<%BANNER%>

Wakhi Agreement Clitics

Permanent Link: http://ufdc.ufl.edu/UFE0042855/00001

Material Information

Title: Wakhi Agreement Clitics
Physical Description: 1 online resource (70 p.)
Language: english
Creator: HUGHES,TODD R
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: AGREEMENT -- CLITIC -- EUROPEAN -- INDO -- IRANIAN -- LINGUISTICS -- MARKER -- PAMIR -- PASHTO -- PAST -- POSITION -- SECOND -- SHUGNI -- SYNTAX -- TENSE -- WACKERNAGEL -- WAKHI
Linguistics -- Dissertations, Academic -- UF
Genre: Linguistics thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Wakhi past tense agreement clitics are classified as second position, or Wackernagel, clitics, similar to those in Pashto, Tagalog and Ancient Greek. Various approaches in the literature analyze these clitics through phonological, syntactic or hybrid perspectives. In some proposals the clitics are treated as arguments of the verb, while in others as agreement clitics affixed to an element other than the verb. Of particular interest are accounts of Pashto clitics, with which Wakhi has much in common. In this study Wakhi clitics are analyzed as subject agreement morphemes, rather than phonologically deficient pronominal arguments. The complementary distribution of non-clitic present tense agreement morphemes and the co-occurrence of full subject NPs with subject clitics lead to this conclusion. A proposal is made regarding a base generation site for the clitics. Based on that, certain phrase structure projections and syntactic movement operations are proposed to account for surface structure data.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by TODD R HUGHES.
Thesis: Thesis (M.A.)--University of Florida, 2011.
Local: Adviser: Henderson, Brent Mykel.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0042855:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042855/00001

Material Information

Title: Wakhi Agreement Clitics
Physical Description: 1 online resource (70 p.)
Language: english
Creator: HUGHES,TODD R
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: AGREEMENT -- CLITIC -- EUROPEAN -- INDO -- IRANIAN -- LINGUISTICS -- MARKER -- PAMIR -- PASHTO -- PAST -- POSITION -- SECOND -- SHUGNI -- SYNTAX -- TENSE -- WACKERNAGEL -- WAKHI
Linguistics -- Dissertations, Academic -- UF
Genre: Linguistics thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Wakhi past tense agreement clitics are classified as second position, or Wackernagel, clitics, similar to those in Pashto, Tagalog and Ancient Greek. Various approaches in the literature analyze these clitics through phonological, syntactic or hybrid perspectives. In some proposals the clitics are treated as arguments of the verb, while in others as agreement clitics affixed to an element other than the verb. Of particular interest are accounts of Pashto clitics, with which Wakhi has much in common. In this study Wakhi clitics are analyzed as subject agreement morphemes, rather than phonologically deficient pronominal arguments. The complementary distribution of non-clitic present tense agreement morphemes and the co-occurrence of full subject NPs with subject clitics lead to this conclusion. A proposal is made regarding a base generation site for the clitics. Based on that, certain phrase structure projections and syntactic movement operations are proposed to account for surface structure data.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by TODD R HUGHES.
Thesis: Thesis (M.A.)--University of Florida, 2011.
Local: Adviser: Henderson, Brent Mykel.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0042855:00001


This item has the following downloads:


Full Text

PAGE 1

1 WAKHI AGREEMENT CLITICS By TODD R. HUGHES A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS UNIVERSITY OF FLORIDA 2011

PAGE 2

2 2011 Todd R. Hughes

PAGE 3

3 To my wife

PAGE 4

4 ACKNOWLEDGMENTS Acknowledgements typically point out those without whom the work at hand would have be en impossible. With that in mind, I would like to recognize the following individuals. My wife, Sara h, allowed me to quit my job and take over her better gas mileage car for the daily trips between Jacksonville and Gainesville She listened to me as I recounted my good and less than good days as a graduate student. She was patient when I ignored her in order to do school work and reminded me to take a break when I needed it. Without her support, I could not have completed this thesis Drs. Brent Henderson and James Essegbey, my committee members, have modeled what good linguists look like. They have found a healthy mix between their role s as teachers, researchers and family members Individually, Dr. Henderson spent hours helping me hash out ideas and understand the data. His ability to edit my writing with a smile made the process much less stressf ul, and is much appreciated. Dr. Essegbey allowed me to present at a workshop he organized, which in turn helped narrow my focus for th e current project Their combined help completing grant proposals has been invaluable. Without their support, I could not have completed this thesis Habiba Begum, a Wakhi speaker living in the United States, spent hours answering questions, both in person and via e mail. She allowed me to pose often bizarre sentences, trying to tease apart the workings of her language. And through all of it, she was patient and gracious. Without her support, I could not have completed this thesis

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ............................. 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION TO CLITICS ................................ ................................ ................ 11 2 TYPOLOGICAL INTRODUCTION TO WAKHI ................................ ....................... 14 2.1 Basic Word Order ................................ ................................ ........................... 14 2.2 Morphology ................................ ................................ ................................ .... 16 2.2.1 Nominal Morphology ................................ ................................ ............ 16 2.2.1.1 Plurals ................................ ................................ ..................... 16 2.2.1.2 Deverbal nouns ................................ ................................ ....... 17 2.2.2 Verbal Morphology ................................ ................................ .............. 17 3 DETERMINING HOW SECOND POSITION CLITICS GET THERE ...................... 22 3.1 Better Clitic Placement through Phonology ................................ .................... 22 3.1.1 Anderson and Optimality Theory ................................ ......................... 22 3.1.1.1 Pure syntax ................................ ................................ ............. 23 3.1.1.2 Halpern and a little PF movement ................................ ........... 24 3.1.1.3 Anderson gets optimal ................................ ............................ 26 3.1.2 Pashto Second Position Clitics Meet Prosodic Structure ..................... 27 3.1.2.1 The stress of language ................................ ........................... 27 3.1.2.2 Halpern, again ................................ ................................ ........ 29 3.1.2.3 Prosodic structure ................................ ................................ ... 30 3.2 Better Clitic Placement through Syntax (With a Little Phonology) .................. 31 3.2.1 Clitics or Agreement Morphemes ................................ ........................ 32 3.2.1.1 Distribution of clitics ................................ ................................ 32 3.2.1.2 Coordination of clitics ................................ .............................. 34 3.2.2 Base Position of Clitics ................................ ................................ ........ 35 3.2.3 Just a Little Phonology ................................ ................................ ........ 36

PAGE 6

6 4 WAKHI CLITICS ................................ ................................ ................................ ..... 38 4.1 What Wakhi Clitics Are Not ................................ ................................ ............ 38 4.2 What Wakhi Clitics Are (The Nature of Wakhi Clitics) ................................ .... 39 4.2.1 What They Represent ................................ ................................ .......... 40 4.2.2 Where Clitics Attach ................................ ................................ ............ 40 4.2.2.1 Allowed hosts ................................ ................................ ......... 41 4.2.2.2 Disallowed hosts ................................ ................................ ..... 43 4.2.2.3 Disallowed environments ................................ ........................ 46 4.3 The True Nature of Wakhi Clitics ................................ ................................ ... 48 4.3.1 Arguments ................................ ................................ ........................... 49 4.3.2 Agreement Morphemes ................................ ................................ ....... 49 4.3.2.1 Complementary distribution ................................ .................... 49 4.3.2.2 Coordination ................................ ................................ ........... 50 4.3.2.3 Phonological form ................................ ................................ ... 51 4.3.2.4 Clitic doubling ................................ ................................ ......... 51 4.4 A Home for Wakhi Clitics ................................ ................................ ................ 52 5 FUTURE RESEARCH AND CONCLUSION ................................ ........................... 61 5.1 Dialectical Variation ................................ ................................ ........................ 61 5.2 Phrase Structure ................................ ................................ ............................ 62 5.3 Inanimates ................................ ................................ ................................ ...... 63 5.4 Conclusion ................................ ................................ ................................ ..... 63 APPENDIX WAKHI LANGUAGE MAPS ................................ ................................ ....... 64 REFERENCES ................................ ................................ ................................ .............. 67 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 70

PAGE 7

7 LIST OF TABLES Table page 2 1 Present tense agreement morpheme paradigm ................................ ................. 20 2 2 Past tense agreement morphem e paradigm ................................ ...................... 21

PAGE 8

8 LIST OF FIGURES Figure page A 1 Kyrgyzstan and Tajikistan language map ................................ ........................... 64 A 2 Afghanistan language map ................................ ................................ ................. 65 A 3 Northern Pakistan language map ................................ ................................ ....... 66

PAGE 9

9 LIST OF ABBREVIATIONS 1 first person 2 second person 3 third person ACC accusative AR T article COMP complementizer CL clitic DAT dative FEM feminine FUT future tense IMPF imperfective aspect INTR intransitive MASC masculine NEG negative NOMIN nominalizer OBJ object PAST past tense PERF perfective aspect PL plural PRES non past tense /presen t PRN pronoun SG singular TRANS transitive

PAGE 10

10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts WAKHI AGREEMENT CLITICS By Todd R. H ughes May 2011 Chair: Brent Henderson Major: Linguistics Wakhi past tense agreement clitics are classified as second position, or Wackernagel, clitics, similar to those in Pashto, Tagalog and Ancient Greek. Various approaches in the literature analyze these clitics through phonological, syntactic or hybrid perspectives. In some proposals the clitics are treated as arguments of the verb, while in others as agreement clitics affixed to an element other than the verb. Of particular interest are accounts of Pashto clitics with which Wakhi h as much in common. In this study Wakhi clitics are analyzed as subject agreement morphemes, rather than phonologically deficient pronominal arguments. The complementary distribution of non clitic present tense agreemen t morphemes and the co occurrence of full subject NPs with subject clitics le a d to this conclusion. A proposal i s made regarding a base generation site for the clitics. Based on that, certain phrase structure projections and syntactic movement operations are proposed to account for surface structure data.

PAGE 11

11 CHAPTER 1 INTRODUCTION TO CLITICS According to Payne (1997:22) phrasal or clausal level, but which binds phonologically to some other word, known as The se morphemes are syntactically independent, but must be bound to a phonological host. For example, the F rench clitic me in 1 is not an affix derivation al or inflectional. (1) Elle me regarde 3SG. PRN 1SG.OBJ.CL look.at.PRES is looking at It is syntactically contentful but phonologically deficient, requiring a phonol ogical host One could not, for example, during a night out in a Parisian nightclub answer me when asked Elle regarde qui ? is she looking at That would be ungrammatical ; it has no host ( and might generate a strange look in reaction ) Rather one would answer moi The nature of clitics then, is problematic for linguists trying to determine their role in the grammar This split identity has inspired res earch going back decades which focus es on the Phonology Morphology Syntax interface ( Richardson, Marks, and Chukerman 1983) Among these recent research topics i s the study of s econd position ( 2P ) clitics also known as Wackernagel clitics (Wackernagel 1892; C ysouw 2004) a phenomenon in which the clitic attaches to either the first phonological word of a clause or the first clausal constituent regardless of its hosts lexical or syntactic properties. Occurring in a wide variety of languages, including Serbo Croatian, Pashto, Tagalog and Warlpiri, t hey

PAGE 12

12 have received much attention in the literature (Zwicky 1977; Anderson 1993; 1996; Serbo Croatian is one of the most intensely studied languages for the properties of 2P clitics. Its clitics, such as those in the cluster smo mu je in 2 always appear after the first phonological word in their clause whether a full pronoun as in 2 a, a n interrogative in 2 b, or a complementizer in 2 c d (2) a. Mi smo mu je predstavili ju e. we are him.DAT her.ACC introduced yes terday b. Za to smo mu je predstavili ju e. why are him.DAT her.ACC introduced yesterday c. Ona tvrdi da smo mu je mi predstavili ju e. she claims that are him.DAT her.ACC we introduced yesterday d. Predstavili smo mu je ju e. introduced are him.DAT her.ACC yesterday 9) Most recent analyses follow a variation of one of two paths: phonological or syntactic placement. (2001) further subdivides these two camps by how strong or weak their adherence is to the chose n approach. A strong syntactic approach, for example, w ould work to avoid any movement at PF, while a weak approach might allow for highly constrained movement after spell out for the sake of phonologically

PAGE 13

13 well formed sentences. Strong and weak phonological approaches would allow, rather than restrict, various amounts of movement in the phonological component of grammar. This study will analyze pronominal clitics in Wakhi a southeastern Iranian language native to 32,000 speakers in the Pamir Mountains of Pakistan, Afghanistan, Tajikistan and China (Lewis 2009) 1 While t hese clitics have been described as second position (Bashir 2009:835; Backstrom 2009:29) the empirical eviden ce req uires a different analysis. Chapter 2 will offer a basic typology of Wakhi syntax, followed by an overview of previous second position clitic research in Chapter 3. In Chapter 4 I will propose a new analysis for Wakhi agreement clitics that rely pr imarily on syntactic operations but also allow constrained phonological operations a weak syntactic approach in C onclusions and future research possibilities are discussed in the Chapter 5 1 For a more detailed view of where Wakhi is spoken see maps in Appendix A.

PAGE 14

14 CHAPTER 2 TYPOLOGICAL INTRODUC TION TO WAKHI 2.1 Basic Word Order Wakhi clauses are overwhelmingly head final. This is demonstrated most concretely i n basic declarative sentences like 3 which are SOV in both present and past tenses (3) a. saka yi parinda winetu. 1PL. PRN the bird see.PAST.PERF W e saw the b. y ezi yi ghar yav dietu. yesterday the rock 3 PL. PRN hit.PAST.PERF c. Rehberesh parinda wind. R ehber bird see.PRES a 2 While genetically and geographically proximate languages exhibit scrambling tendencies (Toosarvandani 2007:3) and therefore various word orders, Wakhi shows little surface evidence of this. Both matrix and embedded clauses show th e same basic word order (4) maj k eshen ki sav a shapik yikt 1SG.PRN hear .PAST.IMPF COMP 2 PL .PRN bread eat.PAST.IMPF 2 Unless cited otherwise, all Wakhi examples are from personal communication with Habiba Begum, a Wakhi speaker. Th e transcriptions are mine.

PAGE 15

15 The language is strongly verb final, and as such it would be expected that other constructions are also head final. This is not the case, however, with nouns and adjectives. The expected order is adjective noun which is attested in Wakhi as in 5 a However, t he reverse is also attested as in 5 b 3 (5) a. y ezi maja raftch baf shapik xetu yesterday 1SG. PRN very fine bread make.PAST.PE RF b. y ezi maja shapik raftch baf xetu yesterday 1SG. PRN bread very fine make.PAST.PERF a Many verb al forms in Wakhi (all but about 300, according to Bashir (2009:833) ) (6) yava yark xetu they work make.PAST.PERF (7) ya yix hob vite the ice melted become.PAST.PERF As with the above examples, the structure is verb final. This is the overwhelming tendency in Wakhi. In fact, only one structure has been found which places anything after the verb T his will be discussed in Chapter 4. 3 Lorimer (1958a:para. 202) claims that adjectives always precede their NP except the quantifier t

PAGE 16

16 2.2 Morphology This section will examine nominal and verbal morphology, which is central to a study on clitics. Before describing each of those areas, though, it is important to note that Wakhi does not hav e any gender or noun class di stinction. 2.2.1 Nominal M orphology 2.2.1.1 Plurals Plurals are created by adding one of two suffixes to the noun stem. Plural subjects of both transitive and intransitive verbs take isht (8) a. parind asht shapik go en bird PL bread make 3PL b. ya kash isht hob= ov vite the boy PL melted=3PL become.PAST.PERF Plural objects take the other plural marker, ov (9) yava parind av 4 winetu they bird PL see.PAST.PERF There is no difference between animate and inanimate plurals for number agreement. (10) kitob isht tsa mezen raben pervetu book PL from table down fall.PAST.PERF 4 When a stem is vowel final and a suffix vowel initial, one vowel deletes. The remaining vowel has the qualities of the stem vowel.

PAGE 17

17 2.2.1.2 Deverbal nouns Deverbal nouns are for med by adding the nominaliz er kuzg Note the verbal form in 6 above, repeated here as 11 (11) yava yark xetu they work make.PAST.PERF The addition of the nominalizer following the verb changes it to a noun. (12) yark xek kuzg shapik yitu work make.PERF NOMIN bread eat.PAST.PERF 2.2.2 Verbal Morphology The discussion of Wakhi verbal morphology will center around present and past tense agreement morphemes. 5 Present tense forms are made from the present base 6 The present base is the second person singular form which is unmarked To the base is added a subject agreement morpheme which encodes number (singular or plural) and person (first, second or third) ( Table 2 1 ). For example, the verb in the sentence below has the third person plural agreement morpheme, en 5 In most of the literature (Lorimer 1958a; Bashir 1986; Backstrom 2009) the past tense agreement morphemes are referred to throughout Chapters 3 and 4. 6 Past tense bases, while not central to a discussion of Wakhi clitics, are usually formed from the third person singular present form.

PAGE 18

18 (13) parindash shapik go en birds bread make.IMPF 3PL Past tense agreement morphemes have a similar paradigm, and even show some syncretism with the present tense ( Table 2 2 ) The past tense agreement morphemes do not typically suffix to the verb. Instead they suffix to other constituents in the sentence, as in 14 (14) yezi kuie parind= av wine tu 7 yesterday someone bird=3PL see.PAST.PERF Yesterday s Only one clitic is allowed per clause. Compare 15 with 14 W hen I proposed 15 to the consultant, she said it sounded (appropriately) like someone who was trying to learn the language. (15) *yez= iv kuie parind= av winetu yesterday=3PL someone bird =3PL see.PAST.PERF Complex verbal structures, like those in 6 and 7 allow clitics to affix to the nominal element. Since clitics only attach to phrasal constituents, the nominal element is a constituent separate from the verb, rather than part of a phrasal verb. (16) yezi yark=ov xetu yes ter day work=3PL make.PAST.PERF 7 All following Examples from other authors will use original conventions.

PAGE 19

19 Sentences with embedded clauses do not all ow clitics to move outside their clause. (17) a. maj keshen [ ki tou shapik yikt ] 1SG.PRN hear.PAST.IMPF COMP 2 SG .PRN bread eat.PAST.IMPF b. keshen= o m [ ki tou shapik yikt ] hear.PAST.IMPF=1SG COMP 2 SG .PRN bread eat.PAST.PERF c. maj keshen [ ki shaipk= et yikt ] 1SG.PRN hear.PAST.IMPF COMP bread=2SG eat.PAST.PERF d. keshen= o m [ ki shapik= et yikt ] hear.PAST.IMPF=1SG COMP bread =2SG eat.PAST.PERF e yez=im keshen ki shapik=et yikt yesterday=1SG hear.PAST.IMPF COMP bread=2SG eat.PAST.PERF f keshen= o m ki yez=it shapik yikt hear.PAST .IMPF=1SG that yesterday=2SG bread eat.PAST.PERF

PAGE 20

20 g yez=it maj keshen yesterday=2SG 1SG.PRN hear.PAST.IMPF ki shapik yikt COMP br ead eat.PAST.PERF h *yezi keshen=om ki shapik=it yikt yesterday hear=1SG COMP bread =2SG hear.PAST.PERF The ungrammatical sentences above, 17 g and 17 h demonstrate the limitations of clitic placement. In 17 g =it from the lower clause appearing in the upper clause. The ungrammaticality of 17 h is due to the affixation of the clitic to the verb. I n 17 b and 17 d the structure was licit because there was no preceding host, requiring PI to suffix it to the first phonol ogical word, in this case, the verb. I t is illicit in 17h because there is a preceding phonological host within the clause yezi 8 The typological details given in this chapter are sufficient for the analysis offered in Chapter 4. Before that proposal, however, we will look at other treatments of clitics in the literature to gain insight into the nature of Wakhi clitics. Table 2 1. Present tense agreement morpheme paradigm person sg pl 1 em en 2 it 3 t/ d en 8 The exact syntactic structure will be outlined more clearly in Chapter 4. At that point the relations hip of yezi

PAGE 21

21 Table 2 2. Past ten se agreement morpheme paradigm person sg pl 1 em en 2 et ev 3 t/ d/ ev

PAGE 22

22 CHAPTER 3 DETERMINING HOW SECOND POSITION CLITICS GET THERE The two broadly competing views of placement for 2P clitics, syntactic or phonological, are rarely held to dogma tically. Most researchers acknowledge that, empirically, neither explanation is sufficient by itself One recent notable exception to (2008) investigation of Warlpiri 2P clitics. In her paper, she maintains that Warlpiri clitics can be explained almost exclusively by appealing to syntactic movement with only minimal dabbling in morphological functions to produce licit surface order. Importantly, though, she discounts the possibility of movement at PF. As view is, she makes no claims that her analysis is appropriate for other languages. Fortunately for a study of Wakhi clitics, an examination of both sides of th e argument can provide hints that help work out their nature and p lacement. In this chapter we will look at clitics through a mostly phonological lens based on Anderson (1996) and Robert s (1997) Then, t o provide the syntactic perspective, we will turn again to Roberts whose 200 0 dissertation on Pashto r everses his earlier claims 3.1 Better Clitic Placement through Phonology 3.1.1 Anderson and Optimality Theory Deviating slightly from his a morphous framework of morphology, Anderson (1992) suggests that 2P clitic placement is managed by the ranking of specific Optimality Theoretic constraints (as opposed to the Word Formation Rules of his earlier work). Before looking at the constrain t system, we must first understand his criticism of a purely syntactic perspective and (1992) hybrid treatment of clitics, o n which (perhaps against which) Anderson leans heavily for his argument.

PAGE 23

23 3.1.1.1 Pure syntax To account for 2P clitic placement with syntax alone requires one of two operations: either the clitic is generated in first position and something is then moved to its left, or the clitic is a head and a phrase is then generated to its left, in the specifier position. Both of these can produce a structure in which the clitic occupies the second syntactic position. By this account, then, whatever appears before t something that can be moved, adjoined, or base generated as a single syntactic (Anderson 1996:17 0) This observation is purely syntactic points of view. from (1995) 18 shows the different firsts that clitics ( in 18 ) can follow in Serbo Croatian (SC). (18) a. Moja mladja sestra u utorak my younger sister FU T come on Tuesday b. Moja mladja sestra u utorak my FUT younger sister come on Tuesday (Anderson 1996:173) The first sentence seems compatible with the idea that whatever precedes the clitic must be a syntactic constituent. In this case moja mladja sestra clearly a well formed DP. The problem comes in 18 b, where follows the first word of the same DP, but with no corresponding change in semantic content. The only response that maintains the syntactic interpretation of this phenomenon is that moja is itself a constituent.

PAGE 24

24 Anderson counters with data from Browne (1975) in which a clitic ( je two parts of a phrase that could not be considered phrases separately. (19) a. Lav Tol stoi je veliki ruski pisac Leo Tolstoi is great Russian writer b Lav je Tolstoi veliki ruski pisac Leo is Tolstoi great Russian writer (Anderson 1996:174) This cannot be satisfactorily explained using only the syntactic tools of base generation and movement. 3.1.1.2 Halpern and a little PF movement The above conclu sion provides the basis for moving away from advocating only syntax to acknowledging some interaction with the phonology, an innovation advocated by Halpern (1992) For the most part, placement of clitics occurs in the syntax module of the grammar. Specifically they are generated or adjoined to the first position of their clause (IP) logically sub (Anderson 1996:175) In cases when clitics are base generated to the left of this position, l anguages then have one of two rules which can apply to satisfy the sub categorization The first simply moves a constituent to a position preceding the clitic The clitic is then in second position by means of a syntactic o peration and all is well. (1992:17) Prosodic Inver sion moves the clitic only as much as is necessary for phonological well formedness, specifically one word or one phonologically well formed unit to the right from its initial position in the clause. Languages which require clitics to follow entire phrase s would

PAGE 25

25 appeal to the first rule but lack the second. Conversely, those which define second position as following only the first word would lack the first rule, but employ the second. Anderson again draws data from Brown e (1975) to undermine Halpern and PI. As mentioned above, a language which fronts entire phrases above the clitic should be precluded from also moving the clitic by prosodic inversion. The sentences in 20 however, are examples of both rules applying. (20) a. Sovjetske goste je primio i predsjednik Republike Soviet guests PAST received also president republic Austrij e Jonas Austria Jonas b. Sovjetske je goste primio i predsjednik Republike Soviet PAST guests received also president republic Austrije Jonas Austr ia Jonas Soviet guests. (Anderson 1996:178) (Serbo Cr oatian) The direct object Sovjetske goste s canonical surface position but has been fronted, fulfilling the phonological requirements of the clitic. It should not be the case that the clitic could also be moved by PI after a ph rase was fronted. Anderson leaves this as a challenge to proponents of PI.

PAGE 26

26 3.1.1.3 Anderson gets optimal In response to the perceived inadequacies of these other possibilities, Anderson posits a solution that takes part of his a morphous morphology and me rges it with the tools of Optimality Theory. Word Formation Rules at work in PF can put the phonological form of the clitics in their proper place when the first position (clitic host) is occupied by a phonological word. Otherwise the syntactic rules can move a syntactic phrase into first position. All sentences, therefore, must be interpreted either phonologically or syntactically. This is a weakness for languages similar to SC represented in 20 where according to this fr amework, both phonological insertion and syntactic movement is occurring If a sentence must be uniformly interpreted either phonologically or syntactically, then the sentences in 20 should not be possible. Anderson (1996) us e s ranked constraints to derive the positions of 2P clitics. The morphous morphology. The most basic constraints at work with all 2P clitics are E DGE M OST (cl i L ) and N ON I NITIAL (cl i ) These expres s a preference for candidates with clitics at the far left edge of their clause without being in first position, assuming E DGE M OST (cl i L ) outranks N ON I NITIAL (cl i ). To account for the relative position of the clitics with respect to words and clauses ano ther constraint s I NTEGRITY (XP) must be considered. Th is constraint express es the preference that nothing from o utside the XP can be inserted. For clitics that must follow a phrase, I NTEGRITY (XP) would dominate E DGE M OST (cl i L ) Clitics that occur after the first word would rank E DGE M OST (cl i L ) more highly than I NTEGRITY (XP). In all cases N ON I NITIAL (cl i ) is undominated due to the requirement that clitics have a host.

PAGE 27

27 In response to the issue proposed with the data in 2 0 A nderson suggests that the relative ranking of E DGE M OST (cl i L ) and I NTEGRITY (XP) in some languages (including this SC dialect) is optional. Either constraint can dominate the other. This analysis is interesting due to its application of phonological tools to syntactic processes. That this application has not been widely extended to other areas within syntax brings doubt as to its universality. While the constraints were derived from similar processes in word formation (specifically addressing infixes wit h the constraint I NTEGRITY (W ORD ) ), their response to the problem of clitics that occur in environments with both phrasal fronting and PI is less than fully satisfactory. Anderson leans heavily on optionality in OT to explain what is essentially a syntacti c phenomenon. Generativist syntactic frameworks a re less amenable to optionality than is OT. 3.1.2 Pashto Second Position Clitics Meet Prosodic Structure Roberts (1997) propose s a much stronger phonological analysis of 2P clitics He does this using Pash to data specifically the interaction between stress patterns and clitics, and comparisons between Pashto and SC clitic structures. Applying an Optimality Theoretic framework to phonological issues makes for a much stronger argument 3.1.2.1 The stress of l anguage Pashto clitics are strongly second position. The sentences in 21 taken from Tegey (1977) clearly bear this out. The mo dal de occurs second position (21) a. tor de n n xar n r wali Tor should today donkey not bring

PAGE 28

28 b. n n de xar n r wali today should donkey not bring c. xar de n r wali donkey should not bring (Roberts 1997:371) This pattern continues until only the verb and clitic remain in V O order There is some variation, however. In the above examples, the clitic seems to attach to the first word in the clause. The following sentences in 22 however, show clitics following first constituent, rather than t he first word, as seen in some of the SC examples above. (22) a. [ NP a el kal na danga aw kh ysta pe la] me n n by w lida that 20 year tall and pretty girl I today again saw year old tall and pretty girl again today. b. [ NP xu al aw patang] ba ye d r ta r w i Khosal and Patang will it you to bring (Roberts 1997:372) An appeal to syntactic constituents, however, is empirically insufficient. Pashto clitics are also sensitive to stress. In 23 an affix precedes the verb. The main word stress can optional ly fall on the penultimate verb root syllable or the prefix. Note the change in clitic placement.

PAGE 29

29 (23) a. a xist l me PREFIX buy I b. me xist l PREFIX I buy (Roberts 1997:374) While the clitics can intervene between a prefix and root, they cannot come between syllables of a root, in this case a verb composed of a single mor pheme in 24 (24) a. me beat I b. me beat I (Roberts 1997:374) It is apparent that Pashto 2P clitics are sensitive to phonological, spe cifically prosodic, elements. They are also somewhat sensitive to syntactic structures, as briefly outlined below. 3.1.2.2 Halpern, again Again referring to Halpe rn Roberts focuses on SC subordinate clauses, in which clitics attach to the complementizer introducing the clause. Pashto clitics may not do that. Recalling 19 above, SC clitics can attach either to the first phonological word or the first constituent of the clause, depending on whether the sentence invokes a synta ctic or phonological (with PI) ordering of the clitics Pashto, however, does not hav e this option. Since Pashto clitics may attach neither to complementizers nor to the

PAGE 30

30 first prosodic word, but rather to the first constituent, pure synta x and PI are una ble to account for the pattern. Roberts suggests that rather than appealing to syntactic operations for clitic placement, a process he contends occurs primarily for phonological purposes, the tools of phonology should be used. 3.1.2.3 Prosodic structur e Having demonstrated both prosodic and syntactic sensitivity, Roberts suggests that Pashto 2P clitic placement is entirely phonological. The syntactic structure is mapped to a prosodic structure as in 25 (25) a. [ NP x o al aw patang] ba ye d r ta r w i Khosal and Patang will it you to bring b. [ IP [ NP x o al aw patang] [ VP [ PP d r ta] r w i ] ] Khosal and Patang you to bring c. xo al ) aw patang )) d r ta ) r w i )) Khosal and Patang you to bring (Roberts 1997:384) 9 The prosodic structure then provides the skeleton on which clitics can be attached. Three constraints determine the attachment location. (26) a. ALIGN (cl,L,PPh,R) aligns the left edge of the clitic to the right edge of a phonological phrase. b. EDGEMOST (cl,L) ensures the clitics placement to the far left o f the clause. 9 (33) below.

PAGE 31

31 c. NON INITIAL (cl), an undominated constraint, precludes a clitic from occupying first position (Roberts 1997:385 386) In th e case of 25 the constituent prosodic phrase allowing the clitic to follow. Note that patang are considered to be co occurring for the purposes of prosodic mapping, since they are coordinated DPs. That is, the two a re treated as one prosodic unit, ignoring Otherwise, the clitic could follow which would be ungrammatical. One weakness of this proposal relates to 23 above, repeated here as 27 (27) a. a me buy I b. me xist l PREFIX I buy (Roberts 1997:374) To ac count for the grammaticality of 26 b, Roberts stipulates that Pas h to prefixes can map directly to full prosodic phrases when they bear stress. 3.2 Better Clitic Placement through Syntax (W ith a Little Phonology) Because of the above mentioned weakness inherent in the phonological approach (Roberts 1997), Roberts (2000) re evaluate s his proposal. Rather than advocating an almost totally phonological analysis he moves to the other extreme, advocating an al most exclusively syntactic approach. The change brings with it two innovations not directly addressed by other linguist s First, he considers clitics not as arguments but

PAGE 32

32 agreement morphemes that do not move, depending on Pashto scramblin g for word order effects. The second innovation is discussion of the base position of clitics 3.2.1 Clitics or Agreement Morphemes The evidence for Pashto 2P clitics as agreement morphemes rather than a phonologically emaciated argument is strong. I pr esent two components of Roberts rationale which are salient to our current discussion : distribution and coordination of clitics. 3.2.1.1 Distribution of clitics Pashto verbs bear agreement morphemes that agree with subject in the present tense as in 28 a and object in the past tense as in 28 b. These agreement markers occur whether the corresponding DP arguments are full NPs or pronouns as in 28 c (28) a ahmad ghwa lwesh i Ahmad(MASC) cow(FEM.SG) milk 3PRES b. ahmad ghwa lwash el a Ahmad(MASC) cow(FEM.SG) milk PAST FEM3SG c. (ze) hara wrez pe baagh kee gerz em (PRN1SG) every day at garden in walk 1SG (Roberts 2000:96) The overt pronoun ze 27 c is optional because the verbal a greement morpheme licenses pro drop. When the pronoun is present, however, it is located in the same syntactic position as a full NP. Specifically the pronoun is not found in second position, rather it is found where a clitic could never occur, initially According to Roberts

PAGE 33

33 (2000:98) i f clitics were pronouns (arguments), one would expect them to be positioned like pronouns, which they are not. Pronouns also occur with agreement morphology. On e would likewise expect clitics as pronouns to appear with verbal agreement morphemes, but they do not, as seen below in 29 Clitics, like mee yee occur with verbal agreement morphemes indicating the same features. (29) a. gad eed em (* mee ) dance INTR 1SG(PAST.I MPF) 1SG b. khkol ew I mee (* yee ) kiss TRANS 3SG(PRES.IMPF) 1SG 3SG c. ahmad (* mee ) khkol ew em Ahmad 1SG kiss TRANS 1SG(PAST.IMPF) d. ahmad (* yee ) gad ig i ahmad 3SG danc e INTR 3SG(PRES.IMPF) (Roberts 2000:97) We see from the sentences in 29 that w here verbal agreement morphology exists, the corresponding pronominal clitic is ungrammatical, whether subject in present tense or object in past tense. They are in complementary distribution. Agreement suffixes and pronominal clitics also serve the same function. Both mark the argument, either internal or external, for person and number.

PAGE 34

34 complementary distribution between verbal agreement suffixes and 2P pronominal clitics is evidence that both kinds of morp hology serve to identify pro (2000:98) To put it another way, the complementary distribution and parall el function of pronominal clitics and verbal agreement markers are evidence that the pronominal clitics are agreement morphemes. 3.2.1.2 Coordination of clitics Considering additional consequences of the claim that pronominal clitics are pronouns leads to coordination of clitics. Full pronouns can coordinate as in 30 while clitics cannot coordinate as in 31 either with themselves or with DPs. (30) parun taa aw maa kitab olwelewu yesterday PRN2SG and PRN1SG book read (Roberts 2000:103) (31) a. *parun dee aw mee kitab olwelewu yest erday 2SG and 1SG book read b. *parun Aman aw mee kitab olwelewu yesterday Aman and 1SG book read (Roberts 2000:104 5) Clitics considered as full arguments should be able to coordinate, just as other arguments do. Agreement markers are not expected to coordinate. The sentences above seem t o demonstrate Pashto clitics having more in common with agreement markers tha n full argument pronouns.

PAGE 35

35 3.2.2 Base Position of Clitics Having derived the Pashto clause structure in 32 and assuming agreement marker nature of clit ics detailed above, Roberts places clitics as the head of a projection below TP, but above AspP. This is based on empirical evidence like the sentences below as in 33 (32) [ TP AspP NegP v P EXT ARGUMENT [ VP INT ARGUMENTS (33) tor mee we lid e Tor 1SG PERF see MASC3SG [ CP magar [ TP spin mee we ne lid e ] ] but Spin 1SG PERF NEG see MASC3SG (Roberts 2000:81; from Tegey 1977:127) Assuming the perfective morpheme we f an aspectual between the head of AspP and Tor in spec,TP indicates that it s projection must also be between TP and AspP. Pashto 2P clitics are heads of an agreement projection which Sportiche (1996) calls CliticP. Roberts adopts the more general AgrP, Agreement Phrase, believing that clitics do not actually form a class or group. Differen t clitics, even within Pashto, play different roles, and in this case, they are serving as agreement morphemes. Assuming the AgrP is found below TP and above AspP, we can derive for Pashto the clause structure in 34 What foll ows from this is a natural second position structure for clitics. (34) [ TP AgrP AspP NegP v P EXT ARGUMENT [ VP INT ARGUMENTS There is no overt subject in 33 leaving spec,TP empty. In order to satisfy EPP, the o bject moves to spec,TP. The clitic is now just below a DP that can serve as its

PAGE 36

36 phonological host. While AgrP is not a syntactic spot designed for clitics, the role these clitics play put them in the perfect location to nearly always be second position. This is derived through purely syntactic means. One note before looking at the small role of phonology in this structure: On the surface the clitic is interpreted as the external argument. That is not the case, however, since its base position is the he ad of AgrP It has not moved up from spec, v P In reality the external argument is the phonologically null pro which is base generated in spec, v P, and moves to spec,AgrP to be interpreted. The clitic Agrees with pro making it appear to be the overt sub ject. 3.2.3 Just a Little Phonology The only phonology Roberts allows is quite familiar by now: Prosodic Inversion by Halpern. Roberts makes a strong case for a syntactic account of 2P clitics/agreement markers, but realizes that in rare cases syntax mig ht output a derivation that converges at LF, but not PF. pro drop makes this a distinct possibility. A simple sentence with only subject pronoun and verb could create a situation in which, after pro drop, only clitic an d verb remain, in that order. The clitic would have no phonological host, therefore motivating the call for PI (Roberts 2000:78) Roberts is explicit that this and other phonological constraints that apply at PF only do so as a last resort although he does not specify the circumstances under which this might be necessary. After PI, the derivation that would converge at LF would also converge at PF. In this chapter we have seen that although Anderson and Roberts make a compelling case for a phonological analysis of this phenomenon, the syntactic approach is superior with respect to Pashto, a language closely related to Wakhi. Wh ile both

PAGE 37

37 Serbo Croatian and Pashto have strong 2P tendencies, t he constraints advocated for a phonological analysis ultimately answer a syntactic question by ignoring what are syntactic processes in Pashto. As will be seen in Chapter 4, Wakhi clitic place ment is also syntactic in nature. Considering the above discussions, and applying their ideas to Wakhi data, this project will assume a syntactic approach the best option for Wakhi clitics. We will see in C hapter 4 why that choice is made for this langua ge.

PAGE 38

38 CHAPTER 4 WAKHI CLITICS There is much in the literature about clitics and we have looked closely at some analyses of 2P clitics in other languages. In this chapter we will examine pronominal clitics in Wakhi. The first task will be to make clear wh at these clitics are not Once completed, we will take a detailed look at what Wakhi clitics are, giving attention to a comparison with Pashto presented above. Finally an attempt will be made to establish where the clitics are generated. 4.1 What Wakhi Clitics Are Not Wakhi pronominal clitics are not 2P clitics. While there are some Pamiri languages which do have Wackernagel clitics, including Shugni, a dominant language near the Wakhan (Barie 2009) the position of Wakhi clitics in relation to other cl ausal constituents is less restrictive. (35) a. yez= im shapik raftch baf xetu yesterday=1SG bread very fine make.PAST.PERF b. yezi shapik= om ra f tch baf xetu yesterday bread=1SG very fine make.PAST.PERF c yezi shapik raftch baf= om xe tu yesterday bread very fine=1SG make.PAST.PERF The sentences in 35 show the various hosts which can take the pronominal clitic. In 34 a the temporal adverb yezi host, while shapik the AP raftch baf 34 b and 34 c respectively. Under a 2P analysis, the clitics attachment to yezi would not be a surprise, and a case could be

PAGE 39

39 made for shapik as the host, discounting the tempo ral adverb. For om to cliticize onto raftch baf makes a Wackernagel analysis untenable. While most examples in this paper give the past tense verb in perfective form, the patterns of agreement clitic placement hold for the imperfective as well in 36 (36) a. yez= im shapik raftch baf goxt yesterday=1SG bread very fine make.PAST.IMPF b. yezi shapik= om raftch baf goxt yesterday bread=1SG very fine make.PAST.IMPF c. yezi shapik raftch baf= om goxt yesterda y bread very fine=1SG make.PAST.IMPF Some linguists treat Wakhi clitics as 2P with idiosyncracies. For example, Bashir most (2009:835) Statements like this put Wakhi clitics in the discussion with 2P clitics, but take no firm stand. Erschler, during a recent presentation, boldly proposed (AWC), a class of clitics that often appear after the first word or constituent, but can also appear la ter in the sentence (Erschler 2010:3) By putting aside the intuition that Wakhi clitics are 2P, the discussion and analysis of this phenomen on can be that much freer. 4.2 What Wakhi Clitics Are (The Nature of Wakhi Clitics) While it is true that Wakhi clitics are not 2P, there is much from the general discussion of 2P clitics that can help in determining thei r nature in this language.

PAGE 40

40 4.2.1 Wh at They Represent As mentioned in Chapter 2, Wakhi clitics are only grammatical in past tense. While the present and past tense agreement paradigms exhibit some syncretism, the y act differently. P resent tense markers must suffix to the verb, but past ten se clitics must not attach to the verb, although they can suffix to other elements in the clause. (37) fruitstones break 1SG (Lorimer 1958a:no. 9) (38) yez= iv yava yark kert yesterday=3PL 3PL.PRN work do.PAST.IMPF Wakhi clitics, unlike Pashto clitics, only express features of the subject, never the object. In 39 the first person singular clitic =im represents t he pro dropped subject. In 40 the subject is maja =iv ya kashve (39) yez= im ya kashve tse vinetu yesterday=1SG the boys COMP see.PAST.PERF (40) yez= iv maja ya kashve tse vinetu yesterday=3PL 1SG.PRN the boys COMP see.PAST.PERF We see, then, that Wakhi clitics represent subjects of past tense constructions. 4.2.2 Where Clitics Attac h At first glance, pronominal clitics seem to attach to nearly any phrasal constituent in the sentence. Upon closer inspection, though, a pattern emerges. There is a

PAGE 41

41 syntactic basis for cliticization. In this section I will demonstrate the syntactic fou ndation for clitic location by showing that clitics may only attach to phrasal constituents, except in rare circumstances. This is contrary to Pashto, which allows clitics to intervene within constituents and even within words as in 21 and 23 4.2.2.1 Allowed hosts Among licit hosts for clitics are objects, both full NP s and pronouns. (41) a. kashv= em dietu boys=1SG hit.PAST.PERF b. yav= em dietu 3PL.PRN=1SG hit.PAST.PERF Clitics can also attach to adjectiv al elements as in 35 repeated here as 42 for convenience. (42) yezi shapik raftch baf= om xetu yesterday bread very fine=1SG make.PAST.PERF 10 Note, though, that in this case in 42 shapik could not take the clitic Only baf could grammatically host the clitic as in 43 There is some optionality in the relative order of noun and adjective phrase. There does not seem to be optionality, however, regarding 10 It is worth noting here that (42) and (43) provide evidence against a phonological analysis of these clitics. It is difficult to conceive of a well motivated constraint that would preclude (43) (b), but allow the other two sentences. A syntactic view is much simpler.

PAGE 42

42 serve as its host. (43) a. yezi raftch baf shapik= om xetu yesterday very fine bread=1SG make.PAST.PERF b. *yezi shapik= om raftch baf xetu yesterday bread=1SG very fine make.PAST.PERF Another possible host is a temporal adverb such as yezi This cl ausal modifier is a phrase, AdvP, on its own, and therefore maintains the proposal that clitics can only suffix to constituents. (44) yez= iv yark xak kuzgve shapik yitu yesterday=3PL work make NOMIN bread eat.PAST.PERF For each of these positions in the clause, the clitic s presence does not change the types and degree of emphasis are conveyed by their positioning and optional (Bashir 1986:16) The lack of semantic effects could be due to dialect differences. Also possible is diachronic c hange. The data in this study were collected in 2011, while sometimes twenty years before publication. On e of these, combined with the relative dearth of recent data, is likel y the source of the differences concerning semantic effects. A larger body of data will clarify this issue.

PAGE 43

43 A study of all the allowed positions reveals a consistent theme. The clitics may only attach to phrasal constituents, rather than individual words This is consistent with the above data. One comment is necessary before looking at elements which may not host clitics. In Chapter 2 sentence 17 we saw that clitics may not move out of their clauses. So while clitic atta chment is phrasal, it is only for phrases within the clause. 4.2.2.2 Disallowed hosts While there are a number of acceptable positions for clitics to use as hosts, there are also several i llicit hosts includ ing subjects. Wakhi clitics cannot attach to an y overt subject DP, either full NP or pronoun. (45) a. Alia ghar winetu Ali y a rock see.PAST.PERF y b. Alia= d ghar winetu Ali y a=3SG rock see.PAST.PERF y This prohibition also includes interrogative pronou ns. (46) a. chiz pervetu what fall.PAST.PERF

PAGE 44

44 b. *chiz= ov pervetu 11 what=3PL fall.PAST.PERF Neither can clitics attach to articles which follows from the general prohibition aga inst clitics intervening within constituents. (47) a. yez= iv Ali at Rehbere yi parinda winetu yesterday=3PL Ali y a.and Rehber the bird see.PAST.PERF y b. *yezi Aliat Rehbere y= iv parinda winetu yesterday Ali y a.and Rehber the=3PL bird see.PAST.PERF In the same vein, quantifiers cannot host clitics. As part of a DP, they are internal to their phrase. (48) a. yezi chok shapik= om winetu yesterday much bread=1SG see.PAST.PERF b. *yezi chok= om shapik winetu yesterda y much=1SG bread see.PAST.PERF Verbs cannot serve as hosts for clitics Recall 44 above. While the clitic may attach to the temporal adverb, it may not generally attach to the verb, as below in 49 (49) yezi y ark xak kuzgve shapik yit= u v yesterday work make NOMIN bread eat.PAST.PERF=3PL 11 While the clitic tested here, =ov not work here either. Wakhi can use 3PL to obtain a general meaning. yez=iv maj dixt yesterday=3PL 1SG.PRN hit

PAGE 45

45 The exception to this prohibition is when the verb is the only other phonologically realized element. The clitic must follow a phonological host. Below we will see that clitics are base generated higher than verbs. When they are the only phonologically realized elements, PI must move the clitic in PF so that it has a host to its left. In this case that host is an element that would otherwise be disallo wed, the verb. (50) a. maj yit 1SG.PRN eat.PAST.PERF b. Yit= om eat.PAST.PERF=1SG We saw a similar construction above in 17 Example 17 h is repeated here as 51 Without yez i the clitic to keshen yezi at the beginning of 51 b a suitable host for the clitic, makes the placement of the clitic clause fina l l y illicit. (51) a keshen=om [ ki shapik=it yikt ] hear=1SG COMP bread=2SG eat.PAST.PERF b. *yezi keshen=om [ ki shapik=it yikt ] yesterda y hear=1SG COMP bread=2SG eat .PAST.PERF While clitics can attach to clausal adve rbs like yezi they cannot attach to intensifier adverbs like those that modify adjectives. In 43 the clitic could attach to both shapik baf (whether

PAGE 46

46 head initial or head final) Here, though, it may not attach to the phrase internal raftch (52) *yezi shapik raftch = om baf xetu yesterday bread very=1SG fine make.PAST.PERF Nor can clitics attach to verbal modifiers. (53) a. yezi shapik= om raftch jald xetu yesterday bread=1SG very quickly make.PAST.PERF b. *yezi shapik raftch jald= om xetu yesterday bread very quickly=1SG make.PAST.PERF This is consistent with other clitic placements under the assump tion that this AdvP adjoins as part of the VP. Since AdvP adjoins to the VP, it is phrase internal to that VP. It cannot, therefore, be host to the clitic. The same pattern emerges whether studying allowed or disallowed clitic host positions. Clitics ma y only attach to phrase final words (except the VP ) and may not attach to phrase internal elements. 4.2.2.3 Disallowed environment s In addition to restrictions on hosts, clitics are also restricted by environment. For example, co occurrence of overt subj ects and clitics is restricted. Clitics can co occur in the same clause with an explicit NP subject as in 47 above They cannot, however, do so with a pronominal subject as in 54 (54) a. yezi maja s hapik raftch jald xetu yesterday 1SG.PRN bread very quickly make.PAST.PERF

PAGE 47

47 b. yezi shapik= om raftch jald xetu yesterday bread=1SG very quickly make.PAST.PERF c. yez= im shapik raftch jald xetu yesterday=1SG bread very quickly make.PAST.PERF d. *yezi maja shapik= om raftch jald xetu e. *yez= im shapik raftch jald xetu The pronoun is acceptable without a clitic as in 53 a, and clitics are acceptable without the pronoun as in 53 b c. To combine them in the same clause is illicit in 53 d e Clitics and pronouns are in complementary distribution. O n the surface th is might seem to be evidence that Wakhi clitics are pronouns, but we will see in the next section that this is not the case, based on a dditional evidence which indicates that these clitics are agreement morphemes. Clitics are precluded from clauses with inanimate subjects. (55) a. yezi parindasht tsa mezen raben pervetu yesterday birds from table down fall.PAST.PERF b. yezi parindasht= ov tsa mezen raben pervetu yesterday birds=3PL from table down fall.PAST.PERF c. yezi kitobisht tsa mezen raben pervetu yesterday books from tab le down fall.PAST.PERF

PAGE 48

48 d. yezi kitobisht= ov tsa mezen raben pervetu yesterday books=3PL from table down fall.PAST.PERF The unaccusative subject parindasht normally inanimate kitobisht for anthropomorphized. The subjec t ceases to be inanimate, becom ing intentional or volitional in a poetic sense. While this construction is possible, it is odd to a native speaker. T here are no agreement morphemes for inanimate subjects in the present tense either. (56) a. kitobish ta mezen raben pervte books from table down fall.PRES b. kitobish ta mezen raben pervt en books from table down fall.PRES 3PL The motivation for excluding inanimate subjects from hosting clitics is not immediately ap parent. This is an area for further research. 4.3 The True Nature of Wakhi Clitics Two issues are now left concerning Wakhi pronominal clitics. In this section we will determine whether they are arguments (phonologically deficient pronouns) or agreement morphemes as suggested by Roberts (2000) for Pashto In the final section of this chapter we will then turn to their ori gin, that is, where they are base generated.

PAGE 49

49 4.3.1 Arguments In this section I will present data which might indicate that Wakhi clitics are arguments. There are two main bits of evidence to support the notion that clitics are arguments. First, and str ongest as mentioned earlier they are in complementary distribution with full subject pronouns. While clitics may be present in clauses with NP subjects such as Aliat Rehbere y in 47 repeated here as 57 they are excluded from clauses with overt pronominal subjects like those in 54 (57) a. yez= iv Aliat Rehbere yi parinda winetu yesterday=3PL Ali y a.and Rehber the bird see.PAST.PERF li y b. *yezi Aliat Rehbere y= iv parinda winetu yesterday Ali y a.and Rehber the=3PL bird see.PAST.PERF The second, albeit weaker, piece of evidence is that NPs and full pronouns appear in the same position in their cl auses. If a full NP is inserted into subject position, a subject p ronoun is unable to be inserted in the position. In this case, the pronoun still appears, but in phonolog ically deficient form, as a clitic. 4.3.2 Agreement Morphemes In this section I wil l demonstrate that Wakhi clitics are not arguments, but are in fact agreement morphemes. Included will be concepts such as complementary distribution, coordination, phonological form, and a discussion of clitic doubling. 4.3.2.1 Complementary distribution Just as clitics are in complementary distribution with full pronouns so too are they in complementary distribution with verbal agreement markers. In this case the division is between present tense, which generally requires verbal agreement markers, and past

PAGE 50

50 tense, which precludes them. Roberts (2000:98) notes that Pashto clitics, if they were arguments, w ould be expected to co occur with verbal agreement morphology Neither Pashto nor Wak hi clitics exhibit this behavio r Because clitics are in complementary distribution with both pronouns and agreement morphology additional evidence is necessary. 4.3 .2.2 Coordination Among this additional evidence is the fact that pronouns can be part of coordinated constructions, while agreement clitics and agreement morphology cannot. The coordinated structure in 58 a is acceptable. The attempt to coordinate clitic s, however, as in 58 b and 5 8 c is not acceptable. While 5 8 d is grammatical, it does not communicate the same information as 5 8 a. The subject of 5 8 (58) a. to et maja kitob winetu 2SG.PRN and 1SG.PRN book see.PAST.PERF b. *kitob=ov=om winetu book=2SG=1SG see.PAST.PERF c. *kitob=ov t=om winetu book=2SG and=1SG see.PAST.PERF d kitob=on winetu book=1PL see.PAST.PERF

PAGE 51

51 4.3.2.3 Phonological form Like agreement morphemes, clitics are phonologically weak. They cannot stand on their own as words A ffixes cannot attach to clitics to form words and they must attach to roots or stem s. 4.3.2.4 Clitic doubling Wakhi clitics appear on the surface to act similarly to the phenomenon of clitic doubling (CD). is a construction in which a clitic co occurs with a full DP in (Anagnostopoulou 2007:520) Th is description of CD seems to refer to sentences such as 59 Here we see the f ull DP Aliat Rehbere the clitic =iv (59) yez Aliat Rehbere yi parinda = a v winetu yesterday Aliya.and Rehber the bird =3PL see.PAST.PERF Clitics like this one also look like agreement morphemes that happen to attach to some constituent other than the verb, where agreement markers are often found. Within t he liter ature on clitic doubling ( for example Preminger 20 09; Anagnostopoulou 2007 ) there are three common notions regarding the nature of doubled clitics. The fi rst is that clitic doubling occurs only for non subject DPs. This is obviously not the case in Wakhi. Only subjects are represented through agreement clitics. There is no agreement marking for other arguments. The second is semantic effects due to the presence and placement of clitics in a clause. Such effects are expected in clitic doubling constructions. Wakhi agreement clitics create no semantic effects making a discussion of clitic doubling difficult for Wakhi Third is the notion that clitics a bsorb case of the related NP, requiring the NP to appear in some other, often

PAGE 52

52 oblique, case. This requirement often manifests as a preposition. This does not occur in Wakhi, where subject NPs that co occur with agreement clitics are identical to the NPs in clauses without the clitics. Out of the research of clitic doubling, however comes an innovation from Sportiche (199 6) and Koopman (1996) which provides a mechanism for explaining the optionality of clitics just mentioned They propose application of the doubly filled voice filter and generalized doubly filled COMP filter, respectively. These filters both seek to shed light on the discussion between whether clitics are base generated (agreement) or subject to movement (arguments). They argue that clitics are base generated, with the NP arguments subject to movement when necessary. We will examine the COMP filter in more detail below. The evidence indicates that Wakhi clitics are not arguments, but rather non verbal agreement morphemes. In the final section of this chapter I will offer a proposal of where these agreement clitics are base generated and the mechanism that accounts for their various surface positions. 4.4 A Home for Wakhi Clitics The determination that clitics are not arguments bu t rather agreements, eliminates much of the clause structure as a possibility for where they are generated. Borrowing from Roberts (2000) we will assume that these clitics, like their Pashto counterparts, are heads. As such, t hey almost certainly have to be generated above v P A look at the data shows us how far above v P to look. While sentences like 60 m ight suggest that they are generated low in the sentence (just above the verb), movement could easily account for that construction as we will see below

PAGE 53

53 (60) yezi shapik= ov raftch jald xetu yesterday bread=3PL very quickly make.PAST.PERF s It seems reasonable, then, to posit that clitics are base generated in an Agreement Phrase (AgrP) above TP, la Pollock (1989) The adverb yezi could then adjoin within that projection (or any place above it, as will be apparent in a moment), and serve as the clitic host if necessary. Som e possible evidence for base generation high in the tree is the ability of yezi to host the clitic above the subject ( 61 and its syntactic tree in 62 below ) In this example, the object undergoes object shift and raises to spec,vP, thus producing the correct surface order for the AdvP and the verb. The subject, base generated in another specifier of vP, deletes a weak uninterpretable feature (that is, one which does not require movement) on Agr th rough A GREE by virtue of the c command relationship between the two. Thus the clitic =ov Yezi adjoins above Agr. The derivation converges. Unlike Pashto syntax (Roberts 2000:84,88), Wakhi does not have an EPP feature to attract the s ubject to spec,TP. The scrambling that is apparent in Pashto occurs because of a strong feature on T. Th is movement allows an analysis of Pashto that independently derives the 2P structure. Because the clitic projection is below TP, the subject raising to spec,TP places it above the clitic. The clitic is thus located in the second position of the clause through standard syntactic operations rather than a separate clitic movement operation.

PAGE 54

54 The lack of an EPP feature in Wakhi partly explains the differen ce between Wakhi and Pashto clitic positions. In Wakhi, the movement (or not) of arguments above the clitic in its base generated site is more complicated As seen in the following examples, the position of the clitic depends on movement operations which can render a 2P clitic in the surface structure, but can also find the clitic following the third element in the clause. (61) yez= iv Aliat Rehbere shapik yesterday=3PL Aliya.and Rehber bread raftch jald xetu very quickly make.PAST.PERF (62)

PAGE 55

55 Working with the clitic base generated high in the tree, we can evaluate that claim by examining other structures and at the same time provide insight into the larger (Brent Henderson, p.c.) Consider 63 and 64 for example. As before, the object shifts to vP. Unlike the previous sentence, however, the subject moves to spec,A grP to check a strong uninterpretable feature. The object moves again, to spec,TopicP. To avoid violat ion of the generalized doubly filled COMP filter (DFCF), discussed in greater detail below, the subject moves out into spec,TopicP. Yezi adjoins above the subject and the derivation converges. The object is in position to serve as clitic host. (63) yezi Aliat Rehbere shapik= ov raftch jald xetu. yesterday Alia.and Rehber bread=3PL very quickly make.PAST.PERF (1996) DFCF is derived from the Linear Correspondence Axiom (LCA) by Kayne (1995) Similar to the doubly filled Voice filter by Sportiche (1996) the DFCF allows a language to have an overt head and silent specifier in a given projec tion, or vice versa Both of them cannot be overt. B ased on the premise that there must be asymmetric c command to have linear order Koopman derives the filter by allowing segments (X ) to participate in c command. The subject in sentence 63 having moved to spec,AgrP (t SUBJ in 64 ), is not in asymmetric c command relationship with Agr removing the possibility of determining linear order. In order to avoid crashing, one of the elements in the d erivation must be covert (silent) or one of them must move as happens here, when the subject moves to spec,TopicP.

PAGE 56

56 (64) The DFCF accounts for many of the structures common to Wakhi clitics. The following sentence 65 has no c litic This is easily explained by the subject raising to spec,AgrP to discharge a strong uninterpretable feature. There is no Topic feature to motivate another movement, leaving two phonologically overt elements in the specifier and head position as in 66 Th e DFCF violation must be dealt with for the derivation to converge so the head is phonologically covert, leaving a sentence with no clitic.

PAGE 57

57 (65) yezi Aliat Rehbere shapik raftch jald xetu yesterday Aliya.and Rehber bre ad very quickly make.PAST.PERF (66) Pashto clitics, however, occur whether the subject is overt or not. Both NPs and pronouns can co occur with agreement clitics, as in 28 Bec ause subjects move to generated location, there will not be DFCF violations, allowing both overt subject and clitic. Sentences without any overt subject like 67 are also easily understood in light of DFCF. Because of the agreement system, Wakhi is a pro drop language. In this case, pro moves to spec,AgrP to delete a strong uninterpretable feature, but because it is phonologically null, its co occurr ence with the clitic in the projection does not violate

PAGE 58

58 DFCF as in 68 The object moves to spec,TopicP allowing it to host the clitic. Yezi adjoins above the object and the derivation converges. (67) yezi shapik= ov raftch jald x etu yesterday bread=3PL very quickly make.PAST.PERF Yesterday they (68) Overt pronouns are in complementary distribution with agreement clitics. W h ile on the surface this might lead one to claim the clitics are pronouns, DFCF provid es an analysis that accounts for an absent clitic. In 69

PAGE 59

59 precludes the appearance of an agreement clitic. The pronoun moves t o spec,AgrP to delete a strong uninterpretable feature as in 70 Wakhi s ubject pronoun s do not move out the projection to avoid a DFCF violation. The head must be silent when a pronoun is present, therefore deriving the complementary distribution of overt subject pronouns and agreement clitics (69) yezi yava shapik raftch jald xetu yesterday 3PL.PRN bread very quickly make.PAST.PERF (70) These sentences and their syntactic trees present a clear explanation for the varied agreement structures in Wakhi. An a nalysis following DFCF accounts for the data.

PAGE 60

60 Only one issue is left unsatisfied at this point. What to make of a sentence like 71 which has only two phonologically realized components: verb and clitic. Wakhi is strongly ve rb final. But clitics must have a host to their left. There is no motivation for V to raise above Asp, leaving no satisfactory syntactic solution. (71) yit= om eat.PAST.PERF=1PL Since syntax cannot account for the grammaticality of 71 the only recourse is to appeal to phonology. Like many other treatments of clitics, Prosodic Inversion (Halpern 1992) will co initial clitic right ward until it finds a suitable host, in this case the verb itself. While Wakhi agreement clitics are not second positio n, the understanding of their nature, base gen erated position and position relative to other constituents benefits from the study of true 2P clitics in other languages. Much more needs to be done, however, to fully understand this phenomenon. Future research will give greater enlightenment.

PAGE 61

61 CHAP TER 5 FUTURE RESEARCH AND CONCLUSION Wakhi clitics are complex To understand them is to understand many aspects of the grammar beyond clitics. Their study is enriched by the study of other similar structures within languages genetically and geographical ly near (Shugni and Pashto, respectively), and those more removed (ancient Greek and Serbo Croatian, to name two). Wakhi also opens the possibility that second position clitics are less frequently found than first thought, leading to a more improved under standing of this phenomenon in which syntactically free elements are not free phonologically and must rely on a host for expression. While the present study addresses some theoretical questions, it is still less than wholly satisfying. More research is ne eded to better understand the phenomenon. Three areas have emerged from the current research that would be promising veins for further investigation: dialectical variation, understanding of phrase structure on a more detailed level, and a specific questio n about why inanimates are precluded as clitic hosts. 5.1 Dialectical Variation The vast majority of Wakhi data in this paper came from one speaker, Habiba, originally from northern Pakistan, currently living in the United States with her Wakhi husband and children. Other than this consultant the only available Wakhi data were from a grammar written by a British linguist, D.L.R. Lorimer, in 1958 (Lorimer 1958a; 1958b) and another grammar written by Soviet linguists in 1976, then later translated into French (Grunberg and Steblin Kamensky 1988a; 1988b) Lorimer did field work in northern Pakistan, the Soviets in Tajikistan.

PAGE 62

62 Since Lorimer gat hered data in the same area that my consultant was from I was often surprised at the differences between the grammatical structures represented in his work and those derived from recent elicitations Most noticeable was the number of times a clitic could appear in a clause (multiple times for Lorimer, but only once for Habiba). This and other variations were puzzling, and I have no explanation, but only research (which took place more than a decade before his grammar was published) and the recent elicitation sessions, significant diachronic change has occurred. There were also differences between the Pakistan data that from Soviet Tajikistan. Bashir, one of the f ew linguists to study Wakhi, has documented some differences in her writings (Bashir 1986; 2009) but there is still little with which to compare these differences. Only field work in Tajikistan, Pakistan and Afghanistan 12 can remedy the dearth of modern data availability 5.2 Phrase Structure The least satisfying component of this study is the treatment of movement operations to account for various surface structures. This weakness came from a lack of suitable data from which to derive the underlying phrase structure. While the available data were marg inally suitable for an M.A. thesis beyond that much more is needed. One possib le solution was to use data from Lorimer (1958a) and Grunberg Steblin Kamensky (1988a) Their data were so suspect compared to the more recent 12 I am unaware of any linguistic research on Afghan Wakhi.

PAGE 63

63 data (for reasons mentioned above), that any conclusions drawn from those data would not have been compatible with conclusions drawn from the recently elicited data. 5.3 Inanimates While the two issues mentioned above for further study are broad, there is one topic for w hich I ha ve no satisfactory answer. Why are inanimate NP subjects unable to host clitics, while animate subjects are able to do so? The pattern does appear in at least one other part of Wakhi grammar, verbal agreement markers in the present tense. I am unable to hazard even a guess regarding the reason behind this difference. For now it is sufficient to say that inanimate subjects in Wakhi do not trigger agreement. The answer will likely come from an intimate knowledge of the language gained by a lingu ist spending a great deal of time in the field. 5.4 Conclusion In this study I take useful innovations from previous studies, some phonological and others syntactic in their framework, and propose a nearly syntax only approach to Wakhi clitics based on em pirical evidence This approach eliminate s all but the most necessary phonological operation. The resulting proposal outlines the syntactic nature of Wakhi agreement clitics. This includes their base generated location in the phrase structure, then cont inues that notion by outlining syntactic operations that derive the placement of the clitic in relation to other constituents in the clause, something not always undertaken in studies of clitics in other languages.

PAGE 64

64 APPENDIX A WAKHI LANGUAGE MAPS Figure A 1. Kyrgyzstan and Tajikistan language map [Reprinted with permission from Lewis, M. Paul (ed.). 2009. Ethnologue, 16th edition. Dallas : SIL International Online version: http://www.ethnologue.com.]

PAGE 65

65 Figure A 2. Afghanistan language map [Reprin ted with permission from Lewis, M. Paul (ed.). 2009. Ethnologue, 16th edition. Dallas: SIL International. Online version: http://www.ethnologue.com.]

PAGE 66

66 Figure A 3. Northern Pakistan language map [Reprinted with permission from Lewis, M. Paul (ed.). 2009. Ethnologue, 16th edition. Dallas: SIL International. Online version: http://www.ethnologue.com.]

PAGE 67

67 REFERENCES A NAGNOSTOPOULOU E LENA 2007. Clitic doubling. The Blackwell Companion to Syntax ed. by M. Everaert and H. van Riem sdijk, 519 581. Malden, MA: Blackwell Publishing. A NDERSON S TEPHEN R 1992. A Morphous Morphology Cambridge: Cambridge University Press. A NDERSON S TEPHEN R 1993. Wackernagel's revenge. Language 69.68 98. A NDERSON S TEPHEN R 1996. How to put your cliti cs in their place, or why the best account of second position phenomena may be something like the optimal one. The Linguistic Review 13.165 191. B ACKSTROM P ETER C 2009. Grammar sketch for Gojal Wakhi. Dallas, TX, MS B ARIE A MANDA E LIZABETH 2009. Explor ing cleft sentences and other aspects of Shugni syntax. Lexington KY: University of Kentucky thesis. Online: http://web.as.uky.edu/linguistics/shughni/Full Thesis.pdf Accessed o n December 27, 2010. B ASHIR E LENA 1986. Beyond split ergativity: Subject marking in Wakhi. Proceedings of the 22nd Regional Meeting (CLS 22) 14 35. Chicago: Chicago Linguistic Society. B ASHIR E LENA 2009. Wakhi. The Iranian Languages ed. by Gernot Win dfuhr, 825 862. New York: Routledge. B ELJKO 1995. Participle movement and second position cliticization in Serbo Croatian. Paper presented at Annual Meeting of the Linguistic Society of America, New Orleans. B ELJKO 2001. On the Nature of the Syntax Phonology Interface: C liticization and Related Phenomena Oxford: Elsevier. B ROWNE W AYLES 1975. Serbo Croatian enclitics for English speaking learners. Contrastive Analysis of English and Serbo Croatian ed. by R. Filipovic, 1:105 134. Zagrab: University of Zagreb. C YSOUW M I CHAEL 2004. Morphology in the wrong place: A survey of preposed enclitics. Morphology and its Demarcations ed. by Wolfgang U. Dressler. Amsterdam: John Benjamins. Online: http://www .taleshyar.com/Articles/418_cysouwMORPH.pdf Accessed on February 28, 2011.

PAGE 68

68 E RSCHLER D AVID 2010. On optionality in grammar: The case of East Iranian almost Wackernagel clitics. Paper presented at the Syntax of the World's Languages IV, Lyon, France. G R UNBERG A.L., AND I.M. S TEBLIN K AMENSKY 1988a. La Langue Wakhi 1: Corpus de Littrature O rale (Trans.) Dominique Indjoudjian. Vol. 1. 2 vols. Paris: Editions de la Maison des Sciences de l'Homme. G RUNBERG A.L., AND I.M. S TEBLIN K AMENSKY 1988b. La Lang ue Wakhi 2: Essai Grammatical et Dictionnaire (Trans.) Dominique Indjoudjian. Vol. 2. 2 vols. Paris: Editions de la Maison des Sciences de l'Homme. H ALPERN A ARON 1992. Topic in the placement and morphology of clitics Palo Al to, CA: Stanford University dissertation. K AYNE R. 1995. The Antisymmetry of Syntax Cambridge, MA: MIT Press. K OOPMAN H ILDA 1996. The spec head configuration. UCLA Working Papers in Syntax and Semantics 1.37 64. L EGATE J ULIE A NNE 2008. Warlpiri and the theory of second positio n clitics. Natural Language and Linguistic Theory 26.3 60. L EWIS M. P AUL (ed.) 2009. Wakhi 16th ed. Dallas, TX: SIL International. Online: http://www.ethnologue.com/show_language.asp?cod e=wbl Accessed on February 28, 2011. L ORIMER D.L.R. 1958a. The Wakhi Language, Volume I: Introduction, Phonetics, Grammar and Texts Vol. 1. 2 vols. London: School or Oriental and African Studies. L ORIMER D.L.R. 1958 b The Wakhi Language, Volume II: V ocabulary and Index Vol. 2. 2 vols. London: School or Oriental and African Studies. P AYNE T HOMAS E. 1997. Describing Morphosyntax: A Guide for Field Linguists Cambridge: Cabmbridge University Press. P OLLOCK J EAN Y VES 1989. Verb movement, Universal Gra mmar and the structure of IP. Linguistic Inquiry 20.365 424. P REMINGER O MER 2009. Breaking agreements: Distinguishing agreement and clitic doubling by their failures. Linguistic Inquiry 40.619 666. R ICHARDSON J OHN F.; M ITCHELL M ARKS ; AND A MY C HUKERMAN ( eds.) 1983. Papers from the Parasession on the Interplay of Phonology, Morphology, and Syntax Chicago: Chicago Linguistic Society.

PAGE 69

69 R OBERTS T AYLOR 1997. The optimal second position in Pashto. PF: Papers at the Interface 30. MIT Working Papers in Linguist ics.369 394. R OBERTS T AYLOR 2000. Clitics and agreement Cambridge, MA: Massachus etts Institute of Technology dissertation S PORTICHE D OMINIQUE 1996. Clitic constructions. Phrase Structure and the Lexicon ed. by Johan Rooryck and Laurie Zaring, 213 27 6. Dordrecht: Kluwer Academic Publishers. T EGEY H ABIBULLAH 1975. A study of Pashto clitics and implications for linguistic theory. Studies in the Linguistic Sciences 5.154 196. T EGEY H ABIBULLAH 1977. The g rammar of c litics: Evidence from Pashto and o th er l anguages. Urbana, IL: University of Illinois at Urbana Champaign dissertation T OOSARVANDANI M AZIAR 2007. Ellipsis in Farsi complex predicates. Berkeley, CA: University of California, Berkeley, MS Online: http://ling.ucsc.edu/~hank/toosarvandani_vpe.pdf Accessed on February 28, 2011. W ACKERNAGEL J ACOB 1892. ber ein Gesetz der indogermanischen Wortstellung. Indogermanische Forschungen 1.333 436. Z WICKY A RNOLD 1977. On Clitics Bl oomington, IN: Indiana University Linguistics Club.

PAGE 70

70 BIOGRAPHICAL SKETCH Todd R Hughes was born in Ohio. After living in Ohio, Kentucky and Texas, Stant on College Prepar atory School Following high school, he earned a B.A. in History at Cumberland Colleg e in Williamsburg, Kentucky As a gr a duate student Todd earned separate M.A.s in Education and Linguistics from t he College of William and Mary in Williamsburg, Virginia and the University of Florida in Gainesville respectively. Building on a summer spent in West Africa during college, Todd left Virginia and initially to run a guest house. During four years in West Africa, his jobs were varied, from logistics coordination in Burkina Faso to roving photographer These tasks took him from the west coast of Africa to the Plateau of Nigeria, and most places in between. In 2003 he was dispatched to northern Iraq to document projects in a sma ll town west of Mosul during the aftermath of the U.S. invasion of Iraq. For a few weeks i n 2004 he found himself in Khorog, Tajikistan introduc ing him to the beaut iful Pamir Mountains and the Wakhi people who call them home. Upon returning to the U.S., Todd was asked to give back to his high school alma mater by teach ing U.S. History and French, which he did for four years. During that time he spent two summers at the University of North Dakota, studying at the Summer Institute of Linguistics. It was t here he met Dr. Sarah Titkemeier whom he eventually marr ied