, 649, 650, 651, 652,
675, 699
colloquial, 211212, 214, 223224,
226, 595596, 596 n. 6, 597, 605,
621
Damascene, 224 n. 31
of Granada, 222
Northern Sudanese, 608
Northern
Creole, 621, 639
Juba, 611, 698
Damascene, 238
corpus, 532 n. 5
dialect(s), 189, 578, 595, 677, 687,
690, 693
dialectology, 72
dialects, 104, 595 n. 3, 596597, 604,
604 n. 17, 658
dictionary, 221
diglossia, 654, 656, 671
doxology, 107
early, 201
educated Cairene, 7
Egyptian, 675, 692693, 698699
grammar, 23, 25, 35, 42, 75, 200, 438
grammatical
literature, 79
sources, 23
terminology, 213, 219 n. 19
terms, 210212, 224
theory, 216, 219
tradition, 23, 227, 237
early, 23
Granadan, 212
graphemes, 540
history, 189
history of, 192 ,211 n. 2
Jewish, 296, 299
literary, 370
Maghribian, 291, 296297
Magribian, 309
Moroccan, 292293, 295297, 308
Modern Standard, 248249, 251,
253, 367369, 372 n. 19, 373,
376, 376 nn. 31, 35, 377 n. 39,
379380, 384, 387388, 390394,
398, 568569, 586 n. 70, 598
urban colloquial, of Damascus, 211
vernacular, 210
imperfect, 70
706
Ismls, 197
Jewish, 101
Juba, 128, 128 n. 19, 607612, 616,
618621, 692693, 698
Judaeo, 68, 7172, 72 n. 2, 100, 100
n. 30
dialect of Tunis, 100 n. 30
Juna, 617
Jurhum, 200
Khartoum, 608
Koranic, 531
Kuwaiti, 563
language, 47, 89, 654655, 661,
669672
history, 639, 649, 651
Lebanese, 99
letters, 220
lexical items, 668
lexicography, 196
linguistic history, 652
linguistics, 28
literature, 544
literary document, 189
literary, 84
Maltese, 100
masdar, 484
media, 530531, 532 n. 5, 536, 540,
543
medieval, 181182
middle, 72, 72 n. 2, 118
modern, 541
modern Cairene, 72
modern literary, 95 n. 28
Moroccan Jewish, 81 n. 10
mother tongue, 662
native speakers, 9, 9 n. 7
North, 201
neo, 7576, 95, 103, 118, 639
new, 649, 651
newspapers, 654
Nigerian, 639641
Northern Sudanese, 609
old, 118119, 639, 640 n. 1, 650,
651
orthography, 189
Ottoman, 619
poetry, 204
popular, 544
pre-diasporic, 640 n. 1, 641, 649
pre-modern, 79
pre-neo-, 77
Prophet Muhammads, 205
of prophetic language, 189
Quraishi, 200
religious pre-history of, 192
index
scholars, 194
script, 200201
sentences, 181
sources, 211 n. 2
South, 201
speaking countries, 656, 661
speech, 67
spoken, 652
standard, 621
Egyptian, 679
standardization of, 189
studies, 528, 565 n. 1
Sudan, 685 n. 14
Sudanese, 608
sudanic, 648
syntax, 88
Syrian, 86
terminology, 224, 227, 231, 233234,
236, 236 n. 53, 237, 239240
terms, 214, 216, 218, 226, 231, 233,
237
tradition, 211 n. 2, 216, 220, 227
Tunisian, 100101, 104
Upper Egyptian, 697, 697 n. 31
Uzbekistan, 651 n. 13
vocabulary, 189, 661
Western Sudanese, 614
Western Sudanic, 611, 640
writing, 200
written, 528, 541
Yarubs-, 199, 201
arabicization, 235
arabicizing, 661
Arabie, 120
centrale, 120 n. 4
mridionale, 563, 698
arabique
est-, 125
ouest-, 125
arabisants, 113, 118, 121, 123, 125
arabisation, 130
Arabisch, 697
Alt-, 118, 119
Mittel-, 118
Neu-, 118, 119
Arabists, 67, 527
Western, 104
al-arabiyya, 53, 119120, 121 n. 7
Arabized, 620
Arabness, 189
Arabs, 26, 33, 40, 86, 106107, 191, 194,
197199, 201, 203, 205, 207
North, 202, 217, 227, 230, 235236,
240, 528, 534, 543, 620, 642
non-, 620
index
arafa, 660
Araic
Classical, 602
Aram, 70
Aramaic, 72, 189, 197, 201, 595, 597,
600, 605
Biblical, 597, 599601
Jewish, 600601
Old, 601
Arazi & Masalha, 110
Arberry, A.J., 87
arbre
conceptuel, 358, 359
archaisms, 595
Archimde, 501
ard sr, 194
Arfaxad b. Sm, 195
rid, 169
al-Ar
dialect of, 570 n. 21
Aristotelian
dialectic, 34
logic, 25, 41
Aristotle, 33, 41
arrangement
morphological, 437
orthographical, 437
article(s), 209 n. 1, 210, 211 n. 2, 217
n. 15, 226, 571, 573574, 576, 591 n. 88
al-, 573574, 576
il-, 573574, 576
definite, 214215, 217, 306, 373, 411
indfini, 509, 515 n. 33
indefinite, 305
separate [= not suffixed], 229
an, 76
Ashtor, Eli, 82, 82 nn. 1314, 110
Asia, 232, 639
Asn, Palacios, 222
asl, 438, 448, 507, 507 n. 18
marf, 169
asm al-af l, 178
ASP, 72 n. 2
aspect, 294, 370372, 384 n. 55, 389,
395
perfect, 378, 378 n. 42, 520, 522
Assaf, Simha, 84, 110
Assemani, Joseph Simon, 220, 220 n. 22
assimilate, 574
assimilating, 574
assimilation, 127, 568, 579 nn. 43, 48,
581 n. 55, 586 n. 73, 604
of + h, 568
of initial h of suffixes, 568
Assise, Saint Franois d , 121 n. 8
707
Assyrian, 601
Astarbd, 162 n. 19, 163 n. 20, 164,
164 nn. 2223, 169 n. 26, 170 n. 28,
171172, 175177, 186
Ar, 681
aswad, 659, 664
Ataba, 538
tr, 30 n. 3
athnach, 219 n. 19
atnach, 219 n. 19
Atif Efendi, 44
attention, 398
attribute, 71
attributive
clauses, 6869
relative clauses, 69
Audebert, Claude, 335, 363
audience, 249, 405, 424, 432433
Auezova, Zifa, 437, 446, 453
augment(s), 440444, 446, 448
internal, 444
augmentation, 441
Austronesian, 210
auxiliaries, 293, 302 379
finite, 293
Auz, 199
avant
vnementiel, 325
Avi Shivtiel, L.J., 595 n. 1, 600601
Avokaya L., 609
AwaJ, L., 403 n. 1, 434
awil
-collections, 198
-literature, 198
awmil, 1011, 140, 141 n. 32, 142, 160,
185, 232, 234
8operants, 10
Awld Sad, 566 n. 9
Awld Silmiy, 567 n. 3
Aws Ab d, 195
awwaliyya, 507 n. 18
awzn, 314
al-Axfa, 45, 46 n. 3, 47 n. 4, 62, 51 n. 8,
64, 154 n. 10, 157, 167168, 175176,
180182
ya, 29, 46, 46 n. 1, 5054, 56, 58, 60, 191
ayn, 540
ayna, 16, 20
Ayoub, G., 378 n. 41, 400
el-Ayoubi, 374 n. 25, 375 n. 29, 381
n. 51, 385 nn. 56, 59, 386, 386 nn.
6061, 398 n. 91
ayy, 14 n. 10
Ayyoub, Georgine, 654, 655 n. 1,
664665, 669, 671
708
index
al-A'am, Muhammad Musta f, 200,

200 n. 31, 201, 207
Azz, Araf, 696697
al-Azraq, 195, 195 n. 12, 207
Baalbaki, Ramzi, 3, 6, 7 n. 5, 8 n. 6,
1223, 230 n. 39, 235 n. 49, 242
Bb Tma, 222 n. 26
bba, 658
**ai, 658
Bbil, 196, 202
Babylon, 196, 203204
Babylonian
Old Egyptian, 692 n. 23
Tower, 191, 196
back consonant, 658
Backus, A., 294, 310
bada llad, 79
badal, 169 n. 26, 182
badam, 17, 19
Badrah, 566 n. 9
Badawi, El Said, 158 n. 13, 187, 313,
314 n. 3, 363, 371 n. 17, 374 n. 25,
379 n. 47, 381 n. 51, 385 n. 56, 386
n. 64, 387388, 388 n. 67, 400, 542,
655656, 661, 664665, 671, 698
Badawi-Hinds, 532, 538 n. 23, 542
al-bdiya, 115
Badrn, 557 n. 35, 559
Bagdad, 113
Baghdad, 667, 669, 671
Bahariyya, 680, 684, 694 n. 25, 697
Bahg Isml, 699
Bahga, 683
Bahloul, Maher, 368 n. 4, 371 n. 17, 400
Bahr al Ghazal, 610, 614
Bahri, A., 272, 287
Bailey, Clinton, 544, 549 n. 11, 563,
566 n. 4, 566 n. 7, 566, 570 n. 18,
577578, 585 n. 68, 586 n. 75
Bailly, A., 276 n. 37, 287
Bakhtin, M.M., 408, 434
bala, 23
balbala, 194, 196
Bangala, 609
Ban Ati yya, 561562
ban firn, 545 n. 2
Ban H asan, 546
Ban Sad, 206
Ban Sad b. Bakr b. Hawzin, 206
Baniy Wsi, 567 n. 9
Ban H m, 203
Ban Hshim, 204205
Ban Isml, 195
Ban Qaht n, 195

Ban Rala bint Yajub b. Yarub b.
Ldan b. Jurhum b. mir b. Saba b.
Yaq tn b. bir b. lix b. Arafxaad
b. Sm b. Nh, 199
Ban Yarub, 204
baqa, 660
al-Bqilln, 34, 38, 43
baqiya, 660
Baram, Amatzia, 197 n. 17
Barhebrus, 107
Bari, 609611, 618620, 635, 638
Barrk Dish Ab Tyih, 562
base
bi-consonantal, 256
basilectal, 608, 611612, 614, 615, 617,
620
Basra, 30 n. 3, 64, 114, 117, 133
Basran(s), 160 n. 15, 175176
Basrien, 117
Bassiouney, R., 407, 434
Bath-party, 197
Baumgartner, Walter, 605
Baumstark, Anton, 105107, 111
Bay, 610, 614, 617, 620, 626, 631
bayn, 38
al-Bayhaq, 203
Bayram at-Tnis, 563, 685, 700
Bedja, 693 n. 23, 699
bedouin
Arabic, 599
dialects, 565 n. 2, 567, 573, 575, 586
n. 70, 657658
Sinai, 568, 571
speech, 658
tribe, 565566
usage (qawl al-arab), 47
Bedouin(s), 25, 30, 46, 543 n. 1,
544547, 548 n. 6, 550 n. 22, 552,
553 n. 27, 557 n. 35, 560 n. 40,
560563, 565, 565 n. 2, 566568,
571, 573, 575, 577578, 586 n. 70,
586 n. 75, 657
badu, 545
contemporary poetic tradition, 544
Egyptian, 578
in Iraq, 546
Iraqi, 546
Jordanian, 562
law, 578
Nothern Arabia, 563
of Jordan, 546
poet, 545 n. 4, 560
poetry, 544545, 560 n. 38, 563
index
poets, 544545, 560, 563
popular poetry, 544, 559
Rwala, 698
Sinai, 545 n. 2
society, 544, 563
tribal poetry, 561
Bdouins, 114115
Beersheva, 543 n. 1
Behnstedt, Peter, 567578, 652, 690, 696
n. 30, 697
Beja, 685 n. 14
Bekir, M., 43
Belot, J.B., 313, 314 n. 3, 363
Ben Cheneb, Mohammad, 113, 232
n. 43, 242
Benares, 209
beneficiary, 394396
Bengal, 209
Bengali, 679 n. 7
Benjelloun, Fauzia, 674
Benveniste, mile, 320, 335337, 350,
356, 363
Berk, 105106, 108109
Bergman, 375 n. 27, 387 n. 66, 400
Bergstrer, G., 33 n. 4, 44
Bri, 677, 678 n. 6, 697, 690 n. 20
Berrendonner, 506 n. 16
Beyrouth, 113
Beziehungsnomen, 69
Bhatia, K., 671, 673
Bible, 189, 196, 611, 623
Arab, 223 n. 28
Hebrew, 219 n. 19
Jewish, 189
Biblical, 106, 198
Aramaic, 597, 599601
Hebrew, 595, 595 n. 4, 596, 599, 599
n. 13, 602603
material, 192
Bickerton, Derek, 698
Bihr, 209
Biir, 618
bilabial stop, 595 n. 3
bilingual(s), 291, 293, 295, 301, 308, 408,
433
bilingualism, 291, 295, 297, 299, 301,
307, 309, 655 n. 2
homeland, 308
immigrant, 308
biqi, 660
Br anNasb, 566 n. 8
Bir Nasb, 565 n. 3
biradical verbs, 449
Blachre, Rgis, 125, 131, 313, 363
709
Blair, Tony, 561

Blanc, Haim, 121 n. 10, 131, 567 n. 13,
570 n. 20, 570 n. 22, 573 n. 31, 574 n.
32, 575 n. 37, 578579, 579 n. 49, 586
n. 71, 654, 656, 658659, 664, 671
Blas Francisco de Salamanca, 210 n. 2
Blau, Joshua, 68, 68 n. 1, 6972, 74, 77,
80, 89, 9394, 99, 111, 119, 131, 654,
655, 671
bleaching process, 675, 687
blending, 256259, 259 n. 20, 265, 272,
277 n. 38, 283, 285, 661
Bobzin, H., 109, 111, 671, 673
Bohas, George, 4, 23, 255, 255 nn. 2,
45, 256, 257 n. 14, 258, 259 n. 18,
260 n. 22, 262, 264 n. 27, 265, 269
n. 30, 272, 279 n. 40, 283, 285 nn.
4849, 287288, 371 n. 15, 400, 448,
453
Bonaventura da Molazzana, 211 n. 2
Boormans, Maurice, 313, 314 n. 3, 327
n. 18, 363
Booth, Marilyn, 544, 563
Borer, Hagit, 494, 498
Borg, Gert, 527
borrowing
foreign lexemes, 662
terms, 661
Bosworth, C.E., 200 n. 28
Boucherit, Aziza, 653, 670, 671
Boumans, Louis, 291, 292 n. 12, 293,
309310
Boussofara-Omar, Naima, 404, 404 n. 2,
407, 434, 654, 671
Brague, Rmi, 502 n. 2, 522
Braham, A., 315, 363
Braslavski, 83, 110
Bravmann, Josef, 28 n. 1, 6870, 111
Bravmann, Meir, 44
Brazilian, 304305
British, 544, 546, 560
Brockelmann, Carl, 86, 86 n. 23, 87
n. 23, 313, 313 n. 3, 327 n. 18, 363,
599
Broselow, Ellen, 671
Bubenik, Vit, 595 n. 1, 601602
Buburuzan, Rodica, 28, 44
al-Buxr, , 78, 90, 110, 199
bukara-Dialekte, 578
bukarra, 578, 580 n. 52, 583 n. 65
Blq, 161
bunya, 435
Burton, John, 65
Bush, George, 560
710
index
CA, 72 n. 2, 540, 659

Caballero, 210 n. 2, 224, 224 n. 31,
225231, 233, 238, 240241
Caballero, Lucas, 209 n. 1, 210 n. 2, 211
n. 2, 222, 224, 224 n. 31, 237
Cachia, P., 434, 403 n. 1
cadi, 117, 221 n. 26
Cadora, Frederic, 655, 667, 671
CAI, 659 n. 4
Caire, 113, 131
Cairene, 75, 80, 9495, 98, 100, 568, 586
n. 70, 591 n. 87, 650
Arabic, 76, 78, 86, 568, 586 n. 70
modern, 72, 94
Cairo, 83, 101, 528, 598 n. 12, 684, 686,
687 n. 17
Arabic, 696
Genizah, 101
Old, 79, 80 n. 6
caliph Al, 31
calque, 655, 661, 668
Cameroon, 640
Camilleri, A., 296297, 310
Canadian, 302
Caes, Francisco, 210, 210 n. 2, 211,
212, 222, 224227, 229233, 238,
240241
canonical, non-, 61
Cantarino, 374 n. 25, 381 n. 51, 386
cantillation marks, 219 n. 19
Cantineau, Jean, 283, 288, 314, 316, 316
n. 5, 335 n. 20, 363, 690, 697
capacit
de parole, 502
caractre
abstrait, 519
Carter, Michael G., 4, 4 n. 2, 6 n. 3, 23,
25, 3031, 35 n. 8, 39 n. 9, 44, 65, 140
n. 30, 147, 155, 162 n. 19, 187, 209
n. 1, 228, 228 n. 35, 234, 240, 243,
313, 313 n. 3, 363, 503 n. 6, 522
case(s), 368 n. 4, 370, 372373, 382384
389, 394, 399
accusative, 248251, 253, 280,
382383, 476, 484, 484 n. 13, 493
definite, 375
indefinite 252
ending(s), 8, 9, 11, 18, 30 n. 2, 141,
216, 251, 253
feature, 373
genitive, 248251, 253, 375, 383, 484,
494
marking, 374, 384
nominative, 383, 248251, 253

role(s), 369 n. 7, 391, 393394, 394
n. 83, 395, 395 n. 85
system, 217
Caspari, 363
Caspari-Uricoech, 313, 314 n. 3, 327 n. 18
Castilian
alphabeth, 213 n. 7
letters, 220
Castilians, 227
Castro, 532
casuistique, 503, 503 n. 6
casus accusativus, 228 n. 35
cataphora, 154 n. 8
catgorie
intermdiaire, 321
lexicales, 359
moyenne, 321
noun, 391, 399
particle, 399
smantiques, 344, 359
syntactic, 370
verb, 391, 399
categories
mixed, 475, 478, 484, 498
nominal, 477, 485
verbal, 477, 485
categorization
verb, 399
Cauallero, Lucas, 224, 228
Caubet, D., 292 nn. 12, 310
causal
clause, 88, 92, 108109
non-coreferential, 92
subordinate, 89, 91
conjunction, 76, 91
connection, 76
function, 87 n. 24
non-, 8788
relation, 8788, 104, 109
structure, 87, 55
causative, 441
function, 68
xalla clauses, 694 n. 26
causativit, 313
causativity, 450
cause(s)
consequence relation, 263, 267268,
278, 284
de transformation, 317318, 318 n. 7
Central Arabia, 563
Chad, 640641, 648
Chadian, 605
index
chanes de raisons, 501
Chambers, Jack, 648, 652
changement dtat, 326331, 333335,
339340
changes
educational, 404
political, 404
social, 404
technological, 404
chercheur, 504
chesmantes, 230
China, 223
Chisarik, Erika, 697
Chomsky, 503
Chrtiens, 123
Christian
Baghdadi speakers, 669
doxology, 106, 109
faith, 221, 221 n. 26
liturgy, 105106
Maronites, 224
scholars, 34
speakers, 657
Christian(s), 79, 93, 106, 109, 195, 209
New, 212, 213 n. 7
Old, 212, 213 n. 7
circumstantial(s), 391, 393, 395, 399
local, 395
optional, 395
clan, 545, 565, 565 n. 3, 570 n. 20, 571,
586 n. 71, 589
classement, 503504
grammatical, 504
classer, 503504
classes
nominales, 503 n. 5
classical, 213, 428, 640 n. 1
Arabic, 549 nn. 9, 13, 555 n. 31,
567568, 595596, 596 nn. 56,
597604, 639641
elements, 224 n. 31
language, 98, 639
non-, 79
registers, 223
post-, 70
clause, 69, 100, 152, 164, 177 n. 35, 180
complement, 369, 376, 376 nn. 30,
33, 377379, 379 n. 45, 380, 386,
388389, 391
Clauson, Gerard, 443 nn. 79, 453
cleft sentence, 76, 687 n. 17, 695
clusters
prepositional, 386
711
coda, 487
position, 488 n. 19, 489
code
mixing, 618, 659, 665, 655 n. 2, 670
mixture, 653
switching, 291 n. 1, 295, 297 n. 7,
298 n. 8, 300301, 307, 406408,
417, 433, 665, 655 n. 2, 670,
672
codes
linguistic, 403, 410
Codoer, Carmen, 243
Coelho, Paulo, 535
cognate(s), 597, 597, 602, 653, 655656,
659 n. 4, 660, 666, 667
lexemes, 653
cognitum, 235 n. 51
Cohen, David, 296, 310, 314, 321, 335
n. 20, 344 n. 26, 363, 370 n. 14, 400,
523, 697, 697 n. 31
Colin, Georges S., 690, 698
collectivity, 397
collocation, 391, 393
embedded, 309
foreign, 309
language
colloquial Arabic, 595596, 596 n. 6,
597, 604, 621, 653655, 655 n. 2,
656670
Chadian, 605
dialect(s), 656658, 667
dictionairies, 658
lexemes, 655, 668
lexical items, 656
pairs, 656
Proto-, 597
speakers, 659, 664, 665
Sudanese, 605
text, 659
vocabulary, 656
words, 666
colloquial, 211, 213214, 222, 224 n. 31,
403, 403 n. 1, 422, 529, 531, 533, 536
n. 13, 538, 540
elements, 224 n. 31
Egyptian, 529, 540
features, 614
modern, 529
speech, 227
urban speech, 224 n. 31
written, 540
colloquialisms, 224 n. 31
Colossoi of Memnon, 678 n. 6
712
index
combinatoire
binaire, 521522
comitative, 393395
commands, 397
comment, 368 n. 4, 376377, 393, 397
n. 89
Q+ allad + topic, 96
Qcompound, 5, 14, 15 nn. 1112,
1720, 22
Qconditional, 17, 2021
Q-topic, 76
QQstructure, 96
Qxabar, 216
communication, 520, 522
quotidienne, 122
savante, 122
community
Dutch-Moroccan, 293
Companion(s) of, 193, 206
the City (ash b al-qarya), 195
the Prophet, 206
competence
language, 294, 308309
complement, 376 nn. 30, 32, 378, 379,
381, 381 n. 50, 382, 386389
function, 385, 388
phrase
interjectional, 387
prepositional, 375, 385, 390391,
396
structure, 391
complementizer(s), 406407, 419, 676,
692693
complments, 518
collocational, 309
completive(s), 380, 391, 393, 395, 399
composites
binary, 287
ternary, 287
compound(s), 456
Dutch, 456
verbal, 293
words, 456
Comrie, Bernard, 324 nn. 1314, 326
n. 17, 363
concept
religious, 471
technical, 470
concord, 212, 368, 368 n. 4, 374, 376 n.
31, 377, 383384, 389, 397
concrete, 370, 383, 396397
conditional clause, 177
conditionnements
phontiques, 350
confirmation, 398
conjugaison, 316 n. 5, 317, 343, 357,
505, 517
arabe, 332
conjugated, 644
conjugation, 486, 640
verbal, 436
of verbs, 613
conjunction(s), 6769, 7173, 75, 7779,
86, 87 n. 24, 8889, 95104, 107, 109,
213, 229, 374375, 388, 533, 541
causal, 88, 103, 108109
of comparison, 77
coordinating, 388389
final, 80 n. 6
h ukm arab, 190
lisn arab, 190
subordinating, 388389
conjunctional, 6769
allad, 69
function, 71
connaissance, 501
conocimiento, 212, 212 n. 7, 214, 217,
217 n. 14
conqute islamique, 114, 116, 120, 128
consonant(s), 236 n. 53, 247, 258, 286,
368, 368 n. 6, 436437, 441444,
448451, 567568, 571, 611
augmented, 436, 441, 445
basic, 436, 444445, 449
diffusion, 256
doubled, 443
emphatic, 614
final, 437, 451
geminate, 437, 451
initial, 437, 451, 572
medial, 437, 451
middle, 446
non-basic, 444, 450
non-regular, 437
pharyngeal, 614
root, 371
sound, 437, 451
uvular, 256 n. 10
velar, 256 n. 10
vocalized, 447
voiceless, 579 n. 48, 592 n. 91
weak, 437
consonantal, 276
bi-, 596, 598, 598 n. 10
mono-, 596, 596 n. 6, 598
phonemes, 567
roots, 644
tri-, 598
index
consonantalism
bi-, 598
mono-, 598
tri-, 598
consonante, 215 n. 11
consonne(s), 319, 506508, 508
n. 21, 509, 509 n. 22, 510511, 511
n. 26, 512514, 515 n. 31
darrire, 318
double, 509 n. 21
glottales, 319
implosives, 512
longue, 508 n. 21
proto-, 510 n. 24, 511 n. 25, 515 n. 31
vocalique, 515 n. 31
constant
phonetic, 271
phono-semantic, 272
semantic, 271
constituency
immediate, 369
constituent(s), 367368, 376, 383384, 390
phrasal, 398399
order
higher, 291
lower, 291, 292 n. 1
constitution
de la langue
gnrale, 502
construccin
de concordancia, 212
de rgimen, 212
construct state, 574
construction(s)
auxiliary, 293
impersonelle, 343
periphrastic, 293, 295, 296 n. 4,
297, 297 n. 6, 299304, 304 n. 11,
305309
possessive, 493
contextual, 5, 13
elements, 5
continuant(s), 256, 256 n. 8, 257,
260261, 266270, 275276, 278,
280282, 285286
non-, 256 n. 8
continuity, 247
Contossopoulos, N.G., 298, 310
contour principle
obligatory, 259 n. 19
contractual language, 25
contrainte(s)
formelles, 361
morphologiques, 361
713
phontique, 510
smantiques, 361
contrle, 321, 343, 348349, 351 n. 32
imparfait, 321
convention
structure, 521
conventionalization, 294
conversational implicature, 36
coordination, 388, 517518, 520
asyndetic, 388 n. 67
coordinator(s), 388
adversative, 388
coordonnants, 519
Coran, 125127, 130
du Caire, 127
langue du, 120, 125
du Maghreb, 127
coranique, 125, 128 n. 18
coreferential, 24, 2627, 33, 43
non-, 109, 159, 165, 180
subject, 92
coreferentiality, 92
Cornips, L., 297 n. 6, 310
coronal, 255257, 259261, 265266,
266 n. 28, 267, 270, 272, 275, 279
n. 39, 280 n. 41, 281282, 284286
correspondence
input, 480, 482484, 493, 495496,
498
linear, 479480, 483484, 493496,
498
quantitative, 479 n. 4
Corriente, Frederico, 213, 215, 215 n. 11,
217 nn. 14, 15, 243
countable, 396
couplets, 548 n. 5
Cowan, William, 210, 213 n. 7, 214, 215,
215 n. 11, 217218, 218, n. 17, 243,
365
Cowley, 320 n. 9
crement
initial, 262, 262 n. 25
inset, 262 n. 26, 265
crementation
initial, 264
creole, 692, 698
Arabic, 639
Juba-, 611
languages, 692
crole(s), 128129, 607, 608, 620621
tudes, 128
crolisation, 129
crolistes, 129
creolistic, 608
714
index
creolization, 607, 621, 652, 699

crosslinguistically, 698
Crystal, David, 27, 44
culture
Arab, 424
Islamic, 468
curvature, 257, 264, 270, 279280
Cuvalay, Martine, 378 n. 41, 400
Cuvalay-Haak, Martine, 400
Czapkiewicz, Andrzej, 192 n. 2, 207
D ahab, 569 n. 16
Dahbi, 674
colloquial Arabic, 659, 662
discourse, 665
dim, 56
Dakhla-Oasis, 690
dlika, 15 n. 11
Damas, 113
Damascene, 650
Franciscan tradition, 211
Damascus, 210, 211 n. 2, 221, 221 n. 26,
222, 239, 532
damir, 210, 217218, 239
damm, 227
damma, 14, 218, 218 n. 16, 230, 239
damma, 224, 231
dammatn, 218
Dankoff, R. and K., 437, 453
Dannenfeldt, 210 n. 2, 219
Dante, Alighieri, 121 n. 8
darf, 216
Darfouf, 255 n. 5, 287
ad-Drim, 89, 110
Darwin, Charles, 274, 274 n. 35
Dat, 255 n. 4, 257 n. 14, 272, 288
Dat na, 690, 697 n. 31
datiuo, 214, 227
dative, 214
dativus, 221
Dawsir, 563
Day of Judgment, 190
Dbr, 566 n. 9
de Jong, Rudolf, 565, 567 nn. 10, 1213,
569 n. 17, 570 nn. 21, 23, 572 nn. 25,
2728, 573 nn. 3031, 574 n. 32, 575
nn. 34, 37, 577 n. 40, 578, 583 n. 63,
586 n. 70, 591 n. 88, 652, 675 n. *
de Sousa, 211 n. 2
dclaration, 517 n. 44, 518520, 520 n. 47
du lien, 517
declarative sentence, 169
declension, 373, 436
declensional, 227228
endings, 227
non-, 227
decreolization, 608
dcrochage, 361
deductive application, 25
dfaut, 345346
defective, 252253
deficiency (illa), 53
dfinir, 503504, 506
definite, 250251, 253, 172, 373
marifa, 47
article, 17 n. 13, 611613
definiteness, 175, 368 n. 4, 370, 382, 384,
389
dfinition(s), 455456 504, 504 nn. 89,
505, 505 n. 12, 506 n. 16
deixis, 27
Dell, F., 256 n. 7, 288
demonstratives, 373, 406407, 419
denativization, 308
Deng, 612, 622, 624625
dnombrements; entiers, 501
denominatives, 373
dental, 262
dependencies, 368369, 376, 381382
Derenbourg, Hartwig, 43, 143 n. 43, 187
derivation, 370, 372, 374, 395, 486
feature, 372373
droulement
du temps, 505, 512
Descartes, Ren, 501, 503, 522
Descls, Jean-Pierre, 313 n. 1, 321, 324,
324 n. 14, 325, 325 n. 15, 326, 326 n.
17, 330, 332333, 336 n. 21, 341, 347,
362363
description(s), 456458, 465467,
469472
paraphrased, 455456
formal, 367369, 369 n. 10, 370, 372
nn. 1819, 374, 376 n. 35, 377
nn. 3637, 39, 387, 391
linguistic, 367, 376, 383, 385
phonemic, 368
translations, 467, 469
descriptive
analysis, 48
compounds, 696
dterminants, 513, 516
determination, 373, 374
Dvnyi, Kinga, 23, 46 n. 1, 44, 45, 56
n. 12, 65
dhamma, 233
Diab, 257 n. 11, 288
diachronic, 67, 74, 656
index
dialect(s), 10, 97101, 216, 223, 403,
403 n. 1, 404, 404, n. 2, 405408, 410,
427, 543, 545, 567, 573, 570 n. 20,
571, 573, 577578, 586 n. 71, 591 nn.
8889, 595, 595 n. 3, 596597, 604,
640 n. 1, 641 n. 4, 650, 650 n. 12, 642,
653, 657, 665, 669674, 687
of the Agylah, 567
of the Ahaywt, 567
Arabic, 640641, 677, 690
of the Arabs, 543
of al-Ar, 570, 570 n. 21
bedouin Sinai, 571, 575, 578, 586
n. 70
Bri, 690 n. 20
Cairene, 540
contemporary, 640
of Damascus, 222
dependent, 658
differences, 657 n. 3
Eastern Arabian, 555 n. 30
Egyptian Arabic, 675, 697 n. 31, 699
English, 596 n. 6
features, 658
of the Jarjrah, 567
of the H amdah, 565
H amdiy, 567
H ijz, 16
Jewish, of Tripolis in Libya, 100, 104
of the Lgt, 565
Lgiy, 567
local, 403
modern, 70, 75, 94, 100, 102, 527, 639
of the Mznah, 567, 567 n. 12, 568
Mzniy, 575
of Negev, 570 n. 22, 565, 567, 574
n. 32, 575, 575 n. 37, 8
Neo-Aramaic, 595 n. 3
North African, 659
rural, 676
of the Samnah, 572 n. 25, 574 n. 32, 567
of Sinai, 568, 573, 574 n. 32, 575, 575
n. 34
bedouin, 568
of southern Sinai, 567, 568, 568 n. 14,
572 n. 25, 573, 573 n. 29, 575 n. 35
speakers, 657
Tamm, 16
of the Tarbn, 567, 569 n. 17, 575
of the Tayhah, 567
of Tunis, 100, 104
Yemeni, 640 n. 1
dialectal, 80, 81 n. 10, 403 n. 1, 405, 405,
n. 3, 414, 422, 608
715
inventory, 667
Marocain, 698
dialecte(s), 120, 563
ancien, 126
arabe, 100, 128 n. 19, 129
moderne, 126
de larabie mridionale, 698
dialectic(s), 22
Aristotelian, 34
dialectical, 34, 42
dialectique, 114
dialectologie historique, 129
dialectologists, 211
dialectology, 72, 605
diaspora communities, 302, 306
diathse, 316 n. 5, 335, 336
externe, 336337, 352, 353 n. 33,
354360, 362
objective, 513, 519
subjective, 513, 519
diatopic, 74
Daz, Lourido, 210 n. 2, 211 n. 2,
243
dichotomy, 639
Dichy, Joseph, 313, 315, 317, 318 n. 7,
319, 323, 327, 341 n. 24, 343 n. 25,
344, 344 n. 26, 347, 353 n. 33,
362363, 394 n. 81, 396397, 400
dictionaries, 455459, 462, 465473
Arabic, 455
Arabic-Dutch, 455, 472
Arabic-English, 578, 605
bilingual, 455456, 471
classical and MSA, 532 n. 7
Dutch-Arabic, 472, 455
Egyptian Arabic, 604, 698
English-Arabic, 457
etymological, 443, 443 n. 7, 446, 449,
455456, 472
French-Arabic, 457
German-Arabic, 457
Hamito-Semitic, 605
Hebrew, 605
lexicographical, 437
Modern Arabic, 605
monolingual, 455
Moroccan Arabic, 605
multilingual, 455
dictionnaire(s), 315, 317, 326 n. 16, 340, 347
dArabe Dialectal Marocain, 698
informatis de larabe, 315
didactic, 220
Diem, Werner, 8182, 110, 130131,
654, 664, 665 n. 8, 666, 671
716
index
Dietrich, Albert, 93, 110

difl (oleander), 21
diglossia, 189, 403404, 415, 421, 527,
653, 654, 655 n. 2, 656, 664, 671672
diglossic, 212
differences, 653
makeup of Arabic, 670
situation, 403, 528
diglossie, 121
diglossique, 125, 129
DIINAR, 391, 391 n. 72
diminutive(s), 373, 444, 450
dimuqrtiyye, 657
ad-Dnawar, 202
aJ-Xindn, 563
Dinka, 610, 612613, 616, 622, 624
Nilotic, 610
diphthong(s), 406407, 419, 491 n. 23,
568569, 576, 650 n. 12, 657
diptote, 124, 127
drah, 544, 566 nn. 78
direct speech, 534, 537, 539, 676,
678679, 683
disambiguation, 367 n. 2, 369, 390, 398
discourse, 368, 656, 672
functions, 300, 403, 417
marker(s), 619, 680681, 683, 685,
687, 691
political, 405, n. 3
record, 681, 681 n. 9
manipulation, 681 n. 9
routine, 680682
strategies, 410
structure, 27
dispensable part, 136138, 142, 146
dispersion, 350, 356357
semantique, 313, 315, 321, 356
dissent, 397
Ditters, Everhard, 367, 368 n. 4, 369 nn.
8, 10, 374 n. 26, 376, 376 nn. 31, 35,
379 n. 43, 382 nn. 5253, 385 n. 58,
386 n. 62, 400401
divinity, 404
diviser, 501
Dwn al-Adab, 436438, 445, 446 n. 11, 447
Dwn al-HudalByBn, 110
Dwn Lut at-Turk, 437, 441447, 449,
450452
dwns, 545
do-construction(s), 301, 307
Dombay, Franciscus de, 223, 241
Dominicus Germanus de Silesia, 124
n. 13, 221, 223
domma rafo, 225
Doniach, 473
dorsal, 256 n. 10, 257, 264266, 270,
279, 279 n. 39, 280
non-, 256, 275
Doss, MadBha, 529
doubt, 398
Doxologie, 106107
doxology, 106109
Driver, G.R., 597 n. 8
dropping (ilq), 53
Drost, 218 n. 18
Druze speakers, 657
\ l-Qarnayn, 190
dual, 82 n. 15, 91, 153, 227, 383, 397
n. 86
dure, 324325, 327328, 338, 341,
351352
Dutch, 293, 295297, 297 n. 6, 299
n. 9, 309, 455458, 465482, 565 n. 2
non-standard, 297
standard, 297
dzhamma, 226
Edman, Johan, 222 n. 27
Educated Spoken Arabic, 665
education, 404
Egyptian, 403, n. 1, 405406, 407,
n. 5, 407423, 425, 429, 433
higher, 404
public, 404
Egypt, 81, 83, 83 n. 18 197, 224 n. 31,
528530, 536, 544, 545 n. 2, 563, 566
n. 6, 578, 655656, 662, 666, 671673;
686, 690
Upper, 690 n. 20
lEgypte, 130
Egyptian(s), 105, 190, 537, 544, 545
n. 4, 578, 590 n. 84, 598 n. 12,
600604, 675, 679, 687
Ambassador, 529
Arabic, 675, 679, 692693, 698699
Arabic dialects, 675
Babylonian Old, 692 n. 23
history, 528
egyptianisation, 539
Eid, Mushira, 402403, 403 n. 1, 405, n.
3, 407, 410, 434, 498, 671
Eisele, John, 371 n. 17, 384 n. 55, 401
El Aissati, 309310
El Assal, 403, n. 1
El Escorial, 210 n. 2, 211 n. 2
elatives, 373
lment(s), 504505
embedded, 292
index
Elgibali, Alaa, 402, 671
El-Hassan, 665
elision, 23, 572
Elzeiny, Nagwa, 375 n. 28, 401
emphatic(s), 284, 569, 642 n. 6, 653,
656658, 571
consonants, 658
lateralized, 283
sounds, 651 n. 13
enantiosemy, 272, 272 n. 33, 286
Encarnacin, Juan de la, 210 n. 2, 211
n. 2
ending(s)
feminine, 492, 250
inflectional, 216218, 225, 231233,
239
English, 9294, 291, 296299, 299 n.
9, 300304, 304 n. 11, 305309, 391,
395, 399, 456, 465, 483 n. 10, 493, 548
n. 5, 569, 578, 583 n. 63, 595, 596
n. 6, 605
Old, 596 n. 6, 609611, 618, 626,
661662, 668, 673, 681, 681 n. 9, 682,
686, 692693
ensemble, 503506
entertainment, 403, 425, 432
enthusiasm, 397
entit(s), 503, 505, 508, 512 n. 27,
520
abstraites, 512 n. 27
communes, 516
multidimensionnelles, 512
universelle(s), 516, 516 n. 41
entity(ies)
animate, 396397
non-, 396
human, 396397
non-, 396397
perceptible, 396
pseudo-animated, 397
entry(ies), 435437, 449, 456, 465466
derived
non-, 375
noun-, 375
double, 446
Dutch, 455
lexical, 369, 381, 383, 395
verbal, 369 n. 7, 377, 379380, 391,
395, 397 n. 86, 391392, 394396,
399
epenthetic, 489
poque prislamique, 120
Equatoria, 609, 611, 621
Ermers, Robert, 435, 437, 441, 449, 453
717
Erpenius, Thomas, 211, 231234, 236,

236 n. 52, 238, 240241
Escavy, R., 243
espace, 512, 512 n. 27, 517 n. 41
lespagnol, 115
Ess, Josef, van, 192 n. 3, 194 n. 6
tat(s), 317, 321, 323324, 324 n. 13,
325326, 326 n. 17, 327333, 337,
339340, 342, 347359
accompli, 334
prsent, 332, 325 n. 15, 341
rsultatifs, 330331, 347
non-caractristiques, 329
Ethio-Semitic, 697 n. 31
etymological, 13, 14, 17, 2021, 91
etymology, 5, 8, 13, 17 n. 13, 2022, 203,
262, 436, 440, 444, 696 n. 30
etymon(s), 256258, 258 n. 15, 259262,
262 n. 26, 264268, 270275, 277, 277
n. 38, 278281, 283285, 287
homonymic, 261
lEuphrate, 128
Euphrates, 196
Europe, 209210, 219220, 121 n. 8,
223, 232
Northern, 231, 240
European(s), 221 n. 26, 608
language, 213
scholars, 222
vnement(s), 321, 324, 324 n. 13,
325326, 326 n. 17, 327, 330334,
339340, 342343, 347, 349, 351352,
359, 505
event, 476
nominal, 475
vora, 219 n. 20
Evripidou, D., 300, 303, 310
Evstatiev, Simeon, 44
exceptive, 19, 21
sentences, 171
excitative (tah /:/), 16
exclamation(s), 375, 387, 397398
exegesis, 49 n. 5
Quranic, 438
exegetes (ash $b at-t;wl), 54, 193
exegetical
methods, 63
tradition, 63
grammatical, 64
works, 45
Exodus, 105
exprience(s), 505, 508, 511510, 520
non-physical, 396
experiencer, 394, 396 n. 85
718
index
explicit, 368 n. 4, 378380, 389, 392

n. 77, 397 n. 86, 394
expression(s), 455456, 466, 469
adverbial, 484
frozen, 375, 375 n. 27, 387
extending meanings, 661
extensions, 518
verbal, 379, 380
extra-textual linguistic source, 46, 61
al-Ezbekiyah, 538
fa- after amm, 60
f, 7 n. 5, 21, 38
faala, 313, 314 n. 3, 314, 316323, 337,
350351, 353358, 360361, 435, 440
factitif, 115
factitivit, 313
fil, 151152, 154, 154 nn. 7, 9,
155158, 158 n. 13, 161, 164, 167,
175184, 323, 329
faila, 314, 314 n. 3, 314, 316318,
320323, 327, 327 n. 18, 337338,
344347, 349350, 351 n. 32, 356,
358, 360361
fal pattern, 659
falsafa, 114
familles morphologiques, 315, 324325
n. 15
far structure, 151
al-Frb, 4243, 120, 131, 436438, 445,
447, 453, 504 n. 8, 513 n. 29, 522
Farj, Muhammad, 247, 249, 253254
Farazdaq, 7, 8 n. 6
al-Fris, 11, 143 n. 43, 144 nn. 4950,
147, 164 n. 23, 165, 165 nn. 2425,
166168, 171, 177, 180, 185186
al-Farr, 4546, 46 n. 3, 4748, 49 n. 6,
50, 50 n. 7, 5156, 56 n. 12, 57, 57
n. 13, 5862, 62 n. 15, 6364, 64
n. 17, 125, 129, 175 n. 33
fasd, 114, 129, 194, 207
al-lua, 118, 124, 128
fash a, 120
Fasold, Ralph, 655 n. 2, 670671
Fassi Fehri, Abdelkader, 386 n. 63, 401,
475, 475 n. 1, 484485, 498
fatah ahu, 659
fataho, 659
fath , 227
fath a, 1415, 218, 218 n. 16, 224225,
226, 230231, 239
fath atn, 218
Fathi RaJwn, 699
Fathiyya l-Assl, 699
Ftihah, 105
fatina, 668
faula, 313314, 314 n. 3, 316317, 320,
322, 327329, 337338, 338 n. 23,
340341, 343344, 350, 353, 353
n. 34, 357358, 360361, 435436
al-Fayrz bd, 353, 353 n. 35
feature(s), 368370, 370 n. 12, 371, 371
nn. 1517, 372 n. 18, 373, 376, 381,
383, 389390, 392 n. 75, 394, 396398
clusters, 369, 383, 389
ideal, 386
inherent, 367 n. 1
inherited, 367 n. 1
linguistic
universal, 368
manner, 386
morphological, 371, 373
names, 367 n. 1, 370, 372, 372 n. 18,
373 nn. 20, 23, 385, 391
noun, 396
place, 386
prepositional, 396
semantic, 368, 368 n. 5, 369, 369 n. 7,
383, 386, 390, 397399
semantico-syntactic, 369 n. 9
time, 386
values, 367 n. 1, 369 n. 9, 370, 370
n. 11, 371372, 372 n. 18, 373 nn.
20, 23, 383, 385386, 390, 391, 392
n. 77, 393, 398
local, 390
semantic, 399
variables, 373, 389, 397398
featuring, 367, 367 n. 3, 380, 393
morphological, 368
nominal, 369
phoneme, 371 n. 15
semantic, 368
syntactic, 368
verbal, 369
Feghali, Michel, 99, 111
feminine
ending, 231
marker, 182
Ferguson, Charles A., 121 n. 9, 131, 404,
654, 656, 664, 667, 671
Fernandez, Mauro, 654, 672
Ferrando, Ignacio, 126, 131
Fez, 232
figement lexical, 362
fil, 151, 154, 154 n. 9, 156157, 163
n. 21
+fil, 177
index
Filip IV of Spain, 223 n. 28
filiyya, 149, 177, 158 n. 13, 165 n. 24
fillers, 377379, 382, 385389
Fillmore, Charles, 369, 401
finite
forms, 488
verb, 163
fiqh, 31
Firanescu, Daniela, 387 n. 65, 401
firaq islmiyya dlla, 530
Fischer and Jastrow, 656, 664
Fischer, A. & A.K. Irvine, 208
Fischer, Wolfdietrich, 672, 698699
fiten, 668
Fleisch, Henri, 314, 316, 316 n. 5, 320 n.
9, 321, 321 n. 10, 328329, 335, 335
n. 20, 344 n. 26, 364, 371 n. 15, 373 n.
22, 378 n. 40, 401, 513 n. 29, 522
Fleischer, Heinrich Leberecht, 118119,
121, 121 n. 7, 131, 132
flexion
casuelle, 125126
dsinentielle, 117118, 120123, 130
triptote, 130
language, 662, 670
lexemes, 655656, 661
loanwords, 661662, 668
terminology, 661
words, 662
form, 380
pausal, 248
prejunctural, 250252
prepausal, 250252
formal, 403405, 423
literally Arabic, 658
formalism, 369 n. 10, 370, 396
formality, 405, 410
format, 369
formation
external, 373
internal, 373
forme(s), 315 n. 4, 332, 357358,
360362
augmentes, 314
construites, 316
drive(s), 313, 313 n. 3, 314
verbale, 332, 362
forms
bound, 374, 384
finite, 369, 488
free, 374, 384
hispanicised, 229-230
non-finite, 488
paradigmal, 449
719
fraccion, 225
franais, 331, 339, 343355, 355 n. 36,
511 n. 25
Franciscain, 123
Franciscan(s), 210, 211 n. 2, 212, 221,
221 n. 26, 222, 231232, 232 n. 44,
237, 239240
authors, 211
Spanish, 211
College, 222
tradition, 222
Frank, Armin Paul, 243
Frayha, Ans, 666 n. 9, 672
French, 119, 292, 292 n. 2, 295296, 297
n. 7, 298, 299 n. 9, 300, 302, 308, 456,
465, 661
fronted, 169 n. 26
fricative(s), 260, 282, 407, 597, 657
final glottal, 639
interdental, 407, 407 n. 5, 409410,
413, 418420
voiced, 419
voiceless, 268, 419
Fritz Paul, & Horst Turk, 243
Fck, Johann, 119, 124 n. 13, 128, 132,
210 n. 2, 219220, 233, 243, 654, 672
Fuentes, Pedro, 223 n. 30
function, 380
word(s), 675, 676 n. 4, 689, 690
Funktionsschwche, 75, 94, 96
Funktionsschwchung, 69, 86
fush $, 403404, 404, n. 2, 405, 405,
nn. 23, 406407, 407 n. 5, 408410,
410, n. 8, 411416, 416, n. 12,
417423, 427433, 529
al-Fust ^t , 7984, 101, 130
Gabriel, John, 615, 618620, 626, 635
gahawa(h)
complex, 641 n. 5
syndrome, 572, 588 n. 81
Galloway, George, 561
gap(s)
lexical, 455456, 458, 465472
Garawiyya, 609612, 614, 619620,
622
Gardner-Chloros, P., 303, 310
azla, 690
Gedichte, 563
Geez, 597, 599603
Geiger, Ludwig, 231 n. 41, 243
el-Gemei, Dalal M., 391 n. 71, 401
geminates, 611
gemination, 489 n. 21, 490, 491
720
index
Gen(esis), 191, 194, 196, 198, 202

gender, 154 n. 6, 249250, 367, 368
n. 4, 370374, 378, 380, 382383, 389,
395, 615
affixes, 611
genealogical traditions, 192
genealogists, 203, 205
gnitif, 130
genitiuo, 214, 227
genitive, 7, 10, 13, 106, 135 n. 2, 144,
214, 216, 227228, 230, 230 n. 39
genitivus, 221
Genizah, 7980
genre, 513
Georgakopoulu, Alexandra, 677 n. 5,
679 n. 8, 698
German, 81 n. 8, 9294, 98, 104, 456,
465, 679 n. 7, 698
German-Arabic, 458, 462, 466468
gerund(s), 299, 301, 305, 476, 476 n. 2,
477479, 482, 492
English, 475476, 482, 493
nominal, 477
Gesenius, 320 n. 9
Geva-Kleinberger, Aharon, 658, 672
Ghassn Surr a-baylt (Ab Surr),
546548, 551556, 559
Ghawnmah, M., 548, 563
Ghazeli, et Hassour, 315
ghost words, 472
Giargianius, 235
idir, 660
Giesbers, H., 297 n. 6, 310
Gil, Moshe, 80, 80 nn. 4, 6, 81, 81 n. 8,
82, 84, 110
Gilliot, Claude & Larcher, Pierre, 195
n. 11, 208
glide(s), 368, 437, 441, 446450
inserted, 444
glissement(s)
de sens, 338, 342
smantique(s), 321, 326, 334335,
338341, 354358, 360362
global, 405, 410, 417, 433
glottal, 657
catch, 571
stop, 595 n. 3, 596, 596597, 597 n. 8,
598, 598 n. 12, 639640 nn. 12
final, 642
goal, 380, 394395
God, 26, 3537, 3942, 189, 190191,
193194, 196200, 202204, 207
Godefroy-Demombynes, Maurice, 313,
314 n. 3, 363
Goitein, Dov, 80, 80 nn. 4, 6, 8182, 110111

Golato, Andrea, 684, 698
Goldenberg, Gideon, 7679, 111,
153154, 163 n. 21, 173 n. 31, 174,
187, 595 n. 1, 596 n. 6, 597 n. 8, 598
n. 12, 600
Goldziher, Ignaz, 119, 132, 208
Golius, Jacob, 211, 238
Gonzles Palencia, A., 43
Gonzlez, Bernardino, 210 n. 2, 211 n. 2,
221222, 226
Gonzalez, Francisco, 212, 231, 238
Goral, Mira, 655 n. 2, 673
Gottheil & Worrell, 8184, 111
Goutsos, Dionysis, 300301, 310, 698
grammaticalized, 78
grammaire(s), 113114, 118, 121, 123
n. 12, 502 n. 4, 507
arabe(s), 116117, 126, 130, 314, 348
n. 30
arabisantes, 313, 319, 327
compar, 118
gnrale, 517 n. 44
lhistoire de la, 114
grammairien, 113, 118, 122, 126
arabes, 124125, 129, 505506
non-arabes, 506
grammar(s), 3, 4 n. 1, 23, 2526, 28, 31,
33, 37, 4243, 45, 62, 77, 79, 171, 209,
209 n. 1, 210, 211 n. 2, 212, 213, 215,
217 n. 15, 218 n. 18, 219, 219 n. 19,
220, 223, 224, 226227, 232, 234236,
239240, 370, 527528, 617
of de Alcal, 219, 219 n. 19, 226, 232
Arabic, 248, 252253, 435437, 439,
442, 448449
Classical, 223, 249
Granadan, 212
of Caballero, 209 n. 1, 210, 210 n. 2,
221, 222, 224 n. 31, 238
of Caes, 211, 221, 222
case, 369, 381, 393, 399
Castilian, of Antonio de Nebrija, 215
dependency, 369 n. 8
early, 25, 49 n. 5
formal, 367, 388
of Greek
classical, 223
modern, 223 n. 30
generative, 4
of Hebrew, 209
Indian, 209
of Japanese, 210
Latin, 216, 239
index
of Martelottus, 238
medieval, 213
missionary, 213
North-European, 211
Renaissance, 213
of Sanskrit, 209
Semitic, 605
Spanish, 210, 211 n. 2, 212, 223 n. 30,
237
of Tagalog, 219 n. 19
Tamil works of, 210
traditional, 1112
transformational, 396
of Turkish, 224 n. 30
grammarian(s), 3, 68, 10 n. 8, 1113,
17, 17 n. 13, 20, 22, 26, 33, 3940,
4546, 60, 70, 79, 135, 135 n. 2, 138
n. 21, 142, 143 n. 43, 146, 149, 150,
150 n. 2, 151, 152, 152 n. 3, 153 n. 5,
154, 154 n. 7, 155157, 158, 158 n. 13,
160165, 167170, 172, 173, 173 n. 31,
174175, 177179, 181, 184186, 220,
223, 225, 235, 240, 595 n. 3
Basran, 175
classical, 5
Franciscan, 212, 238
medieval, 150, 154, 160, 162, 176177
Arab, 184
missionary, 238
traditional, 11, 13 n. 9
grammatical, 34, 4 n. 1, 6,8, 10, 12,
15, 1718, 19 nn. 15, 17, 21, 23, 25,
3334, 4546, 46 n. 3, 47, 47 n. 4, 48,
5557, 6164, 64 n. 17, 7879
analysis, 4, 6, 12, 2223, 42, 4648, 49
n. 5, 62, 64, 142, 146
categories, 34, 38
coherence, 655 n. 1
description, 209
features, 33
form, 25, 30
framework, 210
literature, 79
medieval, 186
Arab, 158
medieval Arabic, 181
markers, 679 n. 7
monographs, 239
problems, 25
rule, 49, 5762
of jaz, 59
structure, 48, 152 n. 4, 655
of sentences, 160
subject, 694
721
systems, 4
term(s)
Arabic, 210
exo-, 218
technical, 217
terminology, 233
orientalising Western, 240
endo-, 240
theory(ies), 8, 228, 240
Arab, 3
medieval Arab, 168, 178
non-Western, 239
Western, 239
thinking
medieval Arab, 152, 168169
thought
medieval Arab, 180
tradition, 42, 64, 211 n. 2, 210, 227
Hebrew, 209
Italian, 211 n. 2
Indian, 209
of the Franciscans, 211 n. 2
medieval Arab, 149
works
canonical, 220
writings, 160 n. 15
grammaticalisation, 294, 621, 697699
grammaticalization, 675, 676 n. 4, 678,
681, 681 n. 9, 682, 684, 685 n. 14, 687,
687 n. 16, 688 n. 19, 697699
endo-, 236
exo-, 236
grammaticalized, 581 n. 54, 616, 675,
677, 683684, 689, 690, 692
grapheme(s), 540, 598
Arabic, 371
grec(que), 124, 130
classiques, 124
grco-arabe, 119
Greco-Latin, 219, 226, 228
Greek, 34, 105106, 108, 189, 193, 213,
221, 661
American, 301302, 304
Classical, 223
Cypriot, 300301, 303
doxologies, 107108
ethics, 42
expansions, 108
grammar, 228 n. 35
Modern, 223 n. 30
vernacular, 223, 233 n. 46, 291, 298,
298 n. 8, 299301, 303304,
306307, 309
verbal system, 108
722
index
greetings, 387
formulaic, 375
Gregory James & Emilio Ridruejo, 244
Grice, H. Paul, 2730, 42, 44
Gricean, 28, 29 n. 1
maxims, 33, 42
griechisch, 106
Grimshaw, Jane, 475, 475 n. 1, 476, 476
n. 2, 498
Gruntfest, Y., 123, 132
Guadagnoli, Philip, 211 n. 2, 232235,
235 n. 51, 238241
Guadaoli, Felipe, 212, 231
Guentchva, Zlatka, 324 n. 14, 330,
362363
Guerssel, M., 256, 288
Guillaume, Jean-Patrick, 318, 317 n. 7,
318 n. 8, 364
Gldemann, Tom, 676 n. 3, 697698
Gully, Adrian, 313 n. 3, 314, 363
guttural(s), 256 n. 9, 262, 266 n. 28, 280
n. 41
consonant, 658
Haak, Martine, 379, 401, 578, 652
h abbad, 17
habitudo, 215, 215 n. 12
h add structure, 151
h adf (deletion), 61, 163
al-h dira, 115
H adt , h adt , 25, 30 n. 3, 36, 120, 163
n. 21, 173 n. 31, 192, 198, 206, 530
canonical, 199
scholars, 30 n. 3
H adramawt, 14, 16, 199
Haeri, Niloofar, 404, 434, 654, 657,
665666, 669, 672
H afs an sim, 127
Hagar, 197
al-H akm, 315, 362
al-Hakkak, Ghalib, 313, 364
h l, 50, 50 n. 7, 57, 57 n. 13, 60, 136,
138, 138 n. 19, 139142, 145146,
159160
hall, 17
Halle, M., 256 n. 8, 288
halumma, 13 n. 9, 16, 17
H m, 196
H amdah, 565, 567 n. 9, 571
h amdiy, 566 n. 8, 567, 573, 575576,
576 n. 39, 577
dialect, 568, 568 n. 14, 569, 571573
h amd, 46, 47
H amdalah, 67, 70, 85, 9192, 96, 104,
106107, 109
H amdullh, M., 43
Hamilton, Charles, 31, 44
H ammm Farawn, 566
H amw, Subh, 364
hamza, 214 n. 10, 226, 437, 441,
444445, 598, 598 nn. 11 12
initial, 437, 445446, 451
al-wasl, 598
H anaf, H ., 43
H anafs, 31, 35
h arakat, h arakt, 211, 225, 231
ayn, 513
h aram, 199
h arf, 165 n. 24, 218, 507 n. 18
8jarr, 135
8nid$, 41
8tat niya, 182
8w$id, 14, 1617
Harmon, R.M., 305, 310
Harnell, Richard S., 605
H arrn, 196, 197
Harsusi, 604 n. 17
Hartley, A.H., 298, 301, 303, 310
Hartmann, Regina, 396, 401
Hrn, Salm, 143 n. 43, 161, 368 n. 6,
370 n. 13, 378 n. 40
h asan, 26
Hashemite(s), 549, 552
H assn, Tamm, 247, 249, 250, 254
Hasse, Johann Gottfried, 123
Hasselmo, N., 308, 310
Hassoun, Mohamed, 315, 363
h att, 21
Haugen, E., 308, 310
Hauschild, Richard, 209210, 243
Hava, J.G., 578, 586 n. 77
Hawa, 614615, 620
h wiy, 583
h ayawn
ntiq, 502
h ayt um, 1417, 19
Haywood, 437
head(s), 69, 7374, 76, 8082, 84, 8789,
9091, 9699, 103104, 156, 163, 368
n. 4, 376, 376 nn. 33-34, 377 n. 38,
378379, 38182, 382 n. 53, 383390,
393, 396, 482
function, 376, 381, 381 n. 50, 384,
388
nominal, 89, 480, 485, 494 n. 26
phrasal, 380
phrase
interjectional, 387
verbal, 378380, 393, 480, 482, 485,
494, 496
index
header(s), 386, 390
prepositional, 390
Heath, Jeffrey, 295296, 308, 310, 653,
670, 672
hbreu, 320 n. 9, 321 n. 10, 354355,
355 n. 36
Hebrew, 68 n. 1, 70, 80 n. 6, 108, 189,
193, 195196, 198199, 209, 219
n. 19, 221, 264, 273, 493, 595 nn.
34, 596, 596 nn. 56, 597, 598,
598 n. 10, 604605, 662, 668,
672673
Biblical, 595596, 599, 599 n. 13,
602603
classical, 198
dialectical, 198
grammar(s), 231, 231 n. 41
grammatical tradition, 209
Israeli, 596 n. 6
Middle, 600, 602
posodical-grammatical terms, 219
n. 19
pre-, 599 n. 14
proficiency, 662
Proto-, 596 n. 6, 598599
Qumranic, 600
terminology, 231
Hedjaz, 120, 120 n. 4
hedjazismes, 125
Heine, Bernd, 607, 621, 676 n. 4, 688
n. 19, 692, 698
Heinrichs, Wolfhart, 595, 602
Hell, Joseph, 208
hemistich(es), 548 n. 5
Hernndez Terrs, M., 243
Herrero Muoz-Cobo, B., 296, 310
Hetzron, Robert, 672
heuristics, 255, 286, 368
high vowel, 572, 574
H ijz, 16
h ikya, 1315, 15 nn. 1112, 1617,
1921, 22, 58
particles, 15, 15 n. 11, 2022
particles, non-, 15, 19, 2022
h ikma, 910
H imyar
kings of, 203
Hindw, H asan, 187
Hinds and Badawi, 596
Hinds, Martin, 542, 604, 683, 691 n. 21,
696, 696 n. 30, 698
Hindustn, 209
Hinnenkamp, Volker, 408, 433434
hispanicised forms, 229230
historians, 203
723
history
Arab world, 528
Hoberman, Bob, 595 n. 1, 3, 596 n. 6,
598
Holes, Clives, 128, 132, 248249, 254,
543, 563, 654, 656, 664665, 672
Holy
Congregation of the Propaganda Fide,
211 n. 2
Land, 211 n. 2
hominisation, 522
homme(s), 501502, 502 n. 4, 503504,
506, 512, 520521
homo
loquens, 502, 522
sapiens, 521
homonymy, 255 n. 1, 255, 257259,
261262, 269, 272, 278, 286287,
367 n. 3
homophonie, 515 n. 32
hongrois, 119
Hoogland, Jan, 391 n. 71, 401, 455, 473,
532 n. 5, 542
Hopkins, Simon, 77, 111
Hopper, Paul J., 692 n. 23, 698
host, 297
matrix, 291, 291 n. 1, 292, 295,
307309
minority, 297
superimposed, 291292, 308309
Hd, 195
humain, 345346, 351 n. 32,
non-, 351 n. 32
human, 370, 383, 389, 392, 394397, 397
nn. 8687
humanist learning, 219
Hurst, Nicholas, 305306
h urf, 38, 228, 237
al-h alq, 318
jarr, 228
Hurwitz, S., 262, 271 n. 32, 275, 288
H uwt t, 561562
H uwt , 562
hybrid forms, 224 n. 31, 655 n. 1
hybridity, 403, 405407, 408, n. 7,
409410, 413, 413, n. 10, 417418,
421, 421, n. 14, 422, 432, 433434
hypallage, 512 n. 27, 517 n. 42
hypercorrection, 80 n. 5, 596 n. 6
hyperonym(s), 465466, 469470
translations, 469
iarab, 218, 218 n. 17
Ibdu, 686
ibil, 47
724
index
Ibn Abbs, 193, 200

Ibn Ab d-Dam a-fi, 87, 111
Ibn Ab r-Rab, 150 n. 2, 153 nn.
56, 156, 168, 170171, 173174,
186
Ibn jurrm, 232, 232 n. 43, 233, 236,
238240
Ibn al-Anbr, 124, 151, 152 n. 3, 153,
160 n. 15, 162 n. 19, 163164, 175,
176 n. 34, 186187, 513 n. 29, 522
Ibn Aql, 17 n. 13, 22, 155, 187
Ibn Askir, 194, 198 n. 21
Ibn Durayd, 203
Ibn Fris, 125, 131, 435, 505 n. 12, 507
n. 19, 513 n. 28, 522
Ibn H anbal, 8687, 111
Ibn H azm, 201, 201 n. 32, 208
Ibn Him al-Ansr, 157, 175177, 177
n. 35, 178187
Ibn Hishm, 203
Ibn Jinn, 89, 9 n. 7, 1012, 22, 164,
164 n. 22, 169, 169 n. 27, 170, 170
n. 29, 187, 193, 317, 318 nn. 7-8
Ibn Kathr, 205 n. 40
Ibn MaJ al-Qurt ub, 11, 22
Ibn Man'r, 115, 131
Ibn Masd, 49 n. 6, 56, 56 n. 12, 61
Ibn Mujhid, Ab Bakr, 640 n. 1, 652
Ibn Mutarrif at-Taraf, 196 n. 13, 208
Ibn Mut, 35, 43
Ibn Qutayba, 197 n. 19, 203
Ibn Sad, 196 n. 13, 198 nn. 2021, 23,
199, 199 n. 25, 206 n. 44
Ibn as-Sarrj, 4, 22, 152154, 162163,
163 n. 21, 165 n. 24, 187
Ibn Taymiyya, 39
Ibn Usfr, 154 n. 10, 163, 163 n. 20,
164, 169, 171, 176 n. 34, 187
Ibn Uta, 200
Ibn Walld, 3334, 43
Ibn Xlawayh, 318 n. 8, 353 n. 34, 355
n. 35, 362
Ibn Xaldn, 3 n. 1, 23, 119120, 120
n. 6, 128129, 131, 543, 544, 563
Ibn Ya, 154, 154 n. 9, 157, 157 n. 12,
169, 172173, 178, 180, 187, 254, 316
n. 5, 316 n. 5, 319, 336 n. 22, 342, 362
Ibrhm, 190, 195200
ibrn, 196
ibtid, 137, 139, 144 n. 46, 145, 149,
153154, 156, 158159, 161, 162
n. 18, 167, 173 n. 32, 175176, 178,
185186
Icelandic, Old, 213

Id Abuw Silm (al-Atra at-Turbniy),
565
id, 14 n. 10,
idea, 396
ideal, 396
identit, 503504
identity(ies), 405406, 408, 423424,
428, 430, 433
cultural, 403
personal, 430
ideophones, 696
idm, 127
idiomatic expressions, 611, 618, 696
idm, 1617
idmr, 45, 5161, 63
Idrs, 195
Idriss, Socheil, 473
Igla, B., 294, 310
ignoratum, 235 n. 51
ijtim al-kalm, 56
ilBawti, 684
il, 141, 141 n. 32, 142, 142 n. 37, 143,
145146, 147, 160
ilhm, 192, 192 n. 3
illa, 114, 317
ill, 7, 10 n. 8, 12, 15, 17, 1921
illi, 7276
Ilys, Jsf, 326 n. 16, 331, 345, 364
iml, 63 n. 16, 141, 141 n. 32
imla, 650 n. 12, 658
Imlq, 203
imm, 16, 17
impratif, 343, 343 n. 25, 519
imperatiuo, 217
imperative(s), 34, 59, 217218, 295, 569,
575, 575 n. 36, 681
negative, 597
imperfect(s), 70, 224 n. 31, 570, 574, 575
n. 36, 576577, 581 n. 56, 586 n. 78,
603, 616, 640643, 643 n. 9, 645, 646,
681, 694 n. 27, 695
y-, 679
imperfectif, 315 n. 4
imperfective, 296, 613
implicatures, 27
implicit, 378379
implied, 392 n. 77
Imra al-Qays, 68, 86 n. 23
Imru al-Qays, 324
in, 14 n. 10, 16, 21
inaccompli, 315 n. 4, 332
inaccuracies, 472
index
inceptive, 21
incipient shift, 308
incompatibilities
semantic, 286
incongruities
semantic, 286
incrementation, 256
initial, 275, 277, 281
indefinite 26, 162, 248, 250251, 253, 373
indefiniteness, 217
India, 210
Indian, 209
indicative, 10, 228, 230 n. 39, 378
indices
contextuels, 362
indirect speech, 676, 676 n. 3, 678, 687
n. 16, 679, 683, 689
indispensable
part, 135136, 142, 146
predicate, 135137, 40141, 141 n. 33,
142143, 146147
indo-europennes, 118
Indo-Iranian, 295
inductive derivation of rules, 25
infinitive(s), 292293, 295296, 298, 301,
305306, 373, 383, 386
nominal
Dutch, 483 n. 10
infixes, 436
inflections
verbal, 294
informal, 403, 405, 423, 432
information, 403, 425
Ingham, Bruce, 544, 563
Ing-of construction, 476477, 483 n. 10,
484, 492
inherent, 373
inherited, 373
inn-, 74
inna, 21, 135
Qsentences, 165
innam, 11, 1516, 19, 19 n. 17
in, 28
ind, 126 n. 15
instabilit formelle, 317
instrument, 380, 393, 395
insults, 387
integrating terms, 661
integration
morphological, 292, 295, 297 n. 7,
298300, 304, 306309
morpho-phonological, 308
intensivit, 313
725
intention (qasd ), 9, 396397

interaction, 405, 405, n. 3, 410, 424, 427, 432
interdental(s), 567568, 586 n. 70, 591
n. 88, 657
interjection(s), 229, 374375, 387, 398
interne, 321, 335337, 344, 346347,
349350, 352, 353 n. 33, 354360,
362
interrogatif, 519
interrogation, 398
interrogative(s), 1719, 161, 170, 577
8a-, 156, 170, 184
component, 170
sentence, 169 n. 27
intertextuality, 543
intertwining vocabulary, 670
interview(s), 405, 405 n. 3, 409412,
415, 423433
format, 425, 427
multi-textual, 426, 428
set-up, 424, 427431
intransitifs, 337
intransitive 247
intransitivit, 319, 320, 320 n. 9, 336
invariant(s)
cognitifs, 324 n. 14
notional, 255, 255 n. 5, 256257, 257
n. 11, 259260, 260 n. 21, 261,
263266, 268269, 275, 279281,
283285
inversion, 151, 156
irb, 3 n. 1, 8, 46, 5051, 5860, 117,
118, 121123, 123 n. 12, 124126,
156, 180, 211, 218 n. 17, 228, 230
n. 39, 232, 234235, 237
endings, 46, 5051, 5860, 62 n. 15,
64 n. 17
ram b. Sm, 195, 202
iranienne, 113
Iraq, 197, 529530, 545546, 560, 560
n. 40, 561, 563, 656
Iraqi, 546, 561
iref, 660
Irish, Old, 213
irrgularit, 507
s, 190
arm a-ayh, 530
ib, 448
al-Isfahn, 203
Ishq, 195
Ishmael, 198, 202, 578
Ishml, 197
islh al-lafz, 170
726
index
Islam, 27, 39, 42, 92, 193, 196, 202, 205,

207, 404
heilige Sprache des, 119
medieval, 42
religious language of, 104
Islamic, 109
conquest, 207
data
historical, 42
East, 109
fundamentalists, 561
law, 42
scholars, 41
sciences, 28
society, 42
Islamic times
pre-, 105, 109
history, 192
past, 207
times, 189, 206
world civilization, 207
ism, 58, 155
al-fil, 341
fil, 177
kna, 179
al-maf l, 341
wh id, 14, 18
Isml b. Ibrhm, 190, 195, 197200, 202
ismiyya, 149150, 157, 158 n. 13, 165,
165 n. 24, 177
isoglosses, 608
Isolated Natural Subject, 28 n. 1
Israel, 546, 653660, 662, 667669, 673
sons of, 195
Israeli(s), 561, 595 n. 3
Hebrew, 596 n. 6
Israelite settlers, 197
istaqarra, 159160, 162164, 167, 180,
183185
hypothesis, 162, 164, 167, 180, 185
istidll, 34
istifhm, 4, 117
isti lh , 192
istinf, 49 n. 5, 50
istiqrr, 141, 163 n. 20, 179180
istish b al-h l, 36
istit n, 31
iswad, 659, 664
Italian, 296, 298299, 661
Italie, 124 n. 13
itb, 48, 52, 63
itration, 313
itimd, 152, 152 n. 4, 153, 153 n. 5, 154,
157 n. 12, 172, 179180, 184
izhr, 57
zm mudf, 215
izzayy, 86
Jabal Says, 127 n. 18
Jackendoff, Ray, 479, 498
Jdis, 195, 202203
al-Jaft, 562
Jhiliyya, 190
al-Jhi', 13, 45, 122123, 131, 328
n. 19
jhi'ien, 123
Jahn, G., 136 n. 8, 147, 238
Jir Ab _amd, 195
jiz, 33 n. 4
James, Gregory, 210, 243
Janah, A., 283, 287
Japanese, 210
scholars, 210
Jarajira, 567 n. 9
jarr, 29, 216, 225, 227, 228, 230
Jsim, 203
Jastrow, Otto, 658, 671672, 698
jawb, 53
al-Jawhar, 3
jaz, 59
al-Jazra, 194
jazm, 59, 226228, 230231, 234
Jeremiah, 595 n. 2
Jerusalem, 8182, 84, 221
Jerusalemites, 80 n. 6
Jesuit, 209210
Jesus, 190
Jewish, 82 n. 14, 101, 106, 109, 296
Aramaic, 600601
Jews, 195
jezme, 215 n. 11
Jibrl, 200
Jhn Mahmd, 528
jinn, 190
Jordan, 543, 543 n. 1, 545548, 551,
561562, 601, 661
Jordanian(s), 546, 548, 552, 554, 557,
557 n. 35, 559, 560 n. 40, 560562
society, 547
tribe, 546
Jos de Len, 221
Joseph, 190
Joon, Paul, 264, 288, 314, 320 n. 9, 321,
321 n. 10, 335 n. 20, 344 n. 26, 354,
364
Juba, 607 612, 614, 618, 620622,
637638
Arabic, 607-621, 638, 692693, 698
index
courts, 607609, 620, 693
Rejaf, 609
Judaeo-Arabic, 7980
Jdisch, 105
jumla, jumal, 113, 177
fi liyya, 149150, 157, 165, 165 n. 25,
166167, 177 n. 35, 178186
ismiyya, 149150, 157, 164167, 169,
177186
inverted, 150, 150 n. 2, 162, 182
arziyya, 177 n. 35
zarfiyya, 158, 167168, 179181,
183185
wasf, 158 n. 13
juncture 249
Jurhum, 195, 197200, 202
jurisprudence, 42
al-Jurjn, 8, 1012, 2223, 154 n. 7, 163,
164, 174, 187, 232, 235, 238240
jussive, 10, 16, 56, 228, 230 n. 39,
Juynboll, W.M.C., 243
ka, 78
kaanna, 16
kaannam, 16, 19
kaayyin, 16
Kaba, 196, 197, 206
kabs(a), 668
kad, 1617, 31, 3435, 35 n. 7
Kahle, Paul, 125, 132
Kairenisch, 72, 74, 699
Kakwa, 609
kalm, 27, 47, 4950, 5255, 5759,
177
al-arab, 49
al-arab/badw, 61
Kalb, 202
kalima wh ida, 14, 1617
ka-llad, 7778
kallad, 78
kam al-xabariyya phrase, 171
kam, 16, 78
Kamel Al-Kabir French-Arabic, 458, 462,
466, 468
kna
an-nqisa, 135
at-tmma, 178 n. 36
Kaplony, Andreas, 113 n. 1
al-Kar, Mahmd, 435, 437, 441447,
449450, 453
kasr, 227
kasra, 218, 218 n. 16, 239
kasratn, 218
Kator, 609, 610611, 617620, 634
727
Kaufman, Terrence, 294, 311, 665

Kautzsch, 320 n. 9
Kaye, Alan S., 595, 597, 605, 654, 657,
664, 665, 655 n. 1, 672
kayfa, 20
Kayne, Richard, 494, 499
Kzim Bahr al-Marjn, 187
Kazimirski, A. de Biberstein, 115, 132,
257, 277, 284, 288
Kelly, James, 437, 453
Kembera (Austronesian), 698
Kennedy, J.H., 305, 311
Kenstowicz, M., 256 n. 9, 257 nn. 1213,
288
Kenya, 608, 621
Kerkhof, Maxim, 243
kesra, 224226, 231
kesre, 233
Khadduri, Majid, 26, 44
Khalidi, Tarif, 202 nn. 3334, 208
Kharbush, A., 661, 672
Kharga, 536
Khartoum, 608, 620
Khatef, L., 272, 288
Khim, Alain, 401
Khubchandani, L.M., 308, 311
Kifya movement, 529
Kihm, 368 n. 4
Killean, Cavolin G., 210 n. 2, 243
Kinna, 116
kinya, 34
Kinberg, Naphtali, 50 n. 7, 54 n. 11, 65,
384 n. 55, 401
Ki-Nubi, 607608, 621, 692
Kirsten, Peter, 232, 234, 238
al-Kis, 52 n. 9, 54
Kissling, Hans-Joachim, 679 n. 7, 692
n. 23, 698
Kister, M.J., 205 n. 40
kitb, 141, 316, 318, 320 n. 9, 329, 338,
338 n. 23, 353 n. 34, 504, 504 nn. 8,
10, 507 n. 18
Klamer, Marian A.F., 698
Kleiber, Georges, 506 n. 16, 522
Koehler & Baumgartner, 595 n. 2, 596,
597 nn. 89, 598, 598 n. 10, 599, 599
n. 13, 600602, 602 n. 16, 603605
Koehler, Ludwig, 605
koinsation, 126
Konjunktion, 69, 75, 87 n. 24
Koplewitz, Emanuel, 662, 672
Koptisch, 697
Koran, 105, 223 n. 28, 410, n. 8, 411
530531
728
index
Koranic reading, 640 n. 1

Koster, Cornelis, 369 nn. 8, 10, 370, 389
n. 69, 401
Kouloughli, D.E., 23, 161, 164, 168, 175,
185 n. 38, 187
Kremers, Joost, 475, 475 n. 1, 484485,
485 n. 14, 493, 499
Krio Creole English, 692693
Kroeger, Paul R., 693, 693 n. 24, 695, 698
Kropp, Manfred, 202 n. 35, 204 nn.
3839, 208
Kfa, 117
Kfan(s), 35, 6364, 151, 154 n. 8, 157,
160, 160 n. 15, 162 n. 19, 175176, 181
Kundlzza Rys, 543 n. 1, 563
Kurpershoek, P. Marcel, 544, 560 nn.
3839, 563
Kuwaiti Arabic, 563
l, 14
LA, 659 n. 4, 665 n. 7, 666 n. 9
laalla, 16
laallam, 1617
labels, 367, 373 n. 20
labial(s), 255256, 256 n. 6, 257, 259,
264265, 267268, 270, 279282,
284286
Labd, 18, 86, 87 n. 23, 111
Labov, William, 129, 405, 410, 410, n. 9
ladu, 546
lafz, 151, 169 n. 26, 173, 176
structure, 151, 171
lafz, 10, 10 n. 8, 11
lafziyya, 235, 235 n. 49
Lgiy clan, 565 n. 3
law, 14, 14 n. 10, 1516, 136, 138, 146, 160
lah n, 117118, 123, 206
lam, 20
lm, 21
al-lm al-friqa, 21
lamh al-asl, 17, 17 n. 13, 20
lamm, 16
Lammens, 200 n. 28
lan, 15 n. 12, 123 n. 12, 124
Landberg, Carlo le Comte de, 544, 563,
690, 697 n. 31, 698
Landsberger, Benno, 601
Lane, E.W., 147, 581 n. 57, 600, 605
langage naturel, 521
Langhade, Jacques, 120 n. 5, 132
Landsberger, Benno, 601
language(s)
academies, 661
as behavior, 4
body-, 29
community, 291292, 308

contact(s), 291, 293298, 306309,
653, 670
dominant, 610, 662
embedded, 291, 291 n. 1, 295, 297,
307308
exposure, 309
faculty, 479
foreign, 308
institutes, 661
interference, 653, 670
matrix, 292, 307
natural, 367
shift , 307
socially, 291292, 307
specific pharyngeals, 657
structure, 398
superimposed, 295
usage, 27
metaphorical, 369
ordinary, 369
varieties, 408, 410, 418, 420, 433
langue(s), 502, 502 n. 4, 503, 505 n. 12,
506, 506 n. 16, 508, 511512,
517521, 521 n. 49, 522
arabe, 124, 121, 316, 317, 319, 349,
355, 501, 505, 508 n. 21, 509-511,
515 nn. 31, 32, 519-521
commune, 521
historique(s), 515 n. 31, 519 n. 46
humaines, 521
littraires, 121 n. 8
maternelle, 128
naturelles, 521
ngro-africaines, 503 n. 5
smitiques, 506, 508, 508 n. 20, 511, 514
vulgaire, 121, 121 n. 8, 124 n. 13
Larcher, Pierre, 28, 44, 113, 120 n. 5, n.
6, 121 n. 9, 126 nn. 14, 17, 128 n. 21,
130, 132, 314, 316 n. 5, 321 n. 10, 335,
341 n. 24, 364
laryngeal(s), 598, 642, 644, 646, 646
n. 10, 647651
final, 639, 641645, 647650,
650 n. 12, 651, 658
lta, 20
lateral, 257, 257 n. 13, 268270,
275276, 278, 280282, 286, 657
Latin(s), 115, 119, 121, 121 n. 8, 124,
211 n. 2, 211, 213217, 221, 229233,
235236, 239240, 661
classiques, 124
letters, 220
sources, 211 n. 2
terminology, 211, 218, 236237
index
latine, 124, 502
no-, 119
latinized form, 231, 233, 236 n. 53
Lavie, Smadar, 578
Lavorel, Pierre-Marie, 506 n. 15, 522
law, 2527, 33, 3840, 42, 209, 213, 243
Islamic, 25
religious, 205
Law, Vivian, 213
lawl, 16
lawm, 16
laysa, 20
layta, 20
Leeman-Bouix, Danielle, 332, 364
Left-to-Right Association, 489
legal, 2526, 33, 33 n. 4, 40, 43
argumentation, 27
contexts
medieval, 42
definitions, 37
implications, 25
methodology, 25
obligation, 32
principle, 37
sciences, 35, 39, 41
semantics, 38
system, 25, 43
terminology, 31
theory, 25
valid speech, 37
works, 33, 41
legalistic, 38, 42
Lgt, 565566, 566 nn. 7, 9, 567 n. 13,
574575, 578
Lgiy, 565 n. 3, 566 nn. 78, 567, 570,
573, 575578
dialect, 567568, 568 n. 14, 569,
571574, 574 n. 32
Lehmann, Christian, 87 n. 24, 111
lengthening, 446449
consonants, 447
prosodic, 446, 448449
Lentin, Jerome, 211, 243
LeTourneau, Mark, 393, 401
Levant, 547
Levin, Aryeh, 135, 137 nn. 1314, 138
n. 19, 140 n. 30, 141 n. 33, 143 n. 41,
147149, 153, 153 n. 6, 158, 162, 163
n. 20, 187, 391, 391 n. 73, 395, 399,
401, 672
lexemes, 568, 586 n. 70, 614, 646,
653655, 655 n. 2, 657, 659 n. 4, 660,
661, 663, 665 n. 8, 666, 670, 672
lexical, 38, 56, 611612, 616, 619
aspect, 654655, 676
729
domains, 662
cultural areas, 662
politics, 662
religion, 662
figement, 362
forms, 641
items, 653, 655, 661, 667, 670
literary Arabic pairs, 656
meaning, 3839, 675
pairs, 656
lexicographer(s), 207
lexicographical, 39
work, 211, 219
lexicography, 435, 448, 450, 471
Arabic, 455, 471
lexicon, 255, 273274, 287, 370, 393,
395, 406413, 415416, 418, 418
n. 13, 420422, 449, 608
Arabic, 255, 286, 399
semantics, 653
verbal
lexique arabe, 326
li + a relative clause, 140
li + genitive, 140
li-ann, 76
Libya, 656
Libyan
Eastern, 640
Lichtenstdter, Ilse, 239, 243
lien, 506
gnral, 515
limite, 504 n. 8
linear
descriptive text analysis, 63
interpretation, 63
linearization, 494495
Lingala, 609
lingua franca, 661
linguist(s), 150, 609
Egyptian, 529
linguiste(s), 115, 118, 130
linguistic(s), 2530, 33 nn. 4, 6, 36,
4042, 4446, 52, 68, 190, 192, 295,
307308, 369, 403404, 408, 410411,
420, 422, 424, 433, 527528, 533, 543,
559, 565 n. 2, 595 n. 1, 596 n. 6, 605,
607, 609, 619621
Arabic, 250, 435, 437, 441, 448449
analysis, 34, 10, 12, 20, 22, 64 n. 17
aspects, 26
awareness, 5, 89
codes, 25
community(ies), 193, 201
corpus, 25
developments, 541
730
exegesis, 63
extra-, 33, 36
forms, 189
history, 191, 195, 200, 203
history of, 209 n. 1, 210 n. 2, 211
ideas
history of, 4
methods, 45
panorama, 201
phenomena, 4, 910
process, 25
sciences, 35
situation, 527
systems, 4
theory, 213
typology, 80
works, 222
linguistique, 113, 118, 133, 502
arabe, 121 n. 9, 129, 317, 317 n. 7,
323
lhistoire de la, 113
mdivale
arabe, 314315, 318320, 329
historique, 118119, 121, 123, 130
no-hellnique, 121 n. 9
Lipinski, E., 273 n. 34, 288
liquids, 262
lisn, 196 n. 15, 257, 288
mm, 119, 121
mm, 121
asl, 120
mudar, 120
arab, 190
Literary
Arab, 658
Arabic, 653655, 655 n. 2, 656670,
672673
articulation errors, 659
and Colloquial, 653657, 659,
662663, 665, 668670
diphthongs, 657
lexemes, 655, 668
lexical items, 656
pronunciation, 658
text, 659
vocabulary, 661, 666
written and spoken, 654
language, 80, 85
literature, 403404
Arabic, 403, n. 1
Littmann, Enno, 126 n. 16, 132
loan
translation, 655
verbal, 302
index
verbs, 299, 303, 652
word(s), 298 n. 8, 299 n. 9, 301,
307308, 472, 540, 602, 655
English, 299300, 304, 307
local, 404405, 417, 433
dialects, 659
localisation, 505
locus, 417
London, 529530
long
monophthongs, 657
vowel phonemes, 568
Lontos, S.S., 301, 303, 311
Lord, Carol, 679, 693, 698, 698 n. 7
Loucel, Henri, 39, 44, 208
Lowenstamm, J., 256, 288
Luffin, Xavier, 607, 621
lua(t)
8mmiyya, 121 n. 7, 122
8drija, 121 n. 7
8fush , 120, 120 n. 3, 121 n. 7, 125
8Mudar, 120
QQuray, 120, 125
Lt , 195
Luxor, West Bank, n. 6, 678
Lyons, John, 321, 324 nn. 1314, 326
n. 17, 364
m, 14 n. 10, 1819, 21, 79
interrogative, 13 n. 9
ma-...-, 688
Maan, 562
Maadd, 201
ma, 667
Maa, Kurt-Jrgen, 243
Mceiaru, Adrian, 595 n. 1
Machrek, 128
marf, 235 n. 51
md, 1718, 18 n. 14, 19, 22
maddo, 226
madh , 555, 558, 561
md, 59, 315, 315 n. 4, 316317, 323,
331332, 334, 339
Madina, 601, 605
Maeda, S., 266 n. 28, 289
al-Mafraq, 546
maf l, 151, 179, 214, 217
maful, 214
maful, 216
Maghreb, 101, 128, 223
Maghrebine, 101
Maghrebis, 80 n. 6
mah all, 169 n. 26, 249
al-waqf, 249
index
Mahdi, 120 n. 5, 131
al-Mahdiyyah, 81, 83, 110
Mahmd Bayram at-Tnis, 544
Mahmd Taymr, 666, 699
Mahmud, Ushari, 607608, 621
Maiduguri, 640, 642, 647648, 652
Maingueneau, Dominique, 324 n. 14,
326 n. 17, 331, 364
majz, 507 n. 18
majrr, 177, 214, 216
Makkah, 200
Makram & Umar, 46 n. 2, 49 n. 6, 52 n. 9
Makram, Abd al-l Slim, 65
Malak al-Araf, 221 n. 26
malaka, 3 n. 1
malapropism, 613, 615, 617
Malbad, 566 n. 9
Malekiyya, 620
Malkiel, Y., 294, 311
Malta, 698
Maltese, 296, 299, 650, 650 n. 12, 651652
Maluk, 612, 622625
Mamluk, 89
maml, 124, 149, 151, 168, 170, 235
man, 38, 151, 157, 176
l-fi l, 176
l-hadt , 38
l-hudt , 342
l-kalm, 38
n-nasb, 38
t -t ubt, 342
manaw(iyya), 10, 10 n. 8, 11, 235,
235 n. 49
mand, 1819, 22
Mandarin, 223
Manda, 680, 684, 693 n. 24, 697
Al Manhal French-Arabic, 458, 462,
466468
maniabilit, 508
manire, 504 n. 8
manner, 392, 394395, 397
al-manqs, 252253, 449
mana, 507 n. 18
Mansouri, W., 256 n. 6, 288
mansb, 124
Manwel Mifsud, 698
mapping, 482
idiosyncratic, 481, 481 n. 6
one-to-one, 481 n. 6, 497 n. 30
phonological, 498
principle, 479, 479 n. 4, 480, 484, 493,
495
general, 480
lexical, 480
731
rule, 479, 497 n. 29, 481 n. 6, 482, 492

idiosyncratic, 481, 497, 497 n. 30,
498
Maqil b. Xuwaylid, 89
al-maqsr 252253
Marais, William, 121 n. 9, 132, 690,
698
Mardin, 640
marf, 59
marker
adjective, 249
adverbial, 249
case, 411412, 476
idiosyncratic, 490
infinitive, 298, 299 n. 9
mood, 409
non-finite, 488
object 249
present tense, 409
relative clause, 406407, 419, 412
stem
verb, 496
stem VII, 488489
verb
foreign, 299
marking
nominal, 476
Maronite, 211, 220, 222 n. 26
Christians, 211
martaba structures, 171
Martelotti, Francesco, 226
Martelottus, Franciscus, 226 n. 33, 234,
236, 238241
Maruyama, Toru, 210, 243
Marwn Qadr, 535
Marx, Emanuel, 578
Masd, 566567, 569 n. 9
masculin, 123 n. 12
masdar(s), 47, 118, 210, 355, 381 n. 50,
383, 436, 475, 484487, 487 n. 18,
488494, 494 n. 27, 495497, 497
n. 28, 498, 513514, 532 n. 7
+acc, 484485, 493
Arabic, 475, 484, 487488, 498
formation, 475, 486, 492, 495496
Arabic, 493
idiosyncratic, 492
+li, 484, 485 n. 14, 492493
mmy, 492
stem II, 490, 492
stem III, 492, 497
stem III, 497 n. 30
stem IX, 489 n. 21
stem V, 490 n. 22
732
index
mashl, 235 n. 51
Masoretic times, 219 n. 19
mass media, 669
al-Masd, 196, 196 n. 16, 202
matching, 377, 380, 390, 390 n. 70, 393,
396397
agent, 395
feature, 393
gender, 397 nn. 8687
number, 397 n. 86
matres lectionis, 540
matrix(ces), 255256, 256 n. 6, 257, 257
nn. 11, 14, 258260, 260 nn. 21, 22,
261266, 266 n. 28, 267269, 269
n. 30, 270271, 273275, 277280,
280 n. 41, 281285, 287, 291
binary, 273
language, 670
segment, 262
source, 263
Mauritania, 223
mawdi, 40
mawd
al-lua, 507 n. 18
Al Mawrid English-Arabic, 458459, 468
Mawsat al-hadt , 111
mawsf, 151, 153
Mawwls, 694
maxim
of manner, 28
of quality, 27
of quantity, 27
of relation, 27
maxrr, 214
maxrr, 215
al-Maydn, 502 n. 3, 523
mazd, 313
Mzin al-Mubrak, 187
Mzin Mubrak, 113, 131
Mazraani, Nathalie, 655657, 659, 666,
672
McAuliffe, Jane, 208
McCarthy, John, 259 n. 19, 288, 421 n.
15, 476, 486, 486 n. 17, 488, 490, 492,
494495, 497 n. 29, 498499, 671
McKenny, J., 304, 311
Mdaxxal Slmn, 565 n. 3
meaning(s)
general, 466, 469
specific, 465467, 469
Mecca, 197, 202203, 206, 561
Meccan period
late, 190
middle, 190
Mecque, 117, 125

mdde, 215 n. 11
media, 403404, 404, n. 2, 405, 432
broadcasting, 403404, 404, n. 2, 405
written, 404, 405 n. 2
Medina, 202
mdio-passif, 335
Mehri, 604 n. 17
Meijer, Hans, 370, 401
Meiseles, Gustav, 654, 656, 664665,
672
Meissner, Bruno, 544, 563
Mejdell, Gunvor, 434, 654, 655 n. 1, 672
mesolectal, 608, 610614, 616617,
619620
Mesopotamian, 658
Messiah, 190
metalanguage, 217 n. 15, 227
non-Western, 231
Spanish, 240
meteg, 219 n. 19
metheg, 219 n. 19
methodological armory, 33
mthodologies, 501503, 504 n. 8
linguistique, 501
methodology, 236
Michaelis, Johann Davis, 123
Middle
Ages, 221 n. 25
Arabic, 655, 655 n. 1, 662, 665, 669
Hebrew, 600, 602
mifal, 249
mif al, 30, 31 n. 3
Mifsud, Manwel, 296, 311, 650 n. 12,
652, 672, 698
milehal, 219 n. 19
Miller, Catherine, 128, n. 19, 607608,
619, 621, 692693, 698
minb, 218
minenbn, 218
minib, 218
minib, 218
minib, 218
mininbn, 218
minunbn, 218
mirbad, 30
missionary(ies), 210, 211 n. 2, 212, 219
n. 19, 231
linguistics, 209 n. 1, 219, 219 n. 20,
240, 244
sources, 223
Spanish, 211 n. 2, 219
Mitchell, Terence F., 248, 254, 664665,
672
index
mix(ed)
form(s), 611615, 617
language, 655
utterance, 658
m-masdar, 492, 492 n. 25, 497
mnemonic, 217218
terms, 239
words, 220
Moabite, 602
mobility
social
upward, 291
modalit, 516517, 517 n. 41, 519 n. 45
danimit, 517
ditration, 508 n. 21
de temps, 517
de vie, 517
mode(s), 510
of action, 263
irrel, 519
potentiel, 519
prepausal, 247, 252
rel, 505 n. 13, 507, 507 n. 18, 519
model
descriptive, 368
prosodic, 476
modle italien, 121
modern
literary Arabic, 654, 667, 658
Moroccan, 596, 605
proto-colloquial, 597
scholarly terminology, 665
Standard Arabic, 527, 545, 568569,
598, 598 n. 12, 654, 669, 671
standard language, 560, 598 n. 12
technology, 661
version of Middle Arabic, 665
written Arabic, 605
modification
smantique, 334
modifier(s), 367 n. 1, 368 n. 4, 376 n. 34,
adverbial, 378379
aspectual, 378
modal, 378
peripheral, 378379
temporal, 378
modus, 505 n. 11, 512, 512 n. 27, 513,
515, 517 nn. 4243, 44, 519, 519 n.
46, 521 n. 49, 522
communs, 513
dexclamation, 516
dassertion, 515
gnral, 515
infinitif, 514
733
Mohamed Sami, 536

Mokilko (East Chadic), 604 n. 17
Mol, Mark van, 458, 462, 466, 468, 473
Mombasa, 621
monocausal, 6970
monolingualism, 293, 297, 304
monophthongs and , 569
monotheist, 197
monotheistic religions, 196
Monroe, James T., 210 n. 2, 220222, 244
Monteil, Vincent, 654, 672
mood, 248, 251252, 372, 384 n. 55
aspect, 371
imperative, 378
perfect, 371 n. 17
non-, 371 n. 17
subjunctive, 386, 388
mora, 486 n. 16, 489, 491, 491 n. 23
Moravcsik, Julius, 396, 402
Mrner, A.R. & Magnus Mrner, 222,
244
Moro, 609
Moroccan Arabic, 596, 605
Morocco, 232, 690
morpheme(s) 491, 496
autosegmental
content, 421
fem., 570, 574
grammatical, 421; 496
masdar, 492 n. 24, 495, 498
morphological, 435, 435 n. 2,
436442, 442 n. 6, 443450, 452
nominal, 438
nominalizing, 493
non-finite, 490
non-finiteness, 489
paradigmatic, 436
prosodic, 486
root, 489, 495
radical, 435
verbal, 438
vowel, 438, 439
morphmes
dfinisseurs, 519
spcificateurs, 519
morphologic, 9
morphological, 63, 87 n. 24, 91, 568
n. 15, 570, 592 n. 91, 608, 611612,
615616, 642643, 645, 648650
differences, 659
reduction, 687
morphologie, 316
arabe, 508 n. 21, 509 n. 22
mdivale, 344
734
index
morphology, 256, 292, 296297, 299,

307309, 368, 370, 370 n. 12,
373375, 384, 388390, 396, 398,
406407, 412, 418420, 431, 436, 443,
449, 475, 485, 573, 653654, 659
Arabic, 475476, 486, 494
concatenative, 498
non-concatenative, 475, 498
nominal, 573
verbal, 574, 651
morphophonological, 648649, 653
morphosyntax, 212, 479 n. 4, 480
Moscati, Sabatino, 315 n. 4, 316, 364,
597, 605
Moses, 190
mother tongue, 201
motion, 396, 397
Mounin, Georges, 506 n. 16, 523
Mount Jd, 195
Moutaouakil, Ahmad, 28, 44
mouvement smantique, 116
Moyen Age, 121 n. 7
MSA, 527, 529, 529 n. 1, 531533, 535,
537538, 540, 586 n. 70
Muallaqah, 86, 8687 n. 23
(al-)Mubrak, Mzin, 113, 131, 187
al-Mubarrad, 4, 23, 33, 175, 176 n. 34
al-Mubarrid, 329, 335, 343, 362
mubtada, 48, 137, 137 n. 13, 138139,
144145, 149, 151153, 154 n. 9,
55157, 157 n. 12, 158, 158 n. 13,
159164, 168170, 170 nn. 2829,
171173, 173 n. 32, 174176,
178179, 181186, 214, 216
clause, 172
definite sentence-initial, 174
indefinite, 169, 169 nn. 2627, 170,
170 n. 29, 172, 174
al-kalm, 55
position
pre-, 149
post-, 174
xabar, 151152, 169, 170- 173, 177
mubted, 214
mubtede, 215
mudf, 214
mudf, 214, 216
Mudar Badrn, 554, 557 n. 35
Mudar, 120, 129
mudri, 230, 315 n. 4, 316319, 323, 338
mudmar, 151
mufrad, 164
muayyira, 1516
muh addat anhu, 167, 173, 173 n. 31,
181
muh l, 29
Muhammad al-Fanti l al-H ajy, 560
Muhammad b. Al, 198
Muhammad Salm, 199
Muhammad Taymr, 700
Muhammad, 36, 39, 86, 106, 189, 191,
193, 195, 198, 200, 202206
muxtab, 4, 36
Muhazzim Qaht n, 203
Muhtaseb, 396
al-Mujam al-wast, 331, 340
mujarrad, 313, 314 n. 3
al-Muji, 162 n. 19, 164, 164 n. 23, 187
mulan, 136, 138, 146
multicausal, 70
multilingual, 609
Muluk, 616, 632
mund, 214, 217
al-mund, 221
munde, 214
Munra, 683
al-Munjid, 331, 340, 347
muqaddar, 163
murd, 37
Murray, George W., 565 n. 3, 567 n. 9,
567 n. 13, 575 n. 37, 578, 589
Ms, 190, 548, 551552
Musaylima, 204
mush af, 36
Musharrafa, Musta fa, 434
Musil, Alois, 544, 549 n. 9, 563, 690, 698
Muslim, 25, 29, 37, 42, 79, 92, 104, 189,
191192, 193 n. 4, 196, 530
Arab
tradition, 199
scholarship, 192
Baghdadi speakers, 669
dogma and consensus, 189
faith, 207
narratives, 197
revelation, 191
scholars, 34
Spain, 221 n. 25
speakers, 657
speech, 92
musnad, 177
al-mutak, 221
mustaqarr, 136139, 141142, 145146
ayr, 136, 146
mustaqm, 26, 37
al-mustaqti, 221
al-mustat, 221
musulman, 128
mutakallim, 4
al-Muta rriz, 239240
index
al-mutasamm, 221
al-mutawallid, 221
mutawwalt, 3
Mutazil, 39
mutazilites, 11, 193
muwallad, 115
muxbar, 174
Muysken, P., 293, 295, 302, 302 n. 10, 311
Muzeina, 566, n. 3
Myers-Scotton, C., 291 n. 1, 311, 670, 672
Mznah, 566 nn. 7, 9
dialect, 567 n. 12
Mznih, 567568
Mzniy, 575
N projection, 477
Nabat, 201
Nabatean(s), 197, 200201
inscriptions, 197
kingdom, 201
language, 201
script, 201
Nabawiyya, 682
an-Nbiah, 86, 87 n. 23
Nad Tmn Ab Tyih, 562
Nagba, 691
nahw, 12, 23, 113114, 118, 121, 123
n. 12, 133
nahwa, 118
nahy, 34
najr, 9 n. 7
Nairobi, 620
Najd, 544, 560 n. 38, 563
Namrd, 196
nqis, 178
narration(s), 415, 417, 428, 430432
an-Nas, 87, 111
nasal, 257, 259261
nasb, 30, 30 n. 2, 32, 34, 41, 4749, 49 n.
6, 5051, 5556, 58, 60, 124, 158, 160,
162 n. 18, 169 n. 26, 175, 227228,
230, 234, 238
nasbo, 225
nsib, 58, 124
Nsf al-Yzij, 35
Nsf, Jirjis, 326 n. 16, 331, 345, 364
nassuantes, 230
nat, 47, 49
native speaker(s), 72, 94, 101, 527
of Arabic in Israel, 662
Hebrew, 596 n. 6
natural language
processing, 367, 370, 389
Arabic, 367
Understanding System, 396
735
naw, 7 n. 4
nazm, 12
Nebajoth/Nabat, 201
Nederhoff, Mark-Jan, 370, 402
Neeleman, Ad, 475476, 478479, 480
n. 5, 481, 481 n. 6, 482, 482 n. 8, 483,
483 n. 10, 484, 486, 495, 497 n. 30,
498
ngatif, 519
negation, 14, 378 n. 42, 398, 505 n. 13,
578 n. 42, 694
value, 378379, 381, 381 n. 51,
382
negative(s), 14, 21, 406407, 413, 419
participle, 154
sentence, 169 n. 27, 170 n. 29
Negev, 567, 574 n. 32, 575, 575 n. 37,
577578
dialect, 570 n. 22
type of dialect, 567
neoArabic, 75, 77, 639, 649, 651
Aramaic dialects, 595 n. 3
neologism(s), 465467, 469472
Arabic, 471
translations, 467
Nepos, Ferdinandus, 215, 215 n. 12,
242
van Ness, S., 308, 311
Nestorian, 106
nestorianische Kirche, 106
The Netherlands, 565 n. 2
Netherlands-Flemish Institute, 529
neutestamentlich, 105
new
terminology, 655, 661
Testament, 106, 189
World, 212
Newmeyer, Frederick J., 693, 695, 698
news, 405, 412
Neyreneuf, Michel, 313, 364
Niederehe, Josef, 210 n. 2, 242, 244
Nigeria, 640641, 648, 652
Nigerian, 640
Arabic, 639641
Nijmegen, Arabic corpus, 532 n. 5
Nile valley, 545 n. 2
Nilotic
Dinka, 610
speakers, 613
nisba, 194
niyya, 7, 7 n. 4, 151
Nizr, 201
Nizri(s), 202
nodes, 369 n. 8
736
index
Nldeke & Schwally, 105, 107

Nldeke, Theodor, 111, 200 n. 30
nom, 315 n. 4, 362, 504, 504 n. 8, 505,
505 nn. 12, 41, 506
abstrait, 512 n. 27
commun, 508, 512 n. 27
nomad(s), 546
Syrian, 690
nomade(s), 117, 120
dOrient, 697
nomadism, 545, 563
nombre, 513
nominal, 92, 109, 153, 157, 181, 183, 219
complex, 476 n. 2
constituent, 153, 184185
declension, 216
element(s), 157, 165, 165 n. 25, 178,
181
ending, 228
event, 476
simplex, 475 n. 1
indefinite, 172, 173
non-referential, 155
paradigm, 214
phrase, 162
referential, 155
sentence(s), 6, 18, 77, 135138,
144146, 149, 166, 178, 216
inverted, 154 n. 8
result, 475 n. 1, 476 n. 2
simplex, 476 n. 2
system(s), 659
type, 80, 86, 8990, 9293, 98
nominality, 476
nominalization, 293, 483
nominalized
clause, 171
phrase, 58
nominalizer, 488492, 492 n. 24,
495497
idiosyncratic, 490492
nominatif, 124, 130
nomination, 507508, 508 n. 20, 509
n. 22, 520, 522
des entits, 508
des expriences, 508
smitique
nominatiuo, 214
nominatival
constituent, 156, 181, 183
indefinite phrase, 173
nominative, 10, 13, 18, 19 n. 15,
137139, 142, 144145, 160, 178
n. 37, 215, 227228, 230, 238, 374
nominativo, 227
nominativus, 221
nonBedouin Colloquial Arabic, 658
compound, 15, 19
concatenative morphology, 498
contrastive compatibility, 667
coreferential, 92
emphatic
articulation, 658
phonemes, 658
finiteness, 488
ikya, 15, 1922
past verbal patterns, 660
phonetic dictionaries, 657 n. 3
restrictive, 26, 33, 103
clause, 91
Nortier, Jacomine, 297 n. 5, 308, 311
Norwegian, 484 n. 13
Research Council, 209 n. 1
nose, 257, 260261
notificacin, 226
noun(s), 78, 13 n. 9, 14, 17 n. 13, 18,
41, 43, 50, 59, 6869, 84, 88, 92, 95,
102, 108, 135 n. 2, 137139, 141142,
144145, 151152, 155156, 156
n. 11, 158 n. 13, 160, 165167,
177, 180, 213214, 218, 218 n. 17,
225230, 230 n. 39, 231, 235, 235
n. 49, 239, 248252, 299, 302303,
305, 308, 367, 370, 370 nn. 1214,
373, 376 n. 34, 381, 383, 390,
396397, 436437, 439, 441447,
451452; 482 n. 8, 573
abstract, 16, 20, 386, 436
accusative, 8
agentive, 481, 481 n. 6, 482
biradical, 438
clause(s), 6869, 172
collective, 392 n. 78
colloquial Arabic, 226
common, 373, 382383
corroborative, 18
definite, 251
definite subject, 150
deverbal, 373374, 382, 484 n. 13, 488
ends, 227
genitive, 7
indefinite, 169, 173, 175, 251, 381
individuality, 373
multitude, 373, 381383
nominatival, 149, 173, 173 nn. 3031, 181
+noun, 165
phrase, 153, 169, 174175, 368 n. 4,
376, 386, 693
definite, 150
index
nominatival, 159
plural (non-human), 155
singular, 155
proper, 213
quality, 373
relation, 373
tetraradical, 442, 446
triradical, 439
uniradical, 448
verbal, 60, 247, 382, 436
vessel, 373
vocative, 41
NOW, 565 n. 2
noyau, 517518
de la phrase, 518
NP, 695
NSA, 609618, 620
N-to-D movement, 494
Nuayma, Mxl, 673, 666
Nubi, 607608, 621
Nh, 190, 192, 194196
Nuijtens, E.T.G., 297 n. 6, 311
Numn r, 699
number, 368 n. 4, 370, 371374, 378,
382383, 389, 395
numerals, 35, 373, 619
nunated forms, 218
nunation, 225, 248, 251253
Nr ad-Dn, 313, 314 n. 3, 316 n. 5, 364
Nuweiba, 565 n. 3
Nyckees, V., 255 n. 1, 288
Obicini, Thomas, 232, 232 n. 44, 233,
233 n. 48, 234, 238, 242
object(s), 30, 37, 40 n. 10, 56, 6869,
7374, 9192, 109, 214, 217, 380,
392395, 395 n. 85, 476, 484, 484
n. 13, 493, 494 n. 27, 544, 611, 692
accompaniment (maahu), 217
accompanying, 393
clause, 6869, 102
cognate, 393
direct, 179, 217, 230 n. 39, 238, 293,
306, 309, 378 n. 40, 379380,
391392, 682, 697
direct (bihi), 217
from it (minhu), 217
fronted direct, 182
human, 393 n. 80
non-, 393 n. 80
indirect, 293, 380
locative, 217
locative ( fhi), 217
noun
verbal, 293
737
predicate-, 369, 377, 380381,

383384, 389390, 397, 399
preposition-complement, 397399
prepositional, 379, 391
pronoun, 73
reason (lahu), 217
topic-comment, 377, 383384,
389390, 397, 399
objet, 319, 327, 327 n. 18, 335, 337338,
354, 356357
double, 357
externe, 354355
obligatory
constituent, 160
fronting, 169
occlusion, 515 n. 31
official, 404, 414, 423, 427429
language policy, 661
old
Arabic, 639, 640 n. 1, 641,
650651
South Arabian, 597, 599603
Testament, 105, 108, 595 n. 2,
605
Syriac, 108
World, 212
Omayyade(s), 117
one-to-one
equivalent(s), 456, 465
translation(s), 455
onomatope(s), 517 n. 43, 520521
onset, 486 n. 16, 487, 491
operants, 8, 10 11
opration
humaine, 503
operator(s), 151, 160161, 175, 180181,
183, 237
abstract, 178
Oppenheim, Max Freiherr von, 565
n. 3, 578
oppositions
binaires, 515 n. 34, 520
optional constituent, 160
oral
articulation, 656
communication, 665
Order of St. Jerome, 210
ordonner, 511
ordre, 501, 521 n. 49
Orel and Stolbova, 604
Orel, Vladimir, 605
organisation
binaire, 522
des langues
gnrale, 501, 521522
738
index
primitive, 520
primordiale, 522
Orient, 128, 527
Oriental languages, 221, 222 n. 27
orientalising terminology, 211 n. 2
Orientalist(s), 220, 527
orientalistes, 130
orthographic, 595, 596 n. 5, 598 n. 11
sign, 217
orthographical, 540
orthography, 189, 368, 448, 540, 598
n. 11, 684 n. 13
Ottoman, 612
Arabic, 619
ougaritique, 316
Owens, Jonathan, 127, 132, 158, 165,
187, 188, 216217, 228, 228 n. 36,
244, 453, 595 n. 1, 598 n. 11, 607,
621, 639640, 642, 647, 650 n. 12,
652653, 665, 670, 673
Oxford Doniach
English-Arabic, 458459, 466, 468, 473
Oyanguren de Santa Ins, 219 n. 19, 242
palatalization, 611
palatalized variant, 657
Palestine, 113, 197, 201
Palestinian
Ancient South, 72, 72 n. 2
pan-Arab, 404405
Pap, L., 306, 311
Papapavlou, A., 300, 303, 311
paradigm(s), 639, 642, 647648, 650,
650 n. 12
weak final, 642, 646, 651
stability, 651
paradigmatic, 641, 649650, 650 n. 12,
651
paragraph, 367
paraphrases, 455
parser(s), 369370
part/whole relationship, 269
participants, 405, 425, 427, 429431
participe(s), 341, 341 n. 24, 342
actif(s), 341, 341 n. 24, 342, 344,
347348, 348 n. 30, 349350
passif, 341342, 115
participial form, 151, 614
participle(s), 82, 107108, 150156,
156 n. 11, 158 n. 13, 160, 163164,
178179, 181, 183186, 295, 373, 498,
588 n. 80, 602
active, 108, 151152, 152 n. 3,
153154, 154 n. 7, 157158, 167, 646
conditional, 375
coordinating, 375
formation, 492
Greek, 108
Greek attributive, 108
passive, 157, 488, 492, 497, 497 n. 30
present, 482
selective, 375
singular, 155
singular feminine, 155
particle(s), 5, 8, 13, 13 n. 9, 1415, 15
n. 12, 17, 17 n. 13, 1822, 38, 41, 54,
63, 165, 165 n. 24, 167, 170, 172,
177, 210, 214, 227229, 229 n. 37,
230231, 233, 235, 235 n. 49,
237239, 367, 370, 370 n. 12,
374375, 376 n. 32, 378, 384, 388,
397, 579 n. 49, 612, 681682
auxiliary, 292 n. 3, 384 n. 55
conditional, 375
conjunctive
conditional, 388
connective, 388
of comparison, 78
coordinating, 374
cumulative, 374
dual-marking, 182
exceptive, ill, 171
interrogative, 152153, 155156, 170,
179180, 184
m, 238
min, 135 n. 2
negative, 156
relative, 7072, 76, 86, 95, 108
selective, 374
type, 374
verbal, 614
vocative, 30
particularit, 346
particulas charrantes, 230, 237
particule(s), 504505, 505 n. 13, 506
Partikel, 69
ara, 546
pass
compos, 331, 332
simple, 332
passif, 115, 343 n. 25
passive(s), 38, 63, 64 n. 17, 231, 235
n. 51, 281, 282 n. 43, 316, 323,
406407, 419, 441
Pastrana, Juan de, 215
patient, 323, 349, 393, 393 n. 80,
394395, 395 n. 85
pattern, 370
index
Arabic, 444
biradical, 440, 447
heptaradical, 441
hexaradical, 441
morphological, 435437, 439450, 452
nominal, 439
pentaradical, 440
tetraradical, 440, 446
triradical, 439440, 442444, 446
nominal, 439
uniradical, 440
verbal, 436
Paul V, 221
pausal form, 30 n. 2
pause(s) 219 n. 19, 247249, 570
rules 248
syntaxique, 510
Payne Smith, S. and R., 107, 111112
PCCL, 377, 379
Peled, Yishai, 149, 160, 169, 188
Pellat, Charles, 30 n. 3, 123 n. 11
Pellitteri, Antonino, 44
People of the Cave, 190
Perez de Soto, Antonio, 222, 241
perfect, 96, 371, 572, 577, 586 n. 78, 616,
640643, 645646, 680681, 694 n. 27
stems, 645
perfectif, 315 n. 4
perfective, 613
Perfekt
asyndetisch, 699
performative, 28, 41
aspects, 41
Persans, 115, 128
Persian, 193, 661
person, 370374, 378, 380, 382, 389, 395396
verbal affixes, 611
persona(s), 423424, 428429, 431432
hybrid, 431
private, 428, 430, 432
professional, 432
public, 428, 431432
personal, 24, 417, 426432
personne
deuxime, 514515
premire, 509, 515
Peter, William, 616
le Petit Robert, 288
Pfaff, 295
Pharaoh, 105, 190
pharyngeal(s), 256, 256 n. 9, 257, 266
n. 28, 275276, 279 n. 39, 284, 286,
598, 650 n. 12, 658
final, 650
739
philological methods, 219

philology (ilm al-lua), 12
philosophers, 205
philosophie, 502
Phoenician, 597, 599602
phoneme(s), 10, 256, 273, 284, 368, 370
n. 14, 508, 510, 520, 522, 540,
567568, 576, 586 n. 70, 613, 640
n. 1, 657658
Arabic, 368, 371, 371 n. 16
auxiliary, 371
voyelles, 510
consonnes, 510
consonantal, 567
phonemic, 568, 576 n. 38, 598 n. 12, 641 n. 4
phonetic(s), 9, 258, 260261, 273274,
281, 286, 571, 650 n. 12
differences of cognates, 660
evolution, 273
variants, 658
vowel system, 658
phonetical, 63
phonetically-conditioned urban/rural
Colloquial Arabic, 658
phontique, 319
phonological, 602, 608, 611613, 616,
515 n. 34 618619, 641642, 649651
analysis, 656
categories, 655
differences, 656567
features, 653, 655, 656, 668
systems, 657
phonology, 295, 297, 368, 406408, 416,
418420, 442443, 479480, 492496,
497 n. 30, 498, 567, 576, 597, 620,
653654, 656
Arabic, 490
morpho-, 3
phonotactics, 571
phrase(s)
adjective, 376, 376 n. 34, 377,
379380, 382, 388, 397 n. 89, 504,
506, 506 n. 16
indefinite, 379 n. 46
adverb, 376, 376 n. 30, 377379,
384386, 388, 390, 393, 395
adverbial, 485 n. 14
Arabic, 494
complement, 388
conjunctive, 376, 376 n. 30, 384, 388389
inflectional, 386 n. 63
interjectional, 376 n. 30, 384, 387
nominales, 341
noun, 368 n. 4, 369, 370 n. 12, 376,
740
index
376 n. 33, 377381, 381 n. 50, 382,

382 n. 53, 383385, 385 n. 57,
386388, 391395, 397, 477, 485, 494
particle, 376, 376 n. 30, 384
prepositional, 369, 376, 376 n. 30,
377380, 382, 384387, 390395
structure, 517
non-, 517 n. 43
verb, 376380, 386, 388, 393, 397 nn.
86, 88
verbal, 386, 477, 482, 482 n. 7, 483,
484 n. 13, 485, 493, 495, 497 n. 29
phraseology, 548
Piamenta, Moshe, 6768, 112
Pidgins, 128, 607, 621
pidgin-crole, 128
pidginisation-crolisation, 128
Pidginization, 607, 621, 652, 699
pilgrimage, 199
Pisa, 84
place, 371, 373, 378 n. 40, 380, 394397
of pause 249
plaidoirie, 504 n. 7
Plank, Frans, 696 n. 29, 697698
Plato, 547
Platonic, 39
plosives, 657
plural(s), 227, 374, 383, 397 n. 86, 436
affixes, 611
feminine, 224 n. 31
form(s), 153, 660
nominal, 373
patterns, 659
poem(s), 61, 204, 543 n. 1, 543,
545548, 548 n. 5, 552, 554, 556, 559,
560 n. 40, 562
couplets, 548 n. 5
hemistich, 548 n. 5
meter, 548, 559
metrically, 548 n. 5
rhyme, 486 n. 16, 548 n. 5, 550 n. 18,
559
scheme, 548
rhyming, 548 n. 5
syllables, 548 n. 5
verse, 548 n. 5
posie, 123 n. 12, 125126
arabe ancienne, 323
poet, 189, 543 n. 1, 544545, 545 n. 3,
547, 553 n. 27, 559560, 560 n. 38,
561563
Bedouin, 544545, 545 nn. 2, 4, 560, 563
Jordanian, 546
pote(s), 117
modernes, 116
poetic(s), 43, 544545, 547548, 550

n. 18, 559562, 578
genres, 204
poetical, 45, 561
potique, 125, 126 n. 15,
poetry, 3 n. 1, 436, 448, 450, 543,
543 n. 1, 544, 545 nn. 34, 547,
562563
Arabic popular, 543544
Bedouin, 543 n. 1, 544, 560 nn. 38,
40, 563
contemporary, 545
popular, 544, 559
tribal, 561
Jahiliyya, 190
love, 204
popular, 543544
praise, 204
pre-Islamic, 86, 559
political satire, 543, 559
polysmie, 355, 359
polysemy, 255 n. 1, 367 n. 3,
Pope John Paul II, 197
Portius, Simon, 223 n. 30
Portugal, 219 n. 20
Portuguese, 210, 211 n. 2, 291, 294, 298,
304307, 309
American, 306
Poss-ing, 478, 483, 483 n. 10
construction, 476, 483484,
495
postdeterminer, 381, 381 nn. 5051,
382
postmodifier, 381382, 384385
pragmata, 33 n. 6
pragmatic, 2728, 29 n. 1, 33, 33 n. 6,
34, 42
analysis, 36
approach, 33
Gricean, 30
modern, 41
Pragmatist, 40
prceptes, 501
predeterminer, 381382
prdicat(s), 339, 341
verbaux, 339
predicate(s), 7, 2829, 29 n. 1, 68, 74,
135137, 141, 145, 149150, 155, 158
n. 13, 159, 162163, 163 n. 21,
165166, 168, 171173, 173 n. 31,
216, 230, 368 n 4, 376, 378, 389, 397
n. 86
adjectival, 155156
adverbial, 150
clauses, 68
index
indispensable, 138, 140141, 145146
non-verbal, 152
prepositional, 150
pre-subject verbal, 156
subject, 76, 369, 377, 380381,
383384, 389390, 397, 399
order, 162, 172
predicatival
constituent(s), 152, 167,179180
non-verbal, 152
prepositional phrase, 168
relationship, 149, 159160, 162,
165166, 180
sentences, 170 n. 29
predicative constituent, 149, 177,
179181, 184
adverbial/prepositional, 185
prefix(es), 15, 224 n. 31, 262 n. 25, 262,
264, 275, 277, 292, 407, 409, 412, 436,
486, 488, 488 n. 20, 490, 490 n. 22, 492,
492 n. 24, 494, 597, 604, 612, 640
detransitivizing, 488 n. 19
incrementation, 271
prefixation, 17 n. 13
prfixe, 509
prfix, 315 n. 4, 316, 317319, 338
prsent, 331332, 342
prefixed, 17 n. 13
preformative(s), 262, 597
prejunctural, 249
premodifier, 378379, 384385, 387389
prepausal mode 247
prepause 249
preposicion, 215, 230
preposition(s), 7, 63, 79, 89, 93, 97,
101102, 163, 165, 214218, 229230,
374375, 385386, 389, 391, 484, 497
n. 28, 541, 576577, 579 n. 47, 581
n. 54, 598, 598 n. 10, 689690
bi-, 7, 102
dummy, 476
Latin, 216
prepositional, 379 n. 48, 380, 384, 389
n. 68, 391
conjunctive, 389
linker, 385386
phrase, 5455, 152, 153 n. 5, 159160,
162163, 163 n. 20, 164, 164 n. 23,
167, 171172, 177, 179, 184
definite, 150
present tense, 224 n. 31
presentations, 397
prestige, 404
prestigious variety of Arabic, 664
presuppositions, 27
741
pre-vocalic position, 647

Prince, Alan, 421, n. 15, 476, 486, 486
n. 17, 488, 490, 495, 497 n. 29, 499
Priscian, 209
private, 403, 405, 423, 427430
procs, 321324, 324 n. 13, 325326,
328 n. 19, 329330, 332, 335339,
343, 347349, 351 n. 32, 354357
accompli
antrieurement, 331
processing, 369, 377 n. 36
automated language, 367, 367 n. 2,
370 n. 14, 396, 398
processus, 321, 324, 324 n. 13, 325326,
326 n. 17, 327, 329331, 333, 340,
342343, 347, 349, 351352, 357, 359,
361362
accompli, 332
cognitif, 349
professional, 426432
progressif, 341, 341 n. 24, 347
projection
nominal, 485
verbal, 485
pronominal, 7374, 97, 571 n. 24, 573
concatenation, 7374
element, 154155
system, 659
prononciation pausale, 126, 127 n. 18
pronoun(s), 15 n. 12, 73, 88, 91, 106,
151, 154 nn. 7-8, 164, 164 n. 23,
173, 176, 180, 182, 578 n. 41, 613,
615616, 687 n. 16
affix, 612613, 616
affixed, 615
anticipatory, 171, 176
conjunct, 217218
free and bound, 659
implicit, 151
indefinite, 373, 381383
interrogative, 373, 381383
negated, 576
object, 615
suffix, 611
personal, 89, 154, 154 n. 7, 164, 373,
381383, 612613, 617
possessive, 613, 615
prefixed, 379
relative, 18, 67, 71, 78, 87 n. 24, 91,
94 n. 27, 9899, 101, 103104,
107108, 152, 576, 373
resumptive, 73, 87 n. 24
Spanish, 218
suffixed, 239, 379, 381 n. 50
Syriac relative, 108
742
index
pronunciation(s), 249, 253, 409410,

410, n. 8, 411413, 419
proper names 250
properties
phonological, 479
semantic, 479
syntactic, 479, 492
prophet(s), 26, 37, 190192, 195, 198,
200, 204206
Muhammad, 189, 191, 193, 195, 198,
200, 202, 204
prophetic language(s), 195, 204
prose contemporaine, 343
prosody, 486, 490
protasis, 377, 486 n. 16
ProtoEthio-Semitic, 600
Hebrew, 596 n. 6, 598
Semitic, 595, 596 nn. 56, 597599,
601, 604
proverb(s), 3 n. 1, 375, 387, 578, 692
proximity (itb), 48, 52
theory, 56
public, 403, 405, 415, 423424, 427432
purposes (ard), 9
Q. (Qurn), 155
Qabbwa, 318, 364
qabh , 29
qad, 49, 50, 52, 57
qadara, 660
qa/iyye, 657
qfiya
muqayyada, 126
mutlaqa, 126, 126 n. 15
Qaht n, 202, 204
al-Qahtani, 661, 673
qalla, 19
qallam, 17, 19
Qms, 257
al-Qms al-Muht , 288
qnn, 42
qara, 660
qaraa, 660
qat, 49 n. 5, 50, 50 n. 7
Qatar, 532
qawl, 47, 5859
qawmiyye, 657
al-Qayrawn, 84
Qays, 201
q'ltu and g'l't dialects, 658
qillat al-ijtim, 52
qira, 61
qirt, 46, 48, 51, 652
qiri, 660
qisas al-anbiy, 192
qiys lafz, 11
qualit, 320, 327 n. 18, 328, 345346
constante, 337
constitutif, 338
durable, 327328, 337
inhrente
naturelle, 320, 338
permanente, 327328
temporaire, 327
transitoire, 327
quantifiers, 373
quantitatives, 397
quantity, 398
Quechua, 484 n. 13
Queiroz, A., 304305, 311
Quiniou, Y., 274 n. 35, 288
Qumranic Hebrew, 600
Qurn, 2526, 29, 3637, 4549, 5154,
57, 61, 64, 64 n. 17, 77, 79, 8687, 92,
105106, 109, 189192, 196, 205207,
247-248, 611, 657
ajam, 190
Arabic, 198, 205
exegesis of, 192193
language of, 189190
Q 3/180, 36
Q 67:1, 105
Q 7:54, 105
grammatical commentary, 45
vers, 193, 198
Qurnic, 4547, 4950, 5253, 5556,
60, 8788, 105106, 108, 194
excerpts, 45
exegesis, 6263
linguistic outlook, 191
message, 200
readings, 47
revelation, 190, 205206
sciences, 49 n. 5
text, 46, 58, 63, 155
traditions, extra-, 191
verses, 19 n. 17
Quray, 120, 125, 202, 206
Quayr, 117
Qutrub, 126127
Rabbi David (i.e. Qimhi), 209
Rabin, Chaim, 120 n. 4, 133, 640 n. 1, 652
racine(s), 317318, 345, 507508, 508
nn. 2021, 509, 509 nn. 2223, 510,
510 n. 24, 511516, 520, 521 n. 49, 522
consonnes, 513, 516
index
de consonnes, 508 n. 20, 509 n. 22
de syllabes, 508 n. 20, 509 nn. 2223
de voyelles, 508 n. 20
monoconsonantiques, 515, 515 n. 32,
516
primitives, 507
proto-, 515516
triconsonantique, 516
radical(s), 253, 256258, 262, 269270,
274, 277, 277 n. 38, 282, 285, 287,
371372, 435, 440441, 443 n. 8, 444,
446, 450
augmented, 436, 440442
base, 435
basic, 436, 441, 447
final 252
middle, 445446
non-ambiguous, 258 nn. 1617,
primary, 441
second, 446
original, 441
primary, 440
three-, 435
Turkic, 441
radicale(s), 314315, 317, 353 n. 34, 511
n. 26
consonne(s), 313, 318319, 338, 344,
345 n. 28, 353
deuxime, 317319, 345, 345 n. 28
premire, 318
troisime, 318
radio broadcast 251
Syrian 248
raf , 29, 33, 46, 48, 52, 56, 5860, 149,
152154, 156160, 160 n. 15, 161, 162
n. 18, 167, 170 n. 27, 171, 175176,
180181, 183185, 225, 227228, 230,
234, 238
case, 161, 181, 183, 186
ending, 46, 46 n. 3, 49 n. 6, 50, 58,
158
rfi, 55, 5859, 161, 167, 175
Rib, Ysuf, 93, 111
rajaz, 204
Ramadan, 609, 616, 619
ar-Ramlah, 566, 566 n. 7, 578
range, 393, 395
Rappaport Hovav, Malka, 391 n. 73, 401
Rs adr, 566
Rs Sadr, 566 n. 7, 575, 575 n. 37
Rashd al-Kln, 548, 556
rasm, 125
rasl, 191
rteb, 667
743
reading
oral 247
style, 411412, 414, 425
reasons (ilal), 9
recitation 247249
Reckendorf, Hermann, 77, 112
rection, 1011, 1517, 19
Red Sea, 561
Reese, 371 n. 17
reflexive, 271, 277, 441
rflexivit, 313
regular
verb patterns, 660
verbal sentence, 166
rgularit(s), 356, 507, 511
smantique(s), 313, 356
regularities, 368, 389
relation(s)
biunivoque, 514, 517, 517 n. 44,
518520, 520 n. 47, 521 n. 49
causal, 276
univoques, 517520
relationships, 368369, 376, 381382
relative, 2, 6, 13, 26, 3334, 36, 3839,
4243
clause, 6869, 7172, 7677, 87,
87 n. 24, 8889, 9193, 9798,
103105, 107109
asyndetic, 151152
causal, 104
coreferential, 98
non-restrictive, 98
Relativpronomen, 85, 87n. 24
Relativsatz, 69, 75, 106
syndetischer, 75
Relativsatzgefge, 70, 74
religious
arengas (xutbas), 89
language, 104
orders, 221
scholars, 43
Renaissance, 209, 213
Rendsbourg, G.A., 595 nn. 12, 600601,
602 n. 16
reported speech, 676, 676 n. 3, 678, 683,
685, 687689
representation
morphophonological, 479480
morphosyntactic, 479480
phonological, 479480
semantic, 479
syntactic, 479
res, 505 n. 11, 512513, 517, 519, 519
n. 46, 520, 522
744
index
anime, 514
banales, 514
communes, 513
dostension, 515
de reprsentation, 515
gnrales, 515
montres, 515
reprsentes, 515
vagues, 515
restriction, 398
rsultat, 330331
rsultatif, 329
resumptive, 8, 87 n. 24
retrieval
information, 369 n. 8, 389 n. 69
Reuchlin, Johannes, 209, 218, 231 n. 14,
242
Reuland, Eric, 483 n. 10, 499
de Reuse W., 308, 310
revelation(s), 190194, 198, 204205
pre-Arabic, 198
revelationist
theory, 193
tradition, 194
Rhema-Thema, 7576
rhetoricians, 46
rhyme, 486 n. 1, 548 n. 5
scheme(s), 436, 450
Ridruejo, Emilio, 210 n. 2
rime
absolue, 126
interne, 127
lie, 126
Ritchie, C., 671, 673
Ritchie and Bhatia, 653
Ritter, Elizabeth, 493, 499
Rizzi y Franceschi, Mariano, 229 n. 37
Ritter, Helmut, 23
Rmlt, 569 n. 17
Rodrigues, J., 304, 311
Rodrigues, Joo, 210, 242
Roeder, Larry, 565 n. 3
Roldn, A., 243
Roman(s), 121 n. 8, 228 n. 35, 314, 316,
316 n. 5, 321 n. 11, 341 n. 24, 347,
349 n. 31, 353, 364, 501, 505 n. 13,
508 n. 21, 509 n. 22, 523
Romance, 212 n. 7
romanes, langues, 119
Romani, 294
Rome, 223 n. 28, 232, 232 n. 44, 233,
235, 238, 240, 243
Romero, Juan Antonio, 223 n. 30
root(s), 8, 9 n. 7, 13, 56, 203, 216, 231
n. 41, 239, 286, 370, 370 n. 14, 371,
407, 435436, 446, 448, 450, 480,

486489, 491, 496, 498, 568, 580 nn.
50, 53, 581 n. 54, 586 n. 77, 596 nn.
67, 597, 598 n. 10, 599 nn. 13,
600601, 604, 604 n. 17; 650 n. 12
bilateral, 486 n. 15
consonant(s), 489, 489 n. 21, 491
consonantal, 370 n. 14, 395
nominal, 438
mono-, 596
quadri-, 287
tri-, 286, 287
paradigmatical, 435
quadriliteral, 486 n. 15, 497 n. 29
strong, 251
triradical, 442, 488
verbal, 438, 494
Rosenbaum, Gabriel M., 662, 665666,
673
Rosenhouse, Judith, 595 n. 1, 596 n. 6,
653654, 655 n. 2, 656658, 660663,
668, 672673
Rosenthal, Franz, 544, 563
Roth, Heinrich, 209, 242
Roth-Laly, Arlette, 648, 652
rubba, 19
rubbam, 17
Rubin, Aaron D., 70, 112, 595 n. 1, 601
Rubin, Uri, 208
rules
agreement, 367 n. 1
concord, 367 n. 1
ar-Rummn, 3435, 43, 504 n. 8
rural, 657
dialects, 657
Russian, 294, 546, 561
Rwala, 563, 690, 698
ar-Rynih, 566 n. 7
sa, 658
Saad, George Nehmeh, 369, 371 n. 15,
381, 393394, 394 n. 81, 396 n. 85,
399, 402
Saba, 199, 202
sabab, 114
sabat ah ruf, 64 n. 17
Sacy, Sylvestre de, 314, 364
Sad ad-Dn Wahba, 699700
sadaqa, 31
Sadat, Anwar, 544
Saddam Hussein, 561
d arf al-Jazr, 528
f, 667
Saf Abd al-Munim, 699
as-Safad, 200 n. 30
index
a-fi, 26, 205, 205 n. 41
al-Saffr, 369 n. 9, 402
Saguer, Abderrahim, 255, 255 n. 2, 257
n. 14, 258, 262, 262 n. 24, 288289, 400
as-Shib, 125, 131, 505 n. 12, 507 nn.
1819, 513 n. 28
Sahh, 78, 90
Saint-Barthlmy, 129
as-Sakkk, 28
akl, 217
sakrn, 659
sakrne, 659
Salamanca, 221
Slih, Subh, 195, 206 n. 49
salqiyya, 9 n. 7
Saloniki, 82
Sm, 196
Samnah
dialect(s), 567, 572 n. 25
Smih Farag, 699
San Diego, 222
San Pietro di Montorio, 211 n. 2
sand, 658
sane, sini, 658
Sanskrit
grammars of, 209
Santino, J., 618, 637
So Paulo, 219 n. 20,
al-Sq, FJil, 370 n. 13, 402
Sarbt al-Xdim, 565 n. 3, 566 n. 6, 578
Sarah, 197
arf al-Alf, 528
art (cause), 54
a-artn, Rad, 314, 318, 364
arw, 21
satirical, 528
Saudi Arabia, 545, 561562, 661, 673
Saudi(s), 545, 561563
de Saussure, Ferdinand, 502, 523
Sawlhah, 567 n. 9
awq D ayf, 22
Saxon genitive, 476
ay wh id, 16
a-aybn, 40, 43
Schabert, Peter, 100, 112
Schatz, H., 308, 311
Schepper, Hugo de, 243
Schmidt, Richard, 418 n. 13, 420, 434
Schnurrer, 210 n. 2, 212 nn. 35, 220,
220 nn. 2122, 222223
scholars of religious Law, 205
Schregle, Gtz, 473
Schuler, Bernard, 504 n. 7, 523
Schulz, David, 408 n. 7, 434
schwa(s), 219 n. 19, 296
745
Schwenter, Scott, 681 n. 9, 698

science, 661
Seaman, P.D., 302304, 307, 311
Searle and Austin, 28
Seattle, 595 n. 1
Second World War, 544
Segeral, 256, 289
Sellheim, Rudolf, 239
Selkirk, Elizabeth, 486 n. 16, 499
semantic(s), 3, 8, 1415, 22, 4849,
7071, 76, 86, 87 n. 24, 91, 9698,
101, 103104, 255 nn. 1, 5, 258, 261,
262 n. 25, 263265, 269270,
272275, 277278, 281, 284287,
367368, 369 nn. 78, 370, 370 n. 12,
371, 387, 389391, 393, 397399, 421,
479, 496, 656, 684687, 694695
analysis, 48, 49 n. 5
aspects, 3, 22
content, 684, 694 n. 27, 695696
differences, 656, 666
elements, 5
fields, 70, 661, 665, 667, 670
finite, 370, 396
generalization, 93
level, 8
static, 389, 389 n. 69, 390
status of commands, 33
semantically, 15, 54, 70, 96, 103, 103 n. 31
deficient verb, 178
smantique, 314, 316, 319
arabe, 324
semicolon, 219 n. 19
semiconsonants, 448
Semitic, 197, 201, 595, 597598, 598
n. 10, 599, 599 n. 13, 604605
ancient, 597
East, 598
languages, 108
Proto-, 595, 596 nn. 56, 597599,
599 n. 13, 600601, 604, 657, 672
Proto-Ethio-, 600
West, 108
smitiques, langues, 125
Semitist, 595
seal, 217, 217 n. 15, 226
de demostracin, 217 n. 15, 226 n. 34
sens, 313, 315, 319, 322 n. 12, 324, 326,
328, 330, 333, 339344, 349, 352,
354358, 360362
abstract, 508
grammatical, 324, 358, 360362
grammaticaux, 359, 362
lexical, 330, 334, 340
lexicaux, 362
746
index
sense(s), 258259, 260 n. 21, 261,

263265, 267270, 272, 274279,
281282 n. 42, 283286
abstract, 275
contradictory, 272
homonymic, 263, 266, 270
static, 264
sentence, 367
-initial position, 154, 156, 161, 170
n. 29, 172, 174, 185
nominal, 368 n. 4, 376, 376 n. 31, 377,
384, 388389, 393, 397, 399
medieval theory of, 150, 156
meaning, 28, 38
type(s), 149150, 156158, 158 n. 13,
162, 164168, 173 n. 31, 176177,
179181, 183185, 185 n. 38, 376,
398399
Arabic, 157, 185
verbal, 368 n. 4, 376, 376 n. 31,
377378, 384, 387389, 393, 397,
399
separator, 249
Serabit el-Khadim, 565 n. 3
Serhane, R., 257 n. 14, 288, 289
serial verb construction(s), 675, 692, 694
n. 26, 698
SVC, 692696
serial verbs, 692, 698699
serialization, 675
Seuren, Pieter, 696, 698
Seville, 211 n. 2, 221222
x Ima, 680
Sezgin, Fuat, 10, 23
Shawqi Daif, 404, 404, n. 2, 405
Sheba, 603
Shibl, 198 n. 22
Shiites, 193, 206
Shth, 195
Shraybom-Shivtiel, 655, 661, 668, 673
Shuayb, 195
Sbawayhi, 3, 3 n. 1, 4, 4 n. 2, 56, 6
n. 3, 7, 7 n. 5, 8, 10, 1215, 15 nn.
1112, 1718, 18 n. 14, 19, 19 n. 15,
20, 2123, 2528, 28 n. 1, 29, 29 n. 1,
30, 30 n. 3, 3133, 33 nn. 45, 3443,
45, 52, 62, 62 n. 15, 64, 64 n. 17, 129,
135, 135 n. 2, 136138, 138 n. 19,
139, 139 n. 25, 140141, 141 n. 3233,
142, 142 n. 39, 143, 143 n. 43, 144
nn. 4951, 145147, 149, 151152,
152 n. 3, 153, 156, 156 n. 11, 158159,
159 n. 14, 160162, 162 n. 18, 167,
175, 178, 180, 182, 184, 187, 216217,
227, 232, 238240, 316, 318, 320 n. 9,

329, 336 n. 22, 338, 338 n. 23, 345
n. 27, 353, 353 n. 34, 362, 368 n. 6,
370 n. 13, 378 n. 40, 402, 504, 504
n. 8, 505, 505 n. 12, 523
sibilant(s), 262, 407, 407, n. 5, 409, 413,
418419
pronunciation, 409410, 410, n. 8,
420
sifa, 54, 140, 151, 157, 169, 169 n. 26,
173 n. 30, 174
muabbaha, 333, 341342
signifiant(s), 507508, 508 n. 21, 510,
513, 520, 521 n. 49, 522
de cas, 510
de lanimit, 513
de mode, 509
du temps, 513
signifi(s), 509510, 520, 521 n. 49, 522
sila, 59
Silesia, Dominicus Germanus of/de, 211
n. 2, 221, 223, 241
Simon, Udo, 28, 44
sina, 3, 4 n. 1
Sinai, 543 n. 1, 545, 545 n. 4, 560 n. 38,
563, 565, 566 n. 7, 9, 567 n. 9, 568,
572574, 574 n. 32, 575, 575 n. 34,
577578, 586 n. 70, 587, 587 n. 79
bedouin dialect(s), 568, 571
Sinaitic, 567568
Singer, Hans-Rudolf, 100, 100 n. 29, 112
single
phrase, 164
word, 18
singular, 157, 227, 380, 383, 397 n. 86,
659 n. 4, 660
feminine, 224 n. 31
form, 153
as-Srf, 44, 117, 136 n. 8, 143 n. 43,
144 n. 48, 145, 147, 164 n. 23, 336
n. 22
a-irbn, 37, 44
lwh, 563
slots, 377380, 382, 385389
Smith-Stark, Thomas, 219, 244
Sobelman and Harrell, 596
Sobelman, Harvey, 605
social differentiation, 408
sociality, 398
Socin, Albert, 544, 563
sociolinguistics, 291, 297, 307308, 652, 648
sociolinguistique(s), 121, 123124
amricaine, 129
Soden, Wolfram von, 605
index
Somekh, Sasson, 665, 673
sonorant(s), 256 n. 6, 257, 257 n. 12,
269, 282
non-, 257 n. 12
non-voiced, 256 n. 7
voiced, 257 n. 12
Soqotri, 604 n. 17
sorrow, 397
Souchne, 690
Soudan, Sud, 128
Soukhne, 697
sound-related interjections, 696697
source, 380, 392395
language, 455457, 466467
sourdit, 515 n. 31
southern Sinai, 565 n. 2, 566567, 567
n. 12, 573, 573 n. 29, 575 n. 35
dialects, 567, 568 n. 14
southern Sinaitic dialects, 568
Southern Sudan, 609610
Sowayan, Saad Abdallah, 544, 563
Spain, 210, 211 n. 2, 220, 223 n. 28, 639
Spanish, 210, 210 n. 2, 212, 218,
222223, 223 n. 30, 225, 231, 294,
296, 661
tradition, 212
Sparvenfeldius, Johannes Gabriel, 220,
220 n. 23
speakers awareness, 5, 8, 12, 17, 20, 22
speech(es)
acts, 27
community(ies) 404
Arabic, 403
context(s), 403405, 410, 410,
n. 8, 413, 417, 423, 432433
event, 249250
sounds, 657
spontaneous, 427
style(s), 406, 410, 414415, 419420,
422, 426427, 427 n. 17, 429,
431432
religious, 251
Spitaler, Anton, 21, 2729, 3235, 47,
6872, 7475, 78, 78 n. 3, 86, 9294,
97100, 105 n. 33, 112, 596 n. 6, 605
Sprenger, Alos, 125
stabilit
formelle, 317
standard, 404
variety, 403
state construct, 375
statif, 328
status (manzila), 7
stem(s), 371, 395, 436, 441442, 442
747
n. 6, 444, 481, 486, 486 n. 17, 487,

496, 640641, 641 n. 4, 643, 643 n. 9,
644647, 649, 650 n. 12
I, 488, 492
II, 490492, 497
III, 490492, 492 n. 25
V, 491492, 497
VI, 492, 497, 497 n. 29
IV, 489
V, 488, 490492, 492 n. 24, 497 n. 29
VI, 488, 490492, 492 n. 24, 497 n. 29
VII, 488 n. 19, 496
VIII, 488 nn. 1920
X, 488
Arabic, 489
base, 372373, 488, 491
causative, 262
denominative, 262
derived, 372373
extension, 296297, 299
final weak, 647
laryngeal, 646
perfect, 645
perfective, 486487
quadriliteral, 497 n. 29
reflexive, 262
verb, 478, 488
Stephen, 610, 618620, 634, 638
Stetkevych, Jaroslav, 654, 661, 673
Stewart, Frank Henderson, 578, 580
n. 53, 584 n. 67, 586 n. 70, 591 n. 89,
592 n. 90
Stockholm, 222
Stoic, 33, 33 n. 6
Stolbova, Olga V., 605
stop(s) 407, 409, 418419
glottal, 407
uvular, 407
Strngns, 222
strategy
conversational, 417
stylistic, 417
stress, 18 n. 14, 21, 569572, 572 n. 25,
576, 591 n. 87, 617
on the penultimate syllable, 219 n. 19
stressed, 540, 569571, 573574
strike a blow to, 255257, 259260,
263266, 270, 281282, 285286
structuration
binaire, 520521
structure(s)
complement, 369, 369 n. 7, 379380,
393 n. 79, 395396, 398399
verbal, 369, 378380, 393 n. 79
748
index
exclamative(s), 117, 124

interrogative(s), 117, 124
morphological, 475, 495
phonological, 480 n. 5, 481, 483, 488
radicale, 519
syntactic, 480 n. 5, 481483, 492,
494495, 497498
syntaxiques, 506 n. 15
style(s)
casual, 405, 410411, 427
conversational, 418, 427
differentiation, 429
formal, 410411
shift, 415, 417
stylistics (bala), 1112, 368
subject, 18, 2829, 29 n. 1, 29, 38, 68,
7374, 89, 9192, 9698, 102, 104,
136138, 145146, 149, 152154, 154
n. 6, 155156, 158, 158 n. 13,
159160, 162, 165168, 171, 173
n. 31, 175, 181, 184, 210, 214, 230,
230 n. 39, 238, 397 n. 86, 476, 484,
544, 559, 694, 694 n. 26
clause, 68, 102, 154 n. 7
definite, 150, 168
explicit, 394
human, 96
implicit, 394 n. 84
indefinite, 150
nominal, 55, 58
position, pre-, 155
predicate relation, 68
Subjekt, 7375
Subjektsstze, 10
subjonctif, 124
subjunctive, 10, 16, 23, 228, 230
n. 39
subordinate clause(s), 67, 9, 87 n. 24,
89, 9192, 104, 109, 687
causal, 91
substantival clauses, 74
substantive clause, 7172
subordination, 388, 517518, 520
subordinators, 388
subordonnants, 519
succession dtats, 326, 331
Sud Soudan, 621
Sudan, 619, 640641
Arabic, 685 n. 14
southern, 639
Sudanese, 614
Northern, 608609, 611, 620
South, 607
Southern, 608609
Southerners, 608
Sudanic, 640
Arabic, 640, 648
WSA, 649650, 650 n. 12, 651, 651 n. 13
suffix(es), 15, 292, 292 n. 2, 293294,
296, 298, 436437, 441442, 444445,
450, 478, 480481, 481 n. 6, 482483,
486, 494, 494 n. 27, 568 n. 14, 571
n. 24, 572573, 581 n. 56, 592
n. 91, 611, 640 n. 2, 643 n. 9,
645646, 648649, 650 n. 12
agentive
-at, 644, 646
C-initial, 644
C-initial subject, 644
diminutive, 482 n. 8
English, 480481
final V-initial, 643
finite, 488
first person, 541
future, b-, 224 n. 31
inflectional, 249, 294
infinitive, 296
-kiy, 568 n. 14
nominalizing, 484 n. 13
non-finite, 482, 488, 490 n. 22,
491492, 495, 497
non-finiteness, 489, 496
object, 645646
paradigmal, 444
past tense, 442
plural, 645
pronoun, 576
-, 688, 690
subject, 643 n. 8
person, 643
syntactic, 482
template, 488
verbal, 298
V-initial, 644645
suffixation, 19
suffixe(s), 509, 612
C-initial, 644645, 648
C-initial subject, 647
object, 646, 646 n. 10
relateur, 126, 128 n. 18
subject, 643, 646
V-initial, 644, 647649
V-initial subject, 647
verbal, 612, 640, 643
suffix, 315, 315 n. 4, 316317, 331332,
334
suffixed, 19, 568, 572573, 575, 615, 646
n. 10
object bound pronouns, 659
preposition(s), 576577, 579 nn. 4647
index
sujet, 123 n. 12, 316, 321 n. 11, 329,
335337, 347, 521 n. 49
extrieure, 336
grammatical, 321323, 329, 338, 351
n. 32, 354357
humain, 337, 345346, 351, 351
n. 32, 352
non-humain, 346, 351352, 356
intrieur, 335336, 356
sukn, 218, 230, 253, 371
Suleiman, Yasir, 9 n. 7, 23, 661, 673
Sumerian, 197
summarizing
text-, 369 n. 8
sundq, 658, 664
sunletters, 574
sunna, 42
support, 398
Sura, 190191, 193194, 198
srat Ysn, 195
surprise, 397
Suryna, 194
suryniyya, 194
as-Suyt , 10 n. 8, 17, 17 n. 13, 23,
120 n. 5, 131, 192, 194 n. 89,
195 n. 10, 198 n. 24, 199 n. 26,
200 nn. 2930, 203 n. 37, 205 n. 42,
208
Swadesh, Morris, 667, 673
Swanson, D., 298, 311
Sweden, 222
Sheri, 604 n. 17
syllabe(s), 508, 509 n. 23, 510, 511 n. 26,
520, 522, 573
open, 57, 570, 570 n. 21, 571, 574,
598
closed, 570 nn. 2223, 574
syllable(s), 18 n. 14, 219 n. 19, 548 n. 5
closed, 598
CVC, 490
CVV, 490
extrametrical, 489, 491
final, 617
penultimate, 645
synonyms, 38
syntactic, 9, 16, 33, 50, 62 n. 15, 68, 73,
77, 79, 9093, 96, 101104, 135, 141,
144 n. 44, 611, 675, 677, 687, 692, 694
n. 27, 695696
analysis of sentences, 174
constructions, 135, 141, 144
context, 86, 91
feature, 92
non-human, 96
object, 92
749
positions, 214
relations, 12
shift, 91
structure, 50, 62 n. 15
subject, 96
syntactical, 8, 1114, 19 n. 17, 30,
42, 72, 76, 9394, 106107, 676,
695
analysis, 11, 13
constructions, 79
ramifications, 18
rules, 12
structure, 687
syntactico-semantic, 3
syntaktisch, 69, 75
syntax, 12, 17, 21 28 n. 1, 29, 33, 35,
62 n. 15, 72, 104, 212, 229, 232, 235,
367369, 369 n. 7, 370, 371, 373376,
378, 383384, 387391, 393, 396, 398,
406407, 412, 416, 418420, 431, 475,
477, 479, 492, 495496, 497 n. 30,
527, 654
morpho-, 3
syntaxe, 120, 133, 316, 510
de position, 120
syntaxique, 126127
Syria, 532, 546, 557 n. 35
Syriac, 4144 105, 105 n. 32, 106109,
193196, 201203, 600601
doxology(ies), 107108
New Testament, 107
Nhs, 192
perfect, 103
relative, 43
script, 200
nomads, 690
Syrian, 220
Syrie, 113
Syrien, 697
syrisch, 105
Syro-Lebanese
dialects, 667
region, 667
varieties, 667
system(s)
generative, 479
natural language, 368
nominal, 487
verbal, 487
systme(s)
de communication, 510511, 518520,
520 n. 47
de nomination, 508 n. 21, 510512,
514, 516, 519, 520 n. 47
750
matriels, 506 n. 14
syllabique, 509 n. 23, 510, 513
universel, 518
Systemzwang, 569, 596 n. 6
t marbta, 128 n. 18, 248250, 571,
658659
t mabsta, 250
taajjub, 117
at -_alib, 199
at-Tabar, 45, 5357, 6364, 193, 196,
196 n. 14
taba, 39
tches, 510511
smantiques, 505
Tckholm, Vivi, 578
tadiya, 158, 178 n. 37, 185
tadmn, 116
tafsr, 37, 51, 156, 193
tah qq, 507 n. 18
tah rf, 194195
Taine-Cheich, Catherine, 653, 670, 673
tajwd, 126
talabtak, 659
talabtuka, 659
tlam, 19 n. 16
tall, 9 n. 7
Talmon, Rafael, 62 n. 15, 65, 152 n. 4,
159, 161 n. 17, 173 n. 30, 175 n. 33,
188, 444, 453
_amnn, 195
Tamil language, 210
Tamm, 120, 126 n. 15
Tamm, 9, 16
Tamis, A., 302, 311
tamyz, 138, 138 n. 21, 140, 146, 217
_amd, 202203
tanwn, 126 n. 15, 126127, 217, 225,
509, 515 n. 33
taqdm, 120, 149, 161, 168169
wa-taxr, 149, 161
wa-tarx, 120
taqdr, 56, 6 n. 3, 7, 7 n. 5, 8, 11,
142143, 143 n. 41, 146147, 151,
156, 164
_aqf, 199
Tarbn, 566 n. 9, 575, 575 n. 37
dialects, 567, 569 n. 17
target, 393, 395
language, 456, 466
Trx, 186 n. 14, 198 n. 21, 199 n. 22
Tasm, 202, 203
Tassa, 277 n. 38
index
tat niya, 4
tawdu, 39, 192
tawahhum, 8 n. 6
Tawfiq al-Hakim, 666
tawld, 115
tawqf, 39, 192, 192 n. 3
Tayhah, 566 n. 9
dialects, 567
tayn, 170
at-Tayyib, Muhammad Sulaymn, 565 n.
3, 566 nn. 4, 7, 577
technical and scientific vocabulary, 656
Teixidor, J., 287 n. 50, 289
television
broadcast 251
news 248
Tell Amarna, 316
tem, 217 n. 14
temiz, 210, 217
template(s), 486, 486 n. 17, 490491
CV-, 486
prosodic, 486487, 494
stem, 488490, 497 n. 29
stem I, 488
stem II, 488, 491, 497
stem V, 497
verbal, 490
temple of Hathor, 566 n. 6
temporal markers, 619
temps, 504507, 511, 512 n. 27, 513,
517 n. 41, 517, 517 n. 41, 519520,
522
gnral, 515
tenifa, 210
tenivofa, 210
tense, 248, 251252, 294, 370371, 371
n. 17, 372, 384 n. 55, 389, 395
mood, 384
past, 437
present, 294
Terabin, 567 n. 13
terminals
non-, 367
terminological, 62
terminology, 47 n. 4, 62
Texas, 560, 560 n. 41
text data, 376, 390 n. 70
Arabic, 367, 395, 398399
textes arabes
mdivaux, 315
texts
religious, 411
textuality, 425, 432
index
Theodory, Constantin, 473
theologians, 205
thorie
du levier, 501
Theory of Matrices and Etymons, 255,
272
Thom, 506 n. 15
Thomason, S.G., 294, 311, 665, 673
Thomason and Kaufman, 665
Tibriade, 113
tier
autosegmental, 488, 498
prosodic, 494495
segmental, 488489, 495496
syllabic, 489
tightening, 256, 284, 286
Tigre, 128
Th, 566
Tihama, 650
time, 373, 380, 394396, 398, 522
Tingstadius, Johan Adam, 210 n. 2, 222,
222 n. 27
TMA, 611, 614616
token, 29
Toledano, Moses, 81, 111
Toledo, 82
tongue, 256 n. 10, 257, 257 n. 13,
269270, 275278, 280283, 286, 371
topic(s), 368 n. 4, 376, 376 n. 33, 377,
377 nn. 3738, 405, n. 3, 414417,
421, 423429, 431433
agent, 376 n. 31
personal, 414
shift , 417, 427
Torre, Patricio Jos de, 220
Tosco, Mauro, 607, 621
Tottoli, Roberto, 208
Toulouse, 221
traction, 257, 260261
tradition(s), 7778, 87, 90, 9293
arabe, 114, 118, 507, 509
Arab-Islamic 247
grammatical
Arabic, 373, 378, 378 n. 40, 380
grammaticale
arabe, 503 n. 6, 504 n. 10, 506507,
507 n. 19, 513, 519 n. 45
non-, 509
orientaliste, 507
sunan, 205
traditional Literary Arabic
articulation, 657
phonological system, 659
751
traditionists, 199
trait commun, 503
transitif(s), 115, 337, 343
transitive, 247
transitivit, 313314, 314 n. 3, 319320,
336, 336 n. 22, 343, 344, 349350,
521 n. 49
directe, 320, 337
indirecte, 320, 337
transitivity, 379, 379 n. 44, 380382, 391,
435, 441
translation(s), 456458, 465467,
469470, 473
Arabic, 456459, 462, 471
English, 457, 467468
equivalent(s), 457458, 466, 471472
of foreign words, 661
French, 457
German, 457
profile(s), 467470
unidirectional, 456
tree
conceptual, 396
lexicogenetic, 269270, 276, 278,
280281, 286
semantic, 396
structures, 477
syntactic, 494
Treffers-Daller, 299 n. 9, 311
tribal, 544545, 555, 558559, 561563
community, 201
history, 201, 204
tribales, 129
tribe(s), 189, 195, 199200, 559,
561, 563, 565, 566 nn. 34, 569,
571, 586 n. 71, 589 n. 82, 601,
626
Arab, 192, 201
Arab Canaanite, 197
Arabic-speaking, 198
bedouin Sinai, 577
Canaanite, 197
Jordanian, 546
Jurhum, 197200
South Arabic, 202
sub-, 203
Yamani, 202
tribu
de la rgion de la Mecque, 117
triptote, 124125, 127
Troupeau, Grard, 6 n. 3, 7 n. 4, 23, 38,
44, 141 n. 32, 148, 234, 244, 504
n. 8, 523
752
index
Truckenbrodt, Hubert, 486 n. 16, 499

Tubayq, 561562
Tubba
kings of, 203
Tunis, 84
Tunisia, 101, 690
Turbniy, 575 n. 37, 576577, 580 n. 53
Turcs, 128
Turkey, 640
Turkic, 435438, 442443, 446, 446
n. 10, 448450
Turkish, 294295, 568, 661, 690, 693
n. 23, 679 n. 7
Turku, 607, 621
Twi, 679 n. 7
Ubayl, 195
Ubayy b. Kab, 49 n. 6, 56, 56 n. 12, 61
Uganda, 608, 621
Ugaritic, 597 n. 8, 597, 599603
uira, 667
UK, 561
ulqiyat, 61
Umayyim, 203
Umm al-Jiml, 126 n. 16
Umm Anwar, 683
umma, 404, 404, n. 2
Unz Ab Slim at-Turbn, 545 n. 4
unification-principle, 370, 383
unit(s), 508, 511512
amorphes, 519
de nomination, 506, 509, 509 n. 23,
510, 512, 515 n. 33, 516518
banale, 514
commune, 514
gnrale, 514
de nomination des entits, 508
de nomination des expriences, 508
flchies, 519
units, 367
lexical, 456
Upper Egyptian Arabic, 697
Uppsala, 222 n. 27
Ur, 197
urban
dialects, 657, 658
speakers, 657
Ursprache, 118
US, 531532, 536
uskitat, 61
al-Umn, 17 n. 13, 23
u<l al-=qh, 2628, 33, 4243
u<l:(s), 27, 3438, 4043

Qsemantics, 40
Qsources, 37, 41
`Utayba, 563
utterance, 28, 38
uvular(s), 266 n. 28, 657658
al-Uwayst, 546
Uygur, 449
Uzbekistan, 651
Arabicism, 651 n. 13
V projection, 477
valeur(s)
aspectuo-temporelles, 315 n. 4, 325
n. 15, 327, 341, 362
attributive, 340341, 343
causative-factitive, 357
dacquisition dtat, 333
dimmutabilit smantique, 342
doccurrence smantique, 342
morale, 338
physique, 338
smantique(s), 313314, 319320,
322, 324, 325 n. 15, 327, 329, 334,
337, 339, 341 n. 24, 343344,
346347, 350, 354, 358359
morpho-lexicale, 327
stative, 330, 333334
Valladolid, 219 n. 20
van Dijk-Wittop Koning, A.M., 298, 311
Vanhove, Martine, 100, 112, 685 n. 14,
693 n. 23, 697, 699
van Staden, M., 292, 311
variety(ies)
Arabic, 418, n. 13
dominant, 408, 419422
literary, 403
mixed
Arabic, 420
non-dominant, 422
superposed, 404
velarization, 371, 571, 576, 578 n. 41,
591 n. 88
verb(s), 68, 13 n. 9, 16, 18, 30, 41,
47, 4951, 5354, 5657, 59, 70,
82, 83 n. 18, 84, 86, 89, 93, 9799,
102103, 105, 108109, 139, 142, 142
n. 37 144, 144 n. 51, 149, 151152,
152 n. 3, 153, 154 n. 6, 154, 154 n. 7,
155157, 158 n. 13, 159, 163, 165
167, 177178, 178 n. 36, 179180,
183186, 196, 213, 218 n. 17, 221 n.
24, 225, 228, 230, 230 n. 39, 231, 231
index
n. 41, 233 n. 47, 235, 235 n. 49, 239,
258260, 263, 270271, 282, 284, 292,
292 n. 2, 293297, 297 n. 7, 298304,
304 n. 11, 305, 307309, 367, 369, 369
n. 7, 370, 370 nn. 12, 14, 371, 379,
391, 393394, 397, 407, 435437,
439442, 447, 449, 451452, 475 n. 1,
475476, 481, 482 n. 7, 484 n. 13, 486,
487 n. 18, 490, 497 n. 29, 498, 532
n. 5, 572, 575 n. 36, 576577, 577
n. 40, 578 n. 42, 580 nn. 50, 53, 581
n. 56, 589 n. 83, 592 n. 91, 597 n. 8,
612613, 616, 619, 639641, 641 n. 4,
642644, 646, 646 n. 10, 647, 650
n. 12, 649, 651, 676 n. 4, 679
-acting constituent, 156
of action, 394
l, yil, 675678, 681, 692696
of approximation, 378 n. 41
banish, 391, 395, 399
base, 371
of beginning, 378 n. 41
bitransitive, 380
C- final, 649
of change, 394
classes, 391, 396, 399
cognitive, 178
conjugation, 617
construction
bilingual periphrastic, 291, 297
coreferential, 92
delocutive, 698
embedded, 307, 309
English, 305306
of esteem, 378 n. 41
of experience, 394
final
//, 650
/h, 648649
// or /h/, 642, 647
laryngeal-, 646 n. 10, 648649, 650
n. 12
pharyngeal-, 650 n. 12
weak, 641
finite, 97, 155, 163, 178, 183, 294, 296,
371372, 372 n. 19, 378, 378 n. 42,
379380, 384, 394 n. 84, 395, 491
foreign, 292293, 295, 306309
imperfect, 228
embedded, 307
morphologically, 303304, 306
transitive, 293
form(s), 294, 407, 409, 640
753
gl, yigl, 675, 677678, 697
of the heart, 378 n. 41
Hebrew, 596 n. 6
of hope, 378 n. 41
imperative, 34
incomplete, 378 n. 41
incorporation, 294
integration, 292293, 295, 296 n. 4,
297298, 306309
intransitive, 391, 435, 482, 695
lexical, 675
li- +, 54
-like constituent, 152
-like element, 160
loan, 297, 652
locutive and speech-related, 693
modal, 371 n. 17
monotransitive, 380
+noun dichotomy, 165
+noun type, 165
patterns, 660
of praise and blame, 378 n. 41
prepositional, 63
prohibitive, 34
pseudo-, 73
remove, 392, 395
serial, 675
stem(s), 292, 294, 297, 299, 301, 306,
641, 645, 649
foreign, 292, 306
I, 407
IV, 406407, 419
strong, 640, 644, 646 n. 10, 649
Syriac, 109
system, 660
temporal, 371 n. 17
of transformation, 394
transitive, 391
bi-, 391
mono-, 391
tri-, 391, 380
triliteral, 487
Turkic, 437, 440, 442 n. 6
type, 369, 372, 395
weak, 369, 372, 644, 649
final, 252253, 640, 644, 647650,
650 n. 12, 648649
verbal, 5455, 57, 72, 86 n. 23, 89,
108109, 608, 611614, 616, 640
analysis, 151, 153154, 154 n. 8,
155157
clause, 163
construction, 155
754
index
effect, 156
element, 165 n. 25, 181
ending(s), 210, 228
force, 155156, 168, 176
form(s), 531, 679, 613, 615616
function, 151, 153156, 166
measures, 572
modals, 695
paradigm(s), 224 n. 31, 230 n. 39
phrase, 55
prepositional phrases, 54
sentence(s), 6, 77, 139, 149, 151,
154 n. 8, 157
status, 178
structures, 54
system, 230 n. 39
type(s), 10, 81, 86, 8990, 92-93,
98, 109
Verbalphrasen, 699
verbe(s), 115, 118, 134, 314315,
318320, 323, 326330, 332335,
339343, 345352, 353 n. 35, 354356,
358362, 504505, 506, 506 n. 15, 509,
513, 515 n. 35, 573, 575, 621
dacquisition dtat, 335, 344, 351,
358360, 362
caractristique, 352
non-caractristique, 334
actif(s), 320, 336337, 350
daction, 314, 319321, 323, 336, 359
agentif(s), 354
entirement, 359
non-, 322-323 336, 344, 346, 351,
351 n. 32, 355, 359
partiellement, 322, 359
pleinement, 322, 328 n. 19,
350351
dagentivit neutralise, 328 n. 19
arabe, 313
augment(s), 313, 324
non, 315
de changement dtat, 328329, 331,
333
commun, 508
concaves, 353354, 356
caractristique, 322, 330, 334,
341343, 352, 357, 359, 361
dtat, 314, 319, 320, 320 n. 9 321,
323, 326, 328, 334336, 342344,
354, 359, 362
externe(s), 336337, 350
intransitifs, 320, 338, 344, 346348,
351, 356
mdio-passif(s), 335336, 344, 346,

352, 354356, 358360
moyen(s), 320321, 335, 335 n. 20,
336337, 344, 346347, 349350,
352, 354, 359360
intransitif(s), 355, 337
transitif(s), 328 n. 19, 338 n. 23
direct, 337
indirect, 337
passif, 343
de perception, 323
primitif, 314
pronominaux, 354
de qualit, 321 n. 11, 328329
redoubls, 353, 353 n. 34, 354, 356
regular, 575
de sentiment, 323
simple(s), 313, 315316, 316 n. 5,
317322, 322 n. 12, 323, 324,
326327, 337338, 344, 350, 353,
357360, 362
statifs, 320
transitifs, 320, 337, 338 n. 23, 341,
349, 352, 354, 356357
directs, 348351
indirects, 347351
vernaculaire(s), 125
vernacular(s), 223, 403, 527, 559560,
607, 609610, 613
African, 609, 611
Non-Arabic, 610
speech, 227
vernacularization, 608, 618
vernacularized, 611, 620
non, 612
Arabic, 607
verse(s), 46 n. 2, 189, 204, 548 n. 5
pre-Islamic, 89
Versteegh, Kees (C.H.M.), 33, 4445, 65,
114, 117, 120 n. 6, 121 n. 10, 124, 126,
128, 128 n. 20, 133, 209 n. 1, 227, 364,
371 n. 15, 402, 434, 436, 441, 453, 455,
473, 503 n. 6, 504 n. 9, 523, 565 n. 1, 578,
607, 621, 639, 652, 654, 656, 664665,
671, 673, 675, 675 n. 1, 692, 693, 699
vocabulary, 654, 656, 662, 665666,
668670
vocalisation, 233 n. 45, 445
vocalism, 604
vocatiuo, 214
vocative(s), 41, 214, 217, 375, 387,
397398, 676, 696
noun, 41
index
vocativus, 221
voice(s), 370372, 395, 426432
active, 379
passive, 379380, 397 n. 88
voiced, 256 n. 7
non-, 256, 256 n. 7, 267, 275
voicing, 371
Voigt, Rainer, 595 n. 1, 599
volition, 370, 383, 396397
Vollers, Karl, 125, 130
Vopadeva, 209
vowel(s), 9, 15, 47, 214 n. 10, 217219,
219 n. 19, 224 nn. 3132, 225228,
230, 233, 239240, 247, 250, 256, 368,
435, 445, 446 n. 10, 447450, 540,
543, 568570, 570 n. 23, 572, 572
n. 27, 573575, 575 n. 34, 577, 579
n. 44, 582 n. 60, 595, 597598, 601,
612613
back, 296
Canaanite, 598
declensional, 230 n. 39
deletion 251
empty, 371
ending(s), 4647, 63 n. 16, 252253,
543
short-, 63, 64 n. 17
final, 16, 63, 253, 295, 646 n. 10
stem, 643644, 649
front, 296
high stem, 641, 643 n. 9
infixed, 371
inflectional, 216, 224227
-initial, 572, 572 n. 27
length, 449, 488
lengthened, 448
long, 252253, 407, 407 n. 4, 440, 449,
488489, 491, 491 n. 23, 611, 614,
640, 641 n. 4, 649
low stem, 643 n. 9, 641
marks, 251
mutation, 9
omission, 253
pattern(s), 370, 411412, 447, 488,
490, 492
prefix(ed), 371, 641 n. 4
preformative, 262, 597
quantity, 488
rear, 256 n. 10
semantic, 224
semi-, 641
sequence, 436, 439
stem, 641 n. 4, 644645
755
short, 247, 251, 449, 568, 569

final, 248249, 251
short high front and back, 641 n. 4
system, 224, 224 n. 32, 233, 657658
terms, mnemonic, 217
omission of, 224 n. 31
zero-, 227
voyelle(s), 314315, 316 n. 5, 317,
319, 338, 345 n. 28, 510, 510 n. 24,
511, 511 nn. 2526, 512513, 515
n. 31
dsinentielle, 509
premire, 513
mdiane(s), 313, 314 n. 3, 315,
317319
wa-, 38
wa/, 39
w/i (creator), 10
Wdiy
sSa haw, 566
Frn, 566
arandal, 566
Isla, 566
Lihyn, 566
Wahhabi scholars, 200
Wahd, 685
wahy, 192, 193
wajh, 48, 249
wjib, 169
Walid Irfa, 535
Wald Ysif, 699
wallada, 115
Wallin, Georg August, 544, 563
waqf, wuqf, 85, 227, 247, 247253
War an Nfi, 127
wasfiyya, 158
wasl, 127
Watson, Richard, 607, 619, 621
wafa, 249
wazn, 435, 444, 509
Wehr, Hans, 344, 365, 599, 600, 601,
602, 603, 604, 605, 668, 673
Wehr-Cowan, 315, 331, 339, 340, 343,
353
Weinreich, Uriel, 665, 674
Weiss, Bernard G., 193, 208
Wei, Josef, 228, 234, 244
Wellens, Ineke, 607, 621
Werner, John Henry, 220
Wernitz, C.J., 292, 311
West Germany, 546
Western
756
index
languages, 104
Sudanese Arabic, 614
Sudanic, 611
tradition, 219
Wetzstein, Johann Gottfried, 126, 133
wi (and), 677, 679, 694, 695
wi, 163
Wild, Stefan, 435, 436, 437, 438, 453
Winograd, 393, 402
wish, 398
Woidich, Manfred, 7278, 80, 85, 86,
9498, 100-102, 112, 224, 224 n. 31,
402, 455, 473, 565 n. 2, 675, 675
n. 2, 678, 683, 683 n. 12, 687, 687
n. 17, 688, 688 n. 18, 692, 692, n. 22,
694696, 699
word
classes, 213, 236
-final position, 657
formation, 475, 479, 498
order, 12
Wright, William, 314, 320, 327, 338, 365,
373, 374, 378, 385, 386, 388, 435,
436, 453, 487, 499
written tradition, 210
WSA, 610, 616
Wurff, Wim van der, 613, 614, 697
Wurzelbegriff, 219
Wstenfeld, Ferdinand, 195, 203, 207
xabar, 48, 49, 138, 144 n. 19, 149, 150,
151, 152, 153, 154, 155, 156, 157,
158, 159, 160, 161, 162, 163, 164,
168, 169, 170, 171, 172, 173, 174,
176, 178, 179, 180, 182, 183, 184, 216
accusatival, 159
adverbial, 169
adverbial/prepositional, 164, 168
clausal, 164, 180
fronted, 153, 182
kna, 179
non-adverbial, 169
muqaddam, 153
mubtada, 180
obligatorily fronted, 168, 173, 174
obligatory fronting, 170, 171
preposed obligatorily, 173
as-sifa, 173
xaf/, 60, 234
xfi/, 58, 60
Xlid b. Sinn al-Abs, 195
al-Xall b. Ahmad, 28, 29 n. 1, 33
151, 152, 155156, 238, 444, 453
Xawrij, 530
al-Xidr, 195
xiftu, 660
xilf (or muxlafa), 162
XP, 478
xucla, 210, 217
xuft, 660
al-Xwrazm, 33
y, 41
Yahwe, 105
Yaman, 202, 203
Yamani(s), 202
yaqdiru, 660
al-Yaqb, 199
Yqt, 202, 203
yarifu, 660
Yarub, 202, 203, 204
Yarub b. Qaht n, 196, 199, 201204
Yajub b. Qaht n, 195, 199, 202
Yajub/Yaman, 202
Yassin, M.A., 555, 563
yxud, 660
yaxudu, 660
Yemen, 195
Yemeni dialects, 640
Yeou, M., 266, 289
yieraf, 660
yidar, 660
Yishml, 198
Yngve, Victor, 389, 402
Yoda, Sumikazu, 100, 112
Yokwe, Elisai M., 607, 621
Yoon, James, 483, 499
Youssi, Abderrahim, 674
yxed, 660
yxud, 660
Yucatn, 219
Yumn Bassiouni, 533, 536
Ynus, 195
Ysuf al-Qad, 699
Ysuf f, 699
Zaborski, Andrezej, 402
hir, 151
|hirite(s), 11
az-Zajjj, 57, 58, 64, 113
az-Zajjj, 113, 114, 121124, 126, 131,
153, 154, 157, 186, 187, 318, 362
zakh, 31
Zakariyy, Michel, 3, 23
Zakia Iraqui Sinaceur, 698
az-Zamaxar, 89, 111, 157, 177, 180,
287, 320, 329, 338, 342, 343, 345, 362
index
Zande, 609
arf, 135, 136, 137, 138, 139, 140, 141,
142, 143, 144, 145, 146, 147, 162,
177, 181, 214, 217.
dispensable part, 135, 136
indispensable part, 135
indispensable predicate, 136
maa, 144
mulan, 142, 147
arfiyya, 177, 184

Zarpetea, P., 303, 311
az-Zarq, 546
zawid, 509
Zayd, 13
zayy, 78, 689
illi, 74, 77, 78
zero endings, 230
757
STUDIES IN SEMITIC
LANGUAGES AND LINGUISTICS
3. Corr, A.D. The Daughter of My People. Arabic and Hebrew Paraphrases of
Jeremiah 8.13-9.23. 1971. ISBN 90 04 02552 9
5. GrandHenry, J. Les parlers arabes de la rgion du Mza b (Sahara algrien). 1976.
ISBN 90 04 04533 3
6. Bravmann, M.M. Studies in Semitic Philology. 1977. ISBN 90 04 04743 3
8. Fenech, E. Contemporary Journalistic Maltese. An Analytical and Comparative Study. 1978. ISBN 90 04 05756 0
9. Hospers, J.H. (ed.). General Linguistics and the Teaching of Dead Hamito-Semitic Languages. Proceedings of the Symposium held in Groningen, 7th-8th
November 1975, on the occasion of the 50th Anniversary of the Institute
of Semitic Studies and Near Eastern Archaeology of the State University at
Groningen. 1978. ISBN 90 04 05806 0
12. Hoftijzer, J. A Search for Method. A Study in the Syntactic Use of the Hlocale in Classical Hebrew. With the collaboration of H.R. van der Laan
and N.P. de Koo. 1981. ISBN 90 04 06257 2
13. Murtonen, A. Hebrew in its West Semitic Setting. A Comparative Survey of
Non-Masoretic Hebrew Dialects and Traditions. Part I. A Comparative
Lexicon.
Section A. Proper Names. 1986. ISBN 90 04 07245 4
Section Ba. Root System: Hebrew Material. 1988. ISBN 90 04 08064 3
Section Bb. Root System: Comparative Material and Discussion. Sections C,
D and E: Numerals under 100, Pronouns, Particles. 1989.
ISBN 90 04 08899 7
14. Rets, J. Diathesis in the Semitic Languages. A Comparative Morphological
Study. 1989. ISBN 90 04 08818 0
15. Rouchdy, A. Nubians and the Nubian Language in Contemporary Egypt. A Case of
Cultural and Linguistic Contact. 1991. ISBN 90 04 09197 1
16. Murtonen, A. Hebrew in its West Semitic Setting. A Comparative Survey of
Non-Masoretic Hebrew Dialects and Traditions. Part 2. Phonetics and
Phonology. Part 3. Morphosyntactics. 1990. ISBN 90 04 09309 5
17. Jongeling K., H.L. Murre-van den Berg & L. van Rompay (eds.). Studies in
Hebrew and Aramaic Syntax. Presented to Professor J. Hoftijzer on the Occasion of his Sixty-Fifth Birthday. 1991. ISBN 90 04 09520 9
18. Cadora, F.J. Bedouin, Village, and Urban Arabic. An Ecolinguistic Study.
1992. ISBN 90 04 09627 2
19. Versteegh, C.H.M. Arabic Grammar and Qur"a nic Exegesis in Early Islam.
1993. ISBN 90 04 09845 3
20. Humbert, G. Les voies de la transmission du Kita b de Sbawayhi. 1995. ISBN
90 04 09918 2
21. Mifsud, M. Loan Verbs in Maltese. A Descriptive and Comparative Study.
1995. ISBN 90 04 10091 1
22. Joosten, J. The Syriac Language of the Peshitta and Old Syriac Versions of Matthew.
Syntactic Structure, Inner-Syriac Developments and Translation Technique. 1996. ISBN 90 04 10036 9
23. Bernards, M. Changing Traditions. Al-Mubarrads Refutation of Sbawayh
and the Subsequent Reception of the Kita b. 1997. ISBN 90 04 10595 6
24. Belnap, R.K. and N. Haeri. Structuralist Studies in Arabic Linguistics. Charles A.
Fergusons Papers, 1954-1994. 1997. ISBN 90 04 10511 5
25. Talmon R. Arabic Grammar in its Formative Age. Kita b al-"Ayn and its Attribution
to ]all b. Ah.mad. 1997. ISBN 90 04 10812 2
26. Testen, D.D. Parallels in Semitic Linguistics. The Development of Arabic la- and
Related Semitic Particles. 1998. ISBN 90 04 10973 0
27. Bolozky, S. Measuring Productivity in Word Formation. The Case of Israeli
Hebrew. 1999. ISBN 90 04 11252 9
28. Ermers, R. Arabic Grammars of Turkic. The Arabic Linguistic Model Applied to
Foreign Languages & Translation of #Abu- ayya-n al-#Andaluss Kita-b al-"Idra-k liLisa-n al-"Atra-k. 1999. ISBN 90 04 113061
29. Rabin, Ch. The Development of the Syntax of Post-Biblical Hebrew. 1999.
ISBN 90 04 11433 5
30. Piamenta, M. Jewish Life in Arabic Language and Jerusalem Arabic in Communal Perspective. A Lexical-Semantic Study. 2000. ISBN 90 04 11762 8
31. Kinberg, N. ; Versteegh, K. (ed.). Studies in the Linguistic Structure of Classical
Arabic. 2001. ISBN 90 04 11765 2
32. Khan, G. The Early Karaite Tradition of Hebrew Grammatical Thought. Including a
Critical Edition, Translation and Analysis of the Diqduq of "Ab Ya#qb
Ysuf ibn N on the Hagiographa. 2000. ISBN 90 04 11933 7
33. Zammit, M.R. A Comparative Lexical Study of Qur"nic Arabic.
ISBN 90 04 11801 2 (in preparation)
34. Bachra, B.N. The Phonological Structure of the Verbal Roots in Arabic and Hebrew.
2001. ISBN 90 04 12008 4
35. kesson, J. Arabic Morphology and Phonology. Based on the Mar al-arw by
Amad b. #Al b. Mas#d. Presented with an Introduction, Arabic Edition,
English Translation and Commentary. 2001. ISBN 90 04 12028 9
36. Khan, G. The Neo-Aramaic Dialect of Qaraqosh. 2002. ISBN 90 04 12863 8
37. Khan, G., ngeles Gallego, M. and Olszowy-Schlanger, J. The Karaite Tradition of Hebrew Grammatical Thought in its Classical Form. A Critical Edition and
English Translation of al-Kitb al-Kf f al-Lua al-#Ibrniyya by "Ab al-Faraj
Hrn ibn al-Faraj. 2 Vols. 2003. ISBN 90 04 13272 4 (Set), ISBN 90 04
13311 9 (Vol. 1), ISBN 90 04 13312 7 (Vol. 2)
38. Haak, M., De Jong, R., Versteegh, K. (eds.). Approaches to Arabic Dialects.
A Collection of Articles presented to Manfred Woidich on the Occasion of
his Sixtieth Birthday. 2004. ISBN 90 04 13206 6
39. Takcs, G. (ed.). Egyptian and Semito-Hamitic (Afro-Asiatic) Studies in Memoriam
W. Vycichl. 2004. ISBN 90 04 13245 7
40. Maman, A. Comparative Semitic Philology in the Middle Ages. From Sa#adiah
Gaon to Ibn Barn (10th-12th C.). 2004. ISBN 90 04 13620 7
41. Van Peursen, W.Th. The Verbal System in the Hebrew Text of Ben Sira. 2004.
ISBN 90 04 13667 3
42. Elgibali, A. Investigating Arabic. Current Parameters in Analysis and Learning.
2004. ISBN 90 04 13792 0
43. Florentin, M. Late Samaritan Hebrew. A Linguistic Analysis of Its Different
Types. 2004. ISBN 90 04 13841 2
44. Khan, G. The Jewish Neo-Aramaic Dialect of Sulemaniyya and \alabja. 2004.
ISBN 90 04 13869 2
45. Wellens, I. The Nubi Language of Uganda. An Arabic Creole in Africa. 2005.
ISBN 90 04 14518 4
46. Bassiouney, R. Functions of Code Switching in Egypt. Evidence from Monologues. 2006. ISBN 90 04 14760 8
47. Khan, G. Semitic Studies in Honour of Edward Ullendorff. 2005.

ISBN 90 04 14834 5
48. Mejdell, G. Mixed Styles in Spoken Arabic in Egypt. Somewhere between Order
and Chaos. 2006. ISBN-10: 90 04 14986 4, ISBN-13: 978 90 04 14986 1
49. Ditters, W.E. and Motzki, H. (eds.). Approaches to Arabic Linguistics. Presented to
Kees Versteegh on the Occasion of his Sixtieth Birthday. 2007.
ISBN 978 90 04 16015 6

Approach To Arabic Linguistics PDF

Uploaded by

Copyright:

Available Formats

Approach To Arabic Linguistics PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Approach To Arabic Linguistics PDF

Uploaded by

Copyright:

Available Formats

Approaches to Arabic Linguistics

Studies in Semitic Languages

Everhard Ditters and Harald Motzki

This book is printed on acid-free paper.

Bibliography Kees Versteegh ............................................................

Pragmatics and Contractual Language in Early Arabic

Idmr in the Man of al-Farr: A Grammatical Approach

Arabic allad as a Conjunction: An old Problem and a New

Les origines de la grammaire arabe, selon la tradition:

Sbawayhis View of the zarf as an mil .........................................

Problems in the Medieval Arabic Theory of Sentence Types .........

Arabic avant la lettre. Divine, Prophetic, and Heroic Arabic .......

The Explanation of Homonymy in the Lexicon of Arabic .............

The Periphrastic Bilingual Verb Construction as a Marker of

Faula, faila, faala: dispersion et rgularits smantiques dans

Featuring as a Disambiguation Tool in Arabic Natural

Arabic on the Media: Hybridity and Styles .....................................

Do They Speak the Same Language? Language Use in Juba

Paradigmatic Stability and Final Laryngeals in Nigerian Arabic:

Some Aspects of Diglossia as Reflected in the Vocabulary of

Everything you Always Wanted to Know about l, yil to say

From when we began to compile a Festschrift for Kees Versteegh on the

apparent in his search for Hellenistic elements in Arabic linguistics.

al anna afsah ah ajamu

On the tunes of birds that speak true Arabic,

BIBLIOGRAPHY KEES VERSTEEGH

bibliography kees versteegh

bibliography kees versteegh

bibliography kees versteegh

bibliography kees versteegh

bibliography kees versteegh

Geschiedschrijving in de klassieke Arabisch-Islamitische samenleving.

bibliography kees versteegh

bibliography kees versteegh

New approaches towards the history of Arabic. Taqaddum al-lisniyyt

bibliography kees versteegh

bibliography kees versteegh

Robins, Robert H. 1993. The Byzantine Grammarians: Their place

bibliography kees versteegh

honor of El-Said Badawi. Edited by Alaa Elgibali. Cairo: American

bibliography kees versteegh

Bekkum, Jan Houben, Ineke Sluiter and Kees Versteegh. Amsterdam:

bibliography kees versteegh

bibliography kees versteegh

bibliography kees versteegh

C. E. Bosworth, E. van Donzel and W. P. Heinrichs. Leiden: E. J.

bibliography kees versteegh

by Geert Boo, Christian Lehmann, Joachim Mugdan and Stavros

bibliography kees versteegh

bibliography kees versteegh

INSIDE THE SPEAKERS MIND:

speakers awareness as arbiter of usage

of operations performed by the speaker in order to achieve a specific

2. Meaning and speakers awareness

speakers awareness as arbiter of usage

yasruxu) before the accusative noun (Kitb I:355f.). The intention of

speakers awareness as arbiter of usage

Jinn stands out as a unique scholar whose innovative approach in his

over others (II:33), or to the onomatopoeic nature of phonemes within

regarded as single words) while kad wa-kd stands for compound

case, e.g. in Q 2:19 (Man I:17, 1ff.): yajalna asbiahum . . . h adara

3.5.5 amm . . . fa-: Man I:228 regarding Q 3:106: fa-amm lladna