Hristov, B. (2020) - Grammaticalising The Perfect and Explanations of Language Change

Grammaticalising the Perfect and Explanations of Language Change
Brill’s Studies in Historical

Linguistics
Series Editor
Jóhanna Barðdal
(Ghent University)
Consulting Editor
Spike Gildea
(University of Oregon)
Editorial Board
Joan Bybee (University of New Mexico) – Lyle Campbell (University of Hawai’i

Manoa) – Nicholas Evans (The Australian National University)
Bjarke Frellesvig (University of Oxford) – Mirjam Fried (Czech Academy of
Sciences) – Russel Gray (University of Auckland) – Tom Güldemann
(Humboldt-Universität zu Berlin) – Alice Harris (University of Massachusetts)
Brian D. Joseph (The Ohio State University) – Ritsuko Kikusawa (National
Museum of Ethnology) – Silvia Luraghi (Università di Pavia)
Joseph Salmons (University of Wisconsin) – Søren Wichmann (MPI/EVA)
volume 10
The titles published in this series are listed at brill.com/bshl

Grammaticalising the Perfect and
Explanations of Language Change
Have- and Be-Perfects in the History and Structure of
English and Bulgarian
By
Bozhil Hristov
LEIDEN | BOSTON
Library of Congress Cataloging-in-Publication Data
Names: Hristov, Bozhil, 1984- author.

Title: Grammaticalising the perfect and explanations of language change :
have- and be-perfects in the history and structure of English and
Bulgarian / by Bozhil Hristov.
Description: Leiden ; Boston : Brill, 2020. | Series: Brill’s studies in
historical linguistics, 22114904 ; vol. 10 | Includes bibliographical
references and index.
Identifiers: LCCN 2019046039 (print) | LCCN 2019046040 (ebook) |
ISBN 9789004414327 (hardback) | ISBN 9789004414051 (ebook)
Subjects: LCSH: English language—Tense. | Bulgarian language—Tense. |
English language—Grammar, Comparative—Bulgarian. | Bulgarian
language—Grammar, Comparative—English.
Classification: LCC PE1301 .H75 2020 (print) | LCC PE1301 (ebook) |
DDC 425/.62—dc23
LC record available at https://lccn.loc.gov/2019046039
LC ebook record available at https://lccn.loc.gov/2019046040
Typeface for the Latin, Greek, and Cyrillic scripts: “Brill”. See and download: brill.com/brill-typeface.
ISSN 2211-4904
ISBN 978-90-04-41432-7 (hardback)
ISBN 978-90-04-41405-1 (e-book)
Copyright 2020 by Koninklijke Brill NV, Leiden, The Netherlands.

Koninklijke Brill NV incorporates the imprints Brill, Brill Hes & De Graaf, Brill Nijhoff, Brill Rodopi,
Brill Sense, Hotei Publishing, mentis Verlag, Verlag Ferdinand Schöningh and Wilhelm Fink Verlag.
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.
This book is printed on acid-free paper and produced in a sustainable manner.

For Maggie and Kalina,
who came knocking on the door
when uncle didn’t have time to play.
∵
Contents
Acknowledgements xi
List of Abbreviations xiii
Transliteration from Cyrillic XV
1 Introduction 1
2 Theoretical Preliminaries 9
2.1 Evolutionary and Functionalist Models of Language Change 9
2.1.1 Evolutionary Linguistics 9
2.1.2 The Concepts of Function and Functional Load, the Invisible Hand
and Teleology 12
2.1.3 Problems and Challenges 16
2.1.4 Interim Summary and Preview 19
2.2 Grammaticalisation and Reanalysis 20
2.2.1 Grammaticalisation 20
2.2.2 Reanalysis in the Context of Grammaticalisation 25
2.2.3 The Role of Frequency and Contact in Grammaticalisation 27
2.3 Other Preliminaries 28
2.3.1 Progress 28
2.3.2 Randomness 29
2.3.3 Justification 30
2.4 Conclusion 30
3 The Story of the English Perfect 32

3.1 Perfect with Have 32
3.1.1 Terminological and Etymological Preliminaries 32
3.1.2 The Have-Perfect in Old English: Morphological Marking,
Ambiguity, and Reanalysis 34
3.1.3 Perfect with OE agan 43
3.1.4 Increased Use and Greater Degree of Grammaticalisation of the
Have-Perfect in Middle and Modern English 43
3.2 The Origin and Status of the Be-erfect 45
3.3 Competition between the Be- and the Have-Perfect 48
3.4 Traditional Accounts for the Decline of the Be-Perfect 52
3.4.1 Frequency 52
3.4.2 Functional Load and Ambiguity 54
3.4.3 Functional Load and the Disappearance of OE weorðan 58
3.4.4 Ambiguity in the Contexts of Coordination and Contraction 60
viii Contents
3.5 Problems for the Traditional View 61

3.5.1 Agreement and Perfect Readings 62
3.5.2 Proliferation of the Functions of Have 63
3.5.2.1 Passive Have 63
3.5.2.2 Causative Have 65
3.5.2.3 Modal Have 67
3.5.2.4 Summary 71
3.5.3 Development of Alternative Passives and Perfects 72
3.5.3.1 Passives with Get 72
3.5.3.2 Perfects and Causatives with Get 73
3.5.3.3 New Be-Perfects 75
3.5.4 Corpus Data and Statistics 77
3.5.5 Alternative Explanations 80
3.6 Conclusion 82
4 The Development of the Perfect in a Selection of Old English Texts 85

4.1 Objectives of the Study 85
4.2 Issues in Corpus Work and Compilation 86
4.3 Choice of Text Samples and Approach 88
4.4 The Perfect in the Anglo-Saxon Chronicles 90
4.4.1 Introduction and Background 90
4.4.2 Analysis and Discussion 91
4.4.2.1 Manuscript A: Eighth Century 91
4.4.2.2 Some Notes on OE Agreement 99
4.4.2.3 Manuscript A: Ninth Century 102
4.4.2.4 Manuscript A: Late Ninth and Early Tenth
Centuries 110
4.4.2.5 Manuscript E (Peterborough Chronicle): Late Tenth and
Early Eleventh Centuries 130
4.5 Conclusion 144
5 Further Development of the Perfect Based on a Selection of Texts – from

Middle into Modern English 146
5.1 Introduction 146
5.2 The Fourteenth Century: Chaucer 147
5.3 The Fifteenth Century: The Second Shepherds’ Play (Secunda
Pastorum) 160
Contents ix
5.4 The Sixteenth Century: Shakespeare 167

5.5 The Seventeenth Century: Restoration Comedy 183
5.6 Late Middle English Correspondence: the Paston Letters 195
5.6.2.1 William Paston I 196
5.6.2.2 William Paston III 200
5.7 Seventeenth-Century Correspondence: Samuel Pepys 209
5.8 Epilogue 218
5.9 Conclusion 221
6 How Things Could Have Been: a Glance at German 223

6.2 Analysis and Discussion of Eighteenth-Century German Drama 228
6.3 Conclusion 237
7 The Perfect in the History and Structure of Bulgarian 238

7.1 Overview of the Bulgarian Temporal and Aspectual System 239
7.2 The Bulgarian Be-Perfect 241
7.3 Have as an Auxiliary in the History of Bulgarian 244
7.4 The Have-Perfect in Bulgarian 245
7.4.1 Formation 245
7.4.2 Agreement 249
7.4.3 Status of the Have-Perfect 251
7.4.4 Functional Load as an Explanation for the Rise of the Have-Perfect
in Bulgarian 254
7.4.5 The Have-Perfect in Macedonian 260
7.4.6 Agreement and Grammaticalisation Revisited 262
7.4.7 Mixing of Active and Passive Participles and Language Contact as
Explanations for the Have-Perfect in Bulgarian and
Macedonian 265
7.5 Conclusion 270
x Contents
8 The Development of the Perfect in a Selection of Middle and Modern

Bulgarian Texts 272
8.2 The Tale of Troy (14th c.) 273
8.3 Damascenus Troianensis (17th c.) 287
8.4 A Slavonic-Bulgarian History (1762) 302
8.5 Epilogue: the Have-Perfect in Modern Bulgarian 318
8.6 Conclusion 326
9 Conclusions 329
References 335
Author Index (Modern Authors) 356
Language Index 360
Subject Index 362
Index of Sources 367
Acknowledgements
First of all, I would like to thank Mira Kovatcheva, my teacher and esteemed
colleague, for her generous professional guidance, and more specifically for
feedback on abstracts, book recommendations and the unstinting expendi-
ture of time and effort to improve the draft (which aroused the curiosity of
Charlie the cat, too). I am indebted to Mitko Sabev, Mark Hadley, Michael
Everdell, Vladimir Bondar, Florian Dolberg and Steven Kaye for supplying me
with papers and/or various insights, as well as to Evgenia Pancheva for ad-
vice on picking mediaeval literary texts. During pleasant conversations about
things literary, philosophical and philological, she and I were unable to decide
whether we should opt for the spelling Hilendar or Hilandar. Stefan Stefanov
and Theodore Todorov readily and enthusiastically helped with Latin even
during unsocial hours. The anonymous reviewers for the LAGB, ISLE 2016 and
ICEHL XX conferences made useful contributions and raised some very per-
tinent issues. The financial support and travel grants from Sofia University’s
Faculty of Classical and Modern Languages and Doctoral School, as well as
the Erasmus+ exchange programme, are gratefully acknowledged. Part of the
research was finalised and made possible by a Fulbright grant at the University
of Texas at Austin. Thanks go to the audiences at the LAGB 2016 meeting in
York, the ‘Gradience and Constructional Change in English’ workshop in
Edinburgh (2017), the 20th International Conference on English Historical
Linguistics (2018), as well as the Syntax/Semantics and Historical Linguistics
discussion groups at UT Austin, for their intelligent questions and comments,
especially David Beaver, John Beavers, Theresa Biberauer, Mary Blockley,
Vladimir Bondar, Bernard Comrie, Hubert Cuyckens, Patience Epps, Michael
Everdell, Teresa Fanego, Carole Hough, Cristian Juares, Danny Law, Benjamin
Molineaux, Muriel Norde, Peter Petré, Susan Pintzuk, Tamara Bouso Rivas,
Hendrik De Smet, Graeme Trousdale, Turo Vartiainen and Steve Wechsler. I
would also like to express my deep sense of gratitude to the participants in the
2017 Winter School at the New Bulgarian University, especially Elena Savova,
Maria Neikova, Maria Stambolieva, Milka Hadjikoteva and Zhivko Hristov,
for their warm reception and thought-provoking feedback. I am grateful to
the members of staff at the University of Roehampton, London, who made
it to the Research Seminar in February 2017, especially Eva Duran-Eppler and
Evangelia Sifaki. As ever, Alexandra Bagasheva was a knowledgeable point of
contact for references and many fruitful and enjoyable discussions. Andrei
Stoevsky, John H. McWhorter and Matti Kilpiö kindly provided me with some
xii Acknowledgements
very important material, going to a lot of trouble posting it and making sure it
reached me. Members of Sofia University’s Faculty of Slavonic Languages, in-
cluding Tatyana Slavova, Anna-Maria Totomanova, Gergana Ganeva and Petko
Petkov, allowed me to sit in on their classes on Old Bulgarian and the History
of Bulgarian. My interest in these subjects was first sparked by Catherine
Mary MacRobert at Oxford. Yoana Sirakova from the Department of Classics
let me join her introductory Latin course. Thomas Kohnen was a lovely host
and granted me access to the library collections at the University of Cologne.
Andrei Stoevsky, Tatyana Slavova, Gergana Ganeva and Mary Dalrymple made
helpful comments and suggestions on earlier drafts. Gergana Ganeva further
assisted with various matters, especially those related to Middle Bulgarian and
the history of the language. Relying on his meticulous eye for detail and native
intuitions, Steven Kaye read the text and made numerous stylistic and linguis-
tic improvements. Colleagues from the English Department at the University
of Sofia, including A. Bagasheva, E. Slavova, M. Kolarova, M. Kovatcheva, N.
Tincheva, N. Yakimova, R. Ishpekova, and T. Venkova, provided feedback on the
first draft and/or words of encouragement during a stimulating departmental
discussion. Denitsa Bozhilova double-checked some of the basic maths and
statistics. Last but not least, I would like to thank Brill’s editors Elisa Perotti
and Maarten Frieswijk, as well as the series editor Jóhanna Barðdal, for their
high professional standards and commitment, including the numerous help-
ful suggestions on improving the draft. I am likewise much obliged to Bridget
Drinka for taking on the task of reviewing the complete manuscript and for
her constructive feedback and valuable pointers. Like the families of many
other academics, mine have also been unfailingly patient, supportive and un-
derstanding during the long hours of writing, including even my little nieces.
Needless to say, all mistakes, omissions and flaws remain my own.
Abbreviations
A adjunct
acc accusative (case)
act active (voice)
adj./adj adjective
adv adverb
AdvP adverb(ial) phrase
AmE American English
aor aorist (tense)
aux auxiliary (verb)
BNC British National Corpus
BrE British English
CO object complement
COCA Corpus of Contemporary American English
COHA Corpus of Historical American English
comp complementiser
conj conjunction
CONTAmE Contemporary American English
CONTBrE Contemporary British English
conv conversation
dat dative (case)
def definite (article)
DP determiner phrase
EME Early Middle English
EModE Early Modern English
F feminine (gender)
fut future
gen genitive (case)
HPSG Head-Driven Phrase Structure Grammar
imperf imperfect (tense)
impers. impersonal
inf/inf. infinitive
injc interjection
ipfv imperfective (aspect)
L Latin
LFG Lexical Functional Grammar
LOE Late Old English
m masculine (gender)
xiv Abbreviations
ME Middle English
ModE Modern English
n. noun
neg negative
nom nominative (case)
NP noun phrase
nt neuter (gender)
O object
OB Old Bulgarian
OCS Old Church Slavonic
OE Old English
OED Oxford English Dictionary
ON Old Norse
pass passive (voice)
PDE Present-Day English
pfv perfective (aspect)
pl plural (number)
PP prepositional phrase
pptcp past participle
Pred predicate
pref prefix
pres present (tense)
ptcp participle
refl./refl reflexive
S subject
sb. somebody
sbjv subjunctive (mood)
SG singular (number)
V verb
voc vocative (case)
VP verb phrase
Transliteration from Cyrillic
following Comrie and Corbett (1993: xii–xiii)
Cyrillic Latin Cyrillic Latin Cyrillic Latin
а a п p ы y
б b р r ѣ ĕ
в v с s ѧ ę
г g т t ѫ ǫ
д d у u
е e ф f
ж ž х x
з z ц c
и i ч č
й j ш š
к k щ št
л l ъ ă
м m ь ’
н n ю ju
о o я ja
Note: The jer vowel symbols, ъ and ь, are retained for earlier stages of
Bulgarian. Alternatively, they can be transliterated as follows: ъ = ŭ, ь = ĭ.
A Greek-style ω (omega) appears in some of the Cyrillic texts and is
transliterated as o.
chapter 1
Introduction
As the title suggests, this book aims to cover ground at the intersection of three
broad subfields of linguistics: grammar (in the sense of morphology and syn-
tax), the linguistic history of English and Bulgarian, and historical linguistics.
I set out to investigate aspects of the morphosyntactic structure of the two
languages selected, as well as how their grammars came to be the way they
are today. Unsurprisingly, the discussion of each enlightens the discussion of
the other in ways unattainable by considering them separately. Throughout
the text, I highlight parallel explanations of what appear to be nearly identi-
cal phenomena in a pair of rather distant relatives, one of them much less
studied than the other. It is therefore attempted to fill some of the research
gaps (even for a language as extensively investigated as English) and situate the
work in the context of a more general theory of language change as applied to
the events of interest here. Marshalling extensive data from two languages as
opposed to one, and at the same time touching upon other varieties such as
Latin, German and Macedonian, shifts the focus away from English, which still
dominates the literature, and also lends more empirical, areal and typological
support to the general theoretical conclusions. The book builds on an enor-
mous amount of scholarship accumulated so far, reassesses and revises some
long-held notions and shines the spotlight on certain areas that have received
less attention, such as the role of ambiguity in actual usage. The detailed anal-
ysis of rich, contextualised material from a selection of texts dovetails with
large-scale corpus studies, complementing their findings and adding value to
our understanding of the processes under discussion. This monograph thus
presents a happy marriage of traditional philological techniques, which retain
their worth and usefulness, and recent advances in theoretical linguistics and
corpus work.
The following chapter addresses the theoretical preliminaries. It contains
a brief overview of evolutionary models of language change, adopted into lin-
guistics after the momentous breakthroughs of 19th-century biology, includ-
ing some functionalist ideas associated with the same theoretical affiliation.
A cautionary note is sounded to the effect that, neat and attractive though
these models may seem, there are certain problems arising from the transfer
of biological ideas to the study of language. Special attention is paid to central
notions such as function and functional load, as well as the ‘invisible hand’
in language and teleological (i.e. purpose-driven) change, some of which are
© Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004414051_002

2 chapter 1
taken up in later chapters. I then proceed to flesh out the basic principles of
grammaticalisation and reanalysis, since this monograph is informed by the
broad field of grammaticalisation studies. Some of the pertinent properties of
grammaticalisation are singled out and briefly explained, among them the role
of frequency and language contact in grammaticalisation, as well as some well-
established tenets with a bearing on its nature and modus operandi.
Having thus set the scene by going through the major points of theoretical
interest, in Chapter 3 I trace the story of the perfect in English. Both have- and
be-perfects are considered, starting from Old English times, and are compared
to other periphrases, such as the passive. The origins of the English perfect
(and passive) go back to copular and possessive constructions with verbs
meaning ‘be’, ‘become’ and ‘have’, combined with past participles which origi-
nally showed agreement either with the subject or with the object (depend-
ing on the construction). The earlier participial (and hence adjectival) subject
or object complement gradually came to be reanalysed as a lexical verb. The
emphasis here falls on the morphological marking on the participle, includ-
ing ambiguous cases which are often said to have triggered this reanalysis and
the subsequent loss of agreement. Crucially, I stress that there are attestations
which demonstrate that overt agreement on the participle is not incompatible
with a bleached meaning of have serving as a perfect auxiliary (rather than a
lexical verb indicating possession).
The chapter’s second concern is with the origin and status of the be-perfect
in English, as well as the competition between be and have in this domain, cul-
minating in the ultimate disappearance of be as a viable perfect auxiliary for
so-called ‘mutative’ intransitive verbs. This is interesting because it will throw
into sharp relief the properties of both auxiliaries, at the same time inviting
some more general conclusions regarding the nature of language change.
Tying these points to the content of the opening chapter, I recount some of
the traditional explanations offered in the literature for the displacement of
be by have. One explanation relies on the greater type and token frequency
of have in this function. A particularly popular hypothesis rests on claims that
be was ill-suited for the role in question since it came to be overloaded with too
many functions – in perfect, passive, as well as resultative stative clauses. This
could lead to rampant ambiguity, which is thought to be generally disfavoured
in some evolutionary-based functionalist models of language change. The ulti-
mate victory of have is assumed to have resolved such ambiguities and relieved
be of some of its undesirably heavy functional load.
And yet, many problems beset such a functionalist story, neat though it
may seem at first sight. One major issue is the unexpected proliferation of the
functions of have itself. If it won the ‘battle’ for the domain of expressing the
Introduction 3
perfect because it was less multi-functional and potentially ambiguous, why

do its own functions proliferate over time? Emphasis is laid on the rise of pas-
sives and causatives with have, as well as modal have. A related puzzle has to
do with the emergence of alternative passives and perfects. If getting rid of be-
perfects ‘optimised’ the system by making have the auxiliary par excellence for
the perfect and be the auxiliary for the passive, why would alternative perfects
and passives need to develop at all? Nevertheless, English has evolved passives,
and perhaps even perfects, with get, in addition to more lexical causatives with
get. Even more perversely, numerous varieties of English have seen the birth of
brand new be-perfects after the loss of the original one.
These considerations suggest that functional adaptiveness is to be dis-
preferred as the main factor involved in the developments in question, even
though it is given the greatest prominence in the specialist literature. It is pro-
posed that the frequency of the auxiliaries in the various roles is more like-
ly to have played a decisive part, as corroborated by some corpus evidence.
Frequency might indeed be seen as an indirect measure of how appropriate
speakers find a certain item. Another plausible contributing factor might be
the general meanings of verbs like be, have and get, which easily lend them-
selves to abstraction and reanalysis. Language contact is a further force to be
reckoned with. All of those probably interacted in subtle ways and reinforced
each other.
Chapter 4 deals with the Old English period in more detail. As the have-
perfect was being grammaticalised, the reanalysis of the participle as part of
the verbal group, rather than as an adjectival object complement, is suspected
by many earlier scholars to have first gained a foothold in constructions with
neuter accusative singular objects, which had no overt accusative marking on
the object or on the participle. Since they were ambiguous, it was possible for
the reinterpretation to take place. In other words, due to the lack of overt expo-
nents, it was not morphologically clear whether the participle ought to be seen
as an adjectival complement/modifier or a lexical verb.
To verify this proposal, in Chapter 4 I examine the perfects in a selection
of entries from the Anglo-Saxon Chronicles (MSS. A and E). The aim here and
elsewhere in the book is to study a sequence of coherent and reasonably self-
contained passages of non-translated (where possible) prose which allow for
diachronic comparison of similar samples from different stages in the history
of the language and which are not far apart in subject-matter or genre. Working
through a text (or rather, a compilation of ‘micro-texts’) in its entirety also af-
fords insights into the context and potentially the preferences and idiolects
of individual scribes/authors, as opposed to harvesting a wide range of occur-
rences from large-scale electronic corpora with less emphasis on extensive
4 chapter 1
context or individual usage elsewhere in the language user’s output. The mate-
rial thus collected has been divided into three categories – examples where
overt agreement is present, examples where overt agreement is missing, and
examples where zero marking is expected anyway. Both have- and be-perfects
are taken into account and comparison is sought with passive or potentially
stative constructions with beon ‘be’ and weorðan ‘become’.
The study confirms that there is a substantial amount of expected zero
morphology in the perfect, as well as in (proto-)passives built with beon and
weorðan. However, while the prevalence of expected zero exponence may ap-
pear to have led to an almost complete loss of agreement in the perfect, par-
ticularly evident in the plural, this is certainly not the case in the passive. Both
types of passive construction, with beon and weorðan, demonstrate a situation
in which zero morphology is used in the singular (legitimately with the mascu-
line and neuter, not so with the feminine), whereas -e appears consistently in
the plural (unlike the perfect). This state of affairs is replicated by the ambigu-
ous passives or statives. Crucially, these findings call into question the claims
that cases of expected zero morphology were responsible for the loss of agree-
ment across the board, and possibly point to different rates at which the gram-
maticalisation of the passive and the perfect proceeded. Finally, this chapter
also offers some corroborating evidence that the displacement of perfect be by
have was already underway in Old English and might be at least in part due to
contact with Old Norse.
Chapter 5 charts the further development of the perfect from Middle into
Modern English based on a selection of texts. In view of the primacy of speech
in language change, I have sought texts which would approximate the spoken
medium. The first genre of proven suitability can roughly be summed up as
comedies or comic tales and comprises excerpts from Chaucer (14th c.), a 15th-
century mystery play, a Shakespearean comedy (16th c.), and a Restoration
comedy (17th c.). I additionally examine the late Middle English correspon-
dence of the Paston family and the 17th-century correspondence of Samuel
Pepys. A quick peek is taken at Victorian English, analysing some material from
Dickens. The developments in English are compared to German, represented
in Chapter 6 by a famous 18th-century play by Lessing. Standard German is a
close relative which took a distinctly different course from English, but offers
some curious analogues nonetheless. The examination (albeit cursory) of an-
other Germanic language might help to disentangle genetic ‘predispositions’
from ‘external’ factors such as language contact – English and German started
from a similar starting point, but something must have caused them to come
out rather differently with regard to their be-perfects. At least in part, the di-
vergent developments must be due to the different historical pressures and
contacts that the two languages experienced.
Introduction 5
As expected, the take-home message of Chapters 4, 5 and 6 is that things are

not as straightforward as the foregoing functionalist account might suggest.
The two main pillars of functionalist doctrine are that be-perfects were lost
because be was functionally overloaded and was therefore often ambiguous
(e.g. It was grown could be perfect or passive). Crucially, there are no signs of
ambiguous be-perfects significantly increasing or decreasing over time if ac-
tual usage is monitored. Context is frequently instrumental in disambiguating
too and neither are there that many cases of uncertainty of interpretation to
start with. Moreover, analogous potential ambiguity persists to this day, as in
He was frightened, which can double as passive or stative; this is in fact the
type of ambiguity encountered most often throughout the selection of pas-
sages, but it never seems to have bothered speakers. Based on the case study
of a sample of texts, it is concluded that the reasons for the demise of the be-
perfect in English cannot be straightforwardly attributed to functional load,
but lie elsewhere, including frequency effects (of certain auxiliaries being pre-
dominant in certain roles, as confirmed by statistics), natural cognitive pro-
cesses of reanalysis (most likely prompted by the generic meanings of verbs
denoting existence and possession, but unaided by zero morphology in the
relevant syntactic constructions), and language contact. There are indications
in the texts I examined that the loss of be-perfects was more advanced in the
North of England (the area settled by Viking invaders), which would support a
contact scenario.
Chapter 7 discusses the perfect in the history and structure of Bulgarian.
After a cursory survey of the Bulgarian temporal and aspectual system, I in-
troduce the traditional Bulgarian perfect auxiliary be, which combines with
all classes of verb (including transitives and intransitives). I also outline the
traditional auxiliary roles of have in Bulgarian, involving the encoding of mo-
dality and futurity, but not perfectivity. I then investigate the fairly recent (at
least according to some) formation of a new have-perfect in Bulgarian, as well
as in the closely related Macedonian, which has gone further in that innova-
tive direction, pointing to possible areal diffusion of this grammatical innova-
tion from the west to the east. It is additionally noted that Bulgarian does offer
instances in which a participle in a periphrastic construction (either a long-
standing or an emergent one) might default to neuter singular agreement,
even though this language has no ambiguous zero-marked contexts compa-
rable to those in Old English. This comparison casts extra doubt on the cre-
dentials of ambiguous zero marking as a cross-linguistically valid explanation.
The sloughing off of agreement morphology appears to be a natural conse-
quence of advanced grammaticalisation irrespective of the aid of ambivalent
zero marking. As much is confirmed by the juxtaposition of two typologically
different systems – one with abundant null exponence leading to uncertainty
6 chapter 1
of interpretation (Old English), as opposed to one where there is little compa-

rable uncertainty (Bulgarian/Macedonian). The same reanalysis has occurred
anyway, blind to the type of morphological exponence.
Similarly to English, many earlier researchers have been tempted to come
up with functional accounts for the appearance of the new Bulgarian have-
perfect, which is now supposedly in competition with the established perfect
with be. Following contact with Turkish during the centuries of Ottoman rule
over the Balkan peninsula, the old be-perfect came to specialise in expressing
non-witnessed events (the so-called evidential signalled by the (inconsistent!)
omission of the perfect auxiliary in the third person and the use of a stand-
alone past participle). Besides, the perfect has retained its erstwhile meanings
of resultativeness and completion, conclusivity, etc. (all of them inherent in
the past participle). The old construction has thus become excessively bur-
dened and now have is coming to its rescue. Again, there might be a few snags
for such a chain of cause and effect, some of them having to do with the rela-
tive chronology or the distribution of functional load. Other scholars have in-
stead appealed to language contact with Greek and/or Romance as the source
of the have-perfect in Bulgarian/Macedonian, though this hypothesis is harder
to verify or reject. As pointed out above, the geographical pattern of more fully
grammaticalised have-perfects in Macedonian, especially in certain areas, sug-
gests an eastward areal spread of this construction.
Chapter 8 zooms in on the behaviour of the perfect in a modest selection of
mediaeval and early modern Bulgarian texts. The first piece is a 14th-century
Middle Bulgarian version of the Tale of Troy composed and adapted prior
to the Ottoman conquest. It does turn up some breakdowns in agreement
with periphrastic verbal constructions, which however are in a clear minor-
ity; moreover, some of the instances of the be-perfect already contextually
denote non-witnessed events, though the auxiliary is never omitted and the
synthetic aorist can report second-hand events too (something rare and un-
usual today). No have-perfects were found, nor were any structures that could
be perceived as their precursors. The second sample is from a post-Ottoman-
contact 17th-century damascene book which contains edifying sermons and
the life of a saint. By the time of this representative of 17th-century Bulgarian,
no significant shifts can be detected in the functional load of the be-perfect as
compared to the 14th century, including ‘evidential’ uses. And yet, there are in-
stances of have as a delexical transitive verb, albeit without a participial object
complement.
Overall stability over long periods of time is what characterises both the
English and the Bulgarian data. Like the 14th- and 17th-century samples,
the preface to Paisius’s 1762 Slavonic-Bulgarian History, a centrepiece of the
Introduction 7
Bulgarian National Revival, still allows be-perfects and aorists to narrate non-
witnessed events. What is new here is the frequent omission of the perfect
auxiliary in order to express non-witnessed events with a stand-alone parti-
ciple terminating in -l, i.e. the appearance of evidentials proper. Crucially, the
advent of evidentials has not led to an immediate deployment of have-perfects
in order to alleviate the burden of be. As in the earlier centuries, there are
nevertheless examples of delexical have which might be seen as intermediate
stages in the grammaticalisation of this verb. It would be perilous to posit a
causal relationship between the rise of evidentials and the new have-perfect, in
view of the rather long period of reasonable stability in a system which coped
marvellously without have-perfects for centuries, traditionally relying on be-
perfects (with or without auxiliaries) for mediated reports, as well as in other
situations. That is to say, the material does not show a significant widening or
narrowing in the functional scope of the be-perfect for hundreds of years – we
only see structural omission of the auxiliary and shifting frequencies, but the
range of available functions remains the same. I finally conduct searches in the
Bulgarian National Corpus and mid-20th-century fiction in order to ascertain
how well established the have-perfect is in contemporary Bulgarian. Judging
by my findings, this construction still proves to be rather rare and insufficiently
entrenched.
My general conclusion regarding functional explanations is that, in English
and Bulgarian alike, the grammaticalisation paths of be and have are more
likely to have been steered by their non-specific meanings, which easily lend
themselves to reanalysis and greater abstraction (instigated by contact or
just everyday monolingual interaction), rather than by any perceived ‘func-
tional needs’ or ‘preferences’ of the languages or their speakers. One should
be mindful of the dangers of equating categories from different languages or
even different stages in the history of the same language, since the nature of
each depends on links and oppositions within the synchronic system. And yet,
some thought-provoking parallels could tentatively be drawn. Despite instanc-
es of default agreement, agreement still appears more robust in Present-Day
Bulgarian compared to Old English, which might have to do with the bet-
ter preserved overall morphology of Bulgarian and/or the large-scale loss of
Germanic morphology due to lack of phonological prominence on inflectional
endings. Language-internal and cross-linguistic comparisons show that am-
bivalent marking does not play as important a role of bridging context in re-
analysis and grammaticalisation as has been claimed.
In sum, the two language groups chosen for the present study, Germanic
and Slavic, exhibit both fascinating similarities and palpable differences
with regard to building their perfects. They have been selected as typical
8 chapter 1
representatives of the east/west split in perfect formation (see Drinka 2017),

with a different participial profile in the two groups: the availability of past ac-
tive and past passive participles in Slavic gave these languages the opportunity
to create active periphrastic constructions that were less easily crafted in the
west, where the array of participles is more limited (with a single all-purpose
past participle). Hence, be + past active participle has been traditionally domi-
nant in the east, while there has been a division of labour between be and have
combined with a past participle in the west, though be has lost ground in lan-
guages like English. In the east, be combines with active participles in order to
build perfects and with passive participles in order to build passives, whereas
western languages like French and German employ have + transitive past parti-
ciples for perfects in order to distinguish them from the passive (be + transitive
past participle), with perfect be restricted to intransitives (where the passive is
not an option). Perfect have has by now completely taken over in English and
its analogue is making incursions into Bulgarian and Macedonian, even though
it is, strictly speaking, not ‘needed’. As noted above, all these processes seem
largely unaffected by functional considerations and the reanalysis/grammati-
calisation is most likely due to the inherent properties of such generic verbs,
further prompted by prolonged contact between speakers and cultures on the
Old Continent. The main conclusions and contributions are summarised in
Chapter 9, which also highlights some directions for future work.
chapter 2
Theoretical Preliminaries
This chapter lays the theoretical groundwork on which the rest of the book will
be based. Section 2.1 traces the adoption into linguistics of ideas from biology
and economics, including functional adaptiveness and functional load, teleol-
ogy, as well as non-teleological intentional mechanisms such as the invisible
hand in language change (2.1.2). Problems and challenges for these models
are discussed in Section 2.1.3. Section 2.2 then goes on to outline the central
concepts of grammaticalisation and reanalysis. Also discussed are some of the
features of grammaticalisation that are particularly relevant to the processes
surrounding the rise of the perfect. Special attention is paid to the roles of fre-
quency and language contact in grammaticalisation (2.2.3). Finally, Section
2.3 briefly touches upon the notions of progress and randomness in language
change, asserting that a great deal of change is simply arbitrary; the chapter is
rounded off by justifying the need for a study of this kind (2.3.3). The conclu-
sion summarises the main points and looks ahead to the rest of the book.
2.1 Evolutionary and Functionalist Models of Language Change
2.1.1 Evolutionary Linguistics

Following the celebrated breakthroughs of 19th-century biology, linguists did
not take long to adopt and adapt some of the central notions first pioneered
in the field of biology (see Pinker and Bloom 1990, Newmeyer 1991, Durie 1995:
esp. 301–302, with references, Haspelmath 1999, 2008a: Section 4.5, Croft 2000:
esp. Chapter 2, Bischoff and Jany 2013, Norde and Van de Velde 2016, among
others). Even Charles Darwin himself noted that ‘[t]he formation of different
languages and of distinct species, and the proof that both have been devel-
oped through a gradual process, are curiously parallel …’ (The Descent of Man,
quoted in Croft 2000: 9).
There are several ways in which biological evolution has been linked to lin-
guistics: firstly, there has been debate surrounding the evolution of the human
linguistic faculty or capacity which ultimately resulted in humans being a
speaking species (including Chomsky’s innate Universal Grammar claims –
seminal, revolutionary and still controversial; see Pennisi and Falzone 2016
for recent discussion); secondly, there has been interest in and controversy
around the genetic origins and spread of families of earlier and contemporary

10 chapter 2
human languages (culminating in genealogical classifications and family trees

of related language varieties); thirdly, the theory of biological evolution has in-
spired the construction of evolutionary theories of language change (see Croft
2000, Traugott 2008, with references). Here I will ignore the first two connec-
tions between the disciplines of biology and linguistics, and will only focus on
the way evolutionary theory has been applied to language change. This cross-
pollination of ideas and central tenets rests on the following premise:
Evolution is recognized as a process that occurs with certain types of

entities. The process is probably best understood as it occurs with popu-
lations of biological organisms; that is evolutionary biology. The hypothe-
sis is that language change is an example of the same process, or a similar
process, occurring with a different type of entity, namely language.
Croft 2000: 10
Language change began to be accordingly thought of as originating in a single

innovation which, like a mutation, engenders variation in the linguistic com-
munity.1 After a period of competition between the old and the new variant,
change ensues if the new one is selected and the old one dies out.2 A great deal
of the variation in a linguistic community can be sociolinguistically shown to
represent language change in progress. Variation can reflect the propagation of
a novel variant in the linguistic community, i.e. the selection of this variant as
the conventional sign for the concept in question in the relevant community.
Crucially, not all innovations become changes. It is conceivable that an in-
novation loses the battle and perishes itself, so in the end the status quo might
prevail. In order to make it, an innovation has to be in some sense suited to the
job. According to some evolutionary models, the selection of a competing vari-
ant might ‘cause change to be directed towards increasing fitness’ (Croft 2009:
81, cf. Haspelmath 1999, 2008b: Section 6.2). Innovation, which supplies the raw
material of subsequent variation and potential change, might be consciously
or subconsciously intended to avoid ambiguity and misunderstanding. Such a
view rests on the premise that ambiguity somehow renders a system less fit.
All innovations are in a sense suited to the job, otherwise speakers would
not be motivated to produce them. The ‘birth’ of an innovation is a cognitive
1 Of course, the possibility of multiple independent innovations is in no way precluded.

2 Croft (2000) offers much more detailed refinement and elaboration than the rough sketch
outlined here. Important issues skirted around in my text include who or what the interac-
tors are in those evolutionary processes (i.e. adult speakers, children acquiring their native
language, or both?), as well as the locus of replication: i.e. is the linguistic equivalent of the
DNA an actual utterance, a speaker’s mental grammar, the variants of a linguistic variable (in
the sociolinguistic sense), or something else?
Theoretical Preliminaries 11
process, dependent on various aspects of the context as the speaker sees them
(the microlevel). Selection, however, is not an entirely cognitive process but a
predominantly social one (the macrolevel). The contexts are no longer in the
control of the individual speaker, hence an invisible controller takes over (to
be discussed below; cf. Mathieu and Truswell 2017). Speakers and their utter-
ances can thus be viewed as objects or microlevel actants, while language is a
self-organising system on the macrolevel.
In essence then, a linguistic change is often taken to comprise several stages,
including the actuation or innovation stage, i.e. the beginning of the change
when someone produces a novel form/meaning/sound etc., followed by its
propagation/implementation or selection – the spread in the linguistic com-
munity. In at least some functionalist evolutionary models, the motivations
and mechanisms for innovation are believed to be functional, whereas the
mechanisms responsible for propagation are evolutionary selection mecha-
nisms, and they are social (Croft 2000: 8, 31–32, 38, 166ff., 184–185, with ref-
erences). These social factors, essentially sociolinguistic concepts such as
prestige, social status or identifying with a social group, have been compared
to the ecological factors in biology. Propagation can be seen as the adoption
of a convention in society. Other functionally-minded scholars, however, seek
functional explanations for the selection stage too, assuming an analogy be-
tween functional explanations in linguistics and adaptive explanations in evo-
lutionary biology (see Haspelmath 1999). In defence of this latter view, it could
be stressed that adoption is also a cognitive process to a certain extent because
the new element needs to be integrated in the individual speaker’s grammar.
As a supporter of the former hypothesis, Croft (2000: 39) counters that
the empirical evidence indicates that linguistic selection is governed

largely if not exclusively by social forces that have little or nothing to do
with functional adaptiveness for communication.
Croft (2000: 195, n. 10) backs up this statement by pointing to the existence of
social situations which might favour linguistic changes that are not considered
to be ‘optimal’ or ‘natural’ on functional grounds. This furnishes evidence that
functional considerations are unimportant in selection (propagation). Rather,
complex older forms, or elaborate ‘dysfunctional’ innovations, might end up
surviving for a long time or even spreading through a speech community.3
3 The dividing lines between the stages are often blurred. For instance, we could think of the
gradual displacement of be by have (discussed below in the context of the English perfect) as
both propagation and also innovation (or a series of micro-innovations). This is highlighted
12 chapter 2
2.1.2 The Concepts of Function and Functional Load, the Invisible Hand
and Teleology
As noted above, fitness in language has to do with function. If a construction
has too many functions, i.e. if it has too great a functional load, it will be too
ambiguous and might be subjected to change (see Durie 1995: 276ff.). As Lass
(1980: 65) reminds us, the idea that function is a significant factor in change
largely goes back to the pioneering work of the Prague School, whose pro-
ponents tended to see change as ‘purposeful’ or ‘teleological’. Of course, the
intellectual roots of functionalism can be traced back much further, as synthe-
sised by Givón (2013). Function is often associated with Humboldt’s Universal,
which requires that (ideally) one meaning should correspond to one form.4
Croft (2009: xiii) states the functionalist position quite succinctly:
one cannot deny the existence of grammar, that is, knowledge of linguistic
conventions in a speaker’s mind; one cannot disentangle this knowledge
from the act of using language; language function influences language
form through the dynamics of language use and language change.5
Speakers can hardly be said to create optimal grammars intentionally; rather,

they consciously, rationally and purposefully select among available variants
and create new ones, opting for the ones that best suit their needs (Haspelmath
2008b: Section 6, with references; cf. Durie 1995: 276, 290). So in effect, func-
tionally optimal grammars are the subconscious macrolevel result of con-
scious micro-choices, an effect compared to an invisible hand in seminal work
by Keller (1994). Both natural causes and end-oriented intentions can be ac-
commodated in this model too (see Pinto de Lima 1995 and discussion below).
in the framework of Construction Grammar (see Traugott and Trousdale 2010, 2013, inter
alia).
4 See Lass (1980: 71–75) for counterexamples as well as further references. Itkonen (2013) con-
tains modern echoes of Humboldt’s view, and argues against them.
5 Durie (1995) explores what is meant when we say that a linguistic structure has a certain func-
tion. On the divide (and bridges) between formal and functional linguistics, see Newmeyer
(1991), Hill (1993), Beaugrande (1994), Łęcki (2010: 44ff., with references), Vincent and Börjars
(2010), Bischoff and Jany (2013), Hengeveld et al. (2017). In a nutshell, functionalists see com-
munication as the primary raison d’être of language, to which it is geared and which in turn
shapes it, while formalists view the communicative function of language as largely irrelevant
to its analysis. Haspelmath (2008a: esp. Section 4, 2018) provides a good overview of current
functional explanations in typology, focusing on ways of accounting for language universals
and comparing the functionalist approach to generative syntax, including the Principles and
Parameters framework and Optimality Theory. Bischoff and Jany (2013) and Hengeveld et al.
(2017) are collections of recent work within the functionlist paradigm.
In Keller’s theory, when speakers innovate, they do not consciously intend

to change their language. When they use language, they might be trying to
achieve certain goals in the pursuit of which innovations, i.e. violations of
pre-existing linguistic conventions, might occur. If an innovation results in a
language change, the change will be the unintended causal effect of an inten-
tional human action. Following Popper, Keller (1994: 57) calls this a phenome-
non of the third kind, distinguishing it from artifactual phenomena (occurring
by intentional human design and corresponding to teleological mechanisms
of language change), as well as from natural phenomena (purely natural phe-
nomena where human will plays no part, exemplified roughly with ‘mechani-
cal’ views of language change). The propagation of an innovation in the speech
community is compared to a process controlled by an invisible hand, which
might be seen as analogous to evolutionary drift:
an invisible hand explanation explains its explanandum, a phenomenon

of the third kind, as the causal consequence of individual intentional
actions which are based on at least partially similar intentions.
Keller 1994: 71, quoted in Croft 2000: 59
Nevertheless, it is exceedingly tempting, sometimes convenient, or perhaps

just less boring, to personify languages.6 According to some evolutionary ac-
counts, languages, not unlike living organisms, can be seen as ‘volitional
agents’ somehow constantly ‘striving’ to attain greater fitness (see Lightfoot
2002, with references). This invests them with some inexplicable will or single-
mindedness of their own. However, as Croft (2000: 4) correctly admonishes,
‘languages don’t change; people change language through their actions’, as al-
ready discussed above. Lass (1980: 113) warns that language should be seen as
a ‘cultural institution’ or a ‘cultural artefact’, and since it is not a thing of the
natural world, evolutionary explanations simply do not work (see below). In
other words, languages are certainly not biological systems or living organisms,
and this might invalidate at least some of the parallels with biology (cf. Dresher
and Idsardi 1999 for more scepticism).
It might be a popular misconception of evolutionary processes that they are
necessarily teleological and should be explained by invoking the idea of an or-
ganism that aims towards some goal. Croft (2000: 64) does not see selection as
6 Whenever such personification occurs in the rest of this book, the caveats from this section
should kick in. For instance, wherever I have used loose metaphorical turns of phrase such as
to recruit an item for grammaticalisation, no purposeful single-mindedness or grand designs
on the part of the language or its speakers have been implied.
14 chapter 2
a teleological process, since the goal of the process needn’t be built in – what is
needed is simply a mechanism that causes different interactors to end up with
differential survival rates. Adaptation is such a mechanism – in it certain prop-
erties of a given organism interact with properties of the environment in such
a way that the likelihood of the organism’s survival is increased or decreased.
Adaptation thus ultimately results in directed selection, whereby a greater pro-
portion of organisms in possession of adaptive properties survive.
On this view, adaptation is neither teleological nor deterministic, since
being in possession of adaptive properties merely increases the chances of
survival, but without guaranteeing survival. However, Croft still believes that,
in a population of ‘organisms’, given enough time, the laws of probability will
direct change in favour of the better adapted organisms. Although this stance
is more sophisticated, in the spirit of the literature on self-organising complex
adaptive systems, it also seems to indirectly amount to a teleological explana-
tion, albeit not a deterministic one. Even when you see directional language
change as probabilistic, rather than deterministic, you still attribute to it the
same functional causal mechanisms.
A careful treatment will thus distinguish different intentional mechanisms.
Teleological intentional mechanisms assume that a speaker innovates in order
to change the linguistic system – essentially Keller’s artifactual phenomena
whereby the linguistic system is shaped by the speaker to have the structure
that it does, as well as to change the way it does. Croft rejects those. He is in fa-
vour of non-teleological intentional behaviour – the speaker produces an inno-
vation because he or she is trying to achieve some other goal in using language.
Intentional processes do not necessarily aim to produce linguistic innovations.
Croft (2000: 65) states that he ‘will use the term intentional to describe mecha-
nisms that are not teleological but involve the intention of a speaker to achieve
some other goal in language use’. He refers to teleological mechanisms as ‘sys-
temic functional explanations’, reserving the term ‘functional explanations’ for
non-teleological intentional mechanisms.
There exists yet another type of causal mechanism proposed for language
change which is neither teleological nor intentional, since language change
is not seen as an intended means aimed at achieving some other goal in lan-
guage use. On this view, a change simply happens as a result of the process
of utterance production. This phenomenon has sometimes been labelled ‘me-
chanical’, especially when applied to production in phonetic articulation. Such
explanations could be proposed for higher-level cognitive processes in the use
of language as well. This kind of causal mechanism is therefore often labelled
as broadly non-intentional (corresponding to Keller’s natural processes). The
distinction can be clarified as follows:
The difference between intentional and nonintentional mechanisms for

language change is whether or not the innovation is a means towards
the intended goal. In intentional mechanisms, the innovation is a means
towards the intended goal; in nonintentional mechanisms, it is not.
Croft 2000: 66
Rejecting teleological mechanisms as highly suspect based on the evidence

he has reviewed, Croft finds it hard to rule out either intentional or non-
intentional mechanisms a priori, since neither is teleological and both can
participate in replication and selection. Furthermore, they might be hard to
distinguish, especially at the level of grammar.
Typical intentional mechanisms of altered replication, i.e. innovation, in-
clude expressiveness or creativity, avoidance of ambiguity or misunderstand-
ing, as well as economy of expression. Non-intentional mechanisms that have
been proposed in the literature include speech errors, articulatory phonetic
overshoots or undershoots of a target, as well as auditory reanalysis (see Croft
2000: 74–78, with references). There have been suggestions that features of lan-
guage of which a speaker is less aware or less in control, such as sounds, will
tend to change mechanically, whereas levels such as morphology and syntax,
of which speakers are arguably more aware, might tend to obey the laws of
intentional mechanisms (Vincent, cited in Croft 2000: 78). Croft (2000: 78),
however, proposes that many grammatical changes might be motivated by a
non-intentional reanalysis of the mapping between grammatical form and
conventional meaning or function.
A brief example will shed some light on the difference between teleological
and non-teleological, intentional and non-intentional. Sometimes, teleology
might appear to be at work where in fact it is not, strictly speaking, directly
responsible for what has happened. Teleological mechanisms are defined as
mechanisms in which innovations happen in order to change or preserve the
linguistic system, either because this is the goal intended by speakers, or be-
cause this is some mysterious law that governs linguistic systems in the course
of their evolution. An example of such a teleological process comes from the
deletion of intervocalic /s/ in Ancient Greek. This deletion was apparently
prevented from taking place when /s/ was the sole marker distinguishing the
present and the future forms of verbs (McMahon 1994: 332, cf. Croft 2000: 66–
67). This teleological mechanism therefore aims to avoid homonymy and to
preserve the distinctions of the language system. However, the avoidance of
homonymy can actually be described as intentional rather than teleological, if
we recognise that it takes place in order to avoid ambiguities, breakdowns and
misunderstandings in communication.
16 chapter 2
The subtle difference between the two hypotheses lies in their different as-
sumptions about what speakers actually ‘wanted’ – to preserve the linguistic
system for its own sake or to avoid misunderstanding (and thereby indirectly
preserving the linguistic system). In other words, does avoidance of homon-
ymy happen for the sake of propping up the system as such or does it follow
from the intention of interlocutors to avoid being misunderstood? This is an
example of how a teleological and an intentional mechanism would basically
amount to the same explanation, despite the fine line that divides them.
2.1.3 Problems and Challenges

There are a number of problems for teleological mechanisms. Most of these
would indirectly apply in equal measure to their more sophisticated cousin,
intentional mechanisms. For instance, if the goal of a change is to restore or
introduce symmetry in the system, one might ask: if it is so desirable, why was
symmetry disrupted in the first place? Perhaps due to other, non-teleological
changes? Besides, languages are rife with homonymy, allomorphy and asym-
metry, as well as ambiguity of words or grammatical morphemes and con-
structions, but this does not seem to get in the way of efficient communication
(see Wasow et al. 2005, Piantadosi et al. 2012, Fortuny and Corominas-Murtra
2013, Winter-Froemel 2013). Going back to the Ancient Greek example above,
it should be noted that many languages do not distinguish present and future
(e.g. Japanese), but seem to get by perfectly fine nonetheless. Even Greek al-
ready had homonymy between indicative and subjunctive active first-person
singular present forms; moreover, the intervocalic /s/ also survived in the first
aorist, although it was not strictly needed in order to distinguish the aorist
from other forms (Lass, cited in Croft 2000: 68). Empirical studies furnish no
consistent evidence to support teleological mechanisms which would lead us
to believe that segments are less likely to be lost if they mark an inflectional
category (Croft 2000: 68, with references). Quite the contrary – the histories
of numerous languages demonstrate that they have lost their inflections for
various categories.
Similarly, a systemic-functional, teleological account would predict that
new forms will tend to be grammaticalised where there are gaps in the sys-
tem. Such is the case in English: you, which was originally plural (and formal/
polite), can now be either singular or plural as a result of the loss of singular
(and familiar) thou (see Baugh and Cable 2013: 235–236). This has purportedly
led to the grammaticalisation of relatively recent colloquial plural forms such
as you guys, you all/y’all, etc. Those are said to have arisen in order to fill the
gap and restore symmetry in the pronominal paradigm. However, the empiri-
cal facts are that such grammaticalisation processes occur independently of
whether or not a gap exists in the system. On the one hand, there are many pro-
nominal systems worldwide that have number gaps but do not show any signs
of changing. On the other hand, there are instances of new plural forms aris-
ing in addition to pre-existing ones, i.e. where no gaps existed (see Croft 2000:
69–70, with references). So linguistic systems are not always as symmetrical as
we might imagine; rather, they tolerate a high degree of messiness. To be more
accurate and careful, we should say that speakers exhibit a high level of ‘toler-
ance’ to ambiguity. Croft himself (2000: 71) concedes that
[a]lthough not being misunderstood may contribute to language change

as an intentional mechanism for innovation, avoiding homonymy plays
at best a marginal role.
As one of the most outspoken critics of function-based accounts of change,

Lass (1980: 69) sees functional explanations as ‘post hoc (in the sense of being
totally non-predictive)’. He states that ‘any teleological argument in which a
goal is not identifiable independently of the behavioural manifestations lead-
ing to it is not explanatory’ (Lass 1980: 83; cf. Haspelmath 1999: Section 4, as
well as Dresher and Idsardi 1999). Lass (1980: 90, 92) rejects functional expla-
nations, because there are no principled definitions of ‘dysfunction’, and we
therefore cannot define what counts as a ‘function’ in the first place. This relies
on the assumption that there are no dysfunctional languages, and in all their
synchronic states, natural languages which are not dying are perfectly well
suited for what is meant to be achieved with them. For those reasons, it may
be concluded that functional load cannot qualify as an explanatory concept.7
In a similar vein, Morris Halle (1975: 528, quoted in Lass 1980: 136) writes
that, since ‘language is not, in essence, a means for transmitting information,
it is hardly surprising to find in languages much ambiguity and redundancy, as
well as other properties that are obviously undesirable in a good communica-
tion code’ (cf. Pinker and Bloom 1990: 719, Newmeyer 1991: 6, 24). Even those
who believe that language is primarily designed to transmit information would
agree that this is not the sole use it is put to. Lass (1980: 143) sums up that, given
the fact that language is many things besides a ‘means of communication’
(including a formal system, a plaything, a set of interactions between partially
open systems, etc.), it is little wonder that ‘function’ as a motivation for lan-
guage change is not an explanatory notion.
7 See Lass (1980: Chapter 4) for more philosophical arguments against the explanatory power
of functional load.
18 chapter 2
Further evidence against functionalist teleology comes from Newmeyer

(1991: 26), who reports instances of phonological change without any func-
tional motivation whatsoever. Labov’s (1987: 319) study of near-mergers sup-
ports this point. In a near-merger, previously distinct phonemic contrasts are
neutralised perceptually, but not physically. For instance, in Philadelphia, the
stressed vowels in the pairs merry ~ Murray, ferry ~ furry have become so close
to each other that speakers of the Philadelphia dialect cannot distinguish them
reliably. However, the spectrographic evidence confirms that the two vowels
are distinct after all. It is now believed that such near-mergers have occurred
frequently in language history and are responsible for apparent mergers in pro-
nunciation which have later been ‘undone’. The existence of near-mergers has
led Labov to conclude that the communicative role of phonemic contrasts can
be suspended for quite some time, and yet this does not disturb the integrity of
the word classes and the systemic oppositions in which they participate. While
phonemes certainly do distinguish words, their historical development is not
narrowly motivated by this communicative function, which justifies joining
the anti-functionalist camp.8
With a focus on change resulting in grammaticalisation, and asking them-
selves why grammaticalisation should occur at all, Bybee et al. (1994: 297–298)
aver that their approach to explanation does not appeal to motivations or to
functional teleology. Grammaticalisation can roughly be defined as the pro-
cess whereby a lexical item becomes a grammatical item or a grammatical
morpheme in the context of a particular construction (see Section 2.2 below).
Essentially, languages do not tend to develop grammatical categories because
they really need them, as has frequently been claimed before. A number of ar-
guments can be adduced against this need-based position. In the first place, it
makes no sense to state that a language needs a certain grammatical category,
because most such categories are doubtless not universal. While many languag-
es possess a future tense as a grammatical category, many simply do not, which
gives us no reason to disparage them as dysfunctional, because their speakers
get by perfectly fine (as already stated above). Secondly, two or more markers
which serve similar communicative goals could potentially arise. This is not at
all infrequent, as we shall see in the case studies below, and it should not occur
8 Apart from near-mergers, Ringe and Eska (2013: 101–103) report studies that demonstrate
the failure of functional load to account for mergers in phonology more generally. As they
conclude, “[t]hat confirms the consensus of opinion among rigorous historical linguists that
functional load has no demonstrable effect on mergers and should be disregarded in framing
hypotheses about sound change (Ringe and Eska 2013: 102)”. Below, I extend this to morpho-
syntactic change.
if grammaticalisation were driven by communicative needs only. Thirdly, by

their very nature, the inflections produced by grammaticalisation, and inflec-
tions more generally, are often redundant and not strictly necessary in context,
and speakers can often cope without them. Finally, Bybee et al. (1994: 298) re-
port case studies which demonstrate that change can in fact damage fitness
and foster ambiguity: ‘the spread of the Conditional marker in Spanish at the
expense of the Imperfect Subjunctive actually creates ambiguity rather than
increasing expressiveness’ (emphasis original). The same applies to the loss of
Bulgarian aorist passives with биде/bide ‘was’ in favour of the more ambigu-
ous беше/beše ‘was’ – aorist or imperfect (Gerdžikov 1984/2003: 243 fn. 3). An
examination of data in the same spirit will be undertaken in the chapters to
follow.
The non-teleological nature of grammaticalisation can be attributed to the
mechanisms of change involved in it (more on these mechanisms is given in
the next section). Firstly, the semantic, syntactic and other changes along the
grammaticalisation path are not individually directed toward creating a gram-
matical morpheme. In other words, ‘when a verb “want” is first used with a
verb phrase complement, it is not because the language has the ultimate goal
of creating a future tense several centuries later’ (Bybee et al. 1994: 298). Even
though the addressee may easily and naturally infer an intention from an ut-
terance that expresses desire (such as I want to), the addressee can hardly be
said to do so with the goal of creating a new future tense. In turn, a prediction
inference which may arise from a statement of intention cannot be motivated
by a necessity for a language to have a future tense either. In the light of these
arguments, Bybee et al. (1994: 298) see grammaticalisation as a phenomenon
which is much more mechanistic or automatic than functional, stressing that
‘the relation between grammar and function is indirect and mediated by dia-
chronic process’. The conclusion they reach is that the processes that moti-
vate grammaticalisation occur in the actual use of language for their own sake.
The cumulative effect of those processes just happens to accidentally lead to
the development of grammar, with the new constructions (innovations) then
propagated in the linguistic community in line with the sociolinguistic and
other factors noted above. This non-teleological view of the evolution of gram-
mar will be empirically confirmed later in this book for the two languages that
I consider.
2.1.4 Interim Summary and Preview

To sum up, it is still a widely held belief (though also rightly disputed) that
language change might be mysteriously and teleologically directed towards
achieving greater fitness (mostly through the subconscious involvement of
20 chapter 2
speakers, ignoring prescriptive interventions or language planning). Greater

‘linguistic fitness’ is often associated with avoiding ambiguity and misunder-
standing, thereby ultimately maintaining a manageable functional load. If the
functional load of a construction gets out of hand, it could lead to rampant
ambiguity and therefore potentially damage ‘fitness’. Below, I will demonstrate
that teleological and intentional explanations involving the avoidance of am-
biguity as a result of excessive functional load do not prove fully satisfactory if
one traces the grammaticalisation of the perfect in English and Bulgarian. The
development of the perfect is a textbook example of grammaticalisation and
reanalysis, and explanations invoking ambiguity and excessive functional load
have been put forward in numerous influential accounts. However, they leave
many questions unanswered and are in need of reassessment in view of the
empirical facts. Before I move on to the perfect in subsequent chapters, in the
next section I flesh out the basics of grammaticalisation and reanalysis.
2.2 Grammaticalisation and Reanalysis
2.2.1 Grammaticalisation
Grammaticalisation studies had their beginnings in the 18th century. The term
itself is credited to Meillet (1912: 131), dating back to the early 20th century.
However, this field of research really picked up only in the 1970’s, when Givón
(1971: 413) famously observed that ‘[t]oday’s morphology is yesterday’s syn-
tax’ (see Heine 2003: 575–578, 599 n. 4, Hopper and Traugott 2003: Chapter 2,
Traugott 2003: 624, 646 n. 2, 2010, Łęcki 2010: 3ff., Macleod 2012: 5, for an over-
view of the history of research into grammaticalisation).
Grammaticalisation can be regarded as a process (mechanism) of innova-
tion in grammatical change. In it, constructions which are made up of specific
lexical items develop grammatical meanings, which leads to reinterpreting the
lexical items as serving grammatical functions (Croft 2000: 156, cf. Traugott
2002: 22ff., 2003: 645, Macleod 2012: 5ff.). In addition, grammatical forms can
give rise to the development of even more grammatical forms, i.e. ones even
more firmly embedded in the morphosyntactic system. It is also important to
stress that grammaticalisation does not happen to items in a vacuum but to
entire constructions in context, including not just the syntactic environment
but an all-encompassing idea of context as situation and discourse, paradig-
matic relationships, etc. (see Bybee 2003: 602–603, Heine 2003: 575, 580–581,
Hopper and Traugott 2003, Traugott 2003, 2008, Traugott and Trousdale 2013:
Chapter 5). For instance, English has the constructional schema in (1).
(1) [[movement verb + Progressive] + purpose clause (to + infinitive)]
Several movement verbs can be inserted in this template: I am going/travel-

ling/riding to see the king. Out of those, only go has experienced grammaticali-
sation as a future marker in this specific syntactic context.9
Grammaticalisation is often said to be a cover term for a set of inter-related
phonological, morphosyntactic and semantic/pragmatic changes. These in-
clude a wide range of processes: attrition or reduction and erosion of pho-
nological substance, a rigidification of word order and loss of independent
syntactic status or of morphosyntactic properties (e.g. free morphemes under-
going cliticisation and subsequently becoming affixes), context generalisation
(i.e. use in new contexts), extension of the semantic range, semantic bleach-
ing, etc.
In order to illustrate this umbrella process, I take up the well-known ex-
ample of the English future with go.10 As mentioned above, the construction
initially contained an infinitival subordinate clause of purpose, as in I am going
(into town) (in order) to see my grandparents. This syntagm gradually evolved
into a marker of futurity, as in I’m gonna see my grandparents. The phonologi-
cal attrition that accompanies grammaticalisation has resulted in the coales-
cence of originally independent words into gonna. The erstwhile subordinate
clause of purpose has become obligatory and the infinitival to has lost its in-
dependent status, having fused in gonna. Semantically and pragmatically, the
restriction of what was originally a verb expressing motion to motion contexts
and animate subjects has been lifted, and now no motion or animacy/volition
on the part of the subject is implied (as in It’s gonna rain). The meaning has
shifted from denoting motion and purpose to denoting future.
The first step in a grammaticalisation process could be referred to as a
periphrastic stage, at which a new periphrastic construction is recruited for a
9 This view is highly compatible with synchronic models of language analysis such as
Construction Grammar (with its definition of construction as a ‘form-meaning pairing’),
LFG, HPSG, as well as hybrids such as Sign-Based Construction Grammar (SBCG) (see
Hollmann 2003: 17–21, Traugott 2003: 625, 2008, 2010: 277–279, Bergs and Diewald 2008,
Sag 2010, Traugott and Trousdale 2010, 2013, Vincent and Börjars 2010, Hoffmann and
Trousdale 2013). Trousdale (2008, with references) and Coussé et al. (2018), among others,
provide diachronic discussion of grammaticalisation in Construction Grammar. Bergs
and Diewald (2008), Hoffmann and Trousdale (2013) and Traugott and Trousdale (2013)
deal with constructions and language change more generally.
10 The discussion here is primarily based on Croft (2000: 156–157) – cf. Danchev and Kytö
(1994), Heine (2003: 580, 584–586), Traugott and Trousdale (2013: 102, 105–107, 116ff., 217ff.),
among others.
22 chapter 2
particular function. It has been argued that this might be driven by expressivity –
a speaker comes up with a novel periphrastic way of expressing a function
which stands out and is thus more effective (see Croft 2000: 159; cf. Heine 2003:
578 and the preceding discussion of the place of expressivity in evolutionary
models). There has been a long tradition of claims that expressivity could be an
important driving force behind grammaticalisation. It was discussed as early
as Meillet (1912, cited in Łęcki 2010: 15), who attributed the emergence of new
grammatical constructions to the loss of the vividness (i.e. the bleaching in
modern parlance) of the older, well-established ones, coupled with the desire
of language users to be expressive. Traugott and Trousdale (2013: 124ff.) provide
critical assessment of such proposals, remarking that speakers are often un-
aware that they might be initiating or propagating a change.
Another related mechanism offered to explain the rise of periphrasis is the
avoidance of misunderstanding (see Heine 2003: 578 and Section 2.1.2 above).
All too often, in fact, it has been proposed that speakers introduce a certain
amount of redundancy in interaction in order to avoid being misunderstood
(Croft 2000: 159, with references). As Keller (1994: 109) puts it, ‘to increase re-
dundancy beyond the acoustically possible, one must use lexical means’, that
is, a periphrastic construction. As the subjunctive inflections in English be-
came eroded to a point where they were hardly distinguishable, speakers re-
sorted to expressing the same non-factual semantics with modals instead.
Semantically, when a linguistic expression A is recruited for grammaticali-
sation, it acquires a second pattern of use – B; there is then an overlapping
stage of ambiguity between A and B, before A is potentially lost and only B
remains (Heine 1993: 48–49, 2003: 579, 587, 589–590, Macleod 2012: 8–11). So
ambiguity is important in this overlap model, as it enables the shift of mean-
ing. Such overlaps will be pivotal in discussing the morphosyntactic marking
of Old English periphrastic constructions in the chapters to follow.
Grammaticalisation can be demonstrated to involve pragmatic inferenc-
ing, also labelled as ‘context-induced reinterpretation’ (see Croft 2000: 160–161,
Traugott 2002, 2003: 634–636, 2010: 279–280, Heine 2003: 580, 586–588, 591–
592; cf. Levinson 2000: Chapter 4, Traugott and Trousdale 2013: 26 fn. 19, 217ff.).
There is normally a continuum of situation types which bridge the gap be-
tween the original meaning of a construction and the new meaning or function
which has arisen as a result of grammaticalisation (with significant overlaps).
Pragmatic inferencing allows speakers to gradually extend the meaning from
one situation type to another (contiguous one). In Henry is going to town/to
the library,11 we have a purely allative situation type of motion from point A
11 The examples in this paragraph come from Croft (2000: 161), sometimes with only cos-
metic modifications.
to point B, but when the subject is human, it is easy to infer an intention on

the part of the agent. If intention is inherently associated with the meaning of
prepositional and infinitival to, an intended action can be inferred from the
spatial destination, as in I am going to eat. In this latter eating example, future
time reference is contextually inferred in addition to the expression of inten-
tion and spatial destination. If you say I am going to do my best, there may be no
motion meaning and the (intentional) future meaning might be reanalysed as
an inherent property of the construction. Prediction is a contextual inference
in I am going to make you happy, but it can in turn be reanalysed as inherent
and applied to inanimate subjects, as in The rain is going to come. In It’s gonna
rain, the meanings of motion, spatial destination and intention have been
completely lost – the speaker is making a prediction concerning the near fu-
ture. The pragmatic inferencing involved in the shift from resultative to perfect
in the history of English is traced by Carey (1995) and I return to it in Chapter 3.
From among the numerous grammaticalisation tendencies uncovered by
various scholars, I shall single out some of the principles formulated by Hopper
(1991: 22; cf. Tagliamonte 2000, Heine 2003: 588ff., Łęcki 2010: 3ff., 12), as they
are relevant to my discussion of the diachronic developments in English and
Bulgarian. Some of the properties of grammaticalisation are roughly as follows:
– Layering: new layers or methods of encoding meanings will keep emerging
in a functional domain as a result of grammaticalisation. Older layers, how-
ever, will not necessarily be discarded; they may conceivably coexist and
interact with the new ones.
– Divergence: when a lexical form grammaticalises, say, into a clitic or an affix,
the original lexical item may continue to exist as an autonomous element.
– Specialisation: at a given stage, a functional domain might have a variety of
forms with different semantic nuances. As a result of grammaticalisation,
this variety of formal choices will typically narrow and the smaller number
of selected/grammaticalised forms will assume more general grammatical
functions/meanings.
– Persistence: when a form passes from the lexical to the grammatical do-
main, some traces of the form’s original lexical meaning(s) will tend to ad-
here to it, and traces of its lexical history might be reflected in constraints
on its grammatical distribution.
– De-categorialisation: forms subjected to grammaticalisation will tend to
lose or neutralise the morphological markers and syntactic properties
typical of full/lexical nouns and verbs, thereby acquiring properties typi-
cal of secondary categories such as adjectives, participles, prepositions,
auxiliaries etc.
Layering in the tense/aspect domain can be illustrated with three different stra-
ta employed to this day in order to refer to anteriority in English: periphrasis
24 chapter 2
(I have used it – newest layer), affixation (I admired it – older layer), ablaut

(They sang – oldest layer) (Hopper 1991: 24, cf. Łęcki 2010: 12–13). English ab-
laut originated from Indo-European vowel gradation and is therefore archaic.
The weak preterite dental suffix (-ed and allomorphs thereof) is a Germanic
innovation, hence later than the morphophonemic alternation known as ab-
laut. The periphrasis with have probably emerged a short time before the West
Germanic languages were first recorded in the 7th and 8th centuries, though
the process was by no means complete. Bulgarian too has an aorist which can
be traced all the way back to Proto-Indo-European, a perfect with be that can-
not be dated back further than Proto-Slavonic, as well as a have-perfect which
is arguably a Modern Bulgarian creation (see Duridanov et al. 1991: 288, 299,
Ganeva 2015: 8–9, Drinka 2017: 288, as well as Chapter 7 herein).
Divergence is illustrated by the grammaticalised indefinite article a(n) of
Modern English, versus the non-grammaticalised numeral one, both of them
going back to the same Old English numeral ān. In a similar vein, the gram-
maticalised motion-based future can undergo attrition to gonna, but lexical
go cannot (e.g. *He’s gonna the park for He’s going to the park), so the two have
clearly diverged (see Heine 1993: 51–52, Hopper and Traugott 2003: 1–3). The
same has happened to had: He’d left the table [perfect], but *He’d a green car
[lexical] (see Ammann 2005: 267 fn. 25), though present tense have might be
cliticised even when it denotes possession. Old English willan ‘want’ and its
Old Bulgarian/Old Church Slavonic translation equivalent хотѣти/xotĕti were
recruited to serve as future markers, but their meanings of willingness and in-
tention still persist (to some extent) in Modern English and Modern Bulgarian
(see Hopper 1991: 29, Heine 2003: 584–586, Łęcki 2010: 16). Similarly, English
have and its Bulgarian counterpart can mark the perfect, as well as modality,
in addition to retaining their lexical meanings of possession and association.
There was also competition between alternative auxiliaries for the perfect
(have v. be) – competition being customary in most types of language variation
and change. Specialisation comes about when one of these ‘rivals’ ousts the
other(s) (see Smith 2001 and later chapters herein).
When decategorialised as part of grammaticalisation, original nouns no
longer identify participants in an event, and original verbs no longer report
events. Thus, auxiliary have no longer denotes an event separate from that of
the ‘main’ verb. Finally, while charting the course of the perfect in the chap-
ters to follow, we shall see that originally adjectival past participles came to
shed their adjectival morphology, which enabled them to be perceived as more
verbal in character, arguably demonstrating decategorialisation or at least a
functional shift.
Importantly for present purposes, it has been observed that a grammaticali-

sation cline or path can branch off – for example, a verb indicating possession
can end up expressing obligation, futurity, as well as the perfect (see Heine
2003: 590–591, Łęcki 2010: 43, with references). This will become evident below
from my Latin, English and Bulgarian data. Such a phenomenon of divergence
has sometimes been labelled polygrammaticalisation.
Last but not least, as hinted above, some scholars do not believe in the ex-
istence of grammaticalisation as a unified process, instead saying that it is
merely a cover term for a ragbag of syntactic, semantic and phonetic chang-
es which are all capable of applying independently (see Traugott 2002: 25ff.,
Heine 2003: 583–585, 600 n. 6, Łęcki 2010: 44–45, with references). However,
the phonetic, morphosyntactic and semantic/pragmatic developments that
make up grammaticalisation are not really independent and isolated in the
context of grammaticalisation – rather they are interrelated and trigger each
other. One of them is reanalysis, which is examined in the following section.
2.2.2 Reanalysis in the Context of Grammaticalisation

Syntactic or structural reanalysis involves a realignment of the syntactic
structure of a construction in terms of its constituency and/or dependency.
Croft (2000: 120, 164) describes it as ‘the introduction of an alternative syntac-
tic structure for a construction, even if there is not any visible change to the
construction (i.e. an innovated structure)’. However, it is hard to establish that
reanalysis has occurred until one sees the production of a novel, previously
unattested grammatical pattern made possible by the reanalysis. For instance,
the original allative [I am going [to see the king]], with its subordinate non-
finite clause of purpose, has been rebracketed to give the following constitu-
ent structure: [I [am going to] see the king], with going to as (part of) a unit
marking futurity and reducible to gonna (Hopper and Traugott 2003: 3, 68–69,
Trask and Millar 2015: 134). It was possible to establish that the reinterpretation
had occurred only when utterances appeared where an allative reading was
ruled out (as in It’s going to rain). A related concept, form-function reanalysis,
involves remapping the relations between form and function in combinations
of syntactic units and the semantics they express.
As already indicated on several occasions, reanalysis is deemed impor-
tant in some theories of grammaticalisation (see Hopper and Traugott 2003:
Chapter 3, Traugott 2008, 2010a: 280–282, Traugott and Trousdale 2010, Macleod
2012: 7, Itkonen 2013: 49). There have been claims that all instances of gram-
maticalisation necessarily include reanalysis; a more moderate stance taken
by other linguists is that some instances of grammaticalisation might involve
26 chapter 2
reanalysis, but not all (see Heine 2003: 592–593, 600 n. 6, Hopper and Traugott
2003: 58ff., 69, Łęcki 2010: 32–33, Traugott and Trousdale 2010, with references).
It has been contended, notably by Haspelmath (1998), that grammaticalisation
and structural reanalysis are two distinct processes which should in principle
be kept apart. According to Haspelmath, genuine structural reanalysis involves
no loss of syntactic autonomy or of phonological substance, and it is poten-
tially reversible, unlike grammaticalisation, which involves loss of autonomy
as well as substance, and is usually unidirectional.12
So, some scholars are inclined to see reanalysis as synonymous, or near syn-
onymous, with grammaticaliastion. Others maintain that the two are distinct
and grammaticalisation involves, but is not coextensive with, reanalysis. That
is to say, reanalysis is a mechanism for grammaticalisation, as well as for other
changes. Yet others believe that grammaticalisation does not require reanaly-
sis and the two are not necessarily related. I tend to side with the second group.
Form-function reanalysis is likewise a pervasive mechanism of change
and it is often crucial in innovation, its causes being an integral part of
communication:
the complexity of the morphosyntactic encoding of meaning, combined

with the variability of the experiences communicated on particular oc-
casions of use, may lead to the reanalysis of the mapping between mor-
phosyntactic form in constructions and the semantic content that they
denote.
Croft 2009: 82
This entails potential ambiguity for the reinterpretation to be able to take

place (see Haspelmath 1998: 322–323, 326–327, Hopper and Traugott 2003: 52,
Kilpiö 2013). For instance, Croft (2009: 87) points out that the periphrastic fu-
ture marker be going to most likely evolved in order to avoid misunderstand-
ing, perhaps due to the volitional meaning of will or its compression to barely
perceptible ’ll. This would be a functional motivation for the initiation of this
change in line with the communicative scenarios put forward in earlier sec-
tions. Crucially, it was possible to reinterpret the original sense of I am going/
travelling [somewhere] [in order] to do something, as ‘I intend to do something’
and later simply as ‘I am about to do something [in the future]’, since all those
readings were originally entailed and allowed. Given the right pragmatic cir-
cumstances, it was possible to interpret the input construction in more than
12 On the unidirectionality hypothesis, see Heine (2003: 582–583), Hopper and Traugott
(2003: Chapter 5), Traugott (2003: 627ff., 646 n. 9, 2010a: 272–274), Danylenko (2005: 351ff.).
one way (see Croft 2000: Chapter 5, Traugott and Trousdale 2013: 217ff.). As with
the material I present in later chapters, the duality of motion and/or future
readings of be going to persists to this day.13
This may often be the case, but ambiguity is not a necessary condition.
Although it is assumed by many that reanalysis presupposes the existence
of alternative surface structural parses, such as ambivalent case marking or
alternative constituency structure possibilities, ‘remappings between gram-
matical form and conventional function may violate existing unambiguous
conventional mappings’ (Croft 2000: 120, cf. Traugott and Trousdale 2013: esp.
199ff.). Therefore, it might be too strong a claim to require formal grammatical
ambivalence for reanalysis to be able to occur. Both reanalysis and ambiguity
will play a central part in the account of the development of the perfect below.
2.2.3 The Role of Frequency and Contact in Grammaticalisation

It has long been known that grammatical morphemes and the constructions
they appear in occur much more frequently in language output than lexical
morphemes. Grammatical morphemes commonly arise out of lexical ones via
the process of grammaticalisation, so this process involves a dramatic rise in
frequency as a result of the greater range of contexts in which the grammatical
morpheme is appropriate (Bybee 2003: 602, Hopper and Traugott 2003: 126ff.).
However, frequency is not just an outcome of grammaticalisation but also con-
tributes to it – Bybee (1988, 2003, 2013) sees frequency as an active force which
instigates the various phonological, morphosyntactic and semantic/pragmatic
changes that make up grammaticalisation (enumerated above).14
In this respect, repetition is an important mechanism, since it has an effect
on the cognitive representation of grammaticalising constructions (see Bybee
2003: 603–604, 609, with references). Due to repetition, a frequently occurring
sequence of words or morphemes will become automated as a single pro-
cessing unit. Giving a taste of the role of frequency in the recruitment of OE
13 On grammaticalisation, reanalysis and implicatures, see Haspelmath (1998), Hopper and
Traugott (2003: Chapter 4), and also Levinson (2000: Chapter 4, esp. 263ff.), where he dis-
cusses the pragmatics of the development of the be-going-to future construction, as well
as the pragmatic aspects of grammaticalisation more broadly, as already noted in the pre-
vious sub-section.
14 Also consult Bybee and Hopper (2001); cf. Hollmann (2003: 102–104, 108 fn. 9, 110 fn. 11,
116–117, 125–126, with references), for alternative and complementary usage-based mod-
els, where (type and token) frequency is paramount too. See Traugott (2010a: 280) for
a brief critical assessment. Traugott (2010b) warns that, although frequency tends to be
construed as a contributing factor, it is not actually a cause. Hoffmann (2005) points to
grammaticalised forms which remain infrequent even post-grammaticalisation. I return
to problems with frequency in Chapter 9.
28 chapter 2
cunnan ‘to know’ as a modal auxiliary – ModE can, Bybee suspects that cunnan
began to grammaticalise as an auxiliary because it was already quite frequent,
having undergone some bleaching of its semantic content. The frequency of
the various competing auxiliaries can likewise be regarded as a crucial ingredi-
ent in their survival during the development of the perfect, as will emerge in
chapters to come.
Finally, it has been suggested that constructions (in the Construction
Grammar sense) can be transferred from one language to another in contact
situations, with areal and contact-based research on grammaticalisation car-
ried out for a number of languages (see Traugott 2010a: 282–283, Drinka 2017:
14ff.). Thus, it has been confirmed that grammaticalisation can sometimes be
motivated externally. In such cases, it is potentially not always easy to tell if
one is dealing with externally-prompted grammaticalisation or with the bor-
rowing of structure. Below we shall see that contact-induced scenarios have
been proposed for the development of the perfects in English and Bulgarian,
as well as Germanic and Slavonic more widely. For instance, the (relatively re-
cent?) have-perfect of Bulgarian is less grammaticalised than those of contigu-
ous Macedonian and Greek in that it does not normally occur with intransitive
verbs (see Chapter 7). Since grammaticalisation has not been implemented to
the same extent, this might suggest that Bulgarian did not straightforwardly
copy the structures themselves but replicated the model of its more advanced
neighbours in an externally-motivated round of grammaticalisation, imitating
the movement towards more grammatical forms (see Aikhenvald 2002, Heine
and Kuteva 2005, Drinka 2017: 285–286).
2.3 Other Preliminaries
Before closing off this chapter, I ought to quickly tie three more loose ends –
discrediting progress as a legitimate way of thinking about language change
and endorsing randomness instead (as avowed already), as well as justifying
why it is worthwhile to embark on the present exploratory journey.
2.3.1 Progress
Croft (2000: 78–79, 82) notes that teleology in evolution and in language change
can be associated with the idea of progress, which can be seen as movement
towards some specified (positive) goal. Even if change were progressive (at
least locally, if not globally) in order to achieve better adaptation to the en-
vironment, it is usually hard and subjective to specify what is desirable in the
first place. Furthermore, the conditions (both in biology and in language use)
are so changeable that today’s adaptive trait that confers a distinct advantage
might turn out to be rather dysfunctional tomorrow. There are competing mo-
tivations, such as economy of expression, a preference for shorter or even zero
forms, as opposed to iconicity – expressing concepts overtly so that no mis-
understanding occurs. Although it was widely popular in the 19th century, the
idea that there is (global) progress in language is universally condemned by
serious linguists today. As Croft (2000: 80–81) cautions, linguistic subsystems
are all ‘well enough adapted to effective communication that no one structural
type has a decided functional advantage over the others’.15
2.3.2 Randomness
Above I outlined one possible evolutionary model of language change, devel-
oped and articulated by Croft (2000). It can accommodate both teleological
and non-teleological mechanisms, as well as intentional and non-intentional
behaviour. Croft also envisages other alternatives. A fairly simple alternative
theory to his evolutionary account involving non-teleological intentional
behaviour is that language change is random. Randomness can easily be
subsumed under a generalised theory of selection, since it is a mechanism un-
derlying replication. Croft (2000: 42) goes on to make the following admission:
I am inclined to believe that randomness does play some role in language

change. That is to say, randomness may be one of the causal mechanisms
underlying altered replication, and perhaps even selection. But there is
enough evidence of regularity in language change, in both innovation
and propagation, to lead historical linguists to assume that language
change is not solely a random process.
Croft’s conclusion sounds reasonable and he is laudably open-minded in al-

lowing for a certain degree of unpredictability a priori. Fløgstad (2016), among
others, reaches a similar conclusion regarding the expansion of the preterite
and demise of the perfect in several varieties of Spanish and Portuguese – the
situation in those varieties is the exact opposite of the tendency for perfects
to replace and push out preterites, so things can go either way; the tendency is
not immutable. My empirical survey below likewise supports a view whereby
a great deal of change is due to randomness as opposed to direct or indirect
design or a subconscious desire for orderly systems or communication free
of ambiguity. Even where plausible motivations are registered in subsequent
15 See Croft (2000: 78–84) for further refinements and a more sophisticated treatment of the
topic. Cf. Sampson et al. (2009) and Aitchison (2013).
30 chapter 2
chapters, they do not appear to follow from such intentional human behaviour
as has been suggested.
2.3.3 Justification
Some of the anonymous reviewers for the LAGB and ISLE 2016 conferences
cast doubt on the seriousness of the issues involved here, arguing that an ap-
peal to (naïve) teleology might be a straw man. One of them states that
[t]he suggestion that teleological explanations of language change are

still ‘widely held’ seems an exaggeration in the light of modern theories
of language change. It appears therefore that the problem addressed […]
is not as serious as the abstract indicates.
According to another, the place the teleological approach holds in present-

day (historical) linguistics appears to be overestimated. Indeed, there have
been many voices against attaching too much importance to teleology, some
of them already echoed above. However, it was also shown above that some
serious researchers still uphold functional (un)desirability, and a more sophis-
ticated view of teleology (recast as non-teleological intentional behaviour), as
explanatory factors in (historical) linguistics in general (see also Frajzyngier
2005, Gorlach 2005: 364, Ringe 2006: 181, or the references in Drinka 2017: 89, 91,
among many others). In a recent publication, Berdicevskis and Eckhoff (2016)
likewise suggest that the functionality (non-redundancy) of morphological
features can predict the survival of Common Slavic grammemes in the mod-
ern Slavic languages. Below, I demonstrate that explanations in the same spirit
have been proposed, and continue to be proposed, by numerous researchers
regarding the specific phenomena that I will be examining. Thus, the issue still
merits discussion and empirical testing.
2.4 Conclusion
Function and functional load, including the avoidance of ambiguity, have been
central in this introductory overview. It was demonstrated that accounts ap-
pealing to linguistic fitness, in the sense of avoiding ambiguity and/or func-
tional overload, are highly suspect, even though they have been, and continue
to be, invoked on countless occasions. Their predictive power will be chal-
lenged in the chapters to come, where I zero in on the role of ambiguity and
functional load both on the level of the (abstract) linguistic system, as well as
in actual output. By now, the intellectual climate has shifted: naïve teleology
(i.e. changing or maintaining the system for system-internal reasons, such as

the preservation of a contrast that is threatened with ‘extinction’) has been
superseded by non-teleological intentional behaviour of the invisible-hand
type (i.e. speakers may preserve or create contrasts sub-consciously, not for
the sake of propping up or streamlining the system, but for the sake of avoid-
ing misunderstanding in interaction); nevertheless, I have shown how the two
views essentially make the same predictions (in this case, that contrasts might
be preserved, created or eliminated due to considerations of functionality and
(ill-)adaptiveness). I reiterated many of the arguments against the overreliance
on such explanations (Section 2.1.3), suggesting instead that a change can often
occur simply mechanistically and randomly, regardless of functional consider-
ations. In the chapters to come, I set out to prove that this is by and large what
happened with perfects in English and Bulgarian.
To give me the toolkit I need in order to describe the paths taken by the vari-
ous perfect formations in several languages, in the preceding sections I fleshed
out the basics of grammaticalisation theory and reanalysis. Ambiguity is seen
as important in reanalysis and grammaticalisation (in a way different from the
one sketched in the light of functionality): this time, an ambivalent string can
provide a vital ‘bridging context’ between perceiving a construction in two
ways – the old one (prior to grammaticalisation/reanalysis) and the new one
(post-grammaticalisation/reanalysis). Other crucial notions from this chapter
which will be revisited later involve layering, divergence, specialisation, persis-
tence and de-categorialisation, as well as the curious effect of polygrammati-
calisation. Extra emphasis was likewise placed on the roles of frequency and
language contact in grammaticalisation, since explanations relying on them
will keep resurfacing throughout this text. With the essential tools in place, I
am now in a position to commence the discussion of the grammaticalisation
of perfect periphrases in English.
chapter 3
The Story of the English Perfect
Synthesising and reassessing a vast body of research, the present chapter docu-
ments the development of the perfect in English. Section 3.1 deals with per-
fects built with have, paying special attention to the morphological marking
and the ambiguity which accompanied the reanalysis of have as an auxiliary in
the Old English period (3.1.2). Section 3.1.3 is an excursus into a short-lived rival
construction relying on a verb for ‘own’, before I consider how have really took
off from the Middle English period onwards (3.1.4). Section 3.2 is dedicated
to perfects with be, whereas Section 3.3 chronicles the competition between
the two auxiliaries, culminating in the loss of be. Echoing the major themes
of the previous chapter, in Section 3.4 I recount some of the explanations for
the demise of perfect be from the specialist literature, in particular: the greater
frequency of perfect have; the functional overload and the ambiguity of be as
a copula and an auxiliary for the perfect, the progressive and the passive; the
disappearance of alternative passive auxiliaries, which ‘conspired’ to overload
be even further; as well as unclear morphosyntactic contexts involving coordi-
nation, contraction and ellipsis. Section 3.5 raises certain concerns about the
validity of some of the earlier explanations: I dwell on the interaction between
morphological marking and perfect readings (3.5.1), the polygrammaticalisa-
tion of have itself, which resulted in an explosion in the number of functions
it took on (3.5.2), as well as the development of alternative perfect and passive
constructions, notably with get, in addition to a comeback of perfect be (3.5.3).
In the light of frequency explanations, Section 3.5.4 presents some statistical
trends which emerge from the analysis of corpus data and lend credibility to
frequency as a contributing factor. In Section 3.5.5, I propose other convincing
scenarios about the displacement of be from the perfect domain, including
explanations appealing to language contact, before finally wrapping things up
in the conclusion.
3.1 Perfect with Have
3.1.1 Terminological and Etymological Preliminaries

In this chapter, we embark on an exploration of the development of the per-
fect in English. Before we set off, however, some preliminary terminological
issues need to be got out of the way. The term perfective will be used here to
refer to aspectual semantic interpretation rather than tense and it designates

The Story of the English Perfect 33
a completed event, e.g. eat up the food, as opposed to normally imperfective

[they] eat food. Resultative will be taken to indicate a state which obtains as a
result of an action in the past, e.g. ModE he is gone (cf. Brown and Miller 2016:
249). A morphologically marked past tense, i.e. a preterite, of course charac-
teristically locates an event as taking place prior to the moment of speech,
whereas a perfect designates an event which occurs before the reference time
but has an effect on or is relevant to the temporal reference point (the defini-
tions above are roughly based on Łęcki 2010: 147).
When I use the term ‘perfect’ in the rest of this book, I may panchronical-
ly and cross-linguistically mean a family of related constructions with some
shared core properties, including their prehistoric, mediaeval or modern
sources, without necessarily implying that the ‘perfect’ in Old English is equiv-
alent to that in Middle or Modern English or mediaeval or Modern Bulgarian
(cf. Haspelmath 2007 for words of caution, as well as McFadden and Alexiadou
2010 for a similar view). Neither do I always imply that the grammaticalisation
of what I label a perfect is necessarily complete. The designations ‘past perfect’
and ‘pluperfect’ are interchangeable in this book.
The Modern English present perfect normally indicates ‘past/anteriority
with current relevance’, as in Each of you has taken an oath [?*back then] to
return a true verdict according to the evidence [BNC] [perfect: current relevance
implied, focus on present state of affairs] vs. Each of you took an oath [back
then] to return a true verdict according to the evidence [preterite: no current rel-
evance need be implied, focus more on past action, at least in Standard British
English]. The perfect additionally has the following major meanings (elabo-
rated on in Quirk et al. 1985: 192ff.; cf. Michaelis 1994, Miller 2004a, McFadden
and Alexiadou 2010, Macleod 2012: 41ff., Johannsen 2016, as well as Mitchell
1985: §§731–733, for the meanings already available in OE):
– state leading up to the present (That house has been empty for ages),
– indefinite event(s) in a period leading up to the present (Have you (ever)
been to Florence?),
– habits (i.e. recurrent events) in a period leading up to the present (Mr Terry
has sung in this choir ever since he was a boy).
Other linguists might opt for slightly different labels. Miller (2000, 2004a), for
instance, talks about interpretations of result, experience/indefinite anterior,
recent past/hot news and persistent situation/extended-now. Other languag-
es do not invariably use a perfect in all those contexts. For example, many
European languages, including Bulgarian, employ the (simple) present for a
persistent situation leading up to the moment of speech.
The perfect in Modern English, consisting of a form of have and a past par-
ticiple, has variously been regarded as a type of (secondary) tense, a type of as-
pect, a combination of the two, or an altogether separate category, sometimes
34 chapter 3
labelled ‘phase’ or ‘correlation’ (see Quirk et al. 1985: Chapter 4, Stoevsky

1999, Elliott 2001: 10, Smith 2001: 379 n. 1, Huddleston and Pullum et al. 2002:
Chapter 3, Chankova 2008: 6–9, 20–38, Macleod 2012: 28–30, among others; for
Old English, cf. Mitchell 1985: §865ff.). Those matters of classification will not
concern us here and the reader is referred to the sources listed above for per-
tinent discussion.
In order to build its perfect, English has grammaticalised the once lexi-
cal verb have, which originally used to denote possession. Of course, the
lexical verb with its possessive meaning survives too, in line with the ten-
dencies enumerated in Chapter 2. Modern English have is a continuation of
Old English habban, from Proto-Germanic *haf-jan, which comes from the
Proto-Indo-European root *kap- ‘to grasp’ (cf. Latin capere ‘to seize’, ultimate
source of English capture) (Łęcki 2010: 49–50, OED). Aspects of have’s gram-
maticalisation and history have been traced in ample detail by Łęcki (2010).
The remainder of this section will deal with its role in marking the perfect and
how it came to acquire this meaning.
3.1.2 The Have-Perfect in Old English: Morphological Marking, Ambiguity,

and Reanalysis
The origins of the English perfect construction are to be sought in Old English
times, or perhaps even earlier, though scholarly opinion is still divided on
this issue.1 According to many researchers, at that stage the perfect was not
yet fully grammaticalised, a view which I regard as convincing (see Mitchell
1985: §§729–730, §733). The OE verb habban ‘have’ was used as a full/main
verb frequently followed by an accusative object and a past participle which
functioned as an object complement agreeing with the object in case, gender
and number, as in (1), where the idea is probably along the lines of ‘I have you
in a state of being set’.
(1) Ic hæbbe ðe nu todæg ge-set-ne ofer

I have thee.acc.2sg now today pref-set-acc.m.sg over
rice …
kingdoms
‘Today I have set you over kingdoms…’ [King Alfred’s West-Saxon Version
of Gregory’s Pastoral Care, cited in Mitchell (1985: 287, 294), Denison
(1993: 346)]
1 See Denison (1998: 135); cf. Mustanoja (1960: 499), Strang (1970: 352), Smirnickaja et al. (1977),
Bybee and Dahl (1989: 67ff.), Rydén (1991: 345–346), Elsness (1997: Section 4.2.2), Łęcki (2010:
Chapter 4, esp. 204), Macleod (2012: 18).
The syntactic origins of the perfect clearly illustrate the point made in
Chapter 2 that it is constructions that grammaticalise, not isolated lexical
items (cf. Heine 1993: 30). Clauses like (1) are the prototype of what was later
to become the periphrastic verbal construction which is our topic here. The
stative meaning of possession gradually gave way to a dynamic reading, and
what used to be a transitive clause with an object and a participial object
complement, as a result of reanalysis, became a periphrastic perfect with an
auxiliary and a main verb.2 After the auxiliation of have, the original object of
have came to be the object of the participle (now the main verb). The semantic
and pragmatic change behind this development was from a meaning of pos-
sessing/having something in a certain state, via denoting a state as a result of
previous action, to the meaning of completed action.3
The different stages of this development can be schematically represented
as follows:
table 1 Stages in the development of the English perfect (based on Łęcki 2010: 149)
Stage I: He (subject, possessor) has (possessive verb) meat (direct object,

possessee)
Stage II: He (subject, possessor) has (possessive verb) meat (direct object,
possessee) cooked (adjectival modifier/object complement)
2 Current syntactic theory is increasingly coming to treat traditional auxiliary verbs as heads
which take what is traditionally labelled the main verb as a dependent complement (see
Warner 1993: 19ff., Huddleston and Pullum et al. 2002: 104, 1209–1220; cf. Heine 1993: 6–7,
99, 106, Chankova 2008: 40ff., Hristov 2015). On this analysis, has written is not part of the
paradigm of write and, according to many contemporary researchers, the two verbs project
their own separate clauses or phrases – i.e. has and written appear in their own separate VPs.
Although there is solid evidence in favour of this analysis, I will stick to the traditional and
constructionist view, whereby the auxiliary forms a unit with the main verb, of which it is
a dependent. This choice has been dictated mostly by reasons of expository convenience,
sacrificing syntactic faithfulness and rigour (cf. Bowie et al. 2013: Section 1.2). In an elegant
synthesis, Brown and Miller (2016: 89–90) reconcile the two views.
3 See Jespersen (1931: 29–30), Mincoff (1958: 131), Visser (1963–1973: §§2001–2003, §2028, §2043
(on the analogous origin of the infinitival construction to have killed him), §2116, §2189),
Mitchell (1985: §§702–733, esp. §§724–728), Rydén and Brorström (1987: 16–17), Brinton
(1988: 99–100, 107–108), Kovatcheva (1989: 135ff.), Traugott (1992: 190–192), Denison (1993:
364), Warner (1993: 97), Bybee et al. (1994: Chapter 3, esp. 68), Fischer and van der Wurff
(2006: 139ff.), Kilpiö (2007), Trask and Millar (2007: 175–177), Chankova (2008: 71, 218, 2009:
425), Łęcki (2010: 145ff.), Macleod (2012: 11–12), Johannsen (2016, for a recent reassessment),
as well as the Dictionary of Old English (http://tapor.library.utoronto.ca/doe/, accessed on
27 Feb 2017). OE adjectival paradigms can be found in Brinton and Arnovick (2011: 208–209)
and in Table 2 below.
36 chapter 3
table 1 Stages in the development of the English perfect (based on Łęcki 2010: 149) (cont.)
Reanalysis
(ultimately leading to a change in word order)
Stage III: He (subject, agent) has (auxiliary) meat (direct object, patient) cooked
(main verb)
Stage IV: He (subject, agent) has (auxiliary) cooked (main verb) meat (direct object,
patient)
Analogy
(extension to intransitive verbs)
Stage V: He (subject, agent/patient/undergoer) has (auxiliary) cooked/died (main
verb)
Mitchell (1985: §24) stresses that in the proto-periphrasis with habban, the par-
ticiple was often declined to agree with the object and could be regarded as a
passive participle used as an adjective. When the periphrasis became a genuine
verbal combination, we no longer had an adjectival use of a passive participle,
but an active verbal use instead. In other words, in the original proto-perfect
complex transitive construction, the transitive participle retained its passive
meaning (e.g. ‘He had the song [in the state of having been] sung’), which was
lost with the transition into a purely verbal, active combination (e.g. ‘He had
sung the song’).
A Modern English construction which has been claimed to be directly de-
scended from the original ‘proto-perfect’ is sometimes dubbed the conclusive
perfect (see Denison 1993: 343–344, 364, 1998: 135):
(2) The way he kept on, he had me almost convinced Hickock and Smith were
innocent. [1965, from Denison (1993: 343)]
According to the traditional view, the conclusive have-perfect is a direct con-

tinuation of the pattern involving the original meaning of possession which is
believed to be the precursor of the ordinary perfect, though without reanaly-
sis (cf. Jespersen 1909–1949: V 16, Visser 1963–1973: §§2116–2118, Miller 2000,
Brown and Miller 2016). Lee (2003) also argues that there is continuity between
the OE and ModE constructions, demonstrating that there are no major dif-
ferences at least as far as semantic continuity is concerned. Another way of
thinking about the construction in (2) is as a causative, as will emerge below,
i.e. something along the lines of ‘He almost caused me to be convinced’.
As is often the case in linguistics, the road gets bumpy even before we have
started our exploratory journey: criticism has been levelled at the received
wisdom presented above, casting doubt on the evidence in favour of the OE
stative meaning of possession, in addition to the weak chronological evidence
for the spread to actional meanings (see Brinton 1988: §3.1.3; cf. Denison 1993:
365, Macleod 2012: esp. 234). For the time being, however, I shall uphold the
traditional account, in spite of the validity of such concerns.
Interestingly, in Old English the participle did not always agree with the ob-
ject, as becomes apparent from (3). The accusative masculine singular agree-
ment marker -ne is missing from befangen (cf. (1) above and Table 2 below), in
which case it is an open question whether we have an SOVC O clause or wheth-
er it is better interpreted as SOV AUX V MAIN (see Denison 1993: 340–341; cf. Strang
1970: 311, Traugott 1992: 190).
(3) Hraðe heos æþeling-a an-neo hæf-d-eV/AUX | fæste

quickly she noble-gen.pl one-acc.m.sg have-past-3sg fast
befang-enCo/V
seiz-ed
‘Quickly she grasped firmly one of the nobles’ [Beowulf, 1294, cited in
Mitchell (1985: 288), Denison (1993: 347)]
The traditional endings of the strong declension in the nominative and accusa-
tive are set out in the following table:
table 2 Partial strong adjectival declension in OE (based on Brinton and Arnovick 2011:
208, Kovatcheva and Sabev 2015: 145)a
Partial strong adjectival declension

(gōd ‘good’ is a heavy stem [long vowel + consonant], blæc ‘black’ a light one [short
vowel + consonant])
Masculine Neuter Feminine

Singular Plural Singular Plural Singular Plural
Nominative – (gōd, blæc) -e (gōde, – (gōd, –/-u (gōd, –/-u (gōd, -a, -e (gōda,
blace) blæc) blacu) blacu) blaca, gōde,
blace)
Accusative -ne (gōdne, -e (gōde, – (gōd, –/-u (gōd, -e (gōde, -a, -e (gōda,
blæcne) blace) blæc) blacu) blace) blaca, gōde,
blace)
38 chapter 3
table 2 Partial strong adjectival declension in OE (cont.)

a The adjectival endings which Germanic inherited from Proto-Indo-European came to be
used when an attributive adjective modified the noun alone or when an adjective/participle
was used predicatively, as is the case in (proto-)perfect constructions. This is referred to as
the strong (or indefinite) declension of Germanic (and OE). In addition, Germanic evolved
one more paradigm of adjectival endings, the so-called weak (or definite) declension, used
in the presence of demonstratives or possessive determiners, with comparative or superla-
tive adjectives, or in instances of direct address (see Mitchell 1985: §§102–141, Brinton and
Arnovick 2011: 134, 207, among others).
Elsness (1997: 241–242, with references) reviews claims and counterclaims that
the grammaticalised OE perfect was to be distinguished from its precursor on
the basis of the presence or lack of agreement. Denison (1993: 341) offers the
following criteria for determining whether a construction is best analysed as
an SVOC O clause with lexical have or as a genuine perfect syntagm with auxil-
iary have (SV AUX OV MAIN ). SVOC O presupposes the following:
– a sentence brace in main clauses, with non-adjacency of finite have and the
participle;
– accusative adjectival inflection on the participle;
– a stative context where have can mean ‘possess’ and both have and the par-
ticiple can refer to states.
By contrast, the analysis of have as an auxiliary presupposes the following:
– adjacency of have and the participle;
– no adjectival inflection on the participle;
– similar patterns in a non-stative context;
– similar patterns with subjects that cannot be possessors;
– similar patterns with objects that cannot be possessed;
– similar patterns with participles that do not take an accusative direct object.
For problems arising from such rigid distinctions, see Mitchell (1985: §727) and
Kilpiö (2007: 337ff.) – inflection, for instance, can be present or absent in what
is essentially the same clause with the same meaning (cf. (6) below).
In fact, Mitchell (1985: §§709–710) and Denison (1993: 346) report studies
which show that agreeing inflected forms were much fewer in number than
non-agreeing ones, and became even less common over time. Brinton (1988:
101) notes that the majority of participles in OE are uninflected (cf. Fridén 1948:
41, Mitchell 1985: 283–284, 292, Brinton and Arnovick 2011: 229). It can even be
claimed that OE participles are ‘normally invariable’ (Quirk and Wrenn 1957:
75). Strang (1970: 305) points out that the participle was not inflected in the
period between 970 and 1170, though inflection was still present between 770
and 970. Traugott (1992: 190) concurs that inflected forms became less frequent
during the OE period, although they were never predominant in the first place.
Mitchell (1985: §710) quotes publications which record between fourteen and
twenty-five percent of participles as being declined in Old English text sam-

ples, whereas the overall number of declined forms in the Old English sec-
tion of the Helsinki Corpus is approximately ten percent (Łęcki 2010: 172, with
references). According to the Dictionary of Old English (http://tapor.library
.utoronto.ca/doe/, accessed on 27 Feb 2017), out of ca. 2,440 transitive have-
perfects, only 220 show agreement. In his sample, Kilpiö (2007: 329) has found
that 11.6% of participles received inflection between the years 850 and 950,
and 5.9% between 950 and 1050. By early Middle English, the figure drops to
0.4% (Kilpiö 2007: 340). The participle without agreement could have come to
be seen as an active form approximating a supine or a converb (an adverbial
participle) – a stage that would have enabled the reinterpretation.
The reanalysis of the participle as part of the verbal complex, rather than as
an adjectival object complement, is suspected to have first gained a foothold in
constructions with neuter accusative singular objects, which had no overt ac-
cusative marking on the object or on the participle (Traugott 1992: 192, further
endorsed by Kilpiö 2007: esp. 337, 341). Mitchell (1985: §709, quoted in Denison
1993: 364, Łęcki 2010: 152) also sees these zero-inflected forms as ‘an analogical
factor in the ultimate disappearance of the inflected forms’. Since they were
ambiguous, it was possible for the reinterpretation to take place; hence the
grouping in (4a) must have given way to that in (4b).
(4) a. we habbaþ [geweorc geworht]

we have the.stronghold in.a.state.of.builtness
b. we [habbaþ] geweorc [geworht]

we have the.stronghold built
‘we have built the stronghold’ [from Traugott (1992: 192)]
An attested example in which there would have been no inflection (or zero in-
flection) in any case is provided in (5). The neuter singular object interrogative
pronoun triggers no overt ending on the past participle gedon. It is impossible
to tell if this is zero agreement (i.e. agreement works but the exponent in this
slot of the paradigm happens to be a meaningful zero morpheme in paradig-
matic opposition to overt markers in other slots), or whether it is a breakdown
in agreement (i.e. agreement no longer functions and the speaker/writer did
not intend to signal the relationship between the two elements by copying the
features of one onto the other).4
4 Haas (1957) is a classic on the theoretical implications of positing zeros in linguistic descrip-
tion. In Chapters 7 and 8, it will emerge that, although there are few such morphosyntactically
40 chapter 3
(5) Hwæt hæf-st þu ge-do-n?

what.acc.nt.sg have-pres.2sg thou.nom.2sg pref-do-pptcp
‘What have you done?’ [Ælfric, cited in Mitchell (1985: 284), Denison
(1993: 347)]
In general, the modifying participial adjective (object complement) remained

legitimately uninflected when combined with accusative neuter singular
nouns, with non-accusative objects, with prepositional and clausal objects or
when the object was missing (Mitchell 1985: §709, §712, §720, §727, Łęcki 2010:
151–152). As Denison (1993: 347) notes, the number of additional examples of
‘proper breakdowns’/‘defaults’ outside the above-mentioned categories is in
fact larger, as adjectival inflections in general had an increased tendency to
disappear from polysyllables, at least by Middle English (cf. Mossé 1952: §74).
This was the case with past participles, which normally have a minimum of
two syllables. There exist examples of conjoined inflected and uninflected
participles – in (6), gehyred bears no agreement marker (expected -e), whereas
gesewen is inflected for the plural with the usual -e ending.5
(6) Fela God-es wundr-a we habbað ge-hyr-ed

a.lot God-gen.sg wonder-gen.pl we have pref-hear-pptcp
and eac ge-sew-en-e
and also pref-see-pptcp-pl
‘Many of God’s miracles we have heard and also seen.’ [Ælfric, cited in
Mitchell (1985: 284), Brinton (1988: 101), Traugott (1992: 191), Denison
(1993: 347)]
We are thus dealing with two types of lack of agreement – audible/visible mor-
phology is legitimately missing in (5) because of zero exponence for this slot in
the paradigm, so agreement can be thought of as ‘covertly present’; (5) cannot
have overt agreement (cf. Table 2 above). In (6) on the other hand, the overt
plural -e marker fails to appear on one of the two participles. We can tell that
a breakdown has occurred. The first type is said to have opened the floodgates
for the second. The above sources imply that reanalysis is to a great extent due
to the syntactic ambiguity of the construction, which in turn existed because
there was no overt morphology to dispel it. Trask and Millar (2007: 177) are
ambiguous cases in Bulgarian and Macedonian, the same reanalysis has also occurred or is
occurring. So it cannot be just form that matters – there must be something else too.
5 More examples like this one can be found in Mitchell (1985: §711), Kilpiö (2007: 328) and
Łęcki (2010: 173, 229 n. 29).
rather more cautious in stating that ‘reanalysis may be favoured by the absence
or loss of overt morphology which is inconsistent with the reanalysis’. The cru-
cial difference boils down to whether the reanalysis was caused or possibly just
favoured by ambiguity.
Conversely, inflections could appear when they were not expected, perhaps
due to scribal errors or analogical levelling. In (7), hus is a neuter plural object
and it should not trigger inflection on the past participle (which should surface
as berypt); however, the (originally) accusative plural masculine form berypte
turns up instead (see Łęcki 2010: 172, Table 2 above and Section 3.2 below).
(7) & we habbað Godes hus inne & ute

and we have God’s house(nt)[acc.pl] inside and outside
clæne berypt-e ælcra gerisena
fully spoiled-acc.pl(.m?) of.all honour
‘And we have completely spoiled God’s houses both inside and outside of
all honour.’ [ca. 1014, from Łęcki (2010: 172)]
At the end of the day, it remains a moot point whether habban was a genu-
ine auxiliary in Old English (see preceding discussion, as well as Mitchell
1985: §§729–730, §733, Fischer 1992: 250, Denison 1993: 340, Łęcki 2010: 164ff.).
Crucially, perfect progressives and perfect passives did not appear until Middle
English, pointing to an incomplete grammaticalisation cycle in Old English.6
According to Mitchell (1985: §753) and Łęcki (2010: 188), at least one of these
gaps might be due rather to the later development of the progressive. What is
more, the OE verb agan ‘own’ is also to be met with in perfect-like construc-
tions instead of habban, suggesting uncertainty and lack of a fully fixed and
grammaticalised periphrasis (see Brinton and Arnovick 2011: 228; cf. the gram-
maticalisation tendencies outlined in Chapter 2 above and the brief discussion
of agan in Section 3.1.3 below).
On the other hand, Traugott (1992: 191) reminds us that the use of the hab-
ban construction with intransitive verbs from very early on, as well as its use
with transitive verbs which had non-accusative objects (where there was no
6
See Mustanoja (1960: 440, 516–519 on perfect infinitives in ME, 590–591), Mitchell
(1985: §702, §753, §§782–785, §922 on perfect infinitives in OE), Fischer (1992: 276–277),
Chankova (2008: 138, 185–186), Brinton and Arnovick (2011: 373); cf. Łęcki (2010: 184–188 on
perfect infinitives in OE). Denison (1998: 183–184) notes that even in earlier Modern English
the combination of perfect and passive seems to have frequently been avoided, (i), as it still
is in Dutch and German (cf. Jespersen 1909–1949: IV 102–104, Visser 1963–1973: §793, §1909).
(i) but that objection is done away with [1818, Keats, Letters 66, p. 146 (21 May), quoted in
Denison (1998: 183)].
42 chapter 3
agreement either), suggests that OE habban could serve as an auxiliary, though

we should not forget the potentially adjectival status of inflected participles
(cf. Mustanoja 1960: 500, Mitchell 1985: §730, Los 2015: 73). Already in Old
English, the subject did not have to be an animate possessor, and a possessive
context might be ruled out, as in the case of ‘deprivation’ or taking something
away – (8). Inanimate and abstract subjects are both attested – (8)–(9) (see
Mitchell 1985: §726, Łęcki 2010: 169–170).
(8) me hæfþ ðeos gnornung ðære gemynde benumen

me has this grief of.the memory deprived
‘This grief has deprived me of memory.’ [ca. 880, King Alfred’s Old English
Version of Boethius, De Consolatione Philosophiae, from Łęcki (2010: 169),
also quoted in Mitchell (1985: 287)]
(9) gimmas hæfdon bewrig-en-e weorðlice <wealdendes> treow

gems had cover-pptcp-pl? splendidly ruler’s tree
‘Gems had splendidly covered the ruler’s tree.’ [950–1000, Dream of the
Rood, from Łęcki (2010: 169), also quoted in Mitchell (1985: 285)]
As pointed out by Łęcki (2010: 173), the problematic inflection on the participle
bewrigen-e cannot be used as evidence in favour of a non-perfect stative/ad-
jectival reading because of the semantics of the clause. The strong plural (?) -e
might be due to attraction to plural gimmas or it might somehow agree with
neuter singular treow. In that case the form should probably be emended to
strong neuter singular bewrigen, since bewrigene is unlikely to be a weak neuter
singular form (cf. Table 2 above). Mitchell (1985: §712) is more inclined to treat
the whole thing as a scribal error, noting that the participle is preceded by a
plural.
Going back to the debate surrounding the status of OE habban, Denison
(1993: 352) lists the following stages in the grammaticalisation of the perfect
construction with have as possible indicators of the establishment of the per-
fect category:
a) when the have-perfect became available for any lexical verb which did
not conjugate with be (late Old English?)
b) when it had come to be a pure tense equivalent (late Old English?) – see
Section 3.1.4 below
c) when it had developed approximately its present-day meaning (seven-
teenth century?) – which involved the loss of b)
d) when it became available for every non-auxiliary verb (late Modern
English)
Denison settles on stage a) as the point at which have had attained true
auxiliary status, since it was used ‘without an argument structure or selec-
tional restrictions of its own’ – decategorialisation being a typical feature of
the grammaticalisation process, as is the bleaching of the original meaning,
in our case the meaning of possession (see exx. (8) and (9) and Chapter 2,
Section 2.2.1).
3.1.3 Perfect with OE agan

Old English had another verb of possession, agan (giving ModE owe, own,
and ought, the latter from the past tense). Apart from indicating possession,
in combination with an inflected infinitive, agan could mark obligation; it
could also act as a (proto-)perfect auxiliary (see Mitchell 1985: §743, Łęcki 2010:
54, 155–156, as well as the remarks on polygrammaticalisation in Chapter 2,
Section 2.2.1). However, as Łęcki (2010: 54) notes, OE habban tended to become
specialised as a perfect auxiliary, while agan was more frequently used to indi-
cate obligation. Readers will remember that it is typical in grammaticalisation
for one variant to be selected for a given function from a pool of several origi-
nal candidates. The attestations of perfect agan are only sporadic:
(10) forþon him gelyfeð lyt, se þe ah lifes wyn

therefore him believes little he who has life’s pleasure
gebiden in burgum
experienced in cities
‘Therefore he little believes, he who has experienced life’s pleasure in cit-
ies’ [The Seafarer, 950–1000, from Łęcki (2010: 155), also quoted in Mitchell
(1985: §743)]7
3.1.4 Increased Use and Greater Degree of Grammaticalisation of the

Have-Perfect in Middle and Modern English
In Middle English, the perfect saw a rapid increase in frequency.8 As pointed
out in Chapter 2, when a construction grammaticalises, it becomes appropri-
ate in more and more contexts due to its generalised meaning and consequent-
ly experiences a rise in frequency. Combinations of auxiliary have and full/
lexical have appear for the first time in this period, as in (11). Since there are two
7 There may be alternative interpretations of these verse lines whereby ah lifes wyn and ge-
biden do not belong together – see http://www.anglo-saxons.net/hwaet/?do=get&type=text
&id=Sfr&textOnly=false (accessed on 1 June 2017).
8 See Mustanoja (1960: 504–505), Fischer (1992: 250, 257), Elsness (1997: 252–253, 264–270, 277,
340–341), Smith (2001: esp. 375, also tracing later periods), Chankova (2007a: 44, 2007b: 510,
2008: 102, 129ff., 152, 201, 216).
44 chapter 3
instances of this verb in (what is traditionally considered) the same clause, one
of them must be an auxiliary (see Łęcki 2010: 171 and the remarks in Chapter
2, Section 2.2.1 above concerning the survival of the original lexical item post-
grammaticalisation; cf. am being, going to go, there is something there).
(11) We habbeð ihaued moni burst, moni hunger, & moni þurst.
we have had much need much hunger and much thirst
[c. 1275(?1200) Lay. Brut, from Łęcki (2010: 171)]
In Middle English, the plural -e ending on the participle could even be used in
relation to singular nouns, which is taken to indicate that the grammaticalisa-
tion process had more or less reached completion by that stage (Zimmermann
1968: 36, Fischer 1992: 256; cf. example (9) above). More usually, however, in-
flection is generally lost, as noted in the preceding section. On the other hand,
the perfect often alternated in its functions/meanings with the preterite,
sometimes in different manuscripts of the same text, suggesting that it was not
yet fully grammaticalised and ‘regulated’.9 In the mediaeval period, the perfect
was still compatible with specific past time reference, as will be stressed and
illustrated in Chapter 5, Section 5.2.2. Conversely, Chaucer used a simple past,
rather than a perfect, with adverbs like never: A fairer saugh I nevere ‘I have
never seen anyone more beautiful’ (Benson 1987a: xli). Essentially both the
perfect and the preterite could indicate a past event connected to the moment
of speech. Therefore, Middle English is indeed a transition period in which the
perfect has started to come into its own more fully than in Old English but has
not yet attained the state of distinctness from the preterite and idiomatic use
it has in Modern English.
The ambiguity between a periphrastic verb phrase and a complex transitive
clause with main verb have plus a participial object complement persisted into
Middle English, and in fact the two constructions survive into the modern peri-
od, now distinguished by word order (Fischer 1992: 257).10 This is illustrated in
9 See Denison’s stage b) in Section 3.1.2 above and Denison (1993: 352–353), as well as
Mustanoja (1960: 498–499, 504, 506–507), Visser (1963–1973: §766, §772, §800, §§805–
811, §2004, §2008), Zimmermann (1968: 36), Mitchell (1985: §723, §726 fn. 173), Fischer
(1992: 250, 256–258), Elsness (1997: 247ff., 291), Tagliamonte (2000), Fischer and van der
Wurff (2006: 139), Chankova (2008: 119–124, 128, 157–158, 169, 200–203); for the preterite
and the perfect in Shakespeare’s English, consult Kakietek (1976: 50–51).
10 See Visser (1963–1973: §1964, §1979, §2001, §§2116–2122), Brinton (1988: 102, 1994),
Kovatcheva (1989: 135–137, 139), Denison (1993: 348, 358, 365). The survival of both con-
structions and our recurrent themes of ambiguity and indeterminacy resonate with the
constructionist idea of synchronic gradience resulting from constructional change (see
Traugott and Trousdale 2010, 2013, inter alia). Such indeterminacy in this book involves
the analysis of OE perfects and passives as complex transitive or copular constructions as
(12), where (12b) is reminiscent of the so-called conclusive perfect or causative

construction mentioned above.
(12) a. TheyS hadAUX repairedV the houseo . [ordinary perfect]

b. TheyS hadV the houseo repairedCo. [a complex transitive construction]
There are modern instances where word order cannot help to distinguish
between these two constructions and we have to rely on contextual cues. In
other words, the duality of have + past participle still persists and might even
result in double-marking, as seen in (13), with two tokens of have and two past
participles.
(13)
In one year the Abbey Theatre has had submitted 600 plays by Irish
authors. [1960, Daily Telegraph, quoted in Visser (1963–1973: §2120)]
Instead of coming after the object at the end of the clause, submitted appears
medially with the other two verbs – the perfect auxiliary has is thus followed
by two past participles, the first one part of the perfect construction, the sec-
ond more like an object complement. Essentially, have + past participle occurs
twice. I return to similar cases of double-marking below.
3.2 The Origin and Status of the Be-Perfect
Have was by no means the only perfect auxiliary in Old, Middle and Early
Modern English. Be could build perfect periphrases too, mainly, though not
necessarily, with intransitive verbs denoting a change of place or state.11 Both
opposed to verbal periphrases, as well as uncertainty of interpretation arising from equiv-

ocal perfects, passives or (copular) statives. The terms ambiguous, equivocal, ambivalent,
indeterminate etc. are meant as synonyms here. In semantics a distinction is sometimes
drawn between an equivocal sentence, i.e. underdetermined for both speaker/writer and
hearer/reader, and an ambiguous sentence, whose author intended only one possible
reading out of two or more (see Wasow et al. 2005). In our context, it is rarely feasible to
pinpoint with absolute certainty what was intended by the producer of a sentence.
11 See Jespersen (1931: 30ff.), Mincoff (1958: 131), Mustanoja (1960: 499ff.), Visser (1963–
1973: §§1898–1904, §2052), Mitchell (1985: §702, §§734–742), Brinton (1988: 99–100, 108),
Kovatcheva (1989: 135), Fischer (1992: 260), Denison (1993: 358ff., 1998: 135), Warner (1993:
96–97), Elsness (1997: Section 4.2.2), Kilpiö (1997: 104ff.), Kytö (1997), Fischer and van der
Wurff (2006: 139–141), Brinton and Arnovick (2011: 371), Macleod (2012: 12–16), as well as
Denison (1993: 344, 349–352) for verbs with have. For an overview of be and have as per-
fect auxiliaries in the Germanic languages, see Shannon (1995). Wischer (2010) and Laing
(2010) discuss the suppletive paradigm of OE and early ME be, comprising forms of the
46 chapter 3
auxiliaries are found from Old English on, when have occurred mostly with
transitive verbs and be was restricted to intransitives, as in Modern Dutch or
German, although some intransitive verbs already combined with have. The
original meaning of the be-construction is widely believed to have been that of
a stative use of linking/copular be followed by a predicative adjective in order
to refer to a state which resulted from an earlier action (Denison 1993: 366;
cf. Visser 1963–1973: §1898, Rissanen 1999: 213). This is shown in the following
simple OE example:
(14) hie wær-on ge-cum-en-e

they.nom.3pl be.past-pl pref-come-pptcp-nom.pl
‘They were/had come’ [from Rydén (1991: 353, n. 11), glossing and transla-
tion mine]
Unlike the have‑perfect, which gradually came to indicate action/process,

the one with be is traditionally said to have indicated a state (Visser 1963–
1973: §1898, Mitchell 1985: §729, §§741–742, Denison 1993: 366, Elsness 1997:
247). Fischer (1992: 260), however, rightly observes that
[it] is unlikely that such a clear dichotomy exists in view of the facts that
the two constructions often seem to be interchangeable and that in the
later development have ousts the be forms. Such a dichotomy would also
mean that a structure like he is come always referred to state, never to
completed action.
Occurrence with OE past participles like that of come arguably indicates some
degree of grammaticalisation, since they cannot be so naturally interpreted
as denoting resultant states which persist at the time of speaking: cf. ModE
a fallen tree and *a come man (Los 2015: 75), though both phrases would be
equally acceptable in Bulgarian (see the Epilogue to Chapter 5 too). In addi-
tion, Traugott (1992: 193) notes that the participle occurring with be is always
semantically active, because the verb is intransitive.12
originally separate verbs beon and wesan, in fact combining three Indo-European roots,
with traces of a fourth one – see Prokosch (2009[1939]: 219–222), Mitchell (1985: §§651–
664), Laing (2010: 238), Łęcki (2010: 8, 212 n. 8).
12 It will be demonstrated below that Bulgarian and Macedonian possess morphological
distinctions between active and passive past participles. This, however, has not prevented
the formation of a new have-perfect in addition to the old one which relies exclusively
on be.
In the early stages, if the participle in a be-perfect construction was in-

flected, the agreement was with a nominative NP, as in (14) – just what one
would expect from a typical subject complement. Mitchell (1985: §737) notes
that in general the participle in a be-perfect construction remains uninflected
in the singular, and receives an -e in the plural (see Table 2 above). According
to Denison (1993: 359), this is to be expected, because the OE masculine and
neuter nominative singular of the strong adjectival declension was zero (-Ø)
anyway, and the masculine plural -e ending was frequently extended to femi-
nine and neuter plural adjectives (in addition, this probably explains the prob-
lematic example with the neuter plural hus ‘houses’ which triggers -e on the
participle in (7) above). By and large, there is hardly any participial inflection
left by the Middle English period.
Elsness (1997: 261–263) observes that, in comparison with present perfects
using have, a somewhat larger proportion of the recorded be-perfects occur
with inflected participles (which seems to be borne out by the OE evidence
presented in Chapter 4). Elsness considers this a sign that their transition to a
genuine perfect construction is sometimes a little less obvious. In Chankova’s
(2008: 98–99) corpus of OE, 7.06% of participles exhibit agreement in con-
structions with habban, vs. 92.94% of lack thereof, while in be-perfects, 21.43%
of participles agree, and 78.57% fail to do so. The higher number of agreeing
participles with be-perfects supports the hypothesis that these constructions
were being grammaticalised at a slower rate compared to structures with hab-
ban. As a baseline, according to some estimates already mentioned earlier, in
total 11.81% of past participles showed agreement in Old English, as opposed to
88.19% for uninflected participles.
Furthermore, the existence of participles in apparent be-perfect construc-
tions which corresponded to no finite verbs, as well as coordination with
adjectives, and modification by intensifiers like swiþe ‘very’, all testify to the
incomplete grammaticalisation of the be-perfect in OE (see Denison 1993:
360–361). Mitchell (1985: §734, §§741–742, §842) is not even convinced that
the be-perfect was ever fully grammaticalised (a similar view is expressed in
McFadden and Alexiadou 2006, 2010).
As is generally the case in syntax, most of the tests listed in the preceding
paragraph have to be applied with caution. For instance, the present participle
in a progressive construction can still be conjoined with an adjective, as in (15),
but this does not mean that the progressive has not been fully grammaticalised
in Present-Day English.
(15) Tens of thousands of phantom azalea bushes and geraniums are believed
to be alive and growing in the gardens of Northern Ireland. [news, from
Biber et al. (2002: 331)]
48 chapter 3
And yet, when applied in combination, the tests above do suggest that the
grammaticalisation of the be-perfect had not reached completion in OE.
3.3 Competition between the Be- and the Have-Perfect
Have started to drive be out of its domain as early as Middle English or even
earlier, in spite of occasional instances of be where have would be expected, as
in (16), where be is used in a transitive clause.13
(16) for þou art passed thy paynes alle [Tundale, 1464, quoted in Mustanoja
(1960: 501)]
Cases like (16) are rather rare and exceptional (though they survive until late
and are to be revisited below – see Chapter 5 and Rydén and Brorström 1987:
133ff.). Traugott (1992: 191) notes that throughout the Old English period al-
ready, and of course much more so later, habban could occur with intransi-
tive participles in contexts which had nothing to do with possession (cf. Łęcki
2010: 153–154, 157–159, and the Dictionary of Old English, at http://tapor.library.
utoronto.ca/doe, accessed on 27 Feb 2017). Quite a few verbs were happy with
either auxiliary, so we observe the typical variation that is to result in language
change. The following are early examples of have used with intransitive verbs
of motion:
(17) Syððan hie togædere gegan hæfdon

after they together gone had
‘after they had come together’ [Beowulf 2630, quoted in Jespersen (1931:
34), translation and emphasis mine]
(18) Wæs se feond full neah || þe on þa frecnan fyrd

was the devil very near who in the dangerous journey
gefaren hæfde ofer || langne weg
travelled had over long way
13 See Jespersen (1931: 30), Mustanoja (1960: 500–501), Visser (1963–1973: §§2002–2003),
Strang (1970: 207), Zimmermann (1973: 109ff., for be with transitive verbs), Mitchell
(1985: §722ff.), Rydén and Brorström (1987: 16–17, 195ff., 197 fn. 11, 200), Rydén (1991: 346,
esp. Fig. 2), the last two showing a typical S-curve of the progression of have between
1500 and 1900; consult also Fischer (1992: 260), Denison (1993: 344, 363–364, 1998: 135),
Kytö (1994: 179, 1997: 17–18), Elsness (1997: 246, 271–272, 325–327, 340), Rissanen (1999:
213), Fischer and van der Wurff (2006: 142), Chankova (2008: 136–137).
‘The devil was very near who had travelled in the dangerous journey over
a long way.’ [Genesis B 688–90, quoted and translated in Lee (2003: 390),
also quoted in Mitchell (1985: 289), emphasis mine]
By the sixteenth century, only have was allowed with transitive verbs and was
preferred with non-mutative intransitives, with variation in the case of mu-
tative ones (Rissanen 1999: 213; cf. Strang 1970: 149, Kakietek 1976, Rydén and
Brorstöm 1987: 17–18, Chankova 2008: 181–186, 205, Brinton and Arnovick 2011:
371–373). Curiously, Nakamura (1987: 43) notes that be-perfects are statistically
still decidedly preferred with mutative verbs in Pepys’s Diary from the seven-
teenth century.14
It was mainly in the eighteenth century that have gained a lot of ground,
although be was still more common with intransitive verbs (Rissanen 1999: 215;
cf. Rydén and Brorström 1987: esp. 195ff., Shannon 1989, Rydén 1991: 346–347,
Kytö 1997: 32ff.). Interestingly, prescriptive grammarians in the 18th century
picked up on the variation that existed back then. Some of them actually pro-
scribe the use of be with certain mutative intransitive verbs, as in (19), which
Lowth denounces as ‘improper’ (see Rydén and Brorström 1987: 206–209 and
Straaijer 2010: 65–66).
(19) The rules of our holy Religion, from which we are infinitely swerved [Robert
Lowth. 1762. A Short Introduction to English Grammar. p. 63, quoted in
Straaijer (2010: 66), emphasis mine]
Presumably, Lowth would have preferred have here. These views make Lowth,
and other eighteenth- and early-nineteenth-century grammarians who ex-
pressed similar opinions, innovators in terms of this change ongoing at the
time, though more research is needed to verify if they were also innovators
in practice, i.e. if they really practised what they preached. Faithful to his
more descriptive (rather than prescriptive) inclinations, in the 1768 edition
of his Rudiments of English Grammar, Joseph Priestley (1768: 127–128, cited in
Straaijer 2010: 66–67) comments on the choice of auxiliary, suggesting that I
am fallen refers more to the present (state?), whereas with I have fallen ref-
erence is also made to the past (action?) (cf. Rydén and Brorström 1987: 208
for similar contemporary accounts). According to Straaijer (2010), Priestley’s
private correspondence demonstrates that in his actual linguistic output he
adhered to the principles of differentiation which he postulated in his work.
14 A mutative verb is one which indicates a transition from one place, location, state or
condition to another, e.g. go, come, return, become, grow, fade, disappear, etc. (Mustanoja
1960: 500 fn. 2, Rydén and Brorström 1987: 22).
50 chapter 3
Furthermore, his usage to a great extent chimes in with the more general usage
of the time.
The nineteenth century saw the most rapid switchover to have in all remain-
ing contexts, especially in informal language. By the end of the nineteenth cen-
tury, have had won out as the sole marker of the perfect, despite some fossilised
leftovers such as She is gone; those are arguably no longer taken to be perfects
at all, but are once again to be thought of as copular clauses with predicative
complements, devoid of actional meaning, as suggested by so-called ‘double
perfects’ (to be discussed below).15 (20) and (21) are two late examples, from
the 19th and the 20th centuries. Both of them must be interpreted as perfect,
not passive, since the verbs are intransitive (see Denison 1998: 137).
(20) … the extremity at which we are arrived [1864?, The Federalist Paper,
COHA]
(21) The warm sea wind was risen and blew over them now. [1963, Iris Murdoch,
The Unicorn, quoted in Visser (1963–1973: §1900), emphasis added]
Of course, syntactic phenomena should be investigated in their wider con-

text, including semantic considerations, though the scope of my survey will
necessarily have to remain more restricted. It was likewise noted in the intro-
ductory Chapter 2 that a change has its cognitive as well as its social facets.
Discussion of the linguistic and extra-linguistic factors influencing the process
of displacement of be by have can be found in Rydén and Brorström (1987: esp.
198–199), Shannon (1989), Rydén (1991: esp. 350), Kytö (1997: 31ff., 64), Smith
(2001), McFadden and Alexiadou (2010: esp. 396–397), contesting some of the
widely accepted conditioning factors to be listed below, Straaijer (2010), and
Los (2015: 75–77), as well as in Rydén and Brorström (1987: 18ff.) and Shannon
(1995: esp. 138ff., 142–143), for Germanic more widely. Linguistic factors include
the semantics and the frequency of the verbal lexeme: be persists longest with
frequent words like come and go. On the other hand, verbs of motion like come
and fall tend to host have more often than verbs which denote processes, like
become and grow. So do verbs of durational and prolonged action, such as
15 See Jespersen (1931: 31, 34), Mincoff (1958: 131–132, 188), Visser (1963–1973: §2052, §§2061–
2062), Strang (1970: 100), Mitchell (1985: §740), Rydén and Brorström (1987: esp. 116, 195ff.,
210–213), Rydén (1991: 346–347, 350, 353, n. 8, n. 11), Denison (1993: 344, n. 6, 359, 363, 367–
368, 1998: 135–136), Kytö (1994: 180, 1997: 19, 30), McFadden and Alexiadou (2006: 270),
Mugglestone (2012: 352–353); also see Harris (1984) for modern non-standard varieties.
travel and walk, unlike change-of-state lexemes such as change and recover,
which prove more resistant to have. Also of importance are:
– the tense: present vs. past perfect, the latter reportedly favouring have
(Shannon 1995: 150, though this is disputed by McFadden and Alexiadou
2010: 396–39),
– finiteness: with perfect infinitives and (less certainly) present participles
favouring have (Mustanoja 1960: 502, Shannon 1995: 148, 150; cf. McFadden
and Alexiadou 2010: 398–399 for caveats),
– modality,
– complementation patterns,
– etymology: apparently, borrowed main verbs contributed to the rise of
have, etc.
Extra-linguistic and social factors include time, region, type of text, level of
formality, and gender. As is to be expected, there is variation depending on
the individual author’s idiolect (regarded cognitively as well as socially).
Idiolectal peculiarities stand out clearly in the material presented in Rydén and
Brorström (1987: 201–206) and Rydén (1991: 351). For instance, Dickens appears
to be innovative concerning this ongoing change, while Jane Austen is more
conservative, as are other prominent female Victorian writers, like Charlotte
Brontë and George Eliot (cf. Cortes 2019). The various factors (e.g. age, gender
and social status) interact in complex ways. This can also be seen in Straaijer’s
(2010) study of the idiolect of the 18th-century grammarian Joseph Priestley.
Kytö’s (1994: 184–186) is a particularly valuable and detailed survey of the
distribution of be and have with intransitives in Early Modern English, featur-
ing discussion of relative frequencies and linguistic and extra-linguistic factors
at work in determining the choice of auxiliary. The Bible favours the traditional
use of be with intransitives, which is hardly surprising, but strangely enough,
private letters, comedy and fiction from the Early Modern period do not favour
the use of the rising form have either. In addition, the verbs come and go show a
clear preference for the construction with be more than other verbs, as already
noted in the preceding paragraphs. For the same period, the past perfect and
the perfect infinitive, both of which highlight the perfective aspect of ‘action’,
were among the constructions that paved the way for the rise of have with
intransitive verbs (cf. Straaijer 2010; Rydén and Brorström 1987: 189ff. cover
the 18th and 19th centuries, when this still obtains to a large extent, though
with some fluctuations due to complex interactions with the other factors).
In the language of Shakespeare, be-perfects are incompatible with adverbials
of duration and with modals (Kakietek 1976: 47; cf. Łęcki 2010: 159 for similar
observations concerning Old English). In Kakietek’s (1976: 49–53) corpus of
Shakespeare’s work, be is also more restricted than have in terms of the range
52 chapter 3
of meanings of the construction – the restriction is to the ‘current relevance/

result/effect’ meaning of the perfect (refer back to Section 3.1.1). In the next
section, I list further factors that favoured have and thus made it more frequent
and hence more viable as a perfect auxiliary.
3.4 Traditional Accounts for the Decline of the Be-Perfect
3.4.1 Frequency
Apart from the factors discussed in the previous section, have-perfects were
normally preferred to be-perfects in the following contexts during the ME and
EModE periods, while they were still in competition: the presence of durative
meaning, (22), recurrent/repeated actions, (23), statements containing adver-
bial expressions, (24), and hypothetical statements, (25), as well as negated or
questioned propositions.16
(22) and while I had a while goon [14th c., The Romaunt of the Rose, 135, from
Mustanoja (1960: 502), emphasis mine]
(23) a. þis þretty winter … hath he gone and preched [14th c., Piers Plowman
B xviii 293, from Mustanoja (1960: 502), emphasis mine]
b. The letters have come so regularly of late … [1714, Wentworth, cited in

Rydén and Brorström (1987: 65, 184), Rydén (1991: 350)]
16 Mustanoja (1960: 502), Shannon (1989: 677, 1995: 147–150), Huber (2019); cf. Fridén (1957),
Johannisson (1958), Rydén and Brorström (1987: 27–29, 184ff.), Rydén (1991: 350), Fischer
(1992: 260), Fischer and van der Wurff (2006: 141–142), and McFadden and Alexiadou
(2010), who express some doubts about negation. See Łęcki (2010: 159–162) for similar
examples and observations concerning tendencies in Old English, including adverbials
(or perhaps more accurately, locative complements), (i), and conditional/hypothetical
contexts, (ii).
(i) þa Scipia hæfde gefaren to ðære niwan byrig Cartaina …
when Scipio had gone to the new city Carthage
‘When Scipio had gone to the new city of Carthage …’ [ca. 880, from Łęcki (2010: 160)]
(ii) hæfde ða forsiðod sunu Ecgþeowes …
had then died [lit. for-journeyed] son Ecgtheow’s
‘Ecgtheow’s son would then have perished …’ [Beowulf, from Łęcki (2010: 161)]
Similar preferences regarding the usage of Samuel Pepys in the 17th century are noted by
Nakamura (1987: 43). Rydén and Brorström’s (1987) data covers the 18th and the 19th cen-
turies. Denoting hypothetical events indicates further grammaticalisation that has gone
beyond the encoding of temporal and aspectual relations.
(24) ye han entred into myn hous by violence [14th c., Chaucer, Canterbury
Tales, B The Prologue and Tale of Melibee 3001, from Mustanoja (1960:
502), emphasis mine]
(25) a. I wolde […] that I, Aurelius, Hadde went ther nevere I sholde have come
agayn [14th c., Chaucer, Canterbury Tales, F The Franklin’s Prologue and
Tale 971, from Mustanoja (1960: 502), emphasis mine]
b. I should not have been sorry if you had entered a little more into
Peninsular politics [1837, Wordsworth, from Rydén and Brorström
(1987: 186), Rydén (1991: 350)]17
Furthermore, when the mutative verb emphasised the action, this favoured
have, as opposed to emphasis on states, which favoured be (Fridén 1957,
Johannisson 1958, Mustanoja 1960: 502, Mitchell 1985: §722, §729, Rydén and
Brorstöm 1987: 17, 26, Shannon 1989: 677, 1995: 148, Rydén 1991: 350, Fischer
1992: 260, Rissanen 1999: 213; cf. Gräf 1888: 81ff.). All of this made the have-
perfect more frequent than the one with be, and since frequency is crucial
in grammaticalisation and entrenchment, this may be why have ultimately
pushed out be (Fischer and van der Wurff 2006: 142; cf. Bybee and Thompson
1997, Bybee 2003, Łęcki 2010: 24ff., and the discussion in Chapter 2). More
globally, since non-mutative verbs outnumbered mutative ones, making have
the more frequent marker of the perfect from the outset, it was poised to take
over anyway (see Traugott 1972: 145, cited in Denison 1998: 136; cf. Chankova
2008: 184).18
The token frequency of an item is how many times it occurs in a given
corpus – for our purposes, the occurrences of be or have as perfect auxiliaries
in a selection of running text. While token frequency was certainly important,
type frequency must have played a part as well. Smith (2001) notes that even in
the earliest OE, the displacement was already underway and it was influenced
by the fact that perfect have combined with many more types of verbal lexeme
than its underdog rival.
17 With regard to this last (and also rather late) example, Rydén (1991: 350, quoting Rydén
and Brorström 1987: 28, 186) notes that have is resorted to in order to ‘prevent potential
ambiguity in time-reference’, lamented as ‘disastrous’ by earlier researchers.
18 Frequency as a functional explanation of syntactic universals is discussed in Haspelmath
(2008a: Section 4, 2008b); Haspelmath (1999: Section 5) also considers the role of frequen-
cy in functional explanations more generally. See Bybee (2003, with references) on the
entrenchment of schemas with high token frequency.
54 chapter 3
By contrast, when be served as an auxiliary, it was used to build the passive

in the majority of cases in all periods of the history of English examined by
Kilpiö (1997: 115). This quantitative advantage of passive be over other types of
auxiliary be explains why be receded in its other auxiliary functions, including
its replacement by have in the perfect (though be is robust in the progressive).
Based on his statistics, Kilpiö (1997: 112) sums up his findings as follows:
Unlike the use of be as the auxiliary of the progressive, its use as a tense
[i.e. perfect] auxiliary […] is a receding feature. Although have had been
gaining ground at the expense of be with mutative intransitives since
the late Middle English period, the decisive steps in the process where
have superseded be were taken in the 18th and 19th centuries. […] One
possible conclusion to be drawn is that considering the preponderance
throughout the periods studied (1) of the copular uses of be and (2) of the
passive auxiliary uses of be compared to all the remaining auxiliary uses,
it becomes evident that the use of be as a tense auxiliary is marginal and
one which thus was perhaps more liable to be superseded by the have
perfect than not.
This explanation, however, does not tell us why be did not recede as a progres-
sive auxiliary, since the numbers of the progressive were initially not that great
either. Perhaps it did not because there was no other alternative. I return to this
and similar issues in Chapter 9.
3.4.2 Functional Load and Ambiguity

Apart from frequency, numerous researchers invoke functional load and am-
biguity as decisive factors for the demise of the be-perfect. These explanations
are similar to functional explanations involving the avoidance of homophony
in phonology (for a very critical appraisal of those, see Lass 1980: 75ff., as well as
the introductory discussion in Chapter 2). In the realm of diachronic grammat-
ical description, be is often said to have been ousted from the role in question
because of its greater functional load as an auxiliary for the perfect, progressive
and passive, which could lead to potential ambiguity.19 The combination of be
19 See Jespersen (1909–1949: IV 41), Mustanoja (1960: 501), Traugott (1972: 145, cited in
Kakietek 1976: 47–48), Zimmermann (1973), Rydén and Brorström (1987: 17, 197),
Kovatcheva (1989: 111ff.), Shannon (1989: 677), Rydén (1991: 347), Fischer (1992: 261–262),
Denison (1993: 366, 1998: 136, 183–184), Kytö (1994: 182, 1997: 18, 28), Elsness (1997: 246),
Kilpiö (1997), Rissanen (1999: 215), Fischer and van der Wurff (2006: 142), Chankova (2008:
137–138), Łęcki (2010: 163–164), Brinton and Arnovick (2011: 372). The references range
followed by the past participle of a verb that had both transitive and intran-
sitive uses could be construed as perfect, as well as passive, or a resultative
stative, as in (26) and (27).20
(26) Our hopes are again revived of seeing the Viceroy of Mexico. [1797, Nelson,
Letters, ed. Naish (1958) 190 p. 328 (30 Jun.), cited in Rydén and Brorström
(1987: 24), Denison (1998: 184)]
(27) … this house is turn’d bawdy-house, they pick pockets [Shakespeare, Henry
IV, Part I, III 3.69, quoted in Kakietek (1976: 46), emphasis mine]
Rydén (1991: 347) sums up this position rather succinctly:
The chief actuating force in the elimination process of be as ‘perfect’

marker, giving the directional impetus to the whole process, was appar-
ently the functional diversity of be – as copula and as passive and perfec-
tive marker, which could give rise to functional and systemic ambiguities.
Earlier on, the increasing use of be as passive marker favoured have as
perfective marker. Another factor was the polyconstructional character
of certain verbs.
An analogous scenario is sketched out by Fischer (1992: 261–262), once again

giving a taste of this type of popular explanation:
from introductory textbooks to more serious scholarly publications, spanning more than
a century, which shows how widespread and deeply entrenched this view is. Kovatcheva
(1982: 14), among others, proposes an analogous explanation for the shift from impersonal
to personal constructions.
20 Again, see Jespersen (1909–1949: IV 41), Mincoff (1958: 187–188), Kakietek (1976: 46–47),
Mitchell (1985: §735), Rydén and Brorström (1987: 24, 100 fn. 11, 147), Fischer (1992: 261),
Denison (1993: 345–346, 1998: 183–184), Kytö (1994: 181), Elsness (1997: 243, 339), Rissanen
(1999: 215), Fischer and van der Wurff (2006: 142), Straaijer (2010: 75), Brinton and Arnovick
(2011: 372). It remains unclear, however, why English seems to ‘have no objections’ to the
verb be still serving as the auxiliary of the progressive and the passive – probably as a
result of the lack of ambiguity in the form of the following non-finite verb? The availabil-
ity of separate non-finite participial shapes in Bulgarian/Macedonian probably likewise
enabled be to function unproblematically as perfect and passive auxiliary. On the use of
the Germanic so-called ‘past (or second) participle’ in the passive and the perfect, see
Anderson (1997).
56 chapter 3
Finally, it remains to consider the reason(s) why have ousted be in the

formation of the perfect. Various factors are at work (see Zimmermann
1973): (a) the greater functional load of be (used as an auxiliary of the
passive, progressive and perfect) compared to have (at this stage only
perfect) and the ambiguity that could arise because of this (i.e. be + past
participle of a transitive verb could be perfect as well as passive; the pro-
gressive in -ing is sometimes found written as -en, the same ending as the
past participle of strong verbs) made have a more suitable candidate for
the perfect; (b) be had become the auxiliary par excellence of the passive
voice.
The relevant passage from Rissanen (1999: 215) is pretty much in the same spir-
it, where Rissanen is confident that the reasons are rather self-evident:
The reasons for the loss of be are fairly easy to find. The functional load
of be was heavy as this verb was not only used as the copula but also
in the be + -ing structure and in the passive. It was particularly the last-
mentioned function that easily caused ambiguity in expressions such as
was grown, was developed, etc.
Denison (1998: 136) adds that functional load and ambiguity may look a lit-
tle post hoc as explanatory factors, but he nevertheless finds them ‘plausible
enough’. The following examples are also ambiguous in the same way. Although
Visser (1963–1973: §2073) treats them as perfects, they could be passives too.
(28) how thou art altered with thy travel! [1598–9, Ben Jonson, The Case Is
Altered, quoted in Visser (1963–1973: §2073), emphasis added]
(29) The whole aspect of the room was now altered. [1911, A. Bennett, Hilda
Lessways, quoted in Visser (1963–1973: §2073), emphasis added]
Kakietek (1976: 47) notes in passing that this type of ambiguity was a rare phe-
nomenon in Old English; according to him, this was due to the fact that in
Old English most transitive and intransitive verbs had distinct shapes, and in
addition, resultative stative past participles were arguably adjectival and often
inflected (as already pointed out in earlier sections).
Confusion of analysis might be caused even by verbs which normally belong
to the class of mutative intransitives. Quirk et al. (1985: 170 n. [a]) actually refer
to cases like her friend was gone as pseudo-passives, as did many eighteenth-
century grammarians (see Rydén and Brorström 1987: 208, Denison 1998: 136,
Rissanen 1999: 215, Straaijer 2010: 65–66). In addition, as Mustanoja (1960: 500)
notes, an important difference between OE wesan/beon ‘be’ and OE habban
‘have’ is that the former never fully lost its capacity to express states, as op-
posed to actions; this was mentioned above and can be seen clearly in (30),
where ambiguity between a passive and a copular reading persists to this day.21
(30) we could not go in because the door was closed. [Mustanoja (1960: 500),
emphasis added]
Crucially, be is said to have been temporally ambiguous between past action

and present state resulting from a past action, so it sometimes needed to be
strengthened by have in what was essentially double-marking, as shown in
(31).22 Have acts as a proper perfect auxiliary and be more as a copula. This
double-marking occurs in Middle and Modern English, probably in order to lay
emphasis on the result or current relevance. In Present-Day English, using both
perfect have and perfect be + an intransitive past participle only works with
the verb go, as in (31b), and by now gone is probably analysable as an adjectival
subject complement (Denison 1998: 137–138). Interestingly though, even the
participle of a phrasal verb, which should arguably retain more verbal than ad-
jectival properties compared to its non-phrasal counterpart, could participate
in such a construction – (31a) and (31c).
(31) a. he has been come over about ten days [Swift, Journal to Stella II 625,
from Rissanen (1999: 215)]
b. Cher Frere has been gone since four o’clock this morning to a private con-
ference. [1788, Betsy Sheridan, Journal 42 p. 131 (21 Nov.), from Denison
(1998: 137)]
c. Richard had been gone away some time, when a visitor came to pass a
few days with us. [1853, Charles Dickens, Bleak House, Chapter XXX]
21 Compare Denison (1993: 361) and Chankova (2008: 87) for counterarguments. The ambi-
guities between passive (actional/dynamic) and copulative (statal/stative) be are aptly
illustrated in Mincoff (1958: 187–189) and Quirk et al. (1985: 167–170).
22 Further discussion can be found in Jespersen (1931: 36), Visser (1963–1973: §2162), Mitchell
(1985: §735), Rydén and Brorström (1987: 25), Denison (1993: 363, 1998: 137–138), Rissanen
(1999: 215), Łęcki (2010: 164), McFadden and Alexiadou (2010: 415). Ammann (2005) deals
with double perfects in German. Also see Visser (1963–1973: §2023) and Bowie et al. (2013:
Section 3.2.3) for other double perfects with have:
(i) I have done told you. [from Visser (1963–1973: §2023)]
58 chapter 3
Only two pages after the example in (31c), the chapter’s opening sentence,
Dickens gives an indication that this construction was most probably adjecti-
val for him:
(32) However, as he is now gone so far away, and for an indefinite time, and as he
will have good opportunities and introductions, we may consider this past
and gone. [1853, Charles Dickens, Bleak House, Chapter XXX]
Placing gone in object complement position strongly suggests adjectival status

for both occurrences of the participle in (32) – subject complement (or loca-
tive predicative complement) in the first instance and object complement in
the second. A few lines down, Dickens uses have with gone when the meaning
is more dynamic and therefore verbal:
(33) Not to be always selfish, talking of my son, who has gone to seek his
fortune, and to find a wife… [1853, Charles Dickens, Bleak House,
Chapter XXX]
(34) also confirms that be gone was a copular construction for Dickens. He uses
it in a temporal subordinate clause introduced by while to indicate a stretch
of time, where the dynamic perfect would not be allowed (cf. *Take care of
Pa while I have gone). The meaning is clearly ‘while I am away’, as in (31)–(32)
above.
(34) Take a little care of Pa while I am gone, mama! [1853, Charles Dickens,
Bleak House, Chapter XXX]
3.4.3 Functional Load and the Disappearance of OE weorðan

Apart from be, Old English (as well as other Germanic varieties) could build
passives with a verb meaning ‘to become’ – OE weorþan/weorðan, cognate
with German werden ‘become’, Latin verto ‘turn’, and Bulgarian въртя/vărtja
‘turn’:
(35) He wearð ofslegen from his agnum monnum

he became slain from his own men
‘He was slain by his own men’ [Alfred’s Orosius, from Brinton and
Arnovick (2011: 228), emphasis and gloss mine]
More broadly, OE weorðan could be used as an intransitive full verb, as a

copula with nouns, adjectives or PPs headed by to, or as a perfect and pas-
sive auxiliary, with the past participles of intransitive and transitive verbs
respectively.23 It has been proposed (and also argued against) that, when
combined with the past participle of a transitive verb to form a passive(-like)
construction, beon and wesan denote a resulting state, whereas weorðan de-
notes a process/action.24 There are also attestations of wearð geworden ‘had
happened’, testifying that this verb can occur as an auxiliary and a lexical verb
in the same clause.
Unlike what happened in other Germanic languages, OE weorðan ‘become’
for some mysterious reason disappeared as an alternative passive auxiliary,
leaving be as the only marker of the passive. According to some scholars, the
disappearance of passive weorðan thus created a ‘conspiracy’ to make be even
more unpopular in the context of the perfect, again due to excessive functional
load.25 Rissanen (1999: 215) notes that German, which has no actional passives
with sein ‘be’, retains the sein/haben ‘be/have’ distinction in the perfect, unlike
standard Swedish and English, both of which rely on be-passives (cf. Fischer
and van der Wurff 2006: 140, and Chapter 6 herein).
The Middle English reflex wurthen could have played a more significant
role, just as it does in English’s West Germanic relatives German and Dutch.
However, even in the earliest attested Old English, weorðan was arguably some-
what infrequent in the passive, becoming even more so in Early Middle English,
when the most significant shifts in the newly emerging auxiliary system were
taking place. It remains unclear why exactly ME wurthen disappeared: poten-
tial reasons have been sought in the nature of the verb itself or perhaps there
might have been some foreign influence – Scandinavian or French (see Fischer
1992: 262; cf. Mincoff 1958: 188, Mustanoja 1960: 616–619, Mitchell 1985: §801,
McWhorter 2002, Petré 2010, 2013, 2014).
Kovatcheva (1989: 109–114) puts forward three scenarios for the demise of
OE weorðan/ME wurthen: the first is that OE beon acquired a dynamic inter-
pretation (which, however, it always had, as established above!); alternatively,
the whole thing was arbitrary (the second scenario); the final scenario, which
is the view Kovatcheva seems to subscribe to, is that, unlike combinations with
23 On weorðan as an alternative to the be-perfect from OE to the early fourteenth century, as
well as on ambiguous perfects/passives with OE beon and OE weorðan, see Visser (1963–
1973: §1897), Mitchell (1985: §671, §734ff.). Denison (1993: 344–346) points out that perfect
weorðan ‘is fairly uncommon in Old English and rare in Middle English’, making its status
as perfect auxiliary uncertain (cf. Łęcki 2010: 227 n. 15).
24 See Mitchell (1985: §§673–674, §§786–787, §§789–801), who doubts the universal validity
of this distinction (cf. the discussion of Modern German and of get-passives below): there
is evidence that OE beon/wesan + past participle could denote an action or process, in
addition to denoting states (more details will be given in the discussion below); likewise,
weorðan could express both states and actions.
25 See Mustanoja (1960: 437–440, 616–619), Strang (1970: 351), Zimmermann (1973), Fischer
(1992: 261–262), Chankova (2008: 185).
60 chapter 3
OE beon, those with OE weorðan were not being grammaticalised – they were
just free syntactic combinations (but that ignores the collocation of wearð
geworden). Arbitrariness might be best squared with the facts.
3.4.4 Ambiguity in the Contexts of Coordination and Contraction

Lack of explicitness also thrives in the context of coordination. In (36), gedon
‘done’ is most likely a passive participle and geworden ‘happened’ (from
weorðan) an intransitive perfect participle.26 The two participles are ‘coordi-
nated complements of a single occurrence of be’ and such examples under-
mine the distinction between passive and perfect (Denison 1993: 345–346).27
(36) & man cydde Harolde cyng hu hit wæs þær gedon
and one informed Harold king how it was there done
& geworden
and ?happened/?made
‘and King Harold was informed how things had gone there’ [ChronE
(1066), from Denison (1993: 345–346), emphasis mine]
Moreover, Fischer (1992: 262) suspects that the loss of perfect be was perhaps
additionally facilitated by the fact that ‘in co-ordinate constructions the aux-
iliary was usually not repeated so that have was often used where be was ex-
pected’. However, it is not entirely clear why this should not work vice versa, in
favour of be. This is probably due to the far greater frequency of have (and of
transitive verbs), but the point needs to be studied and quantified in more de-
tail. Be that as it may, coordination is one more ambiguous context in which re-
interpretation might sneak in – a construction intended by the speaker/writer
26 On the properties of the ‘past’ participle in English, perfect as well as passive, see Anderson
(1997: esp. 8), who points to its adjectival origins and relates it to the development of
the perfect periphrasis (cf. Mitchell 1985: §§23–24, §728, §734, §777). As remarked above,
in earlier English (proto-)perfects the past participles of intransitive verbs are active in
terms of meaning, while those of transitive verbs are passive (cf. Sections 3.1.2, 3.2).
27 More constructions of this kind are discussed in Mitchell (1985: §735). Compare a similar
flexibility of progressive and passive (stative?) be, (i), or progressive and perfect be, (ii),
in Victorian English, as well as of the contraction ’s in PDE, (iii). Also see the Epilogue to
Chapter 5.
(i) … lest the object of his distrust should be looking over, or hidden on the other side.
[1853, Charles Dickens, Bleak House, Chapter XLVI]
(ii) … people are going or gone to bed [1853, Charles Dickens, Bleak House, Chapter LVI]
(iii) to make sense || of all that’s changed and changing [poem by Tom Phillips, http://
msvstp.blogspot.bg/2016/12/trees-in-portishead.html, accessed on 30 Dec 2016]
to contain/imply be might get reanalysed by the hearer/reader as containing/

implying have.
Also conducive to reinterpretation is the oft-cited influence of the contract-
ed form ’s, which can stand for is as well as has, although it is a little worrying
that the evidence for that comes from post-Middle English times.28 (37) is one
such ambiguous example from 1955. Note that in addition it contains asyndet-
ic (i.e. conjunctionless) coordination, with the last verb being transitive and
therefore expected to take have in the perfect (also see fn. 43).
(37) Anyhow, she’s gone, walked out, slung her hook. [1955, L. P. Hartley,
A Perfect Woman, cited in Visser (1963–1973: §1902, p. 2072), emphasis
mine]
More radically, ambiguities may arise in the so-called clipped perfect. What
defines the clipped perfect is its missing auxiliary, as illustrated in (38). An am-
biguous context is presented in (39), where He come may be expanded either
as ‘is he come’ or as ‘has he come’ (Visser 1963–1973: §2054), unless come is an
infinitive/base form.
(38) You ever been to the Riviera? [from Visser (1963–1973: §2054)]
(39) ‘Tis miraculous that you tell me, sir. He come to woo our lade mistress for
his wife! [1612, Chapman, The Widow’s Tears (ed. Smeak) I, i, quoted in
Visser (1963–1973: §2054)]
3.5 Problems for the Traditional View
Things are not as straightforward as the foregoing account might suggest. There
are problems of four kinds: firstly, there are counterexamples to the assump-
tion that agreement marking on OE past participles is incompatible with a
perfect reading, as noted in earlier sections; secondly, instead of being ‘neatly’
restricted to the perfect, have seems to have dramatically increased its func-
tional load over time, fostering a greater diversity of constructions it can par-
ticipate in (including a have-passive); thirdly, alternative passive and perfect
28 See Jespersen (1931: 30, 41), Mincoff (1958: 131), Visser (1963–1973: §1898), Rydén and
Brorström (1987: 197), Shannon (1989: 677), Fischer (1992: 262), Denison (1993: 366, 1998:
136), Kytö (1994: 181, 1997: 18, 30), Fischer and van der Wurff (2006: 142), Brinton and
Arnovick (2011: 372).
62 chapter 3
structures appear in puzzling competition with be and have, including con-

structions relying on get, although be and have are not expected to face such
competition now that they are arguably less ambiguous and less overloaded
than they used to be (due to the loss of the be-perfect); finally, the demise of be
as a perfect auxiliary does not seem to go hand in hand with a higher propor-
tion of passive be.29 Neither do we see any significant increase or decrease in
ambiguity over time. All of these concerns cast doubt on attaching such great
importance to functional load and ambiguity in order to explain the grammat-
icalisation and subsequent development of the English perfect.
These setbacks will each be discussed in turn below. Furthermore, if this
switchover was so functionally desirable, one might legitimately wonder why it
took such a long time to complete. As observed by Rydén (1991: 352), the accep-
tance of have as the only marker of the perfect ‘took an unusually long time,
i.e. about 1,000 years’. It remains mysterious how the language and its speakers
(especially when acting as hearers) ‘coped’ at all during this inordinately pro-
tracted period of ‘indeterminacy’ and ‘non-optimal functionality’.30
3.5.1 Agreement and Perfect Readings

It was stated above that lack of agreement on OE participles facilitated the
grammaticalisation of the perfect, in contrast to the earlier construction with
an agreeing participial object complement and have’s original stative meaning
of possession. However, agreement on the participle in OE is not incompatible
with a perfect interpretation, as demonstrated by the following example:
(40) Ac hie hæfdon þa … hiora mete genotud-ne

but they had then … their food(m)[acc.sg] used.up-m.acc.sg
‘But they had then used up their food.’ [Anglo-Saxon Chronicle, MS A,
894, from Trask and Millar (2007: 177), also quoted in Mitchell (1985: 293);
cf. Elsness (1997: 262)]
As Trask and Millar (2007: 177) observe, since the food is all gone, it cannot
be available or in someone’s possession (see Mitchell 1985: §726, who pro-
vides additional examples). This example, therefore, cannot receive a stative
29 An anonymous LAGB reviewer objects that the fourth of these points “does not appear
to problematise the traditional functional load account. This is because the frequency of
the passive (regardless of how it is constructed) presumably depends on speakers’ com-
municative needs; why should they all of a sudden wish to express the passive function
more often when be recedes from the perfect paradigm?”
30 Kovatcheva (1982) raises similar concerns regarding the transition from impersonal to
personal constructions.
interpretation; instead, it must be a perfect, despite the unambiguous agree-

ment on genotudne. Furthermore, Brinton (1988: 101) notes that ‘have + past
participle constructions with possessive meaning (with the object in the pos-
session of the subject) are at best rare in Old English’. In essence, there are
indisputable perfects showing agreement, and conversely, there are clearly
adjectival statives where the participle is not inflected (see Los 2015: 73).
Agreement is therefore not a reliable guide to grammaticalisation. At best, the
ultimate falling off of agreement morphology might be seen as a consequence
of the grammaticalisation process (following a period of variation), but not so
much as a trigger, as evidenced here and in the following chapter.
3.5.2 Proliferation of the Functions of Have

3.5.2.1 Passive Have
Just as be was being displaced as a marker of the perfect in the eighteenth
century, supposedly because of its greater involvement in the passive, have,
which is believed to have won out because it had been restricted to the perfect,
actually came to participate in the formation of passive-like constructions,
thereby confusingly becoming more ambiguous itself (Denison 1993: 342–343,
Moessner 1994: 20, Rissanen 1999: 215–216). Ever since Old English, be has been
the default passive auxiliary. In Early Modern English, however, both have and
get came to be employed in a kind of rival passive structure in certain circum-
stances. In (41) and (42) I illustrate have and return to get below.
(41) If they had any parte of their liberties withdrawne [1568, Grafton, Chron. II
141, OED s.v. have 18, cited in Rissanen (1999: 216), emphasis added]
(42) Another had one of his hands … burnt. [Defoe, Robinson Crusoe II 10, OED
s.v. have 18, cited in Rissanen (1999: 216), emphasis added]
This passive-like construction, of the type He had a book given (to) him, is
paraphrasable with established passives, e.g. ‘A book was given (to) him’ or
‘He was given a book’. It can be traced back to the middle of the fourteenth
century, or arguably even earlier. Initially relatively infrequent, it became fully
established by the beginning of the Modern English period, and might now
sometimes be preferred to the regular passive, He was given a book (Visser
1963–1973: §1964, §1979, §§1980–1984, §§2116–2122). As mentioned above, the
earliest example in Visser dates back to the 14th c.:
(43) Bot of Oðer wommen tua [he] Had four suns geten him. [13.. Cursor M.,
3900, quoted in Visser (1963–1973: §1979), emphasis added]
64 chapter 3
Łęcki (2010: 196) gives the following as the earliest attested have-passive, on
the cusp between Old and Middle English.
(44) He hæfde an fet to ðam anum iwroht.

he had a vessel for that alone worked
‘He had a vessel made for that purpose alone’ [c. 1175 (?OE), from Łęcki
(2010: 196), emphasis and translation added]
However, (44) could receive a causative interpretation (see comments in previ-

ous sections, as well as those below). (45), unless it merely expresses posses-
sion (i.e. ‘they had names which were produced on the basis of their fathers’
names’), could serve as another ME example from the 14th century:
(45) In þe same manere kynges sones of Engelond hadde

in the same manner kings’ sons of England had
names ischape by hir fader names
names given (shaped) after their father names
[a1387, Trev. Higd., from Łęcki (2010: 196),
emphasis added]
The role of the subject in this type of construction with have is more active
than in the be-passive, and this structure often has a causative meaning too,
as in (44) (see next section). The emergence of the so-called have-passive can
be associated with the subjectivisation of the recipient argument. The noun
phrase which denotes the person in the clause gets topicalised in either type of
construction: the passive with be, He was given a book, or the rival with have, He
had a book given to him. The rise of the have-passive has likewise been linked to
dispelling ambiguity. As Rissanen (1999: 216) observes:
Moessner points out that in [He had a book given to him] there is no risk of
even momentary ambiguity as to the semantic role of the subject; theo-
retically speaking, he in [He was given a book] could be analysed either as
the direct or the indirect object of the corresponding active clause until
the post-verbal elements are heard or seen.
However, why should the have-passive come into being and gain in frequency
when it adds to what is essentially a greater functional load and potential for
ambiguity in the roles of have? As it were, it goes back to the original mean-
ing of the ancestor of the perfect – the complex transitive clause indicating
possession. It could therefore be thought to have branched off as another de-
scendant of the original prototype (see Chapter 2, Section 2.2). Even if the two
structures have coexisted all along, what we see is hardly a reduction in the
range of functions; quite the contrary, the duality of perfect- and passive-like
uses is still with us, including intermediate cases. This ties in well with the
points made in Chapter 2 about layering and coexistence of input and output
constructions after they undergo grammaticalisation.
There can even be double-marking, as in (46), which shows the perfect and
the quasi-passive with have. The string of three verbs means that the first two
must be auxiliaries. Since it represents a new ‘layer’ of auxiliary, the second
have is still more ‘lexical’ than the first one in terms of NICE properties, in-
cluding the ability of auxiliaries to host negation and invert with the subject,
among others (more on this is given below). As the context indicates, the mem-
bers of staff do not have their passes any more, so a meaning of possession is
ruled out. There can be no causative meaning either, as they certainly did not
want to have their passes revoked (see below for the causative construction).
Similarly, the referents of the subject in (47) sadly do not have legs any more,
so the have-passive can surely exhibit semantic bleaching.
(46) Several members of staff have had passes revoked. [Sky News, 25 March
2016]
(47) They had legs amputated. [Sky News, 22 April 2016]
Such examples are not at all rare and can easily be multiplied (e.g. … had prop-
erty confiscated – from a History Channel documentary, etc.).
3.5.2.2 Causative Have

Have also came to participate in causative constructions such as (48), again
mostly from the eighteenth century onward. In its causative sense, have can
be followed either by a past participle or a bare infinitive, among some other
options (see Quirk et al. 1985: 132, 1412 n. [b], Hollmann 2003: Chapter 3, Łęcki
2010: 189–190). Like (46), (48) also contains two auxiliary uses of have, one per-
fect and one causative.31 It is antedated by the earliest example of causative
have recorded by Visser, given in (49), from the Middle English period – the
year is 1205. A much later example, from the 20th century, in (50), confirms
31 As Łęcki (2010: 189) correctly notes, “causative HAVE in English is only grammaticalised
to some extent in that it could be treated as an idiomatic use of HAVE rather than as an
auxiliary since it virtually does not exhibit any NICE properties […], nor are there any
signs of erosion or fusion …”
Grammaticalisation is only incipient here, involving the pragmatic-semantic dimension,
with no concomitant morpho-syntactic and phonological changes (cf. Krug 2001).
66 chapter 3
that the construction is still pretty much alive and well and shows no sign of
fading away.
(48) He … would have had me dined [edn: dine] with him [1711, Swift, Jnl. to
Stella 230. 27 (3 Apr), cited in Denison (1993: 342), emphasis added]
(49) a. He hadde þare tweie castles bi-walled swiðe faste. [c. 1205, Layamon,
quoted in Visser (1963–1973: §2118), emphasis added]
b. He hafde þer tweie castles biclused [Otho: bi-walled] swiðe uaste

he had there two castles fortified with walls very fast [version quoted
in Łęcki (2010: 193), who dates it c. 1275 (?a1200)]
(50) one went there to have his teeth pulled, his wagon fixed, his heart listened
to … [H. Lee, To Kill a Mockingbird, quoted in Visser (1963–1973: §2118)]
All of the sentences in (49)–(51) can be paraphrased with ‘X causes something

to be done’. Interestingly, (49) and (50) contain undeniable traces of passive
as well as causative meaning, in line with the nature of the past participle.
More generally, The guard patrol had two men shot can mean either ‘The pa-
trol caused two men to be shot’ or ‘The patrol suffered the loss of two men by
shooting’ (Quirk et al. 1985: 1207).32
A similar construction with a causative meaning is actually also attested
from Old English (see Kilpiö 2013 and the Dictionary of Old English, http://
tapor.library.utoronto.ca/doe/, accessed on 27 Feb 2017):
(51) and hæbbe him gæworht of cwicbeame feower Cristes mælo

and have him worked of quickbeam four Christ’s crosses
‘And let four crosses of aspen wood be made for him’/‘And let him
have four Christ’s crosses of rowan made for him’ [Ags. Metric. Charms,
quoted in Visser (1963–1973: §1985), Kilpiö (2013: 104); emphasis and
glossing mine, second translation from Kilpiö]
32 Exemplifying each type of construction, Łęcki (2010: 196) comes up with the following
sentences (though the first three are very close in meaning and are in practice often dif-
ficult to tease apart; see also Mitchell 1985: §704, and Johannsen 2016):
(i) I had my testicles removed when I was a child. [experiential; alternatively treated as
passive here]
(ii) I had my testicles removed by the doctor. [passive]
(iii) I had my testicles removed to spite my wife. [causative]
(iv) I had the doctor (?to) remove my testicles. [causative]
Causative have + past participle evolved from the same stative/resultative

structure which gave rise to the perfect. Łęcki (2010: 192) confesses that ‘it is
not easy to provide unequivocal examples of causative HABBAN in Old English
because of the sameness of the surface structure of the causative, experien-
tial, stative and perfect constructions’. So, have was by no means a stranger to
uncertainty of interpretation, fostering a diversity of semantic readings. The
following is given as a convincing example of the causative in OE:33
(52) hæfde Hæsten ær geworht þæt geweorc æt Beamfleote …

had Hasten before made the fortress at Benfleet
‘Hasten formerly had the fortress built at Benfleet …’ [Anglo-Saxon
Chronicle, 900–950, from Łęcki (2010: 192)]
I close this sub-section with an interesting question to ponder: if passive and

causative have are attested from so early on, why have they not developed
NICE properties, similarly to perfect have?
3.5.2.3 Modal Have

Have was additionally recruited as a semi-modal, also referred to as a lexical
modal, with yet more meanings – those of obligation (deontic, sometimes re-
ferred to as ‘root’), (53), and logical conclusion (epistemic), (54) (see Quirk et
al. 1985: 145–146, Krug 2001, Huddleston and Pullum et al. 2002: 205–206, Łęcki
2010: 81ff.). The epistemic meaning of a modal generally evolves later from its
deontic uses.34
(53) You have (got) to come in now. [Huddleston and Pullum et al. (2003: 205)]
33 The causal implication of Latin fecit might also be rendered with hæfþ in OE (among
other constructions), suggesting possible causative overtones for OE habban from very
early on, as in (i) (see Mitchell 1985: §726 fn. 173). Alternatively, hæfþ in (i) is simply a
marker of the (proto-)perfect and the causality is inherent in the semantics of heal, ‘to
make whole/healthy’.
(i) fides tua te saluum fecit [Latin, Mark 10:52, from
faith thine thee healthy has.made Mitchell (1985: 293)]
þin agen geleafa þe hæfþ gehæledne [OE, Blickling Homily, from
thine own faith thee has healed Mitchell (1985: 293)]
‘thy faith hath made thee whole’ [Mark 10:52, King James Bible]
34 This is similar to the story of ought to, which derives from the OE verb agan ‘to own’ (cf.
Kovatcheva 1989: 143, Łęcki 2010: 132, and Section 3.1.3 above).
68 chapter 3
(54) This has (got) to be the worst restaurant in town. [Huddleston and Pullum
et al. (2002: 205)]
The auxiliation of have to is traced by Łęcki (2010: Chapter 3) and Fischer (2015)
(cf. Mitchell 1985: §§950–953, and the Dictionary of Old English http://tapor.
library.utoronto.ca/doe/, accessed on 27 Feb 2017). According to a traditional
grammaticalisation account (disputed by Fischer), the development went
through stages reminiscent of the formation of the perfect: from I [haveV [a
book to recommend]NP:Object], with a full transitive have and an infinitival ad-
junct/post-modifier which depends on the object noun, to I [[have to recom-
mend]VP [a book]NP:Object], with the newly established unit (see Łęcki 2010: 80,
86–91, Fischer 2015: 121).35 In the course of late Old English and Middle English,
the possessive meaning is believed to have slowly bleached and given way to
a deontic colouring of obligation in conjunction with the to-infinitive, accom-
panied by a concomitant syntactic reanalysis. As a result of the reanalysis, the
infinitive is now the complement of the deontic semi-auxiliary have and is no
longer a modifier/adjunct in the NP, while the original object of have is now
the object of the infinitive.36 The word order has been ‘reshuffled’ according-
ly. The following Old English example contains shades of necessity as well as
possession:
(55) uton we forþon geþencean hwylc handlean we him forþ

let us therefore consider what recompense we them forth
to berenne habban
to bear have
‘Therefore, let us think what recompense we have to offer them.’ (i.e.
‘… we have recompense to offer them’ or ‘... we have to offer them recom-
pense’) [971, from Łęcki (2010: 88, 219–220 n. 10)]
(56) is a less ambiguous example from the Canticles of the Cambridge Psalter
(which, unsurprisingly, must have been heavily influenced by the Latin
original):
35 Alternatively, the grammaticalised construction can be rebracketed as follows: I [have

[to recommend a book]]. This bracketing is more consistent with current syntactic theory
than the traditional one above.
36 Have and recommend can be said to share their object in I have a book to recommend. Or
rather, the implied object of recommend is the same as the overt object of have.
(56) Ad cuius aduentum omnes homines resurgere habent cum

To þæs tocyme ealle men to arisanne hi habbað mid
to his coming all men to rise they have with
corporibus suis …
heora lichaman …
their bodies
‘With his [i.e. Jesus’] advent all men have to arise with their bodies …’
[from Łęcki (2010: 108); the Dictionary of Old English gives arisan habbað]
There is no NP object here that could serve as an argument of either habbað

or the infinitive to arisanne. Since habbað takes no NP object, it cannot denote
possession. Judging by the context, any meanings of ability or possibility are
also ruled out, since, in Christian doctrine, all human beings, dead or alive, are
to be judged, so the judgement is neither a matter of possibility nor of ability.
As Łęcki (2010: 109) sums up:
This example shows beyond doubt that HABBAN must have operated as
an auxiliary of necessity accompanied by an inflected infinitive already
in Old English.
The final stage of this grammaticalisation process is conventionally placed

during the Middle or Early Modern English period, when inanimate subjects
appear, as do intransitive infinitives in greater numbers (Łęcki 2010: 90–91, 129–
131, Fischer 2015: 121–122). Actually, both the purported original construction
and the one with the deontic (and epistemic) quasi-auxiliary coexist today.37
The frequency of have + to-infinitive increased from 17 occurrences in 413,000
words in Old English (0.004%), through 150 occurrences in 608,500 words in
Middle English (0.025%), to 339 occurrences in 551,000 words in Early Modern
English (0.06%) (Łęcki 2010: 99, quoting Fischer). Focusing on the modern pe-
riod, Close and Aarts (2010) report a dramatic decrease in frequency of the core
modal must and a significant increase of the frequency of the semi-modal have
to (but not have got to)38 in British English between the 1960s and the 1990s,
at least in its deontic/root uses (cf. Krug 2001, Tagliamonte 2004). Recent data
37 Though Fischer (2015) raises some legitimate objections to the traditional outline above,
they would take us too far afield in our pursuit.
38 Have got to was grammaticalised later than have to – see Łęcki (2010: 138–142) and Section
3.5.2.4 below.
70 chapter 3
from 21st-century American English, where have (got) to/(have) gotta is again
much more popular than must, can be found in Łęcki (2010: 225 n. 54).39
Finally, there are indications that, apart from being omissible, as in gotta
(you got to laugh at me, 1911, from Łęcki 2010: 142), have in this construction can
now be mixed up with be, as in (57)–(58). I return to this interchangeability of
be and have in Sections 3.5.3.2 and 3.5.3.3.
(57) What I’m gotta do first … [BNC, cited in Łęcki (2010: 142), quoting Krug,
emphasis mine]
(58) I’m gotta finance industry so … [BNC, cited in Łęcki (2010: 142), quoting
Krug, emphasis mine]
Another modal use of have is in the had better idiom (see Quirk et al. 1985:
141ff.). Strangely enough, had better replaced an Old English construction
made up of the OE verb beon ‘be’ accompanied by bet(ere) ‘better’ and a dative
NP followed by a that-clause (Łęcki 2010: 69–70), again showing paradigmatic
and cognitive links between verbs of being and possession:40
(59) betere him is þæt he þæs dæges hit forga

better him.3sg.m.dat is that he.3sg.m.nom that day’s it forgo
‘It is better for him to refrain from it this day.’ [Łęcki (2010: 70)]
39 Interestingly enough, be also developed (semi-)modal uses in combination with an infini-
tive, or maybe had them all along, as in (i). Such uses, though with an active instead of a
passive infinitive, are attested as early as Old English, as in (ii), perhaps at least in part in-
fluenced by Latin originals (see Fischer 1991: 146–151, 174ff.; cf. Mitchell 1985: §§934–944).
Latin influence or reinforcement is also possible for the use of have as a deontic auxiliary
in OE (see Łęcki 2010: 114–118). In Latin, habere + infinitive could refer to the future, as
well as to obligation, (iii) (cf. the translation example above, as well as Chapter 7 for a
have-based future in Bulgarian).
(i) They are to be blamed therefore. [from Fischer (1991: 146)]
(ii) quanti psalmi dicenda sunt [Latin]
hu fela psealma … to singenne synt [Old English]
how many psalms to sing are
‘how many psalms are to be sung’ [from Fischer (1991: 149)]
(iii) Haec canta-re habe-o
these sing-inf have-1sg
‘I have these to sing’ or ‘I have to sing these’ [from Łęcki (2010: 114)]
40 Kovatcheva (1982: 17, 25), Bauer (2000: 197–203, 211), Danylenko (2005), as well as
Grković-Major (2011), highlight a similar interchangeability of generalised possessive
schemas involving ‘I have something’/‘Something is at/to me’, e.g. Latin pugnatum mihi
est ‘fought is to me’, i.e. ‘I have fought’ (cf. Heine 1993: 31–32, 1997). In Hiberno-English,
possession can be expressed with be + near/at/by, as in There wasn’t any money by that boy
‘That boy didn’t have any money’ (Brinton and Arnovick 2011: 465).
Unlike (59), in the had-better construction, the noun phrase is in the nomi-
native and a bare infinitive is used instead of a that-clause. The infinitive,
which could appear in this context as early as OE, started to be used in ME
with a greater frequency than the that-clause. The Middle English example
below still has the verb be and an oblique argument:
(60) Forr miccle bettre iss to þe mann … To don all hiss

for much better is to the man to do all his
unnþannkess god þan ifell hise þannkess
against.will’s good than evil his will’s
‘For it is much better for man to do good unwillingly than evil willingly’
[c. 1200, Ormulum, from Łęcki (2010: 71)]
Have, in its past subjunctive form had, first appeared in this construction in
late Middle English; the subject (though missing here) is now nominative and
better is followed by an infinitive rather than a that-clause (Łęcki 2010: 71–72):
(61) & so had better haf ben þen britned to nogt

and so had better have been then cut to nothing
‘and so (he) had better have been killed then’ [c. 1400/1390, Gawain, from
Łęcki (2010: 72)]
The original construction with be survived throughout Middle English, though

it gradually declined as the period wore on.
3.5.2.4 Summary
At this point, I should repeat that, in addition to the other uses enumerated so
far, have still occurs in complex transitive clauses with an undeniably adjecti-
val object complement, as in (62), which comes from colloquial Present-Day
English:
(62) I had it right the first time, didn’t I? [conv, from Biber et al. (2002: 201),
emphasis mine]
In such contexts, it can exhibit bleached semantics, whereby the object is

not actually in the subject’s possession, as in the example from a DIY manual
from the 1980s or the 1990s given in (63). This time the object complement is a
present participle (most likely adjectival), not a past one, and it clearly shows
that have can lose its possession meaning even when it is not in perfect or
perfect-like constructions with a past participle. So semantic bleaching does
not seem to have been reserved exclusively for the grammaticalising perfect
72 chapter 3
construction or for the other offshoots of (quasi-)auxiliary have. It is simply a

natural consequence of its generic meaning.
(63) Make sure the plug casing is in one piece, and replace any that are cracked,
chipped or, worst of all, have bits missing that could expose live parts. [BNC]
Summarising his findings, Łęcki (2010: 207) gives the following approximate
dates for grammaticalising the various functions of have (though there might
be periods of uncertainty for most of these):
– perfect have + past participle – from 800
– causative have + past participle – from 1100
– causative have + infinitive – between 1300 and 1400
– obligative (i.e. deontic) have to – from 1300
– have got to – late 19th c./1900
– had better – from 1500
Based on these somewhat conservative dates and also adding passive have into
the mix, it is right to talk about the polygrammaticalisation of have (cf. discus-
sion in Chapter 2). If the whole point of displacing be from the perfect was to
restrict the functional load of both auxiliaries, this ‘extravagant’ increase (or
maintenance) of have’s own functional load is rather problematic, as is the
creation of yet more alternative passives in competition with be, as well as per-
fects in competition with have itself.
3.5.3 Development of Alternative Passives and Perfects

3.5.3.1 Passives with Get
In addition to the traditional passive with be and the passive-like construction
with have, passives with get are first attested from 1652 and into the eighteenth
century (Strang 1970: 150–151, Rissanen 1999: 215–216, Hollmann 2003: 178 fn. 2,
Baugh and Cable 2013: 335, citing the OED). However, this construction is rare
and unusual before the nineteenth century, when it really came into its own
(see Schwarz 2019). The early example in (64) might still be more akin to a
causative than a bona fide passive:
(64) Insteade of mentioninge his name: Jo: fox the presbyterians gott his name
changhed: & putt in George ffox ye quaker. [[HC] Fox 155, cited in Rissanen
(1999: 216), emphasis mine]
Maintaining a functional explanation, one might wonder why get is being

grammaticalised as a passive auxiliary, when be should have become less am-
biguous as a result of the disappearance of the be-perfect. According to Baugh
and Cable (2013: 335), he got hurt arose because he is hurt is too static in mean-
ing, whereas he became hurt is too formal.41
Indeed, some might say that the new get-passive has evolved in order to
dispel the existing lack of clarity, since potential ambiguity persists to this day
in They were frightened (passive or copular be – see Quirk et al. 1985: 167ff.).
Mitchell (1985: §796) makes similar claims regarding both OE weorðan and
ModE get (cf. Section 3.4.3 for analogous issues with weorðan-passives, which
could be dynamic as well as stative). The use of get instead of be normally gives
a dynamic, as opposed to stative, interpretation, as in (65) (see Biber et al. 2002:
171–172, among others). Get, however, is not completely free of ambiguity ei-
ther. X got frightened can describe an action, as in (66a), or a transition to a
state in a copular clause with an adjectival subject complement, as indicated
by typically adjectival modifiers such as very – (66b). Furthermore, the case
studies below demonstrate that ambiguity between passive and copular be
existed for a long time without any ‘intervention’ from get, which once again
casts doubt on the validity of functional explanations.
(65) a. They were frightened. [dynamic passive or stative copular]

b. They got frightened. [dynamic only]
(66) a. What happened? Some theologians got frightened by this hypothesis.

[COCA, 2015]
b. I got very frightened sometimes. [COCA, 2006]
3.5.3.2 Perfects and Causatives with Get

Interestingly enough, though the grammaticalisation cycle of the perfect seems
to have come full circle in Modern English, another round might be starting
anew. In some varieties of English, the perfect of get, have got, has ceased to be
41 As already noted above, it is not terribly accurate to apply the term ‘auxiliary’ in these
circumstances. Although get has come to participate in passive and perfect constructions,
it shares none of the NICE properties characteristic of the earlier generation of auxiliaries
(see Huddleston and Pullum et al. 2002: 92ff.). By the time get was recruited for the roles
in question, the group of earlier auxiliaries proper had been established for a long time,
distinguished by their unique set of special features which do not seem to be extended
to ‘newcomers’ (cf. Heine 1993: 71–72, Krug 2001, Bybee 2003: 619–621). The same applies
to the later ‘auxiliary’ uses of have (see Łęcki 2010: 85–86, 219 n. 5). ‘Semi-modal’ have to,
for instance, serves as operator in subject-auxiliary inversion and hosts negative not only
exceptionally in some varieties of English. The following would be unacceptable to many
speakers:
(i) Why has it to be me, Jimmy? [M. Munro, Whispering Sands, 1961, from Łęcki (2010: 219)]
74 chapter 3
seen as a perfect, and has come to be synonymous with the stative possession
meaning of present-tense have, as in (67).
(67) The Amphibicar. It’s got little propellers in the back. <cf. It has little
propellers.> [from Biber et al. (2002: 112), emphasis added]
The erstwhile perfect auxiliary is now omissible in speech from the have got
phrase, and we appear to be witnessing the birth of something approximating
a new get-perfect (in addition to the get-passive mentioned above). In (68), get
heads a complex transitive clause with a participial object complement very
much reminiscent of the Old English ‘proto-perfect’ (Miller 2000: 344ff., 2004a:
239, and Brown and Miller 2016: 249–250 talk about resultatives with have got,
get and other verbs).
(68) You got your homework done, Jason? <cf. Have you got…?> [example from
Biber et al. (2002: 112), emphasis added; treated as causative-resultative
by Miller (2004a: 240)]
(69) would normally represent a remnant of a be-perfect, now reanalysed as a

stative copular construction (you’re done); curiously, get has been substituted
here, perhaps on analogy of other copular structures, such as get tired. This also
looks like a nascent perfect.
(69) Maybe Ernest Hemingway there could run me in a beer when you get done
maulin’ him. [1989, from The Dark Half by Stephen King]
Such a development would remain mysterious and outright aberrant on an

ambiguity-cum-functional-load account. If now the system is supposedly
less ambiguous, with have streamlined and more ‘strictly’ associated with the
perfect, why can get be used in such embryonic perfects, in addition to more
established get-passives (themselves encroaching on what should be the terri-
tory of be)? What is more, perfects with other non-standard auxiliaries, includ-
ing done in I done been to Miami, are reported in Miller (2004a: 233). A more
plausible story about this ‘extravagant’ proliferation of functions and ever-
growing versatility would be to say that using general verbs like be, have, do or
get in such constructions is simply the outcome of routine cognitive processes
of reanalysis which do not seem to tie in with functional load at all. Rather, this
proliferation stemming from recurrent spontaneous reanalyses happens with
little regard to functional load; it is more plausibly prompted by the gener-
ic meanings of ‘delexical’ verbs. The evolution of nascent get-perfects might
have gone through a possessive stage (have got) or through a copulative (even
causative) stage, so in essence the possessive/copulative/causative schemas
have been recycled multiple times. The result of such repeated rounds of gram-
maticalisaion of generic and interchangeable lexemes is multiple functions for
a single auxiliary and multiple alternative auxiliaries serving the same func-
tion, which violates the ‘one form, one meaning’ principle (cf. Chapter 2, Heine
1993: 43).
Support for this hypothesis comes from the fact that, in addition to get-
passives and perfects, get is also being recruited as a causative in the construc-
tions ‘get someone to do something’ or ‘get something done’ in yet another
instance of polygrammaticalisation (see Hollmann 2003: Chapter 4, Miller
2004a: 240, as well as sentences (64) and (68) above; cf. also modal gotta in
Section 3.5.2.3, including the interchangeability of be and have there too). In
the following Facebook post, it is clear that someone else came to pick up the
boxes, so the implied agents of got and picked up are different:
(70) Finally got all the empty boxes picked up today for recycling/reuse! Our
shed is so much more spacious now [Facebook post, American English
speaker, 25 July 2017]
3.5.3.3 New Be-Perfects

Even more in tune with the conclusion above is the appearance of brand-new
be-perfects in various dialects of Modern English, including be-perfects with
transitive verbs in Canadian English or in Southern American English – the
latter illustrated in (71) (see Łęcki 2010: 67, 217 n. 32, Werner 2016, Fryd and
Giancarli 2017).
(71) a. He must be finished it by already.42

b. I’ll be broken another tennis rocket [sic].
c. He must be got a new one [car].
[examples attributed to Stephen Nagle, cited in Łęcki (2010: 67), empha-
sis added]
Jucker and Taavitsainen (2013: 26) report similar examples recorded in

Cambridgeshire in the 1970s:
42 Bernard Comrie (p.c.) notes that he also has a be-perfect with transitive finish, as in I’m
finished my dinner (see Rydén and Brorström 1987: 93–94, for earlier examples and an il-
lustration of the rise of this construction). B. Comrie further shares that in Modern Dutch
be has come to be used in the perfect with transitive forget.
76 chapter 3
(72) a. I’m had days digging fires out.

b. Days I’m had doing that.
[from Jucker and Taavitsainen (2013: 26), emphasis added]
The structures in (71) and (72) above are likely to be judged as degraded by
many native speakers, but there is also a more widespread ‘flavour’ of new be-
perfect that occurs in varieties of English spoken on the British Isles, including
Scottish English and also Northern English more generally (see Miller 2000:
345–349, 2003, 2004 a, b, Brown and Miller 2016: 251–252, Weir 2016). As evident
from (73), this pattern takes the form: that be NP Pred (that followed by a form
of be, followed by a noun phrase [determiner phrase (DP) in some theories],
and then followed by some kind of predicate). The predicate can be a past
participle, among other things. The more passive-like construction (73a) is dis-
tributed more widely in the UK than the more active type, (73b)–(73d), which
is restricted to the north and Scotland.
(73) a. That’s him consulted. (≈ He has been consulted, I’ve consulted him)
[Miller (2000: 346, 2004 a, b), quoted in Brown and Miller (2016: 251)
and Weir (2016)]
b. … that’s me seen it. (≈ I’ve seen it now) [Miller (2000: 347, 2004 a, b),
quoted in Brown and Miller (2016: 251) and Weir (2016)]
c. [Is] that you left the school now? (≈ Have you left school?) [Miller (2004
a, b), quoted in Brown and Miller (2016: 251) and Weir (2016)]
d. that’s Ian arrived [Miller (2004a: 243)]
It cannot be denied that multiple be-perfects have arisen independently (or

survived) in various dialects. Łęcki (2010: 67–68) draws the following conclu-
sion, which can be considered quite apt in the given circumstances:
Such exchangeability of HAVE and BE must be due to a very high fre-

quency of occurrence of these verbs, which in turn has resulted in their
semantic content being worn off to the point where they are nothing
more than meaningless phonetic substance.
In other words, it is natural for perfects to arise with such general-purpose

verbs, both of which can have existential as well as possessive uses, and these
grammaticalisation processes do not seem to be too mindful of functional
(un)desirability (cf. Asenova 1987, as well as Section 3.5.2.3). As remarked
in Chapter 2, we simply witness blind/mechanistic (as opposed to purpose-

driven) persistence, layering and divergence. Some modern be-perfects even
restore (or retain) supposedly ‘undesirable’ ambivalence between perfect and
passive/copular: e.g. This country is all descended from the old immigrants, The
town is changed, when in the same variety you also get I’m pass a lot of trouble,
I’m forgot all them things and If a child is taken measles…, side by side with have-
perfects (data from Tagliamonte 2000: 331, 336).
3.5.4 Corpus Data and Statistics

By way of illustrating the rise of the perfect against the backdrop of the much
more widespread and ancient preterite, I supply the relative frequencies cal-
culated by Elsness (1997) for the various periods in the history of English (cf.
Drinka 2017: 251–252). The results are based on past-referring present perfects
and preterites in a comprehensive diachronic databank and are presented in
Fig. 1 and Fig. 2. These figures demonstrate how the perfect gradually carved
out more and more territory from the consistently dominant preterite, though
it has dwindled somewhat over the last two and a half centuries.43 I collected
further frequency data from Early English Books Online, comprising 755 mil-
lion words of British English between the 1470s and the 1690s (https://corpus.
byu.edu/eebo/, accessed on 09 Sept 2018), the Hansard Corpus, comprising 1.6
billion words of speeches from the British Parliament between 1803 and 2005
(https://www.hansard-corpus.org/, accessed on 09 Sept 2018), as well as COHA,
400 million words of American English between 1810 and 2009 (https://cor-
pus.byu.edu/coha/, accessed on 09 Sept 2018). I searched for preterites and
present perfects of two of the commonest lexemes, be and do, since they pos-
sess distinct shapes for the finite preterite and the non-finite past participle,
and my results replicate those obtained earlier by Elsness (see Appendix 1).44
The statistics below give an idea of the growth and consolidation of the
perfect as a category. Statistical data can also shed light on the dynamics of
the relevant auxiliaries. Relying on the Helsinki Corpus, Kilpiö (1997: 106) has
studied the functions of be in the history of English, and his results indicate a
remarkable stability in the distribution of the different uses, with ‘the relative
shares of copular, auxiliary and non-copular main verb uses remaining almost
the same over some 800 years’ (Kilpiö 1997: 111) – see Table 3a.
43 Consult Elsness (1997) for a much more detailed discussion of the distribution of these
past-referring forms, including various kinds of perfects, as well as further refinement of
the data sources and other considerations.
44 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
78 chapter 3
Development to 1600
100%
80%
60%
40%
20%
0%
Old English Early Middle English 1350‒1400 1550‒1600
Present perfect Preterite Other
figure 1 Relative frequencies of major past-referring verb forms from Old English
to 1550–1600
source: Elsness (1997: 270)
Development since 1750

80%
60%
40%
20%
0%
1750‒1800BrE 1750‒1800AmE CONTBrE CONTAmE
Present perfect Preterite Other
figure 2 Relative frequencies of major past-referring verb forms since 1750.

British and American English
source: Elsness (1997: 270)
table 3a Uses of be in the history of English [source : Kilpiö (1997: 106)]
OE ME (early) ME (late) EModE
Copular uses 65% 68% 64% 64%

Auxiliary uses 28% 23% 28% 28%
Main verb, non-copular 7% 9% 8% 8%
The same can be said about the breakdown of auxiliary roles themselves, as
shown in Table 3b.
table 3b Auxiliary uses of be in the history of English [source : Kilpiö (1997: 106)]
OE ME (early) ME (late) EModE
Passive auxiliary 88% 88% 80% 90%

Passive or tense auxiliary 8% 2% 6% 4%
Tense (perfect) auxiliary 4% 10% 10% 6%
Progressive n/a 0% 4% 0%
As becomes evident from this data, the passive auxiliary function of be has
been its most important auxiliary function throughout the history of English.
The figures for the perfect are negligible in all the periods, and there are no
signs of ambiguity increasing or decreasing significantly over time. Kilpiö
(1997: 107–108) notes that the relative stability in the roles of be goes against
‘the expectations that the relative share of auxiliary uses at the expense of the
remaining two uses would rise’. Furthermore, the statistics invite the conclu-
sion that be died out as a (traditional) auxiliary of the perfect just because it
was vanishingly rare in this job from the very outset, not because of any pre-
sumed excessive functional load or any rife ambiguity.45 By contrast, 26% of
all the instances of habban in the Old English section of the Helsinki Corpus
involve the perfect, which makes the perfect periphrasis the most common
function of habban in Old English (Łęcki 2010: 184, citing Kilpiö).
45 Compare the statistical data in Elsness (1997: 264–265), where the proportion of present
perfects with be does not exceed 1.2% throughout the periods he studied. Drinka (2017:
252) reports similar results. Refer back to Chapter 2 and Section 3.4.1 for the role of fre-
quency in grammaticalisation.
80 chapter 3
McFadden and Alexiadou (2010: 415–417) confirm the stable percentages of

perfect be between the years 1150 and 1710 (though measured against the total
of all clauses in the samples). They explain this stability by claiming that, un-
like have, be was never grammaticalised into a proper perfect but remained
a resultative stative: have was developing as a perfect so its use increased (at
the expense of the simple preterite rather than be-periphrases), whereas be
was not being recruited as a perfect in the same way so it remained stable for
a long time; it only declined between 1700 and 1900. McFadden and Alexiadou
(2010) see the later loss of so called perfect be as separate from the rise of per-
fect have in Middle and Early Modern English, but this position does not seem
tenable. If be remained a mere copulative stative paradigmatically unrelated
to have, why would have come to displace it at all (recall the arguments from
Section 3.2 above)? It is therefore hard to agree that the disappearance of the
be-perfect and the takeover by have are two unrelated phenomena. Rather, it
appears that both auxiliaries were undergoing grammaticalisation as perfects
but one was lagging behind the other from the outset. I return to this issue in
Chapter 4.
3.5.5 Alternative Explanations

Although the literature has been and, to a great extent, still seems to be domi-
nated by functional explanations, alternative proposals have also been elabo-
rated, sometimes side by side with the dominant functional ones, as became
evident from the sample of quotations in Section 3.4. Of course, the alternative
explanations do not have to be mutually exclusive with functionalist scenarios
or with each other. The point is that functional accounts are at least in part dis-
credited by the evidence presented above, so they are not likely to have been
the major driving force behind the developments of interest here. According to
the findings from the previous sections, the greater frequency of have has most
certainly played a role in the loss of the be-perfect. Another non-functional
factor might be language contact. Indeed, there have been proposals that lan-
guage contact is one of the reasons for the disappearance of be as a perfect
marker.
McWhorter (2002: 236–238, 258) argues for Scandinavian influence on the
gradual spread of have at the expense of be in the role of perfect auxiliary
during the Middle English period. He seeks support from the fact that Old
Norse, Modern Icelandic and Swedish use a cognate of have with both transi-
tive and intransitive verbs, while be is employed as a resultative in those lan-
guages with only a tiny fraction of intransitives (cf. Rydén and Brorström 1987:
19–20, for a brief diachronic overview of the perfect in the Scandinavian lan-
guages). Chankova (2007a, 2008: 153ff., esp. 164–165, 174, 216, 219) also detects
Scandinavian influence on the use and development of the English perfect.46

The contact between two communities speaking distinct but mutually intel-
ligible varieties (Old English and Old Norse) might conceivably have resulted
in the adoption of the more salient auxiliary in their bilingual interactions
(have rather than marginal be), especially if have was also the more frequent
default option in the language of the invaders. This contact scenario might
neatly explain why there are so few be-perfects in The Second Shepherds’ Play
(Secunda Pastorum), a Middle English text of northern provenance exam-
ined in Chapter 5. Conversely, in discussing the development of the perfect in
Middle English, Fischer (1992: 257) states that it is ‘unlikely that either Latin or
French played any significant role’ (cf. Zimmermann 1968: 17–26). Convinced
that both periphrastic perfects are due to pan-European contact ultimately
emanating from Greek and Latin/Romance (see below), Drinka (2017: 252), on
the other hand, believes that Chaucer’s usage of the perfect was inspired by
French models.
More fundamentally, the very rise of the perfect and progressive periphra-
ses has been associated with the preservation of function. It has been claimed
that the emergence of the analytic/periphrastic perfect and progressive could
have been triggered by the decline of the earlier Germanic aspectual markers –
numerous verbal prefixes which were capable of expressing subtle aspectual
distinctions or distinctions of Aktionsart (see Kovatcheva 1989: 133, Elsness
1997: 245–246, 347, Fischer and van der Wurff 2006: 141, Macleod 2012: 105–107,
190–191).47 Caution has also been advised, sometimes by the same linguists
who propose such causal links. Fischer and van der Wurff (2006: 141), for in-
stance, note that ‘the development of a perfect from a possessive verb like have
is quite a natural grammaticalisation path, occurring independently in many
other languages’, including Romance and Slavic. At the same time, recent re-
search has shown that this construction is actually rare outside of Europe,
which might point to areal spread on the Old Continent (cf. Cysouw 2011, who
sees the have-perfect as a ‘quirk’).
Thus, in addition to being natural, the development of perfects with cop-
ulas and with verbs meaning ‘possess’ may well be an areal affair too (see
Danylenko 2005, Drinka 2017). Like the growth of articles, the formation of be-
and have-perfects might be regarded as a European phenomenon because the
46 It might be problematic that earlier Swedish had a somewhat greater amount of be-
perfects, but perhaps the important thing is that they shrank over time (see Drinka 2017:
248–250). On substratal contact in Hiberno-English as linked to the retention of earlier
English patterns, see Harris (1984).
47 Mitchell (1985: §870) casts doubt on the aspectual connotations of prefixes in OE.
82 chapter 3
distant ancestors of two of Europe’s major language families, Germanic and

Romance, originally did not have those constructions, but they evolved in the
daughter languages over time, normally recruiting a full verb denoting have/
own in transitive constructions and be in intransitive ones (Strang 1970: 100).
The perfect in the third major European language family also sprang up on
‘Slavic soil’ – the Slavic perfect is not traceable back to the old Indo-European
perfect forms either (Duridanov et al. 1991: 299). A pan-European areal spread
through contact does not sound implausible at all, though it is a little worry-
ing in this connection that Bulgarian, for example, has only recently started
to grammaticalise have-perfects, having for centuries relied solely on be (see
Chapter 7). Have-perfects seem to have taken a rather long time to arrive.
3.6 Conclusion
This chapter has charted the development of the perfect in English. In order
to ensure a more comprehensive treatment, the perfect has been considered
in the context of other constructions, such as the passive, and also against the
background of the polygrammaticalisation of generic/delexical verbs such as
be, have and get. Special attention has been paid to the explanations provided
in the literature, including functional load, frequency and contact. Ambiguity
has also figured prominently as an enabling factor for reanalysis.
The opening sections traced the grammaticalisation of have in Old English –
an original complex transitive clause with lexical have and a participial object
complement came to be reinterpreted as a perfect construction with auxiliary
have and a main verb in the past participle. Contexts involving legitimate zero
inflections on the initially adjectival participle have been deemed instrumen-
tal in enabling the reinterpretation due to the lack of unequivocal marking.
Consequently, agreement was later lost across the board from the verbal per-
fect. The second major concern in this chapter has been the competition be-
tween have and be in the perfect domain, and the reasons why be ultimately
became defunct. Contributing factors that have been put forward by earlier
scholars include:
– the disparities in the relative frequencies of the two constructions;
– the functional ‘overload’ of be as copula, as well as perfect, passive and pro-
gressive auxiliary;
– the undesirable ambiguity which the above ‘functional promiscuity’ could
lead to (the latter two obviously go together and are by far the most popular
explanations in the specialist literature);
– the role played by the ancillary confusion-fostering contexts of coordina-
tion, contraction and auxiliary ellipsis.
Problems for this neat and orderly traditional story come from several di-
rections. Firstly, the non-straightforward interaction between morphological
marking and perfect readings is damaging to the causality inherent in the OE
reanalysis account – presence vs. absence of morphological marking and pos-
sessive vs. perfect interpretations do not match neatly as expected. Secondly,
the functionalist take on the competition between the two auxiliaries faces
challenges from the polygrammaticalisation of have itself (arguably, have ought
to have remained restricted to the perfect rather than acquiring a range of ad-
ditional functions, because supposedly its ‘monofunctionality’ is what gave it a
competitive edge in the first place). Further challenges are posed by the devel-
opment of new rival perfect and passive constructions, including be-perfects.
The actual picture is very far from Humboldt’s ‘one-form-one-meaning’ ideal.
In a series of repeated and overlapping waves of grammaticalisation, English
has seen the birth of both perfects and passives built with be, have and get.
It is therefore concluded that the various cases of polygrammaticalisation in
the domains of the perfect and the passive do not obey functional laws at all;
instead they are just natural consequences of the generic meanings (often in-
terchangeable) of verbs like be, have and get, coupled with common cognitive
processes of routine reanalysis.
Corpus evidence indicates a remarkable stability in the relative shares of
the auxiliary functions of some of the relevant verbs under consideration and
suggests that ambiguity tended not to fluctuate wildly, consistently remain-
ing fairly low. Rates of perfect be also remain low throughout the history of
English, while have has seen a dramatic rise and was much more common in
that role from the outset. This lends credibility to the frequency hypothesis.
Another plausible-sounding suggestion might be that be-perfects receded as
a result of contact with Scandinavian. These claims will be substantiated in
subsequent chapters.
For now, it might be instructive to conclude with a parallel from another
area of grammar. It is a well-known fact that syncretism is omnipresent
even in richly inflected languages. Even in a highly synthetic variety such
as Old English, in many cases the ‘apparently crucial’ distinction between
nominative and accusative can itself be neutralised; in Old English, this hap-
pens with all neuter nouns, with all plurals, in the singular of consonant-stems,
as well as with heavy i-stems. Lass’s (1994: 138–139) moral from those observa-
tions is:
that there is a great difference between what a language has and what it
does with it; this should make one suspicious of any kind of facile argu-
ment suggesting that changes are ‘caused’ by the growth of morphologi-
cal ambiguity.
84 chapter 3
The evidence which I adduce below supports such a conclusion. Under

specification appears to be the norm in language, rather than an undesirable
exception or obstacle. The focus on indeterminacy and underspecification in
the rest of this book involves the analysis of OE perfects and passives as com-
plex transitive or copular constructions as opposed to verbal periphrases, as
well as uncertainty of interpretation arising from ambivalent perfects, passives
or copular statives in both English and Bulgarian.
chapter 4
The Development of the Perfect in a Selection of

Old English Texts
This chapter marks the beginning of the more empirical part of the book.
In it I follow the evolution of the perfect and the passive in the Anglo-Saxon
Chronicle, before moving on to Middle and Modern English texts in Chapter 5.
The objectives of the study (Section 4.1) have been informed by the earlier
discussion, so I will be trying to unravel the intricacies of frequency, reanaly-
sis, uncertainty of interpretation, ambiguity and morphological marking (in-
cluding the influence of zero exponence), based on a selection of original Old
English prose. Before I get to the empirical data, I discuss some of the pitfalls
that should not be lost sight of when working with earlier stages of languages
(4.2), and justify my choice of samples and my approach (4.3). Then I proceed
to analyse the actual texts, successively examining Anglo-Saxon Chronicle en-
tries from the eighth century (4.4.2.1), followed by some interim remarks on OE
agreement (4.4.2.2), and then on to the ninth century (4.4.2.3), the late ninth
and early tenth centuries (4.4.2.4), finally supplemented by entries from the
Peterborough Chronicle for the late tenth and early eleventh centuries (4.4.2.5).
In the rich and fascinating material yielded by the texts, I do find support for
potential Scandinavian influence on the development of the perfect, as well
as differential rates of the grammaticalisation of the passive and the various
types of perfect, including curious idiosyncrasies of agreement behaviour.
Morphosyntactic agreement will be a major recurrent theme in the chapter
as it is inextricably linked to the questions of grammaticalisation and reanaly-
sis. Importantly, ambivalent zero exponence is demonstrated to play no sig-
nificant part in the relevant developments, so its role has been overstated in
the existing literature. The chapter’s setting is roughly that of Construction
Grammar, with a special emphasis on gradual changes of auxiliation which
have produced synchronic gradience in terms of the entrenchment of a cluster
of periphrastic schemas in the Old English slice in time.
4.1 Objectives of the Study
The following objectives have been set for a philological study of a selection of
passages, presented in this and subsequent chapters. Naturally, the objectives

86 chapter 4
reflect the major issues which emerged in reviewing the literature and analys-
ing the preliminary data in Chapter 3.
– I will ascertain how frequent the different types of perfect are in a sample
of OE texts and will then compare the rates to frequencies in analogous
ME and ModE text samples, as well as to similar nascent proto-perfects in
Bulgarian.
– An important question to keep in mind will be the following: how often
are the occurrences ambiguous between a perfect syntagm and an SVOC O
clause, taking into account the presence or absence of morphological agree-
ment marking and the other factors identified earlier? Again, comparisons
will be drawn between the selections from the different periods, in an at-
tempt to find out whether ambiguities increase or decrease as time goes by.
– What is the role (if any) of zero exponence as a bridging context enabling
reanalysis and as a contributing factor for the loss of agreement across the
board? Is the loss (or lack) of agreement a cause or a consequence of more
advanced grammaticalisation?
– The corpus research will feature an analysis of be- and have-perfects in
the material, recording any ambiguities between be-perfect and passive or
stative constructions. Middle and Early Modern English will be instrumen-
tal in establishing whether ambiguities proliferate over time. Parallels will
be drawn to Bulgarian and German; the latter retains both types of perfect,
as well as a passive with werden ‘become’ (arguably relieving the functional
load of sein ‘be’). In addition, I will trace occurrences of coordinated past
participles, of auxiliary ellipsis, as well as of ’s which is ambiguous between
is and has, and whether there are any significant diachronic fluctuations.
4.2 Issues in Corpus Work and Compilation
Compiling and working with a diachronic corpus, albeit a modest one, comes
with specific challenges of several kinds. The most important of these are
touched upon by Elsness (1997: 255), and here I summarise Elsness’s main
points. In the first place, the text samples selected for diachronic comparison
can never be one hundred percent parallel. The disparities between texts from
different periods can be due to a number of reasons, such as changing styles
of writing or changing topics of interest (cf. Macleod 2012: 122, 179ff., 226ff., for
similar issues surrounding the complexities of synchronic stylistic variation in
earlier periods). It is virtually impossible, both theoretically and practically, to
draw clear lines of demarcation between changes in language due to purely
linguistic factors and those due to extra-linguistic phenomena. Developments
the Perfect in a Selection of Old English Texts 87
in society and culture can obviously affect the lexicon, but there could be more
subtle influences too. Elsness (1997: 255) notes that the temporal reference ex-
pressed by language might be skewed, since many early texts that survive deal
with events situated in a reasonably distant past, while many contemporary
texts are often interested in the more immediate context, including the imme-
diate and recent past. Such shifts in focus and subject matter could to a certain
extent influence the way verb tenses and other temporal indicators are used
in a passage (cf. Macleod 2012: 72–73). For instance, the Old English samples
to be presented below are dominated by pluperfects, while my Middle English
samples show a number of present as well as past perfects. This must surely
be an epiphenomenon due to the nature of the selected material – the plu-
perfect is to be expected in annals dedicated to historical events such as the
Anglo-Saxon Chronicles.
Placing these concerns about the nature of the evidence on a more solid em-
pirical footing, Szmrecsanyi (2016) provides empirical confirmation that there
are limitations associated with inferring grammatical change from variable
text frequencies in historical corpus data. Based on a case study, Szmrecsanyi
(2016) verifies that fluctuations in the frequencies of grammatical variants in
real time may not only reflect genuine grammatical changes in progress, but
can also be conditioned by other, extraneous ‘environmental’ changes, such as
a change in content or textual subject matter, as conjectured above.
Thinking along similar lines, Mustanoja (1960: 505) too highlights certain
pitfalls associated with our historical sources:
Figures based on two late ME literary texts […] suggest that even at the
end of the period the general frequency of the perfect is many times
smaller than that of the preterite. The notion obtained from these figures
is perhaps not quite accurate, for there is reason to assume that the fre-
quency of the perfect is somewhat higher in everyday speech, where the
speaker naturally tends to look at events from the angle of the moment
of speaking.
Certainly, even the closest literary approximation to spoken language is still

not a transcript of real spoken language, which would doubtless furnish the
most appropriate material on which to base conclusions. Sadly however, writ-
ten approximations with varying degrees of faithfulness are the next best
thing that we have for earlier periods (see also Szmrecsanyi 2016: Section 3). Of
course, our evidence is only indirect and we cannot verify, say, by experimental
work or acceptability judgement questionnaires, what must have been going
on in speakers’ minds (cf. Menn et al. 2013).
88 chapter 4
Another major problem identified by Elsness (1997: 255) is the problem

of continuity. In an ideal world, a diachronic comparison should juxtapose
samples which come from different periods but from the same dialect.
However, we are not always in a position to achieve such laudable but unrealis-
tic goals. This is especially pertinent to research in the history of English, as we
can trace the direct dialectal ancestry of Standard Present-Day English no fur-
ther back than the early 15th century. Out of the multitude of Middle English
dialects, the late 14th century saw the emergence of a standard that over the
course of the following century came to be established and recognised, and
later gradually morphed and was codified into the standard English that we
know today (see Baugh and Cable 2013: 50, 187). This emerging standard was
for the most part based on the language of the East Midlands, in particular
the speech of the capital London. However, our most extensive records of Old
English are in the dialect of the West Saxon kingdom located in the southwest
of England. This variety of Old English had come to approximate the status of
a literary standard, but due to the Norman Conquest of 1066, there is no conti-
nuity between this ‘standard’ Anglo-Saxon and the East Midland standard that
arose in late Middle English times. Bybee (2003: 623 n. 8), while acknowledging
this issue, is still cautiously optimistic:
There are several varieties subsumed under Old English and even more
under the designation Middle English, so it cannot be assumed that there
is necessarily a direct developmental relation between the languages rep-
resented in the texts […]. Still it is clear that in some general sense a type
of diachronic relation exists.
Despite all the aforementioned potential pitfalls and caveats to be wary of,
Elsness (1997: 255) also concludes that broad diachronic comparisons are
nevertheless worthwhile endeavours capable of yielding meaningful, valid re-
sults. Issues related to corpus design and compilation are further discussed in
Jacobs and Jucker (1995: 6ff.), in Chankova (2008: 16–18), and in many other
sources dealing with the diachronic description of English. Conscious of the
limitations of our data, we can still draw some non-trivial conclusions and gain
insights from a corpus study. The choice of samples will be essential in over-
coming at least some of the limitations, at least in part.
4.3 Choice of Text Samples and Approach
As has emerged already in the previous section, it has been noted by schol-
ars (e.g. Zimmermann 1968: 108–158, quoted in Fischer 1992: 258 and Macleod
2012: 230), that the Middle English present and past perfect appears more fre-
quently in colloquial style, while it is rare in purely narrative contexts; it is
also frequent in instructional genres. In the Early Middle English works that
Zimmermann examined (Ancrene Wisse, the ‘Katherine Group’, Layamon/
Laȝamon’s Brut and The Owl and the Nightingale), the perfect is found almost
exclusively in direct speech (cf. Macleod 2012: 229–230, for Old English). This
observation highlights the link between the perfect and the colloquial register.
Therefore, the investigator should seek texts which approximate the spoken
usage of the time.
Analysing be- and have-perfects, Rydén and Brorström (1987) and Rydén
(1991: 343) focused on two types of text, private letters and comedies, since
they believed that those genres would reflect the ‘frontier’ uses. However, some
of the texts Rydén and Brorström examined turned out to be conservative rath-
er than innovative and progressive (Rydén 1991: 343). Overall, comedies proved
to be ‘more genuine reflexes of non-conservative language’ (Rydén 1991: 344;
see Rydén and Brorström 1987: 197, 199, 200–201). This finding is not surprising,
as private letters, depending on stylistic consciousness and levels of literacy,
might be formal as well as informal, though of course in principle they are ex-
pected to be less formal than official letters. As Rydén (1991: 344) recapitulates:
On the whole, our texts can be said to represent informal written lan-
guage, with some speech-based elements, rather than spontaneous spo-
ken discourse, a fact which should be kept in mind when it comes to
assessing the texts (or rather the grammars as extracted from the texts)
as ‘spearheads’ of the paradigmatic development.
Following the experience and knowhow obtained from earlier research, here
I will also be examining genres which have proved to be reasonably reliable
reflections of genuine contemporary usage. My mediaeval and Early Modern
samples will be based on comedies and comic tales, as well as on private letters.
While there are a number of electronic corpora available, synchronic as
well as diachronic, I have instead opted for a more traditional philological ap-
proach, investigating a selection of texts. There are several reasons why I have
chosen not to avail myself of an authoritative and truly invaluable database
such as the Helsinki Corpus (or a later version). First, I am looking for ambigu-
ity, which has been intentionally left out or unwittingly neglected in earlier
empirical research (e.g. Rydén and Brorström 1987: 24, 32). Moreover, ambi-
guity is difficult to code or tag (or sometimes even to detect), so reading the
actual texts in their entirety might work better. Close reading will also make it
easier to immerse oneself in the broader context, which often turns out to be
relevant to the linguistic analysis. As Macleod (2012: 241) admonishes,
90 chapter 4
[i]f research upon a language is based on data that do not reflect actual
usage, but rather a statistical composite reflecting the practices of no
individual speaker, the conclusions reached by such means necessarily
have a less stable foundation than might be wished.
Second, since many previous studies have been based on the same material,
especially the Helsinki Corpus, this might skew the data and the findings;
therefore, I have decided to put together a slightly different, tailor-made sam-
ple which will complement the earlier larger-scale studies. Thus breadth of
coverage and size have been sacrificed for the gains of depth and close reading.
Below I will first examine a number of entries from the Anglo-Saxon and
the Peterborough Chronicles, tracing the use of the perfect in the light of mor-
phosyntactic agreement, ambiguity and semantic interpretation. I will then
move on to the perfect and passive auxiliaries as used in Chaucer, 15th-century
drama, Shakespeare and Restoration Comedy, as well as the correspondence of
the Pastons and of Samuel Pepys (Chapter 5).
4.4 The Perfect in the Anglo-Saxon Chronicles
4.4.1 Introduction and Background

The collection of Anglo-Saxon Chronicles might have originated as Easter
tables which were used to determine the date of Easter each year. Gradually,
entries describing the year’s major events began to be added. Seven versions
of the Anglo-Saxon Chronicle exist, originating from different locations and
sometimes recording different events, though up until 891 they seem to share a
common source, having perhaps been first commissioned at the court of King
Alfred. The last of the chronicles (from Peterborough) was kept up until 1154
and shows the transition into Early Middle English. Since the chronicles were
maintained for a long period of time, this allows linguists to observe how the
language gradually changes. Importantly, these annals are not translations
from Latin and do not aim for special stylistic or literary effects, so they con-
stitute a straightforward native manifestation of plain and unadorned early
English prose (see Brinton and Arnovick 2011: 159, Macleod 2012: 74–75).
The earliest surviving manuscript of the Anglo-Saxon Chronicle (Corpus
Christi College, Cambridge, MS. 173) is known by the name of its donor as the
Parker Chronicle, normally cited as Manuscript A. It shows some connections
to Winchester. The first part was probably written some time after the year
891, which is when the first hand finishes, and it was then continued up to
1093 (see Plummer and Earle 1892: x, Whitelock et al. 1961: xi–xxix). Although
this A version is the oldest, it is not the author’s original and other versions
sometimes preserve better readings. One of these is the so-called Cotton man-
uscript, Tiberius A.vi (London, British Library, Cotton Tiberius A.vi), referred
to as Manuscript B. Manuscript B was copied by a single scribe from a lost
source close to Manuscript A in the late tenth century (between 977 and 1000)
and goes on until the year 977, after which it was not continued. Although this
is disputed, the manuscript is traditionally associated with Abingdon (http://
asc.jebbo.co.uk/intro.html, accessed on 19 Sept 2016).1 The A version will be
consulted for the years between 700 and 950, as it appears in digitised format
on the following website: http://asc.jebbo.co.uk/a/a-L.html (accessed on 19
Sept 2016). In principle, I will be exclusively analysing the A text, only checking
other manuscripts occasionally if and when necessary, for instance if there are
pertinent divergent readings noted in the editions used.
The other manuscript of interest here is most famously known as the
Peterborough Chronicle. It is also designated as Manuscript E (Oxford, Bodleian
Library, MS. Laud 636), and was most probably initially copied in 1121 or 1122
at Peterborough and maintained in various hands until 1154 (see Plummer and
Earle 1892: xii, Irvine 2004). This is a representative of the so-called ‘Northern
Recension’ of the Chronicle, and the transcript I use is the one which appears
at http://asc.jebbo.co.uk/e/e-L.html (accessed on 19 Sept 2016). For both
Manuscripts A and E, the online publications were verified with a more au-
thoritative traditional source, Plummer and Earle’s (1892) printed parallel
edition (which was fed into the Helsinki Corpus too). In order to ensure ac-
curate interpretation, I relied on the translations from Whitelock et al. (1961).
4.4.2 Analysis and Discussion

4.4.2.1 Manuscript A: Eighth Century
The first sample from the Parker Chronicle (Manuscript A) which I examine
includes the entries between the years 703 and 799. The year of the entry (ac-
cording to the manuscript, which is not always historically accurate) is indi-
cated in square brackets after each example. The total number of words is
approximately 1,615, based on http://asc.jebbo.co.uk/a/a-L.html (accessed on
22 Sept 2016). After each example, a Modern English translation is supplied
from Whitelock et al. (1961), with only a few minor modifications in one or two
places. The glosses are mine.
1 See however Plummer and Earle (1892: x–xi), who point out a Canterbury connection, as they
do for MS A.
92 chapter 4
Table 1 Perfect, passive and progressive auxiliaries in Manuscript A of the Anglo-Saxon

Chronicle (years 703–799)
With No Zero TOTAL % (of all

agreement agreement morphology (number) auxiliary
expected occurrences
recorded)
Have-perfects 1 1 0 2 6.7%
with transitive
verbs
Have-perfects with 0 0 0 0 0%
intransitive verbs
Be-perfects with 0 0 1 1 3.3%
intransitive verbs
Progressives 0 0 3? 3 10%
Passives with beon 3 1 12 16 53.3%
Passives with 0 0 2 2 6.7%
weorðan
Ambiguous 0 0 5 5 16.7%
passives/statives
with beon
Ambiguous 0 0 0 0 0%
passives/statives
with weorðan
Ambiguous 0 0 1 1 3.3%
be-perfects
Hardly any perfects were found in this section of the Chronicle. As noted in the
general discussion of corpus planning, such a genre with exact dates from the
reasonably distant past presupposes fewer perfects, since reference is made to
past time without much relevance to the immediate context in the present (i.e.
the scribe or the author’s here and now; cf. Macleod 2012: 72–73). Hence, the
few perfects I register are all pluperfects. In addition, most of this manuscript’s
entries for the eighth century are rather brief and matter-of-fact,2 and most of
2 An especially short, vivid and poignant example comes from the entry for 761:
(i) Her wæs se micla winter. [761]
‘Here [i.e. in this year] was the big winter.’ (translation mine)
the perfects I did harvest (two out of three) come from an exceptionally exten-
sive narrative for the year 755. Of course, the scarcity of the perfects might in
part be due to the early date of composition of the originals from which the
Parker scribe copied. There is ample evidence that this text is too early for the
perfect to have come into its own (this would happen centuries later):
(1) 7 wæs todęled3 in foreweardum Danieles dagum in tua

and was divided in early Daniel’s days in two
biscepscira Westseaxna lond, 7 ær hit wæs an [709]
bishoprics of.West.Saxons land and previously it was one
‘Early in Daniel’s time the land of the West Saxons had been divided into
two dioceses, whereas it had previously been one.’
The pluperfect is not used where it would be expected today – instead, the
Anglo-Saxon in (1) has the simple preterite (for more details, see Mitchell
1985: §611ff., §§634–644, Macleod 2012, Los 2015: 73–74); here and in many
other places elsewhere, only context or lexemes like ær ‘ere, earlier, previ-
ously’ are relied on to signal past in the past, as in (2) (on ær, see Mitchell
1985: §§1112–1115).
(2) Her Æþelburg cuen towearp Tantun þe Ine ær

here Æthelburh queen demolished Taunton which Ine previously
timbrede [722]
built
‘In this year Queen Æthelburh demolished Taunton, which Ine had built’
What is more, both OE (proto-)perfects and OE (proto-)passives exhibit a

relatively free order of constituents in these prose texts uninfluenced by Latin
originals, as even a cursory look at Appendix 2.14 will confirm. This is poten-
tially indicative of incomplete grammaticalisation.5
3 Following the actual scribal practice, the electronic edition used as the source of the ex-
amples distinguishes two forms of the OE character æsc: ę and æ (cf. Irvine 2004: civ).
4 The Appendices of this book have been made available online. They can be accessed through
a QR code which can be found in the References section of this book.
5 On the relative order of elements in perfect and passive constructions, see Mitchell
(1985: §703ff., §§757–758), and the discussion in Chapter 3. Word order might also affect
agreement options, as is evident from Mitchell (1985: §706, §710). Kilpiö (2007: 329–330, 2013:
108) demonstrates some intriguing correlations between word order and agreement in the
perfect – placing the object before the participle decisively favours agreement, while plac-
ing the object after the participle disfavours agreement (cf. the Dictionary of Old English at
http://tapor.library.utoronto.ca/doe, accessed on 27 Feb 2017).
94 chapter 4
Another unresolved issue is the question of how many have-constructions

are genuine perfects and how many are ambiguously stative (see Chapter 3
and Johannsen 2016). Readers should keep in mind that most of the examples
in my have-perfect category can be given a stative interpretation too, including
(3) and (4) (cf. Carey 1995, Macleod 2012: 108–109, 118), but this has not been
described as ‘detrimental’ in the grammaticalisation of the have-perfect in the
way it has been for be. Sentence (4) can even receive a causative interpretation,
so ambiguity was widespread with have as well. Have-perfects could often be
ambiguous statives, but be-perfects could additionally be stative or passive.
(3) 7 hie alle on þone Cyning wærun feohtende oþ þæt hie

and they all on that king were fighting until they
hine of-slæg-en-ne hæfd-on [755]
him.3m.acc.sg pref-slay-pptcp-m.acc.sg had-pl
‘Then they all fought against the king until they had slain him’
(4) 7 þa gat-u him to be-loc-en

and that.acc.pl gate(nt)-acc.pl them to pref-lock-pptcp
hæfd-on [755]
had-pl
‘and [they] had locked the gates against them’
Of the two have-perfects with transitive verbs, both from the account for 755,
example (3) shows the agreement expected of a participial object comple-
ment in the accusative masculine singular (-ne in the strong declension),
while no such agreement is present in (4); for it, neuter plural should be in-
dicated with -u on the past participle belocen, but the -u is not there, so at
least as far as the have-perfects here are concerned, agreement morphology
is dropped even without cases of ambiguous zero marking. Neither does the
context in (3) strictly suggest anteriority in the past (cf. (12) below, again with
oþ ‘until’).
By contrast, the only unambiguous be-perfect with a mutative intransitive
verb, (5), is of the type for which it is impossible to tell whether it does or does
not agree, since the singular morphological marker would be zero anyway. In
terms of idiom, was… agan could be a common set phrase or formula, as could
many other examples in this chapter, e.g. wæl + habban/beon/weorðan + ofslæ-
gen, to ærcebiscepe + gehalgod, gefeoht + gefohten, geweorc + geworht, etc. (see
Danchev 1969: §59ff., Smith 2001, and earlier chapters for the resilience and
survival of frequent archaic set phrases with a high degree of entrenchment).
(5) Her Danihel forþferde, þa was .xliii. wintr-a

here Daniel went.forth then was 43 winter(m)-gen.pl
a-gan siþþan he onfeng biscepdome [745]
pref-gone since he succeeded.to bishopric
‘In this year Daniel died. Forty-three years had then passed since he
succeeded to the bishopric.’
Agreement here might have defaulted to neuter singular (with a zero expo-
nent on agan) because of the numeral + noun subject xliii wintra. Mitchell
(1985: §563) notes that when a finite verb precedes a numeral acting as subject,
the verb may be singular or plural. Wintra is a masculine noun, most probably
appearing in the partitive genitive plural because it follows a numeral, so it is
not nominative and therefore cannot trigger agreement. Since the finite auxil-
iary was is singular, the participle has no option of being in the plural (though
see the next paragraph).
A comparison to the (proto-)passive would be especially instructive. There
are many more prospective passives than perfects in the sample, which might
conceivably suggest that the passive had reached a more advanced stage of
grammaticalisation by this time; on the other hand, the greater number might
be due to the subject matter of this type of narrative. Three of the passives
show agreement, although it is rather defective in one of them, given in (6) (cf.
Chapter 3, Section 3.1.2):
(6) 7 Cuþbryht wæs to ærcebiscep-e

and Cuthbert(m)[nom.sg] was to archbishop(m)-dat.sg
gehalgod(e), 7 Dun biscep to Hrofesceastre. [741]
consecrated(?) and Dunn bishop to Rochester
‘And Cuthbert was consecrated archbishop and Dunn bishop of
Rochester.’
In this example, apparent agreement inflection appears in the manuscript

where it does not properly belong. Plummer and Earle (1892: 44 fn. 5) note
that the manuscript has gehalgode but they emend it to standard zero-marked
nominative masculine singular gehalgod ‘hallowed, consecrated’; the -e could
be a scribal slip, or it might have been influenced by the (legitimate) dative
singular -e on ærcebiscep-e; alternatively, gehalgode could be plural, as there
are two people involved, but Cuthbert and Dunn are certainly not in the same
clause and the auxiliary wæs is definitely singular. The remaining two exam-
ples show no such deviations, -e being straightforwardly nominative plural:
96 chapter 4
(7) eowr-e gefer-an þe mid þam cyninge

your-m.pl comrade(m)-pl who with that king
of-slægen-e wær-un [755]
pref-slain-m.nom.pl were-pl
‘your comrades who were slain along with the king’6
(8) 7 wunderlec-a nędr-an wær-on

and marvellous-f.nom.pl adder(f)-nom.pl were-pl
ge-sewen-e on Suþseaxna londe. [773]
pref-seen-nom.pl on of.South.Saxons land
‘And marvellous adders were seen in Sussex.’7
The majority of passives, however, a total of 12, are morphologically ambiguous

as to presence or absence of agreement due to the zero marking for those slots
of the strong adjectival paradigm. In other words, the marker in such instances
is null anyway, as illustrated below for the nominative masculine singular – the
most frequent type of subject and topic in this portion of the Chronicle:
(9) Her was of-slægen Osric

here was pref-slain[m.nom.sg] Osric(m)[nom.sg]
Norþanhymbra cyning [731]
of.Northumbrians king
‘In this year Osric, king of the Northumbrians, was slain’
Interestingly enough, at this early stage, non-zero agreement markers are not
lost in the perfect and passive on any significant scale (a total of two instanc-
es), even though there are quite a few cases with expected zero morphemes,
especially in the passive (see Table 1 above and Appendix 2.1).8
6 The invariable relativiser þe presumably takes its gender and number features from its ante-
cedent geferan. Note plural wærun.
7 The attributive adjective wunderleca takes the traditional strong nominative feminine
plural -a, whereas the -e on the predicative participle might be influenced by the masculine
nominative plural. It is widely attested as an alternative in the feminine as well, however (see
Chapter 3, Table 2). Might this mean that the two are treated differently? These attributive
and predicative elements are certainly given different markers, suggesting that they were
probably not perceived as equivalent (more on this to follow). On whether gesewene could
be an adjective, ‘visible’, see Mitchell (1985: §766).
Very often it was difficult to tell whether something was a genuine passive
structure with dynamic meaning or a stative copular clause with linking be
(Rydén and Brorström 1987: 27 admit similar difficulties):
(10) 7 þeah he wæs oft ge-wund-ad [755]

and though he.3m.nom.sg was oft pref-wound-ed[m.nom.sg]
‘though he was often wounded’
The example in (10) was regarded as dynamic because of the adjunct of fre-
quency oft, though it could be stative nonetheless (cf. e.g. ‘He was often sad/
in a bad mood/incapable/out of action’, see also Mitchell 1985: §§766–767). By
contrast, the following were treated as ambiguously stative:
(11) þæs cyn is beforan

that.one’s kin(nt)[nom.sg] is before
a-writ-en [716]
pref-writt-en[nt.nom.sg]
‘whose ancestry is given above’
(12) Ac hie simle feohtende wæran oþ hie alle lægon

but they continually fighting were until they all lay
butan anum Bryttiscum gisle 7 se swiþe
but one British hostage and that.m.nom.sg very
ge-wund-ad wæs. [755]
pref-wound-ed[m.nom.sg] was
‘But they continued to fight until they all lay dead except for one British
hostage, and he was severely wounded’
In (12), adjectival status for the participle might be indicated by the modifying
adverb swiþe ‘very, exceedingly’ (see Mitchell 1985: §741). In addition, (10) and
(12) show very clearly the flexibility of word order pointed out already.
Lack of overt morphology where it is expected did not always ensure a pas-
sive reading, as in the following potentially copular clause, in which the parti-
ciple (subject complement?) is missing the traditional -u morpheme to mark
nominative feminine singular agreement with the subject:
98 chapter 4
(13) 7 hio wæs for-gif-en Norþanhymbra

and she.3f.nom.sg was pref-giv-en of.Northumbrians
cyninge Aldferþe [718]
to.king to.Aldfrith
‘She had been married to Aldfrith, king of the Northumbrians’9
After some deliberation, (13) was put down as a passive, though for many
of these examples it was not at all easy to decide with any amount of con-
fidence. This particular decision was prompted by the dynamic semantics of
give, not by the missing agreement morphology.10 It will emerge below that,
in all the periods studied, the grey area between genuine passive and a stative
copular clause remains uncurbed (cf. Mitchell 1985: §735). By contrast, only
one example was judged to be ambiguously perfect or passive/stative (see
Appendix 2.1,11 section on ambiguity between be-perfect and passive/stative) –
it involves the verb for ‘gather’; ‘the synod was gathered/collected’ can be inter-
preted as ‘someone had gathered/collected the synod’ (hence passive) or ‘the
synod had gathered/collected’ (perfect; more on this to follow).
The other OE proto-passive auxiliary, weorðan ‘become’, does not seem to
suffer from such a ‘functional disadvantage’ to the same extent; weorðan is nev-
ertheless used much less frequently than beon ‘be’ here (only twice), despite its
apparent ‘functional superiority’.12
(14) Her Osred Norþanhymbra cyning wearþ

here Osred of.Northumbrians king(m)[nom.sg] became[sg]
of-slægen [716]
pref-slain[m.nom.sg]
‘In this year, Osred, king of the Northumbrians, was slain.’
9 Feminine and neuter -u endings could be dropped because participial stems might be
treated as heavy/long (-u is generally lost after heavy/long syllables). However, the past
participle suffixes (strong -en and weak -ed/-od/-ad/-d/-t – see Mitchell 1985: §983) did
not contain long vowels and did not create clusters of consonants (in my material here),
which is what makes a syllable heavy. Note Mitchell’s examples with feminine singular -u
in the next section.
10 Cf. observations above and the discussion of this sentence in Mitchell (1985: §839), where
it is also treated as a bona fide passive.
12 Weorðan however is not completely immune from ambiguity between a passive and a
copular reading – see Mitchell (1985: §735, §798) and the discussion in Chapter 3, Section
3.4.3.
Quite apart from perfects and passives, the texts turn up a causative con-
struction with a now-lost verb which meant ‘to command’ – OE (ge-)hatan,
preterite singular het.13 This was later supplanted by causative have (among
other options), as indicated by the Modern English translation in (15). First
introduced in Chapter 3, the topic of the increased functional load of have will
be taken up again in subsequent sections and chapters.
(15) Her Offa Miercna cyning het Ęþelbryhte rex

here Offa of.Mercians king commanded Ethelbert king
þæt heafod ofaslean. [792]
that head slay
‘In this year Offa, king of the Mercians, had Ethelbert beheaded’
Finally, there were three (proto-)progressives with present participles termi-

nating in -ende, though the link (if any) to modern progressives is not univer-
sally agreed on (see Mitchell 1985: §§682–701). The e on -ende is probably not
plural agreement, since -ende is the usual shape of the present participle suffix.
(16) 7 hie alle on þone Cyning wær-un feohtende

and they.3nom.pl all on that king were-pl fighting
oþ þæt hie hine ofslægenne hæfdon [755]
until they him slain had
‘Then they all fought against the king until they had slain him’
4.4.2.2 Some Notes on OE Agreement

Before proceeding further, some interim comments are in order regard-
ing agreement with adjectives and participles in Old English, recapitulat-
ing some of the major principles expounded earlier in this book. Adjectives
and participles appearing with beon/wesan and weorðan agree with the sub-
ject in number, case, and gender, as noted in the previous chapter (consult
Table 2, Chapter 3). However, since an adjective or a participle is declined
strong in such circumstances (i.e. when used predicatively), it is impossible to
tell whether it is inflected or not if the subject is masculine or neuter singular
due to the zero endings in those slots of the strong paradigm (e.g. OE Crist
is gōd ‘Christ is good’, Mitchell 1985: §33, §759). A distinctive form in -u can
occur in the nominative feminine singular of adjectives with short/light stems,
as in (17):
13 For more on OE hatan, consult Mitchell (1985: §§679–680).

100 chapter 4
(17) Swæ clæne hio wæs oðfeallen-u

so entirely she.3f.nom.sg was[sg] fallen.away-f.nom.sg
‘It had deteriorated so completely …’ [King Alfred’s West-Saxon Version of
Gregory’s Pastoral Care, 3.13, from Mitchell (1985: §33), glossing and trans-
lation mine]
Mitchell (1985: §33, §738) however adds that the form without -u is more com-
mon, as in (18):
(18) … hu sio lar Lædengeðiodes ær

how that.f.nom.sg learning(f)[nom.sg] of.Latin before
ðissum afeallen wæs
this fallen.away was
‘how the learning of Latin had deteriorated before that (time)’ [King
Alfred’s West-Saxon Version of Gregory’s Pastoral Care, 7.15, from Mitchell
(1985: §33), glossing and translation mine]
In fact, there are no feminine singular forms with -u in the entries of the
Anglo-Saxon Chronicle between 734 and 1001 (Mitchell 1985: §33 fn. 7; cf. previ-
ous section too).
In the plural, the adjective or participle can either remain uninflected or
receive a generalised -e for all genders. Mitchell (1985: §34, §760) observes that
feminine plurals in -a and neuter plurals in -u are rare. Uninflected plurals do
occur but in general appear to be the exception rather than the rule. According
to some sources, only one indisputably uninflected plural is attested in the
Anglo-Saxon Chronicle between 734 and 1001 (Mitchell 1985: §34, with refer-
ences, though see the next sections, where I report more and return to this
example):
(19) 7 þa oþr-e wær-on hungre acwolen

and that.nom.pl other-nom.pl were-pl of.hunger died
‘And the others had died of hunger’ [Chronicle A, 894, quoted in Mitchell
(1985: §34), glossing and translation mine]
To sum up, adjectives or participles used predicatively in the nominative are

declined strong – they usually receive no inflectional ending in the singular
and -e in the plural. Accusative predicative adjectives and participles are also
declined strong, if at all. Only 30 out of 230 have-perfects have inflected parti-
ciples in the Chronicle, and only accusative masculine singular participial (ob-
ject complement?) forms are said to be inflected in MS A of the Chronicle (see
Mitchell 1985: §§128–129, §710, §737, §§759–761, with references; also see my

data presented in subsequent sections).14
A hypothesis might be built on the evidence marshalled above. It could
conceivably be the case that speakers had by that stage simply reshuffled the
morphological system (at least in the nominative), with zero marking in the sin-
gular, and a generalised -e in the plural (cf. the situation in Middle English). All
of this probably happened irrespective of functional load or morphosyntactic
ambiguity. At least in the passive, there is an overwhelming amount of expect-
ed zero morphology in the singular, but that does not make the overt morphol-
ogy disappear from the plural. It would be odd to say that the (non-agreeing?)
singulars are genuine passives, but the agreeing plurals are not (cf. the argu-
mentation and discussion in Mitchell 1985: §762ff., and especially §765).
Mitchell’s (1985: §760) conclusion regarding agreement in the passive relies
on analogy of legitimate zero forms and runs as follows (note that the square
brackets represent my additions to or omissions from the original):
The absence of -u in the feminine singular forms is apparent early and

can be explained by the working of analogy, for the majority of parti-
ciples properly had the nominative singular ending – [i.e. ‘zero’]; […] The
sporadic early examples of pl. – [‘zero’] can be similarly explained by ref-
erence to the existence of both feminine and neuter nouns which had –
[‘zero’] in the nominative plural.
This suggestion, appealing to traditional analogy, will be further discussed

below. It might work for the spread of zero marking in the singular (the major-
ity form) or the generalised -e in the plural. However, there are some problems
arising from it, at least regarding the spread of zero plural forms: those were
never the majority in the samples here and my data prove resistant to such an-
alogical levelling. I shall now seek confirmation for the hypotheses entertained
here in the data collected for the ninth century, where some other interesting
patterns show up as well.
14 Below is a late example from MS E with plural agreement which is not included in my
sample (see Mitchell 1985: §718):
(i) Đas þing we habbað be him ge-writ-en-e, ægðer ge
these thing(nt)[acc.pl] we have about him pref-writt-en-pl both
god-e ge yfel-e
good-pl and evil-pl
‘These things we have written about him, both the good and the bad.’ [Chronicle, MS
E, 1086, quoted in Mitchell (1985: 288), gloss and translation mine]
102 chapter 4
4.4.2.3 Manuscript A: Ninth Century

Continuing with Manuscript A into the 9th-century entries, I examine the
years from 800 to 891 (a total of about 3,850 words), which is when the first
hand finishes, so to a certain extent this portion of the text might be said to
represent a more or less coherent idiolect (despite additions, uncertain au-
thorship and other problems listed above; cf. Macleod 2012: 75). Indeed, the
results presented in Table 2 are almost identical to those from the 8th century,
so the expanse of the Chronicle up to this point is clearly consistent.
Table 2 Perfect, passive and progressive auxiliaries in Manuscript A of the Anglo-Saxon

Chronicle (years 800–891)
With No Zero Total % (of all

recorded)
Have-perfects with 1 1 0 2 4.6%

transitive verbs
Have-perfects with 0 0 0 0 0%
intransitive verbs
Be-perfects with 0 0 0 0 0%
intransitive verbs
Progressives 0 0 6? 6 13.6%
weorðan
Ambiguous 2 0 4 6 13.6%
passives/statives
with beon
passives/statives
with weorðan
be-perfects
The perfect is still extremely rare. Even in clauses like (20), where we would
use it today alongside the passive, Old English only has a preterite passive:
(20) þy hie from his mægum ær mid unryhte

for.that they.3nom.pl from his kinsmen previously with unright
a-nid-d-e wær-un [823]
pref-force-pptcp-m.nom.pl were-pl
‘because they had been wrongfully forced away from his kinsmen’
Of the two have-perfects from this sample, again one shows agreement with an
accusative masculine singular object, (21), while the other one does not, (22)
(appearing further down).
(21) 7 hie hæfd-un hiera cyning aworpen-ne

and they had-pl their king(m)[acc.sg] deposed-m.acc.sg
Osbryht, 7 ungecynd-ne cyning
Osbert and with.no.hereditary.right-m.acc.sg king(m)[acc.sg]
underfeng-on Ęllan [867]
took-pl Ælla
‘and they had deposed their king Osbert and taken [took?] a king with no
hereditary right, Ælla’
It should be noted that the (proto-)perfect with have and an agreeing parti-
cipial object complement closely parallels in morphology and syntax the sec-
ond clause, which exhibits an agreeing accusative modifier (ungecyndne) of
the direct object (cyning), and has the indisputably lexical verb underfon ‘to
receive, get, take’, which appears in the preterite plural – underfengon (the past
participle of underfon is underfangen). It is thus safe to conclude that have is
not fully grammaticalised here – due to the parallel to the second clause and
the adjectival agreement. At the same time, it is clear that the people referred
to as ‘they’ no longer literally ‘have’ Osbert as a king; therefore, the interpreta-
tion should be perfect-like, despite the agreement and the parallel with the
next clause (readers will remember the relevant discussion from Chapter 3;
also see Macleod 2012: 120). The meaning of possession has been bleached in
favour of abstract anteriority.15
15 Yet Mitchell (1985: §728) counters, not entirely convincingly, that the original sense of this
construction was ‘They had (=held) their king Osbryht (having been) rejected’.
104 chapter 4
The other have-perfect from the sample ‘should’ also supply its participle
(afliemed) with the accusative masculine singular agreement morpheme -ne,
but it is now missing:
(22) 7 hine hæfd-e ær Offa Miercna cyning

and him.3m.acc.sg had-sg previously Offa of.Mercians king
7 Beorhtric Wesseaxna cyning afliemed .iii. gear of
and Brihtric of.West.Saxons king banished 3 years of
Angelcynnes lande on Fronclond, ær he cyning wære [836]
of.English.race land on France before he king were
‘Earlier, before he became king, Offa, king of the Mercians, and Brihtric,
king of the West Saxons, had driven him from England to France for three
years.’
In the first place, it is worth pointing out that a perfect is here used with ær
‘previously, earlier’, unlike what we saw above (cf. Mitchell 1985: §638, with ref-
erences). Attention should also be drawn to the word order, with separation
of the auxiliary and the participle; this, according to Denison (see Chapter 3,
Section 3.1.2), should suggest that (22) is not a real perfect yet. Nevertheless,
there is certainly no agreement, so this is obviously a transition stage and there
is a clash between the various criteria. As demonstrated above, the universal
applicability of criteria such as presence or absence of agreement is at best du-
bious. Both examples arguably show semantic bleaching of have, whereby the
object is not around and cannot be had or possessed, but the agreement does
not seem to be affected by that. Probably due to chance, there are no cases in
this sample of perfects with expected zero morphology – the alleged original
foothold for the loss of agreement.
There are six progressive-like combinations. They are important for two rea-
sons. Firstly, some of them show that -ende is simply the unchangeable present
participle ending, even when the subject is singular, as in (23):
(23) 7 þær was .xii. monaþ wuniende [855]

and there was 12 months living
‘and remained there a twelvemonth’
The subject, implied from a previous clause, is singular in number, and so is

the auxiliary was.16 Secondly, even if one does not believe in the continuity
between these forms and modern progressives, it is still notable that the al-
leged precursors rely on the verb to be, as do the far more numerous passives,
16 However, see Mitchell (1985: §101, §974), who notes that both present and past participles
can be declined strong or weak and provides examples of inflected present participles.
whereas there are no traces of be-perfects here. This might have implications
for functional-load explanations. In any case, be indeed seems to have been
extremely rare in the perfect from the very outset (though have-perfects are
not faring much better in terms of statistics at this stage either).
By contrast, passives with be are relatively numerous. Four of them show
agreement with a plural subject:
(24) 7 þæs geares wær-un of-slæg-en-e .viiii.

and that year were-pl pref-slay-pptcp-m.nom.pl 9
eorl-as 7 an cyning [871]
earl(m)-nom.pl and 1 king
‘And that year nine (Danish) earls were killed and one king’
Eorlas here is certainly nominative plural. It is not in the partitive genitive plu-
ral, which might be expected in the context of numerals. The auxiliary in (25)
is missing from the second clause, but could arguably be recovered from the
previous clause, so both occurrences were regarded as be-passives (the first
one ambiguously stative, however):
(25) 7 þær17 was ungemetlic węl

and there was[sg] excessive slaughter(nt)[nom.sg]
ge-slæg-en Norþanhymbra, sume binnan, sume
pref-slay-pptcp of.Northumbrians some inside some
butan; 7 þa cyning-as begen
outside and that.nom.pl king(m)-nom.pl both
of-slæg-en-e [867]
pref-slay-pptcp-m.nom.pl
‘and an immense slaughter was made of the Northumbrians, some inside
and some outside, and both kings were killed’18
17 On passives introduced by ‘expletive’ þær, see Mitchell (1985: §754). Normally, examples
with (of-/ge-)slægen ‘slain’ were treated as bona fide passives, but those containing ex-
pletive þær and wæl were thought to be potentially stative. This is probably a slippery
distinction.
18 Although they share an auxiliary, these were counted as separate instances because of
the different morphological agreement options on the participles. However, whenever I
encountered the same auxiliary with, say, two zero-marked participles, this was counted
only once, as in the following:
(i) 7 þær wearþ monig mon ofslægen 7
and there became[sg] many man(m)[nom.sg] slain &
a-druncen on gehwęþere hond [853]
pref-drunk on either hand
‘and many men [lit. many a man] on both sides were killed and drowned there’
106 chapter 4
The other eight be-passives had nominative masculine singular subjects so

zero morphology is to be expected for them:
(26) Her Herebryht aldormon wæs of-slæg-en

here Hereberht ealdorman(m)[nom.sg] was[sg] pref-slay-pptcp
from hęþnum monnum 7 monige mid him on
from heathen men and many with him on
Merscwarum [838]
Marshes
‘In this year Ealdorman Hereberht was killed by heathen men and many
men with him in the Marsh’
Essentially, what we get across the board is zero marking in the singular and -e
in the plural; despite the numerical advantage of the legitimate zero-marked
forms, this has not led to the loss of morphology where it is expected, as ad-
ditionally seen from a passive with weorðan from the same year (cf. Section
4.4.2.2):
(27) 7 þy ilcan geare eft on Lindesse, 7 on Eastenglum,

and that same year again on Lindsey and on East.Anglia
7 on Cantwarum wurd-on monig-e men
and on Kent became-pl many-m.nom.pl man(m)\nom.pl
of-slæg-en-e from þam herige [838]
pref-slay-pptcp-m.nom.pl from that enemy.army
‘and later in the same year many men in Lindsey, East Anglia, and Kent,
were killed by the enemy’
Weorðan-passives are actually more numerous than passives with be

here, so it is hardly justified to say that they were always marginal and sec-
ondary (see Mitchell 1985: §§800–801, as well as the discussion in Chapter 3,
Section 3.4.3). Seven weorðan-passives show plural agreement with -e, while
zero morphology in the singular is normal for another ten. There is only one
example which is missing what should be an overt agreement exponent –
(28). It might have been dropped because of the zero neuter plural noun ge-
feoht (though suspicion was voiced regarding such analogical extensions in
Section 4.4.2.2).
(28) 7 þæs geares wurd-on .viiii. folc gefeoht

and that year became-pl 9 folk fight(nt)[nom.pl]
ge-foht-en wiþ þone here on þy cynerice be
pref-fought-pptcp against that army on that kingdom by
suþan Temese [871]
south Thames
‘And during that year nine general engagements were fought against the
Danish army in the kingdom south of the Thames’
This is significant because there are more instances of lack of plural agreement
in the Chronicle than the single one quoted in Mitchell (1985: §34, with refer-
ences, see Section 4.4.2.2 above). The finite verb wurdon is inflected for the
plural, but the participle gefohten is not. Importantly, however, example (28) is
still exceptional and we once again see a robust tendency for preservation of
agreement where it should be overt, regardless of the many cases of expected
zero exponence.19
Similarly to be above, there are analogous cases of ellipsis with weorðan:
(29) þær wearþ micel gefeoht, 7 þær begen

there became[sg] great battle and there both
of-slæg-en-e þa aldormen [800]
pref-slay-pptcp-m.nom.pl that.nom.pl ealdorman(m)\nom.pl
‘a great battle took place, and both ealdormen were killed’
In (29), weorðan is first used as a singular lexical verb; it is then only implied in
the second clause, where it should be plural and combined with an agreeing
past participle. Likewise, fela þusenda ofslægenra (lit.) ‘many of thousands of
killed’ in (30) implicitly goes with earlier wearþ, which should be construed
as lexical and existential in this stretch of the sentence, as opposed to the
rest, i.e. ‘there were many thousands of casualties’ [lexical] v. ‘were slain/re-
pelled’ [auxiliary(-like)]. This sharing of forms (one overt, the other(s) ellipti-
cal) blurs the lines between auxiliary and non-auxiliary uses of the same verb.
19 Gefeoht is a heavy-stem neuter noun which is zero-marked for the nominative plural in
(28) above. The partitive genitive could have been used instead because of the presence
of a numeral, but the genitive plural should take the morpheme -a. Compare (24) above,
in which eorlas follows a numeral but is overtly nominative plural. On verbs with ‘cognate’
objects/passive subjects such as gefohten and gefeoht, see Mitchell (1985: §754). In addi-
tion, zero marking is legitimate in the nominative neuter plural of the strong adjectival
paradigm, but again only for long/heavy stems.
108 chapter 4
Furthermore, when the auxiliary is missing like that, it is sometimes not clear
which form should be supplied, as in (30), where gefliemde ‘repelled, put to
flight’ was taken to belong with the weorðan-passives despite the presence of
progressive wæron in a later clause.
(30) 7 þær wearþ Sidroc eorl of-slæg-en se

& there became[sg] Sidroc earl pref-slay-pptcp that
alda, 7 Sidroc eorl se gioncga, 7 Osbearn eorl, 7
old & Sidroc earl that young & Osbearn earl &
Fræna eorl, 7 Hareld eorl, 7 þa herg-as
Fræna earl & Harold earl & that.nom.pl army(m)-nom.pl
begen ge-fliem-d-e, 7 fela þusend-a
both pref-repel-pptcp-m.nom.pl & many thousand-gen.pl
of-slæg-en-ra, 7 onfeohtende wær-on oþ niht [871]
pref-slay-pptcp-gen.pl & fighting were-pl until night
‘there were slain Earl Sidroc the Old, and Earl Sidroc the Younger and Earl
Osbearn, Earl Fræna, and Earl Harold; and both enemy armies were put
to flight and many thousands were killed, and they continued fighting
until night’
(31) also beautifully demonstrates the happy coexistence of zero forms in the
singular and -e in the plural:
(31) 7 þa wearþ þæt rice

& then became[sg] that.nt.nom.sg kingdom(nt)[nom.sg]
to-dęl-ed on .v. 7 .v. kyning-as to
pref-divide-pptcp on 5 & 5 king(m)-nom.pl to
ge-halg-od-e [887]
pref-hallow-pptcp-m.nom.pl
‘the kingdom was then divided into five, and five kings were consecrated
to it’
The ellipsis in the second clause is not unproblematic, however. As it is, it was
taken to imply a form of weorðan for the gap, but another manuscript inserts
wæron in front of to gehalgode (see Plummer and Earle 1892: 80–81 fn. 5). In
sum, beon and weorðan are interchangeable in the passive (cf. the earlier dis-
cussion and Mitchell 1985: §786). In view of the solid numbers of weorðan and
its greater clarity/functional ‘streamlining’ (debatable though it is), it is myste-
rious why it disappeared (especially for a functionalist).
Although the following were treated as unambiguously dynamic passives, it

cannot be ruled out that they might in fact receive an adjectival/copular inter-
pretation (see Mitchell 1985: §735, where a case is made in favour of an adjecti-
val interpretation of wearð gesicclod ‘became sickened’):
(32) 7 hie wurd-on þær be-hors-ud-e [885]

and they.3nom.pl became-pl there pref-horse-pptcp-nom.pl
‘and they were deprived of their horses there’
(33) 7 þær wearþ micel węl

and there became[sg] great slaughter(nt)[nom.sg]
ge-slæg-en [833]
pref-slay-pptcp
‘and a great slaughter was made there’
However, I would claim that weorðan gave the clause a more dynamic reading,
unlike ambiguously stative counterparts with beon (but recall the discussion
above and Mitchell’s objections to the rigidity of this distinction):
(34) 7 þær wæs micel węl

and there was[sg] great slaughter(nt)[nom.sg]
ge-slæg-en [823]
pref-slay-pptcp
‘and a great slaughter was made there’
It is worth stressing again that this ambiguity between passive and stative
seems to have persisted throughout the history of the language, ‘blind’ to any
functional considerations and ultimately unresolved by the apparent func-
tional advantage of weorðan. In addition, future work will need to assess the
influence of frequent formulae like the one in (34) or geweorc geworht, among
others (cf. Danchev 1969 and the earlier discussion).
In conclusion, Table 2 confirms that the perfect is very rare and agreement
is either randomly present or absent from a clause where it ought to be overt.
There are no cases of expected zero morphology. Semantics seems not to mat-
ter, in that an unequivocally perfect reading can be enforced in both agreeing
and non-agreeing contexts. Passives and statives with beon and weorðan are at-
tested in greater numbers and suggest that loss of non-zero agreement cannot
be tied to ‘rampant’ zero exponence. Although legitimate zero exponence in
the singular is predominant, this has hardly resulted in loss of plural -e.
110 chapter 4
table 3a Perfect and passive auxiliaries in Manuscript A of the Anglo-Saxon Chronicle

(892–924)

recorded)

transitive verbs
intransitive verbs
Be-perfects with 1 2 7 10 15.9%
intransitive verbs
Weorðan-perfects 0 1 0 1 1.6%
with intransitive
verbs
Progressives 0 0 0 0 0%
weorðan
Ambiguous 5 3 5 13 20.6%
passives/statives
with beon
passives/statives
with weorðan
Ambiguous 2 1 0 3 4.8%
be-perfects
4.4.2.4 Manuscript A: Late Ninth and Early Tenth Centuries

The entries for these years from the late ninth and early tenth centuries are
much longer and more extensive. In a way, they can also be said to form
something of a coherent unit (see Macleod 2012: 75). The total word count is
about 4,524 words (based on http://asc.jebbo.co.uk/a/a-L.html, accessed on
16 Oct 2016). Most likely due to the different scribes involved, the style is accord-
ingly slightly different after the second hand commences, and that might have
to do with some of the differences in grammar that I record – more perfects
and fewer bona fide passives than in earlier samples (see Plummer and Earle
1892: 85 fn. 9, 99 fn. 2, 102 fn. 1, 103 fn. 1, where changes of hand are indicated).
The genre and subject-matter is still indisputably the same, however, so the
observed differences must for the most part be genuinely linguistic. Although
most of the earlier entries are much shorter, there is no reason why perfects
should not have been used so much, for instance. It is significant that, by the
tenth century, the perfect is employed (instead of the earlier simple preterite)
to indicate anteriority in the context of adverbs like ær, (35), or even without
needing their temporal adverbial reinforcement, (36).
(35) Þa berad mon þæt wif

then pursued one that.nt.acc.sg wife(nt)[acc.sg]
þæt he hæfd-e ær ge-num-en [901]
that.nt.acc.sg he had-sg earlier pref-take-pptcp
‘Then the woman was seized whom he had taken …’20
(36) 7 ahredd-on eall þæt hie ge-num-en

and rescued-pl all that.nt.acc.sg they pref-take-pptcp
hæfd-on [917]
had-pl
‘and rescued all that they had captured’
As with the other samples so far, the years given after each example are as
they appear in Plummer and Earle (1892), and the translations are again from
Whitelock et al. (1961). At first glance, Table 3a summarising the data may look
rather messy, but once the material is examined more carefully, some interest-
ing patterns emerge.
It is immediately striking that agreement has virtually disappeared from the
perfect, but not from the passive. There are only two unambiguous perfects
that still retain agreement, both of them from the early part of the period, the
890s.21 The first one is already familiar from the literature review in Chapter 3:
20 In line with the remarks on style, note also the non-use of a passive and the preference
instead for an impersonal construction with mon ‘one’ (I return to this issue later in this
section).
21 It should be stressed that overt agreement is lost equally early, however (observe the dates
of the entries in this section).
112 chapter 4
(37) Ac hie hæfd-on þa heora stemn

but they had-pl then their term(m)[acc.sg]
ge-set-en-ne, 7 hiora mete
pref-sit-pptcp-m.acc.sg and their meat(m)[acc.sg]
ge-not-ud-ne [894]
pref-use-pptcp-m.acc.sg
‘but they had completed their term of service and used up their
provisions’
As already noted in the earlier discussion, this example supplies solid confir-
mation that overt agreement is not incompatible with a perfect reading. Both
objects are masculine, accusative and singular, and both participles indicate
overt masculine accusative singular agreement. However, the term of service is
already up, i.e. there is none left, and neither do they have any food, so the only
plausible reading is (at least partially) perfect, with bleached lexical seman-
tics. (37) stresses the resultativeness of the events and may or may not further
indicate anteriority – i.e. it might simply mean that at that past moment the
service was up/complete and the supplies were depleted, without bringing to
the fore the temporal precedence. The comparable PP adjunct in (38) below
suggests that the giving of oaths had a result relevant ‘in this year’, i.e. similar
resultativeness without emphasis on anteriority. (37) was the only agreeing
have-perfect. In ten attestations, overt agreement was expected but was not
there. This is especially noticeable in the plural, as in (38) and (39).
(38) On þys geare, þæt wæs ymb twelf monað þæs þe

on this year that was about twelve months from.that that
hie on þæm eastrice geweorc
they on that east.kingdom work(nt)[acc.sg]
ge-worh-t hæfd-on, Norþhymbre 7 Eastengle
pref-work-pptcp had-pl Northumbrians and East.Angles
hæfd-on Ælf( f )rede cyninge aþ-as ge-seal-d [894]
had-pl to.Alfred king oath(m)-acc.pl pref-sell-pptcp
‘In this year, that was twelve months after the Danes had built the fortress
in the eastern kingdom, the Northumbrians and East Angles had given
King Alfred oaths.’
(39) 7 he him hæfd-e ge-seal-d gisl-as 7

& he him had-sg pref-sell-pptcp hostage(m)-acc.pl and
að-as [894]
oath(m)-acc.pl
‘and he had given him [the king] oaths and hostages’ [NB! hæfde is an
insertion!]
The semantics here is again doubtless perfect. If you give someone your hostag-
es, you no longer have them, but as we saw above, this perfect reading does not
have to preclude overt agreement. In (38), geseald, with its plural morpheme -e
missing, coexists in the same sentence with another zero-marked participle,
geworht, whose zero morphology is however legitimate (see below). According
to the traditional hypothesis proposed in Chapter 3, this might have been
the reason why agreement was ultimately lost from the perfect. ‘Bare’ forms
sneaked in in contexts like this. However, the picture is actually more complex
when different facets of it are examined. Below is another example of missing
overt agreement:
(40) Hæfd-e se cyning his fierd on tu

had-sg that king his army(f)[acc.sg] on two
to-num-en [894]
pref-take-pptcp
‘The king had divided his army into two’
If it agreed, the participle tonumen should take an accusative feminine

singular -e, but it does not. Interestingly, the noun fierd/fyrd itself does not
have the usual accusative singular -e inflection of a strong feminine noun, i.e.
one belonging to the class of o-stems. Could this be another instance of anal-
ogy spreading from the noun to the modifier or could it be gender fluctuation?
In the examples below, collected from Bosworth and Toller, both inflected, (41),
and uninflected, (42), forms of fyrd serve as direct object (cf. Irvine 2004: cxliii):
(41) Claudius, se casere, fyrd-e gelædde on Breotone

Claudius that emperor army(f)-acc.sg led on Britain
‘Claudius, the emperor, led an army into Britain’ [Source: http://bosworth
.ff.cuni.cz/012891, accessed on 03 Oct 2016]
114 chapter 4
(42) Gesomnade he his fyrd wið West-Seaxum

assembled he his army(f)[acc.sg] against West-Saxons
‘he assembled his army against the West Saxons’ [Source: http://bosworth
This noun occurs with a feminine demonstrative determiner (sio) elsewhere in

these Chronicle entries, confirming its gender:
(43) ær sio fierd gesamnod wære [894]

ere that.f.nom.sg army(f)[nom.sg] assembled were
‘before the English force was assembled’ [nom.f.sg -u expected but not
found on gesamnod]22
It is important to point out that agreement and lack thereof can be found in
very close proximity, as in (44):
(44) þa wær-on hie mid metelieste

then were-pl they with want.of.food
ge-wæg-d-e, 7 hæfd-on micl-ne
pref-afflict-pptcp-m.nom.pl and had-pl great-m.acc.sg
dæl þara hors-a fret-en [894]
part(m)[acc.sg] that.gen.pl horse(nt)-gen.pl eat-pptcp
‘they were oppressed by famine, and had eaten the greater part of their
horses’
This sentence also involves ‘provisions’ being eaten and therefore impossible to
literally have in one’s possession (this time the unfortunate horses). The pred-
icative participle gewægde in the first clause does agree with its subject, and
the attributive adjectival modifier miclne shows accusative masculine singular
agreement with the head noun dæl. One might expect to find the same -ne
morpheme on freten (as in the analogous example in (37) above). Coupled
with the semantics of the sentence, the lack of agreement in (44) might thus
testify to the more advanced grammaticalisation of what is already morphing
into a perfect, in comparison with the predicative construction from the first
clause, and also compared to proper adjectival modifiers like attributive mi-
clne, to which the participle used to be more similar before it ceased to agree.
22 It should additionally be noted that there is no perfect passive in this context yet. Due to
the meaning of the verb gesamnod ‘collected, assembled’, this sentence can be perfect or
passive.
The presence of the clear and unambiguous -ne ending on the attributive mod-
ifier invites the conclusion that the agreement was lost due to the more highly
grammaticalised nature of the perfect, not because people simply forgot the
rules and markers of agreement, and perhaps not so much because of zero
morphology or ambiguity elsewhere. As is plain to see, this speaker (writer)
obviously still uses the -ne suffix, and also extensive plural agreement, but not
in the (proto-)perfect. At this stage, it should be reiterated that the absence of
agreement is especially striking in the plural:
(45) 7 seofon ærendrac-an he him hæfd-e to

and seven messenger(m)-acc.pl he them had-sg to
a-send [905]
pref-send[pptcp]
‘and he had sent seven messengers to them’
(46) 7 se cyng hæfd-e ge-gadr-od sum hund

and that king had-sg pref-gather-pptcp some hundred
scip-a [911]
ship(nt)-gen.pl
‘and the king had collected about 100 ships’
(47) þara wicing-a þe hie him to fultume

that.gen.pl Viking(m)-gen.pl whom they them to assistance
a-span-en hæfd-on [921]
pref-entice-pptcp had-pl
‘of the Vikings whom they had enticed to their assistance’
All of these examples could technically have a marker for plurality on the parti-
ciple but they do not. The first one, (45), is the clearest: the object is accusative
masculine plural; morphologically, the form ærendracan cannot be genitive
plural (terminating in -a), despite the preceding numeral which might be used
with a partitive genitive (cf. scipa in (46)); with a great degree of confidence,
asend can thus be expected to receive a plural -e in order to match acc.pl
ærendracan. In (46), agreement might have defaulted because this time scipa
is certainly in the genitive plural after the numeral. In (47), the invariable rel-
ativiser þe is ‘entitled’ to pick up the masculine gender and the plural num-
ber of the antecedent wicinga, and the participle aspanen might have copied
those features (see below for such examples). And yet, the adjectival agree-
ment markers are not there, most likely pointing to the verbal status of those
participles.
116 chapter 4
Zero morphology was expected for 11 of the transitive have-perfects. One

was already given in (38) above and is also the prototypical example cited on
numerous occasions in the literature (see Chapter 3). The OE word for fortress,
geweorc, is neuter and no overt agreement exists for it in this syntactic context.
The subject matter of these annals presupposes that this word and this type of
construction will feature prominently, and examples of this kind are indeed
frequent in my sample – 3 of the 11 legitimately inflectionless participles modi-
fied the noun geweorc ‘work, fortress’, as in (48). The influence of this lexeme
and the entire idiomatic collocation might have been felt because it was re-
peated often due to what these passages describe (see discussion of corpus
design above).
(48) Hæfde Hæsten ær ge-worh-t þæt

had Hæsten previously pref-work-pptcp that.nt.acc.sg
geweorc æt Beamfleote, 7 wæs þa ut afaren
work(nt)[acc.sg] at Benfleet and was then out travelled
on hergaþ [894]
on raid
‘Hæsten had previously built that fortress at Benfleet; and he was then
out on a raid’
In terms of time reference, the linearity of the narrative presupposes that

Hæsten first had the fortress built and then went on a raid, which is to say
that (48) may involve just a straightforward sequence of events, rather than
the ‘past in the past’ expected of today’s ‘genuine’ pluperfect. Ær … þa ‘first/
earlier … (and) then’ indicates that had + past participle happened before the
event denoted by was + past participle. Alternatively, one might enforce a plu-
perfect reading on both constructions, denoting events that precede the past
moment of orientation designated by þa ‘then’ – ‘H. had previously built the
fortress and had then gone on a raid’. Another point that needs to be made
regarding this example is that Hæsten most probably did not build the fortress
himself, but rather had it built. In other words, this may well be a causative
use of have. In the next chapter, I demonstrate that ambiguity between per-
fect and causative have resurfaces as late as the 17th century. There were many
more rival (potential?) causatives with the preterite form het, however – 15 in
total (again numerous due to what the scribe/compiler was interested in talk-
ing about):
(49) 7 se cyng het ridan æfter [901]

and that king ordered ride after
‘And the king ordered them to pursue [him]’
This indeed seems to be the causative construction of choice in the passages.

However, it is only used in reference to a king as the syntactic subject, in a
rather formulaic way, so it is perhaps not fully grammaticalised (and it prob-
ably never was).
More of the subtleties of agreement with neuter nouns are demonstrated
in (50):
(50) 7 þa Deniscan hæfdon hira wif befæst

& those Danes had their woman(nt)[acc.pl] secured
innan Eastengle [896]
in East.Anglia
‘And the Danes had placed their women in safety in East Anglia’
Accusative neuter plural -u is expected only with short/light stems, but this
one, befæst, is long/heavy due to the consonant cluster, so zero is customary for
it anyway (at least in attested, as opposed to prehistoric/reconstructed, OE).
In addition, the zero plural of heavy neuter a-stems like wīf might have further
reinforced zero marking (as noted above). And yet, this is not all there is to it,
as becomes evident from (51):
(51) 7 hæfd-e eall-e þa geat-u for-worh-t

& had-sg all-pl that.acc.pl gate(nt)-acc.pl pref-work-pptcp
in to him [901]
in to him
‘and he had barricaded all the gates against him’
In (51), eall is also a heavy stem, but it is marked for the plural with -e.23 True,
this is not the ‘standard’ accusative neuter plural -u, but it is plural nevertheless
(unless it is an adverb, which seems unlikely). The heavy-stemmed forworht,
by contrast, is not inflected, so it cannot be just the type of stem or analogy on
zero-marked plural nouns that matters (cf. the -u on geatu in (51), as opposed
to zero-marked plural wīf in (50)). Things are more complex and it is not easy to
23 If the double consonant had by then been degeminated, eall might perhaps have been
treated as a light syllable.
118 chapter 4
say why ealle is inflected in (51), but forworht is not, since both of them contain
a consonant cluster. It is perhaps reasonable to suggest that this is not to do
with the influence of ‘expected zero morphology’, because these expectations
seem to have been overridden. It appears that ealle agrees in (51), despite its
heavy stem, because of the type of construction it is in, i.e. it is in an attribu-
tive plural environment. If it attributively modifies a neuter singular noun/pro-
noun, as in (52)–(53) below, it remains (legitimately!) uninflected, but this has
not prevented it from agreeing in the plural, and neither has its stem type. By
contrast, forworht in (51) might be uninflected due to its predicative position
and/or due to the fact that grammaticalising the perfect was already underway.
(52) 7 oferhergade eall hira land [905]

and harried all their land(nt)[acc.sg]
‘and harried all their land’
(53) 7 ahredd-on eall þæt hie ge-num-en

& rescued-pl all that.nt.acc.sg they pref-take-pptcp
hæfd-on [917]
had-pl
‘… and rescued all that they had captured’24
Other reasons to expect no overt agreement are clausal objects, (54), or no ob-
jects at all, (55), both of which are attested (once each) in the sample:
(54) 7 se cyng hæfde fund-en [þæt him mon sæt

and that king had find-pptcp that them one sat
wiþ on suþhealfe Sæfernmuþan…] [918]
against on south.half of.Severn.mouth
‘and the king had arranged that men were stationed against them on the
south side of the Severn estuary’
(55) Þa hie ða þæt geweorc furþum ongunnen hæfdon, 7

when they then that work just begun had and
þærto ge-wic-od hæfd-on [896]
thereto pref-camp-pptcp had-pl
‘When they had just begun that work and had encamped for that
purpose …’
24 Land in (52) could conceivably be a zero plural as well, and eall might serve as the head of
the NP in (53).
(55) shows that have can already occur with an intransitive verb (gewicod).
No agreement can be triggered here because there is no accusative object. It is
also curious that this appears in the same sentence as a zero-inflected transi-
tive past participle (ongunnen in the context of geweorc).
The situation of have-perfects is in some ways similar to that of the other
perfect (proto-)auxiliaries in OE – beon and weorðan. Only one instance of
agreement (in the plural) was recorded, given in (56).
(56) wær-on þa men uppe on londe of

were-pl that.nom.pl men(m)\nom.pl up on land of
a-gan-e [897]
pref-gone-m.nom.pl
‘the men from them had gone up on land’
Here too, agreement is much more frequently missing from the perfect, as
in (57), indicating that grammaticalisation must have been underway.
(57) 7 þa oþr-e wær-on hungre

and that.nom.pl other-m.nom.pl were-pl of.hunger
a-cwol-en [894]
pref-die-pptcp
‘and the rest had died of starvation’
Once more, it should be noted that there is agreement on attributive oþre, but
not on the participle (as was the case with the have-perfect).25 Another ques-
tion that arises is whether it is possible to give (57) a passive or stative interpre-
tation. Judging by Bosworth and Toller’s dictionary entry and examples quoted
below, the strong verb a-cwelan (pptcp a-cwolen/a-cwelen) is intransitive,
meaning ‘to die, perish’, a translation equivalent of Latin mori ‘die’ (a deponent
verb, i.e. passive in form but active in meaning):
(58) pisces morientur Latin

ða fixas acwelaþ Old English
those fish die
‘The fish shall die’ [Ex. 7, 18; source: http://bosworth.ff.cuni.cz/000222,
accessed on 16 Oct 2016]
25
O E oþer is always declined strong, even after a demonstrative (see Mitchell 1985: §508, §568,
and below). It is, strictly speaking, used (semi-)independently here, without a following
noun.
120 chapter 4
(59) Ofercumen biþ he ær he acwele

overcome is he ere he die
‘he will be overcome ere he dies’ [Exon, 90b; Th. 340, 10; Gn. Ex. 114; source:
http://bosworth.ff.cuni.cz/000222, accessed on 16 Oct 2016]
(60) Monige men hungre a-cwæl-on

many men of.hunger pref-died-pl
‘many men died of hunger’ [Chr. 918; Erl. 104, 13; source: http://bosworth
No transitive uses were found in Bosworth and Toller, and (60) is an active
intransitive clause in the preterite with a structure otherwise similar to that
of (57), so (57) must be a perfect. Mitchell (1985: §795) supplies the follow-
ing potentially passive example: … mid hungre wurdon acwealde ‘(lit.) with
hunger became dead/killed’, where the participle is weak and belongs to a
separate lexeme. There is indeed a transitive verb a-cwellan, ‘to kill’, which is
weak, with a past tense a-cwealde, and past participle a-cweald (see http://
bosworth.ff.cuni.cz/000223, accessed on 14 Jan 2017). Thus, pairs of verbal
lexemes (strong intransitives alongside weak transitives) serve to differentiate
between be-perfects and passives.
There was another interesting be-perfect for which agreement had failed:
(61) Þreo a-sæt-on on ða healfe þæs deopes ðe

three pref-sat-pl on that half of.that deep which
ða Denisc-an scip-u a-set-en
that.nom.pl Danish-nom.pl ship(nt)-nom.pl pref-sit-pptcp
wær-on [897]
were-pl
‘three were aground on that side of the channel on which the Danish
ships were aground’
As the translation indicates, the original does not necessarily possess the
meaning of ‘past in the past’ – it may still be viewed as preterite be with a
participial subject complement (i.e. an adjective), as it were painting a picture
of what could be seen at that past moment. Yet again, the attributive adjective
Deniscan agrees (in the weak declension), but the predicative participle aseten
does not bear nom.nt.pl -u (this state of affairs might be similar to what
we find in Modern German, for which see Eisenberg 1994: 366). Crucially, this
sentence from the Chronicle goes on as follows:
(62) 7 þa oðr-u eall on oþr-e

and that.nom.pl other-nt.nom.pl all on other-f.acc.sg
healf-e
half(f)-acc.sg
‘and all the others on the other side’
As already mentioned, oþer/oðer is always declined strong, even in attribu-

tive position after demonstratives (see Mitchell 1985: §508ff., §568, and fn. 25
above); here it receives the expected nom.nt.pl -u agreement morpheme (rare
though it may be in general); eall, however, remains uninflected in this manu-
script, receiving an -e in MSS B., C. and D. (see Plummer and Earle 1892: 91 fn. 2).
Eall may well be an adverb, as suggested by the word order (cf. eall þa oðru).
With the be-perfect, zero morphology was expected in seven of the in-
stances, some of which might be ambiguously stative. As elsewhere, there was
much more ambiguity between ambivalent passives and statives (see below),
but this has not prevented the grammaticalisation and retention of be-passives
(despite the greater ‘suitability’ of weorðan). The proportions of ambiguous
perfects and passives here are similar to those from later periods (cf. Table
3a and Chapter 5). While later texts (ME, EModE) rely predominantly on the
syntactic and semantic context to disambiguate, verbal morphology is often
enough in OE to dispel uncertainty of interpretation, as with acwolen above,
as well as (63) below:
(63) 7 him cierde eall þæt folc to

and him submitted all that.nt.nom.sg folk(nt)[nom.sg] to
þe on Mercna lande ge-set-en wæs [922]
that on of.Mercians land pref-sit-pptcp was[sg]
‘and all the people who had settled in Mercia submitted to him’
The example in (63) is morphologically disambiguated as perfect and not pas-

sive. Geseten must be a form of the strong, usually intransitive verb sittan (past
singular sæt, past plural sæton, past participle seten), not the mostly transitive
derived (via umlaut) weak verb settan (past singular sette, past participle seted,
set[t]). This is confirmed by the example with preterite plural asæton and the
past participle aseten in (61) above. A potentially transitive use of the strong
verb might be supplied from (37): hæfdon heora stemn gesetenne ‘(lit.) [they]
had sat [i.e. completed] their term [of service]’. It is nevertheless not inconceiv-
able to think of the object as a quasi-adverbial modifier, i.e. ‘they had sat for the
entire term of service’. Similarly, in (66) below, wurdon aseten will be treated
122 chapter 4
as an intransitive weorðan-perfect, not a passive. The participles gefaren and

geseten are also listed as mutative intransitives in Mitchell (1985: §740). By con-
trast, Mitchell (1985: §748) gives an example of a passive with transitive settan:
wearð (to abbode) geset ‘became/was (as abbot) set/ordained/appointed’ (cf.
Mitchell 1985: §784, §795).26
And yet, despite such morphological clues, the incipient displacement of
be by have is already underway:
(64) Þa he þærto ge-far-en wæs [894]

when he.3m.nom.sg thereto pref-travel-pptcp was[sg]
‘when he arrived there’
This is the entry from Manuscript A. Manuscript D of the same text, how-
ever, has gefaren hæfde (see Plummer and Earle 1892: 87 fn. 1; cf. Mitchell
1985: §740). Significantly, MS. D is another representative of the so called
26 Here I give a taste of the intricacies of classification and analysis, this time regarding faran
‘travel’. Bosworth and Toller list the following forms and definititions for the instransi-
tive strong verb faran (to farenne; ic fare, ðu farest/færest/færst/færsþ, he fareþ/færeþ/færþ,
present plural faraþ; past singular for, past plural foron; past participle faren): “A word
expressing every kind of going from one place to another, hence ‘to go, proceed, travel,
march, sail; īre, vādĕre, incēdĕre, transīre, migrāre, nāvīgāre’ (http://bosworth.ff.cuni.
cz/010118, accessed on 17 November 2016)”
By contrast, the weak counterpart feran has transitive (alongside intransitive) uses, as
cited in Bosworth and Toller (cf. also Mitchell 1985: §735):
feran, to ferenne; past singular ferde, past plural ferdon; past participle fered [ fer ‘a
journey’]
‘to go, make a journey, set out, travel, march, sail; īre, ĭter făcĕre, proficisci, transīre,
migrāre, nāvĭgāre’
A transitive collocation involves mycelne siþfæt feran ‘go so great a journey’ (http://bo-
sworth.ff.cuni.cz/010498, accessed on 17 November 2016). Finally, there is a third entry
from Bosworth and Toller (cf. Mitchell 1985: §735): ge-fered (adj./ptcp.) ‘associated, band-
ed together’ (http://bosworth.ff.cuni.cz/047936, accessed on 17 November 2016).
According to Mitchell (1985: §781), geferede can be active or passive in Beowulf 361 Her
syndon geferede, feorran cumene … ‘(lit.) Here are/have? travelled, far come …’, though
it is hard for me to get how it can be passive. Then Mitchell goes on to supply an ex-
ample of intransitive and transitive gefaren too, so figuring out valency is often far from
straightforward:
(i) he hæfde on þæm emnete gefaren
‘he had travelled on the level ground’ [intransitive, Orosius 186.22, quoted in Mitchell
(1985: §781), translation mine]
(ii) Ða he ða þreo burga gefaren hæfde … [þa þreo burga as a quasi-adverbial?]
‘When he had travelled/visited the three cities’ [transitive, LS 34.19, quoted in
Mitchell (1985: §871), translation mine]
‘Northern Recension’ of the Chronicle, so the replacement might be linked

to Scandinavian influence in areas settled by Norse speakers, as surmised in
Chapter 3 (this point will be revisited in the discussion of MS. E below).
The disambiguating properties of OE derivational morphology are neatly
illustrated by (65), where ahebbad is from intransitive a-ebbian ‘to ebb away, to
recede’, and is hence perfect, whereas beebbade is from the transitive counter-
part be-ebbian ‘to strand or leave aground by ebbing’, derived with the transi-
tivising prefix be-, and hence passive or potentially copular/stative (see entries
in Bosworth and Toller):
(65) Ac ða þæt wæter

but when that.nt.nom.sg water(nt)[nom.sg]
wæs a-hebb-ad fela furlanga from þæm scipum.
was[sg] pref-ebb-pptcp many furlongs from those ships
þa eodan ða Deniscan from þæm þrim scipum to þæm
then went the Danes from those three ships to those
oðrum þrim þe on hira healfe be-ebb-ad-e
other three which on their side pref-ebb-pptcp-pl
wær-on [897]
were-pl
‘But when the water had ebbed many furlongs from the ships, the Danes
from the remaining three ships went to the other three which were
stranded on their side’
Overt agreement and legitimate zero morphology coexist side by side.

Importantly, this example further demonstrates that there is agreement in
a relative clause with the invariable relativiser þe. The antecedent scipum
should be neuter, but the agreement is indicated by the originally masculine
plural -e morpheme, which however can be used for all genders (cf. Section
4.4.2.2 above). The relativiser þe has picked up the plural feature of the an-
tecedent and passed it on to the auxiliary and the participle. Once again, it
should be underscored that the rate of ambiguity of perfect vs. passive has re-
mained more or less stable, as has the rate of ambiguous passives/statives, for
an incredibly long time, in the periods that I have studied (though see further
Chapter 5).
One (unambiguous) perfect with weorðan was also attested, with no agree-
ment where it could have been overt (see Chapter 3 for a brief mention of
weorðan-perfects). The auxiliary in (66) is plural but the participle remains
uninflected. This behaviour makes perfect beon and perfect weorðan different
from their passive auxiliary counterparts.
124 chapter 4
(66) Þa comon forðy onweg ðe ðara

that.nom.pl came therefore away that that.gen.pl
oþer-ra scip-u asæton, þa
other-gen.pl scip(nt)-nom.pl ran.aground that.nom.pl
wurd-on eac swiðe uneðelice a-set-en [897]
became-pl also very awkwardly pref-sit-pptcp
‘These got away because the ships of their opponents ran aground.
Moreover, they had run aground very awkwardly’
Probably due to the stylistic preferences of the new scribes from this period,
(proto-)progressives, numerous in the previous century, are now not used at
all. Similarly, passives proper are much rarer too. Instead, those scribes seem to
favour the formally active construction with indefinite/impersonal man/mon
‘one’, (67) (cf. Irvine 2004: clviii; see also examples earlier in this section; the
uses of OE man/mon are dealt with in Mitchell 1985: §§363–377, §747).
(67) Þa slog hie mon [918]

then attacked them one
‘then they were attacked’ [lit. ‘then one struck/attacked them’]
As mentioned above, despite the low numbers, the agreement trends from the
earlier entries are in evidence in this section too. Two be-passives show plural
agreement, one of them is given in (68):
(68) þæt wæs forðy þe hie wær-on

that was because they.3nom.pl were-pl
be-num-en-e ægðer ge þæs ceapes, ge þæs
pref-take-pptcp-m.nom.pl both of.that cattle and of.that
cornes, ðe hie gehergod hæfdon [895]
corn which they ravaged had
‘That was because they were deprived both of the cattle and the corn
which had been ravaged’
No overt agreement is expected for another two, (69)–(70), whereas agreement

breaks down only once, with a feminine singular subject, (71), where nom.f.sg
gehalgodu fails to show up. Note in addition that the combination of perfect
and passive is absent from the OE originals.
(69) forþæm hiora cyning wæs ge-wund-od on

because their king(m)[nom.sg] was[sg] pref-wound-pptcp on
þæm gefeohte [894]
that fight
‘because their king had been wounded in the battle’
(70) 7 þær wæs Ecwils cyng of-slæg-en [911]

and there was[sg] Eowils king(m)[nom.sg] pref-slay-pptcp
‘and there was killed King Eowils’27
(71) forðon ðe heo wæs ær to nunnan

for.that that she.3f.nom.sg was[sg] previously to nun
ge-halg-od [901]
pref-hallow-pptcp
‘for she had been consecrated a nun’
The same scenario is replicated with passive weorðan. The singulars (a total
of five) have the expected zero marking, as in (72), and the plurals (a total of
three) take overt -e, as in (73)–(74), with no defaults in overt exponence.
(72) 7 þær wearð Ordheh cyninges þegn

and there became[sg] Ordheah king’s thegn(m)[nom.sg]
of-slæg-en [894]
pref-slay-pptcp
‘and the king’s thegn Ordheah was killed’
(73) 7 se dæl þe þær aweg com

and that.m.nom.sg part(m)[nom.sg] that there away came[sg]
wurd-on on fleame ge-ner-ed-e [894]
became-pl on flight pref-save-pptcp-m.nom.pl
‘and the part that escaped were saved by flight’
27 This type of clause was treated as a dynamic passive, as opposed to a copular stative, be-
cause of the meaning contribution of the adjunct of place, either a PP or an AdvP. In (70),
this is unlikely to be the (potentially) expletive þær which was encountered above in the
context of wæl – see esp. example (25) in Section 4.4.2.3.
126 chapter 4
(74) 7 þær wurd-on ge-fliem-d-e,

and there became-pl pref-repel/put.to.flight-pptcp-m.nom.pl
7 sume feower cyninges þegn-as
and some four king’s thegn(m)-nom.pl
of-slæg-en-e [896]
pref-slay-pptcp-m.nom.pl
‘and there [they] were put to flight and four king’s thegns [were] slain’
(73) is especially interesting, because a collective singular noun, dæl, triggers

singular agreement in the relative clause and plural agreement in the main
clause. On agreement with singular collective nouns, for which targets can
switch between singular and plural, see Mitchell (1985: §§78–86). The example
in (75) below, on the other hand, might be a little problematic:
(75) 7 þara Denisc-ena þær wearð ma

and that.gen.pl Danish-gen.pl there became[sg] more
of-slæg-en [905]
pref-slay-pptcp
‘but more of the Danes were killed’
This is probably default neuter zero agreement. (75) is a continuation of anoth-

er clause so the (implied) subject might be neuter singular wæl. Importantly,
the auxiliary wearð is singular anyway, which could be because the subject is
actually ma; the phrase þara Deniscena is in the partitive genitive plural so it is
not expected to trigger agreement.
One of the three ambiguous be-perfects/passives/statives is given in (76).
Both a perfect (‘had assembled/gathered’) and a passive reading (from ‘some-
body had assembled/gathered them’) are conceivable, as is a more stative read-
ing (i.e. ‘they were together/ready’).
(76) þa herg-as wær-on þa

that.nom.pl enemy.army(m)-nom.pl were-pl then
ge-gader-od-e begen to Sceobyrig on
pref-gather-pptcp-m.nom.pl both to Shoebury on
Eastseaxum [894]
Essex
‘the (other) two Danish armies were assembled at Shoebury in Essex’
Finally, there were many more cases of what were judged to be ambiguous pas-
sives or statives. This type of ambiguity remains rife throughout the periods
studied. There were three agreement categories for those sentences too: five
of them showed overt agreement, all in the plural; zero morphology was nor-
mal for another five, which had masculine or neuter singular subjects; there
were also three breakdowns in agreement (see Appendix 2.128 and Table 3b).
However, all three of them contain feminine singular subjects, as in (77).
(77) 7 hie gebette 7 geedneowade þær heo ær

& her repaired & restored where she.3f.nom.sg ere
to-broc-en wæs [921]
pref-break-pptcp was[sg]
‘and repaired and restored it where it had been broken’29
Several points need to be reiterated at this juncture. Firstly, lack of agree-

ment is not incompatible with a stative interpretation, in the same way that
presence of agreement does not preclude a dynamic perfect interpretation.
Secondly, agreement here does not default in the plural, unlike the perfect (cf.
Table 3b and Kilpiö 2007: 330ff. for the have-perfect). According to a tentative
suggestion by Kilpiö (2007), in later periods the have-perfect apparently came
to show more plural than masculine singular agreement, thus realigning itself
with the passive and be-perfect. The next section herein, however, does not
seem to bear this out, and neither does Kilpiö’s data, at least not conclusively.
In conclusion, there is a great deal of expected zero morphology in the per-
fect with habban, beon and weorðan, as well as in beon- and weorðan-passives
(see Table 3b). However, while the ubiquity of zero exponence may appear to
have led to an almost complete loss of agreement in all three types of perfect,
including in the plural, this is certainly not the case in the passive. Both types
of passive construction, with beon and weorðan, demonstrate a situation in
which zero morphology is used in the singular (legitimately with the mascu-
line and neuter, not so with the feminine), whereas -e appears consistently in
the plural. This state of affairs is confirmed by the ambiguous statives. It is con-
ceivable that agreement in the nominative feminine singular was lost because
of analogical pressure from the masculine and neuter, but no such loss occurs
in the plural. This might have to do with the make-up of the strong adjectival
paradigm. A crucial difference is that zero is the predominant marker in the
nominative singular – only light/short feminine stems could take non-zero
29 On the non-occurrence of the perfect passive in such contexts, see Mitchell (1985: §§782–
785), the discussion above, and Chapter 3.
128 chapter 4
table 3b Different morphosyntactic behaviour of have-perfects, beon/weorðan-perfects

and beon/weorðan-passives (and/or statives) in Manuscript A of the Anglo-Saxon
Chronicle (892–924)

agreement agreement morphology auxiliary
recorded)
Have-perfects with 1 (m.sg) 10 (m.sg, 11 (nt.sg, 22 34.9%

transitive verbs f.sg, pl) nt.pl,
clausal obj)
Have-perfects with 0 0 1 (no obj) 1 1.6%
intransitive verbs
Be-perfects with 1 (pl) 2 (pl) 7 (m/nt.sg) 10 15.9%
intransitive verbs
Weorðan-perfects with 0 1 (pl) 0 1 1.6%
intransitive verbs
Passives with beon 2 (pl) 1 (f.sg) 2 (m.sg) 5 7.9%
Passives with weorðan 3 (pl) 0 5 (m.sg) 8 12.7%
Ambiguous passives/ 5 (pl) 3 (f.sg) 5 (m/nt.sg) 13 20.6%
statives with beon
Ambiguous passives/ 0 0 0 0 0%
statives with weorðan
Ambiguous be- 2 (pl) 1 (f.sg) 0 3 4.8%
perfects/passives/
statives
marking, while zero is found in all the other nominative slots. By contrast, zero
is legitimate only with neuters in the accusative singular, and only with heavy
neuters in the nominative/accusative plural of the same paradigm.
For both have- and be-perfects, there are many examples where zero mor-
phology is indeed expected – the overwhelming majority with be-perfects, and
in the case of have-perfects, as many as the examples in which overt morphol-
ogy should be present but is nevertheless missing. However, if that is believed
to be the reason why agreement disappeared across the board in the perfect
(and this sample does demonstrate virtually no agreement at all, with two ex-
ceptional survivors), some questions still await answers. In a single sentence,
agreement is often present in non-perfect constructions where it is absent

from the perfect, but this might be because these other agreeing constructions
were not being grammaticalised in the same way. More worryingly, the pas-
sive (here and in the previous sample) also had a great deal of legitimate zero
morphology, at least as much as the perfect, but overt agreement is much more
robust in the passive, particularly in the plural. According to some, it might
be too far-fetched to state that the passive was not being grammaticalised to
the same extent just because of those disparities in agreement behaviour be-
tween the perfect and the passive, though it is not implausible a priori. On this
view, more advanced loss of agreement is a consequence, and hence indica-
tive, of more advanced grammaticalisation. In the parlance of Construction
Grammar, the gradual diachronic process of grammaticalising the perfect and
passive schemas (both instances of a more abstract verbal periphrastic sche-
ma) has resulted in synchronic gradience regarding the extent of entrench-
ment of the two schemas in the OE slice in time (see Traugott and Trousdale’s
2010 programmatic paper, as well as the other contributions to that volume,
Traugott and Trousdale 2013, and Chapters 2 and 3 herein).
In either case, the reason both constructions were ultimately grammati-
calised and the agreement gradually disappeared from them should not be
identified with the influence of legitimate zero morphology. Rather, it is more
probable (despite my small, statistically insignificant numbers) that overt
marking was first lost in the have-perfect because marking in the accusative
might be expected to vanish before marking in the nominative. Be-perfects
might have been influenced by have-perfects, but they lag slightly behind
and do retain one agreeing nominative plural, placing them somewhere be-
tween have-perfects and passives (though certainly gravitating to the former).
Readers should recall the statistics confirming the larger number of declined
participles in be-perfects from Chapter 3, Section 3.2, the claims of the lesser
extent of grammaticalisation of the be-perfect from Chapter 3, Sections 3.2 and
3.5.4, as well as the idea of a network of interlinked constructions. The gradi-
ence stands out even more clearly when ambiguous be-perfects are taken into
account, this time patterning with passives/copulars. Interestingly, the sole
weorðan-perfect has likewise shed the plural adjectival declension on the par-
ticiple, but not the passives with the same auxiliary. Passives, which are often
ambiguously stative and therefore closer to genuine copulars, preserve agree-
ment longest.
Constructions with beon and weorðan were still seen as closer to adjectival/
copular structures, due to the ambiguity in them and perhaps even due to a
lesser extent of grammaticalisation – hence the greater amount of agreement.
130 chapter 4
In both copulars and proto-passives there is an overt copula linking the subject
and the (original) subject complement, indicating the relationship between
them, whereas the link between an object and an object complement (as in the
proto-have-perfect) is not overtly expressed, making it easier to reanalyse the
object complement as a non-agreeing verbal element. In other words, the pas-
sive makes explicit the relationship between the active object and the comple-
ment participle, which is implicit in SVOC O structures, including those with
have – cf. I had him bound v. He was bound (by me). Admittedly, this scenario
is also based on a kind of marking (instantiated in the overt/covert copula),
but this kind of marking is much more ubiquitous and with arguably more far-
reaching consequences than non-explicit morphology in certain slots of the
paradigm. Thus, zero morphology and its impact appear to have been over-
rated in the specialist literature and there are other plausible reconstructions
of how things must have proceeded.
4.4.2.5 Manuscript E (Peterborough Chronicle): Late Tenth and Early

Eleventh Centuries
This extract from the Laud Manuscript (MS. E) of the Anglo-Saxon Chronicle,
the so-called Peterborough Chronicle, covers the late tenth and the early elev-
enth centuries, more specifically the entries for the years 991 through 1012 (cf.
Irvine 2004, Macleod 2012: 75, for textual background). The sample contains ap-
proximately 3,640 words (based on http://asc.jebbo.co.uk/e/e-L.html, accessed
on 28 Oct 2016). The later date of composition/copying is immediately evident
from its linguistic features. By now, the morphological system of Old English
is undergoing massive transformations as a result of the pervasive influence of
sound change. In the material below, unstressed endings are often mixed up,
and there is confusion of -on, -an, -en, and -um, for instance (see Irvine 2004:
cxxiv, cxxxix ff.). Word-final /m/ and /n/ are likewise sometimes confused (cf.
spellings like for-þam/for-þan ‘for that, because’ in (79) and (95) below), as are
the vowels of the unstressed demonstratives se and seo. Even against this back-
drop of cataclysmic metamorphosis, agreement still seems to be working as it
did in the previous centuries. The same contrast shows up again between the
behaviour of the perfect and the passive, as highlighted in Table 4.
Table 4 confirms that agreement is more or less completely missing from the
perfect, but it still holds its ground (in the plural) in passives (including poten-
tially stative ones). Only one transitive have-perfect retains agreement and it is
noteworthy that the controller is a feminine singular noun (burh), with agree-
ment indicated by the ending -e:
table 4 Perfect, passive and progressive auxiliaries in Manuscript E of the Anglo-Saxon

Chronicle (991–1012)

recorded)
Have-perfects 1 (f.sg) 4 (m/f.sg, 1 (genderless 6 14.6%

with transitive pl) obj)
verbs
Have-perfects 0 0 1 (no obj) 1 2.4%
with intransitive
verbs
Be-perfects with 0 0 0 0 0%
intransitive
verbs
Weorðan- 0 0 0 0 0%
perfects with
intransitive
verbs
Progressives 0 0 3 3 7.3%
Passives with 2 (pl) 1 (f.sg) 13 (m/nt.sg, 16 39%
beon clausal subj)
Passives with 1 (pl) 1 (f.sg) 3 (m.sg) 5 12.2%
weorðan
Ambiguous 1 (pl) 0 3 (m/nt.sg) 4 9.8%
passives/statives
with beon
Ambiguous 0 0 1 (m.sg) 1 2.4%
passive/copular
clauses with
weorðan
Ambiguous 0 2 (f.sg) 3 (m.sg) 5 12.2%
be-perfects/
passives/statives
132 chapter 4
(78) 7 þa hi hæfd-on þa burh

and when they had-pl that.f.acc.sg borough(f)[acc.sg]
eall-e a-sme-ad-e.
all-f.acc.sg?/adv? pref-ransack-pptcp-f.acc.sg
wend-on him þa to scip-on [1011]
turned/went-pl them.refl then to ship(nt)-dat.pl
‘and when they had then ransacked the whole borough, they turned/
went to their ships’
As pointed out above, the confusion of endings is evident from cases like to
scipon, which should actually be to scipum (dative plural). However, this late
entry contains overt accusative feminine singular agreement on the participle
asmeade, which might be deemed strange and unexpected, especially in the
singular, considering the date of this text. Perhaps the -e was somehow still felt
to be a more distinctive ending, and ealle might suggest adjectival status for
the participle too, which arguably increases the likelihood of agreement. Ealle
could be an adjective or perhaps an adverb (see the previous section and the
conclusions in Chapter 9). In addition, burh is a zero-marked feminine singular
noun, but there is an -e on the adjective/participle, so any purported analogy
on zero-marked nouns has failed to operate here (cf. Section 4.4.2.2, and the
opposite scenario in (79) – -e on the noun and zero on the participle).
The partially conservative character of the manuscript might have to do with
the fact that this kind of language was probably transmitted as a learned writ-
ten register influenced by earlier exemplars rather than being a pure reflection
of the naturally acquired spoken language. Or rather, the two were most likely
mixed – the earlier written documents serving as conservative bookish models
and the actual spoken language at the time (see Irvine 2004 for textual trans-
mission). Still, the consistency in agreement patterns is striking, even more so
against the background of general ‘confusion’ elsewhere.
Not surprisingly, cases in which agreement has failed in the have-perfect
are more numerous (as before) – four of the six transitive have-perfects here
are not inflected where they could have been. In (79), the accusative feminine
singular -e is missing from the predicative participle gemarcod, but it is pres-
ent on the attributive modifier ælce, perhaps suggesting that, unlike ælce (and
maybe asmeade above), gemarcod was not felt to be a modifier related to the
direct object to the same extent.
(79) forþan þe hi hæfd-on ælc-e scir-e on

for.that that they had-pl each-f.acc.sg shire(f)-acc.sg on
West Sexum stiðe ge-marc-od mid bryne. 7 mid
West Saxons cruelly pref-mark-pptcp with burning and with
hergunge [1006]
harrying
‘for they had cruelly left their mark on every shire of Wessex with their
burning and their harrying’
These sentences demonstrate consistent use of inflection (-e) on genuine

modifiers (ealle and ælce), or where the participle is potentially felt to be
more of an adjectival modifier than anything else (asmeade); by contrast, if
the participle is seen as more verbal and less of a modifier (arguably the case
of gemarcod), agreement is dropped, and this does not seem to be linked to
the influence of expected zero morphology or to the scribe/author not using
agreement elsewhere. This hypothesis receives support from the masculine:
(80) Her wæs Eaxeceaster abrocen þurh þone

here was Exeter broken through that.m.acc.sg
Frencisc-an ceorl Hugon þe seo hlefdige
French-m.acc.sg churl(m)[acc.sg] Hugh that that lady
heafd-e hire ge-set-t to gerefan [1003]
had-sg her pref-set-pptcp to reeve
‘In this year Exeter was stormed on account of the French ceorl [churl]
Hugh, whom the queen had appointed as her reeve’
The antecedent of the invariable relative particle þe is the ‘churl Hugh’, so gesett
could legitimately have hosted the accusative masculine singular ending -ne.
But it does not. Interestingly enough, a little later in the same entry, an adjecti-
val object complement does agree in the accusative masculine singular:
(81) þa gebræd he hine seoc-ne [1003]

then feigned he.3m.nom.sg him.3m.acc.sg sick-m.acc.sg
‘he feigned him[self] sick’ [i.e. ‘he pretended to be sick’]
In another similar construction from the year 1009, agreement is present on

both adjectival object complements which describe the object:
134 chapter 4
(82) þæt he Wulfnoð cucon-ne oððe

that he Wulfnoth(m)[acc.sg] quick-m.acc.sg or
dead-ne begytan sceolde [1009]
dead-m.acc.sg get should
‘to capture Wulfnoth alive or dead’ [lit. ‘that he would get Wulfnoth alive
or dead’]
Therefore, it should be concluded that the scribe or author fails to use the
agreement markers in the perfect because he does not feel that those parti-
ciples modify the object, not because he has forgotten the appropriate mor-
phemes, since he uses them in comparable constructions elsewhere.
The most striking departure from the state of affairs in the passive is that
perfects do not host plural agreement in this sample either (though see Section
4.4.2.2 above for an example from MS. E, 1086; cf. Kilpiö 2007: 332). In (83) and
(84), the objects are plural but the participles have no overt plural morphology
(acc.pl -e). The word order is also noteworthy.
(83) Þa sum siðe heafd-e se cyng hi

then some time had-sg that king them.3acc.pl
fore-be-gan mid ealre fyrde [1009]
pref-pref-gone with all army
‘Then on one occasion the king had intercepted them with all his army’
(84) Hi heafd-on þa ofer-gan East Engla .i. and East Seaxe .ii.
they had-pl then pref-gone East Angles and East Saxons
7 Middel Seaxe .iii. 7 Oxenafordscire .iiii. … [1011]
and Middle Saxons and Oxfordshire
‘They had then overrun: (i) East Anglia, (ii) Essex, (iii) Middlesex, (iv)
Oxfordshire …’
Even if the objects in (84) were all singular, there could still have been plural
agreement because of the list of conjoined place names. In a passive clause
below, (87), a conjunction of two singular nouns indeed triggers resolved plu-
ral agreement. The marking on East Engla is not clear, so the zero morphology
on the participle might be legitimate agreement with the nearest conjunct if
East Engla was treated as a neuter singular perhaps, or this could be a geni-
tive plural noun (characteristically terminating in -a), though there is hardly
any justification for having a genitive here. East Engla is far more likely to be
an accusative plural direct object, like those with the component Seaxe, and
agreement has simply broken down. The following example from Bosworth
and Toller, again from the Chronicle, including MS. E, suggests that both types
of name are actually grammatically plural, triggering plural verb agreement
(on plural names like Seaxe, see Prokosch 2009[1939]: 246):
(85) Her Middel-Seaxe (MS. E. Middal-Engla) onfeng-on ryhtne

here Middle Saxons (Middle Angles) received-pl right
geleafan
faith
‘In this year the Middle Saxons (Middle Angles) were converted to the
right faith.’ [Chr. 653, from http://bosworth.ff.cuni.cz/022812, accessed on
27 Oct 2016]
Whatever the correct analysis may be, it is important that there could have
been plural agreement in (84), even if all the enumerated objects were sin-
gular. By contrast, zero morphology is legitimate in only one transitive have-
perfect, (86). Though legitimate zero morphology is widespread in passives,
plural agreement is still healthy there, so the loss of plural agreement in the
perfect cannot be attributed to this lone instance of expected lack of overt
marking.
(86) ac þonne hi mæst to yfele ge-don hæfdon. þonne nam

but when they most to evil pref-done had then took
man grið. 7 frið wið hi [1011]
one truce and peace with them
‘but when they had done most to our injury, peace and truce were made
with them’
There is arguably no real accusative object here, at least not a noun; mæst
might be taken to be a genderless object (or an adverb serving as adjunct, or
a determiner used independently in the neuter singular), so agreement has
defaulted legitimately anyway. As an aside, an alternative modern translation
might involve did in place of had done, laying no emphasis on the anteriority
and acting as a reminder that the perfect has not yet attained its modern state.
As before, passives show robust agreement in the plural and zero in the sin-
gular. The examples in (87) and (88) below demonstrate that zero morphology
and overt agreement can coexist happily side by side in the same sentence.
Moreover, they confirm that overt marking survives until quite late, as late as
1012.
136 chapter 4
(87) 7 Wulfgeat-e wæs eall his are

& Wulfgeat(m)-dat.sg was[sg] all his area(m)[nom.sg]
of-ge-num-en. 7 Wulfeah 7
pref-pref-take-pptcp & Wulfheah &
Ufegeat wær-on
Ufegeat [conjoined m.nom.pl NP] were-pl
a-blend-e [1006]
pref-blinded-m.nom.pl
‘Wulfgeat was deprived of all his property, and Wulfheah and Ufegeat
were blinded’
(88) Ða þet gafol ge-les-t

when that.nt.nom.sg tribute(nt)[nom.sg] pref-pay-pptcp
wæs. 7 þa frið að-as
was[sg] and that.nom.pl peace oath(m)-nom.pl
ge-swor-en-e [1012]
pref-swear-pptcp-m.nom.pl
‘When that tribute was paid and the oaths of peace [were] sworn’
It should be reiterated that, in the passive, agreement (again!) only fails in the
feminine singular (expected nom.f.sg -u), never in the plural:
(89) Her on ðissum geare wæs Bæbbanburh

here on this year was[sg] Bamburgh(f)[nom.sg]
to-broc-on 7 mycel herehuðe þær
pref-break-pptcp and much booty(f)[nom.sg] there
ge-num-en [993]30
pref-take-pptcp
‘In this year Bamburgh was sacked and much booty was captured there’
The ‘wrong’ spelling of tobrocon showcases the confusion of -on (normally

preterite plural) and -en (strong past participle) which is characteristic of Late
Old English. Interestingly, this time there is no overt agreement on the attribu-
tive modifier mycel either (no f.nom.sg -u in (89) because -u is rare?), though
overt morphology resurfaces in the entry from the same year:
30 Unlike some of those above, this example is treated as a dynamic passive because of the
meaning contribution of the PP adjunct of time.
(90) Þa gegaderode man swiðe mycel-e fyrd-e [993]

then gathered one very large-f.acc.sg army(f)-acc.sg
‘Then a very large English army was collected’
As mentioned already, cases of legitimate zero morphology, with masculine

and neuter singular subjects, are numerous (a total of 13), but have neverthe-
less not led to loss of agreement in the plural. Zero morphology is also expect-
ed if the subject has no gender or number features, as is the case with clausal
subjects:
(91) forþon þam cyng-e wæs

for.that that.m.dat.sg king(m)-dat.sg was[sg]
ge-cyd-d þæt hi woldon hine besyrewian
pref-make.known-pptcp that they would him deceive
æt his life. 7 syððan ealle his witan [1002]
at his life and since all his councillors
‘because the king had been informed that they would treacherously
deprive him, and then all his councillors, of life’
The same situation is replicated with weorðan-passives: overt participial

agreement in the plural, (92), agreement missing when the subject is femi-
nine singular, (93), with zero anyway for masculine singular subjects, (94) (see
Appendix 2.231 and Table 4).
(92) 7 þær wurd-on ge-hors-od-e [1010]

and there became-pl pref-horse-pptcp-m.nom.pl
[the subject is þa Dæniscan ‘the Danes’ from the preceding clause] ‘and
there they [the Danes] were provided with horses’32
(93) Ðær wærð East Engla folces seo

there became[sg] of.East Angles of.folk that.f.nom.sg
yld of-slag-en [1004]
age(f)[nom.sg] pref-slay-pptcp
‘There the flower of the East Anglian people was killed’
32 As already noted above, some researchers argue that analogous forms, such as wearð ges-
icclod ‘became sickened’, could be potentially adjectival (see Mitchell 1985: §735, and ex-
ample (97) below).
138 chapter 4
(94) þonne wearð þær æfre þurh sum þing

then became[sg] there ever through some thing
fleam a-stih-t [998]
flight(m)[nom.sg] pref-instigate-pptcp
‘a flight was always instigated by some means’
Things are essentially the same with examples judged (admittedly, sometimes
arbitrarily) to be ambiguous between passive and stative. (95) once more
shows plural agreement (on druncene) and expected zero morphology (on sin-
gular gebroht) side by side in the same sentence. Therefore, it can be concluded
that the agreement system which operated in the previous annals is still intact
at this late date, ‘oblivious’ to any influence coming from legitimate zero mark-
ing or to other large-scale changes in the morphology of Late Old English.
(95) Wær-on hi eac swyðe drunc-en-e.

were-pl they.3nom.pl also very drink-pptcp-m.nom.pl
forþam þær wæs ge-broh-t win
for.that there was[sg] pref-bring-pptcp wine(nt)[nom.sg]
suðan [1012]
south
‘They were also very drunk, for wine from the south had been brought
there’33
Adjectival status for druncene is suggested by the modifier swiðe and by the
argument structure of the verb – (95) does not come from ‘X drank them’. The
same reasoning applies to the combination swiðe gehindred ‘much hindered’
from another example for 1003 (see Appendix 2.2).34 Swiðe likewise points
to adjectival status for the participle in the only combination with weorðan
which was judged to be ambiguous:
33 The adverbial clause of reason which contains gebroht is treated as a bona fide passive. It
remains to determine whether the presence or absence of the perfective prefix ge- on the
participles is significant.
(96) Þa on þone Sæternesdæg wearð swiðe

then on that Saturday became[sg] very
ge-stir-ed se here ongean
pref-excite-pptcp that.m.nom.sg army(m)[nom.sg] against
þone biscop [1012]
that bishop
‘Then on the Saturday the army became greatly incensed against the
bishop’
Due to another type of cue, this time lexicographic, the example in (97) below
was deemed ambiguously stative too. Bosworth and Toller’s dictionary entry
gives both a dynamic and a stative translation: ge-siclian (past -ode, past
participle -od) is defined either as ‘to be taken sick or ill’ (dynamic), or ‘to be
infirm’ (stative) (http://bosworth.ff.cuni.cz/016005, accessed on 20 Oct 2016).35
(97) 7 cweð þet he ge-sicl-od

and said that he.3m.nom.sg pref-sicken-pptcp
wær-e [1003]
were-sbjv.sg
‘and said that he was taken ill’
The clause in (98), by contrast, was assigned to this category a little more
arbitrarily:
(98) Swa hit ge-cweð-en is [1003]

as it.3nt.nom.sg pref-say-pptcp be.pres.3sg
‘As the saying goes’
Finally, it remains to consider ambiguity involving be-perfects vs. be-passives

or statives, which seems to be more widespread in this portion of text. No un-
ambiguous be-perfects were found in this sample, but there was one instance
of have used with an unquestionably intransitive verb:
35 Cf. the note above regarding the treatment of wearð gesicclod ‘became sickened’ as adjec-
tival in Mitchell (1985: §735).
140 chapter 4
(99) Þa hi swa feor ge-gan hæfd-on swa

when they.3nom.pl as far pref-gone had-pl as
hi þa wold-on [1010]
they.3nom.pl then would-pl
‘When they had thus gone as far as they pleased’
Readers will remember that there is already an intransitive have-perfect for

the year 896 – gewicod hæfdon, in (55), so this is not a precedent. A similar dis-
placement was noted in comparing MS. A to MS. D in the previous section –
see example (64). Regarding (99), attention should also be given to the
quasi-object-like adverbial expression (or rather, locative complement) swa
feor, in the light of the discussion in Chapter 3, Section 3.4.1. This might have
favoured the choice of have. Crucially, the use of have as opposed to be does
not serve to dispel ambiguity and might have been additionally influenced by
the use of have with transitive prefixed derivatives of gan elsewhere in these
entries, as shown in (83) and (84) above. Future work ought to ascertain what,
if anything, have might add to the meaning that be does not, or whether both of
them are simply sufficiently generic and bleached to a point when it does not
matter which one is deployed.
Of the ambiguous be-perfects, two have feminine singular subjects with
no overt agreement (nom.f.sg -u missing from gesomnod in (100) and from
geendod in (102)), while another three have masculine singular subjects, so
zero morphology is legitimate for them. (100), for instance, is treated as an
ambiguous passive or perfect, since the verb ge-somnian, apart from being
transitive – ‘to gather, collect; join, unite’, has an intransitive use too. Bosworth
and Toller define intransitive ge-samnian/-somnian (past -ode/-ade/-ede,
past participle -od/-ad/-ed), as follows: ‘to collect, come together; congrĕgāri,
convĕnīre’, providing the example in (101). Incidentally, although the Latin
original in (101) is in the perfect passive, the Old English version has the pres-
ent active. Another thing to highlight are the similarities between the Latin
perfect passive in (101) and its possible OE counterpart in (100) – both are con-
structed with a form of be and a past participle.
(100) Þa wæs þær fyrd ge-somn-od

then was[sg] there army(f)[nom.sg] pref-collect-pptcp
æt Cynetan [1006]
at Kennet
‘The English army was then gathered at the Kennet’
(101) congrĕgāti sunt [Latin]

Hi gesamniaþ [Old English]
‘They gather/come together’
[from Bosworth and Toller, http://bosworth.ff.cuni.cz/015813,
accessed on 28 Oct 2016, translation mine]
It is certainly conceivable that (100) is a passive resultative construction in

the preterite, semantically equivalent to active ‘Then the English army gath-
ered at …’/‘Then someone gathered the English army at…’ – cf. adverbial
þa ‘then’. Some might even go as far as saying that the pluperfect category
developed only much later, as an analogical formation after the complete
grammaticalisation of the present perfect, so (100) is then simply a preterite
passive (M. Kovatcheva, p.c.). And yet, a (proto-)perfect reading cannot be
ruled out with absolute certainty. A similar ambiguity exists with ‘be ended’
in (102) – has the campaign ended or did someone put an end to it? Mitchell
(1985: §749, §752) also notes the double-membership of ge-endian as a transi-
tive and an intransitive verb, while Los (2015: 74–75) talks about analogous wæs
begunnen ‘was begun’.
(102) Þa ðeos scip-fyrd ðus

when this.nom.f.sg ship-levy(f)[nom.sg] thus
ge-end-od wæs. [1009]
pref-end-pptcp was[sg]
‘When this ship-levy had ended thus’
The first of the examples with ‘licensed’ zero morphology is especially curious:
(103) 7 his scipu wendon ut abuton Legceastre and sceoldan

& his ships went out round Chester and should
cumon ongean hine, ac hi ne mihton. […] 7
come against him but they not might &
seo unfriðflot-a wæs þæs sumeres
that enemy.fleet(m)-nom.sg was[sg] that summer
ge-wend to Ricardes rice. [1000]
pref-turned to Richard’s kingdom
‘and his ships went out round Chester and should have come to meet
him, but they could not. […] And the enemy fleet had gone to Richard’s
kingdom that summer’
142 chapter 4
Unfriðflota is a masculine noun, despite the seemingly feminine demonstra-

tive (cf. Bosworth and Toller, http://bosworth.ff.cuni.cz/032979, accessed on
19 Oct 2016). The confusion of masculine se and feminine seo probably stems
from the unstressed nature of the demonstrative and the late date of this text
(se cyng and seo cyng ‘that king’ are both employed almost side by side in the
annal for 1006; cf. Irvine 2004: cxlv).36 This was a time when unstressed vowels
were beginning to be mixed up, as demonstrated elsewhere in the sample and
noted above, including in this very sentence – sceold-an should be sceold-on
(preterite plural, not an infinitive), and conversely, cum-on should be cum-an,
an infinitive in -an after the pre-modal (see Irvine 2004: cli). Crucially for pres-
ent purposes, Bosworth and Toller confirm that the verb ge-wendan ‘to turn’
has both transitive and intransitive uses (http://bosworth.ff.cuni.cz/016779,
accessed on 19 Oct 2016; Irvine 2004: clix further illustrates reflexive uses of
wendan). Mitchell (1985: §735, §752, §775) also notes that wurdon awende ‘is
genuinely ambiguous and can be either active or passive, because awendan
can be either intransitive or transitive’. This verb appears in another sentence
from my material:
(104) Þa wæs se cyng ge-wend

then was[sg] that.m.nom.sg king(m)[nom.sg] pref-turned
ofer Temese into Scrobbesbyrigscire [1006]
over Thames into Shropshire
‘Then the king had gone across the Thames, into Shropshire’
Indeed, (104) probably features the intransitive perfect use of this verb, as
suggested by Whitelock et al.’s (1961) translation and since this is an example
Bosworth and Toller cite to illustrate intransitive ge-wendan. The usage of wen-
don ut in (103) also favours a perfect interpretation, but the potential for am-
biguity is still there nevertheless (despite the undeniable disambiguation in
context). The final sentence, coupled with the other ones adduced here, might
suggest that, after all, OE morphology did not keep ambiguity between perfect
and passive readings at bay as much as has been claimed.
36 Alternatively, it could be argued that unfriðflota got transferred to the feminine gender
class because of influence from semantically related feminines like scipfyrd. This explana-
tion is less attractive.
(105) þa toferde se here wide swa he ær

then dispersed that army wide as he.3m.nom.sg ere
ge-gader-od wæs [1012]
pref-gather-pptcp was[sg]
‘the Danish army then dispersed as widely as it had been collected’
In terms of percentages, ambiguity in my OE samples has ratcheted up steadi-

ly from 3.3%, (then 0%), through 4.8%, to 12.2% of all recorded auxiliary oc-
currences (though it should be kept in mind that the numbers are still rather
small). At the end of the day, if there are so many ambiguous cases already by
the beginning of the eleventh century, with hardly any clear be-perfects, why
was the be-perfect not eliminated as early as that stage?37 As the current sam-
ple demonstrates, the switchover has certainly begun – there is gegan hæfdon
in the entry for 1010, (99), as well as similar cases in the earlier samples too. So
it could perhaps be said that mounting ambiguity has led to the displacement
of be by have. However, why did be last so many centuries after this period?
Could it be just down to conservative inertia acting as a deterrent and/or slow
geographical diffusion from area to area? What is more, had gone does not re-
ally dispel any potential ambiguity between perfect and passive, and be is used
with gone as late as Dickens, and indeed to this very day. In a similar vein, the
Middle English text of Secunda Pastorum in the next chapter shows an unusu-
ally high degree of displacement which, one might argue, is likewise related to
the elimination of ambiguity. There, however, the level of ambiguity is compar-
atively low (1.9%), which at the same time goes along with hardly any remain-
ing be-perfects. The correlation between ambiguity and loss of be is therefore
problematic, because now we see the opposite: a good deal of ambiguity with
hardly any non-ambiguous be-perfects. This makes it extremely precarious to
attempt to establish any directionality and causation – is the loss of be-perfects
supposed to go with rampant or curbed ambiguity?
37 The absence of unequivocal be-perfects from this stretch of MS. E is probably just a fluke.
It would be illuminating to draw a parallel to ‘proto’-progressives. More plentiful in the
early entries, progressives were absent from the previous sample. However, they have now
made a comeback (there are three of them in the annals under investigation). So their
absence from the earlier annals must be a coincidence and can scarcely be taken to indi-
cate that the progressive had disappeared. The same applies to the observed fluctuations
between bona fide passives and competing constructions with man/mon ‘one’. Thus it is
not advisable to make too much of the non-occurrence of clear be-perfects here.
144 chapter 4
The explanation might actually lie elsewhere. Van Gelderen (2018: 97) notes
that ‘the language of the Peterborough Chronicle is traditionally seen as repre-
senting the change from Old to Middle English and is often seen as represen-
tative of changes contact with Scandinavian brought’ – see for instance (86),
where the preposition wið is already used with its Old-Norse-influenced sense
of ‘with’, rather than the earlier ‘against’ (cf. (96) and the earlier annals). It is
insightful to note that both the Peterborough Chronicle and Secunda Pastorum
are of Northern provenance, as is MS. D discussed in the previous section. All
of them exhibit displacement of perfect be by have, so the advanced loss of
perfect be might in fact be associated with contact with Old Norse speakers
in the north of England, as suggested in Chapter 3, Section 3.5.5. What these
Chronicle entries reveal might be the earliest seeds of this development. In ad-
dition to documenting the Viking raids and the ensuing clashes and atrocities,
this section of the Chronicle could also be a testament to the Viking influence
on the incipient retreat of be from the perfect. This is probably a contact-
induced change whereby speakers started to rely on the default auxiliary in
bilingual interactions, reinforced by the greater frequency of have-perfects
in the language of the Scandinavian invaders. This process is reflected in the
consistently low numbers of perfect be in my northern samples, with reliance
placed on have instead, and will be further traced in the Middle English survey
to follow in the next chapter.
4.5 Conclusion
Examining in detail a number of entries from the Anglo-Saxon Chronicle and

its continuation known as the Peterborough Chronicle (up to 1012), I have tried
to trace the use of the perfect in the light of morphosyntactic agreement, am-
biguity and semantic interpretation. The samples have been grouped roughly
according to the years they report on and according to changes of hand, in an
attempt to get as close as possible to a true picture of idiolectal conformity
and representativeness. The material reaffirms that the grammaticalisation of
the perfect is incomplete, judging (among other things) by the free word order
coupled with the impossibility to rule out a stative interpretation for most of
the examples (the latter is of course more solid as evidence). Likewise, it was
often difficult to tell if a sentence was a dynamic passive or a copular stative.
At the same time, the annals illustrate how these budding constructions are
starting to slowly come into their own.
One important lesson from this and from the previous chapter is that overt
agreement morphology cannot be used as a reliable correlate of semantic in-
terpretation (stative or dynamic) – neither in the perfect nor in the passive.
For instance, the stative meaning of possession does not rule out absence of
overt agreement, and conversely, a dynamic perfect reading does not rule out
presence of agreement. And yet, there is a robust tendency throughout, even in
the late entries already on the cusp between Old and Middle English, for agree-
ment to appear more in non-grammaticalised adjectival constructions, often
side by side with non-agreeing perfects. I therefore believe that, rather than
acting as a trigger for grammaticalisation and reanalysis, disposing of agree-
ment is an outward sign and a consequence of ongoing grammaticalisation.
Scribes do not appear to have simply forgotten the requisite endings – instead,
they use them surprisingly consistently in constructions which are perceived
as genuine modifiers of nouns and drop them in perfects which they arguably
regard as more verbal in character.
One of the objectives was to test empirically the role of legitimate zero ex-
ponence as a bridging context which supposedly enabled reanalysis and sub-
sequent loss of agreement across the board. Both have- and be-perfects were
examined and comparison was sought to passive or potentially stative con-
structions with beon ‘be’ and weorðan ‘become’. The material was divided into
three categories – examples where overt agreement is present, examples where
overt agreement is missing, examples where zero marking is expected anyway.
I confirm that there is a substantial amount of expected zero morphology
in the perfect, as well as in beon- and weorðan-passives. However, while the
prevalence of expected zero exponence may appear to have led to an almost
complete loss of agreement in the perfect, particularly evident in the plural,
this is certainly not the case in the passive. In both types of passive, with beon
and weorðan, zero morphology is used in the singular (legitimately with the
masculine and neuter, not so with the feminine), whereas -e appears consis-
tently in the plural (unlike the perfect). These findings call into question the
claims that ambiguous cases of expected zero morphology were responsible
for losing agreement across the board, and point to different rates at which the
grammaticalisation of these constructions proceeded.
As for the ambivalence between be-perfects and passives, a characteristic
of OE is that its verbal derivational morphology can often (but not always)
disambiguate a given verb as transitive or intransitive, unlike later periods,
where context and semantics assume a more important role. The rates of am-
biguity between perfect, passive and stative or between passive and stative are
for the most part comparable to those from Middle and Modern English (see
Chapter 5), though a tendency for the former to gradually creep up with the
gradual spread of the perfect makes itself felt. These issues will be addressed
again in the next chapter, as will be the clues for potential Scandinavian in-
fluence on the loss of perfect be discernible in the northern recension of the
Chronicle.
chapter 5
Further Development of the Perfect Based on a

Selection of Texts – from Middle into Modern
English
Picking up where the previous chapter left off, Chapter 5 charts the further
development of the perfect from Middle into Modern English. As agreement
on the participle is no longer at issue, the focus is on the competition between
the two auxiliaries and the role ambiguity had to play in the displacement of
be, as compared to some of the other explanations outlined in Chapters 3 and
4. The material comes from some of Chaucer’s comic tales, a 15th-century mys-
tery play, comedy from the Renaissance and the Restoration period, as well as
the correspondence of the Paston family and Samuel Pepys. The chapter con-
cludes with an epilogue featuring commentary on Victorian English, and the
next chapter draws some parallels to German.
5.1 Introduction
My main concern here will be to concentrate on the ambivalence and the inter-
actions of the various auxiliaries, also trying to find more evidence for a contact
scenario. The diverse selection of passages provides a series of vivid snapshots
of how the perfect construction is gradually edging towards the state at which
it finds itself today, in addition to being an eloquent testimony to the multiple
waves of grammaticalisation discussed in Chapter 3. Section 5.2 examines ma-
terial from the 14th century (Chaucer), Section 5.3 is about a 15th-century play
from the north which might confirm the influence of Scandinavian, Section
5.4 tackles a 16th-century play by Shakespeare, while Section 5.5 analyses
Restoration comedy from the 17th century. To bolster the database, in Section
5.6 I supply data from the Late Middle English correspondence of the Paston
family, while in Section 5.7 I do the same for the 17th-century correspondence
of Samuel Pepys. The Epilogue (5.8) adduces material as late as Dickens in
order to consolidate the discussion of English and to convey one final time
the message that the functionalist story from earlier chapters crumbles when
subjected to close scrutiny. The changes involved must have been much more
mechanistic and/or contact-driven than driven by functional considerations.

The perfect from Middle into Modern English 147
As in the previous chapter, each section is based on a representative work

which was selected before extracting the data in accordance with the criteria
from Chapter 4, Sections 4.2 and 4.3. The texts are distributed roughly at one-
century intervals and each section opens with some relevant background be-
fore proceeding to the linguistic analysis of the distribution of the auxiliaries
and their grammatical behaviour.
5.2 The Fourteenth Century: Chaucer

Starting with ‘classical’ Middle English, I have decided to examine Chaucer’s
Miller’s and Reeve’s Prologues and Tales, neither of which is included in the
Helsinki Corpus. The choice of tales has been prompted mainly by the fact that
they belong to the genre of fabliau, or comic tale, and are therefore expected
to be closer to the language spoken around Chaucer than other types of text
might be, in line with the generalisations drawn in Chapter 4 (cf. Crystal 2004:
177, 241). The most authoritative edition of Chaucer’s work available to me is
The Riverside Chaucer, which takes a scholarly approach and records different
manuscript readings and critical interpretations (see Benson 1987a: ix, 1987c).
Page and line references here are to the third edition. The different parts of the
Canterbury Tales can roughly be dated to the 1380s and the 1390s (see Benson
1987a: 3, 1987b: xxix). Discussion of the perfect in Chaucer’s output more wide-
ly can be found in Gräf (1888).
As stated above, it was deemed that the fabliau genre is a good enough re-
flection of and approximation to colloquial language. This view finds support
in what the critics have to say. To quote Larry Benson (1987a: 4):
Much of this lifelike quality is due to Chaucer’s eye for authenticating

detail and his ear for the rhythms of colloquial speech, but, even more,
it is the product of his dramatic method. Chaucer pretends merely to re-
port what he sees and hears; he is content to let his characters speak for
themselves, and when he does tell a tale in his own person, he does so as
their equal, merely another teller.
The tales are often quite suited to their tellers and they seem like dramatic
soliloquies, which makes them similar to comedy on the stage. Benson (1987a:
7) sums up the fabliau style as simple, vigorous and straightforward; fabliaux
deal with the present time, relying on realistic settings in familiar places; the
148 chapter 5
characters are ordinary people such as tradesmen, peasants, priests, students,

and restless wives; the plots normally revolve around tricks and ruses. This can
ensure that they are linguistically progressive rather than conservative (cf. Gräf
1888: 38–39).

The size of the sample is roughly 9,857 words. As in the previous chapter, I was
not interested in the full-verb uses of be and have and they have therefore been
omitted from the material collected (unless they appear in the passive or the
perfect as main verbs or unless there is ambiguity with a passive/perfect). By
far the largest group of examples includes the use of auxiliary have to form the
perfect of prototypical transitive verbs. There were sixty such cases, such as (1),
which clearly marks anteriority.
table 1 Perfect, passive and modal auxiliaries in the Miller’s and Reeve’s Prologues and
Tales
Number % (of all auxiliary

occurrences recorded)
Have-perfects with transitive verbs & with the 60 35.5%

copula
Have-perfects with intransitive verbs 14 8.3%
Modal have 1 0.6%
Be-perfects with (mutative) intransitives 19 11.2%
Be-perfects with transitives 1 0.6%
Be-passives 47 27.8%
Ambiguity between passive and a stative 20 11.8%
copular clause
Ambiguity between be-perfect and passive/ 6 3.6%
stative
Missing auxiliary in a non-coordinated context 1 0.6%
(1) Whan that the Knyght had thus his tale ytoold [3109]
Here I also included typically transitive verbs with implied objects which were
not overtly expressed, (2),1 as well as verbs with prepositional complements
(a.k.a. prepositional objects), (3), and the copula be, (4).
(2) as I have told biforn [3302]
(3) As he had kiked on [‘gazed at’] the newe moone [3445]
(4) if she hadde been a mous [3346]
As noted by Mustanoja (1960: 501), the verb be has always formed its pres-
ent and past perfect by means of the auxiliary have (cf. Mitchell 1985: §1099
and Macleod 2012: 120–122 for OE; an example involving heafde gebeon from
the Anglo-Saxon Chronicle entry for 1096 is provided in the Dictionary of Old
English, at http://tapor.library.utoronto.ca/doe, accessed on 27 Feb 2017).
A question arises whether instances in which the auxiliary and the main
verb are separated might not be carriers of stative meaning, rather than being
genuine perfects. It was concluded that they were simply ME perfects (though
see discussion and caveats in Chapter 3, esp. Sections 3.1.2, 3.1.4, and Chapter
4, esp. Section 4.4.2.1). The free word order in the perfect, including different
permutations of the relative positions of the auxiliary, the main verb and the
object(s), as in (5)–(8), is most likely due to the poetic form and the less strict
word-order constraints during this period, less strict than ModE that is (cf.
Kilpiö 2007: 335–336 for the flexibility of the word order in perfects from ME
prose texts too). Moreover, the same flexibility is evident in the passive, (9),
where adjacency of the finite and non-finite verb is certainly not a must.
(5) He auntred hym [‘took a risk’], and has his nedes sped [‘accomplished his
purpose’] [4205]
(6) This joly lyf han [‘have’] thise two clerkes lad
Til that the thridde cok bigan to synge [4232–4233]
1 See Nakamura (1987: 31–32), who does the same.

150 chapter 5
(7) Thus have I quyt [‘repaid’] the Millere in my tale [4324]
(8) I have thee broght a ryng [3794]
(9) Ther was hir whete and eek [‘also’] hir malt ygrounde [3991]
In general, there are now much clearer cues of anteriority such as til ‘till’, (6),
or after, (10), compared with the ubiquity of clauses with þa ‘then/when’ in the
Anglo-Saxon Chronicle, which do not lay particular emphasis on anteriority. By
Middle English, the semantics of the perfect has edged closer to what we cur-
rently have in Modern English. The more abstract meaning of modality/coun-
terfactuality, illustrated in (12)–(13), further testifies to even more advanced
grammaticalisation.
A substantial number of intransitives – 14 – formed their perfect with have.
Those included verbs like crow, itch, run, misgo, creep, wake (‘stay awake’). Gräf
(1888: 69) confirms that misgo, for instance, is only used with have in Chaucer’s
work and I return to the likes of wake below. The following are typical exam-
ples of intransitive have-perfects:
(10) A litel after cokkes hadde ycrowe [3358]
(11) My mouth hath icched [‘itched’] al this longe day [3682]
It is striking that the intransitives that use be and have are normally in com-
plementary distribution. This is consistent with the constructionist view (see
Chapters 2, 3 and 4), and a similar observation regarding other aspects of gram-
mar is also made by Danchev (1969). Each verb prefers one or the other aux-
iliary, but does not tend to occur with both. Run, for instance, prefers have,
whereas go (but not misgo ‘go astray’) generally takes be. In the only case where
go combines with have in my sample, have seems to indicate a hypothetical
context, suggesting that the action was not accomplished (see discussion in
Chapter 3):
(12) I hadde almoost goon to the clerkes bed. [4219]
In at least one example, (13), have was probably used with an intransitive due
to the non-finite (and hypothetical/counter-factual?) environment (the read-
er is again referred to the preceding discussion, esp. Chapter 3, Sections 3.3
and 3.4.1).
(13) He wende have cropen [‘crept’] by his felawe John [4259]
It is natural to wonder whether the high frequency of have with intransitives

serves the purpose of dispelling the ambiguity between be-perfects and be-
passives. Most of the time, substituting be for have would not have produced
any ambiguity. It makes little difference in (14)–(16), as demonstrated by the
bracketed alternatives invented by the author:
(14) A litel after cokkes hadde ycrowe [or ‘were(n) ycrowe’?] [3358]
(15) Thise sely [‘innocent, hapless, foolish’] clerkes han [‘have’] ful faste yronne
[or ‘are(n)/be(e)n yronne’]
Toward the fen, bothe Aleyn and eek [‘also’] John. [4090–4091]
(16) I hadde almoost mysgoon [‘gone astray’] [or ‘was … mysgoon’]; [4218]
Be would perhaps be inappropriate in a non-mutative context anyway, as in

(17) below or with the verb ‘crow’ in (14) above (cf. Los 2015: 75–76).2
(17) My mouth hath icched [‘itched’] al this longe day [3682]
Nevertheless, if the subject of run is inanimate, misunderstanding might arise,

as in (18):
(18) And ever sithe hath so the tappe yronne [‘is… yronne’ potentially passive?],
Til that almoost al empty is the tonne [‘barrel, cask’] [3893–3894]
What might have disambiguated this as non-passive is the syntactic and se-
mantic context, with ever sithe ‘ever since’ favouring a perfect interpretation.
In any case, out of 14 examples of have used with intransitives, substituting
be for have would have created potential misunderstanding in only three or
four at most. Apart from run in (18), such might be the case with had waked
2 Judging by the OED and the Online Middle English Dictionary entries, the verb for ‘crow’ only
takes have in the perfect (https://quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=
31869142&egdisplay=compact&egs=31875371&egs=31876210, accessed on 10 Jan 2017). By con-
trast, there are attested combinations of ‘be’ and ‘run’: see Gräf (1888: 69) for examples from
Chaucer, as well as phrases like ben runnen ‘to have run or gone’ from the Middle English
Dictionary (https://quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=159568672&e
gdisplay=compact&egs=159590540, accessd on 10 Jan 2017).
152 chapter 5
‘had stayed awake’ in (19) – if it was not for the adjunct al nyght ‘all night’, was
waked could conceivably have been interpreted as the ambivalent ‘was awo-
ken/woken up’ (Rydén and Brorstöm 1987: 49 provide later examples of be- and
have-perfects with related (a)wake(n)). However, the adjunct and the context
would have supplied ample disambiguation anyway and there are moreover
clear signs that had here does not serve any disambiguation purposes at all.
Contrary to what one would expect if disambiguation was paramount, Gräf
(1888: 65) notes that when ME wake means ‘awake, wake up’, it combines ex-
clusively with be in Chaucer’s oeuvre to form its perfect (But whan that he was
waked out of his sleep), while it takes have in the non-mutative sense of ‘stay
awake’, as in (19).
(19) For she was falle aslepe a lite wight [‘a short time’]
With John the clerk, that waked hadde al nyght [4283–4284]
A third context with potential for confusion might involve the verb for ‘work’,
as in (20).
(20) Aleyn wax wery in the dawenynge,

For he had swonken [‘worked’] al the longe nyght [4234–4235]
In conclusion, it appears that have is not predominantly used with intransitives

in order to prevent potential misunderstandings due to confusing grammatical
constructions. It is plausible to speculate that, since have could already take on
the function of forming the perfect of intransitives anyway, it displaced be sim-
ply because it had this ‘universal applicability’ or ‘meaning potential’ to denote
anteriority, not because it was in any sense ‘needed’ in order to disambiguate.
There is also one lonely instance of what seems to be an incipient modal
use of have with a to-infinitive expressing purpose (cf. discussion in Chapter 3):
(21) I have therwith to done [3777]
Without many surprises, mutative intransitives in be-perfect constructions

included fall, go, enter, rise, pass, come, grow:3
3 It is interesting to observe that Chaucer’s representation of a Northern dialect features have-
as well as be-perfects with intransitives, among the latter (i) and (ii). This state of affairs
should be compared to the texts of actual Northern provenance – the Peterborough Chronicle
and Secunda Pastorum. Inconsistencies in Chaucer’s use of Northern dialectal forms in this
tale are noted in Crystal (2004: 165–167).
(i) And forthy is I come, and eek [‘also’] Alayn [4031]
(ii) What, whilk [‘which’, Northern dialect] way is he geen [‘gone’, Northern dialect]? [4078]
(22) But sith that he was fallen in the snare [3231]
Apart from fall with a spatial sense where it indicates motion, it turns up in
a copular sense with the be-perfect too, as in (23) (cf. Chapter 2, Section 2.2,
for remarks on pathways of pragmatic inferencing during grammaticalisation
which lead to greater abstraction).
(23) For she was falle aslepe a lite wight [‘a short time’]
With John the clerk, that waked hadde al nyght [4283–4284]
As suggested by the time adjunct of duration a lite wight, by the neighbour-

ing intransitive have-perfect, and the possible sequential relations between
the two events in (23), was falle(n) in (22) and (23) might simply represent a
preterite be with a participial adjective/complex predicative unit, while the
‘real’ pluperfect would have been had(de) falle(n) (also refer to Chapter 3,
Section 3.1.4, as well as the discussion below, for the interchangeability of per-
fects and preterites in Middle English). And yet, sequential coordinations like
In an instant she was gone, and he had shut the door are found as late as Dickens
(1853, Bleak House, Chapter LIX), where both types of perfect appear to be on
a par. It is thus equally plausible that both was falle and waked hadde are plu-
perfects used to talk about events preceding an implied temporal reference
point of the narrative, i.e. with anteriority and result as grammaticalised infer-
ences (see Chapter 2, Section 2.2). Likewise, example (24) might potentially be
seen as copular and stative, but that might be giving it too modern a reading.
Therefore it was also put down as a bona fide be-perfect.
(24) Whan that the grete shour is goon away [3574]
Similarly, (25) was interpreted as a clear be-perfect, not a passive (cf. Gräf 1888:
76). What disambiguates it is the temporal subordinate clause (with syn ‘since’)
and the adverb meaning ‘hence’, which provide a typical perfect context, as
confirmed in (26) and in the earlier discussion.
(25) As many a yeer as it is passed henne [‘hence, away’]

Syn [‘since’] that my tappe of lif [‘life’] bigan to renne [3889–3890]
(26) And ever sithe hath so the tappe yronne,

Til that almoost al empty is the tonne [3893–3894]
(27) is also clearly not a transitive grow, so this copular use of grow can form
nothing but a be-perfect (or act as an adjective, cf. Gräf 1888: 64):
154 chapter 5
(27) This wenche [‘girl’] thikke and wel ygrowen was [3973]
Interestingly, be is also licensed for the perfect of a verb that appears to be

used transitively, with hym serving as an object in (28). This is probably under
the influence of the root go, which is a prototypical mutative intransitive (now
prefixed with transitivising bi-; cf. Chapter 4 for OE). (28) demonstrates that
the story is not simply one of straightforward expansion of have from transi-
tives to intransitives. There were occasional incursions in the other direction,
and Chaucer seems to be fine with such a transitive be-perfect (cf. Chapter 3,
Section 3.3, as well as similar examples with ME tyde and passe in Gräf 1888:
62, 74, 76).
(28) Fro day to day this joly Absolon

So woweth hire that hym is wo [‘woe’] bigon [3371–3372]
Although there is a ME compound adjective wo-bigon ‘deeply grieved, dis-

tressed, sorrowful, despondent’ (see https://quod.lib.umich.edu/cgi/m/mec/
med-idx?type=byte&byte=14857611&egdisplay=compact&egs=14863637&egs=
14862294&egs=14870945, accessed on 10 Jan 2017), also in evidence in my sam-
ple, (28) must be a transitive be-perfect because of the accusative/‘objective’
case on hym.4
There were as many as 47 passives, which confirms Kilpiö’s (1997) conclu-
sion that when be served as an auxiliary, it was mostly used to form the pas-
sive (see Chapter 3). The decision whether something was a proper dynamic
passive or perhaps a stative copular construction was often an arbitrary one.
The reader can double-check the choices that were made by consulting
Appendix 3.5 The rule of thumb was to treat a construction as potentially
stative and copular if the participle could be replaced with a non-participial
4 There is reasonably solid evidence that (28) is perfect: firstly, have is also attested in this con-
text, as in (i); secondly, the adjectival construction goes with a nominative subject, as in (ii).
And yet, an impersonal adjectival construction with an accusative experiencer, creative word
play and departure from general norms cannot be ruled out a priori (Gräf 1888: 77 indeed
treats wo-bigon as mostly adjectival, including in (28)).
(i) Gyle haþ bi-gon hire so, heo graunteþ al his wille. [(c1390) Chaucer CT.ML.(Manly-Rickert),
https://quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=14857611&egdisplay
=compact&egs=14863637&egs=14862294&egs=14870945, accessed on 10 Jan 2017]
(ii) That sche was wo bego withal. [(c1395) Chaucer CT.WB.(Manly-Rickert), https://quod
.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=14857611&egdisplay=compact&
egs=14863637&egs=14862294&egs=14870945, accessed on 10 Jan 2017]
adjective or a noun (without changing the meaning), or when the participle

was coordinated with an adjective or a noun phrase (though there are dan-
gers associated with this test, as noted in Chapter 3). Thus, (29) was taken to
be a clear passive, standing for present-day ‘Our corn has been stolen’ (note
the non-use of the perfect passive in the original, as pointed out in Chapters 3
and 4; cf. Gräf 1888: 16ff.), since the two students’ corn has been stolen by the
thieving miller; in context, the sentence does not mean that the Cambridge
students are in possession of stolen, i.e. illegal, corn.
(29) Oure corn is stoln [4111, 4183]
Example (30) is, admittedly somewhat arbitrarily, regarded as a potentially

stative construction, as it is paraphrasable with an adjective, e.g. ‘be well aware’:
(30) Be wel avysed [‘well warned’] on that ilke nyght

That we ben entred into shippes bord [3584–3585]
(31) and (32) show coordination with an adjective and a noun phrase,
respectively:
(31) His hoote love was coold and al yqueynt [‘quenched’] [3754]
(32) For Symkyn wolde no wyf [‘wife’], as he sayde,

But [‘unless’] she were wel ynorissed [‘bred, raised, educated’] and a
mayde [‘virgin’] [3947–3948]
It is worth asking whether a perfect reading might not be present in the fol-
lowing type of construction, treated as ambiguously stative or passive, but not
perfect:
(33) This tale is doon [3854]
Crucially, (33) is different from, say, I’m done/finished, where the subject is an
animate agent and where the sense is ‘I have done/finished [whatever I had
to do/finish]’. This, however, is not the case with This tale is done, where *This
tale has done would be ungrammatical, or rather #This tale has done [some-
thing] is semantically/pragmatically anomalous (cf. however, This tale/game
is/has begun, to which I return below). Certainly, even at a conservative es-
timate, and even if other scholars dispute some of the choices made as to
category-assignment, the current sample demonstrates that there is much
156 chapter 5
more ambiguity between a passive and a stative reading than there is between
a passive and a be-perfect.
Before moving on to what was considered genuine ambiguities between
passive and be-perfect, several additional borderline cases should be men-
tioned. (34) is regarded as passive (or potentially stative) (cf. Gräf 1888: 63 for
similar intuitions):
(34) Thou art a fool, thy wit is overcome! [3135]
While thy wit has overcome [something] would be grammatically and semanti-
cally well-formed, it simply makes no sense in this context, where the person
referred to as a ‘fool’ has (ostensibly) ‘taken leave of their senses’. (35) was like-
wise treated as passive (though some scholars might suggest otherwise, e.g.
Gräf 1888: 72–73; also see the Middle English Dictionary entry, which indirectly
corroborates my analysis, https://quod.lib.umich.edu/cgi/m/mec/med-idx?ty
pe=byte&byte=46261898&egdisplay=compact&egs=46337024&egs=46277303,
accessed on 16 Jan 2018):
(35) Now are we dryve til hethyng [‘to contempt’] and til scorn. [4110]
The context indicates passive here, because the two students have been fooled
by the cunning miller; i.e. they do not get themselves into that mess; rather, it
is the miller that drives them to contempt and to scorn. In addition, though not
an argument-clincher, a parallel passive construction appears in the following
line (repeated from (29) above):
(36) Oure corn is stoln [4111]
The example in (37) was also somewhat problematic:
(37) The sely [‘innocent, foolish’] tonge may wel rynge and chymbe [‘chime’]
Of wrecchednesse [‘wretchedness’] that passed is ful yoore [‘long ago’].
[3896–3897]
Upon careful consideration, it was taken to belong to the group of be-perfects.

The key might be supplied by the phrase ful yore ‘a long time ago’. Crucially,
the perfect could be used with specific past time reference in Middle English,
where ‘narrative of past events can be told with the simple past tense, the his-
toric present, or the perfect’ (Burrow and Turville-Petre 2005: 46; cf. Chapter 3,
Section 3.1.4, Gräf 1888: 20ff.).6 Mustanoja (1960: 504) confirms that in Middle
English, the perfect could occur with adverbs of past time, as in adventures that
whilom han bifalle from the General Prologue to the Canterbury Tales. In (30)
above, the act of entering (ben entred ‘are/have entered’) took place on that ilke
nyght ‘on that same night’. The following analogous case from my corpus also
shows that an indisputable have-perfect (hath take) can certainly be used with
specific past time reference, as indicated by the ‘when’ temporal subordinate
clause:
(38) And whan the millere saugh [‘saw’] that they were gon
He half a busshel of hir flour hath take [4092–4093]
Be-perfects were compatible with specific past time reference as late as the
17th and 18th centuries, as shown in the following example from Pepys’s Diary
(cf. Rydén and Brorström 1987: 162 for late-18th-century attestations).7
(39) This day I have news ... that the fleet is sailed yesterday from Harwich
to the coast of Holland, to see what the Duch will do. [Samuel Pepys, Diary,
22 April 1665, quoted in Nakamura (1987: 40)]
The remaining ambiguous cases, which could be perfects or passives, are then
relatively few – initially, there appear to be six of them, as noted in Table 1. The
following are the most convincing ones:
(40) For trewely the game is wel bigonne. [3117]
(41) Heere is ended the Reves Tale.
(42) Myn herte [‘heart’] is broken; help! [4289]
6 It is unlikely that Chaucer has opted for a historic-present-tense passive to report a past event
in (37) due to what he uses in the surrounding context, as in the following preterite passives,
just a few lines above and below the example in (37):
(i) For sikerly, whan I was bore, anon … [3891]
(ii) This dronke Millere hath ytoold us heer
How that bigyled was a carpenteer [3913–3914]
7 (Have-)perfects can occur with specific past time adverbials to this day, as in the following
example from the conversation section of a modern corpus (see Quirk et al. 1985: 195 n. [a],
Michaelis 1994, Miller 2000, 2004a, Bowie et al. 2013, Brown and Miller 2016: 229, 245–247):
(i) I think I’ve had a meal there years ago. [from Biber et al. (2002: 369)]
158 chapter 5
(43) but al his fantasye

Was turned for to lerne astrologye [3191–3192]
Has the game or tale begun/ended or has it been begun/ended (cf. Chapter 4,
Section 4.4.2.5 for OE parallels)? Has the character’s heart broken or has it been
broken? Damagingly for functionalists, such supposedly ill-suited examples as
the Campaign is begun, Christmas is ended or it is broke into factions and partys,
are rather long-lived and are attested as late as the 18th and 19th centuries (see
Rydén and Brorström 1987: 54–57, 81, where they are treated as perfect only).
Turn is also discussed at length in Rydén and Brorström (1987: 175–180), with
an analogous ‘ill-adapted’ example as late as the 1830s from Darwin himself: his
mind is now turned to exertion … (Darwin).
In fact, the above number might have to be revised downwards; if turn in
(43) is conceived of as an intransitive verb in the perfect, i.e. ‘all his fantasy
had turned to learn astrology’, the sentence does not seem particularly well-
formed, since it hardly warrants a pluperfect time frame (though it cannot be
completely ruled out). As it is, was turned is probably adjectival and means
‘was inclined/prone’. The example in (44), and another identical one, were like-
wise put down as ambiguous perfects/passives, though they are most likely
just copular clauses comparable to today’s ‘we are agreed/accorded’.
(44) And thus they been accorded and ysworn

To wayte a tyme [3301–3302]
Gräf (1888: 77, 81) believes that be(e)n accorded and ysworn is simply a per-
fect. ME accorden can mean ‘to agree, to become reconciled’ or ‘to reconcile,
make friends of (persons)’, so it has both transitive and intransitive uses (see
the Middle English Dictionary). In addition to the phrase been accorded, the
participle also appears with another link verb: fallen accorded ‘to become rec-
onciled’, as in (45), which suggests copular status for both (44) and (45).
(45) But at the laste, with muchel care and wo, We fille acorded by vs seluen
two. He yaf me al the brydel in myn hond. [(c1395) Chaucer CT.Mch.
(Manly-Rickert); Source: https://quod.lib.umich.edu/cgi/m/mec/med-id
x?type=byte&byte=1169670&egdisplay=compact&egs=1178495, accessed
on 10 Jan 2017]
Taking out the two instances of accorden and the one with turn would leave
only three genuinely ambiguous be-perfects or passives, but I keep them
in this category so that I am not vulnerable to criticism of being too exclusive.

In addition, Gräf (1888: 79) treats is (y)laft ‘is left’ itself as perfect too, so this
might have to be added here (it is currently in the ambiguous passive/stative
section in the Appendix).8 Notably however, none of the examples in the
Middle English Dictionary illustrating leave’s intransitive sense of ‘stay, remain,
tarry, dwell’ contain a be + left construction. By contrast, there are examples of
the transitive senses where be + left clearly corresponds to a Latin passive: It is
lefte þerfore to the worse [L relinquitur… ad peius]; Gen.45.11: Fyue ȝere ben leuyd
[WB(2): ben residue (probably adjectival/nominal); L residui sunt (passive)] of
hunger (https://quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=9
9594093&egdisplay=compact&egs=99697945&egs=99683945&egs=99734509,
accessed on 20 Jan 2018). I am therefore not convinced of the credentials of is
laft as a perfect, but even if I did count it, the statistics would not be greatly al-
tered. Albeit impressionistically, this does not seem an overwhelming number
of ambiguous cases. Though no statistically sound algorithm has been applied
in order to establish whether this constitutes some critical mass that might
trigger functionally-motivated change, it seems hardly likely. Nor is there much
ambiguity due to ellipsis of the auxiliary verb for the perfect, which could be
have or be. Only one indisputable occurrence of that was recorded, in what is
a reduced/elliptical relative clause (or simply a post-modifying past participle
clause):
(46) A wyf he hadde, ycomen of noble kyn [3942]
Morphologically, come in (47) could also be a past participle, so is this a second

instance of ambiguity due to a missing auxiliary?
(47) Hym thynketh verraily [‘truly’] that he may see

Noees flood come walwynge [‘surging’] as the see
To drenchen [‘drown’] Alisoun, his hony deere. [3615–3617]
The answer hangs on whether come is a past participle or an infinitive. The

following present participle and the future-time context suggest that it is prob-
ably an infinitive. In addition, Davis (1987: xli) reminds us that, in the language
of Chaucer (and Middle English more widely of course), the plain infinitive ‘is
frequent (as it is later) with a noun or pronoun object of verbs of perception:
160 chapter 5
[…] they herde a belle clynke’. However, Gräf (1888: 92) provides Chaucerian
examples with past participles after see.
It is important to reiterate the main points that have emerged in this section.
Firstly, the majority of intransitive verbs which combine with have instead of
be do not seem to do so in order to avoid ambiguity. Secondly, although ambi-
guity between perfect and passive might at first sight appear to be widespread,
the syntactic and semantic context significantly reduces the number of un-
clear cases. Thirdly, ‘undesirably’ unclear constructions persist for an astound-
ingly long time. Having thus analysed Chaucer, I now move on to a text from
the following century.
5.3 The Fifteenth Century: The Second Shepherds’ Play

(Secunda Pastorum)

This play, written some time in the fifteenth century by an anonymous author,
owes its name to being the second of two pageants in the Towneley/Wakefield
Cycle. It mixes religious themes with contemporary realistic social issues
and comic elements (Walker 2000: 42, Sanders 2004: 76–77, cf. http://people.
ucalgary.ca/~scriptor/towneley/plays/second.html, accessed on 18 July 2016),
which makes it comparable in style and subject matter to Chaucer’s fabliaux
analysed above. Sanders (2004: 76) points out that Secunda Pastorum features
many realistic elements, and at the same time might at first sight appear to
be ‘a harshly comic farce’, though of course it relies on a great deal of spiri-
tual symbolism. We should also note the distinctive use of Yorkshire dialect,
which sets the language of Secunda Pastorum apart from Chaucer’s London
English, though in the Reeve’s Tale there is a celebrated attempt to represent
the Northern speech of the two students (see Crystal 2004: 163–168). Likewise,
in the Towneley/Wakefield text, the central character Mak briefly tries to adopt
a Southern dialect, so both authors demonstrate an appreciation of regional
diversity.

This fifteenth-century drama sample is considerably smaller than the sample
from Chaucer’s works (9,857 words). Secunda Pastorum has 754 lines and con-
tains about 5,825 words. Citations and references to line numbers are from
Walker’s (2000) edition. Despite the imbalance in terms of size, the overall
Table 2 Perfect, passive and modal auxiliaries in the Second Shepherds’ Play (Secunda
Pastorum)


copula
Modal have 2 1.9%
Be-perfects with mutative intransitives 4 3.7%
Ambiguity between passive and a stative copular 17 15.9%
clause
Ambiguity between be-perfect and passive/stative 2 1.9%
Missing auxiliary in a non-coordinated context 0 0%
results for both specimens are strikingly similar, as demonstrated in Table 2:

32.7% of transitive have-perfects as opposed to 35.5% in Chaucer, 10.3% of
intransitive have-perfects compared to Chaucer’s 8.3%, 1.9% for modal have
(0.6% in Chaucer), 33.6% for passive be (27.8% in Chaucer), 15.9% of ambigu-
ous passives or statives (11.8% in Chaucer). The only palpable difference con-
cerns be-perfects, to which I return below.
In Secunda Pastorum, have was used to form the perfects of transitive verbs
and the copula in 35 clauses. Again, the word order varied, as in (48)–(49), but
this was not taken into account or deemed significant in determining the stage
of grammaticalisation.
(48) My parte have I fun [‘found/played’] [78]
(49) abyde unto syne [‘until after’]

We have mayde it [148–149]
As with Chaucer, here I included verbs with unexpressed objects which how-
ever are normally transitive:
(50) Have done! [i.e. ‘Get it over with! Finish up!’] [663]
162 chapter 5
Have was employed as many as 11 times to form the perfect of intransitive

verbs.9 Some of those intransitives are not ‘particularly mutative’, e.g. nap, (51),
sleep, (52), or lie, (53).
(51) so long have I nappyd [‘slept’] [2]
(52) Lord, what I have slept weyll [355]
(53) A, my nek has lygen wrang

Enoghe [380–381]
More dynamic is run in (54), which is used with have by Chaucer too.
(54) When we have ryn [‘run’] in the myre [160]
The most interesting use involved a construction with have + gone in (55),
which did not express any hypothetical meaning (cf. esp. Chaucer’s examples
(12) and (13)). Be could have been employed instead without any ambiguity
in (55).
(55) Mak where has thou gone? [199]
Notably, this occurs side by side with a question by another shepherd, who
asks: Is he commen?, see (56). It is puzzling that be was not selected for (55),
and ambiguity cannot have been a determining factor. The combination
of have + gone with a non-hypothetical meaning is innovative compared to
Chaucer, and even to the later texts below. It is tentatively suggested that this
might be due to the northern provenance of the Second Shepherds’ Play. It is
well known that the northern dialect of Middle English is more progressive in
terms of grammar, and this is where most grammatical innovations originate
before they make their way south (e.g. changes in verbal morphology, as well
as loss of inflection or noun gender; see Danchev 1997, Irvine 2004: cliv, Brinton
and Arnovick 2011: 288ff., 296ff., 311, Dolberg 2012, 2018, forthcoming, Lutz 2012:
514, among many others). It seems that be-perfects first started to be lost in
9 This number includes three instances of the past participle farne, probably from ex-
tended senses of the OE verb faran, ‘go, travel’, ME faren, today’s fare. See Appendix 4
and the Middle English Dictionary entry (https://quod.lib.umich.edu/cgi/m/mec/med
-idx?type=id&id=MED15284, accessed on 11 Jan 2017).
the north too. This coincides with the hypothesis involving language contact
with Scandinavian as a contributing factor to the loss (see Chapter 3), as well
as with the data from MSS. D and E of the Anglo-Saxon Chronicle presented in
Chapter 4. Contact with Old Norse in the north of England is highly likely to
have resulted in more and more preference being given to the prominent have
auxiliary at the expense of marginal be. Therefore northern sources from the
Danelaw area, including Secunda Pastorum, show innovative instances of sub-
stituting have for be with common intransitives such as go and fare(n) ‘travel’,
in addition to exhibiting consistently low rates of surviving be-perfects com-
pared to more conservative material from the south.
In line with what was said above, the use of be with intransitives in this
sample is severely curtailed. There are only 4 instances in the entire play, and
those are with the usual last survivors, come and go. Come and go carried on
combining with be as late as the Victorian period (though see the discussion of
Dickens’ language in Chapter 3 above and the Epilogue below; cf. Rydén and
Brorström 1987: 61ff.). Again this 15th-century text proves to be quite ahead of
its time.
(56) Is he commen? [200]
(57) Here shall we hym hyde to [‘until’] thay be gone [333]
(58) All glad were ye gone! [568]
(59) Thay ar gone full clene, that have thay lorne. [‘they have lost the chance’]
[694, translation from Walker 2000: 56]
For (56), Walker (2000: 46) gives the translation ‘Is he here?’, which might sug-
gest subject-complement status for the participle, but this might be anachro-
nistic. (56) is still best regarded as a perfect, though perhaps already undergoing
some fossilisation into a fixed expression. Another important observation
is that not many of the intransitives which combine with have instead of be
would have been potentially ambiguous if used with be – they would be either
completely inappropriate with it (?*so long am I nappyd; see (51)), or absolutely
unambiguous, as in (60), where, incidentally, the pluperfect serves a modal
function of indicating non-fulfilment.
(60) I wald [‘wish’] I had ryn [‘run’] to [‘until’] I had lost hir [108]
[OR: ?‘I was/were ryn’ – still unambiguous because of context?]
164 chapter 5
However, the clauses in (61), semantically a ‘future perfect’, and (62), where
Present-Day English could resort to a perfect progressive (‘has been lying’),
might create confusion if be were substituted in them, and I will return to such
cases below. There were only about three such potentially unclear instances
out of the eleven have-perfects with intransitives.
(61) When we have ryn [‘run’] in the myre [160]
(62) A, my nek has lygen wrang

Enoghe [380–381]10
More evidence will be adduced below that the advanced loss of be-perfects
and the widespread use of have with intransitives in Secunda Pastorum does
not seem to have been prompted primarily by the avoidance of ambiguity. In
the light of this, language contact emerges as a more credible explanation.
Contrary to what would be expected if this displacement was mostly due to
a drive to achieve a one-to-one correspondence between form and function/
meaning, have, in addition to virtually taking over the role of perfect auxiliary
across the board, starts to appear in contexts suggestive of some modal colour-
ing, as already hinted in the discussion in Chapter 3 and illustrated here with
authentic material in context. There are two or three potential examples in
the text of the incipient recruitment of have for expressing modal (or at least
delexical) meanings:
(63) Have hald my hand [350]
(64) I had lever [‘rather’] be dede [‘dead’] or [‘than’] she had any dyseasse.
[486]
(65) We have it not to lose. [673]
Admittedly none of them is indisputably modal. The first one, in (63), is partic-
ularly unclear and should be discarded. It appears to showcase delexical have,
i.e. ‘have/take hold of my hand’ (Steven Kaye, p.c.). (64) is the most convincing.
Here, the context suggests the translation ‘I would rather be dead [now] than
10 The distinction between intransitive lie and transitive lay must have helped, though it has
been lost in many dialects (since?). Van Gelderen (2014: 130) mentions confusion of lie
and lay in Middle English. I treat (62) as potentially ambiguous in order to be generous to
the hypothesis that is argued against.
[that] she had any dis-ease/discomfort’, a volitional and hypothetical meaning

expressed by a past-tense form with present-time reference – exactly the way
modality works in many languages, including English. Therefore, this is prob-
ably modal have followed by a bare infinitive, though be might be an n-less past
participle (see Łęcki 2010: 69–78). Regarding the emergence of this construc-
tion in late Middle English, Łęcki (2010: 71–72) notes that have almost always
appears in its past subjunctive form had, the subject is nominative, and levere
or bettre is followed by an infinitive with or without to, as in (66) below, which
closely parallels (64) (see Chapter 3):11
(66) Thei had welle lever haue bene stille.

they had well rather have been still
[c. 1340, Hampole, Prose Tr. (1866) 25, cited in Łęcki (2010: 72)]
For (65), Walker (2000: 56) supplies the reading ‘We have nothing to lose’, so it
is not at all certain if have has a modal colouring, but an interpretation where-
by it deontically expresses ‘We shouldn’t lose it’ is not that far-fetched, though
probably not the best reading in context. Nevertheless, the meaning of have,
on either interpretation, is slowly drifting towards modality.
As before, it was very difficult to tease apart bona fide passives from copular
statives. In context, a dynamic reading is favoured for (67) because of the verb
yede ‘went, walked’ in the subordinate clause, and because the entire play is
about the act of stealing the sheep (compare a similar example in the discus-
sion of Chaucer’s tales above).
(67) oure shepe that we gett [‘guard’]

Ar stollyn as thay yede [‘walked’] [505–506]
Similarly, though slain might be paraphrased with the adjective ‘dead’, I still
included (68) in the passive section, so that ambiguous cases did not multiply
exponentially. This choice is in no way indisputable or foolproof, but it was
thought safer to err on the side of caution, so that the number of potential
ambiguities would not be artificially inflated to prove my point.
11
M E lever(e) goes back to a comparative form of OE leof ‘desirable, beloved, dear, precious’,
related to OE lufu ‘love’, from PIE root *leubh- ‘love’, with cognates in German lieb, Gothic
liufs, Old Icelandic ljūfr, Latin libīdō ‘desire, pleasure’, Old Church Slavonic/Old Bulgarian
ljubъ ‘pleasant’ (Łęcki 2010: 69–70). It is this word, in addition to the present-time refer-
ence and the simple preterite in the second clause introduced by or in (64), that prevents
a perfect reading of ‘I had rather been dead’/‘I would rather have been dead’, which is less
likely here.
166 chapter 5
(68) I trow [‘believe’] oure shepe be slayn [541]
Even with this conservative approach, most of the ambiguity again involved
genuine passives, 36 in number, vs. potential statives, 17. By contrast, there
were hardly any ambiguous be-perfects (see below). Stative copular status was
indicated by adjectival, (69), or nominal paraphrase, (70), or by coordination
with nouns or adjectives, (71); so for (71), coordination with adjectives is the
main reason why it was not counted as a genuine passive.
(69) my fyngers ar chappyd [‘sore’] [3]
(70) These men that ar wed have not all thare [‘their’] wyll [73] [‘are (part/
members of) a couple’]
(71) He was poorly arayd [‘clothed’],

Both mener [‘poor’] and mylde. [690–691]
The example in (72) was treated as passive, but this choice needs to be justified:
(72) With you will I be left [623]
Only a passive reading seems viable in this context. ‘With you will I have left’
does not make much sense at all, because of the anomalous temporal orienta-
tion and the meaning which this interpretation would entail; the context and
the preposition with here suggest that the speaker and addressee(s) will/wish
to remain together, not that the speaker will/wishes to leave a place or a person
in the addressee’s company (though cf. the section on Chaucer and Gräf 1888:
79–80, who would perhaps admit a perfect reading in the sense of ‘remain’).
Once more, when all things have been considered, the result is a negligible
number of only two ambiguous be-perfects, (73) and (74):
(73) I am wery, for-rakyd [‘exhausted’] and run in the myre [256]
(74) I am sett for to spyn [298]
Again, (73) might have to be removed and perhaps treated as adjectival or

passive only, because the context disambiguates it – syntactically (coordina-
tion with an adjective and an adjectival or passive past participle, though this
is not conclusive), as well as semantically/pragmatically, suggesting that the
speaker has been driven into the mire, rather than having run there himself
(cf. Chaucer in the previous section). What is more, the verb run appears three
times in the perfect elsewhere in the play, all of them with have and none with
be, as in (75), making it unlikely that (73) is a be-perfect.
(75) Ye have ryn in the myre, and ar weytt yit [‘still wet’] [494]
Be that as it may, run is attested in analogous ambiguous be-perfects, like But

we are not yet quite run down to the old level, as late as the 1870s (see Rydén and
Brorström 1987: 160, who treat this as perfect only), further undermining func-
tional streamlining as a decisive factor. Similarly, Rydén and Brorström (1987:
160) provide another ambivalent example from the 1890s which is pretty much
identical to Chaucer’s from the previous section: until the sap is run out of our
lives. No cleaning up of such supposedly woefully undesirable combinations
appears to have taken place.
As ambiguous be-perfects go, (74) is a far better candidate (cf. Gräf 1888: 78).
Mak’s wife says that she has sat down, or maybe set herself (down), in order to
do some spinning. Alternatively, she is set (passive or adjectival, i.e. ‘ready’?) to
do some spinning. Gräf (1888: 78–79) supplies analogous cases from Chaucer.
For the confusion that used to (and sometimes still does) exist between sit
and set, as well as for 18th- and 19th-century ambiguous examples parallel to
(74), see Rydén and Brorström (1987: 168–170). Crucially, no other ambigui-
ties involving the perfect were detected, so this second Middle English case
study confirms that ambiguity does not appear to have played a major part in
the historical shifts that marked the development of the perfect – already far
advanced in this innovative northern text. Thus, language contact emerges as a
more viable explanation. Our next stop is the Elizabethan era.
5.4 The Sixteenth Century: Shakespeare

The Taming of the Shrew most probably appeared in 1593 or 1594, having been
written between 1590 and 1594 (Barton 1997: 138, Evans and Tobin 1997: 80).
Like the previous texts examined above, it is also farcical and light-hearted
in tone and involves playing pranks, so its subject matter, style and contents
were judged appropriately compatible with the works previously selected
(see Barton 1997, who also mentions links between the tradition of Chaucer,
mystery plays and Shakespeare’s comedy). It was appreciated as felicitous for
present purposes that The Taming of the Shrew is considered ‘short on poetry
and deep emotion’ (Barton 1997: 140). All line references and citations here
168 chapter 5
are from Evans and Tobin (1997). The following website was also used: http://
shakespeare.mit.edu/taming_shrew/full.html (accessed on 28 July 2016). Only
the first three acts were examined (as they appear in Evans and Tobin 1997),
because otherwise this sample would have grown disproportionately big com-
pared to the others. The first three acts amount to approximately 13,247 words,
making this the largest sample analysed so far in this chapter. Importantly, this
text is not included in the Helsinki Corpus.

Due to the larger sample, the figures for the 16th century are higher than those
for the previous two centuries, but the proportions for the most part remain
strikingly similar. Have-perfects with transitives, broadly construed, were the
most numerous – 74. Word order is still flexible in declarative clauses with cen-
tral transitive verbs which take have to form their perfect:
(76) Seeing too much sadness hath congeal’d your blood [Induction, Scene II, l.
132]
(77) for I have Pisa left

And am to Padua come [I.i.21–22]
table 3 Perfect, passive, progressive and modal auxiliaries in the first three acts of
Shakespeare’s The Taming of the Shrew

Have-perfects with transitive verbs & with copula 74 37%

and auxiliary be
Have-perfects with intransitive verbs 2 1%
Modal and causative have 7 3.5%
Progressives 11 5.5%
Modal be 2 1%
clause
Ambiguity between be-perfect and passive/stative 8 4%
Missing auxiliary in a non-coordinated context 2 1%
’s contraction 1 0.5%
(78) And therefore has he closely mew’d her up [‘placed her in close
confinement’],
Because [‘so that’] she will not be annoy’d with suitors. [I.i.183–184]
(79) Therefore this order hath Baptista ta’en [I.ii.126]
Throughout the periods studied so far in this chapter, word order proves of
little value in indicating the extent to which the perfect is grammaticalised.
Another typical feature of Shakespeare’s English is that a relative clause might
be missing its subject relativiser:
(80) and I have met a gentleman

Hath promis’d me to help [me] to another [I.ii.171–172]
Today, we would expect ‘a gentleman who/that has promised [me] to help

me …’, at least in Standard English.
There were objectless clauses which were nevertheless treated as transitive
because their verbs have the potential to take objects. Therefore, dream in (81)
below is taken to be transitive whereas sleep is considered intransitive.12
(81) they say that I have dream’d

And slept above some fifteen year or more. [Induction, Scene II, l. 112–113]
Other examples include woo [somebody], (82), and tune [something], (83).
12 H
ave applies to a (potentially) transitive and to an intransitive verb in this example. There
is hardly any possibility of finding be, however, as it is doubtful that sleep would normally
combine with it anyway; as confirmed by (52) above, as well as by the OED and the follow-
ing examples from the Middle English Dictionary, sleep usually takes have:
(i) Þo hi hadde alle islepe [Ashm: islept; vrr. slepe, slepen] ynouȝ, sone hi gonne arise &
wende to here schip. [a1325(c1280) SLeg.Pass. (Pep 2344), https://quod.lib.umich.edu/
cgi/m/mec/med-idx?type=byte&byte=182957819&egdisplay=compact&egs=1829883
22&egs=182999307, accessed on 12 Jan 2017]
(ii) Þat niȝt he hadde litel yslape. [c1350(a1333) Shoreham Poems (Add 17376), https://
quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=182957819&egdisplay=
compact&egs=182988322&egs=182999307, accessed on 12 Jan 2017]
(iii) When thei hadde scleped & saw tyme, Thei ros vp be-fore the prime. [c1440(a1401) Life
Bridlington in NM 71 (Yale 331), https://quod.lib.umich.edu/cgi/m/mec/med-idx?
type=byte&byte=182957819&egdisplay=compact&egs=182988322&egs=182999307,
accessed on 12 Jan 2017]
170 chapter 5
(82) Yet never means to wed where he hath woo’d. [III.ii.17]
(83) His lecture will be done ere you have tun’d. [III.i.23]
Line 25, only two lines down from sentence (83), contains the words tune your
instrument, thereby confirming the transitive nature of this verb. Verbs with
finite, (84)–(85), and non-finite, (86), clausal complements were treated as
transitive too, even when incompatible with NP direct objects.
(84) your father hath consented

That you shall be my wife [II.i.269–270] [that-clause complement]
(85) we have ‘greed so well together,

That upon Sunday is the wedding-day [II.i.297–298] [that-clause
complement]
(86) As [‘as if’] had she studied to misuse me so [II.i.159] [to-infinitive

complement]
Prepositional verbs were also grouped with the transitives proper:
(87) You have but jested with me all this while [II.i.20]
(88) yourself and all the world,

That talk’d of her, have talk’d amiss of her [II.i.290–291]
Light in (89) was likewise taken to licence a prepositional complement, but it

is possible to regard it as a mutative intransitive which indicates directed mo-
tion and combines with have in order to avoid potential ambiguity due to the
multi-functionality of be.
(89) And by good fortune I have lighted well

On this young man [I.ii.167–168]
(90) approximates a set expression, along the lines of ‘finish up, get it over
with’, like modern ‘have done with it’, but it was nevertheless included in the
count:
(90) therefore ha’ [‘have’] done with words [III.ii.116]

Have got was accepted as a bona fide perfect construction too – (91)–(92).
Judging by the context, it is unlikely that it had by that stage been reinterpret-
ed as a possessive stative equivalent of ‘have/possess/own’ (at least in these
examples).
(91) That none shall have access unto Bianca

Till Katherine the curst have got a husband. [I.ii.127–128]
(92) No doubt but he hath got a quiet catch. [II.i.331]
For both of these, the context points to a non-stative, non-possessive meaning

akin to ‘he/she has caught/obtained’.
As before, this section included the perfects of the verb be. Apart from copu-
lar be, however, (93), I now register the perfects of auxiliary be. Perfect passives,
(94), and perfect progressives, (95), appear for the first time in the database,
confirming a much more advanced stage of grammaticalisation.
(93) Persuade him that he hath been lunatic [Induction, Scene I, l. 63]
[copular be]
(94) We have not yet been seen in any house [I.i.199] [perfect passive]
(95) this young scholar,

that hath been long studying at Rheims [II.i.79–80] [perfect progressive]
The most surprising finding, compared to the texts from the 14th and the 15th
centuries, is the virtual absence of intransitives which form their perfect with
have, despite the larger size of this sample. There were only two such instances:
(96) they say that I have dream’d

And slept above some fifteen year or more. [Induction, Scene II, l. 112–113]
(97) Whom would to God I had well knock’d at first,

Then had not Grumio come by the worst. [I.ii.34–35]
(96) features a conjoined structure where the auxiliary have applies to two
past participles, the first taken to be potentially transitive and the second a
non-mutative intransitive (as already mentioned above). (97) can be explained
away by reminding readers that have tends to combine with intransitives of
172 chapter 5
motion such as come in hypothetical contexts, i.e. ‘then G. wouldn’t have come
by the worst’ or ‘if then G. had not come by the worst’ (see Chapter 3, esp.
Sections 3.3 and 3.4.1, as well as Rissanen 1999: 229–231 for Early Modern condi-
tionals). In the light of this, it is remarkable that Shakespeare prefers be to form
the perfects of (mutative) intransitives, a total of 17, including a surprisingly
conservative counter-factual conditional clause with the be-perfect of come,
(98). This goes against the tendencies noted in Chapter 3 that hypothetical
contexts and the past tense both favour have over be.
(98) If, Biondello, thou wert come ashore,

We could at once put us in readiness, [I.i.42–43]
The use of be with intransitives is much more extensive than it is in Secunda

Pastorum, where it was confined to come and go; indeed, Shakespeare approxi-
mates Chaucer in applying be to a wider range of verbs, with even fewer oc-
currences of have instead of be in such structures. Apart from the staples come
and go, (99), mutative intransitives which here select be included arrive, (100),
grow, and step, (101), the latter being especially unusual and interesting, dem-
onstrating that the be-perfect is pretty much alive for Shakespeare (cf. Rydén
and Brorström 1987: 172). This further bolsters the contact scenario outlined in
the previous section, with Shakespeare representing more conservative south-
ern usage outside the former Danelaw.
(99) you are come to me in happy time [‘opportunely’] [Induction, Scene I,

l. 90]
(100) I am arrived for fruitful [‘fertile’] Lombardy [‘have come here on my

way to’] [I.i.3]
(101) since we are stepp’d thus far in,

I will continue […] [I.ii.83–84]
Regarding (102), only a perfect reading, i.e. ‘until the sun have/hath/has set’
makes sense; (102) cannot be the passive of an active *Someone (has) set the
sun (cf. (74) from the previous section for another treatment of set).
(102) Thrice noble lord, let me entreat of you

To pardon me yet for a night or two,
Or, if not so, until the sun be set: [Induction, Scene II, l. 118–120]
Likewise, the passive would make no sense for (103). What is unequivocally
meant here is ‘My master has grown/become quarrelsome’ (cf. *Someone grew/
has grown my master quarrelsome).
(103) My master is grown quarrelsome. [I.ii.13]
There were numerous instances of be welcome, but they were discarded as ad-
jectival and not recorded, apart from the following one, in which the dialogue
might indicate that contemporary native speakers could perhaps feel a certain
link between welcome as an adjective and a perfect of having/being [well] come:
(104) Baptista: You are welcome, sir.

Petruchio: And yet I come not well. [III.ii.88]
Petruchio’s remark might suggest that, to a contemporary speaker, a link to a

be-perfect of come was perhaps alive to some extent, though it has to be noted
that Shakespeare is especially sensitive to linguistic matters and fond of lan-
guage puns, as indicated elsewhere in the play and in his work more generally,
probably much more so than the average speaker.13
Similarly, the phrase be gone might have been undergoing fossilisation as a
single fused and non-decomposable item, but it was nevertheless counted as
a genuine combination of perfect be + the past participle of a mutative verb:
(105) Fellows, let’s be gone. [I.ii.278]
(106) I must be gone [III.i.85]
(107) For me, I’ll not be gone till I please myself. [III.ii.212]
Admittedly, let’s have gone and I must have gone would sound markedly anom-
alous if inserted above, so be gone is probably a set phrase already, though
13 Etymologically, welcome as a verb derives from Old English wilcumian ‘to welcome, greet
gladly’, ultimately from wilcuma (n.) ‘a welcome guest’, literally ‘will-comer’, i.e. ‘one
whose coming suits another’s will or desire’, from willa ‘will, desire, pleasure’ + cuma
‘comer, guest’, related to cuman ‘come’ (see the OED). The first element of this compound
word was subsequently altered to the adverb wel(l) and the second element cuma came
to be identified with the imperative or infinitive of the verb come, under the influence
of Old French and Latin. This folk etymology must have served as the popular basis of
Shakespeare’s word play and confirms that the general English-speaking public did per-
ceive a connection between the compound welcome and its components well + come.
174 chapter 5
I will not have gone till I please myself is somewhat more convincing. Rydén
and Brorström (1987: 104 fn. 4) exclude constructions like imperative be gone or
I must be gone ‘I must go’ from the category of genuine perfects since there is
no opposition with have. The salient semantic component in be gone must be
the result, whereas have gone would probably highlight the anteriority.
Importantly for present purposes, this snapshot of the history of English
vividly demonstrates the proliferation of functions of both auxiliaries under
investigation. Already present here are modal and causative uses of have, as
well as modal and progressive be. The modal construction had lever/rather,
previously encountered in the 15th-century Townley play, finds continuation
in (108):
(108) but I had as lief take her

dowry with this condition: to be whipt at the high
cross every morning [I.i.131–133]
Have is also used twice with a to-infinitive in interrogative clauses, (109)–(110):
(109) what have you to do? [I.ii.224]
(110) I will be angry; what hast thou to do?

Father, be quiet [III.ii.216–217]
For both of these, Evans and Tobin (1997: 151, 159) provide the translations
‘what is it to you’ or ‘it’s no concern of yours’. No matter whether they repre-
sent deontic obligation, as it would appear to a modern reader, or whether
they stand for ‘it’s none of your business’, as per the translations provided in
the explanatory notes, modal overtones are present anyway. Note also the in-
terchangeability of have and be in the originals and the translations, once more
showing how one generic verb can easily substitute another. (111), on the other
hand, is only faintly modal, mostly because of the imperative mood, and might
need to be discarded:
(111) and then have to’t afresh [I.i.139]
As indicated elsewhere in the play (see e.g. I.ii.194), to’t stands for ‘to it’. Have
to it does not exist in English now, but have at it does, in the meaning ‘get on
with it’, ‘get down to it’, ‘take it away!’ (Steven Kaye, p.c.).
(112) is likewise not unproblematic, but it might illustrate the rise of caus-
ative have:
(112) we’ll have thee to a couch

Softer and sweeter than the lustful bed
On purpose trimm’d up for Semiramis. [Induction, Scene II,
l. 37–39]
In Shakespeare’s English, verbs of motion are often omitted under certain cir-
cumstances, as amply illustrated elsewhere in the play (e.g. I’ll in to counsel
them [Induction, Scene I, l. 136], i.e. ‘I’ll go in to counsel them’). So a verb of mo-
tion might be reinserted in (112) above too, something along the lines of ‘we’ll
have thee taken to a couch’. Be that as it may, the following are less contro-
versial causatives, demonstrating how have has been branching out into new
(quasi-)auxiliary territories:
(113) I’ll have them [‘books’] very fairly [‘handsomely’] bound [‘I desire them
to be, i.e. see that they are’] [I.ii.145]
(114) Take your paper too,

And let me have them [‘books’] very well perfum’d [I.ii.150–151]
The proliferation of functions of be is showcased by the rise of the modern pro-

gressive, appearing for the first time in my Middle and Early Modern English
corpus – 11 occurrences, some of them showcased in (95), (115)–(118), including
perfect progressives (see (95) above), past progressives, (117), and progressives
with modals, (118).14
(115) Trow [‘know’] you whither I am going? [I.ii.164]
(116) besides an argosy [‘large merchant vessel’]

That now is lying in Marsellis road [‘harbour of Marseilles’]
[II.i.374–375]
14 It should be noted, however, that despite the emergence of the progressive, there are
many instances in which Shakespeare uses a simple form where a progressive would be
required today (see Rissanen 1999: 216, Beal 2004: 78ff., and example (129) below). (Proto-)
progressives were mentioned in the sections on the Anglo-Saxon Chronicle in Chapter 4.
On the (disputed) continuity between the OE progressive and the one that I now record,
consult Chapter 4, Fischer and van der Wurff (2006: 135ff.), Brinton and Arnovick (2011:
229, 373).
176 chapter 5
(117) as he was drinking [III.ii.176]
(118) I should be arguing still upon that doubt [III.i.55]
There are two instances of be, (119)–(120), in which it combines with a to-
infinitive and has undeniable modal overtones, just like have above. The modal
be-to construction expresses futurity combined with deontic colouring, as it
still does today. (120) can be paraphrased as follows: ‘I need to/should/have to
get a man’, once more showing the synonymy of modal be and have. The shift
to deontic and future meanings was illustrated in Chapter 3, Section 3.5.2.3.
(119) I am thus resolv’d: on Sunday next you know

My daughter Katherine is to be married. [II.i.393–394]
(120) I am to get a man – whate’er he be,

It skills [‘matters’] not much, we’ll fit him to our turn [III.ii.131–132]
The passive still accounts for the greatest share of auxiliary uses of be – as
many as 57 dynamic passive clauses. (121) and (122) attest to the freedom with
which Shakespeare could employ conversion and coin passivisable verbs out
of other parts of speech, even proper nouns, (122) (see Crystal 2004: 332–333).
(121) being mad herself, she’s madly mated [III.ii.244]
(122) I warrant him, Petruchio is Kated. [III.ii.245]
Apart from the perfect passives noted above, there were also non-finite pas-
sives: with a present participle, (123), and a to-infinitive, (124), including a
passive bare infinitive after a modal, (125).
(123) being close convey’d [Induction, Scene I, l. 127]
(124) to be noted [III.ii.14]
(125) O, that a mighty man of such descent,

Of such possessions and so high esteem,
Should be infused with so foul a spirit! [Induction, Scene II, l. 14–16]
Although such complex combinations are overall on the increase, there are
nevertheless instances where we would now expect, say, a perfect passive, but
we find a simple form instead, so grammaticalisation has not quite progressed
to its modern state (see Chapter 3):
(126) So workmanly the blood and tears are drawn. [Induction, Scene II, l. 60]
[instead of ‘have been drawn’, in a painting]
Due to other large-scale morphosyntactic realignments in the language, active

indirect objects appear as passive subjects for the first time in my dataset too,
as in (127), which ensures one more option for passive-building, thereby poten-
tially indirectly boosting its numbers.
(127) shall I be appointed hours [I.i.103]
Mostly because of the context, (128) is (tentatively) believed to represent a

genuine passive construction, though it might potentially mean either ‘until
she gets/has got married’ (dynamic) or ‘until she is no longer single’ (stative).
(128) Until the elder sister first be wed [I.ii.261]
More convincingly, (129) is best read dynamically as: ‘To me she is getting/
being married’, since they haven’t got married yet at this stage in the play. Note
incidentally that now we would use the progressive to convey the meaning of
arrangement for the future, as pointed out earlier. Also compare the previous
samples, where other category-assignment decisions were made for the same
participle due to context. This vindicates taking the approach outlined at the
beginning of Chapter 4.
(129) To me she’s married, not unto my clothes [III.ii.117]
Again due to context, it is clear that (130) is passive too – the gentleman has
been grieved by his shrewish daughter, and grieve is not really mutative. ‘Had a
gentleman thus grieved’ would not fit in context, and also perhaps temporally,
as the past perfect might introduce anomalous time reference.
(130) Was ever gentleman thus griev’d as I? [II.i.37]

178 chapter 5
Once again, my study demonstrates the proven usefulness of taking a tra-

ditional philological approach. Reading the entire work, as opposed to just
machine-harvested chunks, affords insights into context which would other-
wise remain elusive and which enable and enhance in-depth analysis (albeit
with a numerically less impressive database). Confirming the same point, the
word curst in (131), though it might at first sight resemble a verbal past par-
ticiple, is actually conceived of as an adjective, ‘shrewish’, and was therefore
excluded from all counts. It is found inflected for the superlative degree of
comparison elsewhere in the text.
(131) ‘Tis bargain’d ‘twixt us twain, being alone,

That she shall still be curst in company. [II.i.304–305]
A total of nineteen sentences were judged to be ambiguously passive or stative.

Again, this was determined based on the availability of an adjectival para-
phrase, in brackets after examples (132)–(136), though often it was a matter of
pure intuition and individual discretion. The very difficulty of these choices
confirms how pervasive this type of ambiguity is. When in doubt, I was more
likely to include something as a bona fide passive (as before), so the number of
‘unambiguous passives’ might artificially be slightly higher.
(132) I am not bestraught [‘distracted, mad’] [Induction, Scene II, l. 25]
(133) But art thou not advis’d [‘aware’], he took some care [I.i.186]
[‘Aren’t you aware/don’t you know?]
(134) would ‘twere done! [I.i.254] [‘I wish it were/was over/complete’]15
(135) But be thou arm’d for some unhappy [‘hateful’] words. [II.i.139]
[‘prepared, ready’]
(136) Yet you are wither’d. [II.i.237] [‘old’]
Similarly, a perfect reading (‘His lecture will have done’) is ruled out for (137),
for which there is a ready adjectival/adverbial paraphrase – ‘His lecture will be
over’ (cf. (134) above and Chaucer in Section 5.2.2):
15 There were other combinations of be + done here which were treated as genuinely passive
because no such adjectival paraphrase was available – see Appendix 5.
(137) His lecture will be done ere you have tun’d. [III.i.23]
The example in (138) is different. It can be passive because strike is typically a

transitive verb. A perfect of the form I/Myself have struck in years would now
be decidedly strange, but the Middle English Dictionary does give a since-lost
intransitive sense for ME strike(n) (inf.), ‘to go, make one’s way, proceed’, so a
perfect interpretation is rather plausible, i.e. ‘I have gone/advanced in years’, a
leftover from ME which persists in Shakespeare but which is by now defunct
(see https://quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=198059
501&egdisplay=compact&egs=198129127, accessed on 13 Jan 2017).
(138) Myself am strook [‘struck, i.e. advanced’] in years, I must confess

[II.i.360]
In addition to (138), there were another seven instances of ambiguity which

had a possible be-perfect interpretation. The slightly higher number might be
due to the larger sample, but the proportion (4%) is more or less like the one
calculated for Chaucer (3.6%). If be-perfects were disappearing due to ram-
pant ambiguity, one would expect ambiguity to gradually and monotonically
decrease over time, as opposed to increasing, stagnating or wildly fluctuating.
This should be the palpable result produced by the displacement of be by have,
which had started centuries before Shakespeare’s day and which supposed-
ly got rid of ambiguity and functional overload. It would be quite an orderly
story: be is gradually being displaced by have and with it ambiguity drops as
well. However, the historical record suggests that this is not quite how things
proceeded, at least not straightforwardly and at least not in the drama samples.
Ambiguous be-perfects constitute 3.6% of all recorded auxiliaries in Chaucer’s
selected texts (plus 0.6% due to missing auxiliaries), 1.9% in Secunda Pastorum,
4% in Shakespeare (+ 1.5% due to missing or contracted auxiliaries) and 0.7%
in Restoration comedy below (+ 2.6% due to missing or contracted auxiliaries),
so Shakespeare disrupts the pattern due to his southern conservatism which
makes him comparable to Chaucer. No clear trend emerges after adding up
the percentages from ambiguous be-perfects and ambiguity due to contrac-
tion or omission of the auxiliary. If the categories are kept apart, ambiguous
be-perfects decline over time (ignoring Shakespeare), whereas contraction- or
omission-related ambiguity increases (ignoring Secunda Pastorum). By con-
trast, the less representative and wide-ranging correspondence material in
subsequent sections does appear to follow a trend of diminishing ambiguity.
There might be another correlation – Chaucer, who has be-perfects, also
exhibits some ambiguity; ambiguity drops in the linguistically progressive
180 chapter 5
Secunda Pastorum, where be-perfects are severely curtailed, and then returns
with Shakespeare, who uses be-perfects more freely.16 In effect, be-perfects
tend to be accompanied by ambiguity. If this is so, though, Shakespeare does
not seem to resolve uncertainty of interpretation by resorting to have-perfects
with intransitives, as the earlier authors had done, including Chaucer, whose
sample contains more mutative have-perfects (disambiguating at least to a
certain extent). Those are virtually absent here and might suggest that a cor-
relation between unclear be-perfects and the rise of have-perfects with mu-
tatives is at best tenuous. Lack of clarity might just appear to thrive where
be-perfects thrive themselves. Secunda Pastorum could indicate that eliminat-
ing be-perfects eliminates ambiguity. So does ambiguity ‘thrive’ in Shakespeare
because he does not rely on have-perfects with mutative intransitives (instead
of be) and the be-perfect has not been suppressed to a sufficient degree?
In fact, thriving might be an overstatement – out of the eight instances in
the first three acts of The Taming of the Shrew, not that numerous as they are,
four are only partially ambiguous, (139)–(142):
(139) For how I firmly am resolv’d you know:

That is, not to bestow my youngest daughter
Before I have a husband for the elder [I.i.49–51]
(140) I am resolv’d [I.i.90]
(141) I am thus resolv’d: on Sunday next you know

My daughter Katherine is to be married. [II.i.393–394]
(142) I am agreed, and would I had given him the best

horse in Padua [I.i.142–143]
None of these would make a very convincing passive today – compare the
modern ill-formedness of ?*Someone resolved/agreed me, though resolve had
transitive senses such as ‘inform’ or ‘satisfy’ in Shakespeare, and Wycherley’s
Country Wife does have Thou hast stared upon her enough to resolve me [II.i,
p. 172] (see Appendix 6,17 as well as the OED for earlier transitive uses of agree
16 Here, the temporal dimension should be compounded with the dimension of regional
variation – Chaucer and Shakespeare seem to represent conservative southern usage,
whereas Secunda Pastorum is innovative for its time because it originated in the north. By
the Restoration period, the change has presumably made its way south too.
and resolve with human objects). Resolved can take a to-infinitive clausal
complement, as in (139), so its perfect should be built with have. In addition,
a perfect reading would be strange in context for I am agreed. I have agreed
would indicate some sort of temporal precedence, which is hardly appropri-
ate, though it should not be ruled out completely (i.e. (142) might focus on
the present state).18 Despite the theoretical ambivalence potential, examples
(139)–(142) should perhaps be analysed as mostly adjectival. This is confirmed
by Shakespeare’s usage elsewhere. As evident from (85) above, when he forms
the legitimate perfect of the verb agree, he selects have, as we would today:
we have ‘greed so well together... What is more, both adjectival constructions,
with agreed and resolved, survive intact to this day, making it hardly likely that
they were ever thought of as real be-perfects, or at least they conceivably be-
came ‘fossils’ like be gone. The adjectival status of PDE resolved, for instance,
is confirmed by compatibility with modifiers like very in (143). In any case, the
survival of these expressions makes it unlikely that ambivalence was ever an
issue that needed to be fixed.
(143) Well, the president is very resolved, looking for a – a way to have a diplo-
matic solution [COCA, 1990s]
This leaves only four indisputable cases which could be perfect, as well as
passive – (138) above and (144)–(146) below:
(144) Tranio is chang’d into Lucentio. [I.i.237]
(145) My lord, ‘tis but begun. [i.e. ‘the play is but begun’] [I.i.252]
(146) Petruchio:
Myself am mov’d to woo thee for my wife.
Katherine:
Mov’d! in good time [‘indeed, forsooth’]! Let him that mov’d you hither
Remove you hence. I knew you at the first
You were a moveable. [II.i.194–197]
18 Nakamura (1987: 34–35) treats they are not yet agreed to impeach him and the Duke of York
and Duchess and my Lady Castlemaine are now all agreed in a strict league from Pepys’s
Diary as be-perfects. Crucially, Nakamura often fails to take into account potential ambi-
guity, e.g. phrases like [X] is altered, the war is begun, or the river is frozen are regarded as
perfect only (see e.g. Nakamura 1987: 35ff.).
182 chapter 5
(146), for instance, could be an intended perfect (‘I have moved to woo thee …’)
which is then turned into a pun by Katherine, who takes it to be a passive in-
stead. Rydén and Brorström (1987: 147–148) note that this verb tends to favour
have ‘obviously’ in order to avoid ‘interpretative difficulties’, though they quote
ambiguous examples from the 19th century. As pointed out above, we know for
a fact that Shakespeare was fond of such linguistic play and loved puns based
on the availability of more than one reading – allowing multiple interpreta-
tions appears to be favoured rather than disfavoured in this text. It turns out
that ambivalence is embraced and cherished, rather than avoided. Even so,
upon consideration ambiguity seems hardly more widespread here than it was
in the previous samples (considering their sizes), as highlighted already. Once
again, it is worth stressing that such ‘interpretative difficulties’ persisted well
into the modern period (cf. also Chaucer’s earlier use of begin), with similar
examples of be changed from the 18th and 19th centuries provided by Rydén
and Brorström (1987: 58–61). Despite its supposedly ‘inferior fitness’, Rydén
and Brorström (1987: 59) note that be is still clearly favoured for the perfect of
change as late as the 19th century.
In the same vein, there were only two instances of lack of clarity as to the
choice of be or have due to a missing auxiliary – (147) and (148). The second
of those, (148), is rather dubious, as it is surrounded by unequivocal passives
or adjectives – the unfortunate horse in question is troubled, infested, defiled
with and full of various diseases (see further Section 5.6.2.2).
(147) A merchant of great traffic [‘business’] through the world,

Vincentio, come of the Bentivolii. [I.i.12–13]
(148) [his horse] sped [‘far gone’]

with spavins [‘a disease’] [III.ii.52–53]
Finally, there was only a single contracted ’s in the excerpt, (149). In the light of
all the evidence presented here, we can be fairly confident that is should be re-
covered in both instances involving come, (147) and (149), as far as Shakespeare’s
usage is concerned. Moreover, there are indications that ’s can only stand for is
in pre-18th-century English more widely (Rydén and Brorström 1987: 32 fn. 10).
(149) I am glad he’s come, howsoe’er he comes. [III.ii.74]
In conclusion, the three text samples investigated so far in this chapter dem-
onstrate that cases of ambiguity in context are not as many as might have
been expected – they consistently do not tend to exceed the 5% mark in all
periods investigated. Even if some of my judgements as to category assignment

were challenged or revised by other scholars, those percentages would not be
greatly altered. Nor have I detected any indisputable straightforward trends in
ambiguity increasing or decreasing consistently over time – considering the
size of the samples, the two cases in Secunda Pastorum are not that incommen-
surate with Shakespeare’s eight, which are actually closer to four. The material
demonstrates that there is no linear diachronic decrease in the number of un-
clear clauses without intermittent dips or peaks. Even Kilpiö’s more extensive
data cited in Chapter 3, Section 3.5.4, arguably shows an overall decrease in
ambiguity (OE 8%, EME 2%, late ME 6%, EModE 4%), but with a trough in
early ME followed by a peak in late ME (a similar pattern to the one observed
in the present chapter). Thus, attendant ambivalence does not appear to have
followed a steady trend or to have been particularly damaging for the lifespan
of the be-perfect. To the extent that ambivalent cases did gradually decline
with the decline of the be-perfect, one can never be certain whether that was a
contributing factor/cause/desideratum or just a natural side effect of the dying
out of the construction itself. The data suggests that ambivalence was not at
all dysfunctional or detrimental to efficient communication. On the contrary,
it appears to have been even desirable in certain circumstances, skillfully em-
ployed and manipulated by a linguistic genius such as Shakespeare. It is now
time to proceed to a comedy written a century later than The Taming of the
Shrew.
5.5 The Seventeenth Century: Restoration Comedy

Moving into the 17th century, the next sample comes from a Restoration come-
dy, William Wycherley’s 1675 play The Country Wife. The source used is Salgādo’s
(1968a) edition. Like Chaucer before him, Wycherley had connections to the
court and moved in high circles and so might have written with an upper-class
audience in mind. However, when introducing the genre, Salgādo (1968b: 25,
28, 34) notes that there is a realistic side to it, even though it might have some
escapist elements too. According to critics, this type of comedy gives a ‘vivid
sense of contemporary life’ and has a ‘consistently secular temper’, as well as
realistically portraying ‘the social conflict between the Royalist aristocracy and
the increasingly prosperous Puritan citizenry’ (Salgādo 1968b: 25). Although
much Restoration comedy deals with high society, there aren’t any especially
high-born characters in The Country Wife, in keeping with the other texts ex-
amined so far; once more, we are in for some tricks and contrivances, as well
184 chapter 5
table 4 Perfect, passive, progressive and modal auxiliaries in the first two acts of
Wycherley’s The Country Wife

Have-perfects with transitive verbs & with copula 52 33.8%

and auxiliary be
Ambiguous have-perfects 1 0.7%
Progressives 5 3.3%
Modal be 5 3.3%
clause
’s contraction 2 1.3%
as plenty of wittiness, obscenity, sexual relationships, adultery, and innuendo.

The Country Wife is additionally distinctly satirical, even somewhat grotesque,
keeping up the tone of the three pieces investigated above.

Only the first two acts of the play were examined, which supplied approxi-
mately 10,083 words, based on the following website: http://quod.lib.umich
.edu/e/eebodemo/ALJ0561.0001.001/1:4.2?rgn=div2;view=fulltext (accessed on
11 August 2016). Older spelling conventions were used on the website, though
here I follow Salgādo’s (1968a) modernised orthography. Page number refer-
ences are to Salgādo’s (1968a) edition. Importantly for present purposes, this
text is not featured in the Helsinki Corpus, on which some previous research is
based, but the figures from it confirm the results obtained by earlier scholars,
as shown in Table 4. They are also strikingly consistent with the previous data-
sets presented above, which they all but replicate in terms of relative shares of
each type of structure.
By the 17th century, the have-perfect should have come into its own, and for
the most part it has, though there are still occasional examples of non-perfect
constructions where a perfect would be expected today. (150) illustrates the

non-occurrence of Modern English’s so-called inclusive perfect for persistent
situations (see Chapter 3, Section 3.1.1; cf. languages like German or Bulgarian,
which would still use the present (simple) in equivalents of (150)). Sentences
like this have of course not been counted but they confirm that the temporal
and aspectual system was still undergoing major shifts in the 17th century.
(150) and he’s a sign of a man, you know, since he came out of France. [I.i,
p. 162] [instead of ‘has been a sign of a man’]
The subordinate temporal clause introduced by since in (150) creates an expec-

tation for a perfect in the main clause, but instead we find simple ‘is’. Similarly,
today a perfect would be usual in standard English with never, in the sense of
‘not once up until now’, also coupled with till now in the same clause (see Quirk
et al. 1985: 194–195; cf. Miller 2000, Brown and Miller 2016: 247, for exceptions).
Instead, in (151) Wycherley opts for the simple past, which would still be ac-
ceptable in certain varieties, probably including American English.
(151) and till now I never thought I should have envied you or any man about
to marry [II.i, p. 172]
Similarly to the texts examined earlier, the present simple might still be used
in the passive, as in (152), in order to avoid a combination of a perfect passive,
despite the meanings of anteriority and current relevance (cf. Chapter 3).
(152) The writings are drawn, sir, settlements made [II.i, p. 174] [instead of
‘have been drawn’]
Conversely, specific past time reference is allowed with a present perfect, as

perhaps demonstrated in (158) below, though yesterday might more plausibly
be confined in scope to the subject NP (see also (39) above).
As in the previous samples, there were numerous instances of have-perfects
of typically transitive verbs like fight, buy, and tell, (153)–(154), copular or aux-
iliary be, (154), as well as verbs governing prepositional complements, as in
(155)–(156) – a total of 52.
(153) But though our Bayes’s battles oft I’ve fought,

And with bruised knuckles their dear conquests bought [Prologue,
p. 149]
186 chapter 5
(154) Madam, you would not have been so little generous as to have told him.
[II.i. p. 175]
(155) Thou hast stared upon her enough to resolve me. [II.i, p. 172]
(156) Have you agreed on matters? [II.i, p. 184]
Example (154) further demonstrates a case of ‘agreement’ or ‘attraction’/

‘contamination’ – the infinitival to have told has been influenced by the preced-
ing perfect, where simple to tell would work, too (more on this to follow). This
‘vacuous’ use of a non-finite perfect construction is indicative of a high degree
of grammaticalisation.
Unlike the earlier texts, where word order was rather flexible, it is now
rather fixed and, as a rule, objects tend not to separate the auxiliary and the
main verb. The only partial exception to this involves a case of ellipsis in a
verse passage, (153) above, as well as (163) below. (Readers will remember that
occurrences where one auxiliary applies to two participles were counted as
one token.) In the prose dialogue, I do not record any objects that have been
inserted between the auxiliary and the main verb, even when the word order
is unusual from a present-day perspective, (157), or when there are other par-
enthetical items, (158).
(157) I have left at Whitehall an еarl to dine with you [I.i, p. 162]
(158) your appearance at the play yesterday has, I hope, hardened you for the
future [I.i, p. 158]
Another feature that sets this 17th-century drama sample apart is the fact that
for the first time the material yields the perfect of auxiliary or substitute do:
(159) for he would but ruin you, as he has done hundreds [II.i, p. 171]
Greater grammaticalisation is demonstrated by the occurrence of perfect pas-

sives, (160) (as in Shakespeare), including non-finite ones, (161), and also the
perfect in combination with the progressive, (162).
(160) Well, I have been hired by young gallants [I.i, p. 154]
(161) What dost thou blush at nine and forty, for having been seen with a
wench? [I.i, p. 166]
(162) besides he has been making love to me. [II.i, p. 175]
In the light of the comments regarding word order, (163) was deemed ambigu-
ous. It comes from one of the few lines of verse in the text, so the alignment in
which the object intervenes between the auxiliary and the main verb might be
due to metric or rhyming considerations:
(163) Who for his business from his wife will run,
Takes the best care, to have her business done. [II.i, p. 185]
Therefore, this could be a genuine perfect with a word order that is not found
elsewhere in the first two acts of this play (of the type characteristic of the ear-
lier verse samples). Alternatively, it could be a causative or quasi-passive con-
struction. In support of this reading, the context implies that someone else will
do the wife’s ‘business’ instead of the husband. Crucially, this deviance in my
drama database confirms that, even at this late stage, it is not only be-perfects
that were plagued by uncertainty of interpretation (for the ambiguity of have-
perfects, cf. the sections above and the previous chapter on OE).
In general, here and in the other samples, transitives by far outnumber
intransitives, which might have contributed to the demise of be-perfects, as
already discussed in the previous sections and chapters. It is interesting to
ponder whether people talk more about events involving more than one par-
ticipant. There were only four properly intransitive have-perfects. The first
three, (164)–(166), are of verbs which are not typically mutative. What is more,
the hypothetical (and non-finite, after the modal) context in (165) and (166)
favours the use of have, as noted in Chapter 3.
(164) Pinchwife: I have dined already.

Horner: Come, I know thou hast not. [I.i, p. 166]
(165) but I would have ventured for all that. [II.i, p. 168]
(166) and I would as soon have lain with him, as have named him.
[II.i, p. 185]
The sentence in (167) shows that have can displace be even with verbs such
as go, though I found only one instance of this, as opposed to many more in
which go combines with be (see below and compare similar examples from
the earlier case studies of Chaucer and Secunda Pastorum, so this can hardly
be called a precedent).
188 chapter 5
(167) I’ve gone too far, and slipped before I was aware [II.i, p. 171]
There were still quite a few be-perfects, three times more than intransitive per-
fects with have. No fewer than four of those contain the phrase let’s be gone, as
in (168)–(169).
(168) Pray, Sir Jasper, let us be gone from this rude fellow. [I.i, p. 155]
(169) Why do we stay? Let’s be gone; I can’t indure the sight of him. [I.i, 156]
No anteriority seems to be implied here and by this stage be gone is in all prob-
ability a fossilised fixed phrase which is no longer a genuine perfect formed on
productive syntactic principles. Indeed, it was often written as a single word
in this period (Steven Kaye, p.c.). Apart from the four instances of let’s be gone,
two similar examples might at first suggest that the perfect here is not that
moribund, since it is not invariably preceded by let’s, as in (170) (cf. imperative
have done [with words] from example (90), Section 5.4.2).
(170) Do not be frightened; but pray, be gone ladies. [II.i, p. 177]
Nevertheless, speakers may have felt that be gone was no longer a bona fide
perfect, but a fixed phrase which is hard to decompose as a sequence of verbs,
making it a copular adjectival construction with a stative meaning, or even
a compound verb. This much is suggested by substituting get for be in lines
which appear in close proximity, giving the clause a more dynamic feel:
(171) Pray get you gone ladies [II.i, p. 177]
(172) get you gone to your business together [II.i, p. 185]
(171) and (172) have an adjectival participial object complement, which might
hint that be gone was already adjectival in status too, with the participle acting
as subject complement (cf. previous section). Again, no anteriority is indicated
by be gone in (173), and the modal must arguably shows that be gone behaves
like one (compound?) verb.
(173) Come, is your ladyship reconciled to him yet? Have you agreed on
matters? For I must be gone to Whitehall. [II.i, p. 184]
As expected, there are also more typical be-perfects with come and go,
the mainstay of this type of construction which can be found as late as the
19th century (cf. Chapter 3 and the Epilogue below):
(174) I can be sure she that shews an aversion to me loves the sport, as those
women that are gone, whom I warrant to be right [I.i, p. 157]
(175) We are come to wait upon her to the new play. [II.i, p. 177]
(176) She has locked the door and is gone abroad. [II.i, p. 177]
Abroad ought to be the predicative complement in (176), ensuring a verbal

(as opposed to adjectival) interpretation for gone. There were also instances
where no literal motion is implied, (177)–(178), suggesting that the be-perfect is
still a viable category at this time, though there are already signs of its gradual
fossilisation and ever shrinking domain.
(177) Fie, fie upon ‘em! They are come to think cross breeding for themselves
best, as well as for their dogs and horses. [II.i, p. 178]
(178) and it is come to that pass, we dare not take your words, no more
than your tailor’s, without some staid servant of yours be bound with you
[II.i, p. 184]
Finally, it should be noted that (179) can be nothing but a perfect of copular
grow, equivalent to ‘has become’. A passive reading is easily ruled out in con-
text (as in the previous periods).
(179) So the little fellow is grown a private person with her. [II.i, p. 179]
The proliferation of functions of both be and have is demonstrated by the

use of the progressive, a total of 5 instances, including the perfect progressive
showcased in (162) above, as well as by modal have and modal be. As with the
perfect, the progressive is sometimes not used where it would be expected
today – cf. (180) and (181), as well as the analogous comments from the previ-
ous section.
(180) Horner’s coming to carry you away. [I.i, p. 153]

190 chapter 5
(181) Come, Horner does not come [II.i, p. 176] [instead of ‘is not coming’]
Modal have is represented by had better, (182), and a similar construction with
as good in (183).
(182) and he had better employ her, than let her employ herself [I.i, p. 156]
(183) Well, I see one had as good go to law without a witness, as break a jest
without a laughter on one’s side. [I.i, p. 162]
(184) is a credible causative, paraphrasable with ‘I would therefore cause you to

go with …’/‘I would therefore arrange for you to go with …’.
(184) Sir Jasper: Well, well, in short I have business at Whitehall, and
cannot go to the play with you, therefore would have you go –
Lady Fidget: With those two to a play? [II.i, p. 180]
(185) and (186) also contain some tinge of (semantic) modality in them, most
plausibly volition. These instances of have do not take another verb and are
best treated as more or less delexicalised, albeit without being accorded aux-
iliary status.
(185) So, so, why, ‘tis as I’d have it. [II.i, p. 185]
(186) And as I’d have it. [II.i, p. 185]
Modal be was represented by the be-to construction. It has distinctively deon-

tic overtones, (187), or might indicate (im)possibility (188)–(189).
(187) Well, but am I to give thee joy [I.i, p. 163]
(188) Shy husbands and keepers like old rooks [‘swindlers’] are not to be
cheated, but by a new unpractised trick [I.i, p. 154]
(189) Do not teach my wife where the men are to be found. [II.i, p. 169]
Sometimes, it might predominantly serve to mark futurity, as in (190)–(191):
(190) Why, the next thing that is to be heard is, thou’rt a cuckold. [I.i, p. 163]
(191) This is one of those, my pretty rogue, that are to dance at your wedding
tomorrow [II.i, p. 172]
Even if the construction is taken to express pure futurity, without deontic over-
tones, futurity is still in some sense related to modality, since prediction can
hardly be divorced from expressing the speaker/writer’s attitude to the propo-
sition, rather than merely reporting a fact.
Once again, it was often exceedingly difficult to judge whether something
was a dynamic passive or a clause that is indeterminately passive or copular,
so the decisions made are sometimes arbitrary and open to debate. I finally
settled on 50 dynamic passives vs. 15 instances of ambiguity (see Appendix 6).19
Very frequently, the same verb or construction might be classified differently.
For (192)–(196), the context imposes a dynamic passive reading, where there is
a single occurrence of an activity on a particular occasion:
(192) Some unbelieving sisters of my former acquaintance, who I am

afraid expect their sense should be satisfied of the falsity of the report.
[I.i, p. 154]
(193) Won’t you be acquainted with her, sir? [I.i, p. 155]
(194) … wife, who before would not be acquainted with me out of jealousy
[I.i, p. 157]
(195) And so go to her, begin thy new employment. Wheedle her, jest with her,
and be better acquainted one with another. [II.i, p. 183]
(196) First, I shall be rid of all my old acquaintances [I.i, p. 157]
(194), for instance, was taken to mean that the wife refused to be introduced
to the speaker (the punctual action), rather than not wanting to be his ac-
quaintance (the state). For (196), a stative ‘I shall be free of …’ does not sound
convincing in context. By contrast, (197)–(199) were treated as ambiguously
stative, since no comparable dynamic readings were unequivocally imposed
by the context, and adjectival paraphrases were readily available, as indicated
in the brackets after each example sentence:
192 chapter 5
(197) I’m satisfied you are of the society of the wits and railleurs [‘banter-
ers’] [II.i, p. 173] [= ‘I’m willing to believe that you are …’ or, less likely,
‘I’m happy/glad that …’]
(198) Nay, by this means you may be the more acquainted with the husbands,
but the less with the wives. [I.i, p. 157] [= ‘more familiar’]
(199) And next to the pleasure of making a new mistress is that of being rid of
an old one [I.i, p. 157] [= ‘of being free of’]
(200) was treated as closer to the adjectival end of the scale mainly because of
the adverb well (though compare better in (195)), and the available paraphrase
with the adjective popular:
(200) I have given you security already to save you harmless, my late
reputation being so well known in the world, madam. [II.i, p. 184] [= ‘so
popular in the world’]
A passive reading might be flagged up by the presence of a PP introducing the

agent, sometimes with of instead of the by usual today:
(201) a woman is hated by ‘em as much for loving her husband as for loving
their money [I.i, p. 155]
(202) I think that was prettily said of me, ha, Harcourt? [II.i, p. 173]
When it was nearly impossible to decide, it was usually preferred to err in

favour of the passive (as before; cf. (195) too):
(203) ‘Tis an errant shame women of quality should be so slighted

[II.i, p. 178] [i.e. verbal ‘offended’, on a single occasion]
(204) and the crime’s the less when ‘tis not known. [II.i, p. 179]
Moving on to the rivalry between passive and be-perfect, the following

pair of sentences were judged to be exclusively passive, since they cannot
be perfect:
(205) Well, that’s spoken again like a man of honour [II.i, p. 184]
(206) If I did, nobody would believe me. The reputation of impotency is as

hardly recovered again in the world as that of cowardice, dear madam.
[II.i, p. 184]
The context (semantico-syntactic, i.e. the inanimate subjects, as well as

semantico-pragmatic, i.e. the overall meaning in the discourse) disambigu-
ates them and they can only be passive, not perfect. (207) is also an unlikely
candidate for a be-perfect due to the meaning it conveys in context (cf. earlier
discussion and note again the persistence of such supposedly poorly adapted
items):
(207) we are your friends, and will not take it ill to be left, I do assure you.
[I.i, p. 162]
I found only a single instance of ambiguity between be-perfect and passive or

stative – (208). Even this is highly suspect.
(208) Hold, do not rail at him, for since he is like [‘likely’?] to be my husband,
I am resolved to like him. Nay, I think I am obliged to tell him you are not
his friend. [II.i, p. 175]
As emerged from the discussion in Section 5.4 above, example (208) should
perhaps be removed from the count, since it is most probably adjectival.
Resolved is clearly parallel to the more convincingly adjectival obliged, though
Wycherley uses it with an object too (to resolve sb., see Appendix 6),20 so (208)
might be passive. As it is, resolved also takes a complement – the to-infinitive
clause. So, were this to be a perfect, in all likelihood it should be formed with
have anyway. As in the previous sections, this was still put down as ambiguous,
despite the remote possibility, in order to ensure that my data is not skewed in
favour of the hypothesis I am trying to prove.
There were only two instances of missing auxiliaries in a non-conjoined
environment:
(209) for you look upon a friend married as one gone into a monastery, that is,
dead to the world [II.i, p. 173]
194 chapter 5
(210) my wife has just now the smallpox come out upon her [II.i, p. 177]
Based on the material above, the chances are that these ought to be construed
as combining with be, though there was a minority example of have gone too.
There were another two cases of ambiguity due to contraction of the auxiliary,
(211)–(212):
(211) the vizard-masks [‘prostitutes’], you know, never pitty a man when all’s
gone [I.i, p. 158]
(212) He’s come newly to town it seems, and has not heard how things are with
him. [I.i, p. 165]
Elsewhere in this portion of the text, ’s only instantiates different uses of is,
never of has (cf. the comment about ’s in pre-18th-century English made in the
previous section and Rydén and Brorström 1987: 32). I have not found a single
instance of ’s in the first two acts of the play where it unambiguously stands for
has; it could conceivably stand for has only in the two examples in (211)–(212),
but in all the other numerous cases, it certainly represents is, mostly the copula
or, on a few occasions, the progressive or passive auxiliary (see (180) and (205)
above). Secondly, as has been pointed out already, come and go tend to almost
exclusively combine with be to form their perfect. All of these factors contrib-
ute to disambiguation in the context of the discourse and favour be over have.
Needless to say, the loss of be as perfect auxiliary has of course in no way af-
fected the continued usage of such contracted or elliptical constructions: The
late visitors were Aunt and Uncle Israel Fish, come straight from the opera, still
in their evening clothes … [2003, The Great Husband Hunt by Laurie Graham].
It probably never mattered to speakers whether there was only one option for
spelling out the ‘complete’ version (with have), or more than one (be or have).
Finally, it should be mentioned that some combinations were excluded
from all the counts. For instance, do not be frightened does not belong in any of
my categories because the context rules out even a passive reading (?do not be
frightened by anyone). Similarly, (213) is not included in any of the categories,
as it is neither a passive derivable from Someone mistook the man, nor a perfect
like A man has often mistaken. It is adjectival only (be resolved from (208) above
probably belongs here too!). The same applies to (214).
(213) a man is often mistaken [I.i, p. 157] [= ‘is often wrong’]

(214) to try whether I was concerned enough for you [II.i, p. 176]
[= ‘was anxious’]
In conclusion, most of the ambiguity again lies in the grey area between genu-
ine passive vs. stative copular clauses. In the 17th century too, ambiguous be-
perfects seem to be statistically negligible, with the number of unclear cases
reduced even further by context. The relative proportion of contracted and
missing auxiliaries has risen, but the overall picture is the same, with indica-
tions that contraction should actually favour be. What is more, there is some
uncertainty attending the supposedly clearer and functionally superior have-
perfect, a residue of ambivalence paralleled in Old English (see Chapter 4).
Against this backdrop of consistently underwhelming amounts of ambiguity
of perfect be, The Country Wife offers one more sample which demonstrates
the proliferation of the functions of both auxiliaries under investigation. In line
with the findings of previous scholars, I document four centuries of reasonable
stability in the relative shares of the major functions of the two auxiliaries,
with the role of ambiguity being at best indecisive at all times and with little
perceptible fluctuation or indisputable directionality in one way or another:
4.2% of unclear be-perfects in Chaucer (from all sources combined), 1.9% in
Secunda Pastorum, 5.5% in Shakespeare, and 3.3% in Restoration comedy.
5.6 Late Middle English Correspondence: the Paston Letters

In order to broaden the database, I decided to inspect a sample of the Paston
Letters from the 15th century. This series of family and business letters begins
in the 1420s and spans four generations, or about one hundred years. The fami-
ly originated in the village of Paston, in East Anglia, and gradually rose through
the social ranks. William Paston, a.k.a. William Paston I (1378–1444), went to
school and founded the family fortunes after a successful career at law. His
son John Paston, a.k.a. John Paston I (1421–1466), studied in Cambridge and
the Inner Temple and, like his father, had a career in law too (see Burnley 1992:
177). I will be examining William Paston I’s letters from the 1420s and 1430s, and
then those of his grandson, William Paston III, son of John Paston, which were
written between 1478 and 1492. The texts used here are the ones which ap-
pear in the Corpus of Middle English Prose and Verse, at http://quod.lib.umich
.edu/c/cme/Paston?rgn=main;view=toc (accessed on 22 August 2016).
196 chapter 5

5.6.2.1 William Paston I
William Paston I’s letters written or drafted between 1425 and 1436 usually deal
with domestic or legal matters, and run to a total of 5,343 words (http://quod
.lib.umich.edu/c/cme/Paston/1:1?rgn=div1;view=fulltext, accessed on 22 August
2016). Table 5 confirms the relative rates that are quite familiar by now, and
also that be serves predominantly as a passive auxiliary, 41 times, while have is
the perfect auxiliary par excellence – there are a total of 34 have-perfects, only
two of which involve an intransitive verb. Intransitive verbs are vanishingly
rare in this sample when compared to transitives, especially if the opposition
is conceived of as one between mutitative intransitives on the one hand ver-
sus all the rest of the verbs on the other. Ambiguity is again mostly limited to
dynamic passive vs. stative copular readings, 9 instances, with the number of
ambiguous be-perfects more or less the same as in the drama samples – 3 cases
of indeterminacy between a be-perfect and a passive or stative clause, and
2 missing auxiliaries. Crucially, although there are hardly any be-perfects,
ambiguity stays the same, i.e. it seems to be unaffected by the apparent elimi-
nation of be-perfects from this collection of correspondence.
table 5 Perfect, passive, progressive and modal auxiliaries in the letters of William
Paston I


copula be
Progressives 0 0%
Modal be 0 0%
clause
’s contraction 0 0%
In these non-verse texts, the auxiliary and the main verb in have-perfect
constructions tend to stay together and not to be separated by the object.
There are a couple of exceptions in which nought comes between the auxiliary
and the lexical verb, but it quickly becomes clear that nought has by this stage
been decategorialised as a marker of negation, no longer the original reinforc-
ing noun/pronoun which meant ‘nothing, not a bit’. The object in (215) below is
probably the preposed more or perhaps this is the have-[something]-to-do-with
pattern with the object missing and more simply an adjunct, while trespass in
(216) is intransitive, with the PP headed by against possibly acting as a prepo-
sitional complement.
(215) and more I haue nought hadde to do with þe seyd John [PROBABLY TO
MASTER JOHN URRY: DRAFT 1425, 11]
(216) I haue nought trespassed a-geyn noon of these iij [TO WILLIAM
WORSTEDE, JOHN LONGHAM, AND PIERS SHELTON 1426, 03, 01]
Apart from those, there was only one more example of unusual word order in
a non-relative and non-interrogative clause:
(217) þere hem for to have hangyd; [MEMORANDUM TO ARBITRATORS

1426–7]
The object hem ‘them’ is preposed in (217), whereas a potential reflexive pro-
nominal object is altogether missing in (218) below:
(218) And also þe Priour of þe seyd hous [unclear] hath resigned in-to yowre
worthy handes by certeins notables and resonables causes [TO THE
VICAR OF THE ABBOT OF CLUNY: DRAFT 1430, 04?]
An example of a verb with a PP complement which was grouped with the tran-
sitives is given in (219), whereas (220) shows the perfect of causative do, haue +
doon, followed by another past participle, examyned: a causative construction
which was unambiguous but was outlived by the likes of causative have, pre-
sented in (221) (see Denison 1993: Section 10.2.2.5).
(219) and þat þe same John, atte reuerence of yowr right worthy persone,
hath cesed of his sute of certeins processes ageyns me [PROBABLY TO
MASTER JOHN URRY: DRAFT 1425, 11]21
21 It cannot be ruled out that this is actually a phrasal verb, i.e. cease off.
198 chapter 5
(220) I haue, after þe aduys of yowr lettre, doon dwely examyned þe jnstru-
ment by þe wysest I coude fynde here [TO WILLIAM WORSTEDE, JOHN
LONGHAM, AND PIERS SHELTON 1426, 03, 01]
(221) and how sone I myght haue the seid stoon caried to Moneslé a-forn seid
[To JOHN STANFORD 1425]
(221) is most likely causative and not perfect in the light of the observations
about word order made above. Crucially, neither of the two intransitive have-
perfects, with trespass in (216) above and the one in (222) below, would have
been in competition with be.22
(222) monkes clad and vn-professyd, þat have abedyn þere wythow [unclear] t
abbyte ix or x yeere, [TO THE VICAR OF THE ABBOT OF CLUNY: DRAFT
1430, 04?]
The only unambiguous be-perfect with a mutative intransitive is given in (223):
(223) Myn aduersarie is become bysshop of Cork in Irland [TO WILLIAM

WORSTEDE, JOHN LONGHAM, AND PIERS SHELTON 1426, 03, 01]
By contrast, there were three seemingly ambiguous be-perfects. The first of

these should probably be removed, however:
(224) ne yet ne dar not, rydyn ne goo abowte swyche occupacion as he arn vsed
and disposed [MEMORANDUM TO ARBITRATORS 1426–7]
They are used and disposed would make no sense as passives. The anteriority
entailed by the perfect would not fit here either, as the reference is to the pres-
ent time. Used and disposed should therefore be regarded as adjectival. The
remaining two cases are equally suspect:
(225) And here aunswere is þat al þis processe, þough it were in dede proceded
as þe instrument specifieth, is not suffisant in þe lawe of Holy Cherche
[TO WILLIAM WORSTEDE, JOHN LONGHAM, AND PIERS SHELTON
1426, 03, 01]
22 If the PP headed by against in (216) is a genuine complement, this example had better be
transferred to the group of transitives.
(226) in whiche sute in þe seyd court it is proceded ageyn þe seyd Adam

in other maner þanne othyr lawe, conscience, or good feyth wolde
[MEMORANDUM TO ARBITRATORS 1426–7]
Proceed does not normally take an object, but the dictionary entry below might
suggest that (225) and (226) are impersonal passives, which means that they are
not strictly ambiguous. According to the electronic Middle English Dictionary
(http://quod.lib.umich.edu/cgi/m/mec/med-idx?type=id&id=MED34752, ac-
cessed on 23 August 2016), in its most common sense, proceed means ‘to go or
move in a certain direction’ or ‘to continue in an action or process’ – a typical
mutative intransitive. The dictionary then gives the passive-looking phrases:
be hit proceded ‘let (the incision, ligature, ?procedure) be continued’, as well as
hit is proceded ‘it (a ligature) is continued’; that ... hit be proceded ‘that ... it (a
procedure) be carried on’. That this type of proceed is most probably passive is
indicated by the Latin original supplied in (227). Latin procedatur is a 3rd per-
son singular passive (in the present subjunctive) ending in -tur.
(227) Þe inscision forsoþ ydone, be yt proceded [Ch. (2): be it iprocedede;

L procedatur], puttyng in a tent infused in hote oile of roses. [?a1425
*Chauliac(1) (NY 12) 76b/b, http://quod.lib.umich.edu/cgi/m/mec/
med-idx?type=id&id=MED34752&egs=all&egdisplay=open, accessed
on 24 August 2016]
As confirmed by the same dictionary, it is proceeded can also mean ‘it is argued’,
including in a legal sense, which is probably the sense intended in William
Paston’s correspondence. Indeed, the dictionary explains the self-same exam-
ple in (226), ‘it is proceeded against the said Adam’, as follows: ben proceded ‘of
legal actions: to be carried out, put into effect; -- also impers.’ (http://quod.lib.
umich.edu/cgi/m/mec/med-idx?type=id&id=MED34752&egs=all&egdisplay=
open, accessed on 24 August 2016). So, upon careful examination, even those
cases of possible uncertainty turn out to be spurious, but they are in this cat-
egory due to the potential of proceed to build perfects and to passivise.
Finally, below are the two instances of ambiguity due to a missing auxiliary
(with the participles (?) passed and n-less falle[n]), neither of which is very
certain:
(228) A prest of Norffolk þat spak with yow in Jull or August last passed told me
þat … [TO AN UNIDENTIFIED LAWYER IN ROME: DRAFT 1425, 11, 05]
200 chapter 5
(229) where þe seyd William þe tyme of þe seyd enformacion was, wyth Ser John
Jermy, knyght, and othre of þe counseill of þe seyd Duk of Norffolk in hys
lordshipes in Norffolk and Suffolk þanne to hym falle by þe deth of þe right
worthy and noble lady hys modyr [MEMORANDUM TO ARBITRATORS
1426–7]
In sum, this correspondence from the early 15th century confirms my findings
above in terms of the relative share of the uses of be and have, as well as the in-
significant part that ambiguity seems to play (cf. esp. the similar percentages in
the contemporary tables for Chaucer and the Second Shepherds’ Play). I now
move on to the correspondence of William’s grandson.
5.6.2.2 William Paston III

These letters were written or drafted between 1478 and 1492, the year when
Columbus discovered America. William Paston III’s correspondence deals
with both private and public matters, and one of the letters is actually a copy
of a message from King Henry VII to the Earl of Oxford. The total word count is
4,616 words (http://quod.lib.umich.edu/c/cme/Paston/1:13?rgn=div1;view=fullt
ext, accessed on 25 August 2016). Table 6 is another confirmation of the stable
table 6 Perfect, passive, progressive and modal auxiliaries in the letters of William
Paston III


and auxiliary be
Progressives 4 3.5%
Modal be 1 0.9%
Ambiguity between passive and a stative copular 16 14%
clause
Missing auxiliary in a non-coordinated context 2? 1.8%
relative shares of the auxiliary uses of the two verbs under investigation. The
proportions are strikingly similar to everything analysed so far. It is once again
easy to discern the proliferation of functions mentioned above (also compared
to the correspondence of the earlier generation of Pastons), though the hege-
mony of perfect have and passive be remains undeniable. It should be noted
that ‘hot news’ contexts typical of letters might favour the use of the present
perfect, but its numbers do not seem to be significantly greater than they are
in other types of comparable text. There are 35 have-perfects of transitive or
copular verbs here, including auxiliary be. The perfect auxiliary and the past
participle are not normally separated by the object, as in (230), with only one
exception, (231):
(230) letynge yow wete that I haue reseuyd of Alwedyr a lettyr and a nobyll in
gowlde þer-in. [TO JOHN PASTON III 1478, 11, 07]
(231) Also ȝe sende me worde in the letter of xij li. fyggys and viij li. reysons.
I haue them nott delyuered, but I dowte not I schal haue [TO JOHN
PASTON III 1479, 02, 23]
The example in (231) might be a curious stative, i.e. ‘I don’t have them in my
possession yet, in a delivered state’ or a have-based passive; it also illustrates
an auxiliary in an elliptical construction. A PP complement (as opposed to a
direct object) is illustrated in (232):
(232) and they ferde as thow they wolde not haue medylde wyth them
[TO JOHN PASTON III 1488, 05, 13]
In the earlier discussion (see Sections 5.4.2 and 5.5.2), I cast doubt on the vi-
ability of some be-perfects due to anomalous temporal reference in a context
where no anteriority is entailed, e.g. I must be gone [to Whitehall] (attested in
the 16th and 17th centuries), with be gone behaving like a single compound
verb. Here I find confirmation that time-reference issues can arise with have-
perfects too (already mentioned in passing while discussing example (154)
from Section 5.5.2 above). According to one possible reading of (233) below,
he sent his lordship a sum of money via a servant [in order] to have excused
him[self], but the perfect is not strictly necessary; [in order] to excuse him[self ]
would do just as well.
(233) He sent my lorde be a seruaunt of hys xl li. to haue excusyid hym, and it
wolde not be takyn; [TO JOHN PASTON III 1489, 05, 06–10]
202 chapter 5
I further record combinations of modal + perfect + passive, again with the

perfect marked (copied?) redundantly in a following infinitival clause without
implying proper anteriority, as in (234). Denison (1998: 140–142) notes that the
perfect in such contexts signals unreality.
(234) in hope þat they schuld haue ben lycensyd to haue gone ouer
It turns out that both auxiliaries face questions of defective temporal mark-
ing. However, while must be gone has lost all traces of anterior meaning and is
hardly a perfect any more, seemingly redundant have-perfects still signal
unreality or anteriority, albeit vacuously, arguably retaining more of their
functionality.
There were six have-perfects with intransitive verbs, demonstrating the dis-
placement of be by have in this domain. Four of them are with the verb go,
(234)–(237), and one with come, (238) – those could hardly have caused much
ambiguity if constructed with be. There is also one with sail, (239).
(235) where as it was seyde that my lord Woddevyle and other schulde haue
gone ouer in-to Breten to haue eyded the Duke of Breten [TO JOHN
PASTON III 1488, 05, 13]
(236) and soo whan he was countyrmaundyd thos that resortyd thedyr
to haue gon ouer wyth hym taryde there styll [TO JOHN PASTON III
1488, 05, 13]
(237) And also I had went to haue had folkys a mette wyth me at Hedyngham
whyche dednott. [TO JOHN PASTON III 1492, 02, 18]
(238) how that I was sore disesyd, notwythestondyng I was welewyllyd to haue
come to fulfyll my promesse [TO JOHN PASTON III 1492, 02, 18]
(239) And they had nott seylyd not paste vj legys butt they aspied a Frencheman
As pointed out already (see Chapter 3) and aptly illustrated by the material
here, contexts of modality and non-finiteness favour have over be. The same
abnormal temporal reference is detectable in ‘resorted to have gone over’ in
(236), and in most of the others. (237) and (239) have a past perfect, another
context which paved the way for the displacement. It is possible, however, that
went might be a form of the verb wenen ‘to believe, opine, suppose’ (see http://
quod.lib.umich.edu/cgi/m/mec/med-idx?type=id&id=MED52207, accessed
on 26 August 2016). This much is suggested by the fact that the past participle
of go used by William Paston III is otherwise gone, not went.
There were four unambiguous be-perfects, three of them with the resilient
verb come:
(240) there was ij c of them that gete them in-to a Breten schyppe þe whyche
was late come ouer wyth salte [TO JOHN PASTON III 1488, 05, 13]
(241) we wryte vnto you of them as thay [i.e. ‘tidings’] be comen to oure know-
lage, and as foloueth. [TO JOHN PASTON III, SENDING COPY OF A
LETTER FROM HENRY VII TO THE EARL OF OXFORD 1489, 04]
(242) And where as ye desyre that I schulde sende yow worde of suche tydyng
as Phylyp Lewes and Wyndesor bryngythe fro the corte, they be come
thens bothe [TO JOHN PASTON III 1489, 05, 06–10]
(243) and Phylyp Lewes is redyn ageyn to the Kyng, and schall brynge wyth
hym money for all ther wagys that schall be in my lordys retynew
[TO JOHN PASTON III 1489, 05, 06–10]
It is worth noting that (240) takes be even in the pluperfect. (241) uses come in a
metaphorical, non-spatial sense. Crucially, (243) has a less usual verb – to ride,
pointing to a certain productivity of the be-perfect. And yet, it might again be
tentatively suggested that the advanced rate of displacement here might be
due to geography and (earlier) contact. The Pastons come from Norfolk, which
is an area settled by Viking invaders – note the use of ON they, them, their in
(240)–(243) (though, of course, those Scandinavian forms had been diffused
to most areas by the 15th c. and the earlier generation of Pastons has OE he
‘they?’ and hem, e.g. in (224) and (217), respectively). Perhaps more tellingly, the
ON preposition fro ‘from’ shows up in (242), though alongside the Southern/
Midland -eth ending for the third person singular and plural present. Like the
Peterborough Chronicle and Secunda Pastorum, the Pastons’ periphrastic out-
put appears innovative in comparison with later texts such as The Country Wife
(cf. Rainer 1989 for the innovative use of the perfect in letters from the north).
This fits the contact scenario outlined in earlier chapters, with the Pastons
being on the border of the northern territory, which seems to be reflected in
their somewhat mixed language. The data on the rates of be-perfect loss in my
samples consistently points to a diffusion of this change from north to south.
204 chapter 5
Perfects aside, arguably there were four examples of nascent causative have:
(244) And where as ȝe reqwere to haue it grauntyd wyth-owt fyne or fee, I wold
ȝe had it wyth the payment of þe fyne or fee [TO JOHN PASTON III 1487,
00, 307]
(245) And he had many lordys and gentylmen to aunswere for hys trowthe
and hys demenynge afore-tyme [TO JOHN PASTON III 1487, 00, 307]
(246) Also he schewyd me that Herry Wyott wholde fynde the mene to haue yow
condemnyd and recouer þe obligacion of xl li. ageyns yow. [TO JOHN
PASTON III 1488, 05, 13]
(247) And also I had went to haue had folkys a mette wyth me at Hedyngham
whyche dednott. [TO JOHN PASTON III 1492, 02, 18]
The first of those, (244), might ambiguously be a have-perfect, which means

that the have-perfect is not 100% immune to ambiguity, as already noted above.
However, the observations on word order and the context (suggesting that it
will be someone else who will grant it, probably the king, but not the second-
person subject of the sentence) both tip the balance in favour of a causative
interpretation. (245) could simply feature lexical/‘possessive’ have, but the
context again suggests some sort of causation – i.e. he asked many lords and
gentlemen to put in a word for him. If (247) is taken to contain causative have
+ the past participle of meet, it will be an example of perfect have in combina-
tion with causative have. What is more, causative have coexists in this sample
with causative let, as in (248) below, so functionally speaking there is certainly
no gap in the system that would need to be filled.
(248) And if Bayard be on-solde I pray yow late hym be made fatte ageyns the
Kynge come in-to the contré, what so euer I pay for the kepyng of hym.
[TO THE BAILIFF OF MAUTBY 1487]
The functions of be have been enriched with the advent of the progressive, of
which there was no trace in the correspondence of William’s grandfather, but
there are now four occurrences. And yet, the example in (249) demonstrates
that both passive progressives and passive perfects are still unusual:
(249) hys tentys and alys [alleys? balustrades?] be a-makyng [instead of later
‘are being made’] faste and many of them be made [instead of later
‘have been made’]. [TO JOHN PASTON III 1492, 02, 18]23
There was, however, a progressive following modal will, (250), and just one
instance of a modal be structure, (251).
(250) soo that be lykelyod hys grace wolbe goyng sone vpon Ester. [TO JOHN
PASTON III 1492, 02, 18]
(251) ye be xxti s. in hys dette, for a monthe was to pay for when [deleted in
MS] he had mony laste [TO JOHN PASTON III 1478, 11, 07]
Unsurprisingly, dynamic passives were numerous, coming to 39 in total,

though some borderline cases like the one in (252) might have been left out –
(252) could be an existential there-clause with a participial modifier or it could
be intended to read non schall be grauntyd (passive proper).
(252) he aunswerde me that he wott well ther schall be non grauntyd of a gode
whyle [TO JOHN PASTON III 1487, 00, 307]
The following are disambiguated as dynamic by the temporal subordinate

clause, (253), and by the adjectival predicative complement/modifier dry,
(254).
(253) for I tolde hym how that ȝe were begylyd whan ȝe were at London.
(254) that þe towne schalbe dronkyn drye as Yorke was [unclear] whan the
Kynge was there. [TO JOHN PASTON III 1487, 00, 307]
In (255), coordination actually disambiguates things, rather than being confus-

ing, confirming that driven participates in a passive construction.
23 Although ‘be made’ can be paraphrased with ‘be ready’, I still regarded this as a bona fide
passive, in view of the ‘perfect passive context’ and, most of all, in order to keep the cat-
egory of genuine passives as inclusive as possible.
206 chapter 5
(255) The garnison of the towne of Concarnewe, which is oon of the grettest
strenghes of all Bretayn, was besieged in like wyse and drevyn to that ne-
cessité that thei wyth-in offerid [TO JOHN PASTON III, SENDING COPY
OF A LETTER FROM HENRY VII TO THE EARL OF OXFORD 1489, 04]
In (256), the by-phrase disambiguates one of the instances (‘is agreed’), where-
as comynd might be a present participle (‘coming’?) and therefore progressive,
or it might be passive (‘communed’?). A passive is suggested by the use of -ing
for the present participle elsewhere and by the preposition at, as opposed to to.
Communication and discussion meanings are also supported by the context.
(256) Syr Herry Heydon schewyd me that it is agreyd be Syr Edmond Bedyngfeld
that the mater betwyx hym and my brodyr Yeluerton schalbe comynd at
Norwyche, and there a dyreccion to be takyn in the same mater mete for
them bothe. [TO JOHN PASTON III 1492, 02, 18]
Again, most of the ambiguity had to do with genuine passives vs. copular
statives – 16 such cases (see Appendix 7 for examples).24 There were only three
potentially ambiguous be-perfects. The first is only marginally appropriate as
an illustration of ambivalence:
(257) latynge yow wete that, as for John Petyrsons mater, as ȝett it is nott
spedde. [TO JOHN PASTON III 1487, 00, 307]
The Middle English Dictionary gives the following explanation of the central
meaning of the verb: ‘to achieve one’s goal, accomplish one’s purpose; fulfill
one’s expectation; also, refl. be successful’ (http://quod.lib.umich.edu/cgi/m/
mec/med-idx?type=byte&byte=189697457&egdisplay=open&egs=189714944,
accessed on 25 August 2016). There are plenty of examples there in which speed
forms its perfect with have in this first sense (cf. (261) below too). Speed also
takes have in its second (intransitive) sense of ‘fare, get along’, as confirmed by
the dictionary corpus:
(258) How þou hast sped, sey me now. [a1400 (c1303) Mannyng HS (Hrl 1701)
7798; http://quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byt
e=189697457&egdisplay=open&egs=189714944, accessed on 25 August
2016]
(259) Somme wer glad þat Iason hath sped wel. [c1425 (a1420) Lydg. TB
(Aug A.4) 1.3513; http://quod.lib.umich.edu/cgi/m/mec/med-idx?type
=byte&byte=189697457&egdisplay=open&egs=189714944, accessed on
25 August 2016]
(260) By þat tyme I sall wete þe at saye How þat I hafe spedde. [c1440 (a1400)
Eglam. (Thrn) 630; http://quod.lib.umich.edu/cgi/m/mec/med-idx?
type=byte&byte=189697457&egdisplay=open&egs=189714944,
accessed on 25 August 2016]
This verb is transitive in a third sense of ‘to grant, fulfill (a prophecy, etc.)’, as
in (261)–(263). In (261), mater acts as an object (cf. (257)). Speed is also attested
in the passive phrase he shal be sped ‘his affairs shall be concluded’, while
(262)–(263) are passives with inanimate subjects and are structurally analo-
gous to (257).
(261) The legatis took her leue, as þouȝ þei had sped a grete mater.
[(a1464) Capgr. Chron. (Cmb Gg.4.12) 143/14; from the Middle English
Dictionary]
(262) Youre desyre shall be spedde. [(a1470) Malory Wks. (Win-C) 413/29;
from the Middle English Dictionary]
(263) This prophecye is now spad. [?a1475 Ludus C. (Vsp D.8) 148/58; from the
Middle English Dictionary]
This means that (257) must be passive, with (261) representing its active
counterpart. Nakamura (1987: 41–42, 48) confirms that the mutative intransi-
tive verb speed only occurs with perfect have in Samuel Pepys’s Diary. I found
no be-perfects for the intransitive senses listed in the second edition of the
OED, and this item is given as ambiguous here just because it has both tran-
sitive and intransitive uses. Taking all of the evidence into account, this first
case of ambiguity is highly suspect, as is the same verb in earlier samples (see
Section 5.4.2). The other two cases of ambiguity in the current correspondence
sample, (264)–(265), are much more convincing:
208 chapter 5
(264) As for my lord Treserer, he was not wyth the Kyng of all the counsell tyme,
þe whyche was endyd on the iijde day of Marche. [TO JOHN PASTON III
1487, 00, 307]
(265) my cosyn John Heydon tolde me that þe Prior of Waburnes horse was ri-
ally amendyd, and that þe Abott of Seynt Benetys schewed hym ther was
a bay hors off a parsons nygh onto Seynt Benetis [TO JOHN PASTON III
1492, 02, 18]
The likes of (264) have been resurfacing continuously ever since the survey of
Chaucer (and even earlier); so if this pattern was so detrimental to commu-
nication, it should have been discarded by now, rather than coming up again
and again. (265) is an ambiguous be-perfect because amend can have a transi-
tive meaning, ‘to restore (sb.) to health, make well; cure (a disease, etc.)’, or an
intransitive one, ‘to recover from illness, get well’; the ME phrase ben amended
is explained as ‘be cured, be well’ (https://quod.lib.umich.edu/cgi/m/mec/
med-idx?size=First+100&type=orths&q1=amenden&rgxp=constrained,
accessed on 17 Jan 2017).
More radically, both purported illustrations of ambiguity caused by auxil-
iary ellipsis should arguably be removed. In the first one, (266), come is not
terribly clear and against probably has the meaning of ‘in preparation for/in
relation to/regarding’ (cf. a parallel etymology for vis-á-vis), so a perfect might
make no sense, though (266) could potentially be read as ‘made fat in prepara-
tion for the King [who is/has] come into the country’ (or ‘against/in prepara-
tion for the King’s coming/arrival?’).
(266) And if Bayard be on-solde I pray yow late hym be made fatte ageyns the
Kynge come in-to the contré, what so euer I pay for the kepyng of hym.
[TO THE BAILIFF OF MAUTBY 1487]
As regards (267), all the signs, especially the final PP adjunct of time, point to
preterite forms with specific past time reference:
(267) The Lord Malpertuis, now late wyth vs in ambassade from oure dere
cousine the Duchesse of Bretayne, shippid at oure porte of Dortmouth
and arriued at Saynt Powle de Lyon in Bretayn on Palmesonday at iiij
aftere noone [TO JOHN PASTON III, SENDING COPY OF A LETTER
FROM HENRY VII TO THE EARL OF OXFORD 1489, 04]
Finally, in (268), there is no compelling reason (perhaps apart from the ad-
verb last) to consider the adjective (?) past as verbal (i.e. standing for ‘passed’),
and past has accordingly not been counted. Elsewhere in the letters, this type
of past corresponds to the clearer adjective next (but compare a different de-
cision regarding passed in (228), where the (unreliable?) spelling is arguably
‘more verbal’, though hardly making that token of passed more convincing as
an actual verbal form – probably they should both be excluded).
(268) Howbeyt they haue made proclamacions in þe cuntrey to mete wyth oder
of ther affynyté as on Tuesday last past [TO JOHN PASTON III 1489, 05,
06–10]
For the sake of comparison, in (269) I provide an example from Pepys’s Diary,
where past appears as a verb or adjective in a perfect or adjectival construc-
tion (see also Rydén and Brorström 1987: 133–140). It is now time to move on to
Pepys himself.
(269) We fearing the canonicall hour would be past before we got thither,
did with a great deal of unwillingness send away the Licence and
wedding-ring. [Samuel Pepys, Diary, 31 July 1665, quoted in Nakamura
(1987: 39)]
5.7 Seventeenth-Century Correspondence: Samuel Pepys

Since I have both comedy and correspondence sub-sections of my database
for earlier periods, it is prudent to supplement the Restoration comedy genre
with a non-fiction sample too. Samuel Pepys was a prominent figure and a suc-
cessful public official in the Restoration Period who worked his way up the
social and professional ladder. He famously kept a diary between 1660 and
1669, providing an unprecedented insight into day-to-day life in seventeenth-
century England. Sim (1997: v–vi) remarks that ‘Pepys writes well, in a simple,
unaffected manner that draws the reader effortlessly into his concerns, and the
diary has something of the impact of a novel’. Pepys’s prose has been further
described as ‘chatty and often breathless’ (Sim 1997: vii), which makes it suit-
able for linguistic research into the language of the period. It has been shown
that the Diary reflects informal speech of the time and Pepys seems to have
210 chapter 5
been rather innovative in at least some of his usage (though perhaps not of the
perfect) compared to more ‘serious’ genres (Nakamura 1987: 26).
However, it has to be noted that the Diary was originally written in short-
hand and it was only later decoded by scholars. There arise some non-trivial
issues pertaining to this – Pepys’s shorthand is not as precise as a modern
counterpart would be, which can sometimes result in ‘a degree of ambiguity as
to meaning’ (Sim 1997: viii; see also Nakamura 1987: 33 for issues to do with lack
of clarity in perfect constructions due to the original shorthand). Furthermore,
the Diary has already been the subject of pertinent research into the use of
have-perfects with mutative verbs.
Investigating the Diary and confirming earlier studies, Nakamura (1987: 23,
49, cf. Rydén and Brorström 1987: 10) has found little evidence for significant
change in the formation of the perfect of mutative verbs for almost four cen-
turies from Chaucer’s time to the end of the 17th century; consequently, the
switchover should be dated to the 18th century.25 In Nakamura’s publication
readers can find valuable data for a range of periods comparable to mine, in-
cluding statistics on Chaucer, Shakespeare and the Pastons. Nakamura (1987:
48) concludes that Pepys’s usage of the perfect with mutatives is more or less
the same as that during the periods from Chaucer to Shakespeare – with the
exception of the verbs walk, proceed, escape, speed, ride, pass and meet, the be-
perfect is preferred. This ascertainment of stability lasting for several centuries
has been corroborated here (in the samples examined so far). In addition, I
have discovered no great fluctuations in ambiguity.
In view of the original shorthand and the previous research on the Diary, it
is more expedient to instead focus on some of Pepys’s correspondence, which
will additionally make it easier to compare this Restoration correspondence
sample with the results obtained from the Pastons, as well as with Nakamura’s
findings based on the Diary.

Here I examine letters exchanged between Samuel Pepys and his friend John
Evelyn between 27 April and 14 October 1665, mostly on business matters con-
nected with the navy, more specifically the care that is to be taken of sick and
wounded soldiers and prisoners. The following source was used for the texts:
http://www.pepysdiary.com/letters/ (accessed on 01 Sept 2016). The size of this
25 Due to space and time constraints, confirmation with 18th-century material will have to
await future investigation; see especially the book-length study by Rydén and Brorström
(1987).
Table 7 Perfect, passive, progressive and modal auxiliaries in the letters of Samuel Pepys
and John Evelyn


and auxiliary be
Passive have 2 1.6%
Be-perfects 1 0.8%
Progressives 5 4.1%
Modal be 5 4.1%
Be-passives 48 39%
clause
sample is smaller, approximately 3,651 words, but it yielded about the same
amount of material due to the more advanced development and diversifica-
tion of the auxiliaries under investigation. The figures in Table 7 are again simi-
lar to everything seen so far.
There was a total of 39 have-perfects with transitive verbs and copular or
auxiliary be. In none of these was the object inserted between the auxiliary
have and the past participle. Again, a minority contained prepositional objects
or were used without an overt object, as in (270):
(270) Take notice also, I pray, how few have miscarried [Most of those who
dyed perished for want of covering] [E to P, Saturday 14 October 1665]
Miscarry can in fact be an intransitive verb here, which in ME and EModE had
senses like ‘suffer harm’, ‘come to grief’, ‘perish, die’, or (later) ‘fail to reach its
proper destination’ (Rydén and Brorström 1987: 132, cf. the OED). There were
another three have-perfects with intransitives. One of them is with the raising
verb seem, which does not really qualify as mutative:
212 chapter 5
(271) as you have lately seem’d to impute it to me [E to P, Tuesday 3 October

1665]
The other two, (272)–(273), featured come in the present as well as in the past
perfect, which would have been completely unambiguous with be.
(272) For finding divers Chyrurgeons, and Sick-persons at my dores who had
come from Several places with sad complaints that they could not procur
quarters for them [E to P, Tuesday 3 October 1665]
(273) What has come collateraly on you (not through my fault) ought not be
imputed to me [E to P, Tuesday 3 October 1665]
By contrast, there was only one be-perfect in this 17th-century correspondence:
(274) (long before this as having prospect sufficient of what is befallen us)
[E to P, Tuesday 3 October 1665]
Although us may look like an object, (274) is probably the result of confusion
with mutative intransitive fall, of which befall is a derivative, or with the obso-
lete intransitive sense of befall as ‘fall’ (see OED entry for befall, as well as Petré
and Cuyckens 2008 for the properties of the prefix; cf. Rydén and Brorström
1987: 23–24, 53, 81–82, for such patterns involving ‘quasi-objects’ of verbs like
befall). (274) can only be a perfect and no passive, since no argument is missing
or has been promoted or demoted – there is a subject (thematically a force or
cause) and a quasi-object (patient/affected, or better a ‘deep-structure’ loca-
tive adverbial/adjunct or goal complement, i.e. ‘something has fallen/come
(up)on us’, as suggested by Rydén and Brorström 1987: 23, 81). Crucially, both
participants in the event are there. This can be compared to an unusual pas-
sive, in (275), which also looks like it has an extra object (again a thematic
locative/source), but the subject is actually a promoted patient/affected object
from the active and the agent is missing altogether. In the entry for the verb
discharge, the OED supplies more examples of the omission of the preposition
from, which was possible in earlier periods and resulted in a second ‘locative
quasi-object’ for discharge.
(275) When they are discharg’d our Hospitals as cured … [E to P, Thursday

12 October 1665]
Both of these sentences are unusual from a present-day perspective due to

changes in the way the complementation patterns of these verbs work, more
specifically the way the locative complements/adjuncts are expressed: some-
thing befalls somebody/something (be)falls on somebody (with possible confla-
tion of the two); somebody discharges somebody from a place. The difference
between the time of Pepys and the present day lies in the fact that these loc-
ative complements did not need to be expressed by a PP in earlier English,
therefore coming to resemble objects, but that did not affect the perfectivisa-
tion and passivisation strategies which would otherwise be selected for those
verbs (i.e. a be-perfect for intransitive (be)fall and a be-passive for monotransi-
tive discharge). In the light of this, it is still noteworthy that the sole surviving
be-perfect here, (274), should occur in what at first sight appears to be a transi-
tive clause, even though with an untypical ‘object’.
Is then the ambiguity gone (see Table 7 above and discussion below) be-
cause be-perfects have virtually disappeared, having been displaced by have?
Or is it, as I think more likely, because mutative intransitives are very rare in
the first place? I will return to this shortly. Also, has have ousted be from this
domain because of the functional overload of the latter? There are already
many modal and progressive uses of be in this sample, but not many more than
in the previous ones, where be-perfects are still more productive. Comparison
with The Country Wife (Section 5.5.2) would be especially instructive. There,
be-perfects still outnumber have-perfects with intransitives – 12 to 4. The pro-
liferation of functions of be is in evidence to the same extent (including per-
fect, progressive, modal, and passive), yet ambiguity remains more or less the
same relative to text size (3.3% in The Country Wife, of which 0.7% for ambigu-
ous be-perfects/passives vs. 1.6% here, of which 0% for ambiguous be-perfects/
passives). The statistics suggest that there are hardly any firm causal links be-
tween functional load, ambiguity and the disappearance of be-perfects.
Furthermore, Pepys’s 17th-century correspondence sample shows the great-
est proliferation of functions of have, adding one more category where it is
employed, namely in a quasi-passive construction. (276) and (277) are the two
examples of what looks like an incipient passive with have. As observed in
Chapter 3, it is puzzling (for a traditional functionalist) that have should come
to compete with passive be just as the two have supposedly been disentangled
as ‘exclusively’ perfect and passive respectively.
(276) a Shipp of ours (the little Guift) that in a Conflict with a Hollander on
the Irish Coast (wherein shoe though much over matched hath acquitted
her selfe very well) hath had severall Men wounded [P to E, Thursday
27 April 1665] [i.e. ‘several men were/got wounded’]
214 chapter 5
(277) at Chatham, and Gravesend, where I was threatnd to have our sick all
expos’d [E to P, Saturday 23 September 1665] [i.e. ‘that our sick would
all be/get exposed’]
As mentioned above, none of the bona-fide have-perfects had objects com-

ing between the auxiliary and the lexical verb. Moreover, perfect and (quasi-)
passive have co-occur in (276). In addition, there was one example, (278),
where have expressed an incipient modal meaning, though with an earlier
word order, so the Hospitals of London might still function as the object of ‘pos-
sessive’ had, which is in the process of being bleached of its possessive sense
(cf. Chapter 3 and previous discussion in this chapter):
(278) and these coming under my particular cognizance, as I have had (to
the greate increase of my Trouble) the Hospitals of London to look after
during the absence of my Brother Commissioners [E to P, Thursday
12 October 1665]
The sentence in (279) contains two constructions of interest:
(279) By what I have sent you, you will have a Specimen of the Method
observed where I have any-thing to do [E to P, Saturday 14 October 1665]
The first one seems like a potential candidate for causative have, if observed is
construed as related to have, i.e. ‘have this observed, make sure it is observed’.
However, observed might also be a (reduced relative) post-modifier of Method,
i.e. you will receive a specimen of the method which is observed (the latter is
far more likely). Crucially, this once again shows have in an ambiguous context.
The second construction, with the clause-final infinitive, might conceivably
carry a modal meaning comparable to the one in (278). Alternatively, it is just
the have-something-to-do-(with) pattern.
The functions of be have been enriched with five instances of the progres-
sive and five more modal (deontic) uses with a to-infinitive (some of which
probably expressed futurity more than anything else – see Appendix 8).26 In
one of the examples of the progressive, the present participle was conjoined
with an adjective (albeit with an ultimately present participial etymology in
Romance) – in (280), assistant is almost certainly an adjective and is unlikely

to be a noun.
(280) I was forc’d to dispatch Warrants to the Connestables and other Officers
to be ayding and assistant to my Deputyes [E to P, Tuesday 3 October
1665]
There is indeed a wide range of recorded functions of be (perfect, progressive,

modal and passive), though have hardly lags behind (perfect, passive, modal
or causative). Despite the proliferation of functions of be, the passive remains
dominant, with 48 occurrences. Of these, some were admittedly doubtful:
(281) where there will be care taken for them [E to P, Saturday 23 September
1665]
About four such instances were found, which might in fact contain simply an
existential there construction with copular be and a participial post-modifier
to the noun. This much is suggested by (282), where the post-modifier is clearly
adjectival:
(282) there were none of those Vessells ready you were pleas’d to name [E to P,
Tuesday 3 October 1665]
Essentially, this reaffirms that ambiguity is inherent in the passive too, as also
demonstrated by the fifteen clauses which could be passive as well as stative.
(283) should be treated as exclusively passive, however:
(283) They all this while concealing their having pensions, enjoy the Kings
Super-allowance in the Hospitals, which formerly was not continud
[E to P, Thursday 12 October 1665]
Formerly and the context indicate past time which favours a preterite reading –
this non-specific, generic ‘beforeness’, however, is not the type of past time that
is wholly incompatible with the perfect. Be that as it may, continue probably
never combined with be to form its perfect. Even the intransitive senses of
continue in the second edition of the OED contain have-perfects only. Eight
of the fifteen clauses in my dataset which are passive or ambiguously stative
involve the formulaic construction be pleased, as in (282) or (284), which might
be merely adjectival:
216 chapter 5
(284) and what his Majestie shall be pleas’d to superadd to his other Commands
[E to P, Monday 1 May 1665]
And yet, it sometimes had a somewhat verbal force, as in the following:
(285) Be pleasd to seale [this?] [E to P, Saturday 30 September 1665]

[= ‘be/make sure to seal’?]
(286) For prevention whereoff you may be pleas’d to order that none be
admittd from any our Infirmitories into Pensions [E to P, Thursday
12 October 1665] [= ‘it may please you to order …’?]
To be pleased is paraphrasable with ‘to be happy/willing; to like; to have the

will or desire; to have the humour; to think proper’, as confirmed by the OED
and by example (287), where please is certainly an active verb (cf. Allen 1995).
To be pleased, further explained as ‘to be gratified, delighted, or agreeably satis-
fied’, is indeed classified as a passive in the OED, but it is potentially adjectival
according to my criteria.
(287) The other Accoumpt, keepe by you as long as you please [E to P,

Saturday 14 October 1665] [= ‘as long as you wish/want/like’]
Likewise, (288) is passive or stative, but not perfect:
(288) and some of these concernd me as far as Deale and Sandwich, where we
are so overlayd, that they send them back upon us, and they perish in the
returne [E to P, Tuesday 3 October 1665]
Overlaid evidently stands for ‘overwhelmed [with sick and wounded soldiers
and prisoners]’. So might additionally suggest adjectival status for overlaid.
Cure is also normally transitive, so (289) is unlikely to be a perfect, not least
because of the hospital context too (though the OED lists an obsolete intransi-
tive sense of ‘be cured, get well again’, attested between the 16th and the 18th
centuries, which is marked as rare).
(289) First, Our men in the London Hospitals steale downe to Chatham before
they are Cur’d [i.e. ‘healthy again’] [E to P, Thursday 12 October 1665]
Even if (289) were treated as potentially perfect (though this would be hardly
justifiable), the statistics would remain more or less the same. There was no
uncertainty of interpretation coming from any other ambiguous be-perfects or

from contracted ’s. In this sample, all the (persistently insignificant!) ambiguity
arises from two instances of ellipted auxiliaries:
(290) From a letter this day come to my hand from a Shipp of ours [P to E,
Thursday 27 April 1665]
(291) but how shall we (when recoverd) secure them from running away?
[E to P, Saturday 23 September 1665]
The latter is especially noteworthy: it could certainly take have; alternatively,

if they are is inserted in front of recovered, it would still allow a perfect as well
as a passive reading. Apparently, the writer did not consider this problematic
enough to necessitate disambiguation by insertion of have.
In conclusion, we once again observe about the same amount of (negligible)
ambiguity in relation to be-perfects in a sample which demonstrates roughly
the same proportions for the different uses of have and be, possibly with the
debut of passive have in the dataset. Most of the ambiguity throughout my
samples, including this last one, is between a passive and a stative interpreta-
tion for be, but passive get has not yet ‘come to the rescue’, although this am-
biguity persisted for centuries. Be-perfects have been almost completely lost
here, unlike what Nakamura found in the Diary, but this does not seem to have
been driven (primarily) by functional considerations. Despite the small num-
bers, there was an interrupted trend of rising ambiguity to go with the rise of
the (be-)perfect in the different periods from the Anglo-Saxon Chronicle: 3.3%
(then 0%!) < 4.8% < 12.2% (see Chapter 4). The Middle and Early Modern cor-
respondence demonstrates the reverse trend of declining ambiguity accom-
panying the ever-dwindling be-perfect: 5.5% > 4.4% > 1.6% of ambivalence at
each successive stage. This monotonic fall, however, is not corroborated by the
drama sample and might be just a coincidence due to extraneous factors, so
further research with larger databases is clearly needed (cf. Kilpiö’s data cited
in Chapter 3, Section 3.5.4, again without any readily discernible trends). The
rise and fall of this tense-aspect construction may have been accompanied by
a concomitant rise and fall in inherent ambivalence, but that does not auto-
matically mean that this ambivalence is what triggered the demise of the be-
perfect (cf. the passive/stative and the earlier discussion). I now fast forward to
the 19th century in a brief epilogue where I tie up some of the main threads of
the argument, followed by a summary of the findings from this chapter in the
more general conclusion.
218 chapter 5
5.8 Epilogue
As noted throughout this chapter, although the unclear cases I have recorded
are not that numerous, at the same time the availability of multiple readings
does not seem to have been detrimental or undesirable in discourse. In order
to illustrate once again that speakers throughout the history of English could
tolerate and obliviously live with a large amount of ambiguity in close proxim-
ity, I give a final quotation from Shakespeare (cf. Chapter 3, Section 3.4.4):
(292) The bonds of heaven are slipp’d, dissolv’d, and loos’d [Shakespeare,
Troilus and Cressida V.ii, line 156]
Presumably, are slipp’d is an intransitive perfect (today’s ‘have slipped’) or a

passive in the sense of ‘have been stripped or taken off’, a sense attested during
Shakespeare’s time and recorded in the OED, [are] dissolv’d is also ambiguously
perfect (‘have dissolved’) or passive (‘have been dissolved’), while [are] loos’d
is perhaps more of a passive (‘have been loosened’), but it could conceivably
(though less convincingly) be read as ‘have loosened’ as well. All of them are
next to each other and share the same auxiliary, but this diversity does not
seem to have troubled Shakespeare or his contemporaries.27 As pointed out in
earlier chapters and highlighted in the constructionist literature (Traugott and
Trousdale 2010, 2013), synchronic gradience and indeterminacy between differ-
ent interlinked constructions is a natural consequence of gradual diachronic
change.
As stressed throughout the previous sections, ambiguity and variability lin-
gered on even after the almost total obliteration of the be-perfect. While exam-
ining his usage in Chapter 3, it was confirmed that that Dickens had only some
vestigial be-perfects, as in (293):
(293) By that time I had made up my mind that the period was come when I
must tell my guardian what I knew. [1853, Charles Dickens, Bleak House,
Chapter XLIII]
27 A similar case is provided by the version of the Lord’s Prayer in Ayenbite of Inwyt, a Kentish
text from 1340. The same token of by ‘be’ applies to y-halȝed ‘hallowed’, an adjective or a
transitive past participle, cominde ‘coming’, a present participle, and y-worþe ‘become’, a
typically intransitive past particple with the prefix y- (from OE ge-), along with loss of
final -n:
(i) Vader oure þet art ine hevenes, y-halȝed by þi name, cominde þi riche, y-worþe þi wil …
[1340, Ayenbite of Inwyt, quoted in Baugh and Cable (2013: 417), emphasis mine]
At the same time, the narrator of Bleak House opts for non-finite have in
I should have come, as well as plurperfect she had come elsewhere (both from
the same chapter).28 Double marking is in evidence too: should have been gone
in another hour (ibid.), indicating that be + gone is a copular structure with a
participial adjective. Nevertheless, Chapter LVII has non-standard after they
was gone to bed, with a to-PP, said by a police officer and presumably suggest-
ing that (vestigial) be-perfects were still part of colloquial English, at least to
a certain extent. In addition, there is the following example with a missing
auxiliary:
(294) Some of these vessels were of grand size: one was a large Indiaman just
come home [1853, Charles Dickens, Bleak House, Chapter XLV]
The overall distribution of the be- and have-perfect with the verbs come and go
in the 19th century is discussed by Rydén and Brorström (1987: 66–71, 109–115),
including Dickens. Be come and have come receive an average of 21% v. 79%,
respectively, for all authors, but with only 4.6% for be come in Dickens’s let-
ters. Dickens likewise uses have gone (as opposed to be gone) 76% of the time
in his correspondence. By the Victorian period, ambiguity ‘should’ have argu-
ably disappeared with the almost complete disappearance of be as a perfect
auxiliary, especially in the language of a progressive user like Dickens (with
87% of have with intransitives in his correspondence, according to Rydén and
Brorström 1987: 197, 203), but there are instances in Dickens which confirm
that it persisted nevertheless. This makes it difficult to claim that the demise
of the be-perfect had to do with streamlining the system and getting rid of un-
clear cases. Are burst open, was passed and am changed in (295)–(297) below
can be perfect or passive/stative. It is also worth noting the neighbouring have-
perfect as well as the clear be-passive at the end of (295) – even in the 19th cen-
tury they all seem to coexist happily together, unperturbed by considerations
of what is ‘functionally desirable’. In the context of (296), we would probably
use had (been) passed today. The adverb in how changed might further suggest
adjectival status for the participle in (297).
28 It is noteworthy that Dickens also opts for had when a past event is clearly over and does
not continue into the past reference moment, as in (i). He was come would probably have
implied that he was still there at the time (cf. Smith 2001: 369; as well as the Modern
English resultative stative is gone, which can no longer indicate an anterior event that
does not go on into the present: *He is gone and then is back again vs. He has left and then
has come back again, Elliott 2001: 63).
(i) We agreed that he was either prevented from coming, or that he had come, and gone
away; [1853, Charles Dickens, Bleak House, Chapter LXI]
220 chapter 5
(295) then upon my honour, upon my life, upon my reputation and principles,
the floodgates of society are burst open, and the waters have – a – oblit-
erated the landmarks of the framework of the cohesion by which things
are held together! [1853, Charles Dickens, Bleak House, Chapter XL]
(296) I could not venture to approach her, or to communicate with her in writ-
ing, for my sense of the peril in which her life was passed was only to
be equalled by my fears of increasing it. [1853, Charles Dickens, Bleak
House, Chapter XLIII]29
(297) I know full well how changed I am [1853, Charles Dickens, Bleak House,
Chapter LXI]
As pointed out during the discussion of unclear examples in the previous

sections, and as reaffirmed here, this long-lived ambiguity, unabated despite
the loss of perfect be, is rather difficult to account for by having recourse to
the traditional functionalist hypothesis. Processing studies provide further
evidence that production is not always motivated by the perceived needs of
addressees – speakers do not shy away from ambiguity, which is resolved by
addressees with the aid of contextual information (see Traugott 2008: 241 fn. 14,
with references). Sometimes it hardly needs to be resolved at all as it does not
get in the way of communication (see Ammann 2005: 268–269 fn. 28, Wasow
et al. 2005, Piantadosi et al. 2012, Winter-Froemel 2013). In this and in the previ-
ous chapter, I have found much more support for explanations appealing to
frequency and language contact.
Going back to Chapter 2, I need to revisit and disentangle in the light of
my data the different ways of thinking about functional load. The traditional
accounts tend to conceive of language as an idealised abstract system that
seems to run its own course and regulate itself, ridding itself of communica-
tively undesirable imperfections. By contrast, as argued by more sophisticated
later scholars who dismiss naïve teleology, ambiguity primarily matters to
the speaker/hearer in actual communication, not so much on the level of the
system (i.e. in parole/performance, not in langue/competence). The speaker
presumably uses whatever means are available to suit his or her immediate
goals of getting messages across, which can have only an indirect effect on the
system, hence Keller’s vivid metaphor of the invisible hand. My data suggest
that, judging by the written evidence, even in actual communication, language
29 Chaucer’s sentence containing is passed in Section 5.2.2 above was treated as an unambig-
uous perfect because of syntactic clues; no such clues are present in Dickens’s example.
users do not appear to have been guided by functional considerations or ‘led

astray’ by unclear or missing marking as much as has been purported. Like
Lehmann (1985: 316), I am inclined to conclude, rather anticlimactically, that
‘[t]here is much change just for the sake of change’.
5.9 Conclusion
This chapter has documented the competition between the two perfect auxil-
iaries in earlier English written samples which can be argued to approximate
the spoken medium, showing in ample detail how have gradually came to be
selected as the only auxiliary for the perfect, ousting be from that domain (but
only after centuries of ‘resistance’). The material from a collection of represen-
tative texts (14th to 17th century) demonstrates that ambiguity should not be
considered an important contributing factor in this process. Most instances
of replacement of be by have with intransitives would have been unambigu-
ous with be anyway. Furthermore, the syntactic and semantic context is often
sufficient to disambiguate, attenuating any prospective ‘communicative dam-
age’ arising from the availability of multiple interpretations (perfect, passive
or stative). There are even indications that the availability of multiple inter-
pretations was actually exploited as creatively desirable, rather than avoided.
Supposedly ill-suited hybrid perfects/passives such as X is begun or ended keep
turning up and survive until the 19th century, even after the almost total oblit-
eration of be as a perfect auxiliary. If perfect be was dispreferred because of its
inherent ambivalence, these should not have enjoyed such remarkable lon-
gevity. The greatest amount of ambiguity in all periods studied is between a
dynamic passive and a copular stative reading (e.g. PDE He was frightened),
by far outnumbering ambiguous perfects, but this has not harmed passive be’s
chances of survival. Neither do I record any imminent ‘relief’ from passive get.
In terms of frequency, the statistics confirm that have has always been pre-
dominant in the perfect and be in the passive. Moreover, perfects of transitive
verbs are much more common than intransitive ones.
The diachronic statistical trends are also problematic and inconclusive.
Ambiguous be-perfects account for 3.6% of all recorded auxiliaries in Chaucer’s
selected texts (plus 0.6% due to missing auxiliaries), 1.9% in Secunda Pastorum,
4% in Shakespeare (+ 1.5% due to missing or contracted auxiliaries) and 0.7%
in Restoration comedy (+ 2.6% due to missing or contracted auxiliaries), so
Shakespeare disrupts the pattern. No clear trend crystallises after collating
the percentages from ambiguous be-perfects and ambiguity due to contrac-
tion or omission of the auxiliary. If the categories are kept apart, ambiguous
222 chapter 5
be-perfects decline over time (ignoring Shakespeare), whereas contraction- or

omission-related ambiguity increases (ignoring Secunda Pastorum). By con-
trast, the less extensive correspondence material does appear to follow a trend
of diminishing ambiguity. So, the be-perfect tends to be accompanied by some
inherent ambiguity, but it should not be seen as detrimental. To the extent
that ambivalent cases did gradually decline with the decline of the be-perfect,
that was most likely not a contributing factor/cause, but just a natural effect of
the dying out of the construction itself. Another question which is difficult to
answer is whether the decreasing rate of unclear be-perfects somehow made
it possible to resort to more and more ellipsis and contraction (cf. percentages
above), but this is unlikely. The material shows that ambiguity did not particu-
larly matter to language users and ellipsis and contraction might have acted
in favour of be rather than have. In German, where the be-perfect never fully
receded, ellipsis is also widespread (see Chapter 6).
The texts additionally illustrate the proliferation of the functions of both
auxiliaries under consideration, as outlined in Chapter 3, so a great deal of lan-
guage change, including reanalysis and grammaticalisation of generic verbs,
happens automatically and mechanistically, as a natural cognitive process (e.g.
of pragmatic inferencing in context, as discussed in Chapters 1 and 2). The ma-
terial here provides vivid illustrations of some of the intermediate steps along
the way (for instance, from expressing possession to expressing obligation and
futurity). Like the northern redaction of the Chronicle examined in Chapter 4, a
15th-century northern play and the correspondence of the Pastons testify that
the loss of perfect be is more advanced in areas of Viking settlement (Sections
5.3 and 5.6), so contact may well have conditioned and/or accelerated it. This
much is suggested by the innovative usage in the north in the context of overall
stability for centuries in the more conservative south, with the switchover of
the auxiliaries happening there only later.
chapter 6
How Things Could Have Been: a Glance at German
In Chapters 2–5, I defended the idea that a good deal of language change is
simply random, in the sense of non-functionally motivated. Here, I return to
this claim by turning the spotlight on two Germanic varieties (English and
German) which took a very different course from each other, despite shar-
ing the same point of departure and despite their near-identical ‘grammati-
cal ecologies’. Chapter 3, Section 3.4.3, mentioned the loss of the Old English
weorðan-passive, which may have conspired to overload passive be and push be
out of its perfect role. It was stated there that this never happened in German,
hence both types of perfect survive. Things are hardly likely to have been as
simple as that (and it remains mysterious why OE weorðan disappeared but
German cognate werden did not).
In Section 6.1 below, I introduce the German auxiliary system, which is in-
deed very much reminiscent of that of earlier English outlined in Chapters 3–5.
I also emphasise that the potential for misunderstanding existed in German,
just as it did in English. In Section 6.2, I demonstrate that lack of clear mark-
ing is widespread in a German text from the 18th century comparable to the
English ones above. And yet, German has not experienced loss of perfect be,
although the same purported functional triggers were there. Granted that the
two languages started from a similar starting point, they must have come out
so differently with regard to their be-perfects either because of the essentially
arbitrary nature of language change and/or because of the different historical
pressures and contacts that they experienced.
6.1 Introduction
As noted in the opening paragraphs, this chapter aims to take a cursory glance
at the way German works in comparison with English. The two languages
are closely related – both of them belong to the West Germanic group of
Indo-European. Investigating a Modern German text within the same frame-
work as the English samples above can therefore provide an insight into how
things might have looked in English if it had preserved both types of perfect, as
well as its weorðan-passive.1
1 An overview of the Modern German temporal system can be found in Eisenberg (1994,
2004: 106ff.). A brief survey of the history of the German perfect is available in Rydén and

224 chapter 6
Modern German builds its perfect with haben ‘have’ or sein ‘be’ in combina-
tion with a past participle. In addition, the perfect has now taken over as a past
tense in that it is compatible with specific past time reference, such as gestern
‘yesterday’ (for more details, see Smirnickaja et al. 1977, Zifonun et al. 1997:
1686ff., 1702–1708, 1711, Atanasova and Sugareva 2004: 32–34, Eisenberg 2004:
108, 110–114, esp. 113, Macleod 2012: 241–242, Gillmann 2016). Haben is used
for the perfect of all transitive and reflexive verbs, including transitives with
implied objects, impersonal verbs (Es hat geblitzt und gedonnert. ‘There was
thunder and lightning’ – Atanasova and Sugareva 2004: 26), as well as modals.2
As shown in (1) and (2), haben has also been extended to intransitives which
do not express movement or a change of state, i.e. non-mutatives such as sit-
zen ‘sit’, stehen ‘stand’, liegen ‘lie’ etc., which were once compatible with either
auxiliary (see Rydén and Brorström 1987: 19, Eisenberg 2004: 108–109; as well
as Drinka 2017: 241, for exceptions in modern southern German, where be is
preferred for these verbs).
(1) Ich hab-e in der ersten Reihe ge-sess-en.

I have-pres.1sg in the first row pref-sit-pptcp
‘I sat/was sitting in the front row.’ [original from Atanasova and Sugareva
(2004: 26)]3
(2) Er ha-t an der Tür ge-stand-en.

he have-pres.3sg at the door pref-stand-pptcp
‘He stood/was standing at the door.’ [original from Atanasova and
Sugareva (2004: 26)]
Sein is restricted to intransitive verbs of motion or change of state, such as

gehen ‘go’, fahren ‘fare, travel’, fliegen ‘fly (away)’, kommen ‘come’, fallen ‘fall’,
einschlafen ‘fall asleep’, aufwachen ‘wake up’, aufstehen ‘get up’, erkranken
‘fall ill’, genesen ‘get better, recover’, and the like. Verbs of motion at one time
Brorström (1987: 18–19, with references), while Drinka (2017: Chapter 9) outlines the situa-
tion in Old High German and other early Germanic dialects. Petré (2013) is a comparative
diachronic study of the passive in English and German. Modern German (stative) passives
and perfects are compared to earlier English by McFadden and Alexiadou (2010).
2 As is the case in English, there are related stative forms with the past participle, such as (i):
(i) Das Pferd hat die Fesseln bandagiert. [from Zifonun et al. (1997: 1792)]
the horse has the pasterns bandaged
‘The horse has its pasterns bandaged’
These are not always distinguishable from the perfect and I return to them at the end of this
section (see Kostov 1982: 162ff.).
3 Unless otherwise specified, the glosses and the translations in this section are mine.
a Glance at German 225
allowed either marker, depending on certain conditions, including a state vs.

action distinction similar to that in earlier English, but now sein is universal
with them. Now sein is also used for the verbs bleiben ‘remain’, which stresses
that there has been no change of state, geschehen ‘happen, occur’, passieren
‘happen, occur’, as well as begegnen ‘meet’ and folgen ‘follow’, among others.
Finally, the auxiliaries sein ‘be’ and werden ‘become’ themselves build their
perfects with sein (cf. Rydén and Brorström 1987: 19 fn. 2, who note that the
structure er hat gewest ‘he has been’ is recorded in Luther).
The intransitive blühen ‘bloom, be in bloom’ does not indicate a change
of state, so it forms its perfect with haben (Die Blume hat geblüht ‘The flower
(has) bloomed’), but aufblühen ‘come into bloom, begin to bloom’, does entail a
change of state and therefore selects sein (Die Blume ist aufgeblüht ‘The flower
has come into bloom’) (Atanasova and Sugareva 2004: 26, Eisenberg 2004: 108–
109). Furthermore, there are a number of verbs which can take either auxiliary,
depending on whether they are used transitively, (3), or intransitively, (4).
(3) Er ha-t die Maschine nach Berlin ge-flog-en.

he have-pres.3sg the machine to Berlin pref-fly-pptcp
‘He flew the plane to Berlin.’ [transitive; from Atanasova and Sugareva
(2004: 27)]
(4) Er ist nach Berlin ge-flog-en.

he be.pres.3sg to Berlin pref-fly-pptcp
‘He flew to Berlin./He has flown to Berlin.’ [intransitive; from Atanasova
and Sugareva (2004: 27)]
Some verbs of motion can take haben if the emphasis falls on the activity
itself, including its duration, (5), rather than on the goal or destination, (6)
(Atanasova and Sugareva 2004: 27, Eisenberg 2004: 109; cf. Los 2015: 76–77 for
Dutch and earlier English, as well as Chapter 5, Section 5.5.2, example (167),
and elsewhere, for similar English material).
(5) Er ha-t eine ganze Stunde ge-schwomm-en.

he have-pres.3sg one whole hour pref-swim-pptcp
‘He swam for an hour.’ [emphasis on duration; from Atanasova and
(6) Er ist ans andere Ufer ge-schwomm-en.

he be.pres.3sg to other bank pref-swim-pptcp
‘He swam to the other bank.’ [emphasis on destination; from Atanasova
and Sugareva (2004: 27)]
226 chapter 6
There is therefore variation in Sie ist/hat in der Ostsee geschwommen ‘She

swam in the Baltic Sea’, but not Sie ist/*hat an die Ostsee gefahren ‘She has gone/
went to the Baltic Sea’ (Eisenberg 2004: 109–110). Based on its wider distribu-
tion, haben is the default perfect auxiliary in German. Crucially, some parti-
ciples in a be-perfect construction can potentially be conceived of as adjectival
in a copular clause. All of these peculiarities do indeed make Modern German
very similar to earlier English.
In the passive too, the cognates of both auxiliaries already familiar from
the discussion of Old English, weorðan and beon/wesan, survive in Modern
German. The so-called dynamic passive, Vorgangspassiv, indicates a dynamic
process and is built with werden ‘(originally) to become’, (7);4 the stative pas-
sive, Zustandspassiv, indicates a state as a result of an action/process and is
built with sein ‘be’, (8) (see Brinker 1990: 399–400, 403–404, Zifonun et al. 1997:
1789ff., Atanasova and Sugareva 2004: 46ff., Eisenberg 2004: 124ff.).
(7) Die Uhr wird (von ihm) reparier-t.

the clock become.pres.3sg (by him) repair-pptcp
‘The clock gets/is (being) repaired (by him).’ [Vorgangspassiv, from
Atanasova and Sugareva (2004: 46)]
(8) Die Uhr ist reparier-t.

the clock be.pres.3sg repair-pptcp
‘The clock is/has been repaired.’ [Zustandspassiv, from Atanasova and
The classification of the so-called stative sein-passive, the Zustandspassiv, as a

genuine passive construction has been called into question (see Zifonun et al.
1997: 1808–1809, 1817–1823, Eisenberg 2004: 132–133). It is not always straightfor-
ward to distinguish bona fide sein-passives from copular clauses, as in English.
Furthermore, the opposition between the two types of passive in German is
undermined because a state can be indicated both by werden and by sein, as
in (9) (Helbig 1990: 478, 480, Zifonun et al. 1997: 1810–1816; cf. Chapters 3 and 4
for Old English).
4 Werden is also an auxiliary for the future in German (see Zifonun et al. 1997: 1688–1689,
Atanasova and Sugareva 2004: 28, Eisenberg 2004: 107), in addition to being a lexical copu-
lar verb – wird geteilt ‘is/gets divided’ is a passive construction, whereas wird ärgerlich ‘gets
angry’ is copular (see Eisenberg 2004: 124–125).
(9) Die Stadt wird/ ist von 2 Millionen

the city become.pres.3sg be.pres.3sg by 2 million
Menschen be-wohn-t.
people pref-inhabit-pptcp
‘The city is inhabited by 2 million people’ (cf. ?*gets inhabited) [original
from Helbig (1990: 478)]
There is no significant meaning difference between werden and sein in (9).

Arguably, wird repariert in (7) could then express a process, as well as (a tran-
sition to) a state. Conversely, Das Haus ist 1920 gebaut ‘The house was [lit. is]
built in 1920’, can approximate an action-oriented meaning in a similar way
to Das Haus wurde 1920 gebaut ‘The house got/was built in 1920’ (see Zifonun
et al. 1997: 1821).5
It is worth noting that even a short introduction to German grammar for
non-native learners such as Atanasova and Sugareva’s (2004: 50) deems it pru-
dent to stress that students should distinguish the Zustandspassiv with sein ‘be’
from the active perfect forms with the same auxiliary (cf. Zifonun et al. 1997:
1809). This might suggest that there is potential for ambiguity after all, though
Eisenberg (2004: 110) points out that the be-perfect and passivisation are nor-
mally incompatible for the same verb, due to the different transitivity con-
straints. Helbig and Buscha (1991: 177), on the other hand, flag up the existence
of ‘homonymy’ between the Zustandspassiv and the active perfect in cases like
Das Kind ist verzogen, something vaguely like ‘The child is slipped’, which can
mean either Das Kind ist falsch erzogen, ‘The child has been/is raised badly’
(passive), or Das Kind ist (an einen anderen Ort) weggezogen, ‘The child has
moved (to another town)’ (perfect). Further difficulties in differentiating be-
tween Zustandspassiv and a stative reflexive, the so-called Zustandsreflexiv, are
discussed in Helbig and Buscha (1991: 177ff.). Finally, as established for English
in Chapter 3, German too has a passive-like construction with haben, so a
sentence like Sie haben hier die schönsten Steine vereint ‘(lit.) They have here
5 In addition, German seems to be recruiting more passive auxiliaries (see Zifonun et al. 1997:
1824–1829, Eisenberg 2004: 133–134, Askedal 2005), including bekommen and kriegen ‘to get,
receive’, which might be unexpected on a functionalist account (cf. Chapter 3). German al-
ready has two passives, so it hardly ‘needs’ another one. The new formations make it possible
to turn the dative indirect object or the sole dative complement into a subject in the passive,
as in (i):
(i) Renate krieg-t von Rainer ge-droh-t. [from Eisenberg (2004: 134)]
Renate get-pres.3sg by Rainer p ref-threaten-pptcp
‘Renate gets threatened by Rainer.’
228 chapter 6
the most beautiful stones gathered’ can get a perfect (active) reading, ‘They
have gathered the most beautiful stones here (themselves)’, as well as a passive
one – ‘The most beautiful stones have been/are gathered here for them (by
someone else)’ (Askedal 2005: 217–218). I set out to establish in my small-scale
corpus study below whether there is any ambiguity in a sample of (earlier)
Modern German, and if so, what it is.
6.2 Analysis and Discussion of Eighteenth-Century German Drama
The German play selected for this study is Minna von Barnhelm by Gotthold
Ephraim Lessing, first published in 1767. The original text used here is based
on the following electronic publication: http://www.gutenberg.org/cache/
epub/9187/pg9187.html (accessed on 12 Nov 2016). The idiomatic English
translations that appear below are those of Ernest Bell, available at http://
www.gutenberg.org/files/2663/2663-h/2663-h.htm (accessed on 12 Nov 2016),
whereas the glosses are mine.
In the preface to Bell’s translation (http://www.gutenberg.org/files/2663/
2663-h/2663-h.htm, accessed on 12 Nov 2016), the comedy is introduced as ‘the
beginning of German national drama’; it is characterised by a patriotic inter-
est in the historical background, as well as a ‘sympathetic treatment of the
German soldier and the German woman’; it has also been said to felicitously
blend ‘the amusing and the pathetic’, like its English counterparts above. In
that sense, it was deemed a fortunate choice for the purposes of comparison.
table 1 Perfect, passive, and modal auxiliaries in the first eight scenes of Minna von
Barnhelm

Have-perfects with transitive verbs & with modals 30 44.1%

Be-perfects with intransitive verbs 7 10.3%
Modal have 4 5.9%
Modal be 1 1.5%
Werden-passives 5 7.4%
Sein-passives/statives 5 7.4%
Missing auxiliary 15 22.1%
Due to space constraints, only the first eight scenes of Act I were examined,
a stretch of approximately 3,222 words. Nevertheless, this supplies enough ma-
terial to go by. The results are laid out in Table 1 above. For the most part they
are very similar to the ones obtained for English. By far the largest majority of
perfects are have-perfects with transitive verbs (in a broad sense). There is only
one intransitive have-perfect and only seven perfects with be. Again, mutative
intransitives are comparatively rare. Like English, German also demonstrates
the extension of the field of operation of both auxiliaries into the modal do-
main. Dynamic passives with werden are as many as stative passives with sein
‘be’, for which there are indications to assume adjectival status (see below).
The most palpable difference from the general situation in English is the total
absence of ambiguous be-perfects/statives and the large number of missing
auxiliaries, which however is in some respects similar to some of the later
English samples. Pepys’s correspondence in Chapter 5, Section 5.7.2, and The
Country Wife in Section 5.5.2, both from the 17th century, seem to have more
or less eliminated ambiguous be-perfects too, though neither of them has as
much ambiguity from ellipsis and contraction.
Starting with perfects, there are no major surprises, with the usual division
of labour between have (with transitives) and be (with intransitives), as illus-
trated in (10). As demonstrated by the material, the perfect is of course well
established by this late stage in the history of German.
(10) Was gilt’s, der Herr Major ist nicht nach Hause
what counts.it the sir Major be.pres.3sg not to house
ge-komm-en, und Er ha-t hier auf ihn
pref-come-pptcp and he have-pres.3sg here for him
ge-lauer-t? [I.ii]
pref-keep.watch-pptcp
‘What will you bet the Major has not returned home, and you have been
keeping watch for him?’
Strictly speaking, gelauert does not take a direct object, but a prepositional
complement with auf; I nevertheless group it with the transitives proper (as I
have done for English throughout). As noted in the introductory remarks, have
is used with modals, (11). Have also forms its own perfect, (12).
(11) Ich hab-e ihm aus Not ein ander Zimmer

I have-pres.1sg him from need one other room
ein-räum-en müss-en. [I.ii]
pref-prepare-inf must-inf
‘I was obliged to have another room prepared for him.’
230 chapter 6
(12) ein Beweis, daß ich nie eine ge-hab-t

one proof that I never one pref-have-pptcp
hab-e [I.vi]
have-pres.1sg
‘a proof that I never had it’
There is only one intransitive have-perfect, (13), with a verb that does not ex-
press motion or change of state and which also has transitive senses.
(13) Du ha-st ge-wein-t? [I.viii]

thou have-pres.2sg pref-cry-pptcp
‘You have been crying?’
Be-perfects are comparatively rare, only seven. These include verbs for ‘come’
(kommen) and ‘fare/travel’ ( fahren), and/or their prefixed derivatives, as well
as the verb for ‘remain’ (bleiben) – see (10) above and (14)–(16) below. Bleiben,
(15), denotes lack of change of state. Arguably, the absence of change of state is
also a type of change of state (see Eisenberg 2004: 109).
(14) Und ich glaube nicht einmal, daß sie sonstwo

and I believe neg once that they elsewhere
unter-komm-en wäre. [I.ii]
pref-come-pptcp were
‘Besides, I don’t believe they could have got a lodging elsewhere.’
(15) Nein, Marloff ist mir nichts schuldig

no Marloff be.pres.3sg me neg in.debt
ge-bleib-en [sic.] [I.vi]
pref-remain-pptcp
‘Marloff does not owe me anything’
(16) Würd-en Sie höflicher mit mir ver-fahr-en

would-3pl nom.3pl more.politely with me pref-fare-pptcp
sein. [I.iii]
be.inf
‘You would have treated me rather more civilly.’
The last of these, (16), has non-finite sein, and also a PP-dependent in the
clause (mit mir ‘with me’). This meaning of verfahren does not have to do with
literal travelling and it can hardly be said to be mutative (or transitive) in this
sense.6 The verb sein ‘be’ also ‘self-sufficiently’ builds its own perfect, (17), tes-
tifying to complete grammaticalisation of sein as a tense/aspect auxiliary (un-
like OE weorðan in the combination wearð geworden ‘had happened’, English
be has always taken auxiliary have for its perfect; see Chapter 3, Section 3.4.4;
Chapter 5, example (4)).
(17) Ich wüßte mich auch nicht zu erinnern, daß er mir jemals
I knew me also neg to remember that he me ever
etwas schuldig ge-wes-en wäre. [I.vi]
something in.debt pref-be-pptcp were
‘nor can I remember that he ever did owe me anything’
In the modal realm, there are four instances of modal have, and one of modal
be. Have seems to be more wide-spread in this function. Zifonun et al. (1997:
1897ff.) discuss modal haben zu in German, where it can be synonymous with
müssen ‘must’, as in (18), or können ‘can’, perhaps as in (19).
(18) ich hab-e mit meinem Bedienten zu sprech-en [I.iii]

I have-pres.1sg with my servant to speak-inf
‘I wish to speak with my servant.’ [lit. ‘I have to speak with my servant.’]
(19) Geschwind, gnädige Frau, was hab-en Sie zu

quick gracious lady what have-pres.3pl nom.3pl to
be-fehl-en? [I.vi]
pref-request-inf
‘Quick, Madam, what have you to request?’
Some readers might find the modal credentials of (19) a little dubious, but it
can be seen as modal. Zifonun et al. (1997: 1897) provide a similar paraphrase
with kann ‘can’: Sie hat immer etwas zu verschenken. ‘She always has something
to give away’ = Sie hat immer etwas, das sie verschenken kann. ‘She always has
something that she can give away’ (cf. the discussion of the grammaticalisa-
tion of English modal have to in Chapter 3). Modal be, on the other hand, has a
passive overtone, as in (20).
6 A similar case was encountered in Pepys’s correspondence in Chapter 5.

232 chapter 6
(20) Sie finden mich in einer Stunde, wo ich leicht zu

you find me in an hour where I lightly to
ver-leit-en wäre, wider die Vorsicht zu murren. [I.vi]
pref-lead-inf were against the Providence to murmur
‘You come to me at a time when I might easily be misled to murmur
against Providence.’
Eisenberg (2004: 131–132) talks about a modal infinitive and a modal passive,
including borderline cases, e.g. Karl ist leicht zu betrügen/Es ist leicht, Karl zu
betrügen ‘Karl is easy to deceive/It is easy to deceive Karl’, analogous to ex-
ample (20), which could be a modal passive or alternatively it could be seen as
an instance of ‘tough-movement’ (cf. Zifonun et al. 1997: 1897ff.). If modal, the
interpretation is one of possibility (see Zifonun et al. 1997: 1898).
By their very nature, werden-passives are usually dynamic, as in (21), while
those with sein normally express a state (see Section 6.1). There are five of each.
(21) Sie müss-en be-zahl-t werd-en [I.iii]

nom.3pl must-pres.3pl pref-pay-pptcp become-inf
‘You must be paid’
There are clear indications that the constructions with sein should be treated
as adjectival. Adjectival status is suggested by coordination with a genuine ad-
jective such as galant, (22), or by modifying adverbs such as stark, (23), gewiß,
(24), or sehr, (25). As was the case with English, this potential for ambiguity
does not seem to have bothered German speakers too much (additionally con-
firmed by the equal numbers of the stative and dynamic constructions).
(22) Das Zimmer ist doch sonst galant und

the room be.pres.3sg still otherwise gallant and
tapezier-t— [I.ii]
paper-pptcp
‘The room is otherwise very nice, and is papered’
(23) Die Wirtshäuser sind jetzt alle stark

the inns be.pres.3pl now all strongly
be-setz-t. [I.ii]
pref-fill-pptcp
‘The inns are all now quite full.’
(24) Denn ich bin gewiß ver-sicher-t, daß … [I.iii]

for I be.pres.1sg certainly pref-make.sure-pptcp that
‘For I feel quite sure that …’
(25) Ich bin dir sehr ver-bund-en, Just. [I.iv]

I be.pres.1sg thee very pref-bind-pptcp Just
‘I am much obliged to you, Just.’
The two most fascinating discoveries here are the total lack of ambiguous be-
perfects from the data and the numerous instances of auxiliary ellipsis – a sim-
ilar rise in the number of ellipted or contracted auxiliaries in the early modern
period was noted in Chapter 5 for English. The first of these discoveries might
prompt a functionalist to rejoice and attribute the preservation of be-perfects
to this lack of ambiguity, but the second finding is rather problematic. This
passage has many more missing auxiliaries than any other passage analysed
so far, a total of 15, half the number of transitive have-perfects, three times the
number of be-passives, and more than twice the number of be-perfects.
The first example of this kind, (26), can be interpreted in two ways due to
the missing auxiliary:
(26) Just (nachdem er ge-trunk-en). [I.ii]

Just after he pref-drink-pptcp
‘JUST. (after drinking).’
This could be Nachdem er getrunken hat ‘After he has drunk’ (perfect) or

Nachdem er getrunken ist ‘After he (has) got/is drunk’ (passive/stative). Both
make sense in this context. Admittedly (26) does come from a stage direction,
but there were plenty more examples from the dialogue:
(27) Selbst ge-mach-t, Herr Wirt? [I.ii]

self pref-make-pptcp sir landlord
‘Made at home, Landlord?’
Selbst probably suggests a perfect: Haben Sie das selbst gemacht? ‘Have you
made it yourself?’ The passive/stative is not completely ruled out though – Ist
das selbst gemacht? ‘(lit.) Is that self-made?’ (28) is also noteworthy:
234 chapter 6
(28) Einem Manne wie meinem Herrn, der Jahr und Tag bei
a man like my master who year and day by
Ihm ge-wohn-t, von dem Er schon so manchen
him pref-live-pptcp from whom he already so many
schönen Taler ge-zog-en, der in seinem Leben keinen
nice thalers pref-draw-pptcp who in his life no
Heller schuldig ge-blieb-en ist [I.ii]
heller in.debt pref-remain-pptcp be.pres.3sg
‘a man like my master, who has lodged at your house above a year; from
whom you have had already so many shining thalers; who never owed a
heller in his life’
The ellipted auxiliaries with gewohnt and gezogen would have been forms of
haben ‘have’. The only one which is not omitted is the final one – a form of be
in geblieben ist. This however can hardly be said to have worked in favour of
displacing have in German, in the same way that similar arguments were made
for the displacement of be in English (see Chapter 3, Section 3.4.4).
Conversely, in (29), ist is again characteristically missing (from worden [ist]),
but a have-perfect is now present in an earlier clause – just the type of context
which can be said to favour displacing be, which however never happened in
German.
(29) Wenn ich sie nicht habe, so ist es ein Beweis, daß ich nie
if I it not have so is it a proof that I never
eine ge-hab-t hab-e, oder daß sie getilgt
one pref-have-pptcp have-pres.1sg or that it honoured
und von mir schon zurückgegeben word-en. [I.vi]
and by me already returned become-pptcp
‘If I have not got it, it is a proof that I never had it, or that it has been hon-
oured and already returned by me.’
The same kind of ellipsis can of course occur when the expected form is a form
of have, as in geteilet [hat] in (30), where both verbs essentially select the same
auxiliary:
(30) Ich hab-e nie etwas tu-n könn-en, mich

I have-pres.1sg never anything do-inf can-inf me
mit einem Manne abzufinden, der sechs Jahre Glück und
with a man to.compensate who six years luck and
Unglück, Ehre und Gefahr mit mir ge-teil-et. [I.vi]
unluck honour and danger with me pref-share-pptcp
‘I have never been able to do anything to repay a man who shared with
me good and ill luck, honour and danger, for six years.’
In (31), the perfect auxiliary can be recovered from the context – from the
stative/passive in the landlord’s turn. Importantly, this shows the multi-
functionality and unity of these words – copular, passive and perfect be are
perceived as the same lexeme (see (22) and the literature review in Chapter 3,
as well as the Epilogue to Chapter 5).
(31) Wirt: […] Das Zimmer ist doch sonst galant und
Landlord the room be.pres.3sg still otherwise gallant and
tapezier-t— Just: Ge-wes-en! [I.ii]
paper-pptcp Just pref-be-pptcp
‘LAND: […] The room is otherwise very nice, and is papered’ ‘JUST:
Has been!’
The next two examples, (32)–(33), are likewise missing their perfect have
(cf. gehabt habe in (29) above), while (34) is missing a form of perfect be:
(32) welches Ihro Gnaden in dem Schreibepulte steh-en

which your honour in the writing.desk stand-inf
ge-hab-t [I.iii]
pref-have-pptcp
‘which your honour had in your writing-desk’
(33) ein Beutel mit fünfhundert Taler Louisdor, den der

a purse with five.hundred thalers louis.d’ors which the
Wirt in Ihrem Schreibpulte ge-fund-en [I.iv]
landlord in your desk pref-find-pptcp
‘a purse with five hundred thalers’ worth of louis d’ors, which the Landlord
found in your desk’
236 chapter 6
(34) Das ist Geld, welches mir aufzuheben gegeben

that is money which me to.keep given
word-en. [I.iv]
become-pptcp
‘That is money given into my charge.’
Overall, there are many missing auxiliaries. (34) should have a perfect passive,
but the perfect auxiliary is not there (cf. (29) and the note on the avoidance
of perfect passives in Chapter 3). As mentioned already, both sein and haben
can be thus omitted: gefunden hat vs. worden ist. Morphologically, forms like
gefunden etc. can only be past participles because of the prefix ge- and the
suffix -en. Worden, albeit prefixless, is also certainly a past participle ending
in -en – -en cannot be a finite marker of preterite plural agreement as the sub-
ject welches is singular, and the ablaut vowel grade would also be amiss if this
were a preterite.
Finally, it is again not obvious whether the participle in (35) should be con-
strued as participating in a passive (ausgelegt ist/ausgelegt wurde/ausgelegt
worden ist ‘was/has been spent/laid out’) or perfect (ausgelegt hat ‘[my master]
has spent/laid out’).
(35) Seit dem Ersten dieses an Kleinigkeiten aus-ge-leg-t

since the first of.this on sundries pref-pref-lay-pptcp
1 Taler 7 Gr. 9 Pf. [I.viii]
1 thaler 7 gr. 9 pf.
‘During the first part of this month, laid out in sundries – 1 thaler 7 gro-
schen 9 pfennigs.’
Quite importantly, this example comes from a brief note written by one of the
characters. The same applies to another four. Even if these are discarded due
to the nature of the genre, the number of such ambiguity-producing omis-
sions remains astounding. As with the English material above, such ambigu-
ous contexts have persisted in German as well. A short random example from
the opening paragraphs of Chapter 3 of Remarque’s Im Westen nichts Neues
‘All Quiet on the Western Front’ (1928) is the one-word utterance Verstanden?
‘(lit.) Understood?’, ‘Got it?’ (addressed to a newly recruited young soldier in
WWI). It does not matter much whether this stands for Hast du das verstanden?
‘Have you understood that?’/i.e. ‘(Do) you get it?’, or Ist das verstanden? ‘Is that
understood/clear?’.
Although no other ambiguous cases were detected, the ambiguity fostered
by auxiliary ellipsis is still significant enough and provides solid grounds to
conclude that German is not devoid of uncertainty of interpretation and
unclear cases. This snapshot of 18th-century German shows that, if functional

considerations were taken seriously and at face value, German too had very
good reasons to ‘streamline’ its perfects by shedding, let’s say, its be-auxiliary
from this role. Yet lack of clarity and proliferation of functions for both auxil-
iaries are obviously not seen as sufficiently detrimental for this to take place.
6.3 Conclusion
This short intermezzo between the discussion of English and the next two
chapters on Bulgarian has been intended to show that English and German
started out with the same system of perfects and passives, with parallel dis-
tributions of the various auxiliaries, as well as with the same potential for
misunderstanding. Like English, German has also seen a number of cases of
polygrammaticalisation of be and have, as well as the birth of new passive
constructions (coexisting with the original two – based on be and become).
Although I detected no ambiguous be-perfects in a randomly selected sample
of 18th-century German drama, interpretative difficulties due to omission of
the auxiliaries abounded. In all important respects, the picture is therefore
identical to the one in earlier English, including relative frequencies of the
auxiliaries. Nevertheless, German has not taken the same course, retaining
both flavours of passive and both flavours of perfect. Consequently, the (non-)
changes in these two closely related Germanic varieties cannot be due to in-
herent functional motivations, but are probably mostly arbitrary and/or partly
conditioned by external factors such as language contact.
In support of contact-induced change, Drinka (2017: Chapters 7 and 9)
notes a number of shared features characteristic of languages from the realm
of Charlemagne’s mediaeval empire, including the semantic transformation
of the perfect into a pure past, as well as an especially well developed and
maintained be/have dichotomy in the perfect. Being central members of the
Carolingian Sprachbund, German and French exhibit these developments,
whereas English, which sits on the periphery, does not. Thus, the preservation
of both auxiliaries in German is at least in part due to its core membership of
the Carolingian Sprachbund (additionally correlating with a plethora of other
linguistic and extra-linguistic factors attributable to the Sprachbund, e.g. the
reverence for and special use of Latin, for which see Drinka). While German
might have experienced areal pressure to maintain the dichotomy in the per-
fect, as a result of contact with Viking invaders English arguably came under
pressure to simplify the system by retaining only the more prominent of the
two auxiliaries, in line with analogous shifts in the other languages peripheral
to this Sprachbund.
chapter 7
The Perfect in the History and Structure of

Bulgarian
It was stated in the introductory chapters that the choice of Bulgarian and
English for a comparative study has been prompted by the fact that they are
typical representatives of the Eastern and Western European methods of
building perfect constructions. Drinka (2017: Chapter 16) is a concise summary
of the east-west split in perfect formation, where readers can find more infor-
mation. In terms of structure, this split has to do with the different participial
profiles of western languages like the Germanic ones, which possess only one
all-purpose past participle, as opposed to eastern languages like those of the
Slavonic group, where two past participles are available – an active and a pas-
sive one. The existence of the two types of participle in Slavic is arguably what
enables be to function unproblematically as a universal perfect and passive
auxiliary, combining with a past active participle in the first case, and a past
passive one in the second. This is not the case in the west due to the more
limited participial inventory; hence have emerged as the dominant perfect
auxiliary, while be has always been the preferred choice for passives, with only
a secondary (or defunct) role in the perfect with intransitives, which do not
normally passivise.
Despite the traditional hegemony of be-perfects, Slavonic languages have
seen the birth of perfect-like resultatives with have too. Although functional
explanations analogous to those of English have been proposed for the evolu-
tion of have as a perfect auxiliary in Bulgarian, it will be argued below that
Bulgarian did not really ‘need’ a have-perfect (in line with the arguments from
the preceding paragraph); so its appearance must be due to the generic mean-
ing of have, a suitable and likely candidate for even greater generalisation and
hence reanalysis and grammaticalisation, most probably aided by language
contact, considering the rarity of have-perfects outside of Western Europe and
the geographical distribution outlined in the next paragraph. In addition, the
morphological agreement marking on Bulgarian participles is comparable to
that of Old English. Rather conveniently, Modern Bulgarian does not exhibit
the same type of ambiguity due to zero marking, which enables profitable ty-
pological comparison and theoretical generalisations as to the role of agree-
ment in reanalysis and grammaticalisation.

The Perfect in the History and Structure of Bulgarian 239
Examining English and German as compared to Bulgarian and Macedonian

further allows an insight into the bigger geographical picture on the Old
Continent (for which see Drinka 2017). It was established in the preceding
chapters that German belongs to the core area of the be/have perfect dichot-
omy, whereas English sits on the western periphery of it and has according-
ly lost be as a perfect auxiliary (at least in the standard language). As noted
above, be has always been predominant in the east of Europe, including the
Slavic-speaking regions, though have has started to make inroads eastwards.
It will emerge in the body of this chapter that these inroads must have pen-
etrated the South Slavic Macedonian/Bulgarian territory from the west, slowly
moving eastwards. This much is suggested by the more advanced grammati-
calisation of have-perfects in Macedonian, especially in particular locations,
compared to Bulgarian, where this construction is much less entrenched.
This and the next chapter thus continue to probe into the global matters
from the preceding parts of the book in the light of data outside of Germanic.
Importantly, Bulgarian/Macedonian are distant Slavic relatives of English and
German, but the development of periphrastic perfects and passives is not part
of their common Indo-European stock inherited from the proto-language. In
Section 7.1, I provide an overview of the Bulgarian temporal and aspectual
system, focusing on the well-established universal be-perfect (7.2). Section 7.3
enumerates the traditional auxiliary functions of have in Bulgarian, which do
not include building a perfect. The rest of the chapter is dedicated to the emer-
gence of a new have-perfect in Bulgarian, starting with its formal aspects (7.4.1),
morphosyntactic agreement (7.4.2), as well as its status and level of entrench-
ment (7.4.3). Explanations have been put forward for its appearance which
rely on functional load, very much in the spirit of the explanations previously
advocated for English – I outline those in 7.4.4. I then compare the situation
in Bulgarian to the state of affairs in the closely related variety of Macedonian
(7.4.5). Section 7.4.6 revisits questions of agreement and entrenchment in the
grammatical system, while Section 7.4.7 turns to language contact as a pos-
sible source of the new have-perfects in Bulgarian and Macedonian, as well as
perhaps even further afield.
7.1 Overview of the Bulgarian Temporal and Aspectual System
Bulgarian has an elaborate temporal and aspectual system, built around the lo-
cation of the event in time (tense), as well as its internal organisation (aspect),
incorporating distinctions such as (a-)telicity/boundedness (the presence/
240 chapter 7
absence of an inherent end point), (non-)durativity, etc. Verbs come in as-

pectual pairs of perfective and imperfective. Perfectives carry a meaning of
completion, while imperfectives either have a meaning of lack of completion
or are neutral as to perfectivity. Morphologically simple verbs are often imper-
fective (e.g. пиш-а/piš-a ‘write.ipfv’), whereas prefixation derives perfectives,
sometimes in addition modifying the lexical meaning (e.g. на‑пиш-а/na‑piš-a
‘write.pfv’, до‑пиш-а/do-piš-a ‘finish.writing.pfv, write up’, пре‑пиш-а/
pre-piš-a ‘copy.pfv (in writing)’). Further suffixation can create ‘secondary
imperfectives’ from the prefixed verbs (e.g. на-пис-в-а-м/na-pis-v-a-m ‘write.
ipfv’, до-пис-в-а-м/do-pis-v-a-m ‘finish.writing.ipfv, write up’, пре-пис-в-а-м/
pre-pis-v-a-m ‘copy.ipfv (in writing)’).1
Apart from those distinctions, the past tense exhibits an opposition between
an imperfect and aorist, again with imperfective and perfective meaning,
respectively (see Nicolova 2008: 283–287, 290–291, 2017: 367ff., among others).
As Comrie (1976: 126) points out,
such combinations as Perfective Imperfect and Imperfective Aorist are

possible, and either represent combinations of different submeanings of
perfective and imperfective meaning, or, in the case of the Imperfective
Aorist, can also represent the combination of perfective meaning (whence
Aorist) with aspectually unspecified meaning (whence Imperfective).
cf. Scatton 1993: 213
For instance, the aorist of a perfective verb will usually express a finished ac-
tion in the past, as in (1). The imperfect tense of the same perfective verb will
not express action in progress in the past, but a series of complete activities,
often in a subordinate adverbial clause; it is the series as a whole for which
no temporal limits are supplied, not the individual events which make up the
series, as in (2).
(1) На-писа писм-о

Na-pisa pism-o
pref-write.pfv[aor.3sg] letter(nt)-sg
‘He/She wrote a letter’
1 A good English-language introduction to basic Bulgarian verbal morphology can be found in

Scatton (1993: 210ff.); also see Comrie (1976: 125–126, as well as 7n., 13, 21, 23, 26, 31ff., 66–71,
74ff., 118ff., 131n.); for a more detailed description, see Nicolova (2008: 247–261, 2017: 347ff., for
aspect, 2008: 261ff., 2017: 367ff., for tense); cf. Stambolieva (2008); Ganeva (2015) provides a
diachronic survey.
(2) Като на-пише-ше писм-о …

Kato na-piše-še pism-o …
when pref-write.pfv-imperf.3sg letter(nt)-sg
‘Every time/Whenever he/she wrote a letter …’
The imperfect tense of an imperfective verb denotes an activity in progress in

the past or a past habit, as in (3). The aorist of an imperfective verb will indi-
cate that there was some of the action in question (on a single occasion), but
without necessarily setting any limits on it or making the action concrete, as
in (4).
(3) Купува-ше плод-ове

Kupuva-še plod-ove
buy.ipfv-imperf.3sg fruit(m)-pl
‘He/She was buying fruit.’/‘He/She used to buy fruit.’
(4) Купува плод-ове

Kupuva plod-ove
buy.ipfv[aor.3sg] fruit(m)-pl
‘He/She bought fruit.’/‘He/She did some buying of fruit.’
7.2 The Bulgarian Be-Perfect
There is also the category of the perfect, in all tenses and both in the perfec-
tive and imperfective aspect (see Comrie 1976: 62, 107–110, 126, Scatton 1993:
211–212). Most standard reference works on Modern Bulgarian treat the perfect
as a (type of) tense and describe its formation as a combination of a form of
the auxiliary съм/бъда / săm/băda ‘to be’ and the perfective past active parti-
ciple, which ends in -l, e.g. написа-л съм/napisa-l săm ‘I have written’, lit. ‘I am
having-written’. It is this dedicated active non-finite form that enabled ‘be’ to
function as a universal marker of the perfect, as well as the passive. The avail-
ability of this l-participle, as distinct from the past passive participle in -t or -n,
gave Slavonic languages the opportunity to create active periphrastic construc-
tions that were less easily crafted in the west, with its limited array of past
participles (see Drinka 2017: Chapter 16).2
2 On the Bulgarian perfect, consult Tilkov et al. (1983a: 318, 335, 348–350, 378–379), Nicolova
(2008: 268, 292–304, 313–316, 2017: 413ff.); also see Bojadžiev et al. (1999: 382ff., 389–392) and
242 chapter 7
The Bulgarian present perfect typically combines present and past mean-
ing, relating a past event to a present state, not unlike its English counterpart
(Comrie 1976: 52, 107). Like the English perfect, this is achieved by combin-
ing a present form of the auxiliary verb and a past (active) participle – the
present tense conveys the present relevance, while the past participle denotes
the past action. Like Germanic perfects with be, the Bulgarian construction
also originated in mutative intransitive contexts based on the /X is Y/ equation
schema with a subject complement (Kuteva 1998: 295). Notably, however, all
Slavonic varieties came to use be both for intransitive and for transitive verbs,
as in (5). The Germanic languages, including earlier stages of English, normally
show a division of labour between have and be, as discussed in earlier chapters.
As stated above, to a large extent this must be due to the different participial
profiles in the two language groups. Be could unproblematically serve as aux-
iliary for both the perfect and the passive in Slavic, since the perfect features
a specialised past active participle, while the passive has its own dedicated
non-finite form.
(5) Иван е купи-л книга-та

Ivan e kupi-l kniga-ta
Ivan be.pres.3sg buy-pptcp[m.sg] book-def
‘Ivan has bought the book’ (literally ‘is having-bought’)
In the singular, the l-participle has endings for all three genders in order to
agree with the subject: masculine (купил/kupil – see (5)), feminine (купил-а/
kupil-a), and neuter (купил-о/kupil-o). The distinct gender inflections which
Old Bulgarian/Old Church Slavonic had in the plural (m.pl -и/-i, f.pl -ъi/-ъi,
nt.pl -а/-a), are now collapsed into one generalised plural form in -и/-i
(купил-и/kupil-i) (cf. (6)–(9) below).3
Old Bulgarian/Old Church Slavonic (OB/OCS) essentially had the same
perfect with бьiти/byti ‘to be’ plus the past active participle ending in -l. This
construction was not inherited from Proto-Indo-European, but appeared as a
Slavonic innovation from which the Modern Bulgarian forms are descended.4
Kucarov (2007: 249–253), where the perfect is rightly accorded its own category, separate
from tense.
3 See Nicolova (2008: 292), Duridanov et al. (1991: 299).
4 See Chapter 2, Section 2.2, Dobrev (1973), Duridanov et al. (1991: 299–301), Rikov (1993/1994),
Mirčev (2000: 117); cf. Huntley (1993: 152–153), Schenker (1993: 106). Doubts whether the OB/
OCS perfect had the same meaning as in Modern Bulgarian and occupied the same slot in the
system are expressed by Totomanova (2004), MacRobert (2013) and Ganeva (2015: 12 fn. 10).
As underscored above, in contrast to English, the verb ‘to be’ was the exclusive
marker of the perfect even in the earliest Old Bulgarian/Old Church Slavonic.
OB/OCS бьiти/byti was equally happy with intransitive verbs of motion, (6)–
(7), or with ordinary transitives, (8)–(9).
(6) ñѫ пришь-л-и
sǫ prišь-l-i
be.pres.3pl come-pptcp-nom.m.pl
‘they have come’ [original from Duridanov et al. (1991: 300)]5
(7) i бѣ-ахѫ сѣд-ѧшт-є фарисѣ-и. i

i bĕ-axǫ sĕd-ęšt-e farisĕ-i i
& be-past.3pl sit-ing-m.nom.pl Pharisee(m)-nom.pl &
законооучител-е. iже бѣ-aхѫ
zakonoučitel-e iže bĕ-axǫ
doctor.of.the.law(m)-nom.pl who be-past.3pl
пришъ-л-и отъ вьсѣкоѩ вьси
prišъ-l-i otъ vьsĕkoię vьsi
come-pptcp-nom.m.pl from all villages
[Luke 5:16; from Duridanov et al. (1991: 301)]
‘there were Pharisees and doctors of the law sitting by, which were come
out of every town’ [Luke 5:17; King James Bible]6
(8) г[оспод]-и б таланъта ми еси прѣда-л-ъ

gospod-i b talanъta mi esi prĕda-l-ъ
Lord-voc 2 talents me be.pres.2sg give-pptcp-nom.m.sg
[Matthew 25:23; from Duridanov et al. (1991: 299)]
‘Lord, thou deliveredst unto me two talents’ [Matthew 25:22; King James
Bible]
As in earlier English, the perfect seems to have been interchangeable with the aorist in earlier
Bulgarian (cf. Chapter 3, the following chapter, Dobrev 1973 and Totomanova 2004). Although
they were not absolutely identical, often either of the two could appear in the same context
without any obvious semantic differentiation, and writers would frequently use them indis-
criminately, as will become evident in the samples from the next chapter.
5 Unless otherwise indicated, the transliterations, glosses and translations are mine.
6 It is noteworthy that the King James Bible uses a be-perfect here, but have in (9).
244 chapter 7
(9) сѫсѣд-и же и иже и бѣ-ахѫ

sǫsĕd-i že i iže i bĕ-axǫ
neighbour(m)-nom.pl particle & who him be-past.3pl
видѣ-л-и прѣжде
vidĕ-l-i prĕžde
see-pptcp-nom.m.pl before
[John 9:8; from Duridanov et al. (1991: 301)]
‘The neighbours therefore, and they which before had seen him …’ [John
9:8; King James Bible]
7.3 Have as an Auxiliary in the History of Bulgarian
In Old Bulgarian/Old Church Slavonic, имѣти/imĕti ‘have’, a transitive verb

originally denoting possession, could be used with an infinitive to express
futurity and modality (obligation and necessity), as it still does in Modern
Bulgarian, though primarily in negative future forms.7 One difference is that
have now takes a finite da-clause as its complement, as opposed to the infini-
tive, which disappeared in the history of Bulgarian (see Balan 1957: 23–24 and
MacRobert 1981). The Modern Bulgarian modal/future construction in (10) can
be compared to its OB/OCS predecessor in (11).
(10) Има-м да взима-м/ да уч-а/ да

Ima-m da vzima-m/ da uč-a/ da
have-pres.1sg comp take-pres.1sg comp learn-pres.1sg comp
чака-м
čaka-m
wait-pres.1sg
‘I (will) have to take/learn/wait’ [original from Balan (1957: 23–24)]
(11) Има-мь бьра-ти

Ima-mь bьra-ti
have-pres.1sg bear-inf
‘I will pick up’/‘I have to pick up’ [original from Balan (1957: 23–24)]
7 See Balan (1957: 23–24), Georgiev (1957: 43, 57), Tilkov et al. (1983a: 339–340), Minčeva (1987:
27–30, including notes on Greek and Romance influence on the development of the future
periphrases), Duridanov et al. (1991: 303–304), Huntley (1993: 154–155), Scatton (1993: 211),
Mirčev (2000: 117–118), Grković-Major (2011), who also cites abstract and delexical uses in
OB/OCS, such as ‘have peace’, ‘have faith’, ‘have pain’ (i.e. ‘be sick’). Načeva (1988) outlines the
wide range of meanings and functions of this verb in Modern Bulgarian.
Probably the major strand of grammaticalisation for this verb involves

the expression of futurity. As mentioned above, before it was fully grammati-
calised as a future marker, the OB/OCS auxiliary имѣти/imĕti inevitably had
some additional connotations of necessity or obligation, among others, not
unlike semi-modal English have to.8 Though they have not been completely
wiped out, those connotations have now been toned down in the negative
uninflected particle няма/njama, which is the regular pure future marker in
negative clauses. Няма/Njama is a contraction of the negator не/ne and the
third person, singular, present tense of имам/imam. It is used for all persons
in both numbers. At the same time there are still combinations in present-day
Bulgarian where имам/imam ‘have’, either marked for person and number or
appearing as a default 3sg form, simultaneously expresses futurity and modal-
ity without much trace of its original meaning of possession, as in (12), which
is analogous to (10) above.9
(12) има/ има-м да ход-я

ima/ ima-m da xod-ja
have[pres.3sg] have-pres.1sg comp go-pres.1sg
‘I will go’/‘I (will) have to go’
7.4 The Have-Perfect in Bulgarian
7.4.1 Formation
Apart from the traditional be-perfect and the auxiliary functions of have dis-
cussed above, Modern Bulgarian has evolved another, quite distinct, perfect-
like construction which consists of a form of имам/imam ‘have’ and a passive
past participle of a transitive verb, which ends in -n or -t and agrees with the
object – (13a). This newcomer is practically interchangeable with the older be-
perfect with an active past participle – (13b).10
8 See Balan (1957: 23–24), Duridanov et al. (1991: 304), Huntley (1993: 154), Mirčev (2000:
118). Cf. Chapter 3, Section 3.5.2.3, as well as the grammaticalisation tendencies from
Chapter 2, Section 2.2.
9 Georgiev (1976: 310) treats this as a composite future tense.
10 See Georgiev (1957/1976: esp. 294, 300), Kuteva and Heine (2004: 43–44), and Nicolova
(2008: 270, 2017: 379–381); on forming the past passive participle, consult Tilkov et al.
(1983a: 380–382) and Danylenko (2005: 352–357). This construction has been around for
a while, especially in some dialects – see Elliott (2001), Xaralampiev (2001: 144–145), and
the sections below. Bybee and Dahl (1989: 70–73) supply similar examples from Czech,
Middle High German and Romance. Kuteva and Heine (2004) and Danylenko (2005) dis-
cuss analogous constructions in Northern Russian dialects and other languages. Chapter 3
246 chapter 7
(13) a. има-м написа-н-о писм-о

ima-m napisa-n-o pism-o
have-pres.1sg write-pptcp-nt.sg letter(nt)-sg
‘I have a written letter’/‘I have a letter written’/‘I have written a letter’
b. написа-л съм писм-о

napisa-l săm pism-o
write-pptcp[m.sg] be.pres.1sg letter(nt)-sg
‘I have written a letter’11
Apart from the passive past participle in -n or -t, an имам/imam-construction

is also available with an active past participle which ends in -l – (14).
(14) Той има остана-л-а от баща му

Toj ima ostana-l-a ot bašta mu
he have[pres.3sg] remain-pptcp-f.sg from father his
едн-а къщ-а.
edn-a kăšt-a.
one-f.sg house(f)-sg
‘He has a house left from his father.’ [original from Penčev (1968: 177)]
This is different from (13), however, in that the subject (he) in (14) is not inter-
preted as the agent of the participial verbal element (left/having remained).
By contrast, if има/ima participates in an impersonal existential construction
functionally equivalent to English there is, the have-(proto-)perfect with an ac-
tive past participle in -l, (15a), has the same meaning as the established be-
perfect, (15b) (Bojadžiev 1968: 461–462).
presented a parallel state of affairs in OE. Have-perfects in Polish are mentioned in Łęcki
(2010: 168–169):
(i) Mam ten artykuł napisany. [Polish]
I.have this.m.sg article(m)[sg] written.m.sg
‘I have this article written.’/‘I have written this article.’ [from Łęcki (2010: 168)]
11 The word order of the auxiliary and the lexical verb is relatively free, though also governed
by the principles of clitic placement. See Elliott (2001: 102ff. and elsewhere), Nicolova
(2008: 292–294, 2017: 415–418).
(15) a. има постъпи-л-о оплакван-е

ima postăpi-l-o oplakvan-e
have[pres.3sg] arrive-pptcp-nt.sg complaint(nt)-sg
‘A complaint has been received’/‘There has been received a complaint’
[from Bojadžiev (1968: 461–462)]
b. постъпи-л-о е оплакван-е
postăpi-l-o e oplakvan-e
arrive-pptcp-nt.sg be.pres.3sg complaint(nt)-sg
‘A complaint has been received’ [from Bojadžiev (1968: 461–462)]
Extensive reports of the new perfect with имам/imam and the past passive
participle in the specialist literature date back to the 1930s and onwards, but
the pattern receives a mention even in the late 19th century.12 Georgiev (1976:
294) describes these combinations as typical of spoken, as opposed to written,
Bulgarian, in spite of their being frequent in officialese, reports, memos, etc.
They are also attested in the fiction of late 19th and early 20th-century writers,
such as Hristo Botev and Ivan Vazov, as in (16).
(16) Трет-и има-т вече втораче-н орлов-и-я

Tret-i ima-t veče vtorače-n orlov-i-ja
third-pl have-pres.3pl already fix-pptcp[m.sg] eagle-m.sg-def
си поглед в спокойния си стол.
si pogled v spokojnija si stol.
own stare(m)[sg] in peaceful own chair
‘And a third group have already set their aquiline sights on their peaceful
chairs.’ [Vazov, Draski i šarki, IХ, 214; quoted in Georgiev (1976: 295)]
As a perfect, (17) is still embryonic at best – it might more plausibly be analysed

as an impersonal existential construction with a passive meaning due to the
past passive participle втъкната/vtăknata. It is worth noting that the object
consists of a coordination of two NPs.13 One of them is feminine singular, and
the other neuter singular. In those circumstances, the participle agrees with
the controller closest to it.
12 See Popova (1931), Balan (1957), Georgiev (1957/1976); cf. Kostadinova (2006: 153, 2009: 487,
with references) and Stoevsky (2006).
13 Impersonal existential constructions with има/ima ‘have’ take objects and there is no
overt subject. See Tilkov et al. (1983b: 89–90).
248 chapter 7
(17) Над куностас-а има-ше в-тъкна-т-а

Nad kunostas-a ima-še v-tăkna-t-a
above iconostasis-def have-imperf.3sg in-tuck-pptcp-f.sg
освете-н-а босилков-а китк-а и върбов-о
osvete-n-a bosilkov-a kitk-a i vărbov-o
consecrate-pptcp-f.sg basil-f.sg bunch(f)-sg and willow-nt.sg
клонче от цветната неделя
klonče ot cvetnata nedelja
twig(nt)[sg] from Palm Sunday
‘Above the iconostasis had been tucked in a bunch of consecrated basil
and a willow twig from Palm Sunday.’ [Vazov, Under the Yoke; quoted in
Georgiev (1976: 295)]
The sentence in (18) is characteristic of the administrative register:
(18) Какв-и научн-и труд-ове има-те завърше-н-и

Kakv-i naučn-i trud-ove ima-te zavărše-n-i
what-pl academic-pl paper(m)-pl have-pres.2pl finish-pptcp-pl
от 1944 г. до днес?
ot 1944 g. do dnes?
from 1944 until today
‘What academic papers have you written since 1944?’ [memo from 1948,
quoted in Georgiev (1976: 295)]
Like many of the other sentences above, (19) is interesting due to its word
order. The participle has moved next to the finite verb. The adjacency of the
two verbal forms can be taken to indicate a greater degree of ‘amalgamation’,
as discussed in the sections on OE in Chapter 3. More advanced grammati-
calisation could further be deduced from the future time context. Likewise
non-factual, a negative clause is provided in (20), although the subject of ‘have’
is not necessarily the implied agent of the participle (see Kostov 1982: 159).
(20) might show that semantic bleaching had commenced by 1863, because
no orders have been issued yet. However, one could still take ‘have’ here in
its original meaning of possession, albeit of something abstract, to the ex-
tent that orders can be given, taken or received: ‘I don’t have any orders yet’
(Steven Kaye, p.c.). In either case, ‘have’ is rather dynamic in the context of
receiving or having orders and is a delexical verb at the very least. It must also
be significant that ‘nothing’ and the participle are not adjacent as one object
constituent.
(19) Да не си представя-ме, че през месец

Da ne si predstavja-me, če prez mesec
comp neg refl imagine-pres.1pl comp during month
ноември ще има-ме изпълне-н-и тия
noemvri šte ima-me izpălne-n-i tija
November fut have-pres.1pl implement-pptcp-pl these
план-ове.
plan-ove.
plan(m)-pl
‘Let us not imagine that we will have implemented those plans by
November.’ [from Georgiev (1976: 296)]
(20) Да прощава-те за писм-о-то, зачтото

Da proštava-te za pism-o-to, začtoto
comp forgive-pres.2pl for letter(nt)-sg-def because
ничто н-яма-м още нареде-н-о
ničto n-jama-m ošte narede-n-o
nothing(nt)[sg] neg-have-pres.1sg yet order-pptcp-nt.sg
‘Please forgive me for the letter, because I haven’t received any orders yet’
(lit. ‘I don’t have anything ordered yet’) [Hristo Botev, letter to N. Gerov,
1863; quoted in Georgiev (1976: 297)]
7.4.2 Agreement
Georgiev (1957/1976) has an extensive collection of examples of the fledgling
have-perfect, almost all of which show agreement between the object and the
participle. This is apparent in the sentences provided so far. One of the rare
exceptions with a breakdown in agreement is cited in (21), where the potential
controller (the direct object) is feminine singular, but the participle is neuter
(cf. Danylenko 2005 for East Slavonic analogues).
(21) Ние има-ме от минало-то констатира-н-о

Nie ima-me ot minalo-to konstatira-n-o
we.nom.1pl have-pres.1pl from past-def record-pptcp-nt.sg
едн-а практик-а.
edn-a praktik-a.
one-f.sg practice(f)-sg
‘We have recorded a certain practice from the past.’ [from Georgiev
(1957: 36)]
250 chapter 7
An intermediate stage between clear examples where agreement is present,

and examples where there is obviously no agreement, is probably in evidence
in (22)–(23), where words are cited, not used. In those cases, the participle
again defaults to its neuter singular form.
(22) Има-м отбеляза-н-о „чиновник” вместо

Ima-m otbeljaza-n-o “činovnik” vmesto
have-pres.1sg note-pptcp-nt.sg clerk(m)[sg] instead.of
„държавен служител”.
“dăržaven služitel”.
civil[m.sg] servant(m)[sg]
‘I have noted down “a clerk” instead of “a civil servant”.’ [from Georgiev
(1957: 37)]
(23) Тук има-ме посоче-н-о елен-рогач …

Tuk ima-me posoče-n-o elen-rogač …
here have-pres.1pl point.out-pptcp-nt.sg red-deer(m)[sg]
‘It says here “red deer”’ [Georgiev (1957: 38)]
Sometimes the ‘object’ is not realised by a NP, which indicates further gram-
maticalisation, since lexical ‘have’ does not license such non-NP dependents.
The PP in (24) is less closely related to the participle than a direct object NP
would be; moreover, PPs themselves possess neither gender nor number. The
complement in (25) is a clause, again with no gender or number features,
which inevitably triggers default neuter singular morphology on the participle
(cf. Balan 1957: 23–24). It is interesting to point out that when the participle
fails to agree, it does not resort to an ambiguous zero marker (which was the
case in Old English). Instead, the non-agreeing option utilises the neuter sin-
gular, which is clearly and unambiguously indicated with -o.14
(24) Ние има-ме предвиде-н-о като

Nie ima-me predvide-n-o kato
we.nom.1pl have-pres.1pl envisage-pptcp-nt.sg as
производствен-а практик-а
proizvodstven-a praktik-a
industrial-f.sg practice(f)-sg
‘We have envisaged placements in industry’ [from Georgiev (1957: 38)]
14 I will return to attested examples demonstrating lack of agreement in the established pas-
sive and be-perfect/renarrative (see Section 7.4.6).
(25) Има-ме предвиде-н-о да направи-м …

Ima-me predvide-n-o da napravi-m …
have-pres.1pl envisage-pptcp-nt.sg comp do-pres.1pl
‘We are planning to do …’/‘We have made plans to do …’ [from Georgiev
(1957: 38)]
7.4.3 Status of the Have-Perfect

There is no consensus regarding the status of the combinations of have + past
passive participle in Bulgarian. Georgiev (1957/1976: 299), Penčev (1968: 174)
and Nicolova (2008: 270, 2017: 378–379) believe that this construction has not
yet attained full grammaticalisation. Members of this camp maintain that
the verb имам/imam ‘have’ still retains at least some of its original concrete
meaning of possession, and the participle is still felt to be closely connected to
the object. Nicolova (2008: 270) defines this construction as a syntactic com-
bination with a predicative complement, similar to the one from the (pre-)
OE stage. After examining the question whether имам нещо направено/
imam nešto napraveno ‘have something done’ is an analytic verbal form or an
(S)VOC O clause, Penčev (1968: 173) likewise decides in favour of the latter posi-
tion, because (26a) and (26b) are arguably equivalent: they are interchangeable
and both of them show agreement, so Penčev claims that both the participle
and the adjective must serve as object complements.
(26) a. Той има отпечата-н-и дв-е

Toj ima otpečata-n-i dv-e
he.nom.m.3sg have[pres.3sg] print-pptcp-pl two-f.pl
стати-и
stati-i
paper(f)-pl
‘He has two papers printed’
b. Той има готов-и дв-е стати-и

Toj ima gotov-i dv-e stati-i
he.nom.m.3sg have[pres.3sg] ready-pl two-f.pl paper(f)-pl
‘He has two papers ready’
By the same token, however, the analogous equivalence of (27a) and (27b)
could be taken to prove that the long-established be-perfect is nothing more
than a free syntactic combination with a subject complement which can be
either participial or adjectival. There is plural agreement in both cases.15
15 See Penčev (1968: 176), Hristov (2015, with references), as well as the discussion of the
ambiguities between passive and copulative be in English in preceding chapters.
252 chapter 7
(27) a. Дв-е стати-и са излез-л-и

Dv-e stati-i sa izlez-l-i
two-f.pl paper(f)-pl be.pres.3pl come.out-pptcp-pl
‘Two papers have come out’
b. Дв-е стати-и са готов-и

Dv-e stati-i sa gotov-i
two-f.pl paper(f)-pl be.pres.3pl ready-pl
‘Two papers are ready’
In a similar vein, Georgiev (1957: 58) points out that, if such parallels are taken
seriously, then Bulgarian has not grammaticalised a passive either. In the fol-
lowing examples, the past participle also alternates with an adjective, and both
forms host agreement inflections:
(28) Обущ-а-та са изработе-н-и/ готов-и

Obušt-a-ta sa izrabote-n-i/ gotov-i
shoe-pl-def be.pres.3pl make-pptcp-pl ready-pl
‘The shoes have been made/are ready’ [from Georgiev (1957: 58)]
(29) Има-ме изработе-н-и/ готов-и обущ-а

Ima-me izrabote-n-i/ gotov-i obušt-a
have-pres.1pl make-pptcp-pl ready-pl shoe-pl
‘We have made some shoes/got some shoes ready’ [from Georgiev
(1957: 58)]
Heralding the birth of a proper have-perfect at least in some dialects and in

the colloquial language, Penčev (1968: 174) notes that the appearance of a defi-
nite article on the object suggests an analytic verbal form as opposed to a free
syntactic combination in (30), an utterance that would still be more typical of
colloquial or dialectal Bulgarian.
(30) Той го има виде-н-о

Toj go ima vide-n-o
he.nom.m.3sg it.acc.nt.3sg have[pres.3sg] see-pptcp-nt.sg
море-то
more-to
sea(nt)[sg]-def
‘He has seen it, the sea’/‘He has seen the sea’
In addition to indicating the gradual semantic extension of this construc-

tion into the domain of definite NPs (i.e. the context generalisation mentioned
in Chapter 2, Section 2.2.1), the article is useful as a ‘syntactic litmus test’ as
well. Due to the placement of -то/-to in (30), the participle is proved to belong
with the verb, and not with the object NP. In Bulgarian, the definite article at-
taches to the first nominal element in the NP, including adjectival/participial
modifiers (see (37) below). Since the article goes on морето/moreto ‘the sea’,
it must be the only constituent in the object NP in (30). Many other examples
with definite object NPs remain ungrammatical, however, so the process has
not reached its endpoint (see Kostov 1982: 164, Elliott 2001: 88ff.).
Moreover, this construction has evolved what seems to be a fully-fledged
paradigm, with past tense, (31), and renarrated/evidential forms, (32) (Georgiev
1957: 42–43, Bojadžiev 1968: 463). The latter are used to recount second-hand
information and will be revisited below.
(31) Те има-ха предвиде-н тук

Te ima-xa predvide-n tuk
they.nom.3pl have-imperf.3pl envisage-pptcp[m.sg] here
един асистент
edin asistent
one[m.sg] assistant(m)[sg]
‘They had made plans for an assistant’ [from Georgiev (1957: 42)]
(32) Той има-л предвиде-н

Toj ima-l predvide-n
he.nom.m.3sg have-pptcp[m.sg] envisage-pptcp[m.sg]
доклад по животно-въдство
doklad po životno-vădstvo
paper(m)[sg] on animal-farming
‘Apparently he was planning to give a paper on animal farming’ (renar-
rated, i.e. someone else told me) [from Georgiev (1957: 43)]
The sentence in (32) is especially interesting, since the renarrated form with
the -l participle goes back to the original be-perfect (е имал/e imal), from
which the auxiliary has simply been dropped. Essentially, two diachronic lay-
ers or generations of perfect formations coexist side by side (see Chapter 2,
Section 2.2; more on this will follow too).
Just as in Old English, there are examples which show that the lexical mean-
ing of the verb for have has been bleached:
254 chapter 7
(33) Има-ме поръча-н-и (въглищ-а) още през

Ima-me porăča-n-i (văglišt-a) ošte prez
have-pres.1pl order-pptcp-pl coal-pl already during
май, но не са ги донес-л-и
maj, no ne sa gi dones-l-i
May but neg be.pres.3pl them.acc.3pl bring-pptcp-pl
‘We already ordered some coal in May, but it hasn’t arrived yet’ [from
Georgiev (1976: 299)]
As Georgiev (1976: 299) notes, the coal is not there yet, so имаме/imame must
be an auxiliary verb. Possession in the strict physical sense is out of the ques-
tion. Имаме поръчани/Imame porăčani is more or less semantically equivalent
to the established be-perfect поръчали сме/porăčali sme.16 Such observations
lead Georgiev (1976: 300, 311) to conclude that the grammaticalisation process
is already at a rather advanced stage as of the 1950s, and that a new periphras-
tic form has arisen in Bulgarian. As already established, however, the idea of
‘having’ can be rather vague itself. Even in English, we could say We’ve got some
coal ordered, but it hasn’t arrived yet, with the tense showing that this is not the
same as the causative-like we had some coal ordered. Attested English examples
are: I’ve got 1500 Harry and Angela mugs due in this morning (attributed to a
souvenir merchant; http://newsthump.com/2017/11/28/confusion-in-germany-
at-news-that-prince-harry-will-marry-angela-merkel/, accessed on 29 Nov
2017) or She has a book coming out soon (from a linguistics talk). There is thus
clearly a sense in which the coal is theirs, they just do not have it in front of
them yet. The important point in (33) is not that the coal is not there yet. The
crucial thing is the reference to May, which does show that the construction
is used to talk about something done in the past, rather than just how things
stand now (I am grateful to Steven Kaye for alerting me to this).
7.4.4 Functional Load as an Explanation for the Rise of the Have-Perfect in

Bulgarian
Before proceeding, I summarise the major points made so far in the chapter.
In addition to possessing a perfect based on be with universal compatibility,
Bulgarian has started to grammaticalise a new perfect based on have. Georgiev
16 Balan (1957: 25), however, describes the traditional be-form съм струпал (камъни)/săm
strupal (kamăni) ‘[I] have piled up (stones)’ as an active structure focusing on the agent,
with the have newcomer имам струпани (камъни)/imam strupani (kamăni) ‘[I] have
(stones) piled up’ as its passive counterpart which takes the focus away from the agent (cf.
Nicolova 2017: 380–381, as well as example (36) below). This is reminiscent of the distinc-
tion between the have-perfect and the ‘pseudo-passive’ with have in English, as well as the
distinction between dynamic and stative.
(1976: 303) traces a development of this new construction which corresponds

step for step to the chain of events in English (cf. Kuteva and Heine 2004: 42ff.,
Stoevsky 2006: 145ff.). As outlined in earlier sections, the first stage involves a
participle which functions as a predicative complement. It is not adjacent to
the finite verb and it agrees with the direct object in number and, in the singu-
lar, gender. Incipient grammaticalisation is flagged up by moving the participle
in front of the object and placing it right after the verb, though it may still agree
with the object, as in (34). The participle is outside the boundaries of the ob-
ject NP, because it precedes the determiner/quantifier някои/njakoi ‘some’; so
it gravitates towards the verb.
(34) Аз има-м взе-т-и няко-и изпит-и

Az ima-m vze-t-i njako-i izpit-i
I.nom.1sg have-pres.1sg take-pptcp-pl some-pl exam(m)-pl
‘I have passed some exams’ [from Georgiev (1976: 303)]
The second stage sets in when the verb имам/imam loses its lexical meaning
of possession, thereby turning into an auxiliary. The book in (35) has been or-
dered but is strictly speaking not yet in the speaker’s possession, as already dis-
cussed in relation to (33) above (cf. Stoevsky 2006: 145, 149, for similar Modern
Bulgarian examples of the type ‘have something confiscated’, as well as for ob-
jections to the view that these examples really represent desemantisation, in
line with the arguments concerning (33)).
(35) Има-м поръча-н-а вече такава книг-а

Ima-m porăča-n-a veče takava knig-a
have-pres.1sg order-pptcp-f.sg already such.f.sg book(f)-sg
‘I have already ordered a book like that’ [from Georgiev (1976: 303)]
The final stage is when the participle stops agreeing with the object, instead
appearing in the invariant neuter singular form. Such structures are also cited
in Balan (1957: 25) and illustrated once again in (36).
(36) има/ има-м струпа-н-о камън-и

ima/ ima-m strupa-n-o kamăn-i
have[pres.3sg] have-pres.1sg pile.up-pptcp-nt.sg stone(m)-pl
‘There are/I have some stones piled up’ (‘Some stones have been piled
up.’/‘I have piled up some stones.’) [original from Balan (1957: 25)]
As is visible from the material, not all examples necessarily show all the gram-
maticalisation traits. Sentence (36), for instance, lacks agreement and the
256 chapter 7
participle is adjacent to the finite verb, but the possession meaning of first-
person имам/imam is arguably still there, so no semantic bleaching is in evi-
dence. Two further steps, which are yet to take place in Standard Bulgarian,
would be the extension of the construction to intransitive verbs and to inani-
mate subjects.17 The latter step might already have been initiated (at least in
embryonic form) – Stoevsky (2006: 148 fn. 1) notes Bulgarian examples of the
type ‘All computers have antivirus programs installed on them’, though agent-
hood is problematic here (cf. Kostov 2004: 128, Kostadinova 2006, 2009).
Georgiev (1976: 303) singles out the use of 3sg има/ima as an impersonal ex-
istential verb functionally equivalent to there is in English, (36)–(38), as a fac-
tor which contributed to the grammaticalisation of the new have-perfect, here
verging on a passive (though now this verb will have to serve more functions).18
(37) В рецензирани-я учебник има (=са)

V recenzirani-ja učebnik ima (=sa)
in reviewed-def textbook have[pres.3sg] be.pres.3pl
допусна-т-и и фактическ-и грешк-и.
dopusna-t-i i faktičesk-i grešk-i.
make-pptcp-pl and factual-pl error(f)-pl
‘There are also a number of factual errors in the textbook under review.’/
‘A number of factual errors have also been made in the textbook under
review.’ [1950, from Georgiev (1976: 303)]
(38) Минчо трябваше да знае какв-о има

Minčo trjabvaše da znae kakv-o ima
Minčo should comp know what-nt.sg have[pres.3sg]
закопа-н-о под воденица-та.
zakopa-n-o pod vodenica-ta.
bury-pptcp-nt.sg under mill-def
‘Minčo had to know what had been buried under the mill.’ [Ivan Vazov,
Under the Yoke, quoted in Georgiev (1976: 304)]
17 See Friedman (1976: 97–98) and Elliott (2001) for a comparison between Bulgarian,
Macedonian and Serbian, regarding the progress each of them has made along this path.
Cf. Kostov (1982: 164–165), Bužarovska and Mitkovska (2010), as well as Kostov (2004: 127–
128) and Stoevsky (2006: 147–148, 150), who additionally deal with the extension of the
pattern from perfective to imperfective verbs.
18 On impersonal има нещо/ima nešto ‘there’s something’, see Balan (1957: 24). On the
links between possessivity, existentiality, and impersonal ‘have’, see Kostadinova (2006:
154–155) and the comments in Chapter 3 about the interchangeability of have and be.
Have-passives in Bulgarian are discussed by Nicolova (2017: 380–381).
Having thus laid out the stages along have’s grammaticalisation path,
Georgiev (1976: 300–301) suspects a functional trigger for these processes. He
attributes the rise of the new have-perfect to the fact that, between the 17th
and the 19th centuries, the old be-perfect started being employed to express
non-evidential, renarrated events, under the influence of Turkish.19 The same
pattern emerges here: the old be-perfect became excessively burdened with
expressing the current result of a past action, which some Slavic scholars
label ‘relativity’ or ‘time-correlation’, as well as with renarrated, non-witnessed
events. Therefore, a new construction had to come to the rescue and make the
functional load of be somewhat lighter, thereby avoiding potential ambiguity
between resultativeness and renarration.20
Mirčev (1976), on the other hand, objects to the conclusions and the chro-
nology proposed in Georgiev’s grammaticalisation story, providing Old and
Middle Bulgarian material which predates the 17th century and features the
same putative have-perfect constructions. Mirčev (1976: 315) claims that it can
hardly be supposed that those structures have undergone any development in
Bulgarian towards the formation of a new analytic complex, as he finds it hard
to detect any semantic differences between the Old Bulgarian/OCS phrase in
(39a), and its Modern Bulgarian translation in (39b) (cf. Stoevsky 2006). An at-
tested OB/OCS example is provided in (39c).
(39) a. ѭ-же има-мь

jǫ-že ima-mь
acc.f.sg-particle have-pres.1sg
въ-ложе-н-ѫ
vъ-lože-n-ǫ
pref-place/invest-pptcp-acc.f.sg
‘I have it placed/invested’/‘I have placed/invested it’ [from Mirčev
(1976: 315)]
19 On renarration in Bulgarian, consult Gerdžikov (1984/2003). Gerdžikov (1984/2003: 252–
254) argues against the widely accepted view that Bulgarian evidentials calqued the cor-
responding Turkish forms. See Bybee and Dahl (1989: 73ff.) and Izvorski (1997) for more
general pathways of development from perfect to evidential. Narrative uses are reported
even for the perfect in English (Bowie et al. 2013, with references).
20 Cf. Gallis (1960: 184ff.) and Georgiev (1976: 300–302) for more details and some additional
points and refinements of this scenario. Similar scenarios for other languages are out-
lined in Drinka (2017: 91).
258 chapter 7
b. има-м я в-ложе-н-а
ima-m ja v-lože-n-a
have-pres.1sg acc.f.3sg pref-place/invest-pptcp-f.sg
‘I have it placed/invested’/‘I have placed/invested it’ [from Mirčev
(1976: 315)]
c. се мънасъ твоѣ ѭ-же им-ѣ-хъ

se mъnasъ tvoĕ jǫ-že im-ĕ-xъ
here pound thy acc.f.sg-particle have-past-1sg
по-ложе-н-ѫ въ оуброусѣ
po-lože-n-ǫ vъ ubrusĕ
pref-place-pptcp-acc.f.sg in napkin
‘here is thy pound, which I have kept laid up in a napkin’ [Codex
Marianus, Luke 19:20, quoted in Kostov (1982: 164) and Xaralampiev
(2001: 144), English translation from the King James Bible]
I see the early attestations of such proto-perfects as no reason to assume that

the grammaticalisation cycle has not been set in motion and come a long way
by the time of contemporary Bulgarian. Similarly to analogous sentences in
English (see Chapters 3 and 4), the Middle Bulgarian example in (40) demon-
strates the use of this construction without an NP object, which does testify to
greater grammaticalisation. The participle naturally defaults to neuter singular
when combined with a ‘clausal object’ (cf. (25) above). Still, the subject is prob-
ably not an agent in (40) (cf. Stoevsky 2006).
(40) аzъ имамь пiсан-о в с[вѧ]тѣм еvг[е]ли где

azъ imamь pisan-o v svętĕm evgeli gde
I have written-acc.nt.sg in Holy Gospel where
е(с) разпѧт
es razpęt
be.pres.3sg crucified
‘I have it written in the Holy Gospel where he was crucified’ [1359,
Германов сборник/Germanov sbornik, quoted in Mirčev (1976: 314),
Stoevsky (2006: 143)]
If the seeds of this phenomenon had already been sown prior to the Turkish in-
vasion of the Balkan Peninsula, it would be hard to motivate the appearance of
the have-perfect by saying that it arose because the old be-perfect had started
to express evidentiality as a result of contact with Turkish. Yet the have-perfect
could have existed in a low-key way, never really taking off, but it was only
when Turkish influence affected the reading of the be-perfect that it had a rea-
son to be taken up more substantially. It does not need to have been invented
after the Turkish-influenced change to the be-perfect (Steven Kaye, p.c.). The
interactions between these categories, the diachronic developments attend-
ing them and the chronology will be further examined below and in the next
chapter.
Modal uses of the original be-perfect seem to have existed very early on,
long before the modality-based category (or additional meaning) of renarra-
tion/evidentiality developed (cf. Chapter 8 and Dobrev 1973). According to
Huntley (1993: 153), if we take out of the equation doubtful examples and po-
tential scribal errors, we have attested about eighty instances of the l-participle
without an auxiliary in OB/OCS – the form now used as evidential. Huntley
(1993: 153) notes that, unlike the present and past perfect proper, when the OB/
OCS past participle stands alone without an auxiliary, it is not marked for rela-
tivity/time correlation; it is instead opposed to the aorist and imperfect tenses
in expressing modality, i.e. the narrator’s attitude to the event:
(41) ѥша и не събра-л-и събор-а … нъ

ješa i ne sъbra-l-i sъbor-a … nъ
particle and neg convene-pptcp-pl council(m)-gen.sg but
събра-шѧ събор-ъ …
sъbra-šę sъbor-ъ …
convene-aor.3pl council(m)-acc.sg
‘Would indeed they had not convened the council! … But they did
convene the council …’ [Codex Suprasliensis 386.6; cited in Huntley (1993:
153); Huntley’s translation]
Whereas the aorist provides an objective account of the event, the past par-
ticiple събрали/sъbrali expresses the speaker’s attitude, which could be a
feeling of regret (as in (41)), or rejoicing, adding a rhetorical tone to the mes-
sage (Huntley 1993: 153, cf. Dobrev 1973). The significance of this for my the-
sis is that perfect-like forms already had more than one function in OB/OCS,
including non-temporal/aspectual overtones, even before the appearance of
the modal category/meanings of renarration associated with the l-participle
(cf. Gerdžikov 1984/2003: 112–113 fn. 7, Elliott 2001: 20–21). It is therefore prob-
lematic to see the emergence or spread of the new have-perfect as a backlash
against any presumed increase in the functional load of the old construction.
According to Gerdžikov (1984/2003: 259–260), renarrated/non-witnessed
forms themselves first appear in the Bulgarian charters of Wallachia and the
Prayers of Cserged, so these forms must have originated in the late 12th and
260 chapter 7
early 13th centuries, i.e. before contact with Turkish (though it is highly de-
batable whether the Prayers of Cserged are a truthful reflection of 12th- and
13th-century Bulgarian, G. Ganeva, p.c.). Antedating non-witnessed forms to a
period before the traditional date (i.e. the 17th century) casts even more doubt
on the widely accepted chronology and causal relationships pointed out above.
7.4.5 The Have-Perfect in Macedonian

Macedonian, a South-Eastern Slavic variety which is very close to Bulgarian,21
still keeps the Proto-Slavonic perfect with сум/sum ‘be’. In addition, it has
also developed a perfect with има/ima ‘have’. First attested in 1706, this new
perfect is typical of Western Macedonian and again expresses present stative
resultativity (Friedman 1993: 270, Danylenko 2005: 367; cf. Gallis 1960, Elliott
2001: 26 for the first attestation, as well as the rest of the thesis, Bužarovska and
Mitkovska 2010, Drinka 2017: 280–282). In Eastern Macedonia, the има/ima-
perfect is used less than in the West or not at all, so the сум/sum-perfect has a
broader range of functions there, expressing both resultativeness and renarra-
tion, a distinction based on the opposition between witnessed and reported,
as noted for Standard Bulgarian above. In the south-west, the сум/sum con-
struction has been limited to the expression of renarrated events. In the rest of
Western Macedonia, the forms with сум/sum retain some of their perfect uses,
in addition to marking evidentiality/renarration.
The emergence of the Macedonian have-perfect has also been attributed
to functional load (see Elliott 2001: 43). According to Friedman (1993: 270), the
Macedonian be-perfect lost its marking for resultativity because of the rise of
the modal category of evidentiality/renarration (also known as status). The
newer have-perfect thus came to fill the void, taking over the earlier function
of be. Like Bulgarian, Macedonian now has a fully-fledged paradigm of analytic
perfect forms with both auxiliaries, with differentiated or (partially) overlap-
ping roles depending on the region. Samples of the paradigm are given below:22
(42) a. има-м моле-н-о

ima-m mole-n-o
have-pres.1sg ask-pptcp-nt.sg
‘I have asked’ [from Friedman (1993: 280), note the missing/implied
object]
21 I consciously eschew any form of political or sociolinguistic debate regarding the status of
the two varieties.
22 Cf. Balan (1957: 27–28); for a more complete verbal paradigm of Macedonian, including
be- and have-perfects, see Friedman (1993: 280–281).
b. сум моли-л
sum moli-l
be.pres.1sg ask-pptcp[m.sg]
‘I have asked’ [from Friedman (1993: 280)]
(43) a. има-в моле-н-о

ima-v mole-n-o
have-past.1sg ask-pptcp-nt.sg
‘I had asked’ [from Friedman (1993: 280)]
b. бев моли-л
bev moli-l
be.past.1sg ask-pptcp[m.sg]
‘I had asked’ [from Friedman (1993: 280)]
etc.
Macedonian is much more advanced than its eastern neighbour in the gram-
maticalisation of the have-perfect.23 Example (44) illustrates the first stage
along the grammaticalisation path, where the participle still agrees with the
object. Note though that the finite verb and the participle are already next to
each other. However, the later stage in which agreement is absent by now pre-
vails in Macedonian (Georgiev 1957: 44; cf. Balan 1957: 28). This is illustrated
in (45).
(44) има-м бележа-н-и белешк-и

ima-m beleža-n-i belešk-i
have-pres.1sg note.down-pptcp-pl note(f)-pl
‘I have made some notes’ [original from Georgiev (1957: 44)]
(45) има-ме сторе-н-о клетф-а мегьу нас

ima-me store-n-o kletf-a meg’u nas
have-pres.1pl do-pptcp-nt.sg oath(f)-sg between us
‘We have sworn an oath between us’ [original from Georgiev (1957: 44)]
Interestingly, Macedonian has gone a step further than Bulgarian in extending

the have-construction to intransitive verbs, including verbs of movement, as in
23 See Georgiev (1957: 44), Gallis (1960), Elliott (2001) and Kuteva and Heine (2004: 45); for
more on the perfect in Macedonian, consult Friedman (1976, 1993: 270–272), as well as
Bužarovska and Mitkovska (2010).
262 chapter 7
(46). This would be impossible in Standard Bulgarian (cf. Section 7.4.7 below,
as well as Chapter 2, Section 2.2.3).
(46) яска има-м оде-н-о

jaska ima-m ode-n-o
I.nom.1sg have-pres.1sg go-pptcp-nt.sg
‘I have gone/walked’ [original from Georgiev (1957: 44)]
As Georgiev (1957: 44) observes, there are only rare examples of lack of agree-
ment in Bulgarian (as of the 1950s), and there are no occurrences whatsoever
of the have-construction with intransitive verbs (at least in the standard liter-
ary language). As pointed out above, the have-perfect is most fully grammati-
calised in the south-western dialects of Macedonian, from where it spread
towards the north-east (see Elliott 2001: 43ff. and elsewhere, Bužarovska and
Mitkovska 2010, with references), and perhaps only then into Bulgarian, where
it is still much less entrenched. This geographical distribution, coupled with
the rarity of have-perfects outside of Western Europe (Drinka 2017, with refer-
ences), lends credibility to a contact scenario regarding the emergence of this
pattern on South Slavonic territory. It must have entered from the south-west
and diffused eastwards.
7.4.6 Agreement and Grammaticalisation Revisited

As stated above, the original construction from which the South-East Slavic
have-perfect evolved was an SVOC O clause in which the past participle function-
ing as object complement agreed with the object. This is still evident in many
of the examples above and is once again illustrated with the OB/OCS material
in (47), an OVSA PLACE C O configuration with matching in number, gender and
case between the word for ‘planted’ (CO) and the noun denoting the fig tree (O).
(47) O V S APLACE
... смоковьниц-ѫ имѣаше единъ въ виноградѣ своемь
… smokovьnic-ǫ iměaše edinъ vъ vinogradě svoemь
fig.tree(f)-acc.sg had one in vineyard own
CO
въсажде-н-ѫ
vъsažde-n-ǫ
plant-pptcp-acc.f.sg
‘A certain man had a fig tree planted in his vineyard’ [Luke 13:6, English
translation from the King James Bible]24
24 The adjunct of place might actually be a dependent of the participle, not a separate con-
stituent on the level of the clause. It is also worth noting the non-adjacency of the finite
Both Bulgarian and Macedonian later lost their case declensions, but ad-
jectival number and gender survive. Agreement in this type of periphrastic
construction sometimes disappears and the participle can default to neuter
singular even when the object is, for instance, plural or feminine singular,
as in the Modern Bulgarian example in (36), or the Macedonian one in (45).
Unlike the material from English, the morphological marking in Bulgarian
and Macedonian is usually clear and there are hardly any cases which are am-
biguously interpretable as showing either presence or absence of agreement
(though see (51)). Lack of matching adjectival inflection was mentioned in
the previous sections as one of the factors which point to greater grammati-
calisation of the have-perfect. It has also been observed in other, arguably well
established, structures; they are said to be fully grammaticalised but neverthe-
less might on rare occasions admit mismatched/default agreement, although
normally standard agreement would be typical of them.
Agreement can default in the established Bulgarian be-perfect, as in (48)–
(49). The intransitive clause in (48) is especially curious, since the participle
has neuter singular -o (probably under the influence of subjectless neuter sin-
gular impersonal constructions; cf. Kovatcheva 1982: 5), while the expected
standard ending would have been the zero of the masculine singular in order
to agree with the subject ‘rain’. This is the exact opposite of what we saw in
English – here, a zero exponent showing agreement has been replaced by non-
zero marking clearly indicating a mismatch. In (49) and (50), a neuter singular
participle fails to copy the features of its feminine singular or feminine plural
subject.
(48) Навън е валя-л-о дъжд

Navăn e valja-l-o dăžd
outside be.pres.3sg rain-pptcp-nt.sg rain(m)[sg]
‘It has rained outside’/‘It has been raining outside’ [from G. Gospodinov]
(49) мина-л-о е годин-а

mina-l-o e godin-a
pass-pptcp-nt.sg be.pres.3sg year(f)-sg
‘A year has passed’ [Facebook post from 30 March 2016]
verb and the participle in both the OB/OCS and the Early Modern English translations.
Importantly, the Greek original from which the OB/OCS was translated, following it close-
ly, has the same type of construction with have (see Drinka 2017: 97, 296–297). I return to
this in Section 7.4.7.
264 chapter 7
(50) Като се мина-л-о едн-а-дв-е годин-и ...

Kato se mina-l-o edn-a-dv-e godin-i …
when refl pass-pptcp-nt.sg one-f.sg-two-f.pl year(f)-pl
‘After one or two years had passed …’ [original from Mladenov
(1979: 337)]
As demonstrated in (50) above and (51) below, the same breakdown might
happen in the context of stand-alone l-participles used for the purpose of re-
narration (ultimately derived from the perfect construction), as well as in the
well-established passive voice – (52).
(51) Това би-л-о едн-а от причин-и-те

Tova bi-l-o edn-a ot pričin-i-te
this.nt.sg be-pptcp-nt.sg one-f.sg of reason(f)-pl-def
‘That was one of the reasons’ [New Symphony Orchestra poster,
April 2016]25
(52) голям грях е сторе-н-о.

goljam grjax e store-n-o.
big[m.sg] sin(m)[sg] be.pres.3sg do-pptcp-nt.sg
‘A grave sin has been committed’ [Elin Pelin, Zanemelite kambani
‘The Silent Bells’]
In (52), the attributive adjective mirrors the features of the noun, but the pred-
icative participle does not. As in (48), it chooses neuter singular -o in preference
to masculine singular zero (cf. analogous disparities between the attributive
and predicative domain in OE from Chapter 4).
All of the above considerations cast doubt on the influence of agreement on
the grammaticalisation of a new have-perfect in Bulgarian, or its unassailable
usefulness as a diagnostic. Maybe the abandonment of matching adjectival
25 It is conceivable that било/bilo might agree with the neuter singular demonstrative това/
tova, which is also a potential subject candidate in this copular clause. Certainly, how-
ever, feminine singular била/bila is standardly acceptable. A more general case could be
made that neuter singular participles with neuter singular agreement controllers are not
clear and can in theory represent either presence of agreement or default neuter singular
non-agreement where the (irrelevant) controller just happens to match the features by
accident (see Elliott 2001: 27); the latter position strikes me as unjustifiable in the light of
the rest of the material.
inflections is not exclusively a signal of greater amalgamation in this specific

construction, but a phenomenon which happens naturally across the board
in analogous syntactic environments. This will have to be verified with cor-
pus research aimed to ascertain whether agreement markers are dropped to
the same extent in the various relevant contexts (see the next chapter, where
this appears to be the case). Potential qualitative and quantitative differences
will have to be detected and confirmed – for instance, the clauses in (48)–(52)
are all in a sense ‘impersonal’ or ‘existential’, and the subjects are inanimate.
Follow-up research will help to find an answer to the question whether and to
what extent the disappearance of agreement is due to greater grammaticalisa-
tion of the new have-perfect or whether and to what extent it is a more general
phenomenon driven by other linguistic or extra-linguistic forces.
In either case, going from perfectly clear marking to equally clear neuter
singular defaults here, without any intermediate stage of ambiguity, suggests
that agreement in English would have decayed anyway as grammaticalisation
went on, and the role of the ambiguous zero marking as an enabling factor
has been overstated. In other words, grammaticalisation and reanalysis are not
necessarily conditional on ambivalent ‘bridging’ contexts of the Old English
type. It is much more likely that reanalysis and subsequent grammaticalisation
were triggered by natural cognitive abilities to see things in context from more
than one perspective, regardless of morphological cues, but probably aided by
external events such as contact (cf. next section). Both the language-internal
analysis of Old English (Chapter 4) and the cross-linguistic comparison be-
tween English and Bulgarian/Macedonian discredit the role of unclear zero
marking in this process. Loss of inflection is an outward consequence of gram-
maticalisation, not a necessary precondition for it.
7.4.7 Mixing of Active and Passive Participles and Language Contact as

Explanations for the Have-Perfect in Bulgarian and Macedonian
In Chapter 3, I recorded non-functional explanations for the development of
the English perfect, citing researchers who detect or suspect contact-induced
changes. Similar scenarios have been proposed regarding South Slavic. Popova
(1931), Georgiev (1957: 45), Mladenov (1979/1929: 292) and Elliott (2001: 49),
among many others, see Romance influence on the have-perfect constructions
in Bulgarian and Macedonian. This influence is linked to the mixing of the past
active and past passive participles evident in (53a), where past passive suffixes
(-t/-n) are attached to the stems instead of the usual active l-shapes in (53b).
266 chapter 7
(53) a. той е падна-т/

toj e padna-t/
he.nom.m.3sg be.pres.3sg fall-pass.pptcp[m.sg]
дойде-н
dojde-n
come-pass.pptcp[m.sg]
‘He has fallen/come’ [only in dialectal Bulgarian]
b. той е падна-л/
toj e padna-l/
he.nom.m.3sg be.pres.3sg fall-act.pptcp[m.sg]
дошъ-л
došă-l
come-act.pptcp[m.sg]
‘He has fallen/come’ [standard Bulgarian]
The meaning is active in either case and this contamination of the two types of
participle in the South Slavic be-perfect might have occurred since there were
no separate active and passive past participial forms in Latin/Romance. It is
certain that ambiguity and zero marking cannot have played a part in this re-
placement due to the transparent morphology. If anything, the mix-up disrupts
hitherto functionally and formally clear categories, blurring the distinction be-
tween perfect and passive. This change flies in the face of ‘optimal function-
ality’ since it has created be-perfects formally identical to be-passives (both
constructions are also available with transitive verbs, as in the Macedonian
яден сум/jaden sum, which can now mean ‘have eaten’ or potentially ‘be eaten’,
though a prefixed stem, изеден/izeden ‘eaten’, might be preferred for the pas-
sive, D. Jangelovski, p.c.; similarly, сум омажен/sum omažen ‘have/be married’
can be active or passive – see Elliott 2001: 50ff., 64). Once again, functionality
cannot be a cross-linguistically explanatory notion since distinctions have been
dismantled here rather than upheld by relying on the hitherto formally distinct
participles. This formal coalescence further undermines any suggestions that
English be-perfects were sacrificed in order to achieve more optimal form-to-
function mapping of perfects and passives and points to grammatical change
that is much more mechanistic (i.e. indifferent to functional considerations),
driven by natural cognitive abilities to reinterpret linguistic structure and/or
by contact. For comparison, the reader is referred to Chapter 3, esp. Sections
3.1.2, 3.2, 3.4.4, where remarks were made about the nature of participles in OE,
including their reinterpretation in terms of voice from adjectival and passive
(‘I had him bound’) to verbal and active/perfect (‘I had bound him’). A pertinent
concept regarding such reanalysis is that of a converb, an adverbial participle

(деепричастие/deepričastie in the Slavic tradition; Haspelmath 1995). Before
becoming fully verbal in an active perfect construction, an originally adjectival
(passive) participle might go through an adverbial stage.
Going back to Bulgarian morphology, почерня-л/počernja-l ‘[has] turned
black’ is a past active participle, while почерне-н/počerne-n ‘[has been] turned
black’ is a past passive participle. Past active participles are formed from intran-
sitive verbs (почернея/počerneja ‘become/turn black’ > почерня-л/počernja-l
‘[has] become/turned black’), as well as transitive ones (почерня/počernja
‘make/turn something black’ > почерни-л/počerni-l ‘[has] made/turned some-
thing black’), whereas past passive participles are normally formed from tran-
sitive verbs only (почерня/počernja ‘make/turn something black’ > почерне-н/
počerne-n ‘[has been] made/turned black’),26 though of course a verb can have
both transitive and intransitive uses. Горя/Gorja ‘burn’, for instance, can fit
both valency patterns and thus has the two participial shapes: горял/gorjal
‘[has] burned’ and горен/goren ‘[has been] burned’. According to Mladenov
(1979/1929: 292), this type of verb, reinforced by the lack of such a morphologi-
cal voice distinction in Romance, is the source of dialectal Modern Bulgarian
intransitive participles in -n/-t instead of -l, as in (53a). This in turn could have
facilitated or at least reinforced the new have-perfect. In addition, Mladenov
(1979/1929: 292) makes a more general case for Romance, more specifically
Aromanian,27 influence on the emergence of dialectal perfects with имам/
imam ‘have’ + past passive participle typical of the Macedonian area, suppos-
edly transferred or calqued from Romance and displacing the more traditional
pattern made up of съм/săm ‘be’ + past active participle.28
26 The discussion and the examples are based on Mladenov (1979/1929: 292). See Nicolova
(2008: 437–438) for some more special cases and exceptions. Modern Bulgarian verbal
morphology is in certain respects reminiscent of that of OE, especially in the marking of
transitivity.
27 Aromanian is a member of the Eastern Romance group spoken as a minority language
in Greece, Macedonia, and Thrace. See https://www.ethnologue.com/language/rup (ac-
cessed on 29 April 2016).
28 See Gallis (1960), Kuteva (1998), Kuteva and Heine (2004: esp. 41, 46, 51–56, 66–67),
Danylenko (2005) and Drinka (2017) for the role of contact in the development of an
analogous construction in Northern Russian and more generally across Europe (includ-
ing critical appraisal). Croft (2000: 146) mentions a have-perfect in Polish (used as a past
tense), which according to him arose due to interference from German. There have also
been suggestions that the Germanic have-perfect itself evolved under the influence of
Romance (or ultimately Greek) – see Drinka (2003, 2017), Kuteva and Heine (2004: 41),
Danylenko (2005), Łęcki (2010: 226 n. 12), Macleod (2012: 18–19, also providing convinc-
ing evidence against this hypothesis), Johannsen (2016: Section 3). While doubting the
more radical hypothesis that the Germanic perfect is wholly due to Latin, Macleod (2012:
268 chapter 7
Trying to shift the focus further east, Bojadžiev (1968) deplores the fact that
so much attention has been paid to the have-perfect in western, especially
Macedonian, varieties of South-Eastern Slavic and neglected elsewhere. He
stresses that it also exists in the (eastern) Thracian area of Bulgarian and is
indeed all-pervasive in this dialect continuum. Bojadžiev (1968: 462–463) re-
ports similar confusion of the two participles (past active and past passive)
in the Bulgarian dialects of Thrace and joins Mladenov (1979/1929: 292) in
seeking a foreign influence on the development of the have-perfect in Thrace,
this time influence coming from Greek (consult Gallis 1960, Asenova 1987,
Xaralampiev 2001: 144–145, Stoevsky 2006: 142, as well as Drinka 2017: 284, fea-
turing intransitive have-perfects with past passive participles from Thrace –
имам дойдено/imam dojdeno ‘I have come’; cf. Elliott 2001: 79ff., 94, for some
Romanian-influenced dialects of Bulgarian with a more grammaticalised pos-
sessive perfect, hence multiple points of entry for this construction are quite
possible).
Dialectal data with non-agreeing have-perfects from the Thessaloniki region
recorded in the 1970s is provided in (54) and (55). The participle ‘hung’ in (54)
might be given an adverbial interpretation, providing information about how
(in what state) the bag was ‘had’ – i.e. it was ‘had’ in a ‘hanging state’, so some
people would treat this as a converb stage in the grammaticalisation of the
have-perfect (see above). (55) is arguably a little more verbal.
(54) Турб-аʹ-та ут лаʾʹбут чуфчиʹйта йа имаʾʹши

Turb-аʹ-tа ut lаʾʹbut čufčiʹjta ja imаʾʹši
bag(f)-sg-def from the.bread the.farmer f.sg.acc had
закаʹцин-у на идноʹ драʾʹву.
zakаʹcin-u na idnоʹ drаʾʹvu.
hung-nt.sg on one tree
‘The farmer had hung the bag from the bread on a tree’ [original from
Bălgarska dialektologija. Xristomatija by Jordan Ivanov, Plovdiv, 1978,
quoted at https://e-medievalia.uni-sofia.bg/moodle/mod/page/view
.php?id=2031, accessed on 28 Feb 2017]
170–172) concedes that there might be indications of at least some influence, albeit lim-
ited, of mediaeval Latin on European vernacular usage, including OE. The emergence of
perfects with habere ‘have’ in Latin itself (basically identical in every step of the process)
is discussed in Hopper and Traugott (2003: 62–65) and Drinka (2017: Chapter 6).
(55) А бе, жеʹну, пъʹй тас раʹбут-а йа

А bе, žеʹnu, păʹj tas rаʹbut-а jа
injc injc wife injc this.f.sg job(f)-sg f.sg.acc
иʹма-ме стоʹрин-у
iʹma-me stоʹrin-u
have-pres.1pl done-nt.sg
‘Hey, wife, but we’ve done this job’ [original from Bălgarska dialek-
tologija. Xristomatija by Jordan Ivanov, Plovdiv, 1978, quoted at https://
e-medievalia.uni-sofia.bg/moodle/mod/page/view.php?id=2031,
accessed on 28 Feb 2017]
The orthography in (54) and (55) demonstrates a replacement of unstressed

Standard Bulgarian о/o with у/u, e.g. турбаʹта/turbáta – Standard Bulgarian
торбáта/torbáta ‘the bag’, ут/ut – Standard Bulgarian от/ot ‘from’, лаʾʹбут/
l’abut – dialectal or earlier хлябот/xljábot ‘the bread’, драʾʹву/dr’avu – Standard
Bulgarian дървó/dărvó ‘tree’ (with metathesis and different stress placement),
feminine vocative singular жеʹн-у/ žеʹn-u – standard жéн-о/žén-o ‘wife’, etc.
This evidence points to neuter singular participles in both (54) and (55), which
in Standard Bulgarian spelling would be represented as ending in -o, here re-
duced and merged with /u/ in unstressed position (cf. Section 7.2).29
Tracing the auxiliation of have back to the Middle Ages, one finds (proto-)
have-perfects in OB/OCS translations of analogous Greek originals with εχω/
ekho + past participle, including (39c) and (47) above (see Kostov 1982: 160,
164; cf. Drinka 2017: 291ff. regarding the extent to which OB/OCS translations
consistently match the Greek originals). In a similar vein, Minčeva (1987: 27–
30) talks about possible Greek and Romance influence on the development of
future periphrases with verbs meaning ‘want’ and ‘have’ in Bulgarian (cf.
Section 7.3 above), so such prospective influence scenarios have a wider basis.
It is not at all implausible that this type of perfect periphrasis spread in the
Balkans, or in fact in all of Europe, as a result of language contact, though this is
hard to test empirically (cf. Drinka 2017, as well as Chapter 2, Section 2.2.3). The
bulk of the evidence in favour of contact is supplied by the variable geographi-
cal distribution and entrenchment of have-perfects on Bulgarian/Macedonian
territory (see the section on Macedonian above), as well as the areal patterning
29 In principle, certain instances of /u/ in some dialects could be a reflex of the OB/OCS
back nasal, including remnants of accusative feminine singular inflections, but this is
unlikely here.
270 chapter 7
of different perfects in the rest of Europe (as noted in Chapter 6), coupled with
the fact that this construction is rare outside of Western Europe. The differ-
ent participial profile of Eastern languages with two past participles (active
and passive) is significant too, as is mixing them up despite their morphologi-
cal distinctness – they are thus brought into line with those of the Romance
and Germanic languages. All of this is suggestive of Western European and/
or Greek influence on the emergence of this new perfect. The textual record,
including translation practice, furnishes further proof for external models.
7.5 Conclusion
For centuries, Bulgarian has relied on a construction with be to form the per-
fect of all verbs, transitive and intransitive alike. It has more recently started to
grammaticalise have for the same role, as has the closely related Macedonian,
where the process has gone further. Formally, the new South Slavic have-perfect
construction looks almost identical to the complex transitive proto-perfects in
Old English. However, Bulgarian and Macedonian normally distinguish past
active from past passive participles and do not exhibit cases of ambivalent
zero exponence comparable to Old English. The fact that the same reanaly-
sis is nevertheless underway in these Slavic varieties and the agreement has
started to default (especially in Macedonian) indicates that grammaticalisa-
tion/reanalysis does not depend on formal prerequisites such as ambiguous or
missing morphosyntactic marking. The dialectal confusion between the for-
mally distinct active and passive past participles confirms this impression and
probably points to Western European influence, where only one all-purpose
past participle is available.
Earlier scholars have attributed the rise of the new have-perfect to the fact
that Bulgarian/Macedonian came to use the old be-perfect forms to renarrate
non-witnessed events, probably under the influence of Turkish. Be supposedly
became too overloaded with expressing perfect and evidential meanings so the
time was ripe for change. This functional overload account faces serious prob-
lems, just as it did for English. Firstly, have itself already served multiple func-
tions at the time it was recruited for the role of perfect auxiliary (see Section
7.3). Secondly, there are attestations of proto-have-perfects which predate con-
tact with Turkish, so the chronology of the functionalist story does not quite
add up. Furthermore, the Old Bulgarian be-perfect had modal uses even before
it acquired the modal connotations of evidentiality, so contact with Turkish
did not make it much more overloaded. Finally, the appearance of evidenti-
ality itself might actually predate the Ottoman conquest, which additionally
undermines the relations of cause and effect. A number of researchers have

instead appealed to another form of language contact as the source of the new
have-perfect construction, which might have originated from Greek and/or
Romance and been diffused by multilingual speakers across the Balkans (and
perhaps in all of Europe). The evidence for this scenario comes primarily from
the geographical distribution of these constructions, as well as the textual re-
cord, including mediaeval translation practice.
chapter 8
The Development of the Perfect in a Selection of

Middle and Modern Bulgarian Texts
In the previous chapter, I outlined the major trajectories in the develop-

ment of the different types of perfect in Bulgarian, including some of the
explanations offered in the specialist literature. The traditional all-purpose
be-perfect has been retained but it has also come to mark evidentiality (pos-
sibly under Turkish influence), and a new construction with have might be
being pressed into service. Here I examine a selection of texts in order to as-
certain if have-perfects appear on a regular basis in materials pre- and post-
dating the Ottoman conquest, how frequent and grammaticalised they are,
and to what extent they have started to shed their agreement as compared to
other analytic constructions, including passives and established be-perfects/
evidentials. It will emerge that have-perfects remain rare, elusive and insuf-
ficiently entrenched to this day, even next to non-existent, and loss of agree-
ment is a phenomenon with a wider scope across the spectrum of relevant
syntactic environments, though still statistically negligible and unrelated to
grammaticalising this specific incoming perfect construction. An attempt will
also be made to trace the interactions between perfects and evidentials in the
light of functional load. Further comparison will be sought between perfects
and other preterital forms, namely the aorist and the imperfect tenses (both of
them synthetic rather than periphrastic).
8.1 Introduction
The problems of corpus design and analysis outlined in Chapter 4 apply in

equal measure to the mediaeval Bulgarian data. The evidence is often scanty
and far from well suited to testing linguistic hypotheses. While acknowledging
these at times insurmountable obstacles (including the ways in which they
could undermine the conclusions drawn on the basis of deficient data), as
well as other limitations of my study (notably the size of the samples), I shall
scrutinise a selection of mediaeval and early modern Bulgarian documents in
order to chart the development of the have- and be-perfects. Special attention
will be given to the interactions of the perfect with evidentiality, functional
load and language contact, as well as agreement marking and the other fac-
tors identified as potentially significant in the preceding chapter. Four types of

The Perfect in Middle and Modern Bulgarian 273
text will be investigated – a Middle Bulgarian tale composed or adapted be-

fore the Ottoman conquest (Section 8.2), a Damascene book translated in the
17th century, after several centuries of contact with Turkish (Section 8.3), as
well as an 18th-century original work of historiography on the cusp of Modern
Bulgarian (Section 8.4). I shall then fast forward to some Modern Bulgarian
corpus evidence (8.5).
8.2 The Tale of Troy (14th c.)

In the middle of the 14th century, Bulgarian Tsar Ivan Alexander (1331–1371)
commissioned a translation of the Greek Chronicle of Constantine Manasses.
It has been preserved in two contemporary manuscripts from Ivan Alexander’s
reign – one is now kept in Moscow, and the other in the Vatican Library (see
MacRobert 1981: 151, Dujčev and Lixačev 1988, Ruseva 2011: 4–6). One of the sub-
stantial additions to the Chronicle text is the so-called Tale of Troy (Bulgarian
Троянска притча/Trojanska pritča or Троянска повест/Trojanska povest).
As MacRobert (1981: 151–152) notes, the language of this Tale ‘differs strikingly
from that of the Bulgarian translation of the Chronicle’ in that the Chronicle
translation is conservative in comparison with some of the usages in the Troy
passage. The innovative nature of many of the grammatical features of the Tale
has prompted some scholars to see it as a reliable source for Middle Bulgarian
popular usage.
However, this story of Troy can also be found in other Slavonic literatures,
including two Croatian Glagolitic manuscripts from the 15th century, in addi-
tion to an early-sixteenth-century Serbian version. Although all of them are
obviously related, none can be said to have straightforwardly originated from
any of the others. There are indications that the text of the Tale ultimately de-
rives from the Glagolitic tradition. The consensus is that the original must have
been Western European, based on Latin, so it is probable that the Tale initially
sprang from a Croatian version, in favour of which hypothesis there is some
textual evidence too (see MacRobert 1981: 152–156, 179, Ruseva 2011: 8, with
references). Investigating the distribution of infinitival vs. finite subordinate
clauses, MacRobert (1981: 183) sums up that the Middle Bulgarian text
was produced from a Serbo-Croat Tale of Troy mainly by phonological

(or orthographical) and morphological adaptation, and […] grammatical
changes were made only in the rare instances where the grammar of the
source text conflicted with the rather conservative norms of fourteenth
century Bulgarian Church Slavonic.
274 chapter 8
Therefore I concur with MacRobert (1981: 156) that the Tale ‘is not to be treat-
ed uncritically as reliable evidence for fourteenth century popular Bulgarian’. It
is safer to take it to represent more general South Slavonic usage from the 14th
century, with an intermixture of Bulgarian and Serbian/Croatian features, both
literary and perhaps some spoken (see Vasilev 1972, MacRobert 1981: 156–157,
179ff., with references). MacRobert (1981: 183–185) concedes that the text may
still betray some of the developments in spoken syntax, albeit somewhat indi-
rectly, recapitulating as follows:
The Pritcha is isolated in Bulgarian literature. The few other secular nar-
ratives extant in MSS of Bulgarian redaction, the Tales of Alexander and
of Aesop, are thought to derive entirely from Greek sources and are writ-
ten in a Bulgarian Church Slavonic which lacks the modernistic pecu-
liarities of the Pritcha. In the absence of a body of comparable material,
and of detailed information about the development of Bulgarian Church
Slavonic syntax, the Pritcha can be used as a source for the history of the
Bulgarian language, literary or colloquial, only with the greatest caution.
The text examined here will be based on the Vatican MS (Cod. Vat. slav. 2),
made electronically available on the Cyrillomethodiana website (http://hist-
dict.uni-sofia.bg/textcorpus/show/doc_165, accessed on 06 March 2017) and
in Ruseva (2011), with a glossary. The results are presented in the following
section.
table 1 Perfects, conditionals, passives, aorists and imperfects in the Middle Bulgarian
Tale of Troy
With No Total % of all forms

agreement agreement recorded
Be-perfects 36 2? 38 (29 pres. perf. + 9 4.72%

pluperfects)
Be-conditionals 18 3? 21 2.61%
Be-passives 4 1 5 0.62%
Aorists 603 0 603 74.91%
Imperfects 138 0 138 17.14%
The Tale of Troy contains roughly 6,876 words (based on http://histdict

.uni-sofia.bg/textcorpus/show/doc_165, accessed on 06 March 2017). The story
is generally written in the aorist (a total of 603 forms) or the imperfect (138
tokens), though the sample yields a total of 38 perfects too, past – 9, as well
as present – 29 (see Table 1). It is noteworthy that the two synthetic preterital
forms (aorist, 74.91%, and imperfect, 17.14%) vastly outnumber the perfects
here, with only a 4.72% share of the recorded uses of the various constructions.
The periphrases with be and the past active participle were all provisionally
treated as perfects because all of them had auxiliaries. Even when more than
one participle shared an auxiliary, every participle was counted as a separate
token (unlike the English data), as in (1). It is safe to treat each participial token
as a separate (be-)perfect here, whereas doing so for earlier English would in-
evitably raise issues of category assignment, as it is not always advisable to
assume that the same auxiliary should be supplied for all ellipted forms (see,
for instance, Chapter 4, examples (30)–(31), keeping in mind the wider issue
of be vs. have competition in English, where no distinct past active and past
passive participles are available). A more practical reason for counting each
participle separately in Bulgarian is to achieve a unified treatment for stand-
alone participles acting as evidentials (see following sections, esp. the one on
the Slavonic-Bulgarian History).
(1) и по том тѧ е вьзѧ-л-ъ

i po tom tę e vьzę-l-ъ
and after that thee be.pres.3sg take-pptcp-m.nom.sg
Менелаоушь ц(а)рь, а сези си вь третиѧ
Menelaušь carь, a sezi si vь tretię
Menelaus king and now be.pres.2sg in third
постелѧ прѣлюбьі сьтвори-л-а, а мене
postelę prěljuby sьtvori-l-a, a mene
bed adultery make-pptcp-f.nom.sg and me
изгна-л-а
izgna-l-a
drive.away-pptcp-f.nom.sg
‘and afterwards King Menelaus took you, and now you have been com-
mitting adultery in a third bed, and have driven me away’1
1 The transliteration of Middle and Early Modern Bulgarian adopted here follows the same
conventions that are generally familiar from the rendition of Old Bulgarian/Old Church
Slavonic. Yet Slavonicists will be aware that some of the letters (especially those representing
276 chapter 8
As is to be expected, the narrative contains quite a few past perfects (a total

of nine), which characteristically denote past in the past, as in (2) and (3):
(2) И обрѣте его ѡвчарь старъ, емоуже бѣ

I obrěte ego ovčarь starъ, emuže bě
& found him shepherd old to.whom be[past.3sg]
жен-а роди-л-а с(ьi)на
žen-a rodi-l-a syna
wife(f)-sg bear-pptcp-f.nom.sg son
‘And an old shepherd found him, whose wife had given birth to a son.’
(3) И бѣ-хѫ плѣни-л-е гръчьстiи

I bě-xǫ plěni-l-e grъčьstii
& be-past.3pl capture-pptcp-nom.pl Greek
витез-и троискьiѫ страньi и приве-л-е
vitez-i troiskyǫ strany i prive-l-e
knight(m)-nom.pl Trojan land & take-pptcp-nom.pl
бѣ-шѫ Рижеоуш-а поп-а дьщере
bě-šǫ Rižeuš-a pop-a dьštere
be-past.3pl Chryses(m)-gen.sg priest(m)-gen.sg daughter
‘And the Greek knights had captured the Trojan land and had taken
Chryses the Priest’s daughter’
The present perfects (a total of 29) typically appear in direct speech and ex-
press resultativeness, as in (4)–(6). The same link between perfects and direct
speech, hence more colloquial language, was noted for English in Chapter 4.
This association apparently holds true of (Middle) Bulgarian as well, where
the perfect is widespread in representation of conversation (cf. the following
section too).
(4) како ми ѥс(т ҆) б(ог)-ъ пособи-л-ъ

kako mi jest bog-ъ posobi-l-ъ
as me be.pres.3sg god(m)-nom.sg help-pptcp-m.nom.sg
‘as God has helped me’
the nasal vowels, the jers and the jat vowel) probably were either no longer pronounced or no
longer had their original phonetic values in the later periods.
(5) Поусти-л-а мѧ ста оба

Pusti-l-a mę sta oba
release-pptcp-m.nom.dual me be.pres(aor?).3dual both
ц(а)р-ѣ, да мѧ прiиме-шь на опѧт
car-ě, da mę priime-šь na opęt
king(m)-nom.dual comp me take-pres.2sg on back
‘The two kings have (had?) released/sent me, so that you may take me
back’
(6) где того смь и взѧ-л-ъ

gde togo smь i vzę-l-ъ
where that be.pres.1sg and take-pptcp-m.nom.sg
‘where I took him from’
For (5) and (6), like many others, it is clear from the context that the speaker
witnessed the event described by the verb. The present perfect is also typically
used to express past experiences which could affect the present, situated in a
time period leading up to the moment of speech, as in (7):
(7) ѡт млада того н-ѣсмъ сътвори-л-ъ

ot mlada togo n-ěsmъ sъtvori-l-ъ
from young that neg-be.pres.1sg do-pptcp-m.nom.sg
‘I haven’t done this since I was young’
For at least eight or nine of the examples, however, perhaps even more (see
Appendix 10;2 cf. Vasilev 1972: 348ff., with references), the context might pro-
vide some indication that the speaker did not witness the event he or she is
describing, as in (8).
(8) тамо сѧ е обрѣ-л-о и тамо да

tamo sę e obrě-l-o i tamo da
there refl be.pres.3sg find-pptcp-nt.nom.sg & there comp
сѧ и расѫди-тъ
sę i rasǫdi-tъ
refl & judge-pres.3sg
‘that’s where it [the apple?] was found and that’s where it is to be judged’
278 chapter 8
Although this is listed as a third person singular perfect in Ruseva (2011: 96),
it might simultaneously be treated as a potential (proto-)evidential reporting
where the apple of discord was found. Arguably the gods to whom the words
in (8) are attributed were not present when the apple was discovered (though
they are probably omniscient). The perfect here is at the very least conclusive/
inferential, i.e. the speaker deduces/infers something they have not seen and
draws some conclusion based on the evidence. This is the bridge to the system-
atic non-witnessed evidential function which is to evolve later (cf. Vasilev 1972:
348ff., Gerdžikov 1984/2003: 29, 229, 231–236, 255–257, Izvorski 1997, Nicolova
2013). Gerdžikov (1984/2003) indeed notes that the perfect had contextual
non-witnessed uses as early as OB/OCS. Likewise, (9) could foreground the
resultativity but it additionally carries evidential and conclusive/inferential
meanings. It is clear from the context that the person who reports these events
did not witness them, at least not directly.
(9) Оуби-л-е с[ѫтъ] витез-и обрѣтше

Ubi-l-e sǫtъ vitez-i obrětše
kill-pptcp-nom.pl be.pres.3pl knight(m)-nom.pl found
кошоутѫ вильі пророчицѫ Фелешѫ г(оспо)ждѫ, и
košutǫ vily proročicǫ Felešǫ gospoždǫ, i
doe of.sorceress prophetess P. lady &
прогнва-л-а сѧ е на них ҆, и
prognva-l-a sę e na nix, i
anger-pptcp-f.nom.sg refl be.pres.3sg on them and
поусти-л-а е влъньі на море, да
pusti-l-a e vlъny na more, da
release-pptcp-f.nom.sg be.pres.3sg waves on sea comp
погоубить вьсѧ гръчьскьіѫ кораблѧ
pogubitь vьsę grъčьskyǫ korablę
destroy all Greek ships
‘The knights have killed the doe they found which belonged to the proph-
etess P., and she has got angry at them and has released the waves on the
sea in order to destroy all the Greek ships’
In (10) below, presumably the female speaker did not see Paris take Helen, and
only concludes that he has indeed left her for Helen (indirectly, based on the
available evidence). The aorist and the perfect appear side by side, both ex-
pressing second-hand information, rather than being harnessed for first-hand
(aorist) vs. second-hand reporting (perfect). Dobrev (1973: 11–12) too rejects
earlier claims that the aorist here simply denotes a witnessed event, whereas
the perfect denotes a non-witnessed one. This is not borne out by the story-
line. Instead, Dobrev believes that the choice of the perfect is motivated by
expressive reasons due to its greater emotional colouring and emphasis on the
relevance of the event (for the expressivity of periphrastic circumlocutions, cf.
Chapter 2).
(10) Тогьi вьзѧ Еленѫ ц(а)р(и)цѫ, а мене

Togy vьzę Elenǫ caricǫ, a mene
then took[aor.2sg] Helen queen & me
еси остави-л-ъ
esi ostavi-l-ъ
be.pres.2sg leave-ppctp-m.nom.sg
‘Then you took Queen Helen, and have [now?] left me’
As becomes clear from the contexts of the sentences or from general knowledge
(e.g. about when prophets lived), the examples in (11) do confirm that the aor-
ist can be used freely in the text to report non-witnessed events. Those include
reports based on typical hearsay evidence, (11a), counterfactual events which
would certainly require evidential forms in Modern Bulgarian, (11b), as well
as other mediated information, including the retelling of dreams, (11c)–(11d).
These uses of the aorist would all sound rather unusual in Modern Bulgarian.
(11) a. И сльiшавъ Мен[е]лаоушь ц(а)рь, iако

I slyšavъ Menelaušь carь, jako
& having.heard Menelaus king comp
вьзѧ Алеѯандръ Фарижь Еленѫ ц(а)р(и)цѫ, и
vьzę Alezandrъ Farižь Elenǫ caricǫ, i
take[aor.3sg] Alexander Paris Helen queen &
ѡтринѫ сѧ вь море, и отиде под҆ Троѫ
otrinǫ sę vь more, i otide pod Troǫ
let[aor.3sg] refl in sea & go[aor.3sg] under Troy
‘And when King Menelaus heard that Alexander Paris had taken Queen
Helen, he sailed across the sea and went to Troy …’ [typical hearsay
evidence]
280 chapter 8
b. раду-аше сѧ, мнѧ, iaко Ацилеш-а

radu-aše sę mnę, jako Acileš-a
rejoice-imperf.3sg refl thinking comp Achilles(m)-acc.sg
оуби
ubi
kill[aor.3sg]
‘he rejoiced, thinking that he had killed Achilles’ [counterfactual killing
event, which would require an evidential form in Modern Bulgarian;
cf. the same counterfactual role of the English pluperfect here]
c. И сказа емоу сънъ, еже видѣ

I skaza emu sъnъ, eže vidě
and tell[aor.3sg] him dream which see[aor.3sg]
‘And told him the dream that she’d had’ [the listener denoted by ‘him’
obviously did not experience the dream first-hand]
d. И оувѣдѣ то Полинещерь краль, како сѧ

I uvědě to Polinešterь kralь, kako sę
and find.out[aor.3sg] that Polymnestor king how refl
сконча Троа
skonča Troa
finish[aor.3sg] Troy
‘And King Polymnestor found out how(that) Troy had come to an end’
[again, second-hand information about the fall of Troy]
e. iaкоже прор(о)кь провьзвѣсти

jakože pror(o)kь provьzvěsti
as prophet announce[aor.3sg]
‘as the prophet made known’ [prophets lived at a time much earlier
than that of the (14th-century) narrator]
Thus, both perfects and aorists are unproblematically selectable for non-
witnessed events in this Middle Bulgarian text. This is certainly not typical of
the aorist today. Only the perfect retains this function now, while things were
much more indiscriminate earlier. It once again becomes clear from (12) that
the mediaeval perfect had at least a contextual use in which it could be recruit-
ed for non-witnessed events, apparently in free variation with the aorist (cf.
Gerdžikov 1984/2003: 254 for an analogous state of affairs in OB/OCS, which
lacks a morphological category of evidentiality). It might be anachronistic to
spot evidential uses at a time when both the aorist and the perfect can have
them, as the evidential is not yet a fully-fledged category but just a fact about
the situation (Steven Kaye, p.c.). In English too, it would be possible to point
to examples where the speaker has seen an event for themselves and others
where they have only heard about an event, but that barely interacts at all with
which tenses are used (Steven Kaye, p.c.). These are legitimate objections but,
despite the anachronism, I record uses like the one in (12) in view of the later
developments on the way to Modern Bulgarian.
(12) А когьі би-х҆ видѣ-л-а,

A kogy bi-x vidě-l-a,
and when would-1sg3 see-pptcp-f.nom.sg
поѫ-л си Еленѫ ц(а)р(и)цѫ и
poǫ-l si Elenǫ caricǫ i
drug-pptcp[m.nom.sg] be.pres.2sg Helen queen &
иде-шь, то пакьі умоли-л-а би-х҆
ide-šь, to paky umoli-l-a bi-x
come-pres.2sg then again ask-pptcp-f.nom.sg would-1sg
вилѫ, та би-хѫ тво-и корабл-е
vilǫ, ta bi-xǫ tvo-i korabl-e
sorceress comp would-3pl your-m.nom.pl ship(m)-nom.pl
потонѫ-л-е
potonǫ-l-e
sink-pptcp-m.nom.pl
‘And if I had seen that you had drugged Queen Helen and were coming, I
would again have asked the sorceress to sink your ships’
So the potential evidential use of the perfect can already be detected even be-
fore contact with Turkish. It is true that the auxiliary is never omitted in this
mediaeval passage the way it regularly is in Modern Bulgarian; but it should
also be noted that even in Modern Bulgarian a third-person perfect can act as
an evidential with or without omission of the auxiliary, so these alternatives
are still in variation.
3 The form glossed as ‘would’ here is a special conditional form of the verb ‘to be’. I return to it
later in this section.
282 chapter 8
Traditionally, non-witnessed evidential forms are used in Modern Bulgarian

for events that have been related by someone else (see Gerdžikov 1984/2003,
Fielder 1995: 586). The only formal difference between the perfect and the so-
called evidential construction is the absence of the auxiliary in the third per-
son. The other persons are identical. However, Fielder (1995: 587) notes that
the presence vs. absence of the auxiliary does not consistently encode the
distinction between witnessed and non-witnessed in Modern Bulgarian (cf.
Stankov 1980, Fielder 1995, and Elliott 2001: 74ff., for extended examples and
further discussion). This is confirmed by the Modern Bulgarian fiction extract
in (13), where the first l-participle is without an auxiliary, but the second one
is preceded by a form of ‘be’. Presence or absence of the auxiliary thus appears
to be random.
(13) След сватбата дядо ми не отишъ-л, както

Sled svatbata djado mi ne otišă-l, kakto
after the.wedding grandfather my neg go-pptcp[m.sg] as
винаги, по своя гурбет. Просто не е
vinagi, po svoja gurbet. Prosto ne e
always on own work.abroad simply neg be.pres.3sg
иска-л да остави сама жена си още
iska-l da ostavi sama žena si ošte
want-pptcp[m.sg] comp leave alone wife own already
през първата година.
prez părvata godina.
during the.first year
‘After the wedding my grandfather did not, as always, go abroad to work.
He simply did not want to leave his wife alone during the first year.’ [Pavel
Vežinov, quoted in Fielder (1995: 588–589), where the example continues;
her translation with my gloss]
Some speakers might feel that the first case (with omission of ‘be’) approxi-
mates something along the lines of ‘My grandfather told me so’ (renarrative),
whereas the second is more like ‘I guess so’ (conclusive/inferential), though it is
not always easy to keep this demarcation up in investigating actual usage. This
is evident in (14), taken from a corpus of contemporary spoken Bulgarian – the
first and the second attestations in bold have an auxiliary, while the third does
not, but all three of them report the same event:
(14) едно нещо, което вчера са говори-л-и и

edno nešto, koeto včera sa govori-l-i i
one thing which yesterday be.pres.3pl speak-pptcp-pl and
което е обясни-л на германската
koeto e objasni-l na germanskata
which be.pres.3sg explain-pptcp[m.sg] to German
посланичка, която попита-л-а ...
poslanička, kojato popita-l-a …
ambassador who ask-pptcp-f.sg
‘something which they discussed yesterday and which he explained to
the German ambassador, who asked …’ [source: http://www.slav.uni-sofia
.bg/bgspeech/sites/indexbg.htm, accessed on 24 July 2009]
Going back to Middle Bulgarian, example (15) from my corpus again appears
in direct speech. It demonstrates a further, less clearly defined, modal use of
the present perfect (cf. Chapter 7, Section 7.4.4, for the emotional colouring
of the OB/OCS l-participle). This is a conditional context with future time
reference and modality is further suggested by the (optative?) subordinator/
complementiser да/da. This attestation fits the description of emotional uses
of l-forms from the previous chapter, as well as Dobrev’s claim above that the
perfect might impart greater expressiveness to the utterance.
(15) И ко-и сѧ сѧ двигнѫ-л-е

I ko-i sę sę dvignǫ-l-e
& what-m.nom.pl be.pres.3pl refl move-pptcp-m.nom.pl
гръцкьiѫ воевод-ьі, и [...] витез-и, да
grъckyǫ voevod-y, i […] vitez-i, da
Greek leader(m)-nom.pl & knight(m)-nom.pl comp
сие себѣ сѫтъ наиш-л-е съмръть, а
sie sebě sǫtъ naiš-l-e sъmrъtь, a
those refl be.pres.3pl find-pptcp-m.nom.pl death &
тебѣ срамотѫ, и наиш-л-о сѧ е
tebě sramotǫ, i naiš-l-o sę e
thee shame & find-pptcp-nt.nom.sg refl be.pres.3sg
мног-о г(оспо)жд-ь вь вдовичьство, аще
mnog-o gospožd-ь vь vdovičьstvo, ašte
many-nt.nom.sg lady(f)-gen.pl in widowhood if
284 chapter 8
толико не поусти-шь за своѫ дьщеръ

toliko ne pusti-šь za svoǫ dьšterъ
so.much neg let-pres.2sg for own daughter
‘And those Greek leaders that have set off and […] the knights, may they
meet their deaths, and bring you shame, and many a lady become a
widow, if you don’t give up your daughter’
In addition, the participle наишло/naišlo probably hosts neuter singular in-

flection because of the neuter quantifier много/mnogo ‘many’. OB/(O)CS
мъногъ/mъnogъ (m.nom.sg citation form) can act as an adjective with
regular declension or as a substantive/noun in the neuter singular (мъного/
mъnogo), which is followed by a partitive genitive plural – ‘a multitude of la-
dies’. This is one of the rare contexts where agreement could be said to have
defaulted, but strictly speaking, there is no genuine breakdown here and the
overwhelming majority of examples do show consistent use of agreement
markers, as illustrated in the material examined so far (cf. Table 1). Another ex-
ample which might demonstrate a more convincing situation of a mismatched
default is (8) above, repeated as (16). The referent of the missing subject is
most likely the feminine noun for ‘apple’ from the preceding discourse, but the
participle is in the neuter singular.
(16) тамо сѧ е обрѣ-л-о и тамо да

tamo sę e obrě-l-o i tamo da
there refl be.pres.3sg find-pptcp-nt.nom.sg & there comp
сѧ и расѫди-тъ
sę i rasǫdi-tъ
refl & judge-pres.3sg
‘that’s where it [the apple?] was found and that’s where it is to be judged’
Apart from the perfect, there is another structure which involves a combina-
tion of ‘be’ and a past active participle ending in -l. ‘Be’ appears in a special form
(either an old optative or an old aorist), here glossed as ‘would’, and the whole
combination expresses conditional meaning, again in the realm of modality
(for the conditional in Old Bulgarian/Old Church Slavonic, see Duridanov et
al. 1991: 312–314). There were 21 such instances in my sample (see (12) above, as
well as Table 1), confirming the multiplicity of functions of ‘be’ + past participle
constructions (though the different form of the auxiliary must surely matter).
(17) shows masculine agreement with the name of Troy. Although the actual
name ends in -a in the nominative and is therefore expected to be feminine
(cf. (11a), where Troy takes the accusative feminine singular -ѫ/-ǫ), the generic
word for city is masculine, so this might be the reason for the masculine par-
ticiples in (17), essentially an instance akin to semantic or (better) associative
agreement.4
(17) како би Тро-а не стоа-л-ъ

kako bi Tro-a ne stoa-l-ъ
how would[3sg] Troy(f?)-nom.sg neg stay-pptcp-m.nom.sg
до врѣмене, нѫ да сѧ бi
do vrěmene, nǫ da sę bi
until time but comp refl would[3sg]
разори-л-ъ вь Прѣiамош-ево врѣмѧ
razori-l-ъ vь Prějamoš-evo vrěmę
ruin-pptcp-m.nom.sg in Priam-adj time
‘[they were plotting] how Troy wouldn’t survive for a time [for very long?],
but it should be ruined in the time of Priam’
The second clause in (18) exhibits strictly grammatical agreement with the
neuter quantifier ‘many’ (in the comparative degree), not the partitive genitive
dependent governed by it (as with the perfect in (15) above):
(18) Аще би съи фариж-ь бьі-л-ъ

Ašte bi sъi fariž-ь by-l-ъ
if would[3sg] this horse(m)-nom.sg be-pptcp-m.nom.sg
вь Трои, помног-о моу би лѣпот-ьi
vь Troi, pomnog-o mu bi lěpot-y
in Troy more-nt.nom.sg him would[3sg] beauty(f)-gen.sg
при-бьi-л-о
pri-by-l-o
pref-be-pptcp-nt.nom.sg
‘If this horse were in Troy, it would be even more beautiful’
Finally, there were five periphrastic passives which could alternatively be

seen as copular statives with participial adjectives (see Table 1). Interestingly
enough, one of them shows a much clearer breakdown in agreement of the
type described in Chapter 7, Section 7.4.6. The past passive participle in (19) is
neuter singular, but the nearest controller (the word for ‘image’) is masculine.
Going back to questions posed in the previous chapter, loss of agreement in
4 We might be dealing with a mismatch between the declensional class and the grammatical
gender of Troy.
286 chapter 8
periphrastic constructions appears to be a wider phenomenon which does not

affect the have-perfect only (it being non-attested here anyway).
(19) и прѣложи на немъ свѣтлое ѡрѫжие и щитъ, на

i prěloži na nemъ světloe orǫžie i štitъ, na
and gave to him bright armour and shield on
коем бѣ писа-н-о образ-ъ
koem bě pisa-n-o obraz-ъ
which be[past.3sg] write-pptcp-nt.nom.sg image(m)-nom.sg
пространного свѣт-а, сл(ъ)нц-е и
prostrannogo svět-a, slъnc-e i
wide world(m)-gen.sg sun(nt)-nom.sg and
м(ѣ)с(ѧ)ц-ъ, и звѣзд-ьі, и боур-ѣ
měsęc-ъ, i zvězd-y, i bur-ě
moon(m)-nom.sg and star(f)-nom.pl and storm(f)-nom.sg
‘and gave him shining armour and a shield on which was painted an
image of the wide world, a sun and a moon, and stars, and a tempest …’
In conclusion, this Middle Bulgarian text has turned up some breakdowns in

agreement with periphrastic verbal constructions, which however are in a clear
minority. No have-perfects were found, nor were any structures that could be
perceived as their immediate precursors, so be is the only auxiliary in this role
here. Some of the instances of the be-perfect are contextually used for non-
witnessed events, though the auxiliary is never omitted and the much more
widespread aorist can report non-witnessed events as well. The be-perfect and
the aorist are often interchangeable, so the question is when the perfect be-
came a fully-fledged perfect, which in turn would allow evidentiality to evolve
(cf. a similar interchangeability in Old and Middle English). In fact, this reshuf-
fling of the system involves two distinct processes: a) the move of the perfect
away from the aorist, which never happened in the other Slavonic languages
(where the original perfect became/remained just a past tense and encroached
on the aorist), and b) the rise of systematic evidentiality (G. Ganeva, p.c., cf.
Drinka 2017: Chapters 12 and 13). It was this move differentiating the two past
tenses that enabled non-witnessed/evidential perfect-like forms/uses to come
into their own. There have even been suggestions that the emergence of the
perfect as an evidential marker in turn allowed the retention of the other pret-
erital forms (aorist and imperfect) in Bulgarian, as opposed to the rest of the
Slavonic languages, where this did not happen (see Vasilev 1972: 347; cf. the
functional scenarios from the preceding chapter about the interaction of func-
tional load, evidentials and the emergence of have-perfects).
8.3 Damascenus Troianensis (17th c.)

A number of compilations of edifying texts were produced or copied in
Bulgarian between the sixteenth and the nineteenth centuries. Their language
and content is by no means uniform, but they are generally referred to as dam-
ascenes (Bulgarian дамаскини/damaskini), since they usually contain transla-
tions of the Greek sermons of Damascene the Studite (see MacRobert 1981:
296–297, with references, D’omina et al. 2012). Those sermons were originally
published in 1557–1558, and had seen several translations into Church Slavonic
by the seventeenth century. One of the Bulgarian Church Slavonic texts was
in turn translated into ‘varieties of Bulgarian not far removed from the mod-
ern language’, including a version with typically Central Bulgarian dialectal
features (MacRobert 1981: 296). From this version were derived several redac-
tions, one of which is represented by the Trojanski damascene/Damascenus
Troianensis.
Damascene literature can be used as a ‘source of evidence for the state of the
Bulgarian language in the early seventeenth century’, though the texts could be
subject to influence from Church Slavonic and (perhaps less directly) Greek
(MacRobert 1981: 297, cf. Georgieva et al. 1989: 20, 22, 25–29, 36–39, D’omina et
al. 2012: x ff.). They might even exhibit some potential influence from Serbian
Church Slavonic. Nevertheless, MacRobert (1981: 297) concludes that
it is sufficiently apparent that in most aspects of syntax and morphology

these versions departed from Church Slavonic norms and adhered to us-
ages closely resembling those familiar in the Modern Bulgarian spoken
language.
Here, I have chosen the Trojanski damascene from the 17th century, which is
made available electronically at http://histdict.uni-sofia.bg/textcorpus/show/
doc_11 (accessed on 21 March 2017). It is designated in Latin as Damascenus
Troianensis, and appears under MS number № ІІ, 11 or № 88 at the Bulgarian
Academy of Sciences. Ivanova’s (1967) is a print edition with a glossary and
notes.

Only the first two parts of the transcript were examined, as published online:
http://histdict.uni-sofia.bg/textcorpus/show/doc_11 (accessed on 22 March
2017). The page numbers after each example are as they appear on the website –
from 1 to 27. The word count is roughly 8,372 words, though this figure might
288 chapter 8
table 2 Perfects, conditionals, passives, aorists and imperfects in the 17th-century

Damascenus Troianensis

Be-perfects 48 0 48 (44 pres. perf. + 4 10.37%

pluperfects)
Be-conditionals 5 0 5 1.08%
Aorists 314 0 314 67.82%
Imperfects 81 0 81 17.49%
in fact be smaller due to issues of symbol compatibility when the text from
the website was pasted into MS Word for word processing. The first part, a
sermon on the Ten Commandments, is mostly not about narrating past events,
so this might explain the lower incidence of the perfect there. The following
story about the life and miracles of Saint Simeon does narrate past events and
it also has direct speech, where many of the present perfects are to be found.
This once again reaffirms the relationship between (present) perfects and
direct speech already noted in the preceding section and in earlier chapters.
There are only four pluperfects in the entire sample as opposed to 44 pres-
ent perfects, which are common in dialogue because dialogue naturally views
(recent) past events through the prism of the moment of speaking, so current
relevance is more prominent and brought to the fore by perfect forms. There
is much less need for these perfect highlighting devices in narratives of purely
past events characteristic of chronicle entries which are interested in (more
distant) history (see Chapter 4). In the light of this, a pertinent question to ask
would be: are there more perfects in Damascenus Troianensis than in the Tale of
Troy because the category is more grammaticalised in 17th-century Bulgarian
or because there happens to be more dialogue in this sample?
Another issue has to do with the potential influence of Turkish on the
Bulgarian verbal system, since this is a post-contact text. Although lexical
borrowings do not guarantee structural ones, contact with Turkish is abun-
dantly evident from Turkish loanwords such as арти́сува/artisuva ‘be left
over’, файдъ́/fajdъ ‘benefit, advantage, use’ (ultimately Arabic, via Turkish),
душма́нин/dušmanin ‘enemy, tormentor’ (and derivatives), among others
(cf. (31) below). There is also the typical Balkan Sprachbund use of the da-
tive for the genitive: и тогiа стро́шишь глава́ дiа́вол-у/i togia strošišь glava
djavol-u ‘and then you’ll break the devil’s head’ [11] (see (31f) below); женѫ́
бли́жн-ому и съсѣ́ д-а твое́го/ženǫ bližn-omu i sъsĕd-a tvoego ‘[thou shalt not
covet] the wife of thy friend or of thy neighbour’ [12], with a dative (in -omu)
and genitive (in -a) side by side. In the following (almost minimal) pair, the
same dative -u ending serves to express the indirect object and the possessor:
ре́коха iгумен-у/rekoxa igumen-u ‘they said to the abbot’ [20] [dative indi-
rect object]; посте́лiaта С[ве]т-му Сѷмеѡ́н-у/posteljata Svet-mu Simeon-u
‘Saint Simeon’s bedding’ [18] [dative possessor].
Despite the obvious contact with Turkish and structural developments
typical of the Balkan Sprachbund,5 this stretch of text does not yield eviden-
tials of the type one might have expected to find, i.e. with missing auxiliaries
and supposedly due to Turkish influence. The perfect sometimes denotes
non-witnessed events, as confirmed by the context, and the overall situation
appears more or less identical to that in the pre-conquest Tale of Troy. The dis-
tribution of the different patterns is similar, as is the way they are used.
And yet, some crucial shifts deserve special mention. Although the synthet-
ic preterital forms (aorist and imperfect) remain dominant, the percentages of
passive and perfect periphrases with be have increased more than twofold: be-
passives accounted for 0.62% of all recorded verbal constructions in the Tale
of Troy, whereas they now have a 3.24% share. Perfects have gone from 4.72%
in the Middle Bulgarian Troy sample to 10.37% in this early modern text. It will
emerge below that the perfect periphrasis has started to be consolidated not
only in numerical terms but also in terms of typical usage, though there are
still many similarities to the mediaeval period and it has not yet attained its
present-day status. It would come as no surprise that this Early Modern text is
somewhat of a bridge half way between the Middle Bulgarian passage exam-
ined above and the modern day.
As noted in the previous section and earlier in the book, a characteristic
usage is the so-called experiential perfect, focusing on the expression of one’s
own past experiences which may affect the present, as in (20a). Similarly, it
is clear in (20b) that the speakers report their own actions and experiences,
which additionally have current relevance.
5 For the Balkan Sprachbund, see Trask and Millar (2015: 303–304, with references).
290 chapter 8
(20) a. а а́зь си́чко тако́ва зло́ съм

a azь sičko takova zlo sъm
conj I.1sg.nom all such evil be.pres.1sg
стори́-л-ь. [19–20]
stori-l-ь
do-pptcp-m.sg
‘and I have done all that evil’
b. и не ѻста́ ни́где мѣ́ сто где́ нѣ сме́

i ne osta nigde mĕsto gde nĕ sme
and neg remained nowhere place where neg be.pres.1pl
ходи́-л-е. ть́кмо оу въ́бель Сухьi нѣ сме́
xodi-l-e tьkmo u vъbelь Suxy nĕ sme
go-pptcp-pl just in well dry neg be.pres.1pl
ходи́-л-е [20]
xodi-l-e
go-pptcp-pl
‘and no place is left where we haven’t been; we only haven’t been to
the dry well’
For many perfects, though, it was not clear whether they were meant more as
highlighters of resultativity or whether they were intended primarily to de-
scribe non-witnessed events:
(21) Защо́ тi е да́-л-ь богь же́нъ́

Zašto ti e da-l-ь bogь ženъ
because thee.2sg.dat be.pres.3sg give-pptcp-m.sg god wife
да ра́ждашь дѣца́ [14]
da raždašь dĕca
comp bear children
‘Because God has given you a wife to have children with’
The subordinating conjunction of reason in (21) might point to a meaning of

resultativity. Another typical diagnostic of the resultative perfect is a clause of
result, as in (22):
(22) ами сє е въ-вонiа-л-ь, не мо́же

ami se e vъ-vonja-l-ь, ne može
conj refl be.pres.3sg pref-stink-pptcp-m.sg neg can
человeкь да пристъ́пи бли́зу при не́го ѿ
čelovekь da pristъpi blizu pri nego ot
one comp step near to him.3sg.m.acc from
смра́дь [18]
smradь
stench
‘and he has started to stink, so no one can get near him because of the
smell’
The writer could hardly have witnessed the creation of the world or what went
on between God and Joseph, so the material in (23) certainly describes non-
witnessed events, though arguably it might still be intended to foreground the
resultativeness of what is reported:
(23) a. єдн-ому бог-у, дето́ е

edn-omu bog-u, deto e
one-m.dat.sg god(m)-dat.sg that be.pres.3sg
стори́-л-ь небо и землю [2]
stori-l-ь nebo i zemlju
create-pptcp-m.sg heaven and earth
‘… to the one God, who created the heaven and the earth…’
b. [господи] що си́ наоу́чи-л-ь Iѡ�сифа [23]

[gospodi] što si nauči-l-ь Josifa
Lord who be.pres.2sg teach-pptcp-m.sg Joseph
‘[Oh Lord] thou that hast taught Joseph …’
For (24), it is not clear that the events are not witnessed, since the spirit that
speaks these words is probably omniscient (cf. a similar example from the Tale
of Troy). In any case, the 3sg auxiliary would not be omissible today from the
final clause in (24), which might suggest that this is a bona fide perfect, not a
proto-evidential.
292 chapter 8
(24) Защо го си мъчи́-л-ь, и

Zašto go si mъči-l-ь, i
why him.3sg.m.acc be.pres.2sg torment-pptcp-m.sg and
защо́ го си испъ́ди-л-ь, Кое́
zašto go si ispъdi-l-ь, Koe
why him.3sg.m.acc be.pres.2sg expel-pptcp-m.sg what
тi е зло́ стори́-л-ь [20]
ti e zlo stori-l-ь
thee.2sg.dat be.pres.3sg evil do-pptcp-m.sg
‘Why have you tormented him and why have you driven him away; has he
done you any wrong …’
Probably the most typical and convincing example of reporting, where evi-
dential forms with an ellipted auxiliary would be appropriate in Present-Day
Bulgarian, is provided in (25). The narrator most likely presents the event of
the main character telling the monks as witnessed by himself too, so he uses
the aorist of ‘say’. However, the narrator did not witness the character’s dream
for himself, hence the l-participle of ‘see’. The two preterital forms are now
functionally differentiated in this material, though the perfect auxiliary is re-
tained (cf. the discussion of (10) and (11) above, especially the analogous (11c),
again reporting a dream but with two aorists).
(25) и каза́ на кало́герете как е

i kaza na kalogerete kak e
& say[aor.3sg] to the.monks how be.pres.3sg
видѣ́ -л-ь съ́нь стра́шень [20]
vidĕ-l-ь sъnь strašenь
see-pptcp-m.sg dream frightening
‘and he told the monks that he had had a frightening dream’
At least a quarter, and probably more, of the 48 perfects here indeed de-
scribe non-witnessed events, as suggested by the context (see Appendix 11).6
However, as noted above, not all of them could translate into renarrated forms
without an auxiliary today. For instance, the context suggests that the Saint’s
actual death in (26a) was not witnessed by anyone, but an evidential construc-
tion with missing ‘be’ would hardly be resorted to for that subordinate clause
in 20th/21st-century Bulgarian. The same applies to the death of St. Simeon’s

mother in the analogous (26b). Even today, speakers would opt for an inferen-
tial perfect with an explicit form of ‘be’ identical with that in (26a)–(26b). Both
sentences in (26a) and (26b) would be typical non-witnessed perfects with
retention of the auxiliary in contemporary Bulgarian. Likewise supporting
the modernity of this text, (26c) offers a rare glimpse into the words of a first-
person narrator and participant in the events who explicitly reports what he
witnessed. The aorist-perfect dichotomy is indeed coming into its own – (26c)
has the expected witnessed aorists, as opposed to the non-witnessed perfects
in (26a)–(26b), the way it would be today. This might testify to the incipient
differentiation between the two categories.
(26) a. тога́ва позна́-ха си́чкьi лю́дiе че съ

togava pozna-xa sičky ljudje če sъ
then know-aor.3pl all people comp refl
е́ прѣста́ви-л-ь [27]
e prĕstavi-l-ь
be.pres.3sg present-pptcp-m.sg
‘and that’s when the multitude realised that he had passed away’
b. и то́ѧ час ́ разумѣ́ стьi Сѷмеѡ�н че

i toę čas razumĕ stьi Simeon če
& this hour find.out[aor.3sg] St. Simeon comp
ѥ� оумрѣ́ -л-а ма́йк-а му [22]
je umrĕ-l-a majk-a mu
be.pres.3sg die-pptcp-f.sg mother(f)-sg his
‘and at that hour St. Simeon found out that his mother had died’
c. И а́зь антѡ�нie видѣ-x ́ какь не

I azь antonje vidĕ-x kakь ne
and I.1sg.nom Anthony see-aor.1sg how neg
отво́ри, почю́ди-х се що му
otvori, počjudi-x se što mu
open[aor.3sg] wonder-aor.1sg refl what him
бьi [26]
by
be.aor.3sg
‘And I, Anthony, saw that he didn’t open [the door], and I wondered
what was the matter with him’
294 chapter 8
Despite the incipient differentiation between aorist and perfect in (25) and
(26), this post-conquest text also exhibits aorists in non-witnessed/evidential
function, just like the Middle Bulgarian Tale of Troy, written before contact
with Turkish. Such ‘evidential’ aorists were counted and recorded on the first
three pages only, just to illustrate this point.7 The word count for the first three
pages is ca. 727 words, and there are also imperfects throughout the text which
describe non-witnessed events, though they have not been recorded systemati-
cally. In the brief stretch of only 727 words, I found many more ‘non-witnessed’
aorists, apparently much more numerous than the perfects. Pages 1 to 3 con-
tain only one perfect, as opposed to at least eighteen aorists describing non-
witnessed events (though the reader should keep in mind the introductory
remarks about genre). This should additionally be placed in the wider context
of perfect vs. aorist usage in the entirety of this sample compared to the previ-
ous one. While the imperfect has remained fairly constant (at roughly 17% in
both passages), compared to the Tale of Troy, the perfect has carved out more
territory from the aorist here (an increase from 4.72% to 10.37% for the perfect,
accompanied by a drop from 74.91% to 67.82% for the aorist). Nevertheless, the
aorist remains the dominant preterital form (cf. Tables 1 and 2). Crucially, it can
still relate second-hand information.
The two constructions might in fact serve subtly different functions, ex-
pressing slightly different meanings, so a future investigation could shed
more light on the differentiation between them (cf. MacRobert 2013). Let us
consider (27):
(27) то́й напра́ви нeбо и зе́млю, то́й

toj napravi nebo i zemlju, toj
he.3sg.m.nom made heaven and earth he.3sg.m.nom
ре́че и бьiха а́ггльi [2–3]
reče i byxa aggly
said and were angels
‘he [God] created/made heaven and earth, he spake and there were
angels’
7 Participial passives with aorist be were ignored.

The reason for using aorists in (27) (cf. the perfect in (23) above) could be that
the writer wanted to make what he describes in (27) sound more ‘real’, as if he
had been there and seen it himself. This might make his message more vivid
and therefore more convincing (though some scholars ascribe greater vivid-
ness to the periphrases, e.g. Dobrev 1973; see preceding section). All things
considered, however, one can conclude that there is no systematic correlation
between form and meaning at this stage. Non-witnessed meaning can be ex-
pressed both by the perfect and by the aorist (and the imperfect too), so contra
Gerdžikov (1984/2003: 256, 260), it is perhaps too early to talk about a fully-
fledged grammatical category of evidentiality, even though evidential and
‘double-evidential’ (admirative/dubitative/incredulous) forms do exist else-
where in damascene literature, including in this very collection (as described
by D’omina; G. Ganeva, p.c.).8 In some respects, this 17th-century document
is little different from the 14th-century Tale of Troy or from the earliest OB/
OCS records. Thus, Jesus’s life can be summarised in the aorist, as in (28a) (cf.
Dobrev 1973: 13–14, 17–18, and MacRobert 2013: esp. 391, 397–399, for classical
OB/OCS analogues where aorists alternate with perfects (with or without an
auxiliary) in what appears to be more or less free variation). The aorist in (28b)
is also non-witnessed and might even indicate a certain amount of resulta-
tiveness. Perfects and aorists coexist in (28c) and arguably both constructions
indicate resultativeness. Unlike the parallel examples in (26), (28d) has an aor-
ist in a counterfactual context, where a perfect would be obligatory today (cf.
previous section too). All of this suggests that the perfect-aorist dichotomy has
not yet attained its present status, though the specialisation of the two might
be underway.
8 See Gerdžikov (1984/2003: 104 fn. 2, 256, 260–261, with references), D’omina et al. (2012: xii),
inter alia. Below is one such random example from the same collection but outside the scope
of my sample. Vindicating my thesis, the stand-alone evidential participle in -l coexists with
an aorist (случи/sluči) which reports the same event. Both forms would be expected to get -l
in Present-Day Bulgarian.
(i) И тоги́ва случи́ се ѿ индiѧ нѣ́ кой члкь купе́ць
I togiva sluči se ot indię nĕkoj člkь kupecь
& then happen[aor.3sg] refl from India some person merchant
до́шь-л-ь вь iерслимь
došь-l-ь vь jerslimь
come-pptcp-m.sg in Jerusalem
‘And then it happened that a merchant from India came to Jerusalem’ [Source: https://e
-medievalia.uni-sofia.bg/moodle/mod/page/view.php?id=2839 (accessed on 11 Sept 2017)]
296 chapter 8
(28) a. то́й съда́де … слѣ́ зе … и вьпль́ти се …

toj sъdade … slĕze … i vьplьti se …
he.3sg.m.nom created came.down and incarnated refl
и бьi члкь съврь́шень … и поживѣ́ на́ земе …
i by člkь sъvrьšenь … i poživĕ na zeme …
and was human perfect and lived on earth
на́поконь распе́ се и умрѣ́ ... [3]
napokonь raspe se i umrĕ
finally crucified refl and died
‘he [God] created [the human race] … and came down [to earth] …
and was incarnated … and was a perfect human being … and lived
for a while on earth … and was finally crucified and died’ [aorists re-
porting non-witnessed past events]
b. Tié са голѣ́ мите Заръча́нiа божiа. дето́ гьi

Tie sa golĕmite Zarъčania božia. deto gy
these are the.big Commandments Lord’s which them
преда́де на свѣтъ́ть богъ [14]
predade na svĕtъtь bogъ
give[aor.3sg] to the.world god
‘These are the Lord’s great Commandments which God gave to the
world’ [aorist additionally expressing resultativeness, since the
Commandments are now available and this is what the text is about?]
c. щото́ ѻ�к-о не ѥ� видѣ́ -л-о, и

štoto ok-o ne je vidĕ-l-o, i
what eye(nt)-sg neg be.pres.3sg see-pptcp-nt.sg and
оу́х-о, не ѥ� чю́-л-о, и на срдце
ux-o, ne je čju-l-o, i na srdce
ear(nt)-sg neg be.pres.3sg hear-pptcp-nt.sg and on heart
члку не ѥ� ѻтиш-л-о́, дето́ им
člku ne je otiš-l-o, deto im
of.man neg be.pres.3sg go-pptcp-nt.sg what them
оуго́тви богь щото́ нѥ�го ѡбьiча-т. [14]
ugotvi bogь štoto njego obyča-t.
prepare[aor.3sg] god that him love-pres.3pl
‘[But as it is written,] Eye hath not seen, nor ear heard, neither have
entered into the heart of man, the things which God hath prepared
for them that love him. [Corinthians 2:9, King James Bible] [both
perfect and aorist arguably expressing resultativeness; note that the
King James version has a perfect hath prepared]
d. и си́чкьi мнѣ́ -ха чѥ� оумрѣ́ [20]

i sičky mnĕ-xa čje umrĕ
and all think-imperf.3pl comp die[aor.3sg]
‘And everyone thought that he’d died’ [counterfactual non-witnessed
context, where a perfect would be obligatory today]
In addition to its other functions, the perfect in this sample also demonstrates
a certain amount of emotional colouring, indicating immediacy and greater
vigorousness, even impatience and annoyance, compared to the plainer/less
marked simple present which might have occurred in (29) instead:
(29) а́ко ли́ го не на́йде-те, тука вѥ�ке да

ako li go ne najde-te, tuka vjeke da
if particle him neg find-pres.2pl here no.longer comp
нѣ е дош-л-ь ни́кой. [20]
nĕ e doš-l-ь nikoj.
neg be.pres.3sg come-pptcp-m.sg nobody
‘If you don’t [manage to] find him, no one dare come here anymore’
None of the perfects, conditionals or participial passives in this part of

Damascenus Troianensis demonstrate any default agreement (see Table 2).
The closest it gets to a default scenario is in a passive with a clausal subject,
but the neuter singular participle in (30) is absolutely legitimate due to the
genderless and numberless subject, the subordinate clause introduced by the
complementiser да/da:
(30) и не бьi и́ да́де-н-о да

i ne by í dade-n-o da
and neg be[aor.3sg] her.3sg.f.dat give-pptcp-nt.sg comp
го ви́ди [22]
go vidi
him.3sg.m.acc see[3sg]
‘and she wasn’t (allowed) to see him’
As an aside, this might be a good point to note that so-called reflexive passives,
i.e. semantically passive constructions with an active verb and a reflexive clitic,
are treated as morphologically active (see (28a)). The number of passives fea-
tured in Table 2 and in Appendix 119 only includes combinations of be and a
298 chapter 8
past passive participle. Thus, it might be speculated that the perfects here (and
in the previous Troy sample) are much more numerous than the participial
passives due to the existence of ‘reflexive’ passives as an alternative option for
marking voice in Bulgarian. It is hard to tell if this is one of the reasons why
be-perfects survive, though so do the (underdog?) be-passives. This line of rea-
soning is reminiscent of the ruminations involving the two passive auxiliaries
of early Germanic (Chapter 3, Section 3.4.3, and Chapter 6).
Finally, I detected no traces of perfects with have in this sample, at least
no obvious ones. However, the verb for ‘have’ did appear delexicalised in
constructions with object nouns, such as those in (31), similar to earlier and
modern English combinations like have need/shame/a fight etc. (31a)–(31g)
demonstrate the wide range of nouns that can combine with ‘have’ in these
delexicalised constructions: душманлъ́кь/dušmanlъkь ‘enmity’, файдъ́/fajdъ
‘benefit, advantage, use’, кара́зь/karazь ‘fight, argument’, любо́вь/ljubovь ‘love’,
проще́нiе/proštenie ‘forgiveness’, as well as добринiа/dobrinja ‘good(ness)’.
(31f), (31h) and (31i) demonstrate that other generic verbs than ‘have’ are
available in such delexicalised constructions, including ‘do’ and ‘hold’. This
indicates that it is such generic verbs that, due to their non-specific mean-
ings, are likely to be delexicalised even further and to experience grammati-
calisation. The appearance in (31i) of the synthetic verbal form душма́нишь/
dušmanišь ‘(lit.) [thou] tormentest’, related to a noun in the delexicalised V +
N combinations, confirms that these patterns are equivalent to verbs in that
they convey dynamic meanings. The same applies to the verb and the noun
for ‘love’ in (31b) and (31e). Once a stative verb denoting possession, delexical
have has started to acquire a more dynamic interpretation arguably condu-
cive to onward grammaticalisation. The same equivalence between the verb
надзе́мувай/nadzemuvaj ‘take too much, overcharge’ and the V + N combina-
tion зе́ма артъ́кь/zema artъkь ‘take a surplus/take too much, overcharge’ is
in evidence in (31j), where we additionally see that these verb-like, dynamic V +
N constructions can undergo subsequent nominalisation into deverbal (com-
pound?) nouns: артъ́кь зема́нiе/artъkь zemanie ‘surplus-taking, taking too
much, overcharging’.
(31) a. да и́ма-шь душманлъ́кь на/ кара́зь съсь

da ima-šь dušmanlъkь na/ karazь sъsь
comp have-pres.2sg enmity on fight with
нѣ́ кого [10]
nĕkogo
someone
‘to hate someone, to feel hatred for/animosity towards someone’/
‘to have a fight/argument with someone’
b. чѥ� ако́ лю́би-те кои́то вьi лю́б-ът, ко́ѧ

čje ako ljubi-te koito vy ljub-ъt, koę
comp if love-pres.2pl who you love-pres.3pl what
файдъ́ и́ма-те; нъ́ когьi лю́би-те душма́нете
fajdъ ima-te; nъ kogy ljubi-te dušmanete
benefit have-pres.2pl but when love-pres.2pl enemies
ва́ши, тогиа файдъ́ мнѡ�го и́ма-те ѿ бога. [10]
vaši, togia fajdъ mnogo ima-te ot boga.
your then benefit a.lot have-pres.2pl from god
и има́ ще-ш файдъ́ [14]
i ima šte-š fajdъ
and have[inf] fut-2sg benefit
‘… that if you love those who love you, what benefit do you have/what
use is it to you; but if you love your enemies, then you will have a lot of
benefit/a great reward from God. […] and you will have some benefit/
reward’
c. Защо́ богь душманль́кь не́ма [10–11]

Zašto bogь dušmanlьkь nema.
because god enmity not.have[pres.3sg]
‘because God has/feels/harbours no animosity’
d. ако и́ма-шь кара́зь на нѣ́ кого хр҇тiа́нина [11]

ako ima-šь karazь na nĕkogo xrtianina
if have-pres.2sg fight on someone Christian
‘if you have a fight/a bone to pick with a Christian’
e. да и́ма-ме съ си́чкьiте любо́вь [15]

da ima-me sъ sičkyte ljubovь
comp have-pres.1pl with all love
‘Let’s be on friendly terms with everyone’/‘May there be love between
everyone’ [cf. the verb for ‘love’ from the same root in (31b)]
300 chapter 8
f. мака́рь и да не си́ прегрѣ́ ши-л-ь

makarь i da ne si pregrĕši-l-ь
even and comp neg be.pres.2sg sin-pptcp-m.sg
душма́нин-у тво-е́му тьi. нъ́ па́кь тьi
dušmanin-u tvo-emu ty. nъ pakь ty
enemy(m)-dat.sg thy-m.dat.sg thou but again thou
стори́ нѥ�му мета́нiе. и и́ма-шь проще́нiе. и
stori njemu metanie. i ima-šь proštenie. i
do him bow and have-pres.2sg forgiveness and
тогiа стро́ши-шь глава́ дiа́вол-у. [11]
togia stroši-šь glava djavol-u.
then break-pres.2sg head devil(m)-dat.sg
‘even though you may not have sinned against your enemy, still bow
to him and you will be forgiven, and then you’ll break the devil’s head’
g. нъ́ а́ко го́ тьi ѻста́виш тогiа му ще ѿврь́на

nъ ako go ty ostaviš togia mu šte otvrьna
but if him thou leave then him fut retaliate
богь. а те́бѣ и́ма голѣ́ ма добринiа,
bogь. a tebĕ ima golĕma dobrinja,
god and thee have[pres.3sg] great goodness
чѥ си́ прости́-л-ь душма́нина свое́го [11]
čje si prosti-l-ь dušmanina svoego
comp be.pres.2sg forive-pptcp-m.sg enemy own
‘But if you let him be, then God will retaliate against him, and it’ll do
you a lot of good that you have forgiven your enemy’ [cf. the verb for
‘forgive’ to earlier ‘forgiveness’]
h. И ко́йто не дрьжи́ душманлъ́к ни́ на кого́, тiе

I kojto ne drьži dušmanlъk ni na kogo, tie
and whoever neg holds enmity neg to one those
сѫ синѡ�ве божiй [11]
sǫ sinove božij
are sons god’s
‘And those who do not hold grudges against anyone, they are the sons
of God’ [cf. also English collocations with more or less delexical verbs
like hold/bear/have a grudge]
i. А́ко ли́ тьi душма́ни-шь нѥ�му и

Ako li ty dušmani-šь njemu i
if particle thou torment-pres.2sg him and
стори́-л му си мета́нiе. [11]
stori-l mu si metanie.
do-pptcp[m.sg] him be.pres.2sg bow
‘If you torment him and have bowed to him’
j. ѿ завидѣ́ нiе бьiва артъ́кь зема́нiе. а́ артъ́кь

ot zavidĕnie byva artъkь zemanie. a artъkь
from envy is surplus taking and surplus
зе́манiе що е́; че сѣкьi ко́й зе́ма артъ́кь, та
zemanie što e; če sĕky koj zema artъkь, ta
taking what is that everyone who takes surplus that
е́ и́долослужи́тель. Зарад ́ това́ не завиди́, и не
e idoloslužitelь. Zarad tova ne zavidi, i ne
is idolater because this neg envy and neg
над-зе́мувай. [14]
nad-zemuvaj
over-charge
‘Envy leads to overcharging, and what is overcharging; everyone who
overcharges is an idolater; that’s why you shouldn’t envy and shouldn’t
overcharge’ [semantic equivalence between a single verbal lexeme
for ‘overcharge’, the V + N combination ‘take + surplus’, and the nomi-
nalisation ‘surplus taking’; cf. also the nominal (zavidĕnie) and verbal
(zavidi) forms of ‘envy’, paralleling the nominal and verbal use of ‘take’]
Analogous collocations are also attested as early as OB/OCS, e.g. вражьдѫ

имѣти/vražьdǫ imĕti ‘to have a fight/enmity’, where, just as in (31), the meaning
of have is already reasonably bleached and this can be a step in its grammati-
calisation as an auxiliary. Trousdale (2008) discusses comparable constructions
in English in the light of both grammaticalisation and lexicalisation processes.
All of this ties in well with the point made earlier in the book that generic verbs
like do, hold and have are likely candidates to be even further generalised and
grammaticalised due to their abstract, non-specific meanings. The delexical
constructions in (31) might be an intermediate stage in this process.
As I concluded for English, there is hardly any evidence that this step towards
greater grammaticalisation of have, if it is perceived as such, was prompted by
the functional overload of the traditional Bulgarian be-perfect – rather, it is
most likely just a natural consequence of the general meaning of the verb for
302 chapter 8
‘have’ which was prone to get even less concrete and more dynamic over time.
By the time of this representative of 17th-century Bulgarian, I have not detect-
ed any significant shifts in the distribution or functional load of the be-perfect
as compared to the 14th century, including evidential uses (despite the incipi-
ent functional differentiation of the various preterital forms). Stand-alone re-
narrative l-participles exist outside of my sample, but they are still in variation
with aorists. In the next section, I look for realignments within the system in
the 18th century.
8.4 A Slavonic-Bulgarian History (1762)

The Slavonic-Bulgarian History (История славянобългарска/Istorija
slavjanobălgarska) was completed by monk Paisius of Hilendar on Mount
Athos in 1762. This gem of the National Revival period is one of the first works
of Bulgarian historiography and, at least in the view of some scholars, the
first work of modern literary Bulgarian. Its language is mostly 18th-century
Bulgarian but it is heavily influenced by Church Slavonic, probably to a greater
extent than damascene literature (see Georgieva et al. 1989: 20, 24, 34, 37–40,
43–45, 52ff., with references, Peev et al. 2015: 39). Except for the original passag-
es and autobiographical notes, Paisius’s sources for the rest of the book include
Russian translations of European historians as well as Middle Bulgarian docu-
ments, in different redactions of Church Slavonic (see Georgieva et al. 1989:
57–58, Peev et al. 2015: 80, 81 fn. 31, fn. 33).
An overview of Paisius’s language can be found in Georgieva et al. (1989:
58–68), with special reference to the vernacular basis of his output, but also
acknowledging the mixed Church Slavonic and modern character of the book
(cf. Gešev 2009). Touching upon a topic of interest for the present monograph,
Georgieva et al. (1989: 61) note that the narration in Istorija slavjanobălgarska
is done predominantly by means of the so-called evidential/non-witnessed
forms in -l without an auxiliary. Citing Mirčev and Andrejčin, Georgieva et
al. (1989: 61–62) endorse the view that this construction was established in
Bulgarian around the 16th century (though see Chapter 7).
Paisius’s original text was copied multiple times, sometimes with the ad-
dition of features more typical of the spoken and dialectal vernacular, rather
than Church Slavonic (see Radeva 2013, Peev et al. 2015: 39ff.). Paisius himself
most likely came from the town of Bansko, so he must have spoken a south-
western/Macedonian variety of Bulgarian and it would be interesting to see
if that is reflected in the usage of have-perfects in his writing. The edition
consulted here is that of Peev et al. (2015), which is based primarily on what is
believed to be the original draft, but the editors supply readings and emenda-
tions from the later copies too. In the preparation of this chapter, parallel use
was made of the electronic publication available on the Cyrillomethodiana
website (http://histdict.uni-sofia.bg/textcorpus/show/doc_137, accessed on 04
April 2017). Each example is identified with folio and line numbers.

Only the preface to the History was examined – up to folio 5v, where a self-
contained section of the book ends. The preface is least likely to have been
influenced by earlier sources. Since it is addressed to contemporary Bulgarian
readers and (illiterate) hearers, this portion of the text is most likely to ap-
proximate vernacular practices. The total number of words in this excerpt is
modest, about 1,999 (based on http://histdict.uni-sofia.bg/textcorpus/show/
doc_137, accessed on 11 April 2017). This is less than a quarter of the size of
the sample from Damascenus Troianensis and less than a third of the Tale of
Troy, but there is nevertheless plenty of material to go by. In some respects
the results do not seem that far removed from the earlier ones, as borne out
by the figures in Table 3 relating to analytic forms (perfects, conditionals and
passives). Similarly to what was ascertained for English, one could conclude
that there is a fair amount of stability as compared to the previous samples
table 3 Perfects, evidentials, conditionals, passives, aorists and imperfects in the preface
to Paisius’s Slavonic-Bulgarian History (1762)

Be-perfects (with auxiliary)a 15 0 15 16.48%

Evidentials (no auxiliary) 25 2 27 29.67%
Be-conditionals 3 0 3 3.30%
Aorists 36 0 36 39.56%
Imperfects 4 0 4 4.40%
a Of the 15 past participles counted as instances of the perfect, only 8 are combined with an
overt occurrence of the auxiliary be. As long as there was an overt auxiliary that could com-
bine with multiple conjoined participles, all were taken to implicitly go with that auxiliary.
This was done in order to achieve a principled way of categorising the forms, in line with the
introductory remarks at the opening of this chapter (see Section 8.2.2 and the discussion
below).
304 chapter 8
in terms of the distributions of the various periphrastic constructions relative

to one another, their main functions and the presence or absence of agree-
ment. Notably, periphrases have gradually grown in frequency: 4.72% > 10.37%
> 16.48% for the perfect; 2.61% > 1.08% > 3.30% for conditionals; and 0.62% >
3.24% > 6.59% for the passive (cf. Tables 1–3). What stands out most, however,
is the abundance of evidentials, essentially perfect constructions without an
auxiliary used to express non-witnessed events. They have now branched off
from perfects proper and have won a great deal of ground from aorists and im-
perfects when it comes to narrating past events. The balance in the preterital
domain has been tipped in favour of the perfect(-like) forms. The radically dif-
ferent proportions of perfects plus evidentials on the one hand, vs. aorists and
imperfects on the other, suggest a major shift in the system which cannot be at-
tributed to genre alone. This must be a genuine reflection of language change,
though of course there are remnants of earlier patterns and types of use.
As before, perfects with the auxiliary be can definitely denote non-witnessed
events, as they do on a regular basis in this 18th-century piece, perhaps also
contributing varying degrees of resultative semantics:
(32) и̑ли̇ не̑ са и̇ма-л-и̇ болгари̇ црс͠тво и̑

ili ne sa ima-l-i bolgari crstvo i
or neg be.pres.3pl have-pptcp-pl Bulgarians kingdom and
гсподсво [sic] за толiко лета црс͠вува-л-и̇ и̑
gspodsvo za toliko leta crsvuva-l-i i
realm for so.many years reign-pptcp-pl and
би̇-л-и̇ славни̇ и̇ чуе-н-i по сва земла и̇
bi-l-i slavni i čue-n-i po sva zemla i
be-pptcp-pl famous and hear-pptcp-pl on all earth and
много пути̇ ѿ сiлнi ри̇млане и̇ ѿ мудрi
mnogo puti ot silni rimlane i ot mudri
many times from powerful Romans and from wise
грци̇ данъ възи̇ма-л-и̇ и̇ [f. 3v, l. 16–21]10
grci danъ vъzima-l-i i
Greeks tribute take-pptcp-pl and
‘Or didn’t the Bulgarians have a kingdom and a state, reigning for so many
years, known and famous all over the world, and on multiple occasions
the powerful Romans and the wise Greeks paid them tribute, and …’
10 It should be noted that the line numbers given in square brackets after the examples refer
to the numbers in the original editions, which do not correspond to the lines as they ap-
pear here.
Importantly, all past active participles in (32) above are treated as separate
tokens of the perfect (as opposed to the evidential), because there is an aux-
iliary which arguably applies to all of them. Although the events are not wit-
nessed, the author may have intended to additionally foreground and reinforce
the resultativeness arising from the situation described.
The decision to treat such participles as perfects with an auxiliary recover-
able from elsewhere is debatable. Strictly speaking, of the 15 past participles
counted as instances of the perfect, only 8 are combined with an overt oc-
currence of the auxiliary be, as indicated underneath Table 3, so the number
of actual perfects might have to be revised downwards. This applies in equal
measure to example (33).
(33) ѡ̑ни̇ правѣднi праѡ̑ц͠и̇ свi са би̇-л-и̇

oni pravĕdni praoci svi sa bi-l-i
those righteous forefathers all be.pres.3pl be-pptcp-pl
землоделатели̇ i ѡ̇вчаре и̇ би-л-и̇ богати̇ съ
zemlodelateli i ovčare i bi-l-i bogati sъ
farmers and shepherds and be-pptcp-pl rich with
съкоти̇ и̇ с плоди̇ землнi и̇ би-л-и̇ прости̇ на
sъkoti i s plodi zemlni i bi-l-i prosti na
animals and with fruits earthly and be-pptcp-pl simple on
землi и̇ незлоби̇ви̇ [f. 4v, l. 16–19]
zemli i nezlobivi
earth and meek
‘the righteous forefathers were all farmers and shepherds and were rich
in livestock and the fruits of the earth and they were simple [men] on
earth and meek’
Other manuscripts omit the auxiliary in (33) (see Peev et al. 2015: 76, as well
as Appendix 12,11 for further similar editorial and scribal interventions). It
is therefore questionable whether all analogous participles should be taken
to covertly combine with a single instance of auxiliary be. These scribal in-
terventions also bear testimony to the variability which existed at the time
306 chapter 8
(and probably still does), as well as to the interchangeability of the two

types of construction (or rather, the two instantiations of a single construc-
tion). This apparent randomness and lack of differentiation of neighbouring
forms with and without be also surfaces in the next example, (34), where only
the highlighted expression is treated as perfect; the other two l-participles
were deemed tokens of the evidential since they do not follow the auxiliary be.
(34) за толiко лѣта црствува-л-и̇ и̇ господсвува-л-i на

za toliko lĕta crstvuva-l-i i gospodsvuva-l-i na
for so.many years reign-pptcp-pl and rule-pptcp-pl on
землi и̇ма-л-и̇ са̇ iстори̇и̇ цар͠ски̇ и̇ конди̇ки̇
zemli ima-l-i sa istorii carski i kondiki
earth have-pptcp-pl be.pres.3pl histories royal and annals
архiереи̇ски̇ све iзвесно i жи̇ти̇а и̇ прави̇ла за
arxierejski sve izvesno i žitia i pravila za
clerical all known(?) and lives and canons for
с͠тъiхъ болгарск-и̇. мног-и̇ [f. 5r, l. 11–14]
stъixъ bolgarsk-i mnog-i
saints Bulgarian-pl many-pl
‘they reigned and ruled on earth for so many years and had royal
histories and clerical annals on everything in detail, as well as lives and
canons for many Bulgarian saints’
In addition, (34) could be seen as containing имам/imam ‘have’ in a clause

with objects and something approximating object complements, though one
of the candidates looks more like an adverbial, translated as ‘on everything in
detail’ following the interpretation given in Peev et al. (2015: 79). As regards the
second candidate, the original draft has plural многи̇/mnogi, but other MSS opt
for invariable много/mnogo (see Peev et al. 2015: 78), and it is not clear which
word многи̇/mnogi modifies – the saints or the lives and canons. In standard
Church Slavonic, the adjectival morpheme -i should indicate masculine nomi-
native plural so this might point to the saints, rather than the neuter plural
lives and canons (though the case is inappropriate anyway). Yet, in (49) below,
многи/mnogi is used with the feminine plural nouns for ‘books’ and ‘histories’.
Another non-witnessed be-perfect is given in (35). In the space of only a few

lines above and below it, there is the same variation between ‘perfect’ (with an
auxiliary) and ‘evidential’ (without) (see Appendix 12);12 once again, other MSS
omit the auxiliary here (see Peev et al. 2015: 80).
(35) едва се наiде нiхни̇ и̇мена и кои̇ по коего

edva se najde nixni imena i koj po koego
hardly refl find their names and who after whom
е црствува-л-ъ [f. 5v, l. 10]
e crstvuva-l-ъ
be.pres.3sg reign-pptcp-m.sg
‘one could hardly find their names and who reigned after whom’
It is worth noting two more relevant points regarding (35). Assuming that
наiде/najde ‘finds/found’ is an aorist (rather than present tense, with which
it is syncretic), there is a contrast between the event expressed by the aorist
(experienced first-hand) and the second-hand information reported with the
perfect. Secondly, singular наiде/najde (also part of a so-called reflexive pas-
sive construction) fails to agree with its plural subject, probably because of the
impersonal flavour associated with such ‘passives’. As pointed out in earlier
sections, shedding agreement is a wide-ranging phenomenon which applies to
the whole spectrum of verbal constructions (see further below).
Virtually absent from the earlier samples, evidentials are now abundant.
Compared to other verb forms that could serve the same role, there were many
more evidentials with no auxiliary employed to report second-hand informa-
tion here – at least 27, even more if the unclear cases above are included for
which it is not certain whether to relate a single auxiliary to multiple parti-
ciples. In either case, this type of construction now seems fully established
and is indeed the predominant way in which Paisius narrates historical events
(see Section 8.4.1). As with the perfect proper, there is often a clear-cut contrast
between the non-witnessed evidentials (l-forms highlighted in (36a)) and the
witnessed aorist (напи̑сахъ/napisaxъ ‘I wrote’ at the end of (36a) and another
two aorists in (36b)).
308 chapter 8
(36) a. и̇ ѿ всѣго славѣнскаго народа наи̇ славнi

i ot vsĕgo slavĕnskago naroda naj slavni
and from all Slavonic people most glorious
би̇-л-и̇ болгари̇ прво се ѡ̇нi царовѣ нарек-л-i
bi-l-i bolgari prvo se oni carovĕ narek-l-i
be-pptcp-pl Bulgarians first refl they kings call-pptcp-pl
прво ѡ̇нi патрiарха iме-л-и̇ […] како за то по
prvo oni patriarxa ime-l-i kako za to po
first they patriarch have-pptcp-pl as about that in
реду. све. въ си̇ю iстори̇ю напи̑са-хъ [f. 4r, l. 1–7]
redu sve vъ siju istoriju napisa-xъ
turn all in this history write-aor.1sg
‘of all the Slavonic peoples the Bulgarians were most glorious; they
were the first ones to proclaim kings, they were the first ones to have a
patriarch, they were Christianised first and they conquered the great-
est amount of land; thus, of all the Slavonic nations, they were most
powerful and most venerated, and the first Slavonic saint came shining
from the Bulgarian race and tongue, as I wrote about all of that in turn
in this book’13
b. азъ и̇зли̇ха поревнова-хъ по рода и̇

azъ izlixa porevnova-xъ po roda i
I a.lot be.passionate-aor.1sg about nation and
по ѡ̇течесво болгарское и̇ много трудъ
po otečesvo bolgarskoe i mnogo trudъ
about homeland Bulgarian and a.lot work
съвори-хъ [sic] [f. 3v, l. 1–2]
sъvori-xъ
do-aor.1sg
‘I was inordinately passionate about the Bulgarian land and nation
and did a lot of work’
Numerous 1sg aorists describe Paisius’s personal experiences of compiling

his work. However, things are by no means straightforward and always neat-
ly differentiated. In (37), the l-participle and the aorists оувѣщава/uvĕštava
and рече/reče coexist side by side, all arguably recounting second-hand
13
The unabridged quotation on which this translation is based can be found in
Appendix 12.
information.14 Moreover, Peev et al.’s (2015: 67) Modern Bulgarian translation

has е показал/e pokazal ‘has shown/demonstrated’, with an auxiliary in place
of Paisius’s stand-alone participle и̑зѧви́лъ/izęvilъ (here and on many other
occasions elsewhere – see Appendix 1215 and Peev et al. 2015: 81).
(37) въ кра́тце то и̑зѧви́-л-ъ василiѧ ке́саръ

vъ kratce to izęvi-l-ъ vasilię kesarъ
in short that show-pptcp-m.sg Basil Caesar
въсточнiи си́ну своему лъву премудрому оувѣщава-е-го̀
vъstočnij sinu svoemu lъvu premudromu uvĕštava-e-go
eastern to.son own Leo Wise counsel him
глаголетъ iсто́рiю рече [f. 2r, l. 8–10]
glagoletъ istoriju reče
says history said
‘that is what, in short, the eastern Emperor Basil told his son Leo the
Wise; counselling him, he says “History, he spoke, …”’
The -лъ/-lъ of прославiлъ/proslavilъ in (38) was inserted later above the line
in the draft. Without it, the remainder is an ordinary aorist, so this might con-
firm the interchangeability and coexistence of evidentials and aorists with the
same functions in the minds of speakers (unless it is merely a scribal oversight).
(38) наi прво нiхъ вз-луби̇-л-ъ на земли̇ и̇

naj prvo nixъ vz-lubi-l-ъ na zemli i
most first them pref-love-pptcp-m.sg on earth and
прославi-л-ъ [f. 5r, l. 1]
proslavi-l-ъ
glorify-pptcp-m.sg
‘[God] first came to love and glorify them [the ploughmen and shep-
herds] on earth’
14 It is not absolutely certain that оувѣщава/uvĕštava is an aorist, and some versions sup-
ply an unambiguous present-tense form here (see Peev et al. 2015: 66). The hyphenation
in that unclear word in (37) is from the original. The interchangeability of aorists and
evidentials in damascenes and in the National Revival period is noted by Gerdžikov
(1984/2003: 182 fn. 4), possibly attributable to Church Slavonic influence; like OB/OCS,
Church Slavonic does not possess evidentiality as a grammatical category.
310 chapter 8
And yet, evidentials predominate. Another example, in (39), illustrates l-

participles in a typically reported context, as well as a possibly delexical use of
have, to which I shall return below:
(39) самъ туи̇ маврубиръ написа-л-ъ тако кажтъ [sic]

samъ tuj mavrubirъ napisa-l-ъ tako kažtъ
alone this Mavro Orbini write-pptcp-m.sg that say
грци̇ ѿ завистъ и̇ ненави̇стъ що и̇меѧ-л-и̇
grci ot zavistъ i nenavistъ što imeę-l-i
Greeks from envy and hatred that have-pptcp-pl
на богари̇ [sic] [f. 5v, l. 10–12]
na bogari
on Bulgarians
‘Mavro Orbini wrote himself that the Greeks say so because of the envy
and hatred they had/harboured against the Bulgarians’
Finally, two of the stand-alone l-participles default to neuter singular in spite

of having plural subjects – (40). As documented in Peev et al. (2015: 78), other
MSS emend one of the forms to plural ѡбретали/obretali. Showcased here are
one existential (било/bilo) and one passive reflexive structure (се ѡбретало/
se obretalo), both with an ‘impersonal flavour’.
(40) но не би́-л-о въ ѡ̇но време щамб-и̇

no ne bi-l-o vъ ono vreme štamb-i
but neg be-pptcp-nt.sg in that time press(f?)-pl
славѣнск-и̇ а чловѣци̇ ѿ небреженiе не
slavĕnsk-i a človĕci ot nebreženie ne
Slavonic-pl and people from carelessness neg
препiсува-л-и̇ н̇а мало се места таквiа кнiг-и̇
prepisuva-l-i na malo se mesta takvia knig-i
copy-pptcp-pl on few refl places such book(f)-pl
ѡбрета-л-о [f. 5r, l. 15–17]
obreta-l-o
find-pptcp-nt.sg
‘but there weren’t any Slavonic printing presses at that time, and people
were careless and didn’t copy [them], so such books were to be found at
very few places’
As already mentioned above, despite the more numerous evidentials, both

aorists and imperfects in ‘evidential’ functions are abundantly attested in my
material too (see Appendix 12),16 pointing to an incomplete grammaticali-
sation and regulation of this category. There are three imperfects in folio 2v
which describe God’s treatment of the Jews, in addition to other such imper-
fects elsewhere, including беше/beše in folio 4v. Furthermore, no fewer than
14 aorists for second-hand experiences were recorded, which is a sizeable num-
ber. With regard to (41), one could claim that the aorist can still encroach on
the territory of the non-witnessed and resultative perfect (judging by its mean-
ing in context):
(41) на до́лгое времѧ̀ пiса́ниѧ ѡ̑ сих ѡ̑ста́ви̇-ша [f. 1r, l. 16–17]

na dolgoe vremę pisanię o six ostavi-ša
on long time writings about these leave-aor.3pl
‘they left long-lasting writings on those [things]’
One reason for using ‘second-hand’ aorists suggested in Section 8.3.2 was to
lend vividness to the retelling of Biblical events. As in the earlier periods, in
(42a)–(42b) the aorist and the imperfect describe non-witnessed events in a
Biblical context, so one wonders if the Bible might have acted as a conserva-
tive force to retain more archaic ‘non-witnessed aorist uses’ in the face of rising
evidentials, or as a context encouraging ‘vivid, as-if-witnessed’ aorists in order
to lend credibility to Christian doctrine. In (42c), however, the vividness effect
is not connected to the life of Jesus or other figures in Christian belief, so this
is a more general strategy which is not prompted solely by religious doctrine.17
Then again, the ‘incoming’ perfects/evidentials are used for similar material
in (42d), repeated from (33). In this respect, the situation, with its unabated
characteristic variation, is not too far removed from the picture in the earlier
periods. After all, language change is a gradual and incremental process.
17 This can occasionally be encountered today, especially in the speeches of politicians
when they describe historic events. Gerdžikov (1984/2003: 34–36) supplies modern ex-
amples of the coexistence of evidentials and aorists for non-witnessed events (cf. Drinka
2017: 16, with references).
312 chapter 8
(42) a. ѿ адама до давiда i праведного iѡаки̇ма и̇ѡ̑сифа

ot adama do davida i pravednogo joakima josifa
from Adam to David and righteous Joachim Joseph
ѡ̑бручнiка колiко би̇-ха правѣднi и̇ ст͠ьi
obručnika koliko bi-xa pravĕdni i sty
fiancé how.many be-aor.3pl righteous and holy
пророци̇ патрiарси̇ нареко-ха се велiки̇ на земли̇
proroci patriarsi nareko-xa se veliki na zemli
prophets patriarchs call-aor.3pl refl great on earth
и̇ предъ бо͠га̇ не бе-ше ѿ нiхъ
i predъ boga ne be-še ot nixъ
and before God neg be-imperf.3sg of them
нiкои̇ трговѣцъ и̇ли̇ прехiтръ и̑ грделивъ [f. 4v, l. 10–14]
nikoj trgovĕcъ ili prexitrъ i grdelivъ
no.one merchant or cunning and proud
‘From Adam to David, and righteous Joachim and St. Joseph, how
many were righteous, and [how many] holy prophets and patriarchs,
[how many] called themselves great on earth and before God; not one
of them was a merchant or cunning and proud’
b. и̇ самъ хрстосъ оу̑ домъ простаго и̇ нiщетнаго

i samъ xrstosъ u domъ prostago i ništetnago
and alone Christ in home of.simple and poor
iѡси̇фа слезе и̇ пожи̇ве [f. 4v, l. 19–20]
josifa sleze i požive
Joseph come.down[aor.3sg] and live.for.a.while[aor.3sg]
‘And Christ himself came down and lived for a while in the home of
simple and poor Joseph’
c. ка́ко гсдъсвуюшхъ [sic] и̑ го͑рдещи́х се ѡ̑ наро́дѣхъ

kako gsdъsvujušxъ i gordeštix se o narodĕxъ
how dominant and proud refl of nations
iазьiкъ […] вънезапу и̑знемо́го-ша смири́-ша се
jazykъ vъnezapu iznemogo-ša smiri-ša se
tongue suddenly weaken-aor.3pl calm-aor.3pl refl
падо́-ша поги̇бо́-ша и̑счезо́-ша [f. 1v, l. 12–16]
pado-ša pogibo-ša isčezo-ša
fall-aor.3pl perish-aor.3pl disappear-aor.3pl
‘[if you would like to see…] how dominant and proud nations […] all of
a sudden grew weak, became quiet, fell, perished and disappeared …’
d. ѡ̑ни̇ правѣднi праѡ̑ц͠и̇ свi са би̇-л-и̇

oni pravĕdni praoci svi sa bi-l-i
those righteous forefathers all be.pres.3pl be-pptcp-pl
землоделатели̇ i ѡ̇вчаре и̇ би-л-и̇ богати̇
zemlodelateli i ovčare i bi-l-i bogati
farmers and shepherds and be-pptcp-pl rich
съ съкоти̇ и̇ с плоди̇ землнi и̇ би-л-и̇
sъ sъkoti i s plodi zemlni i bi-l-i
with animals and with fruits earthly and be-pptcp-pl
прости̇ на землi и̇ незлоби̇ви̇ [f. 4v, l. 16–19]
prosti na zemli i nezlobivi
simple on earth and meek
‘the righteous forefathers were all farmers and shepherds and were
rich in livestock and the fruits of the earth and they were simple [men]
on earth and meek’
Apart from periphrastic perfects, there are three conditional structures and
six be-passives in this sample. Two of the past passive participles, in (43), ap-
pear on their own without an auxiliary (in a quasi-impersonal post-modifier
clause); they fail to agree with the feminine singular antecedent they appar-
ently modify.
(43) Истори́-ѧ славѣноболгарск-аѧ … събра-н-о и̇

Istori-ę18 slavĕnobolgarsk-aę sъbra-n-o i
History(f)-sg Slavonic.Bulgarian-f.sg collect-pptcp-nt.sg and
нарежде-н-о паи́сіе́ мъ [f. 3r, l. 1–3]
narežde-n-o paisiemъ
order-pptcp-nt.sg by.Paisius
‘Slavonic-Bulgarian History […] compiled and put together by Paisius’
One of the purported passives is actually coordinated with an adjective, (44),

partially repeated from (32) (cf. English data in earlier chapters). While coor-
dination with adjectives may indicate incomplete grammaticalisation, there
18 The grapheme ѧ, originally employed to represent the Old Bulgarian/Old Church
Slavonic front nasal vowel, is probably used here (and elsewhere in the material, includ-
ing Damascenus Troianensis) to represent /ja/ under the influence of Russian Church
Slavonic. The pronunciation of this letter in Bulgaria during the National Revival is dis-
cussed in passing by Peev et al. (2015: 48).
314 chapter 8
is one occurrence of a perfect passive, in (45), showing an advanced stage of

grammaticalisation of both constructions.
(44) и̑ би̇-л-и̇ славн-и̇ и̇ чуе-н-i по сва

i bi-l-i slavn-i i čue-n-i po sva
and be-pptcp-pl famous-pl and hear-pptcp-pl on all
земла [f. 3v, l. 18–19]
zemla
earth
‘weren’t they famous and known (lit. heard) all over the world’19
(45) и̇ нiнѣ не и̇ма ѡнi кнiг-и̑ и̇

i ninĕ ne ima oni knig-i i
and now neg have[pres.3sg] those book(f)-pl and
летопи̇с-н-и̇ що са би̇-л-и̇ пространо
letopis-n-i što sa bi-l-i prostrano
chronicle-adj-pl that be.pres.3pl be-pptcp-pl in.detail
напи̇са-н-и̇ ради̇ свего народа [f. 5v, l. 2–4]
napisa-n-i radi svego naroda
write-pptcp-pl because.of all nation
‘now those chronicle books don’t exist which were written in detail on
account of the entire nation …’
As with the previous samples, no have-perfects proper were found in this 1762
excerpt, but instances of delexical have abound, perhaps paving the way for
the rise of the Modern Bulgarian have-perfect. Examples from Paisius’s History
outside the scope of my sample quoted in the literature include the following:
а болгари имеяли распра между собою .../a bolgari imejali raspra meždu sobo-
ju ‘and the Bulgarians had an argument with one another’, да има разговоръ/
da ima razgovorъ ‘(lit.) to have a conversation’ (Georgieva et al. 1989: 63, 66). In
addition to these instances of have with objects denoting actions, my material
yields (46a) and (46b), though here и̇меютъ/imejutъ and немашъ/nemašъ
could merely indicate possession of a tangible or non-tangible thing (cf.
Timberlake 1993: 874–875 for Russian, where the cognate of Bulgarian ‘have’
appears to have specialised in expressing the possession of abstract qualities
such as honour, influence or authority, unlike the cognate of ‘be’).
19 One might argue that, as in Modern Bulgarian, the adjective славни/slavni ‘glorious,
famous’ is distinguished from the participle славени/slaveni ‘worshipped, honoured,
glorified’.
(46) a. i зато и̇меютъ болгари̇ ѿ много iстори̇и

i zato imejutъ bolgari ot mnogo istorii
and therefore have Bulgarians from many histories
сви̇детелсво [f. 4r, l. 7–9]
svidetelsvo
evidence
‘that’s why the Bulgarians have evidence from many histories’
b. н-ема-шъ нiкои̇ прибитокъ ѿ гръчка

n-ema-šъ nikoj pribitokъ ot grъčka
neg-have-pres.2sg no benefit from Greek
мудростъ [f. 4r, l. 17–18]
mudrostъ
wisdom
‘you won’t have any benefit from Greek wisdom’
The meanings in (47) are arguably more dynamic. In (47a)–(47b), the abstract
objects of love, peace and hatred presuppose feelings/relations which one has/
experiences, rather than physical objects (concrete or otherwise) which one
can possess (similar examples were recorded in Damascenus Troianensis). In
(47c), Paisius exhorts readers to keep his book so that it does not get lost – the
2pl imperative form indicates a more dynamic sense of have, with the address-
ees capable of acting and taking care to preserve Paisius’s work. Static posses-
sive have would not typically appear in the imperative.
(47) a. ненави̇стъ що и̇меѧ-л-и̇ на богари̇ [f. 5v, l. 12]

nenavistъ što imeę-l-i na bogari
hatred which have-pptcp-pl on Bulgarians
‘the hatred which they had for the Bulgarians’
b. да би̇ имеѧ-л-и̇ ми̇ръ i лювъ [sic] с

da bi imeę-l-i mirъ i ljuvъ s
comp would have-pptcp-pl peace and love with
ца͠рi болгарскiе [f. 3v, l. 22]
cari bolgarskie
kings Bulgarian
‘so that they would have peace and love with the Bulgarian kings’
316 chapter 8
c. и̇ и̇ме-и-те ю да се не
i ime-j-te ju da se ne
and have-imperative-2pl 3f.acc.sg comp refl neg
погуби̇ [f. 3v, l. 9–10]
pogubi
perish
‘and have [i.e. keep] it [the book] so that it doesn’t get lost’
It is natural to wonder if these examples could be tied to Paisius’s south-western

origin, as surmised in the opening of this section. Are there so many delexical
uses of have, perhaps anticipating further grammaticalisation of perfect have,
because be is becoming more overburdened with evidential uses (though be
is now omissible in the third person), because this is a later period and/or be-
cause of the region which Paisius hailed from? In addition, living and working
on Mount Athos and relying on various Church Slavonic and (ultimately) other
European sources presupposes multilingual interaction (oral as well as writ-
ten), hence language contact and foreign influence. In (48), и̑машъ/imašъ is
already bleached of some of its possessive meaning and it takes a PP object
complement:
(48) O S V
сегашни̇ хiтрци̇ що гi ти̇ и̑ма-шъ
segašni xitrci što gi ti ima-šъ
present cunning.men that them thou have-pres.2sg
CO
за почестъ [f. 4v, l. 14–15]
za počestъ
for honour
‘our own cunning men, whom you hold in high regard’
Existential have in the default third-person singular has also been recorded
in (45) above; it arguably bears some resemblance to a nascent proto-have-
perfect (though identified with the benefit of a great deal of hindsight and
power of the imagination and far from convincing as evidence). At least it
could be viewed as a stepping-stone to the new perfect, especially with the pas-
sive participle in the relative clause and the existential use of have in the main
clause, as already noted in the previous chapter, though the link remains rather
tenuous. The final example does not showcase have at all (not in the relevant
first clause, that is). In (49) is a construction with another existential verb of
similar ontological category alongside a passive past participle as modifier
which (again with some hindsight) might remind the observer of certain prop-
erties of the later have-perfect, which was just beginning to be attested in the
varieties of Macedonia at about that time (see Chapter 7, Section 7.4.5). The
participle ‘written’ could be taken to modify the PPs ‘by little’ and ‘in short’,
but it could alternatively be a substantivised adjective acting as subject (itself
modified by those PPs).
(49) ако се и̇ находи̇ оу̑ многи̇ кнiги̇ i исторiи̇

ako se i naxodi u mnogi knigi i istorii
if refl and find[pres.3sg] in many books and histories
по мало и̇ вакратце напiса-н-о за
po malo i vakratce napisa-n-o za
by little and in.short write-pptcp-nt.sg about
болгари̇ но не можетъ сваки̇ чловѣкъ и̇мети̇ ѡ̇нi
bolgari no ne možetъ svaki človĕkъ imeti oni
Bulgarians but neg can every person have those
книги̇ [f. 5v, l. 19–21]
knigi
books
‘even if there is in many books and histories a little briefly written about
the Bulgarians, it is not possible for everyone to have those books’
Like the 14th- and 17th-century samples, the preface to Paisius’s 1762 master-
piece allows the use of be-perfects and aorists to express non-witnessed events.
What is new here is the frequent omission of the perfect auxiliary in order
to express non-witnessed events with a stand-alone participle in -l. In other
words, evidentials proper already turn up in great numbers in this text. Unlike
the aorist and the ‘real’ perfect, they are restricted to second-hand reports and
can hardly be met with in first-hand accounts. However, there is a good deal of
variation when it comes to second-hand reporting and all three constructions
can naturally be employed in such contexts, often in close proximity to each
other. Different copyists exhibit different preferences too, with a marked ten-
dency for later copyists to eliminate the auxiliaries (which then often resurface
in the modern translations). Crucially, the advent of evidentials has not led
to an immediate deployment of have-perfects in order to alleviate the burden
of be. As in the earlier centuries, there are nevertheless examples of delexical
have which might be seen as intermediate stages in the grammaticalisation
of this verb. And yet, it would be risky to posit a causal relationship between
the rise of evidentials and the new have-perfect because of the rather long
period of reasonable stability in a system which functioned happily without
318 chapter 8
have-perfects for centuries, traditionally relying on be-perfects (with or with-

out auxiliaries) for mediated reports, as well as in other situations.
8.5 Epilogue: the Have-Perfect in Modern Bulgarian
In order to assess to what extent the have-perfect has become grammati-

calised today, more than 250 years after Paisius, searches were conducted in
the Bulgarian National Corpus (http://dcl.bas.bg/bulnc/, accessed on 15 April
2017). Each present-tense form in the paradigm of имам/imam ‘have’ was in-
vestigated separately, extracting the first 120 occurrences yielded by the cor-
pus. The results are laid out in Table 4, demonstrating that this construction is
still embryonic at best. The evidence adduced below suggests that the level of
grammaticalisation of this incoming grammatical category remains quite low
in the present-day language.
First of all, potential have-perfects are rather few (even on a generous view
as to what should qualify as a perfect), so this construction does not appear
to be well established yet. Most of the occurrences were in the singular, with
only one instance in the first-person plural. Secondly, none of the examples
had default neuter singular agreement. In all relevant cases, the past passive
participle is placed right in front of the object noun, so it looks like a regular
agreeing adjective in attributive position. The four examples in (50a)–(53a) are
the most convincing ones treated as bona fide (proto-)have-perfects. They can
easily be paraphrased with a be-perfect, preserving the same salient meaning.
Those paraphrases are supplied by the author in (b) after each relevant attest-
ed clause in (a).
table 4 Have-perfects in Present-Day Bulgarian
Word form Total number of Have-perfects Have-perfects

occurrences examined with agreement without agreement
1sg имам/imam 120 2 0

2sg имаш/imaš 120 1 0
3sg има/ima 120 0 0
1pl имаме/imame 120 1 0
2pl имате/imate 120 0 0
3pl имат/imat 120 0 0
(50) a. Има-м подготве-н план.

Ima-m podgotve-n plan.
have-pres.1sg prepare-pptcp[m.sg] plan(m)[sg]
‘I have prepared a plan.’/‘I have a plan prepared(?)’ [have + passive
participle]
b. Подготви-л съм план.

Podgotvi-l săm plan.
prepare-pptcp[m.sg] be.pres.1sg plan
‘I have prepared a plan.’ [be + active participle]
(51) a. Има-м запазе-н танц.

Ima-m zapaze-n tanc.
have-pres.1sg reserve-pptcp[m.sg] dance(m)[sg]
‘I have reserved a dance.’/‘I have a dance reserved(?)’ [have + passive
participle]
b. Запази-л съм танц.

Zapazi-l săm tanc.
reserve-pptcp[m.sg] be.pres.1sg dance
‘I have reserved a dance.’ [be + active participle]
(52) a. вече има-ш спечеле-н-а позици-я

veče ima-š spečele-n-a pozici-ja
already have-pres.2sg win-pptcp-f.sg position(f)-sg
‘You have already won a position’ [have + passive participle]
b. вече си спечели-л позиция

veče si spečeli-l pozicija
already be.pres.2sg win-pptcp[m.sg] position
‘You have already won a position’ [be + active participle]
(53) a. има-ме уговоре-н-а срещ-а

ima-me ugovore-n-a srešt-a
have-pres.1pl arrange-pptcp-f.sg meeting(f)-sg
‘We have arranged a meeting’/‘We have a meeting arranged(?)’ [have
+ passive participle]
320 chapter 8
b. уговори-л-и сме среща

ugovori-l-i sme srešta
arrange-pptcp-pl be.pres.1pl meeting
‘We have arranged a meeting’ [be + active participle]
By contrast, no such paraphrase is naturally available for (54) below, a less

prototypical candidate which does not make the cut for Table 4. The sentence
in (54) does not saliently entail that the commission is in charge of granting
access; it is much more likely that someone else grants it. In the same vein,
phrases like вродено чувство/vrodeno čuvstvo ‘innate feeling’ in (55) are sta-
ble collocations whose components are often encountered together. Neither
is the subject of има/ima ‘have’ in (55) the agent of вродено/vrodeno ‘innate,
in-born’ (assuming participial, as opposed to purely adjectival, status), so this
example is a borderline case at best, too. (54) and (55) might even resemble
have-passives (see Chapter 7, Nicolova 2017: 380–381), but they hardly meet any
robust criteria for perfecthood.
(54) Комисията трябва да има осигуре-н

Komisijata trjabva da ima osigure-n
the.commission should comp have[pres.3sg] grant-pptcp[m.sg]
достъп до …
dostăp do …
access(m)[sg] to
‘The commission should be granted access to…’
(55) Той има в-роде-н-о чувств-о

Toj ima v-rode-n-o čuvstv-o
he.nom.3sg.m have[pres.3sg] in-bear-pptcp-nt.sg feeling(nt)-sg
за вкарване на голове.
za vkarvane na golove.
for scoring of goals
‘He has an innate ability (lit. in-born feeling) to score goals.’
The sentences in (56) and (57) are also rather dubious as have-perfects, again
because the participles behave syntactically like attributive modifiers within
the object NP (preceded by determiners). Once again, it is plain to see in (56)
that the subject of has (Slavija) is not the same as the implied agent of the
participle (cf. Chapter 7, Section 7.4.1). Even though they were not included
in Table 4, clauses like (56)–(57) might be seen as (remote) potential precur-
sors in the cline leading to the formation of a real have-perfect, especially the
impersonal (57).
(56) Трима контузени титуляри има Славия

Trima kontuzeni tituljari ima Slavija
three injured first.team.players have[pres.3sg] Slavija
‘Slavija has three first-team players who are injured’/‘Slavija has three in-
jured first-team players’
(57) трябва да има поне един или повече

trjabva da ima pone edin ili poveče
should comp have[pres.3sg] at.least one or more
квалифицирани лекари
kvalificirani lekari
qualified doctors
‘There should be at least one or more qualified doctors’
The sentence in (58) was discarded since the past participle is itself modified
by an adverb. Moving to the plural, (59) and (60) were likewise excluded be-
cause it is again obvious that the pro-dropped first-person plural subject of
имаме/imame is not the agent of the activity expressed by the passive past
participle. In other words, it was not us that supplied the woman in question
with weapons, and it was not us that wounded the person. All the examples so
far have been with passive past participles suffixed with -n or -t. A similar case
but with an active l-participle is illustrated in (61) (cf. Chapter 7).
(58) Има добре развит мозъчен дял.

Ima dobre razvit mozăčen djal.
have[pres.3sg] well developed brain area
‘He/She has a well developed area of the brain.’
(59) Има-ме въоръжена жена.

Ima-me văorăžena žena.
have-pres.1pl armed woman
‘We’ve got an armed woman.’
(60) Има-ме ранен човек тука!

Ima-me ranen čovek tuka!
have-pres.1pl wounded guy here
‘We’ve got a wounded guy here!’
322 chapter 8
(61) Така че сега има-ме стотици заседнали кораби из

Taka če sega ima-me stotici zasednali korabi iz
so that now have-pres.1pl hundreds stuck ships in
слънчевата система.
slănčevata sistema.
the.Solar System
‘So that we now have hundreds of ships stuck all over the Solar System.’
Based on this corpus evidence, I conclude that the have-perfect is still not en-
trenched enough in Bulgarian, and the language community does not seem to
‘feel a need’ to use it, despite the neat picture painted in the functionalist ac-
count whereby have-perfects are expected to ‘come to the rescue’ and diminish
the functional load of be in this domain. Interestingly enough, the material ex-
amined demonstrates more modal uses for have, some of them listed in (62)–
(67). (63) is noteworthy because there is no NP object of have or of the lexical
verb. The word order in (66) favours a grammaticalisation account (unlike in
the perfect above; cf. (64) and (65), with a more conservative word order). In
(67), the clubs are not yet in possession of the money, so semantic bleaching
has set in. There were of course plenty of delexical uses as well, not unlike
those from earlier centuries, but it is not necessary to dwell on them again.
This diversity of ‘polygrammaticalisation’ paths for various auxiliary roles does
indicate that generic verbs like have are the most suitable candidates for re-
cruitment as function words.
(62) Има-м доста да готв-я

Ima-m dosta da gotv-ja
have-pres.1sg quite.a.lot comp cook-pres.1sg
‘I have quite a lot to cook’/‘I have to cook quite a lot’
(63) Има-м да наваксва-м.

Ima-m da navaksva-m.
have-pres.1sg comp catch.up-pres.1sg
‘I have to catch up.’
(64) Има-ш нещо да ми каже-ш?

Ima-š nešto da mi kaže-š?
have-pres.2sg something comp me.dat.1sg say-pres.2sg
‘Have you got something to tell me?’
(65) Има-ме поне 20 сгради да евакуира-ме.

Ima-me pone 20 sgradi da evakuira-me.
have-pres.1pl at.least 20 buildings comp evacuate-pres.1pl
‘We have at least 20 buildings to evacuate.’/‘We have to evacuate at least
20 buildings.’
(66) Има-те да води-те битка ...

Ima-te da vodi-te bitka …
have-pres.2pl comp lead-pres.2pl battle
‘You have to fight a battle…’
(67) … които има-т да получава-т клубовете

… koito ima-t da polučava-t klubovete
which have-pres.3pl comp receive-pres.3pl the.clubs
‘[money] which clubs are to get’
My corpus results point to a rather incomplete and rudimentary grammati-

calisation of have-perfects. Finally, I conducted a small-scale pilot investi-
gation of the possible routes via which have-perfects could have entered
Bulgarian. It was noted in the previous chapter that the have-perfect is much
more fully grammaticalised in Macedonian, so it is plausible to assume that
it has been spreading from the south-west to the north-east. In order to test
this hypothesis, I extracted output from two Bulgarian authors. Dimitar Talev
was born in 1898 in Prilep, in today’s Republic of Macedonia, though he later
lived and worked in Sofia and elsewhere in Bulgaria. His novel Железният
светилник/Železnijat svetilnik ‘The Iron Lamp’ appeared in 1952. It is set in
late-19th-century Macedonia and traces its efforts to gain independence from
the Ottoman Empire. Dimitar Dimov was born in 1909 in Lovech, in central/
northern Bulgaria, though he studied in Dupnitsa, in the south-west, close to
today’s Republic of Macedonia, until he was ten years of age, before moving to
the capital Sofia. His novel Тютюн/Tjutjun ‘Tobacco’ was completed in 1951. It
is set in early-20th-century Bulgaria.
I examined the first 20,514 words of Talev’s Iron Lamp and the first 18,142
words of Tobacco by Dimov, searching for the way have was used in relevant
constructions. Neither excerpt produced any bona-fide have-perfects. There
were only instances of have where it appeared delexicalised to varying degrees.
(68) presents some of the relevant material from the first 20,514 words of The
Iron Lamp. (68a) approximates a modal use, whereas ‘have a wedding’ in (68b)
represents a more dynamic use. The sentences in (68c)–(68d) are interesting
324 chapter 8
because possessive have combines with something vaguely resembling an ob-

ject complement.
(68) a. – Има-ме дърв-а за цепене. – Харно ... Ще ги

– Ima-me dărv-a za cepene. – Xarno … Šte gi
have-pres.1pl wood-pl for cutting good fut them
на-цепи-ме.
na-cepi-me.
pref-cut-pres.1pl
‘We’ve got wood to cut.’ ‘Alright … we’ll cut it.’
b. Какво чува-м, Султано, сватба сте

Kakvo čuva-m, Sultano, svatba ste
what hear-pres.1sg Sultana wedding be.pres.2pl
има-л-и, невеста си стана-л-а
ima-l-i, nevesta si stana-l-a
have-pptcp-pl bride be.pres.2sg become-pptcp-f.sg
‘I hear, Sultana, that you’ve had a wedding, and you’ve become a bride’
c. Нели ти дадо-х вчера една кесия, да

Neli ti dado-x včera edna kesija, da
particle thee gave-past.1sg yesterday one purse comp
я има-ш за пари.
ja ima-š za pari.
it have-pres.2sg for money
‘Didn’t I give you a purse yesterday, to keep money in’
d. в кесията той имаше и двайсетина гроша свои

v kesijata toj imaše i dvajsetina groša svoi
in the.purse he had and twenty.something pence own
пари
pari
money
‘in the purse he also had about twenty pence of his own money’
Dimitar Dimov’s work also featured idiomatic combinations such as имам

доверие/планове/намерения/среща/враждебно отношение към … / imam
doverie/planove/namerenija/srešta/vraždebno otnošenie kăm … ‘have trust/
plans/intentions/a meeting/a hostile attitude towards …’. Future research
might elucidate if there is a link between collocations of have + a meeting
and potential (proto-)have-perfects with ‘similar ingredients’, like the one in

(53a) above. In (69) are other relevant examples from the first 18,142 words
of Tobacco, where the meaning is a little more dynamic. ‘He has a verdict’ in
(69a) is semantically equivalent to ‘he was/got sentenced’, whereas ‘we had a
regional conference’ in (69b) is about holding an event.
(69) a. има присъда от септемврийските събития

ima prisăda ot septemvrijskite săbitija
have[pres.3sg] verdict from the.September events
‘he has a verdict from the events in September’
b. Преди две седмици има-хме областна конференция на

Predi dve sedmici ima-xme oblastna konferencija na
before two weeks have-past.1pl regional conference on
Витоша.
Vitoša.
Vitoša
‘Two weeks ago we had a regional conference on (Mount) Vitoša.’
In (70), again from Tobacco, have retains its meaning of possession but the ob-
ject has a past passive participle as a pre-modifier. It is clear from the adverb
добре/dobre that разработено/razraboteno is an adjectival attributive modi-
fier of дюкянче/djukjanče, but this construction could still be perceived as a
rather remote precursor/‘distant relative’ of the have-perfect (cf. the analogous
sentence in (58) above). This is the closest to a possessive perfect that one gets
in these fiction samples.
(70) … шурея си, който имаше добре раз-работено

… šureja si, kojto imaše dobre raz-raboteno
brother-in-law own who had well pref-worked
дюкянче на главната улица.
djukjanče na glavnata ulica.
little.shop on the.main street
‘… his brother-in-law, who had a well developed little shop on the main
street.’
In sum, neither author exhibits any particularly innovative tendencies to use

have-perfects in the opening portions of their work which was mostly done in the
1940s and appeared in the early 50s. Interestingly enough, the Macedonia-born
novelist here is even a little less progressive than his fellow-writer born in
326 chapter 8
central Bulgaria (though they both moved around a good deal). This outcome
might in part be due to the supposedly initially ‘colloquial nature of the ima
construction, which [causes] it to be avoided in the literary language’ (Elliott
2001: 46). Despite these inconclusive results from an admittedly small-scale
study, it can still be surmised that the geographical spread of these new tense/
aspect forms must have followed a trajectory from west to east, in line with
the fuller grammaticalisation in the modern standard and dialectal varieties
in the Republic of Macedonia as compared to those of Bulgaria, where have
remains at an embryonic stage as a perfect auxiliary. Much more extensive
work is needed in order to shed light on the actual transmission routes. This is
outside the scope of the present book and will have to await future research.
8.6 Conclusion
As I did for English in the previous chapters, here too I have tried to analyse
small chunks of the history of Bulgarian which give insights into the bigger
picture, including inter- and intra-systemic relations, as well as the impact of
external factors such as contact. The selection included representative texts
from the 14th, 17th and 18th centuries. The 14th-century Middle Bulgarian text
yielded some breakdowns in agreement with periphrastic verbal construc-
tions, which however are in a clear minority (and remain so throughout). No
have-perfects were found, nor were any structures that could be perceived as
their immediate precursors, so be is the only auxiliary in this role here. Some
of the instances of the be-perfect are contextually used for non-witnessed
events (and have been since the earliest records), though the auxiliary is never
omitted and the aorist can report non-witnessed events as well. There are also
indications that even in Present-Day Bulgarian, omission of the auxiliary in
third-person evidentials is not consistent, so some things have changed very
little since mediaeval times. Even today, speakers might occasionally use aor-
ists for non-witnessed events. The same range of options is still available but
their distribution has shifted dramatically by now.
By the time of the representative of 17th-century Bulgarian post-Ottoman-
contact, I detect no significant shifts in the distribution or functional load of
the be-perfect compared to the 14th century, including the same contextually
evidential uses. Although evidentials proper exist in damascene literature
more widely (and probably predate it), they do not show up in my sample,
where the perfect and aorist are still in free variation, so an obligatory category
of evidentiality has, strictly speaking, not yet come into being. No have-perfects
were employed either, though there were instances of delexical transitive have
(available since earliest times), albeit without a participial object complement.

Arguably, they are an intermediate stage in the grammaticalisation of have, a
consequence of its generic meaning more than anything else.
18th-century monk and historian Paisius of Hilendar already deploys copi-
ous amounts of evidentials without an auxiliary. A contrast between witnessed
aorists and non-witnessed evidentials is beginning to take shape. However, evi-
dentials are still in variation with be-perfects that retain the auxiliary, as well
as with non-witnessed aorists. A few defaults in agreement are also present,
but they are unrelated to have-perfects. The advent of evidentials has not ‘un-
leashed’ have-perfects in order to rein in the multifunctionality of the earlier
generation of perfect. As in the previous centuries, there are nevertheless ex-
amples of delexical have which might be perceived as intermediate stages in
the grammaticalisation of this verb. It is precarious to posit any causal links
between the rise of evidentials and the new have-perfect because of the rather
long period of stability in a system which functioned without have-perfects for
centuries, traditionally relying on be-perfects (with or without auxiliaries) for
mediated (and non-mediated) reports.
Judging by the results from the Bulgarian National Corpus and fiction from
the 1950s, have-perfects are still exceedingly rare, and agreement tends to stay
robust – this is conditioned (1) by the well-preserved agreement morphology
of Slavic non-finite forms in general, where even highly entrenched be-perfects
and passives still exhibit agreement in the overwhelming majority of cases, and
(2) by the rather incomplete grammaticalisation of have-perfects in Bulgarian
compared to Macedonian, where participles in have-perfects no longer inflect
to agree. So, Bulgarian speakers do not seem to have caved in to any potential
functional pressures to make greater use of a have-perfect and to grammati-
calise it further. In fact, I came across more grammaticalised modal uses in
the corpus material. Therefore, similarly to English (cf. Chapter 3, esp. Section
3.5.3.2), the grammaticalisation paths of have in Bulgarian are more likely to
have been conditioned by its generic meaning, which naturally invites reanaly-
sis and greater abstraction in line with the human species’ cognitive endow-
ment, rather than by any perceived ‘functional needs’ of the language or its
speakers. The naturalness of grammaticalising generic verbs like be and have
is supported by the (re-)emergence of be and have perfects in both languages
under investigation, including new waves of be-perfect in English, alongside
the various other auxiliary uses for both verbs. The numerous light-verb con-
structions reviewed here confirm this conclusion too.
And yet, considering the sparsity of have-perfects outside of Europe, cou-
pled with the geographical patterns of distribution outlined in this and in
previous chapters, contact must also have been instrumental in promoting
328 chapter 8
the developments in question. With all things considered, English remains an

overwhelmingly ‘have language’, whereas be still reigns supreme in Bulgarian
in accordance with the east/west split in perfect formation in Europe. Its geo-
graphical location and contact with Scandinavian must have pushed English
to lose be as a viable perfect auxiliary and it thus remains a typical representa-
tive of the west, with its predominant have-perfect. This have-perfect might
have been spreading from the west into Bulgaria, but this construction is still
only rudimentary and underdeveloped here and the two languages remain dia-
metrically opposed in terms of building their perfects. In the closing chapter, I
survey some of the more global issues that have emerged from the discussion
in the preceding text, in an attempt to tie together all the main threads of the
argument.
chapter 9
Conclusions
In this brief final recapitulation, I sum up some of the major contributions

of this monograph, draw some general conclusions and point to limitations
and directions for future research. This study builds on a very long tradition of
scholarship and presents a systematic synthesis and evaluation of a substantial
amount of previous literature. It aims for a marriage of more abstract general
theory and the nitty-gritty of in-depth analysis of a wide range of linguistic
data. The analysis of English is based on texts spanning the period between
the 9th and the 19th centuries, with a sneak-peek comparison to a German
play from the 18th century. The Bulgarian material comes from the 14th, 17th
and 18th centuries. I also seek to elucidate the current state of the relevant con-
structions in both languages by adducing appropriate contemporary material.
In reviewing and assessing earlier work and collecting a customised and
well-balanced database, this survey offers some reinterpretations of data and
revisions of previous analyses, which moves the debate forward. It is demon-
strated that some of the longest-held and most entrenched views, including
the role of ambiguity and functional load, seem to have been massively over-
stated. Both scholarly treatments and teaching materials (which have thus far
cemented and perpetuated the centrality of functional factors) ought to be
updated accordingly. This monograph additionally moves the debate forward
by providing empirical support for broad-brush programmatic claims made
in the existing literature, though left untested, such as the claims regarding
the Scandinavian influence on the development of the perfect in English. All
the Old and Middle English texts from the north which I examined show con-
sistently more progressive loss of be-perfects compared to material from the
south, and Chapter 4 contains an authentic example of direct replacement
of be with have in a manuscript of northern provenance. As with many other
grammatical features, the bilingual interaction with Scandinavian speakers
must have encouraged the selection of the auxiliary which was more salient in
both those related Germanic varieties, crucially also the predominant auxiliary
in the language of the invaders.
Despite long years of research, some aspects of the phenomena described
in this book have been ignored or remain insufficiently explored. The spotlight
has been shone on them here and some interesting patterns and fascinating
discoveries have emerged, e.g. a new potential get-perfect in English, differ-
ent trends in losing agreement in OE passive and perfect constructions, or

330 chapter 9
the role of morphosyntactic marking and ambiguity in actual output. To the

best of my knowledge, these points have not been considered in great detail in
the specialist literature and it is hoped that this study will provide an impetus
for more extensive work. Fresh insights were also offered regarding the links
between Bulgarian be-perfects, evidentials and the advent of a have-perfect,
including its (insecure) status in Present-Day Bulgarian. This incoming con-
struction is still only in the very earliest stages of grammaticalisation and its
appearance cannot be justifiably linked to shifts in the functional load of the
earlier be-perfect/evidential.
Although I have argued against purely or predominantly functional expla-
nations, it is probably the case that the developments were due to the complex
interactions of a multitude of factors, such as contact, frequency and indeed
ambiguity that can lead to reinterpretation. For instance, the infrequency
of be-perfects in English might have worked in tandem with contact with
Scandinavian to bring about their ultimate ‘downfall’. Above all, I have tried to
debunk popular myths and facile accounts which simply say that this process
was set in motion mainly (or obviously) because of ambiguity and functional
load. The linguistic facts do not seem to corroborate such statements.
The relevant linguistic facts are actually much more intractable than anyone
writing a book about them would wish them to be. Contemplating the bigger
picture might undermine even the credibility of frequency as an explanatory
notion. Have may well have displaced be from the perfect in Standard English
because it was more frequent, in line with the scenario familiar by now. But
why then have be-perfects been retained or created anew in numerous (stan-
dard as well as non-standard) varieties of English and extended to transitive
verbs (see Chapter 3)? The German be-perfect also survives, in spite of being
numerically less common than have (Chapter 6). To a certain extent, the same
logic might apply to Standard Bulgarian – in early Proto-Slavonic, be must have
started out its career as a perfect auxiliary for intransitives, but then it was ex-
tended to all verbs, irrespective of valency (see Chapter 7). The initial valency
restrictions did not hamper its ultimate ascendancy, but perhaps it faced no
competition comparable to the one in English. As noted in earlier chapters,
Slavonic uses a form of be + an active participle. The perfect is thus built in a
more straightforward fashion than the western be-perfects, which are, in some
sense, periphrastic deponents, not just intransitives (B. Drinka, p.c.). This
equation between Slavic and Germanic is therefore not unproblematic, but it
is still a valid point that an initially restricted ‘minority’ pattern came to gain
ground and be extended to more and more contexts. Likewise, South Slavic
have-perfects are also currently underdogs, but they seem to be spreading from
west to east, especially in the Slavic varieties of Macedonia, suggesting that
Conclusions 331
contact might work in a way that overrides frequency. Similarly, in Old English
the ubiquity of legitimate zero marking was proved to play no part in the loss
of overt agreement (Chapter 4). Agreeing plural passives held their ground in
spite of being in the minority, and so did perfect be in English, for a very long
time. Rare(r) items are not necessarily doomed. Though this might be hard to
admit, the messiness and unpredictability of language change often thwarts
even the most ingenious attempts at imposing neatness and orderly patterns
on its vagaries. Delineating violable tendencies is sometimes the best one
can do.
I hope that this study gives a taste of how complex the developments under
consideration are, especially when approached from different angles. Mapping
out parallel developments in two languages has helped to highlight that and
to cross-pollinate ideas from the two scholarly traditions. Another possible
cross-pollination might involve an exciting link between some of the regulari-
ties that have emerged for both languages under investigation, as pointed out
to me by M. Kovatcheva (p.c.). A connection was noted in Chapter 3, Section
3.5, between schemas based on I have something and something is at/to me,
both featuring generic and extremely common lexemes such as be and have
(cf. Drinka 2017: 89–90, with references). In addition, it should be stressed that
past participles can have an adverbial function, and it is precisely the adver-
bial suffixes that appear to act as defaults in certain contexts. In Slavonic, -o
serves both as neuter singular marker on adjectives and as a marker of adverbs,
whereas -e, apart from frequently indicating nominative/accusative plural on
OE adjectives, also marked adverbs in early Germanic, including Old English
(see Chapter 4). These observations would account for the interchangeabil-
ity of the auxiliary verbs in the perfect and the possibility of one ousting the
other, as well as for the changes in the morphological exponence. Due to the
above-mentioned equivalence of the schemas, it no longer seems strange that
a passive participle can acquire an agent (in the perfect), or that an active par-
ticiple (e.g. arrived) can come to combine with have as opposed to be (or vice
versa: see below for transitives with perfect be). Grammaticalisation is indeed
accompanied by loss of agreement, but rather than being a trigger, the loss of
agreement (or the collapse of other morphological distinctions) is the ultimate
outward sign that the construction has been reanalysed. This would enable
the transition from (1a) to (1b) to (1c), all of them paradigmatically related.
The transition must have proceeded via an ‘adverbial’ stage which bestowed
more verbal features on the originally adjectival participle. Adverbs are not
expected to agree, which would in turn account for the shedding of inflections
or switching to petrified neuter singular forms. Essentially, no zero morphol-
ogy is needed as a bridging context. To the extent that agreement in Modern
332 chapter 9
Bulgarian tends to survive much better than it did in late Old English and early
Middle English, this probably has to do with the better preserved overall mor-
phology of Slavic compared to that of Germanic, as suggested in Chapter 1.
Additionally, this has to do with the lesser degree of grammaticalisation of
have-perfects, as confirmed by comparing Bulgarian to Macedonian (where
agreement is dropped).
(1) a.
Бележк-и-те са написа-н-и
Beležk-i-te sa napisa-n-i
note(f)-pl-def be.pres.3pl write-pptcp-pl
‘The notes are written’/‘The notes have been written’
b. Има-м написа-н-и бележк-и

Ima-m napisa-n-i beležk-i
have-pres.1sg write-pptcp-pl note(f)-pl
‘I have written notes’
c. Има-м написа-н-о бележк-и

Ima-m napisa-n-o beležk-i
have-pres.1sg write-pptcp-nt.sg note(f)-pl
‘I have written notes’
The endowment of the participle with more verbal as opposed to adjectival

features further explains the extension of this pattern to intransitive verbs,
which would otherwise not form past passive participles in -n. Thus, the ap-
pearance of be- and have-perfects with intransitive past passive participles in
South Slavonic, e.g. сум дойден/имам дойдено / sum dojden/imam dojdeno
‘I am/have come’ (Chapter 7), as well as the appearance in dialectal English of
transitive be-perfects such as I’m had days … (Chapter 3), both reaffirm that,
given enough bleaching and generalisation of meaning, the morphological dis-
tinctions between active and passive past participles (the latter normally ex-
clusively transitive) can be overridden, as can the distinct uses of perfect have
and be (the latter normally restricted to intransitives in Germanic). All of this
is due to deriving ever more abstract patterns and uses of already interchange-
able delexical verbs and the abstract schemas in which they participate.
In sum, this study set out to examine two distant Indo-European relatives
which share some typological traits (e.g. advanced analyticity), for which they
stand out among their closest sister-languages. Bulgarian and English are fur-
thermore typical representatives of the east/west split in perfect formation in
Europe, making them an ideal pair for contrastive analysis. At the end of the
Conclusions 333
day, they remain basically faithful to what is geographically expected of them –

Standard English continues to rely almost exclusively on have, whereas be still
holds its ground in Bulgarian. It was demonstrated that, albeit coming from
very different perspectives and academic traditions, analogous functional
explanations have been offered regarding the development of be- and have-
perfects in both languages. The loss of perfect be in English has been linked to
the multiple other functions with which this verb came to be ‘overburdened’.
Similarly, the emergence of nascent have-perfects in Bulgarian/Macedonian
has been linked to the ‘functional overload’ of be due to its newer role as evi-
dential. These facile explanations have been discredited as untenable here for
both languages alike.
Bulgarian and English both prove to be essentially indifferent to ambiguity –
morphological or constructional. Ambiguous zero-exponence of agreement
morphology played no pivotal part in the reanalysis of Old English proto-
perfects, and the structural ambivalence of perfects/passives/statives based
on be was almost certainly not detrimental to the survival of the be-perfect.
As confirmed by German, which started out from a similar starting point to
English (Chapter 6), English did not really ‘need’ to discard its be-perfect. The
predominance of have in the west is more plausibly conditioned by the limited
array of participles there – only one all-purpose past participle doubles as ac-
tive perfect and passive participle, hence the be/have dichotomy plays a key
role in marking the separate constructions. Still, as demonstrated by German
or dialectal English with be-perfects, be could have persisted as a perfect un-
derdog in Standard Modern English too, so there must be another reason for
things to have turned out as they did. Slavonic languages, on the other hand,
possess dedicated past active and past passive participial forms, so there was
no need for have to step in as perfect auxiliary at all. If anything, the construc-
tion with have + past passive participle disrupts the otherwise neat pattern of
using be with past active participles for the perfect and past passive ones for
the passive. The ensuing confusion between the two types of participle further
collapses a hitherto robust formal distinction.
Therefore, the changes that took place in both languages must have been
much more mechanistic than functionally-driven, a natural consequence of
the generic meanings of verbs like be, have and get, which predispose speak-
ers to reanalyse them and give them increasingly more abstract, semantically
bleached, interpretations and hence more grammatical roles. This much is
suggested by the cyclic (re)birth(s) of be- and get-perfects (and passives) in
English, by the numerous overlapping auxiliary roles for such verbs in vari-
ous languages in Europe and beyond, as well as by their frequent participation
in delexical light-verb constructions, even their semantic interchangeability,
334 chapter 9
e.g. in be at to mean ‘have’ in certain varieties of English or impersonal има/

ima ‘have[pres.3sg]’ to mean ‘exist, there is’ in Bulgarian (cf. also the other
examples above). The ontological categories of generic existence and posses-
sion were thus instrumental in the cognitive processes of reinterpretation and
greater grammaticalisation, as it were ‘inviting’ them to get underway.
And yet, geography and contact cannot be ignored as decisive stimuli. The
areal distribution outlined in the preceding chapters is quite telling. English
be-perfects appear to have been lost first in the Viking-settled north, align-
ing English with the languages on the western periphery of the Charlemagne
Sprachbund. This western periphery is characterised by the predominance of
have and marginality/non-existence of be as temporal/aspectual marker of
the perfect, whereas the languages at the centre of this linguistic area, French
and German, have stayed staunch pillars of a healthier be/have dichotomy.
Although Bulgarian remains a regular representative of the east in terms
of perfect formation, relying more or less exclusively on be, neighbouring
Macedonian has grammaticalised have-perfects much more fully (especially
in its south-western dialects). Since there was no functional need to do so in
either of these South Slavic varieties, it must be concluded that the newer per-
fect configuration is coming in from the west, prompted by contact. The typo-
logical oddity of have as a perfect auxiliary outside of Europe also supports this
scenario. The naturalness of reanalysing and recruiting generic verbs for ever
more grammatical roles must have been aided by contact.
In order to shed light on all these intricate processes, my aim has been
to qualitatively and quantitatively probe the usages of individual authors
or scribes, such as Dickens or the copyists of portions of the Anglo-Saxon
Chronicles, offering windows into their idiolects and unravelling the principles
that must have guided the linguistic choices they made. The benefit of this
close-reading approach is that the surrounding context often turned up clues
that would have remained elusive if I had just trawled through a large-scale
corpus without extensively examining the sources. Although my samples here
are not large, and there is perhaps not enough material to allow the establish-
ment of ironclad statistical significance, some tendencies do seem to manifest
themselves for this arbitrary (sub-)group of language users across time. An av-
enue for future research would be to confirm whether those tendencies persist
when a larger body of data is investigated and subjected to statistical analysis
with multiple variables.
References
Aikhenvald, Alexandra. 2002. Language Contact in Amazonia. Oxford: Oxford

University Press.
Aitchison, Jean. 2013. Language Change: Progress or Decay? 4th ed. Cambridge:
Cambridge University Press.
Allen, Cynthia L. 1995. On doing as you please. In: Jucker, Andreas H. (ed.), Historical
Pragmatics: Pragmatic Developments in the History of English, 275–308. Amsterdam/
Philadelphia: John Benjamins.
Ammann, Andreas. 2005. Abbau und Anschwemmung: Doppelte Perfektformen und
Grammatikalisierung im deutschen Tempussystem. In: Leuschner, Torsten, Tanja
Mortelmans and Sarah De Groodt (eds.), Grammatikalisierung im Deutschen, 251–
275. Berlin: Walter de Gruyter.
Anderson, John M. 1997. Remarks on the structure and development of the have per-
fect. Folia Linguistica Historica 18(1–2): 3–23.
Asenova, Petja. 1987: Асенова, Петя. 1987. Относно имам-перфектните форми в
българския език [On have-perfect forms in Bulgarian]. Български език XXXVII:
149–151.
Askedal, John Ole. 2005. Grammatikalisierung und Persistenz im deutschen
“Rezipienten-Passiv” mit bekommen/kriegen/erhalten. In: Leuschner, Torsten, Tanja
Mortelmans and Sarah De Groodt (eds.), Grammatikalisierung im Deutschen, 211–
228. Berlin: Walter de Gruyter.
Atanasova, Vera and Tekla Sugareva. 2004: Атанасова, Вера и Текла Сугарева. 2004.
Немска граматика [A German Grammar]. Пето издание. София: Наука и
изкуство.
Balan, Aleksandăr Teodorov. 1957: Балан, Алексаднър Теодоров. 1957. Особит
състав с глагол имам [A peculiar construction with the verb have]. Известия на
Института за български език V: 23–29.
Barton, Anne. 1997. The Taming of the Shrew. In: Evans, G. Blakemore and J. J. M.
Tobin (eds.), The Riverside Shakespeare. 2nd ed., 138–141. Boston: Houghton Mifflin
Company.
Bauer, Brigitte. 2000. Archaic Syntax in Indo-European. The Spread of Transitivity in
Latin and French. Berlin and New York: Mouton de Gruyter.
Baugh, Albert C. and Thomas Cable. 2013. A History of the English Language. 6th ed.
London and New York: Routledge.
Beal, Joan C. 2004. English in Modern Times: 1700–1945. London: Hodder Education.
Beaugrande, Robert de. 1994. Function and form in language theory and research: The
tide is turning. Functions of Language 1/2: 163–200.
336 References
Benson, Larry D. (ed.) et al. 1987a. The Riverside Chaucer. 3rd ed. Boston: Houghton
Mifflin Company.
Benson, Larry D. 1987b. The canon and chronology of Chaucer’s works. In: Benson,
Larry D. (ed.) et al., The Riverside Chaucer. 3rd ed., xxvi–xxix. Boston: Houghton
Mifflin Company.
Benson, Larry D. 1987c. The texts. In: Benson, Larry D. (ed.) et al., The Riverside Chaucer.
3rd ed., xlv–xlvii. Boston: Houghton Mifflin Company.
Berdicevskis, Aleksandrs and Hanne Martine Eckhoff. 2016. Redundant features are
less likely to survive: Empirical evidence from the Slavic languages. In: Roberts, S.
G., C. Cuskley, L. McCrohon, L. Barceló-Coblijn, O. Feher and T. Verhoef (eds.), The
Evolution of Language: Proceedings of the 11th International Conference (EVOLANG11),
69–77.
Bergs, Alexander and Gabriele Diewald (eds.). 2008. Constructions and Language
Change. Berlin: Mouton de Gruyter.
Biber, Douglas, Susan Conrad and Geoffrey Leech. 2002. Longman Student Grammar of
Spoken and Written English. Harlow: Longman.
Bischoff, Shannon T. and Carmen Jany (eds.). 2013. Functional Approaches to Language.
Berlin: De Gruyter Mouton.
Bojadžiev, Todor. 1968: Бояджиев, Тодор. 1968. Аналитични форми с глагола имам в
тракийските говори [Analytic forms with the verb have in the varieties of Thrace].
Известия на Института за български език XVI: 459–465.
Bojadžiev, Todor, Ivan Kucarov and Jordan Penčev. 1999: Бояджиев, Тодор, Иван
Куцаров и Йордан Пенчев. 1999. Съвременен български език: фонетика,
лексикология, словообразуване, морфология, синтаксис [Modern Bulgarian:
Phonetics, Lexicology, Word-Formation, Morphology, Syntax]. София: ИК Петър
Берон.
Bowie, Jill, Sean Wallis and Bas Aarts. 2013. The perfect in spoken British English. In:
Aarts, Bas, Joanne Close, Geoffrey Leech and Sean Wallis (eds.), The Verb Phrase in
English: Investigating Recent Language Change with Corpora, 318–352. Cambridge:
Brinker, Klaus. 1990. Das Passiv im heutigen Deutsch. In: Petkov, Pawel and Diana
Sliwkowa (eds.), Ausgewählte Theoretische Literatur zur Morphologie der Deutschen
Sprache, 398–435. Sofia: Sofioter Universität “Kliment Ochridski”.
Brinton, Laurel J. 1988. The Development of English Aspectual Systems: Aspectualizers
and Post-verbal Particles. (Cambridge Studies in Linguistics 49.) Cambridge:
Brinton, Laurel J. and Leslie K. Arnovick. 2011. The English Language. A Linguistic
History. 2nd ed. Oxford: Oxford University Press.
Brown, Keith and Jim Miller. 2016. A Critical Account of English Syntax. Grammar,
Meaning, Text. Edinburgh: Edinburgh University Press.
References 337
Burnley, David. 1992. The History of the English Language: A Source Book. London:
Longman.
Burrow, J. A. and Thorlac Turville-Petre. 2005. A Book of Middle English. 3rd ed. Oxford:
Blackwell Publishing.
Bužarovska, Eleni and Liljana Mitkovska. 2010. The grammaticalization of habere-
perfect in standard Macedonian. Balkanistika 23: 43–66.
Bybee, Joan L. 1988. The diachronic dimension in explanation In: Hawkins, John A.
(ed.), Explaining Language Universals, 350–379. Oxford: Blackwell.
Bybee, Joan L. 2003. Mechanisms of change in grammaticization: The role of frequency.
In: Janda, Richard and Brian Joseph (eds.), The Handbook of Historical Linguistics,
602–623. Oxford: Blackwell.
Bybee, Joan L. 2013. Usage-based theory and exemplar representations of construc-
tions. In: Hoffmann, Thomas and Graeme Trousdale (eds.), The Oxford Handbook of
Construction Grammar, 49–69. Oxford: Oxford University Press.
Bybee, Joan L. and Östen Dahl. 1989. The creation of tense and aspect systems in the
languages of the world. Studies in Language 13: 51–103.
Bybee, Joan L. and Paul J. Hopper (eds.). 2001. Frequency and the Emergence of Linguistic
Structure. (Typological Studies in Language 45.) Amsterdam/Philadelphia: John
Benjamins.
Bybee, Joan L., Revere D. Perkins and William Pagliuca. 1994. The Evolution of Grammar:
Tense, Aspect and Modality in the Languages of the World. Chicago: University of
Chicago Press.
Bybee, Joan L. and Sandra A. Thompson. 1997. Three frequency effects in syntax.
Berkeley Linguistics Society, 378–388.
Carey, Kathleen. 1995. Subjectification and the development of the English perfect.
In: Stein, Dieter and Susan Wright (eds.), Subjectivity and Subjectivisation, 83–102.
Cambridge: Cambridge University Press.
Chankova, Yana. 2007a. Possible Scandinavian influence on the grammar of Middle
English: The uses of the perfect in Early Middle English reconsidered. Contrastive
Linguistics XXXII(1): 44–55.
Chankova, Yana. 2007b: Чанкова, Яна. 2007b. Употреба на перфекта в превода на
Уиклиф и превода на Тиндейл в сравнение с оригиналите [Use of the perfect
in Wycliffe’s and Tyndale’s translations compared to the originals]. В: Багашева,
Александра (съст.), За човека и езика. Сборник научни статии, посветени на
60-годишнината на проф. д.ф.н. Майя Пенчева, 510–524. София: Университетско
издателство “Св. Климент Охридски”.
Chankova, Yana. 2008: Чанкова, Яна. 2008. Произход и развитие на аналитичен
перфект в английския език [Origin and development of an analytic perfect in
English]. Дисертация за присъждане на образователната и научна степен
“доктор”. ЮЗУ “Неофит Рилски”, Благоевград.
338 References
Chankova, Yana. 2009. From SOVv to SvVX – A Minimalist approach: An account

based on a study of the perfect periphrasis in Old English. In: Trendafilov, V. and
I. Vassileva (eds.), Boundaries, Boundary Crossing, Cross-Boundary Transfer, 425–
435. Blagoevgrad: South Western University Press.
Close, Joanne and Bas Aarts. 2010. Current change in the modal system of English: A
case study of must, have to and have got to. In: Lenker, Ursula, Judith Huber and
Robert Mailhammer (eds.), English Historical Linguistics 2008: Selected Papers from
the Fifteenth International Conference on English Historical Linguistics (ICEHL 15),
Munich, 24–30 August 2008. Volume I: The History of English Verbal and Nominal
Constructions, 165–181. Amsterdam: John Benjamins.
Comrie, Bernard. 1976. Aspect. An Introduction to the Study of Verbal Aspect and Related
Problems. (Cambridge Textbooks in Linguistics.) Cambridge: Cambridge University
Press.
Comrie, Bernard and Greville G. Corbett (eds.). 1993. The Slavonic Languages. London:
Routledge.
Cortes, Nuria Calvo. 2019. Conservatism or the influence of the semantics of mo-
tion situation in the choice of perfect auxiliaries in Jane Austen’s letters and nov-
els. In: Claridge, Claudia and Birte Bös (eds.), Developments in English Historical
Morpho-Syntax. (Current Issues in Linguistic Theory 346), 175–198. Amsterdam:
John Benjamins.
Coussé, Evie, Peter Andersson and Joel Olofsson (eds.). 2018. Grammaticalization Meets
Construction Grammar. Amsterdam: John Benjamins.
Croft, William. 2000. Explaining Language Change: An Evolutionary Approach. London:
Pearson.
Croft, William. 2009. Evolutionary models and functional-typological theories of lan-
guage change. In: van Kemenade, Ans and Bettelou Los (eds.), The Handbook of the
History of English, 68–92. Oxford: Wiley-Blackwell.
Crystal, David. 2004. The Stories of English. New York: The Overlook Press.
Cysouw, Michael. 2011. Quantitative explorations of the world-wide distribution of
rare characteristics, or: the exceptionality of northwestern European languages.
In: Simon, Horst J. and Heike Wiese (eds.), Expecting the Unexpected: Exceptions in
Grammar, 411–431. Berlin: Mouton de Gruyter.
Danchev, Andrei. 1969. The parallel use of the synthetic dative instrumental and peri-
phrastic prepositional constructions in Old English. Annuaire de l’université de
Sofia, Faculté des lettres, LXII(2): 40–100.
Danchev, Andrei. 1997. The Middle English creolization hypothesis revisited. In: Fisiak,
J. (ed.), Studies in Middle English Linguistics, 79–108. Berlin: Mouton de Gruyter.
Danchev, Andrei and Merja Kytö. 1994. The construction be going to + infinitive in Early
Modern English. In: Kastovsky, Dieter (ed.), Studies in Early Modern English, 59–77.
Berlin: Mouton de Gruyter.
References 339
Danylenko, Andrii. 2005. Is there any possessive perfect in North Russian? Word 56(3):
347–379.
Davis, Norman. 1987. Language and versification. In: Benson, Larry D. (ed.) et al., The
Riverside Chaucer. 3rd ed., xxix–xlv. Boston: Houghton Mifflin Company.
Denison, David. 1993. English Historical Syntax: Verbal Constructions. London/New
York: Longman.
Denison, David. 1998. Syntax. In: Romaine, Suzanne (ed.), The Cambridge History of
the English Language. Vol. IV: 1776–1997, 92–329. Cambridge: Cambridge University
Press.
Dobrev, Ivan. 1973: Добрев, Иван. 1973. Произход и значение на старобългарския
перфект [Origin and meaning of the Old Bulgarian perfect]. Известия на
Института за български език ХХII: 3–20.
Dolberg, Florian. 2012. Gender change from Old to Middle English. In: Hegedűs, Irén
and Alexandra Fodor (eds.), English Historical Linguistics 2010: Selected Papers from
the Sixteenth International Conference on English Historical Linguistics (ICEHL
16), Pécs, 23–27 August 2010. (Current Issues in Linguistic Theory 325), 263–288.
Amsterdam: John Benjamins.
Dolberg, Florian. 2018. Looking further afield: An interdisciplinary approach to
language contact and grammatical change. Paper presented at the Twentieth
International Conference on English Historical Linguistics (ICEHL 20), Edinburgh,
27–31 August 2018. Powerpoint available at http://www.conferences.cahss.ed.ac.uk/
icehl20/handouts-and-powerpoints/ (accessed on 19 Sept 2018).
Dolberg, Florian. forthcoming. HITting on migration in the murky Middle Ages –
advocating an interdisciplinary approach. In: Thomson, Simon C. (ed.), Strangers
at the Gate! The (Un)welcome Movement of People and Ideas in the Medieval World.
(Explorations in Medieval Culture.) Leiden: Brill.
D’omina et al. 2012: Дьомина, Евгения и съавт. 2012. Речник на книжовния български
език на народна основа от XVII век (върху текст на Тихонравовия дамаскин)
[A Dictionary of Written Bulgarian Based on the 17th-century Vernacular (Based on
the Text of the Tixonravov Damascene Book)]. София: Издателство “Валентин
Траянов”.
Dresher, B. Elan and William J. Idsardi. 1999. Prerequisites for a theory of diachronic
adaptation. Zeitschrift für Sprachwissenschaft 18(2): 212–215.
Drinka, Bridget. 2003. Areal factors in the development of the European periphrastic
perfect. Word 54: 1–38.
Drinka, Bridget. 2017. Language Contact in Europe: The Periphrastic Perfect through
History. Cambridge: Cambridge University Press.
Dujčev and Lixačev 1988: Дуйчев, И. С. и Д. С. Лихачев (ред.). 1988. Среднеболгарский
перевод Хроники Константина Манассии в славянских литературах [The
340 References
Middle Bulgarian Translation of the Chronicle of Constantine Manasses in the

Slavonic Literatures]. София: БАН.
Duridanov et al. 1991: Дуриданов, Иван, Екатерина Дограмаджиева и Ангелина
Минчева. 1991. Граматика на старобългарския език [Old Bulgarian Grammar].
София: Издателство на Българската академия на науките.
Durie, Mark. 1995. Towards an understanding of linguistic evolution and the notion
‘X has a function Y’. In: Werner, Abraham, Talmy Givón and Sandra A. Thompson
(eds.), Discourse Grammar and Typology: Papers in Honor of John W. M. Verhaar.
(Studies in Language Companion Series 27), 275–308. Amsterdam: John Benjamins.
Eisenberg, Peter. 1994. German. In: König, Ekkehard and Johan van der Auwera (eds.),
The Germanic Languages, 349–387. London: Routledge.
Eisenberg, Peter. 2004. Grundriß der deutschen Grammatik. Band 2: Der Satz. Zweite
Auflage. Stuttgart, Weimar: Verlag J. B. Metzler.
Elliott, Elisabeth M. 2001. The sum (‘be’) and imam (‘have’) resultative constructions in
Macedonian and Bulgarian within a typology of resultative constructions in Slavic.
Doctoral thesis. University of Toronto.
Elsness, Johan. 1997. The Perfect and the Preterite in Contemporary and Earlier English.
Evans, G. Blakemore and J. J. M. Tobin (eds.) et al. 1997. The Riverside Shakespeare. 2nd
ed. Boston: Houghton Mifflin Company.
Fielder, Grace E. 1995. Narrative Perspective and the Bulgarian L-Participle. The Slavic
and East European Journal 39: 585–600.
Fischer, Olga. 1991. The rise of the passive infinitive in English. In: Kastovsky, Dieter
(ed.), Historical English Syntax, 141–188. Berlin: Mouton de Gruyter.
Fischer, Olga. 1992. Syntax. In: Blake, Norman (ed.), The Cambridge History of the English
Language. Vol. II: 1066–1476, 207–408. Cambridge: Cambridge University Press.
Fischer, Olga. 2015. The influence of the grammatical system and analogy in processes
of language change: The case of the auxiliation of HAVE-to once again. In: Toupin,
Fabienne and Brian Lowrey (eds.), Studies in Linguistic Variation and Change:
From Old to Middle English, 120–150. Newcastle upon Tyne: Cambridge Scholars
Publishing.
Fischer, Olga and Wim van der Wurff. 2006. Syntax. In: Hogg, Richard and David
Denison (eds.), A History of the English Language, 109–198. Cambridge: Cambridge
University Press.
Fløgstad, Guro Nore. 2016. Preterit Expansion and Perfect Demise in Porteño Spanish
and Beyond: A Critical Perspective on Cognitive Grammaticalization Theory. (Brill’s
Studies in Historical Linguistics 6.) Leiden/Boston: Brill.
Fortuny, Jordi and Bernat Corominas-Murtra. 2013. On the origin of ambiguity in
efficient communication. Journal of Logic, Language, and Information 22(3):
249–267.
References 341
Frajzyngier, Zygmunt. 2005. The principle of Functional Transparency in language

structure and in language evolution. In: Frajzyngier, Zygmunt, Adam Hodges
and David S. Rood (eds.), Linguistic Diversity and Language Theories, 259–283.
Amsterdam/Philadelphia: John Benjamins.
Fridén, Georg. 1948. Studies on the Tenses of the English Verb from Chaucer to Shakespeare
with Special Reference to the Late Sixteenth Century. Uppsala: A.-B. Lundeqvistska
Bokhandeln.
Fridén, Georg. 1957. On the use of auxiliaries to form the perfect and the pluperfect in
late Middle English and early Modern English. Studia Linguistica 11: 54–56.
Friedman, Victor A. 1976. Dialectal synchrony and diachronic syntax. The Macedonian
perfect. Papers from the Parasession on Diachronic Syntax. Chicago Linguistics
Society, 96–104.
Friedman, Victor A. 1993. Macedonian. In: Comrie, Bernard and Greville Corbett (eds.),
The Slavonic Languages, 249–305. London: Routledge.
Fryd, Marc and Pierre-Don Giancarli (eds.). 2017. Aorists and Perfects: Synchronic and
Diachronic Perspectives. Leiden: Brill/Rodopi.
Gallis, Arne. 1960. Die neuen slavischen Perfekte vom Typus factum habeo und *casus
sum, casus habeo. Scando-Slavica 6: 176–188.
Ganeva, Gergana. 2015: Ганева, Гергана. 2015. Темпоралните основи на
старобългарския глагол [The temporal stems of Old Bulgarian verbs].
Palaeobulgarica XXXIX(4): 3–16.
van Gelderen, Elly. 2014. A History of the English Language. Revised edition. Amsterdam:
John Benjamins.
van Gelderen, Elly. 2018. Analyzing Syntax through Texts: Old, Middle, and Early Modern
English. Edinburgh: Edinburgh University Press.
Georgiev, Vladimir. 1957: Георгиев, Владимир. 1957. Възникване на нови сложни
глаголни форми със спомагателен глагол “имам” [Development of new peri-
phrastic forms with auxiliary ‘have’]. Известия на Института за български
език V: 31–59.
Georgiev, Vladimir. 1976[1957]: Георгиев, Владимир. 1976. Възникване на нови
сложни глаголни форми със спомагателен глагол “имам” [Development of new
periphrastic forms with auxiliary ‘have’]. В: Пашов, Петър и Руселина Ницолова
(съставители), Помагало по българска морфология. Глагол, 294–313. София:
Издателство Наука и изкуство.
Georgieva et al. 1989: Георгиева, Елена, Стоян Жерев и Валентин Станков (ред.).
1989. История на новобългарския книжовен език [History of Modern Standard
Bulgarian]. София: БАН.
Gerdžikov, Georgi. 1984/2003: Герджиков, Георги. 1984/2003. Преизказването на
глаголното действие в българския език [Renarration of Verbs in Bulgarian].
342 References
Второ фототипно издание. София: Университетско издателство “Св. Климент

Охридски”.
Gešev, Valentin. 2009: Гешев, Валентин. 2009. Книжовноезикови стиловe на
българския XVIII в. [Literary language styles of 18th-century Bulgarian].
Съпоставително езикознание XXXIII(1): 23–53.
Gillmann, Melitta. 2016. Perfektkonstruktionen mit haben und sein. Eine
Korpusuntersuchung im Althochdeutschen, Altsächsischen und Neuhochdeutschen.
(Studia Linguistica Germanica 128.) Berlin: De Gruyter Mouton.
Givón, Talmy. 1971. Historical syntax and synchronic morphology: An archeologist’s
field trip. Papers from the Chicago Linguistic Society 7: 394–415.
Givón, Talmy. 2013. On the intellectual roots of functionalism in linguistics. In: Bischoff,
Shannon T. and Carmen Jany (eds.), Functional Approaches to Language, 9–29.
Berlin: De Gruyter Mouton.
Gorlach, Marina. 2005. Resultativeness in English: A sign-oriented approach. In:
Frajzyngier, Zygmunt, Adam Hodges and David S. Rood (eds.), Linguistic Diversity
and Language Theories, 363–377. Amsterdam/Philadelphia: John Benjamins.
Gräf, Adolf. 1888. Das Perfectum bei Chaucer. Eine syntactische Untersuchung.
Frankenhausen: Krebs.
Griffith, Tom (ed.). 1997. The Concise Pepys. London: Wordsworth Classics of World
Literature.
Grković-Major, Jasmina. 2011. The development of predicative possession in Slavic
languages. In: Nomachi, Motoki (ed.), The Grammar of Possessivity in South Slavic
Languages: Synchronic and Diachronic Perspectives, 35–54. Sapporo.
Haas, Walter. 1957. Zero in linguistic description. In: Studies in Linguistic Analysis.
Special Volume of the Philological Society, 33–54. Oxford: Basil Blackwell.
Halle, Morris. 1975. Confessio grammatici. Language 51: 525–535.
Harris, John. 1984. Syntactic variation and dialect divergence. Journal of Linguistics 20:
303–327.
Haspelmath, Martin. 1995. The converb as a cross-linguistically valid category. In:
Haspelmath, Martin and Ekkehard König (eds.), Converbs in Cross-Linguistic
Perspective. Structure and Meaning of Adverbial Verb Forms – Adverbial Participles,
Gerunds, 1–55. Berlin/New York: Mouton de Gruyter.
Haspelmath, Martin. 1998. Does grammaticalization need reanalysis? Studies in
Language 22(2): 315–351.
Haspelmath, Martin. 1999. Optimality and diachronic adaptation. Zeitschrift für
Sprachwissenschaft 18(2): 180–205.
Haspelmath, Martin. 2007. Pre-established categories don’t exist: Consequences for
language description and typology. Linguistic Typology 11(1): 119–132.
References 343
Haspelmath, Martin. 2008a. Parametric versus functional explanations of syntactic

universals. In: Biberauer, Theresa (ed.), The Limits of Syntactic Variation, 75–107.
Haspelmath, Martin. 2008b. Creating economical morphosyntactic patterns in lan-
guage change. In: Good, Jeff (ed.), Language Universals and Language Change, 185–
214. Oxford: Oxford University Press.
Haspelmath, Martin. 2018. The moving parts and fixed parts of our theories: Why
functional-adaptive explanations are more testable. Draft. Source: https://www
.academia.edu/37183389/The_moving_parts_and_fixed_parts_of_our_theories_
Why_functional-adaptive_explanations_are_more_testable (accessed on 06 August
2018).
Heine, Bernd. 1993. Auxiliaries: Cognitive Forces and Grammaticalization. Oxford:
Oxford University Press.
Heine, Bernd. 1997. Possession: Cognitive Sources, Forces, and Grammaticalization.
(Cambridge Studies in Linguistics 83.) Cambridge: Cambridge University Press.
Heine, Bernd. 2003. Grammaticalization. In: Joseph, Brian D. and Richard D. Janda
(eds.), The Handbook of Historical Linguistics, 575–601. Oxford: Blackwell.
Heine, Bernd and Tania Kuteva. 2005. Language Contact and Grammatical Change.
(Cambridge Approaches to Language Contact.) Cambridge: Cambridge University
Press.
Helbig, Gerhard. 1990. Bemerkungen zum Zustandspassiv (Ein Diskussionsbeitrag).
In: Petkov, Pawel and Diana Sliwkowa (eds.), Ausgewählte Theoretische Literatur zur
Morphologie der Deutschen Sprache, 477–494. Sofia: Sofioter Universität “Kliment
Ochridski”.
Helbig, Gerhard and Joachim Buscha. 1991. Deutsche Grammatik. Ein Handbuch für den
Ausländerunterricht. Leipzig: Langenscheidt.
Hengeveld, Kees, Heiko Narrog and Hella Olbertz (eds.). 2017. The Grammaticalization
of Tense, Aspect, Modality and Evidentiality: A Functional Perspective. Berlin: Mouton
de Gruyter.
Hill, Jane H. 1993. Formalism, functionalism and the discourse of evolution. In: Foley,
William A. (ed.), The Role of Theory in Language Description, 437–455. Berlin:
Mouton de Gruyter.
Hoffmann, Sebastian. 2005. Grammaticalization and English Complex Prepositions: A
Corpus-Based Study. London: Routledge.
Hoffmann, Thomas and Graeme Trousdale (eds.). 2013. The Oxford Handbook of
Construction Grammar. Oxford: Oxford University Press.
Hollmann, Willem B. 2003. Synchrony and diachrony of English periphrastic caus-
atives: A cognitive perspective. PhD dissertation. University of Manchester.
344 References
Hopper, Paul J. 1991. On some principles of grammaticization. In: Traugott, Elizabeth

Closs and Bernd Heine (eds.), Approaches to Grammaticalization: Focus on
Theoretical and Methodological Issues. Vol. 1. (Typological Studies in Language 19(1)),
17–35. Amsterdam-Philadelphia: John Benjamins.
Hopper, Paul J. and Elizabeth Closs Traugott. 2003. Grammaticalization. 2nd ed.
(Cambridge Textbooks in Linguistics.) Cambridge: Cambridge University Press.
Hristov, Bozhil. 2015. The atoms of language. Review article on Kibort, Anna and
Greville G. Corbett (eds.), Features: Perspectives on a Key Notion in Linguistics
(Oxford Linguistics.) Oxford: Oxford University Press, 2010. Pp. xi + 333. Journal of
Linguistics 51: 644–676.
Huber, Judith. 2019. Counterfactuality and aktionsart: Predictors for BE vs. HAVE +
past participle in Middle English. In: Claridge, Claudia and Birte Bös (eds.),
Developments in English Historical Morpho-Syntax. (Current Issues in Linguistic
Theory 346), 149–173. Amsterdam: John Benjamins.
Huddleston, Rodney and Geoffrey K. Pullum et al. 2002. The Cambridge Grammar of
the English Language. Cambridge: Cambridge University Press.
Huntley, David. 1993. Old Church Slavonic. In: Comrie, Bernard and Greville Corbett
(eds.), The Slavonic Languages, 125–187. London: Routledge.
Irvine, Susan. 2004. The Anglo-Saxon Chronicle: A Collaborative Edition. Volume 7. MS.
E. Cambridge: D. S. Brewer.
Itkonen, Esa. 2013. Functional explanation and its uses. In: Bischoff, Shannon T. and
Carmen Jany (eds.), Functional Approaches to Language, 31–69. Berlin: De Gruyter
Mouton.
Ivanova, Ana. 1967: Иванова, Ана. 1967. Троянски дамаскин. Български паметник
от XVII век [Trojanski Damascene. A Bulgarian Document from the 17th Century].
София: БАН.
Izvorski, Roumyana. 1997. The present perfect as an epistemic modal. In: Lawson, Aaron
and E. Cho (eds.), SALT VII Proceedings, 222–239. Ithaca, NY: CLC Publications,
Cornell University.
Jacobs, Andreas and Andreas H. Jucker. 1995. The historical perspective in pragmatics.
In: Jucker, Andreas H. (ed.), Historical Pragmatics: Pragmatic Developments in the
History of English, 3–36. Amsterdam/Philadelphia: John Benjamins.
Jespersen, Otto. 1909–1949. A Modern English Grammar on Historical Principles.
7 vols. Heidelberg: Carl Winters Universitätsbuchhandlung/Copenhagen: Ejnar
Munksgard/London: George Allen & Unwin.
Jespersen, Otto. 1931. A Modern English Grammar on Historical Principles. Part IV.
Syntax. Third Volume. Time and Tense. Heidelberg: Carl Winters Universitätsbuch-
handlung.
Johannisson, Ture. 1958. On the be and have constructions with mutative verbs. Studia
Linguistica 12: 106–118.
References 345
Johannsen, Berit. 2016. From possessive-resultative to perfect? Re-assessing the mean-

ing of [hæbb- + past participle] constructions in Old English prose. In Werner,
Valentin, Elena Seoane and Cristina Suárez-Gómez (eds.), Re-Assessing the Present
Perfect: Corpus Studies and Beyond, 23–42. Berlin: De Gruyter Mouton.
Jucker, Andreas H. and Irma Taavitsainen. 2013. English Historical Pragmatics.
Edinburgh: Edinburgh University Press.
Kakietek, Piotr. 1976. The perfect auxiliaries in the language of Shakespeare. Studia
Anglica Posnaniensia 8: 45–53.
Keller, Rudi. 1994. On Language Change: The Invisible Hand in Language. London and
New York: Routledge.
Kilpiö, Matti. 1997. On the forms and functions of the verb to be from Old to Modern
English. In: Rissanen, Matti, Merja Kytö and Kirsi Heikkonen (eds.), English in
Transition. Corpus-based Studies in Linguistic Variation and Genre Styles, 87–120.
Kilpiö, Matti. 2007. Auxiliation in progress: Diachronic grammaticalisation changes in
Old English and Early Middle English HAVE perfects. In: Rissanen, Matti, Marianna
Hintikka, Leena Kahlas-Tarkka and Rod McConchie (eds.), Change in Meaning and
the Meaning of Change: Studies in Semantics and Grammar from Old to Present-Day
English. (Mémoires de la Société Néophilologique de Helsinki 72), 323–343.
Vaajakoski: Gummerus Kirjapaino Oy.
Kilpiö, Matti. 2013. Causative habban in Old English: Tracing the development of a
budding construction. In: Diewald, Gabriele, Leena Kahlas-Tarkka and Ilse Wischer
(eds.), Comparative Studies in Early Germanic Languages: With a Focus on Verbal
Categories, 101–125. Amsterdam: John Benjamins.
Kostadinova, Petja. 2006: Костадинова, Петя. 2006. Имам + минало страдателно
причастие. Едно семантично ограничение [Have + past passive partici-
ple. A semantic restriction]. В: Куцаров, Иван, Васил Райнов, Мери Лакова,
Владко Мурдаров и Светла Коева (ред.), Езиковедски изследвания в чест на
75-годишнината от рождението на ст.н.с. І ст. д.ф.н. Йордан Пенчев Пенчев,
153–156. София: Артграф.
Kostadinova, Petja. 2009: Костадинова, Петя. 2009. Към въпроса за резултативността
в българската темпорална система [On the question of resultativity in the
Bulgarian temporal system]. В: Станков, Валентин и Калина Викторова (ред.),
Проблеми на граматичната система на българския език – глагол. (Българско
езикознание 5), 487–501. София: Академично издателство “Проф. Марин
Дринов”.
Kostov, Kiril. 1982[1972]: Костов, Кирил. 1982. Семантични наблюдения върху
съчетанието имам + минало страдателно причастие в българския език
[Semantic observations on the combination have + past passive participle in
Bulgarian]. Във: Валтер, Хилмар, Карл Гутшмит и Светомир Иванчев (съст.),
346 References
Езиковедската българистика в ГДР, 157–167. София: Наука и изкуство. Original:

Kostov, Kiril. 1972. Semantische Beobachtung über die Verbindung von imam mit
dem Participium Perfekti Passivi im Bulgariaschen. Zeitschrift für Slawistik 17(3):
371–379.
Kostov, Kiril. 2004: Костов, Кирил. 2004. Мисли за структурните ограничения
в употребата на синтактичното съчетание имам + минало страдателно
причастие от преходен глагол [Remarks on the structural restrictions on using
the syntactic construction have + transitive past passive participle]. Български език
LI(2–3): 124–129.
Kovatcheva, Mira. 1982: Ковачева, Мира. 1982. Безлични конструкции в
средноанглийски и техните еквиваленти в съвременния английски език
[Impersonal constructions in Middle English and their Modern English equiva-
lents]. Автореферат на дисертация за присъждане на научната степен
“Кандидат на филологическите науки”. СУ “Климент Охридски”.
Kovatcheva, Mira. 1989. Typological significance of the grammaticalization of the
object in the history of English. Habilitation monograph. University of Sofia.
Kovatcheva, Mira and Mitko Sabev. 2015. The History of English in Texts and Tables.
2nd ed. Sofia: University of Sofia Press.
Krug, Manfred G. 2001. Frequency, iconicity, categorization: Evidence from emerging
modals. In: Bybee, Joan L. and Paul J. Hopper (eds.), Frequency and the Emergence of
Linguistic Structure, 309–335. Amsterdam/Philadelphia: John Benjamins.
Kucarov, Ivan. 2007: Куцаров, Иван. 2007. Теоретична граматика на българския
език. Морфология [Bulgarian Theoretical Grammar. Morphology]. Пловдив:
Университетско издателство “Паисий Хилендарски”.
Kuteva, Tania. 1998. Large linguistic areas in grammaticalization: Auxiliation in Europe.
Language Sciences 20(3): 289–311.
Kuteva, Tania and Bernd Heine. 2004. On the possessive perfect in North Russian. Word
55(1): 37–71.
Kytö, Merja. 1994. BE vs. HAVE with intransitives in Early Modern English. In:
Fernández, Francisco, Miguel Fuster and Juan José Calvo (eds.), English Historical
Linguistics 1992. Papers from the 7th International Conference on English Historical
Linguistics, 179–190. Amsterdam/Philadelphia: John Benjamins.
Kytö, Merja. 1997. Be/have + past participle: The choice of the auxiliary with intransi-
tives from Late Middle to Modern English. In: Rissanen, Matti, Merja Kytö and Kirsi
Heikkonen (eds.), English in Transition. Corpus-based Studies in Linguistic Variation
and Genre Styles, 17–85. Berlin: Mouton de Gruyter.
Labov, William. 1987. The overestimation of functionalism. In: Dirven, R. and
V. Fried (eds.), Functionalism in Linguistics, 311–322. Amsterdam/Philadelphia: John
Benjamins.
References 347
Laing, Margaret. 2010. The reflexes of OE beon as a marker of futurity in early Middle
English. In: Lenker, Ursula, Judith Huber and Robert Mailhammer (eds.), English
Historical Linguistics 2008: Selected Papers from the Fifteenth International Conference
on English Historical Linguistics (ICEHL 15), Munich, 24–30 August 2008. Volume I:
The History of English Verbal and Nominal Constructions, 237–254. Amsterdam/
Philadelphia: John Benjamins.
Lass, Roger. 1980. On Explaining Language Change. (Cambridge Studies in Linguistics
27.) Cambridge: Cambridge University Press.
Lass, Roger. 1994. Old English. A historical linguistic companion. Cambridge: Cambridge
University Press.
Łęcki, Andrzej M. 2010. Grammaticalisation Paths of Have in English. (Studies in English
Medieval Language and Literature 24.) Frankfurt am Main: Peter Lang.
Lee, Jeong-Hoon. 2003. The “have” perfect in Old English: How close was it to the
Modern English Perfect? In: Minkova, Donka and Robert Stockwell (eds.), Studies
in the History of the English Language: A Millennial Perspective, 373–398. Berlin, New
York: Mouton de Gruyter.
Lehmann, Christian. 1985. Grammaticalization: Synchronic variation and diachronic
change. Lingua e stile 20: 303–318.
Levinson, Stephen C. 2000. Presumptive Meanings: The Theory of Generalized
Conversational Implicature. Cambridge, Massachusetts: MIT Press.
Lightfoot, David W. 2002. Review of William Croft, Explaining Language Change: An
Evolutionary Approach, 2000. Journal of Linguistics 38(2): 410–414.
Los, Bettelou. 2015. A Historical Syntax of English. Edinburgh: Edinburgh University
Press.
Lutz, Angelika. 2012. Language contact in the Scandinavian period. In: Nevalainen,
Terttu and Elizabeth Closs Traugott (eds.), The Oxford Handbook of the History of
English, 508–517. Oxford: Oxford University Press.
Macleod, Morgan D. 2012. The perfect in Old English and Old Saxon: The synchronic
and diachronic correspondence of form and meaning. Doctoral thesis. University
of Cambridge.
MacRobert, Catherine Mary. 1981. The decline of the infinitive in Bulgarian. DPhil the-
sis. University of Oxford.
MacRobert, Catherine Mary. 2013. The competing use of perfect and aorist tenses in
Old Church Slavonic. Slavia 82(4): 387–407.
Mathieu, Eric and Robert Truswell (eds.). 2017. Micro-change and Macro-change in
Diachronic Syntax. Oxford: Oxford University Press.
McFadden, Thomas and Artemis Alexiadou. 2006. Pieces of the be perfect in German
and Older English. In: Baumer, Donald, David Montero and Michael Scanlon
(eds.), Proceedings of the 25th West Coast Conference on Formal Linguistics, 270–278.
Somerville, MA: Cascadilla Proceedings Project.
348 References
McFadden, Thomas and Artemis Alexiadou. 2010. Perfects, resultatives, and auxiliaries
in earlier English. Linguistic Inquiry 41(3): 389–425.
McMahon, April M. 1994. Understanding Language Change. Cambridge: Cambridge
University Press.
McWhorter, John H. 2002. What happened to English? Diachronica 19: 217–272.
Meillet, Antoine. 1912. L’évolution des formes grammaticales. Scientia (Rivista di
Scienza) 12(26): 6. (Reprinted 1951, Linguistique historique et linguistique generale,
130–148. Paris: C. Klincksieck.)
Menn, Lise, Cecily Jill Duffield and Bhuvana Narasimhan. 2013. Towards an experimen-
tal functional linguistics: Production. In: Bischoff, Shannon T. and Carmen Jany
(eds.), Functional Approaches to Language, 207–245. Berlin: De Gruyter Mouton.
Michaelis, Laura A. 1994. The ambiguity of the English present perfect. Journal of
Linguistics 30: 111–157.
Miller, Jim. 2000. The perfect in spoken and written English. Transactions of the
Philological Society 98(2): 323–352.
Miller, Jim. 2003. Syntax and discourse in Modern Scots. In: Corbett, John, J. Derrick
McClure and Jane Stuart-Smith (eds.), The Edinburgh Companion to Scots, 72–109.
Edinburgh: Edinburgh University Press.
Miller, Jim. 2004a. Perfect and resultative constructions in spoken and non-standard
English. In: Fischer, O., M. Norde and H. Perridon (eds.), Up and Down the Cline: The
Nature of Grammaticalization, 229–246. Amsterdam: John Benjamins.
Miller, Jim. 2004b. Problems for typology: Perfects and resultatives in spoken and
non-standard English and Russian. In: Kortmann, Bernd (ed.), Dialectology Meets
Typology: Dialect Grammar from a Cross-Linguistic Perspective, 305–334. Berlin:
Mouton de Gruyter.
Minčeva, Angelina. 1987: Минчева, Ангелина. 1987. Старобългарският език в
светлината на балканистиката [Old Bulgarian in the Light of Balkan Studies].
София: Наука и изкуство.
Mincoff, Marco. 1958. An English Grammar. Second revised edition. Sofia: State
Publishing House “Science and Art”.
Mirčev, Kiril. 1976[1973]: Мирчев, Кирил. 1976. За съчетанията на глагол имам +
минало страдателно причастие в български език [On combinations of have
+ past passive participle in Bulgarian]. В: Пашов, Петър и Руселина Ницолова
(съставители), Помагало по българска морфология. Глагол, 312–315. София:
Издателство Наука и изкуство.
Mirčev, Kiril. 2000: Мирчев, Кирил. 2000. Старобългарски език. Кратък граматичен
очерк [Old Bulgarian. A Short Grammatical Outline]. Велико Търново: Фабер.
Mitchell, Bruce. 1985. Old English Syntax. 2 vols. Oxford: Oxford University Press.
References 349
Mladenov, Stefan. 1979[1929]: Младенов, Стефан. 1979. История на българския

език [History of the Bulgarian Language]. София: Издателство на Българската
академия на науките.
Mugglestone, Lynda. 2012. English in the Nineteenth Century. In: Mugglestone, Lynda
(ed.), The Oxford History of English. 2nd ed., 340–378. Oxford: Oxford University
Press.
Mustanoja, Tauno F. 1960. A Middle English Syntax. I, Parts of Speech. Helsinki: Société
Néophilologique.
Načeva, Mira. 1988: Начева, Мира. 1988. Семантично-функционална характеристика
на глагола имам ‘habere’ (върху материал от българския и чешкия език)
[Semantic-functional description of the verb ‘habere’ (based on material from
Bulgarian and Czech)]. Съпоставително езикознание XIII(4–5): 59–68.
Nakamura, Fujio. 1987. Auxiliary have in Samuel Pepys’s Diary, with special reference to
mutative verbs: (1) A statistiacal study. Studies in Modern English 3: 23–51.
Newmeyer, Frederick J. 1991. Functional explanation in linguistics and the origins of
language. Language and Communication 11: 3–28.
Nicolova, Ruselina. 2008: Ницолова, Руселина. 2008. Българска граматика.
Морфология [Bulgarian Grammar. Morphology]. София: Университетско
издателство “Св. Климент Охридски”.
Nicolova, Ruselina. 2013: Ницолова, Руселина. 2013. Перфект и связанная с ним
грамматическая категория эвиденциальности в болгарском языке [The
perfect and the related grammatical category of evidentiality in Bulgarian].
Съпоставително езикознание XXXVIII(2–3): 50–67.
Nicolova, Ruselina. 2017. Bulgarian Grammar. Translated by Christo Stamenov. Berlin:
Frank & Timme.
Norde, Muriel and Freek Van de Velde (eds.). 2016. Exaptation and Language Change.
Peev et al. 2015: Пеев, Димитър, Маргарет Димитрова, Петко Петков. 2015. Паисий
Хилендарски. История славянобългарска. Критическо издание с превод и
коментар [Paisius of Hilendar. Slavonic-Bulgarian History. Critical edition with
translation and commentary]. Трето поправено издание. Света гора Атон:
Военно издателство.
Penčev, Jordan. 1968: Пенчев, Йордан. 1968. Конструкции с глагола имам
[Constructions with the verb have]. В: Славистичен сборник, 173–177. София:
Издателство на БАН.
Pennisi, Antonino and Alessandra Falzone. 2016. Darwinian Biolinguistics: Theory and
History of a Naturalistic Philosophy of Language and Pragmatics. Springer.
Petré, Peter. 2010. The functions of weorðan and its loss in the past tense in Old and
Middle English. English Language and Linguistics 14(3): 457–484.
350 References
Petré, Peter. 2013. Passive auxiliaries in English and German: Decline versus grammati-
calisation of bounded language use. In: Diewald, Gabriele, Leena Kahlas-Tarkka
and Ilse Wischer (eds.), Comparative Studies in Early Germanic Languages: With a
Focus on Verbal Categories, 71–100. Amsterdam: John Benjamins.
Petré, Peter. 2014. Constructions and Environments: Copular, Passive, and Related
Constructions in Old and Middle English. Oxford: Oxford University Press.
Petré, Peter and Hubert Cuyckens. 2008. Bedusted, yet not beheaded: The role of be-’s
constructional properties in its conservation. In: Bergs, Alexander and Gabriele
Diewald (eds.), Constructions and Language Change, 133–169. Berlin: Mouton de
Gruyter.
Piantadosi, Steven T., Harry Tily and Edward Gibson. 2012. The communicative func-
tion of ambiguity in language. Cognition 122(3): 280–291.
Pinker, Steven and Paul Bloom. 1990. Natural language and natural selection. Behavioral
and Brain Sciences 13: 707–784.
Pinto de Lima, José. 1995. Pragmatic maxims in explanations of language change?
In: Jucker, Andreas H. (ed.), Historical Pragmatics: Pragmatic Developments in the
History of English, 197–215. Amsterdam/Philadelphia: John Benjamins.
Plummer, Charles and John Earle (eds.). 1892. Two of the Saxon Chronicles Parallel
with Supplementary Extracts from the Others. A Revised Text. Vol. I. Oxford: Oxford
University Press.
Popova, Genoveva. 1931: Попова, Геновева. 1931. Употреба на минало страдателно
причастие с глагола “имам” в българския език [The use of a past passive par-
ticiple with the verb ‘have’ in Bulgarian]. Известия на семинара по славянска
филология VII: 490–501.
Prokosch, Eduard. 2009[1939]. A Comparative Germanic Grammar. Richmond: Tiger
Xenophon.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik. 1985. A
Comprehensive Grammar of the English Language. Harlow: Longman.
Quirk, Randolph and C. L. Wrenn. 1957. An Old English Grammar. 2nd edn. London and
New York: Methuen.
Radeva, Nija. 2013: Радева, Ния. 2013. От езика на “История славянобългарска” към
езика на нейните преписи. Част 1 [From the Language of the ‘Slavonic-Bulgarian
History’ to the Language of its Copies. Part 1]. Велико Търново: УИ “Св. св. Кирил
и Методий”.
Rainer, Eva Maria. 1989. Das Perfect im Spätmittel- und Frühneuenglischen: eine
Frequenz- und Funktionsanalyse anhand von Brieftexten. Innsbruck: Institut für
Anglistik, University of Innsbruck.
Rikov, Georgi T. 1993/1994: Риков, Георги Т. 1993/1994. Следи от индоевропейските
перфектни формации в славянските езици [Traces of Indo-European perfect
formations in the Slavonic languages]. Български език XLIII/XLIV(1): 1–8.
References 351
Ringe, Don. 2006. From Proto-Indo-European to Proto-Germanic. A Linguistic History of

English. Volume I. Oxford: Oxford University Press.
Ringe, Don and Joseph F. Eska. 2013. Historical Linguistics. Toward a Twenty-First
Century Reintegration. Cambridge: Cambridge University Press.
Rissanen, Matti. 1999. Syntax. In: Lass, Roger (ed.), The Cambridge History of the English
Language. Vol. III: 1476–1776, 187–332. Cambridge: Cambridge University Press.
Ruseva, Dimitrina. 2011: Русева, Димитрина. 2011. Електронно издание с речник
на Троянската притча [An electronic edition with glossary of the Tale of Troy].
Дипломна работа. СУ, Катедра “Български език”.
Rydén, Mats. 1991. The be/have variation with intransitives in its crucial phases. In:
Kastovsky, Dieter (ed.), Historical English Syntax. (Topics in English Linguistics 2),
343–354. Berlin and New York: Mouton de Gruyter.
Rydén, Mats and Sverker Brorström. 1987. The Be/Have Variation with Intransitives
in English with Special Reference to the Late Modern Period. (Stockholm Studies in
English 70.) Stockholm: Almqvist & Wiksell.
Sag, Ivan A. 2010. Feature geometry and predictions of locality. In: Kibort, Anna and
Greville G. Corbett (eds.), Features: Perspectives on a Key Notion in Linguistics, 236–
271. Oxford: Oxford University Press.
Salgādo, Gāmini (ed.). 1968a. Three Restoration Comedies. London: Penguin.
Salgādo, Gāmini. 1968b. Introduction. In: Salgādo, Gāmini (ed.), Three Restoration
Comedies, 11–40. London: Penguin.
Sampson, Geoffrey, David Gil and Peter Trudgill (eds.). 2009. Language Complexity as
an Evolving Variable. Oxford: Oxford University Press.
Sanders, Andrew. 2004. The Short Oxford History of English Literature. 3rd ed. Oxford:
Oxford University Press.
Scatton, Ernest A. 1993. Bulgarian. In: Comrie, Bernard and Greville Corbett (eds.), The
Slavonic Languages, 188–248. London: Routledge.
Schenker, Alexander M. 1993. Proto-Slavonic. In: Comrie, Bernard and Greville Corbett
(eds.), The Slavonic Languages, 60–121. London: Routledge.
Schwarz, Sarah. 2019. Signs of grammaticalization: Tracking the get-passive through
COHA. In: Claridge, Claudia and Birte Bös (eds.), Developments in English Historical
Morpho-Syntax. (Current Issues in Linguistic Theory 346), 199–221. Amsterdam:
John Benjamins.
Shannon, Thomas F. 1989. Review of Mats Rydén and Sverker Brorström, The Be/Have
Variation with Intransitives in English with Special Reference to the Late Modern
Period, 1987. Language 65: 676–678.
Shannon, Thomas F. 1995. Toward a cognitive explanation of perfect auxiliary variation:
Some modal and aspectual effects in the history of Germanic. American Journal of
Germanic Linguistics & Literatures 7(2): 129–163.
352 References
Sim, Stuart. 1997. Introduction. In: Griffith, Tom (ed.), The Concise Pepys, v–xi. London:
Wordsworth Classics of World Literature.
Smirnickaja et al. 1977: Смирницкая, О. А., Мирра Моисеевна Гухман, Л. С.
Ермолаева. 1977. Историко-типологическая морфология германских языков:
Категория глагола [Historical-Typological Morphology of the Germanic Languages:
Verbal Categories]. Москва: Институт языкознания (Академия наук СССР).
Smith, K. Aaron. 2001. The role of frequency in the specialization of the English an-
terior. In: Bybee, Joan L. and Paul J. Hopper (eds.), Frequency and the Emergence of
Linguistic Structure, 361–382. Amsterdam/Philadelphia: John Benjamins.
Stambolieva, Maria. 2008. Building Up Aspect: A Study of Aspect and Related Categories
in Bulgarian, with Parallels in English and French. Frankfurt am Main: Peter Lang.
Stankov, V. 1980: Станков, В. 1980. За характера на опозицията преизказност в
българския език [On the nature of the evidentiality opposition in Bulgarian]. В
чест на академик Владимир Георгиев: Езиковедски проучвания, 485–491. София:
БАН.
Stoevsky, Andrei. 1999. Remarks on the status of the English perfect. In: Mantcheva,
Alexandra (ed.), Mélanges de linguistique, sémiotique et narratologie: dédiés à la
mémoire de Krassimir Mantchev à l’occasion de son 60e anniversaire, 375–380. Sofia:
Colibri.
Stoevsky, Andrei. 2006. The collocation “имам ‘have’ + the past passive participle” in
Bulgarian. В: Куцаров, Иван, Васил Райнов, Мери Лакова, Владко Мурдаров
и Светла Коева (ред.), Езиковедски изследвания в чест на 75-годишнината
от рождението на ст.н.с. І ст. д.ф.н. Йордан Пенчев Пенчев, 142–152. София:
Артграф.
Straaijer, Robin. 2010. Prescription or practice? Be/have variation with past participles
of mutative intransitive verbs in the letters of Joseph Priestley. In: Lenker, Ursula,
Judith Huber and Robert Mailhammer (eds.), English Historical Linguistics 2008:
Selected Papers from the Fifteenth International Conference on English Historical
Linguistics (ICEHL 15), Munich, 24–30 August 2008. Volume I: The History of English
Verbal and Nominal Constructions, 63–78. Amsterdam: John Benjamins.
Strang, Barbara M. H. 1970. A History of English. London: Methuen.
Szmrecsanyi, Benedikt. 2016. About text frequencies in historical linguistics:
Disentangling environmental and grammatical change. Corpus Linguistics and
Linguistic Theory 12(1): 153–171.
Tagliamonte, Sali A. 2000. The grammaticalization of the present perfect in English:
Tracks of change and continuity in a linguistic enclave. In: Fischer, Olga, Anette
Rosenbach and Dieter Stein (eds.), Pathways of Change: Grammaticalization in
English, 329–354. Amsterdam: John Benjamins.
Tagliamonte, Sali A. 2004. Have to, gotta, must. Grammaticalisation, variation and
specialization in English deontic modality. In: Lindquist, Hans and Christian Mair
References 353
(eds.), Corpus Approaches to Grammaticalisation in English, 33–55. Amsterdam:

John Benjamins.
Tilkov, Dimităr, Stojan Stojanov and Konstantin Popov. 1983a: Тилков, Димитър,
Стоян Стоянов и Константин Попов (ред.). 1983. Граматика на съвременния
български книжовен език. Том II. Морфология [Grammar of Standard Modern
Bulgarian. Volume 2. Morphology]. София: БАН.
Tilkov, Dimităr, Stojan Stojanov and Konstantin Popov. 1983b: Тилков, Димитър,
Стоян Стоянов и Константин Попов (ред.). 1983. Граматика на съвременния
български книжовен език. Том III. Синтаксис [Grammar of Standard Modern
Bulgarian. Volume 3. Syntax]. София: БАН.
Timberlake, Alan. 1993. Russian. In: Comrie, Bernard and Greville G. Corbett (eds.), The
Slavonic languages, 827–886. London: Routledge.
Totomanova, Anna-Maria. 2004: Тотоманова, Aнна-Мария. 2004. Към историята на
българската темпорална система [Towards a history of the Bulgarian temporal
system]. Palaeobulgarica XXVIII(4): 82–87.
Trask, Robert Lawrence and Robert McColl Millar. 2007. Trask’s Historical Linguistics.
2nd ed. London: Hodder Arnold.
Trask, Robert Lawrence and Robert McColl Millar. 2015. Trask’s Historical Linguistics.
3rd ed. London: Routledge.
Traugott, Elizabeth Closs. 1972. A History of English Syntax. New York: Holt, Rinehart
and Winston.
Traugott, Elizabeth Closs. 1992. Syntax. In: Hogg, Richard M. (ed.), The Cambridge
History of the English Language. Vol. I: The Beginnings to 1066, 168–289. Cambridge:
Traugott, Elizabeth Closs. 2002. From etymology to historical pragmatics. In: Minkova,
Donka and Robert Stockwell (eds.), Studies in the History of the English Language: A
Millennial Perspective, 19–50. Berlin: Mouton de Gruyter.
Traugott, Elizabeth Closs. 2003. Constructions in grammaticalization. In: Joseph, Brian
D. and Richard D. Janda (eds.), The Handbook of Historical Linguistics, 624–647.
Oxford: Blackwell.
Traugott, Elizabeth Closs. 2008. Grammaticalization, constructions and the incre-
mental development of language: Suggestions from the development of degree
modifiers in English. In: Eckardt, Regine, Gerhard Jäger and Tonjes Veenstra (eds.),
Variation, Selection, Development – Probing the Evolutionary Model of Language
Change, 219–250. Berlin/New York: Mouton de Gruyter.
Traugott, Elizabeth Closs. 2010a. Grammaticalization. In: Luraghi, Silvia and Vit
Bubenik (eds.), Continuum Companion to Historical Linguistics, 269–283. London:
Continuum Press.
Traugott, Elizabeth Closs. 2010b. Revisiting subjectification and intersubjectification.
In: Davidse, Kristin, Lieven Vandelanotte and Hubert Cuyckens (eds.),
354 References
Subjectification, Intersubjectification and Grammaticalization, 29–70. Berlin: De

Gruyter Mouton.
Traugott, Elizabeth Closs and Graeme Trousdale. 2010. Gradience, gradualness and
grammaticalization: How do they intersect? In: Traugott, Elizabeth Closs and
Graeme Trousdale (eds.), Gradience, Gradualness and Grammaticalization, 19–44.
Traugott, Elizabeth Closs and Graeme Trousdale. 2013. Constructionalization and
Constructional Changes. Oxford: Oxford University Press.
Trousdale, Graeme. 2008. Constructions in grammaticalization and lexicalization:
Evidence from the history of a composite predicate construction in English. In:
Trousdale, Graeme and Nikolas Gisborne (eds.), Constructional Approaches to
English Grammar, 33–67. Berlin: Mouton de Gruyter.
Vasilev, Christo. 1972. Die Trojanska Priča als Belegquelle des Mittelbulgarischen.
Zeitschrift für Slavische Philologie 36(2): 340–352.
Vincent, Nigel and Kersti Börjars. 2010. Grammaticalization and models of language.
In: Traugott, Elizabeth Closs and Graeme Trousdale (eds.), Gradience, Gradualness
and Grammaticalization, 279–299. Amsterdam: John Benjamins.
Visser, F. Th. 1963–1973. An Historical Syntax of the English Language. 3 vols. Leiden:
Brill.
Walker, Greg (ed.). 2000. Medieval Drama: An Anthology. Oxford: Blackwell.
Warner, Anthony R. 1993. English Auxiliaries: Structure and History. (Cambridge Studies
in Linguistics 66.) Cambridge: Cambridge University Press.
Wasow, Thomas, Amy Perfors and David Beaver. 2005. The puzzle of ambiguity. In:
Orgun, Orhan C. and Peter Sells (eds.), Morphology and the Web of Grammar: Essays
in Memory of Steven G. Lapointe, 265–282. Stanford: CSLI Publications.
Weir, Andrew. 2016. That’s that construction analyzed: that be ‘resultatives’ in Scottish
English and beyond. Presentation at the LAGB 2016 meeting, University of York.
Werner, Valentin. 2016. Rise of the undead? Be-perfects in World Englishes. In: Werner,
Valentin, Elena Seoane and Cristina Suárez-Gómez (eds.), Re-Assessing the Present
Perfect: Corpus Studies and Beyond, 259–296. Berlin: De Gruyter Mouton.
Whitelock, Dorothy. 1967. Sweet’s Anglo-Saxon Reader in Prose and Verse. Fifteenth edi-
tion. Oxford: Oxford University Press.
Whitelock, Dorothy, David C. Douglas and Susie I. Tucker (eds.). 1961. The Anglo-Saxon
Chronicle. A Revised Translation. London: Eyre and Spottiswoode.
Winter-Froemel, Esme. 2013. Ambiguität im Sprachgebrauch und im Sprachwandel:
Parameter der Analyse diskurs- und systembezogener Fakten. Zeitschrift für franzö-
sische Sprache und Literatur 123(2): 130–170.
Wischer, Ilse. 2010. On the use of beon and wesan in Old English. In: Lenker, Ursula,
Judith Huber and Robert Mailhammer (eds.), English Historical Linguistics 2008:
Selected Papers from the Fifteenth International Conference on English Historical
References 355
Linguistics (ICEHL 15), Munich, 24–30 August 2008. Volume I: The History of English
Verbal and Nominal Constructions, 217–235. Amsterdam: John Benjamins.
Xaralampiev, Ivan. 2001: Харалампиев, Иван. 2001. Историческа граматика на
българския език [A Historical Grammar of the Bulgarian Language]. Велико
Търново: Фабер.
Zifonun, Gisela, Ludger Hoffmann and Bruno Strecker et al. 1997. Grammatik der
deutschen Sprache. Band 3. Berlin: Walter de Gruyter.
Zimmermann, Rüdiger. 1968. Untersuchungen zum frühmittelenglischen Tempus
system. Doctoral dissertation. Kiel.
Zimmermann, Rüdiger. 1973. Structural change in the English auxiliary system: On the
replacement of BE by HAVE. Folia Linguistica 6: 107–117
Appendices
The Appendices of this book have been made available online

and can be accessed through this QR code and the following link:
https://doi.org/10.6084/m9.figshare.9675644.
Author Index (Modern Authors)
Aarts, Bas 69 Bužarovska, Eleni 256n17, 260, 261n23, 262

Aikhenvald, Alexandra 28 Bybee, Joan L. 18–20, 27n14, 28, 34n1, 35n3,
Aitchison, Jean 15, 29n15 53n17, 73n41, 88, 245n10, 257n19
Alexiadou, Artemis 33, 47, 50n15, 51, 52n16,
57n22, 80, 224n1 Cable, Thomas 16, 72, 87, 219n27
Allen, Cynthia L. 216 Carey, Kathleen 23, 94
Ammann, Andreas 24, 57n22, 220 Chankova, Yana 34, 35n2, 35n3, 41n6, 43n8,
Anderson, John M. 55n20, 60n26 44n9, 47, 48n13, 49, 53, 54n19, 57n21,
Arnovick, Leslie K. 35n3, 37, 38, 41n6, 45n11, 59n25, 80, 88
49, 54n19, 55n20, 58, 61n28, 70n40, 90, Close, Joanne 69
162, 175n14 Comrie, Bernard xv, 75n42, 240n1, 241–242
Asenova, Petja 76, 267 Corbett, Greville xv
Askedal, John Ole 227–228 Corominas-Murtra, Bernat 16
Atanasova, Vera 224–225, 226n4, 227 Cortes, Nuria Calvo 51
Coussé, Evie 21n9
Balan, Aleksandăr Teodorov 244n7, 245n8, Croft, William 9, 10n1, 11–17, 20, 21n10, 22n11,
247n12, 250, 254n16, 255, 256n18, 25–28, 29n15, 267n28
260n22, 261 Crystal, David 147, 152n3, 160, 176
Barton, Anne 167 Cuyckens, Hubert 212
Bauer, Brigitte 70n40 Cysouw, Michael 81
Baugh, Albert C. 16, 72, 87, 219n27
Beal, Joan C. 175n14 Dahl, Östen 34n1, 245n10, 257n19
Beaugrande, Robert de 12n5 Danchev, Andrei 21n10, 94, 109, 150, 162
Benson, Larry D. 44, 147 Danylenko, Andrii 26n12, 70n40, 81, 245n10,
Berdicevskis, Aleksandrs 30 249, 260, 267n28
Bergs, Alexander 21n9 Davis, Norman 160
Biber, Douglas 47, 71, 73–74, 157n7 Denison, David 34n1, 35n3, 36–40, 41n6,
Bischoff, Shannon T. 9, 12n5 42–43, 44n9, 44n10, 45n11, 46–47,
Bloom, Paul 9 48n13, 50n15, 53, 54n19, 55n20, 56,
Bojadžiev, Todor 241n2, 246–247, 253, 268 57n21, 57n22, 59n23, 60, 61n28, 63, 66,
Börjars, Kersti 12n5, 21n9 104, 197, 202
Bowie, Jill 35n2, 57n22, 157n7, 257n19 Diewald, Gabriele 21n9
Brinker, Klaus 226 Dobrev, Ivan 242n4, 243n4, 259, 279, 283,
Brinton, Laurel J. 35n3, 37–38, 40, 41n6, 295
44n10, 45n11, 49, 54n19, 55n20, 58, Dolberg, Florian 162
61n28, 63, 70n40, 90, 162, 175n14 D’omina, Evgenija 287, 295
Brorström, Sverker 35n3, 48n13, 49n14, Dresher, B. Elan 13, 17
50n15, 51, 52n16, 53n17, 54n19, 55n20, Drinka, Bridget 8, 24, 28, 30, 77, 79n45,
56, 57n22, 61n28, 75n42, 80, 89, 97, 152, 81n46, 224n1, 237–239, 241, 257n20, 260,
157–158, 163, 167, 172, 174, 182, 194, 209, 262, 263n24, 267n28, 268n28, 269, 286,
210n25, 211–212, 219, 223, 224n1, 225 311, 330–331
Brown, Keith 33, 35n2, 36, 74, 76, 157n7, 185 Dujčev, Ivan 273
Burnley, David 195 Duridanov, Ivan 24, 82, 242n34, 243, 244n7,
Burrow, John A. 156 245n8, 284
Buscha, Joachim 227 Durie, Mark 9, 12n5
Author index (modern authors) 357
Earle, John 90–91, 95, 108, 111, 121–122 Harris, John 50n15, 81n46

Eckhoff, Hanne M. 30 Haspelmath, Martin 9–11, 12n5, 17, 26,
Eisenberg, Peter 120, 223, 224n1, 225, 226n4, 27n13, 33, 53n17, 267
227n1, 230, 232 Heine, Bernd 20, 21n10, 22–25, 26n12, 28,
Elliott, Elisabeth M. 34, 219n28, 245n10, 35n2, 70n40, 73n41, 75, 245n10, 255,
246n11, 253, 256n17, 259–260, 261n23, 261n23, 267n28
262, 264n25, 265–266, 268, 282, 326 Helbig, Gerhard 226–227
Elsness, Johan 34n1, 38, 43n8, 44, 45n11, Hengeveld, Kees 12n5
46–47, 48n13, 54n19, 55n20, 62, 77n43, Hill, Jane H. 12n5
78, 79n45, 81, 86–88 Hoffmann, Sebastian 27n14
Eska, Joseph F. 18n8 Hoffmann, Thomas 21n9
Evans, G. Blakemore 167–168, 174 Hollmann, Willem B. 21n9, 27, 65, 72, 75
Hopper, Paul J. 20, 23–27, 268n28
Falzone, Alessandra 9 Hristov, Bozhil 35n2, 251n15
Fielder, Grace E. 282 Huber, Judith 52
Fischer, Olga 35n3, 41n6, 43n8, 44n9, 45n11, Huddleston, Rodney 34, 35n2, 67–68, 73n41
46, 48n13, 52n16, 53, 54n19, 55n20, Huntley, David 242n4, 244n7, 245n8, 259
59n25, 60, 61n28, 68, 69n37, 70n39, 81,
88, 175n14 Idsardi, William J. 13, 17
Fløgstad, Guro Nore 29 Irvine, Susan 91, 93n3, 113, 124, 130, 132, 142,
Fortuny, Jordi 16 162
Frajzyngier, Zygmunt 30 Itkonen, Esa 12n4, 25
Fridén, Georg 38, 52n16, 53 Ivanova, Ana 287
Friedman, Victor A. 256n17, 260n22, 261n23 Izvorski, Roumyana 257n19, 278
Fryd, Marc 75
Jacobs, Andreas 88
Gallis, Arne 257n20, 260, 261n23, 267n28, Jany, Carmen 9, 12n5
268 Jespersen, Otto 35n3, 36, 41n6, 45n11, 48n13,
Ganeva, Gergana 24, 240n1, 242n4, 260, 50n15, 54n19, 55n20, 57n22, 61n28
286, 295 Johannisson, Ture 52n16, 53
van Gelderen, Elly 144, 164n10 Johannsen, Berit 33, 35n3, 66n32, 94,
Georgiev, Vladimir 244n7, 245n9, 245n10, 267n28
247n12, 248–256, 257n20, 261n23, 262, Jucker, Andreas H. 75, 76, 88
265
Georgieva, Elena 287, 302, 314 Kakietek, Piotr 44n9, 49, 51, 54n19, 55–56
Gerdžikov, Georgi 19, 257n19, 259, 278, 280, Keller, Rudi 12–14, 22, 220
282, 295n8, 309n14, 311n17 Kilpiö, Matti 11, 26, 35n3, 38–39, 40n5,
Gešev, Valentin 302 45n11, 54n19, 66, 77, 79, 93n5, 127, 134,
Giancarli, Pierre-Don 75 149, 154, 183, 217
Gillmann, Melitta 224 Kostadinova, Petja 247n12, 256n18
Givón, Talmy 12, 20 Kostov, Kiril 224n2, 248, 253, 256n17, 258,
Gorlach, Marina 30 269
Gräf, Adolf 53, 147–148, 150, 151n2, 152–153, Kovatcheva, Mira 35n3, 37, 44n10, 45n11,
154n4, 155–160, 166–167 54n19, 55, 59, 62n30, 67n34, 70n40, 81,
Grković-Major, Jasmina 70n4, 244n7 141, 263, 331
Krug, Manfred G. 65n31, 67, 69–70,
Haas, Walter 39n4 73n41
Halle, Morris 17 Kucarov, Ivan 242n1
358 Author index (modern authors)
Kuteva, Tania 28, 242, 245n10, 255, 261n23, 103n15, 104n16, 105n17, 106, 107n19,
267n28 108–109, 119n25, 120–121, 122n26, 124,
Kytö, Merja 21n10, 45n11, 48n13, 49, 50n15, 126, 127n29, 137n32, 139n35, 141–142, 149
51, 54n19, 55n20, 61n28 Mitkovska, Liljana 256n17, 260, 261n23, 262
Mladenov, Stefan 264–265, 267n26, 268
Labov, William 18 Mugglestone, Lynda 50n15
Laing, Margaret 45n11, 46 Mustanoja, Tauno F. 34n1, 41n6, 42, 43n8,
Lass, Roger 12n4, 13, 16, 17n7, 54, 83 44n9, 45n11, 48n13, 49n14, 51, 52n16, 53,
Łęcki, Andrzej M. 12n5, 20, 22–26, 33, 34n1, 54n19, 57, 59n25, 87, 149, 157
35n3, 36, 39, 40n5, 41n6, 42–44, 46n11,
48, 51, 52n16, 53, 54n19, 57n22, 59n23, Načeva, Mira 244n7
64, 65n31, 66n32, 67n34, 68, 69n38, Nakamura, Fujio 49, 52n16, 149n1, 157,
70n39, 71–72, 73n41, 75–76, 79, 165n11, 181n18, 207, 209–210, 217
246n10, 267n28 Newmeyer, Frederick J. 9, 12n5, 17–18
Lee, Jeong-Hoon 36, 49 Nicolova, Ruselina 240, 241n2, 242n3,
Lehmann, Christian 221 245n10, 246n11, 251, 254n16, 256n18,
Levinson, Stephen C. 22, 27n13 267n26, 278, 320
Lightfoot, David W. 13 Norde, Muriel 9
Lixačev, Dmitrij S. 273
Los, Bettelou 42, 46, 50, 63, 93, 141, 151, 225 Peev, Dimităr 302–303, 305–307, 309n14,
Lutz, Angelika 162 310, 313n18
Penčev, Jordan 246, 251n15, 252
Macleod, Morgan D. 20, 22, 25, 33, 34n1, Pennisi, Antonino 9
35n3, 37, 45n11, 81, 86–90, 92–94, Petré, Peter 59, 212, 224n1
102–103, 110, 130, 149, 224, 267n28 Piantadosi, Steven T. 16, 220
MacRobert, Catherine Mary 242n4, 244, Pinker, Steven 9, 17
273–274, 287, 294–295 Pinto de Lima, José 12
Mathieu, Eric 11 Plummer, Charles 90–91, 95, 108, 111,
McFadden, Thomas 33, 47, 50n15, 51, 52n16, 121–122
57n22, 80, 224n1 Popova, Genoveva 247n12, 265
McMahon, April M. 15 Prokosch, Eduard 46n11, 135
McWhorter, John H. 59, 80 Pullum, Geoffrey K. 34, 35n2, 67–68, 73n41
Meillet, Antoine 20, 22
Menn, Lise 87 Quirk, Randolph 33–34, 38, 56, 57n21,
Michaelis, Laura A. 33, 157n7 65–67, 70, 73, 157, 185
Millar, Robert McColl 25, 35n3, 40, 62,
289n5 Radeva, Nija 302
Miller, Jim 33, 35n2, 36, 74–76, 157n7, 185 Rainer, Eva Maria 203
Minčeva, Angelina 244n7, 269 Rikov, Georgi T. 242n4
Mincoff, Marco 35n3, 45n11, 50n15, 55n20, Ringe, Don 18n8, 30
57n21, 59, 61n28 Rissanen, Matti 46, 48n13, 49, 53, 54n19,
Mirčev, Kiril 242n4, 244n7, 245n8, 257–258, 55n20, 56, 57n22, 59, 63–64, 72, 172,
302 175n14
Mitchell, Bruce 33–34, 35n3, 36–39, 40n5, Ruseva, Dimitrina 273–274, 278
41n6, 42–43, 44n9, 45n11, 46–47, 48n13, Rydén, Mats 34n1, 46, 48n13, 49n14, 50n15,
49, 50n15, 53, 55n20, 57n22, 59n24, 51, 52n16, 53n17, 54n19, 55n20, 56,
60n26, 60n27, 62, 66n32, 67n33, 68, 57n22, 61n28, 62, 75n42, 80, 89, 97, 152,
70n39, 73, 81n47, 93n3, 95, 96n7, 97, 157–158, 162–163, 172, 174, 182, 194, 209,
98n9, 98n10, 98n12, 99n13, 100, 101n14, 210n25, 211–212, 219, 223, 224n1, 225
Author index (modern authors) 359
Sabev, Mitko 37 Traugott, Elizabeth Closs 10, 12n3, 20, 21n9,

Sag, Ivan A. 21 21n10, 22, 24–28, 35n3, 37–41, 44n10, 46,
Salgādo, Gāmini 183–184 48, 53, 54n19, 129, 218, 220, 268n28
Sampson, Geoffrey 29n15 Trousdale, Graeme 12n3, 20, 21n9, 21n10, 22,
Sanders, Andrew 160 25–27, 44n10, 129, 218, 301
Scatton, Ernest A. 240n1, 241, 244n7 Truswell, Robert 11
Schenker, Alexander M. 242n4 Turville-Petre, Thorlac 156
Schwarz, Sarah 72
Shannon, Thomas F. 45n11, 49–51, 52n16, 53, Van de Velde, Freek 9
54n19, 61n28 van der Wurff, Wim 35n3, 44n9, 45n11,
Sim, Stuart 209–210 48n13, 52n16, 53, 54n19, 55n20, 59,
Smirnickaja, Olga A. 34n1, 224 61n28, 81, 175n14
Smith, K. Aaron 24, 34, 43n8, 50, 53, 94, Vasilev, Christo 274, 277–278, 286
219 Vincent, Nigel 12n5, 15, 21n9
Stambolieva, Maria 240n1 Visser, Frederik Th. 35n3, 36, 41n6, 44n9,
Stankov, Valentin 282 44n10, 45–46, 48n13, 50, 56, 57n22,
Stoevsky, Andrei 34, 247n12, 255, 256n17, 59n23, 61, 63, 65–66
257–258, 268
Straaijer, Robin 49–51, 55n20, 57 Walker, Greg 160, 163, 165
Strang, Barbara M. H. 34n1, 37–38, 48n13, Warner, Anthony R. 35n2, 35n3, 45n11
49, 50n15, 59n25, 72, 82 Wasow, Thomas 16, 45n10, 220
Sugareva, Tekla 224–225, 226n4, 227 Weir, Andrew 76
Szmrecsanyi, Benedikt 87 Werner, Valentin 75
Whitelock, Dorothy 90–91, 111, 142
Taavitsainen, Irma 75–76 Winter-Froemel, Esme 16, 220
Tagliamonte, Sali A. 23, 44n9, 69, 77 Wischer, Ilse 45n11
Thompson, Sandra A. 53 Wrenn, Christopher L. 38
Tilkov, Dimităr 241n2, 244n7, 245n10,
247n13 Xaralampiev, Ivan 245n10, 258, 268
Timberlake, Alan 314
Tobin, J. J. M. 167–168, 174 Zifonun, Gisela 224n2, 226–227, 231–232
Totomanova, Anna-Maria 242n4, 243n4 Zimmermann, Rüdiger 44n9, 48n13, 54n19,
Trask, Robert L. 25, 35n3, 40, 62, 289n5 56, 59n25, 81, 88–89
Language Index
Arabic 288 German 4, 8, 41n, 46, 57n22, 58–59, 86, 120,

Aromanian 267 165n, 185, 222–237, 239, 267n28,
329–330, 333–334
Bulgarian 5–8, 19, 28, 33, 40n4, 46, 46n12, Middle High German 245n10
55n20, 58, 82, 86, 185, 238–334 Old High German 224n1
(Early) Modern 6, 24, 267, 272–273, 275n, Germanic 7, 24, 28, 34, 38, 45n11, 50, 55n20,
280–283, 287–328 58–59, 81–82, 223, 224n1, 237–239, 242,
Middle 6, 257–260, 272–281, 283–286, 267n28, 270, 298, 329–332
289, 294–295, 302, 326 Gothic 165n
Old 24, 165n, 242–245, 257–259, 262, Greek 6, 15–16, 28, 81, 244n, 263n, 267n28,
263n, 269–270, 275n, 278, 280, 283–284, 268–271, 273–274, 287
295, 301, 313n (see also Old Church
Slavonic) Icelandic 80
Old 165n
Croatian 273–274 Indo-European (see Proto-Indo-European)
Czech 245n10
Japanese 16
Dutch 41n, 46, 59, 75n, 225
Latin 34, 58, 67n33, 68–69, 70n39, 70n40, 81,
English 2–5, 8, 16, 21–24, 28, 32–34, 45–223, 90, 93, 119, 122n, 140–141, 159, 165n, 173n,
224n1, 225–229, 231–234, 236–239, 199, 237, 266, 267n28, 268n, 273
242–243, 245–246, 251n, 254–255,
257n19, 258, 263, 265–266, 270, Macedonian 5–6, 8, 28, 40n4, 46n12, 55n20,
275–276, 280–281, 298, 300–301, 303, 239, 256n17, 260–263, 265–270, 302, 317,
326–334 323, 325–327, 330, 332–334
(Early) Modern 24, 36, 41n, 42, 44–45,
51–53, 57, 69, 72–73, 75–80, 86, 121, (Old) Church Slavonic 24, 165n, 242–245,
145–146, 157–158, 163, 167–195, 201, 257–259, 262, 263n, 269, 273–274, 275n,
209–221, 229, 233 278, 280, 283–284, 287, 295, 301–302,
Middle 4, 39–41, 43–44, 52–53, 57, 59, 306, 309n14, 313n, 316 (see also Old
65–66, 69, 71, 78–81, 86–90, 101, 121, Bulgarian and Slavic/Slavonic)
143–167, 179, 195–209, 211, 217, 218n, Old Norse 4–5, 59, 80–81, 83, 85, 123,
286, 329, 332 144–146, 163, 203, 237, 328–330, 334
Old 3–7, 24, 27–28, 34–43, 44n10, 46, 51,
52n, 53, 56, 58–63, 66–70, 78–79, Polish 246n10, 267n28
81–145, 149–150, 154, 158, 162n, 165n, Portuguese 29
173n, 175n, 187, 195, 203, 218n, 223, 226, Proto-Germanic (see Germanic)
231, 238, 246n10, 248, 250–251, 253, (Proto-)Indo-European 24, 34, 38, 46n11, 82,
264–266, 267n26, 268n, 270, 286, 329, 165n, 223, 239, 242, 332
331–333
Romance 6, 81–82, 215, 244n, 245n10,
French 8, 59, 81, 237, 334 265–267, 269–271
Old 173n Romanian 268
Russian 245n10, 267n28, 302, 313n, 314
Language index 361
Scandinavian (see Old Norse) Spanish 19, 29

Serbian 256n17, 273–274, 287 Swedish 59, 80, 81n46
Slavic/Slavonic 7–8, 24, 28, 30, 81–82,
238–239, 241–242, 249, 260, 262, Turkish 6, 257–260, 270, 272–273, 281,
265–268, 270, 273–274, 286, 327, 288–289, 294, 326
330–334 (see also (Old) Church
Slavonic)
Subject Index
ablaut 24, 236 Carolingian/Charlemagne Sprachbund/

action(-al) (meaning) 37, 46, 49–53, 57, 59, Linguistic Area 237, 334 (see also
191, 225–227, 240–242, 314 (see also language contact)
dynamic) causative (construction) 36, 45, 64–67, 72,
agree(-ment) 2, 4, 7, 34, 36–42, 47, 61–63, 74–75, 94, 99, 116–117, 168, 174–175, 184,
82, 85–86, 90, 92, 93n5, 94–96, 98–107, 187, 190, 196–198, 200, 204, 211, 214–215,
109–121, 123–138, 140, 144–145, 186, 236, 254
238–239, 242, 245, 247–252, 255, clitic/cliticisation 21, 23–24, 246n11, 297
261–265, 268, 270, 272, 274, 284–286, collective (noun) 126
288, 297, 303–304, 307, 310, 313, 318, complement 35n2, 50, 52n, 58, 60, 68, 130,
326–327, 329, 331–333 140, 149, 170, 181, 185, 189, 193, 197, 198n,
Aktionsart 81 201, 205, 212–213, 227n, 229, 244,
ambiguity/ambiguous 10, 12, 15–17, 19–20, 250–251, 255
22, 26–27, 29–32, 39–41, 44–45, 53n17, object complement 6, 34–35, 39–40,
54–58, 59n23, 60–64, 67–68, 72–74, 77, 44–45, 58, 62, 71, 74, 82, 94, 100, 103, 130,
79, 82–86, 89–90, 92, 94, 96–98, 133, 188, 251, 262, 306, 316, 324, 327 (see
101–102, 105, 109–111, 114n, 115–116, 121, also complex transitive)
123, 126–129, 131, 138–146, 148, 151–153, subject complement 47, 57–58, 73, 97,
155–168, 170, 178–184, 187, 191–200, 202, 120, 130, 163, 188, 242, 251
204–208, 210–223, 227–229, 232–233, conclusive/inferential 6, 36, 45, 278, 282,
236–238, 250, 251n, 257, 263, 265–266, 293
270, 307, 309n14, 329–330, 333 conditional 19, 52n, 172, 274, 281n, 283–284,
ambivalence/ambivalent (see ambiguity) 288, 297, 303–304, 313 (see also
analogy 36, 39, 41, 74, 101, 106, 113, 117, 127, counter-/non-factual and
132, 141, 142n hypothetical)
anterior(-ity) 23, 33, 94, 103, 111–112, 135, Construction Grammar 11–12n3, 21n9, 28,
148–150, 152–153, 174, 185, 188, 198, 35n2, 44n10, 85, 129, 150, 218
201–202, 219n, 254 contact (see language contact)
aorist (tense) 6–7, 16, 19, 24, 240–241, contraction 60n27, 61, 82, 86, 168, 179, 182,
243n4, 259, 272, 274–275, 278–281, 284, 184, 194–195, 217, 221–222, 229, 233, 245
286, 288–289, 292–297, 302–304, converb (adverbial participle) 39, 267–268,
307–309, 311, 317, 326–327 (see also 331
imperfect and preterite) coordination 40, 47, 60–61, 82, 86, 134, 136,
arbitrariness/arbitrary 59–60, 223, 237, 331 153, 155, 166, 171, 205–206, 214–215, 232,
(see also randomness) 247–248, 303n, 313
auxiliation 35, 68, 85, 269, 285 (see also copula 32, 55–58, 81–82, 130, 148–149, 161,
grammaticalisation) 168, 184, 194, 196, 200, 211 (see also
copular)
Balkan Sprachbund/Linguistic Area 289 copular (construction/clause) 44n10, 46,
(see also language contact) 50, 54, 57–58, 73–75, 77, 79–80, 84,
bleaching 21–22, 28, 43, 65, 68, 71, 76, 97–98, 109, 123, 125n, 129–131, 144, 148,
103–104, 112, 140, 214, 248, 253–256, 153–154, 158, 161, 165–166, 168, 171,
298–302, 316, 322, 332–333 184–185, 188–189, 191, 195–196, 200–201,
boundedness (see telicity) 206, 211, 215, 219, 221, 226, 235, 251n,
264n, 285 (see also copula)
Subject index 363
copulative (see copular) expletive 105n17, 125n

correlation 34, 257, 259 extension 21–22, 36, 47, 73n, 106, 253, 256,
counter-/non-factual 22, 150, 172, 248, 261, 330, 332
279–280, 295, 297 (see also conditional
and hypothetical) formal(-ism/-ist) 12n5, 266, 270
current relevance 33, 44, 52, 57, 185, 242, frequency 27–28, 32, 43, 50–54, 60, 62, 69,
257, 277, 279, 288–290 76–83, 85–87, 89, 94, 98, 105–106, 109,
116, 144, 151, 187, 196, 213, 220–221, 237,
decategorialisation 23–24, 43, 197 272, 275, 304, 318, 330–331
declension 36–39, 47, 94, 99–100, 104n, 113, token frequency 27n14, 53
117–118, 119n, 120–121, 129, 263, 284, type frequency 27n14, 53
285n (see also agreement and functional(-ism/-ist) 11–19, 26, 29–31,
paradigm) 53n18, 54, 62, 72–73, 76, 80–81, 83, 98,
deep structure 212 108–109, 146, 158–159, 164, 167, 170,
delexical (verb) 6–7, 74, 82, 164, 190, 244n, 182–183, 193, 195, 202, 204, 213, 217,
248, 298–301, 310, 314–317, 322–323, 219–221, 223, 227n, 233, 237–238, 257,
326–327, 332–333 259, 265–266, 270, 286, 322, 327,
deponent (verb) 119, 330 329–330, 333–334
derivational (morphology) 121–123, 140, 145, functional load 12–17, 18n, 20, 30, 32,
176, 212, 230, 240, 266–267, 288, 298 54–59, 61–62, 64, 72, 74, 79, 82, 86,
(see also verbal prefix) 99, 101, 105, 179, 213, 220, 223, 239,
dynamic (reading/interpretation) 35, 254–255, 257, 259–260, 270, 272, 286,
57n21, 58–59, 73, 97–98, 109, 125n, 127, 301–302, 316–317, 322, 326, 329–330, 333
136n, 139, 144–145, 154, 162, 165, 176–177, future/futurity 15–16, 18–19, 21–27, 70n39,
188, 191, 196, 205, 221, 226, 229, 232, 248, 159, 164, 176–177, 190–191, 214, 222,
254n, 298, 302, 315, 323, 325 (see also 226n, 244–245, 248, 269, 283
action(-al); cf. stative)
gradience 44n10, 85, 129, 218
ellipsis (including omission of the gradualness 85, 129, 218, 311
auxiliary) 61, 82, 86, 105, 107–108, 148, grammaticalisation 18–28, 31, 34–35, 38,
159, 161, 168, 175, 179, 182, 184, 186, 41–44, 46–48, 52n, 53, 59–60, 62–63, 65,
193–196, 199–201, 208, 211–212, 217, 219, 68–69, 72–73, 75–76, 79n, 80–83,
221–222, 228–229, 234–237, 253, 259, 85–86, 93–95, 103, 114–115, 117–119, 121,
275, 281–283, 286, 289, 291–293, 129, 141, 144–146, 150, 153, 161, 169, 171,
302–307, 309, 313, 316–317, 326–327 177, 186, 222, 231, 238–239, 245, 248,
(see also clipped perfect and 250–252, 254–258, 261–265, 268, 270,
evidential) 272, 288, 298, 301, 311, 313–314, 316–318,
entrenchment 53, 85, 94, 129, 239, 262, 269, 320, 322–323, 326–327, 330–332, 334
272, 318, 322, 327–328 (see also polygrammaticalisation)
equivocal (see ambiguous)
evidential(-ity) 6–7, 253, 257–260, 264, 270, HPSG 21n9
272, 275, 278–282, 286, 289–297, Humboldt’s Universal 12, 75, 83, 164, 266,
302–311, 316–317, 326–327, 330, 333 295
double-evidential/admirative/ hypothetical 52, 150, 162, 165, 172, 187
dubitative 295 (see also conditional and counter-/non-
evolutionary (linguistics) 9–11, 13, 28–29 factual)
364 Subject index
imperfect (tense) 19, 240–241, 259, 272, 196, 198–200, 207, 210–213, 224–225,
274–275, 286, 288–289, 294–295, 229–231, 242–243, 261–262
303–304, 311 (see also aorist and
preterite) NICE (properties) 65, 67, 73n41
imperfective (aspect) 33, 240–241, 256n17 non-witnessed (see evidential)
impersonal (construction) 55n19, 62n30, null exponence (see zero marking/zero
111n20, 124, 143n, 154n4, 199, 224, exponence)
246–247, 256, 263, 265, 307, 310, 313,
320–321, 334 Optimality Theory 12n5
indeterminacy/indeterminate (see
ambiguity) paradigm 35n2, 35n3, 39–40, 80, 96, 99,
inferential (see conclusive) 107n, 127–128, 130, 253, 260, 318, 331 (see
intransitive (verb/clause) 28, 36, 41, 45–46, also declension)
48–51, 54–58, 60, 69, 80, 82, 92, 102, 110, parole (performance) 220 (see also langue)
119–123, 128, 131, 139–142, 145, 148, participle (see converb, past participle and
150–154, 158–164, 168–169, 171–172, present participle)
179–180, 184, 187–188, 196–198, 200, passive 19, 36, 44n10, 50, 54–66, 73, 76–77,
202, 206–208, 211–213, 215–216, 218–219, 79, 82–86, 92–98, 101–102, 104–111,
221, 224–225, 228–230, 238, 242–243, 119–131, 134–145, 148–161, 165–168,
256, 261–263, 267–268, 270, 330, 332 172–173, 176–182, 184–185, 189, 191–196,
(see also mutative and transitive) 198–201, 205–207, 211–219, 221, 223,
invisible hand 12–13, 31, 220 224n1, 226–229, 231–233, 235–239, 247,
252, 254n, 256, 264, 266, 272, 274,
language contact 5–6, 28, 59, 68, 70n39, 285–286, 329–331 (see also past passive
80–83, 85, 93, 123, 144–146, 163–164, 167, participle)
172, 203, 220, 222–223, 237–239, 244n, be 59, 105–106, 108–109, 121, 124, 127–129,
257–260, 262, 265–274, 281, 287–289, 139–140, 145, 151, 161, 168, 176, 184, 201,
294, 302–303, 316, 326–331, 334 (see 213, 217, 219, 221, 223, 226–229, 232–233,
also Balkan Sprachbund and 235, 238, 241–242, 251n, 266, 274,
Carolingian Sprachbund) 288–289, 294n, 297–298, 303–304, 313,
langue (competence) 220 (see also parole) 327, 333 (and general entries above)
layering 23–24, 65, 77, 253 become (weorðan/werden) 58–60, 73, 86,
lexicalisation 301 98, 106–109, 121–122, 125–128, 137–138,
LFG 21n9 145, 223, 226–229, 232
light verb (see delexical verb) have 61, 63–65, 67, 72, 187, 201, 213–214,
217, 227–228, 256n18, 320
mechanical/mechanistic 13–15, 19, 77, 146, get 72–75, 217, 221, 227n, 333
222, 266, 333 (see also random) perfect passive 41, 103, 114n, 124, 127n29,
merger/(near-)merger 18, 269 140, 155, 171, 176–177, 185–186, 202,
modal(-ity) 24, 28, 43, 51, 67–72, 73n, 75, 142, 204–205, 219, 236, 314
148, 150, 152, 161, 163–165, 168, 174–176, pseudo-passive 56, 254n
184, 187–191, 196, 200, 202, 205, 211, reflexive passive 297–298, 307, 310
213–215, 222, 224, 228–229, 231–232, Vorgangspassiv 226–229
244–245, 259–260, 270, 283–284, Zustandspassiv 226–229
322–324, 327 past participle 33–42, 44–48, 55–63, 65–67,
mutative (intransitive verb) 45, 48–49, 51, 71–72, 74, 76–77, 82, 86, 93–101, 103–105,
53–54, 56, 94, 122, 148, 151–152, 154, 107, 112–116, 119–123, 129–130, 132–134,
161–162, 168, 170–173, 177, 180, 184, 187, 136–140, 153–155, 158–160, 162n, 163,
Subject index 365
165, 171, 173, 177–178, 186, 188, 197, 199, perfective (aspect) 32–33, 51, 138n33,
201, 203–205, 211, 215, 218n, 219, 224, 240–241, 256n17
226, 236, 238, 241–242, 248–253, phase 34
255–256, 258, 263, 266, 269–270, 331, phonological change (see sound change)
333 phrasal verb 57, 197n
past active participle 8, 36, 46n12, 60, pluperfect (see past perfect)
238, 241–242, 245–247, 259, 264–268, polygrammaticalisation 25, 43, 72, 75,
270, 275, 282–285, 292, 295n, 302, 303n, 82–83, 146, 174–175, 189, 195, 201, 213,
305–310, 317, 319–322, 330–333 215, 222, 237, 322, 327, 333 (see also
past passive participle 8, 36, 46n12, 60, grammaticalisation)
166, 238, 241–242, 245–247, 251, possession/possessive (construction) 24–
261–262, 264–268, 270, 275, 285–286, 25, 34–38, 42–43, 48, 62–65, 68–71,
297–298, 313, 314n, 316–321, 325, 327, 74–76, 81, 83, 103–104, 114, 145, 171, 204,
331–333 214, 222, 244–245, 248, 251, 254–256,
perfect 24–25, 29, 32–34, 48–63, 65, 71–87, 268, 289, 298, 314–316, 322, 324–325,
89–90, 92–96, 98, 102–105, 109–116, 333–334
119–124, 126–135, 139–146, 148–164, pragmatic inferencing/implicatures 22–23,
165n, 166–174, 178–189, 192–204, 26–27, 153, 222
206–231, 233–239, 245, 257, 259, Prague School 12
264–267, 272, 274–284 (see also perfect prefix (see verbal prefix)
passive and perfect progressive) prescriptive (grammar/grammarian) 20,
be 45–48, 52–54, 59–61, 74–77, 79–83, 86, 49, 56
94–95, 98, 105, 119–123, 126–129, present participle 47, 51, 71–72, 99, 104, 159,
139–140, 143–146, 148, 150–154, 156–159, 176, 206, 214–215, 218n
161–164, 166–168, 172–174, 179–184, preterite (past tense) 24, 29, 33, 44, 77–78,
187–189, 192–196, 198, 201, 203, 80, 87, 93, 103, 111, 116, 120–121, 136,
206–208, 210, 212–213, 215, 217–227, 139–142, 153, 156, 157n6, 165, 172, 185,
229–231, 233–239, 241–247, 251–254, 208, 215, 224, 236–237, 267n28, 272, 275,
257–261, 263, 266–267, 270, 272, 286, 289, 292, 294, 302, 304 (see also
274–275, 286, 288–298, 301–307, 311, aorist and imperfect)
317–319, 322, 326–334 Principles and Parameters 12n5
become (weorðan) 58–59, 119, 121–124, progress 28–29
127–129, 231 progressive (aspect) 32, 41, 47, 54, 55n20,
clipped perfect 61 (see also ellipsis) 56, 60n27, 79, 81–82, 92, 99, 102, 104,
double perfect 57 108, 124, 131, 143n, 168, 174–177, 184,
get 73–75, 329, 333 189–190, 194, 196, 200, 204–206, 211,
have 28, 34–45, 65, 67, 72, 79–83, 86, 94, 213–215
100, 103–104, 112, 116, 119, 122, 127–130, passive progressive 204–205
132–135, 139–140, 143–146, 148–154, 157, perfect progressive 41, 164, 171, 175,
161–164, 167–171, 174, 179–182, 184–188, 186–187, 189
195–198, 201–202, 204, 206–207,
210–215, 219, 221, 224–231, 233–239, 242, random(-ness) 29, 31, 221–223 (see also
245–265, 267–272, 286, 296, 298, 302, arbitrariness/arbitrary and mechanical/
314, 316–323, 325–334 mechanistic)
own 41, 43 reanalysis 15, 22–23, 25–27, 31, 35–36, 39–41,
past perfect 33, 51, 87, 89, 92–93, 116, 120, 60–61, 68, 74, 82–83, 85–86, 130, 145, 171,
141, 149, 153, 158, 163, 177, 202–203, 212, 222, 238, 265–267, 270, 327, 330–331,
219, 253, 259–261, 274–276, 280, 288 333–334
366 Subject index
reinterpretation (see reanalysis) transitive (verb/clause/construction)

relevance (see current relevance) 35–36, 39, 41, 46, 48–49, 55–56, 58–61,
renarrated/renarrative (see evidential) 68, 75, 80, 82, 92, 94, 102, 110, 116,
resultative(-ness) 6, 23, 33, 46, 55–57, 59, 67, 119–123, 128, 130–132, 135, 140–142, 145,
74, 80, 112, 141, 153, 174, 219n, 238, 257, 148–149, 153–154, 158–159, 161, 164n,
260, 276, 278, 290–291, 295–296, 168–171, 179–180, 184–187, 196–197,
304–305, 311 198n, 200–201, 207–208, 211, 213, 216,
218n, 221, 224–225, 227–231, 233,
S-curve 48n 242–245, 266–267, 270, 326, 330–332
Sign-Based Construction Grammar (see also intransitive)
(see Construction Grammar) complex transitive 36, 44–45, 64, 71, 74,
sound change 7, 11, 14–15, 18, 21, 24–27, 54, 82, 84, 86, 130, 262, 270, 306 (see also
65n, 130, 142, 218n, 269, 273, 276n, 313n object complement)
(see also ablaut, merger and umlaut)
standardisation (of English) 88 umlaut 121
statal (see stative) underspecification 84 (see also ambiguity)
stative (reading/interpretation) 33, 35, unidirectional(-ity) 26
37–38, 42, 45n10, 46, 49, 53, 55–57, 59,
60n27, 62–63, 67, 73–74, 80, 84, 86, 94, verbal prefix 81, 123, 138n33, 140, 154, 212,
97–98, 105, 109, 119, 121, 123, 125n, 218n, 230, 236, 240, 266 (see also
126–131, 138–139, 144–145, 148–149, derivational morphology)
153–156, 159, 161, 165–166, 168, 171,
177–178, 184, 188, 191, 193, 195–196, word order 36, 44–45, 68, 93, 97, 104, 121,
200–201, 206, 211, 215–217, 219, 221, 134, 144, 149–150, 161, 168–169, 186–187,
224n1, 224n2, 225–229, 232–233, 235, 197–198, 201, 204, 211, 214, 246n11, 248,
254n, 260, 268, 285, 298, 315, 333 (see 255–256, 261, 262n, 322–323
also Zustandspassiv, cf. dynamic)
supine 39 zero marking/zero exponence/zero
suppletion 45n11 morphology/zero inflection 4–6,
syncretic/syncretism 83, 307 39–41, 47, 82, 85–86, 92, 94–96, 99,
101–102, 104, 105n18, 106–110, 113,
teleological/teleology 12–20, 28–31, 77, 220 115–119, 121, 123, 125–135, 137–138,
telicity 239–241 140–141, 145, 186, 238, 250, 263–266, 270,
tough-movement 232 331, 333
Index of Sources
Ælfric 40 diary 49, 181n

All Quiet on the Western Front (see Im Pepys, Samuel 49, 52n, 157, 181n, 207,
Westen nichts Neues) 209–210, 217 (see also letters)
Ancrene Wisse 89 Dickens, Charles 4, 51, 57–58, 60n27, 143,
Anglo-Saxon Chronicle(s) 3, 60, 62, 67, 85, 153, 163, 218–220, 334 (see also letters)
87, 90–145, 149–150, 163, 175n, 217, 222, Dictionary of Old English 35n3, 39, 48, 66,
334 68–69, 93n5, 149
Peterborough Chronicle 85, 90–91, Dimov, Dimitar 323–326
130–145, 152n, 203
Austen, Jane 51 Early English Books Online 77
Ayenbite of Inwyt 218n Eliot, George 51
Beowulf 37, 48, 52n, 122n Facebook 75, 263

Bible 51, 67n33, 159, 218n, 243–244, 258, 262,
296, 311–312 Germanov sbornik 258
King James Bible 243–244, 258, 262, 296
BNC 33, 70, 72 Hansard Corpus 77
Bosworth and Toller 113–114, 119–120, 122n, Helsinki Corpus 39, 77, 79, 89–91, 147, 168, 184
123, 135, 139–142
Botev, Hristo 247, 249 Im Westen nichts Neues 236
Brontë, Charlotte 51
Bulgarian National Corpus 318–323, 327 King, Stephen 74
Canterbury Tales 53, 147–160 (see also Layamon 44, 66, 89

Chaucer, Geoffrey) Lessing, Gotthold Ephraim 4, 228–237
Charters of Wallachia 259 letters 51, 89, 179, 217, 222
Chaucer, Geoffrey 4, 44, 52–53, 81, 147–162, Dickens, Charles 219 (see also Dickens,
166–167, 172, 178–180, 182–183, 187, 195, Charles)
200, 208, 210, 220n, 221 Paston 195–210, 222
Chronicle of Constantine Manasses 273 Pepys, Samuel 210–217, 229, 231n
COCA 73, 181 (see also diary)
Codex Marianus 258 Priestley, Joseph 49
Codex Suprasliensis 259 Luther, Martin 225
COHA 50, 77
Corpus of Middle English Prose and Malory, Thomas 207
Verse 195–196, 200 Minna von Barnhelm 228–237 (see also
correspondence (see letters) Lessing, Gotthold Ephraim)
(The) Country Wife 180, 183–195, 203, 213,
221, 229 OED 34, 63, 72, 151n, 169n, 173n, 180, 207,
211–212, 215–216, 218
Damascene (Damascenus Troianensis) 6, Online Middle English Dictionary 151n, 154,
273, 287–303, 309n14, 313n, 315, 326 156, 158–159, 162n, 169n, 179, 199, 203,
Darwin, Charles 158 206–208
368 Index of Sources
Ormulum 71 Shakespeare, William 4, 44n9, 51, 55,

(The) Owl and the Nightingale 89 167–183, 186, 195, 210, 218, 221–222
Slavonic-Bulgarian History 6–7, 275,
Paisius (see Slavonic-Bulgarian History) 302–318, 327
Parker Chronicle (see Anglo-Saxon
Chronicle(s)) (A/The) Tale of Troy 6, 273–281, 283–286,
Pastoral Care (King Alfred’s Version) 34, 100 288–289, 291, 294–295, 298, 303,
Pelin, Elin 264 326
Peterborough Chronicle (see Anglo-Saxon (A) Tale of Two Cities (see Dickens, Charles)
Chronicle(s)) Talev, Dimitar 323–326
Prayers of Cserged 259–260 (The) Taming of the Shrew 167–183 (see
also Shakespeare, William)
Restoration comedy (see Country Wife)
Vazov, Ivan 247–248, 256
Second Shepherds’ Play (see Secunda Vežinov, Pavel 282
Pastorum)
Secunda Pastorum 81, 143–144, 152n, Wycherley, William (see Country Wife)
160–167, 172, 179–180, 183, 187, 195, 200,
203, 221–222

Hristov, B. (2020) - Grammaticalising The Perfect and Explanations of Language Change

Uploaded by

Copyright:

Available Formats

Hristov, B. (2020) - Grammaticalising The Perfect and Explanations of Language Change

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

Hristov, B. (2020) - Grammaticalising The Perfect and Explanations of Language Change

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What languages does it compare?

What languages does it compare?

Grammaticalising the Perfect and Explanations of Language Change

Brill’s Studies in Historical

Joan Bybee (University of New Mexico) – Lyle Campbell (University of Hawai’i

The titles published in this series are listed at brill.com/bshl

Names: Hristov, Bozhil, 1984- author.

Copyright 2020 by Koninklijke Brill NV, Leiden, The Netherlands.

This book is printed on acid-free paper and produced in a sustainable manner.

3 The Story of the English Perfect 32

3.5 Problems for the Traditional View 61

4 The Development of the Perfect in a Selection of Old English Texts 85

5 Further Development of the Perfect Based on a Selection of Texts – from

5.4 The Sixteenth Century: Shakespeare 167

6 How Things Could Have Been: a Glance at German 223

7 The Perfect in the History and Structure of Bulgarian 238

8 The Development of the Perfect in a Selection of Middle and Modern

following Comrie and Corbett (1993: xii–xiii)

Cyrillic Latin Cyrillic Latin Cyrillic Latin

© Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004414051_002

perfect because it was less multi-functional and potentially ambiguous, why

As expected, the take-home message of Chapters 4, 5 and 6 is that things are

of interpretation (Old English), as opposed to one where there is little compa-

representatives of the east/west split in perfect formation (see Drinka 2017),

2.1 Evolutionary and Functionalist Models of Language Change

2.1.1 Evolutionary Linguistics

© Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004414051_003

human languages (culminating in genealogical classifications and family trees

Evolution is recognized as a process that occurs with certain types of

Language change began to be accordingly thought of as originating in a single

1 Of course, the possibility of multiple independent innovations is in no way precluded.

the empirical evidence indicates that linguistic selection is governed

Speakers can hardly be said to create optimal grammars intentionally; rather,

In Keller’s theory, when speakers innovate, they do not consciously intend

an invisible hand explanation explains its explanandum, a phenomenon

Nevertheless, it is exceedingly tempting, sometimes convenient, or perhaps

The difference between intentional and nonintentional mechanisms for

Rejecting teleological mechanisms as highly suspect based on the evidence

2.1.3 Problems and Challenges

[a]lthough not being misunderstood may contribute to language change

As one of the most outspoken critics of function-based accounts of change,

Further evidence against functionalist teleology comes from Newmeyer

if grammaticalisation were driven by communicative needs only. Thirdly, by

2.1.4 Interim Summary and Preview

speakers, ignoring prescriptive interventions or language planning). Greater

2.2 Grammaticalisation and Reanalysis

(1) [[movement verb + Progressive] + purpose clause (to + infinitive)]

Several movement verbs can be inserted in this template: I am going/travel-

to point B, but when the subject is human, it is easy to infer an intention on

(I have used it – newest layer), affixation (I admired it – older layer), ablaut

Importantly for present purposes, it has been observed that a grammaticali-

2.2.2 Reanalysis in the Context of Grammaticalisation

the complexity of the morphosyntactic encoding of meaning, combined

This entails potential ambiguity for the reinterpretation to be able to take

2.2.3 The Role of Frequency and Contact in Grammaticalisation

2.3 Other Preliminaries

I am inclined to believe that randomness does play some role in language

1 Of course, the possibility of multiple independent innovations is in no way precluded.

b. The letters have come so regularly of late … [1714, Wentworth, cited in