The Broken Plurals in the Muscat Dialect of Omani Arabic

Khalsa Hamed al-Aghbari

B.A., Sultan Qaboos University, 2001

A Thesis Submitted in Partial Fulfillment of the

Requirement for the Degree of

MASTER OF ARTS

Department of Linguistics, University of Victoria

0Khalsa Hamed al-Aghbari, 2004

University of Victoria

All rights reserved. This thesis may not be reproduced in whole or in part,
by photocopy or other means, without permission of the author.
Supervisor: Dr. Suzanne Urbanczyk

ABSTRACT

This thesis examines one of the most intriguing and much studied phenomena in

Semitic known as the broken plural formation. It has a twofold goal. It documents the

diverse shapes of broken plurals in the Muscat dialect of Omani Arabic. Furthermore, it

provides a formal analysis to the shapes and vocalism contained in these word forms

within Optimality Theory framework (Prince and Smolensky 1993; McCarthy and Prince

1993a & 1993b).

Following proposals by McCarthy (2000), this thesis assumes that the distinction

between the singulars and broken plural shapes is better represented as 'affixed mora

(p)' attached at a certain locus in broken plural forms. The analysis of the vocalism

characterizing broken plural forms addresses two distinct types of fixed vocalism:

phonological and specified. Fixed vocalism is demonstrated to result from an interaction

between conflicting alignment and CrispEdge constraints (It6 and Mester 1999) together

with *Place markedness constraints.

 TABLE OF CONTENTS

Broken Plurals in the Muscat Dialect of Omani Arabic

Khalsa Harned al-Aghbari

Abstract
..
...............................................................................................
11

Table of contents....................................................................................
...
111

Acknowledgments ................................................................................... vi
Chapter One: Introduction
1.0 Statement of Intent ...................................................................... 1
1.1 The patterns .............................................................................. 2
1.2 Overview of thesis .................................................................... 14
Chapter Two: Overview of Omani Arabic
. .
2.1 Genetic affihation .................................................................... 17
. . . .
profile ............................................................... 18
2.2 Soc~o-lingu~stic
2.2.1 Dialectal variations and languages in Oman .............................18
2.2.2 Situating Muscat dialect ..................................................... 19
2.2.3 Data of study .................................................................. 20
2.3 Previous research on Omani Arabic ............................................... 21
2.4 Discussion of sound system ......................................................... 25
2.4.1 Segmental inventory ........................................................25
2.4.1.1 Consonant inventory ................................................ 25
2.4.1.2 Vowel inventory ..................................................... 27
2.4.2 Syllable structure ............................................................ 29
Chapter Three: Literature review on broken plurals
3.1 Omani Arabic Vs . classical Arabic broken plurals ...........................34
3.2 Linear transformational-generative framework ...............................34
3.3 Historical and comparative approach ..........................................37
3.4 Melodic transfer ...................................................................38
. . . .
.........................................................39
3.5 Prosodic c~rcumscript~on
3.6 Optimality Theory .................................................................40
 Chapter Four: Theoretical assumptions
4.1 Optimalily Theory Framework ................................................. 43
4.1.1 Principles of Optimality Theory ....................................... 44
4.1.2 Markedness constraints .................................................. 45
4.1.3 Faithfulness constraints .................................................. 46
4.1.3.1 Correspondence Theory .......................................... 46
4.1.3.1.1 Output-Output Correspondence .......................47
4.1.3.1.2 Positional Faithfulness .................................49
4.1.4 Alignment constraints .................................................... 51
4.2 Generalized Template Theory ................................................... 54
Chapter Five: Shapes
5.1 Introduction ........................................................................... 55
5.2 OT analysis of the shapes of Muscat broken plurals ............................56
5.2.1 Typical (canonical) broken plurals ....................................... 56
5.2.2. Shapes with epenthetic glides ............................................. 63
5.2.3 Shapes with medial geminates............................................. 69
5.2.4 Plurals with even iambs .................................................... 72
5.3 Exceptional Broken Plural Shapes................................................ 80
5.4 Summary of chapter five ........................................................... 82
Chapter Six: Vocalism
6.1 Introduction ........................................................................... 84
6.2 Overwriting of vocalism (McCarthy and Prince 1990a)........................ 87
6.3 Fixed Segmentism (Alderete et a1 1999)......................................... 88
6.4 Analysis of Fixed vocalism .........................................................90
6.4.1 Fixed [a] ........................................................................ 90
6.5 Analysis of phonological vocalism ................................................ 95
6.5.1 Variable vowels ................................................................95

6.6 Vocalism in the exceptional shapes of the broken plurals ..................... 104
6.7 Summary of chapter six ............................................................ 106
Chapter Seven: Conclusions and implications ..............................................
108
 .......................................................................................111
References...
Appendix A: Muscat broken plural nouns.................................................... 117
Appendix B: Glossary of technical linguistic terms ......................................... 129
 ACKNOWLEDGEMENTS

I cannot retrieve from a lexicon how indebted I am to my supervisor Suzanne

Urbanczyk for her insightful arguments, constructive comments and most importantly
unfailing trust. Her scholarly supervision is not easily repaid or soon forgotten. Not only
did she give me long hours of professional guidance, she also taught me with her sharp
skills how to write linguistics with clarity and how to stand for my analysis and get my
ideas across. Many great thanks, Su.
A deep debt of gratitude is also due to my committee members for exposing to me
the wonders of linguistics with immense enthusiasm during my Master's program. Leslie
Saxon has deeply touched me with her kindness and dedication to syntactic theories since
I stepped my feet in this friendly department. Dr. Andrew Rippin, my outside member,
has as much as I did contributed to my thesis. His generous help with the references is
highly acknowledged and most appreciated. I promise not to dismiss such an
unforgettable and nice touch. I am also lucky to have crossed paths with Ewa
Czaykowska-Higgins who has impressed me about her orderly ideas and excellent skills
in teaching phonology. I cannot imagine such a rewarding learning experience without
these scholars.
I dedicate my thesis with profound admiration and eternal respect to my family who
continuously encourages me to pursue my higher education abroad and boosts up my
spirits whenever despair succeeds to creep into my heart. Countless thanks are due to my
parents for always instilling in me the passion for knowledge and persistent determination
for excellence in life.
I owe tremendous thanks to my teachers, colleagues and students at Sultan Qaboos
University (SQU). Thanks for keeping in touch with me while I'm away from home. It
has definitely opened new horizons of hope and success to me. My deepest thanks go to
Dr. Nafla al-Kharousi who taught me my first linguistics and encouraged me to do my
MA in theoretical linguistics. In linguistics, I surely found myself.
I'm also grateful to my beloved country Oman for bestowing on me this precious
opportunity to pursue my master's in Canada. My language, Omani Arabic, is certainly a
rich ocean of wonders in which how far you dive, there is still more to be exposed. I am
glad to have done my research on it. Again and forever, and to all, Sukran gazi:lan.
 CHAPTER ONE

Introduction

1.0 Statement of Intent

In Semitic languages, 'plural' meaning is expressed most often by the broken

'internal' plural. This phenomenon, where an internal change rather than a fixed affix is

imposed on the singular forms, can be seen in the data below which represent broken

plurals in the Muscat dialect of Omani Arabic. The stem of the singular forms in (1)

particularly the C I V C ~shape

' at the left edge of these forms as in [dafl in (1.a) is altered

into a different shape ClaCza: in the broken plurals.

(1) Broken plurals2 in the Muscat dialect of Omani Arabic

Sing. PI. Gloss
a. daftar dafa:tar 'notebooks'
b. kursi kara:si 'chairs'
c. misgid masa:gid 'mosques'
d. tannu:r-ah tana:ni:r 'skirts'

The intricacies involved in the broken plural formation of Semitic languages have

long intrigued linguists who expressed vigorous interest in proposing different models to

explain this phenomenon. In forming broken plurals, the stem of the singulars modifies in

a different shape and the vowel changes into a fixed [a] to indicate meaning. This

morphological change attracts broad interest and leads to understanding the principles of

word-formation. Although broken plural formation is a highly complicated word

formation, it constitutes the most common type of plural formation in Semitic languages

I
C and V stand for consonants and vowels contained in a word respectively.
2
The broken plural forms are phonemically transcribed following previous scholars on this phenomenon
and Arabic in general (McCarthy and Prince 1990a and McCarthy 2000).
(Levy 1971, McCarthy and Prince 1WOa, Abd-Rabbo 1990, Abu-Mansour 1995, among

others).

This study is a linguistic attempt to document the shapes of broken plural nouns in

the Arabic dialect of Oman. More specifically, it explores the diverse shapes observed in

the formation of this morphological phenomenon in the spoken language of educated

Omanis who specifically relate to Muscat, the capital of Oman.

It is the aim of this thesis to also offer a formal analysis of the shapes and vocalism,

(vowel quality) of broken plurals within the Optimality Theory framework (OT) (Prince

and Smolensky 1993; McCarthy and Prince 1993a &b). This thesis basically builds on

proposals made by McCarthy (2000) to account for the shapes of broken plurals in the

Muscat dialect. On the other hand, the analysis of the fixed vocalism exhibited in broken

plurals is inspired by proposals and arguments made in Alderete et a1 (1999) and

Urbanczyk (1999). This thesis is not limited to investigating the typical, most productive

broken plurals' shapes and vocalism but also presents the exceptional patterns which

diverge from the general mode of broken plural formalism.

1.1 The patterns

In the Muscat dialect of Omani Arabic, there is immense diversity in the shapes

which result from mapping singulars onto broken plurals. The diversity in shapes often

stems from the distinct shape of the singular forms from which broken plurals are

derived. Discussion of broken plural shapes first outlines the typical and most common

broken plural patterns and then moves to describe the peculiarity of the exceptional

shapes.
 Before describing the shapes of broken plurals, it is worth mentioning that

adjectives in the Muscat dialect of Omani Arabic also take broken plurals, for example

[ko:'i' kbi:r] 'big elbow, sing.' is pluralized as [ki'i'a:n kba:r] 'big elbows,pl.'. Just like

the noun preceding it, the adjective [kbi:r] takes the broken plural [kba:r]. However, the

focus of this study is limited to broken plurals of nouns and leaves the discussion of

'adjective broken plurals' for future research.

The generalization that governs the formation of the typical broken plurals

(McCarthy and Prince 1990a) is that the left foot ClVlC2 ([dafl as in form 1.b below) or

CVV ([Sa:] as in form 6.a below) of the singular forms is extracted and mapped onto a

typical iamb3 ClVC2VV ([dafa:], [Sawa:] respectively. In forms (5), the C I V C ~ V(e.g.

[maki] in form 5.a below) which belongs to the first iamb (ma.ki:) of the singular form

[(ma.ki:).n-ah] is extracted and expanded into a typical iamb [ma.ka:] when forming

broken plurals. In a nutshell, the formation of broken plurals targets the first two morae

of the singulars and maps them onto an iamb with three morae. The iamb constructed

involves a sequence of two syllables; heavy H preceded by a light syllable L and resides

at the left edge of the broken plural. This iamb is, then, concatenated to the rest of the

singular form, maintaining the weight of the final syllable (McCarthy and Prince 1990a;

Glover 1988: 1 5 9 ~ )McCarthy

. and Prince (1990a) conclude that, as a basic requirement

for the formation of broken plurals in classical Arabic, a typical iamb L H has to be

' lamb is a type of foot structure which is a right-headed foot. It can be either disyllabic or monosyllabic. If
disyllabic, then it has a light-heavy syllable or light-light syllable with the second syllable more prominent.
If monosyllabic, then it is always heavy.
4
Glover's research reports that one of the characteristics of plurals in Muscat Arabic is that the final
syllable maintains the same length ofthe final syllable in the singular forms.
realized in broken plural shapes. I provide the following prosodic representation to

illustrate the mapping o f singulars onto broken plurals. This representation shows that the

first trochaic foot [mis] o f the singular form [mis.gi.d] 'mosque' is expanded into an

iamb [ma.sa:]in the broken plural.

(2) Prosodic representation o f m i s g i d h masa:gid 'mosques'

Sing. PI.
FI FI
I A
0 0 0 D (r

h/?
m i s g i d
(mis). gi. d (ma.sa:).gi.d

Broken plurals listed below are considered to he the typical shapes. There are many

details to be explored in a full account regarding the extra morphology taking place in

some forms in addition to the general mechanism applied to formalize broken plurals.

(3) The typical shapes of the broken plurals

Sing. PI. Gloss
1. (CVC).CV.C Singular forms (CV.CV:).CV.C
a. (mis).gi.d (ma.sa:).gi.d 'mosques'
b. (daf).ta.r (da.fa:).ta.r 'notebooks'
2.(CVC).CV.C- ah Singular forms (CV.CV:).CV.C
a. (rnas).ta.r- ah (ma.sa:).ta.r 'rulers'
b. (mal).Ya.q- ah (ma.la:).Ya.q 'spoons'
3. (CVC).(CV:).C Singular forms (CV.CV:). (CV:).C
a. (qan).(di:).l (qa.na:),(di:).l 'lanterns'
b. (mis).(ma:).r (ma.sa:).(mi:).r 'nails'
c. (kar).(fu:).n (ka.ra:).(fi:).n 'knuckles'

The singular forms are shaped into CVC.CV(V).C. The suffix-ah attached to the

singular forms in ( 2 ) and other singular forms below indicates that these nouns belong to
the feminine gender. It is not considered as part of the basic consonants which are

assumed to indicate the meaning of the stem of the singular forms (McCarthy and Prince

1990a; Abd-Rabbo 1990, among many traditional Arab grammarians). The first foot of

the singulars C I V C ~
as in form (1.b) [dafltar is changed into [dafa:]tar to indicate

plurality. The final syllable of both the singulars and plurals is preserved and surfaces in

broken plurals without a change. This is mostly manifested in forms (3) where the length

of the final syllable in the singular forms is carried over to broken plurals with no

shortening.

Broken plurals in (4) below contain only three consonants coming from the base

singular with the shape CVC.CV. Again, the shape of the first foot [ C I V C ~is] expanded

into a different shape [CIVC~V:];[mak] becomes [maka:] (form 4.a). The final syllable

CV in these broken plurals corresponds to the final syllable in the singular form where

both the length and quality of the vowel are kept unaltered.

4. (CVC).CV Singularforms (CV.CV:).CV

a. (mak).si (ma.ka:).si 'dresses'
b. (kur).si (ka.ra:).si 'chairs'

The singular forms in (5) and (6) have fewer than four base consonants and their

broken plurals expand by inserting a default glide to conform to the mapped shape. For

example, in forms (6), the first foot of the singular [Sa:]aar contains only one consonant

and a long vowel. [Sa:] requires another consonant for shaping this foot into (CIVC2VV);

[Sawa:]aar results. The singulars in (5) have a long high vowel in their second syllable

[Ci:] like in [makkn-ah]. To form broken plurals, CIVCzV [maki] in the first iamb of the

singular form (5.b) maps onto C1VC2V: [maka:]. The residue of the singular [in] which is
left after the mapping happens requires a glide to form the onset of the third syllable in

the broken plural [maka:jin].

5. (CV.CV:).C- ah Singular forms (CV.CV:). jV.C

a. (ma.ki:).n- ah (ma.ka:).ji.n 'machines'
b. (ha.di:).q- ah (ha.da:).ja.q 'parks'

6. (CVV).CV(V).C Singular forms (CV.wV:).CV.C

a. (Sa:).aa.r (Sa.wa:).aa.r 'bed blankets'
b. (Sa:).ra.Y (Sa.wa:).ra.T 'roads'
c. (ta:).(bu:).r (ta.wa:).(bi:).r 'line-ups'
d. (ta:).(bu:).q (ta.wa:).(bi:).q 'cement blocks'

Forms (7) below have medial geminates [sakki:n] with the first half of the

geminates closing the first foot of the singular and the second half filling the onset

position of the second syllable (sak)l(ki:)2.n. The geminates being adjacent to each other

in the singular split into two feet in broken plurals [(saka:)l(ki:)2n] as a result of the

expansion of the left foot of the singulars (sak) into a different shape (saka:), maintaining

the length of the final syllable of the singular forms. McCarthy (2000: 178) notes that the

original geminates in the singulars become 'long-distance' in broken plurals. The

adjacent geminate of the singulars is separated by a vowel in the broken plurals.

7. (CVCZ).CZV:CSingular forms (CV.C2V:).(C2V:).C

a. (sak).(ki:).n (sa.ka:).(ki:)n 'knives'
b. (8al).(la:).g- ah (@a.Ia:).(Ii:).g 'refrigerators'

There are singular forms which have four consonants in their underlying forms but

map onto a broken plural shape with only three consonants.

8.(CVC)CVj-ah S i n g h r f o r m s (CV.CV:).CV
a. (su0.ri.j- ah (sa.fa:).ri 'cook pots'
b. (kur)fa.j- ah (ka.ra:).fi 'beds'
 The final consonant in the singular forms in (8) is consistently the glide /j/

preceded by an open syllable with a short vowel [i] or [a]. The trochaic foot of the

singular form like in (8.b) [kur] expands into [kara:] in the broken plural. The length of

the final syllable is the same in both the singular and plural forms. Glover (1988: 63)

discusses an active phonological process in Muscat Arabic which coalesces a

tautosyllabic sequence of a short vowel and glide into a long vowel, mostly /e:/. In these

broken plurals, there is a tendency of merging the glide /j/ and short vowels of the final

syllable of the singular forms into the high front vowel /i/.

Some singular forms in the Muscat dialect of Omani Arabic have free variations of

shaping into different broken plurals (Shaaban 1977). The singular forms whose shape is

exactly similar to those in forms (I), (2) and (3) above map onto a different shape of

broken plurals (CCV:).CV(V).C. They thus differ from the canonical broken plurals

(CV.CV:).(CV(V)).C in that they collapse the first two syllables of the first foot at the

left-edge into one heavy foot by deleting the first vowel. Shaaban (1977) formalizes the

following rule to capture the deletion of the vowel between ClC2 in broken plurals.

(9) VI [short] -+ 0 / # CIVI.C~VVresulting in a cluster and heavy syllable CCVV.

Glover (1988) observes a general rule of short vowel deletion from the first syllable

in Muscat Arabic which results in a consonant cluster at the left-edge of words. She states

that the left edge of words is the only environment where a cluster of consonants is

observed in Muscat Arabic. If short vowels are located in the middle or end of words,

then they are likely to he maintained. Thus, syllables in Omani Arabic don't have clusters

at their margins as in most environments consonants are followed by vowels.

10. (CVC).CV(V).C Singular forms ( C.CV:). CV(V). C
a. (maq).Ja.t m.(qa:).Ja.t 'matches'
b. (maq).sa.f m.(qa:).~u.f 'small vendors'
c. (maS).xa.l m.(Sa:).xi.l 'sieves'
d. (muy).(ba:).r m.(ya:).(bi:).r 'curing spells'
e. (muY).(la:).q m.(Ya:).(li:).q 'hooks'
f. (muf).(ta:).h m.(fa:).(ti:).h 'keys'

The first foot of the singular in (lO.b) [maq] maps onto [mqa:] and not the usual

expected pattern [maqa:]. The final syllable in these forms maintains its length when the

mapping to broken plural takes place just like the typical broken plurals. Although broken

plurals patterning like this all begin with a sonorant /m/, this observation is later

countered by broken plurals clustering various types of consonants at their left edge.

There is a wide range of broken plurals with highly different shapes from the

typical broken plurals described above. They are distinguished by the distinct shapes of

their singular forms which are characterized by a long vowel in the only syllable they

have and fewer consonants than in the typical broken plurals.

11. C.(CV:).C Singularforms (CVC).(CV:).C

a. n(Ya:)l (nuT).(la:).n 'sandals'
(tur).(ba:).n 'sands'
c. g(da:)r (gid).(ra:).n 'walls'
12. CVV.C Singular forms CVC.CV:.C
a. ko:b kiba:n 'cups'
nira:n 'fires'
kiYa:n 'elbows'

The singular forms in (1 I ) have only one heavy syllable and three consonants. They

are shaped into C.CVV.C. This thesis proposes that when these singulars are pluralized,

the consonant clusters CC word-initially are broken up by an epenthetic vowel. The

inserted vowel can either be [i] or [u] depending on the place features of the cluster
 consonants in the singular forms; if one of the two consonants is labial, then the

epenthetic vowel is [u]. When one of these consonants is coronal, [i] surfaces as the

epenthetic vowel. This is a second piece of evidence which reflects the fact that syllable

structure in Omani Arabic is simplex and clusters at the margins of syllables are

disfavored and broken up by epenthetic vowels. Because the broken plural shape requires

four consonants to fill the shape CVC.CVV.C, I propose that an epenthetic In1 is inserted

to fill the final consonant position. Despite the fact that an extra consonant is added in the

broken plurals' shape, the inserted consonant always fills the final C position, allowing

other C positions in the broken plural shape to be filled contiguously by the consonants

from the singular forms. The inserted I d can be understood as a remnant from the 'dual'

in the classical Arabic's dual marker [un, genitive] or [an, accusative] (Andrew Rippin,

personal communication; Jayakar, 1889: 659). Glover (1988) lists the insertion of In/ in

front of pronominal clitics in the participles of Omani Arabic like in [ka:tb-in-ha]

'written it' as one of the unique characteristics of Omani Arabic. Shaaban (1977)

considers this In/ as a remnant of nunation5 from classical Arabic.

In forms (12), the singular forms only have two consonants and a long vowel in

between them, shaping into CVV.C. When these shapes form broken plurals, they map

onto the shape CV.CVV.n. The first vowel position is filled by a fixed vowel ti1 while the

long vowel is always realized as [a:]. As is the case in forms (1 I), the final C position is

supplied by an epenthetic In/. These two shapes of broken plurals, specifically the second

5
Nunation is the marker of indefiniteness- accomplished by adding In/ following the last vowel (Guda
1988: 186)
type, are unique to Omani Arabic and there are no similar shapes attested in the literature

of classical Arabic broken plurals. To illustrate, [ko:b] 'cup' is pluralized as [?akwa:b]

and not *kiba:n in classical Arabic.

What is inconsistent about the shapes of these broken plurals is that they conflict

with the general mechanism applied to form broken plurals. They more specifically

contradict McCarthy and Prince's (1990a) proposal that a C I V C or

~ CIVC~V
or CIV: of

the singulars is mapped onto C I V C ~ V(typical

: iarnbl6 as in fact these forms map ClCz V:

in (1 1) and CV: in (12) plus an extrametrical' consonant onto CIVC~.C,V:.~

(two even

iambs Hs) in (1 1) and Ci.CV:.n in (12) and epenthesize In/ as a final consonant.

A wide range of broken plurals exhibits patterns that are notoriously inconsistent

with their singular forms. The shape of the singular forms can't be regarded as a sole

determinant for which shape of broken plural the singular is mapped onto. The following

singulars map their C.CVC or CVC.C onto C.CV:.C with consonant clusters word-

initially and a long vowel in the only syllable of these broken plurals.

13. C.CV.C- (ah) Singular forms C.(CVV).C

a. m.1a.f m.(la:).f 'folders /binders'
b. m.xa.m- ah m.(xa:).m 'brooms'
c. m.5u.t m.(su:).t 'combs'
d. 1n.ka.b- ah m.(ka:).b 'bowls'

14. CVC.C- (ah) Singular forms CCVV.C

a. xur.q x.(ru:).q 'holes'
b. hur.q h.(ru:).q 'bums'
c. har.f h.(ru:).f 'letters'

6
Footnote (c.f #4)
7
Extrametricality as a notion of the metrical theory was first brought up by Liberman and Prince (1977). It
"designates a particular prosodic constituent such as a Foot or syllable as invisible for the purpose of rule
application" (Hayes 1995).
 n.(xi:).l 'palm trees'
S.(mu:).T 'candles'

Although the singular shapes vary in that forms (13) have consonant clusters word-

initially and forms (14) have the clusters word-finally, both these diverse shapes map

onto the shape C.CV:.C where the singular forms in (13) maintain the clusters and

lengthen the vowel following them and the singular forms in (14) delete the short vowels

in between C I C and
~ collapse the consonants into one heavy syllable.

Singular forms with the shape CV.CV.C also take the same shape of broken plurals

in (13) and (14) above. Observe the following shapes of broken plurals.

1S.CVCVC Singular forms C.(CV:).C

a. sahan s.(hu:).n 'plates'
b. tabal t.(bu:).l 'drums'
c. gabal g.(ba:).l 'mountains'

The vowel contained in the only heavy syllable of the broken plurals (13-15) varies

among /a:/, /i:/ and /u:/.The quality of the vowel here is ad hoe and can't be determined

from the place features of any of the surrounding consonants. Both (1 5.b) and (15.c) have

a labial /b/ preceding the long vowel but each gets different vocalic qualities, lu:l and /a:/

respectively.

These forms have also lumped together the first two syllables contained in the first

foot at the left edge of the broken plurals into a single heavy syllable. Instead of shaping

into CV.CV:.C as expected, they map onto C.CV:.C. This conclusion is drawn through

observation of similar broken plurals in classical Arabic. Broken plurals of the Muscat

dialect surface with a consonant cluster at the left edge of these forms. For example,
Omanis pluralize [nafs] as [nfuus] and not *[nufuus] as attested in classical Arabic in the

forms in (1 6).

16. CVCC Singular forms CV.CVV.C

a. nafs nufuus 'souls'
b. qidh qidaah 'arrows'
c. lasad ?usuud 'lions'
(McCarthy and Prince 1990: 217)

There are also singulars of the shape Ce:C which map onto C.jV:.C when forming

broken plurals in the Muscat dialect of Omani Arabic. Observe the following singular and

broken plural nouns.

17.CV:C Singular forms C.(iV:).C

a. Ye:n Y.(ju:).n 'eyes'
b. se:R s.(ju:).h 'barren lands'
c. se:f s.(ju:).f 'swords'

The long mid vowel /e:/ is the only vowel in the singular forms. The singulars have

only two base consonants. The broken plural's shape is C.C*V:.C with three consonant

positions. These broken plurals surface with a glide /j/ filling the second C position and

preceding the long vowel /u:/. They show up with a consonant cluster and long vowel.

More striking are broken plurals which shape into C.CV.C, a light syllable with a

consonant cluster word-initially. The singular shape, while often CVCC-ah resists

mapping its CIVCZat the left-edge onto a CIVC;IV: as expected and the broken plural

thus surfaces with a light syllable.

18.CVCC-ah Singularforms C.CV.C

'trolleys'
'earrings'
'scarves'
d. yar.5- ah y.ra.5 'bottles'

McCarthy and Prince (1990a) propose as part of the broken plural shape

requirement that they surface with a typical iamb. This follows from the principle of

'Template Satisfaction Condition' which requires that forms satisfy the minimal

requirement of the target shape. Broken plurals in the Muscat dialect of Omani Arabic

seem to surface with any expansion of iambs. They either surface with L H or only H to

meet the 'Template Satisfaction Condition'. The exception to this requirement is shown

by the idiosyncratic behavior of broken plurals in (18) which surface with a short syllable

instead.

The final interesting shape of broken plurals involves partial reduplication of the

last consonant in the singular forms which have the ClVC2 shape. These singular forms

have only two consonants in their underlying structure. When they form broken plurals,

the second consonant spreads to fill the third consonant position in the shape C.CZV:.C~

where C2 is the same consonant.

19. CVC-(ah) Singularforms CCzV:Cz

a. xat xtu:t 'lines'
b. Sat Stu:t 'covers'
c. kum- ah kmi:m 'Omani traditional caps'
d. yub-ah ybi:b 'swamps'

The vowel between the first and second consonant of the singular form drops and

the first two consonants form a cluster in broken plurals. The vowel in these broken

plurals varies between /i:/ and /u:/.

To sum up, broken plurals in the Muscat dialect of Omani Arabic are highly diverse

in their shapes. Partially, the diversity is attributed to the distinct shapes of the singulars

from which broken plurals are derived. However, quite a large number of broken plurals
don't relate to their singular forms, either because the same singular shape is licensed to

map onto the canonical broken plural shape or else broken plurals exhibit idiosyncrasies

that can't be explained by attributing their shapes to their singular form.

The typical shapes of broken plurals have [a] filling the vowel positions in the first

foot at the left edge and [a], [i] or [u] in the final short syllable. When the final syllable of

these broken plurals is long due to the length of the final syllable of their singulars, ti:/

always surfaces in the final syllable of broken plurals. Exceptional broken plurals have a

variety of vowels which randomly fill the vowel positions of broken plurals.

1.2 Overview of thesis

This thesis is structured as follows: chapter two deals briefly with the socio-

linguistic aspects of Omani Arabic, highlighting the major languages and dialects in

Oman. More relevant for the sake of this thesis, it situates the Muscat dialect and

describes how data are obtained for this study. Research done on Omani Arabic is also

reviewed to cast light on previous studies on this variety of Arabic. Finally, this chapter

presents the necessary background information on the segmental and syllabic inventories

of the Muscat dialect of Omani Arabic.

Chapter three reviews past scholarship on the broken plural formation. Four

morphophonological models and a historical approach are investigated to report their

basic analyses and assumptions on the broken plural formation. These are the linear

generative approach, the historical and comparative model, melodic transfer, prosodic

circumscription and Optimality Theory. The presentation of these models equips the

readers with the necessary background on previous analyses so that a just evaluation

among them and the model presented here can later be made. Moreover, the chapter
shows how broken plurals in the Muscat dialect relate to broken plurals of classical

Arabic.

The fourth chapter presents the basic theoretical assumptions the research hinges on

for its crucial arguments and analyses. It offers a description of the main principles of

Optimality Theory, a constraint-based theory of Grammar, and introduces the major

constraints used in the analysis of the diverse shapes and invariant vocalism of broken

plurals. This thesis adopts two families of faithfulness constraints couched in

Correspondence Theory (McCarthy and Prince 1995). These are Output-Output

Correspondence (Benua 1995; 1997) which demands identity between related

independent words and Positional Faithfulness Constraints (Beckrnan 1998) which target

special positions in a word as having unique identity requirements. It also relies for its

pivotal assumptions on Generalized Template Theory (GTT) in which there are no

templatesper se (McCarthy & Prince 1994a, 1995, 1999; Urbanczyk 1995, 1996a &b;

Ito, Kitagawa, & Mester 1996; Gafos 1996 & 1997a; Spaelti 1997 among others). This

theory does not entail reference either to the resultant CV-shape or to the type of foot

structure of broken plurals. Despite the fact that an iamb is realized in broken plurals, this

thesis assumes that interaction of constraints of Universal Grammar yield the outcome

shapes. Therefore, no template is needed.

Chapter five includes a formal analysis of the shapes of broken plurals using the

framework of Optimality Theory (Prince and Smolensky 1993; McCarthy and Prince

1993a & b). The analysis adopted in this chapter builds on proposals made by McCarthy

(2000). The analysis first presents the basic arguments by elaborating on the technical

details of McCarthy's (2000) analysis and analyzes the typical shapes. It then extends the
same analysis to capture forms with default glides, medial geminates and even iambs.

Finally, it explores the exceptional shapes of broken plurals.

Chapter six introduces two distinct types of fixed vocalism characterizing the

formation of broken plurals in the Muscat dialect of Omani Arabic. It reviews past

approaches to vocalism in broken plurals of classical Arabic (McCarthy and Prince

1990a). It also discusses how Optimality Theory has recently offered an adequate

analysis to capture fixed segmentism (Alderete et a1 1999). Finally, it analyzes the

phonological fixed vocalism in relation to the basic tenets of Optimality Theory and

*Place Markedness constraints.

Chapter seven addresses how this research ties in with previous work done on

Optimality Theory and research on Arabic in general. It also demonstrates the basic

findings and conclusions made from the study of broken plurals in the Muscat dialect of

Omani Arabic.
 CHAPTER TWO

Overview of Omani Arabic

This chapter presents background information on Omani Arabic, which is basically

a descriptive sketch of both the segmental and syllabic inventories of the dialect of

Arabic spoken in Oman. More preliminarily, the discussion here acquaints readers with

some of the socio-linguistic aspects of Omani Arabic, highlighting the main languages

and dialects existing in Oman up to this point in time and situating the data used in this

study. This chapter also reviews key work done on Omani Arabic, enumerating the basic

aspects of these studies and their findings.

2.1 Genetic Affiliation

As is widely known, Omani Arabic (OA, henceforth) is a daughter dialect of the

mother language 'Arabic', a Semitic language which belongs to the Afro-Asiatic family.

It specifically relates to the Central South Semitic family (Watson 2002; Rose 1997).

OA is spoken in different areas, including the Sultanate of Oman, Kenya, Tanzania

and parts of the United Arab Emirates. There are about two million speakers of Omani

Arabic in Oman alone.

One might ponder the existence of Omani Arabic in African areas, but looking at

the history of Oman elucidates this puzzlement. During the seventeenth and early

eighteenth centuries, Oman largely extended its territories to huge parts of Africa, taking

over the coastal city of Zanzibar which had become a second capital of Oman for a period

of time. This results in the migration of Omanis attracted to the clove and rice plantations

of the Island of Zanzibar in East Africa (Glover 1988). Vestiges of the Omani culture and
 language can be traced in Zanzibar to this day. People in Africa still preserve the Omani

costume and traditions.

2.2 Socio-linguistic Profile

2.2.1 Dialectal variations and languages in Oman

Oman is a linguistically rich country. There are plenty of languages and dialectal

variations of OA spoken across its lands (Holes 1989, Glover 1988, Shaaban 1977). For

instance, in the Southern part of Oman, Omanis speak Jabbali, Mehri and Harsusi

languages8. Rose (1997:3) refers to these languages as 'South-Arabian' and points out

that they are spoken in Yemen as well; Oman and Yemen share southern borders.

The northern part of Oman is a different story. Due to the closeness of this part to

Iran, Omanis there are exposed through trading to ~ u m z a r i(Indo-Iranian,

~ western)

which is widely spoken in Musandam and ~hasab".

Swahili, Lawati and Belushit Balushi are minority languages in Oman. There are

quite a large number of speakers of these minority languages (about 382,000 according to

the1993 census). The minority groups of these languages are assimilated Omanis of

different origins. To illustrate, the Lawati group composes ~ a h a r i n a " merchants of

Persian descent who are Shiite Muslims. The Swahili group migrated from Africa during

the time when Oman and Africa had strong ties and merged as one country (Glover

1988). The Belushi group constitutes the largest minority in Oman (Shaaban 1977). They

- -- -

8
An extensive body of research is done to explore these Ethiopian Semitic languages. The most
comprehensive of which is Johnstone's dictionary of Harsusi (1977) and Jibbali (1983) containing
enormous number of forms used in Omani Arabic.
9
Jayakar (1903) publishes a description of the Arabic spoken in Musandam supplemented with notes on
Kumzari.
I0
Musandam and Khasab are cities in the northern part of Oman.
I1
Prochazka, T. (1981) does an interesting research describing the relation between Shi'i dialects of
Bahrain and Bahraini (Sunni) dialects and their relevance to the Arabic spoken in Ristaq, Oman.
were brought to Oman during the reign of Sayyid Saeid bin sultan1*and have continued

to settle in Oman ever since.

Two major varieties of Arabic are in use in Oman today: the Modem Standard

Arabic (MSA)'~and the spoken Arabic referred to as Omani Arabic (OA). MSA is the

language of mass media, official ceremonies and religious and public speeches (Shaaban

1977). More significantly, it is the medium of instruction in Omani schools, colleges and

educational institutions.

OA, on the other hand, is the native spoken language of Omanis. It is diverse in its

phonology, morphology, and utilizes different lexical terms (Shaaban 1977). Shaaban

makes interesting observations with regard to the Arabic spoken in Oman. He notices that

the linguistic interactions with minority groups as well as expatriates have left traces in

the speech of Omanis, especially in the vocabulary (Shaaban 1977: 11). He reports some

foreign words in the speech of young Omanis of Muscat. Foreign words include [du:bi]

'washer'-originally a Hindi word and [ko:S] 'shoes'-originally a Persian word (Shaaban

1977:13)14. Glover (1988: 17-18, 90-91) also lists some of the technical foreign terms

that are in wide use in the spoken Muscat dialect. For example, she reports [Rafi:~]

'office' borrowed from ~ n g l i s h ' ~ .

2.2.2 Situating Muscat dialect

The data in this study only pertain to OA spoken in the Sultanate of Oman with a

special emphasis on the dialect spoken in Muscat, the capital of Oman. More specifically,

I2
Sayyid Saeid bin Sultan is a ruler of al-Saeid family, the current royal family in Oman.
''Glover (1988) claims that a better term for this variety of Arabic is 'Literary Arabic'.
I4
For more examples of foreign words in the speech of Omanis, readers are referred to Shaaban (1977)
who provides a comprehensive lists of these words.
l 5 Glover (1988: 16, 90-91) lists some of the interesting foreign terms observed in the speech of Muscat
people.
the data represent the dialect of Arabic spoken by Omanis who are native to Muscat and

do not belong to any of the minority ethnic or religious groups found within Oman.

Different Islamic sects exist in Muscat; however, this study focuses on the speech of

1badhiI6people who relate to Muscat.

Shaaban (1977) classifies dialects in Oman into two major types: coastal and

interior. He considers the Muscat dialect to be coastal due to the strategic location of

Muscat, overlooking the Gulf of Oman. The Muscat dialect reflects the speech of

educated people. It also demonstrates the impact of modernization and education on the

speech of young generations. Muscat is exposed to expatriates from other Arab and Asian

countries like 1ndiaI7 more often than other areas in Oman. The influence of these

expatriates on the speech of Muscat people is noticeable since traces of expatriates'

speech do not manage to insinuate its way deep into the country.

Aspects of the distinction between the Omani Arabic spoken in Muscat and other

varieties of OA encompass differences in phonology and lexical words. As reported by

Shaaban (1977) and Glover (1988), the uvular stop /q/ is preserved from classical Arabic

in the speech of Muscat people as opposed to the people in the interior who pronounce it

as /y/. Moreover, while /g/ and /j/ are alternates in the speech of Muscat people, only / g /

is observed in the interior speech.

2.2.3 Data of study

The data in this study document the shapes of broken plural nouns observed in the

speech of educated Omanis in the capital of Oman, Muscat. They are generated from the

--

16
lbadhi is one of the Islamic sects. Most Omanis are lbadhi people.
17
A very pervasive Hindi word observed in the speech of Muscat Arabic is [bannad] 'close'. This word has
been completely assimilated to Omani Arabic that one hardly doubts it is native Omani.
researcher's native knowledge of the language. It is important to note that the researcher

was born and brought up in Muscat. Because the researcher has lived in Muscat since

birth, the researcher's speech is assumed to be a plausible representation of the speech of

contemporary Muscat people.

Anticipating the linguistic change Omani Arabic underwent during and after the

time of his study, Shaaban (1977) predicts a variety of OA moving from the interior of

Oman more towards Muscat speech, saturated with learned forms. This study discloses

how the dialect has changed since Shaaban did his research on the Muscat dialect of

Omani Arabic in 1977.

In 1988, Glover already notices that Arabic-speaking Omanis of regions outside the

capital are likely to speak a fonn of Omani Arabic that resembles to a large extent the

Muscat dialect especially when these speakers address urbanized people.

2.3 Previous research on Omani Arabic

Despite the paucity of research on the theoretical aspects of Omani Arabic, it has

been pinpointed in a number of socio-linguistic studies that Omani Arabic exhibits

unique phonological and morphological variations that set it apart from other dialects of

Arabic spoken in peninsular Arabia (Webster 1991, Holes 1989; Glover 1988; Broket

1985). In his classification of the dialects of the peninsular Arabia, Johnstone (1967: 1-3)

treats Oman as a single dialect area. Given the fact that Oman has varied topography

which includes massive deserts, impassable mountain ranges and long coasts (Holes

1989: 447), a considerable degree of dialectal variations can reasonably be expected.

These variations demonstrate that OA should be considered a distinct dialect of its own.
 Holes (1989) successfully illustrates the diverse nature of Omani Arabic by

presenting a number of morphological and phonological variables distributed across

Oman. He discusses the social and geographical distinctions of a number of phonological

and morphological tendencies characterizing the speech of Omani Arabic. He neatly

supplements his classification of the Omani areas and the dominant speech variant with

thorough discussions and arguments. He assumes that there are basically four dialectal

groups in Oman. The primary division is between Hadhari (modem) dialects and Bedouin

dialects. Each of these core divisions is further classified as H I , H2 and B1 and B2, H

and B standing for Hadbri and Bedouin respectively. This classification relates to

Shaaban's division of the dialects in Oman since it associates the modern dialects of

Oman with the coastal areas and the interior ones with the Bedouin.

Broket's (1985) monograph constitutes a detailed glossary of the technical and

agricultural terms used in the Batinah coastal town of Khabura. A meticulous reader of

Broket's monograph would notice that Broket's main purpose is to survey and

characterize the speech patterns of his informants whom he manages to classify into

groups according to place of family origin. It goes without saying that Broket's study

reflects the social life of Omanis in Khabura because the coastal, agricultural and 'crafts'

terms observed in Khabura are highly reflected in the speech of Khabura people.

Early research on Omani Arabic is mainly attributed to Jayakar (1889)" and

Reinhardt (1 894)19 who initiated efforts into the exploration of Omani Arabic. Jayakar

constructs a concise English1 Omani Arabic dictionary of about 600 words and 320

18
Jayakar was an officer in the British army who wrote about the language and culture of the Omani
people.
(9
Carl Reinhardt came to Oman with an expedition from the Oriental Society of Berlin at the time when
Oman is trying to establish bilateral relations with other Arab countries (Shaaban 1977: 21).
proverbs. He notes all the peculiarities of Omani Arabic. Although his work is regarded

as valuable and mostly accurate, he dispensed with the agricultural vocabulary and makes

no division between inland and coastal dialects (Broket 1985:l). Shaaban (1977)

evaluates Jayakar's work as a haphazard description of the phonology and morphology of

OA, ignoring the dialectal variations reflected in this variety of Arabic.

Reinhardt (1894), on the other hand, focuses on the inland dialect of Wadi Beni

Uarus2' providing a thorough grammar followed by about a hundred pages of

transliterated texts and two hundred proverbs. Because this study is more systematic as it

describes a single dialect, it proves a good reference for future Omani dialect studies.

Shaaban's (1977) study of the phonology and morphology of Omani Arabic

constitutes a hallmark in the systematic study of this variety of Arabic. It extensively

elaborates on the morphological and phonological processes that verbs and nouns have

undergone due to social and political changes. He introduces the language in historical,

social and linguistic terms, outlining the peculiarity of this dialect of Arabic when

compared to other dialects in the region.

Shaaban (1985) investigates the phonological and morphological changes that have

taken place in the verbs of Omani Arabic in a period of about one hundred years. He

basically compares the alternations between the perfective and imperfective forms of the

tri-consonantal verbs. He argues that the opening up of Oman, increased contacts with

other countries, and the spread of education have conspired to introduce to Omani Arabic

the simplification and regularization tendencies observed in the speech of the educated.

His study falls into two major parts: the variations in the perfective and imperfective

Wadi beni Kharus is part of the Rustaq region in Oman, according to the current regional division of
Oman.
forms of the verbs as cited by Reinhardt (1894). The second part deals with the modem

Omani Arabic. It also discusses the factors contributing to the regularization process,

Glover (1988) describes the phonology and morphology of the Omani Arabic

spoken in Muscat. Based on fieldwork done in the old city of Muscat, her study

highlights the basic phonological and morphological characteristics that distinguish this

dialect from other varieties of Arabic. For example, she studies the influence of adjacent

consonants on short vowels, vowel syncope, stress rules and syllabic alternations, to

mention a few. The linguistic frameworks applied in Glover's (1988) work are lexical

phonology and the non-linear approach of non-concatenative2' theory proposed by

McCarthy (I 979).

websteZ2 (1991) offers a description of the ~ l - w a h i b a ~Bedouin

' dialect of Oman,

highlighting the linguistic and ethnographic context of South-Arabian Bedouin dialects.

His previous familiarity with Al-Murra Bedouin of Saudi Arabia enables him to draw the

basic distinctions between the Bedouin of Oman and Saudi Arabia. He recognizes that the

dialects of present-day Oman share common features with both eastern Arabia along the

Gulf littoral of the United Arab Emirates and Arabs of Hadhramaut, Yemen (Webster

1991: 473). Depending on the classification offered by Holes (1989) of the dialects of

Oman. Webster classifies Al-Wahiba Bedouin dialect of Oman as a B2 dialect. He

justifies this division on the basis of some dominant phonological features such as the

lack of velar fronting or affrication of /q/ exhibited in B1 dialect. He further illustrates

21
As proposed by McCarthy (1979), non-concatenative morphology addresses root and pattern languages
that express morphological distinctions by imposing a definite shape in which consonants and vowels are
interleaved together rather than the typical fixed aftix attached to words to signal meanings.
22
Webster participates in a multi-disciplinary study of the Wahiba sands of Oman organized by the Royal
Geographical Society, London and the Diwan Royal Court, Muscat between 1985 and 1987.
23 The Al-Wahibi tribe exists in Oman up to this time and resides in East and North of the main sand desert
in Oman. Other tribes co-existent with Al-Wahiba are Al-Amr, Al-Buiesa, Hikman and Hishm (Webster,
1991: 473)
some interesting features which don't exist in B2 dialects, among which is the

palatalization of /k/ in front vowel environments in forms such as [rikJab] 'camels'

Webster is also able to spot the major features of the social life of Al-Wahiba Bedouin in

Oman as illustrated in the scripts he makes with an informant from this tribe.

2.4 Discussion of the sound system

2.4.1 Segmental inventory

The discussion below introduces the consonant and vowel inventories in the Muscat

dialect of OA. It also sketches a description of the syllable structure of this variety of

Omani Arabic.

2.4.1.1 Consonant inventory

Following mainly the International Phonetic Alphabet (IPA) transcription, the chart

below displays the consonant inventory in the Muscat dialect of OA. The total number of

consonants is 27 sounds including the glottal stop and emphatic segments. Glides, nasals

and liquids comprise two sounds each.

(1) Consonant inventory

bi- labio- dental alveolar palatal velar uvular pharyngeal glottal
labial dental
Stops b

Fricatives f

Nasals m
Liquids
Glides w

Segments represented by a dot underneath them /C1 are the emphatic sounds

(phonetically known as pharyngealized). The consonant Id/ which is present in classical

and other dialects of Arabic is missing from this dialect. Omani Arabic lumps /a/ and Id/

into one; both are represented by 141 (Shaaban 1977; Glover 1988). Hence, Omani Arabic

has three coronal emphatic sounds, the dental 141, the alveolars It/ and IS/.According to

Shaaban (1977), all the non-emphatic sounds with the exception of /r/ have emphatic

allophones governed by the rule of emphatic spread which he allots a whole chapter in

his dissertation (1977) to discuss. The emphatic I!! argued to be present in some dialects

of Arabic occurs only in one form: [allaah] 'God' (Shaaban 1977; Glover 1988) in OA.

Similar to the dialect of Egyptian Arabic, the voiced velar stop /g/ is more

pervasive in Omani Arabic than the affricate I$ which characterizes most varieties of

Arabic like Gulf Arabic, Lebanese, Jordanian and Syrian Arabic. Shaaban (1977) argues

that the preference of this consonant is attributed to the speech of an old tribe of Arab

called Azed of Yemen which settled in Oman a hundred years ago and to which most

Omanis belong. However, the exposure of Omanis to other dialects as well as the

extensive interaction with foreigners admit the alveopalatal 41to freely co-occur with the
velar /g/.

Shaaban (1977) and Glover (1988) note that /x/ and /y/ are u v u l a r ~in~ ~
Omani

Arabic which correspond with x and u occurring in the inventory of classical Arabic and
pervasively in other dialects of Arabic. In current Omani Arabic, these back consonants

-- -- --

24 Although Shaaban (1977) and Clover (1988) consider these back consonants as uvular~, their
transcription of these sounds is consistent with their transcription as velars in this thesis.
are basically velars with a little tendency towards the uvular in some forms. Little trill is

produced while articulating them, but they aren't purely u v ~ l a r s ~ ~ .

Although /p/ and /v/ are not phonemes in Arabic dialects, they occur as allophones

of /b/ and If1 which occur before voiceless and voiced obstruents respectively. This

observation also accords with Shaaban's discussion of the consonants in Omani Arabic.

(2) Allophonic variations of If1 and lbl

a. Iwafdl [wavd] 'delegation'
b. /jahbt/ ljahpt] 'descend'
(Shaaban 1977: 42)
Moreover, [rj] occurs as an allophone of In/when followed by /k/ and /g/ as the following

examples illustrate.

(3) Allophonic variations of In1

a. /nkarax/ [rjkarax] 'bumped into'
b. lngarahl [ ggarah] 'injured'

2.4.1.2 Vowel inventory

In addition to the three fundamental vowels assumed to be in the phonemic vowel

inventory of Arabic (these are /a/,/ul and /i/), Omani Arabic has long mid phonemic

vowels (Shaaban 1977; Glover 1988). These mid vowels are /e:/ and lo:/. They result

from pronouncing words that underlyingly have diphthongs. Shaaban (1977) and Glover

(1988) assume the round mid vowel to be lo:/; however, this study proposes that both mid

round vowels lo:/ and 1x1 are possible and can occur in free variation (John Esling,

personal communication). However, to be consistent with previous research done on

25
This conclusion is primarily made by a phonetician observing how the researcher produces these sounds
In her normal speech.
Omani Arabic and particularly the Muscat dialect, lo:/ is used in the transcription of this

vowel in the data of this study.

(4) Long mid vowels in Omani Arabic

a. /jawnl/ bo:m] 'day'
b. /mawt/ [mo:t] 'death'
c. /Yajn/ [Ye:n] 'eye'
d. /bajt/ [be:t] 'house'

The phonemic vowel system in Omani Arabic is represented in the following chart.

(5) Phonemic vowels

ili: u/u:

da:
Following Shaaban (1977) and Glover (1988), there are thus three short phonemic

vowels /a/, /u/ and /i/ and five long ones /a:/, /u:/, /i:/, /e:/ and lo:/. These vowels exhibit

allophonic variation as follow. These allophonic variations represent a common pattern of

centralization which occurs in closed syllables in Arabic.

1. In closed short syllables, Iil is lowered to [I] as in /misgid/+ [mrs.g~.d]

'mosque'.
2. /a/ raises to [A] and [a] in short, unstressed syllables and basically in the
environment of emphatic (pharyngealized) sounds as explored below.
3. /u/ becomes [u] in closed and short syllables as in Ikurfaj-ah/ -t [kur.faj-ah]
'beds'.
The following sketch provides a succinct description of the position of the tongue when

articulating the fundamental short phonemic vowels in Omani Arabic.

(6) Description of vowels

/a/ low tongue position, open jaw
/u/ raised, back tongue position
/i/ raised, front tongue position
 Vowels often tend to retract (lower) especially in languages that have pharyngeal

sounds. OA has a lot of pharyngealization which retracts vowels after pharyngeal and

emphatic sounds (John Esling, personal communication; Shaaban 1977; Glover 1988).

Most of these vowels can retract to schwa or further back to [a]26 when followed by a

pharyngeal or emphatic sound.

(7) Words with vowel retraction

a. Imatbaxl [matbax] or [matbax] 'kitchen'
b. Imal7aq-ah/ [mal?aq-ah] 'spoon'
c. lhadi:q-ah/ [hadi:q-ah] 'park'
d. lhilml [helm] 'dream'

Phonemic short vowels in OA contrast for length. Observe the forms in (8).

(8) Vowel length contrast

a. Qarab 'hit' 3a:rub 'person who hits'
b. h u ~ 'dirty' hu:s 'rewind, verb'
c. 7 a ~ i r 'afternoon' 7a~i:r 'juice'
d. mara:sim 'artists' clubs' mara:si:m 'royal decrees'

2.4.2 Syllable structure

Reference to syllable structure is indispensable in explaining the shapes of broken

plurals in the Muscat dialect of Omani Arabic. This section sketches a description of the

syllable structure, highlighting the basic syllable shapes and making generalization of

syllabification. The Muscat dialect of Omani Arabic has three basic syllable types.

26
The symbol beneath the schwa phonetically represents lowering of these vowels
 OA bans onsetless syllables as the three basic syllable shapes all surface with

onsets. There is a process of a default glide insertion to resolve onsetless syllables in the

broken plural formation,

(10) Default glide insertion

a. salf-ah sawa:11f 'gossips'
b. Ja:raT Jawa:raT 'roads'
c. Ji:ta:n Jaja:ti:n 'devils'
The syllable shape CVC is heavy non-finally but light finally (McCarthy 1979).

Thus, the final C is said to be extrametrical and does not have weight in word final

position. Thus, the assignment of p (unit of weight) to the final C is blocked, making the

CVC sequence light when occurring word finally; it is syllabified as CV.C.

In the Muscat dialect, what appear to be superheavy syllable shapes such as

CVVC and CVCC most often occur word-finally in which the final C is weightless. This

exempts the final consonant from participation in weight. Thus, the syllable always has

two morae. However, in some rare singular forms, superheavy syllables (syllables with

three morae) show up as the first syllable in the singular forms as seen below. This shape,

however, shortens in broken plurals to confirm to the restricted bimoraic syllable. The

long vowel in the first syllable of these forms results from merger of /a/ and /w/

underlyingly to produce the long mid vowel phoneme lo:/, as shown in forms (4) above.

(11) Superheavy syllables of the singular forms

Sing. PI. Gloss
a. m0:x.r- ah mawa:xar 'noses'
b. m0:q.Y-ah mawa:qaY 'grinders'

Although consonant clusters occur at the left and right periphery of words in Omani

Arabic, the peripheral consonants are considered extrametrical and syllable shapes are

assumed to be simplex. Glover (1988: 60) draws a similar conclusion regarding the basic
 syllable shapes of Muscat Arabic. She attributes the clusters at the left-edge of words to a

phonological rule of short vowel deletion which deletes the short vowel in the first

syllable, resulting in a cluster at the left boundary of words. She further illustrates that

this rule only applies to short vowels in the first syllables as clusters are never attested at

b. CCV:
-
margins of syllables word medially.

(12) Syllabification of clusters word-initially

a. CCV ktub [k.tu.b]
hru:z +[h.ru:.z ]
'book-pl.'
'curing spell-pl.'

Extrametrical elements are only found at word edges (left and right edges), so this

supports the treatment of the initial C in CCV words as extrametrical. Intervocalic single

consonants are syllabified with the following vowel while with intervocalic consonant

clusters of two, the first belongs to the preceding syllable and the second makes up the

onset of the following syllable (Shaaban 1977). Words with a CVCC shape are

syllabified as CVC.C, with the final consonant extrametrical.

The absence of CCCV pattern from medial and final positions is attributed to

syllabification well-formedness which bans consonant clusters at the syllable margins and

restricts the first C in CCV to close the first syllable and the second C to be the onset of

the following syllable. Thus, a CVCCVC sequence is syllabified as CVC.CV.C as

illustrated below2' and there are no VCCCV sequences as complex syllable margins

aren't permitted.

(13) Syllabification of intervocalic clusters

a. xan.ga.r * xa.nga.r 'dagger'
b. mar.ha.m * ma.rha.m 'lotion'
c. mar.ga.Y * ma.rga.T 'reference'

27
For more examples of how this shape of words is syllabified, readers are referred to Glover (1988: 59)
 Like other dialects of Arabic, the Muscat dialect of Omani Arabic has a broad

distribution of geminates in its surface forms. More relevant for the sake of this thesis are

broken plurals with geminates. McCarthy (2000) notes that when broken plurals are

mapped from singular forms with medial geminates, they preserve the length of the

singular consonants and often create long-distance geminates (these are geminates

separated by a vowel). This stems from the tendency of maintaining the same quantity of

root geminates in surface forms (Gafos 2003). Geminate codas are heavy and bear a

mora. The weight of gemination needs to be realized in broken plurals if they are derived

from singulars with geminates. McCarthy (2000) claims that although the geminates are

preserved from the singular forms, they become 'long distance linked structure' p. 178. In

the words below, each geminate consonant belongs to a different foot.

(14) Forms with geminates

Sing. P1. Gloss
a.eal.la:.g- ah (a.la:).(i:).g 'refrigerators'
b. kar.ra:.s- ah (ka.r:).(ri:). 'drawing pads'
c. Sub.ba:.k (Sa.ba:).(bi:).k 'windows'
d. duk.ka:.n (da.ka:).(ki:).n 'shops'

McCarthy (1998) and Gafos (2003) have done elaborate research to prove that

geminates can't occur in the edge of syllables. Thus, geminates at the syllable margins,

o[C,C, or C,C,], in which geminates (similar consonants) belong to the same syllable and

occur at the periphery of this syllable, is hence impossible to surface in Arabic. That is

why the parsing of adjacent geminates licenses the first geminate as a coda to the first

syllable and the second as an onset to the second syllable.

 CHAPTER THREE

Literature review on broken plurals

The complications displayed in the broken plural formation of Semitic languages

have captured the interest of linguists who get inspired in proposing different models to

account for this interesting word formation. The broken plural formation is definitely a

major phenomenon which has received considerable attention in the literature of Semitic.

Although it constitutes a highly complicated word formation, it is considered, by and

large, to be the most productive type of plural formation in Semitic languages (Levy

1971, McCarthy and Prince 1990a, Abd-Rabbo 1990, Abu-Mansour 1995, among

others).

Past work in the broken plural phenomenon includes Levy and Fidelholtz (1970),

Levy (1971), Wright (1971), Benhallam (1980), Hammond (1988); McCarthy and Prince

(1990a); Abdo-Rabbo (1988), (1990), Ratcliffe (1998); (1988), Kager and Zonneveld

(1999), Lombardi and McCarthy (199 I), Kager (1993), Abu-Mansour (1995), and more

recently McCarthy (2000), among others. The purpose of this chapter is to review these

works and highlight their basic premises and findings.

For the sake of this thesis, broken plural formation is reviewed from five basic

frameworks. These include the transformational-generative approach, the historical and

comparative model, melodic transfer, prosodic circumscription and Optimality Theory.

My reason for exploring these frameworks is their wide acceptance as distinguished

approaches in addressing a range of phonological and morphological phenomena. Before

embarking on the detailed review of these approaches, I show how Omani Arabic broken

plurals relate to classical Arabic broken plurals.

3.1 Omani Arabic Vs. classical Arabic broken plurals

Although a large portion of broken plural shapes in the Muscat dialect of Omani

Arabic resemble broken plurals in classical Arabic, Glover (1988) recognizes a few

irregularities in Muscat plurals. She notices alternations between the glides /j/ and /w/ in

the singular + plural mapping. For example, she notices that the second consonant /jl in

singular forms such as [lajsuw] 'women's head cloth' becomes /w/ whereas the final

consonant /w/ turns into /j/ after a short vowel /i/ as in the broken plural [lawa:si]. This

alternation of /j/ and /w/ in the singular -+ plural mapping is interesting.

In classical Arabic, there are two major types of broken plurals: plurals of paucity

and plurals of multiplicity1 multitude. Plurals of paucity denote the number from two to

ten while plurals of multitude encompass numbers which are more than ten. In the

Muscat dialect of Omani Arabic, such a fine distinction is not made (Jayakar, 1889).

Instead singular nouns can either take the 'sound plural' expressed by normal suffixation

of -aat 'feminine' and -uun 'masculine' to singular forms, or broken plurals with no

reference to the number of items. Abd-Rabbo (1988) notes that many dialects of Arabic

don't distinguish between plurals of paucity and multiplicity. His general observation is

that broken plurals are usually formed to express multitude in a wide range of dialects of

Arabic.

3.2Transformational generative framework

Earlier scholarship on broken plurals in Arabic explores this interesting

phenomenon within the linear transformational-generative framework. Research on

broken plurals within this model dates back to 1970s and includes works by Brame
(1970), Levy and Fidelholtz (l970), Levy (1971), Benhallarn (1980), Abd-Rabbo (1990),

and finally Abu-Mansour (1995). According to Levy and Fidelholtz (l970), Levy (1971),

and Abd-Rabbo (1990), the plural of any particular noun is predicted on the basis of the

phonological shape of its singular form, with direct reference to its semantics. These

semantic features are encoded in linear rules such as [radical] and [derived] among others

(Abd-Rabbo 1990). For example, singular forms which are characterized as being derived

from tri-consonantal roots in Arabic are assumed to take a particular plural shape which

is different from the broken plural shape of singulars that aren't derived.

Levy (1971) proposes two types of rules to account for the broken plural formation:

basic and minor rules. The difference between these rules centres on the applicability of

the basic rules to account for a broader range of forms than the minor rules. The

structural configurations of these rules are annotated with features such as [-Fern,

+Rational] and [+Adjective, colour or defect], restricting the application of these rules to

forms having these semantic and morpho-syntactic specifications. The major rules of

Levy are ordered linearly.

Abu-Mansour (1995) stipulates, based on the diversity of plural formation in

Arabic, that plural assignment is based on ". .. phonological, morphological and semantic

properties of nominals, combined with a certain amount of arbitrary marking" (1995:

326). He chooses a set of morphological rules to represent the generalization that reflects

the overall broken plural mechanism in Arabic. His account requires that three

morphological rules be applied.

(1) Abu-Mansour's linear rules of broken plural formation

(A) Feminine suffix deletion

{-x2')+ 0 I [plural]
(B) Infixation (after either the second or third consonant of the stem)
0 - + a : / # C V C - C ...
[+ plural]

(C) Raising
V+ [+ hi] I # CVCa:C - C...#
[+ plural] (Abu-Mansour 1995:326)

As he extends these three basic rules to account for nominals with biliteral roots

and loan words, he incorporates other independently motivated rules in the language such

as C? metathesis and vowel deletion rules to complement his rules and to account for

further phonological changes in broken plurals.

Ratcliffe (1988) proposes a two-level morphology to account for the Arabic plurals

pretty much the same as the two levels29assumed to derive inflectional and derivational

affixes in English. He examines a corpus of about 27 plurals and classifies them into two

levels. Broken plurals belong to level one as they involve 'internal' change while 'sound'

plurals belong to level two.

Abd-Rabbo (1 988) argues that the variations in the vowel combinations observed in

the triliteral broken plurals are the result of homophony avoidance. He claims that

blocking the productive [a] in some of the broken plurals results from the fact that they

would coincide with words with the same shape but with different meaning in Arabic. An

28
Abu-Mansour uses X to stand for the suffix [-at] attached to the singular forms. This suffix is a marker of
feminine gender in classical Arabic which gets deleted in the broken plural formation.
29 Level 1 affixes impose a change to the bases they are attached to. For example, in English, they might
cause stress shift or trisyllabic shortening while in Level I1 affixes are neutral. Level I1 affixes get attached
to the base forms without effecting any change.
example of homophony avoidance is the singular [mu'lmin] 'believer' whose broken

plural, if it existed, would be *[ma?aamin] which is identical to the broken plural of

[ma'lman] 'place of safety' (Abd-Rabbo 1990: 72). To avoid homophony with the broken

plural of [ma?man], [mu'lmin] takes the sound plural [mu?mm-uun130.

3.3 Historical and Comparative study

Ratcliffe (1998) presents a historical and comparative analysis of the broken plural

formation in Arabic and Semitic. He thoroughly examines the nominal morphology of

Arabic, comparing it with available root-and-pattern morphological formations prevailing

in other Semitic languages. As a diachronic study, Ratcliffe's study depends on research

on other fields such as philology, descriptive and theoretical linguistics, compiling

evidence for constructing a proto-language for Semitic languages.

Most relevant for the sake of this thesis is Ratcliffe's discussions of the Arabic

noun plural system in chapter three of his dissertation (1998) where he elaborates on the

various patterns obtained through pluralization of nouns in Arabic. Not only does he

provide a neat classification of the plural shapes, he also delves into the major problems

occurring when attempts are made to derive these plurals from other noun classes. The

discrepancies are cited when a specific noun shape maps onto quite a large number of

plural shapes that vary considerably from another singular of the same shape mapping

onto distinct shapes. He basically deals with the general mechanism of forming broken

plurals, pinpointing previous research into this phenomenon.

The derivational history of these forms differs and this may contribute to the distinct ways of forming
plural forms.
 Greenberg (1955) suggests a genetic relationship between the broken 'internal'

plurals of south Semitic and plurals expressed by internal [a] vowel in the non-Semitic

Afroasiatic languages. He establishes three common features of the internal [a] plurals in

languages throughout the Afroasiatic group. The first of these is the plural stem, which is

generally expressed by a vowel after the second consonant. The second is the continuous

appearance of the vowel [a]. The third of these is the existence of dissimilation from [a]

in the plural whenever the singulars have this vowel quality.

Ratcliffe (1996) argues that plural forms that exhibit partial reduplication occur as a

morpho-phonologically conditioned variant of the prosodically extended stem (or

internal-a). He considers this phenomenon of broken plural formation as a sporadic and

occasional marker of plurality in a large number of Semitic languages including Cushitic

and Chadic groups.

3.4 Melodic transfer (Hammond 1988)

Harnmond (1988) advocates "melodic transfer" to account for the broken plural

formation. In melodic transfer, segmental materials of the singular forms are

copiedltransferred to the broken plural shape without change in their positions.

Hammond first addresses the major problems relating to broken plurals in Arabic which

are the length of the final vowel being identical to the vowel length of the singular, a

consonant that is spread in the singular is also spread in the plural, insertion of lwl which

occurs when there is a long vowel in the first syllable of the singular and finally

truncation which occurs when the number of consonants are too many to accommodate to

the target shape of broken plurals (Hammond 1988: 254). He elaborates on McCarthy's
(1982) approach which comprises three rules: infixation rule of a VV syllable, vowel

melody which spreads the appropriate vowels and the filter rule which ensures that all

broken plurals surface with the canonical [CV.CV:].CV(V)C shape. He touches on the

problem of joining together two separate mechanisms (truncation and infixation) which

McCarthy's rules imply. He also observes a problematic process proposed by McCarthy

which basically converts the V dot into C slot in cases where /w/ is inserted.

3.5 Prosodic Circumscription (McCarthy and Prince 1990a)

McCarthy and Prince (1990a) have done an extensive research study to examine

broken plural shapes of classical Arabic. They identify the general broken plural

formation rule as an extraction of a trochee C1VC2, CIVV or CIVCzV from the singulars

and mapping it onto iamb C I V C ~ V Vin the broken plurals. This operation is dubbed

Prosodic Circumscription. As formalized by McCarthy and Prince, Prosodic

Circumscription is based on a factoring function Q (C, E, B) which parses out the

prosodic constituent C standing at edge E of a base form B (McCarthy and Prince 1990a:

226). This theory restricts the application of a morphological process to a specific

prosodic constituent (p, 0,F, etc.. .13' in a morphological word and identifies two types of

Prosodic Circumscriptional operations: positive and negative. If the morphological

operation targets the parsed portion of a word, notated as (B: Q) and known as the kernel,

then we have positive prosodic circumscription. However, if the morphological operation

targets the rest of the word, notated as (Bl 0)and known as the residue, then we have

negative prosodic circumscription.

3' Prosodic Morphology establishes the need to define templates in terms of the authentic units of prosody
which are mora fl, syllable o, foot F and prosodic word PrWd, hierarchically ordered from the smallest
constituent making up a syllable to the prosodic word consisting of all the elements of prosody (McCarthy
and Prince 1990a: 209).
 McCarthy and Prince (1990a) analyze the broken plural formation using positive

Prosodic Circumscription which extracts the trochaic foot from the singular forms and

maps it onto an iamb. It retains the weight of the final syllable of the singular which is

then concatenated to the rest of the constructed iamb.

The insertion site of the default glide is also addressed. It is assumed to follow

from the position of the long vowel in broken plurals. A long vowel in the first syllable of

the singular form such as in [xa:tam] requires the /w/ to form the onset of the second

syllable of the expanded first foot in broken plurals, thus yielding [xawa:tim]. Due to the

fact that there is only one consonant in the circumscribed foot, glide insertion serves to

provide an onset to the second syllable. When the singular forms have a long vowel in

their second syllable like in [saha:b-at], a default glottal stop /?/ is inserted in the third

syllable of broken plurals. The resultant broken plural shape is [saha:?ib] (McCarthy and

Prince 199Oa: 2 18).

3.6 Optimality Theory (McCarthy 2000)

This thesis builds on proposals made in McCarthy (2000) who shows how a range

of Prosodic Circumscription analyses can be recast into non-derivational OT by retaining

the prosodic structure of output forms. This section briefly reviews the basic assumptions

made by McCarthy (2000) about broken plurals and technical details of his analysis are

addressed in the actual analysis in chapter five of this thesis.

Although McCarthy does not provide a full OT analysis, his assumptions based

on the moraic faithfulness constraints he proposes pave the way for adopting an OT

analysis to account for the diverse shapes of broken plurals in the Muscat dialect of
Omani Arabic. As a start-off point, McCarthy (2000), identifies the major patterns

characterizing the broken plural formation such as preserving the weight of the final

syllable, the insertion sites of the extra consonants and the tendency to preserve

consonantal spreading in geminates.

He argues that high-ranking prosodic faithfulness constraints can obtain

preservation of the weight of the final syllable in broken plurals (McCarthy 2000: 174).

Moreover, he uses constraints against spreading and delinking of association lines to

capture the conservation of consonantal positions in the mapping of the singulars and

broken plurals. In his conclusions, McCarthy suggests the possibility of using positional

faithfulness constraints to obtain the identity or faithfulness effects of certain privileged

positions like 'stem-final' syllables (see McCarthy 2000: 180).

This thesis adopts McCarthy's suggestion to advance this identity by using

positional faithfulness constraints (Beckman 1998). It also builds on the assumption that

the difference between singulars and broken plurals lies in an extra mora 'affixed +p'

attached at a definite position in the singular forms to yield the final shapes of broken

plurals.
 CHAPTER FOUR

Theoretical assumptions

This thesis hinges on a number of theoretical assumptions that govern its overall

arguments and shape the final analyses. This chapter presents the major theoretical

assumptions made to analyze the shapes and vocalism of broken plurals in the Muscat

dialect of Omani Arabic.

The foremost theory it adopts is Optimality Theory, the constraint-based approach

(Prince and Smolensky 1993; McCarthy and Prince 1993a &b) in which linguistic forms

are evaluated by constraints and the optimal form (the one which minimally violates

constraints) is the output. In OT, there are two basic sets of competing constraints:

Markedness and Faithfulness constraints. In syllable theory, CV is more frequent and less

marked than CCV or only V syllable shapes. Markedness constraints, thus, evaluate

marked structures in the candidate outputs while faithfulness constraints monitor identity

in correspondent forms.

This chapter presents the basic theoretical assumptions the thesis adopts to

analyze the shapes and vocalism of broken plurals in the Muscat dialect of Omani Arabic.

5 4.1 summarizes the framework of OT. 5 4.1.3.1 introduces Correspondence Theory and
outlines the premises of two families of constraints: Output-Output Correspondence

(Benua 1995, 1997) and Positional Faithfulness Constraints (Beckman 1998). 5 4.2
introduces Generalized Template Theory (GTT) and shows how it relates to the analysis

of the shapes of broken plurals in the Muscat dialect of Omani Arabic.

4.1 Optimality Theory Framework

OT is a constraint-based theory which assumes that the actual output surfaces as a

result of resolution between a set of conflicting violable constraints through a ranking.

Constraints are ranked with respect to one another and the output violates the lowest

ranked constraints. This section and the subsequent one explain the basic components of

this theory.

Under the tenets of OT, Universal Grammar supplies a set of universal

constraints. These constraints are the same in all languages but ranked differently on a

language specific basis. They are classified into two broad categories: markedness and

faithfulness constraints. While markedness constraints penalize marked structures in the

output forms, faithfulness constraints strive to keep each stringlform consistent with its

correspondent. If the output form exhibits a change from its input, then markedness

prevails over faithfulness. This conflict between markedness and faithfulness is resolved

through a definite ranking which prioritizes either markedness or faithfulness. Through

the function Generator, potential candidates are generated from a particular linguistic

input. These potential outputs compete with the actual output (dubbed as optimal) and are

doomed because they violate high-ranking constraints by the function Evaluator which

checks each candidate against the constraint set. 'Therefore, the optimal output is the one

which is more harmonic and incurs minimal violation. It violates the low ranked

constraints and satisfies high-ranking ones.

Below, the general architecture of OT as proposed by McCarthy (2002) is provided:

- Fl - p]-
(1) OT architecture

input + candidates output

(McCarthy 2002,lO)
 If markedness constraints (M) outrank faithfulness constraints (F), then any

candidate which violates M will receive a fatal violation mark indicated by (!). The

optimal candidate is more harmonic and exhibits less serious violations. It obeys M at the

expense of violating the low ranked constraints F. Solid lines between the constraints in a

tableau of ranking show valid ranking while dotted lines demonstrate that no definite

ranking between the constraints can be established. In other words, they are equal in

ranking. b is placed in front of the optimal (winning) candidate.

This ranking where M prevails over F produces an alternation from the input to the

output form because the identity of the input is changed. The output violates F.

4.1.1 Principles of OT

There are five basic principles that govern OT framework. Universality entails the

assumption that Universal Grammar provides a set of universal constraints. Languages of

the world have the same set of constraints, but they differ in whether these constraints are

active or passive and on whether they are high or low ranked. Violability holds that

constraints are violable. The usual scenario is that the competing suboptimal candidates

are excluded from further evaluation because they violate high-ranking constraints.

Along the same line, the optimal output is not totally obedient. The stipulation stressed

here is that the optimal output can violate constraints minimally and still gets selected as

an optimal output. If constraint A prevails over constraint B, then the optimal output can

violate B as long as it obeys A. Ranking ranks constraints with respect to each other;

some constraints have more priority over others and thus are highly ranked. It determines
how evaluation of constraints proceeds and what the actual output form is. Inclusiveness

entails that any set of candidates are admitted for evaluation since any form is considered

a possible output by OT. OT is proposed to solve the cumbersome serial derivations.

Parallelism guarantees that all phonological, morphological and prosodic changes

encoded in the proposed constraints are applicable simultaneously (Prince and Smolensky

1993; McCarthy and Prince 1993a; Kager 1999).

4.1.2 Markedness Constraints

In Optimality Theory (Prince and Smolensky 1993; McCarthy and Prince 1993a

&b), the grammar of any language is reduced to a ranking of a set of universal

constraints. The ranking resolves a conflict between two major constraints: markedness

and faithfulness. This section is devoted to the discussion of markedness constraints,

leaving faithfulness constraints to be explored in the coming section.

Markedness constraints evaluate marked structures in output forms. They are

violated when outputs produce marked segments or features. For example, they monitor

the syllable structure of output forms. They penalize marked syllable structures such as V

(onsetless syllable) and CCV (syllable with complex onset). Among the constraints

evaluating syllable structure are:

(3) ONSET
Every syllable begins with a consonant. (McCarthy and Prince 1990b; 1993a)

(4)* COMPLEX
Clusters are prohibited from the margins of syllables. (Prince and Smolensky 1993)

( 5 ) *PPPI~
Syllables are not trimoraic. (McCarthy and Prince 1990a, 1990b)

(6Ya ICX,
Geminates are banned from the margins of syllables. (McCarthy 1998 and Gafos 2003)
 4.1.3 Faithfulness constraints

Faithfulness constraints monitor identity between related strings. These strings

shouldn't be thought of as identical because identity is regulated by faithfulness

constraints such as DEP and MAX which militate against epenthesis and deletion of

segments or features between related strings respectively. IDENT (F) ensures that

segments in related strings have the same featural values.

These faithfulness constraints generalize to include larger constituents or units of

phonological representation. For instance, units of prosody such as mora are also

incorporated. For example DEP- p bans epenthesis of morae in surface forms. The full

picture of the effects of these prosodic faithfulness constraints is illustrated in the actual

analysis in chapter five.

4.1.3.1 Correspondence Theory

Correspondence Theory (McCarthy and Prince 1995) posits that strings of words

(input-output pairings or output-output pairings) stand in a correspondence relation. The

general schema of this theory is shown below.

(7) Correspondence Theory

Given two strings S , and Sz, correspondence is a relation R from the elements of SI
to those of Sz. Segments a (an element of S I ) and P (an element of S2) are referred to as
correspondents of one another when a R P. (McCarthy and Prince 1995: 262)

Any two strings of words have a correspondence relation. Segments or features of

one string relate to segments or features of the other string. The following sections

outline the premises of two correspondence families adopted in the analysis of broken

plurals. These are output-output correspondence and positional faithfulness.

4.1.3.1.1 Output-output Correspondence (Benua 1997)

Benua (1997) extends the correspondence reIations from Input-Output and Base-

Reduplicant (McCarthy and Prince 1995) to correspondence between morphologically

related forms in languages.

According to Benua, the identity of words in a paradigm is captured by constraints

regulating output-output relations. These constraints interact directly with other

constraints including phono-constraints (markedness, in other words) that relate to

phonology and whose crucial role is to evaluate marked structures in the output strings.

Quoting Benua, the basic assumption of output-output correspondence is that

"...phonology is sensitive to morphology because phonological identity relations hold

over paradigmatically-related words.. ." (Benua 1997: 227)

Morphological derivation (affixation, truncation, etc ...) is assumed to be of an

output-output correspondence relation between derived output and output forms in

Benua's model. The following diagram represents a simplified version of Benua's model

which is used to illustrate an output-output relation in Omani Arabic.

(8) Output-output ( 0 - 0 ) relations in (maktab] 'office, sing.' and [maka:tub]

'offices, pi.'
Output [maktab] 0-0Correspondence [maka:tub]
+
1-0 Correspondence
t t 1-0 Correspondence

luput /katab/ /ma-katabl

Output forms are linked to Input forms by an 1-0 correspondence relation and the

two words are also related by an output-output correspondence. Benua demonstrates that
any input form, which is fed into a specific grammar of any language, must result in an

output that is well formed in that language.

This proposal reflects how broken plurals in the Muscat dialect of Omani Arabic

which correspond to singulars (independent words) are regulated by the well formedness

constraints in the language. These constraints evaluate output broken plural against

accepted structures in the language. The formalism of the broken plural formation applies

to any singular forms, but it never takes place in isolation from the well formedness

canons of the language as a whole.

Ussishkin (1999) offers cogent arguments for output-output correspondence when

analyzing the denominal verbs3' in Hebrew. He shows that denominal verbs are derived

from independent forms (output nouns) in Hebrew and there is no need to refer to the

consonantal roots3'. His arguments rely on some empirical observations such as the fact

that the vowel and second consonant of the output nouns relate directly to their

denominal verbs. Moreover, consonant clusters in these nouns are also preserved in their

denominal verbs. These crucial observations lead Ussishkin to adopt an output-output

correspondence relation between denominal verbs and their output nouns.

McCarthy (2000) proposes that the correspondence relation observed between the

singular forms and their broken plurals is of the Output-Output type. McCarthy and

Prince (1990a) note some problems of the dependence on the consonantal root as the

input for broken plural formation. They illustrate " ... [Tlhe iambic plural systematically

reflects aspects of the singular that the consonantal root does not determine". Empirical

evidence lies in the total identity of the weight of the final syllables of both the singulars

32Denominal verbs are verbs which are derived from nouns.

" The general trend taken by linguists studying Semitic is to refer to the consonantal roots (typically, three
consonants) in order to analyze the internal structure o f words and how they are derived.
and their broken plurals34. Such identity suggests an Output-Output correspondence

relation between these morphological strings. They are both already existing forms in the

language. In addition, both the singular forms and broken plurals share the same

consonants with the exact order (refer to (9) below). I provide the following

representation to show that the weight of the final syllable of the broken plural is only

derivable from the output singular (not from the input stem alone). Compare the final

syllables of both the singular and broken plural which contains two morae each. This

identity supports that the input stem cannot derive the broken plurals. Moreover, prosodic

structure such as assignment of weight is only reliably present on the output.

(9) Output-Output relation between the singular and broken plural

Sing. [minSa:r] PI. [mana:Si:r]
FI F2 FI F2
I I A I

4.1.3.1.2 Positional Faithfulness

In order to capture the identity in weight between the final syllables of both the

singulars and broken plurals, reference to Positional Faithfulness constraints is made.

Beckman (1998) proposes a set of Positional Faithfulness constraints that capture the fact

that certain positions in a word are more privileged in that they maintain phonological

distinctions. The fact that these positions keep their identity unaltered suggests that they

are more privileged or prominent than other positions. Beckman (1998) outlines a set of

privileged positions sketched below:

'4 This observation is also made by Glover (1988) who briefly discusses the features of plurals in Muscat
Arabic.
(10)
a. Privileged positions b. Non-privileged positions
- Root-initial syllables - Non-initial syllables
- Stressed syllables - Unstressed syllables
- Syllable onsets - Syllable codas
- Roots - Affixes, clitics, function words
- Long vowels - Short vowels
(Beckman 1998: 1)

Beckman (1998:l) also pinpoints the possibility of considering final syllables as a

privileged position. She illustrates "Positions of phonetic prominence include stressed

syllables, syllable onsets, long vowels and possibly final syllables" (emphasis added).

McCarthy (2000: 186) also demonstrates that final-stem is a privileged position and

warrants positional faithfulness constraints to capture its special status. Hayes (1995:59)

claims "In measuring syllable weight, a word final position is likewise often a special

case" when he discusses extrametricality in relation to stress assignment. This leaves us

with three pieces of empirical evidence for the special condition of syllable final position

and supports adoption of Positional Faithfulness constraints to capture the identity of

weight of the final syllables in the singulars and plurals.

(11)Final syllable identity

a. (sak). (ki:)n (sa.ka:). (ki:). n 'knives' H--t H H= Heavy
b. (mux).ba (ma.xa:). bi 'pockets' L- L L= Light

The weight of the final syllable in singular forms is preserved when they form

broken plurals. In forms (1 I), both the final syllable of the singular and broken plural is

equal,

A crucial point to mention: weight concerns the number of morae between elements

in correspondence. So, the prosodic feature used to capture the identity in weight is w.
Beckman proposes the following Positional Faithfulness schema:
(12) I D E N T - P O S(F)
~~~O~
Let p be an output segment in a privileged position P and a the input correspondent of a.
If p is [yF], then a must be [yF].
"Correspondent segments in a privileged position must have identical specifications for
[Fl" (Beckrnan 1998%)

Another vital point which needs to be addressed is the mapping of the first trochaic

foot of the singular onto an iamb in broken plurals. McCarthy (2000) addresses this

mapping by proposing that broken plurals differ from the singular form by an extra mora

('affixed +p', using McCarthy's term) added to the second syllable of the singular forms.

This mapping can be captured by employing a positional faithfulness constraint.

The merit of adopting positional faithfulness constraints is that they check identity

in a certain position in a prosodic word. For example, the weight of the first foot in

broken plurals does not match up with the weight of the first foot of the singular form.

Positional Faithfulness constraints also address the mapping from the singular onto

broken plural straightforwardly as they show that an extra mora distinguishes the first

foot of the broken plural from that of the singular.

4.1.4 Alignment constraints

Alignment constraints demand certain edges of categories/constituents to align with

some edges of other categories/constituents. For example, in the formation of broken

plurals, they require that the right edge of syllables in the broken plural (prosodic

constituent) coincide with a morphological constituent. The general formalism of the

alignment constraints is couched in Generalized Alignment Theory (McCarthy and Prince

1993b).
(13) Generalized Alignment
Align (Catl, edge^, Cat2, Edgez) = def
V Cat, 3 Cat2 such that Edge1 of Catl and Edge2 of Cat2 coincide.

Catl, Cat2 E Pcat u Gcat

Edgel, Edge2 E {Right, Left) (McCarthy and Prince 1993b)

This definition stipulates that Generalized Alignment demands that a particular

edge of every prosodic or morphological constituent of Cat, coincide with a particular

edge of some other prosodic or morphological constituent of Cat2.

The fact that the specified vowel [a] occurs at the right edge of syllables contained

in the first foot of broken plurals as in [maka:tub] 'offices' and [mga:mar] 'incense

burners' is translated into the constraint ALIGN-R(0,[a]) which demands the right edge

of every syllable to coincide with [a].

(14) ALIGN-R(n, [a]) The right edge of every syllable coincide with the right edge of
[a].
This constraint relates to the output broken plurals as it evaluates whether [a] is

aligned to the right edge of syllables contained in broken plurals and not in the singular

forms. It is violated when the right edge of any syllable of the output broken plural form

isn't aligned to [a].

It6 and Mester (1999) refine the notion of alignment to account for strings with

shared features. In words with geminates of which the first half of the geminate belongs

to the first syllable and the second half belongs to the second syllable, the geminates are

contained in two syllables. Thus, these syllables do not have a CrispEdge because of the

shared elements between the two syllables. For example, the form [(sak),~.(ki:),2. n] has

the geminate /kl shared by two adjacent syllables (refer to the representation in (15)).
Such cases warrant refinement of alignment constraints. Therefore, It6 and Mester

propose CrispEdge constraints that successfully capture the fact that shared elements

should be contained within a prosodic constituent. In (IS), the geminate consonant kl is

shared by 01and 0 2 in the word [sakkin]. It functions as the onset of the second syllable

and coda of the first syllable. This violates CrispEdge (0) which requires the shared

segment (geminate) to be included in one syllable.

(15) Geminates violate CrispEdge

01 02

The general formalism of this constraint looks as follow.

(16) CrispEdge [Pcat]

Let /A/ be a terminal substring in a phonological representation, 3 a category of type Pcat

I
and /A/= 13 (the content of 3). Then 3 is crisp (has crisp edges) if and only if A is-a
~:'V'A(/A/=I~I=/A/=~) (It6 and Mester 1999:208)

In broken plurals of the Muscat dialect of Omani Arabic, there is a strict restriction

of [a] linking only to the vowel positions contained in the first foot and not extending to

any vowel position outside the designated foot. The constraint adopted to account for

[al's adherence to the vowel positions within a foot is CrispEdge [Ft] (following

proposals by Urbanczyk 1999). This constraint is violated by any candidate which

extends the shared elements to a position outside the foot domain

4.2 Generalized Template Theory (GTT)

One of the most pivotal theoretical assumptions made in this thesis is Generalized

Template Theory (McCarthy & Prince 1994a; Urbanczyk 1995, 1996a &b; Ito, Kitagawa,

& Mester 1996; Gafos 1996 & 1997a; Spaelti 1997) among others. GTT assumes that

templatic effects such as CV-shapes and type of foot structure are derived from

interactions of other constraints in the Universal Grammar. There is thus no need to

formulate a constraint that checks a specific foot type or CV-shape in the output forms.

In reference to the shapes of broken plurals, GTT basically stipulates that rankings

of related constraints can best generate the optimal outputs without reference to the CV

shapes of broken plurals. Furthermore, despite the persistent tendency of iambs to be

realized in broken plurals, this thesis does not formulate a constraint that monitors the

foot structure of broken plurals. Instead, ranking of well-formedness constraints along

with the general plural formalism (affix +p) derives the output forms without having to

propose a constraint like PLURAL=

IAMB.Such nicety in accounting for the diverse shapes

of broken plurals precludes stipulating an ad hoc set of constraints like TEMPL

to refer to

a variety of resultant CV shapes. It also dispenses with constraints articulating a

particular shape of foot structure.

In conclusion, this chapter has presented the basic theoretical assumptions this

thesis hinges on to offer an integrated analysis to both the shapes of broken plurals and

vowels contained in these word forms.

 CHAPTER FIVE

Shapes

5.1 Introduction

Broken plurals in the Muscat dialect of Omani Arabic exhibit enormous diversity in

their shapes. Broken plurals exhibiting diverse shapes can relate to the particular shapes

of their singulars. However, there are divergent shapes that can't be attributed solely to

the shapes of their singular forms.

In the Muscat dialect of Omani Arabic, there is a set of broken plurals which

resembles in their shapes broken plurals of classical Arabic explored in depth in the

literature (McCarthy and Prince 1990a; Abd-Rabbo 1990; Abu-Mansour 1995; Ratcliffe

1998 and McCarthy 2000, and the literature reviewed in these works). These broken

plurals form typical iambs. Examples of this type include [(ma.ta:).ba.x] 'kitchens' and

[(sa.wa:).Ta.q] 'thunderbolts'. They are dubbed as 'typical or canonical' in this study

because the mechanism applied to form them follows straightforwardly from the general

rule assumed to form broken plurals of classical Arabic.

In addition to the typical broken plurals, Omani Arabic has a wide range of broken

plurals whose shapes are vastly different from the shape of the typical broken plurals. For

example, plurals derived from singulars with a single heavy syllable and a consonant

cluster at the left edge of them like [nTa:l] 'shoe, sing.' contain two heavy syllables

adjacent to one another [(nuT).(la:).n] 'shoes, pl.'. These broken plurals are demonstrated

to follow the same mechanism assumed to map singulars onto the most common shapes.

A last group of broken plurals seem to be notoriously distinct. They do not satisfy

the minimal requirement of plural formation in that they map onto a light syllable. They
 56

resist realizing the extra weight typical broken plurals surface with. An example of this

type is [xnaS] 'trolleys'.

This chapter offers an integrated analysis to the diverse shapes of broken plurals in

the Muscat dialect of Omani Arabic buiIding on proposals made in McCarthy (2000). The

analysis first addresses the typical shapes of broken plurals in 5 5.2.1, presenting the

basic assumptions and technical details of McCarthy's OT analysis to broken plurals in

classical Arabic. The same analysis is then extended to account for the shapes with

epenthetic glides, medial geminates and even iambs in 5 5.2.2, 5 5.2.3 and 5 5.2.4

respectively. Finally, 5 5.3 explores exceptional shapes of broken plurals, highlighting

their exceptionality.

5.2 OT Analysis of the shapes of Muscat broken plurals

5.2.1 Typical (canonical) broken plurals

In the Muscat dialect of Omani Arabic, the majority of broken plurals resemble

broken plurals of classical Arabic, one of the much studied phenomena in Semitic

(McCarthy and Prince 1990a, Abd-Rabbo 1999, Abu-Mansour 1995, McCarthy 2000

among others). The following examples represent the typical shapes of broken plurals in

the Muscat dialect of Omani Arabic.

Sing. Gloss
(1). (CVC).CV.C Singular forms
a. (ma0.ba.x 'kitchens'
b. (daf).ta.r 'notebooks'
(2).(CVC).CV.C-ah Singular forms
a. (mas).ta.r- ah 'rulers'
b. (mal).Ya.q- ah 'spoons'
(3). (CVC).(CV:).C Siqulur forms
a. (qan).(di:).l 'lanterns'
b. (mis).(ma:).r 'nails'
c. (kar).(fu:).n 'knuckles'
(4). (CVC).CV Singular forms
a. (mak).si 'dresses'
b. (kur).si 'chairs'
(5).(CV.CV:).C- ah Singularforms
a. (ma.ki:).n- ah 'machines'
b. (ha.di:).q- ah 'parks'
(6). (CV:).CV.C Singular forms
'bed blankets'
'roads'
'line-ups'
'cement blocks'

Three general observations obtained by examining the shapes in (I) through (6) are

reported below. These observations are basically inspired by similar conclusions made in

McCarthy (2000).

1. The first foot at the left-edge of the singular forms which has the shapes ClVC2 in

forms (1-4) or CtVV in forms (6) is mapped into a typical iamb [C1VC2V:] in

forms (1-4) and [ClVwV:] in forms (6). In forms ( 9 , the shape CIVC2V

contained in the first iamb to the left edge of the singular forms as in (5.a)

[(ma.ki:).n- ah] is extracted from this foot and expanded into an iamb. The iamb

constructed contains a light syllable L followed by a heavy one H.

2. The weight of the final syllable of broken plurals is not affected by the mapping

of the first foot of the singulars onto an expanded foot. Thus, the weight or length

of the final syllable of both the singular and broken plural is identical (McCarthy

2000; McCarthy and Prince 1990a).

3. The position of the epenthetic glide in broken plurals is determined by the

position of the long vowel in the singular forms. In the singular forms in (6), the
 first syllable ([Sa:] in form 6.a) is long and contains just one consonant. Broken

plurals epenthesize a glide /w/ to fill the second consonant position pertaining to

the second syllable of the first foot at the left-edge in the shape CIVC2V:[Sawa:].

However, if the second syllable of the singular forms is long as in forms (5)

([ma.ki:.n- ah], broken plurals epenthesize /j/ to fill the onset of the third syllable

in broken plurals CV.CV:. jV.C [maka:jin].

McCarthy (2000) assumes that the preservation of weight in the final syllable of

both the singular and broken plural is a consequence of the constraints D E P O O ~ ' - ~(against

epenthesis of p) and M A X O O(against

-~ deletion of p).

(7) DEPOO-P For every p in outputz, there is a correspondent p in output,.

This constraint is violated when the broken plural surfaces with an extra mora

compared with the singular form.

(8) M ~ x o o - P For every p in output[, there is a correspondent p in output2.

This constraint is violated when the broken plural surfaces with fewer morae than
its singular form.
The analysis adopted in this study assumes instead Positional Faithfulness

constraints (Beckman 1998). Beckman (1998:l) pinpoints the possibility of considering

final syllables as a privileged position. She illustrates "Positions of phonetic prominence

include stressed syllables, syllable onsets, long vowels and possibly final syllables"

(emphasis added). McCarthy (2000: 186) also demonstrates that the final-stem of the

broken plurals in classical Arabic is a privileged position and warrants Positional

Faithfulness to capture its special behaviour. This leaves us with evidence for the special

35
00 stands for Output-Output correspondence relation between the singulars and broken plurals
condition of syllable final position and justifies adopting Positional Faithfulness

constraints.

The fact that weight in the final syllable of both the singular and broken plural is

identical can be captured using the following Positional Faithfulness constraints.

(9) MAXOO-SF(PI
For every p in the final o of output!, there is a correspondent p in the final 0 of output2.

This constraint militates against the loss of morae from the final syllable of broken

plurals when they are derived from singulars whose final syllable is long (length

retained).

(10) DEPOO-OF (P)

For every p in the final o of outputz, there is a correspondent p in the final o of output1

This constraint bans lengthening of the final syllable of broken plurals when their

singular forms have a final short syllable.

In order to capture the mapping from a trochee CIVC2 or ClVV with two morae

onto a typical iamb C I V C ~ V with

: three morae, a constraint against epenthesis at a

particular position is required. The position which receives the extra weight is attested to

be the first foot at the left-edge of broken plurals. Therefore, the following constraint is

adopted.

(11) DEPoo- FI (P)

For every p in the first F of outputz, there is a correspondent p in the first F of outputl.

This constraint evaluates the output broken plural and is violated when the broken

plural surfaces with an extra mora as shown in the following representations.

(12) Sing.

clTr
 Each initial trochee of the singular forms contains two morae. When broken plural

is formed, the foot gets an extra mora and creates a typical iamb C1V.CzV:.

This constraint is not introduced in McCarthy's (2000) OT analysis of broken

plurals in classical Arabic. He, however, paves the way for adopting this constraint by

assuming that the distinction between the singular and plural forms lies in an affixed p

(DEPOO-p)36 added to a certain locus in the plural forms (McCarthy 2000: 186).

The basic constraints needed to derive broken plurals are Positional Faithfulness

constraints. They include the constraints monitoring identity of weight in the final

syllable of both the singular and plural (DEPOO-OF (p) and MAXOO-
OF (p)) which outrank

the constraint militating against realizing the affixed mora in the first foot of broken

plurals (DEPOO-FI (p)).

To see how these constraints interact, observe the following representations

showing the number of morae in the output singular [mulhaq] and candidates a. and b. in

tableau (1) below.

(13) Sing. Correct plural Wrongplural

PI' '.I p PP C1 P P PIJ

(m u l).ha .q 4 a. (ma. I a:). ha .q * b. (m a 1) (h a:). q

The first foot of the correct form (13.a) surfaces with one more mora than the first

foot of the singular form. The final syllable has exactly the same number of morae as the

final syllable of the singular form. However, the first foot of the wrong form (13.b) has

the same number of morae as the first foot of the singular form. The final syllable affixes

a mora and violates D E P ~OF~ (p)

- (weight of the final syllable should be retained).

36
This assumption conversely entails that if a mora is in the input, then Input-Output-0 is obeyed.
 McCarthy's (2000) DEPOO-pconstraint (translated from his assumption 'affixed-p)

is too general and doesn't capture the fact that a mora is affixed at a designated position.

However, DEPW-FI (p) which this study advocates judges which of these candidates is

the winner as it restricts the added mora to be affixed to the first foot of the singular

forms to form broken plurals. Candidates b. in tableaux (1) & (2) are out because they

violate the high-ranked constraints DEPOO-OF (p) and MAX^- DF (p) fatally because (I .b)
adds a mora to the final syllable of broken plural and (2.b) loses a mora from the final

syllable of broken plural. The optimal output a. in both tableaux completely satisfies the

high-ranking constraints at the expense of violating the low ranked D E P O ~FI

- (p) by

affixing a mora to the first foot.

Adopting Positional Faithfulness constraints offers an elegant analysis that reflects

the contrast in weight between the first foot and final syllable of broken plurals. This

analysis basically shows that where D E P ~FI~ (p)

- is always violated by having an extra

weight realized in the first foot of broken plurals, DEPOO-OF (p) has to be obeyed because

the final syllable surfaces without an extra weight being imposed into it. This is

manifested in tableau (1). Moreover, MAXW-p is strictly adhered to when forming

broken plurals as will be demonstrated in tableau (2) below.

 The idea of affixing a mora to the first foot of the singular forms admits potentially

different shapes of feet to emerge. Observe these kinds of feet.

(14) CiVC2 +p
a. * ClV+p C2 [CIVVC~I
b. *p+ClVC2 [VCIVC~]
c. 4c 1 v c 2 + p [C,VC2V:]
Syllabic well formedness constraints govern the locus of the affixed mora. These

constraints regulate the permissible syllable shapes and prosodic structures in the whole

language in general. To illustrate, candidate (14.b) fatally violates the requirement for

having an onset which requires that every syllable starts with a consonant. Candidate

(14.a), on the other hand, produces a trimoraic3' structure, a highly marked structure

cross-linguistically.

(15) "pppja Syllables don't have three morae. (McCarthy and Prince 1990a; 1990b)

A syllable like CV:C violates this constraint since it has three morae.

(16) ONSET Every syllable has an onset (McCarthy and Prince 1993).

Syllables such as V and VC violate ONSETas they don't begin with a C. The

following tableau illustrates the interaction between syllabic well formedness constraints

and positional faithfulness constraints.

Tableau (3)
0: /(~nax).ba.z/+p'bakery'

6 a. (ma.xa:).bi.z
b. (ma:x).bi.z
c. (max). (bi:).z
d. (max).a.bi.z
e. m.(xab).z

31
Three morae (units of weight) in a syllable
 This tableau illustrates that the locus of the added mora is not random and is

governed by prosodic and syllabic well formedness in the language. The prosodic well

formedness constraints along with the general mechanisms of forming broken plurals

(these are basically two mechanisms: 'preserve the weight of the final syllable' and affix

a mora to the first foot of broken plurals) produce the correct output. The affixed mora

settles in a position that suits the structure of the language. Candidates b., c. and d. violate

one of the high-ranking constraints. To illustrate, candidate b. has a trimoraic syllable

(two morae are associated with the long vowel and one with the coda consonant), the

final syllable of candidate c. is not identical in weight with the final syllable of the

singular output and candidate d. has an onsetless syllable. Candidate e. loses a mora and

violates MAX^,, (p), a highly ranked constraint. Candidate a,, on the other hand, obeys a11

the high-ranking well formedness constraints and violates the low-ranked constraint

D E P O ~FI- (p). It is selected as the winning candidate.

This tableau concludes the discussion of the typical broken plurals. Well-

formedness constraints together with the constraint monitoring identity in weight of the

final syllable of the singular and broken plural dominate the constraint against adding a

mora to the first foot at the left edge of the singulars.

5.2.2 Shapes with default glides

In the Muscat dialect of Omani Arabic, there is a group of broken plurals which

exhibits an additional consonant, specifically an extra glide in their surface forms. This

glide doesn't exist in the singular forms from which these broken plurals are derived.

Observe the following patterns of broken plurals with the extra glide inserted.

(17). (CV.CV).VC-ahSingularforms (Cv.CV:). jV.C

a. (ma.ki:).n- ah (ma.ka:).ji.n 'machines'
b. (Ra.di:).q- ah (ha.da:).ja.q 'parks'
(18).CVV.CV(V).C Singularforms (CV.WV:).CV.C
a. (Sa:).da.r (Sa.wa:).da.r 'bed blankets'
b. (Sa:).ra.T (Sa.wa:).ra.T 'roads'
c. (ta:).(bu:).r (ia.wa:).(bi:).r 'line-ups'
d. (ta:).(bu:).q (ta.wa:).(bi:).q 'cement blocks'

The singular forms in (17) and (18) have fewer consonants than four base

consonants and their broken plurals expand by inserting a default glide to conform to the

broken plural shape. For example, in (1 8), the first foot of the singular [Sa:]dar contains

only one consonant and a long vowel [Sa:] and requires another consonant to create a

well-formed foot (CIVC~V:);[Sawa:]dar results. The singulars in (17) have a long high

vowel in their second syllable [Ci:] as seen in [ma.ki:.n-ah]. After mapping ClVC2V

[maki] contained in the first iamb (ma.ki:) of the singular form (17.a) onto CIVC2V:

[maka:], the residue of the singular [in] requires a glide to form the onset of the third

syllable in the broken plural [ma.ka:.jin]. This behavior is not peculiar to broken plurals

in the Muscat dialect of Omani Arabic. It is discussed in McCarthy and Prince (1990a)

and McCarthy (2000). In classical Arabic, a glottal stop /?Iand not glide /j/ is inserted to

form the onset to the third syllable of the broken plurals.

McCarthy (2000) makes a fourth crucial observation with respect to the mapping of

the consonantal segments from the singular to plural shapes. He attributes the complex

relation between the long vowels in the singulars and epenthetic glide in broken plurals to

a conservation of consonantal positions in singular + plural mappings (refer to forms

(17) and (18) above). He accounts for this relation by proposing undominated
associational faithfulness constraints such as NO- SPREAD^, (p, SEG)and NO-DELINK, (p,

SEG).I define them as follow.

(19)NO- SPREAD^, (p, SEG) segment to mora linkage is preserved in the output-output
mapping.

(20) N O - D E L I N K(p,
~ ~SEG) segment to mora linkage is not delinked in the output-output
mapping.

These constraints ensure that if, for example, the third segment of the singular is

associated with the second syllable (second mora, in this case), then it also should be

associated with the same syllable (likewise, same mora) in broken plurals. The following

represents the correspondence relations adopted from McCarthy (2000) but used to

represent the data under investigation.

5a:dar 4 5awa:i)ar * Sada:war

In these representations, onsets [S, w, 51 are associated with p, forming together

with the vowels that immediately follow them a CV sequence. Final consonant [r] is not

linked to a mora. As seen in the representation above, 82 is linked to p3 in both the

singular form and correct plural. The second broken plural form (incorrect) has 8 2 linked

to p2 instead of p3 as in the singular form, violating NO-DELINK,,(~,SEG). The

superscripts as well as subscripts help keep track of this correspondence relation.

 Therefore, NO-DELINK,,
(p, SEG)is undominated and governs the preservation of the

association lines38.

/3/ which is linked to the final syllable in the singular form is associated with the

second syllable in the broken plural in candidate b. This incurs a fatal violation of NO-

D E L I N K , ~ (SEG).
~ , Therefore, this candidate is out (refer to representation (21)).

An important detail remains; an epenthetic glide is inserted to fill the empty onset

position. The requirement for having an onset is more important than the constraint

against inserting an extra consonant (McCarthy 2000).

(22) D E P ~Glide
~- Every glide in output2 has a correspondent in output,

This constraint is a faithfulness constraint which ensures that glides in the output

broken plural have correspondents in the output singular forms. This constraint is

violated when the output broken plural surfaces with an extra glide.

I h. (Sa. a:). 3a.r I *! I

38
The domination of the association faithfulness constraints accords with McCarthy's (2000) analysis
 The second syllable in candidate b. has no onset. It fatally violates ONSETwhile

candidate a. wins because all its syllables have onsets. It inserts a glide at the expense of

violating the low-ranked constraint D E P O ~Glide.

-

When the second syllable of the singular forms has the vowel /i:/, broken plurals

insert the glide /jl to fill in the onset position of the third syllable of broken plurals as

exemplified below.

(23).(CV.CV:).C-ah Singularforms (cv-cv:). J'V.C

a. (ma.ki:).n- ah (ma.ka:).jin 'machines'
b. (ha.di:).q- ah (ha.da:).ja.q 'parks'
c. (na.ti:).g-ah (na.ta:).ji.g 'results'
d. (ka.ni:).s-ah (ka.na:).ji.s 'churches'

When the singular forms above form broken plurals, the bimoraic shape ClVCzV

[maki] belonging to the first iamb (ma.ki:) of the singular form is mapped onto C,V.C2V:

[rna.ka:~'~,maintaining an iamb in the first foot and keeping the same number of morae

in the singulars (refer to (24) below). Both the singulars and plurals have an iamb with

three morae in their first foot and a mora in their final syllables. These broken plurals

pose problems to the assumption that affixed mora is the basic distinction between

singulars and broken plurals since both the singulars and broken plurals have the same

number of morae. Therefore, M A X ~(p),

O which bans deletion of morae between related

strings, is strictly obeyed. Observe the following moraic representations of both the

singular and broken plural.

39 This approach is similar to McCarthy and Prince's (l990a) and McCarthy's (2000).
(24) Moraie representation of sing. 1xa.rk.t-ah] and pl. [xara:jat] 'maps'
Sing. PI.
FI FI
A A

x a r i:[ a h x a r a:j a t
The singular foot is an iamb, so there is not a straightforward mapping from a

trochee onto an iamb in these forms. Because only the first two morae of the singulars are

mapped onto an iamb, this iambic foot structure is preserved in the plural by mapping the

vowel Iil to a glide in the broken plural.

(25) Representation of the place features

a. L j l

[COT] [+high] (Lombardi 1996; Alderete et at 1999; Ussishkin 1999)

This representation shows that both /i/ and ljl share the same place features; both

are coronal and high. This tendency is translated in OT by the constraint IDENT-COR
or

IDENT-HIGH
(Ussishkin 1999) which requires coronal1 high segments in the singular form

to have coronal1 high correspondents in the output plurals. The following tableau shows

the ranking of the constraints involved in deriving broken plurals with the glide Ijl.
 Candidate b. inserts a glide /w/ to fill in the onset position of the third position of

broken plurals. Thus, it violates IDENT-COR

since the place feature of the glide /w/ is

labial1 round and not coronal1 high. Candidate c. has the third syllable without an onset.

This exhibits a violation of 'have onsets'. Both candidate d. and e. lose one mora and

violate the high-ranking constraint MAXOO(p). The optimal output is structurally well

formed as it obeys all the high ranking constraints at the expense of violating the low

ranked *Pl/cor and DEPOO-GLIDE.

Another interesting candidate to be considered is

(xa:.ra).ja.t. This candidate is ruled out because it doesn't have a proper iamb

The epenthesis of glides results from a need to have onsets as a syllabic well-

formedness requirement in Arabic. However, the inserted glides shouldn't disturb the

existing associational relations in singular 4 plural mappings because of the high-

ranking constraint NO-DELINK,,(~,

SEG).

5.2.3 Shapes with medial geminates

In the Muscat dialect of Omani Arabic, singulars with medial geminates preserve

their geminates when they map onto broken plurals. However, certain syllabification

restrictions pertaining to gemination come into play to govern the acceptable final shape

of broken plurals.

26.(CVC&(CzV:).C Singularforms (CV.C2V:).(C2V:).C

a. (sak).(ki:).n (sa.ka:).(ki:)n 'knives'
b. (8al).(la:).g- ah (0a,Ia:).(li:).g 'refrigerators'

Singular forms in (26) have a medial geminates [sakki:n] with the first half of the

geminates belonging to the first foot of the singular and the second half filling the onset

position of the second syllable. The geminates being adjacent to each other in the
singular maps onto two feet in the broken plural [(saka:)l(ki:)2n] as a result of the

expansion of the left foot C I V C ~of the singulars into a different shape CIVC~V:,

maintaining the length of the final syllable of the singular forms.

The following representation illustrates the effects of the singular-plural mapping

of the word forms [sakki:n]--b [saka:ki:n]:

(2 7) Sing. PI.
0 0 0 0 0
A A I A A
1 2 3
PI P2 P p3 ,P4

.:
sl ":. k2 .:.. i n3 s , :\ k i: ..i .i ;. n3
.... .. ... :: ::
......
.,.. '
ii
:.:
: ::
.
::
:
a 1 a i
When the affixed mora is added to the trochaic first foot of the singular forms to

form broken plurals, several potential output shapes might surface. However, well-

formedness constraints decide which output best adheres to the language as a whole.

(28)
a. sak +p + 4 saka:
b. p + sak -t * asak
c. sa + p k + *sa:k

Shapes b. and c. are ill formed in the Muscat dialect of Omani Arabic. Shape b. is

an onsetless syllable, violating the requirement for having onsets. Shape c. is composed

of three morae, a structure which is highly marked cross-linguistically. A closer look at

the actual broken plural reveals the fact that syllabification canons ensure that geminates

don't surface in one syllable. OT translates this prohibition by the constraint *0 [C,C,

(29) *a [CxCx geminates are banned from the edges of syllables (McCarthy
1998 and Gafos 2003)
 This constraint militates against having geminates occur at the margins of syllables.

The following tableau illustrates the ranking constraints to produce the optimal broken

plural with geminates.

Candidate b. has a trimoraic syllable, violating the constraint against trimoraicity.

Therefore, it is out. Candidate d. violates the high-ranking constraint against loss of

weight from the final syllable of the broken plural since they are derived from singulars

with a final heavy syllable. Candidate e. has an onsetless syllable. This fatal violation

excludes it from further evaluation since the requirement for having an onset is highly

ranked. Candidate c. fatally violates the constraint against having geminates surface at the

margins of syllables. Thus, it is doomed. However, the optimal output obeys all the high-

ranking well-formedness constraints and incurs a violation to DEPOO-FI(p) by adding an

extra mora to the first foot.

McCarthy (2000) concludes that the preservation of consonantal mapping from the

singular to broken plurals is given priority even at the expense of original geminates

becoming long-distance linked; in other words belonging to two different feet (McCarthy

2000: 178).
5.2.4 Plurals with even (H) iambs

The Muscat dialect of Omani Arabic allows the formation of broken plurals with

two expansions of iambs (L H) or (H). Broken plurals can either realize a typical iamb L

H or the even one H.

The mapping of the trochaic foot4' from the singular to the even iamb in broken

plurals is motivated by the unusual shapes of the singular forms which when undergoing

the 'affixed +p formalism' are forced to be shaped in a way that conforms to the

prosodic and syllabic canons of the language. Therefore, the mechanism applied to form

these broken plurals is exactly similar to the proposal advocated to account for the typical

broken plurals in classical Arabic. The general shape of the singular form as well as the

prosodic well-formedness constraints of the language as a whole govern the locus of the

'affixed + p' and the final accepted shape of broken plurals.

(30). C.(CV:).C Singularforms (CVC).(CV:).C

a. n.(Ya:).l (nuY).(la:).n 'sandals'
b. t.(ra:).b (tur).(ba:).n 'sands'
c. g.(da:).r (gid).(ra:).n 'walls'
d. b.(la:).d (bil).(da:).n 'countrysides'

The singular forms have a consonant cluster at the left edge followed by a long

vowel. These broken plurals end in an extra consonant that the singulars don't have due

to the requirement that nouns end in a consonant in Arabic (McCarthy and Prince 1990b).

This final consonant is extrametrical and doesn't participate in the prosody of the

language as whole.

The mechanism applied to form these broken plurals is that a mora is afixed to the

singular form. The resultant shape of the broken plural has to satisfy all the prosody

40
The mapped Lrochaic foot in these forms is of the shape Hand has a cluster at its left edge.
canons of the language in order to be admitted as an actual surface form. The locus of the

affixed mora is highly determined by the prosody of the language.

Moreover, the actual broken plural ends in a consonant In/. The insertion of the

nasal In/ stems from the requirement for words in Arabic to end in a consonant. It is

translated into the constraint Final-C.

(31.a) Final-C A prosodic word doesn't end in a vowel (McCarthy and Prince
1993a: 176)

An important detail remains to be addressed; the singular forms have a final heavy

syllable which remains heavy when they map onto broken plurals. Therefore, MAX^^- OF

(p) which bans deletion of morae from the final syllable of broken plurals is strictly

obeyed.

Before we show the ranking of these constraints, let us observe the moraic
representation of both the singular and the actual output broken plural. It shows that the
actual output broken plurals surfaces with two extra morae than the singular form.
(32) Sing. PI
 Candidates b. and e. fatally violate * COMPLEX
since their first syllables begin with

consonant clusters at their left edge. In candidate b., the second syllable is onsetless.

and ONSET.
Therefore, it is also out because of two crucial violations, namely *COMPLEX

Candidate d. has a syllable with three morae. Such a structure is highly marked and

violates the constraint against trimoraicity. Candidate g. loses a mora from the final

syllable and violates Maxoo- OF (p). Candidate h. fails to add an extra consonant to

confirm to the four consonants in the plural and seems identical to the singular output.

The optimal candidate completely adheres to the well-formedness constraints of the

language and incurs a violation of having two morae. But this constraint is low-ranked.

Candidate f. is ruled out because it has more morae than the optimal output.

This tableau illustrates that the locus of mora insertion complies with the prosodic

structure licensed in the language. Due to the fact that the singular form begins with a

cluster at the left-edge, the mora gets inserted between the two consonants, to eliminate

markedness in the output broken plurals even at the expense of radically changing the

association lines.

Another intriguing shape of broken plurals is derived from singulars with a sole

long syllable and two consonants.

(33). CVV.C Singular,forms CV.CV:.C

a. ko:.b ki.ba:.n 'cups'
b. na:.r ni.ra:.n 'fires'
c. ko:.T ki.Ya:.n 'elbows'

When these singular forms map onto broken plurals, they affix a mora and insert a

nasal In1 to ensure conformity with the Arabic requirement for having all nouns end in a

consonant. These broken plurals also keep the length of the final syllable. Thus, MAXOO-
OF (p) is highly adhered to when forming this shape of broken plurals. Observe the

following tableau:

All candidates have an extra mora; however, the locus of the affixed mora differs

from one candidate to another. The high-ranking well-formedness constraints ensure a

particular position for the extra mora. All the above candidates with the exception of

candidate a. (the optimal candidate) violate one of the high-ranking constraints. Thus,

they are excluded from further violation. To clarify, the first syllable of candidate e. is

onsetless and violates the requirement for having an onset. Candidate b. has complex

clusters at the margin of one of its syllables. Candidate c. has a trimoraic syllable and

candidate d. ends in a vowel and violates Final-C. Candidate g. loses a mora from the

final syllable and violates MAX^^- OF (p), a highly ranked constraint. The optimal output

obeys all the well-fomedness constraints at the expense of violating the low ranked

constraint DEPm- p. Candidate f. is ruled out because of the excessive number of morae

it adds. An interesting candidate to be considered is n.ra:.n which completely obeys all

the above constraints. However, this candidate can't be an optimal output, as it exhibits

no extra mora, a requirement for forming broken plurals.

 In the Muscat dialect of Omani Arabic, some singuIar forms with exactly similar

shapes as those forming the typical shapes map onto diverse shapes of broken plurals. For

example, there is a group of singular forms whose shape lumps together the first two

syllables contained in the foot (CV.CVV) into one heavy syllable C.(CVV). They map

the first trochee of the singulars onto an even iamb H.

34. CVC.CV(V).C Singularforms C.(CV:). CV(V). C

a. maq.Ja.t mqa:Jat 'matches'
b. maq.sa.f mqa:suf 'small vendors'
c. maS.xa.1 mSa:xil 'sieves'
d. muy.ba:r mya:bi:r 'curing spells'
e. muY.la:q m7a:li:q 'hooks'
f. muf.ta:h mfa:ti:h 'keys'

These broken plurals collapse the first two syllables in the expected typical iambic

foot into even iamb and surface with a cluster at the left-edge of broken plurals. Shaaban

(1 977) notices free variation between forming these broken plurals with the typical iamb

and collapsing the two syllables into one even iamb. For example [mqa:Jat] and

[maqa:Jat] 'matches' are both possible broken plurals

According to Shaaban (1977), there is a socio-linguistic factor involved in this

variation. Despite exposure to education and standard Arabic which obviously favours

forms with the typical iamb, young Omanis of Muscat in 1977 still have a great

preference and tendency to retain their parents' way of pronouncing these plurals with the

clusters at the left edge and even iamb.

The tendency for shaping the first foot of broken plurals into CCV: instead of the

typical foot CVCV: can also be analyzed phonologically. The Muscat dialect of OA is

attested to have an active process of vowel deletion (Shaaban 1977; Glover 1988) in
which [a] drops in unstressed positions. Adjectives and plural nouns of OA, as reported

in Shaaban, exhibit free variations between the following patterns:

(35) Plural nouns4'

a. mda:ris/ 1nada:ris 'schools'
b. mra:wiWmara:wih 'fans' (Shaaban 1977: 91)

(36) Adjectives
a. kari:m 'generous' b. sqi:m 'sick'
c. qadi:n~ 'old' d. kbi:r 'large' (Shaaban 1977: 91)

The plural nouns in (35) drop [a] between CI and Cz without exhibiting any shift in

meaning. The adjectives in (36) alternate in whether to have [a] between C I and C2 or to

leave it out. Shaaban (1977) proposes the following linear rule to account for this

variation in the broken plural formation.

(37) VI + 0 I # CVl.CVV resulting in a cluster at the left edge of the plural forms and
heavy syllable CCVV.

In applying the analysis proposed above, one would encounter a dilemma between

having two potential output broken plurals. However, I would suggest that candidate b. in

tableau (10) below violates the constraint against having extrametricality at the left-edge

of words. This constraint is dominated by DEPOO-FI(p). I formulate this constraint as

follow.

(38) NOEXTRAM-L NO extrametrical consonants are allowed at the left-edge of

words. Observe the following ranking:

Both candidates a. and b. are potential outputs since they obey all the high ranking

constraints. I won't delve into the complication of this variation due to space limitation

(see Antilla (1997) who discusses free variation on OT when studying Finnish genitives).

" The transcription of these forms are modified to fit in the IPA system adopted throughout this thesis
 Moreover, the Muscat dialect of Omani Arabic has a group of broken plurals with

only a sole heavy foot and extrametrical C at the left-edge. These broken plurals are

similar to the forms discussed above in that they collapse the expected typical iamb into

the even type.

(39). CCVC- (ah) Singularforms CCV:.C

a. mlaf 'folders /binders'
b. mxam- ah 'brooms'
c. mSut 'combs'
(40). CVCC- (ah) Singular.forms
a. xurq 'holes'
b. hurq 'bums'
c. harf 'letters'
(41).CVCVC Singularforms
a. sahan 'plates'
b. tabal 'drums'
c. gabal 'mountains'

In classical Arabic, forms (39-41) include two syllables (light followed by heavy).

Examples extracted from McCarthy and Prince (1990a) are as follows:

(42) Examples of broken plurals in Classical Arabic

Sing. PI. Glos
a. nafs nufuus 'souls'
b. qidh qidaah 'arrows'
c. ?asad ?usuud 'lions'
 The dilemma imposed in the analysis of these forms is similar in nature and effect

to the problem created from collapsing the first two syllables of the typical iamb into only

one heavy syllable. The only difference between these groups of broken plurals and the

ones discussed above lies in the fact that these contain fewer syllables.

A socio-linguistic factor has also an active role. The variations between classical

Arabic broken plurals surfacing with a typical iamb and broken plurals in the Muscat

dialect of Omani Arabic collapsing the two syllables contained in the typical iamb into an

even iamb supports the tendency of Omanis to retain their parents' way of pronouncing

these forms.

To sum up, McCarthy's (2000) proposal that the distinction between the singular

forms and broken plurals lies in an affixed mora provides an adequate analysis to broken

plurals in the Muscat dialect of Omani Arabic. The affixed mora has to choose a locus

that doesn't disturb the existing association lines in the singular* plural mapping and

conform to the syllabic well-formedness constraints.

For convenience and preciseness of analysis, I will include all the constraints

pertaining to syllabic and prosodic well-formedness constraints such as ONSET, *ppp]o,

*COMPLEX,Final-C, and *G [C,C, under the rubric Syllabic Well-formedness

Requirement (SWR). These constraints have the same priority and they place a pivotal

importance on what shape of the broken plural has to be realized. The general ranking

established to account for the diverse shapes of the broken plurals is MAXOO-aF (p),

SWR, (NO-SPREADoo(p, SEG), NO-DELINKoo(p, sec)), Max00 (p), )) DEPOO- (p),

DEPOO-Glide DEPOO-FI(p).
 The following lattice summarizes the grammar or ranking of the constraints

adopted to offer an integrated analysis to the diverse shapes of broken plurals in the

Muscat dialect of Omani Arabic.

(43)
N o - D E L I N K ~ SEG)
~ ( ~ , NO-SPREAD~,,(~, O ~ MAX^^ (p)
SEG) SWR M A X ~ O -(p)

5.3 Exceptional broken plurals

The discussion above has elaborately analyzed the typical and most productive

broken plurals in the Muscat dialect of Omani Arabic, leaving out the exceptional shapes

for further speculation. This section is dedicated to exploring the divergent shapes of

broken plurals, explaining the exceptionality in their final shape.

There are three exceptional shapes of broken plurals in the Muscat dialect of Omani

Arabic: broken plurals with partial reduplication, broken plurals with a single heavy iamb

preceded by the glide /j/ and more strikingly broken plurals with a sole light syllable.

Noun plurals showing partial reduplication occur sporadically in the list of broken

plurals in the Muscat dialect of Omani Arabic. They are also documented in a wide

variety of Afroasiatic languages particularly in the Chadic, Cushitic and Semitic groups

(Ratcliffe 1996: 296).

(44) Broken plurals with partial reduplication

a. xat xtu:t 'lines'
b. Sat Stu:t 'covers'
c. ha1 Rlu:l 'solutions'
d. saf sfu:f 'classrooms'
 These broken plurals are derived from singulars with two base consonants. The

singulars have a singIe light syllable. When these singulars are mapped onto broken

plurals, the second consonant in the singular form spreads to the third consonant position

in the shape CjC2V:C2 , where similar subscripts indicate reduplicated consonants. The

long vowel in the broken plurals is always /u:/.

Ratcliffe (1996) notes that reduplication occurs only as a feature of noun plurals

whose singulars have one or two root consonants. Singulars with one or two root

consonants consistently expand to a three or four consonant structure in the derivation.

Ratcliffe also claims that reduplication of the final stem consonant is one way in which

this is accomplished (p. 299). More relevant to broken plurals in the Muscat dialect of

Omani Arabic is Ratcliffe's observation that the most common form of reduplication in

Neo-Aramaic dialects and South Arabian languages is sufjxal reduplication of a single

final stem consonant with a predictably fixed vowel appearing between reduplicated

consonants. The predictability of the vowel between the reduplicated consonants is a

feature observed in these broken plurals.

The second group of exceptional shapes of broken plurals encompasses shapes

mapped from singulars with a single heavy syllable /e:P2 and two consonants modulo the

singulars deriving broken plurals with partial reduplication.

(45) Broken plurals with /j/ and heavy syllable

a. se:h sju:h 'barren lands'
b. se:f sju:f 'swords'
c. Ye:S Yju:S 'rices'

42
The singular forms are underlyingly ICjCI. For example, the trilitral verb (verbs with three root
consonants of form (45.c) is ITjSI.
 When the singulars with two consonants and heavy syllable with the vowel 1e:I map

onto broken plurals, the long mid vowel /e:/ in the singulars assimilate to the glide ljl

which precedes the long vowel /u:/. Thus, the final shape of broken plurals surfaces with

consonant clusters, taking the shape CjV:C. The glide Ijl surfacing in broken plurals is

attributed to underlying /j/ in the base forms. For example, the singular form [se:R] is

/sjW in the underlying structure.

More striking are broken plurals which do not realize an extra weight in their

surface shapes. Although these broken plurals are derived from singulars with three

consonants and there might be a potential for the trochee C I V C to

~ map onto the typical

iamb CIV.C~VV,surprisingly enough, these singulars map onto a single light syllable

and consonant clusters at the left edge of broken plurals.

(46) Broken plurals with trochee

a. xanS- ah xnaS 'trolleys'
b. buqY-ah bqaT 'stains'
c. durg- ah drag 'stairwells'
d. Jant- ah bat 'hand bags'
e. dakj- ah dkaj 'pillows'

5.4 Summary of chapter five

This chapter analyzes the shapes of broken plurals in the Muscat dialect of Omani

Arabic building on proposals made by McCarthy (2000). The analysis adopted here

basically derives the shapes of broken plurals from interaction of the Positional

Faithfulness constraint D E P ~ ~ -(p)

F , which maps the trochaic foot at the left edge of the

singulars onto an iamb by affixing a mora to the very first foot and constraints keeping

the weight of the final syllable of both the singular and broken plurals identical.
 The constraint adding a mora doesn't work in isolation from the well-formedness

constraints in the language. These constraints govern the locus of the added mora. This

analysis proves adequate as it successfully accounts for the typical shapes, shapes with

default glides, shapes with medial geminates and shapes with even iambs.

The final discussion presents an overview of some interesting issues relevant to the

formation of exceptional broken plurals. It discusses shapes with partial reduplication,

trochee and heavy syllable preceded by the glide /j/.

 CHAPTER SIX

Vocalism

6.1 Introduction

Broken plurals in the Muscat dialect of Omani Arabic are characterized by two

distinct types of fixed vocalism. The first type is manifested by the vowels contained in

the first foot at the left-edge of the canonical broken plurals which always have the vowel

[a]. It fills all the vowel positions in the first iambic feet (CV.CVV) and (C.CVV). The

forms in (1) have either the typical iamb CV.CVV or the even type C.CVV at their left

edge (bold-faced). In both these kinds of feet, [a] occurs in all the vowel positions. These

broken plurals have a foot and a light 'unfooted' syllable. The final consonant is

extrametrical.

(1) Vowel invariance [a1 in the first foot of the broken plurals
a. (ma.za:).ra.Y 'farms' c. (ma.ka:).si 'dresses'
b. m.(qa:).ba.ij 'door knobs' d. (ma.ta:).ba.x 'kitchens'

The second type is phonological vocalism in which the quality of the vowels in the

final short syllables of these broken plurals is phonologically dependent on the place

features of the consonants that follow them. This chapter provides a formal analysis of

both these types drawing on work done by Alderete et a1 (1999) and Urbanczyk (1999). It

starts with a description of the phonologically determined vowels which are contained in

the final syllables of broken plurals.

In forms (2) below, the underlined vowel in the final syllable agrees in place

features with the consonant that follows it. When the final consonant is back /Y/ or /h/,the
vowel preceding it is also back [a]. Coronal consonants such as In/ and /dl follow front

vowel [i] while round labial vowel [u] precedes labial consonants like Iml and /b/.

(2) The vowels in the final short syllable

la1 Iil [ul
a. ~awa:b& 'fingers' c. gara:jd 'newspapers' f. maka:tu& 'offices'
b. rawa:j& 'bad smells' d. saka:Sb 'sections' e. m a r a : h u 'lotions'

Such a phenomenon is missing from broken plurals of classical Arabic in which the

final consonants don't affect the quality of the vowels in the final syllable. The vowel in

the final syllable of broken plurals in classical Arabic is invariantly [i]. This justifies

McCarthy and Prince's (1990a) proposal to rewrite the vocalic melody [i] to associate it

with the vowel position in the last syllable of broken plurals in classical Arabic.

The following table offers a summary of the types of extrametrical consonants and

the quality of vowels contained in the short final syllable of broken plurals. It also

presents the number of forms containing these consonants and vowels. These forms are

taken from the actual data set in appendix A.

Table (1)
Final C V-cF Gloss Number of
forms
a. With emphatic, pharyngeal consonants- [a] in OF
a mqa:b@ 'door knobs'
1
t mqa:S& 'matches' 3
B i3aba:ja 'kosher meats' 5
7 Sawa:r& 'roads' II
S mara:qax 'casinos' 1
b. With back consonants- [a] in c r ~
x ma@:ba 'kitchens' 1
9 Bada:jag 'parks' 10
'd maba:la 'sums of money' 1
c. With 111- [a] in (JF
r dafa:ta 'notebooks' 14
d. With labial consonants- [u] in OF
b Yaqa:r&u 'scorpions' 7
m mara:h~ 'lotions' 11
e. With labiodental consonant- [u] in OF or [i] in 01:
f mqa:suf 'small vendors'
4
f sawa:lif 'gossips' 1
f. With coronals- [i] in OF
n maka:jb 'machines'
d masa:gg 'mosques'
3
5
s maga:lb 'rooms for guests' 4
z maxa:b~ 'bakeries' 2
S haSa:jiS 'grasses' 1
1 fara:mL 'brakes' 15

The above summary table establishes the fact that both the vowel and the final

extrametrical consonant VC# in broken plurals' final position share the same features for

place. Alderete et a1 (1 999) observe the same phenomenon for ~ a n c o w reduplication

r~~~

where a vowel assimilates to a following coda consonant. This linkage is determined by

language specific tendencies and the prosodic rules of the segments involved which

favors VC# place linkage over CV# place sharing (It6 and Mester 1995; Alderete el a1

1999 and works reviewed there).

Another type of phonological vocalism is shown by the vowel /i:/ in the final

syllable of broken plurals. When the final syllable of singular forms is long, broken

plurals surface with a long final syllable (weight retained). However, the vowel in the

final syllable of broken plurals always surfaces as /i:/ regardless of the featural quality of

43
Nancowry (also known as Nicobarese) is an Anstroasiatic language spoken in the Andaman Islands
(Alderete et a1 1999: 375)
the adjacent consonants and regardless of the quality of the final vowel in the singular

forms.

(3) Fixed /i:l in the last long syllable of the typical broken plurals
a. (muY).(la:).q mYa:kq 'hooks'
b. (man).(du:).s mana:di:s 'traditional Omani boxes'
c. (muh).(ha:).r maha:hcr 'sea shells'

The length of the final syllable in broken plurals corresponds to the long vowel of

the singular forms. However, broken plurals change the quality of the vowels in the

singulars to a fixed long vowel /i:/. These broken plurals have either the typical iamb

(CV.CVV) or the even one (C.CVV) at the left-edge followed by a final heavy syllable

identical in weight with the final heavy syllable in the singular forms.

This chapter is structured as follows: 5 6.4.1 addresses the fixed vowel [a] in the
first foot of broken plurals. The analysis then extends to account for the vowels in the

final short syllables which depend highly on the place features of the consonants

following them in 4 6.5.1. In 3 6.5.2, I analyze the phonologically determined vowel /i:/

in the final syllable of broken plurals when they are derived from singulars with a final

long syllable. Before embarking on the formal analysis of the vocalism, I shall review

McCarthy and Prince's (1990a) vocalic analysis of broken plurals in classical Arabic.

The analysis adopted here is inspired by arguments and thoughts raised in Alderete et a1

(1999). Therefore, 4 6.3 discusses Alderete ei al's (1999) analysis of fixed segmentism in
reduplication and shows how the current analysis relates to proposals made in this work.

6.2 Overwriting of Vocalism (McCarthy and Prince 1990a)

In accounting for the fixed vocalism of broken plural forms in classical Arabic,

McCarthy and Prince (1990a) argue that the invariant vowel [a] showing up in the first
foot (CV.CVV) and the fixed [i] vowel in the second syllable (CV) of these forms result

from overwriting the vowels [a] and [i] in the (CV.CVV) CV.C- shape of broken plurals.

(4) Classical Arabic broken plurals (McCarthy and Prince 1990a: 217)
Sing. PI. Gloss
a. jundub janaadib 'locusts
b. sultaan salaatiin 'Sultan'

In this analysis, McCarthy and Prince propose that the invariant vowels in broken

plurals associate with the vowel positions in the CV-shape of these broken plurals. They

pre-specify a fixed shape of broken plurals (CV.CV:). CV.C onto which singulars are

mapped. While the C positions of this shape are filled with consonants from the singular

stems, invariant [a] and [i] associate with the V positions in the first foot and final

syllable respectively. This analysis is driven by the fact that the vowels appearing in the

vowel positions of broken plurals are always fixed [a] and [i].

The use of melodic overwriting has been proposed to account for various

morphological phenomena attested in a number of the world languages. For example,

English 'Jm' prefixal morpheme is demonstrated to be a consequence of overwriting the

onset with the fixed segments 'Jm' and attaching it to the beginning of words such as

'table-schmable' and 'resolutions-schmesolutions' (McCarthy and Prince 1990a;

Bruening 1997, Alderete et a1 1999).

6.3 Fixed segmentism (Alderete et nl1999)

Alderete et al's (1999) analysis of fixed segmentism constitutes a basic grounding

for the analysis of fixed vocalism of broken plurals in the Muscat dialect of Omani
Arabic. The following paragraph reviews Alderete et al's theory of fixed segmentism and

shows how their proposals relate to the analysis adopted here.

Alderete et a1 (1999) classify fixed segmentism which prevails in patterns of

reduplication into two distinct types: phonological and morphological. Phonological fixed

segmentism refers to phonological defaults which are the least marked segments that

emerge in reduplicative affixes instead of expected copied segments from the base. It

results from the preference of unmarked structures to occur in particular domains

(reduplicative affixes, for example) while permitting marked structures to occur

elsewhere (in bases, for example) in the language. Morphological fixed segmentism, on

the other hand, is shown to exhibit the properties of affixation. One property they observe

is 'morphological fixed segmentism has the alignment properties of an affix' (Alderete et

a1 1999: 357). This leads to adopting alignment constraints to account for the fixed [a] at

the left edge of broken plurals.

As for the phonological vocalism in the final short syllable of broken plurals, the

constraint AGREE-VC#(Place) monitoring identity in place features between the vowel in

the final syllable and final extrametrical consonant dominates alignment constraints

specifying [a] to occur at the right edge of syllables in broken plurals. The vowel /i:/ is

also phonologically driven by the interactions of the constraints filtering identity in the

weight of the final syllable of both the singular and broken plurals and constraints

aligning [a] at the right edge of syllable. By and large, these two types of phonological

vocalism reflect the occurrence of defaults in the final syllable of broken plurals.
6.4 Analysis of the fixed vocalism

6.4.1 Fixed [a]

Fixed vocalism in broken plural forms is manifested by the vowel [a] a

invariably links to all V-positions in the first foot (CV.CVV) or (C.CVV) of broken

plurals. The vowel [a] is [pharyngeal] and [low] (Alderete el a1 1999).

(6) Pharyngeal [a] in the first foot (C.CVV) & (CV.CVV) of the broken plurals
Sing. PI. Gloss
a. maki:n-ah ma.ka:.jin 'machines'
b. maqSat m.qa:Sat 'matches'
c. maqsaf m.qa:suf 'small vendors'

Because [a] surfaces in the final V position when followed by a back and

pharyngeal final extrametrical consonant, this provides evidence that [a] is pharyngeal as

well. Moreover, this fact is encoded in the universal place markedness hierarchy

proposed in Prince and Smolensky (1993) and finalized by Lombardi (1996).

(7) Place markedness hierarchy (Prince and Smolensky 1993; Lombardi 1996)
*Pl/dorsal, *Pl/labial D *Pl/coronal )) *Pl/pharyngeal

There are two pre-OT analyses of the fixed [a] of broken plurals: pre-specification

(Marantz 1982; McCarthy and Prince 1990a, among others) and under-specification

(Pulleyblank 1988, among others). In pre-specification, the vowel [a] is underlyingly

specified in its respective V-places in broken plurals while in under-specification this

vowel is a default and is minimally represented as empty vowel positions which acquire

the vowel specification through markedness constraints on place features and language-

particulars.
 The following discussion adopts quite a different approach than pre-specification

to explain the fixed vocalism occurring in the first foot at the left-edge of broken plurals.

It assumes that [a] is specified through alignment constraints and entails that there isn't

an input [a]. This assumption is crucial to the analysis adopted here since it is consistent

with the assumption that singulars serve as outputs to the plural forms. This analysis also

entails no reference to the shape of broken plural as input to the realized [a] in the output

plurals.

As for the variable vowels occurring in the final short syllable of broken plurals,

they acquire their features from adjacent consonants to their right.

Two types of alignment constraints derive [a] in the first foot of broken plurals.

One constraint aligns [a] to the right edge of syllables in broken plurals.

(8) ALIGN-R (a, [a]) The right edge of every syllable coincides with the right edge of [a]

This constraint prescribes [a] to occur at the right edge of syllables in broken

plurals. It is an alignment constraint couched in the Generalized Alignment Theory

(McCarthy and Prince 1993b) whose general formalism is presented in chapter (3).

Generalized Alignment requires particular edges of categories or constituents to be

aligned to some edges of other categories or constituents. To illustrate, it demands a

particular edge of every prosodic or morphological constituent of Cat, coincide with a

particular edge of some other prosodic or morphological constituent of Catz.

However, because [a] doesn't occur in every 0,Align- R (a,[a]) is dominated by

a constraint requiring a different vowel than [a] to surface in the final syllable of broken
plurals. As demonstrated above, [a] only links to the vowel positions contained in the

first foot of broken plurals. We need a constraint that says 'don't spread outside the foot'.

Such a constraint has been proposed by It6 and Mester (1999) who refine how categories

are aligned when there is a shared feature at one or more prosodic edges. The constraint

they propose is CrispEdge [Peat]. A string has a crisp edge ifand only ifevery element of

that string is contained within the specified prosodic category. For example, a coda

geminate which shares a place node with an onset of a following syllable violates

CrispEdge [cr]. Refer to the representation in (9) below for the CrispEdge effect on

vowels in broken plurals.

(9) a. *[Taqa:rab] 'scorpions' b. [Taqa:rub]

PrWd

V q V r V

The effect of CrispEdge is demonstrated above in (9), where [a] doesn't spread

past the initial foot CV.CV:. The final syllable [ru] of the form [Taqa:rub] 'scorpions' is

not contained within the initial foot (Ta.qa:). Because as seen in (9. a), the vowel feature

[a] is shared by the vowels in the first foot and the vowel outside this foot (V in o,), a

violation of CrispEdge is incurred. When [a] fails to link to the vowels past the first foot,
a crisp foot (9. b) results. Please refer to chapter (3) for the general formalism of

CrispEdge constraint.

Following proposals by Urbanczyk (1999), CrispEdge constraint can be applied to

vowel harmony within a foot as seen in the representation in (9) above. It has to

accommodate the fact that [a] refrains from spreading past the foot in the formation of

broken plurals.

(10) CrispEdge [Ft] Shared features are contained within a foot. (Urbanczyk 1999)
The third required constraint penalizes linking of [a] to many vowel positions,

demonstrating the cost of spreading vowel features or simply maintaining the same place

features in many V-positions. In (12.a), Nolink is violated three times by having [a]

spread to three V positions while in (12.b), [a] is shared by only two V positions and it

violates NoLink only twice.

(11) NoLink No linking between elements (Pullyblank 1988; Urbanczyk 1999)

(12) [a] linking to V-positions

In the typical broken plural forms, the vowels in the even (C.CVV) and typical

(CV.CVV) iambs constructed are filled by [a]. It is more economical to posit that one [a]

is supplied in the first foot of broken plurals but then it links to three V-slots in (12. a)

and two V-slots in (12. b) above.

The following tableau illustrates how [a] occurs in all the vowel positions

contained in the first foot of broken plurals. It shows the ranking of the constraint
aligning [a] at the right edge of syllables in broken plurals and the constraint penalizing

[a] from spreading past the initial foot of broken plurals.

Subscripts indicate shared vowels. For example, in candidate a. all the vowels

which share the same subscript are one vowel which spreads in three V positions. This

incurs only one violation of *Pl/phar.

Both candidates c. and d. violate the high-ranking constraint CrispEdge [Ft] by

having the vowel in the first foot spread in the final syllable. Therefore, their initial feet

have no crisp edge. As for candidate b., it exhibits two violations of ALIGN-R(0,[a]).

Both the second and final syllables have different vowels than [a] being aligned at their

right edges. The initial foot of the output a. has a crisp edge because [a] is only restricted

to fill in the vowel positions contained in the first foot. It violates ALIGN-R(0,[a]) once

because the vowel in the final syllable isn't aligned to [a]. However, ALIGN-R(0,[a]) is

outranked by CrispEdge [Ft] and the optimal output can minimally violate ALIGN-R(0,

[a]) as long as it completely satisfies the high-ranking constraint CrispEdge [Ft]. Both

*Pl/phar and NoLink are low ranked and are violated by the optimal output and other

candidates. For example, since [a] is pharyngeal and shared by three vowel positions, the

winning candidate a. violates *Pl/phar once and NoLink three times.

 The above discussion has demonstrated how the ranking of constraints derives an

output with [a] restricted to the vowel positions in the first syllable by alignment

constraints and a different vowel filling the V-position in the final syllable. 3 6.5 analyzes
the vowels in the final syllable of broken plurals. It shows that these vowels are

phonologically determined.

6.5 Analysis of the phonological vocalism

6.5.1 Variable vowels

Another interesting characteristic of the fixed vocalism of broken plurals in the

Muscat dialect of Omani Arabic pertains to the vowels in the final syllable when it is

short. There is an identity in the place features of both the vowel in the final syllable and

the consonant that comes after it.

(13) VC# in the broken plurals.

a. ra:tub rawa:tub 'monthly allowances'
b. da:faY dawa:faT 'motives'
c. 8adi:d- ah hada:jid 'iron bars'

The quality of the vowel in the final syllable is highly determined by the place

feature of the consonant that follows it. Labial consonants are preceded by round vowels.

The fact that both are articulated using the lips lends support to the shared place of

articulation. When the final consonants in the VC# of broken plurals are back like IT/ in

(15.b), the vowel preceding them is constantly back, pharyngeal [a]. This is another piece

of evidence that [a] is pharyngeal and back since it assimilates to the place feature of the

back consonant that comes after it. Finally, when coronals occupy the final extrametrical

consonant position in broken plurals, the vowel in the final syllable is [i] accordingly.
 Further evidence of VC# place linkage comes from the imperfective verbs of

Omani ~ r a b i c As
~ ~demonstrated
. below, the quality of the vowels in the final syllable is

determined by the place features of the consonants that follows them.

(14) Imperfectives in Omani Arabic

Per- ImperJ Gloss
a. Satab 'cross out'
b. xataf jixwf 'kidnap, pass by'
c. yasal jiysl 'wash'
d. tabax jitbz 'cook'

There needs to be a constraint in the grammar which compels assimilation of place

features between the final extrametrical consonant and the vowel in the final syllable.

Such a constraint would be AGREE(Place) (Lombardi 1996) which basically scans any

pair of adjacent segments in a word and ensures that they share the same place features;

otherwise, a violation would be incurred if discrepancies of place features are observed.

Although this constraint might serve equally well in this particular instance, a more

narrow constraint is needed to capture the fact that only VC# in broken plural final

position share the identical place features. The way to translate this requirement is:

(15) AGREE-VC#(Place) The vowel and consonant in word-final position share the
same place features.

Since the forms *rawa:tib and *rawa:tab fare better in having vowels with the least

marked place feature than the actual form [rawa:tub], AGREE-VC#(Place) must

dominate the constraint aligning the vowel [a] at the right edge of syllables. This conflict

of constraints is shown in the tableaux, subsequent to the discussion.

44
A similar conclusion to the place linkage between the final extrametrical and preceding vowel in the
imperfectives in Omani Arabic is also made by Shaaban (1977).
 Moreover, both V in the final syllable and the following C are place-linked. The

constraint against having structures which are linked to the same place is NoLink (refer to

12 above).

NoLink is dominated by creating structures in the output forms where both the

vowel and the consonant following it are linked through the same place feature. This also

implies that there are no new features introduced by the vowels in the final syllable.

These vowels are clearly defaults, as they have exactly the same place features as the

consonants that follow them.

U
Place

The following tableaux show that both AGREE-VC#(Place) and CrispEdge [Ft]

dominate ALIGN-R(0,[a]) and produce the optimal output in which [a] only occurs in the

vowel positions in the first foot. The vowels filling in the V positions in the final short

syllable are variables and depend on the place features of the consonants that come after

them.

The identical subscripts in the tableaux below refer to shared features. Therefore,

the occurrence of the shared feature only incurs one violation mark regardless of how

many times that feature is observed in the same form. This assumption is crucial to the

analysis adopted here as it shows that [a] filling all V-positions is only one vowel with

the exact place feature and not three distinct ones linked to V-positions, as shown in (14)

above.
Tahleau (3)

The violation of NoLink doesn't determine the winning candidate. The optimal

output violates NoLink four times by having [a] shared by three V positions in the initial

foot and the final vowel linked in place features with the final extrametrical consonants.

The agreement in place features between CV# determines the optimal output. The

optimal outputs in the above tableaux violate *Pl/phar once. Both CrispEdge [Ft] and

AGREE-VC#(Place) are highly ranked and the optimal output obeys them. Where [a]

spreads past the foot level, CrispEdge [Ft] is violated and the candidates of this violation

are doomed.

An important detail remains to be addressed. [a] in the final syllable of some

broken plurals that have a final back consonant is different from the specified [a]

associating to the V positions in the first foot. It results from assimilation to the place
feature with the final extrametrical consonant. Such a fact is translated in different

subscripted [als. Observe the following ranking.

Although the optimal output has [a] filling in all the vowel positions, the [a] in the

final syllable is driven by the assimilatory process which links the final consonant and

preceding vowel in place feature. It is thus distinct from the [a] in the initial foot.

Candidates a. and f. both have the vowel [a] in all their syllables. However, candidate a.

wins because the vowel [a] in the final syllable isn't the same as the vowel [a] in the first

foot at the left edge of broken plurals.

6.5.2 /i:/ in the long final syllable

The second type of phonological vocalism of broken plurals in the Muscat dialect

of Omani Arabic also relates to the vowel in the final syllable. It is manifested by ihe

vowel /i:/ in the final syllable of broken plurals which are derived from singulars with

final long syllables. In forms (17), the final syllable of both the singulars and plurals is

heavy. Regardless of both the quality of the vowels in the final syllable of the singular
forms and featural quality of the final consonants in broken plurals, the final syllable

(underlined) of these broken plurals surface with the vowel /i:/.

(17) Broken plurals with long final syllable

Sing. PI. Gloss
a. qa:mu:s qawa:~s 'dictionaries'
b. kal1a:b kala:hb 'clips /wrenches'
c. Yasfu:r Yasa:filr 'birds'

This pattern clearly incorporates a reduction in phonological markedness in that [i]

is [coronal] and it is less marked cross-linguistically (Prince and Smolensky 1993). This

fact is manifested by the *Place markedness hierarchy (see (7) above). Prince and

Smolensky (1993) suggest that consonants with coronal place feature like Id/ and It/ are

the least marked and most frequent consonants cross-linguistically. Thus, these segments

are selected as defaults in languages. Lombardi (1996) extends this assumption by

designating that the laryngeals, especially I'?/ are the most common default consonants.

She stretches the place markedness hierarchy to the lower end by proposing that

pharyngeals (which include laryngeals (Alderete et a1 1999: 335)) are less marked than

coronals. Smolensky (1993) observes that a marked place of articulation incurs a mark of

violation for every time the feature of that place occurs.

It is important to note that the length in the final syllable of broken plurals is driven

by Positional Faithfulness constraints and relates to the long vowel in the final syllable of

the singulars. McCarthy (2000) shows that preserving the weight of the final syllable of

broken plurals is a consequence of high-ranking Maxoo-p which militates against loss of

morae from the final syllable of broken plurals when they map from singulars with a final

long syllable.
 The constraint MAXOO-o~(p) penalizes the loss of mora from the final syllable of

broken plurals when they are derived from singulars with final long syllable.

(18) MAx00- GF (P)

For every p in the final o of output,, there is a correspondent p in the final 0 of output*.

The vowel 1i:l is coronal and exhibits low ranked violation marks. Therefore,

*Pl/cor is dominated by MAX^^- (p) and CrispEdge [Ft] which monitors identity of

weight in the final syllables of the broken plurals which are derived from singulars with a

long final syllable and restricts the initial foot to having CrispEdge respectively. The

following tableaux show how /i:l is derived in the final syllable of the broken plurals.

An important detail remains to be addressed: despite the fact that a candidate with a

long mid low vowel [a:] in the final syllable of a form like (Ca,. Ca:,). (Ca:,).C satisfies

all crucial high ranking constraints since it retains weight of the final syllable and has a

crisp edge and even fares better in having the right-edge of all its syllables aligned to [a],

it can not be an output in Arabic. This is because of a strict adherence to Obligatory

Contour Principle ( o c P ) ~ ~which the language ensures not to violate (McCarthy 1979,

1986; Gofos 2003, 1997a among others). This principle bans identical adjacent features

from spreading to adjacent positions. [a] doesn't spread to all V positions.

(19) Obligatory Contour Principle (OCP)

At the melodic level, adjacent identical elements are prohibited. (McCarthy 1986:208)

The subscripted vowels show shared elements in that one [a] fills three V positions

and incurs three violation marks of NoLink. Other vowels like [u] and [i] filling more

45
Arabic is famous for obeying OCP as there is a virtually total absence of nominal and verbal stems ofthe
pattern C,VC,VC,. This observation entails two consequences: Arabic roots are subject to OCP and the
spreading in Arabic is rightward (McCarthy 1986)
than one V positions and bearing similar subscripts are assumed to be one vowel linking

to more than one V position modulo to [a].

Tableau ( 6 )
:I man . du: s I+ MAX,,- i CrispEdge j OCP

Candidate d. in tableaux (5) and (6) loses a mora in its final syllable and incurs a

violation to the high-ranking constraint M A X ~ ~ -(p).

O FCandidates b. and e. in tableaux

(5) and (6) violate crispEdge [Ft] by having [a] extends beyond the first foot. Candidates

c. have two syllables whose right edge isn't aligned to [a]. This is one more violation

than the optimal output which only incurs one violation mark to ALIGN-R (0, [a]). Thus,

they are excluded from further evaluation. Candidate f. in tableaux (5) and (6) fatally

violates OCP because all the adjacent vowel positions have an identical vowel [a].

Candidate a,, on the other hand, violates *Pl/phar once by having the same [a] fill in all
V-positions in the left foot of the broken plurals and *Pl/cor by having /il fill the V

position in the final syllable. These two constraints are low ranked.

It is worth mentioning that linking of the extrametrical consonant and vowel in the

final syllable isn't possible in these broken plurals because of the long vowel in broken

plurals which corresponds to the long vowel in the final syllable of the singulars.

Therefore, AGREE-VC#(Place) isn't applicable in these forms.

To sum up, the discussion explores the fixed vocalism observed in the first foot at

the left-edge of broken plurals and the phonological vowel /i:/ attested in the final

syllable of broken plurals when they are derived from singulars with a final long syllable.

The length is thus retained but the quality of the vowel is neutralized to a less marked

vowel (high and coronal). Constraints like NoLink come into play to evaluate that the

fixed [a] is shared by all V-places, thus banning new features from being introduced. In

the case where /i:/ always surfaces in the last syllable of broken plurals, a Positional

Faithfulness constraint M A X ~ - (p)

O ~is needed to make sure that weight is retained from

the singular form.

In summary, the analysis has offered an integrated approach of the two types of

fixed vocalism in broken plurals in the Muscat dialect of Omani Arabic. The fixed vowel

[a] is restricted to spread in the vowels contained within the initial foot allowing vowels

in the final syllable to assimilate in place to the final extrametrical consonant. The

grammar or ranking of the vocalism in broken plurals is summarized below in the form of

a lattice.
(20) M a x o o - a ~(p) AGREE-YC#(Place) CrispEdge [Ft]

ALIGN-R(a, [a])

NoLink

The general ranking is M ~ x o o - a(p~), CrispEdge [Ft], AGREE-VC#(Place) N ALIGN-

R (a, [a]) N *Pllcor N *Pl/phar )) NoLink.

6.6 Vocalism in the exceptional broken plurals

There are large numbers of exceptional broken plural forms which don't conform

with the vocalic invariance exhibited in the typical broken plurals. These exceptional

forms have only either a heavy syllable H or a light syllable L. In the forms with the

heavy syllable, the pervasive vowels linked to VV-positions in CCV:C shape are [a:],

[u:] or [i:]. However, when the vowel in the only syllable of broken plural forms is light,

the invariant vowel [a] surfaces. The following table surveys the vowels and surrounding

consonants occurring in the exceptional broken plural forms.

Table (2)
Cs surrounding V @:I Gloss Number of
forms
a. Between labials and back consonants or Ill, lrl
m T Smu:T 'candles'
1
m x Smu:x 'scratches' 1
f I qfu:l 'locks' 1
m r xmu:r 'wines' 1
b I tbu:l 'drums' 1
b. Between coronals and back consonants
s t mSu:t 'combs'
5
h n $u:n 'plates' 5
c. Between glides and back consonants or coronals
j x Sju:x 'tribal chiefs' 2
j S Tju:S 'rices' 3
d. Between back consonants
x t Sxu:f 'lines' 1
t t xtu:t 'lines' 1
Cs surrounding V /a:/ Gloss Number of
forms
a. Between labials and coronals
m 1 gma:l 'camels'
b. Between back consonants and coronals
S 1 mga:l 'sieves'
c. Between two coronals
1 1 mla:l 'bowls'

-V
Cs surroundinp -li:l
<
Gloss Number of
forms
a. Between labials and coronals
r m hri:m 'women' 2
b. Between back consonants and coronals
x 1 nxi:l 'palm trees' 1

Obviously, there is a high diversity in the quality of consonants that surround the

three long vowels which precludes them from being classified as a natural class

triggering one vowel or the other. Besides, the shapes of these broken plurals are

idiosyncratic, having only one heavy syllable and permitting clusters of consonants at the
edges of forms. These broken plurals are also lexically conditioned as they only map to a

form with a sole heavy syllable. They differ from the typical broken plurals which map

onto a form with an iamb and a second syllable whose length is related to the length of

the final syllable of the singular forms from which they are derived.

The exceptionality of these broken plurals relates to the unusual shapes of the

singular forms from which these plurals are derived. The singulars are characterized by a

sole light syllable and consonant clusters at either the left or right edges of words. The

general two shapes of singular forms are CCVC or CVCC. Observe the following

examples.

(21) Exceptional shapes

Sing. PI. Gloss
a. aarf i)ru:f 'envelopes'
b. xamr xmu:r 'wines'
c. R i g h8u:n 'forts'
d. msal m8a:l 'sieves'
e. mgaz 1nga:z 'grass cutters'

6.7 Summary of chapter six

This chapter offers an analysis of the two distinct types of fixed vocalism of broken

plurals in the Muscat dialect of Omani Arabic. The first type of fixed vocalism is

manifested by the invariant vowel [a] in the first foot at the left-edge of broken plurals.

The phonological vocalism is shown by the vowels in the final short syllable of broken

plurals which are highly dependent on the place features of the consonants that follow

them. When the final syllable of the singular forms is long, broken plurals also surface

with a long final syllable. However, the final syllable of broken plurals always has less

marked vowel /i:/. These facts can be explained if we assume that the fixed vowel in the
first foot of broken plurals is specified by alignment constraints. As demonstrated in the

surface forms of broken plurals, [a] is aligned at the right edge of the syllables contained

in the first foot of the plurals. The constraint CrispEdge [Ft] restricting [a] to link to all

the vowels within a prosodic constituent dominates ALIGN-R((3, [a]) which aligns [a] to

the right edge of all syllables.

Analyzing [a] as a default (under-specified) or pre-specified in the plural shape

would create problems to the correspondence relations assumed between the singulars

and broken plurals. Furthermore, under-specification requires that the underlying

structure of broken plural shape (input) be represented with empty V positions which get

their vocalic specification by supplying the most unmarked vowel by *Place markedness

hierarchy. It doesn't show why vowels other than [a] don't occur in the first foot of

broken plurals since other vowels like [u] and [i] will be ruled out by *Place markedness

hierarchy which excludes any candidate with these vowels from further evaluation. For

example, *matu:bax is ruled out by *Place markedness in which *Pl/lab D *Pl/phar.

As for the vowel /i:/, the length relates to the long final syllable of the singulars.

Therefore, the constraint MAXOO-OF

(p) which penalizes the loss of p from the final

syllable of broken plurals is highly ranked. Moreover, the fact that /i/ is an unmarked

vowel is supported by the *Place markedness hierarchy. The phonological vocalism in

the short final syllable is analyzed by having the constraint AGREE-VC#(Place) that

dooms any candidate whose final C and V don't share place features.
 CHAPTER SEVEN

Conclusions and Implications

This thesis has examined the morphological phenomenon of broken plurals in the

Muscat dialect of Omani Arabic. It documents the various shapes of broken plurals

mapped from singulars with diverse patterns. Canonical broken plurals are demonstrated

to relate in their overall shapes to the singulars. However, the Muscat dialect of Omani

Arabic also exhibits notorious shapes of broken plurals that can't be attributed directly to

the shapes of the singulars. The extra weight distinguishing the singulars from their

broken plurals doesn't apply to form plurality in these forms. They simply resist mapping

onto any expansions of iambs modulo to the typical broken plurals. Some exceptional

shapes also appear with partial reduplication.

This thesis has built on proposals made by McCarthy (2000) who assumes that the

distinction between singulars and broken plurals lies in an affixed p at a particular locus

in the singular forms. Despite the fact that the affixed p enforces a particular foot

structure namely iamb to be realized in broken plurals, the analysis proposed here

assumes along with premises of Generalized Template Theory (GTT) that such a

templatic effect (PLURAL= IAMB)is best interpreted as interactions of markedness

constraints of Universal Grammar. The analysis also entails that well-formedness and

prosodic faithfulness constraints play a major role in determining where the affixed p is

added and how the shapes of broken plurals are finally realized.

Positional Faithfulness constraints (Beckman 1998) encoded in DEPOO-FI (p)

account for the extra weight realized in the first foot at the left-edge of broken plurals.

Prior to McCarthy's (2000) OT proposals, Prosodic Circumscription (McCarthy and

Prince 1990a) addresses this fact by assuming a mapping of a trochaic foot at the left

edge of the singulars C I V C ~or ClVV or C I V C ~ Vonto an iamb CIVC~VV.This

mechanism requires reference to iambs which this thesis dispenses with altogether.

McCarthy's proposal provides a well-established mechanism that basically applies

the general rule for forming broken plurals (affixed +p) not in isolation from the well-

formedness constraints that govern the structure of the language as a whole. Therefore, it

has been demonstrated in the analysis of the shapes of broken plurals that when an extra

p gets attached at the first foot of a singular with an idiosyncratic shape, new well-

formedness constraints are used to basically filter the occurrence of a structure that the

language doesn't permit. For example, in forming broken plurals from singulars with

medial geminates, generally motivated constraints like 'geminates are banned from

margins of syllables' are called upon to license the final shape of the admissible broken

plural.

The basic argument that underlies the analysis of the shapes of broken plurals is

that although forming broken plurals imposes an extra mora to be realized in their surface

structures, Universal Grammar filters how the final shape will be structured by defining

the location of the affixed mora.

Broken plurals in the Muscat dialect of Omani Arabic exhibit interesting vocalism

that differs from the vocalism contained in broken plurals of classical Arabic. The

preference for the least marked vowel [a] at the right edge of syllables contained in the

first foot at the left edge of broken plurals allows alignment constraints to specify this

vowel. As there is a restriction on [a] occurring within just the first foot, a constraint

banning this vowel from spreading past the foot level is required. Such a constraint is
available in the OT literature and dubbed as CrispEdge [Pcat] (It6 and Mester 1999). This

proposal definitely provides an elegant analysis for [a] which is restricted to occurring at

a certain prosodic constituent which is the Foot domain.

Another type of fixed vocalism in broken plurals of the Muscat dialect is

phonological vocalism. The vowels in the final short syllable of broken plurals depend on

the place features of the consonants that follow them. This can be best understood as an

undominated AGREE-VC# (Place) effect which assimilates the place features for both the

vowel in the final syllable and the extrametrical consonant.

The long vowel /i:l in the final syllable of broken plurals is driven kom the length

of the final syllable of the singular forms. Such a relatedness and identity of weight in the

final syllable of both the singulars and plurals is captured by Positional Faithfulness

constraints. McCarthy (2000) shows that stem-final position is privileged in that contrast

in the final position is maintained. Moreover, Beckman (1998:l) refers to final syllables

as a possibly privileged position (using Beckman's word). These empirical pieces of

evidence extracted from the OT literature serve perfectly well in accounting for a whole

range of interesting phenomena in the course of offering an integrated analysis to broken

plurals in the Muscat dialect of Omani Arabic.

To sum up, the analyses of both the shapes and vocalism of broken plurals in the

Muscat dialect of Omani Arabic tie in with previous work on OT. The constraints

adopted as well as the general mechanism of constraint interactions and ranking OT

offers provide a well-motivated mechanism. Reference to Prosodic Circumscription and

pre- or under-specification to account for the shapes and vocalism respectively are

challenged by the elegance of analysis provided by the Optimality Theory framework.

 Bibliography

Abd-Rabbo, M. (1988). Some Morphological Constraints in Classical Arabic. Ph.D

Dissertation. Stanford University.

Abd-Rabbo, M. (1990). Sound plural and broken plural assignment in classical Arabic.
In M. Eid and J. McCarthy (eds.). Perspectives on Arabic Linguistics: Papersfrom the
First Annual Symposium on Arabic Linguistics, Amsterdam: John Benjamins. 55-93

Abu-Mansour, A. H. (1995). The interface between morphology and phonology: The

case of the broken plurals in Arabic. LACUSForum 22,320-339

Alderete, J. 1999. Morphologically Governed Accent in Opfimali@ Theory. Ph.D

Dissertation, University of Massachusetts, Amherst. ROA-309.

Alderete, J., Beckman, J., Benua, L., Gnanadesikan, A,, McCarthy, J., & Urbanczyk, S.
(1 999). Reduplication with fixed segmentism. Linguistic Inquiry 30, 327-64

Anttila, A. (1997). Deriving Variation from Grammar: A Study of Finnish Genitives.

Variation, Change and Phonological Theory. In F. Hinskens, R. van Hout, and L.
Wetzels (eds). Amsterdam: John Benjamins. Rutgers Optimality Archive ROA-63.

Archangeli, D. and Pullyblank, D. (1994). Grounded Phonology. Cambridge: MIT

Beckman, J. (1998). Positional Faithfulness. Ph.D Dissertation. University of

Massachusetts, Amherst. ROA-234-1297

Benhallam, A. (1980). Syllable Structure and Rule Type in Arabic. Ph.D Dissertation.
University of Florida.

Benua, L. (1995). Identity effects in morphological truncation. University of

Massachusetts Occasional Papers in Linguistics 18: Papers in Optimality Theory. In J.
Beckman, L. Walsh Dickey, and S. Urbanczyk (eds.), GLSA, University of
Massachusetts, Amherst, 77-136 [Rutgers Optimality Archive #74]

Benua, L. (1997). Transderivational Identity: Phonological Relations Between Words.

Ph.D Dissertation, University of Massachusetts, Amherst. ROA-259-0498

Brame, M. (1970). Arabic Phonology: Implication for Phonological Theory and

Historical Semitic. Ph.D Dissertation. MIT.

Bruening, B. (1997). Abkhaz Mabkhaz: M-Reduplication in Abkhaz, weightless

syllables, and base-reduplicant correspondence. In B. Bruening, M. McGinnis, and Y.
Kang (Eds.). MIT Working Papers in Linguistics 30, 291-329
Brockett, A. (1985). The spoken Arabic of Khabura. Journal of Semitic Studies 33.
University of Manchester Press: UK. 146-149.

Clements, G. N., Hume, E. (1995). The internal organization of speech sounds. In

Goldsmith (ed.). Handbook of Phonological Theory. Blackwell: Cambridge. 245-307

Crystal, D. (1997). A Dictionary of Linguistics and Phonetics. Cambridge: Blackwell

Publishers Ltd.

Gafos, A. (1 996). The Articulatory Basis of Locality in Phonology. Ph. D Dissertation.

The Johns Hopkins University.

Gafos, A. (1997a). Eliminating Long-distance Consonantal Spreading. Ms., University of

Massachusetts, Amherst. [Revision of Rutgers Optimality Archive #I061

Gafos, A. (2003). Greenberg's asymmetry in Arabic: A consequence of stems in

paradigm. Language 79, 317-355.

Glover, B. (1988). The Morphophonology of Muscat Arabic. Ph.D Dissertation.

University of California, Los Angeles.

Goldsmith, J. (1976). Autosegmental Phonology: Bloomington: Indiana University

Linguistics Club.

Greenberg, J. (1955). Studies in African Linguistic Classification. New Haven: Compass

Press.

Guda, A. (1988). Qur'anic Recitation: phonological analyses. Ph.D Dissertation:

Georgetown University.

Hammond, M. (1988). Templatic transfer in Arabic broken plural, Natural Language

and Linguistic Theory 6,247-70.

Hayes, B. (1979). A Metrical Theory of Stress Rules. Ph. D Dissertation. MIT,

Cambridge, Mass. (Published 1985, New York: Garland Press)

Hayes, B. (1987). A revised parametric theory. In J. McDonough and B. Plunkett (eds.),

NELS 17. Amherst, Mass.: Department of Linguistics. University of Massachusetts. 274-
289

Hayes, B. (1995). Metrical stress theory: Principles and case study. Cambridge, Mass.:
MIT Press.

Hetuon, R. (1974). La division de langues semitiques. In A. Caquot and D. Chen (eds.).

Acres du Premier Congres International d'Etudes Semitiques et Chamito-Semitiques. The
Hague: Mouton, 103-127
Holes, C. (1989). Towards a dialect geography of Oman. London Universify of Oriental
and African Studies Bulletin, 52,446-62

Honvood, Graham. 2001. Anti-faithfulness and subtractive morphology. [Rutgers

Optimality Archive ROA-466-0901]

It6, J., Kitagawa, Y and Mester, A. (1996). Prosodic faithfulness and correspondence:
Evidence from a Japanese argot. Journal of East Asian Linguistics 5,217-294

It& J. and Mester, A. (1995). Japanese Phonology. The Handbook of Phonological

Theory. In J. Goldsmith (ed.). Oxford: Blackwell. 817-838

It& J. and Mester, A. (1999). Realignment. The Prosody Morphology Interface. In R.

Kager, H. van der Hulst and W. Zonnevald (eds.). Cambridge University Press. 189-217

Jayakar, A. (1889). The O'manee dialect of Arabic. JRAS, 649-87 and 81 1-89

Jayakar. A. (1903). The Shahee dialect of Arabic. Journal of the Bombay Branch of the
Royal Asiatic Society 21, 649-880

Johnstone, T. (1977). Harsusi Lexicon. London: Oxford University Press.

Johnstone, T. (1 981). Jibbali Lexicon. London: Oxford University Press.

Kager, R. (1988). Metrical constituency and rhythmic adjustments. Phonology 5,21-71

Kager, R. (1993a). Alternatives to the iambic law. Natural Language and Linguistic
Theory 1I , 38 1-432

Kager, R. (1999). Optimality Theory. Cambridge University Press: UK. 142-190

Kager, R. and Zomeveld, W. (1999). Introduction. The Prosody-Morphology Interface.

In R. Kager and H. van der Hulst (eds.), Cambridge University Press.1-38

Leben, W. (1980). A metrical analysis of length. Linguistic Inquiry 11,497-509

Liberman, M and Prince, A. (1977). On stress and linguistic rhythm. Linguistic Inquiry 8,
249-336

Liven, J. (1985). A Metrical Theory of Syllabicity. Ph.D Dissertation, MIT.

Levy, M. (1971). The Plural of the Noun in Modem Standard Arabic. Ph.D Dissertation.
University of Michigan. Ann Arbor: University Microfilms International.

Levy, M. & Fidelholtz, J. (1970). Arabic plurals, rule features and lexical features.
Glossa 5,57-70
Lombardi, L. (1996). Why place and voice are different, m.s. University of Maryland,

Lombardi, L. and McCarthy, J. (1991). Prosodic circumscription in Choctaw

morphology. Phonology 8, 37-71

McCarthy, J. (1979). Formal Problems in Semitic Phonology and Morphology. Ph.D

Dissertation, MIT, Cambridge, Mass.

McCarthy, J. (1981). A Prosodic Theory of Non-concatenative Morphology. Linguistic

Inquiry 12,373-418

McCarthy, J. (1982). Prosodic structure and expletive infixation. Language 58, 574-590

McCarthy, J. (1985). The phonetics and phonology of Semitic pharyngeals. In Keating

(ed.). Papers in Laboratory Phonology 3: Cambridge: CUP. 191-283

McCarthy, J. (1986). OCP effects: Gemination and Antigemination. Linguistic Inquiry

17,207-263

McCarthy, J. and Prince, A. (1986). Prosodic Morphology, manuscript, University of

Massachusetts and Brandies University.

McCarthy, J and Prince, A. (1990). Foot and word in prosodic morphology: The Arabic
broken plural. Natural Language and Linguistic Theory 8,209-283.

McCarthy, J & Prince, A. (1990b). Prosodic morphology and templatic morphology. In

M. Eid and J. McCarthy (eds.). Perspectives on Arabic Linguistics: Papers from the
second Symposium, Amsterdam: John Benjamins. 1-54

McCarthy, J and Prince, A. (1993a). Prosodic Morphology I: Constraint interaction and

satisfaction. Ms.. University of Massachusetts, Amherst, and Rutgers University. MIT
Press.

McCarthy, J and Prince, A. (1993b). Generalized alignment. In G. Booij and J. van

Marle. Yearbook ofMorphology 1993. Dordrecht: Kluwer. 79-153

McCarthy, J and Prince, A. (1994a). The emergence of the unmarked: Optimality in

Prosodic Morphology. Proceedings of the North East Linguistic Society 24. GLSA,
University of Massachusetts, Arnherst. 333-379

McCarthy, J and Prince, A. (1995). Faithfulness and Reduplicative Identity. In J.

Beckman, S. Urbanczyk and L. Walsh (eds.). University of Massachusetts Occasional
Papers in Linguistics 18: Papers in Optimality Theory, Graduate Linguistic Student
Association, Amherst, Massachusetts. 249-384.
 McCarthy, J. (1998). Morpheme structure constraints and paradigm occultation. Papers
from the Regional Meetings 34, Chicago Linguistic Society. 125-150

McCarthy, J. (2000). Faithfulness and prosodic circumscription. In J. Dekkers, F. van der

Leeuw, and J. van de Weijer (eds.), Optimality Theory: Phonology, Syntax and
acquisition. Amsterdam: HIL. 151-189

McCarthy, J. (2002). A Thematic Guide to Optimality Theory. Cambridge University

Press: New York.

Paradis, C. and Prunet, J. (1990). The coronal vs. placelesness controversy. McGiN
Working Papers in Linguistics 6, 192-228.

Prochazka, T. (1981) The Shi'i dialects of Bahrain and their relationship to the eastern
Arabian dialects and the Omani dialect of al-Ristaq. Journal of Arabic Linguistics 6, 16-
55.

Prince, A & Smolensky, P. (1993). Optimality Theory: Constraint interaction in

generative grammar. Ms., Rutgers University, New Brunswick, and University of
Colorado, Boulder. MIT Press.

Prince, A. (1980). A metrical theory for Estonian quantity. Language 66,511-62

Pulleyblank, D. (1988). Vocalic underspecification in Yoruba. Linguistic Inquiry 19, 233-

270

Ratcliffe, R. (1988). Dejning Morphological Isoglosses: The Broken Plural and Semitic
sub-class~jcation.Tokyo Gaikokugu Daigaku Publishing Company: Japan.

Ratcliffe, R. (1990). Arabic broken plurals: Arguments for a two-fold classification of

morphology. In M. Eid and J. McCarthy (eds.). Perspectives on Arabic Linguistics 11:
Papersfrom the second Symposium, Amsterdam: John Benjamins. 94-1 19

Ratcliffe, R. (1992). The Broken Plural Problem in Arabic, Semitic and Afroasiatic: A
Solution Based on the Diachronic Application of Prosodic Analysis. Ph.D Dissertation.
Yale University.

Ratcliffe, R. (1996). Drift and noun plural reduplication in Afroasiatic. Bulletin of the
School ofOriental and African Studies, LIX (2). 296-3 11.

Reinhardt, C. (1894). Ein Arabischer Dialekt Gesprochen in 'Oman and Zanzibar. Berlin:
W. Spemann.

Rose, S. (1997). Theoretical Issues in Comparative Ethio-Semitic Phonology and

Morphology. Ph. D. Dissertation, McGill University.
Sa'adeddin, M & Thelwall, R. (1999). Handbook of the International Phonetic
Association. Cambridge University Press. 5 1-54

Shaaban, K.A. (1977). The Phonology and Morphology of Omani Arabic. Ph.D.
Dissertation. University of Texas, Austin.

Shaaban, K.A. (1985). Regularization processes in Arabic dialects: The verb in Omani
Arabic. Arab Journal of Language Studies: Journal of the Khartoum International
Institute of Arabic, 4, 157- 165

Spaelti, P. (1997). Dimensions of Variation in Multi-Pattern Reduplication. Ph. D

Dissertation. University of California, Santa Cmz.

Smolensky, P. (1 993). Harmony, markedness and phonological activity. Talk presented at

Rutgers Optimality Workshop 1.

Urbanczyk, S. (1995). Double reduplication in parallel. University of Massachusetts

Occasional Papers in Linguistics 18: Papers in Optimality Theory. In J . Beckman, L.
Walsh Dickey, S. Urbanczyk (eds.). GLSA, University of Massachusetts, Amherst. 499-
532

Urbanczyk, S. (1996a). Morphological templates in reduplication. Proceedings of NELS

26. GLSA, University of Massachusetts, Amherst. 425-440

Urbanczyk, S. (1996b). Patterns of Reduplication in Lushootseed. Ph.D Dissertation.

University of Massachusetts, Amherst.

Urbanczyk, S. (1999). Echo vowels in Coast Salish. In M. Coldecott, S. Gessner and E-S
Kim (eds.). Workshop on Structure and Constituency in the Languages of the Americas
(WSCLA), University ofBritish Columbia Working Papers in Linguistics 2, 165-184

Ussishkin, A. (1999). The inadequacy of the consonantal root: Modem Hebrew

denominal verbs and output-output correspondence. Phonology 19,401-442.

Watson, J. (2002). The Phonology and Morphology of Arabic. Oxford University Press:
UK

Webster, R. (1991). Notes on the dialect and way of life of the Al-Wahiba Bedouin of
Oman. Bulletin ofthe School of Oriental and African Studies 54 (3), 473-485

Wright, W. (1971). A Grammar of the Arabic Language. Cambridge University Press.

Yip, M. (1988). Template morphology and the direction of association. Nntiiral

Language and Linguistic Theory 6, 55 1-577
 APPENDIX A
Muscat Broken Plural Nouns

Note: - ah is a marker of feminine gender in the Muscat dialect of Omani Arabic.

Grouu 11)

Sing. PI. Gloss

CVC.CV.C Singular forms (CV.CV:).CV.C
matbax ma@:bax 'kitchens'
daftar dafa:tar 'notebooks'
sakSan saka:Sin 'sections'
xangar xana:gar 'daggers'
marham mara:hum 'lotions'
margal' mara:gaS 'references'
maxbaz maxa:biz 'bakeries'
Saqrab 7aqa:rub 'scorpions'
talsam tala:sum 'curing spells'
manvad mara:wid 'metal bars of iron used to
put kohl in the eyes'
maktab 'offices'
?arnab 'rabbits'
matSam 'restaurants'
mazbal 'lips'
mathaf 'museums'
manbar 'pulpits'
mashaf ' copies of Holy Qura'an'
misgid 'mosques'
margal 'cauldrons'
xandaq 'ditches'
tahlab 'fungus'
gadwal 'timetables'
manzil 'houses'
markab 'small boats'
funduq 'hotels'
masbah 'showers'
mulhaq 'appendices'
bufdaS 'frogs'
sandal 'blocks of perfumed tree
 woods'
farmal 'brakes'
marsam 'places for artists to draw'
malYab 'play grounds'
mablay 'sums of money'
makaq 'crossroads'
marfa'i 'wooden lifting apparatus
for the Holy book'
mahzam 'belts'
malbaq 'matches'
madxal 'entrances'
maxrag 'exits'
madfa'i 'canons'
ma~raf 'banks'
madrab 'cricket bats'
maghar 'microscopes'
maxba? 'hiding places'
mangam 'mines'
marqas 'casinos'
markaz 'centers
maahab 'Islamic sects'

maqSat 'matches'
maq~af 'small vendors'
maSxal 'sieves'
magmar 'incense burners'
maglas 'rooms for guests'
maqbaa 'door knobs'

mu'iris 'bridegrooms'

Sing. PI. Gloss

CVC.CV.C- ah Singularforms (CV.CV:).CV.C
mastar- ah masa:tar 'rulers'
malTaq- ah rnalx'iaq 'spoons'
 maqlam- ah maqa:lum 'pens' cases'
mahkam-ah maha:kum 'courts'
qunbal-ah qana:bil 'bombs'
mathan-ah mata:hin 'mortars'
mazray- ah maza:raY 'farms'
marhal-ah mara:hil 'stages'
madras-ah madaxis 'schools
mantaq-ah mana:taq 'regions'
marwah-ah marxwah 'fans'
C(CV:).CV.C
maqbar- ah mqa:bar 'graveyards'
maysal- ah mya:sil 'sinks'
(CV.CV:).CV
sufrij- ah safa:ri 'cooking pots'
kurfaj- ah kara: fi 'beds'
CVCVC-ah Singular form
qagij-ah 'cases'

Group (3)

Sing. PI. Gloss

CVC.CV Singular forms (CV.CV:).CV
maksi maka:si 'dresses'
kursi kara:si 'chairs'
muxba maxa: bi 'pockets'
muqla maqadi 'frying pans'
malha ma1a:hi 'casinos'
maqha maqa:hi 'cafes'
marsa mara:si 'harbors'

Group 14)

Sing. PI. Gloss

CV.CV:C- ah Singularforms (cv.cv:). jV.C
maki:n- ah maka:jin 'machines'
hadi:q- ah Rada:jaq 'parks'
gari:d- ah gara:jid 'newspapers'
i)abi:h-ah 8aba:jah 'goats slaughtered in the
 Islamic way, kosher meats'
xari:t-ah xara:jat 'maps'
qabi:l-ah qaba:jil 'tribes'
nati:g-ah nata:jig 'results'
xabi:&ah xaba:jie 'great sins'
kani:s-ah kana:jis 'churches'
8adi:d- ah hada:jid 'iron bars'
Yazi:m- ah Yaza:jum 'invitations /feasts'
risa:l-ah rasa:jil 'letters'
wsa:d- ah wasa:jid 'pillows'
CV.CW.C singularform CV.CVV.jV.C
badi:l bada:jil 'alternatives'
5'asi:r 7asa:jar 'juices'
8asi:r hasa:jar 'carpets'
CVV.CVV.C Singularform (CV. jV:). CVV.C
Ji:ta:n Jaja:ti:n 'devils'

Group (5)

Sing. PI. Gloss

CVV.CV.C Singular forms (CV.WV:).CV.C
Sa:aar Sawa:aar 'bed blankets'
Sa:raY Sawa:raY 'roads'
to:bag tawa:big 'round, flat pans used to
bake Omani bread'
dawa:fay 'motives'
Yawa:mil 'factors'
8awa:die 'car accidents'
dawa:bat 'restrictions'
gawa:maY 'large mosques'

Group 16)

Sing. PI. Gloss

CVV.CVV.C Singularforms (cv.wv:).(cv:).c
ta:bu:r tawa:bi:r 'line-ups'
ta:bu:q tawa:bi:q 'cement blocks'
 'laws'
'missiles'
'dictionaries'
'dates'

Sing. PI. Gloss

CVV.CV Singular forms (CV.wV:).CV
ri:du rawa:di 'radios'
gu:ti gawa:ti 'pairs of shoes'
qo:ti qawa:ti 'cans'
(CV.wV:). jVC
rawa:jah 'bad smells'

Group 18)

Sing. PI. Gloss

CVV.CV- ah Singularforms
sa:lf-ah 'gossips'
qa:jm-ah 'lists'
sa:Yq-ah 'thunderbolts'
qa:Yd-ah 'rules'
mo:xr- ah 'noses'
mo:qS-ah 'grinders'
Sa:jl-ah 'families'
ta:jr- ah 'tires'

ga:rj-ah 'young women'

CVV.CVC- (ah) Singular form
go:nij- ah 'large sacks'
Sa:rub 'mustaches'
ta:baq 'floors'
ha:mil 'pregnant'
Ya:nis 'spinsters'
ra:tub 'monthly allowances'
 C W . C - ah Singular forms
ha:r- ah hawa:jar 'small streets in a village
or town'
cwvc
swar 'pictures'

Group (9)

Sing. PI. Gloss

CVC2CzV:C Singular forms (CV.C2V:).(C2V:).C
sakki:n saka:ki:n 'knives'
0alla:g- ah 0ala:li:g 'refrigerators'
karra:s- ah kara:ri:s 'drawing pads'
Subba:k Saba:bi:k 'windows'
dukka:n daka:ki:n 'shops'
kal1a:b ka1a:li:b 'clips /wrenches'
dabbus daba:bi:s 'pins'
tannu:r-ah tana:ni:r 'skirts'
sugga:d-ah saga:gi:d 'prayer mats'
xajja:t xaja:ji:t 'tailors'
$bba:x taba:bi:x 'cooks'
sannu:r sana:ni:r 'cats'
muhiia:r maha:hi:r 'sea shells'
sal1u:m sa1a:li:m 'sun flower seed'

Sing. PI. Gloss

CVCCV:C-ah Singularforms (CV-CV:)-(CV:).C
kirxa:n- ah kara:xi:n 'machines'
dihda:S- ah daha:di:S 'traditional Omani males'
outfits'
'toothbastes'

Group (I I )

Sing. PI. Gloss

CVC.CV:.C Singularforms (CV.CV:).(CV:).C
barqu:'? bara:qi:T 'veils'
 'kids' toys'
'traditional Omani boxes'
'boxes'
'lanterns'
'nails'
'knuckles'
'powders'
'handkerchiefs'
'small coffee cups'
'interpretations of the Holy
Qura'an'
'SOWS'
'birds'
'royal decrees'
'teachers'
'pairs of pants'
'details'
'pegs'
'structures'
'blankets'
'papers'

'ladles'
'curing spells'
'hooks'
'keys'

Group (12)

Sing. PI. Gloss

CCVCC-ah Singularforms C(CV:).CV.C
mqamS- ah 'spoons'
mbaxr- ah 'baskets with holes used to
perfume the clothes'
C.CV:C-ah Singular form C(CV:).jV.C
dri:S- ah dra:jiS 'windows'
 Group (13)

Sing. PI. Gloss

C.CVV.C Singular forms (CVC).(CV:).n
nYa:l nuYla:n 'sandals'
tra:b turba:n 'sands'
gda:r gidra:n 'walls'
hza:q hizqa:n 'belts'
Ysu: Yiswa:n 'sticks taken from palm trees'
Ra: fu7ja:n 'snakes'
dla:y dilya:n 'pairs of socks'
gla:~ gil~a:n 'glasses'
kra:z kurza:n 'tubes'
qfi:r qufra:n 'baskets made of palm leaves'
y1a:m yilma:n 'slave boys'
Ihxf luhfa:n ' head scarves'
bla:d bilda:n 'countrysides'
xru:f xirfa:n 'sheeps'
yza:l yizla:n 'gazelles'
ydi:r yidra:n 'brooks'
1Ra:m 1uRma:n 'meats'
q$:b qu4ba:n 'twigs'
Y3u:m Yu3ma:n 'bones'
qmi:s qumsa:n 'shirts'
yra:b yirba:n 'crows'
Sda:y Sidya:n 'cheeks'
3fu:r qufia:n 'fingernails'

Group (14 )

Sing. PI. Gloss

CVV.C Singular forms (CV.CV:).n
Ro:S RiSa:n 'backyards'
yu:I yila:n 'snakes'
zo:T ziYa:n 'vomitings'
ba:b biba:n 'doors'
 'bulls'
'coats'
'legs'
'cups'
'fires'
'elbows'
'whales'
'crowns'
'shoes'

Grouu 1151

Sing. PI. Gloss

CCVC-(ah) Singular forms C.(CV:).C
mraS mra:S 'traditional bottles for
spraying perfume'
mla f 'folders /binders'
mnaz 'cots'
mxam- ah 'brooms'
mSut 'combs'
mkab- ah 'bowls'
msal 'sieves'
mgaz 'grass cutters'
qfil 'locks'
sbuY 'fingers'
msar 'turbans'

'fingers'

Group 116)

Sing. PI. Gloss

CVCC-(ah) Singular forms C.(CV:).C
Suxt Sxu:t 'lines'
nad nqu:J 'decorations'
bahe bhu:0 'research projects'
 sign 'prisons'
Sar! 'conditions'
6arf 'envelopes'
xamr 'wines'
hi~n 'forts'
durs 'teeth'
rig1 'legs'
'2x6 'lands'
qalb 'hearts'
sahl 'hills'
nagm 'stars'
ganb 'sins'
Sahr 'months'
Sumx 'scratches'
hurz 'curing spells'
kalb 'dogs'
katf 'shoulders'
8ilg 'snows'
xurq 'holes'
hurq 'burns'
harf 'letters'
kurS 'bellies'
naxl- ah 'palm trees'
qalT-ah 'forts'
SumT- ah 'candles'
gahl- ah ' pieces of pottery'
mall- ah 'bowls'
sahl- ah 'pots for washing hands'
warq- ah 'papers'
Sagr- ah 'trees '
hum-ah 'women'

xalq-ah 'garments'

xanS- ah 'trolleys'
buqY-ah 'stains'
durg- ah drag 'stairwells'
 Jant- ah Snat 'hand bags/ suitcases'
dakj- ah dkaj 'pillows'
xatm- ah xtam 'copies of the Holy Qura'an'
yadf-ah ydaf 'scarves'
hufr-ah hfar 'holes'
hilq- ah hlaq 'earrings'
yarS- ah yraS 'bottles'
hugr-ah hgar 'rooms'
Yulb-ah Ylab 'cans'
bugm- ah bgam 'bulging bruises'
luhj-ah Ihaj 'beards'
bidY-ah '.~nnovations'
bdaY
(CV.CV:). jV.C
watj- ah wata:jah 'sandals'

Sing. PI. Gloss

CVVC-(ah) Singularforms C.dV:).C
Ye:n Yju:n 'eyes'
be:t bju: t 'houses'
ae:1 8ju:l 'tails'
se:h sju:R 'barren lands'
se:f sju:f 'swords'
Ye:S Yju:S 'rices'
Se:x 5ju:x 'tribal chiefs'
CCtV:C2
ain anejn 'ears'
xat xtu:t 'lines'
Sat Stu:t 'covers'
ha1 hlu:l 'solutions'
saf sfu:f 'classrooms'
ra f rfu:f 'shelves'
xad xdu:d 'cheeks'
$1 41a:l 'shades'
sad sdu:d 'dams'
fax 'traps'
had 'limits'
kaf 'palms of the hands'
mux 'brains'
kum- ah 'Omani traditional caps'
yub-ah 'swamps'

Sing. PI. Gloss

CVCVC Singular forms C.(CV:).C
aha an &u:n 'plates'
tabal tbu:l 'drums'
gabal gba:l 'mountains'
gamal gma:l 'camels'
Sabak Sba:k 'nets'
wagaY wga:Y 'pains'

Group (1 9)

Sing. PI. Gloss

CV:CVC Singular forms C.CV.C
xa:turn xtum 'rings'
CV:C-ah Singularforms C.jV.C
sa:Y- ah sjaY 'watches'
 APPENDIX B
Glossary of Technical Linguistic Terms

Below I provide a glossary of the technical linguistic terms and Optimality Theory
constraints used in this thesis.
1. Technical Linguistic Terms
AFFIXRefers to an invariant string of segments attached to a BASE. It denotes a particular
meaning. Affixes are classified into three types depending on their position with
respect to the base. Prefixes are attached to the beginning of the base, suffixes are
attached to the end and infixes are added in the middle of the root.

AUTOSEGMENT Refers to a feature which is considered to be independent and

represented on its own TIER. For example, TONE is seen as an independent feature
of the consonant and vowel contained together with the tone in the same word.

BASERefers to the basic unit of segments in a word to which a morphological operation

can be applied. It is used alternatively with STEM.

BROKENPLURAL Noun plural which is expressed by an internal change of the singular

STEM. An example to illustrate the broken plural is [masa:gid] 'mosques' whose
singular is [misgid]. The ClVC2 portion of the singular stem which corresponds to
[mis] is expanded into C I V C ~ Vwhich
: refers to [masa:] in this particular form.

CANONICAL Refers to a linguistic form or word shape cited as a norm or standard for the
purposes of comparison (Crystal 1997: 51). For instance, the most common
syllable shape is CV wherein C refers to a consonant and V to a vowel.

CODAThe closing consonant of a syllable (Crystal 1997: 375). C2 in a CVCz syllable is a

coda.

DEFAULT
Refers to an UNMARKED segment or feature that emerges in a morphological or
phonological operation either automatically or when certain conditions apply.

EXTRAMETR~CALITY A notion in METRICAL THEORY which designates a consonant or

particular PROSODIC CONSTITUENT such as a FOOT or SYLLABLE as invisible for the
purpose of rule application (Hayes 1995). In stress assignment, the final syllable
can be invisible to the assignment of stress.

FIXEDSEGMENTISM Refers to invariant segments that emerge in certain morphological

or phonological processes. For example, in reduplication, fixed segments can
emerge in reduplicated affixes instead of expected copied segments from the base.
FOOTRefers to the grouping of syllables into metrical units. A foot maximally consists of
two syllables and minimally one syllable. There are two major foot types: iambic
and trochaic.

GEMINATE (GEMINATION) Refers to a string of identical adjacent consonants in a word,

e.g. [sakki:n] 'knife' has a geminate consonant /kt.

HEAVYSYLLABLE A syllable shape which has two units of weight (morae) either
because it has a long vowel CVV or its coda consonant is moraic CVC. Whether
the final consonant is moraic or not depends on language-particulars.

IAMBICFOOT A right-headed foot (the second syllable contained in this foot is more
prominent). It can be either disyllabic or monosyllabic. If disyllabic, it has a light-
heavy syllable (CV.CVV) dubbed as typical iamb or light-light syllable (CV.CV).
If monosyllabic, it is always heavy (CVV) and known as even iamb.

LIGHTSYLLABLE A syllable shape which has a single unit of weight (monomoraic).

Syllables such as CV and V are light. In some languages, CVC is also a light
syllable because the coda consonant is nonmoraic.

MARKEDNESS A crucial notion in linguistics whereby pairs of linguistic forms or

features, seen as distinct, are evaluated as frequent and common (unmarked) and
less common (marked) (Crystal 1997: 233). For example, in syllable theory, a
CCV syllable shape is considered to be marked because it has a consonant cluster
while CV is regarded as unmarked cross-linguistically. In OT, markedness
CONSTRAINTS evaluate marked structures in OUTPUT forms.

MELODICOVERWRITING Allows fixedfinvariant segmentsfmelodies to be overwritten in

the SKELETAL TIER of that form.

METRICAL THEORYOriginally introduced as a hierarchal theory of stress to reflect

relation of prominence between morphological and syntactic constituents. In this
theory, phonological strings are represented using notions of segment, syllable,
foot and prosodic word (Crystal 1997: 241).

MOM A unit of weight. The analysis of segments into morae is applied to syllabic
NUCLEUS and CODAS and not to ONSETS. Heavy syllables are bimoraic (contain
two morae) whereas light syllables are monomoraic (contain a single mora on the
vowel). Codas can be nonmoraic in some languages.

NON-LINEAR PHONOLOGY Assumes that phonological representations such as features

and tones be represented in hierarchical levels and not on a single level.

NUCLEUS The central segment of a syllable (Crystal 1997: 375). V in a CV syllable shape
is the centre of the syllable.
ONSETThe beginninghnitial consonant of a syllable. C in a CV syllable pattern is an
onset.

PROSODVAssumes that phonological words are represented in a metrical hierarchy

which is composed of the units of prosody such as- MORA, SYLLABLE, FOOT and
prosodic word. It also refers to non-segmental aspects of speech such as pitch and
duration.

SOUNDPLURAL Noun plural which is formed by a regular AFFIXATION process whereby

an AFFIX is attached to the singular stem. An example to illustrate sound plurals is
[qibar-aat] 'trains' whose singular is [qitaar]. The plural suffix- aat is attached to
the singular stem Iqitaarl.

REDUPLICATION A morphological process of copying wherein an AFFIX copies segments

of the BASE.

SYLLABLEA unit of pronunciation typically larger than a single sound and smaller than a
word (Crystal 1997:373). The combination of sounds in individual languages
creates sequences of sounds. A consonant-vowel (CV) sequence is a syllable
shape which commonly occurs in all languages. Because this syllable isn't closed
by another consonant, it is named open syllable. Syllables can either be LIGHT as
in V and CV or HEAVY as in CVV.

Vowel deletion.
SYNCOPE

STEMConstitutes the basic element of the structure of a word. It has a distinct meaning.
It refers to the BASE to which a certain morphological process such as affixation is
applied to express different meanings.

TEMPLATE Refers to a fixed phonological pattern (shape-invariant) of words whose

actual segments/melodies are represented as a sequence of consonants C and
vowels V. Arabic is famous for expressing different meanings by specifying a
fixed word shape. For example, verbs of the CVC,C,VC shape are perfect active.

TIER A level of phonological representation wherein segments and features are

represented. SKELETAL TIER is the tier which includes an abstract representation
of the consonants (C) and vowels (V) contained in a particular form. Melodic tier
has the actual segments of a word. The C and V in the skeletal tier associates to
the actual segments by lines called association lines.

TONEA term used in phonology to refer to the pitch level of a syllable.

TROCHAIC FOOT A left-headed foot which has either two light syllables with the stress
falling on the left or a single heavy syllable. There are two types of trochaic feet:
moraic trochee vs. syllabic troches. Moraic trochees as in CVC shape have two
 morae. Syllabic trochees as in CVCV are composed of two light syllables with the
stress falling on the left. Trochaic feet have the shapes CVC, CVCV or CVV.

2. Optimality Theory Constraints

ALIGNMENT CONSTRAINTS A family of constraints in OPT~MAL~TY THEORYwhich

demands particular edges of categorieslconstituents to be aligned to some edges of
other categorieslconstituents.

*ComplexOnset A markedness CONSTRAINT which evaluates the syllable shape of

OUTPUT forms. It bans clusters of consonants at the left-edge of syllables.

CONSTRAINTS
Linguistic filters which evaluate the structures of output forms.

Correspondence Theory Assumes that any two strings of words have correspondence
relations. The general formalism of this theory is "Given two strings SI and Sz,
correspondence is a relation R from the elements of SI to those of S2. Segments a
(an element of S I ) and p (an element of Sz) are referred to as correspondents of
one another when a R (McCarthy and Prince 1995). An example of strings in
correspondence includes input-output pairings.

CrispEdge alignment Refines alignment by requiring that identical segmentdfeatures

(geminates or shared features, for example) in a word be contained within a
PROSODIC CONSTITUENT. The general formalism of this constraint is "Let /A/ be a
terminal substring in a phonological representation, 3 a category of type Pcat and
/A/= 1 3 1 (the content of 3). Then 3 i s crisp (has crisp edges) if and only if A is-a
3:'d A (/A/ = 1313/A/ = 3)"(It6 and Mester 1999).

CrispEdge (F) An example of CrispEdge constraints which requires identical

segrnentslfeatures (such as geminates) be contained within a FOOT.

DEP A general faithfulness constraint which bans epenthesis of segmentslfeatures. It

requires the relation between correspondent strings be complete. If the output
form adds a feature or segment, then this constraint is violated.

DEPOO-F1 (p) A POSITIONAL FAITHFULNESS constraint which bans insertion of morae in

the first foot of the correspondent strings. If a mora is added to the first foot of a
related word, then this constraint is violated. The general formalism of this
constraint is "For every p in the first F of outputz, there is a correspondent p in
the first F of output,".

DEPOO-OF (p) A POSITIONAL FAITHFULNESS constraint which bans deletion of morae in

the final syllable of the correspondents strings. If a mora is added to the final
syllable of a related word, then this constraint is violated. The general formalism
of this constraint is "For every p in the final o of outputz, there is a correspondent
p in the final IS of outputl".
FAITHFULNESS CONSTRAINTS A family of constraints in Optimality theory which
monitors identity between correspondent forms. They penalize epenthesis,
deletion, metathesis and featural change of output forms.

FINAL-CA markedness constraint which requires words to end in a consonant.

GENERALIZED ALIGNMENT Demands particular edges of categories1 constituents to be

aligned to some edges of other categories/constituents. To illustrate, it requires a
particular edge of a prosodic or morphological constituent of Cat1 coincide with a
particular edge of some other prosodic or morphological constituent of Catz. The
formalism of this constraint is "Align (Cat,, Edgel, Cat2, Edge2) = d,f
V Cat, 3 Cat2 such that Edgel of Cat1 and Edge2 of Catz coincide.
Cat!, Cat2 E Pcat u Gcat
Edge!, Edge2 E {Right, Left}" (McCarthy and Prince 1993).

Generalized Template Theory (GTT) Dispenses with constraints that articulate a

particular CV-shape or specific foot type of the surface form. It proposes that
interaction of constraints of Universal Grammar yields the outcome shape of a
word.

IDENT
(F) A faithfulness constraint which requires segments contained in pairs of
correspondent strings of words have identical values for the feature (F). It is
violated when a segment changes its value.

A linguistic form from which potential outputs are generated and filtered by a
INPUT
ranked set of constraints.

MAX A general faithfulness constraint which bans deletion of segments, features or

morae. It stipulates that the relation of correspondent strings must be complete

*ppp]a A markedness constraint which evaluates the syllable shape of output forms. It
penalizes trimoraic syllables (syllables with three morae).

*a [C,C, A markedness constraint which evaluates the syllable shape of output forms. It
prohibits geminates at the left-edge of syllables.

MAXOO- (p) A POSITlONAL FAlTHFULNESS constraint which bans deletion of morae in

the final syllables of the correspondent strings. If a mora is lost from the final
syllable of a related word, then this constraint is violated. The general formalism
of this constraint is " For every p in the final o of outputl, there is a
correspondent p in the final (T of output?.

NO-DELINK(p, SEG)An example of constraints demanding conservation of association

lines. It checks that segment to mora linkage not delinked in the mapping of
 correspondent strings. For example, delinking of a certain mora associated with a
particular segment in the input from the output causes violation of this constraint.

NO-SPREAD (p, SEG) An example of constraints demanding conservation of association

lines. It checks that segment to mora linkage be preserved in the mapping of
correspondent strings. For example if a mora is linked to the second consonant in
the input form, then it has to be linked to the same consonant in the output form.

ONSETA markedness constraints which evaluates the syllable shape of output forms. It
penalizes syllables which have no consonants at the left-edge of them.

OPTIMALThe final OUTPUT is known as optimal; it satisfies high-ranking constraints at

the expense of violating low-ranked ones. It is the most harmonic form.

OPTIMALITY THEORY (OT) A constraint-based model of grammar. It assumes that

linguistic forms are evaluated by a ranked set of constraints supplied by Universal
Grammar.

OUTPUT
is the surface form of a linguistic form.

OUTPUT-OUTPUT CORRESPONDENCE Extends the correspondence relation from

underlying-surface forms to include correspondence relations between
morphologically related forms.

PLACEMARKEDNESS CONSTRAINTS Evaluates place markedness of segments in surfaces

forms. They are contained in a fixed constraint hierarchy which proposes that
coronal place is less marked than labial and dorsal. It is thus more likely to find
more languages with coronals than with labials. The general formalism of this
hierarchy is *Pl/labial, *Pl/dorsal D *Pl/coronal N *PI/ pharyngeal

POSITIONAL FAITHFULNESS Captures the fact that certain positions in a word are more
privileged in that they maintain phonological distinctions.