Noun Plurality in Jebbali


Material Information

Noun Plurality in Jebbali
Physical Description:
1 online resource (243 p.)
Al Aghbari, Khalsa
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Committee Chair:
Wiltshire, Caroline R
Committee Members:
Mclaughlin, Fiona
Henderson, Brent Mykel
Issa, R. Raymond


Subjects / Keywords:
jebbali -- noun -- plurality
Linguistics -- Dissertations, Academic -- UF
Linguistics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


This is a morphophonological study of noun plurality in Jebbali, a Modern South Arabian language spoken in the mountains and coastal plains of Dhofar, Oman. It has a twofold goal: (1) it documents the diverse shapes of noun plurals in Jebbali and (2) provides a formal analysis of the most systematic plural shapes within Optimality Theory (Prince and Smolensky 1993/ 2004). It also examines the peculiar morphophonology of the exceptional plural shapes and discusses how they diverge from the systematic plurals. Finally, it investigates approaches to exceptionality and lexical marking in Optimality Theory. Plurality in Jebbali is expressed by two distinct mechanisms: suffixation and non-concatenative operations such as infixation of Vb, attachment of a suffixal VC template with fixed vocalism and a copy of the final base consonant, mapping geminated singulars onto a specific plural template and ablaut. Jebbali also has irregular plural shapes whose exceptionality is not attributed to their singulars and which cannot be explained phonologically. The analysis of the regular plural patterns reveals that well-motivated constraints can capture their regularity and address their diversity. For instance, the locus of the Vb infix follows from an alignment constraint, a constraint placing restriction on syllable size and a Left anchoring constraint, which together function in harmony with the prosody of the language as a whole. The vowel of the suffixal VC template is prespecified underlyingly and results from MAX-V-SUFFIX outranking MAX-V-ROOT. The final plural shape is reduced to a single syllable and determined by the markedness constraint NO-V (Bakovic 2005: 299). Plurals derived from geminated singulars map onto a specific template; this results from the interaction of constraints such as IDENTQ (Dell and Elmedlaoui 1992) and *VGG# (Muller 2001). Finally, I offer three cogent analyses to the ablaut plurals: (1) positional faithfulness (Beckman 1998), (2) Anti-faithfulness (Alderete 1999a and 2001) and RealizeMorpheme (Kurisu 2001). The last chapter uncovers the exceptionality of the irregular plural shapes and justifies their failure to be captured by a unified Optimality Theoretic analysis. I finally list some of the approaches of dealing with exceptionality in Optimality Theory and apply them to the analysis of these shapes.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Khalsa Al Aghbari.
Thesis (Ph.D.)--University of Florida, 2012.
Adviser: Wiltshire, Caroline R.
Electronic Access:

Record Information

Source Institution:
Rights Management:
Applicable rights reserved.
lcc - LD1780 2012
System ID:

This item is only available as the following downloads:

Full Text




2 2012 Khalsa Al Aghbari


3 For Ahmed


4 ACKNOWLEDGMENTS On one hand, it was a great asset for me to get to do a PhD. On the other hand, it was a real challenge in which I stro ve to keep my sanity while embarking on the biggest project of my life. As I pass ed every phase of the PhD program, my eyes had been set on the torch at the end of the tun n el, waiting anxiously for the time when I would pass my PhD defense. As I reach ed there and held the torch tenaciously I discover ed tha t defending my thes is was not the final destination but a start off point to leaving my own prints in my passionate field of study, phonology. Now that I hold a PhD I do not cease to feel huge burden of disseminating knowledge, accumulated in p recious four years of my lifetime, through teaching at Sultan Qaboos University and doing research. Now that I hold a PhD, I surely savor its sweetness and bitterness, resulting from hard work, patience and determination. Now that I hold a PhD, I do not im agine myself stepping back, stand ing as a passive observer to the long l inguistic wealth and discontinuing research. I promise to be an effective linguist and researcher. Today and forever I am short for words that would appropriately and justly express my gratitude to my wonderful supervisor, Caroline Wiltshire wh o taught me when to slow down and when to speed up. Her challenging exa ms when she taught me phonology and professional supervision coupled with her insightful comments when she was my supervisor are the most fruitful moments of my life I cannot choose the right words to tell you how much you have touched me by your sheer humbleness and immense knowledge I wish I could repay little of what you have done to awaken the researc h p erson in me I promise to follow your path in both teaching and research.


5 I also feel privileged to have learned linguistics from a reputable host of scholars in the Department of Linguistics at U niversity of F lorida in general and from my committee mem bers in particular Diana Boxer, Ratree Wa yland and MJ Hardman have enhanced my love for linguistics. I was also honored to have Fiona McLaughlin and Bren t Henderson in my committee. I will never forget their moral support and constructive feedback while I was writing my qualifying exams and in my PhD defense Last but not least I thank Raymond Issa, my external examiner who was accessible during my candidacy exam and PhD d efense. I wish to extend special thanks to my language consultants Manal Bait Ghar im, Jamila Kashoub for their willingness to share their language with me and their interest in my project I wholeheartedly appreciate the time they generously spared me and the enthusiasm they displayed when knowing that I was working I dedicate my dissertation with profound love to my family members, especially to my beloved grandmother, without whose continuous prayers and phone calls it would have been impossible to get this work done. Inquiring about my progress and h ow much time left to graduate was very sp irit elevating. I know you barely understood what I was exactly doing but it was enough for me that you could cunningly and amusingly mimic Had I not been bl essed with an extremely understanding husband and an adorable daughter, the journey of d oing a PhD would have been flavor constant phone calls and visits were my sole fuel when I was at the end of my tether ily and distracted me from my recurring worries.


6 Naively, I thought that being a single mother in the States with the challenge of completing a PhD would disrupt my life and deprive me of enjoying my daughter However, as I began to enjoy the growing up of my baby in front of my eyes, I realized that I would not be as successful in my education and life without her. Thank you both for asking me only often when I will be done. It certainly motivated me to finish earlier than expected Last but not least, I w ish to express special thanks to my revered university, Sultan Qaboos University, which generously paid for my education and life expenses for four years. I have never felt financially lacking though I would have appreciated little money for conferences S ince I did not have to work to pay for my education, I have had plenty of time to study and submit my assigned works way before a deadline.


7 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 10 ABSTRACT ................................ ................................ ................................ ................... 11 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 13 Statement of Intent ................................ ................................ ................................ 13 Overview of Dissertation ................................ ................................ ......................... 17 Description of the Plural Patterns ................................ ................................ ........... 19 Suffixation ................................ ................................ ................................ ......... 21 Vb Infixation ................................ ................................ ................................ ...... 24 Attachment of a Suffixal VC Template ................................ .............................. 26 Ablaut/ Vocalic Opposition ................................ ................................ ................ 28 Templatic Plurals ................................ ................................ .............................. 29 Plurals derived from gemin ated singulars ................................ .................. 29 Plurals with truncation and templatic expansion ................................ ........ 29 Gender in Singular Plural Mappings ................................ ................................ ....... 32 2 AN OVERVIEW OF JEBB ................................ ................................ ................. 37 Genetic Affiliation ................................ ................................ ................................ .... 40 Dialectal Variations ................................ ................................ ................................ 41 Situating ................................ ................................ .......................... 41 ............ 42 ................................ ................................ 44 ................................ .. 50 ................................ .......... 51 ................................ ............ 57 ................................ ............................... 59 ................................ ................................ ............... 60 Sound Inventory ................................ ................................ ............................... 61 Consonants ................................ ................................ ................................ 61 Vowels ................................ ................................ ................................ ....... 65 Phonological Processes Pertinent to Consonants ................................ ............ 66 Devoicing and aspiration ................................ ................................ ............ 67 Elision ................................ ................................ ................................ ........ 67 Palatalization ................................ ................................ .............................. 67 Insertion ................................ ................................ ................................ ..... 68 Fortition and lenition ................................ ................................ ................... 68 Substitution ................................ ................................ ................................ 69


8 Metathesis ................................ ................................ ................................ .. 69 Supra Segmental Inventory ................................ ................................ .............. 69 Syllabic structure ................................ ................................ ........................ 69 Stress pattern ................................ ................................ ............................. 70 Overview of the Nominal Morphology ................................ ................................ ..... 71 ................................ ................................ ................................ ... 72 ................................ ........................... 75 Concluding Remarks ................................ ................................ ........................ 79 Summary of Chapter 2 ................................ ................................ ............................ 80 3 PLURALITY IN AFRO ASIATIC LANGUAGES ................................ ...................... 81 Classification of the Surveyed Languages ................................ .............................. 84 Observations on Plural Formation in Afro Asiatic Languages ................................ 85 Suffixation ................................ ................................ ................................ ............... 87 Internal Plural ................................ ................................ ................................ .......... 93 Broken Plural ................................ ................................ ................................ .... 93 Affixation with Internal Change ................................ ................................ ......... 96 Reduplication ................................ ................................ ................................ .... 97 Mapping Singulars onto Templates ................................ ................................ .. 98 Infixation ................................ ................................ ................................ ......... 101 Vocalic Opposition ................................ ................................ .......................... 101 Shared Patterns of Plura li Asiatic Languages ..................... 103 Summary of Chapter 3 ................................ ................................ .......................... 105 4 APPROACHES TO NON CONCATENATIVE MORPHOLOGY ........................... 106 Pre Optimality Theory Approches ................................ ................................ ......... 107 Autosegmental and Templatic Approaches ................................ .................... 107 Prosodic Theory of Non Concatenative Morphology ................................ ...... 1 12 Optimality Theory ................................ ................................ ................................ .. 116 Optimality Theory Framework ................................ ................................ ........ 117 Theoretical Assumptions of OptimalityTheory ................................ ................ 120 Markedness versus Faithfulness Constraints ................................ ................. 121 Positional Faithfulness Constraints ................................ ................................ 122 Correspondence Theory ................................ ................................ ................. 124 Alignment Constraints ................................ ................................ .................... 129 Anchoring Constraints ................................ ................................ .................... 130 Generalized Template Theory ................................ ................................ ........ 130 Summary of Chapter 4 ................................ ................................ .......................... 132 5 ANALYSIS OF THE REGULAR PLURAL SHAPES ................................ ............. 133 Analysis of Plurals with Vb Infixation ................................ ................................ .... 134 Analysis of Plurals with Suffixal Template ................................ ............................. 145 Analysis of Templatic Plurals ................................ ................................ ................ 155 Plurals Derived from Gemin ated Singulars ................................ ..................... 155


9 Templatic Plurals Losing the Vowel between C 1 and C 2 ................................ 159 Analysis of Plurals with Ablaut ................................ ................................ .............. 160 Positional Faithfulness ................................ ................................ ................... 160 Anti Faithfulness Constraints ................................ ................................ .......... 168 RealizeMorpheme ................................ ................................ .......................... 173 Summary of Chapter 5 ................................ ................................ .......................... 175 6 EXCEPTIONAL PLURAL SHAPES ................................ ................................ ...... 179 Sub P attern of Vb Infixed Plurals ................................ ................................ .......... 180 Templatically Expanded Plurals ................................ ................................ ............ 182 Truncated Plurals ................................ ................................ ................................ .. 183 Templatic Plurals ................................ ................................ ................................ .. 184 Miscellaneous Shapes ................................ ................................ .......................... 186 Suppletive or lexical plurals ................................ ................................ ...... 186 ................................ ................................ ............. 187 Doubly and triply marked plurals ................................ .............................. 187 Approaches to Excep tionality in Optimality Theory ................................ ............... 188 E liminating Underlying Representation and RealizeMorph ............................ 190 Two level Well formedness ................................ ................................ ............ 193 Multi level Well formedness or Intermediate Levels ................................ ....... 195 Realization Optimality Theoretic Account to Mulitple Plural Markers .............. 198 Specification of the Exceptionality in Lexical Entry ................................ ......... 202 Containment Approach ................................ ................................ ................... 203 Constraint Indexation ................................ ................................ ..................... 204 Selector Constraint and RealizeMorpheme: Plurals with Double Exponents .. 207 Unified Approa ................................ ....................... 212 Summary of Chapter 6 ................................ ................................ .......................... 217 7 C ONCLUSION ................................ ................................ ................................ ...... 218 Results and Contributions ................................ ................................ ..................... 219 Remaining Issues ................................ ................................ ................................ 222 APPENDIX ................................ ................................ ............................... 224 LIST OF REFERENCES ................................ ................................ ............................. 234 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 243


10 LIST OF TABLES Table page 2 1 Phonemic chart of consonants ................................ ................................ ........... 63 2 2 Phonemic chart of vowels ................................ ................................ ................... 65 3 1 Suffixation ................................ ................................ ................................ ........... 92 1 The privileged and non privile g ed positions ................................ ..................... 123 Ablaut Vs. suffixal template in bi consonantal forms ................................ ........ 162 A 1 Plurals with the plural suffix / ................................ ................ 224 A 2 Plurals with the plural suffix Vn whereby V is often /u/ ................................ ... 225 A 3 P lurals with the plural suffix i ................................ ................................ .......... 225 A 4 ................................ ................................ ..................... 225 A 5 Plurals with double and triple plural markers ................................ .................... 226 A 6 Quadri consonantal singulars ................................ ................................ ........... 227 A 7 Bi consonantal and tri consonantal singulars whose plural takes Vb. .............. 228 A 8 Vb Infixed plurals with initial vowels ................................ ................................ .. 228 A 9 Plurals taking a suffixal template ................................ ................................ ...... 229 A 10 Ablaut or vowel opposition ................................ ................................ ................ 230 A 11 Ablaut plurals of CVC shape ................................ ................................ ............. 231 A 12 Plurals derived from geminated singulars ................................ ......................... 231 A 13 Templatically expanded plurals ................................ ................................ ........ 231 A 14 Truncated plurals ................................ ................................ .............................. 232 A 15 Plurals taking the shape CVCVC ................................ ................................ ...... 232 A 16 Plurals taking the shapes CVCC and CCVC ................................ .................... 232 A 17 Varying shapes of plurals ................................ ................................ ................. 233 A 18 Lexicalized plurals ................................ ................................ ............................ 233


11 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy By Khalsa Al Aghbari May 2012 Chair: Caroline Wiltshire Major: Linguistics This is a morphophonological study of noun a Modern South Arabian languag e spoken in the mountains and coastal plains of Dh ofar, Oman. It has a twofold goal: (1) i t documents the diverse shapes of noun li and (2) provides a formal analysis of the most systematic plural shapes within Optimality Theory (Prince and Smolensky 1993 / 2004 ) It also examines the peculiar morphophonology of the exceptional plural shapes and discusses how they diverge from the systematic plural s Finally, it investigates approaches to exceptionality and lexical marking in Optimality Theor y is expressed by two distinct mechanisms: suffixation and non concatenative operations such as infixation of Vb attach ment of a suffixal VC template with fixed vocalism and a copy of the final base consonant, mapping geminated singular s onto a specific plural template and ablaut. plural shapes whose exceptionality is not attributed to their singular s and which cannot be explained phonologically The anal ysis of the regular plural patterns reveals that well mo tivated constraints can capture their regularity and address their diversity For instance, the locus of the Vb


12 infix follows from an alignment constraint a constraint placing restriction on syllable size and a l eft anchoring constraint which together fu nction in harmony with the prosody of the language as a whole. The vowel of the suffixal VC template is prespecified underlyingly and results from M AX V S UFFIX outranking M AX V R OOT The final plural shape is reduced to a single syllable and determined by the markedness constraint N O V ( Bakovi 2005: 299) Plurals derived from geminated singulars map onto a specific template ; this results from the interaction of constraints such as I DENT Q (Dell and Elmedlaoui 1992 ) and *VGG# (Muller 2001 ) Finally, I offer three cogent analyses to the ablaut plurals : (1) positional faithfulness (Beckman 1998), (2) Anti faithfulness (Alderete 1999a and 2001) and RealizeMorpheme (Kurisu 2001) Chapter 6 uncover s the exceptionality of the irregular plural shapes and justifies t heir failure to be captured by a unified Optimality Theoretic analysis I finally list some of the approaches of dealing with exceptionality in Optimality Theory and apply them to the analysis of these shapes


13 CHAPTER 1 INTRODUCTION Statement of Intent plural s Quite a large array of non concatenative morphological mechanisms accompanied with numerous pho nological alternations indicate plurality in the language. For example, the most systematic noun plurals are formed by Vb 1 infixation, attachment of a suffixal VC template with fixed vocalism and a copy of the final consonant of the base mapping singulars with gemination onto a specific plural template and ablaut. The following are r epresentative examples of the most productive noun p lurals in ( 1 ) Plurals with Vb infixation a the top parts of legs b mattresses made of leather ( 2 ) Plurals attaching a suffixal VC template whereby V is invari bly nd C is a copy of the final C of the base a dik roosters b kot towers ( 3 ) Templatic p lurals ( 3.1 ) Plurals derived from geminated singulars a. pots b hilts (of swords) ( 3.2 ) Plurals taking the shape CCVC a ut tears b skun comm unities ( 3.3 ) Plurals taking the shape CVCC a bread b ot stories ( 4 ) Plurals with a blaut a orphans (m.) b s afrir s flowers 1 Pieces of morphology (morphemes) will be italicized throughout the dissertation.


14 In spite of the prevalence of these shapes I observe that suffixation is the default plural marker in It is usually utilized to pluralize loan words from Arabic except when loan words relate to tools or gear In such a case, they take the Vb infix in their plural form t ( V ) /, Vn and i also exhibit noticeable internal phonological changes such as vocalic deletion or insertion, assimilation, and other morpho phonological alternations ( 5 ) Suffixation ( 5.1 ) The plural s uffix t ( V ) a s ti plates b mho t waters (a lot of water) ( 5.2 ) The plural s uffix un a un aunts b un tulchans ( 5.3 ) The plural s uffix i a i cliffs/ mountain edges b i boys is also characterized by plurals that take double and triple plural markers ; t wo to three plural markers are stacked one after the other to indicate plurality in these forms. ( 6 ) Doubly and triply ma rked p lurals a ab l un chameleons b b r in graves c t traditional Despite the int riguing intricacies involved in the noun plural formation of this noun plurals have not been given their due analytical or theoretical linguistic exploration. The only works which briefly touch on plurals Ratcliffe (1992, 1996 1998 a & b ) and Simeone Senelle (1997). T hese are descriptive


15 with the aim of document ing the commo n noun plural shapes in the language using the CV shape s, whereby C stands for consonants and V for vowels. Moreover, they are not solely devoted to the noun plurals is made in order to (Simeone Semitic and Afro A siatic languages (Ratcliffe 1992 1996 and 1998 a & b ). However, these works do offer a good background and lay the fundamentals to understanding the most common patterns of noun plurals in the language with numerous supportive This study is a linguistic attempt to document the diverse shapes of internal and plural shapes observed in the formation of this morphological phenomenon in the current speech of n to also provide a formal unified analysis of the most common and systematic plurals in the language within the framework of Optimality Theory (Prince and Smolensky 1993 / 2004 ; McCarthy and Prince 1993a & b). There is always aspiration to offer a systemati Noun P (1986) Paul Newman (1990) A reliable reference on the recurrent patterns of plural in this la n guage is lacking. is only limited to describing and categorizing the existing plural shapes based on their CV patterns. There is no t any linguistic work that mentions the phonological and morphological


16 mechanisms involved in the process of plural formation in in other Modern South Arabian languages. Moreover, I did not find any study that attempts (even if unsuccessfully) a theoretical framewo rk to offer a cogent analysis of noun plurality in this langua ge. Noun plurals in are very diverse; they exploit many systematic non concatenative morphological processes and exhibit mo rphophonological alternations. A singular shape m a y map onto numerous plural patterns For example, bi consonantal singulars may systematically take a suffixal VC template, exhibit a vocalic change or take two plural markers They may also take a plural suffix or map onto a distinct plural template Thus, despite a level of systemicity, t he relation between the resultan t plural shape and the singular from which this plural is d erived is not always predictable. Using the International Phonetic Alphabet (IPA) to transcribe is a breakthrough contribution of this dissertation The majority of work done on continues to follow the nonstandard notations first employed by the late Thomas Johnstone in 1981 for and other Modern South Arabian language s. This constitutes an obstacle to knowing what exactly the phonemic sounds of the language are Moreover, with the infor mal notations, it was hard to know the phonological processes pertinent to the language and to clearly state how they contribute to th e understanding of Jeb Past works on conducted by the native speakers of the language always insist on the untrue affiliation and relatedness of to Arabic. These studies base this spurious belief on the substantial amount of borrowings has from the dominant and surrounding Arabic and Arabic dialects In this dissertation,


17 noun plurals refute convincingly such beliefs and reveal different plural mechanisms employed by than those found in Arabic. For instance, s not employ the dominant broken plural shape with an extra length in the second syllable and with the canonical iamb ( CV.CV: ), which is widely attested in Arabic. exploits a plural pattern reported to figure in the morphology of Ethiopia n languages. Ratcliffe (1998:196) states shows a pattern of plural formation for underived The language is named Modern South Arabian based on its geographical location in the Arabian P eninsula and not because it relates closely to Arabic. Overview of Dissertation This dissertation is organized as follows. Chapter 1 describes the most common and exceptional noun plurals in It also reveals the role of gender in sing ular plural mappings and uncovers fac ts about how gender functions in the combination of nouns and adjectives. I also discuss the scope and limitations of this dissertation in Chapter 1 The remainder of this dissertation is structured as follows. Chapter 2 presents It provides an overview of through the exploration of its genetic affiliation, dialectal variations and phonemic inventory. It also delineates details on the syllabic structure and stress pattern of information on phonetic tendencies and nominal and verbal morphologies. Since the fo h is thoroughly reviewed in a separate section of Chapter 2


18 to cast light on findings and reveal how this dissertation builds on and supplements previou s works on this pivotal area of linguistics. Chapter 3 reviews the most pervasive mechanisms of plural formation employed in a wide array of Afro Asiatic languages. It surveys the most common mechanisms of plural formation in Arabic, Hebrew, Berber, two E thiopian languages (Tigrinya and Amharic), a host of Chadic languages (most significantly Hausa), Cushitic languages, Modern South Arabian languages and ) and finally Egyptian and Coptic. Because there is not much appreciable distinction in the type or process of plural formation among the languages which belong to the same language family, I limited the discussion of noun plurality in Afro Asiatic languages to representative languages in each family. The last section of C hapter 3 summarizes some of the shared tendencies between and these languages in regards to noun plural formation. C hapter 4 lists some of the significant appro aches to non concatenative morphology. It shows the tenets and assumptions made in previous Templatic and Autosegmental approaches and later the Prosodic Morphology approach. It basically outlines the eff orts of Templatic and Prosodic a pproaches in tacklin g a host of non concatenative processes observed in languages that have non concatenative morphology. Moreover, it presents the basic theoretical assumptions the current study hinges on for its analysis and arguments. It gives a succinct background of the main principles of Op timality Theory, the constraint based theory of phonology, and introduces the major constraints employed for the analysis of the various mechanisms of Chapter 4 briefly discusses


19 General ized Template Theory which the dissertation adheres to in order to analyze Chapter 5 includes a formal analysis of the systematic shapes and regular mechanism s of plural formation in the language using the framework of Optimality Theory (Prince and Smolensky 1993 / 2004 ; McCarthy and Prince 1993a & b). The analysis addresses the Vb infixed plurals, the plurals attaching a suffixal VC template with fixed segmentism and a cop y of the final consonant of the base (partial reduplication), the templatic plurals derived from geminated singulars and plurals marked by ablaut. Chapter 6 discusses the exceptional plural shapes and addresses the phonological and morphological peculiarit y associated with their formation. It also lists the obstacles that hinder accounting for them cogently. Finally, I elaborate on how Optimality T heory addresses such exceptionality and the various approaches it takes to address lexical marking. These appro aches are applied to the analysis of the exceptional noun plurals The final section of Chapter 6 seeks to outline an approach that unifies these exceptional patterns into a single O ptimality T heory analysis. Chapter 7 closes the dissertation by presenting the major contributions of this dissertation and summ arizing the results of the analyses for the various systematic Chapter 7 discusses a number of remaining issues that can be considered for future research. Description of the Plural Patterns Noun plurals in Je various non concatenative operations which occur concomitantly with enormous phonological changes, resulting in immense


20 diversity in the plural shapes For example, plurality can be systematically marked by Vb infixation, attachment of a suffixal VC template with a fixed vowel and a copy of the final stem consonant, mapping singular forms with gemination onto a specific plural template and ablaut. Parallel with these morphological processes, plural forms exhibit phonological alternations like vowel deletion, vowel insertion, metathesis, assimilation and re syllabification. This diversity in plural formation can sometimes be systematic; the resultant plural shapes straightforwardly relate to the particular shapes of their singular forms. For instance, only bi consonantal and mono consonantal singular shapes reduplicate their final stem consonant to indicate plurality; tri consonantal singular shapes are observed to pluralize by processes other than copying a consonant from the base Moreov er, the majority of singular forms that take the infix Vb are quadri consonantal with the canonical shape CVCCVC. Bi consonantal singulars whose second radical is geminated expand their segments and map onto a specific plural template different from that resulting from mapping quadri consonantal singular forms. However, many plural forms can hardly be related to their singulars. To illustrate, since bi consonantal singulars, for example, may take various shapes of plural ( attaching a VC template suffi xati on, vocalic opposition and bearing double plural markers ), it is extremely unpredictable to assign a definite plural shape to a particular singular form. Furthermore, a few singular forms take simultaneously two to three plural markers to mark plurality. T herefore, establishing a general mechanism of plural formation for poses a challenge because there are many divergent plural patterns that cannot be solely attributed to the shapes of their singular forms. However,


21 overall, the diverse ural shapes are phonologically conditioned. They exhibit common morphological and phonological characteristics or tendencies which are indicative of the prosody of the language as a whole. orrowings which have taken place from various dialects of Arabic spoken in Oman and Yemen as well as affected a huge number of plural forms and mechanisms. For instance, simi lar to Arabic, most of the bi consonantal singulars borrowed from Arabic tend to be pluralized by templatic expansion (cop y ing the second or third consonant in the base) although not as previously stated by the canonical broken plurals I will thoroughl y describe the div Discussion of the plural shapes first outlines the typical systematic and most common plural patterns and then moves to describe the peculiarity of the exceptional shapes and shapes that take double and triple plural markers Before embarking o n the description of these patterns, 1 ) the inherent gender of the singular noun, and ( 2 ) the feminine suffix markers ah and ( V ) t Thus, the suffixes (V)t and ah a ttached to some singular forms in the data described below indicate the feminine gender and do not contribute to the consonantal roots of these forms Suffixation Like other Semitic and Afro Asiatic plural suffix es that attach to si ngular forms to mark plurality. However, the resultant plural shape is not purely the plural suffix imposes some sort of phonological change into the final shape of the plural. : t(V) whereby


22 Vn whereby V is mostly /u/, and finally i The last suffix was a marker of duality which no longer seems to be an active and systematic process in the language. Only very few archaic forms continue to take this suffix and bear the dual meaning ( e .g. becomes dual The se plural suffix es attach to various singular shapes ranging from bi consonantal to quadri consonantal singulars. However, the default plural suffix in the language is t(V) which serves as the plural marker for loan and nonce forms. It is also a feminine plural marker, and resembles in shape the Arabic feminine plural suffix a:t Many speakers use this suffix when given a singular form of any shape from Arabic, English or another l anguage. After th e s e suffix es attach to singular forms, a number of phonological alternations affecting the quality of the vowels, syllabic structure or consonantal qualit y of the singular forms occur. For example, vowel d eletion and insertion occur freque ntly. A wide range of vocalic change is also attested in many forms. Moreover, place assimilation of the final nasal consonants in the singular forms to /t/ can easily be detected in the plurals with the suffix t(V) These wide ranging phonological altern ations that accompany suffixation indicate that suffixation does not alone serv e as a sole marker of plurality Below, I show some representative examples of singulars which take t(V) to mark plurality More forms are found in Appendix [A]. ( 1 ) Suffixal p ( 1 .1) T he suffix t(V) a. fef elbows b. bat bat ti beaches ti doors mountains ti grounds/ floors lho ti cows people from dhofari et plait s, tresses of hair


23 The plural forms above exhibit some vocalic changes when the plural suffix t(V) attaches to them. To illustrate, forms (d) and (f) have an extra vowel after the last consonant when the plural suffix attaches. Moreover, a change in the vocalic quality is observed in the plural forms when they are attached to this suffix For example, the plural form (h) has a back rounded vowel unlike its singular form which contains a high front unrounded vowel. Form (a) appears to lose or degeminate an /f/ when the plural suffix gets attached. Another plural suffix common in is un or in Although this suffix is not as common or productive in Jebb as the default one, quite a few singular forms are pluralized by attaching this suffix. Moreov er, this suffix is also similar to the sound plural suffixes u:n a:n and i:n attached to masculine plural nouns in Arabic. Howe ver, does not have length in the plural suffix. Ratcliffe (1998:165), who explores plurality in a number of Afro A siat South Arabian languages] have undergone changes resulting in neutralization of the contrast between long and short, high and low vowels in many environments Moreover, it is worth pointing that the suffix a: n other Arabic dialects; it corresponds to the external suffix of the masculine plural of adjectives and 2009:310). The singulars which attach to Vn suffix are also diverse in th eir shapes. They can have two, three or four consonants in their base Noticeable vocalic change is observed as one moves from the vowels contained in the singulars to those the plural nouns have. An important observation to make is that there is often som e sort of a vocalic contrast that accompanies suffixation to realize plurality. In other words, if the


24 singular form has a back rounded vowel /u/ or any of its variants, the plural noun takes a front unrounded vowel /i/ or any of its variants too, as in ( d ) and ( e ) below. ( 1 .2) T he suffix Vn a un aunts b un tulchans c un children d unub in tails e un doctors The last plural suffix to be described here was formerly a dual marker i. Duality is no and many forms which have the suffix i currently denote plurality. Again, this suffix attaches to bi consonantal ( forms (b), ( c ) and ( e ) below) and tri consonantal singular shapes ( forms ( a ) ( c ) ). The semantics of the forms a ttached to this suffix are diverse and relate to living and non living entities. Therefore, semantics alone cannot se r ve as a clue to signal a correlation between this plural marker and the forms they attach to it. Singular f orms which have the feminine su ffix Vt such as forms (a) to (d) lose the feminine suffix prior to attaching the plural marker i The following examples of plurals take the former dual marker to mark plurality : ( 1 .3) the suffix i a s it s ofor i cooking pans b i boys c at black flies d. i cliffs/ mountain edges e. angels Vb I nfixation of Vb The larg est data corpus collected takes the Vb infix and belongs to the masculine class. This pattern of plurality tends to occur in nouns that relate to tools, gear and equipment in general. Loan words pertinent to tools such as [ mas t r ah / mas abt ]


25 sing./pl. are observed to take this pattern too. Therefore, Vb infixation is productive within this semantic sphere. This plural shape exhibits infixation of Vb exactly after the third segment of the singular form. The infix constitutes the second syllab le from the left edge of the plural form. The majority of quadri consonantal singular forms take this pluralization mechanism ( forms ( a d ) below). However, it is important to note that not every quadri consonantal form will take the Vb infixation since a l arge number of quadri consonantal singular forms instead take the default plural suffix tV to indicate plurality. The shape of the singular form is CVCCVC which becomes CVC Vb CVC after they pluralize. on the p lace features of the preceding consonant. When the consonant is a pharyngeal, pharyngealized or glottalized, the V of the infix is mostly /a/. However, if the preceding C is a coronal velar as in (b) or bilabial, the vowel of the infix is / ( 2 ) Plural s with Vb infixation ( 2.1) Regular Vb i nfixed p lurals a the top parts of legs b mattresses made of leather c s s boxes d um am pots used to keep ghee A number of vowel initial singulars take Vb infixation to mark plurality (examples ( a d) below). All the examples collected begin originally with a nasal /m/ which is deleted word initially in (Johnstone 198 1; Nakano 198 6 ; Hofstede 1998) After /m/ deletes in the singular form the following vowel na s alizes and/ or lengthens The tri consonantal singular shape (underlyingly quadri consonantal) becomes [ : CCVC] and it is in fact the derived version of /mVCCVC/ There are two plural shapes for those


26 singular forms: one plural shape w ith an initial schwa and the other retrieves the deleted /m/. So, the resultant shapes are C Vb CVC and mVC Vb CVC. ( 2 .2) Singulars with an i nitial d eleted /m/ a keys b offices c t restaurants d rooms for guests The last group of singulars that take s the Vb infix is quite esoterically shaped. Some of the singulars have consonant cluster CC word initially ( forms ( a ) and ( b ) below ). I observe that a cluster of two consonants are tolerated word Some of the plurals which belong to this pattern bear the shape CCVC. Others are bi consonantal with the shape CVC. The resultant shapes of the plural forms are also diver se. Plural f orms ( c ) and ( d ) below lose the vowel in the infix and maintain only the b ; they take the shape (V)C b VC. ( 2.3) Other Vb i nfixed p lurals a muzzles b t t Zizyphus spina Christi c t drums d news Attachment of a S uffixal VC Template As a shared anomaly common to many Afro A siatic languages, nouns with one or two stem consonants tend to acquire a third consonant in the plural form by reduplicating a consonant from the base For instance, Belova ( 2009 :310 ) reports a number of Arabic dialects and Ethiopian languages that mark plurality by reduplicating the third or final radical including the Arabic dialects of Upper E gypt ( e.g. [bnitta] for Sudan geria (e.g. [duggunne] for [digin] ) the region of Lake Chad (no example therein is supplied), Amharic (e.g.


27 alaga] ) Ratcliffe (1996) argues that this tendency can be explained in terms of templatic expansion whereby an extra consonant is realized in the plural form in order to meet some templatic constraint required by the language. He further argues that the e xtra normally used as an affix such as /t/ which indicates the feminine gender in Semitic o r a copy of the stem consonant. In reduplicating the final consonant in the base is observed to be a systematic plural formation process. Bi consonantal singular forms of mostly CVC shape exhibit partial suffixal reduplication (V)CC x x Most of the collected plural forms taking th is pattern are, by and large, borrowed fr om Omani Arabic. The single vowel in the singular form varies greatly while most of the plural forms consistently have / / between the last stem consonant and the reduplicated final consonant in the plural form Onl y three forms in the collected data or /e/ in the suffixal reduplicant (forms ( g i ) below ) ( 3 ) Partial suffixal reduplication a fish m b nuf selves m c shelves, racks, bulks m d mus razors m e pal ms of the hand; claws m f pilgrims m g dry leaves f h lavatories f i .hab ot/ hib ot hbeb/ heb songs f The vowel in the plural shape (forms (c f) above) does not occur in all the plurals with the suffixa l template In some forms, the initial inserted vowel harmonizes


28 (form (f) above ) Singular forms taking this pluralization pattern belong to different classes; whether the forms are masculine or feminine, it does not mat ter. In the data collected, there is a single uni c onsonantal form which pluralizes by taking the suffixal template with partial reduplication and a pre specified vowel Th is form bears the shape CV whose single C reduplicates resulting in VC x x The exam ple Ablaut/ Vocalic Opposition One of the most prevalent plural shape s opposition. This tendency toward reversal of vowel quality can also be observed in consonant masculine [nouns] with /e/ or /i/ in the last syllable ha ve the vowel alternation I classify the plurals taking ablaut into two major shapes. The first shape affects singular forms which have thr ee or four root consonants ( forms (a d) below) and the second shape concerns the resultant bi consonantal plural shape CVC ( e g ). In the first shape, the last syllable of the plural form has a vowel different from that in the last syllable of the singular form. In the majority of forms, back vowels appear in the plural form as opposed to front vowels which the singular forms pervasively have. ( 4 ) Ablaut or vowel opposition a orphans (m.) b s af rir s flowers c isolated ho mes d t t dresses e nid nud water skins f ropes g men


29 The second shape of ablaut plurals (examples e g above) is derived from diverse singular shapes which can mostly be bi cosonantal or tri consonantal in shape How ever, the plural is always CVC with an obvious change in the vocalic quality. Templatic Plurals Plurals d erived f rom geminated s ingulars The fourth systematic plural shape concerns the plurals derived from geminated singular forms which take on a defi nite templatic shape In the plural forms, the x C x derive this plural shape The final shape of the plural form is CVC x VC x ( 5 ) Plurals derived fr om geminated singulars a. pots b. hilts (of swords) c. benches outside a house Plurals with t runcation and t emplatic e xpansion which mark plurality in a wide range of words : truncation and templatic expansion. These two processes affect diverse singular shapes ( can be bi tri or quadri consonantal). Templatic expan sion involves an extra syllable or consonant in the plural form as opposed to the singular form which has fewer syllables or consonants. ( 6 ) Templatically e xpanded p lurals a windows b e:kwar chiefs c sweethearts d young bulls On the other hand, the truncated plural form exhibits fewer consonants or fewer syllabic structures than those contained in the singular form Since this language


30 involves a lot of deletion, it is possible to think of the extra syllable or consonant in the plural forms as reappearance or retrieval of the deleted segment or portion in the singular forms. ( 7 ) Truncated plurals a e:s e:s fingers b camel calves c cartridges d t fans e turnings on a path an internal change. However, the change is very eclect ic in nature to the extent that it is very hard to establish a generalization that governs a particular internal change. However, the change can be described as templatic in nature. Plurals belonging to this category are mapped onto three basic templates: CVCVC, CVCC and CCVC. Interestingly enough, the singular forms from which these plurals a re derived are invariantly tr i consonantal. The class to which these forms belong is also diverse and there is no definite class grouping these forms together Obser ve these sets of plural patterns: (8) Templatic p lurals (8.1) Plurals taking the s hape CVCVC a female possessors b cheeks c sahab waves d t rifle bolts (8.2) Plurals taking the shape C C V C a ut dma tears b skun communities (8.3) Plura ls taking the shape CVC C a. bread b. ot stories


31 has some plural forms which cannot be classified into any of the above explored patterns. In the collected data, there are onl y a few plurals which belong to this type. This reveals the rarity of this type. Some of these plurals have metathesis; others have consonantal shift. However, the shift of consonant is not clear or easily identifiable In other words, much morphophonology characterizes these forms Observe the following examples: (9) Miscellaneous s hapes a. at ghosts b. heads c. beads d. months e. old men has a distinct group of noun plurals which take two to three plural markers. These plurals may have two plural suffixes consecu tively following each other (examples ( a c ) below) or can take the Vb infix along with the default plural suffix tV ( forms ( d ) and ( e ) ). The plural f orm ( f ) is the only collected form that bears three distinct plural markers. The set of plural forms mark ed by two to three plural markers is limited. Jebb speakers fail to supply more doubly marked plurals than the ones listed in Appendix [A]. Moreover, t here is nothing special about the singular shapes which can justify why they take double or triple markers: singulars which take double plurals range in their shapes from bi consonantal (forms ( e ) and ( f ) below) to quadri consonantal (form ( a ) below) I observe that the plurals taking double plural markers are native to borrowed from Arabic I also observe that plurals t aking more than one plural suffix do not designate special semantics or add emphasis to these forms. More specifically, they


32 (10) Plurals bearing two to three plural markers a t n b ah un ti traditional wooden boxes c b monitor lizards d b nipples e zol it zol un carpets f et kof caps Like many Afro A siatic languages, has a number of lexicalized plural forms whose singulars and p lurals are vastly unrelated. These plurals, though unsystematic, seem to be semantically interrelated. Most of the collected words relate to humans and living entities. Below I list a sample of suppletive or lexicalized forms. Interested readers are refer red to appendix [A] for more forms. ( 11 ) Suppletive plural f orms a j women b bera boys c bri j ni sons d ti daughters non concatenative morphological mechanisms to indicate plurality. These me chanisms are accompanied by many phonological changes such as vocal ic change, vocalic deletion, insertion and consonantal assimilation The singular forms from which these diverse plural shapes are derived are also very diverse and cannot be solely a clu e about how a singular will be pluralized. It is important to note that some singulars may take multiple plurals and some plurals may be doubly pluralized (i.e. take double plural markers). Gender in Singular Plural Mappings In exploring the diverse plur al shapes I investigate d whether the gender of a particular noun determines what plural pattern it takes. In other words, is gender a


33 direct determina n t fo r the resultant plural pattern ? I also studied the gender of a number of plural forms whe n they combine with descriptive words (adjectives) to check if the re is a difference between the gender of nouns and that of the ad jectives describing them. Do nouns change their gender when they are pluralized? In Arabic and Omani Arabic specifically th ere are two modes of pluralization: sound and broken. Sound plurals take the plural suffix es u:n / i:n for masculine forms and a:t for feminine form s. Most of the feminine singular forms maintain their gender when they pluralize since they attach the femi nine plural suffix a: t Examples from Omani Arabic ah a:t [mall ah ] [mall a:t ah a:t and [ dabba:s ah ] [dabba:s a:t However, since broken plural is the mos t common mode of plural formation in Arabic, many of the feminine singular forms which take broken plurals are prone to alter their gender due to the fact that broken plurals are gender null ah] feminine m uql ah] masculine I also observe that gender may change when singulars are pluralized. For example, a masculine noun such as feminine [ko:b a:t 2 ]. the majority of singular forms pluralize by the default plural marker tV which is considered by many to be a feminine plural marker too Therefore, masculine nouns which take this suffix in their plural formation are, in turn, feminine. This is proven by the fact that quite a large number of masculine nouns which take the plural suffix tV 2 a:t is a feminine plural suffix.


34 consultants 3 I n (12) and (13), I list a number of masculine nouns which become feminine in their plural formation and vice versa : (12 ) Feminine to m asculine : a et / l a beards b bottles, water jars c ot spoons (13) Masculine to f eminine: a. roads b. herum plants c. e:d hands d. eyes Two o bservations about the above forms are in order here. The majority of plural forms taking the t(V) suffixation mode of pluralization are feminine regardless of the gender of the singulars from which these plurals are derived. Masculine plural nouns take diverse plural patterns. However, a large and substantial number of masculine plurals take the Vb infixation and un suffixation (only two exceptional forms are recorded in my data) To conclude, gender is not an influencing feature for the plu ral formation pattern. There is diversity in both the gender and the pattern of pluralization. A singular noun may change its gender in its plural form. However Afro Asiatic and Semitic languages, have a number of plural markers that are gender specific. I will now tackle gender in the combinations of noun plurals and adjectives. In Arabic, adjectives follow nouns and have to agree in gender a nd number with them. In the case of broken plurals, if the noun plural takes a masculine adjective attached with u:n or i:n then its inher i t gender is a masculine. However, if it takes the suffix aat 3 I would vent ure to claim that the default plural suffix Vt is parallel to a:t the feminine plural suffix in Arabic and they may be closely related due to historical contact or borrowing.


35 then it is a feminine. It is worth mentioning that some adjectives also take the broken plural formation. However, the gender of adjectives must accord with that of the feminine plurals ah kabi:r ah [ko:b ta:ris] masc Based on surveying the gender of plural nouns when they combine with various adjectives agre e with the nouns they modify in number and gender noun and the adjective (which always follow s However, there are a number of neutral adjective forms whose shape s tays unaltered whet her the noun they describe is masculine or feminine (e.g. [re cleanliness fatness and strength. These adjectives, thus, have a common gender. On the other hand, I observe that when adjectives are ma rked, they often attach the plural suffix tV and describe feminine plural nouns Observe the following examples: (14) Plural Adjectives Masc. Fem. Gloss a. dark b. tgar business oriented c. badut untruthful min brave In conclusion, based on the data collected and interviews with native are two groups of adjectives in the language The first group takes the same shape for both masculine plural and feminine plural nouns. The other group of adjectives attaches the noun plural suffix tV t o indicate the feminine gender, while the masculine adjective is usually unmarked Adjectives are not observed to pluralize by other plural


36 mechanisms like the Vb infix, attachment of a VC template or ablaut modu lo to the nouns they describe.


37 CHAPTER 2 one of the Modern South Arabian langua ges 1 is widely spoken in the mountains and coastal plains of Dhofar ( in Arabic), a governate in the southern region of the Sultanate of Oman. Geographically, it stretches towns and I slands spoken in sporadic areas situated at the boarder shared between Dhofar and Yemen (Lonnet 1985 :50; Hofstede 1998:13 ). Dhofar receives annually the monsoon rains which transform the entire region into absolute greeneries serving as a tourist attraction and an affordable destination for vacation ers from the Gulf and Euro pe who keep cows and camels; pasturing the cattle depends mainly on the rain the mountains of Dhofar receive from the end of June to mid September during the monsoon season in Arabic). They are also engag ed in cultivating frankincense ) trees whose pedigree gum or incense is known for its fine quality and is exported to the neighboring regions (Johnstone 1981:xi ; Al Tabuki 1982:52 ; Hofstede 1998; Morris 2007 ). Various names designat equally known as Shehri ( voiceless lateral fricative ) in reference to the first inhabited the mountains of Dhofar. have long be lieved that Shehri belongs to them alone and that other tribes have 1 Hobyo t and Socotri. Mehri is originally come from Jidat Al in Hobyot is widely found at and around the border shared between Oman and Yemen. In Yemen, Socotri speakers reside.


38 subsequently acquired it from them. Therefore, they always call the language after their denying that it exclu sively belongs to the Shehri group and arguing that Shehri is originally derived Oman and Yemen. In the past, the language was dubbed as which insinuate reference to old social differences and which sound pejorative to native ; Hofstede 1998; Morris 2007 ). Al Mashani ( 2003 ) argues tha t the best name for this language 2 al ( the contemporary tongue of Himyyar) which indicates that is a shared linguistic wealth and is deeply linked to the glorious civilization built by the ancient people of South Arabia about 5.000 speakers. Hofstede (1998: 13) states that the number might be 50.000. Native Dhofaris who are estimated to be 2 49 ,000 people 3 by the most recent national census c onducted in 20 10 believe that approximately 70% of the population in Dhofar In speakers t o Arabic through modern schools and influential Arabic dialects of local tourists and visitors on one hand and foreign languages on the other hand 2 This tribal group gradually extended its power across the whole of Yemen and eventually exercised authority over the southwestern h alf of the Arabian 256) 3 34.5 % of this figure is expatriates.


39 teach it as a first language to their children This pride coupled with the isolation it enjoys (Al Mashani 1999; Al Sh e hri 2007) enable Jebb until today In fact, it is extremely rare to meet a person from the s outh of the Sultanate of Oman However, since is not written and there is an o ngoing wave of modernization exercised by and formal education for all speakers. Cha pter 2 and dialectal variations. It also situates the data used in this dissertation and provides an overview of the previous scholarship Modern South Arabian l anguages. Rarely is there a reference mentioning the other Modern South Arabian languages This can be attributed to the fact that South Arabian people are often fluent in two or more Modern South Ar abian langu ages (Lonnet 2009: 298) and can readi ly supply field workers and former scholars with forms from the other languages too. Furthermore, the research on these languages is still in its infancy and it is a breakthrough to attempt a comparative linguistic stu dy of all the related Modern South Arabian languages in a single study Chapter 2 also provides a grammatical sketch of which is basically a description of its phonemic consonantal and vocalic inventories, syllable structure and stress. It lists the most common phonological processes pertinent to The last two sections of Chapter 2 discuss nominal and verbal mo rphologies and provide an overview of the major scholarship on plurality in


40 Genetic Affiliation Semitic languages b elong to the Afro A siatic family. More r elevantly, t he South Semitic is divided into two branches ; Modern South Arabian languages occupy an eastern branch ( Rodgers 1991; Hetzron 1997; Faber 1997 ; Ratcliffe 19 92 among a host of Semitists). Faber gives the following classification for South Semiti c : (1) South Semitic Eastern Socotri (emphasis mine) Western Old / Epigraphic South Arabian Ethiopian Semitic (Faber 1997:5 13) As seen in the classification above, Modern South Arabian languages belong to the Southeast Semitic branch. These languages Hobyot and Socotri which are spoken in Oman and Yemen. Due to extreme r esemblance to and shared linguistic features with Ethiopian languages, it is not uncommon for distinguished Semitists and anthropologists to group the Modern South Arabian languages together with the Semitic languages of Ethiopia in the South Semitic branc h. For instance, in Goldenberg (1977), he maintains that the reason for grouping Arabic, Ethiopian and South Arabian languages into one group is due to the fact that they all share broken plurals. Moreover, b ecause it has been established that Arabic (incl uding Sabean 4 Mehri and Socotri) and Ethiopian languages stand in a closely comparative rela tionship (Goldenberg 1977:473), broken 4 Sabean is one of the ancient Old South Arabian languages (also known as Epigraphic or S ayhadic) found in the period between the beginning of the first millennium B.C.E. and the middle of the sixth century C.E (Belova 2009:301).


41 plural formation marks an imp ortant stage in the development of these languages. T hese languages also share vowel lengthenin g in their verbal morphology 5 In his concluding remarks, Goldenberg maintains that South Arabi an and Ethiopian languages constitute the same branch of grouping e xcluding Arabic which is only distantly related to Modern South Arabian languages. This concl usion contradicts claim s made by the native speakers of the language who constantly argue that and Ar abic are highly interrelated. Dialectal Variations geography in Dh ofar: Eastern, Central a nd Western ( Johnstone 1981:xii ; Hofstede 1998:14 represents the original or mother as other dialects have many affinities with and only minor differences between it and other dialects exist Since Central is considered to be the most representative variety of the singular and plural tokens collected in this study pertain to Central i primarily spoken in Salalah ( in Arabic), the main city in the Southern region of Oman. More specifically, the data represent the current spoken by four native speakers whose ages range from 24 to 50. The majority of speakers ar e bilingual with fluency in both and Arabic. Two of the four informants in this study understand both Arabic and while the other two also speak English as a third language. 5 vowels are not contrastive in length.


42 Singular and plural forms were recorded during two principal field work trips to Oman during the summers of 2009 and 2010. The researcher arranged two to three meetings per week with consultants to elicit new forms and verify old ones through corrective feedback and interviews. Singulars and plurals are transcribe d using the International Phonetic Alphabet notation (IPA). To the best of my knowledge, this dissertation is the first in attempting to purely use the IPA for since previous works partially use the IPA in combination with other non standard symbol s. Lonnet (2009) provides the equivalent IPA symbols to some of the informal notation found in the old scholarship of and other Modern South Arabian languages. other Modern S outh Arabian Languages There are two main types of One is based on systematic accounts and studies done by interested European linguists, anthropologists and sociologists. The most significant research is carried out by the late professor Thoma s Johnstone South Arabian language s has long served as the main reference to these languages. In spite of this, the research done is descriptive or anecdotal in nature intending to describe the peculiarities of this language and expose its deviations from the surrounding dominant Arabic or Arabic dialects. Moreover, the data collected in these studies though interesting, are transcribed using confusing symbols and unclear notations. R ecent work on (c.f. Hofs tede 1998) has continued to use this transcription despite the availability of the latest IPA theory of transcription at the time when their research is conducted. The second type of source of information is the recent and growing interest seen in research done by the native speakers of this


43 language. The latter is also descriptive and focus es on proving interrelatedness I also review the majority of descriptive works done on and other Modern South Arabian languages These works delineate the major linguistic and non linguistic features that arouse the interests of sch olars coming from as diverse spheres of study as anthropology, sociology and linguistics and other Modern South Arabian languages differ vastly. Some of these scholars are affiliated with organizations aiming at the revitalization of Modern South Arabian languages while others are engaged in long term research projects that aim at bringing forth the linguistic wealth of these languages. The discussion will first highlight the more miscellaneous and general works o n These works serve as a background to the language and briefly discuss diverse and holistic linguistic aspects of in the same study. Then, I move to discuss the more specialized linguistic works which are considered to be reference works t o the phonetics and phonology, nominal morphology, verbal morphology and syntax of The insufficiently explored despite the wealth of research produced until this point in time The majority of work done is descriptive in nature and cursory in essence. Hardly can a researcher come across any analytic or theoretically systematic study on the language. Moreover, there was uncritical reliance on scripts collected by the early Vienn a Expedition (1898) and certainly can lead to unreliable conclusions (Matthews 1969)


44 Introductory Johnstone ( 1981 ), Simeone Senelle ( 1997 ), Al Mashani ( 1999 ) and Clover ( 1988 ) report that Fulgence Fresnel, a French consul in Jeddah in 1836 makes the first reference to which he calls According to Al Mashani (1999 : 42), Fresnel has written many letters which provide valuable information about this language in 1839. Simeone Senelle (1997) mentions that in 1898 an Austrian expedition known as the Sdarabische Expedition came from Vienna and began studying t he people and languages in the s outhern part of the Arabian Peninsula and Socotra Island. Three pioneer scholars: Dav. Heinr Mller, Alfred Jahn and Wilhelm Hein have significantly eme rged from this expedition and ca me back to Vienna with collected sample texts of explorers in this expedition identify Shkhawri 1981; Simeone Senelle 1997). Thomas (1937) reports the egregious geographical mistake Mller m ade when situating t which is in fact quite far away from D hofar. In 1909, Bittner studied these texts and wrote three articles describing these languages. It is reported in Al Mashani (1999) that Bittner published li until 1937 (Al Mashani 1999: 42). ertram Thomas (1937), a British scholar who speaks Arabic. His study introduces the four strange tongues of the South Arabia revealing social, cultural and linguistic peculiarities of each tongue in addition to intensive comparative lists of numerous wor ds including personal names and


45 names of animals. He explores aspects of the morphology and syntax of these languages such as pronominals, articles, nouns, sentences and more in each l anguage. More relevant, he lists a collection of singular for ms with the ir plurals but offers no discussion about how plural is formed. He classifies which collectively form one group. According to Thomas, a person who se ear is accustomed to the dialects of Modern Arabic is struck by two pervasive fe atures of Modern South Arabian l T h omas 1937: 236) especially those heard in solely devoted to presenting the geography and linguistic features of Mehri and Hobyot, she lists many phonological and morphological end of her article. Such information serves greatly in understanding the phonological ten dencies of this language. Simeone Senelle (1997) offers background information about Modern South Arabian languages. She provides a comprehensive sketch which delineates the phonology and morphology of these languages. For instance, after discussing some sociological and geographic al aspects of these languages, she embarks on giving a detailed description of the phonemic inventory, phonological processes, nominal and verbal morphology and syntax of these languages. She also explores specific linguistic asp ects of these langu ages like numeral and deictic. Her study is broad and the interspersed with information about other Modern South Arabian languages However, it serves as a good linguistic background for these language s.


46 Al 6 assert s that Mehri is the original language of all other Modern South Arabi an l anguages. He also argues that it is the mother of the Semitic languages known today because it is th e oldest among them. His study allocates chapters for three major Modern South Arabian languages ( Socotri and Mehri ) presenting samples of their poetry, texts and the of Dhofar. Then, he intensively talks about Mehri which he names of Mahrah. In both these chapters, he p Arabic verbs to prove relatedness of these languages to Arabic. Finally, he discusses Socotri delineating crucial facts about this language including its location, its ancient name, inhabitan ts and an overview of its poetry. From my perspective, the samples of interrelated. First, the sample supplied includes many borrowings from Arabic. Secondly, one cannot establi handful of ten words or so. published in Semitic based o n the shape of the definite article C(a) where C is one of h She argues that the classification of Semitic can be made based on the ir observed morphological and phonological innovations. These innovations could either result from contact with oth er Semitic languages, especially those of the same grouping, or evidence for similar genetic affiliation. After adopting the fully detailed classification of 6 Mahrah is both the tribal group and place of Mehri.


47 Semitic by Hetzron (1972 et seq.), she presents a number of prominent cues that support such a cla ssification. Most relevantly, Faber (p p .13) mentions that forty nine Sabean nouns form their plural with the prefix the in ternal mode of pluralization). Al Sh e hri (2007) descriptively discusses the phonology, morphology and syntax of briefly examined conversational sp eech transcribed into quasi IPA and translated into Arabic. He also supplements his study with an atlas illustrating comparisons of lexical words found in Johnstone ( 1 981). Moreover, he systematic studies done by native and non native More releva plurals. For instance, he mentions the suffixes at and u:n for the regularly formed plurals. He also presents examples of broken plurals summarizing the canonical C V shapes with which plurals surface. Similar to previous work, his study is descriptive in general and does not include a theoretical linguistic framework to account for the phonology, morphology and syntax of this language other than mere description rese mbling what has been written about the language thus far. The two major contributions that can be acknowledged for Al Sh e hri are the atlas that pinpoints


48 variations in lexical words, phonology and morphology among the three varieties of In a joint meeting of the Anglo Omani and British Yemeni Societies, Morris (2007) presented an exuberant talk that highlights the current situation, earliest work a nd new research on Modern South Arabian languages. In a zealous t one of speech, Morris sheds light on the historical aspects of these languages and traces their origin and relatedness to Old South Semitic and Ethiopian. Her published talk is comprehensive and serves as an excellent sociolinguistic background to these languages as it also presents the social life and traditions of the speakers of these languages. She finally talks about her own project which aims at collecting the poetry of Socotri in order to preserve this pivotal literary wealth and anticipates the future of these languages. In the most recent Encyclopedia of Arabic Language and Linguistics (2009), Lonnet delineates some historical and linguistic facts about Modern South Arabian languages She explores the historical events shaping the development of these lexical borrowin gs, comparing and contrasting the shared roots of some representative forms of these languages with Arabic. Furthermore, she discusses some of the non standard notation of the fricatives and offers their IPA equivalents. She argues that there is a recent t endency for the ejective fricatives to be realized as pharyngealized or uvularized due to contact with Arabic. She also discusses some of the peculiarities in the verbal and nominal morphologies, enumeration system and particles of these


49 languages. Histori cally, Lonnet argues that which she describes as continental, develops away from the influences of Old South Arabic, Arabic and Eastern Modern South Arabian languages. Thus, (pp.297). This con clusion refutes the claims made by native speakers of who insist on the relatedness of to Arabic (Al Mashani 1999 & 2003; Al Shehri 2007). Belova (2009) explores the affinities and differences between South Semitic languages (Old South Ara bic, Modern South Arabian and Ethiopian languages) and Arabic (including Classical Arabic, Modern Standard Arabic and the new dialects of Arabic). She argues that South Semitic languages are heterogeneous and the common only with som e particular group of dialects (of Arabic and South Semitic) which poses a challenge to positioning Arabic within the same sub grouping of South Semitic. She outlines a number of distinguishing phonetic, phonological, morphol ogical and lexical features of Arabic and South Semitic. More relevant is her discussion of the phonological processes prevalent in Modern South Arabian languages and the nominal system, including the exploration of the noun plurality in these languages. B plurality is reviewed in the section on plurality in Chapter 2 Al Mashani (2003) provides a comprehensive introduction to the language and people He proposes the long name for this language based on his belief that it is a shared linguistic wealth and heritage. He reviews the major works of carried out by Arabs and Europeans and provi des commentaries and reflections about these works. Finally, he translated the lexicon he wrote in 1999 into Arabic with some additions and modifications.


50 Below I review the works that deal with the pho netic and phonological aspects of These works are more linguistically focused than the above reviewed works as their objectives are to explore specific phonetic and phonolog ical features of the language. Johnstone (1975) argues that Modern South Arabian languages have glottalized or ejective consonants which were thought to be only a privilege in Ethiopian and some Cushitic languages. His field work in Oman and Socotra documented a number of words which were phonetically tested. It is proven that these languages poss ess such a group of glottalized or ejective sounds. Towards the end of the article, Johnstone involves the opinions and reactions of some prominent scholars who were made to listen to the recordings, about this discovery. Much valuable discussion and interrogation are raised about the origin of glottalization in Modern South Arabian languages. Interestingly enough, this type of discovery reveals interrelatedness between Modern South Arabian languages and Ethiopian languages. It will be worthy to explore the exact linguistic features shared between these two groups of Semitic in future research or ejective sibilant / / (pharyngealized sibilant). She argues that this sound has been found in Mehr presents relevant data for this sound from these languages Her data come fro m two sources: a / /, fourteen of which


51 synchronically correlates with / Looking closely at her data, I observe how / / i n certain plurals and duals corresponds with in the singulars. In quite a large number of forms, the opposite happens. She also includes a detailed discussion of the origin of these roots. She fin / / results from palatalization of I observe that the comparison is not always valid as some of the so called or cognates do not seem to be so. Discrepancies of meaning and distinct radicals in th e roots between the correspondent forms can easily be identified Scholarship on the Nominal Morphology Works that document the morphological behavior of the language are numerous. Much research observes vast differences between the morphology o f nouns and verbs and prefers to discuss them separately Below, I first review the works of nominal morphology and then move to discuss scholarship on verbal morphology. In his attempt to rectify his own mistakes in (Matth ews 1957 7 and 1960), Matthews (1969 ) explores the true phonological nature of the deleted / m / resulting in a nasalized vowel 1969 : 23). Matthews argues that the d eleted / m / before a vowel word initially and medially is a determiner in al in de finite, they surface as [ e sk], [ e 8 (Matthews 1969:25). Thus, 7 A paper g iven at the 24th International Congress of Orientalists, Munich. 8 n / n / where m deletes (e.g. [ isk], [ ol] ) However, I follow the IPA method of transcri ption which shows nasalization on vowels when / m / deletes.


52 remarks, Matthews severely criticizes the hasty deductions and demonstrable inaccuracies made by Brittner and his colleagues of the Austrian expedition in regards to their ignorance about the true nature of the deleted / m / Parts in Modern South Arabian L reveals how Modern South Arabian languages express body parts using words different from those widely used in other Semitic languages. According to Lesalu, these words qualify South Arabian languages data come from various sources including scripts collected by the Vienna Expedition. His study classifies vocabulary into: words common to all Semitic, words found in South Arabic and South Semitic, words exis ting in South Arabic and North Semitic, words common in South Arabic, Akkadian and Ethiopic and finally vocabulary shared by South Arabic dialects only. His way of listing the words extracts the bilateral trilateral and quadrilateral consonantal roots an d lists all relevant words used in each and every language under these types of roots. Johns speak Eastern and Central To begin with he worked with a speaker of Eastern who is a native speaker of Mehri Johnstone o n the basis of the words and texts collected from that informant managed to write a word list. However, after learning that 1981: xiii), he started to re write his list so that it conforms to Central eted a on and published it. This lexicon includes background information about this language, its verbal system with its peculiarities, conjugated prepositions and the


53 lly documents unusual notation used to transcribe the language. Johnstone (1970) observes that previous work carried out by the Sdarabische Expedition on Modern South Arabian lang uages does not make any assertion whether statement made by Matthews (1969 ) who lists three forms for the definite article a ha and but gives no evidence for this proposition. J ohnstone meticulously investigates whether Modern South Arabian l anguages have a definite article by studying numerous examples representing four Modern South Arabian languages which he collected during his fieldwork in Oman. His data reveal that Mehri and Socotri mark definite article by attaching a ha and word other hand, have i je and which he calls a pro s thetic 9 vowel. Johnstone observes that the genitive and partitive forms mentioned in his notes for the first time are not marked by these prefixes. He also notes that this prefix can also be detachable or becomes a radical leading to the conclusion that this prefix has t he tendency to lose its meaning and for a form to be used in its prefixed form only. He also observes that there are etymologically monosyllabic forms i n which the attached prefix has become one of the radicals especially in arsusi and Mehri Johnstone (1980a) argues that gemination which occurs in nouns marked for the definite article and with certain forms of the causative verbs is fairly a recent both these 9 According to the Encyclopedia of Arabic Language and Linguistics (2009: 728), prosthesis involves the addition of a short vowel to prevent the occurrence of impermissible consonant clusters word initially.


54 sets, only certain sounds are geminated. He further claims that gemination as a morphological feature distinguishing meanings of words has long been lost from these dialects. He also believes that gemination cha a f unction of argues that borrowed verbs which have gemination in some forms and not others pro To begin with he discusses forms with the definite article and reveals that the most recurrent form of the definite article in Modern South Arabian languages is e with variation in the q uality of this vowel when a form begins with guttural sounds. However, in some cases, gemination occurs as a marker of definiteness in forms with gutturals. To illustrate his propositions, he lists many nouns that occur with the prefix e in their definite form. He observes that if / b/ or / m/ are the first consonants in a word, then they get elided so that the e prefix is no longer v isible as a marker of definiteness He also observes that the definite article is elided in forms that begin with voiceless co nsonants. Forms that begin with glides behave the same way as words that begin with a glottal stop. In verbs, Johnstone also notes that when an initial radical is deleted by a rule, the medial consonant gets geminated. In his conclusion, he offers no expla nation as to why gemination continues to develop in these two groups. Such a study highlights many interesting phonological and morphological and provides a basic understanding of the interaction betw een morphology and phonology. to the author and done in Oman and Yemen with other research ers from Tokyo


55 University of Foreign Studies. The book is more or less a semi dictionary of Mehri, in which the equivalents of some English words and expressions anguages, this quasi dictionary does not depend on the consonantal roots for the listing s of these for which their equivalent forms in Modern South Arabian languages are given. Before Nakano gave the lists, he provided the readers with the phonemic inventories for these languages followed by some remarks and notes about the notation not shown in IPA. Not all the languages have equivalents for the English words and there are many phonetic affinities in the shapes of the words expressing certain meanings in all of the Mode rn South Arabian languages. Johnstone (1973) explores diminutives in Mehri, Socot ri observes that diminutives have infrequent occurrence in Modern South Arabian languages, he asserts that they may surface in speech only relevant to a few social contexts such as praise, blame, commiseration and when women talk to c hildren. Thus, they may serve a usage. He lists two main patterns for diminutives in each Modern South Arabian language and supplements them with and C e C e C en He lists ample forms for these two patterns and shows their various connotations. Furthermore, he investigates plurals of diminutives and diminutives of body parts. Finally, he shows that the patterns of diminutives in his study deviate from the shap e C u C a C which was thought to be the pattern of diminutives in Modern South Arabian


56 languages. The second type of diminutive relates to CVCVC an The vowel quantity of the diminutive marker an is long but its quality is hard to establish. In 1983, Johnstone wrote an article on the enumeration systems employed in 34 South Arabian languages. He illustrated how these languages differ from each other in their enumeration sy stems. T his work serves to acquaint readers with the enumeration system in these languages. In his succinct article whose length is only three pages, Testen (1998) concludes w ] whereby w / is phonetically realized as a w glide (Nakano 1986:v) This conclusion is based on surveying its prevalent shape in numerous common Semitic languages including Ara bic, Biblical Hebr ew, Akkadian 10 Ugaritic 11 Syriac 12 13 However, the stem of this numeral in Modern South Arabian languages is remarkably different in two respects: the absence of the initial ti w / sound. This res ults in as the constructed stem for Modern South Arabian languages. Testen changes. For example, he argues that the appearance of the sibilant reflects a regular 10 2000:xxi) 11 12 Syriac is a dialect of Aramaic which spread, together with the other Aramaic dialects, from the upper Euphrates (Aram Naharayim) into Syria and Mesopotamia (Rest 2009: 178 182). 13 Semitic group. It dates back to the mid fourth century C.E and continues to exist until approximately the tenth century C.E (Belova 2009:301).


57 phonological change which occurs prehistorically. As for the absence of ti it resembles the widespread phenomenon of the loss of t characterizing certain sets o f Testen claims that this er osion of t reflects an ongoing tendency in the language to abandon t word initially comparable to Ethiopian languages which merge /i/ and Testen ideas require more evidence and intensive survey of numerous data from the language to validat e his conclusions Al Mashani (1999) thoroughly examines the lexical relation between Classical ot and meaning with Classical Arabic, those that correspond in root a nd have similar meaning with Classical Arabic, those whose meaning is similar to Classical Arabic but different in root and finally those that are exactly similar in both root and meaning He also In the verbal paradig ms of Semitic languages, the t prefix marks the 2nd and certain 3r d person forms. Johnstone (1968 and 1980b) observes that there is a t prefix a t prefix loss in certain verb al themes including the indicative, subjunctive and conditional forms of causative verbs, intensive conative verbs, quadri literal verbs and passive verbs, Socotri also elides this prefix in the reflexive and hollow verbs. J ohnstone (1980b) presents lengthy paradigms of these themes proving that there is a systematic loss of the t prefix. He explains that this loss


58 represents a well especially in 3 rd feminine singular, 2 nd subjunctive and conditional passive forms extends this lo ss to the non occurrence of the i / j prefix and the n prefix of the first person plural. He finally draws the conclusion that of the Semitic verb extension an Testen (1992) offers a phonological analysis of the phenomenon of the loss of the t li mentioned in Johnstone (1968 and 1980b). He thoroughly investigates the types of verbs that exhibit truncati on of the t prefix and divides them into two classes: verbs preserving t and verbs lacking t According to Testen, although fication becomes justifiable when comparing them with their stem types cognates of Literary Arabic. He obs erves that t from verbs whose cognates in Literary Arabic have the pre radical vowel /u/ He then explains the loss of the t prefix from phonological and historical perspectives pointing out that the t less f orms result from a change in the initial sequence tu to a simple vowel in the course of the morphological development of the verb. The loss does not happen when the pre radical vowel is /a/. Testen also explains why j and n disappear from the passive pa t prefix before /u/ in open syllables, it also loses j n and perhaps the glottal stop in closed syllables. According to Testen, ongside Arabic and Akkadian,


59 providing evidence that early Semitic distinguishes between the pre radical vowels / u / and / a / in the prefixed tenses of the verbs. H ayword et al (1988) study the vowels contained in the verbal paradigm in Their study w as (1981). That introduction describes two conju gational classes (C(A) C and C(B) : ) of the These classes mainly differ in the shape the third person masculin e singular takes in the perfective, imperfective and subjunctive forms. While most verbs can be classified under these two classes, Johnstone notes that verbs whose consonantal roots are characterized by weak radicals, gutturals or /b/ and /m/ have idiosyn cracies and may thus diverge from the two classes. Hayword et al focus on the effect of gutturals contained in the roots of some verbs on the vowels accent placement and CV shapes of these verbs. They observe that the conjugation taken by these verbs is h ybrid and display features of both classes. After establishing discuss how the vowels contained in verbs with gutturals differ. They also discuss the effect of accent placement of the nature of the surface form of the verb. Towards the end of their study, they manage to linearly derive a small subset of verbs with gutturals. I consider this study to be quite systematic. It also lists a number of phonological processes i n the language which have direct bearing into the verbal paradigm of the language. The syntax of has been examined in a PhD dissertation by Hofstede (1998) who


60 parts of a nominal, prepositional and adverbial phrase a nd explains how relative clauses verbal clause and the simple verbal clause with two crucial aspects relevant to the simple clause: interrogative and comparative syntactic constru ctions. She also offers a description of complex clauses expressing sentential arguments, embedded question complements, indirect quote and adverbial clause. Besides syntax, she investigates negation, degree of comparison and interrogative. Although her study is mainly descriptive in nature with numerous illustrative examples of syntactic and morphological structures, she sporadically ventures into offering a phonological descriptive analysis of the contexts when a particular structure occurs. For example, she describes why a particular structure takes on a plural meaning by listing its syntactic and morphological contexts. Moreover, when a certain morphological function is expressed by distin ct allophonic structures, she explains the phonological contexts of each. Ho with illustrative examples from natural speech. vocalic inventories of the language in separate sections followed by a brief discussion of the phonological processes pertinent to consonants and vowels. I also offer a brief description of the supra segmental inventory, which sketches an overview of the syllabic It is importan t to


61 note that these two linguistic aspects have been poorly understood and little attention was given to them in the literature of Sound I nventory Consonants 5 phonemic cons j (masculin j el at sing. and pl. / is j ] in t he context of j j el at w rounded vowels such as /o/ and w sing. and pl. w sing. and pl. w sing. and pl. and as phonemes (c.f Hofstede 1998:19 for a comprehensiv e list of the consonants in all varieties of with the voiced consonants from a morphological perspective. For instance, he observes that nouns with initial voiced sound s or ejectives when attached to the prefix e a e.g. / e e Hofsted e (1998:27 30) also observes that the definite article precedes a noun if


62 its first root consonant begins with a voiced sound or an ejective; otherwise, the noun occurs without the definite article if it begins with a voiceless sound 14 Thus, they behave as a natural class together. Other evidence comes from the verbal morphology of this language. Johnstone maintains that a morpheme e in the derivation of certain verbs deletes before voiceless consonants and remains before voiced and ejectives/ glottalized sounds. To i absence of e verb initially (Johnstone 1981: xx). The articulated in terminology) and is always aspirated if it occurs word initially and precedes a front h In Table 2 1 I only include the phonemic consonants of Central point and manner of articulation indicated The transcription notation used throughout this dissertation is mainly IPA Due to the fact that most emphatics are realized as ejectives and have glottalization ( and sometimes post glottalization ) in their realization, I transcr instead of the dot beneath the sound which represents velarization or pharyngealization in Arabic emphatic sounds. 14 This observation however, is not without exceptions (c.f Hofstede 1998:27 and 30). For example, when the first root consonant is a semi vowel, it becomes unclear whether t he definite article precedes or follows the noun.


63 Table 2 1 Phonemic c hart of c onsonants bilabial labio dental lamino dental alveolar lamino postalve olar velar pharyngeal glottal Stops voicelesss t k voiced b d g ejective pharyngeal t Fricatives voiceless f s h voiceless ejective rounded sibilant w pharyngeal sibilant voiced z voiced pharyngeal z


64 Table 2 1 Continued bilabial labio dental lamino dental alveolar lamino postalveolar velar pharyngeal glottal voiceless pharyngeal s voiceless lat. fricative voiceless la t. fricative ejective Affricates voiceless ejective lateral Nasals voiced m n Liquids voiced trill r lateral voiced approximant l central voiced approximant j w


65 Vowels Simil [but not in quantitative] 1998:196) because the language has undergone a change shortening long vowels. According to R atcliffe (1998:196), proto Semitic long /aa/ and short stressed /a/ are but also commonly /o/ and /u/. The chart below outlines Table 2 2 Phonemic chart of v owels front mid back high i e u o mid low a has two major phonological processes that relate to its vocalic inventory: raising and backing. A vowel followed by /m/ is usually raised; becomes [u] and /e/ becomes [i] ( Johnstone 1981; Hofstede 1998). To illustrate, in the verbal paradigm, the impefective takes the shape in CV 2 C whereby V 2 is consistently /u/ in the vicinity of a nasal /m/. Compare, for example, [ind u u ( n in the imperfecti ve prefix in is assimilated to a following fricative ), versus [int (Johnstone 1981:xxiii). The vowels / e / and / / have the allophone [ a ] which can be a front or back vowel. Vowel length is not contrastive a nd long vowels are relatively rare, except where elision is involved. To illustrate, long vowels result from the deletion of / w / or intervocalic / b /, or as a merger of the definite article with the first vowel of a noun whose first radical


66 is a glottal sto p, e.g. / /= [ (Simeone Senelle 1997:382; Johnstone 1981: xxx). Long vowels may also result from the deletion of /m/ word initially as in l / [i:nz ] or nz Nasalized vowels are relatively common and they occur when / m / is deleted or re ti ] Nakano (198 6 ) shows that any vowel in can be nasalized. The vowels /i/ and /e/ are diphthongized and realized as [i j ] and [e j ] respectively. The vowel /o/ is realized as [o w ] which can be also realized by some speakers of and /u/ is realized as [u w ]. Simeone phonemic vowel invento ry. However, Hofstede (p p .21) finds the diphthong [ ] in her at], [ [ ] corner, sing. and pl. [ b ] [ ] wooden doll, sing. and pl. Phonological P rocesses P ertinent to C onsonants its consonantal and vocalic inventories. These crucial phonological tendencies are ( 1975:95 104 ) Johnstone ( 1981:xiv ) Lonne t (1985), Simeone Senelle (1997), Hofstede ( 1998 ), Belova (2009) and Lonnet (2009) I will outline the major phonological tendencies that pertain to


67 consonants and then move to give some remarks on the vocalic inventory of the language in a subsequent section. Devoicing and a spiration These phonological processes concern stops. Voiceless stops are aspirat ed word initially and voiced stops are devoiced word finally. To illustrate, /kot/ is realized as [k h ot] [k h ] The pharyngeal phoneme / / is always aspirated if it occurs word initially, and precedes a front h Elision initial and intervocalic position and their loss affects t he length and quality of a following vowel. To illustrate, the vowel following a deleted /b/ lengthens whereas the one following an elided /m/ nasalizes or lengthens [ l ] always delete, e.g. [e 1981; Hofstede 1998). Palatalization Palatalization occurs when a consonant (usually a stop) is followed by a front (high) vowel, typically /i/. For example, in Je j j el at] palatalization may vary depending on the phonetic context, dialectal variety and speaker. Moreover, certain other sounds are palatalized too in the vicinity of a high back vowel like /u/ w when /k/ is followed by a back rounded vowel as observed in [ kabid ] which is pronounced as [ w u bd / is a


68 Belova 2009:303). Insertion insert word finally a glottal stop or /h/ in borrowed words from Arabic which end in a vowe I also observe that and there is no pho nological context that restricts the insertion of /h/ over / or the opposite (/ over /h/) As a matter of fact, some speakers accept the insertion of either /h/ or / in the same form, so these may be interchangeably added in a form. Where the final c onsonant of a word is a liquid or nasal, the combination of the following sounds have hn#, hr#, hl#, hm# and 1981:xiv). Fortition and lenition d lenition) which simultaneously affect the glide /w/ and the stop /b/ in borrowed words from Arabic which underlyingly have these two sounds The glide /w / is normally realized as a glottal stop / ~ / ~ / ~ /~ xiv). At the same time, intervocalic /b/ is substituted by /w/, /j/ or vowel lengthening, e.g. /e bot/


69 Subs titution as well as in other Modern South Arabian languages, there is a free alternation between /f/ and / / (Belova 2009:303). To illustrate, [ is articulated as Metathesis xical forms borrowed from Arabic which have undergone metathesis. Examples include [lat respectively ( more examples can be found in Al Mashani 1999) Supra Segmental Inventory Syllabic structure Unfortunately, Simeone Senelle (1997:382) who devotes a short paragraph stating that the com mon syllabic shap es are CV(C) and CV:. She states that consonant clusters such as CCV(C) or CCV: are not uncommon in word initial position Moreover, in word final position, syllable s consonant clusters word initially by inserting a pro s thetic vowel in front of the first pro s thetic vowel licenses the Howev er, especially singular forms generally begin with a consonant and end in a consonant too Where a form otherwise ends in a vowel, a glottal stop or CCVC and V:C syl lable structures. The syllable shapes VC and V:C are restric ted to word initial position as ] :g ( t he


70 g vowel nasalizes and lengthens) The shapes CVC and CV are very common and occur in word initial, medial and final positions This is illustrated in As for the syllable structure s CCV CCVC and CVCC they are mainly observed in word initial and final positions, as in and [ mhot [mas .t are phonologically derived; they surface after the deletion of a nasal /m/ or /b/ Stress pattern prominent syllable (Simeone Senelle 1997; r is that all syllables, the vowel of which is not anaptyctic tend to be equally prominent, the vowel of an open syllable being half However, Hayward et al (1988) have a different view about stress. In their study of the vowels contained in aim that they did not note any instances of verb s containing more than one accented syllable (pp. 247). They devise an accent rule for the perfective verb in xcept in the case of 3 f eminine singular, 3 masculine d ual and 3 f eminine dual forms (in which accent always falls on the inflectional suffix), the accent falls on the second stem vowel unless this is followed by a guttural, in which case it falls on the first stem vowel. Stressed vowels are slightly longer than unstress ed vowels in open syllables and in final .CVC# syllables. I observe that a noun maximally has two stressed syllables and


71 stress always falls on the last two syllables provided that the vowels in these syllables are full vowels and are not prosthetic. In a form of three syllables, the last two syllables are stressed. w shell used as a receptacle for rule that goes as follows: sta rting from the end of a form, stress falls on the last syllable and the syllable immediately preceding it provided that the vowels in these syllables are full and not inserted Basic stress du e to the fact that only a single syllable has a full vowel. The templatic shapes of C and imperfective C C C (Hayward et al 1988:241) Overview of the Nominal Morphology ine) and three numbers (singular, dual and plu ral). The feminine markers are and Simeone Senelle puts it generally as (V)t whereby (V) can either drop or its quality may vary considerably depending on the phonological context surrounding it. She maintains that some feminine nouns which are borrowed from Arabic take an / h / ending. D ual is externally indicated by the suffix i and is usually followed i t ro] ity is no longer systematic ally marked and dual can be used to signal plurality (Johnstone 1975:113). Plural forms, on the other hand, have diverse shapes and can be externally and internally marked. Simeone Senelle (1997) presents a discussion of the various CV shapes the singular s, duals and plurals take in the nominal morphology of Modern South Arabian languages. For the sake of this dissertation, I will limit the discussion to the shapes of gulars of the biliteral roots have the shapes CVC and CVC + a


72 feminine marker and the triliteral roots take the shapes CVC(V)C CCVC, CV :C VC, C(V)CVC and C(V)CV :C For the quadriliterals, the canonical shape is CVCCV :C which is qu ite common li imposes internal changes on the root to geminated root consonants are phonologically conditioned and do not indicate a sema ntic meaning. Observe the following meanings for the root as different vocalic melodies are sandwiched between the consonants : ( 2 (A l Shehri 2007:174) ges, have two modes of pluralization : external (also known as sound) and internal plurals. Internal plural s involve internal stem change s such as mapping onto a template, redup lication, ablaut and infixation It has also been noted that a singular form may have many plural shapes in (Johns tone 1981; Simeone Senelle 1997 ). Plurali outh Arabian languages has been compared to some plural patt erns of Ethiopian languages and not with Arabic broken plurals (Johnstone 1975:113). Ratcliffe (1998 a ) argues that plurality is a very revealing morphological process. Therefore, it must be taken into account when classifying Semitic languages The diverse patterns of plural formation should be scrutinized as they can be indicative of where a particular language belongs in the classification of Semitic.


73 Ratcliffe (p p .95 97) makes the following observations about plurality in Je discussion of the broken plural and Semitic sub classification: 1. Ethiopi an Semitic shape than to Arabic. 2. T VCCVC (62 of 207 forms in However, /a/ shows in guttural environment only. Therefore, the most common sha 3. T here are 25 plurals with word initial consonant clusters. They neither have an initial vowel nor an epenthetic vowel to break up the consonant clusters; 4. T he third most common shape of plura reflects a 5. T he sound feminine plural is also commonly found but usually derives from weak root or bi radical singulars. This shape exhibits a vocalic stem change; 6. T he femi nine singular suffix has the shapes et, at and between the quality of the vocalic suffix and the plural form. Singulars taking the shape d feminine plural suffix te. On the other hand, an internal plural shape CVCVC (in which vowel in 26 of 27) and CVCCat (11 of 11). 7. Q uadriliteral singulars take three distinct shapes. The first shape they take is the common southern Semitic shape CaCaaCiC but the second syllable is not long and has Vb ( or ab ) betwe en the second and third radical. 8. T he reflex of the quadriliteral shape CoCoCuC is common for the feminine but r are for the masculine singulars. 9. T he prominent plural shapes in other southern Semitic languages CaGaaCiC and fact that both consonantal and vocalic length contrasts have been lost in the language. 10. T first syllable may be /o/ or /u/ and the vowel of the second syllable may be /u/.


74 11. T T he current synchronic based and focuses on the morph o phonologi cal processes involved in the formation of noun plurals It, therefore, differs from diachronic study of plural formation in Semitic (1998) which mainly aims at documenting the most common tendencies observed in the CV shapes of plu rals and finding a proto type plural form in Semitic. However, some of his observations regarding noun plurals are confirmed by this study. For example, I also observe that the most common plural marker for adjectives is the default plural suffix t(V) (ob servation #11) and agree with Ratcliffe about the fact that this suffix is a feminine plural marker and commonly found (observations #5 and #4); whether it attaches to weak or sound roots is not explored in this study. Moreover, this study conforms with Ra literal nouns most often are pluralized by either ablaut or Vb infixation (observation #7). These two processes are other shapes of singular forms too (bi litera l and tri literal) Contrary to Ratcliffe, I did not see the shape CaC ( /o/, ) CiC which he claims to be also common for the quadri literal forms, as prevalent CVCVVC or CVVCVC patterns and this s tudy also confirms the non existence of such plural shapes in the language. Ratcliffe also observes that the most prevalent shapes of VCCVC y rarely do /e/, /i/ surface in these shapes ( observation #2) I list these shapes under common as other plural patterns. In my data, the initial V in the template VCCVC is pro s thetic and does not appea r in many plural forms. While this study also shows that the feminine suffix bears the shapes et


75 at and it does not investigate if t here is a ny correlation between the quality of the vocalic suffix and the plural form Simeone and other Modern South Arabian languag es. The most common pattern of plura l for the trilitral verbs is CCV :C (a plural for many feminine singulars) and for the quadrilitral are CCV:CC and CCV CC. A common patter CCVC Vb CC and there is also vocalic opposition observed in the last syllable of both the singular and plural forms. Simeone Senelle (1997:388) also identifies that some plural patterns correspond to Arabic plural of the plural (emphasis hers). External plural, on the other hand, takes the suffix ( Sime one Senelle 1997:388 and Lonnet 1985:54). Some plurals with the suffix i come from the dual (Johnstone 1975:113). Simi lar to Simeone ion, this study also concludes that Vb infixation and ablaut are, by and large, the most common mechanis m for plural formation in However, this study does not list CCV:C as a common plural pattern for the tri literal verbs and agrees about plural forms do not involve length in their overall shapes. Simeone Senelle claims that is a marker for the external plural. However, this study does not have the (n) included in the default plural marker t(V) and shows that the /n/ belongs to a different plural suffix Vn and is never optional (as shown by the brackets around it in Simeone ) Despite the interesting complexities involved in the plural formation of this language, Je have not been e xplored analytical ly or theoretical ly before this dissertation The only works which briefly to uch on plurals Ratclif fe (1992,


76 1996, 1998 a & b ) and Simeone Senelle (1997). Belova (2009) offers an interesting discussion about plurality in South Semitic in general and briefly mentions some recurrent shapes of plurals in These studies are descriptive and they only document the common plural shapes in the language. Moreover, they are not solely devoted to the study of plurality in n order to either supplement a Senelle 1997) or to compare atcliffe 1992, 1996, 1998 a; Belova 2009 ). of the new theo retical and analytical phonological and morphological frameworks which prove to offer systematic accounts for the morpho phonological particulars of this language. When describing plurality in they only reference CV shapes without mentioning any re levant phonological process. However, these works do give an idea about the most common patterns of plurals in the language with numerous supportive Therefore, they serve a s a background for a study of plurality in the language. This section review s the major work that discusses and describes plurality and and outline s the knowledge ga ps in the literature with respect to this pivotal research area Most specifically, it reviews Ratcliffe ( 1992 ) Ratcliffe ( 1996 ), Simeone Senelle ( 1997 ) and finally Ratcliffe (1998a & b) In his lengthy diachronic study to reconstruct a proto language for the b roken plural formation in Afro A siatic languages and Semi tic, Ratcliffe (1992) surveys quite a large number of languages revealing diverse patterns of plural formation and arguing


77 convincingly that long aa generally characterizes the broken plurals. In the course of surveying plural patterns in Modern South Ara bian languages, he provides insightful then) and Arabic in particular. Most relevant for the s ake of this dissertation, he argues as a result of templatic expansion for bi radical and weak roots. He also states, with els have evolved into short stressed vowels and their quality has been phonologically neutralized. His discussion on remnant sounds /n/ and /l/ of some plural forms provides basic understanding to the otherwise unusual behavior of some derived plurals whos e singulars have no such sounds underlyingly Ratcliffe (1996) brie fly discusses plurals whose second and third radical is exactly the same sound and argues that these plurals are merely templatic expansion. He maintains that Afro A siatic languages do not express plurality by reduplication. However, reduplication surfaces to conform to some templatic restrictions imposed by the relevant language. He provides evidence based on the behavior of similar reduplicated plur als in other Semitic languages. Ratcliffe (1998 a ) presents valuable discussion about patterns of plural formatio n i singulars from which these plurals are derived. He further illustrates the shapes with examples and deep discussion on the ir behaviors and their general phonological tenden


78 Ethiopian than to Arabic. The major observations made in Ratcliffe (1998) about plurality in Ratcliffe was mystified by the large number of different vowel qualities in what he calls group I plurals (i.e. plurals of C V CC masculine). CVCuC, CVCeC, etc. He also questions the plurals with Vb infix (perso nal communication). He states "t he se forms all seem to go back to CVCaaC and but could also reflect forms wi th inserted /u(u)/ or short /a/ (1998b:198). Moreover, in languages where both internal and external plurals co exist, Ratcliffe (1998b: 219 242) ural is either the obligatory or at least the only productive plural for underived, unmarked nouns of three or fewer consonants (stem shapes CVC, CVCC, CVCVC) while the external plural is generally obligatory for productively derived nouns such as partici made in the analysis of plurals, assumes that the shape of the stem (input) determines the shap e of the plurals (output ) instead of the output singulars serving as th e base for the output plurals. However, it is important to remember that Ratcliffe has a different purpose of studying plurality (comparative and historical with the aim of reconstructing a proto plural in Semitic) wh ile this study is phonology oriented. Simeone and other Modern South Arabian languages. She outlines the most common patterns of plural in this language and other Modern South Arabian languages. Her list of the plural shapes in the language serves a good backgr ound for common plurals of Jebb


79 Belova (2009) discusses some plural shapes taken by Jebb and other Modern South Arabian languages. For instance, she observes that the Cu C u:C pattern is so rare in Modern South Arabian languages ). Moreover, som e plural CaCu:C or CiCa:C She also argues that the plural shape [ awr]/ [he weret ] which occurs in is relatively rare in other Modern South Arabian languages. Co ncluding Remarks Johnstone (1975), Simeone Senelle (1997) and Ratcliffe (1998) whose works are reviewed above have made crucial observations about plurality and identified the most he understanding of this morphological phenomenon and highlight certain facts including the closeness of insightful discussion of the study of plurals in Modern S outh Ara b classify Semitic on the basis of the behavior of their plural formation are quite illuminating and revealing. Moreover, the tendencies of these plurals to resemble plural patterns of Ethiopian languages may support their inclusion under the same branch. I particularly acknowledge the organization of the plural patterns in Ratcliffe (1998a & b) and his thorough discussion based on the thoughtful comparison he made among plural patterns in Semitic languages. However, these studies are not without shortcomings. The fact that they are not holistically comprehensive references to this issue. Simeone Senelle (1997) talks briefly about plurals in Modern South Arabian languages since her work focuses mainly on sketching


80 the grammar of these languages and acquainting the readers with the peculiarities of their phonological and morphological aspects. Ratcliffe limits his discussion to certain plurals Summary of Chapter 2 This ch apter serves as an overview of the language under study in this dissertation. It discusses the geographical location of the language, its speakers and genetic affiliation. It also situates the data that will be analyzed in a subsequent chapter and reports major scholarship done on and other Modern South Arabian languages. It also sketches the phonemic inventory of the language using the IPA which previous works on did not fully embrace. It was indeed very hard to follow these works since some non standard notation fi gured prominently with confusing descriptions about the place and manner of articulations of the phonemes in This dissertation will hopefully be a reference to the IPA phonemes of the language and smooth the path for future research on In this chapter, I also provided a description of the prominent syllable structures, stress nominal and verbal morphologies of the language. Finally, I reviewed previous scholarship on plurality in outlining the background information on this area of research. Chapter 3 discusses plurality in a number of Afro Asiatic languages and highlights the most prevalent plural formation processes in these languages. It also explores some salient phonological alternations that accompany the formation of noun plural in these languages.


81 CHAPTER 3 PLURALITY IN AFRO ASIATIC LANGUAGES One of the most intriguing and much investigated morphological phenomena in Afro A siatic languages is plural formation. This phenomenon, which has exposed unusual mechanisms for for ming plurality, has long captivated the interest of linguists who explore the diverse plural patterns in these languages with considerable enthusiasm (c.f. Worrel 1920 for Hamitic 1 ; Vergote 1969 for Coptic and Egyptian; Zaborski 1986 for Cushitic; Arabneh 1978 for Hebrew; McCarthy and Prince 1990 for Arabic; Racliffe 1992, 1998 a & b for Berber and a host of Semitic and Afro Asiatic languages; Newman 1990 for Chadic, Buckley 1990 for Ethiopian; and subsequ ent works on plurality in Afro A siatic languages). Af ro A siatic languages exploit a number of distinct mechanisms to mark noun plurality. For instance, in addition to the usual suffixation mechanism widely attested in Indo European languages 2 noun plural in Afro A siatic languages is also expressed by an int ernal plural which constitutes the most common type of plural formation for a number of Afro A siatic languages including Arabic (Levy 1971; McCarthy and Prince 1990a; Abd Rabbo 1990; Abu Mansour 1995 among others). Suffixation, however, is not purely sound linear or straightforward in many of the Afro A siatic languages (e.g. Hausa, Hebrew, Berber and Amharic). Much allomorphy in the stem happens when a plural suffix attaches to the singulars in order to meet certain phonological or morphological r equirements imposed by the language. 1 Hamitic is no longer a valid language family; however, the term remains in use in academic works done by European scholars. Since the term is considered to be obsolete, I will replace it with AfrasoArabia n, based on the fact that its populations are native to North Af rica, Horn of Af rica and So uth Arabia 2 Indo European languages also mark plurality by non concatenative morphological processes. For example, English has ablaut e.g.


82 Ratcliffe (1998: 86) shows the relative distribution of internal and external plural formation in Afro A siatic languages. He maintains that Hebrew and Aramaic form their plurals mostly by suffixation; the majority of d erived nouns and nouns with more than three consonants take external plural most of the time. On the other hand, Arabic, Old South Arabian languages, Modern South Arabian languages and Tigre utilize internal plural to mark plurality in their nominal morpho logy. Where the internal plural is productive, the external plural for these languages is obligatory in derived nouns only. The two most prevalent mechanisms for plural formation in Afro A siatic languages are suffixation (sound plural) and internal (broke n) plural. In Berber, there is also a mixed plural in which a combination of internal and external plural markers occurs to mark plurality (Ratcliffe 1992: 475). Under internal plural, there are many sub patterns such as affixation with internal change, re duplication, mapping singulars onto diverse plural templates, infixation, and vowel opposition. Ratcliffe (1992), in his diachronic study to re construct a proto Semitic language on the basis of plural formation, has classified these diverse plural mechan isms into five major types of pluralization: Suffix Type, Tonal Type, Reduplication, Internal or /aa/ Type and Lexical Type. The tonal type is observed in tonal languages which al so use tone as a contrastive feature to mark plurality. Because this type is This chapter discusses the major morphological mechanisms employed in the plural formation by different Afro A siatic languages. It also describes the phonological consequences of these mechanisms. More specif ically, it surveys the most common mechanisms of plural formation in Arabic, Hebrew, two Ethiopian languages (Tigrinya and Amharic), Berber, AfrasoArabian languages, a host of Chadic languages (most


83 significantly Hausa), Cushitic languages, Modern South Ar 3 an d finally Egyptian and Coptic. Although the list of languages surveyed is, by no means, complete the discussion herein highlights the major attested mechanisms. I observe that there appears to be no appreciable distinction a mong the languages or dialects classified under a family language in the general patterns of plural formation Therefore, my exploration of plurality in Afro Asiatic languages is limited to representative daughter languages since the same plural patterns m ay be the same in the other daughter languages which belong to the same family language I also observe that the majority of languages surveyed here employ the same mechanism s but the details (i.e. shape of the suffix or the phonological consequence that t he suffix entails) vary greatly. I also note that quite a few languages use similar suffixes to mark plurality; t(V) Berber and Arabic use n in their plural morphemes: Arabic has u:n and Berber utiliz es n for masculine and in for feminine. In order to offer a succinct discussion, I only outline the most prevalent and commonly used mechanisms of plural formation that relate to the patterns of plurality d in Chapter One I classify these patterns into two major mechanisms: Suffixation and Internal Plural. Under Internal Plural, I list a number of relevant mechanisms such as ablaut, mapping onto templates, infixation and reduplicati ng a consonant from the base Before I embar k on the description of these mechanisms, I will present the most widely accepted genetic classification of the surveyed Afro A siatic and Semitic languages, summarized in Robert Hetzron in 1997. 3 The reason for limiting the discussion to these two Modern South Arabian languages is the non availability of literature on plurality in the other languages.


8 4 Then, I will outline a number of observations relevant to plu ral formation in Afro A siatic languages. I conclude the discussion by briefly outlining the shared characteristic A siatic languages. Classification of the Surveyed Languages The Afro A siatic language fa mily includes six main branches: Berber, Chadic, Cushitic, Egyptian/ Coptic, Omotic and Semitic. There are five major nodes within the Semitic branch: Arabic Northwest Semitic Ethiopic South Arabian and East Semitic (Faber 1997). This chapter highlights the plurality mechanisms in Berber, Chadic, Cushitic, Egyptian/ Coptic and the first four nodes of Semitic in the order presented in t he classification of Semitic below. It will also discuss the plural for m in four language groups in the Afro A siatic language family: Berber, Chadic, Cu shitic and Egyptian/ Coptic Hausa. The discussion of Cushitic includes the Bish ari dialect and Afar Saho. Finally, I review plurality in Egypt ian and Coptic As mentioned above, many of these languages display similar tendencies in the general patterns of plural formation. Moreover, the majority of daughter languages classified unde r the same language family show no appreciable distinctions in plural suffixes or internal change Therefore, I only choose representative examples which showcase the shared plural patterns between these languages and the language under study. The language s discussed in this chapter are genetically interrelated and many affinities in the plural formation processes exist among them. The representation below


85 reveals the classification of Semitic, and lists the daughter languages whose plural formation tenden cies will be subsequently explored. (1) Classification of Semitic Arabic Northwest Semitic Canaanite languages Hebrew Aramaic languages South Semitic Western South Semitic Ethiopic Cushitic Bishari Dialect Afar Saho North Ethiopic Tigrinya Tigre South Ethiopic Amharic Gurage Eastern South Semitic Observatio ns on Plural Formation in Afro A siatic Languages Some Afro A siatic languages make fine semantic distinctions between the different types of plural patterns they exhibit. Moreover, different plural patterns may serve d istinct grammatical functions. For example, Arabic has three types of plurals based on their semantic relation (plurals o f paucity, plurals of multitude or multiplicity and collective plurals). Plurals of paucity relates to the numbers from three to ten wh (e.g. Omani Arabic) which no longer references how many numbers a plural denote s. By and large, internal plurals are used to express multitude in a wide range of Arabic


86 dialects. Cushitic languages have singulative, collective and paucal plurals (Zaborski r singular number, often because the bare form is a collective (singulatives and collectives are common in Arabic too) (Kramer 2009: 183). The usual scenario cross linguistically is that singulars are morphologically unmarked. However, the Cushitic family and Arabic mark singulars too. Chadic languages encompass noun plurals and pluractional verbs which mark numbers in verbs (Newman 1990). One of the peculiarities of plurality in Chadic is the identity of markers of nominal plurals with the markers indicati ng the at the same time (Frajzyngier 1977: 37). Yimam (1996) also claims that there are four numbers in Amharic: singulative, singular, paucal and plural. Paucal refers to a number of items. Plurals in Afro A siatic languages can be sensitive to gender. To illustrate, the majority of regular Arabic plurals display sensitivity to gender; singulars which belong to the feminine class take the plural suffix a:t This can also be attested in a number of Afro A whose default plural suffix t(V) is also a feminine plural marker Furthermore, in Cushitic languages, the feminine singulars are attached to a plural suffix different from the plural suffix attached to the masculine sin to masculine and feminine. gender (masculine or feminine), number (singular or plural) and state (absolute or


87 One of the striking characteristics of plurality in AfrasoArabian languages is that the plural of a masculine noun must be feminine while that of a feminine must be masculine (aka polarization or polarity) The feminine ending, which is the marked class for these languages, indicates the plural of a masculine noun. Worrel (1920) argues that this is the case for Somali collective plurals too Thus, i n AfrasoArabian the plural of many masculine nouns such and become mascul Suffixation The most widely employed mechanism for plural formation in Afro A siatic languages is suffixation. Almost all the languages surveyed here are observed to use suffixes to mark plurality. However, the shapes of suffixes us ed vary considerably from language to language. Below, I survey how various Afro A siatic languages use suffixation to signal plurality. In the Arabic sound plural, masculine singulars take u:n in the nominative and i:n u:n] i:n] a:t when marking plurality ( ) Suffixation in Arabic is straight forward and the length in the final syllable of the singular form is not 4 4 The syllable shape CVVC is only allowed word finally. In this example, because the plural suffix begins wi th a long vowel, the new syllabification makes the next to last syllable .CVV. instead.


88 is external plural. Ravid and Schiff (2009: 50 52), who thoroughly discuss plurality in conjunction with a host of phonological alternations in Hebrew, also claim that forming plurals in Hebrew is mainly a stem suffixation of im for the masculine nouns; e.g. ot for the feminine nouns, e.g. [sira] Hebrew are typically e, e.g., [mixe while feminine singular nouns are marked by the stressed e.g., [par [ mor ] [ kon ] t or its variant ut, e.g., In Hebrew, suffixation may keep the stem intact or it can impose morphological and phonologic al changes into the stem to which the suffix attaches There are five major stem changes outlined in the literature of plura lity in Hebrew (Arabneh 1978; Ravid and Schiff 2009). The first phonological consequence is vowel reduction or deletion As stress s hifts, the used to be stressed vowel either shortens or deletes all together, e.g. [p stressed vowel is preceded by a sonorant, then another vowel is inserted. To illustrate in the form [m lon] fter /a/ deletes and the form becomes /mlon/, /e/ is inserted. The final plural shape becomes [melonot] in which the plural suffix ot is attached The second phonological stem change involves vowel opposition/ change For instance, a monosyllabic noun


89 involves an insertion or deletion of t. While some feminine singular nouns ending in t lose it, other feminine nouns ending in a vowel a retrieve it. The fourth phonological change involves a stop/ spirant alternation which is conditioned by many morphological an d lexical contexts. According to Idsardi (1997) the basic generalization th at governs stop/ spirant alternation in Ti berian Hebrew stops appear elsewhere (postconsonantly and at the beginning of word s following a p into spirant .g. [af] Finally, a host of miscellaneous phonological changes One of the phonological consequences for attaching the plural suffix in Hebrew is stress alternation in nouns. The stress moves to the final syllable in plurals. For example, when the plural suffix im is a In the feminine singular stems, if the feminine suffix t precedes a stressed vowel as in [sapar the plural suffix ot attaches to the singular form after the feminine suffix t deletes; [saparijot] results Secondly, if t precedes a non stressed vowel, the final Vt drops and the stressed plural syllable is formed in the same manner, e.g. [malkd et] on take the feminine plural marker ot Zaborski (1986) is a comprehensive reference to the morphology of nominal plurality in Cushiti c languages. He has offered substantive exploration to all the


90 occurring patterns of plurality in Beja (North Cushitic), East Cushitic, Arbore Elmolo Dasenech, Highland East Cushitic, Agaw and South Cushitic along with a comparative discussion of the pecul iarities in each and every language. He observes that Cushitic languages have both external, internal, and reduplication modes of pluralization. One of the most widely attested plural suffixes is a suffix attaches to various singular shapes including bi radical, tri radical singulars, and singulars ending in the suffix ej s the board. In bi syllabic words, the vowels /i/, /e/ and sometimes /a/ elides when the plural of Cushitic, singulars ending in the vowel i also attaches the p lural suffix a and the In Berber, plurality is marked by suffixing n for the masculine plural noun and in for the feminine plural noun. In Afar Saho of Cushitic, there is an opposition between singulative and collective; the latter is marked by the suffixe s to ta and tu In Southern Afar, there is also the suffix n with its variants, i.e. ntu nta with masculine and nto and nta with feminine nouns. The /t/ of these suffixes sometimes assimilates to the root final consonant (unfortunately, no exampl es were provided to illustrate this phonological change). Many Ethiopian languages including Tigrinya and Amharic use suffixation to mark plurality in nouns. For example, in their plural form, many Tigrinya nouns take a plural


91 suffix tat tat] (Buckley 1990: 75). In Amharic, the singulars are unmarked while plurals take the suffix (Kramer 2009: 182), e.g. [bet suffixes in Am haric an and at [ According to Simeone Vt r, some plural nouns appear with the feminine plural suffix ten Ractliffe (1998a: 95) states that the quality of the vowel in the suffix of the singular determines the shape of the plural. For instance, singulars of the shapes CVCC eet CVCC ajt or CVCC iit take the plural shapes CVCaC ten and CVCeC ten e.g. asked to pluralize nonce and loan words, use suffixation (specifically the suffix tV ) most often. Simeone Senelle (1997:388) and Lonnet (1985:54) claim that e xternal Some plurals with the suffix i come from the dual (Johnstone 1975:113). I observe that dual is no longer systematic in the langu age. Very few archaic nouns remain to be considered dual and are used occasionally by native speakers to indicate duality. Nouns bearing the dual marker i indicates plurality Kanakuru, a Chadic language, realize s nominal plural in three distinct ways: suffixation of ngin or its variants suffixation of ijan/ ujan and gemination of the second consonant plus attaching any of the above suffixes. Other Chadic languages such as Jegu and Kera suffix an t


92 by alterning the vowel in the singular into a a plural marker believed to be the proto form of plurality in Semitic (Ratcliffe 1992). The third less widespread mechanism for nominal plurality in Chadic is the attachment of the suffix Vn 5 Many Chadic languages also mark plurality by the morpheme a One of the most studied Chadic languages for its plural formation is Hausa whose various shapes of plurals are highly linked to root expansion (Newman 1990). This expansion is signaled by a range of phonological alternations including gemination of the final consonant in the root and internal stem change. According to Newman (1990), Hausa has twenty eight classes that can be classified into eight classes on the basis of shared phonological and tonal similarities. On the other hand, Schuh (1992) identifies two major classes of plurals in Hausa: final vowel change plurals and suffixed plural s. He further distinguishes four plural vowel endings ( ii, uu, oo and ai) and four suffixes which can be classified into two dimensions (uCa vs. aCi) or the suffixal consonant ( k or n ) which he illustrated in the following table: Table 3 1 Suffixation k in suffix n in suffix uCa tsa:nuka da:kuna aCi go:naki wa:sa ni (Schuh 1992: 3) The masculine plural noun in Egyptian is obtained by suffixing w as metathesis with other consonants, assimilation and elision. It may also be realized as u which surfaces short in closed syllables and long in ope n syllables. The Coptic plurals, on the other hand, relate to four classes: o class, i class, e class and u class 5 un to mark plurality.


93 (Vergote 1969:80). However, Ratcliffe ( 1992 ) listed only two classes for Coptic: e class and u class. The plural ending for the feminine n ouns is wt It is apparent that most Afro A siatic languages mark feminine gender with the suffixal morpheme t Internal Plural The second most widely used mechanism for forming plurality in Afro A siatic languages is internal plural formation. Under Inter nal Plural, a number of non concatenative morphological mechanisms are attested. For example, there exist broken plural, vowel opposition, reduplication, mapping onto templates, and affixation with internal change. Broken Plural Broken plural involves an internal stem change such as forming a typical iamb by lengthening the second syllable contained in the left foot of the plural form Diverse shapes of broken plurals such as ( CVCV: ) 6 CV: C, ( CVCV: ) CVC, ( CVCV: ) C, and CVCVC are attested in Arabic, e.g. [ qirdu n ] [ quruudun ] 1998). Levy (1971) attempts to relate the diverse shapes of the broken plurals to the distinct shape of the singulars. However, a given singular pattern may have two different plural forms which impos e s a challenge to making a definite statement about a direct relation between singular and plural shapes. Despite the fact that Arabic exhibits enormous variation in the shapes of broken plurals, Abu Mansour (1995: 326) proposes that the criterion for map ping singulars onto 6 Foot is enclosed between brackets.


94 concomitant phonological consequences which capture the huge diversity o f broken plural templates in Arabic. (2) Rules of forming broken plurals: 1. Deletion of the feminine suffix which is not part of the consonantal root of the singular form. { 2. / aa/ infixation whose locus is either after the second or third consonant of the stem. [plural] 3. Vowel r aising V [plural] The third phonological consequence for Arabic broken plurals is argued to be a dissimilation process (McCarthy and Prince 1990 a ). It is observed that when the singular form has the vowel /a/ in the las t syllable, the broken plural has /i/ instead in the same syllable. Moreover, the length of the final syllable is maintained when singulars with a long final syllable are mapped onto the plurals. The overall generalization that governs the formation of can onical broken plurals in Arabic is that the left edge of the singular form CVC or CVV is mapped onto a typical iambic foot CVCV V (McCarthy and Prince 1990a; McCarthy 2000). It has been observed that the dominant iamb constructed involves an LH foot, a sequ ence of a light syllable followed by a heavy one. In Hebrew, only two singular shapes CVCC and CVCC at take internal plural and even for these two singular shapes there are external plural forms too (Ratcliffe 1998). Egyptian and Coptic also display some internal change when forming plurals, In Cushitic languages, Zaborski considers the b The plural in these la nguages are characterized by tendencies (mostly templatic in


95 nature) different from those of the broken plural. Fo r example, the plural takes several patterns such as vowel shortening and vowel opposition. For example, in Beja, a long In the Bishari dialect, ablaut is observed, which is in fact a form of vocalic opposition. Many changes in the vowel quality are reported for Cushitic languages including (u:/u), (u:/i), (e/a and o) or (o:/a). In Tigrin ya, the largest number of nouns take s the broken plural which is indicated by two general mechanisms for Tigrinya nouns: infixation or association to a plural template. According to Buckley (1990) the most common exhibited pattern or template is composed of a quadriliteral root which is linked to a disyllabic template in the singular form (CVCCVC) and maps onto a tri 1990: 75). Moreover, the number of the consonants in the singular form forces a number of phonological processes to occur in order to satisfy the designated template. For example, spreading of the medial or final consonant of a triliteral singular occurs to fill the extra consonantal position in the quadriliteral plural template For example, when a singular form with a triliteral root associates to the template of four C slots, the medial co singular form has four root consonants, then spreading proceeds straightforwardly, whereby each C fills in the C slot, e. g. In Amharic, the formation of some plurals includes prefixation, ablaut and change in the prosodic template (i.e. broken plural). However, the nouns that take broken


96 plurals are (pp.185). Affixation with Internal Change Plural formation can display features of both sound and broken plurals. The fact that an affix must occur in conjunction with a stem c hange makes it hard to list these plurals under External or Internal types. To illustrate, the majority of plurals in Amharic t] In Tuareg, a Berber language which includes the dialects of Central Sahara Southern Algeria, Niger and Mali, a plural suffix is accompanied by vowel leng thening (e.g. ) (Ratcliffe 1992:462). In Hausa, suffixation or external pluralization encompasses different allomorphs or internal changes into the singular stem. Surprisingly, these allomorphic suffixes are not straightforw ardly attached to the end of the singular forms. Rather, each suffix imposes an internal change to the root to which it is attached. For instance, where a singular is mapped onto the plural, the consonant in the suffix may be a copy of the final consonant in the root or epenthetic. In some cases, the suffix forces the contiguous consonants of the root to split in order to satisfy templatic requirements. Hausa has three representative suffixal plurals unaa (ba aaCee aa on the shape of the singular forms (Rosenthall 1999:344). Looking at these plural


97 patterns make s one feel they are internal plural. However, s cholars of Hausa regard them as external plurals. It has been observed that the plural in Hausa favors attachment to an iambic base. Particularly, in the suffix aaCee a mix of external and internal plural can be detected due to the pressure of the surf ace form to realize an iambic foot (Rosenthall 1999). When a plural exhibits variations in its final shape, this can be due to violations of a definite set of requirements that govern s the final shape of plurality (prosodic requirements, for example). Besi des the phonological consequences observed when suffixation occurs in if the initial vowel in the root is heavy, the vowel in the reduplicated plurals surfaces short, e.g. [zo : be/ zobaba]. Moreover, in singulars with CVN syllables where N is homorganic to the following consonant, N is not to be thought as a separate consonant. Thus, such a plural will take a different mode of pluralization contrary to what is expected. Reduplication One of the most common mechanism s for forming Internal Plural in Afro A siatic languages is reduplication which is found to apply to a specific set of singulars. Ratcliffe lication as a feature of plural formation is most common in some Neo Aramaic dialects and in some both sets of languages is consistently suffixal reduplication of a single consonant with a predictably fixed vowel appearing between redu 1996: 298).


98 Cushitic also has reduplication which is indicated by a repetition of the last consonant of the singular either after the final vowel or with a cha nge in the final vowel, important to note that, in some Cushitic languages such as Beja, reduplication is limited to plural adject ives. Moreover, Ractliffe (19 92 and 1998) argues that reduplication should not be considered as a valid mechanism for forming plurality in Semitic and In Amharic, a number of nouns may take partial reduplication when they are also observes full or complete reduplication in some Chadic languages. radical singulars form their plurals by reduplicating the last consonant. CVC x singulars take the plural shape CC x VC x whereby t he V in the plural In some cases, the plural initially. There is a correlation between the quality of the vocalic suffix and the singular form in some plural patterns (Ratcliffe 1998). The vowel in the shape varies greatly and determines whether the suffix in the plural morpheme is In Kabyle, most bi radical singular s and some vowel final singulars add an extra consonant in their plural formation. More specifically, the second root consonant gets Mapping Singulars onto Templates Mapping singulars on to a specific plural template is a form of internal plural. It is attested in a number of Afro A siatic languages. For instance, in Amharic, quadri


99 where the final /t/ deletes b efore liquids. an initial vowel to this template. Ractliffe (1998) counts about 14 examples that behave in this way. The second common plural shape in this language is CeCewwe C, which might be a cognate to the Arabic broken plural shape CuCuuC. The feminine suffix in s as et eet (with its phonological variants iit and ajt ), oot aat and eh in words borrowed from Arabic and the original /aa/ vowel usually corresponds to masculine quadriliterals while the plural shapes CeCeeCeC or CeCooCeC r elate to feminine quadriliterals. Finally, singulars which have the CVCVVC shape map onto the patterns CeCiiC, CeCooC or CeCjooC. Other n the most prevalent shapes of plural in Jebb VCCVC or CVCVC, with the very rarely do /e/or /i/ surface in thes e shapes. /a/ appear s in a guttural environment only. Therefore, the most common shapes are third most common (1) the first shape involves Vb infixation, (2) the second shape is CaCaaCiC whi ch is common in southern Semitic /, /u/ or (it is derivable by alternation of the vowel in the final syllable :

PAGE 100

100 CuC) and (3) the third sh ape involves ablaut. Simeone Senelle (1997:388) states that the most common pat tern of plural for the tri consonantal verbs is CCV:C (a plural for many feminine singulars) and for the quadrilitral are CCV:CC and CCVCC. Simeone Senelle (1997:388) also iden tifies that some plural patterns correspond to Arabic plural of the plural (emphasis hers). Templatic plurals derived from geminated singulars : Geminated singular forms map onto distinct templates when they become plural In Arabic, McCarthy (1979) treate In Tigrinya, singulars whose media l consonantal slot is geminated take s a particular template. When forming plurals, the gemination is broken up by a vowel, e.g. explores another templatic plural pa ttern derived from a geminated singular in Tigrinya which involves the spreading of the last consonant instead of the usual medial radical singulars whose second consonant is geminated and which take the shape CVC x C x take the plural shape CVC x x second consonant in the plural

PAGE 101

101 However, this pattern is not very common across other Modern South Arabian languages (Belova 2009). In Chadic languages, the templatic behavior of geminated singulars is c onsidered to be a reduced form of reduplication template taken by a plural derived from a geminated singular, there is also a weakening of intervocalic consonants (Newman 1990: 47). In Kanakuru, for examp le, the geminate may simplify to a single consonant. Infixation evalent plural shapes involves Vb infixation (Johnstone 1981; Simeone Se nelle 1997; Ratcliffe 1998a & b). Ractliffe (1998) considers the shape CVC Vb CVC to have an infix Vb ( or ab ) between the second and third radical. Most of the plurals with the Vb infixation are derived from quadri radical singulars. However, a few bi radical singul ars are also observed to infix Vb or only b when the y become plurals. The locus of Vb infixation is consistent and is observed to occupy the second syllable of the plural form. Newman (1990: 81) recorded some pluractional patterns in Sura of Chadic to have a final / p/ affixation. He argues that this pattern is due to either contact with neighboring non Chadic languages or the /p/ being a r eflex of *t by a morphologicall y restricted dissimilation rule ( e.g. ) Vocalic Opposition Internal vocalic mutation (be it ablaut or apophony) has been attested in many Afro A siatic languages. One interesting peculiarity of Cushitic i s the contrast in suffix es of the singulars and plurals. For example, the singular form [alum

PAGE 102

102 bears one of the suffixes that may mark plural This suffix changes and a different plural suffix like a marks plural Thus, the pl ural for [alum to] is [alu:m to vowel quality are reported for Cushitic including u:/u, u:/i, e/a and o or o:/ a (Zaborski 1986). Quite a large number of tri radical and quadri vocalic opposition when mapped onto their plurals. Singulars with front unrounded vowels form their plurals with back rounded vowels. This mechanism for plural formation Senelle (1997). Pluractionals in Chadic languages exhibit ablaut as one of the common mechanisms for forming plurality. For example, in Bachama plurality is marked by vowel lowering (Newman 1990:72). Another pattern, in Saya, involves vowel lengthening but without a change in quality. Vergote (1969) observes some vocalic changes whereby /a/ becomes /i/ in Egyptian and Coptic languages. In Kabyle (Northern Algeria), a Berber language, masculine singulars with a high vowel (i or u) or zero vowel before the last consonant show the vowel a before the last consonant in the plural form (e.g. [amshish] u this type of plural as internal a plural. Singulars with a in the final syllable exhibit a double vocalic alternation ( e.g.

PAGE 103

103 A siatic Languages ural. This diversity can sometimes be attributed to the shape of the singular form (whether the singular is bi radical, tri radical, quadri radical, or weak). Moreover, Ractliffe (1998) argues that the diversity can also be attributed to the vowel in the s uffix of the singular. Besides the plurals that reveal regularity and can systematically be accounted for, such as boys, girls and babies. The se do not follow a certain pattern and the forms of both the singulars and plurals are highly unrelated. Ractliffe (1992) included a long list of languages that mark plurality by suppletion including Berber, Chadic and Cushtic languages. In spite of the f morphology, it shares many interesting characteristics with other Afro A siatic languages. A siatic langua g es: 1. A plurality more often by imposing an internal change in the stem than by suffixation. This is manifested through the large and highly diverse number of non concatenat ive Ratcliffe (1998) identifies three singular shapes out of five which prefer internal to external plural. He has surveyed similar tendencies in Arabic, Aramaic, Old South Arabian languages, and Tigre. The di stribution of internal plural concentrates more for the tri consonantal singulars taking the shapes CVCC, CVCC at and the quad ric consonantal singular CVCCVC. 2. In the formation of plurality, languages tend to favor a certain pattern over other patterns. the suffixal plural ooCii is the most productive plural pattern.

PAGE 104

104 3. A singular form in Senelle 1997). Hausa, a Chadic language, is also reported to have multiple plurals for a singular form, each with potentially distinct allomorphy and varying tonal melodies (Newman, 1990 and have multiple plurals. Arabic, too, exhibits this fea ture of multiplicity of plurals. 4. I t is not uncommon that some of the mechanisms for plurality are productive and are observed to be employed more often by the majority of Afro A siatic languages. Other mechanisms are archaic and are no longer preferred as plural mechanisms. t he most productive mechanisms include suffixation vowel opposition and mapping onto specific templat es. These patterns are widely attested in Modern South Arabian language s, Chadic and Cushtic languages. 5. B y and large, there is a preference to map singulars with a geminated root consonant onto specific templates Templates are defined as skeletal shape s that are unspecified for segments but have intercalated consonants and vowels. I nfixation of certain consonants is also attested across the board 6. Johnstone (1975: 113) and Simeone Sene lle (1997:389) observe that some plurals in i which is a dual marker. However, they are considered to be plurals in the language. Newman (1990) observes that three or four branches of Chadic exhibit plurals formed by a final i whi ch bea rs a high tone most of the time. 7. A lthough Ratcliffe (1992) eschews that reduplication exhibited widely in the formation should be taken as a result of templatic exp ansion since it occurs mostly in bi consonantal and weak roots, this phenomenon is very prevalent in a number of Chadic languages, Dahalo of the Chushtic family (Zaborski 1986 8. oken plural is the prevalent mechanism for forming plurals in Afro A siatic languages and should be reconstructed as the main plural marker for proto plurals exhibit massive internal changes to the extent that suffixation too is accompanied by internal changes. 9. Zaborski (1976) maintains that Afro Asiatic la nguages have portmanteau plural morphemes. An obvious example would be the Arabic external plural morpheme which takes u:n for the nominative case and i:n for accusative. Socotri, a Modern South Arabian language, marks plural by i:n and 10 Abd R abbo (1990) explores the phonology of vowels in the broken plurals and shows that the variations exhibited in the first and last syllables of various patterns of broken plurals happen as a result of avoiding homophony. Where the productive /aa/ plural is blocked, he shows the possibility of the emergence of similar words but with different meanings in the

PAGE 105

105 multiple plurals are observed to map onto specific templates rather than taking the Vb infix or reduplicating a base consonant. This may happen as a result of homophony avoidance though a thorough scrutiny must be done to justify this claim 11. t mechanism s for forming plurality in the language. I argue that ablaut is, by and la rge, a morphophonological process in which plural forms must observe a change from their morphologically related pair, the singulars. The change can be driven by anti fait hfulness constraints which stipulate that related forms must be different as they belong to different classes. Homophony avoidance can be a reason. Summary of Chapter 3 In conclusion, plural formation in Afro A siatic languages is an intriguing morphologica l phenomenon. It encompasses numerous mechanisms unattested in Germanic languages such as broken plural which involve s some sort of stem change, reduplication, infixation, vocalic opposition and mapping onto plural templates. Each and e very non concatenat ive process outline d above comes with certain phonological changes. In the final section of Chapter 3 I listed some crucial observations that outline where Asiatic languages widely attested mechanisms for forming plurality with these languages. However, it also has a unique plural formation pattern which is the infixation of Vb Chapter 3 outlined the most common non concatenative morphological mechanisms used to indicate plurality in a number of well studied Afro A sia tic languages. It also described the phonological processes that occur conc omitantly with these mechanisms

PAGE 106

106 CHAPTER 4 APPROACHES TO NON CONCATENATIVE MORPHOLOGY In non concatenative morphology, morphological oppositions are most often Thus, a word cannot be divided into smallest contiguous constituents each designating a particular meaning. Rather, the derived meaning is encoded via other non concatenative morphological operations such as infixation, truncation, ablaut, reduplication, mobile morphology and root and pattern (templatic) morphology. In Ar abic, in particular, the meaning is encoded in the template or canonical shape the word takes. To address the complications involved in the formation of words in non concatenative morphology, many morphological approaches have been proposed. These include Autosegmental and Templatic Approaches, Prosodic Morphology and Optimality Theory. The analysis proposed by the earliest models such as Templatic Morphology entails many shortcomings and warrants the exploration of other later models such as Prosodic Mor phology and Optimality Theory to tackle such problems. Below, I will explore the major approaches to analyzing non concatenative morphology. I will specifically address two pre Optimality Theory approaches and then outline the theoretical assumptions of t he Optimality Theory approach. The discussion will sketch the advantages and shortcomings of each approach with respect to other frameworks and discuss their relative success in accounting for non concatenative morphological processes. The last section wi ll delineate the assumptions made by Generalized Template Theory which translated the superfluous templatic effects into universal constraints.

PAGE 107

107 Pre Optimality Theory Appro a ches Autosegmental and Templatic Approaches The stem in non concatenative languages has three constituents that cannot be isolated from each other (McCarthy and Prince 1990b:2), the templatic shape onto which the root consonants are mapped and finally the vowels which indicat e voice and aspect. John McCarthy (1979 and 1981) uses the principles of Autosegmental Phonology which was first proposed by L eben (1973) and Goldsmith (1973 and 1976) for tone and vowel harmony phonological systems in order to account for the three non i solable parts of non concatenative morphology. In Autosegmental or Templatic Morphology, the phonemic segments separated by 1992:32). Rather morphological analysis is given by another simultaneous level of representation (McCarthy 1979: 221). In other words, McCarthy proposes, following proposals of Autosegmental P honology, that words are represented by three tiers: a skeletal tier (also known as the timing tier) which is segmentally unspecified (or uninterpretable, of bare C onson ants and V owels notated as CV The other two tiers have the consonants of the root and vocalic sequences which also occur as independent autosegments and each belongs to separate tiers. Then, the consonants and vowels (melodic elements) associate with the timing tier in accordance with the Universal Association Convention of phonology. The latter stipulates that melodic elements associate with the skeletal tier one to one, left to right. These tiers represent the notion

PAGE 108

108 et of feature matrices dominated by a single To illustrate, the Arabic word [kaa is autosegmentally represented as : (1) k t b {write, Root } CVVCVC {participle, S keletal or Timing Tier } a i {active, Vocalic Melody } [kaatib] is represented by three tiers: the CVVCVC or skeletal tier, the consonantal and vocalic melodies tiers. Though these tiers occur in different planes they are linked by association lines one to one and left to right, except long vowels. For a string of segments to be pronounceable, McCarthy (1986) proposes the convention Tier Conflation which linearizes the morphological tiers into a single tier. Thus, the combi (2) {morpheme} C V V C V C k a t i b {tier conflation} Tier Conflation also operates when a phonolo gical string is composed of a stem and affixes. Thus, it linearizes the stem melodies first and then folds in the remaining affixes. Association lines are also subject to two conditions: No Crossing (first proposed by Leben 1973) which stipulates that asso ciation lines must not cross and Obligatory

PAGE 109

109 Contour Principle (OCP, henceforth) which forbids identical adjacent segments. McCarthy (1979, 1981 and 1986) extends the OCP to account for geminates and double verbs in non concatentive morphology. His formali autosegmental tier, adjacent identical 1979:238). So, in stems with two identical segments, the melody is represented by just one of the identical segments which then spreads to two slots in the skel etal tier. As illustrated above, the vocalic tier for [kaatib] i ncludes one /a/ which associates to two vocalic slots. The Templatic Approach can successfully address a number of non concatenative morphological tendencies especially those pertinent to roo t and pattern morphology. It provides an empirical argument for a language game of Bedouin where a free mutation can only happen to the consonantal root but not to the template nor to non root consonants. The vocalism may vary depending on the neighboring segments (McCarthy 1981:380). This game supports the assumption made by Templ atic Morphology that the consonantal root is a single unit at the level of representation and operates at a different sphere. Any rule could apply to it without affecting the other two tiers. Moreover, stray erasure which stipulates that unassociated skel etal or melodic elements can be elided supports the separation of these three tiers as it could apply to individual tiers without erasing elements from others McCarthy (1981) argues that in deriving Arabic verbs, the rule will have to have access to the root from the vowel quality and from the canonical distribution of consonants and Assimilation rules in Arabic, Akkadian and Hebrew verbal

PAGE 110

110 morphologies support the discontinuous nature of the morpheme in Templatic Morphology. These rules do not target the consonantal root even when their context is met. Rather, they target an infix t (in the eighth binyan of Arabic verbal paradigm but the passive and iterative t in Hebrew and Akkadian). Moreover, Akkadian has a nominal prefix ma which dissimilates to na i f there is a labial root consonant. Only consonants in the root trigger this dissimilation as it fails to apply before a labial stem vowel (McCarthy 1981: 382). This rule provides support for having separate tiers to concatenative morphological systems as it refer s directly to the non concatenative root Geminated roots, reduplication patterns and double verbs also dem onstrate the success of Templatic Morphology. Moreover the Templatic Theory accounts for a number of rules governing the co not occur in the same consonantal root. Such tendencies prove that the mor pheme notion is indeed relevant to the consonantal root. The same conclusion holds true for vocalism. McCarthy observes that there is no linguistic form in Arabic (except some borrowed words) that has the vowels /i/ and /u/ in the same tier and there is no verb that begins with /i/. Couched under the umbrella of Templatic Morphology is melodic transfer (Hammond 1988; Bat El 1994a) which offers an explanation to the behavior of clusters in denominal verbs of Hebrew and Arabic broken plurals. In melodic tran sfer, segmental materials of the stem are copied or transferred to the target derived template without a change in their positions. To illustrate, in Arabic broken plurals the linear order of the singular consonants are transferred to the broken plural tem plate. Moreover, the length

PAGE 111

111 or weight of the final syllable of the singular is copied without shortening or lengthening. As for Hebrew d enominal verbs, when the nouns from which these verbs are derived have clusters, the ver bs also surface with clusters. Despite the crucial breakthroughs Templatic Morphology makes to address non concatenative morphology, it is faced with many problems which the newer proposals of Prosodic Morphology did not escape too One of the major problems with the Templatic Theory t o non and Prince 1993: 18). It does not provide access to information already there in the Universal Gramm ar such as the elements of prosody ( syllables, feet and prosodic word s ) Moreover, the peculiar properties attributed to reduplicative and Templatic Morphology are independently motivated in Prosodic Morphology and can characterize phonological processes, stress and versification (McCarthy and Prince 1990b:2). Templatic Morphology also has ambiguous evidence for segment sized skeletal units. Prosodic Morphology relates words to minimality requirements and explains non concatenative morphological behaviors s uch as insertion of consonants, compensatory lengthening and the like. On the other hand, t he Theory of Templatic Morphology is cumbersome in its nature and always forces some sort of re definition and reference to meaningless templates. To illustrate, whe never there is a need to refer to a specific binyan (template) or verbal paradigm, we specify it using templates that in themselves offer no coherent explanation to the kind of phonological or morphological processes involved in their creation. In criticiz ing Templatic Morphology, McCarthy and Prince

PAGE 112

112 objects (in our case templates) are attached to other objects without recourse to conditions governing such attachment. Finally Prosodic Morphology is more restrictive as it relies on the units of prosody which are independently motivated by Universal Grammar. The template in Templatic Morphology can be expanded without limits. So, template satisfaction in Templatic Morphology is in fact no more than prosodic parsing. Prosodic Theory of Non Concatenative Morphology Prosod ic Morphology (McCarthy 1981; McCarthy and Prince 19 90:209) was proposed to offer access to already existing tools (units of prosody) in the language instead of reliance on a proliferation of unmotivated mechanisms. It basically has three fundamental theses. First and foremost, it establishes the need to define CV templates in terms of the authentic units of prosody which are moras syllables, feet and prosodic wo rds. These are ordered in a hierarchy from the smallest atom making up a syllable to the prosodic word holding all elements together. The phonological word (a domain for stress assignment) contains at least one foot. A foot contains at least one stressed s yllable while a syllable can be light CV with one mora or h eavy CVV and CVC with two moras To illustrate, in reduplication patterns, if the reduplicated element is a CV shape, then Prosodic Morphology does not describe it in terms of the template CV. Rath er, it stipulates that the redupilcant is a syllable length. Below, I show the elements of prosodic hierarchy: (3) Prosodic Hierarchy: Phonological Word Ft Foot Syllable Mora

PAGE 113

113 ( 4) represent s the phonological word writer in prosodic morphological terms: (4) k t b a i The final consonant at the right edge of the above form is extrametrical and does not participate in the prosody of the language. This observation is made by McCarthy and Prince (1990b: 15 17) who support this belief on the basis of the behavior of linguistic forms in Arabic such as str ess assignment Extrametricality is a breakthrough in the analysis of templatic morphologies and is a well motivated device provided by Prosodic Morphology. It is confined to the edges of words only. However, the unifying account spelled out by McCarthy an d Prince (pp.15) is that in non concatenative morphological systems an initial extrametrical consonant has the properties of syllable final position (i.e. moraic) while the final extrametrical consonant is nonmoraic. Thus, it bears the properties of syllab le initial consonants. The second fundamental thesis of Prosodic Morphology is a Template Satisfaction Condition which stipulates that satisfaction of templatic constraints is mandatory and determined by the principles of Prosody, both Universal and langu age specific (McCarthy and Prince 1990:3). The third proposal relates to the domain to which a morphological process applies which needs to be delimited by prosody. In other words, it is assumed that

PAGE 114

114 morphological operations target units of prosody and a pply to them. This is known as Prosodic Circumscription and will be outlined in a separate section. In the 198 0s, Prosodic Morphology was proposed and it succeeded in explaining a huge range of non concatenative m orphological tendencies. McCarthy and Pri nce (1993a) reveal that Prosodic Morphology has successfully discovered independent and general principles that govern the linguis tic properties of reduplication, root and pattern systems, circumscription, truncation, and the like. Prosodic Morphology expr esses generalizations which cannot be expressed in purely templatic terms. It makes use of the information and principles in the grammar and avoids the non concatenative morphological process. They also argue that once analysis starts seeing the higher prosodic units, then reference to what the template CV do es is not important or relevant. T his is of course a very desirable consequence. P rosodic Morpholo gy offers well motivated tools for the description of the maximal size of affixes and weight requirements. Using prosodic units, an affix is observed to be no longer than a syllable. Different languages may also place restrictions on the lightness and heav iness of affixes. Such restrictions are carried over to reduplicants. If we limit ourselves to Templatic Morphology, we will not be able to describe theoretically and convincingly the size and weight of affixes in any non concatenative langu age. Another cr ucial prediction made under Prosodic Morphology is the ability to describe stress patterns and the rules governing the assignments of iambic or trochaic stress by reference to syllables and feet, the units of prosody.

PAGE 115

115 Prosodic Morphology offers cogent ex planation to reduplication patterns. For instance, in Axininca Campa, consonant initial roots reduplicate fully whereas vowel initial roots exhibit divergence from this pattern. If reference is made to the CV pattern of the reduplicant, then the truth of t his type of reduplication is obscured. However, by referring to prosody, one can see that the suffixed reduplicant in this language is consistently consonant initial With this particular example, we are able to divorce the effect of CV templates and util ize the units of prosody readily supplied by Universal Grammar. Now, let us move one step higher and scrutinize the elegance of analyzing Axininca Campa reduplication within a theory like Optimality Theory. By reference to a well motivated syllabic constra int and a faithfulness constraint against deletion, we are able to offer a straightforward analysis to an issue otherwise appearing to be cumbersome. The tenets of Optimality Theory will be discussed in a later section. Also, evidence is offered for how th e satisfaction of well defined templatic and prosodic templates is better translated into violable and competing constraints in Optimality Theory Prosodic circumscription: The third crucial thesis Prosodic Morphology makes is Prosodic Circumscripti on (McC arthy and Prince 1990; McCarthy 2000). The core assumption of Prosodic Circumscription is to delimit the application of non concatenative morphological processes to prosodic constituents like moras syllables, and feet. As formalized by McCarthy and Prince Prosodic Circumscription depends on the edge E of the base B (pp.226). McCarthy and Prince (1990) identify two types of

PAGE 116

116 Prosodic Circumscription that address a wide ra nge of non concatenative morphological operations: Positive and Negative. If the morphological operation targets a prosod ic element at the right or left of a word, then it is referred to as positive prosodic circumscription. However, when the morphological operation applies to the rest of the word, then we have a negative prosodic circumscription. In spite of the fact that Prosodic Circumscription correctly addresses classical non concatenative examples it is not as powerful as Optimality Theory. For examp le, Prosodic Circumscription successfully analyzes linguistic forms and then a morphological operation (in this case reduplication) applies to it. After that, the rest of the word is concatenate d to the changed foot. However, such an analysis is cumbersome and a short cut analysis is available in Optimality Theory. McCarthy (2000: 152 153) argues that the Optimality Theoretic approach based on prosodic faithfulness enjoys conceptual advantages o ver operational circumscription. According to McCarthy faithfulness constraints are independently motivated since these constraints are supplied by Universal Grammar. It reduces dramatically the specific devices of Prosodic Morphology like circumscription templates, or reduplicative copying. Most importantly, prosodic faithfulness eliminates infixation It regards all affixes as either prefixes or suffixes based on their distance from the edge of the root. This dista nce can be translated into well motivat ed constraints in Optimality Theory. Optimality Theory Optimality Theory is a constraint based approach (Prince and Smolensky 1993 / 2004 ; McCarthy and Prince 1993a & b) whose foremost premise is that surface forms are evaluated by a set of conflicting const raints and the optimal form or the actual output is the one that minimally violates constraints. In Optimality theory, there is a

PAGE 117

117 conflict between two major families of constraints: Markedness and Faithfulness constraints. Markendness constraints evaluate how common cross linguistically, easy to perceive or articulate and less marked a linguistic form is whereas Faithfulness constraints monitor identity between underlying forms and actual forms. I will present the framework of Optimality Theory and discu ss its theoretical assumptions. I will then introduce Correspondence Theory and outline the premises of a family of Correspondence Theory constraints : Output Output Correspondence. Finally, I will talk about alignment constraints. Optimality Theory Framew ork Optimality Theory assumes that the actual linguistic form stems from a competition between a set of conflicting violable constraints through a fixed ranking. Constraints are ranked with respect to each other, entailing that the output form must violate the lowest and less important constraints possible in order to be selected as optimal. Under the umbrella of Optimality Theory, Universal Grammar supplies universal constraints which exist in all languages but differ on whether they are active or passive and on the way they are ranked. In other words, constraints are ranked on a language specific basis. These constraints belong to two competing families of constraints: Markedness and Faithfulness. While Markendess constraints penalize marked structures in the surface forms, Faithfulness constraints strive to maintain absolute identity between the underlying forms and surface forms. When the output form exhibits a change from its underlying form, then Markedness constraints take precedence over Faithfulness constraints. On the other hand, when a marked structure is retained in the output form in order to be faithful to the underlying form, then Faithfulness constraints prevail over Markedness constraints.

PAGE 118

118 In any given language, t he conflict between Markedne ss and Faithfulness constrains is settled through a fixed ranking which gives priority to either Markedness or Faithfulness. In Optimality Theory, the Generator is in charge of generating potential and unlimited candidates from a specific underlying form. Thus, it creates an ambience for competition; these potential outputs compete with the actual output (also known as optimal ) and are doomed because they violate higher ranking constraints by the function of Evaluator. The Evaluator checks each and every ca ndidate against the ranked set of constraints. The most harmonic output which violates constraints very minimally or satisfies high ranking constraints at the expense of violating the lowest ones is selected as the winner. Below, I illustrate the mapping of input to output in Optimality Theory grammar as proposed by Kager (1999:8). I add Gen and Eval at t he bottom of the representation. In the representation below from the input form. The Evaluator then evaluates each candidate against the ranked constraints C 1 C 2 C n The candidate which wins the competitio n exhibits satisfaction to the high ranked constraints at the left of the representation and thus is selected as the actual output form. (5) Candidate a Candidate b Input Candidate c Output Candidate d Gen Eval C 1 C 2 C n

PAGE 119

119 If constraint C 1 outranks constraint C 2 then the optimal output form has to be the most harmonic amongst other suboptimal outputs in that it exhibits the least serious violations to the constraints C 1 and C 2 It can violate constraint C 2 at the expense of being faithful to constraint C 1 since C 1 dominates C 2 In a tableau of ranking, C 1 must be to the left while C 2 must be to the right side and a solid line is drawn between them, indicating a strict do mination. The opposite is true if constraint C 2 dominates C 1 In T ableau [1], [2] and [3] when the constraint C 1 prevail s over the constr aint C 2 then each candidate violating C 1 will receive an asterisk and a fatal violation mark indicated by (!). The o ptimal output is not the one that does not violate any constraint. On the contrary, it is the one that is most harmonic in that it exhibits less serious and minimal violations among other candidates. It obeys C 1 at the expense of violating C 2 When both ca ndidates violate C 1 the optimal output must exhibit fewer violations. Thus, the doomed candidate should exhibit more violations to C 1 than the optimal output. However, when both candidates obey C 1 then the optimal output must be more harmonic by exhibiti ng fewer violations to C 2 If the ranking between constraints is established, a solid line between the constraints is drawn in a tableau of constraints. However, a dotted line between the constraints indicates that a definite ranking cannot be established and the constraints are in fact equally ranked. A pointy finger is placed before the winning candidate. The tableaux below illustrate how Optimality Theory functions: Tableau (1) C 1 C 2 Candidates C 1 C 2 a. Candidate A b. Candidate B *!

PAGE 120

120 Tableau (2) C 1 C 2 Candidates C 1 C 2 a. Candidate A b. Candidate B **! Tableau (3) C 1 and C 2 are unrankable with respect to each other Candidates C 1 C 2 a. Candidate A b. Candidate B **! The optimal output (Candi date (a) above) is more harmonic as it exhib its less serious violations Theoretical Ass umptions of Optimality Theory There are five crucial principles that govern the framework of Optimality Theory. Universality entails that Universal Grammar is responsible for supplying the constraints which all the languages in the world have in their reper toire However, as we move from language to language, constraints might be active in one language but dormant in another. They also may be highly adhered to in one language but freely violable in another. Ranking assigns priority or preference to constrain ts with respect to each other. Within a given language, s ome constraints are more important than others and must take precedence over others. Ranking determines how evaluation of constraints proceeds and on what basis the optimal output is selected. Violab ility holds that nothing is perfec t; all constraints are potentially violable. The suboptimal output forms along with the opti mal one violate constraints in one way or another. However, the ranking of the competing constraints is of a paramount importance and it definitely matters if a candidate wants to be optimal or doomed. In order to be optimal, the candidate can violate the low ranked constraints but exhibit s full respect to high ranked constraints. It must also, at all costs, be more harmonic and onl y minimally violate constraints.

PAGE 121

121 Inclusiveness dictates that all candidates are admitted in the competition and no one is excluded. In Optimality Theory, any candidate is a potential output and evaluation should decide the winner. Parallelism solves the c umbersome serial derivations proposed in earlier model. Thus, parallelism ensures that all changes whether in the phonological, morphological or prosodic structures of a linguistic form should be applied simultaneously. Evaluation checks candidates against all these diverse types of changes in a parallel way. Markedness v ersus Faithfulness Constraints Prince and Smolensky ( 1993 /2004 ) argue that the grammar of any language is reduced to a set of competing universal constraints. The resolution in the competit ion is made through the ranking which stipulates t hat one constraint is more important than another. In Optimality Theory, there are two major families of constraints: Markedness and Faithfulness. This section will elaborate more on the function of these c onstraints. Markendess constraints detect marked linguistic structures in the output forms ensuring that unusual or less c ommon features or segments rarely or less often surface in the output form. The sets of constraints adm itted in this family are quite large in scope since the constraints that evaluate features, segments, syllables, feet, and stress are many and the list may go on and on. Let us take a simple example that pertains to syllable structures. The less marked syllable structure has an onset with a nucleus and takes the shape CV. However, languages may admit marked structures. For example, they may permit clusters syllable initially or finally CCVCC. They may also have a sequence of vowels in a syllable CVV. Moreover, they may also have a syll able without an onset VC or a syllable without a vocalic nucleu s but with a syllabic consonant. In Optimality Theory, syllabic markedness constraints penalize any deviance from the

PAGE 122

122 usual unmarked syllable CV. So, a constraint such as O NSET will evaluate ev ery syllable in the surface form for an onset. H IATUS penalizes syllables with a sequence of two or more vowels. C OMPLEX does not prefer clusters syllable marginally and so on. In N O V which bans vowels i n the output forms. This constraint abbreviates many phonotactic constraints and interacts with a number of phonological and prosodic constraints in the language to yield that a plural form has a single vowel. Correspondence constraints (encompassing Fai thfulness, Identity, I nput Output and Output Output constraints) on the other hand, monitor identity between compared forms. Thus, there are constraints which militate against insertion or deletion of features or segments. Furthermore there are constrain ts which ban featural change. For instance, in M AX [+back] SP 1 is crucial. It penalizes deletion of the feature [+back] from the output plural forms. In recent Optimality Theory, prosodic elements such as mora s syllables, feet and prosodic words are also incorporated within Faithfulness constraints. Thus, constraints against the deletion or insertion of mora s are proposed ( c.f. McCarthy and Prince 1995b; Alderete et al 1 999; Kager 1999 ; Crowhust 2004 to list a few) Positional Fa ithfulness Constraints One family of faithfulness constraints is the positional faithfulness set of constraints (Be ckman 1998) which show s that certain positions in a word hold priority over other positions and thus they maintain phonological distinctiven ess. The fact that these positions keep identity or minimally alter identity suggest s their prominence over 1 SP stands for the correspondence relation between the Singular and Plural.

PAGE 123

123 other positions in a word. Beckman (1998) sketches some prominent positions that are more salient in any linguistic form (Table A 4.) Table The p rivileged and non privile ge d p ositions a. Privileged Positions b. Non Privileged Positions Root initial syllables Non initial syllables Stressed syllables Unstressed syllables Syllable onsets Syllable codas Roots Affixes, clitics, funct ion words Long vowels Short vowels (Beckman 1998:1) Beckman state phonological processes and resist the application of otherwise regular alternati (Beckman 1998:52). She further argues that the phonological privilege d status of the initial syllables results from high ranking positional faithfulness constraints. To illustrate, positional neutralization of vocalic contrasts outside the initial syl lable is common in language families that have vowel harmony such as in Turkic, Tungusic, Mongolian, Finno Ugric and Bantu. Initial sylla bles have vowels that come from the whole set of the vocalic inventory of these languages while non initial syllables a re usually a sub set of the whole vowel inventory and are less marked in terms of the available vowel contrasts Furthermore, consonantal contrasts are also restricted to syllable initial positions as in Tamil. Beckman provides a whole set of consonantal r anges in many of the world languages, supporting the idea that initial syllables phonologically play a crucial role. Beckman also discusses that initial syllables play a n important role in the domains of lexical access, word recognition and speech producti on. She listed a number of

PAGE 124

124 psycholinguistic studies that prove that initial syllables are prominent positions in any lexical entry. when plurals with ablaut are formed, the initial syllables maintain a vocalic identity with the root initial sy initial syllable of both the singular and plural forms are identical when plurality is marked in the following singulars and plurals below: (6 ) Plurals with ablaut a. t t dresses b. fag ri fag ru Bedouins c. mo mo flesh of backs Beckman proposes the following schema to address Positional Faithfulness: (7) I DENT Position (F) [F] (Beckman 1998:8) plurals, I adopt the constraint I DENT initial [+back] SP that basically monitor s identit y of the feature [+back] in the initial syllables of both the singular and plural forms. Correspondence Theory As the scope of the investigated linguistic problems expands, new tools are introduced either to fill gaps in the theory or to complement it. For instance, McCarthy and Prince (1995) introduce d Correspondence Theory to cover old Faithfulness constraints (I nput Output constraints ) as well as the new consequences of various Output Output types and to account for identity features in reduplication.

PAGE 125

125 Correspondence Theory then catches on to offer a unified analysis for other sets of non concatenative morphologic al tendencies. It admits a broader range of faithfulness constraints including those pertinent to prosodic constituents such as faithfulness to moras, syllables, feet and the like. Correspondence Theory posits that elements (input output strings, and input reduplicant strings) stand in correspondence. It entails that various phonological strings such as features, segments, syllables or feet can stand in a correspondent relation. The schema of the theory is outlined below: (8) Correspondence Theory Given tw o strings S 1 and S 2 correspondence is a relation R from the elements of S 1 to those of S 2 1 2 ) are referred (McCarthy and Prince 1995:262) One of the major families of constraints which has emerged from Correspondence Theory is Output Output Correspondence. I will outline the tenets of this type of correspondence and review th e major works which illustrate its success. I will also illustrate how Jebb Output output correspondence: The success of Correspondence Theory in addressing shared elements in strings that stand in correspondence inspires Benua (1997) to extend it to correspondence r elations between output forms. The Output Output Correspondence takes correspondence one step ahead by arguing that not only underl ying and surface forms or reduplicants and bases stand in co rrespondence but also morphologically related forms may also sta nd in correspondence. Undoubtedly, words in a morphological paradigm share a number of salient linguistic features. In non concatenative morphology, the consonantal roots along with their linear order are shared all the way down the paradigm. In such a par adigm,

PAGE 126

126 morphologically related words are surface forms and within themselves could serve as related or corresponding forms to each other. Benua (1997) proposes constraints that regulate such output relations. Her constraints function normally as any other set of faithfulness constraints in Optimality Theor y; they interact directly with phono constraints or markendness constraints whose goal is to evaluate marked structures in linguistic form s Benua, who laid down the whole premise which Output Output Corre spondence depends on states in support of her argument that sensitive to morphology because phonological identity relations hold over paradigmatically related words. concatenative morphological derivation s such as internal plural, redupli cation, truncation, affixation and others are all subsumed under Output Output Correspondence. McCarthy (2000) regards the relation between singulars and their broken plurals in Arabic to be an Output Output correspondenc e. Many affinities between the singular forms and broken plurals are salient and support such a correspondent relation. To illustrate, in their discussion of foot and word in prosodic morphology, McCarthy and Prince (1990 a ) highlight some problems invoked by reliance on the consonantal root as an input for total identity of the last syllable weigh t of both the singulars and plurals provides empirical evidence that fully supports adoption of Output Output Correspondence. Furthermore, the consonants along with their linear order are preserved in both the singulars and plurals.

PAGE 127

127 Other evidence comes fr om Hebrew. Ussi shkin (1999), in his analysis of denominal verbs in Hebrew, argues for the success of Output Output Correspondence. Denominal verbs stand in correspondence with the nouns from which they are derived. They retain the consonant clusters of the se nouns and there is a direct relation between the vowel in the nouns and the final shape of the denominal verbs. These theoy oriented pieces of evidence support an Output Output correspondence and exempt any direct relations between denominal verbs and their bare stems. Gafos (2003), who provides a cogent Optimality Theoretic analysis to some of the distinct traits of Arabic verbal morphology, takes Be As he focuses on the paradigmatic consequences of the verbal paradigm ( particularly the asymmetries in the distribution of geminates and identical segments in doubled verbs in Arabic ) he claims that stems in Arabic are also realized in the context of paradigms. The fact that Arabic verbal stems ban initial geminates makes it reasonable to explore the extent to which stem properties, patterns in the lexicon and alternations, derive from this fact rather than being idiosyncratic To relate this quote to doubled verbs in Arabic, Gafos maintains that there is a presen ce of two distinct stem realizations that would imply a violation of some Output Output correspondence whose stipulation requires identity between forms in the perfect paradigm (pp.325). Without Output Output Correspondence, resemblance between output for ms in a paradigm results only from sharing part of the input In Semitic, in particular, this resemblance results from sharing the consonantal root. In the formation of Jebb plurals noun plurals are derived from their singular forms and not solely from the consonantal root. The preservation of m arked structures in the derivation of plurals can

PAGE 128

128 easily be attributed to the output singular. The following representations show the formation of ablaut and Vb infixation) R epresentation (9) below shows the Underlying Representation (UR) /n V V r / the singular form [na rer] and the plural [na The last syllable s of both the singular and plural forms involve ablaut, and reveals how an Output Output correspondence serves to derive plurals from singulars. (9) /n V V r / UR /n V V r+ ablaut/ pl I [na rer ] O [na O O In representation (10) below, the UR / z cannot serve as an input to the plural [ bakery the singular [ ]. If I assume that the plural form [ ] is derived from the consonantal root / VbV z /, it will be hard to explain {ma} a shared sequence of segments in the output singular and plural. It is importance to note that [ma baz] is borrowed from Arabic. This exp lains where the sequence {ma} comes from in the ( 10 ) / z / UR / z + Vb / pl I [ m a baz ] O [ma ] O O

PAGE 129

129 Alignmen t Constraints Another crucial operative set of constraints in Optimality Theory is alignment constraints which demand certain edges in linguistic forms to align with some other edges of categories or constituents. For example, they may require the right ed ge of every syllable in a linguistic form to coincide with a certain morphological constituent. The general formalism of the alignment constraints is couched in Generalized Alignment Theory (McCarthy and Prince 1993): (11) Generalized Alignment Align (Cat 1 Edge 1 Cat 2 Edge 2 )= def Cat 1 Cat 2 such that Edge 1 of Cat 1 and Edge 2 of Cat 2 coincide. Cat 1 Cat 2 Pcat Gcat Edge 1 Edge 2 {Right, Left} (McCarthy and Prince 1993 :4 ) Generalized Alignment demands that the right or left edge of every prosodic or morphological constituent of Cat 1 coincide with the right or left edge of some other prosodic or morphological constituent of Cat 2 McCarthy and Prince (1993a) outline a number of typical alignments such as a. [PrWd [Stem (the left edge of every stem is aligned to the left edge of every prosodic word), b. ]Syllable]S tem (the right edge of every st em aligns with the right edge of every syllable), c. [PrWd [Ft (the left edge of every foot is aligned with the left edge of every prosodic word) and d. ]PrWd Suffix ] (the right edge of every suffix aligns with the right edge of every prosodic word Alig and Mester 1999: 188). It has been argued that although syllable well formedness constraints can be readily translated into a lignment constraints, alignment constraints produce more options in the

PAGE 130

130 typology of observed syllabic structures. They lay the foundation of prosodic morphological analysis. Anchoring Constraints Under Generalized Alignment, there is a set of constraints which demand correspondence between segments standing at the left and rights edges of forms in correspondence. The general formalism of this famil y of constraints is stated in (12) : (12) {Right, Left} Anchoring 1 has a correspondent at the designated periphery of S 2 Let Edge (X, {L, R})= the element standing at the Edge=L, R of X R IGHT A NCHOR 1 R) and = Edge(S 2 L EFT A NCHOR Likewise, mutatis mutandis (Kager 1999: 251) the Vb is infixed medi ally (as in [ ab [ ], so that it does not disrupt the corresponding left edge of the singular and plural forms. Therefore, L EFT A NCHOR is undominated. Finally, I will discuss Generalized Template Theory (McCarthy and Pr ince 1994a; Urbanczyk 1996a; Gafos 1995; among many more) which shows how templatic and prosodic requirements can be explained through the interaction of universal constraints supplied by Optimality Theory. Generalized Template Theory The discussion throug hout Chapter 4 has outlined the advantages and disadvantages of Templatic and Prosodic Morphology approaches. I also discuss the tenets and theoretical assumptions of Optimality Theory. One of the great advancement s of the Optimality Theoretic tools is the Generalized Template Theory that successfully elimi nates templatic constraints since it assumes that the interactions of

PAGE 131

131 phonological and prosodic constraints can successfully yield the desirable templatic form. Below I describe the assumptions of General ized Template Theory (GTT). Generalized Template Theory (GTT) is a powerful theoretical tool offered by Optimality Theory to account for templatic pat terns without reference to them. It assumes that templatic effects including CV patterns and the type of f oot structure in a linguistic form are derived directly from the interactions of motivated constraints in the Universal Grammar. Why refer to superfluous tools when a theory like Optimality Theory has access to information supplied by Universal Grammar? Th e adherence to Generalized Template exempts us from the need to formulate a templatic constraint. Therefore, c onstraints stipulating a parti cular CV shape for an affix / reduplicant or those relating to the size of a truncated form (e.g. T RUNC = or a reduplicant ( R ED are no longer needed. Moreover, there is no need to write constraints referring to the type of foot structure (e.g. I AMBIC or T ROCHAIC ). In Generalized Template Theory, the appropriate ranking of related constraints generates the optimal output without referencing the kind of syllables or feet the linguistic forms have. For example, one of the mechanisms for forming plurality in attaching a suffixal template VC with a fixed vocalism and a copy of the final bas The final shape of the plural output must equal a syllable size. However, following GTT, I do not propose any templatic constraint that stipulates the size of the plural form ( P L a constraint refere ncing the length of the plural marker ( SUFF = interactions of phonological and prosodic constrains yield the syllable size without the need to address it templatically.

PAGE 132

132 Summary of Chapter 4 Non concatenative morphology is e xpressed by imposing a change into the stem resulting in templatically diverse forms which cannot be analyzed into smallest isolable constituents. It can also be expressed by other operations such as ablaut whereby vocalic opposition is imposed on a deriv ed form to keep it contrastive from its base, truncation whereby a prosodic constituent is elided from a stem, infixation in which an extra prosodic or templatic affix is inserted inside a linguistic word, or reduplication which copies elements from the ba se. These mechanisms are highly complicated and impose a challenge to approaches which for long have successfully accounted for concatenative morphological systems. In Chapter 4 I have reviewed two pre optimality Theory approaches namely Templatic and Pr osodic Approaches and discuss their advantages and disadvantages. I then outlined the theoretical assumptions of Optimality Theory whose impetus has marked a new era into the analysis of phonological and morphological processes. Three of the theoretical to ols discussed inclu de Output Output Correspondence, Alignment constraints and Positional Faithfulness I also argued that the assumptions made by Templatic Morphology and Prosodic Morphology which do not benefit from the information in Universal Grammar ar e currently translated into universal constraints in the Optimality Theory framework. Chapter 5 offers an integrated Optimality Theoretic analysis of the regular plural Chapter 4

PAGE 133

133 CHAPTER 5 ANALYSIS OF THE REGULAR PLURAL SHAPES ng singulars onto plurals are enormously diverse. Numerous non concatenative morphological processes with concomitant phonological alternations mark plurality in the language. For example, plurality can be systematically marked by infixation of Vb attachi ng a suffixal template VC with fixed vocalism and a copy of the final base consonant, ablaut and mapping singulars with gemination onto a specific template. The singulars from which these plurals are derived range from bi consonantal to tri consonantal an d quadri consonantal singular forms. In addition to these morphological processes, plural forms exhibit vowel deletion, vowel insertion, assimilation and re syllabification. This diversity in plural formation can sometimes be systematic; the resultant plur al shapes straightforwardly relate to the particular shapes of their singular forms. For instance, only bi consonantal and uni consona n tal singular shapes can reduplicate one of their consonants to indicate plurality; tri consonantal singular shapes are ob served to pluralize by processes other than reduplication. Moreover, the majority of singular forms that take the infix Vb are quadri consonantal w ith the canonical shape CVCCVC( Vt ) Bi consonantal singulars whose second radical is geminated expand their segments and map onto a specific plural template. and triply marker s (i.e. suffixation and Vb infixation together or two suffixes consecutively follo wing one another un ). In Arabic, the latter process is available and widely known as plurality of plurality

PAGE 134

134 sets of plurals. Semantically, they are plurals only. Simeone Senelle (1997) identifies this tendency for Although, as has been mentioned previously, some plural shapes are systematically derived from specific singulars, o thers can hardly be related to their singulars. To illustrate, since bi consonantal singulars, for example, may take various shapes of plural (ablaut shaping into a specific template and attachment of a VC shape with fixed vocalism and a copy of the final consonant in the base ), it is extremely unpredictable to assign a definite plural shape to a particular singular form. Furthermore, establishing a general mechanism of plural forma challenge because there are many divergent plural patterns that cannot be solely plural shapes exhibit common morphological and phonological cha racteristics or tendencies which are indicative of the grammar of the language as a whole. Chapter 5 offers an integrated analysis to the systematically and phonologically conditioned plural shapes. It accounts for a range of diverse shapes of plurals in J within the Optimality Theory framework. The analysis first addresses the plurals with Vb infixation. It is then extended to account for the plurals attaching a suffixal template and plurals derived from geminated singulars. Finally, I offer three an alyses to the plurals with vocalic opposition. Analysis of Plurals with Vb Infixation One of the most prevalent patterns of plural formation in Vb abian languages do not mark plurality by Vb infixation. Moreover, none of the widely studied

PAGE 135

135 Semitic languages is reported to have the Vb infix as a plural marker. Although it is not the default mode of pluralization in the language, it occurs quite freque ntly when pluralizing stationary items, old and new tools and generally loan words which relate to gear ked to pluralize the Arabic word [mas t r Vb Thus, [mas abt *[mas t potential plural with the default plural suffix It also occurs when pl uralizing certain buildings such as offices, restaurants and hotels. I will first present representative examples of this pattern and then describe the locus of the infixation with detailed description of the singulars and resultant plural shapes: (1) Plu rals of Vb infixation a min sieves b mir cauldrons c ab big loads d ab cartridge belts As observed in the above examples, quadri consonantal singulars bearing the shapes CVCCVC or CVCCVC it or CVCCVC ah (the suffixes repr esent the feminine gender) take Vb infixation to mark plurality. The quality of the vowel in the infix can Although there are a velar or bilabial consonant while /a/ is preceded by a pharyngeal(ized), glottalized or a back consonant in general. The Vb infix resides towards the left edge of the plural form. The exact locus of this infix is the second syllable from the left edge of the plura l after the C 1 VC 2 of the base singular form. So, the final plural shape is CVC Vb CVC.

PAGE 136

136 (2) is a prosodic repres ( 2 ) Sing. Pl. 1 2 1 2 3 To govern the locus of the V b infix, I use an alignment constraint. The general formalism of the alignment family of constraints is repeated in (3) : (3) Generalized Alignment Align (Cat1, Edge1, Cat2, Edge2)=def Cat 1 Cat 2 such that Edge 1 of Cat 1 and Edge 2 of Cat 2 coincide. Cat 1 Cat 2 Pcat Gcat Edge 1 Edge 2 {Right, Left} (McCarthy and Prince 1993) Vb is aligned to the left edge of the output plural forms. It occupies exactly the second syllable of these forms as can be clearly seen in the following examples (the [.] indicates syllable boundaries): ( 4 ) Vb infixed plurals with syllabification indicated mi. rifle bolts ma. t caravans, turns, times rat ions, supplies One may argue that the Vb infix appears to be in the middle of these plural forms, and may be right aligned instead. However, if I assume it is right aligned, then The analysis presented below will rule out this potential form because there is an active constraint in and entails it s segments {V and b} should be contained within a syllable. The infix cannot span across tw o syllables. In *miz.nV.bed, the infix segments Vb are contained in two separate syllables.

PAGE 137

137 To address the locus of infixation, I use the alignment constraint in (5) : (5) A LIGN V b L Align Vb to the left edge of the plural form The viol ation of this alignment constraint is gradient. The actual output plural aligns Vb exactly after three segments {C 1 V, C 2 } from the left e dge of the plural form, so three violations of A LIGN V b L are assessed. The Vb resides in the second syllable of the plural form, making C 2 the onset of the Vb infix, and this will prove important to the analysis. This alignment constraint is dominated by the language requirement to keep the right and left edges of the singular forms corresponding to the right and lef t edges of the plural forms. The infix Vb does not disrupt the edges of the singular form when plurality is marked. The set of constraints that keep the edges of the singulars and plurals in a correspondent relation are the anchoring family of constraints whose general formalism stipulates the following: (6) {Right, Left} Anchoring 1 has a correspondent at the designated periphery of S 2 Let Edge (X, {L, R})= the element standing at the Edge=L, R of X (Kager 1999: 251) The actual plural forms have the segments at the leftmost edge and the rightmost edge corresponding with those at the leftmost and right most edges in the singular forms. To address this fact, I use the anchoring constraints (7) and (8) : (7) L A NCHOR PS 1 the segment at the leftmost edge of the plural form corresponds with that at the leftmost edge of the singular form (8) R A NCHOR PS the segment at the rightmost edge of the plural form corresponds with that at the rightmost edge of the singular form 1 PS stands for Plural Singular, following other families of Correspondence constraints.

PAGE 138

138 The violation of the above anchoring constraints is categorical; it occurs when the segments at the edges of both the singular and plur al forms do not match. It stipulates that for the segments at the rightmost and leftmost edges of the plural form, there must be corresponding segments at the leftmost and rightmost edges of the singular form. While L A NCHOR PS is ranked above A LIGN Vb L i to ensure that left edges of the singular and plural stand in correspondence, and stress the fact that plural marking is never observed in the left side of these plurals R A NCHOR PS should be ranked lower than A LIGN Vb L so tha t suffix ation and other plurals marked to the right edge are not doomed. Tableau [1] shows the competition between the alignment constraint and the anchoring constraints. Tableau [1] L A NCHOR PS A LIGN V b L R A NCHOR PS Vb L A NCHOR PS A LIGN Vb L R A NCHOR PS a. mi.z m i z b. miz. .d m i z d c. .miz.n d 2 *! Tableau [1] illustrates the locus of the infixed Vb which is determined by a competition between the alignment and L A NCHOR PS constraints The opt imal output (a) has the infix right after the first three segments, incurring three violations {m, i, z} to the low ranked constraint A LIGN Vb L. It obeys the high ranked constraint L A NCHOR PS and incurs no violation to R ANCHOR PS by keeping the leftmo st and rightmost edges in correspondence. Candidate (c), though it aligns Vb all the way to the left edge and exhibits no violation to A LIGN Vb L it violates L A NCHOR PS which is crucial for plurals. The segment {m} at the leftmost edge of the singular has no correspondent in the leftmost edge of the plural form. Thus, it is doomed. As a matter of 2 Other potential sub optimal candidates such as [ mebizned ] will be dealt with in the next section.

PAGE 139

139 Candidate (b) aligns Vb to the right edge, skipping far more segments in the plural form than the segments skipped in the actual output. Moreover, it violates the right anchoring constraint. Therefore, it is out too. In the actual output plural, the final C of the first sylla ble .C 1 VC 2 of the singular form C 1 VC 2 .C 3 VC 4 makes an onset to the Vb infix (9). (9) Sing. Pl. 1 VC 2 and closes the first syllable. However, in the output plural, it serves a s the onset to the infix Vb a requirement relatively high in the language. So, a potential candidate such as O NSET (10) O NSET Every syllable begins with a consonant. (McCarthy and Prince 1990b a nd 1993a) Tableau [2] illustrates the fact that the locus of the Vb infix must conform with the prosodic requirements of the language. Although the right and the left edges of candidate (d) stand in absolute correspondence with the singular output, it vi olates O NSET Tableau [2] L A NCHOR PS A LIGN V b L, O NSET R A NCHOR PS Vb L A NCHOR PS A LIGN Vb L O NSET R A NCHOR PS a. mi.z m i z c. *! d. miz. m i z *!

PAGE 140

140 The actual output is the most harmonic candidate as it exhibits the fewest violations to t he proposed constraints. It exhibits three violations to A LIGN Vb L by aligning the infix three segments away from the left edge of the plural form. Candidate (d) equally violates A LIGN Vb L three times. However, it fatally violates the constraint requir ing every syllable in the output form to begin with an onset Thus, candidate (d) is doomed in the ranking above. [ m b ] is yet another possible candidate which needs to be considered for the ranking established above. This candidate violates A LIG N V b L only once by skipping the segment {m} at the left of the plural form. Thus, it may seem more harmonic than our actual plural form. However, this candidate has the two segments {V and b} of the infix separated into two syllables. The optimal output has these two as illustrated in Tableau [3] Tab leau [3] L A NCHOR PS A LIGN V b L, O NSET R A NCHOR PS Vb L A NCHOR PS A LIGN V b L O NSET R A NCHOR PS a. mi.z m i z c. *! d. miz. m i z *! e. m b m The ranking above requires the stipulation of a constraint that would favor mi.z over m b infix bold faced ). As seen, the only difference between these candidates that favors the winner is that the segments contained in the infix and b} are contained in the same syllable in the winning candidate. Candidate (e) is

PAGE 141

141 ma kes the onset to the following syllable. So, the infix makes two syllables. The two segments of the infix in the actual output make a syllable and both must belong to that syllable. Representation (11) show s the locus of the Vb infix in the actual output a nd in a potential output : (11) Actual Output Potential Output 2 1 2 z m b i z (11) above show s that the infix in t he optimal output is contained within a single syllable. Thus, it has the weight and size of a syllable In the potential output, the segmental content of the infix gets separated; the vowel belongs to a different syllable from that that has the {b} Crowh urst (2004) who studies the behavior of the reduplicants in Mangarayi, Mokilese and Tzeltal crucially Red[uplicant]s in Mangarayi, Mokilese and Tzeltal may not be syllables in segmental terms, each has the weight of a syllable (Crowhurst 2004:131). She proposes a size restricting constraint developed from the more conventional generalized alignment restricted to no more than a syllable size by the constrai 2004: 129). She stipulates the constraint (12) below to offer a sufficient analysis for the Morphological Category and Prosodic Category misalignment phenomenon: (12) the phonological exponent of an affix i s not larger than a syllable (Crowhurst 2004: 129)

PAGE 142

142 Other evidence for restricting the size of an affix to a syllable comes from Saanich, a dialect of North Straits spoken on the Saanich Penin sula of north Victoria, British Columbia and neighboring islands. According to Kiyota (2003), the plural morpheme of Saanich is expressed by two non reduplicative affixes (more specifically infixes) ess .1) and two reduplication patterns which are dependent on the stress pattern of the root. The infix l has two distinct realizations [ ] and [ of the constrain to rule out plural forms which infix only l constraint rules out candidates whose affixes are larger than a syllable. Moreover, it can be extended to capture the size properties of affixes in general. Violation to this constraint is incurred by: (i) separating the content of an affix or (ii) having an affix that is bigger than a syllable. Incorporating into the analysis of the Vb infixed plural forms Tableau [4] reveals the interaction of the size res tricting constraint with the so far established constraints. Tableau [4 ] L A NCHOR PS, A LIGN V b L, O NSET R A NCHOR PS Vb L A NCHOR PS A LIGN V b L O NSET R A NCHOR PS a. mi.z m i z b. miz m i z n c. *! d. miz. m i z *! e. m b *! m Candidate (e) is now doomed because of the high er which requires the segments of the infix to be contained in a single syllabl e. In this candidate, { of the infix makes the nucleus to the first syllable at the left edge of the plural form while {b} is the onset to the following syllable, forcing the infix to span over two syllables and violating

PAGE 143

143 a sub pattern of Vb infixed plurals which has peculiar morpho phonological properties. Such properties make it diverge from the regular Vb infixed shapes. For instance, some of the Vb infixed plurals begin with a vowel, often a schwa, instead of the systematic in itial #mVC. syllable. They take the shape [ .C Vb. CVC ] and thus differ from the phonologically conditioned shape [ mVC.C Vb. CVC ] in starting with an onsetless syllable plural initially. They are derived from singular forms ( 1 3 a c below), which begin with a nasalized vowel ( traditionally analyzed as a result of deleting a nasal /m/ in Johnstone, ( 1981 ) ; Nakano ( 1986 ) and Hofstede ( 1998). These forms may keep the nasal /m/ or del ete it in their plural form In phonologically c onditioned [ mVC.C Vb. CVC ] shape and the [ V.C Vb. CVC ] with a deleted /m/ and initial vowel. In the singular forms, when /m/ deletes, the following vowel ]. The singular forms may be pronou consultants accep t the two variations as interchangeable and make no difference in semantics between the two options. ( 1 3 ) Plurals with Vb infix and initial vowel keys offices rooms for guests The plurals with the nasal /m/ nicely fit into the proposed analysis as revealed in the above tableaux If we assume that the plurals that start with a vowel are originally derived from a singular whose /m/ is deleted then t hese forms also integrate well in the analysis as Tableau [5] and Tableau [6] show.

PAGE 144

144 Tableau [ 5 ] L A NCHOR PS, A LIGN V b L, O NSET R A NCHOR PS + Vb L A NCHOR PS A LIGN V b L O NSET R A NCHOR PS a. m .g .l g b. c. *! d. e. m b *! m Tableau [6 ] L A NCHOR PS, A LIGN V b L, O NSET R A NCHOR PS + Vb L A NCHOR PS A LIGN V b L O NSET R A NCHOR PS a. m .g .l b. *! c. *! d. *! e. m b *! m ive deletion. It deletes /m/ word initially and /w/ and /b/ word medially and replaces the deleted segments with nasalized or long vowels ( Johnstone 1981; Nakano 1986; Hofstede 1998) although it is unusual to lose an onset and len g then a vowel 3 This trend of deletion also applies to the plural formation and reveals a violation to O NSET Thus, it comes as no surprise that plural shapes for the singulars with a de leted initial /m/ and retained one In T ableaux [5 ] and Tableau [6 ] both candidates (a) and (b) are optimal and admitted in the grammar of the language. However, it is important to note that the {b} of the plural infix never deletes since it is the main element indicating plurality in these forms The retention of the {b} of the infix happens in spite of the fact that it is prone to delete in elsewhere contexts. 3 The usual scenario in phonology is to lose a coda and lengthen the vowel preceding it. This is called 1989 and Clements 1986 to mention very few).

PAGE 145

145 Summary of the ranking for plurals with Vb infixation: To sum up, the analysis of the plurals with Vb infixation reveals the interaction of an alignment constraint and L A NCHOR PS to determine the exact locus of the infix in the output plural forms. As the infix resides in the second syllab le of the plural form, it exhibits three violation marks to the alignment constraint at the expense of obeying the left anchoring constraint The positioning of the infix has to conform with the language requirement to have onsets; thus, O NSET plays a role in Jebb and rules out suboptimal candidate s with an onsetless syllable Moreover, the segments of the plural infix must be contained in a single syllable and be of a syllable size The constraint which add (14) illust rates the overall ranking of the proposed constraints: (14) L A NCHOR PS, A LIGN V b L, O NSET R A NCHOR PS The left anchoring constraint monitoring the segment at the leftmost edge of the singular and plural forms along with the constraint restr icting the size of the infix outrank the alignment constraint and syllabic well formedness constraint Since suffixation and other plural markers occur to the rig ht edge of the plurals R A NCHOR PS is low ranked. Analysis of Plurals with Suffixa l Template suffixal template with fixed segmentism and a copy of the final consonant of the base. This mode of pluralization relates to borrowed nouns from Arabic; approximately 15 out of the 19 collected forms (78%) pertain to borrowings from Arabic and related dialects of Arabic. In Omani Arabic (henceforth OA) these forms pluralize by reduplicating the final consonant in the base preceded by a long vowel /u:/. Similarily, the majority of the

PAGE 146

146 Jebb with fixed vocalism (15) represents examples from Omani Arabic ( OA ) : ( 15 consonantal p lurals Singulars Plurals in Plurals in O A Gloss a feet, soles b d cheeks c raf in O A rfu:f shelves, racks, bulks d kaf in OA kfu:f palms of the hand; claws The semantics of these forms is not restrictive to one particular category and range from animate bein etc. to inanimate objects like axes, books and letters. Moreover, the class of the plurals taking the suffixal template with partial reduplication is diverse in nature and the words collected can equally be divided between the feminine and masculine class groups. Bi consonantal singular forms of mostly CVC shape, whose vocalic quality varies VC template whereby V is / and the final consonant slot is filled by a reduplicated consonant from th e base (V)CC x x (C x C x denotes a reduplicant). The parenthesized vowel in the template (V)CC x x is pro s thetic and often gets inserted word initially if the word begins with a consonant cluster. However, it is hard to establish a pattern which explains wh en exactly this pro s thetic vowel is realized. It is not always the case that certain consonant combination triggers insertion of this vowel In (16) I show how various combination of consonant clusters may and may not surface with the pro s thetic vowel: (1 6 ) Combination of consonants with pro s thetic vowel but not in [kfar] but not in [skun] and

PAGE 147

147 H margins of syllables. In some cases, the pro s thetic suffixal template (form ( f ) below). This pattern of pluralization has three characteristic s that are crucial to the proposed analysis. First, the suffixal template has the size of a syllable .VC. Secondly, it Third, the final shape of the plural form is a sy llable length due to the collapse of the syllable of the base singular form (CVC) into consonant clusters CC+VC. Fixed segmentism has been explored in Alderete et al (1999) and is defined as that are invariant rath from the base nor is it a realization of an unmarked vowel, the segment as such is considered marked cross linguistically and of morphological fixed segme ntism, which is a type of affixal morphology. Ratcliffe (1996: 299) provides at least three pieces of evidence to support that this type of reduplication is not real and thus falls under templatic expansion. First, this type of plural targets bi radical and mono radical consonantal singulars. Second, the quality of the vowel in the templatic suffix is similar to the vowel that occurs between the C 2 and C 3 in three consonant internal plural. To illustrate, in one of the three templatic plural patterns in J few tri consonant rather than the reduplicated consonant in these plural types. Finally, Ratcliffe obs erves that if this pattern occurs in tri consonantal singulars, then one of the

PAGE 148

148 consonant s must be glide (hollow form). I add to these arguments the fact that this type of plurals often maps onto a single syllable, which lends support to its being templati c in nature. The / o the reduplicated consonant is realized simultaneously with th at consonant and they both make a suffix in the final shape of the plural form. It is important to analyze this plural pattern as both a suffixal template with fixed segmentism and reduplication of the final consonant. (17 ) Partial suffixal r eduplication a fish b nuf selves c shelves, racks, bulks d mus razors e palms of the hand; claws f pilgrims Singular forms which are bi consonantal, similar to the aforementioned ones, also or /e/ in the suffixal template instead of ( a ) ( c ) below ) and a copy of the final consonant. In these plural forms, th ere is a vowel between the first two consonants unlike the previous ones, making a total of two syllables in the final shape of this plural type (18 ) Exceptional r eduplicat ed f orms a dry leaves b lavatories c .hab ot/ hib ot h beb/ heb songs lurals with the suffixal template reveal an asymmetry in the distribution of identical consonants. Plural forms with an initial plurals listed above. On the other hand, noun plurals with only a final sequence of identical consonants abound.

PAGE 149

149 To address the fact that initial consonants are never doubled in Semitic in general and in Modern Hebrew denominal verbs in particular, Ussish kin (1999) proposes a type of Anc horing constraint ( S TRONG A N CHOR L ) which disallows doubling of a consonant at the left edge Before I discuss the S TRONG A N CHOR L constraint, let us first recall the general A N CHOR L constraint as a conditional expression : (19 ) A NCHOR L : x y, [(x = Edge(S 1 L)) & (y = Edge(S 2 L))] [x R y] (Ussishkin 1999:413) The above constraint stipulates that i f x is at the left edge of S 1 and y is at the left edge of S 2 then x corresponds to y It is violated when the leftmost segments of S 1 and S 2 do not stand in correspondence. Ussishkin then flips the order of constraint (19) to rule out doubling of the lef tmost segment. He calls constraint ( 20 ) below a S TRONG A N CHOR L and formulates as: ( 20 ) S TRONG A N CHOR L x, y, [(x = Edge (S 1 L)) & (x R y)] [ y = Edge(S 2 L)] (Ussishkin 1999: 414) S TRONG A NCHOR L stipulates that if x stands at the left edge of S 1 and if x corresponds to y, then y must also stand at the left e dge of S 2 According to Ussishkin, S TRONG A N CHOR left edge elements, and in particular, has the effect of disallowing multiple correspondents of a segm ent that is initial element, every correspondent of that element must be initial in the output: the correspondent of an edge element ring a unique

PAGE 150

150 correspondent of element at the left edge of forms. This constraint is cru cial in the VC template as it forbids doubling of the leftmost segment in the plural form I illustrate the effect of S TRONG A N CHO R L in (21) : ( 21 ) S TRONG A N CHOR L satisfied viola ted violated S 1 [ L 1 u 2 t 3 ] R [ L 1 u 2 t 3 ] R [ L 1 u 2 t 3 ] R S 2 [ L 1 t 3 3 ] R [ L 1 1 3 ] R [ L 1 1 3 ] R In order to allow the rightmost segment to double, then I need a comparable ( S TRONG A N CHOR R) constraint, which is outranked by S TRONG A N CHOR L (22): ( 22 ) S TRONG A NCHOR R Let C f = the rightmost consonant of a string: x, y, [(x = (S I C f )) & ( x R y)] [y = Edge(S 2 R)] (Ussishkin 1999 :415) S TRONG A NCHOR R penalizes doubling of rightmost consonant. Therefore, S TRONG A NCHOR L outranks S TRONG A NCHOR R and I NTEGRITY (a faithfulness constraint which bans multiple correspon d ence of a segment ) and this ensures doubling of the rightmost segme nt only. Observe the following tableau: Tableau [7] S TRONG A NCHOR L S TRONG A NCHOR R I NTEGRITY [ V C] S TRONG A N CHOR L S TRONG A NCHOR R I NTEGRITY ut *! *! Candidate (b) is ruled out because the segment at its leftmost edge is doubled, violating the high ranked Strong Anchor L. Candidate (c) also has the leftmost segment

PAGE 151

151 doubled and impinges on S TRONG A N CHOR L Therefore, it is doomed. Candidate (a), on the other hand, has unique correspondence of segments at its leftmost edge. Thus, it wins the competi tion. Since the v o of the vowel in the last syllable of the base singular form, the analysis assumes that the suffixal template VC in the plural f orms comes with a pre specified or fixed segment in the (2 3 ) Output Singular VC Therefore, the constraint M AX V S UFFIX is undominated in these plurals. Since the vowel in the root is lost in these plural forms, then M AX V S UFFIX outranks M AX V R OOT (24 ) M AX V R OOT the vowel in the output singular forms has a correspondent in the output plural forms. (25 ) M AX V S UFFIX the vowel in the pre specified vocalic position in the suffix mus t be realized. Moreover, form above incurs a violation to C OMPLEX (26 ) C OMPLEX *[ CC and *CC] consonants in the clusters to two. Only one collected reduplicated plural form seems to deviate from this rule. In Tableau [8] deletion of the vowel in the root is tolerated at the expense of keeping the vowel in the suffix intact. This is due to the fact that plurality is indicated by both the vowel in the suffix and the C copied from the base. C OMPLEX is violated in the optimal output

PAGE 152

152 Tableau [ 8 ] M AX V S UFFIX M AX V R OOT C OMPLEX [ V C] M AX V S UFFIX M AX V R OOT C OMPLEX *! Candidate (a) the optimal plural output, violates C OMPLEX and M AX V R OOT since it has clusters syllable initially and los es its root vowel These constraint s are low ranked The pre specified vowel in the suffixal template is lost in candidate (b), violating fatally M AX V R OOT There is a conspiracy between M AX V S UFFIX and M AX V R OOT wh ich results in deleting the vowel in the root at the expense of keeping the vowel in the suffixal template intact. Since M AX V S UFFIX which says M AX V R OOT which stipulates the deletion of the vowel in the root, the output form always surfaces with the vowel of the suffixal template and loses the vowel in the root. This stems from the fact that plurality in these forms is marked by the vowel and consonant of the suffixal VC template together. Therefore, the loss of the vowel in the root is less costly as long as plurality is marked by another vowel. Candidate (c) obeys the constraints above but it is not selected as the actual output because it has one more vowel than the optimal output does. The subsequent discus sion will rule it out. The final shape of the output plural is a syllable length as a result of (1) deleting the root vowel and (2) maintaining the vowel in the suffix and preserving its quality in the output form. Thus, I need a constraint that translates the se crucial facts Such a constraint should be able to rule out suboptimal outputs like which copy the entire base string without deleting the vowel in the base

PAGE 153

153 which preserves the two vowels of the base and suffix and extends the final C I adopt the markedness constraint N O 2005: 299) in the output plural forms. This constraint abbreviates a whole set of phonotactic requirements which interact with the constraint M AX V S UFFIX to ensure that the final shape of the plural form has only one vowel, that of the suffix. (2 7 ) N O V No vowels appear in the output form C onstraint (27) interacts with the constraints preserving the vowel in the suffi x at the expense of deleting the vowel in the root (Tableau [9]) Tableau [ 9 ] M AX V S UFFIX N O V M AX V R OOT [ V C] M AX V S UFFIX N O V M AX V R OO T *! **! *! **! Candidate s (c) and (e) lose because they fail to delete the vowe l in the root. They also surface with one more vowel than that in the optimal output. Thus, they impose one N O constraint than the optimal output (a). Candidate (b) deletes the vowel in the suffix and fatally violates M AX V S UFFIX an undominated constraint in the ranking above. Candidate (d) which loses the vowels of the root and suffix incurs a fatal violation to the high ranking constraint M AX V S UFFIX It also violates syllable constraints like the one requiring a nucleus in a sylla ble ( N UC ). Thus, it is out too. Candidate (a) faithfully obeys M AX V S UFFIX at the expense of violating the low ranked constraint M AX V R OO T It incurs one violation to N O V and thus is the most harmonic amongst the other candidates. Th erefore it wins th e competition.

PAGE 154

154 Incorporating the above ranking with the so far established ranking for this shape of M AX V S UFFIX N O V C OMPLEX M AX V R OO T T ableau [10 ] summarizes the effect of the interaction among these constraints. Candidate (a) has a consonant cluster word initially as a result of losing the vowel in the root. It violates C OMPLEX and M AX V R OOT which are low ranked. It also incurs one violation to N O V which penalizes all vowels in the output forms In comparison to candidates ( c ) and ( d ) which have two violations to N O V, candidate (a) is the most harmonic output and it wins the c ompetition. Candidate (b) loses the pre specified vowel in the suffix, incurring a fatal violation to M AX V S UFFIX Thus, it is doomed Candidate (e) has two violations of N O V, a highly ranked constraint Tableau [ 10 ] N O V, M AX V S UFFIX C OMPLE X M AX V R T V C] N O V M AX V S UFFIX C OMPLEX M AX V R T *! **! **! t **! Summary of the ranking for plurals with a suffixal templ ate: The picture of how these constraints crucially interact to yield the optimal output is now clear. The analysis is elegant as it adheres to the principles of Generalized Templatic Theory which disfavors templatic constraints and assumes that interacti on of phonological and prosodic constraints produce the desired result. The analysis here does not refer to templatic constraints such as P L it assumes that a set of phonological constraints and N O V which penalizes vowels in the output forms yield the fact that a syllable has only a single vowel. N O V interacts

PAGE 155

155 with M AX V S UFFIX to li mit the occurrence of any unnecessary vowels in the output plural. Therefore, the full ranking established for these forms is : (2 8 ) M AX V S UFFIX N O V C OMPLEX M AX V R OO T Analysis of Templatic Plurals Plurals Derived from Geminated Singulars Geminated x C x (C x C x denotes a geminate) are mapped onto a particular template when they pluralize. They take the shape CVC x VC x whereby the geminate is broken up by a vowel. (2 9 ) Plurals derived from geminated singulars a t hills b towns; small villages c coffee pots Few noun plurals derived from geminated singulars lose the vowel between the first consonant and the geminate, forming the shape CC x VC x (example s ( 30 .a and b below). some instances, these clusters are resolved by a pro s thetic vowel which gets inserted before the initial consonant clusters #(V)CC (example ( 30 .b ) below). ( 30 ) Plurals losi ng the consonant between C 1 and C 2 a lbeb kernels b heavy wooden bolts of a door Underlyingly, the singular forms contain two consonants; the first consonant is C 1 in the template C 1 VC x C x while the second consonant has two instan ces of /C x /. Thus, singular forms such as form s (30 ) above must be analyzed as bi consonantal /l b/ and mapped to a tri consonantal template from left to right. The geminate consonant in the singular form represents a single melodic segment. In the plural form, however, the second consonant is spread to the final consonant slot of the template; it spans over

PAGE 156

156 two positions with a vowel intervening between the consonants forming a "long distance" geminate, as (31) below illustrates: ( 31 ) Singular Plural l b l b b C V C C C C V C Because the geminate of the singular form is broken up by a vowel in the plural form, it has two correspondents in the plural form violating the constraint calle d I NTEGRITY in Optimality Theory. I NTEGRITY penalizes relations between a form S 1 (here, the Singular) and another related form S 2 (here, the Plural), where a segment in S 1 has more than one correspondent in S 2 The subscripts portray pairs of corresponde nt segments, so that the segment /ll/ in [ ] enters into two (hence violation of I NTEGRITY ) correspondent pair of segments: (ll i i l j ). ( 32 ) I NTEGRITY SP No segment of the singular has multiple correspondents in the plural. (3 3 ) I NTEGRITY SP violation: i S (Si ngular), [ ] 4 i e l j P (Plural), [ ] Noun plurals derived from geminated singulars have an extra vowel which breaks up the consonant cluster, violating D EP V PS which militates against an insertion of a vowel in the plural form: 4 The suffix is excluded as it does not contribute to the consonantal ro ot of the form.

PAGE 157

157 (3 4 ) D EP V SP 5 Vowels in the plural form must have vowel correspondents in the An alternative repair to the geminate is to shorten it as in [ ]. This candidate, compared t o the actual output [ ], is ruled out by the constraint I DENT Q which requires that the skeletal quantity of segments in the stem must be preserved or transferred to the sur face (Dell & Elmedlaoui 1992; Gafos 2003). Specifically, in the actual output [ ], the /ll/ in [ ] is linked to two skeletal C slots, but the correspondent of /l/ in [ ] is linked to a single C slot. I DENT Q penalizes this mismatch. (3 5 ) I DENT Q A segment in S 1 and its correspondent set in S 2 have identica l quantities (number of C slots) (36 ) a l e l /l/ is linked to two C slots C V C But in: /l/ is linked to ONLY one C slot This reveals that I DENT Q outranks both I NTEGRITY SP and D EP V SP. Tableau [11] illustrates the relation among these constraints: Tableau [1 1 ] I DENT Q I NTEGRITY SP, D EP V SP pl / I DENT Q I NTEGRITY SP D EP V SP a. *! The winni ng candidate (a) has /ll/ split into two /l/ correspondents by inserting a vowel to break up the consonant clusters. Thus, it violates the low ranked constraints I NTEGRITY SP and D EP V SP at the ex pense of obeying I DENT Q which requires identical number of C slots for the two identical segments /l/ in the output singular and plural forms. On the other hand, candidate (b) loses one of the C slot devoted for the /ll/ in the 5 This constraint is also violated in the plurals attaching a suffixal VC template.

PAGE 158

158 output singular and surfaces with one C slot for a single /l/; thus, it violates the high ranking constraint I DENT Q Therefore, it is doomed! not broken up by a vowel. This candidate also looks the same as the output singular form except that it does not have the feminine suffix It violates R EALIZE M ORPH a const raint which stipulates that a morpheme ( be it an affix or an internal change) is imposed into a form when a particular meanin g or a morpho syntactic function is expressed by that form. (37 ) R EALIZE M ORPH (Kurisu 2001:39) Typology of Geminates which stipulates that intervocalic geminates VGGV are the most common 6 while geminates which do not have vowels on both sides *VGG# are the most r are (Muller 2001 who did a survey of 40 languages with geminates) (Tableau [12]). Tableau [ 1 2 ] I DENT Q *VGG# I NTEGRITY SP, D EP V SP pl / I DENT Q *VGG# I NTEGRITY SP D EP V SP a. *! *! Candidate (c) is doomed as it fatally violates *VGG#. Candidate (b) loses the competition because it violates I DENT Q by losing one of the C slot s The ranking established for the noun plural forms derived from geminat ed singulars: (38 ) I DENT Q *VGG# I NTEGRITY SP, D EP V SP. 6 The geminate in the output singular is thus evenly syllabified between the preceding and following syllables. They occur intervocalically adhering to the Typology of Geminates.

PAGE 159

159 Templatic Plurals Losing the Vowel between C 1 and C 2 which have consonant clusters word initially freely surface in the language violating the low ranked constraint C OMPLEX The ranking established above still holds true for these plurals. However, t o rule out a sub optimal candidate with a vowel betwee n C 1 and C 2 word initially, I use the markedness constraint N O V which is previously used to account for plurals with the suffixal VC template. I repeat the definition of this constraint in (39) below (3 9 ) N O V No vowels appear in the output form Ta bleau [ 1 3 ] I DENT Q *VGG# I NTEGRITY SP N O V D EP V SP, *C OMPLEX pl/ I DENT Q *VGG# I NTEGRITY SP N O V D EP V SP *C OMPLEX a. lbeb **! *! *! In Tableau [13] the potential does not lose the vowel between C 1 and C 2 in the singular output incurring one more violation to N O V than the optimal output Thus, it is out. Not all the forms with a geminated consonant behave the same with regards to N O V. Hence, t he definite ranking with respect to N O V cannot be established for these forms. I observe that the majority of the plurals derived from geminated singulars retain the vowel between the first two consonant s while a few of them lose it. A justification for s uch a behavior is unknown, but the language may opti o nally have clusters plural initially. Summary of the ranking for templatic plurals derived from geminated singulars: To sum up, the noun plural forms derived from geminated singulars are

PAGE 160

160 mapped onto a p articular template which results from a reconciliation between the constraint requiring identical quantity of consonants and that militating against having multiple correspondents in the plural for the geminate in the singular. The constraint enforcing the typology of geminate syllabification also comes into play; it bans geminates syllable finally. Therefore, the overall ranking for the plurals derived from the geminated singulars: ( 40 ) I DENT Q *VGG# I NTEGRITY SP D EP V SP, *C OMPLEX Analysis of Plura ls with Ablaut Positional Faithfulness The fourth However, the vocalic change is not always systematic or phonologically conditioned. Quite a large number of plurals, which are mark ed by ablaut, are morphologically conditioned or come with extra morphology on them, posing a challenge to integrate To begin with, the majority of tri consonantal and quad ri consonantal singular forms undergo a vocalic change in the final syllable (forms 41 a e below). H owever, there are many irregular plurals which undergo vocalic change in both the first and final syllables (forms 42 a e). It is observed that the majority of plurals with ablaut consistently have one direction for the vocalic change, namely backing. However, back vowels significantly not uncommon to find plurals with /u/, /a/ or /o/.There are very few exceptional forms which surface with a front vowel in the last syllable; at least three forms in the collected data ablaut

PAGE 161

161 into phonologically and morphologically conditioned groups. I observe that while phonologically conditioned ablaut plurals only target one vocalic change, morphologically conditioned plurals may have more than one vowel change. To illustrate, plurals in ( 41 ) ch ange the vocalic quality in the final syllable only, whereas the plural forms in ( 42 ) express plurality by imposing a vocalic change into each vowel in the plural form. Thus, the vowels of the singulars and plurals are distinct. ( 41 ) Phonologically conditi oned ablaut plurals t t dresses b. fag ri fag ru Bedouins c. mo mo flesh of backs mim mum small pieces of wood fef shortened waistcloths (for men) ( 42 ) Morphologically conditioned ablaut plurals a. orphans (m.) b. s af rir s flowers c. isolated homes d. cheeks e. partings Secondly, bi consonantal singular forms, with exactly the same shape as th e one taking a suffixal template with fixed vocalism and a copy of the final consonant in the In Table A 5 below bi consonantal forms (a) and (b) take ablaut as a marker of plurality while the forms (c) and (d) whose shapes are exactly like (a) and (b) attach a s uffixal template to mark plurality. This requires sub categorization for the bi consonantal singular forms because they map onto distinct plural shapes though one might expect a single plural formation process to target all the singulars of the same shape.

PAGE 162

162 Table Ablaut Vs. suffixal template in bi consonantal forms Bi consonantal forms pluralizing by Ablaut Bi consonantal forms pluralizing by suffixal template Sing. Pl. Sing. Pl. a. nid nud c. dik d. ham hmum Previous works on J Senelle (1997) mention ablaut as a pervasive mode of pluralization in the language. According to syllable of the plural forms (fo rms 41 a e above ). Although the collected plurals with are marked by back but not rounded vowels. In these plurals, /a/ occurs in the final syllable. ( 43 ) Ablaut plura ls with /a/ in the final syllable date stones b. gilil t bullets c. sinor t sinar cats sardines As mentioned previously, the type of singular bases ablaut triggers can either be simplex (a sole syllable base of the sha pe CVC) or diverse (tri consonantal and quad ri consonantal singulars). ( 44 ) Singular bases of the shape CVC j t sheep b. nid nud water skins ropes ( 45 ) Singular bases of various shapes a. s afrir s flowers or isolated homes c. es es birds

PAGE 163

163 To reiterate, the plurals with ablaut are characterized by a number of tendencies. I sum up the major observations made about the plurals with ablaut below : 1 of the shape CVC undergo back vowel range s in height. Front vowels in the singular forms shift the quality of their vowels to back vowels. The singular forms have different ranges of vowels but the plurals seem to be character ized by back vowels. 2 In plurals that have more than one syllable, the final syllable undergoes the vocalic shift. In a few cases, the first and last syllables experience a vowel alternation as in [ 3 Sometimes, the vowel chan ge is unpredictable. For example, in a few cases, there is no vocalic change observed; the vowel stays intact as plurality is formed. Morphologically, this is called conversion whereby a different morphological function is assigned to a word, yet the word does not change or take a marker for it ] [ 4 However, by and large, a change in the vowel quality must happen to mark plurality. In other words, the majority of the plurals have some sort of a vocalic change. 5 In so me cases, the vowels in the plurals have the same height as that of the of the plural for ms is identical to that in the initial syllable of the singular forms, I assume that I DENT initial [ back] SP is undominated in these plural forms (4 6 ) I DENT initial [ back] SP [ back] in the initial syllable of the singular form is identical to [ back] in the initial syllable of the plural In Chapter 4 I have outlined evidence supporting the prominence of the root initial syllable cross linguistically. The evidence reveals that initial syllables have mo re priority and more privile ge than othe r positions in a form based on a number of psycholinguistic and phonological studies Moreover, in terms of learning and processing, a root initial syllable remains distinctive from other syllables. Thus, adopting a positional faithfulness

PAGE 164

164 constraint to ac count for the behavior of the initial syllable in these plural forms is justified back vowel in the last syllable, I also assume that there is the feature [+ back] that needs to be parsed in output plural forms. This feature comes associated with the singular forms to derive these plural forms. It is violated when [+back] is not parsed in any syllable of the plural output forms. In Optimality Theory, the constraint that milita tes against the deletion of this feature is: (4 7 ) M AX [+ back] 7 the feature [ + back] must be parsed in the output plural This constraint is highly ranked since [+back] must be realized to mark plurality in these noun plural forms. It is observed that i n many plural forms the first syllable contains a back vowel too. However, the vowel in the last syllable is the one that needs to be distinct from the last sylla ble of the singular form while the vowel in the initial syllable stays intact. Thus, both M AX [+ back] and I DENT initial [ back] M AX [+ back] rules out can didate (c) in T ableau [1 4 ] below This candidate has a back vowel in the first syllable but does not in fact parse the feature [+ back] which should be realized in the final syllable. In order to capture the fact that a t least one syllable in the plural forms undergoes a vocalic change to [+back] from its co rresponding syllable in the singular forms I adopt 7 An alternative analysis would be RealizeMorph, in which [+ back] is the plural morpheme that needs to be realized in the output plural form. I will show how an alternative analysis with RealizeMorph may produce exactly the same res ults reached by Max [+back] in a subsequent section.

PAGE 165

165 a general faithfulness constraint that checks identity of backness of the vowel s in the syllable s of both the singular and plural forms. This constraint will be dominated as there is often that tendency t o alter the quality of the vowel in at least one syllable to The plurals with ablaut have up to two syllables only; no form has more than two syllables. In the majority of these plurals, the last syllable undergoes the vocalic c hange. It specifically I assume a general faithfulness constraint that check s backing of vowels in the syllables of the plural forms This constraint is dominated by the positional faithfulness constraint outlined in [46] (4 8 ) I DENT V [ back] SP [ back] of the vowels of the singular form is identical to [ back] of the vowels of the plural In T ableau [1 4 ] below, candidate (a) alters the vocalic quality of the final syllable to [+back] incurring one violation to I DENT V [ back] SP However, the [ back] of the vowel in the initial syllable of candidate (a) is kept unaltered and matches the [back]ness of the vowel in the initial syllable of the singular. Therefore, it wins the competition. Candidate (b), a loser, has [+back] in the last syllable. However, the first syllable has a front vowel, altering the [back]ness of the initial syllable. Thus, it violates I DENT initial [ back] SP which is undominated, and it is doomed. Candidate (c) stays faithful to the out put singular and violates none of the proposed constraints. What rules it out? The following lines will answer this question. Tableau [ 1 4 ] I DENT initial [back] SP I DENT V [back] SP PL I DENT initial [ back] SP I DENT V [ back] SP b. *! c.

PAGE 166

166 In T ableau [ 15 ] below candidates (c ) and (d) fail to parse [+back] in the last syllable. They both impinge on M AX [+ back], a highly ranked constraint which marks plurality in these forms. Th erefore they are doomed. Candidate (d) goes to the extreme and surfaces with a front vowel in both its initial and final syllables. Thus, it also incurs a violation to I DENT initial [ ba ck] SP. Candidate (b) violates I DENT initial [ back] SP by altering the vowel in the initial syllable. It is out too. The optimal output (a) is faithful to both constraints equally. Tableau [ 1 5 ] M AX [+ back] I DENT initial [back] SP + [+ back] PL M AX [+ back] I DENT initial [ back] SP b. c. *! d. *! Since plural forms must parse [+back] at th e expense of losing identity of the vocalic quality in the final syllable, then M AX [+ back] must outrank I DENT V [ back] SP However, is there a strict ranking between I DENT initial [ back] SP and M AX [+ back]? Tableaux [ 16 ] and Tableau [ 17 ] below ex amine how crucial I DENT initial [ back] SP is in the ranking of the phonologically conditioned ableau [ 16 ], both the first and last syllables of the plural form have back vowels. The optimal output parses [+back] in the last sy llable but leaves the [ back]ness of the initial syllable intact. Thus, both I DENT initial [ back] SP and M AX [+ back] seem to be unrankable with respect to one another. T he optimal output in T ableau [17 ] also parses [+ back] in the last syllable. The init ial syllable has a vowel that matches the vowel in the initial syllable. forms. Ablauted plurals with two syllables do not show whether M AX [+ back] outranks

PAGE 167

167 I DENT initial [back] SP or whether they are unrankable with respect with each other. However, ablauted plurals which have a simplex base reveal a strict ranking between these two constraints. The subsequent analysis of ablauted plurals with a simplex base wil l illustrate that M AX [+ back] outranks I DENT initial [back] SP. Therefore, a strict line must be drawm between these constraints in a tableau of ranking. Tableau [ 1 6 ] M AX [+ back] I DENT initial [back] SP I DENT V [back] SP [+ back] PL M AX [+ back] I DENT initial [ back] SP I DENT V [ back] SP b. *! c. d. rer *! Tableau [ 1 7 ] M AX [+ back] I DENT initial [back] SP I DENT V [back] SP mim + [+ back] PL M AX [+ back] I DENT initial [ back] SP I DENT V [ back] SP mum b. mim *! c. mim *! d. mum *! To reiterate, T ableau [ 17 ] also reveals that the most important thing fo r the output form is to have [+back] in the last syllable; however, it requires identity of the vowel quality in the first syllable of both the singular and plural form. Thus, the overall ranking M AX [+ back] I DENT initial [ bac k] SP I DENT [ back] SP. the shape CVC. Tableau [ 1 8 ] M AX [+ back] I DENT initial [back] SP I DENT V [back] SP j t + [+ back] PL M AX [+ back] I DENT initi al [ back] SP I DENT V [ back] SP a. j t *! Candidate (a) realizes the [+ back] in the final syllable of these plural forms. Thus, it is selected as the optimal output. Candidate, (b), however, does not parse the

PAGE 168

168 featu re [+ back] which is a crucial requirement for marking plurality in these forms. Therefore, it is out. The ranking established for plurals with ablaut using positional faithfulness: This concludes the discussion of the plurals marked with ablaut. I assume that there is a feature [+ back] that targets the final syllable of the plural form and comes associated with the output singular forms. This feature is translated into the constraint M AX [+ back] that ensures the parsing of [+ back] into the output plura l forms. Because the vocalic change affects the last but not the first syllable of the plural, I further posit that I DENT initial [ back] SP outranks I DENT V [ back] SP. So, the final ranking established for these plurals is: (49 ) M AX [+ back] I DENT initial [ back] SP I DENT V [ back] SP Anti Faithfulness Constraints morphophonological tendency to observe a change from their morphologically related pair the singulars in t his case. This morphophonological tendency to be different is pursued by Aldrete (1999 a and 2001) who extends t he notion of faithfulness into a nti opposition between two morp Anti F aithfulness constraints account for morphophonological alternations that both faithfulness and markedness constraints alone may fail to sufficiently address. These include affix driven alternations like accent shift, deletion or retraction. Alderete also argues that encompasses a much wider range of phonological processes than Although he provides no direct analysis for ablaut within

PAGE 169

169 the anti faithfulness model, he assumes that ablaut can definitely be subsumed under this model. His evidence comes from stem vocalism in Arabic (McCarthy 1979 et seq). In discussing accent in Japanese, Aldrete (2001) shows that it has two types of suffixes: dominant accented suffixes outlines three major reasons for arguing that dominant u naccented suffixes follow from Anti F aithfulness: (1) morpholo gically triggered, (2) stem mutating and (3) grammar dependent ( Suffix accent forces contrast in a pair of related words since it makes two word classes contrast It also affects words in a paradigm which means its e ffect is on words sharing the sa me stem. Finally, its specific effects follow from the whole grammar of the language. marks plurality in morphologically related forms (singulars and plurals) in a paradigm. Singular and plural forms share the same stem and the change is induced by a contrast between these two classes. The anti faithfulness constraint integrates well in the ranking established to hold true for In other words, the rest of the grammar determines the surface structure and the constraints assumed to be inviolable in the language are not disrupted by the integration of anti faithfulness constraints. Anti faithfulness constraints impose a change by simply requiring a violati on to a faithfulness constraint. Alderete (1999a and 2001) formulate this theory as follows: (50 ) Anti Faithfulness (Alderete 1999a) For every faithfulness constrain t F, there is a corresponding anti faithfulness constraint F that is satisfied in a string S iff S has at least one violation of F

PAGE 170

170 To apply this theory li, I propose the constraints (51) and (52) : ( 51 ) SP I DENT V [+back]: corresponding vowels in the singular and plural forms agree in the feature of [+ back] ( 52 ) SP I DENT V [+back]: it is not the case that corresponding vowels in the singular and plural forms agree in the feature of [+back] If SP I DENT V [+back] outranks SP I DENT V [+back] and all other related faithfulness constraints then we expe ct a change in all the vowels contained in the plurals ; they are forced to be different than those in the singulars in terms of the feature [+back] the vowel in the first syllable of the singular form is kept unaltered in the first sy llable of the plural. Thus, the positional faithfulness constraint I DENT initial [+back] SP is not outranked and must in fact outranks the antifaithfulness constraint. However, the anti faithfulness constraint should in turn, outrank the general faithfu lness constraint checking the feature [+back] in the vowels of the plural forms Anti Faithfulness, though it very cogently addresses the ablaut plurals of two syllables, may not work so well for the analysis of ablaut plurals with a single syllable. Since the analysis shows that the positional faithfulness constraint Ident [+back] SP is not outranked and must outrank the other proposed anti faithfulness constraints, we are in a dilemma as to what the initial syllable is in those simplex ablaut pl urals. As illustrated, simplex ablaut plurals exhibit a change in the vocalic quality of the only vowel they have. Therefore, those plurals constitute a challenge to the Anti Faithfulness proposals despite its apparent success in accounting for ablaut plur als with two syllables. Therefore, the positional faithfulness approach is more successful.

PAGE 171

171 Tableau [19] and Tableau [20] show how the anti faithfulness constraint produces the surface form but not in isolation from the other crucial constraints in the lan guage. Tableau [ 1 9 ] I DENT initial [+back] SP SP I DENT V [+back] I DENT V [+back] SP I DENT initial [+back] SP SP I DENT V [+back] I DENT V [+back] SP b. **! c. *! ** d. rer *! Tableau [ 20 ] I DENT initial [+back] SP SP I DENT V [+back] I DENT V [+back] SP mim I DENT initial [+back] SP SP I DENT V [+back] I DENT V [+back] SP mum b. mim **! c. mim *! d. mum *! ** The optimal output constraints in Tableau [ 19 ] and Tableau [ 20 ] t olerate a single violation of the anti faithfulness constraint SP I DENT V [+back] at the expense of keeping the vocalic quality of the vowel in the initial syllable intact. On the other hand, the anti faithfulness constraint must outrank the gen e ral faithfulness constraint to ensure a change keeping the two forms distinct from one another. The suboptimal output (b) in both Tableau [19] and Tableau [20] violate SP I DENT V [+back] twice. Therefore, it is doomed. Changing the feature [+back] in all the syllables of the plurals and satisfying the anti faithfulness constraint does not solve the problem either as it incurs a fatal vi o lation to the positional faithfulness constraint which demands retention of the [+back]ness of the initial vowel of both the singular and plural forms. Thus, candidates (d) in Tableau [19] and Tableau [20] are doomed Since Anti F aithfulness constraints are as general as faithfulness constraints, they do not specif y a ce rtain location for the mutation or change. In a footnote, Alderete

PAGE 172

172 transderivational anti faithfulness constraint, and so other constraints in the grammar, including mar kedness, positional faithfulness, and positional antifaithfulness constraints, may have a role in pin This observation is crucial for the and it justifies adoption of a positional faithful ness co nstraint to rule out a change of [+back] in the initial syllable. Alderete presents many advantages for anti faithfulness constraints and shows its ultimate success in capturing a large set of morphophonological processes. Anti faithfulness c onstra ints successfully capture faithfulness constraint integrates well with the already established ranking and does not incur a mutation in the phonology of ablaut plurals. This is manifested through its dependence on the posit ional faithfulnes s constraint which outranks it. Summary of the ranking established for plurals with ablaut using anti faithfulness constraints: In co n clusion, I presented an alternative analysis to ablaut claims that Anti faithfulness provides a cogent analysis to morphophonological processes. They are grammar dependent and should be in harmony with the ranking believed to be true of a particular language. Thus, the y do not disrupt the ranking which assumes to operate in the l anguage as a whole. They are general and do not specify which vowel in the plural form should look distinct from the one in the singular form. A p ositional faithfulness constraint determines the locus of the contrast in the vocalic change. Since a change must be realized in the plural forms, then the anti faithfulness constraint outranks the general faithfulness constraint. The overall ranking is : (53 ) I DENT initial [+back] SP SP I DENT V [+ back] I DENT V [+back] SP

PAGE 173

173 RealizeMorpheme I showed two distinct approaches (Positional Faithhfulness and Anti Faithfulness) Faithfulness, in particular, successfully addres ses two distinct shapes of ablaut plurals: those with simplex bases and those with two syllables. However, as discussed above, Anti Faithfulness, fails to address ablaut plurals with a single syllable. Now I offer a third altenative to the analysis of plu rals with ablaut. It reveals a third mechanism supplied by Optimality Theory that can successfully explain ablaut plurals. This mechanism is RealizeMorpheme 8 which posits that a change (be it a morpheme affix or internal change) must be realized in the out put form when expressing a particular semantic or morpho ablaut plurals, plurality, a morph syntactic function, is expressed by realizing [+back]: the plural morpheme. Therefore, a constraint such as Realize Morpheme or Realiz e[+back] will certainly yeild the optimal output plural. I formalize and define the constraint as follows: (54 ) R EALIZE M ORPH the plural morpheme [+back] mus t be realized in the output plural R EALIZE M ORPH makes sure that the feature [+back] is parsed in the output plural. It outranks the positional failthfulness constraint I DENT initial [back] SP by ensuring that the plural morpheme:[+back] must be realized at the expense of altering the vocalic quality of the vowel contained in the initial syllable This is especially manifested by the change in the only syllable of the simplex ablaut plurals. In the previous analys e s, I have also proven that I DENT initial [back] SP must, in turn, outrank the constraint 8 I supplied a formal definition of RealizeMorpheme in (37) above.

PAGE 174

174 keeping identity of [back]ness in the vowe ls of the output form. Tableau [21] showcase s the two distinct shapes of ablaut plurals: simplex and complex. Tableau [2 1 ] R EALIZE M ORPH I DENT initial [back] SP I DENT V [back] SP j t + [+ back] PL R EALIZE M ORPH I DENT initial [back] SP I D ENT V [back] SP a. j t *! Tableau [2 2 ] R EALIZE M ORPH I DENT initial [back] SP I DENT V [back] SP PL R EALIZE M ORPH I DENT initial [back] SP I DENT V [back] SP r b. *! c. *! ** d. *! Tableau [21] and Tableau [22] show that the winning candidate (a) must realize the plural morpheme: [+back] at the expense of violating I DENT initial [back] SP (simplex ablaut plurals) and I DENT V [back] SP (simplex and complex ablaut plurals). They also show how the involvement of R EALIZE M ORPH serves to provide a unified Although the Positional Faithfulness approach and the Realize Morpheme approach successfully account for simplex and complex ablaut plurals, there is an empirical difference between these approaches. While the Positional Faithfulness constraint checks the feature [back] at a particular position in the plural form, Re alizeMorpheme is vague in that it does not specify a particular change at a certain locus. Alderete (2001) argues against the use of an alternative R EALIZE M ORPH because 2001:13). R EALIZE M ORPH only stipulates that a plural form must be different in [back]ness from its singular form. Thus, a potential plural form, whose any of its vowels

PAGE 175

175 is different in [back]ness from the vowels in the singular form will satisfy RealizeMo rpheme, and may wrongly assumed to be optimal. Summary of the ranking established for plurals with ablaut using realize morpheme: I have argued for a third approach for the analysis of ablaut plurals in which ensures the realization of some sort of change (the feature [+back] in this case) in the output plural. RealizeMorpheme, like Positional Faithfulness, successfully addresses two distinct types of ablaut plurals, and nicely accords with the proposals made for the ablaut plurals. The overall ranking of the ablaut plurals based on the RealizeMorpheme analysis is: (55 ) R EALIZE M ORPH I DENT initial [back] SP I DENT V [back] SP. Summary of Chapter 5 Chapter 5 has offered an integrated analysis to th e most regular and phonologically conditioned the Vb infix, then the analysis is extended to account for plurals with a suffixal template VC and templatic plurals derived from geminated singu lar forms. A set of well motivated constraints in Optimality Theory successfully capture s the systemicity in these plural shapes. Later I presented a Positional Faithfulness theoretic account for the ablaut plurals and offered an alternative analy sis usin g A nti F aithfulness theory. The last section of Chapter 5 uses RealizeMorpheme to tackle ablaut plurals, a third approach

PAGE 176

176 Below I repeat the different rankings established for the regular plural p atterns of present a ranking that holds true for the grammar of noun plurality in the language. (56 ) ( a ) L A NCHOR PS, A LIGN V b L, O NSET R A NCHOR PS ( b ) M AX V S UFFIX N O V C OMPLEX M AX V R OOT ( c ) I DENT Q *VGG# I NTEGRI TY SP N O V D EP V SP, *C OMPLEX ( d ) M AX [+ back] I DENT initial [ back] SP I DENT V [ back] SP ( e ) I DENT initial [+back] SP SP I DENT V [+back] I DENT V [+back] SP (f) R EALIZE M ORPH I DENT initial [back] SP I DENT V [back] SP The const raints which appear to be violated frequently by the analyzed noun plurals are C OMPLEX M AX V R OOT and D EP V SP C OMPLEX and M AX V R OOT are violated by the plurals attaching a suffixal template VC and the templatic plurals derived from geminated singular s D EP V pro s thetic vowel word init i ally to break up a cluster of two consonants and when a vowel makes a geminate in the base singular long distance in the plural form The root vowel is also sacrificed in the ablaut syllables to Max [+back], which supports the fact that M AX V S UFFIX outranks M AX V R OOT One of the faithfulness constraints that is strictly obeyed by the plural forms is M AX V S UFFIX At least two plural types ( Vb infixed plur als and those attaching a VC template) fully adhere to this constraint and forbid the loss of the vowel in the affix. This sometimes happens at the expense of losing the vowel in the root. Finally, the constraint N O V factors in the formation of plurals at taching a VC template and those derived from geminated singulars. It outranks M AX V R OOT tendency to delete the root vowel. However, since this constraint radically gets rid of

PAGE 177

177 any vowel in a plural form its effect has to be minimized by the constraint M AX V S UFFIX which outranks it. The ranking that appl ies to the diverse noun plural patterns is: M AX V S UFFIX N O V *C OMPLEX M AX V R OOT D EP V SP. Tab l eau [23] shows how these common constraints interact in the formation of differe nt noun plurals and the ir sub patterns in Tableau [2 3 ] M AX V S UFFIX N O V C OMPLEX M AX V R OOT D EP V SP Vb infixed *** *** with VC template ** ablauted 3. s ** ** templatic ** 4a. nbeb Ratcliffe (1998: 202) mentions that has the suffix i that take s ablaut in its plural formation (57) : (57 ) a. xarfi monsoonal bedouin In the above examples the final vowel of the adjective plurals is [back]ed similar to the noun plurals that take ablaut. This means that the ranking stipulated for the ablauted noun pl urals may be extended to apply for the adjective plurals too. The feature [+back]

PAGE 178

178 need s to be parsed in the output, yie lding a back vowel in the final syllable of the adjective plurals. I summariz e the ranking I establish for the grammar of noun plurality form of a lattice: (58 ) M AX V S UFFIX N O V *C OMPLEX M AX V R OOT D EP V SP Indeed, Optimality Theory, with th e tools and constraints it entails, can successfully offer an integrated analysis to these diverse noun plural shapes. The final ranking of constraints conforms to the prosody and grammar of the language as a whole. Moreover, I avoid the stipulation of any templatic constraints and assume that the interaction of phonological and prosodic constraints derive the final template, following the proposals outlined in Generalized Template Theory. The analyse s of Positional F aithful ness constraints and RealizeMorpheme elegantly address some of the morphophonological tendencies of these plurals and integrate

PAGE 179

179 CHAPTER 6 EXCEPTIONAL PLURAL SHAPES In Chapter 5 I have present ed a thorough discussion of the most common and motivated constraints in Optimality Theory. The proposed analysis captures the regularity of these shapes, addresses their d iversity and reveals how a set of crucial ly rankable constraints serve s as a clue to the grammar of noun violate M AX V R OOT and C OMPLEX at the expe nse of obeying some other constraints such as M AX V S UFFIX that govern s the shape of the plural marker. The noun plurals accounted for include plurals with Vb infix, plurals attaching the suffixal template VC consonant in the base, templatic plurals derived from geminated singulars and ablaut plurals. Like other plural formation processes world wide, the process of plural formation in volves exceptional and irregular plural shapes that pose a challen ge of incorporating them into the proposed O ptimality T heory analysis. For instance, in addition to the suppletive, templatic and miscellaneous shapes for which it is hard to establish a general mechanism, some of the regular plural patterns have a few sub patterns which do not follow the general procedure for forming their regular patterns. To illustrate, under the Vb infixed plurals, there is a sub pattern which involve s some inexplicable peculiarity which clashes with the phonological properties of the r egular Vb infixed plurals. Moreover, where regular ablauted plurals parse [+back] in the final syllable only irregular ablauted plurals diverge from the prevalent mechanism and parse [+back] in both the initi al and final syllables.

PAGE 180

180 Chapter 6 discusses the addresses the divergent phonology and morphology of the sub patterns for whose patterns an integrated phonological analysis has been proposed. In particular, a discussion of how the unusual shape s which take the Vb infix cannot be incorporated into the analysis of their regular pattern is presented Then, I will discuss the exceptional shapes starting with the templatically expanded plurals. Later, I will deal with the truncated plurals, templat ic shapes and miscellaneous forms I also discuss a group of exceptional noun l offer a thorough discussion of the plurals that stack more than one plural marker in the plural forms to mark plurality Chapter 6 also explor es a number of Optimality Theoretic accounts that deal with exceptionality and lexical marking. Some of these approaches devise extra theoretical tools which will take time and proof to integrate well into the model of Optimality Theory. Others suggest mod ification to the existing Optimality Theory framework. The approaches listed in Chapter 6 have been used to address morphophonological and lexical phenomena observed in some languages. They will be applied to account for exceptionality of plural formation in Sub P attern of Vb Infixed Plurals In addition to the regular Vb infixed noun plurals for which an integrated O ptimality T heory account was supplied in Chapter 5 a sub pattern that greatly diverge s from the regular Vb infixed plu rals and pose s a challenge to the proposed analysis This sub pattern of the Vb infixed plurals is derived from bi consonantal and tri consonantal singulars compared to the fixed quadri consonantal singulars (CVCCVC) from which the regular Vb infixed plu rals a re derived. The singular form (1 .a) has two consonants and map s onto a plural shape C 1 bVC 2 with an infix {b}

PAGE 181

181 after the first consonant. Consonant clusters word initially are allowed but why is the V of the Vb infix lost in these forms? Is it because the singular form from which this plural is derived has an irregular shape too? The singular form (1 .b) is tri consonantal with a consonant cluster word initially. The plural form inserts a pro s thetic vowel and is shaped into VCCbVC, a shape divergent fr om the regular Vb infixed plural. The singular forms (1 .c) and (1 .d) are tri consonantal but each map on to distinct shape s While the plural form (1 .d) takes C CbVC, form (1 .c ) map onto the regular shape of the Vb infixed plural, with an extra /m/ resurfaci ng. I p ropose that the singular form (1 .c) has underlyingly an initial /m/ which deletes and then re appears in the plural form This word is most probabl y borrowed from Arabic [malgam] deletes the initial nasal /m/. However, contrary to the typical trend, it does not nasalize the vowel following it. Instead, it deletes it along with the /m/ preceding it. (1 ) Plurals with (V)b infix of varying shapes news drums muzzles Zizyphus Spina Christi Due to the vast diversity of these shapes, it is hard to tell what the underlying singular shape from which these irregular Vb infixed plurals are derived as there is not a single one. Besides, the V of the Vb infix is lost and there is no obvious phonological explanation that conditions this los s. Moreover, the plural forms (1.a) and (1 .d) have a consonant clusters plural initially which the other shapes do not have. These phonological properties make it impossible to integrate these plurals into the proposed analysis. Various rankings of constraints would be needed for each individual case,

PAGE 182

182 which would undermine a core element of Optimality Theory which stipulates a single ranking in the whole grammar. T emplatically Expanded Plurals In these noun plurals, plurality is marked by the appearance of an extra consonant or the re appearance of a consonant which may be deleted in the singular forms. F or instance, the plural forms (2.a) and (2 .b) below have an ex tra sound /l/ word medially which the singular f orms lack. In the noun plural (2 .c), a /j/ appears after the first closed syllable C 1 VC 2 The plural forms (2 d g ) have eclecti c consonants reappearing or app earing in the plural forms: /m/ word initially whic h is not present in the singular form (form 2d) /w/ after the initial C in the singular forms (form 2e) the re appearance of a deleted /m/ and /n/ (form 2f) and /j/ after an open syllable C 1 V (form 2g) I assume that these forms underlyingly have a conso nant that is deleted in the singular form. However, t here is a variation in the type of consonant s deleted in the singulars or inserted in the plurals, which pose s an insurmountable challenge to offering an integrated O ptimality T heory analysis to these fo rms. It is hard to propose an underlying shape from which these diverse plurals are derived. Moreover, the locus of the insertion is not fixed throughout the whole forms. The singulars, from which these noun plurals are derived, range in shape from bi cons onantal (2 .a, b, and e) to quadri consonantal (2 .g). However, the templatic shape these plurals prefer to map onto match the prosody and the phonological tendencies of the fact that onsetless syllables are licensed to occur word initially and consonant clusters may also occur.

PAGE 183

183 ( 2 ) Templatically expanded plurals windows b. kob kolob dogs young bulls sweethearts r e:kwar chiefs t orphans (f.) nights In Optimality Theory, the fact that there is an underlyingly extra consonant in the plural form which is inserted in the singular form, is translated into the violable faithfulne ss constraint D EP C. However, it will be hard to step beyond this statement; there is no specific and phonologically fixed segment that gets inserted in the singular forms nor is there any obvious motivation for inserting them Another problem lies in the locus of the insertion Although O ptimality T heory supplies theoretical tools for addressing the position of the inserted segments these plurals will require abundant constraints to address their various locations. Thus, no cogent single analy sis will succeed in capturing them all. Truncated Plurals Opposite to the process described above, these plurals involve deletion of some sort to indicate plurality. However, the locus of the deletion is different in each form making it hard to express i t systematically Since the language is known for the vast deletion o f consonants such as the nasal {m}, {w} and {b} it is not uncommon for {m} or {b} to get deleted frequently in the data collected. F or instance, the plural forms (3 a c) del ete {b} which occupies the third C slot in the sin gular forms. The plural forms (3 d f) delete {m} ; forms (d) and (f) also delete {m} which resides in the third C slot of the singular forms. The plural form (f) loses the initial .mo. syllable and retains the lateral fric ative in the plural form (the only plural with a feminine marker). Finally, the plural

PAGE 184

184 forms (g) and (h) below delete one of the long distance geminated consonants in the singular forms. (3 ) Truncated Plurals a. e:s e:s fingers c artridges t fans t pillars masor nails livestock sings marked on the ground h. t t characters Just like the templatically expanded plural forms, the natu re of the truncation plus the locus of the deleted consonants is diverse and there is no fixed and easy to establish pattern that can be used to come up with a cogent O ptimality T heory analysis to these forms. Moreover, it is not obvious what exactly cond itions the deletion of a consonant in these forms. The deletion does not happen to resolve a syllable shape or for a syllabic phonotactic purpose The templatic shapes onto which the singulars are mapped vary greatly. Some forms begin with a vocalic elemen t, others with a consonant clusters and yet a few more with a single consonant plus a feminine marker which is believed not to contribute to the meaning of the consonantal root. The only obvious M AX C. Beyond this aspect, O ptimality T heory cannot handle uncertainties about the indefiniteness of the shape and locus of the deleted segment. Templatic Plurals Singular forms of various shapes can be mapped onto three templates to mark plurality. These templates are CVCVC, CVCC and CCVC. Despite being able to determine the template, it is hard to establish a particular mechanism for forming these plurals. Moreover, no phonological condition is discernable to state what shape of a

PAGE 185

185 singular form maps onto which template. There is also nothing special about the phonology and morphology of the singular forms which make them map onto these templates and not take other systematic plural formation mechanisms. The data below reveal that t hese noun plurals are derived from diverse singular forms which take different templates like VCCVC ( 4.c), CCVC (4.a, 6.d) and CVCC (4b d, 5a; 5c and 6 a; b). However, mostly tri consonantal singulars map onto templatic plural shapes. ( 4 ) Plurals taking the template CVCVC female possessors cheeks sahab waves at palm trees ( 5 ) Plurals taking the template CCVC ut tears skun communities et holes bored in the ear lobes ot honeycombs ( 6 ) Plurals taking the template CVCC bread ot stories There is no correlation between the shapes of the singular and plural forms. Moreover, a statement about whether the consonants contained in these forms play a role in deciding which template a singular map s onto cannot be stated. The only intriguing property of some of the listed words is that the l oan words fro m Arabic (forms 4 .d, 4 .e and 6 .a) are pluralized into the same templatic shapes (CVCVC and CVCC)

PAGE 186

186 M iscellaneous Shapes belong to any of the above plural shapes. These plurals have extra phonology and morphology on them which is inexplicable. For example, some of the plurals belong ing to this group have a consonantal shift (form 7 .c) which changes the /t / in the singular form to /b/ in the plu ral form. In the plural forms (7.b) and (7 .d), a pro s thetic /e/ gets inserted before /r/ word initially and the final shape they take is V 1 CV 2 C, whereby V 1 is /e/. The plural form (7 .a) loses the masculine suffix in and the feminine suffix at in the plural form and gets attached to /o/. ( 7 ) Miscellaneous shapes at ghosts heads beads d. months Such diversity is hard to carve into a unified O ptimality T heory analysis. Just like the above irregular plural patterns, this pattern has no definite underlying form and definite plural marker. The shapes these plurals map onto are diver se in nature which poses a challenge to proposing a unified set of constraints and ranking. Suppletive or lexical p lurals sounds of both the singular and plural forms are h ighly distinct and the semantics of these plurals is restrictive as they mostly relate to human beings such as women, boys, sons, daughters and babies or infants. Moreover, some of these plurals are attached to plural suffixes like i ti and Vn Howeve r, the whole shapes of both the singular and plural form s are unrelated to one another. It is, thus, hard to tell which plural a singular form may take.

PAGE 187

187 ( 8 ) Suppletive plurals a. bri j ni sons ti daughters women Plurals e consonantal ( 9 .d) or tri consonantal ( 9 .a c) singulars. There is no justification for mapping these singulars onto this shape and not other shapes of plural s; the phonology and morphology of these shapes do not condition them to map onto this shape. Besides, some plural forms delete a consonant in the singular forms like the plural forms ( 9 .a) and ( 9 .b) below which delete the final /b/ in the singular forms. The plural forms ( 9 .c) and ( 9 definite shape from which these plurals are derived. Moreover, the plurals do not have a single templatic shape and vary in the number of consonants they each have. Some plurals start with a consonant clusters while others hav e a nice .CV. CV syllable shape With all this intricate diversity, these shapes pose a challenge for an O ptimality T heory to account for them. ( 9 ) corners b carved wooden dolls at monitor lizards et beards Doubly and triply marked p lurals row to mark plurality. T he majority of this type of plurals attach two plural suffixes namely Vn and tV in this order as in forms (10 .d f) below. Other plural forms have the infix (V)b plus the default suffix tV A few other plurals attach the (V)b infix along

PAGE 188

188 with the Vn suffix like form (10 .g). In the collected data, I only find one plural form whi ch is triply marked; it has two plural suffixes and the (V)b infix preceding them. Just like the other irregular shapes, the singular forms from which these plurals are deri ved are so diverse in nature ; they can be bi consonantal (10.a,g), tri consonantal (10 .b,c, d and f) or quadri consonantal (10 .e and h). They may have consonant clusters word initially or simple CV syllable shapes. Moreover, there is no phonological condit ioning which serves as an indicator for these singulars to take more morphologically the singular forms which derive these plurals look the same as other singulars de riving other plural patterns. Forms (10 d f) are loan words from Arabic; these take broken plurals in Omani Arabic (saja:ji:r, kara:fi and zawa:li respectively). The diversity of the bases makes it hard to determine a unified underlying representation for these forms. Also, there is no obviously disti n ct morphology that would urge form (1 0 .h) to take three plural markers and not just two like the other forms. Therefore, these constitute a problem for an integrated Optimality Theoretic analysis to be propose d. ( 1 0 ) Doubly and triply marked p lurals b monitor lizards b nipples c. t t b monitor lizards un cars un beds f. zol it zol un carpets b r in graves mi ab l un chameleons Approaches to Exceptionality in Optimality Theory The previous sections have collaboratively highlight ed the peculiar phonology and morphology of

PAGE 189

189 they impose in proposing a cogent account for them using the principles of Optimality Theory To recapitulate, there are five major problematic issues that these plurals confront us with: 1 T here is no direct underlying form from which these plurals can be straightforwar dly derived. Each plural form seems to be derived from an underlying representation distinct in shape from those deriving other plurals belonging to the same category. 2 The changes happening in the plural forms are eclectic in nature that renders it hard t o propose a definite set of constraints that deals with al l the changes. For example, the inserted and deleted segments are diverse and the locus of the insertion and deletion is not the same across the board 3 There is no motivation for inserting or delet ing segments ; recourse to obvious syllabic phonotactic restrictions to explain the reasons cannot be determined 4 Some singulars which are borrowed from Arabic have undergone co nsonantal deletion {m, b, or w}. The plurals of these forms retrieve these co nsonants for no plausible phonological reason. For these particular plurals, complexities arise. Are their consonants? Is there a n e surface before the d erivation of the plural forms? The answers to these q uestions are complicated and require quite a complex mapping between different forms of singulars and plurals Observe the following representation: (11) /Sing. with C/ /Sing. with C plural marking/ O constraints Constraints requiring C to surface [Pl. formation with C] S P constraints 5 Along the same line of thoughts raised in point (4 ), markedness constraints in Optimality Theory assess the output forms alone and do not care about the structural changes happening to the input. Moreover, the O ptimality T heory account abandons reference to intermediate levels of representations.

PAGE 190

190 is risky since the site of insert ion could be the same as the deletion site in the base singular forms. Satisfactory answers to these broad issues continue to raise debates among phonologists and Optimality theorists. Kager (1999) lists a number of O ptimality T heory approaches that have been proposed to advance cases which cannot be resolved within the current model of Optimality Theory. I will only list some of the relevant approaches to dealing with lexical marking and exceptionality in the Optimality T heory framework The approaches di scussed pertain to the problems faced when dealing with the exceptional noun proposed to tackle these problems are eliminating the underlying representation, R EALIZE M ORPH two level well formednes s and multi level well formedness a realization model of Optimality Theory s pecification of the exceptionality in Lexical Entry in a form of diacritic containment approach, constraint indexation and selector constraint and RealizeMorpheme As I examine these approaches, I will discuss if they are consistent with the analysis of the more regular plurals. Eliminating Underlying Representation and RealizeMorph As Optimality Theory eradicates re write rules and abandons the serial orderings of generative pho nology, it retains the assumptions that one underlying representation derives the surface forms. However, m any surface forms do not show a one to one relationship between a unique underlying representation and its allomorphs. Other surface forms like the e do not seem to be derived from a single underlying representation. Thus, they conflict with the notion of a uniform underlying representation. To solve this issue, attempts to eliminate the underlying representation f rom the m odel have been made ( Russe l 1995; Burzio 1996) Kager

PAGE 191

191 (199 9 ) argues that elim in ation of the underlying representation will consequently eliminate abstractness since the input will be identical in every respect to the output form, ignoring structural change s. The output forms will explain observed alternations and t he output form is then selected based on its being harmonic to the ranking of constraints assumed to be active for the grammar of that language or the grammar of the linguistic phenomenon under st udy This, in turn, will reduce the role of grammar to 199 9 :414) supplies a unique UR for each morpheme, but instead it supplies a set of shape variants of the morpheme, allomorphs, ch unks ready for insertion in various morphological 9 :415). This model is observed to prov ide a conceptual account for neut ralization and allophonic variations. To extend this solution to the truncated and templatically expanded plural f orms in li, I assume that every distinct truncated or templatically expanded plural form is not derived from a concrete u nderlying representation. Thus, the different phonology characterizing each output form is derivable from the competition of avail able which will select the optimal output. Constraints such as *Complex, Onset, M AX C and D EP C are violated for these plural shapes. R EALIZE M ORPH ( Samek Lodovi ci 1994 ) may solve the problem temporarily since there is always a force that coerces the output to be distinct from the input. Besides, R EALIZE M ORPH is a constraint highly ranked in the grammar of plural formation in ; almost Tableau [1] and Tableau [2] below reveal conspiracy between faithfulness constraints and the force to realize a certain change to mark plurality for the truncated and templatically

PAGE 192

192 candidates for the input form. Tableau [ 1 ]: Truncated p lurals R EALIZE M ORPH M AX C Candidates R EALIZE M ORPH M AX C a. b. *! Tableau [ 2 ]: Templatically expanded plurals R EALIZE M OR PH D EP C Candidates R EALIZE M ORPH D EP C a. b. *! Alderete (2001) present s some proble ms of adopting R EALIZE M ORPH in the analysis of irregular phonological processes or morphophon ology in general R EALIZE M ORPH is unable to capture allomorphy and incapable of distinguishing between two distinct allomorphs pertinent to a single morphological phenomenon. For instance, Alderete mentions that there are two distinct patterns of subtracti ve morphology in volved in the formation of Koasati plurals These two patterns have totally different order of constraints. R EALIZE M ORPH in light of being contentless and abstract is unable to describe and distinguish between these two types of allomorph y. Different Koasati plurals end up attaching the wrong allomorph when R EALIZE M ORPH is used since it can license any kind of change in a form Moreover, R EALIZE M ORPH has a number of conceptual problems. For instance, Kurisu (2001) assumes that the change imposed in the output form results from morphology. This entails that morphemes may produce marked structures which cannot be extended to or supported by the phonology of the language as a whole. Consider, for example, the formation of deverbal nouns in Ic elandic which produce s more marked structures that violate *Complex Coda (a

PAGE 193

193 constraint which is completely obeyed elsewhere in the language ) Finally, RealizeMorpheme is abstract and is satisfied by any sort of change in the output form. In the analysis o f the regular plural patterns in Chapter 5 I assume an Output Output correspondence between the singulars and their plurals. I further illustrate that UR is not enough for the derivation of these plurals. For the templatically expanded and truncated plura ls, I eliminate UR completely and assume that the output plural results from a competition between a rankable set of constraints. Ther e is a high possibility that these exceptionally shaped plurals are derived from their Arabic singular forms. So, their U R correlates with their Arabic output singulars. However, this is a very radical idea since in Optimality Theory correspondence always happens between forms in the same language. I argue that elimination of the UR is fairly consistent with my analysis of t he regular plurals. First and foremost, the elimination of UR does not contradict my their URs. Second, the exceptional plural output, similar to the regular plural, res ults from a competition of constraints whose minimal violation determines the winner. However, what seems to be problematic here is the lack of specific Max C and Dep C (i.e. I need to specif y the type of deleted and inserted segments) in order to avoid ra segments is diverse and there is always a risk of an unrestrained function of Max C and Dep C. Two level W ell formedness Kager (1999:378) dis cusses an approach that permi ts reference to the input by direct well formedness constraints. level well that its effects are similar to those in Correspondence Theory with the exception that the

PAGE 194

194 latter admits faithfulness and not markedness co nstraints. In accounting for opacity of vowel harm ony whose triggering vowel is sy ncopated and left a trace on an adjacent vowel in the output form in Tunica Kager proposes a constraint that refer s to a different input vowel because the output vowel resem bles ano own H e formalizes this constraint as follows: (12) H ARMONY IO If input V 1 2 2 agree in backness and rounding 2 Constraint (12) states agreement between vowels in backness and roundness at different levels of representation. This agreement is carried out by the direct correspondence between V 2 2 For the deleted consonant I propose the constraint M AX O O but extend their correspondence to two levels Since the Je have consonants that correspond to the consonants in the Arabic singular s which when this constrai nt as follows: (13) M AX OO the segments in the plural correspond to the segments contained in the Arabic singular To illustrate, instead of having the deleted segment in the output singular s correspond to the segments in t heir output the constraint will evaluate the plurals with the singulars in Arabic. In such a case, the plurals are paralleled to different Underlying singular forms those in Arabic. According to K a ger, these two level constraints are not without any problems. For example, they fail to address types of opacity that neither r elate to the input nor to the output. They also fail to ref e rence opacity occurring at the prosodic level since

PAGE 195

195 prosody is absent at the input level. One renowned example is compensatory lengthening whereby a coda is deleted and the preceding vowel is leng thened Moreover, they function like rules because they stipulate a structural change and a repair at the same time. They are unrestrictive in that they may posit any type of change. Finally, they blur the distinction between markendness and faithfulness c onstraints (Kager 1999:412). F urthermore, f had to extend these two level constraints across two radical idea. Similar to the previous approach to exceptionality, two level well formed ness assumes correspondence between elements belonging to two languages. This certainly removes restrictiveness from the approach, and undermines the Arabic correspondents. This approach is inconsistent with the proposals made for the more regular plurals in the language. M ulti level W ell formedness or Intermediate L evels This approach to exceptionality assumes that the grammar of Optimality Theory is organized into multipl e levels; each has its own functions of Gen and Eval The output of the previous level serves as an input to the subsequent level. The ranking of lving only re ranking of a well formedness cons This model can capture word domain effects, affix ordering, structure preserving and cyclicity (Kager 1999: 382). The plurals with three plura l markers take the Vb infix along with two plural suffixes un and tV in this order while those with two markers stack the two suffixes one after the other. Based on

PAGE 196

196 the multi level I propose that each plural marker gets attached at a different level. Th e order of the affixes in these doubly and triply marked plurals result from well I will present ab l un ] ons (14) Level (1): Underlying Form (1) R A NCHOR PS, L A NCHOR PS A LIGN V b L, O NSET ab Level (2): Underlying Form (2) ab O NSET R EALIZE S UFFIX 1 M AX V ab l un ] Level (3): Underlying Form (3) ab l un ] O NSET F INAL C ab l un t ] The forms belonging to this type of plural formation are not systematic and each doubly marked plural will require a representation different from the above sketched one. Moreover, their phonology varies greatly when compared to the phonology of the r egular plurals attaching the same plural markers. For example, the locus of the Vb infix is different from where it resides in the regular forms ; it is infixed after t he third and not the second consonant at the left edge of the plural form Form (b) in T a bleau [3 ] below is more harmonic than the opt imal output form (a). For candidate (a) to be more harmonic than candidate (b), some constraint must outrank A LIGN V b L. This constraint

PAGE 197

197 cannot be a well formedness constraint ; both candidates have equally one open CV syllable and two close d CVC syllables. So, their syllabic composition is basically the same. Since the Vb infix is aligned to the right edge in these exceptional Vb infixed plural f orms, I use the constraint Align Vb R ( Vb is aligned to the right e dge of plurals), instead of Align Vb L which is used in the regular Vb infixed plurals. Tableau [ 3 ] R A NCHOR PS, L A NCHOR PS A LIGN V b R O NSET Vb R A NCHOR PS L A NCHOR PS A LIGN V b R O NSET a.b b. mi. ab c. ab m *! m d. mi ab *! to avoid making an onsetless syllable and the second level plural marker a ttaches: Tableau [ 4 ] O NSET R EALIZE S UFFIX 1 M AX V un O NSET R EALIZE S UFFIX 1 M AX V ab. l un b. mi. ab un *! c. m *! The output to level (2) is fed as an underlying representation for the third level. No phonological changes are imposed into the final output plural in level (3) except the attachment of a third level plura l marker. un comes linearly before due to the high ranking O NSET The optimal output below violates F INAL C by ending in a vowel Tableau [ 5 ] O NSET F INAL C un + O NSET F INAL C ab. l un b. mi. ab .un *! One desirable consequence of this approach to exceptionality is that the Vb infix is derived from t he same set of constraints that derive the regular Vb

PAGE 198

198 infixed plural. So, it is consistent with the analysis established for the regular Vb infixed plurals. As other plural markers get attached to these exceptional plural markers, I need to refer to the pr osodic and syllabic well formedness canons of the language to rule out potential plural outputs. This is consistent with the proposals made for the more regular plurals. However, I may need to tackle each and every doubly marked plural individually since d ifferent affixes attach to different forms. Therefore, a unified analysis to all the plurals which attach two to three plural markers is impossible to make. Realization Optimality Theoretic Account to Mulitple Plural Markers Xu and Aronoff ( 2011 ) develop a Realization Optimality Theoretic account for and blocking. Extended exponence occurs when multiple exponents in a word realize the same morphosyntactic function Blocking, on the other hand, bans the realization of multiple exponents expressing a single feature value of Xu and and apply it to doubly marked plurals in The core spirit of their analysis relie s on the markedness constraint F EATURE S PLIT which militates against the multiple realization of a single morphosyntactic value. F EATURE S PLIT favo rs simple exponence which is a less marked tendency cross linguistically (pp.2 3) Thus, when F EATURE S PL IT is ranked lower than two or more realization constraints, extended exponence or multiple morphosyntactic markers are realized in a single form. In some instances, F EATURE S PLIT m ay rank between two competing exponents. Xu and Aronoff successfully prov ide a unified account to Tamazight Berber and Classical Arabic extended exponence morphology using an inferential realization model

PAGE 199

199 of morphology within Optimality Theory Accordingly, grammatical functions which realize morphosyntactic features need to be posited through realization constraints. The basic formalisim for realiz ation constraints is shown in (15) : (15) {Morphosyntactic feature} : {Morphophonological form} (Xu and Aronoff 2011 : 7) In their analysis, Xu and Aronoff also c onstraint Ranking realizing a feature set outranks another constraint re alizing a non null subset of the features (16) anking Suppose these constraints are part of a constraint hierarchy CH, and that G is acti ve in CH on some input i Then if G >> S, S is not active on i (Prince and Smolensky 2004: 99) Realization constraints may also specify the position of a morph by encom pas sing both realization and align ment constraint. Thus, a realization constraint like { noun plural} : tV may be decomposed into { noun plural} : tV which states that the plural morph tV must be realized and an alignment or morphotactic constraints which sp ecifies that it must be realized as a suffix. with double plural markers are exemplified in ( 1 9 ) below The majority of these plurals attach the default plural suffix tV and the less common suffix Vn I assume the realization constrai nts to address this particular shape of plurals : (17) {noun plural} : un which stipulates that the noun plural is realized by the morph un which must follow the singular base and precede another morph. (18) {noun plural} : tV which states that the noun plural is realized by the morph tV which must be a suffix.

PAGE 200

200 (1 9 ) Plurals with double e xponents un cars un beds c. zol it zol un carpets Since both these plural markers co occur in a single plural form, t hey both must outrank F EATURE S PLIT The ranking of the above realization constraints with respect to one another is indeterminate as I cannot establish evidence showing that the former outranks the latter. However, the realization constraints conform to the specifity condition which requires a constraint with more specific morphosyntactic or semantic F EATURE S PLIT is ranked lower than those realization con straints to allow for the occurr e nce of doubly marked plural forms. (20) {noun plural} : un {noun plural} : tV F EATURE S PLIT Tableau [ 6 ] { PL }: un { PL }: tV F EATURE S PLIT zol it, un tV { PL }: un { PL }: tV F EATURE S PLIT a. zol un b. zol un *! c. zol un *! d. zol *! Candidate (a) realizes plurality by splitting the morphosyntactic features for plurality into two, violating the low ranked constraint F EATURE S PLIT Candidate (b) realizes the plural marker un as a s uffix (where it should be an infix), while candidate (c) shifts the order of the plural markers Both violate the specifity conditions stipulated by the realization constraints which requires specific positions for the extended exponents. Candidate (d) fa ils to realize the realization constraint { PL }: un Thus, it is out.

PAGE 201

201 Xu and Aronoff also argue that a candidate such as (d) above which does not realize all the required morphosyntactic elements can be ruled out by F AITH constraints. They further stip ulate F AITH 1 and F AITH 2 which associate with each morphosyntactic F AITH 1 relate to un while F AITH 2 relate to Vt Therefore, F AITH 1 must out rank F AITH 2 to rule out the opposite order of these plural markers. Moreo ver, they argue for the constraint P RIORITY which can also rule out any candidate that does not stack the plural m arkers based on their order in the actual surface plural form. According to Xu and Aronoff, this approach to realization morphology is langu age specific in that not all languages allow for extended exponen c e in their grammar. This entails that realization constraints, which are in essence morphologically oriented and langua ge particular and not phonologically determined, play a crucial role in handling morphological phenomena. Along the same line, Kiparsky (2005) propose the constraints E CONOMY and E XPRESSIVENESS to handle blocking and extended exponenece. If E CONOMY ranks higher, then blocking occurs. The opposite holds true for extended expon ence two distinct rankings to handle blocking and extended exponence while the real ization theory can predict a single grammar that handle s both successfully. In summary, the realization optima lity theoretic approach can handle to some extent the cases of doubly marked plurals. It elegantly addresses doubly marked plurals illustrate, I assumed that the output plura ls are derived from their singular outputs and not from their UR s Similar to the regular marked plurals, I argue that the plural markers

PAGE 202

202 come with the singular outputs, and must be parsed in the plural outputs. This is consistent with the analysis establi shed for the regular plural patterns. Specification of the E xceptionality in Lexical Entry The most employed approach for dealing with lexically and morphologically driven phonology is to assume that the inexpl icable segments such as the various deleted an d inserted segments are specified in the lexical entry of the form (Archangeli and Pulleybla nk 1994; Rose 1997; among others) For example, Kiyota (2003) assumes that r realizations of the plural morpheme The invariant shapes of the plural morpheme are then derived by the interaction among generalized templates, markedness constraints and Base Reduplicant Faith constraints. Lexically idiosyncratic plural forms also re quire lexical specification For instance, distinct templates. The singulars from which these templates are derived look similar in the number of consonants and shape; there i s no motivation for mapping onto distinct plurals forms based on the shape of the singular. For these plurals, I assume their singular derivatives come with a template specified in their lexical entry: (21) P L ut + CCVC P L ot + CVCC P L (21) above serves as the underlying representation for the templatic plurals and a set of competing constraints produce the optimal output. In this case, F AITH T EMPLATE is high ranked for these plurals. In Tableau [7] a vowel is inserted at the cost o f obeying

PAGE 203

203 the templatic shape pre specified in the lexical entry of these forms. Candidate (b) stays faithful to the output singular and does not shape into the target template. Thus, it is doomed. Tableau [ 7 ] F AITH T EMPLATE D EP V C P L F AITH T EMPLATE D EP V b. *! This approach to exceptionality undermines Gen e ralized Template Theory which assumes that templatic effects are rather derivable from markedness and prosod ic constraints and not from templatically specified constraints. For the analysis of regular shapes with a suffixal template, I assume that the final templatic shape of the plural which equals a syllable size is derived from markedness and faithfulness con straints. Thus, I did not have to introduce any templatic constraint to rule out potential outputs. Specifying a template for these plurals is not consistent with my analysis to the regular plural patterns. Containment Approach One of the earlier approache s to insertion and deletion in Optimality Theory proposed by Prince and Smolensky (1993) assumes that segments are not literally removed from output structures. Rather, deleted segments are represented as c e able and do not have any phonetic interpretation (Kager 1999:378). Since these elements are still contained in the output forms, they can play a phonological role of some sort. plurals that surface with an extra consonant than its o utput singular may be assumed to have this segme nt contained in the singular out put but not pronounceable. The

PAGE 204

204 underlying representations from which these plurals are derived have these segments unparsed as illustrated in ( 2 2) below: ( 2 2) Templatically e xpanded plurals windows b. kob kolob dogs young bulls sweethearts e:kwar chiefs This approach allows for a direct relation between the plurals with extra li singular forms. Thus, no reference to the Arabic singular forms is made to account for the re appearance of extra consonants in the plural forms. Since the formation of these plural forms with the appearance of inexplicable consonants seems abstract, a reference to a shared input can be a solution. However, not all the plurals collected have equivalent forms in Arabic. So, the problem is really partially solved. This approach definitely solves the problem of correspondence across two languages. It fu r th er shows the exact kind of deleted and inserted segments, restraining the functions of Max C and Dep C and avoiding deleting and inserting random segments in the output plurals. Therefore, it is consistent with the assumptions I made for the regular plural Constraint Indexation Pater ( 2000 and 2004 ) argues that markedness and faithfulness constraints in Optimality Theory can be lexically indexed to capture the exceptional or lexical behavior m orphemes that trigger a process are indexed for the application of a lexically specific markedness constraint, and morphemes that block a process are indexed for the application of a lexically specific faithfulness constraint

PAGE 205

205 (pp.1) According to him, these constrain ts are universal markedness or faithfulness constraints whose application relates to lexical items. H is hypothetical example for the application of these constraints comes from a language with coda deletion which is blocked in certain lexical forms. He ar gues that the exceptional items can be accounted for by a lexically indexed faithfulness constraint ( M AX L whereby L stands for lexical) In this case, the ranking that holds for these lexical items is M AX L N O C ODA M AX (23) M AX L N O C ODA M AX Regul ar form: /pitak/ > [pita] Exception: /timak L / > [timak] ( Pater 2004:1) In the lexicon, the exceptional items are as seen in the example above or the ranking attached to them. This will consequently exclude any plausible ranking variation for the se lexical items. pro s thetic vowel in plural forms attaching a suffixal VC template is not phonologically determined. Some forms that have consonant clusters word initially insert a pro s thetic vowel to break up the cluster while other forms do not. This can be analyzed by the ranking C OMPLEX D EP V Tableau [ 8 ] *C OM PLEX D EP V *C OMPLEX D EP V a. b. *! The blocking of insertion of a vowel to break up the consonant clusters word initially is conditioned by the indexed faithfulness constraint D EP V L which ranks above *C OMPLEX and applies only to lexical items that do not insert a vowel. Grammar: D EP V L C OMPLEX D EP V Lexicon: [ L

PAGE 206

206 Tableau [ 9 ] D EP V L *C OMPLEX D EP V Output Sing. Otput Pl. D EP V L *C OMPLEX D EP V a) b) *! L *! Pater (2004) shows that this approach can successfully capture exceptionality through well motivated constraints wh ose different sub rankings produce typological ranking observed cross linguistically. Furthermore, he also argues that this approach is faithfulness constraint and index it to exceptional forms. Thus, they do not need to memorize the nu merous rankings that hold for a lexical phenomenon. This approach assumes that constraints in Optimality Theory are lexically indexed to account for the shapes that diverge from their regu lar patterns. Plurals that surface with a prosthetic vowel do not follow a certain phonological pattern (for instance, sonority of the initial consonants do not determine whether a form gets a prosthetic vowel or not) Therefore, there is a need to have co nstraints with extra information about these lexically shaped plurals. Since t he output plurals are derived their singular outputs and not from their URs, this approach is consistent with my assumption that plurals are derived from their singular outputs. T here is a competition between the expected (potential and regular) plural and the resultant (exceptional) plural. Only a highly ranked indexed constraint will be able to determine the actual output form (the exceptional shape) The strategy employed by an d consequence reached from this approach are consistent with the analysis of the regular plural shapes in the language.

PAGE 207

207 Selector Constraint and RealizeMorpheme : Plurals with Double Exponents which take double plural morphemes displ ay three distinct shapes based on the plural markers they attach to. The first type ((24) below) takes the infix (V)b and the less common plural suffix (V)n The majority of the singular forms from which these plurals are derived take the shape CaC. The singu lar form (d) has a total of three consonants CCVC, with a cluster of two consonants word initially. The plural shape for all of the forms listed in ( 24 ) is CabCin, in which an infix b and the plural suffix in a re attached to mark plurality. There is no phonological reason why these forms take multiple plural markers to express plurality. Bi are observed to pluralize by either attachment of a suffixal VC ] derived from [nid forms to take double plural markers is quite unknown yet. They can readily attach the suffixal template or be pluraliz ed by ablaut. (24) The infix Vb and plural suffix in Sing. Pl. Gloss in graves in wells in fractures in flue e. al abl in tails The second group of plural exemplified in (2 5) below has the plural infi x (V)b and the default plural suffix Vt Only three forms of this doubly marked plural are found in a pool of 25 forms. These plurals are derived from tri consonantal (form (a) and (c) below) or quadri consonantal (form (b) below ) singulars The final sha pe of the plural is not the same a whereby the first two Cs come directly from

PAGE 208

208 the singular form. The second attested plural shape begins with a pro s thetic vowel (schwa) and has another inserted vowel between the final r epeated C s in the base (25) The infix Vb plus the suffix tV a. t t b monitor lizards b nipples b ot food The final and most common attested pattern of plurals with double exponents includes the plural suffixes Vn an d tV which follow each other in a fixed order. In all the collected forms, the default plural suffix tV occupies the last position in the plural form. The singular forms which take these plural suffixes are eclectic; they range in shape from bi consonan tal to quadri consonantal. It is important to note that some singular forms come with the suffixes t and Vh which are the feminine gender suffixes and subsequently get deleted before the attachment of the plural suffixes take place Semantically it is important to note that the majority of these forms are borrowed words from Arabic. Moreover, many of the consonants that appear in the correspondent plural form and are not originally in the singular form are in fact deleted consonants and retrieved in th e plural (26) Two suffixes Vn and tV Sing. Pl. Gloss t traditional Omani male gowns ah ti traditional Omani boxes ti Dhofari women d cars e. sayings f. clocks g. pillows h. jan shares/ rights i. eye lashes j. ot girls k. pens l. legs

PAGE 209

209 I make the following observations about the formation of doubly marked plurals in Jebbali: concatenative processes including ablaut, attachment of a VC template and mapping singulars onto plural te mplates, only two plural suffixes and an infix participate in double plural marking. It has been attested that for multiple plural exponence, one suffix and one internal change may together mark a morpho t; two suffixes mark double plurality or an infix plus a suffix mark double plurality. 2. The plural markers are not phonologically identical but semantically identical. Interestingly enough, in some Modern South Arabian languages and in Arabic, the exploi no additional meaning is emphasized by the extra plural marker In other words, how many plural markers a form attach es to does not contribute to a special meaning. These forms are still plural. They can easily be attaching one plural marker like the majority of plural forms in the language. But, these have two or even three morphemes to mark 3. For bi consonantal nouns, I assume it occurs to meet some templatic requirements of the language. The bi ablaut to mark plurality or expand by attaching a VC template with a fixed vocalic element and a copy of the final base consonant. So, a total of three p lural marking processes are enjoyed by bi 4. Some sort of fusion occurs when plural markers are attached such as deletion of the infix vowel and retention of the b alone. In some cases, it is not clear if the stem vowel is th e one attached with the plural infix or actually the plural infix assimilates to some sort of vocalic change. More interestingly, where /b/ is deleted elsewhere in the language, it is never deleted in the plural infix despite the fact that in both of these cases, /b/ is intervocalic. tV to mark plurality, I argue that this suffix is a defau l t plural marker for at least two reasons: (1) borrowed and nonce forms always take it as a plural marker ( 2) many diverse plural forms attach this plural suffix to mark plurality. I further argue that in theory, attachment of this plural suffix is a sufficient phonological change to express plurality. However, since another plural marker accompanies this suff ix there is a possibility that this suffix is made

PAGE 210

210 invisible 1 to R EALIZE M ORPHEME and carr ies no plural meaning in these plurals or it is not sufficient for it alone to express plurality There is a ne ed to attach another plural marker or else impose a p honological change on the stem to mark such a morphological function. If I assume that there is a domain for a double phonological change that cannot be applied where the first change has taken place, then we assume that such plural forms will definitely n eed a second plural marker to express the morphological process This idea is inspired by Ku risu (2001) who establishes a selector constraint To explain what the selector constraint does I assume that the i contains a suffix (specifically Vt ) as well as a stem when discussing German plurals which take umlaut and a suffix, if it were not present in t he structure, making only the stem available as a visible violate s R EALIZE M ORPHEME and the stem requires a phonological change (a second plural marker or an internal change ) due to th e pressure of R EALIZE M ORPHEME According to the He proposes a selector constraint t hat can be applicable for the three varieties of affixation observed cross linguistically (prefixation, suffixation and infixation). The relevant morphological process es involved in the formation of doubly marked plurals are 1 Following proposals made in Kurisu (2001:191 194), a second plural marker (be it a non concatenative operation or an affix) is added w hen no suffix exists underlyingly because the affix is the only eligible strategy to avoid a violation of R EALIZE M ORPHEME So, Vt is made invisible to R EALIZE M ORPHEME and its presence alone violates R EALIZE M ORPHEME

PAGE 211

211 infixation and suffixation for which the following selector constraint s ha ve been proposed: (27) (a) Infixation (b) Suffixation PrWd PrWd [Stem [Affix] Stem ] [Stem [Affix]] For example, t he selector constraint (27b) makes the domain of a stem accord with that of a prosodic word. Therefore, an output that satisfies it has the structure in which the suffix is disregarded from the prosodic word domain. Since R EALIZE M ORPHEME is sensitive to the prosodic word domain, then such an output violates R EALIZE M ORPHEME Thus, a second plural marker is needed to satisfy R EALIZE M ORPHEME when a stem wi th an invisible suffix violates it. In other words, a plural form which bears only one plural marker (suffixation only) is doomed since it satisfies the selector constraint which makes it invisible to R EALIZE M ORPHEME and a violation of RealizeMorpheme occ urs. However, the optimal form satisfies RealizeMorpheme since a change is made in the domain prescribed by the selector constraint. T he s elector approach works for plurals that have a suffix and an internal change (such as a Vb infix) but it fails to add ress plurals with two suffixes since both suffixes would be invisible to RealizeMorpheme. In the analysis of the regular plural patterns in Chapter 5 I used well motivated constraints in Optimality Theory. The selector approach refers to a constraint whos e motivation is not well established in the framework of Optimality Theory. However, the constraint appears to successfully capture some of the plurals with double exponence.

PAGE 212

212 The previous sections have explored eight distinct Optimality Theoretic approaches that address exceptionality irreg ularity in the formation of sub patterns of systematic plural and lexical marking characterizing the formation of forms. I also showed that each exceptio nal pattern exploits a different way to realize the plural morpheme which imposes difficulty to the morphoph o nological models proposed to date. As a result, I argued that t he se diverse morphophonological tendencies associated with the formation of plurals may be advanced by misce l laneous approaches to fully capture their richness and intricacy For example, while t emplatic plurals take a specific template there is no driving phonological motive that forces the singulars to take that template, except the n eed for a plural to be distinct from its singular form T emplatically expanded and trun cated plurals insert and delete eclectic segments in contexts that are not phonologically triggered. The potential of integrating these diverse patterns into a unified t heory thus seems far fetched given the fact that the most powerful p honological theories like R EALIZE M ORPH (henceforth; RM) will confront many insurmountable problems when attempting to group these exceptional shapes under th e same u mbrella Below, I will first review the Realizational Morphology Theory (RMT) as advanced in Ku risu (2001) and show the success it enjoys conceptually and applying this theory to address the exceptionality o cogently Kurisu (2001) has made promising theoretical contributions to the constraint R EALIZE M ORPH by delimiting its powerful consequence and offering a better understanding to its interactions with well motivated O ptimality T heory constraints His doctoral dissertation devoted to Morpheme Realization argues that

PAGE 213

213 R ealize M orpheme cou ched under the Realization al Morphology Theory (RMT) can capture a whole range of phonological exponence of morphemes and allow for unification of both concatenative and non concatenative processes in natural languages I will first present the formalism of Re alizational Morphology Theory as proposed in Kurisu (2001), and then illustrate its effects on various morphological processes : (2 8 ) Realize M orpheme (RM): ly. (Kurisu 2001: 262) According to Ku surface phonological manifestation, so it is not a c onstraint special to non concatenative This assertion is suppo r ted by showing how a range of affixational and non concatenative operations nicely fit into the RMT and produce observed typological sy s tems in natural languages. Ku risu ar gues that RM provides a unifying a ccount to subtractive and templatic truncation ( phonological invariable in subtractive morphology whereas the residue remaining after deletion is long thought be unrelated and hence must be captured distinctly According to Ku risu, t he difference lies in the presence or absence of templatic constraints which interact with RM and other faithfulness constraints such as M AX This will subsequently delimit the size of the output form in the case of templatic morphology RM also address es truncation and reduplication without recourse to templatic constraints such as TRUNC and RED. Finally Ku risu shows that RM, alo ng with the sympathy theory, succeeds in accounting fo r double morphemic

PAGE 214

214 correspondence whereby zero to two morphemes mark a morphosyntactic value. RM predicts no language that has more than two morphemes for a singular morphosyntactic phenomenon. It also predicts that affixation and subtractive morphology ne ver co occur. This sums up Ku of anti faithfulness constraints. Based on the above summary of RM, it seems that RM can admit all kinds of processes assum ed to be analyzed by anti fai thfu lness constraints (Alderete 1999) and even go beyond by accounting for cases which ant i faithfulness constraints have failed to address 2 Ku risu stipulates that t he overall ranking RM >> Faith address es morphophonological cases where the c hange is trigger ed by parsing underlying material or exhibited by stem modification As established in phonology, p honological changes result from a competition between markedness and faithfulness constraints while morphoph onological phenomena display features of both pho nological alternations and morphological marking. Therefore, Kurisu assumes that stems are necessarily devoid of any morphosyntactic value. For example, stems which are realized morphosyntactically as singular forms without any phonological change do not v iolate RM. W hen they und ergo a change to mark plurality, they satisfy RM This, according to Ku risu, enhances RM since it accords with O ptimality T heory are violable and that RM is no exception. Ku risu also considers R EALI ZEMORPH to be a morphological faithfulness constraint w hich is supported by the fact that affixational and non concatenative morphology both display allomorphy in their actual forms Ku risu 2 In a d dition to bringing a redundant t ool in Optimality Theory, Kurisu (2001:74) argues that anti faithfulness constraints are only operative in surface surface mappings and cannot be extended to lexical surface dimension. It also fails to account for phonological polarity which is associated w ith morphological conditioning

PAGE 215

215 stipulates that not only does RM require a stem modification but a lso demands a morpheme attached to an underlying form must be phonologically realized in the output form. As Ku risu develops the RMT, he proposes that indexed faithfulness constraints may be sandwiched between RM to produce the actual output of a language that imposes certain restrictions about its shape and prosody. Pursuing Ku undermine his arguments for RM. First, if I assume that templatically expanded plural fo rms result from RM >> D EP and tempatically truncated plurals surface as a result of the competition between RM and M AX then the analysis will need to stop here and no further progress can be made. For one thing, the inserted and deleted elements are not t he same across the board. I cannot delimit the definition of the faithfulness constraints so that the inserted or deleted portions are clear and exact. In some instances, a syllable is chopped from a singular form to mark its plural counterpart. For anothe r, the size of the final plural form is not constant all the way through. This makes a stipulation of a prosodic constraint to forbid excess insertion or deletion impose difficulty t o restrain the power of RM. In fact, any change is admitted by RM. But, what makes the theory worthy of consideration is the kind of constraints that play a role to direct the focus of the change. Ku risu also presents insightful analyses to double marked morphology that can be Since doubly Vb I assume that the plural

PAGE 216

216 suffix is invisible to the singular plural mapping due to 3 (Kurisu 2001:246). According to Ku risu, double marked morphology like German plurals which have both a suffix plural marker and stem change can be accounted for if we assume that the suffix is invisible to R EALIZE M ORPHEME as a plural marker. Th is entails that a stem + the plural suffix alone violates R EALIZE M ORPHEME ensure double realization of plurality in German. Applying RM to templatic plurals produces more desirable results since there is a defi nite template that these plurals take. So, both RM and a templatic constraint stipulating a final template together derive th ese plurals. However, I am conf ronted with the problem of proposing a unified constraint that is ranked lower than RM and T EMPL Th e problem is that there is no phonologically driven cause which entails these singulars to take a templatic pattern and not any other pattern. A whole range of diverse singulars which take a definite template make s this move disc ouraging. Is it possible th at numerous plurals (ranging from bi tri and quadri consonantal) prefer that template for no reason? most powerful devices of O ptimality T heory proposed to date. One may think that RM, being abstract and non stipulative, will be able to offer a unified analysis to these notoriously shaped plurals. However, RM comes with many drawbacks too. Finally, Ku risu predicts that the maximal numbers of morphemes that may mark a morphosyntactic up to three plural markers attached to a form to mark plurality. Therefore, RM fails to unifyingly address these exceptional plural patterns. 3 K u L (stem, PrWd), Anchor

PAGE 217

217 Summary of Chapter 6 ized by a number of peculiar and exceptional shapes which pose a challenge of incorporating them with the proposed Optimality Theory analysis. There is a huge diversity attributed to the singular forms from which these plurals are derived which makes it h ard to propose a unified underlying representation for them. The plural shapes are also distinct from each other; some start with a consonant clusters and others have an initial vowel. There is no specific singular template that derives a specific template Moreover, the plural marker is sometime s not clear or easy to state for these plurals. For instance, plurals may come with a vocalic change or consonantal shift along with a plural marker. Moreover, t hree templatic shapes result from mapping distinct sin gular shapes. It is hard to classify singulars based on their specific shapes into classes taking a particular template and not other templates. All these difficulties constitute a big obstacle in offering a unified O ptimality T heory account for these unus ual shapes. Chapter 6 also outlines some approaches to dealing with exceptionality and lexical marking in Optimality Theory such as eliminating the underlying representation, R EALIZE M ORPH two level well formedness, realization optimality theoretic account intermediate levels, s pecification of the exceptionality in Lexical Entry containment approach, constraint indexation and selector constraint In discussing these approached to exceptionality, I examine d whether they are consistent with the proposals I made for the more regular shapes in the language. I finally explore d the potential success and failure of Realize Morpheme Theory in offering a unified analysis to the exceptional

PAGE 218

218 CHAPTER 7 CONCLUS ION Throughout this dissertation, I ha ve addressed the diversity and intricacy involved In documenting the enormously diverse shapes of plurals, I explored a number of non concatenative morphological proces ses under which these plurals can be classified. Previous work, which shed light on plural formation in the language, concerns only listing noun plurals in accordance with their CV shapes, and does not identify the crucial morphological processes involved in the formation of noun plurality. I further showed that many noun Therefore, they can be analyzed theoretically using a powerful generative framework like Optimality Theory. Nor does th is dissertation stop with the exploration and analysis of the systematically formed plurals ; it also explains, in depth, the exceptional plural patterns which seem to be ad hoc in their overall shapes. These are also formed by processes different from the systematic non concatenative processes triggering the formation of their systematic counter part plurals. They, thus, constitute a challenge to integrating them into the proposed analyses of the regular plural shape s I also examined whether the Optimality Theory approaches to exceptionality and lexical marking are consistent with the proposals I made for the more regular plurals in the language. To close this dissertation, I summarize the major results and contributions made in this study of Plurality in especially in Chapter 5 and finally present ideas for

PAGE 219

219 Results and Contributions tendencies characterizing its phonology and morphology. First and for e has quite a large phonemic inventory when compared to Arabic, a dominant neighboring language. This phonemi c inventory exploits 35 consonants or more in other varieties of has expansive vocali c contrasts and involves an abundant variety of syllable structures. It encompasses two prominent stresses which are assigned in words of two syllables or more. Moreover, the phonology of this language is intricately structured with all the possible phono logical processes being operative; some of thes e processes are contrasting, for example fortition versus lenition and insertion versu s deletion, which operate simul t a neously to produce surface forms in the language. More relevant is that ex hib it many of the attested and non attested non concatenative morphological processes. An example of the non attested shape is the Vb infixed plural since other Modern South Arabian languages and Semitic do not have s uch a plural patter n This rich morphology extends to embrace exuberant exponence manifested by the doubly and triply marked plurals. It has been argued that triple exponence is non existent cross its grammar of plur al formation. Ku risu (2001) confidently states that Multiple Morphemic Expone nce is limited in range: l anguages may have from zero morpheme to maximally two to express a certain morpho exhaustive survey of all huma to K u s He

PAGE 220

220 reaches this conclusi o n b ased on his theoretical assumptions and the system he used to analyze the scope of Multiple Morphemic Exponence. The first contribution of this dissertation has been to document the above mentioned linguistic tendencies and more. Apart from exploring no un plurality of the language, it has been a goal of this dissertation to reveal using modern devices and well es tablished tools available in current phonological theory This has been done through the faithful adoption of the Inte rnational Phonetic Alphabet (IPA) which hopefully clear s all the misconceptions and difficulties made by the confusing and info rmal notations employed in all the previous work on the language. At the very beginning of embarking on this research I was ove rwhelmed by the inconsistent transcription composed of huge number of references It was very hard to identify the exact phonological processes prevalent in ences, considering the fact that the symbols are unclear. I hope that I have laid out the fundamentals in Chapter 2 of this dissertation and will provide a reliable reference on the sounds syllable structures and phonological processes of the language fo r future researchers Careful readers will also notice a comprehensive literature review of the major linguistic and non linguistic work that has been done Arabian languages in the last century or so, s ome of which do not directly relate to the work presented in this dissertation. However, I included everything I could find about futu re researchers find brief synop s e s on the previous w ork, and to encourage building

PAGE 221

221 up on previous work. Needless to say, documentation of previous work will definitely save time and speed up the process of any revitalization project for the language in the future. Last but not least, I myself found the lite rature review section very helpful and I was able to get ideas for future research. Chapter 5 of this dissertation is my major original contribution. First, I identified the four systematic non concatenative processes exploited in the formation of noun plurality in Vb infixation, ablaut, attachment of a VC template and templatic patterns. I described these patterns thoroughly and analyzed them using the framework of Optimality Theory. For some patterns, I showed more than one possible approach which allows the overall analyse s to take a complete shape. To illustrate, in analyzing the plurals with ablaut, I explored three distinct approaches: Positional Faithfulnes s, Anti Faithfulness and RealizeMorpheme. I further showed that while Positional Faithfulness and RealizeMorpheme are able to account for the two distinct shapes of ablaut plurals (simplex and complex), the Anti faithfulness model may encounter some diffic ulty addressing the ablaut plurals with a simplex root. This stems from the fact that ablaut plurals alter the [back]ness of the vowel contained in their initial and only syllable, while the proposed Anti Faithfulness analysis of ablaut plurals stipulates that I DENT initial [back] SP must outrank other Anti Faithfulness constraints. In the Vb infixed plurals an alignment constraint, which shows the locus of the Vb infix in the plural forms, is dominated by the constraint monitoring the left edge of th e plural form. These two constraints are well motivated in Optimality Theory, and along with the constraint restricting the size of the affix to a syllable length produce the actual

PAGE 222

222 Vb with the suffixal VC template with the final reduplicated C of the base is very integrated, and conforms proposals that this pattern is not a true reduplication but rather a templatic expansion. This dissertation now explains theoretically what templatic expansion is since the final analysis of this pattern conforms with the language specific syllabic templatic and prosodic well formedness requirements Finally, the template resulted from mapping singulars with a geminate onto plurals, is also explaine d in terms of the interaction of well motivated constraints in Optimality Theory. have not been theoretically approached before this dissertation. Only a very little description on plurality (though inspiring and organized ) was given in Ratcliffe ( 1996, 1998a & b ) This dissertation hopefully contributes to existing knowledge in phonology and Optimality Theory, by adding yet another set of Semitic language phenomena that Optimality Theory can cogently and elegantly explain Remaining Issues In fo Vb infixation a pattern of plural that has not been attested in Semitic. It will be very illuminating to trace the history of this infix and study diachronically how this infix evolves to be a plural ma rker in the language. It will also be revealing to explain why the b of the Vb plural infix never gets elided while th e language extensively deletes b elsewhere I suggest an exhaustive list of all the forms that have elided b should be compiled and studie d thoroughly. Triple lso a potentially fascinating avenue for future research. First the language seems to admit triple plural markers, a tend ency y rendered to be impossible cross linguistically. Although I observe it is very ra re in light

PAGE 223

223 n ts were not able to generate more than a single form that has triple plural marking more work needs to focus on this pattern and trace back the form that has triple exponents. Secondly there is now an ongoi ng research on double exponence, as linguists are baffled by the huge number of languages employing this linguistic tendency in the grammar. There was an old belief that there is a bi directional relation between a form and meaning. In other words, only a single form may express a particular meaning. But, this wave is now changing and many the oretical frameworks are now devis ed to explain this phenomenon (Kurisu 2001; Xu and Aronoff 2011). is good enough to contribute to this debate. It will also be a contribution to analyze theoretically the other templatic shapes that result from mapping singulars onto plurals. Ratcliffe 1992; Ratcliffe 1996; Simeone Sene lle 1997; Ratcliffe 1998a and b) was too obsessed with the plural templates. Ratcliffe ( 1998b ) in particular, took this obsession to a level Although such an obsession w as justified bearing in mind the goals of the researchers and the paucity of research done on plurality at that time, my biased view envisions more fruitful results from theoretical and analytical linguistic work than from further generalization of descri ptive facts. Finally, all the exceptional patterns of plurals that were described in Chapter 6 should be taken seriously in future research. I believe the Optimality Theory models for dealing with exceptionality and lexical marking will need to go one st ep higher, so that an integrated analysis for these shapes may be built.

PAGE 224

224 APPENDIX A External Plural Table A 1. Plurals with the plural suffix / Sing. Pl. Gloss Class 1 s ti plates m mho t waters (a lo t of water) m men m rem re ti tall persons m fef elbows m bat bat ti beaches m ti doors m mountains m ti grounds/ floors f lho ti cows m blankets m ti mountains m et holes bored in the ear lobes f ti t soil m shops m w on fang ti small coffee cups m people from Dhofari f et plaits, tresses of hair f ti plateaus m ti nails m ut ti moles f ut ti butterflies f ti donkeys m ears m herum ti plants m ti lamps m t slaves m tobacco f traditional house m ti ti friends m e:d adi hands m names m et demons f roads m eyes m it old women f at t guests m, f mouths m 1 Class refers to the class of the singular form.

PAGE 225

225 Table A 1 Continued Sing. Pl. Gloss Class t slaves m ot cars m t green grass m at i Jebbalis (f.) f at i Bedouin women f kub t cups m Table A 2 Plurals with the plural suffix Vn whereby V is often /u/ Si ng. Pl. Gloss Class fudun fidn in stones f u aunts m un aunts m un tulchans 2 m un children m unub in tails m un doctors m losi losun large traditional scarfs m ut un years f gafan eyelids m Table A 3 Plurals with the plural suffix i Sing. Pl. Gloss Class s it s ofor i cooking pans f i boys m at black flies f i cliffs/ mountain edges f angels m Table A 4 Plurals ending 3 Sing. Pl. Golss Class corners f b carved wooden dolls f mar it mirrors f at monitor lizards f et beards f 2 Tulchan was a man appointed as bishop in Scotland 3

PAGE 226

226 Doubly and Triply Marked Plurals Table A 5 Plurals with double and triple plural markers Sing. Pl. Gloss Class ab l un chameleons m b r in graves m t n / f ah un ti traditional wooden boxes f on ti Dhofari women f un cars f un beds f zol it zol un carpets f et kof caps f b t ot food m b r in wells m b r in fractures m s b s in flue m r un sayings m sa un clocks m takj un p illows m jan un shares/ rights m n tongues m un eye lashes f al a b l in tails m segod sigad un carpets f j bg it/ ot ti girls f n ti pens m n legs f t t b monitor lizards f b n ipples m

PAGE 227

227 P lurals with V b Infixation Table A 6 Quadri consonantal s ingulars Sing. Pl. Gloss Class the top parts of legs m mattresses made of leather m s s boxes m um am pots used to keep ghee m sieves m mattresses m hotels m cauldrons m rifle bolts m caravans, turns, times m deserted places m ref mi s rations, supplies m mis grass cutting knives m t caches m police stations m murkus crutches, walking sticks m boats m bakeries m lids m great barren plains m homesteads m shoulders m muns ur wood or bone hairpins m pincers m adzes for digging m tools to take the cover of f something; or to remove nails m pestles and mortars m it complaints m biers m gardens m ports m veils m rifles m pure white waistcloths m feathers for applying kohl m cartridge belts m hobbles m choppers m big flocks of goats or sheep m big loads m

PAGE 228

228 Table A 6 Continued Sing. Pl. Gloss Class big fishing nets m mattresses made of cow leather m leather bags, traveling cases m mufs joints m hairlines (in women) m no tebooks (borrowed from Arabic) m w on coffee cups m dozens m hocks m notebooks m mattresses made of cow leather m karabsi chairs m mas t r ah mas abt rulers f bor rowed from Arabic Table A 7 Bi consonantal and t ri consonantal s ingulars whose plural takes Vb. Sing. Pl. Gloss Class muzzles m news m t ad t bed Zizyphus spina Christi 4 f t drums m Table A 8 Vb Infixed p lura ls with initial v owels Sing. Pl. Gloss Class keys f offices m t restaurants m rooms for guests m mosques m rifles m houses f 4 A type of spiny shrubs and small trees in the buckthorn family

PAGE 229

229 Plurals w ith Suffixal Template VC Table A 9 Plurals taking a suffixal template Sing. Pl. Gloss Class k books m dik roosters m kot towers m fish m nuf selves m t axes f letters m lavatories f feet, soles m cheeks m shelves, racks, bulks m mus razors m palms of the hand; claws m pilgrims m ham hmum conc erns m dry leaves f rights m brothers m dots f hab ot/ hib ot hbeb/ heb songs f

PAGE 230

230 Table A 10 Ablaut or vowel o pposition Sing. Pl. Gloss Class orphans (m.) m s af rir s flowers f isolated homes m s elim s elam nuts m t t dresses f fag ri fag ru Bedouins m cheeks f mo mo flesh of backs m mim small pieces of wood m ded narrow passages leading to the base of a mountain m date stones f partings m shortened waistcloths (for men) m albod sandals m noses m ab teats f m a waistcloth used to heads m s ad s fish m cheeks m at bags f molar teeth m babies, infants m teeth m dry wounds m faces m um enemies m sardin es f gear m t solos chains f es es birds f gilil t bullets f sinor t sinar cats f il ol w orn out dress f sabro jins f/m klin t kilan weddings m w ol eyebrows m combs m

PAGE 231

231 Table A 11 Ablaut p lurals of CVC shape Sing. Pl. Gloss Class j t sheep f nid nud water skins f ropes m men m houses f hab ot/ hib ot heb/ hbeb songs f bottles, water jars f kob kub cups f it rice f tribes f fit fet towels f Templatic Shapes Table A 12 Plurals derived from geminated singulars Sing. Pl. Gloss Class pots f hilts (of swords) f hills f town; small villages f coffee pots f benches outside a house f heavy wooden bolts of a door f lbeb kernels f ot aunts f nib nbeb bees f hab ot/ hib ot hbeb songs f ot flies f Table A 13 Templatically expanded plurals Sing. Pl. Gloss Class windows f kob kolob dogs m skins m bo mot o kitchens m young bulls m sweethearts m e:kwar chiefs m t orphans (f.) f rocks f nights m

PAGE 232

232 Truncation Table A 14 Truncated plurals Sing. Pl. Gloss Class e s e s fingers m camel calves m cartridges m e t e fans f turni ngs on a path f corners f b carved wooden dolls f t pillars f masor nails m livestock f sign s marked on the ground m t papers f t t characte rs f Templatic Plurals Table A 15 Plurals taking the s hape CVCVC Sing. Pl. Gloss Class female possessors f cheeks m sahab waves m t rifle bolts m t schools f t rocks m roofs m presents f sim mats f et grounds f at skin used for milking f palm trees f Table A 16 Plurals taking the s hapes CVCC and CCVC Sing. Pl. Gloss Clas s ut tears f skun communities m bread f kofor kfar non believers f ot stories f et holes bored in the ear lobes f o t honeycomb s f

PAGE 233

233 Miscellaneous S hapes Table A 17 Varying shape s of plurals Sing. Pl. Gloss Class at ghosts m/f heads m beads f months m old men m et beards f ot date stones f ut/ ot spoons f Suppletive/L exical Plurals Table A 18 Lexicalized plurals Sing. Pl. Gloss Class j women f times f ejat gol camels m pieces m bera boys m bri j ni sons m ti daughters f w z goats m et tu r dates f flies f tents f elbows m brothers m jum a:m days m women f berdam jo people f/m te kinds of meat m babies, infants m, f girls f fleas f

PAGE 234

234 LIST OF REFERENCES Abd Rabbo, M ohammed 1990 n M. Eid and J. McCarthy ( e ds.), Perspectives on Arabic Linguistics: Papers from the First Annual Symposi um on Arabic Linguistics Amsterdam: Benjamins pp.55 93 Abu Mansour, Anas 1995 L ingu i stic A ssociation of C anada and the U nited S tates Forum 22 320 339 Alder ete, J ohn 1999. Morphologically Governed Accent in Optimality Theory Ph.D. dissertation, Univ ersity of Massachusetts Alderete, J ohn 2001. D e rivational A ntifaithfulness Phonology 18 201 253. Alderete, John, Jill Beckman, Laura Benua, Amalia Gnanadesikan, John McCarthy, and Suzanne Urbanczyk 1999 Reduplication with F ixed S egmentism Linguistic Inquiry 30 327 364. Al Hafeedh, Ali Mohsin. 1989. Min lahja:t Mahrah wa a:dabiha Nahdha. Al Mas hani, M ohammed 1999 The Lexical relationship between Classical Arabic and Shehri: A comparative analytical study Ph. D dissertation, Universi ty of Manchester Al Mashani, M ohammed 2003 Lisan Dhofar al hemyari al muqaranah Center for Omani Studies, Sultan Qaboos University. Al Shehri, S alem 2007 Al lugha al lugha al muqaranah Al Tabuki, S alem 1982 Arabian Stud ies 6 51 56. Arabneh, Jafar 1978 The M orphophonemics of P luralization in Bibli cal Hebrew and Classical Arabic PhD d issertation University of Utah. Arch angeli, D iana and Douglas Pulleyblank. 1994 Grounded Phonology. Cambridge, Mass: MIT Press. B akovi E ric 2005 Anti gemination, A ssimilation and the D etermination of I dentity Phonology 22 279 315. Bat El, Outi. 1994a odificatio N atural L anguage and L inguistic T heory 12 571 596.

PAGE 235

235 Beckman, J ill .1998 Positional Faithfulness Ph.D d issertation: University of Massachusetts. Belova, A nna 2009 South Semitic Lan guages n K. Versteegh et al (e ds.), Encyclopedia of Arabic Language and Linguistics, volume IV Leiden: Brill, pp. 301 315. Benha llam, Abderrafi. 1980. Syllable Structure and Rule Type in Arabic Ph.D. dissertation, University of Florida. Benua, L aura 1995 runcation n J. Beckman, L. W alsh Dickey, and S. Urbanczyk (e ds.), University of Massac husetts Occasional Papers in Linguistics 1 8 G LSA Amherst, MA, pp. 77 136 Benua, L aura 1997 Transderivational Identity: Phonological Relations between W ords Ph.D dissertation, U niversity of Massachusetts Bittner, Maximilian 1909 Studien zur Lau t und Formenlehre der Mehri Sprache in Sdarabien I. Zum Nomen im engeren Sinn. Alfred Hlder: Wien. Bravmann, M eir 195 3 blaut in Semitic Orientalia 22 1 24. Buckley, Eugene. 1990 Edge in Associations and OCP 'Violations' Proceedings of the West Coast C onference on Formal Linguistics 9 75 90. Burzio, L uigi 1996 Surface C onstraints versus U nderlying n J. Durand and B. Laks (e ds.), Current Trends in Phonology: Models and Methods CNRS, Pa ris X and University of S alford, Uni versity of Salford Publications, pp. 97 122 B urzio, L uigi 1998 Lingua 104 79 109. Clements, George. 1986. n L. Wetzels, an d E. Sezer (eds.), Studies in Compensatory Lengthening: Publications in Language Sciences 23 Dordrecht: Foris pp. 37 76. Crowhurst, M egan 2004 lignment Natural Langua ge and Linguistic Theory 22 127 177. Dell, F rancois and Mohamed Elmedlaoui 1992 on con catnative M orphol Journal of Afroasiatic Languages 3 89 125. Faber, A lice 1997 n R. Hetzron ( e d.), The Semitic Languages Lo ndon, Rutledge pp. 3 15 Place in Language 77 91 106.

PAGE 236

236 Frajzyngier, Z ygmunt 1977 M. Newman (e ds.), Papers in Chadic Linguis tics Leiden, Afrika Studiecentrum pp. 37 56 Frolova, T atiana 2005 ibilant in Modern South Arabian Languages and its Etymological P ersp n L. Kogan ( e d.), Babel und Bibel 2: Memoriae Igor M. Diakonoff Winona Lake, Ind pp. 4 29 455 Gafos, Adamantios 1995 On the Proper Characterization of 'Nonconcatenative' L anguages s Johns Hopkins University. Gafos, A dmantios .2003 Gre A symmetry in Arabic: A C onsequence of S tems in P aradigm Language 79 317 355. Glov er, B onnie .1988 The M orpho phonology of Muscat Arabic Ph.D dissertation, Unive rsity of California Goldenberg, Gideon 1977 angu Bulletin of the School of Oriental and African Studies 40 461 5 07. Greenberg, Gideon 1955b Lukas ( e d.), Afrikanistische Studien Diedrich Westermann zum 80. Geburtstag gewidmet. Berlin: Akademie Velag. Hammond, M icheal 1988 r ansfer in Arabic Broken P lural N atural Language and Linguistics Theory 6 247 70. Harris, J ames .1980. Nonconcatenative Journal of Linguistic Research 1 15 31. Hayes, Bruce. 1989. Compensatory Linguistic Inquiry 20 253 306 Hayward, K athrine, Richard Ha yward and Salem Al Tabuki .1988 Bulletin of the School of Oriental and African Studies 51 240 250. Hetzron, R obert 1972 Ethiopian Sem Journal of Semiti c Studies Monograph 2, Manchester, Manchester University Press pp. xii 145 Hetzron, R obert 1997 The Semitic Languages London: Rutledge. Hofstede, A nje 1998 Syntax of Jibbali Ph. D dissertation, University of Manchester. Horwood, G raham 2001 faithfulness R utgers O ptimality A rchive 466 0901

PAGE 237

237 Huehnergard, J ohn 2000 A Grammar of Akkadian Ha rvard Semitic Museum Studies 45, Eisenbrauns, Winona Lake, Indiana. s and Historical Changes in Hebrew Derivations and constraints in phonology, Clarendon Press, pp.367 392 Ito, J unko and Armin Mester. 1999 Realignment n R. Kager, H van der Hulst and Zonnevald (e ds.), The prosody morphology interface Cambridge University Press pp. 189 217 Jahn, A lfred 1902. Mehri sprache in Sdarabien. Texte und Wrterbuch. Wien:Hlder. Jayakar, A tmaran 1889 Journal of Royal Asiatic Society of Great Britai n and Ireland 649 87 and 811 89. Johnstone, Thomas 1970a rtic Bulletin of the School of Oriental and African Studies 33 295 307. Johnstone, Thomas. 1973 atterns in the Modern South Arabian Journal of Semitic Studies 18 98 107. Johnstone, Thomas. 1975 rticulatio ns in the Modern South Arabian n James and T. Bynon ( e ds.), Hamito Semitica The Hague and Paris, Mouton, pp. 155 159. Johnstone, Thom Journal of Arabic Linguistics 4 61 71. Johnstone, Thomas 1980b occurrence of a t Prefix in Certain Jibbali Verbal F orms Bulletin of the School of Oriental and African Studies XLIII 3 466 470. Johnstone, Thomas 1981 London: Oxford University Press. Johnstone, Thomas 1983 n n R. L. Bidwell and G. R. Smith (e ds.), Arabian and Islamic Studies: Articles prese nted to R. B. Serjeant on the occasion of his retirement from the Chair of Arabic at the University of Cambridge London and New York, Longman pp. 225 228 Johnstone, T homas 1984 honem es in the Modern South Arabian L an guages of Dho n J. Bynon (e d.), Current Progress in Afro Asiatic Linguistics: Papers of the Third International Hamito Semitic Congress Amsterdam and Philadelphia, John Benjamins pp. 389 390 Kager, Rene. 1999. Optimality theory Cambridge Univer sity Press.

PAGE 238

238 (eds.), The Prosody Morphology Interface, Cambridge University Press, pp. 1 38. Yearbook of Morphology 2004 113 135. K u risu, Kazutaka. 2001. The Phonology of Morpheme Realization Ph.D. dissertation, University of California at Santa Cruz. Kiyota, Masaru. 2003 A paper presented in the 38 th International Conference on S alish and Neighboring Languages, University of Victoria, British Columbia. Kramer, Ruth. 2009. Definite Markers, Phi features and Agreement: A Morpho syntactic I nvestigation of the Amharic DP Ph. D dissertation, Unive rsity of Califor nia at Santa Cruz Leben, William 1977a eduplication in n A. Juilland (e d.), Linguistic Studies Offered to Joseph Greenberg on the Oc casion of His Sixtieth Birthday, Saratoga, California pp. 419 39 Lebe n, William 1977b Parsing Hausa ds.), Papers in Chadic Linguistics, Leiden, Afrika Studiecentrum pp. 89 105 Leslau, W illiam 1945 in the Modern South Arabic Language 21 230 249. L eslau, W illiam 1946 Bibliogra Bulleti n of the New York Public Library 50 607 33. Leslau, W illiam 1947a Word 3 180 203 Leslau, W illiam 1947b of the D ialect of Curia Bulletin of the New York Public Library XII 5 19 Leslau, W illiam 1970 Ethiopic and S n T. A. Sebeok (e d.), Linguistics in South West Asia and North Africa : Current Trends of Lingu istics 6 pp. 467 527 Levy, M ary and James Fi delholtz 1970 Glossa 5 57 70 Levy, M ary 1971 The Plural of the N oun in Modern Standard Arabic Ph D dissertation, University of Michigan at Ann A rbor. Lonnet, A ntoine 1985 Proceedings of the Seminar for Arabian Studies 15 49 55.

PAGE 239

239 Lonnet, A ntoine 2009 South Arabian, n K. Versteegh et al (e ds.), Encyclopedia of Arabic Lang uage and Linguistics, volume IV Leiden, Brill pp. 297 300 Matthews, C harles .1960. Noun D etermination in South Arabian University of S outh Florida Language Quarterly VII 41 48. Matthews, Charles. 1969. a Clue T hereto from Journal of the American Oriental Society 89 22 27. McCarthy, J ohn 1979 Formal Problems in Semitic Phonology and Morphology Ph.D dissertation, MIT, Cambridge, Mass. McCarthy, J ohn 1981 A Prosodic Theory of Non concatenative Linguistic Inquiry 12 373 418. McCarthy, J ohn 1982 Language 58 574 590. McCarthy, J ohn 1985 n Keating ( e d.), Papers in Laboratory P honology 3 Cambridge CUP pp. 191 233 McCarthy, J ohn 1986 Linguistic Inquiry 17 207 263 McCarthy, J ohn 1998. Morpheme S tructure C onstraints and P aradigm O ccultation Papers from the Regional Meeti ngs 34 Chicago Linguistic Society pp. 125 150 McCarthy, John.1998. R utgers O ptimality A rchive 252 03981 McCarthy, J ohn 2000. Faithfulness and P rosodic C ircumscription n J. Dekkers, F. van der Leeuw, and J. v an de Weijer ( e ds.), Optimality Theory: Phonology, Syntax and acquisition Amsterdam HIL pp. 151 189 McCarthy, J ohn and Alan Prince. 1986 orphology University of Massachusetts and Brandies University. McCarthy, J ohn and Alan Prince 1990 a Natural Language and Linguistic Theory 8 209 283. McCarthy, J ohn and Alan Prince. 1990b orpholo i n M. Eid and J. McCarthy (e ds.), Perspect ives on Arabic linguistics: Papers from the second symposium Amsterdam, Benjamins pp. 1 54 McCarthy, John and Alan Prince. 1993a. Prosodic Morphology I RUCCS TR 3, Technical Reports of the Rutgers Center for Cognitive Science, New Brunswick, NJ.

PAGE 240

240 M cCa rthy, J ohn and Alan Prince. 1993b n G.E. Booij and J. van Marle (e ds.), Yearbook of Morphology Dordrecht, Kluwer pp. 79 153 McCarthy, John and Alan Prince. 1995a. 'Faithfulness and Reduplicative Identity', in J. Beckman, S. Urbanczyk, and L. Walsh (eds.), University of Massachusetts Occasional Papers in Linguistics 18 GLSA, Amherst, MA, pp. 249 384. Muller, Jennifer. 2001. The Phonology and Phonetics of Word initial G eminates Ph D dissertation, Ohio State University. Morris, Miranda 2007 literate, Non Arabic L ang uages of Oman and Yemen: Their The British Yemeni Society Journal, 39 53 Nakano, Aki'o 1986 Comparative V ocabulary of Southern Arabic: Mahri, Gibbali, and Soqotri Tokyo, Institute for the Study of Languages and Cultures of Asia and Africa. Newman, P aul 1990 Nominal and verbal plurality in Chadic Dordrecht: Foris. Palmer, Frank R obert 1955 B ulletin of S chool of O riental and A rabic S tudies 17 548 565. Pater, J oe 2000. Nonuni formity in English Stress: The Role of Ranked and Lexically Specific C onstrai Phonology 17 237 274 Pater, J oe 2004. Exceptions in Optimality Theory: Typology and Learnability Ha ndout from the Conference on Redefining Elicitation: Novel Data in Phonological Theory, New York University. (Available at exceptions.pdf) Porkhomovsky, V ictor 1997 uth Arabian L anguages fro m a Semitic and Hamito Proceedings of the Seminar for Arabian Studies 27 219 223. Prince, Alan and Paul Smolensky. 1993. Optimality Theory RUCCS TR 2. Technical Reports of the Rutgers Center for Cognitive S cience, New Brunswick, NJ. Prince, A lan and Prince Smolensky. 2004. Optimality Theory: Constraint Interaction in Generative Grammar Malden, MA, and Oxford, UK: Blackwell. [Revision of 1993 technical report, Rutgers University Center for Cognitive Scien ce. Available on Rutgers Optimality Archive 537 ] Ratcliffe, R obert 1990 lurals: Arguments for a Two fold Classification of M orpholo n M. Eid and J. McCarthy (e ds.), Perspectives on Arabic Linguistics II: Papers from the second Symp osium Amsterdam, Benjamins pp. 94 119

PAGE 241

241 Ratcliffe, R obert 1992 The Broken Plural P roblem in Arabic, Sem itic and Afro asiatic: A Solution Based on the Diachronic Application of Prosodic A nalysis Ph.D dissertation, Yale University. Ratcliffe, R ober t 1996 Bulletin of the School of Oriental an d African Studies 59 296 311. Ratcliffe, R obert .1998a Sub Journal of Near Eastern Studies 57 81 123. Ratcliffe, R obert .1998b The Broken Plural problem in Arabic and C ompara tive Semitic: Allomorphy and Analogy Non concatenative M orphology Amsterdam/ Philadelphia: John Benjamins. Ratcliffe, R obert 2002 lural S ystem of M oroccan Arabic: Diachronic and n D. B. Parkinson and E. Benmamoun (e ds.). Perspectives on Arabic Linguistics XIII XIV Amsterdam, Benjamins pp. 87 109 Ravid, D orit and Rachel Schiff. 2009 Phonological Categories of Noun Plurals in Linguistics 47 45 63. Rodgers, Kaye (ed.), Semitic Studies in Honor of Wolf Leslau on the occasion of his eighty fifth birthday Wiesbaden, Otto Harrosowitz, pp. 1323 1335. Rose, S haron .1997 Theoretical Issues in C ompara tive Ethio Semitic Phonology and Morphology, Ph. D. dissertation, McGill University. Rosenthall, S amuell 1999 n R. Kager and H. van der Hulst (e ds.), The Prosody Morphology Interface Cambridge University Press pp. 344 366 Rubin, A aron D.2008a L anguage and Linguistics Compass 2 61 84. Samek Lodovici, V ieri .1 994 n P. Ackeme and M. Schoorlemmer (e ds.), CoSole I Proceedings The Hague, Holland Academic Graphics pp. 265 283 Schuh, R ussel .1992 23rd Annual Co nference on African Linguistics 1 7 Shaaban, K assim Ali .1977 The Phonology and M orphology of Omani Arabic Ph.D. dissertation, University of Texas at Austin.

PAGE 242

242 Simeone Senelle, Marie Claude .1997 n R. Hetzron ( e d.), The Semitic Languages London, Rutledge pp. 378 423 Testen, D avid .1998 Bulletin of the School of Oriental and African Studie s 61 314 17 Testen, D avid .1992 M arker in Jibbali a nd Socotri Bulletin of the School o f Oriental and African Studies 55 445 450. Thomas, B ertram .1937 trange T o Proceedings of the British Academy 23 231 331. Ussishkin, A dam .1999 antal Root: Modern Hebrew Denominal Verbs and Output Output Correspondence Phonology 16 401 442 Urbanczyk, Suzanne. 1996. 'Morphological Templates in Reduplication', in J. Beckman et al. (eds.), NELS 26 GLSA, Amherst, MA, pp. 425 440. Vergote, Louva in 1969 Orientalia 38 77 96. Watson, J anet 2002 The Phonology a nd M orphology of Arabic Oxford University Press. Worrell, William H. 1925 American Journal of S emitic Languages and Literature 41 179 182. Worrell, William H. 1920 hei r Bearing upon the Journal of the Palestine Oriental Society 1 15 20. Xu, Zheng and Mark Aronoff. 2011 A Reali zation Optimality Theoretic Approach to Journal of Linguistics 47 ,673 707 Yip, M oris. 1988 Natural Language and Linguistic Theory 6 551 577 Yimam B aye. 1996 iscourse Journal of African Languages and Linguistics 17 47 83 Zaborski, A ndrzej and Poland Cracow. 1976a Monographic Journals of the Near East: Afro asiatic Linguistics 9 1 9. Zaborski, A ndrzej. 1986 The Morphology of the Nominal Plural in the Cushitic L anguages Vienna: Afro Pub.

PAGE 243

243 BIOGRAPHICAL SKETCH Khalsa Al Aghbari was born in Muscat, e ducation at Sul tan Qaboos University and was appointed as a teacher assistant in the Department of English, College of Arts and Social Sciences. In 2002, she was granted a full scholarship to pursue a M inguistics. She decided to join the Department of Linguis tics at University of Victoria, British Columbia, Canada. She developed immense interest in theoretical l inguistics there and wrote her the sis on phonological analysis to a linguistic phenome non observed in Omani Arabic and other dialects of Arabic too. After obtaining an MA in theoretical l inguistics, she went back to Oman and taught English, a phonology class and two tran slation courses to prospective English teachers and a rts students in a span of approximately four years. Through close interaction with students, she developed teaching and learning skills. During the same pe riod, she became interested in t ran slation and published translations to three religious books and managed to voice som e concerns about translation and its difficulties through writing and publishing a few articles in the Omani newspapers and electronically. In 2007, she was granted a scholarship to do a PhD in linguistics too. She was lucky enough to be accepted at Univer sity of Florida where she has spent four years, expanding and sharpening both her linguistic knowledge and research skills. After graduation, she will be appointed as an assistant professor in the English department at Sultan Qaboos University where she wi ll be under the happy obligation of teaching English and various linguistics courses to arts and e ducation students.