Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The syntax and semantics of gap and resumptive strategies in Iraqi Arabic D-linked content questions Sterian, Laura Andreea 2011

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
24-ubc_2011_fall_sterian_laura.pdf [ 687.96kB ]
Metadata
JSON: 24-1.0105127.json
JSON-LD: 24-1.0105127-ld.json
RDF/XML (Pretty): 24-1.0105127-rdf.xml
RDF/JSON: 24-1.0105127-rdf.json
Turtle: 24-1.0105127-turtle.txt
N-Triples: 24-1.0105127-rdf-ntriples.txt
Original Record: 24-1.0105127-source.json
Full Text
24-1.0105127-fulltext.txt
Citation
24-1.0105127.ris

Full Text

THE SYNTAX AND SEMANTICS OF GAP AND RESUMPTIVE STRATEGIES IN IRAQI ARABIC D-LINKED CONTENT QUESTIONS by  Laura Andreea Sterian B.A. in Linguistics, Simon Fraser University, 2010 M.A., University of Bucharest, 2004 B.A. in Arabic, University of Bucharest, 2002 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES (Linguistics) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) July 2011 © Laura Andreea Sterian, 2011  Abstract This thesis examines Iraqi Arabic D-linked content questions of the type “Which woman saw Ragheb?”. I develop a syntactic and semantic analysis of both the gap and resumptive strategies of such D-linked content questions. Chapter 1 provides background information on Iraqi Arabic. Chapter 2 develops the syntactic analysis: the gap strategy is treated as an instance of full DP-deletion, with the deletion site being structurally ambiguous between a D-N and a D-ϕ-N structure. I further propose that the resumptive strategy is an instance of remnant DP-deletion with a D-ϕ-N structure, and treat the resumptive pronoun as a stranded ϕ-element. Chapter 3 relates the two syntactic structures — D-N versus D-ϕ-N — to the semantic distinction between the pair-list interpretation versus a natural-function interpretation. A pair-list reading is found when a question such as “Which woman did every man invite?” is answered with a list such as: “John, Sue; Bill, Lucy, …”. A natural function reading would answer the same question with a relational noun: “His sister.” In contexts where both the gap and resumptive strategy are possible, we observe the following: the gap strategy is ambiguous between a pair-list and a natural function reading; the resumptive strategy only allows a natural function reading. I propose that the semantic ambiguity of the gap strategy reflects its structural ambiguity: if the deletion site is D-N, this corresponds to the pair-list reading; if the deletion site is D-ϕ-N, this corresponds to the natural function reading. As for the resumptive strategy, in contexts where the gap strategy is also possible, it is unambiguously interpreted with a natural functional reading; this is consistent with the syntactic remnant DP-deletion, which requires a D-ϕ-N structure. I further show that, in contexts where only the resumptive strategy is possible, economy considerations allow syntactic remnant DP-deletion to be semantically ambiguous between a pair-list and a natural function reading. Chapter 4 examines the syntactic and semantic parallels between D-linked content questions and genitive interrogatives and argues that the latter are inherently D-linked.  ii  Table of Contents Abstract.............................................................................................................................. ii Table of Contents ............................................................................................................. iii List of Tables ................................................................................................................... vii List of Abbreviations ....................................................................................................... ix Acknowledgments ............................................................................................................. x 1 Introduction.................................................................................................................... 1 1.1 Background information about Iraqi Arabic............................................................. 1 1.1.1 Geographical and historical classification of Iraqi Arabic ............................ 1 1.1.2 Word order: deriving SVO versus VSO order............................................... 3 1.1.3 Verbal inflection ............................................................................................ 4 1.1.4 Subject pro-drop............................................................................................. 5 1.1.5 Null copula..................................................................................................... 6 1.1.6 Adjectival agreement: human versus non-human nouns ............................... 7 1.2 Background information about resumptive pronouns............................................... 9 1.2.1 Defining resumption ...................................................................................... 9 1.2.2 A syntactic restriction: the highest subject restriction ................................. 11 1.2.3 A semantic restriction: on optional versus obligatory resumption .............. 13 1.3 Organization of thesis ............................................................................................. 15 2 The syntax of D-linked content questions in Iraqi Arabic ....................................... 16 2.1 Overview of the syntactic problem, analysis and consequences ............................ 17 2.1.1 The syntactic problem: A’-dependencies show a complement noncomplement asymmetry ........................................................................................ 17 2.1.2 The syntactic analysis: the extraction site can have two distinct structures 18 2.1.3 Consequences of the syntactic analysis ....................................................... 19 2.2 D-linked versus non-D-linked interrogative expressions ....................................... 19 2.3 The inventory of interrogative expressions in Iraqi Arabic content questions ....... 20 2.3.1 D-linked ya: ‘which’ is unrelated to other interrogative expressions.......... 21 2.3.2 D-linked ya: ‘which’ is in complementary distribution with other Delements ................................................................................................................ 22 2.3.3 D-linked ya: ‘which’ requires an overt lexical noun after it........................ 24 iii  2.3.4 D-linked ya: ‘which’ occurs in a wider range of contexts than bare interrogative expressions ...................................................................................... 26 2.4 Extraction from argument positions, adjunct positions, and islands ...................... 29 2.4.1. Extraction from argument positions: subjects, direct objects, prepositional objects ................................................................................................................... 29 2.4.2 Extraction from adjunct positions: temporal and locative adjuncts............. 33 2.4.3 Extraction from islands: wh-islands and adjuncts islands ........................... 35 2.5 Syntactic analysis of the resumptive strategy ......................................................... 37 2.5.1 Previous analyses of resumption.................................................................. 37 2.5.2 The inventory of pronouns in Iraqi Arabic .................................................. 38 2.5.3 Internal structure of the resumptive pronoun............................................... 39 2.5.4 Resumption as remnant DP-deletion ........................................................... 42 2.6 Syntactic analysis of the gap strategy ..................................................................... 46 2.6.1 Internal structure of the DP.......................................................................... 46 2.6.2 The gap strategy as full DP-deletion............................................................ 47 2.7 Implications of the analysis for extraction.............................................................. 49 2.7.1 Why resumption isn’t possible with subject extraction ............................... 49 2.7.2 Why resumption is possible with prepositional object extraction ............... 53 2.7.3 Why resumption is obligatory with extraction from islands........................ 58 2.8 Implications of the analysis for reconstruction ....................................................... 63 2.8.1 Scope reconstruction with a quantificational antecedent............................. 64 2.8.2 Binding reconstruction with a referential antecedent .................................. 66 2.8.3 Reconstruction with resumption .................................................................. 67 2.8.4 Scope Reconstruction of Iraqi D-linked interrogatives ............................... 69 2.8.5 Binding Reconstruction of Iraqi D-linked interrogatives ............................ 72 3 The semantics of D-linked content questions in Iraqi Arabic ................................. 78 3.1 Overview of the syntax-semantics interface problem and a possible solution ....... 78 3.1.1 The syntax-semantics interface problem ..................................................... 78 3.1.2 A possible solution to the syntax-semantics interface problem................... 80 3.2 Semantic correlates of the gap and the resumptive strategy ................................... 82 3.2.1 The distinction between de dicto and de re readings................................... 82  iv  3.2.2 The distinction between single-individual and multiple-individual readings ............................................................................................................................... 86 3.2.3 The distinction between individual, natural function and paired-list readings ............................................................................................................................... 88 3.2.4 The gap strategy competes with the resumptive strategy ............................ 91 3.3 What competition reveals about the syntax-semantics interface ............................ 93 3.3.1 The natural function reading subsumes the pair-list reading ....................... 94 3.3.2 The gap strategy is always syntactically and semantically ambiguous ....... 95 3.3.3 The weak resumptive strategy is sometimes syntactically and semantically ambiguous ............................................................................................................. 96 3.3.4 Implications of the analysis: partial copying ............................................... 98 4 Genitive interrogatives as inherently D-linked content questions........................... 99 4.1 The problem, the analysis and consequences ......................................................... 99 4.1.1 The contrast between genitive and bare interrogatives................................ 99 4.1.2 The analysis: genitive interrogatives are inherently D-linked ................... 101 4.1.3 Consequence: D-linking arises whenever there is an overt domain restriction ............................................................................................................................. 101 4.2 Comparing bare, genitive and D-linked interrogatives......................................... 103 4.2.1 The resumptive and gap strategy: local extraction .................................... 103 4.2.1.1 Local extraction with bare interrogatives ............................................... 103 4.2.1.2 Local extraction with genitive interrogatives ......................................... 105 4.2.1.3 Local extraction with D-linked interrogatives ........................................ 107 4.2.1.4 Summary of the local extraction data ..................................................... 109 4.2.2. The resumptive and gap strategy: long-distance extraction...................... 110 4.2.2.1 Long-distance extraction with bare interrogatives.................................. 110 4.2.2.2 Long-distance extraction with genitive interrogatives............................ 112 4.2.2.3 Long-distance extraction with D-linked interrogatives .......................... 114 4.2.2.4 Summary of the long-distance extraction data........................................ 116 4.2.3 Superiority effects...................................................................................... 117 4.3 Why genitive interrogatives are inherently D-linked............................................ 120 4.3.1 D-linking arises if there is an overt domain restriction.............................. 120  v  4.3.2 Derivation of the genitive interrogative with the gap strategy .................. 121 4.3.3 Derivation of the genitive interrogative with the resumptive strategy ...... 124 5 Conclusions................................................................................................................. 128 5.1 Results................................................................................................................... 128 5.2 Unsolved problems ............................................................................................... 128 References ...................................................................................................................... 130  vi  List of Tables Table 1. The paradigm of inflected verbs in Iraqi Arabic................................................... 5 Table 2. Distribution of natural function and pair-list readings in Jordanian Arabic content questions (adapted from Malkawi 2009).............................................................. 14 Table 3. Distribution of natural function and pait-list readings in Iraqi Arabic D-linked Content Questions............................................................................................................. 15 Table 4. Complement/ Non-complement Aymmetry with Gap versus Resumptive Strategy ............................................................................................................................. 18 Table 5. Structural Differentiation of Extraction Sites with Iraqi D-linked Content Questions........................................................................................................................... 18 Table 6. The inventory of Iraqi Arabic interrogative expressions .................................... 21 Table 7. The paradigm of the demonstrative pronouns in Iraqi Arabic............................ 22 Table 8. Interrogative strategies in Iraqi Arabic ............................................................... 28 Table 9. Resumption and Gap strategies in Iraqi Arabic mono-clausal and bi-clausal Dlinked content questions.................................................................................................... 29 Table 10. Adjunct extraction with respect to the resumptive strategy and the gap strategy in Iraqi Arabic mono-clausal and bi-clausal content questions ........................................ 35 Table 11. The paradigm of personal pronouns in Iraqi Arabic......................................... 39 Table 12. Scope reconstruction with resumptive pronoun................................................ 72 Table 13. Binding reconstruction with resumptive pronoun ............................................ 76 Table 14. Structural Differentiation of Extraction Sites with Iraqi Arabic D-linked content questions ........................................................................................................................... 79 Table 15. Interpretations associated with Iraqi Arabic D-linked content questions ......... 80 Table 16. Syntax-semantics mapping for Iraqi Arabic D-linked content questions ......... 81 Table 17. de dicto/ de re contrast for Iraqi Arabic bare interrogatives (non D-linked content questions) using gap and resumptive strategies ................................................... 86 Table 18. Single-individual/ multiple-individual contrast in Modern Hebrew................. 87 Table 19.Interpretations associated with Iraqi Arabic D-linked content questions .......... 92 Table 20. Economy hierarchy for gap and resumptive pronouns in Iraqi Arabic and Jordanian Arabic (adapted from Malkawi 2009) .............................................................. 93 vii  Table 21. Extensional/ Intensional contrast and the gap/ resumptive strategy in Iraqi Arabic (Context 1: optional resumption) .......................................................................... 95 Table 22. Internal structure of Iraqi Arabic bare, genitive and D-linked interrogatives 101  viii  List of Abbreviations S  singular  Pl  plural  M  masculine  F  feminine  Acc  Accusative  ix  Acknowledgments This thesis could not have been completed without the help of a number of dedicated people. I am deeply grateful to my thesis committee: Rose-Marie Déchaine (chair), Martina Wiltschko, Michael Rochemont. I especially would like to thank Dr. Déchaine for embarking with me on the journey to explore Iraqi Arabic, also for urging me to be thorough in eliciting the data, for her invaluable guidance, insight and encouragement. I would like to thank Dr. Wiltschko for her illuminating suggestions and to Dr. Rochemont for encouraging me to explore various avenues. I would also like to thank my consultant Suha Al-Sudani for her availability and patience. I would like to thank my father for encouraging me to continue my graduate studies and to the rest of my family for their support.  x  1 Introduction The goal of this thesis is to describe and analyze the syntax and semantics of resumption in content questions in Iraqi Arabic. In this chapter, I give general information about Iraqi Arabic (§1.1) and I describe the phenomenon of resumption with a preview of the thesis (§1.2.).  1.1 Background information about Iraqi Arabic After providing information about the geographical and historical situation of Iraqi Arabic (§1.1.1), I present some of the linguistic background that is relevant to the concerns of this thesis, focusing on word order (§1.1.2), verbal inflection (§1.1.3), subject pro-drop (§1.1.4), occurrence of the null copula (§1.1.6) and adjectival agreement (§1.1.6).  1.1.1 Geographical and historical classification of Iraqi Arabic Iraqi Arabic (‫ عراقي‬ʕiraqi; formally: ‫ اللغة العربية العراقية‬al-luɣat al-ʕarabiyya al-ʕiraqiyya), also known as Mesopotamian Arabic [ISO 639-3]) is a variety of Arabic spoken in the Mesopotamian basin of Iraq, from Baghdad south, as well as in Khuzestan Province of Iran and eastern Syria. The linguistic situation in the Arab countries is largely characterized by diglossia, given by the coexistence of Classical Arabic (or Modern Standard Arabic) and colloquial Arabic. Modern Standard Arabic is the language of the mass-media, universities, conference discussions, lectures and literature across all Arab countries. The unity of the Arab culture across political boundaries is given by the language, hence the privileged role of Modern Standard Arabic. Colloquial Arabic is the language spoken by people in their everyday activities in all social occasions; it is spoken as mother tongue. It displays 1  varieties given by geographical distribution, education, religion. There is a predominant colloquial dialect in each Arab country, given by the variety spoken in the capital. In the case of Iraqi Arabic, the predominant colloquial dialect is the Baghdadi one. A distinction is recognised between Mesopotamian Qeltu Arabic and Mesopotamian Gelet Arabic, the appellations deriving from the form of the word ‫قلت‬ (qeltu/ gelet, meaning ‘I said’). The Qeltu group includes the Anatolian dialect cluster, also known as North Mesopotamian Arabic or Maslawi (Mosul Arabic), and Jewish and Christian dialects. The Gelet group includes a Tigris dialect cluster, of which the bestknown form is Baghdadi Arabic, and a Euphrates dialect cluster, known as Furati (i.e. Euphrates Arabic) which has some affinities to Gulf Arabic. Both the Gelet and the Qeltu varieties of Iraqi Arabic have some speakers in the extreme eastern parts of Syria. Some features of rural Palestinian Arabic, such as the pronunciation of ‫ ك‬kaf as [tʃ], are also shared with Iraqi Arabic. Medieval Iraqi Arabic appears to have been of the Qeltu type; Blau (1965) considers that the gelet features in mainstream Iraqi Arabic today are the result of a process of "re-Bedouinization"; the affinity to Gulf Arabic, and the persistence of qeltu features in the Jewish and Christian dialects, offer some support to this view. In modern day Iraq there are three main varieties of spoken Arabic: the northern (Mosul Arabic), southern (Basra Arabic) and Baghdadi Arabic. The dialect investigated in this thesis is Baghdadi Arabic. The examples I give in this thesis from Iraqi Arabic are first written in the Arabic alphabet, then they are followed by their phonological realization, by the gloss and finally by their translation into English. I adopt an etymological writing in Arabic, which means a writing that is faithful to Modern Standard Arabic, and in the following I explain why. There are phonological differences between Modern Standard Arabic and the Iraqi dialect: for example, the voiceless velar stop /k/ is pronounced as voiceless affricate /ʧ/ in some contexts (Alkalesi 2006). Since this phonological phenomenon is predictable, when I wrote the Iraqi Arabic sentences with the Arabic alphabet, I chose to maintain the writing of the words with ‫ ك‬/k/ as they would appear in Modern Standard Arabic, in order to ease the reading and the understanding for those who can read Arabic, but are not Iraqis. This also makes it possible for the root to be easily identifiable as it appears across all varieties of Arabic. Also, there are other predictable phonological processes, for 2  example where the enclitic pronoun /hu/ attaches to the verb, the /u/ is pronounced in Modern Standard Arabic, but in most vernaculars the /u/ is deleted in the phonology. I also chose to be faithful to the writing of Modern Standard Arabic, because the phenomenon is predictable and because it is relevant to know that the enclitic pronoun exists there regardless of how vernaculars decide to pronounce it (1). (1) ENCLITIC PRONOUN a. Iraqi Arabic enclitic pronoun  ‫درسته‬ darasta=h studied.1S=3MS 'I studied it.'  b. Classical/ Modern Standard Arabic  ‫درسته‬ darastu=hu studied.1S=3MS 'I studied it.' The transliteration in Latin script is however faithful to the phonology, since this is the accepted practice in rendering the gloss of languages.  1.1.2 Word order: deriving SVO versus VSO order The basic word order of Arabic is VSO; however as early as Classical Arabic SVO is also mentioned in the grammars of Arabic. In modern times, spoken Arabic varieties have been reported to show both VSO and SVO. The VSO/SVO alternation in word order in spoken Arabic it itself a lively topic of study, but is not discussed in this thesis. With VSO order, the verb agrees with the subject in gender, but not in number, as shown in (2a). This is sometimes called “weak agreement”, and is associated with the  3  derivation in (2b). In particular, note that the verb ʃa:f ‘saw.3MS’ raises all the way to Infl, while the subject il-asa:tiδa ‘the professors, masculine’ remains in a VP-internal position.  (2) a. Verb-Subject-Object  .‫شاف األساتذة الكتاب بالمكتبة‬ ʃa:f  il-asa:tiδa  saw.3MS the-professors.M  il-kitab  bi-l-maktaba  the-book in-the-library  'The professors saw the book in the library.' (16 FEB 2011, SA offered freely) b. representation of VSO in Iraqi Arabic [IP [I ʃa:f] [VP [DP il-asa:tiδa] [V ʃa:f] [DP [D il] [N kitab]]] [PP bi-l-maktaba]]] With SVO order, the verb agrees with the subject in both gender and number, as shown in (3a). This is sometimes called “strong agreement”, and is associated with the derivation in (3b). In particular, note that as before the verb ʃa:fwu ‘saw.3MPl’ raises to Infl, and the subject il-asa:tiδa ‘the professors’ raises to SpecIP.  (3) a. Subject-Verb-Object  .‫األساتذة شافوا الكتاب بالمكتبة‬ il-asa:tiδa  ʃa:fwu  the-professors.M saw.3MPl  il-kita:b  bi-l-maktaba  the-book in-the-library  'The professors saw the book in the library.' (16 FEB 2011, SA offered freely)  b. representation of SVO in Iraqi Arabic [IP [DP il-asa:tiδa] [I ʃa:fwu] [VP [DP il-asa:tiδa] [Vʃa:f] [DP il-kita:b]] [PP bi-l-maktaba]]]]  1.1.3 Verbal inflection  4  Arabic words have a three consonantal root which carries the meaning. The vowel melody as well as some affixes carry the inflection. The paradigm of an inflected verb in the perfect aspect is given in Table 1 below.  Person and number  The verb ‫ كتب‬kitab "to write"  1S  ‫ كتبت‬kitab=it  2MS  ‫ كتبت‬kitab=it  2FS  ‫ كتبت‬kitab=ti  3MS  ‫ كتب‬kitab=Ø  3FS  ‫ كتبت‬kitb=at  1Pl  ‫ كتبنا‬kitab=na  2Pl  ‫ كتبتو‬kitab=tu  3Pl  ‫ كتبوا‬kitba=w  Table 1. The paradigm of inflected verbs in Iraqi Arabic  1.1.4 Subject pro-drop Iraqi Arabic is a subject pro-drop language. (4a) is an example where the subject is not expressed overtly and (4b) and example where the subject is expressed overtly.  (4) a. subject not expressed overtly  .‫شافت راغب بالمكتبة البارحة‬ ʃa:fət  Ragheb bi= l-maktaba il-ba:riHa.  saw.3FS  Ragheb in the-library yesterday  'She saw Ragheb in the library yesterday.' (8 MAR 2011, SA offered freely) b. subject expressed overtly  .‫إيمان شافت راغب بالمكتبة البارحة‬ 5  Iman ʃa:fət  Ragheb bi= l-maktaba il-ba:riHa.  Iman saw.3FS  Ragheb in the-library yesterday  'Iman saw Ragheb in the library yesterday.' (8 MAR 2011, SA offered freely)  1.1.5 Null copula Arabic is characterized by the absence of the copula when the state described is an unfinished act or state (the Imperfect aspect); this has been rendered by grammarians in the Indo-European languages in the present tense. The Arabic grammarians consider this a nominal sentence, because it starts with a noun. However, the verb to be surfaces when the state described is finished or complete (the Perfect) rendered in Indo-European languages in the past tense. This is illustrated in (5a) for the imperfect state and in (5b) for the perfect state.  (5) a. The copula - Imperfect state  .‫بغداد مدينة بھا ھواية عجائب‬ Baghdad  madiyna bi=ha hawa:ya ʕaʤaɁib.  Baghdad city.F  in=3F many  marvels  'Baghdad is a city in which there are many marvels.' (8 MAR 2011, SA offered freely)  b. The copula - Perfect state  .‫كانت بغداد مدينة بھا ھواية عجائب‬ ʧa:nat  Baghdad  madiyna bi=ha hawa:ya ʕaʤaɁib.  was.3FS  Baghdad  city.F  in=3F many  marvels  'Baghdad was a city in which there are many marvels.' (8 MAR 2011, SA offered freely)  6  1.1.6 Adjectival agreement: human versus non-human nouns Arabic has two grammatical genders, masculine and feminine, and only human nouns show gender agreement. The verbs agrees with the subject in gender, when the verb precedes the subject, i.e., with VSO order, as shown above in (2a). Predicate adjectives also show agreement with their subjects. This is shown in (6) for masculine agreement. In (6a), ʃa:Tir ‘smart, masculine’ agrees with the noun aTTa:lib ‘the student, masculine’. In (6b), ʃa:Tiri:n ‘smart, masculine plural’ agrees with the noun aT-Tulba ‘the students, masculine’. (7) shows the same contrast with feminine agreement. In (7a), ʃa:Tira ‘smart, feminine’ agrees with the noun aT-Ta:liba ‘student, feminine’. In (7b), ʃa:Tira:t ‘smart, feminine plural’ agrees with the noun aT-Taliba:t ‘students, feminine’.  (6) adjective - noun agreement with human noun a. .‫الطالب إللي حكيت ويه شاطر‬ aT-Ta:lib  illyi  Haʧi:t  wuya=h  ʃa:Tir  the-student.MS who.Rel spoke.1Sg with=3MS smart.MS 'The student with whom I spoke is smart.' b. .‫الطلبة اللي حكيت وياھم شاطرين‬ aT-Tulba  illyi  Haʧi:t  the-students.M who.Rel spoke.1S  wuya=huma  ʃa:Tiri:n  with=3MPl  smart.MPl  'The students with whom I spoke are smart.' (8 MAR 2011, SA offered freely)  7  (7) adjective - noun agreement with human noun a. .‫الطالبة إللي حكيت وياھا شاطرة‬ aT-Ta:liba  illyi  Haʧki:t  wuya=ha  ʃa:Tira  the-student.FS who.Rel spoke.1Sg with=3FS  smart.FS  'The student with whom I spoke is smart.' (5 JUL 2011, SA offered freely) b. .‫الطالبات اللي حكيت وياھم شاطرات‬ aT-Taliba:t  illyi  Haʧi:t  wuya=huma  the-students.F who.Rel spoke.1S  with=3FPl  ʃa:Tira:t smart.FPl  'The students with whom I spoke are smart.' (5 JUL 2011, SA offered freely) However, with non-human nouns, gender agreement shows a different pattern. With a singular subject, the predicate adjective shows gender agreement, as in (8a) and (9a). But with a plural subject, the predicate adjective always appears in the in feminine singular, as shown in (8b) and (9b).  (8) adjective - noun agreement with non-human noun a. .‫بيت سھى كبير‬ beyt  Suha ʧibi:r  house.MS  Suha big.MS  'Suha's house is big.' b. .‫بيوت أھل سھى كبيرة‬ buyu:t  ʔahl  house.MPl family.MS  Suha  kabi:ra  Suha  big.FS  'Suha's family's houses are big.' (5 JUL 2011, SA offered freely)  8  (9) adjective - noun agreement with non-human noun a. .‫شفت الجزيرة الصغيرة من السفينة‬ ʃuft  il-ʤazi:ra  iS-Saġi:ra  min is-safi:na  saw.1S the-island.FS the-small.FS from the-ship ‘I saw the small island from the ship.’  b. .‫شفت الجزائر الصغيرة من السفينة‬ ʃuft  il-ʤaza:ʔir  iS-Saġi:ra  min is-safi:na  saw.1S the-islands.FPl the-small.FS from the-ship ‘I saw the small island from the ship.’ (5 JUL 2011, SA offered freely)  1.2 Background information about resumptive pronouns This section presents the resumptive pronoun strategy. After defining resumption (§1.2.1), I introduce a well-known syntactic restriction on resumption, namely the highest subject restriction (§1.2.2) and then illustrate a restriction on resumption that is sensitive to whether resumption is obligatory or optional (§1.2.3).  1.2.1 Defining resumption Rouveret (2011) defines a “resumptive pronoun” as follows: The overt pronominal element found in some languages in the variable position of unbounded A’dependency constructions—the latter include relative clauses, constituent questions, comparative clauses, dislocation and focus constructions. Given this characterization of resumption, for the purposes of this thesis, I adopt the following working definitions:  9  (10) A-position (argument position): a position to which a thematic role is assigned. (11) A’-position (non-Argument position): a position to which a non-thematic role is assigned. (12) A’-dependency: An A’-dependency is a dependency relation between two syntactic objects, one of which is in an A’-position, and the other is in an A-position. (13) A pronominal element is: a. a weak pronoun; b. a strong pronoun; c. a weak pronoun doubled by a strong pronoun; or d. an epithet. (14) variable position: the thematic A-position that corresponds to the extraction site of an A’-dependency. With these definitions in mind, consider the examples in (15), which illustrate an A’-dependency in Iraqi Arabic in the context of relativization. In (15a), the pronominal element –hu ‘him’ (a weak pronoun) occupies the thematic A-position of the moved constituent illyi ‘whom’, which is in an A’-position. Thus, there is an A’-dependency between the interrogative operator illyi and the weak pronoun –hu ‘him’; in other words, –hu is a resumptive pronoun. In (15b), there is a gap in the thematic A-position: in the context of relativization, the gap strategy is ill-formed in Iraqi Arabic.  (15) a. resumptive strategy  .‫الرجال إللي شفته ببيت سھى كاتب عظيم‬ ir-riʤa:l  illyi  ʃuft=hu  bi=beyt  the-man  whom  saw.1S=3MS  in house Suha writer  Suha ka:tib  ʕaδˀyim great  'The man whom I saw [him] at Suha's house is a great writer.'  10  b. gap strategy  * .‫الرجال إللي شفت ببيت سھى كاتب عظيم‬ *ir-riʤa:l the-man  illyi  ʃuft____  whom  saw.1S_____ in house Suha writer  bi=beyt Suha ka:tib  ʕaδˀyim great  'The man whom I saw _____ at Suha's house is a great writer.' Traditionally, the gap strategy is analyzed as the result of movement of a constituent to an A’-position, leaving behind a gap or a trace. The resumptive strategy has often been analyzed as the result of a binding relation between an antecedent in A’position and a pronoun in an A-position. On this view, while the gap strategy involves movement, the resumptive strategy does not (Sells 1984; McCloskey 1990). Typologically, the resumptive strategy is widely attested in VSO languages. Semitic languages, which are VSO, are no exception to this generalization1.  1.2.2 A syntactic restriction: the highest subject restriction The previous literature on resumption has established that, in many languages that use the resumptive strategy, there is an additional syntactic restriction that prohibits resumptive pronouns from occurring in the highest subject position (Borer 1984, McClosky 1990, Rouveret 2011, to appear). This is called the highest subject restriction. For Iraqi Arabic, a question that arises regarding the highest subject restriction concerns the syntactic status of subject agreement on the verb. On the one hand, if agreement is treated as a pronominal element, then subject agreement will have the status of a resumptive element, and Iraqi Arabic would be analyzed as not exhibiting the highest subject restriction. But if agreement is not a pronominal element for the purposes of establishing an A’-dependency, then Iraqi Arabic would be analyzed as exhibiting the highest subject restriction. To chose between these two alternatives, one needs to examine island contexts, which obligatorily require resumption. If, in such island  1  The Celtic languages are another language family that makes regular use of the resumptive strategy, and which also has VSO word order.  11  contexts, subject agreement by itself is sufficient to license an A’-dependency, then one can conclude that subject agreement has the status of a pronominal element for the purposes of resumption. But if an additional pronoun is needed in such contexts, then one can conclude that subject agreement is not a pronominal element for the purposes of resumption. The relevant data are given in (16) and (17) for wh-islands and adjunct islands respectively. In (16a) and (17a), where there is subject agreement only, extraction from the island is illicit. And in (16b) and (17b), where an overt subject pronoun is present, extraction from the island is licit. On the basis of such contrasts, I conclude that subject-verb agreement in Iraqi Arabic does not qualify as a pronominal element for the purposes of resumption.  (16) WH-ISLAND a. no overt pronoun  *‫منو يتساءل راغب ليش باست بھجت بالحفلة؟‬ *minnu: ytasa:ʔil who  Ragheb  wonder.3MS Ragheb  le:ʃ ___ ba:sit  Behjet bi-l-Hafla  why ____ kissed.3FS Behjet at-the-party  '*Who is Ragheb wondering why ___ kissed Behjet at the party ?'  b. overt pronoun  ‫منو يتساءل راغب ليش ھي باست بھجت بالحفلة؟‬ minnu: ytasa:ʔil who  Ragheb  wonder.3MS Ragheb  le:ʃ hyi ba:sit  Behjet bi-l-Hafla  why she kissed.3FS Behjet at-the-party  'Who is Ragheb wondering why she kissed Behjet at the party ?' (17) ADJUNCT ISLAND a. no overt pronoun  *‫منو قال سامر لراغب شوقت راح تمشي لبغداد ؟‬ *minnu: ga:l who  Samer li-Ragheb  ʃw:aget ____ ra:H timʃi:  said.3MS Samer to-Ragheb when____  li-Baghdad  will go.3FS to-Baghdad  'Who did Samer say to Ragheb when ___ will go to Baghdad ?'  12  b. overt pronoun  ‫منو قال سامر لراغب شوقت ھي راح تمشي لبغداد ؟‬ minnu: ga:l who  Samer li-Ragheb  ʃw:aget hyi ra:H timʃi:  said.3MS Samer to-Ragheb when  li-Baghdad  she will go.3FS to-Baghdad  'Who did Samer say to Ragheb when she will go to Baghdad ?'  1.2.3  A  semantic  restriction:  on  optional  versus  obligatory  resumption Previous work on resumption in other varieties of Arabic includes Lebanese (Aoun et al. 2001; Choueiri 2003), Palestinian (Shlonsky 1992), Egyptian (Demirdache 1991) and Jordanian (Guilliot 2006; Malkawi 2009). For Jordanian Arabic, Malkawi (2009) investigates all syntactic contexts where the resumptive strategy is employed — including left dislocation, relativization and content questions — and examines both the syntax and the semantics of resumption, specifically contrasting contexts where resumption is optional versus those where it is obligatory. The interpretive difference that he observes relates to interrogative contexts, where a distinction is made in the semantics literature between the “natural function reading” versus the “pair-list reading” of a question as in (18).  (18)  Which woman did every man invite?  a. NATURAL FUNCTION READING: His sister b. PAIR-LIST READING: John, Sue; Bill, Lucy… As summarized in Table 2, in contexts where resumption is optional in Jordanian Arabic, the gap strategy is ambiguous between the natural functional reading and the pair-list reading. But the resumptive strategy (with a weak pronoun) is compatible only with the natural function reading. In addition, Jordanian Arabic has the peculiarity of allowing two distinct resumptive strategies: resumption with a weak pronoun and resumption with a weak pronoun doubled by a strong pronoun. These two strategies come  13  to the fore in contexts where resumption is obligatory: Malkawi observes that, when resumption is obligatory, resumptive with only the weak pronoun is ambiguous between the two readings, while resumptive with a doubled pronoun is only compatible with the natural function reading. NATURAL  PAIR  FUNCTION  -LIST  READING  READING  OPTIONAL  GAP  √  √  RESUMPTION  WEAK PRONOUN  √  x  (n/a)  (n/a)  (n/a)  (n/a)  WEAK PRONOUN DOUBLED BY STRONG PRONOUN  OBLIGATORY  GAP  RESUMPTION  WEAK PRONOUN  √  √  WEAK PRONOUN DOUBLED BY STRONG  √  x  PRONOUN  Table 2. Distribution of natural function and pair-list readings in Jordanian Arabic content questions (adapted from Malkawi 2009) Malkawi’s findings for Jordanian Arabic are relevant to the present thesis, which focuses on D-linked content-questions. As we shall see in Chapter 3, in such contexts, Iraqi Arabic allows both the gap strategy and the resumptive strategy. In other words, in Iraqi Arabic, D-linked content questions are contexts where resumption is optional. And like Jordanian Arabic, the distribution of the natural function reading and the pair-list reading is sensitive to whether resumption is optional and obligatory. However, the two varieties differ in which resumptive strategies are available: while Jordanian Arabic has two resumptive strategies (weak pronoun and weak pronoun doubled by strong pronoun), Iraqi Arabic only has one resumptive strategy (weak pronoun). My findings for Iraqi Dlinked content questions are summarized in Table 3.  14  NATURAL FUNCTION  PAIR-LIST  READING  READING  OPTIONAL  GAP  √  √  RESUMPTION  WEAK PRONOUN  √  x  OBLIGATORY  GAP  (n/a)  (n/a)  RESUMPTION  WEAK PRONOUN  √  √  Table 3. Distribution of natural function and pair-list readings in Iraqi Arabic Dlinked Content Questions  1.3 Organization of thesis The thesis is organized as follows. Chapter 2 presents the morphology and syntax of interrogative expressions in Iraqi Arabic and develops a syntactic analysis of the gap and resumptive strategies for D-linked content questions. In particular, I argue that the gap strategy involves full DP-deletion and is structurally ambiguous between a D-N structure and a D-ϕ-N structure. In contrast, I propose that the resumptive strategy involves remnant DP-deletion and operates only on a D-ϕ-N structure. Chapter 3 extends the syntactic analysis to account for the interpretive differences between the two strategies. I propose that the D-N structure correlates with the pair-list reading, while the D-ϕ-N correlates with the natural function reading. This correctly accounts for the fact that, in contexts where resumption is optional, the gap strategy is compatible with both a natural function reading and a pair-list reading, while the resumptive strategy only supports the natural function reading. And in contexts where resumption is obligatory, both readings are available: I conclude that this indicates that, as a last resort, the stranded ϕ-element that corresponds to the resumptive pronoun, may be ignored for purposes of interpretation. Chapter 4 turns to genitive interrogative constructions in Iraqi Arabic. I argue that they are inherently D-linked; this accounts for why they parallel overtly D-linked content questions. Finally, chapter 5 concludes the thesis and suggests directions for future research.  15  2  The  syntax  of  D-linked  content  questions in Iraqi Arabic Content questions are questions that contain an interrogative expression. The interrogative expression may be bare (as in who, what, where), or it may be D-linked (as in which person, which table, which city). Examples of Iraqi Arabic bare interrogative expressions are given in (19); examples of D-linked interrogatives are given in (20). (19) BARE INTERROGATIVE EXPRESSIONS a. ‫منو شفتي ؟‬ minnu: ʃeftyi ? who  saw.2FS  Whom did you see ? b. ‫شنو شفتي ؟‬ ʃinnu: ʃeftyi what saw.2FS What did you see ? (20) D-LINKED INTERROGATIVE EXPRESSIONS a. ‫يا مرية شفت ؟‬ ya:  mraya  ʃeftyi  which woman saw.2FS Which woman did you see ? b. ‫يا فلم شفت ؟‬ ya:  film ʃeftyi  which film saw.2FS Which film did you see ? 16  In this chapter I develop a syntactic analysis of the resumptive and gap strategy of Iraqi Arabic D-linked content questions. First, I present an overview of the problem, the proposed analysis and the consequences of the analysis (§2.1). Then I turn my attention to the distinction between bare interrogative expressions and D-linked interrogative expressions (§2.2). This is followed by a discussion of the distribution of D-linked interrogative expression in Iraqi Arabic (§2.3). Next, I present an analysis of the resumptive strategy (§2.4) and the gap strategy (§2.5) in terms of the copy theory of movement. Then I discuss the implications of the analysis for extraction from subject position (§2.6), for extraction from adjunct position (§2.7), and for reconstruction (§2.8).  2.1  Overview  of  the  syntactic  problem,  analysis  and  consequences Before going into the details, I present an overview of the syntactic problem that presents itself with Iraqi Arabic D-linked content questions, the proposed analysis and the consequences of the analysis.  2.1.1 The syntactic problem: A’-dependencies show a complement non-complement asymmetry In Iraqi Arabic, D-linked content questions pattern differently according to whether the extraction site is from a complement or a non-complement position. In particular, extraction from a complement position is compatible with both the gap and the resumptive strategy. In contrast, extraction from a non-complement position (subjects and PP adjuncts) allows only the gap strategy. This is summarized in Table 4:  17  EXTRACTION FROM  EXTRACTION FROM  COMPLEMENT POSITION  NON-COMPLEMENT POSITION  GAP STRATEGY  √  √  RESUMPTIVE STRATEGY  √  x  Table 4. Complement/ Non-complement Asymmetry with Gap versus Resumptive Strategy  2.1.2 The syntactic analysis: the extraction site can have two distinct structures I propose that Iraqi D-linked content questions are associated with two distinct structures. Specifically, the extraction site may be a D-N structure or a D-ϕ-N structure. I argue that the gap strategy is ambiguous between these two structures, permitting either a D-N structure or a D-ϕ-N structure with a covert ϕ. In contrast, the resumptive strategy is not structurally ambiguous: it occurs only with the D-ϕ-N structure and has an overt ϕ. This is summarized in Table 5. INTERNAL STRUCTURE OF EXTRACTION SITE  GAP STRATEGY  D-N  D-ϕ-N  √  √ (with covert ϕ)  RESUMPTIVE STRATEGY  x  √ (with overt ϕ)  Table 5. Structural Differentiation of Extraction Sites with Iraqi D-linked Content Questions  18  2.1.3 Consequences of the syntactic analysis The proposed syntactic analysis — which claims that the extraction site can be a D-N structure or a D-ϕ-N structure — accounts for the complement/ non-complement asymmetry. In particular, it correctly predicts that only complements will permit overt resumption. This is because the overt pro-ϕ which corresponds to the resumptive pronoun is a clitic and cliticization is only possible from a complement position. With noncomplement positions — subjects and PP adjuncts — there isn’t a syntactic host for the pro-ϕ clitic, so the D-ϕ-N structure with an overt pro-ϕ is not permitted. This derives the fact that extraction from a non-complement position is possible only with the gap strategy.  2.2 D-linked versus non-D-linked interrogative expressions Pesetsky (1987) notes a difference in the behaviour of interrogative expressions such as who, what and the behaviour of interrogative expressions of the which-kind. Given the peculiar behaviour of which-phrases, Pesetsky (1987) refers to them as Dlinked (for Discourse-linked); he notes that with which-phrases, the set of felicitous answers is limited to the set of objects which both speaker and hearer have in mind. It is in this sense that wh-phrases are discourse-linked. However, no such requirement is imposed on bare interrogative expressions such as who, what or how many. Pesetsky (1987) notes that which-phrases seem to function pronominally in that they are "familiar" rather than novel. To exemplify this, Pesetsky (1987) gives the examples in (21). ENGLISH (Pesetsky 1987)  (21) a. Some men entered the room. Mary talked to them. b. Some men entered the room. Which ones did Mary talk to ? c. Some men entered the room. Whom did Mary talk to ? Pesetsky's interpretation of (21) is as follows: "In (21b) it is natural, almost obligatory, to assume that the question is asking for a choice among the men who entered the room. In 19  (21c) considerations of textual connectedness make this assumption possible, but much less natural". Pesetsky (2000) also observes that while bare interrogatives show Superiority effects, D-linked interrogative expressions do not. Another notable difference between the behaviour of bare interrogative expressions and the D-linked ones is that Superiority effects appear with bare interrogative expressions and disappear with D-linked ones (Pesetsky 1987, 2000; Aoun et al 2003, Boeckx 2003). Pesetsky (2000) further acknowledges that any attempt to explain the semantic sources of the syntactic properties of D-linking will require more conclusive investigation of this issue. Much of what has been done on resumption in spoken Arabic relies heavily on examples with D-linked interrogative expressions (Aoun et al 2001, Guilliot & Malkawi 2006, 2009; Malkawi 2009). This has to do with the fact that D-linked interrogative expressions appear in questions and the co-presence of the two are used as a way of testing for the semantic distinction between a natural function reading and a pair-list reading. This will be discussed in Chapter 3.  2.3 The inventory of interrogative expressions in Iraqi Arabic content questions A first question that arises is how the D-linked pronoun ya: ‘which’ is related to other pronominal and determiner elements in Iraqi Arabic. I show that: (i) ya: is unrelated to other interrogative expressions (§2.3.1); (ii) ya: is in complementary distribution with D-elements (§2.3.2); (iii) ya: requires an overt lexical noun after it (§2.3.3); (iv) ya: ‘which’ occurs in a wider range of contexts that bare interrogative expressions (§2.3.4).  20  2.3.1 D-linked ya: ‘which’ is unrelated to other interrogative expressions Consider Table 6, which lists the inventory of Iraqi Arabic interrogative expressions. Observe that there are six non-D-linked interrogative forms, three of which contain the initial morpheme ʃ-, namely ʃinu: ‘what’, ʃuwagit ‘when’ and ʃlo:n, ‘how’. The other three non-D-linked forms — namely minnu: ‘who’, we:n ‘where’ and le:ʃ ‘why’ — don’t seem to have internal morpheme structure. As for the D-linked form ya: ‘which’, it doesn’t seem to be transparently related to any of the non-D-linked forms. This establishes that, inasmuch as there is any regular paradigmatic structure with Iraqi Arabic interrogative expressions, the D-linked interrogative form ya: is not part of that paradigm.  non D-linked forms  D-linked form  ARABIC  ROMANIZED  MORPHEME  ORTHOGRAPHY  ORTHOGRAPHY  STRUCTURE  ‫منو‬  minnu:  ???  ‘who’  ‫شنو‬  ʃinu:  ʃ-inu:  ‘what’  ‫شوقت‬  ʃuwagit  ʃ-uwagit  ‘when’  ‫شلون‬  ʃlo:n  ʃ-lo:n  ‘how’  ‫وين‬  we:n  ???  ‘where’  ‫ليش‬  le:S  ???  ‘why’  ‫يا‬  ya:  ???  ‘which’  GLOSS  Table 6. The inventory of Iraqi Arabic interrogative expressions Etymologically, the Iraqi Arabic D-linked interrogative pronoun ya: ‘which’ is related to the Classical and Modern Standard Arabic form ‫ أي‬ʔayy ‘which’. As shown in (22), I speculate that in Iraqi Arabic, the initial glottal stop has been dropped, leaving ayy, which has been resyllabified into CV ya:  21  (22)  POSSIBLE SOURCE OF IRAQI ARABIC  ya: ‘which’  ʔayy  Classical and Modern Standard Arabic  ayy  loss of initial glottal stop  ya:  metathesis/resyllabification into CV.  2.3.2 D-linked ya: ‘which’ is in complementary distribution with other D-elements The D-linked interrogative pronoun does not match with the paradigm of the demonstrative pronouns either. Table 7 shows the paradigm of the demonstrative pronouns in Iraqi Arabic.  Near  ‫ ھذا‬haδa:  this  ‫ ھذاك‬haδa:k  that  ‫ ھذي‬haδyi  this  ‫ ھذيك‬haδa:ʧ  that  MS FS  Distance  ‫ ھذولة‬Haδo:la these  Pl  ‫ ھذوالك‬Haδo:lak those  Table 7. The paradigm of the demonstrative pronouns in Iraqi Arabic Like demonstratives, D-linked ya: ‘which’ precedes the noun it combines with, (23a-b). However, D-linked ya: doesn’t co-occur with demonstratives, (23c-d). (23) a. ‫سھى يا مغني شافت بالمطعم؟‬ Suha ya:  muɣanyy ʃa:fit____ bi-l-maTʕam  Suha which singer  saw.3FS  in-the-restaurant  'Which singer did Suha see in the restaurant?’ (12 SEP 2010, SA 1a, elicited)  22  b. .‫سھى شافت ھذا المغني بالمطعم‬ Suha ʃa:fit  haδa: il-muɣanyy bi-l-maTʕam  Suha saw.3FS this  the-singer in-the-restaurant  ‘Suha saw this singer in the restaurant.’ (5 JUL 2011, SA 2b, freely given) c.* .‫سھى شافت ھذا يا المغني بالمطعم‬ *Suha ʃa:fit  haδa: ya  Suha saw.3FS  this  il-muɣanyy bi-l-maTʕam  which the-singer in-the-restaurant  *’Suha saw this which singer in the restaurant.’ (5 JUL 2011, SA 2c, elicited) d.* .‫سھى شافت يا ھذا المغني بالمطعم‬ *Suha ʃa:fit  ya  haδa:  Suha saw.3FS which this  il-muɣanyy bi-l-maTʕam which the-singer in-the-restaurant  *’Suha saw this which singer in the restaurant.’ (5 JUL 2011, SA 2d, elicited) Similarly, D-linked ya: is in complementary distribution with the proclitic definite article ‫ال‬, il- the (24). (24) a. ‫سھى يا مغني شافت بالمطعم ؟‬ Suha ya:  muɣanyy ʃa:fit____ bi-l-maTʕam  Suha which singer  saw.3FS  in-the-restaurant  'Which singer did Suha see in the restaurant?’ (12 SEP 2010, SA 1a, elicited)  23  b. .‫سھى شافت المغني بالمطعم‬ Suha ʃa:fit  il-muɣanyy bi-l-maTʕam  Suha saw.3FS the-singer in-the-restaurant ‘Suha saw the singer in the restaurant.’ (5 JUL 2011, SA 3b, freely given) c.* .‫سھى شافت ال يا مغني بالمطعم‬ *Suha ʃa:fit Suha  il-ya: - muɣanyy bi-l-maTʕam  saw.3FS the-which-singer in-the-restaurant  *’Suha saw the which singer in the restaurant.’ (5 JUL 2011, SA 3c, elicited) d.* .‫سھى شافت يا المغني بالمطعم‬ *Suha ʃa:fit Suha  ya  il-muɣanyy  bi-l-maTʕam  saw.3FS which the-which-singer in-the-restaurant  *’Suha saw the which singer in the restaurant.’ (5 JUL 2011, SA 3d, elicited) I conclude that that non-occurrence of D-linked ya: ‘which’ with demonstratives or the definite article reflects its status as a D-element2.  2.3.3 D-linked ya: ‘which’ requires an overt lexical noun after it Like other D-elements, ya: ‘which’ requires an overt lexical noun after it. This is 2  In a given context which is known to the speaker and his interlocutor, it is possible to use ya: wahed ‘which one’: Context: the consultant went to a cinema multiplex where she had a choice of watching a movie out of several movies. When she comes out of there, I ask her: ‫يا واحد شفته ؟‬ ya wa:hed ʃuft=hu which one saw.2FS=3MS Which one did you see ? Consultant’s note: This is not really used, even though it is possible to say it.  24  shown in (25) for demonstratives, in (26) for the definite article and in (27) for ya:. (25) a. .‫سھى شافت ھذاك المغني‬ Suha ʃa:fit  haδa:k il- muɣanyy  ‘Suha saw.3FS that  the-singer.’ (5 JUL 2011, SA 4a, freely given)  b.* .‫سھى شافت ھذاك‬ *Suha ʃa:fit  haδa:k  Suha saw.3FS that ‘Suha saw that’. (5 JUL 2011, SA 4a, elicited) (26) a. .‫سھى شافت المغني‬ Suha ʃa:fit  il-muɣanyy  Suha saw.3FS the-singer ‘Suha saw the singer’. (5 JUL 2011, SA 5a, freely given) a. *.‫سھى شافت ال‬ *Suha ʃa:fit  il  Suha saw.3FS the *’Suha saw the’. (5 JUL 2011, SA 5b, elicited) (27) a. ‫سھى يا مغني شافت ؟‬ Suha ya:  muɣanyy ʃa:fit  Suha which singer  saw.3FS  'Which singer did Suha see ?’ (12 SEP 2010, SA 1a, elicited)  25  b. .* ‫سھى يا شافت ؟‬ *Suha ya:  ʃa:fit  Suha which saw.3FS 'Which did Suha see ?’ (5 JUL 2011, SA 6, elicited) I conclude that the obligatoriness of an overt lexical noun after D-linked ya: ‘which’ is indicative of its syntactic status as a D-element.  2.3.4 D-linked ya: ‘which’ occurs in a wider range of contexts than bare interrogative expressions In content questions, the interrogative pronoun is moved to the left periphery, preceded by the subject as left-most topic. Aoun et al (1999) and Malkawi (2009) state that in Arabic there are three strategies to form content questions. Their examples are replicated below for Iraqi Arabic showing in (28a) the gap strategy, in (28b) the resumption strategy and in (28c) the in-situ strategy. (28) INTERROGATIVE STRATEGIES WITH D-LINKED INTERROGATIVES a. the gap strategy  ‫سھى يا مغني شافت بالمطعم؟‬ Suha ya:  muɣanyy ʃa:fit____ bi-l-maTʕam  Suha which singer  saw.3FS  in-the-restaurant  'Which singer did Suha see in the restaurant?’ (12 SEP 2010, SA 1a, elicited)  26  b. the resumption strategy  ‫سھى يا مغني شافته بالمطعم؟‬ Suha ya:  muɣanyy ʃa:fit=hu  Suha which singer  bi-l-maTʕam  saw.3FS=3MS  in-the-restaurant  'Which singer did Suha see [him] in the restaurant ?' (12 SEP 2010, SA 1b, elicited)  c. the in-situ strategy  ‫سھى شافت يا مغني بالمطعم؟‬ Suha ʃa:fit  ya:  Suha saw.3FS  muɣanyy bi-l-maTʕam  which singer  in-the-restaurant  'Suha saw which singer in the restaurant ?' (12 SEP 2010, SA 1c, elicited) However, bare interrogative expressions (i.e. non-D-linked interrogative expressions) only allow the gap strategy. This is shown in (29), where a bare interrogative expression is well-formed with the gap strategy (29a), but ill-formed with the resumptive strategy and the in-situ strategy (29b-c). (29) INTERROGATIVE STRATEGIES WITH BARE INTERROGATIVES a. the gap strategy  ‫سھى منو شافت بالمطعم ؟‬ Suha minnu: ʃa:fit_____ Suha who  bi-l-maTʕam  saw.3FS_____ in-the-restaurant  'Whom did Suha see in the restaurant ?'  27  b. the resumption strategy  *‫سھى منو شافته بالمطعم ؟‬ *Suha minnu: ʃa:fit=hu Suha who  bi-l-maTʕam  saw.3FS=3MS in-the-restaurant  'Whom did Suha see [him] in the restaurant ?'  c. the in-situ strategy  *‫سھى شافت منو بالمطعم ؟‬ *Suha ʃa:fit  minnu:  Suha saw.3FS who  bi-l-maTʕam in-the-restaurant  'Suha saw whom in the restaurant ?' (12 SEP 2010, SA 2, elicited) The patterning of the interrogative strategies described above with respect to the behaviour of the D-linked and non-D-linked interrogative expressions is given below in Table 8.  Non-D-linked  D-linked  WH  WH  √  √  Resumptive Subject - WH - Verb-Resumptive pronoun  x  √  In-situ  x  √  Strategy  Gap  Word Order  Subject - WH - Verb  Subject -  Verb - WH  Table 8. Interrogative strategies in Iraqi Arabic  28  2.4 Extraction from argument positions, adjunct positions, and islands I give an overview of extraction contexts with D-linked interrogative expressions, including extraction from argument positions (§2.4.1), extraction from adjunct positions (§2.4.2), and extraction from islands (§2.4.3).  2.4.1. Extraction from argument positions: subjects, direct objects, prepositional objects Now consider the distribution of D-linked interrogative expressions in contexts that involve argument extraction. The results are summarized in Table 9.  Mono-clausal Gap Resumption Subject Direct Object Object of Preposition  √ √ x  x √ √  Bi-clausal Gap Resumption √ √ x  x √ √  Argument Table 9. Resumption and Gap strategies in Iraqi Arabic mono-clausal and bi-  clausal D-linked content questions With D-linked interrogative expressions, subject extraction allows only the gap strategy, as in (30). In contrast, direct objects allow both the gap strategy and the resumptive strategy, as in (31). As for prepositional objects (32), they allow only resumption, as preposition stranding is ungrammatical in Arabic; also, PP-fronting allows only the gap strategy, because Arabic does not have resumptives for entire prepositional phrases.  29  (30) SUBJECT a. extraction with gap  ‫يا صديقة اشترت شقة ببغداد ؟‬ ya:  ʃigga  Sadi:ga iʃtarat_____  bi-Baghdad  which friend.F bought.3FS_____ apartment in-Baghdad 'Which friend bought an apartment in Baghdad ?' (1 DEC 2010, SA, offered freely)  b. extraction with resumption  *‫يا صديقة اشترت ھي شقة ببغداد ؟‬ *ya:  Sadi:ga iʃtarat  hi: ʃigga  bi-Baghdad  which friend.F bought.3FS she apartment in-Baghdad 'Which friend [she] bought an apartment in Baghdad ?' (1 DEC 2010, SA, elicited) (31) DIRECT OBJECT a. extraction with gap  ‫إيمان يا رجال شافت بالحفلة ؟‬ Iman ya:  riʤa:l ʃa:fit ____  Iman which man  saw.3FS____  bi-l-Hafla at-the-party  'Which man did Iman see___ at the party ?' (1 DEC 2010, SA 1I, offered freely)  b. extraction with resumption  ‫إيمان يا رجال شافته بالحفلة ؟‬ Iman ya:  riʤa:l ʃa:fit=hu  Iman which man  saw.3FS=3MS  bi-l-Hafla at-the-party  'Which man did Iman see [him] at the party ?' (1 DEC 2010, SA 1II, offered freely)  30  (32) PREPOSITIONAL OBJECT a. extraction with gap  *‫سھى يا معلم التقت بالكلية وية ؟‬ *Suha ya:  muʕallim iltagat wu:ya: ___  bi-l-kulli:a  Suha which  professor  at-the-faculty  met.3FS with ___  'Which professor did Suha meet with____ at the faculty ?' (1 DEC 2010, SA 3I, offered freely)  b. extraction with resumption  ‫سھى يا معلم التقت وياه بالكلية ؟‬ Suha ya:  muʕallim iltagat  wu:ya:=h  bi-l-kulli:a  Suha which  professor met.3FS with=3MS at-the-faculty  'Which professor did Suha meet with [him] at the faculty ?' (1 DEC 2010, SA 3II, offered freely)  c. PP-fronting  ‫سھى وية يا معلم التقت بالكلية ؟‬ Suha wu:ya: ya:  muʕallim iltagat  Suha with  professor met.3FS at-the-faculty  which  bi-l-kulli:a  'With which professor did Suha meet at the faculty ?' (1 DEC 2010, SA 3III, offered freely) With long-distance extraction - i.e. extraction from a bi-clausal environment - we observe the same contrasts. While subjects allow only the gap strategy (33), direct objects allow both the gap and resumptive strategy, as in (34). Prepositional objects (35) allow only resumption; as stated before, PP-fronting does not employ resumption.  31  (33) SUBJECT a. extraction with gap  ‫راغب يا صديقة يدري إنو اشترت شقة ببغداد ؟‬ Ragheb ya:  Sadiga  ydry  ʃigga  ennu: ____ iʃtarat  bi-Baghdad  Ragheb which friend.F know.3MS that_____ bought.3FS apartment in-Baghdad  'Which friend does Ragheb know that ___ bought an apartment in Baghdad ?' (1 DEC 2010, SA, offered freely)  b. extraction with resumption  *‫راغب يا صديقة يدري إنو ھي اشترت شقة ببغداد ؟‬ *Ragheb ya:  Sadiga  ydry  ennu: hi iʃtarat  Ragheb which friend.F know.3MS that  ʃigga  bi-Baghdad  she bought.3FS apartment in-Baghdad  'Which friend does Ragheb know that she bought an apartment in Baghdad ?' (1 DEC 2010, SA, elicited) (34) DIRECT OBJECT a. extraction with gap  ‫بھجت يا مرية يعرف إنو إيمان شافت بالحفلة ؟‬ Behjet ya:  mraya  Behjet which woman  yaʕarəf  ennu: Iman  know.3MS  that  ʃa:fit____  bi-l-Hafla  Iman saw.3FS____ at-the-party  'Which woman does Behjet know that Iman saw___ at the party ?' (1 DEC 2010, SA 5I, offered freely)  b. extraction with resumption  ‫بھجت يا مرية يعرف إنو إيمان شافتھا بالحفلة ؟‬ Behjet ya:  mraya  Behjet which woman  yaʕarəf  ennu: Iman  know.3MS  that  ʃa:fit=ha  Iman saw.3FS=3FS  bi-l-Hafla at-the-party  'Which woman does Behjet know that Iman saw [ her] at the party ?' (1 DEC 2010, SA 5II, offered freely)  32  (35) PREPOSITIONAL OBJECT a. extraction with gap  *‫سناء يا معلم تعتقد إنو سھى التقت وية بالكلية ؟‬ *Sena ya:  muʕallim taʕatagid  ennu: Suha iltagat wu:ya: ____  Sena which professor know.3FS that  bi-l-kulli:a  Suha met.3FS with ___ at-the-faculty  'Which professor does Sena know that Suha met with___ at the faculty ?' (1 DEC 2010, SA 7I, offered freely)  b. extraction with resumption  ‫سناء يا معلم تعتقد إنو سھى التقت وياه بالكلية ؟‬ Sena ya:  muʕallim taʕatagid  Sena which professor know.3FS  ennu: Suha iltagat that  wu:ya:=h  bi-l-kulli:a  Suha met.3FS with=3MS at-the-faculty  'Which professor does Sena know that Suha met with [him] at the faculty ?' (1 DEC 2010, SA 7II, offered freely)  c. PP-fronting  ‫سناء وية يا معلم تعتقد إنو سھى التقت بالكلية ؟‬ Sena wu:ya: ya:  muʕallim taʕatagid ennu: Suha iltagat  Suha with  professor think.3FS that Suha met.3FS at-the-faculty  which  bi-l-kulli:a  'With which professor does Sena know that Suha met at the faculty ?' (1 DEC 2010, SA 3III, offered freely)  2.4.2 Extraction from adjunct positions: temporal and locative adjuncts Standard Arabic does not allow resumption with non-D-linked interrogative adjuncts (i.e. non-arguments). Wahba (1984) notes for Egyptian Arabic and Malkawi (2009) for Jordanian Arabic that resumption is possible only with arguments and not possible with non-arguments; the data they show in supporting this argument is non-Dlinked adjuncts. Similarly in Iraqi Arabic, adjunct extraction allows only the gap strategy 33  with non-D-linked interrogative adjuncts, whether the content question is mono-clausal or is bi-clausal. With respect to D-linked interrogative adjuncts, D-linked manner and rationale adjuncts are already beyond the register of spoken Arabic; they belong to the Modern Standard Arabic register. The temporal adjunct of the form ya: wagit ‘which time’ is ill-formed in Iraqi Arabic, hence examples of content questions with D-linked interrogative temporal adjuncts are ungrammatical (36). The D-linked interrogative locative adjunct allows only the resumptive strategy and does not allow the gap strategy, as in (37). (36) D-LINKED TEMPORAL ADJUNCT a. extraction with gap  *‫إيمان يا وقت راحت لألردن ؟‬ *Iman ya:  wagit  Iman which time  ra:Het went.3FS  li=l-Urdun ____ to-the-Jordan ____  'What time did Iman go to Jordan ?' (9 SEP 2010, SA 1, offered freely)  b. extraction with resumption  *‫إيمان يا وقت راحت لألردن ذاك الوقت ؟‬ *Iman Iman  ya:  wagit  which time  ra:Het went.3FS  li=l-Urdun  δa:k il-wagit  to-the-Jordan then  '*When did Iman go to Jordan then ?' (9 SEP 2010, SA 1, elicited) (37) D-LINKED LOCATIVE ADJUNCT a. extraction with gap  *‫إيمان يا محلة شافت سھى ب ؟‬ *Iman ya:  maHalla  Iman which store  ʃa:fit  Suha bi____  saw.3FS Suha in____  'Which store did Iman see Suha in ____ ?' (9 SEP 2010, SA 3, elicited) 34  b. extraction with resumption  ‫إيمان يا محلة شافت سھى به ؟‬ Iman ya:  ʃa:fit  maHalla  Iman which store  Suha bi=h  saw.3FS Suha in=3MS  'Which store did Iman see Suha in [it] ?' (9 SEP 2010, SA 3, offered freely)  c. PP-fronting  ‫إيمان بيا محلة شافت سھى ؟‬ Iman bi=ya:  maHalla ʃa:fit  Iman in=which store  Suha  saw.3FS Suha  'In which store did Iman see Suha in ?' (10 JUL 2011, SA 3, offered freely) Table 10 summarizes the distribution of D-linked interrogative expression from adjunct position.  Adjunct  Temporal Locative  Gap  Resumption  x x  x √  Table 10. Adjunct extraction with respect to the resumptive strategy and the gap strategy in Iraqi Arabic mono-clausal and bi-clausal content questions  2.4.3 Extraction from islands: wh-islands and adjuncts islands Let us move to island contexts. By island (Ross 1986) we understand the syntactic domain from where extraction is not possible. The resumptive pronouns have been observed to be oblivious to island phenomena (Rouveret 2011, to appear). Island constraints also play a role in the scope of semantic operators and quantifiers. In the following examples, I use wh-islands (i.e. embedded CPs introduced by wh-constituents) 35  and adjunct islands (i.e. islands formed from an adjunct clause). Consider first extracting from a wh-island: the gap strategy is ill-formed, but resumption is possible, as in (38). Similarly, with adjunct islands, the gap strategy is ill-formed, but resumption is possible, as in (39).  (38) WH-ISLAND a. gap strategy  *‫يا صورة البنھا سامر يسأل إذا كل مرية شققت ؟‬ *ya:  Su:ra.F  li-bni=ha  Samer ysʔal  which picture.F of-son=her Samer ask.3MS  iδa kull  mraya  ʃagagat____  if every woman tore.3FS___  '*Which photo of her son did Samer wonder if every woman tore ____ ?' (13 NOV 2010, SA 14aI, elicited)  b. resumptive strategy  ‫يا صورة البنھا سامر يسأل إذا كل مرية شققتھا ؟‬ ya:  Su:ra.F  li-bni=ha  Samer ysʔal  which picture.F of-son=her Samer ask.3MS  iδa kull  ʃagagat=ha  mraya  if every woman tore.3FS=3F  'Which photo of her son did Samer wonder if every woman tore [it] ?' (13 NOV 2010, SA 14aII, elicited) (39) ADJUNCT ISLAND a. gap strategy  *‫يا صورة البنھا سامر زعل ألنو كل مرية شققت ؟‬ *ya:  Su:ra.F  li-bni=ha  Samer zaʕal  li-ennu:  which picture.F of-son=her Samer get angry because  kull  mraya  ʃagagat___  every woman tore.3FS___  '*Which photo of her son did Samer get angry because every woman tore___ ?' (13 NOV 2010, SA 14bI, elicited)  36  b. resumptive strategy  ‫يا صورة البنھا سامر زعل ألنو كل مرية شققتھا ؟‬ ya:  Su:ra.F  li-bni=ha  Samer zaʕal  li-ennu:  which picture.F of-son=her Samer get angry because  kull  mraya ʃagagat=ha  every woman tore.3FS=3FS  'Which photo oh her son did Samer get angry because every woman tore [it] ?' (13 NOV 2010, SA 14bII, elicited)  2.5 Syntactic analysis of the resumptive strategy In this section I present an analysis of the resumptive strategy in Iraqi Arabic content questions. I briefly summarize previous approaches to resumption (§2.5.1), and then consider the relation of the resumptive pronoun to the pronoun inventory (§2.5.2), the internal structure of pronouns (§2.5.3), present my analysis of resumption in terms of remnant-DP deletion (§2.5.4).  2.5.1 Previous analyses of resumption With respect to the syntax of the resumptive strategies, the earliest theories differentiate between the gap strategies which are derived by movement and the resumption strategies which are derived without movement (Sells 1984; McCloskey 1990) or they propose resumption as a last resort to save a derivation where movement is blocked (Shlonsky 1992). Other theories of resumption consider it a special kind of movement (Demirdache 1991). Within the Minimalist Program, the Phasal Agree approach analyzes the links in the resumptive chain as connected by the operation Agree (Chomsky 2000, 2007; Adger & Ramchand 2001, 2005; Rouveret 2002, 2008). The most recent analyses of resumption approach this phenomenon by considering that (i) resumptive elements are not considered as a uniform class, but their status is differentiated as strong (i.e. strong pronouns and epithets) and as weak (i.e. weak pronouns and weak pronouns doubled by a strong pronoun) (Guilliot 2006; Guilliot & 37  Malkawi 2006; Guilliot 2008; Malkawi 2009); (ii) resumptive strategies give rise to reconstruction effects (Aoun & Li 2003; Boeckx 2003; Guilliot & Malkawi 2006; Malkawi 2009; Rouveret 2011, to appear); (iii) resumptive pronouns may have different internal structures (Elbourne 2002; Déchaine & Wiltschko 2002).  2.5.2 The inventory of pronouns in Iraqi Arabic The personal pronouns in Arabic are strong and weak. The strong personal pronouns correspond to the forms used with the Nominative case, whereas the weak pronominal forms are the ones used with the Accusative and the Genitive. Arabic allows subject pro-drop and the overt presence of a subject pronoun is highly marked. Moreover, as we have already seen, resumption with subject pronouns is not possible in Iraqi Arabic. I return to this below. Since the Accusative forms are weak pronouns which cliticize to verbs and to prepositions, the resumptive strategy in which they participate is known as weak resumption (Malkawi 2009). A complete paradigm of the personal pronouns with their strong and weak forms in Iraqi Arabic is given in Table 11; of these, the Accusative forms are used in resumption in Iraqi Arabic content questions.  38  Person and number  Nominative  Accusative  Genitive  (strong pronouns)  (weak pronouns)  (weak pronouns)  1S  ‫أني‬  any  ‫ني‬-  -ny  ‫ي‬-  -y  2MS  ‫إنت‬  enta  ‫ك‬-  -ək  ‫ك‬-  -ək  2FS  ‫إنت‬  enti  ‫ك‬-  -iʧ  ‫ك‬-  -iʧ  3MS  ‫ھو‬  hu:wwa ‫ه‬-  -hu  ‫ه‬-  -hu  3FS  ‫ھي‬  hyia  ‫ھا‬-  -ha  ‫ھا‬-  -ha  1Pl  ‫إحنا‬  ʔiHna  ‫نا‬-  -na:  ‫نا‬-  -na:  2Pl  ‫انتو‬  intu:  ‫تم‬-  -tum  ‫تم‬-  -tum  3MPl  ‫ھم‬  humma  ‫ھم‬-  -hum  ‫ھم‬-  -hum  3FPl  ‫من‬  henna  ‫ھن‬-  -hunna  ‫ھن‬-  -hunna  Table 11. The paradigm of personal pronouns in Iraqi Arabic The resumptive strategy in Iraqi Arabic content questions is realized via resumptive clitic pronouns (weak pronouns). According to Malkawi's (2009) classification, the resumptive strategy which involves weak resumptive pronouns (clitics or clitics doubled by a strong pronoun) is a case of weak resumption. In the following, the resumptive strategy discussed refers only to resumptive clitic pronouns (i.e. weak resumption).  2.5.3 Internal structure of the resumptive pronoun Postal (1969) advanced the claim that pronouns are definite articles. Since then, various authors have treated pronouns as DPs (i.e. determiner phrases) and worked on their internal structure (Evans 1980; Reinhart 1983; Cardinaletti & Stark 1999). Elbourne (2002) proposed that pronouns are definite determiners whose NP-complement has undergone deletion in the phonology (i.e. The NP-deletion Theory) as in (40).  39  (40) Elbourne's (2002) internal structure of a pronoun [D pronoun [NP noun] Elbourne's NP-deletion theory is central to Guilliot & Malkawi's (2006) and Malkawi's (2009) analysis of resumptive pronouns in Jordanian Arabic. In essence, we retain for this thesis that pronouns can have an NP-complement. Déchaine & Wiltschko (2002) propose that pronouns are not primitives and that they are decomposable. They argue that there are at least three pronoun types: pro-DP, pro-φP and pro-NP, each one associated with a distinct syntactic projection as in (41). (41) DÉCHAINE & WILTSCHKO'S (2002) INTERNAL STRUCTURE OF A PRONOUN  The DP-structure in (41a) functions as an R-expression, and, according to Déchaine & Wiltschko (2002), always contains ϕP and NP as sub-constituents. The ϕPstructure in (41b) functions as bound variable, while the NP-structure in (41c) has the status of a semantic constant. Relevant to the present analysis is Déchaine & Wiltschko’s claim that the ϕ-element is what allows a pronouns to function as a bound variable. Recall that Rouveret’s (2011) definition of resumptive pronoun (introduced in §1.2.1) equate the resumptive pronoun with an A’-bound variable position. Combining this with the Déchaine & Witlschko analysis predicts that pronominal elements which function as resumptive pronouns should be able to function as bound variables. This is confirmed in Iraqi Arabic: observes that the accusative pronouns that otherwise functions as Condition 40  B pronoun — it is locally free, as shown in (42) — can also be used as a bound variable as in (43). (42) THE PRONOUN CAN BE BOUND FROM OUTSIDE ITS LOCAL DOMAIN a. condition B violation – local domain  *.1‫ شافتھا‬1‫إيمان‬ *Iman1 ʃa:fit=ha1 Iman saw.3FS=3FS ‘Iman1 saw her1.’ b. condition B observed – from outside local domain  .1‫ قالت إني شفتھا‬1‫إيمان‬ Iman1 ga:lit  ʔin=ny ʃuft=ha1  Iman said.3FS  that=1S saw.1S=3FS  ‘Iman1 said that I saw her1.’ (43) THE PRONOUN IS A BOUND VARIABLE  .‫كل واحد يقول إنو إيمان شافته‬ kull waHed1 ygwul everyone  ennu: Iman ʃa:fit=hu1  say.3MS that  Iman saw.3FS=3MS  ‘Everyone1 says that Iman saw him1.’ (SA, 1, 14 APR 2011 elicited) Amending Déchaine & Wiltschko’s (2002) analysis, I propose that D-linked interrogative expressions always have a DP-shell, but differ according to whether the intermediate ϕ-layer is present. In terms of the inputs to the numeration, there are two logical possibilities: a [D-N] structure, and a [D-ϕ-N] structure. In terms of how these structure are spelled out, I show that the gap strategy is structurally ambiguous between (44a) and (44b), while the resumptive strategy is always associated with a [D-ϕ-N], as in (45).  41  (44)  (45)  SYNTAX OF GAP STRATEGY  a.  [D-N]  b.  [D-ϕ-N]  (with covert ϕ)  SYNTAX OF RESUMPTIVE STRATEGY  [D-ϕ-N]  (with overt ϕ)  2.5.4 Resumption as remnant DP-deletion In this section I show how the resumptive strategy is derived with remnantdeletion. I use the minimalist derivational analysis which employs bare phrase structure and the operations Select, Merge, Copy and Delete. In the following, whenever I use the term "movement", I understand it not as a primitive operation, but as the combination of Copy + Merge (Hornstein et al 2006). In content questions with the resumptive strategy and D-linked interrogative expressions, I treat the D-linked interrogative expressions as having a tripartite internal structure with a D- φP - N substructure as shown in (46).  (46) internal structure of D-linked interrogative expressions [D [D-linked interrogative] [φP [φ resumptive pronoun] [N noun]]] Let us unpack this with an example of D-linked content question employing the resumptive strategy (47a) whose numeration is given in (47b).  42  (47) D-LINKED CONTENT QUESTION a. the resumptive strategy  ‫إيمان يا رجال شافته ؟‬ Iman ya:  riʤa:l ʃa:fit=hu  Iman which man  saw.3FS=3MS  'Which man did Iman see [him] ?' (1 DEC 2010, SA 1II, offered freely) b. Numeration: {TOPØ, CØ, IØ, ImanD, Sa:fitV, ya:D, huφ, ridza:lN,} In (48) I show the derivational analysis of (47a). The derivation develops by phases (i.e. the vP phase, the IP phase, etc), where each maximal projection represents a domain for the application of rules (Chomsky 1995; Wojdak 2005; Hornstein 2008). Thus, within the VP3 phase (48b), the DP is built by: (48bI) merging the pronoun hu ‘him’ with the noun riʤal 'man' and (48bII) merging that complex syntactic object with the D-linked interrogative ya: ‘which’. Then the verb V ʃa:fit 'she saw' merges with the DP ya: hu riʤa:l 'which he man' (48bIII), and cliticizaton of the pronoun hu ‘him’ immediately takes place via successive application of Copy and Delete (48b IV-V). Thus the cliticization rule of the resumptive pronoun to the verb occurs in the VP phase; the pronoun has to cliticize to the verb and this determines its early movement in the derivation, during the VP phase. The subject DP Iman then merges with the V at SpecVP (48bVI). At the IP phase (48c), the inflectional head merges with the VP (48cI), then the subject DP Iman is moved to SpecIP via successive application of Copy and Delete (48cII – III). At the CP phase (48d), the remnant of the DP ya: hu riʤa:l is moved to SpecCP via successive application of Copy and Delete (48dI-II). At the TopP phase (48e), the topical head Top merges with the CP (48eI) and the subject DP Iman is moved to SpecTopP via successive application of Copy and Delete (48eII-III).  3  It does not make a difference for my analysis if the verb is a v or a V. To keep matters simple, I label the verb as V.  43  (48) derivational analysis of (47a) a. Numeration: {TOPØ, CØ, IØ, ImanD, Sa:fitV, ya:D, huφ, ridza:lN,}  b. VP phase I. Merge <φ, N> [φ [hu φ] [riʤa:l N ] ] II. Merge <D, φ> [D [ya:D ] [huφ riʤa:lN ] ] III. Merge <V, D> [V [ʃa:fit V] [D ya:D huφ riʤa:lN ] ] IV. Copy huφ & Merge <V, φ> [V [ʃa:fit V huφ] [D ya:D huφ riʤa:lN ] ] V. Delete huφ [V [ʃa:fit V huφ] [D ya:D huφ riʤa:lN ] ] VI. Merge <D, V> [V [ImanD ] [V [ʃa:fit V huφ] [D ya:D huφ riʤa:lN ] ] ] c. IP phase I. Merge <I, V> [I [V [ImanD ] [V [ʃa:fit V huφ] [D ya:D huφ riʤa:lN ]]]] II. Copy ImanD & Merge <D, I> [I [ImanD ] [I [V [ImanD ] [V [ʃa:fit V huφ] [D ya:D huφ riʤa:lN ]]]]] III. Delete ImanD [I [ImanD ] [I [V [ImanD ] [V [ʃa:fit V huφ] [D ya:D huφ riʤa:lN ]]]]]  44  d. CP phase I. Copy [D ya:D huφ riʤa:lN ] & Merge <C, I> [C [D ya:D huφ riʤa:lN] [I [V [ImanD] [V [ʃa:fit V huφ] [D ya:D huφ riʤa:lN]]]]] II. Delete [D ya:D huφ riʤa:lN ] [C [D ya:D huφ riʤa:lN] [I [V [ImanD ] [V [ʃa:fit V huφ] [D ya:D huφ riʤa:lN]]]]] e. TopP phase I. Merge <Top, C> II. Copy ImanD & Merge <Top, C> [Top ImanD [C [D ya: hu riʤa:l] [I [V [ImanD ] [V [ʃa:fit V huφ] [D ya: hu ridZa:l ]]]]]] III. Delete ImanD [Top ImanD [C [D ya: hu riʤa:l] [I [V [ImanD] [V [ʃa:fit  V  huφ] [D ya: hu riʤa:l]]]]]]  To summarize, the resumptive strategy is analyzed as "remnant deletion", because of the deletion of the remnant-DP ya: hu riʤa:l resulted from the cliticization of the φpronoun to V. The brief description of the remnant DP-deletion process is given in (49).  (49) remnant-DP deletion in brief Remnant-DP deletion = cliticization of resumptive pronoun followed by deletion of remnant-DP Thus, in this analysis the resumptive pronoun is a stranded pro-φP deleted from the lower occurrence of the DP as in (50).  (50) the resumptive strategy as remnant DP-deletion [C [D D] [[φ [φ φ] [NP N]]... [I [V V φ] ...[D D] [[φ [φ φ] [NP N]]]]]  45  Observe that the full DP containing all three subcomponents does not surface overtly (i.e. the form ya: hu riʤal is ungrammatical). I stipulated that the motivation for the resumptive pronoun to remain overt in the syntax is that it cliticizes to the verb4.  2.6 Syntactic analysis of the gap strategy I propose that the internal structure of the interrogative DP differs with the gap and the resumptive strategy (§2.6.1), and analyze that the gap strategy involve full DP deletion (§2.6.2).  2.6.1 Internal structure of the DP I show that the gap strategy is derived with full DP-deletion. While the D-linked interrogative in the gap strategy has a D-N structure as in (51a), the D-linked interrogative expression of the resumptive strategy has a D-φ-N structure as in (51b). (51) SYNTACTIC STRUCTURE OF THE D-LINKED INTERROGATIVE EXPRESSION a. the gap structure [D D [N N]] b. the resumptive structure [D D [φ φ [N N]]]  4  Various mechanical solutions have been proposed within the Minimalist framework regarding the phenomenon of resumption, for example Adger and Ramchand (2001, 2005) and Rouveret (2002, 2008) propose that the relation between the resumptive pronoun and its antecedent is established via Agree, while Boeckx (2003) analyzes resumption as subextraction in which resumptive strategies involve Match followed by Move, and not Agree. In future research I intent to pursue these analyses and compare and contrast each one of them with the one proposed in this thesis.  46  2.6.2 The gap strategy as full DP-deletion In content questions where the gap strategy is employed, the lower occurrence of the DP is deleted in full and merged at SpecCP. Let us unpack this by considering the example in (52a) below which is a D-linked content question in which the gap strategy is employed. The numeration is given in (52b). (52) D-LINKED CONTENT QUESTION a. the gap strategy  ‫إيمان يا رجال شافت بالحفلة ؟‬ Iman ya:  riʤa:l ʃa:fit ____  Iman which man  saw.3FS____  bi-l-Hafla at-the-party  'Which man did Iman see___ at the party ?' (1 DEC 2010, SA 1I, offered freely) b. Numeration: {TOPØ, CØ, IØ, ImanD, Sa:fitV, ya:D, ridza:lN,} Notice that the numeration of (52b), which derives the D-linked content question employing the gap strategy, does not contain the φ-element. The derivation is given in (53). At the VP phase (53b), the DP is built by (53bI) merging the D-linked interrogative ya ‘which’ with the noun riʤal 'man'; then the verb ʃa:fit 'she saw' merges with the DP ya: riʤa:l 'which man' (53bII). The subject DP Iman merges with the V at SpecVP (53bIII). At the IP phase (53c), the inflectional head merges with the VP (53cI), then the subject DP Iman is moved to SpecIP via successive application of Copy and Delete (53cII-III). At the CP phase (53d), the DP ya: riʤa:l ‘which man’ is moved to SpecCP via successive application of Copy and Delete (53dIII). At the TopP phase (53e), the topical head Top merges with the CP (53eI) and the subject DP Iman is moved to SpecTopP via successive application of Copy and Delete (53eII-III).  47  (53) derivational analysis of (52a) a. Numeration: {TOPØ, CØ, IØ, ImanD, Sa:fitV, ya:D, riʤa:lN,}  b. VP phase I. Merge <D, N> [D [ya:D ] [riʤa:lN ] ] II. Merge <V, D> [V [Sa:fit V] [D ya:D riʤa:lN ] ] III. Merge <D, V> [V [ImanD ] [V [Sa:fit V] [D ya:D riʤa:lN ] ] ] c. IP phase I. Merge <I, V> [I [V [ImanD ] [V [ʃa:fit V] [D ya:D riʤa:lN ]]] II. Copy ImanD & Merge <D, I> [I [ImanD ] [I [V [ImanD ] [V [ʃa:fit V] [D ya:D riʤa:lN ]]]] III. Delete ImanD [I [ImanD ] [I [V [ImanD ] [V [ʃa:fit V] [D ya:D riʤa:lN ]]]] d. CP phase I. Copy [D ya:D riʤa:lN] & Merge <C, I> [C [D ya:D riʤa:lN] [I [V [ImanD ] [V [ʃa:fit V] [D ya:D riʤa:lN]]]] II. Delete [D ya: hu ridZa:l ] [C [D ya:D riʤa:lN] [I [V [ImanD ] [V [Sa:fit hu] [D ya:D riʤa:lN]]]]  48  e. TopP phase I. Merge <Top, C> II. Copy ImanD & Merge <Top, C> [Top ImanD [C [C [D ya:D riʤa:lN] [I [V [ImanD ] [V [Sa:fitV] [D ya:D riʤa:lN]]]]] III. Delete ImanD [Top ImanD [C [C [D ya:D riʤa:lN] [I [V [ImanD ] [V [Sa:fitV] [D ya:D riʤa:lN]]]]] As can be seen from (53), the verb ʃa:fit 'saw' has a DP-complement ya: riʤa:l 'which man' (53bIII) which is copied and merged at SpecCP (53dI-II); the full-DP is deleted from its lower occurrence and maintained in its upper occurrence. Thus, while the gap strategy is derived with full DP-deletion, the resumptive strategy is derived with remnant DP-deletion.  2.7 Implications of the analysis for extraction I present the implications of the syntactic analysis outlined in the two previous sections — namely remnant DP-deletion (§2.5) and full DP-deletion (§2.6) — as they apply to extraction of a subject (§2.7.1), of a prepositional object (§2.7.2), and of a PP (§2.7.3).  2.7.1 Why resumption isn’t possible with subject extraction Recall that subjects in Iraqi Arabic can only be strong pronouns which are independent pronouns in that they do not cliticize (see §2.2.2). But, as already observed above, Iraqi Arabic only has weak resumption, in that the pronominal element which serves as a resumptive is drawn from the (Accusative) clitic series. Therefore, subject resumption is not possible because partial DP-deletion cannot occur as the pronoun cannot cliticize to the verb. In the following I present this analysis in more detail. In (54a) I show the 49  derivation of a content question with subject extraction employing the gap strategy; the numeration of (54a) is given in (54b). (54) SUBJECT EXTRACTION a. extraction with gap  ‫يا صديقة اشترت شقة ؟‬ ya:  Sadi:ga iʃtarat  ʃigga  which friend.F bought.3FS apartment 'Which friend bought an apartment ?' (1 DEC 2010, SA, offered freely)  b. the numeration of (54a) Numeration: {TOPØ, CØ, IØ, iʃtaratV, ya:D, Sadi:gaN, ʃiggaN} (55) shows the derivation, which, as before, develops by phases (i.e. the vP phase, the IP phase, etc), where each maximal projection represents a domain for the application of rules. Within the VP phase (55b), the VP is built by merging the verb V iʃtarat 'she bought’ with the DP ʃigga ‘apartment’ (55bI). Then, the subject is built by merging the D-linked interrogative pronoun ya: ‘which’ with the noun Sadi:ga ‘friend’, as in (55bII). The subject DP ya: Sadi:ga ‘which friend’ then merges with the V at SpecVP (55bIII). During the IP phase (55c), the inflectional head merges with the VP (55cI). Then the subject DP ya: Sadi:ga ‘which friend’ is moved to SpecIP via successive application of Copy and Delete (55cII-III). At the CP phase (55d), the DP ya: Sadi:ga ‘which friend’ is moved to SpecCP via successive application of Copy and Delete (55dI-II).  (55) derivational analysis of (54a) a. Numeration: {TOPØ, CØ, IØ, iʃtaratV, ya:D, Sadi:gaN, ʃiggaN}  50  b. VP phase I. Merge <V, D> [V [iʃtaratV] [D ʃiggaN ] ] II. Merge <D, N> [D [ya:D ] [Sadi:gaN]] III. Merge <D, V> [V [D [ya:D ] [Sadi:gaN]] [V [iʃtaratV] [N ʃiggaN ]]] c. IP phase I. Merge <I, V> [I [V [D [ya:D ] [Sadi:gaN]] [V [iʃtaratV] [N ʃiggaN ]]]] II. Copy [D ya:D Sadi:gaN] & Merge <D, I> [I [D [ya:D] [Sadi:gaN] [I [V [D [ya:D ] [Sadi:gaN]] [V [iʃtaratV] [N ʃiggaN ]]]]] III. Delete [D ya:D Sadi:gaN] [I [D [ya:D] [Sadi:gaN] [I [V [D [ya:D ] [Sadi:gaN]] [V [iʃtaratV] [N ʃiggaN ]]]]] d. CP phase I. Copy [D [ya:D ] [Sadi:gaN]& Merge <C, I> [C [D [ya:D ] [Sadi:gaN] [I [D [ya:D] [Sadi:gaN] [I [V [D [ya:D ] [Sadi:gaN]] [V [iʃtaratV] [N ʃiggaN ]]]]]] II. Delete [D [ya:D ] [Sadi:gaN] [C [D [ya:D ] [Sadi:gaN] [I [D [ya:D] [Sadi:gaN] [I [V [D [ya:D ] [Sadi:gaN]] [V [iʃtaratV] [N ʃiggaN ]]]]]] Let us consider now the ungrammatical example in (56a) showing a content question with subject extraction employing the resumptive strategy; the numeration of (56a) is  51  given in (56b). Notice that the numeration in (56b) has a resumptive pronoun and the structure of the D-linked interrogative expressions is D-φ-N. (56) SUBJECT EXTRACTION a. extraction with resumption  *‫يا صديقة ھي اشترت شقة ؟‬ *ya:  Sadi:ga  hyia iʃtarat  ʃigga  which friend.F she bought.3FS apartment 'Which friend [she] bought an apartment ?' (1 DEC 2010, SA, offered freely)  b. the numeration of (56a) Numeration: {TOPØ, CØ, IØ, iʃtaratV, ya:D, Sadi:gaN, ʃiggaN, hiyaD} (57) shows the derivation of (56a). Within the VP phase (57b), the VP is built by merging the verb V iʃtarat 'she bought’ with the DP ʃigga ‘apartment’ (57bI). The subject DP (57bII) is built by: merging the pronoun hiya ‘she’ with the noun Sa:diga ‘friend’ and (57bIII) merging that complex syntactic object with the D-linked interrogative ya: ‘which’. At the next step (57bIV), the derivation crashes, because the subject DP merges with V at SpecV and the pro-φ hiya ‘she’ cannot cliticize to the verb iʃtarat 'she bought’ within the VP-phase. Notice that this account crucially assumes that phonological clitic attachments must be resolved within the same syntactic phase that introduces the clitic.  (57) derivational analysis of (56a) a. Numeration: {CØ, IØ, iʃtaratV, ya:D, Sadi:gaN, ʃiggaN, hiyaφ}  b. VP phase I. Merge <V, D> [V [iʃtaratV] [D ʃiggaN ] ]  52  II. Merge <φ, N> [φ [hiya φ] [Sa:diga N ] ] III. Merge <D, φ> [D [ya:D ] [hiyaφ SadigaN ] ] IV. Merge <D, V> *[V [D [ya:D] [hiyaφ SadigaN]] [V [iʃtaratV] [D ʃiggaN ]]]  2.7.2 Why resumption is possible with prepositional object extraction Recall that with D-linked interrogative expressions, temporal adjunct extraction in Iraqi Arabic does not allow either the gap strategy or the resumptive strategy (see §2.3.3). Locative adjunct extraction allows only the resumptive strategy. Given that the resumptive strategy is derived by remnant-DP deletion with cliticization of the resumptive pronoun. It follows that resumption with a prepositional object is possible because the resumptive clitic can cliticize. I present this in more detail by looking at the derivation of the locative adjunct extraction with the gap strategy and respectively with the resumptive strategy. Consider the example in (58a) of locative adjunct extraction employing the resumptive strategy and its numeration in (58b). (58) D-LINKED LOCATIVE ADJUNCT a. extraction with resumption  ‫إيمان يا محلة راحت إليھا ؟‬ Iman ya:  maHalla ra:het  ʔli:=ha  Iman which  store  to=3FS  went.3FS  'Which store did Iman go to [it] ?' (9 SEP 2010, SA 3, elicited)  53  b. numeration of (58a) Numeration: {CØ, IØ, ra:HetV, ya:D, maHallaN, Pʔila, haφ, ImanD} In (59) I show the derivational analysis of (58a). As I explained above, the derivation develops by phases (i.e. the vP phase, the IP phase, etc), where each maximal projection represents a domain for the application of rules (Chomsky 1995; Wojdak 2005; Hornstein 2008). The DP is built (59bI) by merging the pronoun ha ‘her’ with the noun maHalla ‘store’ and (59bII) merging that complex syntactic object with the D-linked interrogative ya: ‘which’. Then the preposition P ʔila ‘to’ merges with the DP ya: ha maHalla ‘which her store’ (59bIII) and cliticizaton of the pronoun ha ‘her’ immediately takes place via successive application of Copy and Delete (51bIV-V). Then the verb V ra:Het ‘she went’ merges with the PP ʔila haφ ya:D haφ maHallaN (59VI). The subject DP Iman then merges with the V at SpecVP (59bVII). At the IP phase (59c), the inflectional head merges with the VP (59cI), then the subject DP Iman is moved to SpecIP via successive application of Copy and Delete (51cII–III). At the CP phase (59d), the remnant of the DP ya: ha maHalla is moved to SpecCP via successive application of Copy and Delete (59dI-II). At the TopP phase (59e), the topical head Top merges with the CP (59eI) and the subject DP Iman is moved to SpecTopP via successive application of Copy and Delete (59eII-III).  (59) derivation of (58a) a. Numeration: {CØ, IØ, ra:HetV, ya:D, maHallaN, Pʔila, haφ, ImanD} b. VP phase I. Merge <φ, N> [φ [haφ] [maHalla N ]] II. Merge <D, φ> [D [ya:D ] [haφ maHallaN ]] III. Merge <P, D> [P [Pʔila] [D [ya:D ] [haφ maHallaN ]]  54  IV. Copy haφ & Merge <P, D> [P [Pʔila haφ] [D [ya:D] [haφ maHallaN ]]] V. Delete haφ [P [Pʔila haφ] [D [ya:D] [haφ maHallaN ]]] VI. Merge <V, P> [V [ra:HetV] [P [Pʔila haφ] [D [ya:D] [haφ maHallaN ]]]] VII. Merge ImanD [V [ImanD] [V [ra:HetV] [P [Pʔila haφ] [D [ya:D] [haφ maHallaN ]]]] c. IP phase I. Merge <I, V> [I [V [ImanD] [V [ra:HetV] [P [Pʔila haφ] [D [ya:D] [haφ maHallaN ]]]]] II. Copy ImanD & Merge <D, I> [I [ImanD] [I [V [ImanD] [V [ra:HetV] [P [Pʔila haφ] [D [ya:D] [haφ maHallaN ]]]]] III. Delete ImanD [I [ImanD] [I [V [ImanD] [V [ra:HetV] [P [Pʔila haφ] [D [ya:D] [haφ maHallaN ]]]]] d. CP phase I. Copy [D [ya:D] [haφ maHallaN ] and Merge <C, I> [C [D [ya:D] [haφ maHallaN ] [I [ImanD] [I [V [ImanD] [V [ra:HetV] [P [Pʔila haφ] [D [ya:D] [haφ maHallaN ]]]]]] II. Delete [D [ya:D] [haφ maHallaN ] [C [D [ya:D] [haφ maHallaN ] [I [ImanD] [I [V [ImanD] [V [ra:HetV] [P [Pʔila haφ] [D [ya:D] [haφ maHallaN ]]]]]]  55  e. TopP phase I. Copy ImanD & Merge <Top, C> [Top [ImanD] [C [D [ya:D] [haφ maHallaN ] [I [ImanD] [I [V [ImanD] [V [ra:HetV] [P [Pʔila haφ] [D [ya:D] [haφ maHallaN ]]]]]]] II. Delete ImanD [Top [ImanD] [C [D [ya:D] [haφ maHallaN ] [I [ImanD] [I [V [ImanD] [V [ra:HetV] [P [Pʔila haφ] [D [ya:D] [haφ maHallaN ]]]]]]] The derivation of the content question with D-linked locative adjunct employing the resumptive strategy converges because the resumptive pronoun φ of the D-linked interrogative adjunct can cliticize to the preposition. Let us now consider the ungrammatical content question with D-linked interrogative adjunct employing the gap strategy in (60a) and at its numeration (60b). (60) D-LINKED LOCATIVE ADJUNCT a. extraction with gap  *‫أمك يا محلة راحت إلى ؟‬ *ʔummiʧ  ya: maHalla ra:hit  mother=your:F which store went.3FS  ʔila___ to ____  'Where did your mother go ____ ?' (9 SEP 2010, SA 3, offered freely) b. Numeration of (60a) Numeration: {CØ, IØ, ra:HetV, ya:D, maHallaN, Pʔila, ImanD} (61) shows the derivation of (60a). The DP is built by: (61bI) merging the Dlinked interrogative ya: ‘which’ with the noun maHalla ‘store’. Then the preposition P ʔila ‘to’ merges with the DP ya: maHalla ‘which store’ (61bII). Then the verb V ra:Het ‘she went’ merges with the PP ʔila ya:maHallaN (61bIII). The subject DP Iman then merges with the V at SpecVP (61bIV). At the IP phase (61c), the inflectional head merges with the VP (61cI), then the subject DP Iman is moved to SpecIP via successive 56  application of Copy and Delete (61cII – III). At the CP phase (61d), the DP ya: maHalla is moved to SpecCP via successive application of Copy and Delete (61dI-II). At this point the derivation crashes, because Arabic does not allow preposition stranding.  (61) derivation of (60a) a. Numeration: {CØ, IØ, ra:HetV, ya:D, maHallaN, Pʔila, ImanD}  b. VP phase I. Merge <D, N> [D [ya:D] [maHallaN ]] II. Merge <P, D> [P [Pʔila] [D [ya:D] [maHallaN ]] III. Merge <V, P> [V [ra:HetV] [P [Pʔila] [D [ya:D] [maHallaN ]]] IV. Merge ImanD [V [ImanD] [V [ra:HetV] [P [Pʔila] [D [ya:D] [maHallaN ]]]] c. IP phase I. Merge <I, V> [I [V [ImanD] [V [ra:HetV] [P [Pʔila] [D [ya:D] [maHallaN ]]]]] II. Copy ImanD & Merge <D, I> [I [ImanD] [I [V [ImanD] [V [ra:HetV] [P [Pʔila] [D [ya:D] [maHallaN ]]]]] III. Delete ImanD [I [ImanD] [I [V [ImanD] [V [ra:HetV] [P [Pʔila] [D [ya:D] [maHallaN ]]]]]  57  d. CP phase I. Copy [D [ya:D maHallaN ]] and Merge <C, I> [C [D [ya:D] [maHallaN ] [I [ImanD] [I [V [ImanD] [V [ra:HetV] [P [Pʔila] [D [ya:D] [maHallaN ]]]]]] *II. Delete [D [ya:D] [haφ maHallaN ] *[C [D [ya:D] [maHallaN ] [I [ImanD] [I [V [ImanD] [V [ra:HetV] [P [Pʔila] [D [ya:D] [maHallaN ]]]]]] CRASH NO PREPOSITION STRANDING  2.7.3 Why resumption is obligatory with extraction from islands Recall that in wh-islands (i.e. embedded CPs introduced by wh-constituents) and adjunct islands (i.e. islands formed from an adjunct clause), resumption is obligatory (see §2.4.3). Consider the grammatical example in (62a) of a wh-island employing the resumptive strategy and its numeration in (62b).  (62) WH-ISLAND a. resumptive strategy  ‫يا صورة سامر يسأل إذا كل مرية شققتھا ؟‬ ya:  Su:ra.F  Samer ysʔal  which picture.F Samer ask.3MS  iδa kull  mraya  ʃagagat=ha  if every woman tore.3FS=3F  'Which photo did Samer wonder if every woman tore [it] ?' (13 NOV 2010, SA 14aII, elicited) b. Numeration of (62a): { CØ, IØ, ʃagagatV, ysʔalV, ya:D, Su:raN, kull mrayaD, haφ} (63) shows the derivation of (62a). During the VP phase, the DP is built (63bI) by merging the pronoun ha ‘her’ with the noun Su:ra ‘picture’ and (63bII) by merging that complex syntactic object with the D-linked interrogative ya: ‘which’. Then the verb V ʃagagatV merges with the complex syntactic object ya: ha Su:ra (63bIII) and cliticizaton of the pronoun ha ‘her’ immediately takes place via successive application of Copy and  58  Delete (63bIV-V). The subject of the embedded clause kull mrayaD ‘every woman’ is merged at SpecV (63VI). During the IP phase, the inflectional head merges with the VP (63cI), then the subject DP kull mrayaD ‘every woman’ is moved to SpecIP via successive application of Copy and Delete (63cII – III). At the CP phase (63d), the DP ya: ha Su:ra is moved to SpecCP via successive application of Copy and Delete (63dIIII). At the VP phase, the verb ysʔalV merges with the embedded clause (63eI), then the subject SamerD is merged as SpecVP (63eII). During the IP phase, the inflectional head merges with the VP (63fI), then the subject DP SamerD is moved to SpecIP via successive application of Copy and Delete (63fII – III). At the CP phase (63g), the DP ya: ha Su:ra is moved to SpecCP via successive application of Copy and Delete (63gIIII).  (63) derivation of (62a) a. Numeration: { CØ, iδaC , IØ, ʃagagatV, ysʔalV, ya:D, Su:raN, kull mrayaD, haφ} b. VP phase I. Merge <φ, N> [φ [haφ] [Su:ra N ]] II. Merge <D, φ> [D [ya:D ] [haφ Su:raN ]] III. Merge <V, D> [V [ʃagagat V] [[D [ya:D] [haφ Su:raN ]]] IV. Copy haφ & Merge <V, φ> [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]] V. Delete haφ [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]  59  VI. Merge <D, V> [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]] c. IP phase I. Merge <I, V> [I [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]] II. Copy kull mrayaD & Merge <D, I> [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]] III. Delete kull mrayaD [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]] d. CP phase I. Merge <C, I> [C [iδaC] [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]] II. Copy [D [ya:D ] [haφ Su:raN ]] & Merge <D, C> [C [D [ya:D ] [haφ Su:raN ]] [C [iδaC] [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]]] III. Delete [D [ya:D ] [haφ Su:raN ]] [C [D [ya:D ] [haφ Su:raN ]] [C [iδaC] [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]]] e. VP phase I. Merge <V, C> [V [ysʔalV] [C [D [ya:D ] [haφ Su:raN ]] [C [iδaC] [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]]]  60  II. Merge SamerD [V [SamerD] [V [ysʔalV] [C [D [ya:D ] [haφ Su:raN ]] [C [iδaC] [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]]] f. IP phase I. Merge <I, V> [I [V [SamerD] [V [ysʔalV] [C [D [ya:D ] [haφ Su:raN ]] [C [iδaC] [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]]]] II. Copy SamerD and merge <D, I> [I [SamerD] [I [V [SamerD] [V [ysʔalV] [C [D [ya:D ] [haφ Su:raN ]] [C [iδaC] [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]]]] III. Delete SamerD [I [SamerD] [I [V [SamerD] [V [ysʔalV] [C [D [ya:D ] [haφ Su:raN ]] [C [iδaC] [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]]]] g. CP phase I. Copy [D [ya:D ] [haφ Su:raN ]] & Merge <C, I> [C [D [ya:D ] [haφ Su:raN ]] [I [SamerD] [I [V [SamerD] [V [ysʔalV] [C [D [ya:D ] [haφ Su:raN ]] [C [iδaC] [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]]]]] II. Delete [D [ya:D ] [haφ Su:raN ]] [C [D [ya:D ] [haφ Su:raN ]] [I [SamerD] [I [V [SamerD] [V [ysʔalV] [C [D [ya:D ] [haφ Su:raN ]] [C [iδaC] [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV haφ] [D [ya:D ] [haφ Su:raN ]]]]]]]] The derivation of the content question with D-linked locative adjunct employing the resumptive strategy converges because the resumptive pronoun φ of the D-linked interrogative adjunct can cliticize to the verb.  61  Consider the ungrammatical example (64a) and its numeration (64b). Notice that the numeration (64b) does not have a resumptive pronoun and the structure of the Dlinked interrogative is D-N. (64) WH-ISLAND a. gap strategy  *‫يا صورة سامر يسأل إذا كل مرية شققت ؟‬ *ya:  Su:ra.F  Samer ysʔal  which picture.F Samer ask.3MS  iδa kull  mraya  ʃagagat____  if every woman tore.3FS___  '*Which photo did Samer wonder if every woman tore ____ ?' (13 NOV 2010, SA 14aI, elicited) b. Numeration: { CØ, iδaC , IØ, ʃagagatV, ya:D, Su:raN, kull mrayaD} (65) shows the derivation of (64a). During the VP phase, the DP is built (65bI) by merging the D-linked interrogative ya: ‘which’ with the noun Su:ra ‘picture’. Then the verb V ʃagagatV merges with the direct object ya: Su:ra (65bII). The subject of the embedded clause kull mrayaD ‘every woman’ is merged at SpecV (65bIII). During the IP phase, the inflectional head merges with the VP (65cI), then the subject DP kull mrayaD ‘every woman’ is moved to SpecIP via successive application of Copy and Delete (65cII – III). At the CP phase (65d), the complementizer iδaC ‘whether’ is merged at C (65dI) and then the DP ya: Su:ra is moved to SpecCP via successive application of Copy and Delete (65dII-III). At this point the derivation crashes, because there is no clitic to remain cliticized to the verb.  (65) derivation of (64a) a. Numeration: { CØ, IØ, ʃagagatV, ya:D, Su:raN, kull mrayaD} b. VP phase I. Merge <D, N> [D [ya:D ] [Su:raN ]]  62  II. Merge <V, D> [V [ʃagagat V] [[D [ya:D] [Su:raN ]]] III. Merge <D, V> [V [kull mrayaD ] [V [ʃagagatV ] [D [ya:D ] [Su:raN ]]]] c. IP phase I. Merge <I, V> [I [V [kull mrayaD ] [V [ʃagagatV ] [D [ya:D ] [Su:raN ]]]] II. Copy kull mrayaD & Merge <D, I> [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV ] [D [ya:D ] [haφ Su:raN ]]]]] III. Delete kull mrayaD [I [kull mrayaD] [V [kull mrayaD ] [V [ʃagagatV ] [D [ya:D ] [haφ Su:raN ]]]]] - CRASH NO CLITIC TO CLITICIZE TO VERB  2.8 Implications of the analysis for reconstruction Resumptive strategies give rise to reconstruction effects (Rouveret 2011, to appear). Let's first define what reconstruction is (66). Given a relation between an XP and its extraction site (66i), "reconstruction" is a term used to describe the interpretation of XP in its in-situ position relative to a c-commanding antecedent YP, as in (66ii), and the interpretation of XP in its in-situ position relative to a c-commanding quantifier YP, as in (66iii) below.  63  (66) DEFINITION OF RECONSTRUCTION (i) [XP]i … [___]i (ii) [XP]i ... [YPantecedent]... [___]i (iii) [XP]i [QP] [___]i The authors who work on the syntax of resumption in either Semitic or Celtic use reconstruction effects to account for the bound variable readings of the resumptive elements (Lebeaux 1990; Aoun et al 2001; Rouveret 2002, 2008; Guilliot 2006; Guilliot & Malkawi 2006; Malkawi 2009). There are two kinds of reconstruction: scope reconstruction and binding reconstruction. After illustrating that the copy theory of movement derives these two types of reconstruction (§2.8.1 - §2.8.2), I show how the two kinds of reconstruction play out in Iraqi Arabic (§2.8.3 - §2.8.4).  2.8.1 Scope reconstruction with a quantificational antecedent Scope reconstruction is illustrated in (67) below with an example adapted from Guilliot & Malkawi (2009). SCOPE RECONSTRUCTION IN ENGLISH (Guilliot & Malkawi 2009)  (67) Which patient did every doctor examine ? (i) 'Every doctor examined a different patient' Ax, Ey [examine (x. y)] (ii) 'There is one patient that every doctor examined' Ey, Ax [examine (x, y)] One of the possible readings of the question in (67) is the one where every doctor  64  examined a different patient: the universal quantifier every has scope over patient. This is possible if patient “reconstructs” to its thematic position as the complement of the verb examine, where it would be in the scope of the quantifier. Reconstruction is automatically available in the copy theory of movement that is assumed in this thesis, as shown in (68).  (68) reconstruction applied to example in (67) a. Which patient did every doctor examine which patient ? b. [CP [C which patient1] [TP [T did] [VP [DP every doctor] [V examine][DP which patient1]]]] In particular, scope reconstruction automatically follows because the constituent which patient can be interpreted in its thematic position as complement of the verb examine, which is under the scope of the quantifier every. Observe that when scope reconstruction applies, the interpretation assigned to the D-linked wh-phrase which patient is that of an indefinite expression: ‘for every doctor there is a patient that the doctor examines’. How is this possible? Here I follow other scholars (Kayne 1994, Agüerro-Bautista 2001, Guilliot & Malkawi 2009) in assuming that the lower copy in (68) can be interpreted as an indefinite. For concreteness, I adopt the mechanics proposed by Guilliot & Malkawi (2009), for whom the indefinite construal is an instance of a skolemized choice function, defined as follows (69):  (69)  a.  skolemization:  the  method  for  removing  existential  quantifiers  from a logical form (Kratzer 1999; Agüerro-Bautista 2001) b. choice function: a function which chooses exactly one element from each set in the domain (Kratzer 1999; Agüerro-Bautista 2001) c. skolemized choice function: a function which takes one individual x and a set of entities P and returns one individual of that set relative to x (Guilliot & Malkawi 2009) As a skolemized choice function, the copy which patient in (68) is bound by the quantifier every doctor, yielding the distributive reading where each doctor is mapped to a different patient. Gulliot & Malkawi (2009) further point out that, according to  65  Agüerro-Bautista (2001), the interpretation of the copy as a skolemized choice function gives rise to the pair-list reading, as follows (69):  (69) deriving the pair-list reading for (68) 1. Which patient did every doctor examine? 2. LF:  WHICH(y), patient(y), ∀(x) doctor (x), [examine (x, y) ]  3. Which patient = indefinite ∃(y), patient(y), ∀(x) doctor (x), [examine (x, y) ] 4. Skolemized choice function ∀(x) doctor (x), [examine (x, f(y), patient(y) ] 5. Pair-list answer: Dr. Johnson examined Victor, Dr. Smith examined Joyce, etc We saw that (69) is an example of scope reconstruction.  2.8.2 Binding reconstruction with a referential antecedent Let us now consider the sentence in (70) which illustrates a case of binding reconstruction. ENGLISH (Guilliot 2006)  (70) Which picture of himself1 does every man1 prefer ? = Ax [x prefer picture of x] (70) is an example of binding reconstruction: picture of himself is reconstructed to its thematic position as the complement of the verb prefer. This allows the pronoun himself to be bound by the quantifier each, resulting in a bound variable reading, which maps a different picture to each man.  (71) reconstruction applied to example (70) a. Which picture of himself1 does every man1 prefer picture of himself1?. 66  The representation in (71) is an example of binding reconstruction, because picture of himself can be reconstructed in its thematic position as the complement of the verb prefers, such that the pronoun him is bound by each man; the result will be a bound variable reading which maps a different picture for each man. This is illustrated in the abstract representation in (72) below.  (72) representation of (71) - binding reconstruction [CP [DP which picture of himself1][TP does [DP each man] [V prefer] [DP picture of himself1]]]]]  2.8.3 Reconstruction with resumption So far we have seen how reconstruction effects appear with the gap strategy. In the following I show Guilliot & Malkawi's (2006) account of reconstruction effects with resumptive strategies in Jordanian Arabic. Both (67) and (70) above are structures with gaps. In reconstruction with gap strategies, we reconstruct a copy of the displaced constituent in its thematic position, as we saw above. Let us now consider the example in (73) which shows a resumptive strategy.  JORDANIAN ARABIC  (Guilliot & Malkawi 2006)  (73) resumptive strategy [Talib-[ha]1  l-kassul]2 ma ziʕlat  student.M=3FS the-bad kaHʃat-uh2  mn  expelled=3MS from  Neg upset.3FS  wala mʕalmih1 laʔannuh l-mudiirah no  teacher  because the-principal  l-madrase the school  'Her1 bad student, no teacher1 was upset because the principal expelled [him] from school'.  67  In the construction with resumptive pronoun in (73), a resumptive pronoun occupies the thematic position which is otherwise occupied by a gap in the gap strategies; apparently, we could not reconstruct a copy of the displaced constituent Talib-ha l-kassul "her bad student" in its thematic position, since that position is occupied by a pronoun. The solution that Guilliot & Malkawi (2006) propose is to consider the pronoun to be a definite description in the spirit of Elbourne (2002). Then the reconstruction the displaced constituent is possible as the pronoun's NP-argument whose copy is elided under identity with its antecedent (Guilliot & Malkawi 2006). The abstract representation of the reconstruction with resumptive strategy in (73) is given in (74) below:  (74) reconstruction of (73) - resumption [DP Talib=[ha]1 l-kassul]2 .... [wala mʕalmih]1 .... [DP -uh [NP Talib=ha1 l-kassul]]2 Therefore, they amend Lebeaux's (1990) formulation of reconstruction, given in (75a), to (75b). (75) RECONSTRUCTION a.  Lebeaux  if  1990:  an  XP  allows  for  reconstruction,  then that XP has undergone movement b.  Guilliot & Malkawi 2006: if an XP allows for reconstruction, then  a  copy  of  that  XP  is  present  in  the  derivation.  For Lebeaux, reconstruction is a diagnostic for movement. For Guilliot & Malkawi (2006) reconstruction is a diagnostic for the presence of a copy. These two approaches make different predictions about the interaction of reconstruction with the resumptive strategy. For analyses of resumption that treat it as a binding relation, the Lebeaux movement-account of reconstruction (incorrectly) predicts that reconstruction won’t be possible with resumptive. But for analyses such as the copy theory of movement, where both the gap strategy and the strategy involve the copy-and-delete operation, reconstruction is (correctly) predicted to apply in both contexts.  68  2.8.4 Scope Reconstruction of Iraqi D-linked interrogatives In Iraqi Arabic reconstruction can be seen in contexts where the resumptive strategy is optional and in contexts where the resumptive strategy is obligatory. Consider the optional resumption examples in (76), where (76a) employs the gap strategy and (76b) employs the resumptive strategy.  (76) optional resumption a. the gap strategy  ‫يا صورة البنھا كل مرية شققت ؟‬ ya:  Su:ra.F  li-bni=ha  kull  mraya  which picture.F of-son=her every woman  ʃagagat____ tore.3FS  'Which photo of her son did every woman tear___ ?' (13 NOV 2010, SA 10a, elicited)  b. the resumptive strategy  ‫يا صورة البنھا كل مرية شققتھا ؟‬ ya:  Su:ra.F  li-bni=ha  kull  mraya  ʃagagat=ha  which picture.F of-son=her every woman tore.3FS=3FS 'Which photo of her son did every woman tear [it] ?' (13 NOV 2010, SA 10b, elicited) The possessive pronoun -ha 'her' in the interrogated expression ya: Su:ra libni=ha 'which photo of her son' can be interpreted as a variable bound by the quantified expression kull mraya 'every woman'. The availability of this interpretation suggests that the possessive pronoun -ha 'her' can be reconstructed in the scope of the quantified expression kull mraya 'every woman'. The representation of the reconstruction of (76a) is given below in (77a) and its representation in (77b).  69  (77) a. reconstruction applied to example in (76a) ya:  Su:ra.F  li-bni=ha  kull  mraya  ʃagagat  Su:ra li-bni=ha  which picture.F of-son=her every woman tore.3FS 'Which photo of her son did every woman tear picture of her son ?'  b. representation of (76a) [DP ya: Sura1 li-bni=[ha]2] .... [kull mraya]2..... [VP [V ʃagagat]... [NP Sura1 li-bni=[ha]2]] When we reconstruct in (76a), Su:ra li-bni=ha 'photo of her son' we consider a copy of it in its thematic position as complement of the verb ʃagagat 'she tore', where it is c-commanded (i.e. in the scope of) the universal quantifier kull 'every' in kull mraya 'every woman'. The reconstruction allows for the bound variable reading in which every woman tore a different picture of her son. Consider now in (78) below how reconstruction applies to a D-linked content question with the resumptive strategy in (76b). We can reconstruct Su:ra li-bni=ha 'photo of her son' in its thematic position as the NP-complement of the resumptive pronoun -ha 'it'.  (78) reconstruction applied to example in (76b) ya:  Su:ra1  li-bni=ha2  kull  mraya2 ʃagagat=[DP ya: [φ ha1 [NP Su:ra1  li-bni=ha2]]  which picture of-son=3FS every woman tore=[DP which [φ it1 [NP picture1 of-son=3FS2]] 'Which picture of her son did every woman tear [DP which [φ it1 [NP picture1 of-her2 son]?' Thus, in the copy theory of movement, reconstruction is correctly predicted to occur with both the gap strategy and the resumptive strategy. We conclude that reconstruction is possible in Iraqi Arabic D-linked content questions with a resumptive pronoun in cases where the resumptive strategy is optional. Let us see how reconstruction is applied to the island contexts in (79) and (80).  70  (79) WH-ISLAND  ‫يا صورة البنھا سامر يسأل إذا كل مرية شققتھا ؟‬ ya:  Su:ra.F  li-bni=ha  Samer ysʔal  which picture.F of-son=her Samer ask.3MS  iδa kull  ʃagagat=ha  mraya  if every woman tore.3FS=3F  'Which photo of her son did Samer wonder if every woman tore [it] ?' (13 NOV 2010, SA 14aII, elicited) (80) ADJUNCT ISLAND  ‫يا صورة البنھا سامر زعل ألنو كل مرية شققتھا ؟‬ ya:  Su:ra.F  li-bni=ha  Samer zaʕal  li-ennu:  which picture.F of-son=her Samer get angry because  kull  mraya ʃagagat=ha  every woman tore.3FS=3FS  'Which photo oh her son did Samer get angry because every woman tore [it] ?' (13 NOV 2010, SA 14bII, elicited) Let us first apply reconstruction to the wh-island example in (79). The representation of the reconstruction applied to (79) is given below in (81).  (81) reconstruction applied to the resumptive strategy in (79) ya:  Su:ra1  li-bni=ha2  Samer ysʔal  which picture.F of-son=3FS Samer ask.3MS  iδa kull mraya2 if  every woman  ʃagagat=[DP ya: [φ ha1 [NP Su:ra1 li-bni=ha2]] tore.3FS=[DP which it1 [NP picture1 of-son=3FS2]] 'Which photo of her son did Samer wonder if every woman tore [DP it1 [NP picture1 of-her2 son]]?' Consider now the representation in (82) where I show how reconstruction is applied to the resumptive strategy in the adjunct island example in (80).  71  (82) reconstruction applied to the resumptive strategy in (80) ya:  Su:ra1  li-bni=ha2  Samer zaʕal  li-ennu:  which picture.F of-son=her Samer get angry because ʃagagat=[DP ya: [φ ha1 [NP Su:ra1  kull  mraya2  every woman  li-bni=ha2]]  tore.3FS=[DP which [φ it1 [NP picture1 of-son=3FS2]] Which photo oh her son did Samer get angry because every woman tore [DP which [φ it1 [NP picture1 of-her2 son]]? The availability of the bound variable readings with the resumptive strategies in the wh-island example in (79) and in the adjunct island in (80) suggest that we could reconstruct the interrogated constituent in its thematic position in the scope of the quantifier. I have shown that scope reconstruction is possible with the resumptive strategy employing a resumptive pronoun in both optional resumption contexts and obligatory resumption5 contexts. These findings are given in Table 12 below.  Optional resumption  Obligatory resumption (island context)  Reconstruction quantificational  with √  √  antecedent Table 12. Scope reconstruction with resumptive pronoun  2.8.5 Binding Reconstruction of Iraqi D-linked interrogatives Before I proceed with the binding reconstruction examples, I show that Condition A holds for Iraqi Arabic. Condition A of binding theory stipulates that a reflexive 5  Obligatory resumption is tied to extraction from islands.  72  pronoun must be locally bound. Consider the example in (83) below which shows a grammatical sentence where condition A holds as in (83a), an ungrammatical example where the reflexive pronoun has no antecedent (83b), and a grammatical example where the reflexive pronoun is locally bound (83c) an ungrammatical example where the reflexive pronoun is not locally bound (83d).  (83) Condition A in Iraqi Arabic a. .‫فارس يحب نفسه‬ Faris yHebb  nafs=hu  Faris love.3MS soul=his 'Faris loves himself'. (13 NOV 2010, SA 22e, offered freely) b. *.‫نفسه يحب أمه‬ *nafs=hu yHebb  ʔumm=hu  soul=his love.3MS mother=his *Himself loves his mother. (13 NOV 2010, SA 22a, elicited) c. .‫أم فارس تحب نفسھا‬ ʔumm  Faris tHebb  nafs=ha  mother Faris love.3FS soul=her Faris' mother loves herself. (13 NOV 2010, SA 22d, offered freely) d.* .‫أم سامر تحب نفسه‬ *ʔumm  Samer tHebb  nafs=hu  mother Samer love.3FS soul=his *Samer's mother loves himself. (13 NOV 2010, SA 22c, elicited)  73  Now consider binding reconstruction with a referential antecedent. Let us first consider the optional resumption context illustrated in (84).  (84) condition A with referential antecedent - optional resumption context a. the gap strategy  ‫يا صورة لنفسه سامر شقق ؟‬ ya:  Su:ra.F  li-nafs=hu  Samer ʃaggag____  which picture.F of-soul=3MS Samer tore.3MS___ 'Which photo of himself did Samer tear ___ ?' (13 NOV 2010, SA 20aI, elicited)  b. the resumptive strategy  ‫يا صورة لنفسه سامر شققھا ؟‬ ya:  Su:ra.F  li-nafs=hu  Samer ʃaggag=ha  which picture.F of-soul=3MS Samer tore.3MS=3FS 'Which photo of himself did Samer tear [it] ?' (13 NOV 2010, SA 20aII, elicited) Considering that Condition A holds for Iraqi Arabic and therefore reflexive pronouns must be locally bound, it follows that the grammaticality of (84a) and (84b) indicates that the reflexive pronoun nafshu "himself" in (84) is bound by its antecedent Samer. This is possible if the interrogative constituent Su:ra li-nafs=hu "picture of himself" is reconstructed. Let us look at how reconstruction works with the gap strategy in (84a). The representation of reconstruction is given below in (85).  (85) a. reconstruction applied to example in (84a) ya:  Su:ra1  li-nafs=hu2  Samer2  ʃaggag  Su:ra1  li-nafs=hu2  which picture.F of-soul=3MS Samer tore.3MS Su:ra1  li-nafs=hu2  'Which photo of himself did Samer tear photo of himself ?'  74  b. abstract representation of (84a) [DP ya: Sura1 li-nafs=[hu]2] .... Samer2..... [VP [V ʃagagat] [NP Sura1 li-nafs=[hu]2]] As seen in (85), Su:ra li-nafs=hu 'photo of himself' reconstructs to its thematic position as complement of the verb ʃagagat 'he tore', where the reflexive pronoun nafshu "himself" is c-commanded by its antecedent Samer. Reconstruction thus allows for the bound variable reading in which Samer tore a different picture of himself (i.e. the photo of his wedding, the photo of himself at the graduation party, etc). In (86) below, reconstruction applies to (84b) where the resumptive strategy is employed. If we reconstruct Su:ra li-nafs=hu 'photo of himself' in its thematic position, we notice that the position in which there was a gap in (84a) is now occupied by the resumptive pronoun -ha 'it' in (84b). We now reconstruct Su:ra li-nafs=hu 'photo of himself' in its base position as the NP-complement of the resumptive pronoun -ha 'it'.  (86) reconstruction applied to example in (84b) ya:  Su:ra1  li-nafs=hu2  Samer2 ʃaggag=[DP ya: [φ ha1 [NP Su:ra1  li-nafs=hu2]]  which picture.F of-soul=3MS Samer tore.3MS=[DP which [φ it1 [NP picture1 of-himself2]] 'Which photo of himself did Samer tear [DP which [φ it1 [NP picture1 of-himself2]]?' Let us now move on to see how binding reconstruction with a referential antecedent works in contexts of obligatory resumption. Consider the example in (87) showing a context of obligatory resumption. The example in (87a) where the gap strategy is employed is ungrammatical, while the example in (87b) where the gap strategy is employed is grammatical.  (87) condition A with referential antecedent - obligatory resumption context a. the gap strategy  *‫يا صورة لنفسه إللي سامر شقق ؟‬ *ya:  Su:ra.F  li-nafs=hu  illyi  Samer ʃaggag_____  which picture.F of-soul=his which Samer tore.3MS____ '*Which photo of himself was the one that Samer tore ___ ?' (13 NOV 2010, SA 20bI, elicited)  75  b. the resumptive strategy  ‫يا صورة لنفسه إللي سامر شققھا ؟‬ ya:  Su:ra.F  li-nafs=hu illyi  Samer ʃaggag=ha  which picture.F of-soul=his which Samer tore.3MS=3FS 'Which photo of himself was the one that Samer tore [it] ?' (13 NOV 2010, SA 20bII, elicited) Let us see how reconstruction can be applied now to the example in (87b) where the resumptive strategy is employed rendering a grammatical sentence. The representation of reconstruction with (87b) is given in (88).  (88) a. reconstruction applied to the example in (87b) ya:  Su:ra1  li-nafs=hu2 illyi  Samer2  which picture.F of-soul=his which Samer  ʃaggag=[DP ya: [φ ha1 [NP Su:ra1 li-nafs=hu2]] tore.3MS=[DP which [φ it1 [NP picture1 of-  himself2]]  'Which photo of himself was the one that Samer tore [DP which [φ it1 [NP picture1 ofhimself2]]?' Thus, the binding reconstruction is possible with the resumptive pronoun in contexts of optional resumption as well as in contexts of obligatory resumption. The generalization is that the binding reconstruction is always possible with the resumptive pronoun. Table 13 below summarizes the findings regarding binding reconstruction.  Optional resumption  Obligatory resumption (island context)  Binding reconstruction  √  √  Condition A Table 13. Binding reconstruction with resumptive pronoun Iraqi Arabic D-linked content questions reveal that the resumptive pronoun allows 76  for binding reconstruction and scope reconstruction in optional contexts as well as obligatory contexts. The reconstructed bound variable reading will allow for different interpretations in the semantics, as it will be detailed in chapter 4.  77  3 The semantics of D-linked content questions in Iraqi Arabic This chapter focuses on the fact that D-linked content questions can be interpreted in one of three ways. I begin by introducing the problem that these interpretations pose for our understanding of how the syntax of D-linked content questions is related to their semantics (§3.1). After discussing the semantic correlates of the resumption and the gap strategy (§3.2), I turn to the question of what the two strategies reveal about the syntaxsemantics interface (§3.3).  3.1 Overview of the syntax-semantics interface problem and a possible solution The contrast between the gap and the resumptive strategy has interpretive consequences. I introduce the problem that this raises for the syntax/semantics interface (§3.1.1) and sketch two possible solutions (§3.1.2).  3.1.1 The syntax-semantics interface problem As discussed by Sharvit (1999), a D-linked content question can be answered in one of three ways. As shown in (89), the answer can consist of an individual-denoting expression (89a), a natural function (89b), or a pair-list (89c). I adopt the convention of referring to these three interpretations as the individual reading, the natural function reading and the pair-list reading.  78  (89) SEMANTIC AMBIGUITY OF D-LINKED CONTENT QUESTION Q:  Which woman did every many invite?  A:  a.  Mary.  INDIVIDUAL  b.  His mother.  NATURAL FUNCTION  c.  John invited Mary; Bill invited Sally,…  PAIR-LIST  Recall that, on independent grounds, in the previous chapter, I argued that Iraqi Arabic D-linked content questions are associated with two distinct structures: [D-N] and [D-ϕ-N]. These findings are given in Table 14.  INTERNAL STRUCTURE OF EXTRACTION SITE D-N  D-ϕ-N  GAP STRATEGY  √  √ (with covert ϕ)  RESUMPTIVE STRATEGY  x  √ (with overt ϕ)  Table 14. Structural Differentiation of Extraction Sites with Iraqi Arabic D-linked content questions For D-linked content questions, a question that arises regarding the three interpretations—individual, natural function, and pair-list reading—and the two syntactic structures—[D-N] and [D-ϕ-N]— is whether and how they are connected to each other. I call this the syntax-semantics interface problem: (90) SYNTAX-SEMANTICS INTERFACE PROBLEM How does the interpretation of D-linked content questions relate to their syntactic structure?  79  3.1.2 A possible solution to the syntax-semantics interface problem A relevant observation is that, in Iraqi Arabic, while the gap strategy is threeways ambiguous, the interpretive pattern of the resumptive strategy is more complex. This is summarized in Table 15. In some contexts — I call this context 1 — the resumptive strategy permits only the individual and the natural function reading. But in other contexts — I call this context 2 — the resumptive strategy permits all three readings.  GAP  RESUMPTIVE STRATEGY  STRATEGY  CONTEXT 1  CONTEXT 2  INDIVIDUAL  √  √  √  NATURAL FUNCTION  √  √  √  PAIR-LIST  √  x  √  INTERPRETATION  Table 15. Interpretations associated with Iraqi Arabic D-linked content questions The two contexts for the resumptive strategy are as follows: (91) RESUMPTIVE CONTEXTS FOR IRAQI ARABIC D-LINKED CONTENT QUESTIONS a.  CONTEXT  1: both the gap and the resumptive strategy are possible; i.e. the  resumptive strategy is optional b.  CONTEXT  2: only the resumptive strategy is possible; i.e. the resumptive strategy  is obligatory One way of approaching the syntax-semantic interface problem, as it presents itself with D-linked content questions, is to ask which syntactic structures correlate with which semantic interpretations. According to the syntactic analysis developed in Chapter 2, all D-linked content questions have a DP structure, but the internal structure of the DP can differ: it can be [D-N] or [D-ϕ-N]. Table 16 summarizes how the three interpretations map onto the syntax.  80  GAP SYNTAX  INTERPRETATION  RESUMPTIVE STRATEGY  STRATEGY  CONTEXT 1  CONTEXT 2  [D…]  INDIVIDUAL  √  √  √  [D-ϕ-N]  NATURAL FUNCTION  √  √  √  [D-N]  PAIR-LIST  √  x  √  Table 16. Syntax-semantics mapping for Iraqi Arabic D-linked content questions The individual reading is sensitive to the external DP-syntax only and so is available in all contexts, with either the gap or the resumptive strategy. The natural-function reading requires a [D-ϕ-N] structure (with ϕ covert or overt), and so is available with the gap strategy (covert ϕ) or with the resumptive strategy (overt ϕ). The pair-list reading shows a blocking effect. In contexts where the resumptive strategy is optional, the pair-list reading is blocked; this is Context 1. In contexts where the resumptive strategy is obligatory, the pair-list reading is available; this is Context 2. I take Context 1 to indicate that the pair-list reading is available only with a [D-N] structure, which is automatically derived by the gap strategy. However, in contexts where the gap strategy isn’t possible (Context 2), then the resumptive strategy can be used for the pair-list reading. I speculate that, in context 2, some additional mechanism makes the resumptive strategy behave as if it were a [D-N] structure. In other words, in contexts where resumption is obligatory, the resumptive strategy is semantically ambiguous. I speculate that this reflects whether or not the ϕ-element is interpreted at LF: if the ϕelement is interpreted this gives a natural function reading; if the ϕ-element is not interpreted, this gives a pair-list reading. From the point of view of the syntactic derivation, the resumptive strategy is potentially structurally (and therefore, semantically) ambiguous when the resumptive pronoun cliticizes, but before the DP is copied, (92a). Normally, the copy operation copies the entire labeled [D-ϕ-N] structure, (92b-i). But nothing prevents the copy operation from copying only the phonological string, which is a [D-N] structure, (92b-ii). Notice that this is only possible if there is early lexical insertion of the PF-formatives. I call this the copy solution to the syntax-semantics interface problem. I assume that 81  copying PF-formatives is a last resort option, and is only allowed in contexts where the resumptive strategy is obligatory. (92) COPY SOLUTION TO THE SYNTAX-SEMANTICS INTERFACE PROBLEM ϕ:  [ϕ ϕ]  [D ϕ N]  a.  CLITICIZATION OF  b-i.  COPY SYNTACTIC FORMATIVES:  [D ϕ N]  [ϕ ϕ]  [D ϕ N]  b-ii.  COPY PF-FORMATIVES:  [D N]  [ϕ ϕ]  [D ϕ N]  3.2 Semantic correlates of the gap and the resumptive strategy The semantics of questions is generally assumed to be revealed by the appropriate answers that they elicit (Hamblin 1973; Kartunnen 1977; Pesetsky 1987; Hornstein 2006). We have seen that a D-linked content question can be answered in one of three ways: the answer can be a single-individual, a natural function, or a paired-list. Here I discuss how these three interpretations are related to earlier observations about the semantic correlates of the gap and the resumptive strategy. I first introduce the distinction made between de re and de dicto readings in the context of relative clauses (§3.2.1). I then show how the de re/de dicto contrast is also found with D-linked content questions: here the relevant distinction is between a single-individual answer and a multipleindividual answer (§3.2.2). I then discuss how the de re/de dicto and the singleindividual/multiple-individual contrasts are related to the three interpretations associated with D-linked content questions (§3.2.3).  3.2.1 The distinction between de dicto and de re readings De dicto (from Latin "of the word") and de re (from Latin "of the thing") are readings used to mark distinctions in the possible interpretation of statements. Consider the example in (93) which can have both a de dicto and a de re reading.  82  (93) Peter believes that someone is out to get him. a. de dicto reading: someone is unspecific, Peter suffers a general paranoia; he truly believes that a person is out to get him, but he doesn't have any beliefs about who this person may be. b. de re reading: someone is specific, picking out a particular person. There is some person Peter has in mind and he believes this person is out to get him. One of the earliest semantic analyses of resumptive pronouns is Doron's (1982) de dicto and de re interpretation for Hebrew, a Semitic language. Hebrew is a language where the resumptive strategy freely alternates with the gap strategy. Doron (1982) shows that the gap strategy has only the de dicto meaning and resumption has the de re meaning. Consider the example in (94). The Hebrew statement in (94a) uses the gap strategy and the reading is de dicto, while the statement in (94b) which contains a resumptive pronoun has a de re reading. HEBREW (Doron 1982)  (94) DE DICTO AND DE RE READINGS IN HEBREW a. the gap strategy: de dicto reading Dani  ymca  et  Dani  will find  the woman.Acc  haiSa1  Se  hu  mexapes _____1  that  he  seeks _______  'Dani will find the woman whom he seeks.' de dicto reading: Dani is seeking and will find a woman, whoever she may be; the woman is unspecific and Dani does not necessarily have any beliefs about who she may be.  83  b. the resumptive strategy: de re reading Dani  ymca  et  Dani  will find  the woman.Acc  haiSa1  Se  hu  mexapes  ota1  that  he  seeks  her  'Dani will find the woman whom he seeks [her].' de re reading: Dani has a specific woman in mind. Dani seeks a particular woman and she will find her. The meaning of (94b) can only be "There is a woman that Dani is seeking and he will find this woman". As seen from (94a), the gap strategy gives rise to a de dicto reading corresponding to a non-specific interpretation in which the gap acts as a bound variable. In (94b), the resumptive strategy gives rise to a de re reading, corresponding to a specific reading in this case, the resumptive pronoun is a definite description. In Iraqi Arabic, resumption is obligatory with relativization, as shown in (95), so the de dicto/de re contrast cannot be observed in that context. However, with bare interrogatives, with extraction from an embedded clauses, de dicto/de re can be observed, as shown in (95). (95) IRAQI ARABIC RELATIVE CLAUSE a. .‫راغب راح يلقى المرية اللي يدور عليھا‬ Ragheb ra:H yilga:  il-mraya illyi  Ragheb will meet.3MS the-girl  ʕale:y=ha  yidu:r  who.Rel seek.3MS  for=3FS  Ragheb will find the girl whom he seeks [her]. (10 JUL 2011, SA 4a, offered freely) b.* .‫راغب راح يلقى المرية اللي يدور علي‬ *Ragheb ra:H yilga:  il-mraya illyi  Ragheb will meet.3MS the-girl  yidu:r  who.Rel seek.3MS  ʕalæ____ for____  Ragheb will find the girl whom he seeks. (10 JUL 2011, SA 4a, elicited)  84  In Iraqi Arabic content questions in contexts where both the gap strategy and the resumptive strategy freely alternate, the same de dicto and de re contrast can be observed (96).  (96) a. the gap strategy - de dicto reading  ‫سھا منو تعتقد راح يعزم أحمد ؟‬ Suha minnu: taʕatqid  ra:H yaʕzim______  Suha who  will invite.3MS _____ Ahmad  think.3FS  Ahmad ?  Whom does Suha think Ahmad will invite _____ ? de dicto reading: Ahmad will invite people to the wedding, whoever they may be. The question being non-specific, it is implied that anybody can be invited  b. the resumptive strategy - de re reading  ‫سھا منو تعتقد راح يعزمه أحمد ؟‬ Suha minnu: taʕatqid  ra:H yaʕzim=hu  Suha who  will invite.3MS=3MS Ahmad  think.3FS  Ahmad ?  Whom does Suha think Ahmad will invite [him]? de re reading: The person asking the question has somebody specific in mind. Recall that, in contexts where both strategies are possible, the gap strategy is structurally ambiguous — it permits either [D-N] or [D-ϕ-N] (with covert ϕ), — while the resumptive strategy only permits a [D-ϕ-N] structure (with overt ϕ). I propose that the de dicto/de re contrast found with the gap and the resumptive strategy respectively is the semantic counterpart to this structural distinction. Thus, while the de dicto gap strategy has a [D-N] substructure, the de re resumptive strategy has a [D-ϕ-N] substructure. This is summarized in Table 17.  85  STRATEGY:  GAP  RESUMPTIVE  SYNTAX:  [D-N∅]  [D-ϕ-N∅]  INTERPRETATION”  de dicto: unspecific  de re: specific  Table 17. de dicto/ de re contrast for Iraqi Arabic bare interrogatives (non D-linked content questions) using gap and resumptive strategies With bare interrogatives, the nominal constant is null (this is notated as N∅ in Table 17). As we shall see when we look at D-linked content questions, a similar interpretive contrast arises when N has overt content.  3.2.2 The distinction between single-individual and multiple-individual readings There is also an interpretive contrast between the gap and the resumptive strategy in sentences with quantified expressions (Sharvit 1999). Consider the examples in (97), where the gap/resumptive element in the relative clause is c-commanded by the quantified expression kol gever ‘every man’. The gap strategy in (97a) is ambiguous between a single-individual reading (where the same woman is invited by every man) and a multiple-individual reading (where every man invites a different woman). In contrast, the resumptive strategy in (97b) is only compatible with the single-individual reading. HEBREW (Sharvit 1999)  (97) SINGLE AND MULTIPLE INDIVIDUAL READINGS IN HEBREW a. the gap strategy ha-iʃa  ʃe  kol  gever hizmin ___  the-woman that every man  invited____  hodeta lo thanked him  'The woman every man invited thanked him'. (i) single-individual reading: The woman every man invited thanked him (ii) multiple-individual reading: For every man x the woman that x invited thanked x  86  b. the resumptive strategy ha-iSa  Se  kol  gever hizmin ota  the-woman that every man  hodeta lo  invited her thanked him  'The woman every man invited [her] thanked him.' (i) single-individual reading: The woman every man invited thanked him For Sharvit (1999), the contrast in (97) reflects a pragmatic difference between the gap and resumptive strategies. The interpretive pattern in (97) is consistent with the syntactic analysis developed in Chapter 2: the gap strategy, which is structurally ambiguous between [D-N] and [Dϕ∅-N], is also semantically ambiguous. And the resumptive strategy, which has only the [D-ϕ-N] structure, shows no semantic ambiguity. This is summarized in Table 18.  STRATEGY:  GAP  RESUMPTIVE  SYNTAX:  [D-N]  [D-ϕ∅-N]  [D-ϕ-N]  INTERPRETATION  multiple-individual  single-individual  single-individual  Table 18. Single-individual/ multiple-individual contrast in Modern Hebrew In Iraqi Arabic, resumption is obligatory with relativization, as shown in (98), so the distinction single-individual versus multiple-individual reading cannot be observed in this context; (98a) is the grammatical example employing the resumptive strategy and (98b) is the ungrammatical example employing the gap strategy. (98) SINGLE AND MULTIPLE INDIVIDUAL READINGS IN IRAQI ARABIC a. .‫شكرته‬  il-mraya  ‫المرية اللي كل رجال عزمھا‬ illyi  kull  riʤa:l ʕazam=ha  the-woman who.Rel every man  ʃakarat=hu  invited.3MS=3FS thanked.3FS=3MS  ‘The woman whom every man invited [her] thanked him.’ (10 JUL 2011, SA 5a, offered freely)  87  b.* .‫المرية اللي كل رجال عزم شكرته‬ kull  riʤa:l ʕazam____  *il-mraya  illyi  the-woman  who.Rel every man  ʃakarat=hu  invited.3MS____  thanked.3FS=3MS  ‘The woman whom every man invited thanked him.’ (10 JUL 2011, SA 5b, offered freely)  3.2.3 The distinction between individual, natural function and pairedlist readings Sharvit (1999) observes that another way in which the gap and the resumptive strategy differ from one another concerns the possible answers that can be given to a Dlinked content question. This can be seen with content questions that also contain a quantificational expression, as in (99). D-linked questions with quantifiers can have three possible answers: an expression denoting an individual as in (99a); an expression denoting a natural function (where the “natural function” names a salient function), as in (99b), or a list of pairs, as in (99c). ENGLISH (Sharvit 1999)  (99) Q: Which woman did every man invite ? A: a. individual denoting expression: Mary b. natural function: His mother. c. pair-list answer: John invited Mary; Bill invited Sally. Sharvit (1999) shows that, in Modern Hebrew, if D-linked questions use the gap strategy, as in (100), then all three answers are possible: the individual reading (100a), the natural functional reading (100b) and the pair-list reading (100c). However, with the resumptive strategy, only the individual and natural function readings are possible (101ab); the pair-list reading isn’t possible (101c).  88  HEBREW (Sharvit 1999)  (100) POSSIBLE ANSWERS TO QUESTIONS WITH QUANTIFIERS - GAP STRATEGY ezyo  iʃa  kol  gever hizmin_____  which woman every man invited____ 'Which woman did every man invite___ ?' a. individual denoting expression: et  Gilla  Acc Gilla 'Gilla' b. natural function: et  im-o  Acc mother-his 'his mother' c. pair-list answer: Yosi et  Gilla; Rami et  Rina.  Yosi Acc Gilla Rami Acc Rina 'Yosi, Gilla; Rami, Rina'. (101) POSSIBLE ANSWERS TO QUESTIONS WITH QUANTIFIERS - RESUMPTIVE STRATEGY  ezyo  iʃa  kol  gever hizmin  ota  which woman every man invited  her  'Which woman did every man invite [her] ?' a. individual denoting expression: et  Gilla  Acc Gilla 'Gilla' b. natural function: et  im-o  Acc mother-his 'his mother' *c. pair-list answer: Yosi et  Gilla; Rami et  Rina.  Yosi Acc Gilla Rami Acc Rina 'Yosi, Gilla; Rami, Rina'.  89  In contexts where both the gap and the resumptive strategy are possible, the same interpretive contrast is found in Iraqi Arabic. Thus, as shown in (102), with the gap strategy all three readings are possible: individual (102a), natural function (102b), and pair-list (102c). But with the resumptive strategy, as shown in (103), only the individual and natural function readings are possible, (103a-b); the pair-list reading isn’t available (103c). (102) POSSIBLE ANSWERS TO QUESTIONS WITH QUANTIFIERS - THE GAP STRATEGY  ‫يا مرية كل رجال عزم ؟‬ ya:  mraya  kull  riʤa:l  which woman every man  ʕazam____ invited.3MS  'Which woman did every man invite ___ ?' a. Individual denoting expression: Suha b. Natural function answer: his sister c. Pair-list answer: Samer, Suha; Ahmad, Najwa; etc (103) POSSIBLE ANSWERS TO QUESTIONS WITH QUANTIFIERS - THE RESUMPTIVE STRATEGY  ‫يا مرية كل رجال عزمھا ؟‬ ya:  mraya  kull  riʤa:l  which woman every man  ʕazam=ha invited.3MS=3FS  'Which woman did every man invite [her] ?' a. Individual denoting expression: Suha b. Natural function answer: his sister *c. Pair-list answer: Samer, Suha; Ahmad, Najwa; etc A question that arises is whether resumption is always incompatible with the pairlist reading. The answer is no. This can be seen by looking at contexts where resumption is obligatory. For example, extraction from an island is only possible if there is resumption: thus the gap strategy is ungrammatical (104) and the resumptive strategy is  90  obligatory (105). Observe that in contexts of obligatory resumption, the resumptive strategy is compatible with all three readings: the individual reading (105a), the natural function readings (105b) and the pair-list reading (105c). (104) THE GAP STRATEGY * ‫يا مرية سامر يسأل إذا كل رجال سلم على ؟‬ *ya:  mraya Samer ysʔal  iδa kull  riʤa:l sallæm  which woman Samer ask.3MS if every man  ʕalæ____  greeted.3MS for____  ‘Which woman did Samer wonder if every man greeted ____?’ (10 JUL 2011, SA 6a, elicited) (105) THE RESUMPTIVE STRATEGY  ‫يا مرية سامر يسأل إذا كل رجال سلم عليھا ؟‬ ya:  mraya  which woman  Samer ysʔal  iδa kull  riʤa:l sallæm  Samer ask.3MS if every man  ʕalæ=ha  greeted.3MS for=3FS  ‘Which woman did Samer wonder if every man greeted [her] ?’’ (10 JUL 2011, SA 6a, offered freely) a. Individual reading: Faten b. Natural function answer: his sister c. Pair-list answer: Behjet, Suha; Ahmad, Iman, etc.  3.2.4 The gap strategy competes with the resumptive strategy Table 19 summarizes the findings for Iraqi Arabic regarding the patterning of D-linked content questions:  91  GAP INTERPRETATION  STRATEGY  RESUMPTIVE STRATEGY CONTEXT 1  CONTEXT 2  OPTIONAL  OBLIGATORY  RESUMPTION  RESUMPTION  INDIVIDUAL  √  √  √  NATURAL FUNCTION  √  √  √  PAIR-LIST  √  x  √  Table 19.Interpretations associated with Iraqi Arabic D-linked content questions The range of interpretations associated with Iraqi D-linked content questions indicate that, in contexts where resumption is optional (context 1), the pair-list reading is blocked with the resumptive strategy. This indicates that the two strategies compete with one another. That this is the effect of competition is confirmed by the fact that, when resumption is obligatory (context 2), all three readings are possible. This competition effect is also observed by Malkawi (2009) for Jordanian Arabic, who proposes the economy hierarchy in Table 20. The hierarchy ranks zero variables higher than pronominal variables. Moreover, in Malkawi’s analysis, pronominal variables are of two types: simple versus complex. This reflects that fact that Jordanian has two distinct resumptive strategies: resumption with a weak (clitic) pronoun versus resumptive with a strong pronoun. If we combine Malkawi’s economy hierarchy with the syntactic structures argued for in Chapter 2, we see that the zero variable is structurally ambiguous: it can be [D-N] or [D-ϕ∅-N]. In contrast, the pronominal variables are not structurally ambiguous: they are always [D-ϕ-N].  92  FORMAL  STRATEGY  SYNTAX  STATUS  zero variable  MOST  gap  simple  weak  pronominal  resumptive  ATTESTED  IN  IN  IRAQI  JORDANIAN  ARABIC  ARABIC  √  √  √  √  x  √  [D-N] [D-ϕ∅-N]  ECONOMICAL  ATTESTED  [D-ϕ-N]  variable LEAST  complex  weak  ECONOMICAL  pronominal  resumptive  variable  pronoun  [D-ϕ-N]  doubled by strong pronoun  Table 20. Economy hierarchy for gap and resumptive pronouns in Iraqi Arabic and Jordanian Arabic (adapted from Malkawi 2009)  3.3 What competition reveals about the syntax-semantics interface We have seen that, in contexts where both the gap and the resumptive strategy are possible (context 1), there is a blocking effect. In context 1, the gap strategy is compatible with both the natural function and pair-list readings, but the resumptive strategy is only compatible with the natural function reading. But in contexts where resumption is obligatory (context 2), it is compatible with both readings. In light of these findings, I assess Chierchia’s (1993) proposal that the pair-list reading is a special case of the natural function reading (§3.3.1) and show how this approach accounts for the ambiguity of both the gap and the resumptive strategy in Iraqi Arabic (§3.3.2 - §3.3.3). A consequence of  93  this analysis is that, in certain marked contexts, partial copying is possible (§3.3.4).  3.3.1 The natural function reading subsumes the pair-list reading Chierchia (1993) defines the distinction between natural reading and pair-list reading as follows: “understanding a “natural function” like [λx [mother of(x)]] means understanding its intension, not its extension. Lists viewed as functions are just the opposite. They cannot be characterized but in terms of the set of their inputs and outputs. They can only be grasped by scanning their graph. They are, as it were, pure extension.” Thus Chierchia (1993) rejects the hypothesis that a question of the form “Whom does every Italian like ?” has three independent interpretations as in (106). (Chierchia 1993) (106) SEMANTIC REPRESENTATIONS OF “WHOM DOES EVERY ITALIAN LIKE?” (i) [p: p is true and for some x, p = every Italian likes x] ‘individual reading’ (ii) [p: p is true and form some f, p = every Italiany likes f(y) ] ‘natural function reading’ (iii) for every Italian y: who does y like? ‘pair list reading’ According to Chierchia (1993), the common structure of natural function and pair-list readings is the one given in (107), where f is a variable ranging over functions. (107) Qx [x loves f(x)] Chierchia (1993) concludes that: “The natural function reading subsumes as a special case the so called pair list reading, given that lists are just functions of a certain kind. Why only universal NPs support lists follows from general semantic and pragmatic considerations (namely to draw a list one needs a domain, which is naturally supplied only by universals)”. In Chierchia’s analysis, the natural function reading is intentional, while the pairlist reading is extensional. I suggest that the intentional/ extensional distinction has a 94  structural counterpart: the [D-N] structure is interpreted extensionally as a pair-list, while the [D-ϕ-N] structure is interpreted intentionally (as a natural function). This captures the fact that the gap strategy — which is structurally ambiguous between [D-N] and [D-ϕ∅N] — is compatible with both the extensional pair-list reading, and the intentional natural function readings. It also accounts for the fact that, in contexts where resumption is optional (Context 1), only the intentional natural function reading is possible with the resumptive strategy. This is summarized in Table 21.  STRATEGY SYNTAX SEMANTICS INTERPRETATION  GAP  [D-N]  RESUMPTIVE  [D-ϕ∅-N]  [D-ϕ-N]  EXTENSIONAL  INTENTIONAL  PAIR-LIST  NATURAL FUNCTION  Table 21. Extensional/ Intentional contrast and the gap/ resumptive strategy in Iraqi Arabic (Context 1: optional resumption)  But recall that in contexts where resumption is obligatory (Context 2), the resumptive strategy is compatible with both the pair-list and the natural function reading. I now show in greater detail exactly how the interpretive difference between Context 1 (optional resumption) and Context 2 (obligatory resumption) is accounted for.  3.3.2 The gap strategy is always syntactically and semantically ambiguous Assume that the pair-list reading has a [D-N] structure and that the natural function reading has a [D-ϕ-N] structure. Given that the gap strategy is structurally ambiguous between a [D-N] and a [D-ϕ∅-N] structure, this correctly predicts that the gap strategy will also be semantically ambiguous between a pair-list reading and a natural function reading. To see this, consider the partial derivations in (108a) and (108b-c):  95  (108) a. OPTIONAL RESUMPTION CONTEXT – THE GAP STRATEGY  ‫يا صورة البنھا كل مرية شققت ؟‬ ya:  Su:ra.F  li-bni=ha  kull  mraya  which picture.F of-son=her every woman  ʃagagat____ tore.3FS  'Which photo of her son did every woman tear___ ?'6 (i) Natural function answer: the photo of his last passport (ii) Pair-list answer: Iman tore the picture of her son Faris, Awatif the picture of her son Adel, etc b. D-N structure – pair-list reading [C [D ya:D Su:raN li-bni=ha] [I [V [kull mrayaD ] [V [ʃagagtV] [D ya:D Su:raN li-bni=ha N]]] c. D-φ-N reconstruction with covert φ – natural function reading [C [D ya:D ϕ∅ Su:raN li-bni=ha] [I [V [kull mrayaD ] [V [ʃagagtV ϕ∅] [D ya:D ϕ∅ Su:raN libni=ha N]]]  3.3.3 The weak resumptive strategy is sometimes syntactically and semantically ambiguous If the natural function reading has a [D-ϕ-N] structure, then this correctly predicts that when there is an overt resumptive pronoun, only the natural function reading will be possible. This corresponds to (109-a) and the partial derivation given in (109-b).  6  There is a possible reading where the pronoun "her" refers to a particular woman, let's say Mary as in "Which photo of [Mary's] son did every woman tear?". In this thesis we are not interested in this reading, but it must be mentioned that it exists.  96  (109) a. CONTEXT 1: OPTIONAL RESUMPTION  ‫يا صورة البنھا كل مرية شققتھا ؟‬ ya:  Su:ra.F  li-bni=ha  kull  mraya  ʃagagat=ha  which picture.F of-son=her every woman tore.3FS=3FS 'Which photo of her son did every woman tear [it] ?' (13 NOV 2010, SA 10, elicited) (i) Natural function answer: the photo of his last passport (ii) *Pair-list answer: Iman tore the picture of her son Faris, Awatif the picture of her son Adel, etc b. D-φ-N reconstruction with overt φ – natural function reading [C [D ya:D haφ Su:raN li-bni=ha] [I [V [kull mrayaD ] [V [ʃagagtV haφ] [D ya:D haφ Su:raN libni=ha N]]] Now consider (110a), which is a context where the resumptive strategy is obligatory. Here we observe that both the natural function and pair-list readings are possible. By hypothesis, the natural function reading arises with the [D-ϕ-N] structure, (110-b). But what accounts for the availability of the pair-list reading? According to the proposed analysis, the pair-list reading arises with the [D-N] structure. I suggest that when resumption is obligatory, the pair-list reading can be forced by copying only the phonological substring that remains after cliticization has applied; this means that only [D-N] is copied, as in (110-c).  97  (110) a. Context 2: OBLIGATORY RESUMPTION  ‫يا صورة البنھا سامر يسأل إذا كل مرية شققتھا ؟‬ ya:  Su:ra.F  li-bni=ha  Samer ysʔal  which picture.F of-son=her Samer ask.3MS  iδa kull  mraya  ʃagagat=ha  if every woman tore.3FS=3F  Which photo of her son did Samer wonder if every woman tore [it] ? (13 NOV 2010, SA 14a, elicited) (i) Natural function answer: the picture of her son's wedding (ii) Pair-list answer: Iman, the picture of her son's wedding; Awatif, the picture of her son with his girlfriend. b. Copy [D-ϕ-N]: natural function reading [C [D ya:D haϕ Su:raN li-bni-ha] [I [V Samer ysʔal iδa kull mraya [V ʃagagat-haϕ [D ya:D haφ Su:raN li-bni-haN ] ] ] c. Copy [D-N]: pair-list reading [C [D ya:D Su:raN li-bni-ha] [I [V Samer ysʔal iδa kull mraya [V ʃagagat-haϕ [D ya:D haϕ Su:raN li-bni-haN ] ] ]  3.3.4 Implications of the analysis: partial copying A consequence of this proposal is that, in contexts where resumption is obligatory, it is possible to only copy the phonological sub-string that contains [D-N]. It is this partial copy that permits the pair-list reading with the resumptive strategy. In Iraqi Arabic, such partial copying is only possible in contexts where the gap strategy is ruled out.  98  4 Genitive interrogatives as inherently Dlinked content questions In this chapter, I document a parallel between genitive interrogatives and D-linked content questions, which to my knowledge has not been investigated before. In particular, I argue that genitive interrogatives are inherently D-linked and I explore the nature of this D-linking property. I first introduce the problem, the proposed analysis, and its consequences (§4.1). I then consider in greater detail how, on the one hand, genitive interrogatives differ from bare interrogatives, and on the other hand, how they parallel Dlinked content questions (§4.2). I argue that the property that ties together genitive and Dlinked interrogatives is the fact that they both have an overt domain restriction (§4.3).  4.1 The problem, the analysis and consequences I introduce the contrast between genitive and bare interrogatives (§4.1.1), I sketch an analysis that account for this contrast (§4.1.2) and I present the consequence of this analysis (§3.1.2).  4.1.1 The contrast between genitive and bare interrogatives  I start with the observation that, given the context in (111), in Iraqi Arabic a content question with a bare interrogative pronoun is not felicitous (111a). In contrast, a content question with a construct state genitive is felicitous (111b), as is a D-linked content question (111c).  99  111. Context: a class of students is defending their theses; they all have different supervisors. The secretary of the board has to talk to each student's supervisor.  a. content question with bare interrogative pronoun  ‫ويا منو حاكت السكريتيرة ؟‬ wu:ya: minnu: Haʧit with  who  is-sikriti:ra  spoke.3FS the-secretary.F  #'Whom did the secretary talk to ?' (26 JAN 2011, SA 1, elicited)  b. content question with genitive interrogative construction  ‫السكريتيرة ويا أستاذ منو حاكت ؟‬ is-sikriti:ra  wu:ya: ʔusta:δ  the-secretary.F with  minnu: Ha:ʧit  professor who  spoke.3FS  'With whose professor did the secretary talk ?' (26 JAN 2011, SA 1, offered freely)  c. content question with D-linked interrogative  ‫السكريتيرة وية يا أستاذ حاكت ؟‬ ʔusta:δ  is-sikri:ti:ra wu:ya  ya:  Ha:ʧit  the-secretary with  which professor talk.3FS  ‘With which professor did the secretary talk ?’ (6 JUL 2011, SA 3, offered freely) (111) establishes that in certain contexts, a content question with a bare interrogative is not felicitous, while the same question becomes felicitous once a genitive interrogative or a D-linked interrogative is used. This is a first indication that bare and genitive interrogatives don't pattern in the same way; it also indicates that there is a parallel between genitive and D-linked interrogatives. The question that I address is the following:  100  (112) What accounts for the contrast between bare interrogatives and genitive interrogatives  on  the  one  hand,  and  the  parallel  between  genitive  interrogatives and D-linked interrogatives on the other hand?  4.1.2 The analysis: genitive interrogatives are inherently D-linked I propose that genitive interrogatives are inherently D-linked. Specifically, I argue that what defines D-linking is the presence of an overt domain restriction in the form of an overt noun. Consider Table 22. Bare interrogatives such as minnu: 'who' lack an overt domain restriction. In contrast, both genitive interrogatives and D-linked content questions have an overt domain restriction. With genitive interrogatives such as ʔustaδ minnu: 'whose professor', the domain restriction is supplied by a head noun. With Dlinked interrogatives such as ya ʔustaδ 'which professor', the overt domain restriction is supplied by the noun that follows the interrogative operator.  SYNTAX  EXAMPLE  BARE INTERROGATIVE  [D WH [N ∅ ] ]  minnu:  ‘who’  GENITIVE  [D [N N [D WH ] ] ]  ʔustaδ minnu:  ‘whose professor  [D WH [N N ] ]  ya: ʔustaδ  ‘which professor’  INTERROGATIVE  D-LINKED INTERROGATIVE  Table 22. Internal structure of Iraqi Arabic bare, genitive and D-linked interrogatives  4.1.3 Consequence: D-linking arises whenever there is an overt domain restriction I take the syntactic parallel between genitive and D-linked interrogatives to  101  indicate that D-linking arises whenever there is an overt domain restriction. This has consequences for our understanding of how D-linking interacts with the resumptive strategy. On the one hand, domain restriction is purely semantic and arises when a quantifier has an overt restriction on its domain of application. Quantifiers with no overt restriction — such as each, who and what in (113) — are generally taken to have a contextually defined domain restriction.  (113) a.  They each attended the lecture.  b.  Who attended the lecture?  c.  What did they attend?  Of course, it’s always possible to introduce an overt restriction: these are the underlined nouns in (114). Semantically, it’s the presence of an overt domain restriction that distinguishes D-linked interrogatives (e.g. which student, which lecture) from bare interrogatives (e.g. who, what).  (114) a.  Each student attended the lecture.  b.  Which student attended the lecture?  c.  Which lecture did they attend?  As we shall see in more detail below, the Iraqi Arabic data from genitive interrogatives indicates that the head noun of a genitive also counts as a domain restriction: these are the underlined nouns in (115). Moreover, the presence of an overt domain restriction has syntactic consequences in Iraqi Arabic. These will be discussed in more detail at the end of this chapter.  (115) a.  Everyone’s student attended the lecture.  b.  Whose student attended the lecture?  c.  Whose lecture did they attend?  102  4.2 Comparing bare, genitive and D-linked interrogatives In this section, I compare the distribution of bare, genitive, and D-linked interrogatives with respect to three diagnostics: local extraction (§4.2.1); long-distance extraction (§4.2.2); superiority effects (§4.2.3). This comparison reveals that genitive interrogatives consistently differ from bare interrogatives, and they consistently parallel D-linked interrogatives.  4.2.1 The resumptive and gap strategy: local extraction I show how the resumptive and the gap strategy pattern with local extraction (extraction from a single-clause), as concerns bare interrogatives (§4.2.1.1), genitive interrogatives (§4.2.1.2) and D-linked interrogatives (§4.2.1.3). The section closes with a summary of the findings (§4.2.1.4).  4.2.1.1 Local extraction with bare interrogatives First consider local extraction. Here bare interrogatives allow only the gap strategy with extracted subjects and objects, as in (116) and (117). As for extracted prepositional objects (118), they permit neither the gap strategy nor the resumptive strategy: this reflects the general prohibition against P-stranding in Arabic. PP-fronting allows only the gap strategy, because Arabic does not have resumptives for entire prepositional phrases (119).  103  (116) SUBJECT EXTRACTION OF BARE INTERROGATIVE a. Subject extraction with gap  ‫منو اشترى الجريدة البارحة ؟‬ minnu: iʃtara: who  il- ʤarida  il-ba:riha ?  bought. 3MS the-newspaper  yesterday  'Who bought the newspaper yesterday ?' (8 JUL 2010, SA 1, offered freely)  b. Subject extraction with resumption  * ‫منو ھو اشترى الجريدة البارحة ؟‬ il- ʤarida  *minnu: hwu iʃtara: who  he  bought. 3MS  il-ba:riha ?  the-newspaper  yesterday  '*Who [he] bought the newspaper yesterday ?' (8 JUL 2010, SA 1, elicited) (117) DIRECT OBJECT EXTRACTION OF BARE INTERROGATIVE a. Direct object extraction with gap  ‫إيمان منو شافت ببيت عواطف ؟‬ Iman minnu: ʃa:fat______ bi-beyt Iman who  saw.3SF  Awatif  in-house Awatif  'Whom did Iman see at Awatif's house ?' (8 JUL 2010, SA 3a, offered freely)  b. Direct object extraction with resumption  *‫إيمان منو شافته ببيت عواطف ؟‬ *Iman minnu: ʃa:fat=hu Iman who  saw.3SF=3MS  bi-beyt  Awatif  in-house Awatif  '*Whom did Iman see [him] at Awatif's house ?' (8 JUL 2010, SA 3a, elicited)  104  (118) PREPOSITIONAL OBJECT EXTRACTION OF BARE INTERROGATIVE a. Object of preposition extraction with gap  *‫راغب منو التقى بالمكتبة ويا ؟‬ *Ragheb minnu: iltaga Ragheb who  bi-l-maktaba wu:ya: ____  met.3MS at-the-library with____  'Whom did Ragheb meet at the library with___ ?' (8 JUL 2010, SA 12, offered freely)  b. Object of preposition extraction with resumption  *‫راغب منو التقى وياه بالمكتبة ؟‬ *Ragheb minnu: Ragheb  who  iltaga  wuya:=h  bi-l-maktaba  met.3MS  with=him  at-the-library  '*Whom did Ragheb meet with [him] at the library ?' (8 JUL 2010, SA 12, elicited) (119) PP-FRONTING WITH BARE INTERROGATIVE  ‫راغب ويا منو التقى بالمكتبة ؟‬ Ragheb wu:ya: minnu: iltaga Ragheb with  who  bi-l-maktaba  met.3MS at-the-library  'With whom did Ragheb meet at the library ?' (8 JUL 2010, SA 12, offered freely)  4.2.1.2 Local extraction with genitive interrogatives The overall pattern with genitive interrogatives differs from that of bare interrogatives. As before, with extraction from subject position, the gap strategy but not resumption is possible, as in (120). In this respect, a genitive interrogative is like a bare interrogative. But local extraction from the direct object position, as in (121), allows both the gap strategy and the resumptive strategy. (Bare interrogatives allow only the gap strategy in this context.) As for extraction from a prepositional object position, the gap  105  strategy is predictably prohibited because of the impossibility of P-stranding, as in (122). Finally, with PP-fronting, the gap strategy but not resumption is possible, (123). (120) EXTRACTION FROM SUBJECT POSITION WITH GENITIVE a. the gap strategy  ‫رجال منو شاف نجوى بالحفلة؟‬ minnu: ʃa:f  riʤa:l  husband who  Najwa bi-l-Hafla  saw.3MS Najwa at-the-party  ‘Whose husband saw Najwa at the party ?’  b. the resumptive strategy  *‫رجال منوھو شاف نجوى بالحفلة؟‬ *[riʤa:l husband  minnu:]1 hwu1 ʃa:f who  he  Najwa bi-l-Hafla  saw.3MS Najwa at-the-party  ‘Whose husband1 [he1] saw Najwa at the party ?’ (121) OBJECT EXTRACTION OF GENITIVE INTERROGATIVE a. the gap strategy  ‫نجوى رجل منو شافت بالمكتبة ؟‬ Najwa riʤal minnu: ʃa:fat Najwa man who  bi-l-maktaba  saw:3SF in-the-library  'Whose husband did Najwa see at the library ?' (8 JUL 2010, SA 8a, offered freely)  b. the resumptive strategy  ‫إيمان رجل منو شافته بالمكتبة ؟‬ Iman  riʤal minnu: ʃa:fat=hu  Iman  man who  bi-l-maktaba  saw:3SF.3MS at-the-library  'Whose husband did Najwa see [him] at the library ?' (8 JUL 2010, SA 8b, offered freely)  106  (122) PREPOSITIONAL OBJECT EXTRACTION OF GENITIVE INTERROGATIVE a. the gap strategy  *‫نجوى صديق منو التقت وية بالحفلة؟‬ *Najwa Sadig minnu: iltagat Najwa friend who  wu:ya: _____ bi-l-Hafla  met.3FS with_______ at-the-party  ‘Whose friend did Najwa meet with __ at the party ?’  b. the resumptive strategy  ‫نجوى صديق منو التقت وياه بالحفلة؟‬ Najwa Sadig minnu: iltagat Najwa friend who  wu:ya:=h  bi-l-Hafla  met.3FS with=3MS at-the-party  ‘Whose friend did Najwa meet with [him] at the party ?’ (123) PP-FRONTING WITH GENITIVE INTERROGATIVE  ‫نجوى وية صديق منو التقت بالحفلة؟‬ Najwa wu:ya: Sadi:g minnu: iltagat  bi-l-Hafla  Najwa with  at-the-party  friend who  met.3FS  ‘With whose friend did Najwa meet at the party ?’  4.2.1.3 Local extraction with D-linked interrogatives Now consider local extraction with D-linked interrogatives. (Similar data was already presented in Chapter 2, but for ease of exposition, I repeat it here.) With a Dlinked interrogative, subject extraction is possible only with gap (124); direct object extraction permits gap and resumption (125), prepositional object extraction permits only resumption (126) and PP-fronting permits only the gap strategy (127).  107  (124) SUBJECT EXTRACTION OF D-LINKED INTERROGATIVE a. extraction with gap  ‫يا صديقة اشترت شقة ببغداد ؟‬ ya:  ʃigga  Sadi:ga iʃtarat_____  bi-Baghdad  which friend.F bought.3FS_____ apartment in-Baghdad 'Which friend bought an apartment in Baghdad ?' (1 DEC 2010, SA, offered freely)  b. extraction with resumption  *‫يا صديقة اشترت ھي شقة ببغداد ؟‬ *ya:  Sadi:ga iʃtarat  hi: ʃigga  bi-Baghdad  which friend.F bought.3FS she apartment in-Baghdad 'Which friend [she] bought an apartment in Baghdad ?' (1 DEC 2010, SA, elicited) (125) DIRECT OBJECT EXTRACTION OF D-LINKED INTERROGATIVE a. extraction with gap  ‫إيمان يا رجال شافت بالحفلة ؟‬ Iman ya:  riʤa:l ʃa:fit ____  Iman which man  saw.3FS____  bi-l-hafla at-the-party  'Which man did Iman see___ at the party ?' (1 DEC 2010, SA 1I, offered freely)  b. extraction with resumption  ‫إيمان يا رجال شافته بالحفلة ؟‬ Iman ya:  riʤa:l ʃa:fit=hu  Iman which man  saw.3FS=3MS  bi-l-hafla at-the-party  'Which man did Iman see [him] at the party ?' (1 DEC 2010, SA 1II, offered freely)  108  (126) PREPOSITIONAL OBJECT EXTRACTION OF D-LINKED INTERROGATIVE a. extraction with gap  *‫سھى يا معلم التقت بالكلية وية ؟‬ *Suha ya:  muʕallim iltagat wu:ya: ___  bi-l-kulli:a  Suha which  professor met.3FS with ___ at-the-faculty  'Which professor did Suha meet with____ at the faculty ?' (1 DEC 2010, SA 3I, offered freely)  b. extraction with resumption  ‫سھى يا معلم التقت وياه بالكلية ؟‬ Suha ya:  muʕallim iltagat  wu:ya:=h  bi-l-kulli:a  Suha which  professor met.3FS with=3MS at-the-faculty  'Which professor did Suha meet with [him] at the faculty ?' (1 DEC 2010, SA 3II, offered freely) (127) PP-FRONTING WITH D-LINKED INTERROGATIVE  ‫سھى وية يا أستاذ التقت بالكلية؟‬ Suha wu:ya: ya Suha with  ʔusta:δ  which professor  iltagat  bi-l-kullyia  met.3FS at-the-faculty  ‘With which professor did Suha meet at the faculty?’  4.2.1.4 Summary of the local extraction data  Table 23 summarizes the patterning of the gap and resumptive strategy in the context of local extraction with bare, genitive interrogatives and D-linked interrogatives:  109  GAP STRATEGY  RESUMPTIVE STRATEGY  BARE  GENITIVE  D-LINKED  BARE  GENITIVE  D-LINKED  SUBJECT  √  √  √  x  x  x  OBJECT OF V  √  √  √  x  √  √  OBJECT OF P  x  x  x  x  √  √  PP-FRONTING  √  √  √  x  x  x  INTERROGATIVE TYPE  Table 23. Comparison of bare, genitive and D-linked interrogatives with respect to local extraction (extraction from a single clause) Table 23 indicates the following. The resumptive strategy is always prohibited with bare interrogatives. But with genitive and D-linked interrogatives, it is permitted with direct objects and prepositional objects. More generally, we observe that, with respect to local extraction, genitive and D-linked interrogatives pattern in the same way.  4.2.2. The resumptive and gap strategy: long-distance extraction I now examine how the resumptive and the gap strategy pattern with longdistance extraction (extraction from an embedded clauses), as concerns bare interrogatives (§4.2.2.1), genitive interrogatives (§4.2.2.2), and D-linked interrogatives (§4.2.2.3). The section closes with a summary of the findings (§4.2.2.4).  4.2.2.1 Long-distance extraction with bare interrogatives With bare interrogatives, long-distance extraction from subject position only allows the gap strategy, as in (128). Long-distance extraction from the object position permits both gap and resumption, as in (129). And long-distance extraction of the prepositional object is ruled out: neither gap nor resumption are possible, as in (130). PPfronting is also possible with long-distance extraction (131).  110  (128) SUBJECT EXTRACTION OF BARE INTERROGATIVE a. Subject extraction with gap  ‫إيمان منو تعتقد شاف أحمد بالحفلة ؟‬ Iman minnu: taʕataqid____ ʃa:f Iman who  think:2S____  Ahmad  saw:3MS Ahmad  bi-l-hafla at-the-party  'Who does Iman think ____ saw Ahmad at the party ?' (8 JUL 2010, SA 9a, offered freely)  b. Subject extraction with resumption  * ‫إيمان منو تعتقد ھو شاف أحمد بالحفلة ؟‬ *Iman minnu: taʕataqid huwwa ʃa:f Iman who  think:2S he  Ahmad  bi-l-hafla  saw:3MS Ahmad  at-the-party  'Who does Iman think [he] saw Ahmad at the party ?' (8 JUL 2010, SA 9b, elicited) (129) OBJECT EXTRACTION OF BARE INTERROGATIVE a. Direct object extraction with gap  ‫سھى منو تعتقد راح يعزم أحمد ؟‬ Suha minnu: taʕatagid Suha who  ra:H yaʕzim____  Ahmad ?  think.3FS will invite.3MS____ Ahmad  'Whom does Suha think that Ahmad will invite____ ?' (8 JUL 2010, SA 10a, offered freely)  b. Direct object extraction with resumption  ‫سھى منو تعتقد راح يعزمه أحمد ؟‬ Suha minnu: taʕatagid Suha who  ra:H yaʕzim=hu  Ahmad ?  think.3FS will invite.3MS=3MS Ahmad  'Whom does Suha think that Ahmad will invite [him] ?' (8 JUL 2010, SA 10b, offered freely)  111  (130) PREPOSITIONAL OBJECT EXTRACTION OF BARE INTERROGATIVE a. Prepositional Object extraction with gap  *‫إيمان من تعرف إنه بھجت كتب الرسالة إلى ؟‬ *Iman man  taʕarif  ennu: Bahjat  Iman who know:3FS that  Bahjat  kitab  ir-risala ʔila___  wrote.3MS the-letter to___  'Whom does Iman know that Bahjat wrote the letter to ?' (8 JUL 2010, SA 12a, offered freely)  b. Prepositional Object extraction with resumption  *‫إيمان من تعرف إنه بھجت كتب الرسالة له ؟‬ *Iman man Iman who  taʕarif  ennu: Bahjat  know:3FS that  Bahjat  kitab  ir-risala la=hu  wrote.3MS the-letter to=3MS  'Whom does Iman know that Bahjat wrote the letter to [him] ?' (8 JUL 2010, SA 12b, elicited) (131) PP-FRONTING  ‫إيمان إلمن تعرف إنه بھجت كتب الرسالة ؟‬ Iman il-man  taʕarif  Iman to-who  know:3FS that  ennu: Bahjat Bahjat  kitab  ir-risala  wrote.3MS the-letter  'To whom does Iman know that Bahjat wrote the letter ?' (8 JUL 2010, SA 12c, offered freely)  4.2.2.2 Long-distance extraction with genitive interrogatives Now consider long-distance extraction with genitive interrogatives. As before, with subject extraction, only the gap strategy is possible (132). With object extraction, both gap and resumption are allowed (133). The same holds of long-distance extraction of a prepositional object: both gap and resumption are allowed (134). And with PPfronting, only the gap strategy is possible (135).  112  (132) SUBJECT EXTRACTION OF GENITIVE INTERROGATIVE a. the gap strategy  ‫سھى رجال منو تعتقد إنو شاف نجوى بالحفلة؟‬ Suha riʤa:l  minnu: taʕatagid ennu: ʃa:f  Suha husband who  think.3FS that  Najwa bi-l-Hafla  saw.3MS Najwa at-the-party  ‘Whose husband does Suha think that saw Najwa at the party ?’  b. the resumptive strategy  *‫سھى رجال منو تعتقد إنو ھو شاف نجوى بالحفلة؟‬ *Suha riʤa:l Suha husband  minnu: taʕatagid ennu:  hu: ʃa:f  who  he saw.3MS Najwa at-the-party  think.3FS that  Najwa bi-l-Hafla  ‘Whose husband does Suha think that [he] saw Najwa at the party ?’ (133) OBJECT EXTRACTION OF GENITIVE INTERROGATIVE a. the gap strategy  ‫سھى رجل منو تعتقد إنو نجوى شافت بالمكتبة ؟‬ Suha riʤal  minnu: taʕatagid ennu: Najwa  Suha husband who  think  that  Najwa  ʃa:fat  bi-l-maktaba  saw:3SF at-the-library  'Whose husband does Suha think that Najwa saw at the library ?' (8 JUL 2010, SA 8a, offered freely)  b. the resumptive strategy  ‫سھى رجال منو تعتقد إنو نجوى شافته بالمكتبة ؟‬ Suha riʤa:l  minnu: taʕatagid ennu: Najwa  Suha husband who  think  that  Najwa  ʃa:fat=hu  bi-l-maktaba  saw:3SF=3MS at-the-library  'Whose husband does Suha think that Najwa saw [him] at the library ?' (8 JUL 2010, SA 8b, offered freely)  113  (134) PREPOSITIONAL OBJECT EXTRACTION OF GENITIVE INTERROGATIVE a. the gap strategy  *‫سھى صديق منو تعتقد إنو نجوى التقت وية بالحفلة؟‬ *Suha Sadig minnu: taʕatagid ennu: Najwa iltagat Suha friend who  wu:ya: _____ bi-l-Hafla  met.3FS with_______ at-the-party  ‘Whose friend did Najwa meet with __ at the party ?’  b. the resumptive strategy  ‫نجوى صديق منو التقت وياه بالحفلة؟‬ Najwa Sadig minnu: iltagat Najwa friend who  wu:ya:=h  bi-l-Hafla  met.3FS with=3MS at-the-party  ‘Whose friend did Najwa meet with [him] at the party ?’ (135) PP-FRONTING WITH GENITIVE INTERROGATIVE  ‫سھى وية صديق منو تعتقد إنو نجوى التقت بالحفلة؟‬ Suha wu:ya: Sadi:g minnu: taʕatagid ennu: Najwa iltagat Suha with  friend who  bi-l-Hafla  think.3FS that Najwa met.3FS at-the-party  ‘With whose friend does Suha think that Najwa meet at the party ?’  4.2.2.3 Long-distance extraction with D-linked interrogatives Long-distance extraction of D-linked interrogatives was already discussed in Chapter 2. Again for ease of exposition, I repeat the relevant data here. With subject extraction, only gapping is possible (136). With object extraction, both gapping and resumption are possible (137). With long-distance extraction of a prepositional object, only resumption is possible (138). And with PP-fronting, only the gap strategy is possible (139).  114  (136) SUBJECT EXTRACTION OF D-LINKED INTERROGATIVE  ‫راغب يا صديقة يعتقد إنو اشترت شقة ببغداد ؟‬ Ragheb ya:  ʃigga  Sadi:ga yaʕatagid ennu: iʃtarat_____  Ragheb which friend.F think.MFS that  bi-Baghdad  bought.3FS_____ apartment in-Baghdad  'Which friend does Ragheb think that bought an apartment in Baghdad ?' (1 DEC 2010, SA, offered freely)  b. extraction with resumption  *‫راغب يا صديقة يعتقد إنو ھي اشترت شقة ببغداد ؟‬ *Ragheb ya:  Sadi:ga yaʕatagid ennu: hyi iʃtarat_____  Ragheb which friend.F think.MFS that  ʃigga  bi-Baghdad  she bought.3FS____ apartment in-Baghdad  'Which friend does Ragheb think that [she] bought an apartment in Baghdad ?' (1 DEC 2010, SA, elicited) (137) OBJECT EXTRACTION OF D-LINKED INTERROGATIVE a. extraction with gap  ‫راغب يا رجال تعتقد إنو إيمان شافت بالحفلة ؟‬ Ragheb ya:  riʤa:l yaʕatagid ennu: Iman ʃa:fit ____  Ragheb which man  bi-l-Hafla  think.3MS that Iman saw.3FS____  at-the-party  'Which man does Ragheb think that Iman saw___ at the party ?' (1 DEC 2010, SA 1I, offered freely)  b. extraction with resumption  ‫راغب يا رجال تعتقد إنو إيمان شافته بالحفلة ؟‬ Ragheb ya:  riʤa:l yaʕatagid ennu: Iman ʃa:fit=hu  Ragheb which man  bi-l-Hafla  think.3MS that Iman saw.3FS=3MS  at-the-party  'Which man does Ragheb think that Iman saw [him] at the party ?' (1 DEC 2010, SA 1II, offered freely)  115  (138) PREPOSITIONAL OBJECT EXTRACTION OF D-LINKED INTERROGATIVE a. extraction with gap  *‫راغب يا معلم تعتقد إنو سھى التقت وية بالكلية ؟‬ *Ragheb ya:  muʕallim taʕatagid ennu: Suha iltagat wu:ya: ___  bi-l-kulli:a  Ragheb which professor met.3FS with ___ at-the-faculty 'Which professor does Ragheb think that Suha met with____ at the faculty ?' (1 DEC 2010, SA 3I, offered freely)  b. extraction with resumption  ‫راغب يا معلم تعتقد إنو سھى التقت وياه بالكلية ؟‬ *Ragheb ya:  muʕallim taʕatagid ennu: Suha iltagat wu:ya:=hu  bi-l-kulli:a  Ragheb which professor met.3FS with ___ at-the-faculty 'Which professor does Ragheb think that Suha met with____ at the faculty ?' (1 DEC 2010, SA 3II, offered freely) (139) PP FRONTING WITH D-LINKED INTERROGATIVE  ‫راغب وية يا معلم يعتقد إنو سھى التقت بالكلية ؟‬ Ragheb wu:ya ya:  muʕallim yaʕatagid ennu: Suha iltagat wu:ya:=hu bi-l-kulli:a  Ragheb with which professor think.3MS that Suha met.3FS with=3MS at-the-faculty 'Which professor does Ragheb think that Suha met with____ at the faculty ?' (1 DEC 2010, SA 3II, offered freely)  4.2.2.4 Summary of the long-distance extraction data Table 24 summarizes the patterning of the gap and resumptive strategy in the context of long-distance extraction with bare, genitive interrogatives and D-linked interrogatives.  116  GAP STRATEGY  RESUMPTIVE STRATEGY  BARE  GENITIVE  D-LINKED  BARE  GENITIVE  D-LINKED  SUBJECT  √  √  √  x  x  x  OBJECT OF V  √  √  √  √  √  √  OBJECT OF P  x  x  x  x  √  √  PP-FRONTING  √  √  √  x  x  x  INTERROGATIVE TYPE  Table 24. Comparison of bare, genitive and D-linked interrogatives with respect to long-distance extraction (extraction from an embedded clause) As with local extraction, with long-distance extraction we observe that genitive and D-linked interrogatives pattern in the same way with respect to whether they use the gap or the resumptive strategy. And as before, bare interrogatives are distinct form genitive/D-linked interrogatives.  4.2.3 Superiority effects Superiority effects arise in contexts where two interrogative expressions are contained in the same clause. As we shall see, in Iraqi Arabic bare interrogatives show superiority effects, while genitive and D-linked interrogatives don’t. The superiority condition (Chomsky 1973), as stated in (140) is an attempt to account for the contrast between the well-formed (141) and ill-formed (142).  (140) Superiority Condition No rule can involve X and Y in the structure …X… […Z… Y…]… where the rule could also apply to X and Z , and Z is superior to Y (Z is superior to Y if Z ccommands Y) (141) a. b.  Who saw what ? I wonder who saw what.  117  *What did who see?  (142) a.  *I wonder what who saw  b.  The superiority condition derives the fact that, in sentences where both the subject and object are interrogative expressions, only the subject (Z) can undergo movement; i.e. the movement rule involves X and Z. It correctly prohibits movement of the object over the subject, as this would be an instance of a rule involving X (the object position) and Y (the A’ landing site), with Z (the subject) superior to Y (the object). As shown in (143), Iraqi Arabic bare interrogatives obey the superiority condition:  143. superiority effects with bare interrogative pronouns a. superiority effect observed  ‫منو شنو قال لسامر؟‬ minnu: ʃenu: ga:l li-Samer who  what said.3MS to=Samer  Who said what to Samer ? (10 JUL 2010, SA 1a, offered freely)  b. superiority effect violated  *‫شنو قال منو لسامر؟‬ * ʃenu: ga:l what=said.3MS who  minnu:  li-Samer  to=Samer  '*What did who say to Samer ?' (10 JUL 2010, SA 1b, elicited) With genitive interrogative constructions, the superiority effect disappears. This is illustrated by the grammaticality of both multiple questions in (144), where (144a) shows SVO word order and (144b) shows OVS word order.  118  (144) Superiority effects disappear with genitive interrogative expressions a. SVO word order  ‫رجل منو شاف ولد منو ؟‬ riʤa:l minnu: ʃa:f man  who  walad minnu:  saw.3MS boy  who  'Whose husband saw whose son ?' (13 JAN 2011, SA 8a, offered freely)  b. OVS word order  ‫ولد منو شافه رجل منو؟‬ walad minnu: ʃa:f=hu boy  who  riʤa:l  minnu:  saw.3MS=3MS husband who  'Whose son did whose husband see ?' (13 JAN 2011, SA 9a, offered freely) With D-linked interrogatives superiority effects also disappear. This is illustrated in the grammaticality of the D-linked questions in (145), where (145a) shows SVO word order and (145b) shows OVS word order.  (145) Superiority effects disappear with D-linked interrogative expressions a. SVO word order  ‫يا طالبة اشترت يا كتاب ؟‬ ya:  Ta:liba  iʃtarat  ya:  kita:b  which student.F bought.3FS which book 'Which student bought which book ?' (13 JAN 2011, SA 3a, offered freely)  119  b. OVS word order  ‫يا كتاب اشترت يا طالبة؟‬ ya:  kita:b iʃtarat  which book  ya:  Ta:liba  bought.3FS which student.F  'Which book did which student buy ?' (13 JAN 2011, SA 4a, offered freely)  4.3 Why genitive interrogatives are inherently D-linked I propose that it is the syntactic structure which causes genitive and D-linked interrogatives to pattern in the same way with respect to local extraction, long-distance extraction, and superiority. In particular, I suggest that the structural parallel between genitive and D-linked interrogatives lies in the fact that they both contain an overt domain restriction (§4.3.1). I then show how the derivation of the gap and the resumptive strategy proceeds with genitive interrogatives (§4.3.1 - 4.3.2).  4.3.1 D-linking arises if there is an overt domain restriction A comparison of the syntax of bare interrogatives, D-linked interrogatives and genitive interrogatives is given in (146).  (146) a. syntax of bare interrogatives [D wh [NØ] ]  the gap strategy  [D wh [φ hu [NØ]] ]  the resumptive strategy (only long-distance)  b. syntax of D-linked interrogatives [D wh [N N ]]  the gap strategy  [D wh [φ hu [N N ]]  resumption (local & long-distance)  120  c. syntax of genitive interrogatives [D [N N [D wh]]]  the gap strategy  [D [N N [D wh] [φ hu [N ]]] resumption (local & long-distance) There are two observations to note here. First, the structure of the genitive in Arabic and Hebrew is known as the construct state, where the head noun is left-adjacent to a noun phrase or interrogative pronoun (Borer 1999; Fehri 1988; Ritter 1988; Shlonksy 2004). Second, with both genitive and D-linked interrogatives there is an overt noun which specifies a domain restriction. It seems that the D-linked interrogative expressions and the genitive interrogative expressions are domain restricted because of the overt noun, whereas the bare interrogative expressions do not have this restriction since they do not have an overt noun. I speculate that the domain restriction is supplied by the overt noun that allows the resumption strategy to be used in a wider range of contexts that is possible with bare interrogatives. This suggests that there is a close connection between domain restriction and the presence of a resumptive pronoun, because resumption selects an element from a domain. The bare interrogative expressions do not have any such domain restriction and therefore resumption is allowed only for syntactic reasons, i.e. with long-distance extraction but not with local extraction. The claim made here is that genitive interrogatives are inherently D-linked. Therefore, they should have a D-N structure in content questions which employ the gap strategy and a D-φ-N structure in content questions which employ the resumptive strategy. With this in mind I now look at the derivation of content questions with genitive constructions in more detail.  4.3.2 Derivation of the genitive interrogative with the gap strategy Consider (147a) which is a genitive interrogative employing the gap strategy. The numeration is given in (147b).  121  (147) GENITIVE INTERROGATIVE CONSTRUCTION a. Genitive interrogative with gap  ‫نجوى رجل منو شافت ؟‬ Najwa riʤal minnu: ʃa:fit Najwa man who  saw:3SF  'Whose husband did Najwa see ?' (8 JUL 2010, SA 8a, offered freely) b. Numeration: {TOPØ, CØ, IØ, NajwaD, Sa:fitV, minnu:D, ridza:lN,} (148) gives the derivation of (147a). At the VP phase (148b), the DP is built by (148bI) merging the interrogative pronoun minnu: ‘who’ with the noun riʤal ‘husband/ man’; in accordance with Ritter’s (1991) N-to-D raising in construct states, N raises to SpecD via a successive application of Copy and Delete (148bII-III). Then the verb ʃa:fit 'she saw' merges with the DP riʤa:l minnu: ‘whose husband’ (148bIV). The subject DP Najwa merges with the V at SpecVP (148bV). At the IP phase (148c), the inflectional head merges with the VP (148cI), then the subject DP Najwa is moved to SpecIP via successive application of Copy and Delete (148cII-III). At the CP phase (148d), the DP riʤa:l minnu: ‘whose husband’ is moved to SpecCP via successive application of Copy and Delete (148dI-II). At the TopP phase (148e), the topical head Top merges with the CP (148eI) and the subject DP Najwa is moved to SpecTopP via successive application of Copy and Delete (148eII-III).  (148) derivational analysis of (147a) a. Numeration: {TOPØ, CØ, IØ, NajwaD, Sa:fitV, minnu:D, ridza:lN,}  b. VP phase I. Merge <D, N> [D [minnu:D ] [riʤa:lN ] ] II. Copy riʤa:lN and Merge <N, D> [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]] 122  III. Delete riʤa:lN [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]] IV. Merge <V, D> [V [Sa:fit V] [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]]] V. Merge <D, V> [V [NajwaD ] [V [Sa:fit V] [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]]] c. IP phase I. Merge <I, V> [I [V [NajwaD ] [V [Sa:fit V] [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]]]] II. Copy NajwaD & Merge <D, I> [I [NajwaD ] [I [V [NajwaD ] [V [Sa:fit V] [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]]]]] III. Delete NajwaD [I [NajwaD ] [I [V [NajwaD ] [V [Sa:fit V] [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]]]]] d. CP phase I. Copy [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]] & Merge <C, I> [C [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]] [I [NajwaD ] [I [V [NajwaD ] [V [Sa:fit V] [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]]]]] II. Delete [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]] [C [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]] [I [NajwaD ] [I [V [NajwaD ] [V [Sa:fit V] [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]]]]]  123  e. TopP phase I. Merge <Top, C> [Top [C [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]] [I [NajwaD ] [I [V [NajwaD ] [V [Sa:fit V] [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]]]]] II. Copy NajwaD & Merge <Top, C> [Top NajwaD [C [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]] [I [NajwaD ] [I [V [NajwaD ] [V [Sa:fit V] [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]]]]] III. Delete NajwaD [Top NajwaD [C [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]] [I [NajwaD ] [I [V [NajwaD ] [V [Sa:fit V] [D [riʤa:lN ] [D [minnu:D ] [riʤa:lN ]]]]]  4.3.3 Derivation of the genitive interrogative with the resumptive strategy Let us now look at the derivation of a genitive interrogative employing the resumptive strategy (149a). The numeration is given in (149b); notice that it contains the φ-element hu ‘him’. (149) GENITIVE INTERROGATIVE CONSTRUCTION a. Genitive interrogative with resumption  ‫إيمان رجل منو شافته بالمكتبة ؟‬ Iman  riʤa:l minnu: ʃa:fat=hu  Iman  man who  bi-l-maktaba  saw:3SF.3MS at-the-library  'Whose husband did Najwa see [him] at the library ?' (8 JUL 2010, SA 8b, offered freely) b. Numeration: {TOPØ, CØ, IØ, NajwaD, Sa:fitV, minnu:D, ridza:lN, huφ}  124  (150) gives the derivation of (149a). At the VP phase (150b), the DP is built by: (150bI) merging the pronoun hu ‘him’ with the noun riʤal 'husband/ man' and (1509bII) merging that complex syntactic object with the interrogative pronoun minnu: ‘who’. In keeping with the N-to-D raising of the construct state, N raises to SpecD via a successive application of Copy and Delete (150bIII-IV). Then the verb ʃa:fit 'she saw' merges with the DP riʤa:l hu minnu: ‘whose him husband’ (150bV) and cliticizaton of the pronoun hu ‘him’ immediately takes place via successive application of Copy and Delete (150bVI-VII). The subject DP Najwa merges with the V at SpecVP (150bVIII). At the IP phase (150c), the inflectional head merges with the VP (150cI), then the subject DP Najwa is moved to SpecIP via successive application of Copy and Delete (150cII-III). At the CP phase (150d), the DP riʤa:l him minnu: ‘whose husband’ is moved to SpecCP via successive application of Copy and Delete (150dI-II). At the TopP phase (150e), the topical head Top merges with the CP (150eI) and the subject DP Najwa is moved to SpecTopP via successive application of Copy and Delete (150eII-III).  (150) derivational analysis of (149a) a. Numeration: {TOPØ, CØ, IØ, NajwaD, Sa:fitV, minnu:D, ridza:lN, huφ}  b. VP phase I. Merge <φ, N> [φ [hu φ] [riʤa:l N ]] II. Merge <D, φ> [D [minnuD ] [huφ riʤa:lN ]] III. Copy riʤa:lN and Merge <N, D> [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]] IV. Delete riʤa:lN [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]  125  V. Merge <V, D> [V [Sa:fit V] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]] VI. Copy huφ & Merge <V, φ> [V [ʃa:fit V huφ] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]] VII. Delete huφ & Merge <V, φ> [V [ʃa:fit V huφ] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]] VIII. Merge <D, V> [V [NajwaD ] [V [ʃa:fit V huφ] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]] c. IP phase I. Merge <I, V> [I [V [NajwaD ] [V [ʃa:fit V huφ] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]]] II. Copy NajwaD & Merge <D, I> [I [NajwaD ] [I [V [NajwaD ] [V [ʃa:fit V huφ] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]]]] III. Delete NajwaD [I [NajwaD ] [I [V [NajwaD ] [V [ʃa:fit V huφ] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]]]] d. CP phase I. Copy [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]] & Merge <C, I> [C [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]] [I [NajwaD ] [I [V [NajwaD ] [V [ʃa:fit V huφ] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]]]]] II. Delete [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]] & Merge <C, I> [C [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]] [I [NajwaD ] [I [V [NajwaD ] [V [ʃa:fit V huφ] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]]]]]  126  e. TopP phase I. Merge <Top, C> [Top [C [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]] [I [NajwaD ] [I [V [NajwaD ] [V [ʃa:fit V huφ] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]]]]]] II. Copy NajwaD & Merge <Top, C> [Top NajwaD [C [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]] [I [NajwaD ] [I [V [NajwaD ] [V [ʃa:fit V huφ] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]]]]]] III. Delete NajwaD [Top NajwaD [C [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]] [I [NajwaD ] [I [V [NajwaD ] [V [ʃa:fit V huφ] [D [riʤa:lN ] [D [minnuD ] [huφ riʤa:lN ]]]]]]]  127  5 Conclusions 5.1 Results In this thesis I described the resumptive strategy and the gap strategy in Iraqi Arabic content questions and I showed that the choice of either the gap or the resumptive pronoun in the syntax gives rise to different interpretations in the semantics. The theoretical devices used in this thesis are not new; rather, I argued that various proposals applied to the resumptive strategies in other Arabic dialects apply to Iraqi Arabic, as well. In chapters 1 I gave some background on the morphology and syntax of Iraqi Arabic; I presented the morphology and syntax of interrogative expressions in Iraqi Arabic and I described the resumptive strategy in Iraqi Arabic content questions. In chapter 2, I presented a syntactic analysis of resumption in Iraqi Arabic content questions and I argued that the gap strategy is derived with full-DP deletion and the resumptive strategy with remnant-DP deletion. In chapter 3, I explored the interpretive differences associated with the gap and the resumptive strategy, and argued that the structural difference between [D-N] and [D-φ-N] correspond to the semantic difference between the (extensional) pair-list reading and the (intentional) natural function reading respectively. In chapter 4, I showed that genitive interrogatives in Iraqi Arabic are inherently D-linked in the sense of Pesetsky (1987).  5.2 Unsolved problems In the course of the thesis I touched on several questions that deserve further investigation. In chapter 3 I showed that in contexts where resumption is obligatory, it is semantically ambiguous between the natural function reading and the pair-list reading. A topic for further research is to give a more precise analysis of these two interpretations. In chapter 4 I showed that bare interrogatives are more restricted in their use of the resumptive strategy. Remains to be seen if the syntactic analysis proposed in this thesis can be extended to non-D-linked interrogative expressions. 128  It would also be interesting to investigate other environments where Iraqi Arabic uses resumptive strategies, such as relative clauses and dislocation structures. In addition, in this thesis, I have focused on weak resumptive pronouns; remaining to be investigated are other types of resumptive elements, such as strong pronouns and epithets. For example, in chapter 4 I showed that resumptive strategies in content questions in Iraqi Arabic do not use strong pronouns. It would be interesting to investigate whether this is a uniform phenomenon in Iraqi Arabic across all clause type environments (i.e. content questions, relative clauses, dislocation structures).  129  References Adger, David, and Gillian Ramchand. 2001. Phases and interpretability. In Proceedings of West Coast Conference on Formal Linguistics, ed. K Magerdoomian and L.A. Bar-El, 20:1-14. Somerville, MA: Cascadilla Press. Alkalesi, Yasin M. 2006. Modern Iraqi Arabic with MP3 files. Washingtom, DC: Georgetown University Press Adger, David and Gillian Ramchand. 2005. Merge and Move: Wh-dependencies revisited. Linguistic Inquiry 36:161-193. Aoun, Joseph, and Lina Choueiri. 1999. Modes of interrogation. Unpublished manuscript. University of Southern Carolina, Los Angeles. Aoun, Joseph, Lina Choueiri and Norbert Hornstein. 2001. Resumption, movement and derivational economy. Linguistic Inquiry. 32:371-403. Aoun, Joseph, and Audrey Li. 2003. Essays on the representational and derivational nature of grammar: the diversity of wh-constructions. Cambridge, MA: MIT Press Aüguero-Bautista, Calixto. 2001. Cyclicity and the scope of wh-phrases. Doctoral Dissertation, MIT Boeckx, Cedric. 2003. Islands and Chains: Resumption as Stranding. Amsterdam: John Benjamins. Borer, Hagit. 1999. Deconstructing the construct. In Beyond Principles and Parameters, ed. Johnson, K. and I. Roberts. Dordrecht: Kluwer publications.  130  Cardinaletti, Anna, and Michal Starke. 1999. The Typology of Structural Deficiency. In Clitics in the Languages of Europe, ed. H. van Riemsdijk, 145-233. Berlin: Mouton de Gruyter. Chierchia, Genaro. 1993. Questions with quantifiers. Natural Language Semantics 1:181- 234. The Netherlands: Kluwer Chomsky, Noam. 1973. Conditions on transformations. In A Festschrift for Morris Halle, ed. S. Anderson and P. Kiparsky, 232-286. New York: Holt, Reinhart & Winston Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: The MIT Press Chomsky, Noam. 1995. Bare phrase structure. In Evolution and revolution in linguistic Theory, ed. Hector Campos and Paula Kempchinsky. Washington DC: Georgetown University Press. Chomsky, Noam. 2000. Minimalist inquiries, the framework. In Step by step, Essays on Minimalist Syntax in Honor of Howard Lasnik, ed. Roger Martin, David Michaels and Juan Uriagereka, 89-156. Cambridge, MA: MIT Press. Chomsky, Noam. 2007. Approaching UG from below. In Interfaces + Recursion =Language?, ed. Uli Saurerland and Hans-Martin Gärtner, 1-29. Berlin: Mouton de Gruyter. Choueiri, Lina. 2003. Issues in the Syntax of Resumption: Restrictive Relatives in Lebanese Arabic. Doctoral dissertation, University of Southern California Déchaine, Rose-Marie and Martina Wiltschko. 2002. Decomposing pronouns. Linguistic Inquiry 33:409-442 Demirdache, Hamida. 1991. Resumptive Chains in Restrictive Relatives, Appositives and Dislocation Structures. Doctoral dissertation, MIT  131  Doron, Edit. 1982. On the Syntax and Semantics of Resumptive Pronouns. Texas Linguistics Forum 19:1-48 Elbourne, Paul. 2002. E-Type anaphora as NP-deletion. The Netherlands: Kluwer Academic Publishers. Evans, Gareth. 1980. Pronouns. Linguistic Inquiry 11: 337-362 Fassi, Fehri A. 1988. Case inflection, VS order and X'-Theory. In Proceedings of the 1st International Conference of the Linguistic Society of Morocco, ed. A. Fassi Fehri Guilliot, Nicolas. 2006. La reconstruction á l'interface entre syntaxe et sémantique. Doctoral Dissertation, Université de Nantes Guillot, Nicolas and Nouman Malkawi. 2006. When Resumption determines Reconstruction. In Proceedings of the 25th West Coast Conference on Formal Linguistics, ed. Baumer, Montero and Michael Scanlon, 25: 168-176. Somerville, MA: Cascadilla Press. Guilliot, Nicolas. 2008. To Reconstruct or Not to Reconstruct: That is the question. In Proceedings of the Workshop on What Syntax Feeds Semantics, published by FoLLI for ESSLLI Guilliot, Nicolas and Nouman Malkawi. 2009. When movement fails to reconstruct. In Merging Features, ed. by J. Brucart, A. Gavarro and J. Sola, 159-174. Oxford University Press Hornstein, Norbert, Jairo Nunes and Kleanthes K. Grohmann. 2006. Understanding Minimalism. Cambridge University Press Hornstein, Norbert and Jairo Nunes. 2008. Adjunction, labelling and bare phrase  132  structure. In Biolinguistics, 2.1: 57-86 Hamblin, Charles L. 1973. Questions in Montague English. Foundations of Language 10:41-53 Karttunen, Lauri. 1977. Syntax and semantics of questions. Linguistics and Philosophy 1:3-44. Kayne, Richard S. 1994. The antisymmetry of syntax. Cambridge, MA: MIT Press Kratzer, Angelika. 1998. Scope or pseudo-scope ?Are there wide scope indefinites ? In Events in Grammar, ed. S. Rothstein, 163-196 Lebeaux, David. 1990. Relative clauses, licensing and derivation. In Proceedings of NELS 20: 318-332 Malkawi, Nouman. 2009. Sur la Syntaxe de Quelques Expressions Anaphoriques: Epithétes et Pronoms Résomptifs. Doctoral Dissertation, Université de Nantes. McCloskey, James. 1990. Resumptive pronouns, Ā-binding and levels of representation in Irish. In The Syntax of the Modern Celtic Languages ed. Randall Hendrick, volume 23 of Syntax and Semantics, 199-248. San Diego: Academic Press. Pesetsky, David. 1987. Wh-in-situ: Movement and Unselective Binding. In The Representation of (In)definiteness, ed. Eric J. Reuland and Alice G. B. ter Meulen, 98-129. Cambridge, MA: MIT Press Pesetsky, David. 2000. Phrasal movement and its kin. Cambridge, MA: MIT Press Postal, Paul M. 1969. On so-called "pronouns" in English. In Modern Studies in English: Readings in Transformational Grammar. Ed. David A. Reibel &and Sanford A.  133  Schaine, 201-224. Englewood Cliffs, NJ: Prentice-Hall Inc. Reinhart, Tanya. 1983. Coreference and bound anaphora: a restatement of the anaphora questions. Linguistics and Philosophy 6:47-88. Ritter, Elizabeth Ann. 1988. A case study in the Syntax of Agreement: Hebrew Noun Phrases and Benoni Verb Phrase. Doctoral Dissertation, MIT Ross, John Robert. 1986. Infinite Syntax ! Norwood, NJ: Ablex Publishing Corporation Rouveret, Alain. 2002. How are Resumptive Pronouns Linked to the Periphery? In Linguistic Variation Yearbook, ed. Pierre Pica, 2:123-184. Rouveret, Alain. 2008. Phasal Agreement and Reconstruction. In Foundational Issues in Linguistic Theory, ed. Robert Freidin, Carlos P. Otero and Maria Luisa Zubizarreta, 167-195. Cambridge, MA: The MIT Press. Rouveret, Alain. 2011. Resumptive pronouns at the interfaces. Amsterdam: John Benjamins Sells, Peter. 1984. Syntax and Semantics of Resumptive Pronouns. Doctoral dissertation, University of Massachusetts at Amherst Sharvit, Yael. 1999. Resumptive Pronouns in Relative Clauses. Natural Language and Linguistic Theory 17:587-612 Shlonksy, Ur. 1992. Resumptive pronouns as a last resort. Linguistic Inquiry. 23:443-468 Shlonsky, Ur. 2004. Enclisis and proclisis. In The structure of CP and IP, ed. Luigi Rizzi, 329-353. New York: Oxford University Press.  134  Wahba, Wafaa. 1984. Wh-constructions is Egyptian Arabic. Doctoral Dissertation, University of Illinois at Urbana-Champaign Wojdak, Rachel. 2005. The linearization of affixes: evidence from Nuu-chah-nulth. Doctoral dissertation, University of British Columbia  135  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0105127/manifest

Comment

Related Items