135. 1.11:: It 5:35.. .1}: .. f.( .v I. . , i: aids! . 1!... $.. I... It. f t I. r.‘ . .I 5... :3 .5 2, .I..0I.L.: .1 3 .I... 31....IL £3; 1. . 3 , .,\ Li, 63...! rlhhu~uy§ “whim?! 3% :41!) 3...... . nan” . #5:. 1 AOL. J} )2 - ~ snfirfi. ‘9 “ma .2» x 3...: .z. .2. 3!. $9,.3023 .II 1.3!! .u-lt album“, :4. 5‘ .- .1.» a3... Ill. no. :12. . .31 35...... z «1 1.3 .unxaiutn yl ”A r {a 1.13).}. $3 $4. a! :1. 2K. 5 0.. .3 1031):..3. . €1.23 ‘ .I: 1.? .z\. 8:} 9.4.3. .ti. 5..)V55ita. to “lutvfiis xv; . 13 .l 7 cf! . 2 .34.... S‘xnkllt‘vvb)! nu? Hi: :1.- . art-fl.- a... :h. 2‘... 5.1.95 A «n :.?.5L:t ‘95....) {tot . (2.5.5 “11' .12: Q .2 i 'V I 4.3 :23! a .1 Matt. I .2: .x , .L‘Il. . g4... . tiybaflhumflaljy . l. J... .. .I at)! I: “3 Asa Jar-"r: .4 4 . _ . 3!... Pi. V I 5905'. 2!: 131:? Y. 1151.21.42: in. . t... 9'. 9t lHlHHllHUIIINIHHHIUllllllllllllllllllllllllllHllHl 31293 02048 3966 This is to certify that the dissertation entitled Parallelism between noun phrases and clauses in Thai presented by Saisunee Visonyanggoon has been accepted towards fulfillment of the requirements for Ph-D- degreeiniinguisljss. Major professor Date 3 “(112000 M5 U is an Affirmative Action/Equal Opportunity Institution 0-12771 LIBRARY Michigan State University —.._. 7’ ~ _.._——‘ PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DA'I’E DUE AUG 1 5 2003 '99 .1 0 0 3 is; m 11/00 e/CIRC/DmoOm.p65-p.14 PARKLLELISM BE PARALLELISM BETWEEN NOUN PHRASES AND CLAUSES IN THAI By Saisunee Visonyanggoon A Dissertation Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Linguistics 2000 This dissertation a rises and clauses in Th The research sho\1 a: the noun alwa} fate to be equivalen :-: :au: phrases comes f1 up,“ . an .i V, ‘11 em across the Cl; [:7‘ . k M4 erbalh, the 2‘:- I‘. ABSTRACT PARALLELISM BETWEEN NOUN PHRASES AND CLAUSES IN THAI By Saisunee Visonyanggoon This dissertation addresses some parallel characteristics of the syntax of noun phrases and clauses in Thai within the Principles and Parameters framework. The research shows that Thai noun phrases and clauses are uniformly head-initial. Although the noun always appears before a classifier, which has been assumed in the literature to be equivalent to a D, the crucial evidence supporting the head-initial analysis of noun phrases comes from instances where nouns are modified by 'referential' adjectives or demonstratives, which the head-final approach fails to account for. Therefore, the apparent head-final structure of the noun phrase results from NP movement across the classifier. Similarly, although some modals and aspect markers occur post-verbally, the post-verbal modals are shown to be heads generated above VP with VP movement across them while the post-verbal aspect markers are analyzed as phrases right-adjoined to some AspPs above VP. The apparent mixed headedness of the clause is thus explainable. The analyses of clauses make use of the two tests: predicator and negation. The predicator test indicates the scope positions of auxiliaries relative to one another as well as their syntactic status as heads or phrases. The negator may 3, analyzed as a NegP generated in the specifier of a verbal projection, helps confirm the syntactic status of the auxiliaries under examination. To my parents and my brother iii lwould like to exprcs Lt. .i Cristina Schmm for 13::Alan llunn the maj or ; 2::: proxide me with valua “"358 patience, encoura-é 1:1: no: haie been accomi lcan'i not mention 1 33;;eton of this work. .\l‘ t parts of the world a: 'i '\«,. I iz-‘I ~15 Was Writing mV diss 2::-4n: ' ' :r my family. sends ch lie-pt me cor D. 11:31! Q Who hElPCd me u QSC‘laj-le « ‘ Utky 10 ha\'e S] y. A. site “'henever my in: Last but not least I E‘Q‘QST' ,: L. “- . o . - 5Q: if ~10 .‘ l 3“ Beretta {0, Q db1:~'both . 0m Tees . - ‘IS : :Ddd . em m the Lin ACKNOWLEDGMENTS I would like to express my deep gratitude to Alan Munn, Alan Beretta, Yen-Hwei Lin, and Cristina Schmitt for their willingness to serve as my committee. Special thanks go to Alan Munn, the major advisor, and Cristina Schmitt for their regular meetings with me to provide me with valuable insightful suggestions and comments. Without their expertise, patience, encouragement and unfailing confidence in me, this dissertation would not have been accomplished. I can’t not mention those who, in one way or another, have contributed to the completion of this work. My family, my boyfriend, my friends at MSU as well as in various parts of the world and my former students never failed to give me mental support while I was writing my dissertation. I am thankful to my younger sister, Ead, who, on behalf of my family, sends me daily e-mail messages to show their love, concern and support, which kept me continuing this work. I will never forget the generosity of John Halliwell who helped me with my TA work while I was busy finishing my dissertation. Also, I am lucky to have Sudawan Ariyasap here who is willing to check some of the Thai data whenever my intuition seems to go astray. Last but not least, I am indebted to all the faculty in the Department of Linguistics at MSU, who have ingrained in me knowledge of various areas of linguistics. I am grateful to Alan Beretta for giving me the opportunity to be his TA. I have learned considerably both from my studies and from being a TA. All in all, I have to say that being a student in the Linguistics program at MSU is one of the memorable experiences in my life. iv Set'u'oles ................... 2:. cf abbreviations .......... .3 :rscnption guide ...... GLtPTER 1: INTRODL’CTI 1.1 Purpose and scope 12 Theoretical framei 1.3. Organization ....... SAFER 2: THAI BARE .\ 2.1 Introduction ....... 22 C lusifrer .......... 2.3 Previous studies. . 2.3.1Longobardi ( 2.3.2 Cheng and Si 3.4 Distribution and i i , ..4.1 COmmOn “0| 7‘ n I a I ' o o TABLE OF CONTENTS List of tables ........................................................................................... x List of abbreviations ................................................................................. xi Data transcription guide ........................................................................... xii CHAPTER 1: INTRODUCTION 1.1 Purpose and scope ...................................................................... 1 1.2 Theoretical fi’amework ................................................................. 4 1.3 Organization ............................................................................. 5 CHAPTER 2: THAI BARE NOUNS 2.1 Introduction .............................................................................. 7 2.2 Classifier ................................................................................. 8 2.3 Previous studies ........................................................................ 13 2.3.1 Longobardi (1994) ............................................................... 13 2.3.2 Cheng and Sybesma (1999) ..................................................... 15 2.4 Distribution and interpretation of Thai bare nouns ................................ 20 2.4.1 Common nouns ............................................................. A ..... 20 2.4.2 Proper names ..................................................................... 22 2.4.3 Pronouns ........................................................................... 23 2.5 Analysis of Thai bare nouns ............................................................... 27 2.5.1 Head-final analysis ............................................................... 27 2.5.2 Head-initial analysis ............................................................. 36 2.6 Conclusion .............................................................................. 42 {3th 3: NOLN Mt 3.11ritroduction ........ 3.2 Existence of Thai 2 3.2.1 Previous . A .2 Argumen 3.3 Prexious liter: 3.3.1 Sproat ar. 3.3.2 Bernsteir (J) 3.3 Kester (1 3.4.1 RESUIar; 3.4.2 ReferemI CHAPTER 3: NOUN MODIFIERS IN THAI 3.1 Introduction .................................................................................. 44 3.2 Existence of Thai adjectives ............................................................... 47 3.2.1 Previous studies on the lack of Thai adjectives .............................. 47 3.2.2 Arguments for the existence of Thai adjectives .............................. 49 3.3 Previous literature ..................................................................... 54 3.3.1 Sproat and Shih (1988) .......................................................... 54 3.3.2 Bernstein (1993) .................................................................. 58 3.3.3 Kester (1996) ..................................................................... 61 3.4 Types and positions of noun modifiers ............................................. 64 3.4.1 Regular attributive adjectives ................................................... 64 3.4.2 Referential adjectives ............................................................ 67 3.4.3 Nominal modifiers ............................................................... 70 3.5 Distribution and interpretation of modified noun phrases in Thai .............. 74 3.5.1 Noun phrases containing regular attributive adjectives .................... 74 3.5.2 Noun phrases containing nominal modifiers ................................ 76 3.5.3 Noun phrases containing referential adjectives ............................. 77 3.6 Conclusion ............................................................................ 79 CHAPTER 4: MODIFIED NOUN PHRASES 4.1 Introduction ........................................................................... 81 4.2 Analyses of modified noun phrases in Thai ...................................... 82 4.2.1 A noun phrase with a regular attributive adj ective/ a nominal modifier ............................................................. 83 vi 4.2.2 A noun P11"? 4.2.3 A noun Phr35 a nominal me. 43 Demonstratii'es. . . 4.3.1 Thai demons 4.3.1 Bernstein l 1 4.3.3 Position of ‘ 4.4 Analysis of 1101.17 4.4.1 A noun phi 4.4.2 A quantifi: 4.4.3 A noun pi: 4.5 Conclusion... Timur 5: Tim Mo; 5.1 Introduction. . 5-3 Preclicator. . .. 5‘3 Characteristic 53" Tints 01 5.3.2 Surface 5.3.3 Distribu 5.4 Analysis Of' 5.4.1 R001 IT) 5. . 42 EPISIen 5.5 . Thal [180“ c b Ll( 4.2.2 A noun phrase with a referential adjective .................................... 88 4.2.3 A noun phrase with a regular attributive adj ective/ a nominal modifier and a referential adjective ............................... 94 4.3 Demonstratives ........................................................................ 98 4.3.1 Thai demonstratives ............................................................. 99 4.3.2 Bernstein (1996) ................................................................ 101 4.3.3 Position of Thai demonstratives .............................................. 103 4.4 Analysis of noun phrases with demonstratives ............................. i ...... 108 4.4.1 A noun phrase with a demonstrative and a classifier ...................... 108 4.4.2 A quantified noun phrase with a demonstrative and a classifier ......... 109 4.4.3 A noun phrase with a demonstrative without a classifier .................. 110 4.5 Conclusion ............................................................................ 112 CHAPTER 5: THAI MODALITY 5.1 Introduction ........................................................................... 114 5.2 Predicator ............................................................................. 1 17 5.3 Characteristics of Thai modals ..................................................... 122 5.3.1 Types of modals ................................................................ 123 5.3.2 Surface positions of Thai modals ............................................. 124 5.3.3 Distribution of Thai modals ................................................... 125 5.4 Analysis of Thai modals ............................................................ 126 5.4.1 Root modals ..................................................................... 127 5.4.2 Epistemic modals ............................................................... 153 5.5 Thai negation ......................................................................... 164 vii 5.5.1 Distribution of may, ............................................................ 165 5.5.2 Analysis of may, ................................................................ 166 5.6 Interaction between modals and negation ......................................... 174 5.6.1 Modals that can be negated by may, ......................................... 174 5.6.2 Modals that cannot be negated by may, ..................................... 175 5.7 naa, and khuan , revisited ............................................................ 178 5.8 Conclusion ............................................................................ 180 CHAPTER 6: THAI ASPECTUALITY 6.1 Introduction ........................................................................... 182 6.2 Aspect ................................................................................. 183 6.3 Characteristics of Thai aspect markers ............................................ 185 6.3.1 Types of aspect .................................................................. 185 6.3.2 Surface positions of Thai aspect markers .................................... 186 6.3.3 Distribution of Thai aspect markers .......................................... 186 6.4 Analysis of Thai aspect markers ................................................... 188 6.4.1 Pre-verbal aspect markers ..................................................... 188 6.4.2 Post-verbal aspect markers .................................................... 207 6.5 Co-occurrences of modals and aspect markers ................................... 222 6.5.1 Surface positions of modals and aspect markers ........................... 223 6.5.2 Distributions of modals and aspect markers ................................ 224 6.5.3 Analysis ......................................................................... 225 6.6 Conclusion ............................................................................ 236 CHAPTER 7: CONCLUSION .................................................................. 238 viii mi?!“ References ........................................................................................ 240 incl Summary of interpr containing differen .5 Does of modals in Title? Positions of modal Teie4Positions of Thai 2 £155 Positions of prey:- .rlsoPositions of post-r LIST OF TABLES Table 1 Summary of interpretations expressed by noun phrases containing different types of noun modifiers .......................................... 80 Table 2 Types of modals in Thai ................................................................ 124 Table 3 Positions of modals relative to one another .......................................... 125 Table 4 Positions of Thai aspect markers ...................................................... 186 Table 5 Positions of pre-verbal modals and aspect markers ................................. 223 Table 6 Positions of post-verbal modals and aspect markers ................................ 223 AspP AuxP BP C 1 C -L C 0MP DEE DemP EXP EocP IMPE ModP NEG PP PRF FROG PRT QP RECP TOP AspP AuxP BP Cl C-L COMP DEF DemP EXP F ocP IMPF ModP NEG PP PRF PROG PRT QP RECP TOP LIST OF ABBREVIATIONS xi Adj ectival Phrase Aspect Phrase Auxiliary Phrase Boolean Boolean Phrase Classifier Classifier-like morpheme Complementizer Definite Demonstrative Phrase Experiential perfect Focus Phrase Irnperfective Modal Phrase Negative marker Preposition Phrase Perfect/Perfective Progressive Particle Question particle Reciprocal Topic particle Tense Phrase lflgrsonaun BilabialTl Stops 1 1d b \1. [mp p T 11.430. ph 1 Fricative T7 Lateral T Trill j— Nasal m T % Tfi Sn .‘ Ari; .‘ 1.5% km DATA TRANSCRIPTION GUIDE Six diphthongs: /ia, iia, ia, iia, ua, uua/ 319m Tones are represented as subscripted numbers afier each syllable. [1 2 3 4 5 l mid low falling high rising I xii 1 Consonant: Bilabial Labio—denta Alveolar Palatal Velar Glottal Stops Vd. b (1 V1. Unasp p t c k ? V1. Asp. ph th ch kh Fricative f s h Lateral 1 Trill r Nasal m n I] Glide w y 2 Em]: Front Central Back High i ii a; u uu Mid e ee 99 o 00 LOW a am 33 o oo 1.1 Purpose and scope .45 its title suggests. '. arses and clauses in Thai. seSVO language containr: 311313-331 below. illkhaws Elk. de‘s he love chrl ‘He loves childre (I) b011; [org-mat" 011 tree son the TICC‘ (3min, “mg m. lhfil he 10V That he loves c1 (4) baa“. khan), home of DESPITE [he pTEVa‘, 11:3, - _ . $3 While lacking 33410 d . Etenmners‘ :34; To ‘ . CHAPTER 1 INTRODUCTION 1.1 Purpose and scope As its title suggests, this dissertation explores syntactic similarities between noun phrases and clauses in Thai. Thai is always claimed to be strongly head-initial because it is an SVO language containing structures whose heads are always to the left as illustrated in (1)-(4) below. (1) khaw, rak, dek2 he love child ‘He loves children.’ i VNP (2) bonl tom-may4 PP = P NP on tree ‘on the tree’ (3) thii3 khaw, rak, dek2 CP = C IP that he love child ' ‘that he loves children’ (4) baan. khooos khaws NP = N PP house of he ‘his house’ Despite the prevalence of head-initial structures, however, there are some structures in Thai which appear at first sight to be head-final. Thai, like many East Asian languages, while lacking determiners, uses classifiers instead. Although classifiers are not identical to determiners, assuming the DP hypothesis (Abney 1987), they are generally thought to be functional heads which select NPs as complements. Classifiers in Thai, ‘ .1,- hwoer. appear to the rig. tf'thether Thai is head-tin. (511333“: 130511114 house Cl this ‘this house' 161baan3 saam. lan- house three Cl ‘those three hous In addition to the ap, aifiajt-lilre elements also : rasarniliaries to the left or metrically. Examples (5) WIN-verbally while 0*. (Tillkhawstgns ma he must cor ‘ 8171115! Come b. . 3W5 maa, lie cOme He can COme.‘ however, appear to the right of the noun as shown in (5) and (6). This raises the question of whether Thai is head-final in the DP functional structure. (5) hams lat], “ii-1 house Cl this ‘this house’ (6) baan3 saam, larj, nan4 house three Cl that ‘those three houses’ In addition to the apparent head-fmal functional heads in the DP, some Thai auxiliary-fike elements also appear to the right of the VP although Thai is predicted to have auxiliaries to the lefi of the main verb as is found in SVO languages cross- linguistically. Examples (7) and (8) show that some modals and aspect markers in Thai appear pre-verbally while others appear post-verbally. (7) a. khawstorj3 maal he must come ‘He must come.’ b. khaw, maal day3 he come can ‘He can come.’ (8) a. khawskamdarj, maa, he PROG come ‘He is coming.’ b. khaw, maaI lteaew4 he come PRF ‘He has already come.’ The post-verbal modals and aspect markers have been treated in the previous literature as either non-auxiliaries or verbs in series (Sereechareonsatit 1984, Thepkanjana 1986, Warotamasikkhadit 1996) or post-verbal auxiliaries (Panupong 1970, Sookgasem 13931.‘ This seems to indie" anon-auxiliaries or verbs-i tuned to be heads gener. Thepkanjana 1986). L'nder t i'luth initially and finally. 1 Era-verbal auxiliaries rm ”Ted to the head of \‘P a 191 1990).’ This seems to indicate that Thai clausal structure is mixed-headed. Treating them as non-auxiliaries or verbs-in-series, the post-verbal modals and aspect markers are suggested to be heads generated as VP adjuncts as in (9) (Sereechareonsatit 1984, Thepkanjana 1986). Under this analysis, Thai VP is assumed to be able to have the head V both initially and finally. Similarly, treating the post-verbal modals and aspect markers as post-verbal auxiliaries implies that they are heads of AuxPs generated to the right as opposed to the head of VP as illustrated in (10). (9) VP /\ VP V /\ Mod/Asp V XP (10) AuxP /\ VP Aux /\ Mod/Asp V XP Although the structure in (10) can take care of the fixed positions of post-verbal modals and aspect markers relative to one another, it is not different from the structure in (9) in being mixed-headed. Although the possibility of mixed-headed languages exists, languages with a uniform head direction are much more common. The apparent head- final constructions in Thai therefore require careful scrutiny to see if they are, in fact, head-final. In what follows, I will explore the possibility that Thai is a well-behaved language ' The claim that post-verbal modals and aspect markers in Thai are verbs-in-series from i the fact that serial verb constructions are abundant in the language. ri‘lrnourr phrases and clau instigate whether Thai nc huge despite the nouns new the head-initial 2 rich analysis can better ac fisher the post-verbal mo asst-Verbally. lfthere is evidence t Ls. in Thai, then the wor Mme that this is indet mrits ln fact. some elen with noun phrases and clauses being parallel in major respects. First of all, I will investigate whether Thai noun phrases are head-initial like all the other structures in the language despite the nouns always appearing initially in the phrases. I will do this by comparing the head-initial and the head-final analyses of Thai noun phrases to determine which analysis can better account for the relevant range of facts. Secondly, I will examine whether the post-verbal modals and aspect markers are base-generated pre-verbally or post-verbally. If there is evidence that supports a head-initial analysis for either the DP or the clause in Thai, then the word order must be derived by NP or VP movement respectively. I will argue that this is indeed the case for some, but not all, post-nominal and post-verbal elements. In fact, some elements will be argued to be right-adjuncts to VP or NP while others will be argued to be heads around which the NP or VP moves. My dissertation thus aims to answer the following research questions to discover whether there is any parallelism between noun phrases and clauses in Thai. 1. Are Thai noun phrases and clauses head-initial like other phrases in the language? 2. Are there instances where NP and VP movements operate in the syntax of Thai? 3. Are noun phrases and clauses alike in allowing right-adjuncts? 1.2 Theoretical framework I adopt for this research the Principles and Parameters framework developed in Chomsky (1986a, 1986b) as well as some aspects of the Minimalist Program (Chomsky t 6:27-41 N4... o93.1995)which is the ct Ii'tltltiCC each theoretical . 130rgan'lm‘l"n The dissertation i5 ”tses while the SECOUd P Chapter 2 discusse' use 1 will adopt somC‘ $th hat their analysis is since for DPs in Thai. ' frettionality of Thai nour imparted throughout. Chapter 3 consider Smears-a 1939, Hum QC t‘erbs, but I will slit 1'55. modifiers in Thai. na W... modifiers and she 7. ‘1 .3 Q p Ositions. C u‘ 3 'd‘ ers elabOraI 3311611. . T101 headfma‘ l Zedl' Illerms or) 1993, 1995) which is the continuation of the former approach to grammar. I will introduce each theoretical aspect I make use of in the section where it is adopted. 1.3 Organization The dissertation is divided into two main parts. The first part deals with noun phrases while the second part deals with clauses. The dissertation is organized as follows. Chapter 2 discusses Thai bare nouns. As Thai noun phrases are very similar to Chinese, I will adopt some proposals made by Cheng and Sybesma (1999), but I will show that their analysis is not sufficient to account for the Thai cases, and that there is evidence for DPs in Thai. To examine my research question regarding the head- directionality of Thai noun phrases, the head-initial and the head-final analyses will be compared throughout. Chapter 3 considers adjectives in Thai. It is often claimed (Bandhumedha 1983, Savetamalya 1989, Hudak 1990, Warotamasikkhadit 1996) that Thai adjectives are stative verbs, but I will show that this is not the case. I will also discuss three types of noun modifiers in Thai, namely regular attributive adjectives, referential adjectives and nominal modifiers and show that the noun modifiers in Thai can be generated in two distinct positions. Chapter 4 concerns the analyses of Thai noun phrases modified by the three types of noun modifiers elaborated in Chapter 3. I will show that noun phrases in Thai are head-initial, not head-final. I will argue that the word order facts in the noun phrase are best analyzed in terms of NP movement to the lefi of the classifier or determiner. ‘J'v ,. .'-*-' 1‘35 Chapter 5 presents 1 {515111. Although Thai mot .3: 1e post-verbal modals 't be assumption that cros -33. Therefore. as in is $11th of Thai. Chapter 6 deals VVl 11‘: precede or follow V TT are “01 generated mimics. I will saga: ChElpter 7 conclud and ClaUses in Th Chapter 5 presents the clausal structure of Thai with an introduction of the modal system. Although Thai modals can occur either pre-verbally or post-verbally, I will argue that the post-verbal modals in Thai are base- generated pre-verbally, thus giving support to the assumption that cross-linguistically auxiliaries precede main verbs in any SVO language. Therefore, as in the analysis of the DP, VP movement is shown to operate in the syntax of Thai. Chapter 6 deals with Thai aspect markers. Although aspect markers in Thai can either precede or follow VP, like Thai modals, I will show that the post-verbal aspect markers are not generated pre-verbally like post-verbal modals due to their different characteristics. I will suggest that they are adjoined to functional projections above VP. Chapter 7 concludes the dissertation by presenting the parallelism between noun phrases and clauses in Thai in the respects under investigation. luill protide in th r"renames and pronour he Ltderinite one is NumP o ”'5‘ ,1; a." need as head-initial. 11 Introduction Assuming the DP h. "N! utter: trorn the other strt can as the first elemen' (llakhaw, kin; pla ‘e eat fish 6 ate fish_’ ' my; 10?. linder table Under the table (T111133, Till, fiSh this ‘ IS fish’ 9 '3 Train table that .10 3.8 Thai and CthleSe t, ,nb .33. Te" 31' ' mllarly IO Chin CHAPTER 2 THAI BARE NOUNS I will provide in this chapter the analyses of Thai bare nouns: common nouns, proper names and pronouns. I will argue that the definite bare noun in Thai is DP whereas the indefinite one is NumP. Moreover, I will show that bare nouns in Thai are better analyzed as head-initial. 2.1 Introduction Assuming the DP hypothesis (Abney 1987), Thai noun phrases seem to be different fi'om the other structures in the language in head-directionality. That is, although Thai is an SVO language (1a) with prepositions (1b), the noun phrase in Thai always has the noun as the first element and the demonstrative last as in (2). (1) a. khaw, kinl plaal he eat fish ‘He ate fish.’ b. tay3 to?3 under table ‘under the table’ (2) a. plaa1 nii4 fish this ‘this fish’ b. to?3 nan, table that ‘that table’ As Thai and Chinese are numeral-classifier languages and Thai noun phrases behave very similarly to Chinese noun phrases, the study by Cheng and Sybesma (1999) a Chinese bare nouns wii an in Chinese is C 1P wh strident to account for th. transpose that the Thai t This chapter is 011111 arms in Thai. ln sectit 4W1 analyses of DPs. Se '22enouns. The comparis esechon 5. The chapter i llClassil'iers Thai is an isolatir l‘xt for number. neitht 3T'Sout'l97631216) th I?" are ' m complememg on Chinese bare nouns will be discussed. Although they suggest that the definite bare noun in Chinese is ClP whereas the indefinite one is NumP, I will show that ClP is not sufficient to account for the definite bare noun in Thai, particularly for pronouns. I will thus propose that the Thai definite bare noun is a DP. This chapter is outlined as follows. I briefly discuss in section 2 the uses of classifiers in Thai. In section 3, I present Longobardi’s (1994) and Cheng and Sybesma’s (1999) analyses of DPs. Section 4 deals with the distribution and interpretation of Thai bare nouns. The comparison of the head-initial and the head-final analyses will be given in section 5. The chapter is concluded in section 6. 2.2 Classifiers Thai is an isolating, non-inflecting language, and nouns in the language do not inflect for number, neither singular nor plural. This conforms with the suggestion made by T’sou (1976:1216) that a language may have either classifiers or plural morphemes or they are in complementary distribution if both exist in a language. Also, since Thai lacks an article system, a noun can be interpreted as definite or indefinite. Furthermore, a Thai noun without any context can be ambiguous between mass and count, and generic and existential readings as illustrated in (3). ~ (3) plaal yuu2 nay, 11am4 fish stay in water ‘(A/The) fish is/are in the water.’ The noun phrase plaa, ‘fish’ above can have either a generic or an existential reading with the noun being interpretable as mass or count, definite or indefinite, and singular or plural. ‘1 As the noun 1“ ST" notation. not obit?“ (H “ r'uncretize’ the noun V1»: eientiahzation and relate. 52:20:25 of classifiers “'11:! 71 hi‘ew languages eg. 1 In terms of indix'rdu. tastier languages is ‘trans :naer. A classifier is thus Titififd. In short, “classi‘ ... assumption is not di f1] ‘3 PM] language or a ,2 are mass. Consequentl «nines cannot directlv qr game the nouns as ev As the noun in such a language as Thai is assumed to carry the conceptual denotation, not objects (Hundius and Kolver 1983, Bisang 1993), a classifier then serves to ‘concretize’ the noun via different operations - individualization, classification, referentialization and relationalization (Bisang 1993). The first two are the primary functions of classifiers while the third is the secondary one. The last operation occurs . only in few languages e. g. Hmong so it is not mentioned here. In terms of individualization, Bisang (1993) states that a noun in numeral . classifier languages is ‘transnumeral’. Therefore, it cannot immediately combine with a number. A classifier is thus needed to individuate the head N before the latter can be quantified. In short, “classifiers make nouns countable” (Hundius and Kolver 1983:33). This assumption is not different from Chierchia’s (1998) proposal that in a NP [+arg, -pred] language or a language with a classifier system, all nouns refer to kinds and thus are mass. Consequently, bare nouns are allowed unconditionally as arguments and numerals cannot directly quantify head nouns. Classifiers are required to individuate and singularize the nouns as evidenced in (4) below. (4) a."'plaaI saam, fish three b.*saam5 plaal three fish c. plaa1| saam, tual fish three Cl ‘three fish’ Regarding classification, classifiers express the intrinsic properties or features of the head nouns'. They do not add any information to the noun. They are used to ' This characteristic applies to only the type of classifiers called ‘unit classifier’ which is defined as a ‘classifier which has a special relationship with one or more concrete noun.’ eegofae or denote some ea As a result. the nouns lazirsanee. the classifier hesitate. chain. thread. agitaped objects e. g. rin 3.20 ungrarnmaticality a [Siasool’3 saam, chain three ‘three chains' b’soo'.’3 saam‘ chain three (61 a \t'aeanf saam. ring three ‘three rines’ bfwzean, Saran nng three Realizing this kind “mans consider ther via-:12. Tang (1990.) has I‘heclasifier Phrase (C categorize or denote some features of the nouns such as size, shape, dimension, function, etc. As a result, the nouns sharing the same features generally take the same classifier. For instance, the classifier sen, is used with nouns referring to long flexible objects such as necklace, chain, thread, rope, etc. and way, is used with nouns denoting round, hollow, ring-shaped objects e.g. ring, bracelet, etc. The misuse or wrong selection of classifiers leads to ungrammaticality as demonstrated in (5) and (6). (5) a. soo?, saams sen3 chain three Cl ‘three chains’ b.*soo?, saam, worjl chain three Cl (6) a. waemn5 saarns worj1 ring three Cl ‘three rings’ b.*waeaen, saam5 sen3 ring three Cl Realizing thiskind of relationship between nouns and classifiers, many syntacticians consider them as having Spec-Head or Head-Complement relationship. For example, Tang (1990) has NP as the complement of the classifier (K), and Gao (1994) has the classifier phrase (ClP) in the specifier of NP as illustrated below. (Noss 1964:105). The other three types of classifiers as categorized by Noss (1964); namely, a metric classifier or a measure word, a general classifier and an imitative classifier do not have strict relationship with the nouns. 10 (7) Tang (1990) DP /\~ SPEC ' / D 18162100994) .33.: ’\ (7) Tang (1990) DP /\ SPEC D’ /\ D KP /\ SPEC K’ /\ K NP /\ /\ Num Cl SPEC N’ /\ XP N (8) Gao (1994) NP /\ ClP N’ /\ PossP N’ /\ NP[de] N Another important fimction of classifiers is to identify a unique referent or to denote definiteness of the head noun (Hundius and Kolver 1983, Bisang 1993, Pacioni 1996). This is evidenced by the facts that classifiers never appear in proverbs (Goral 1987) and that the occurrence of the classifier usually suppresses the generic interpretation (Cheng and Sybesma 1999). Not only do Thai classifiers display this function, but those in other languages such as Vietnamese and Burmese do too. The presence of a classifier in Vietnamese indicates definite reference while its absence in Burmese implies indefinite one as exemplified in (9) and (10) respectively (Goral 1987: 14, 23). ll (913mco 5° T“ cat fear 11“" {315 are afm b. con m60 50! Cl cat fear ‘The cat is a? . (101a thaja: lou-jt: sugar needs. . ,_ I Do you \\ ar‘ i b. hintjou that soup dnni- ‘Do you \\ (11‘. As will be discussed are: ' ' ' ....tial. Hou et er. the pres nuisance. a quantified Ilt (ll)arniil torn. have umbrell ‘There are thi b.‘duul rom- look timbre Look at the Inc general. a quantii ”sins ' .auied by a demonstr display the definite (ITBTduUI tom 1:): uIrrb-rella at thESe ”3,1 duu look Too 1 thtm3 umbrella at the tth Signs ' Jllmg Up, nOum S 'I‘T'T in [h if e bETng arm :4 (9) a. meo so chuot cat fear mouse ‘Cats are afraid of rrrice.’ b. con meo so chuot Cl cat fear mouse ‘The cat is afraid of mice.’ (Vietnamese) (10) a. thaja: lou-jin-dhala: sugar need-want-actual mode-question marker ‘Do you want sugar?’ b. hin:jou thau’hcin-dhala: soup drink-want-actual mode-question marker ‘Do you want some soup?’ (Burmese) As will be discussed, the ‘referentialization’ function of classifiers in Thai is very essential. However, the presence of a classifier does not guarantee the definite reading. For instance, a quantified noun phrase with a classifier is always indefinite. (11) a. mii, rom, saam5 khan, yuu2 bon, to? 4 have umbrella three Cl stay on table ‘There are three umbrellas on the table.’ b.*duu, rom3 saam, khan, si2 suay, car), look umbrella three Cl PRT pretty very ‘Look at the three umbrellas. They are very pretty.’ In general, a quantified noun phrase exhibits the definite interpretation when it is accompanied by a demonstrative as in (12). A quantified noun phrase in Thai, however, can also display the definite reading when an adjective follows the classifier as in (13). (12) duu, rom3 saamS khan, nii4 si2 suay5 car), look umbrella three Cl this PRT pretty very ‘Look at these three umbrellas. They are very pretty.’ (13) duu, rom3 saam, khan, khiawS si2 suay, can, look umbrella three Cl green PRT pretty very ‘Look at the three green umbrellas. They are very pretty.’ Summing up, nouns in a classifier language like Thai are kinds which are not restricted in their being arguments. However, classifiers are needed to individualize them 12 her her are quantified. h. citing them a definit. Lllreyious studies The two relevant s: Sybesma‘s (1999). The Tori 3523.35 whereas the latte :aieinrhesame line as C L“inrigobardi. Therefore. 13.1 Longobardi (1994) Longobardi tries tc Sided (they can only a; $32 an existential inter He accounts for th rial interpretatig when they are quantified, and in some instances to identify as well as referentialize them, thus giving them a definite interpretation. 2.3 Previous studies The two relevant studies on bare nouns are Longobardi’s (1994) and Cheng and Sybesma’s (1999). The former concerns noun phrases in Romance and Germanic languages whereas the latter is about Chinese bare nouns. Although my analysis will be made in the same line as Cheng and Sybesma, they develop some of the ideas proposed by Longobardi. Therefore, both studies are to be introduced. 2.3.1 Longobardi (1994) Longobardi tries to account for the fact that bare nouns in Italian are syntactically restricted (they can only appear in object position) and semantically restricted (they only receive an existential interpretation). He accounts for these facts in the following way. DPS with empty heads receive an existential interpretation by default and empty heads must be lexically governed. The fact that (14a) below is ungrarnmatical, therefore, is because the noun phrase acqua ‘water’ occurs in a non-lexically governed position or in this case a pre-verbal position which lacks a lexical element that can function as the governor of the empty head D. (14) a.*Acqua viene giii dalle colline. (Italian) water comes down from the hills. b. Viene gir‘r acqua dalle colline. comes down water from the hills. c. Ho preso acqua dalla sorgente. I took water from the spring. 13 Since proper name panties the following ltal Dnot'ernent as substitutio (liiall mio Gr. The my Gi. beio Gianni My Gianni C. Gianni mio Gianni my, The grammaticality anesare filled - by the dc Ez-D raising past the genitir msented in (16) and ( l~ ITS 3“ emutt- D 116) I: if th ('17) s Gila; in e 011]), PTOpe, Since proper names are definite, the empty D is illegitimate. Longobardi then provides the following Italian examples to show that there occur instances of overt N-to- D movement as substitution in Italian. (15) a. ll mio Gianni ha finalmente telefonato. (Italian) The my Gianni finally called up. b.*Mio Gianni ha finalmente telefonato. My Gianni finally called up. c. Gianni mio ha finalmente telefonato. Gianni my finally called up. The grammaticality of (15a,c) is due to the fact that the D positions of both proper names are filled - by the definite article 1'] in (15a) and by the noun Gianni in (15c) via N- to-D raising past the genitive which is assumed to be in the specifier position as represented in (16) and (17) respectively. Sentence (15b) is unacceptable because it contains an empty D. (l 6) DP /\ D NP i1 /\ the mic N’ my 1 N Gianni (l 7) DP /\ D NP Gianni,- /\ mio N’ my I N ti Since only proper names can raise to D and generic DPs are names and kinds (Carlson 1977), generic DPs will require an overt definite article as in (18). 14 118‘) a‘Cas'tori cc beavers b M castori the beavers English bare plural ntnpretation and distribu (19) a. Beavers bu b. I foundl 1t Longobardi sugge T22 facts follow from th stated and the distribut Evidence that En Edited proper names t 333111335 in (20)_ CD) a_ Old John bfjohn 01C 131 T Cheng and Si‘bes 2:2, T“ 35 Ptel (18) a.*Castori costruiscono dighe (Italian) beavers build dams b. I castori di questo tipo non costruiscono mai dighe. the beavers of this sort never build dams English bare plurals, in contrast to Romance bare plurals, are unrestricted in their interpretation and distribution as demonstrated in (19) below. (19) a. Beavers build dams. b. I found/I love good wine and fresh oranges. Longobardi suggests that in languages like English, N—to-D raising occurs at LF. Two facts follow from this: both generic and existential readings of bare plurals will be allowed and the distribution of bare plurals will be syntactically unrestricted. Evidence that English lacks overt N-to-D raising comes fiom the fact that modified proper names do not require a determiner and the adjective appears pre- nominally as in (20). (20) 3. Old John came in. b.*John old came in. 2.3.2 Cheng and Sybesma (1999) Developing Longobardi’s ideas of the lexical government restriction of an empty head as well as the N-to-D movement for the definite interpretation, Cheng and Sybesma compare noun phrases in Mandarin and Cantonese, and account for their distribution together with interpretation in syntactic aspects as follows. 2.3.2.1 Wm Cheng and Sybesma propose that bare nouns in Chinese are not simply NPs. Those with definite interpretations are Classifier Phrases (ClPs hereafter) while those receiving indefinite interpretations are Numeral Phrases (NumPs hereafter). They 15 compare the distribution and interpretations of bare nouns in Mandarin and Cantonese. Their finding regarding the differences between the two in terms of interpretations is shown in (21) (1999:528). (21) Mandarin Cantonese Bare noun + + + - c1+ N + - + + Numeral + C1 + N + - + _ Unlike Mandarin (22a), Cantonese cannot use bare nouns to express definiteness (22b).2 The [C1+N] construction must be used (22c). (22) a. gou yao guo malu (Mandarin) dog want cross road ‘The dog wants to cross the road.’ b.*gao soeng gwo maalou (Cantonese) dog want cross road ‘The dog wants to cross the road.’ c. zek gau gamjat dakbit tengwaa Cl dog today special obedient ‘The dog is specially obedient today.’ Similar to D (Longobardi 1994), a classifier is claimed to have the deictic function of individualization and singularization (Bisang 1993, Croft 1994), which enables it to identify or pick out a single individual from what is described by an NP. The 2 However, both Mandarin and Cantonese can use bare nouns to express generic readings as shown in (i) and (ii) respectively. (i) wo xihuan gou (Mandarin) I like dog 'I like dogs.‘ (ii) ngo zung-ji gau (Cantonese) I like dog 'I like dogs' 16 .mce of Cl is thus beli hesitate. suggest that the 2.301220 is a C IP as s (131 Bare nouns with t however. the two lanuua tannese bare nouns ca - nhi a re. the C l-insertic ass 3 .2. tan denote definite reflheads in Cantone p ,m-z .I.'.uently’. unlike Ita' 1! ‘~-.. Nari K‘n . are not restrict “at , - Pd obiect positic ‘zvl presence of Cl is thus believed to yield definiteness in Cantonese. Cheng and Sybesma, therefore, suggest that the [Cl+N] construction in Cantonese which expresses a definite reading (22c) is a ClP as shown in (23). (23) ClP /\ Cl ‘ NP 1 N Bare nouns with definite readings in Mandarin are also claimed to be ClPs. However, the two languages use different operations to display definiteness. Since Cantonese bare nouns cannot carry a definite interpretation, the language uses the default option i.e. the Cl-insertion (=[Cl+N] construction). Mandarin bare nouns, on the other hand, can denote definiteness so the language makes use of the N-to-Cl movement at LF. The C1 heads in Cantonese and Mandarin are thus filled overtly and covertly respectively. Consequently, unlike Italian bare plurals, bare definite noun phrases in Cantonese and Mandarin are not restricted to only lexically-govemed positions; they can occur both in subject and object positions. The quantified noun phrase or the ‘numeral + C1 + N’ construction, on the other hand, can receive only an indefinite reading. Therefore, a numeral is assumed to bring about indefiniteness. It undoes the definiteness expressed by ClP. Accordingly, Cheng and Sybesma propose that indefinite noun phrases (N, [Cl+N], and [Num + C1 + NJ) are NumeralPs as shown below: 17 A bare noun ‘ apnNumeral head 1y an empty Nume ch as indefinite is 5:5381 (35) The ind: of a Nur Due to the er «.0: Occur in the 1 (36:11 gou _\' dog it ‘The t NOT: - 26k g; 0 dt ‘The ( XOT: 1“ Short, f 11 T“ 0an rm 1 "stn (24) NumP Cl NP A bare noun with an indefinite reading thus contains an empty Cl head and an empty Numeral head. A [Cl+N] phrase with an indefinite reading, on the contrary, has only an empty Numeral head. The fact that Mandarin [Cl+N] phrases can be interpreted only as indefinite is accounted for by the language specific restriction stated as follows (p.528). (25) The indefinite interpretation of nominals in Chinese is linked to the presence of a NumeralP (the head of which may be overt or nonovert). Due to the empty Numeral head, indefinite bare nouns (N and [Cl+N]) in Chinese cannot occur in the preverbal position as illustrated below. (26) a. gou yao guo malu (Mandarin) dog want cross road ‘The dog wants to cross the road.’ NOT: A dog wants to cross the road. b. zek gau zung-ji sek juk (Cantonese) Cl dog like eat meat ‘The dog likes to eat meat.’ NOT: Dogs like to eat meat. In short, following Longobardi’s (1994) claim that a nominal expression can be an argument only if its D position is filled, definite and indefinite bare nouns in Chinese can be arguments only if Cl and Numeral are filled respectively. 18 Since proper nami 25ml"). they are beheV Bess-generated in N, “th (37) 2t guojing sh'. Guojing sa ‘Guojing s; b. nei-ge hu that-Cl ll- ‘That Hufc An account of thc 33-53.: Longobardi WhO K ieguges. Cheng and 3.) hemmed N. the noun 1. timbution. Similar to Germa .3: at LF. the N40‘ v y. . ‘h ‘ a PTOper name is F 2.3.2.2 W25 Since proper names in Chinese can take classifiers, numbers, and demonstratives as in (27), they are believed to be structurally like other common nouns. That is, they are base-generated in N, which is consistent with the claim for proper nouns in other languages. (27) a. guojing shuo ta kandao-le liang-ge hufei (Mandarin) Guojing say he see-LE two-Cl Hufei ‘Guojing said that he saw two Hufei’s.’ b. nei-ge hufei zhen bu xianghua! that-Cl Hufei truly not decent ‘That Hufei is really unreasonable! ’ An account of the definite interpretation of a proper name is given in the same line as Longobardi who proposes the N-to-D movement for the Romance and Germanic languages. Cheng and Sybesma, therefore, propose the N-to-Cl raising. When Cl is filled by a moved N, the noun phrase is definite in terms of interpretation and free with respect to distribution. Similar to Germanic languages such as English in which the N-to-D raising operates at LP, the N-to-Cl movement in Chinese is covert as seen in the example below where a proper name is preceded by an adj ectival modifier. (28) wo xihuan xiao Hufei I like small Hufei ‘I like little Hufei.’ 2.3.2.3 Pronouns Cheng and Sybesma assume that pronouns in Chinese are base-generated in N and then N moves to Cl to receive a definite interpretation and to be free of the distributional restriction. Their assumption is drawn from the fact that Chinese pronouns behave 19 enilarly to proper name .333] and. or a demon (29') eong nei-ge from that-C ‘From that n Their analysis of ‘3} many linguists that p mtg others). Since th. - the same role as D 23.".me in D. Howex'e roam. Since Thai xps . 4: same questions can ‘ similarly to proper names; that is, they are definite and can be used with a classifier, a numeral and/or a demonstrative as demonstrated below: (29) cong nei-ge jing-zi, wo keyi kandao wu-ge wo (Mandarin) from that-Cl mirror I can see five-Cl I ‘From that mirror, I can see 5 copies of myself (five I’s/me’s).’ Their analysis of the pronoun is obviously contrary to what is generally claimed by many linguists that pronouns are base-generated in D (Postal 1969, Longobardi 1994, among others). Since they imply that D is not needed in classifier languages because Cl plays the same role as D, it would be inconsistent for them to suggest that a pronoun is generated in D. However, a pronoun cannot be generated in C1 either because the two can co-occur. Since Thai NPs are also bare, and Thai and Chinese are similar in many respects, the same questions can be asked about NPs in Thai: are there syntactic restrictions on indefinite NPs and do pronouns behave like nouns in Thai? 2.4 Distribution and interpretation of Thai bare nouns This section deals with the distribution and the interpretation of the three types of bare nouns in Thai: common nouns, proper names and pronouns. I will show that none of them are restricted to lexically-governed positions although they exhibit similar interpretations to those in Chinese. 2.4.1 Common nouns I will discuss here bare nouns with the definite and the indefinite interpretations as well as the [N+Cl] construction in Thai which displays only an indefinite reading. 20 2.4.1.1 Wm lhai bare commo The can also occur met (30) a khaw. pay he go ‘He went I b. khaw, cho. he llki ‘He likes c (31) 3. mar. Chgg dog hke ‘Dogs like b- maa. nooi dog sleet ‘The dog C: "13% Rat. dOg bite ‘A dOg bl ‘ 1 .‘11‘ ' The [N‘Cl] COns a... Is it can recej‘. e o \i \LiZ‘I} . ‘0‘] IS restricted C“ A-) 1 H1335 gIVQ (mE) FOu No . b-Tbu T. Cal 2m. ~ ~ I flgarene Ci gare 2.4.1.1 Barmmmounmms Thai bare common nouns can be interpreted as definite, indefinite and generic. They can also occur pre-verbally and post-verbally as demonstrated below. (30) a khaws payl sii4 narjssiis he go buy book ‘He went to buy a book/ books/ the book/ the books.’ b. khaws cheap3 maaS he like dog ‘He likes dogs.’ (31) a. maas cheap3 kinl niia4 dog like eat meat ‘Dogs like to eat meat.’ b. maaS noon, yuu2 tay3 to?4 dog sleep IMPF under table ‘The dog is sleeping under the table.’ c. maa5 kat2 dek2 dog bite child ‘A dog bit a child.’ In brief, unlike Chinese, there is no restriction in the distribution of an indefinite bare noun in Thai. 2412 Won The [N+Cl] construction in Thai behaves exactly like Mandarin, not Cantonese. That is, it can receive only an indefmite non-specific singular interpretation (32a) and its distribution is restricted to the post-verbal position (32b). (32) a. khoos buzrii2 muanl give (me) cigarette C1 ‘Can you give me a cigarette?’ NOT: Can you give me the cigarette/ cigarettes? b.“'bu,rii2 muanI tok2 caak2 kralpaw5 khaws cigarette Cl fall from pocket he ‘A cigarette fell from his pocket.’ 21 It must be mentic generally follows verbs l (33) a. (101335) (give let 11 ‘May I ha ‘Can You b. (Rhos) (give let r ‘May 1 sn ‘Can you 2.42 Proper names like all languag their distribution. An _ . name interpretat ‘ireeil -......tier. 3 number and (34‘) awo Zuoti yeste It must be mentioned that this type of construction is very limited in use. It generally follows verbs expressing or implying request as demonstrated below.3 (33) a. (khaas) kin, som3 luuk3 (give/let me)eat orange Cl ‘May I have an orange?’ or ‘Can you give me an orange to eat?’ b. (khaas) suup2 buzrii2 muanl (give/let me) smoke cigarette Cl ‘May I smoke a cigarette?’ or ‘Can you give me a cigarette to smoke?’ 2.4.2 Proper names Like all languages, Thai proper names are definite in their interpretation and free in their distribution. And sirrrilar to Chinese, some common nouns can be used with a proper name interpretation as in (34), and Thai proper names can be combined with a classifier, a number and/or a demonstrative as in (35)-(36). (34) a.wo zuotian peng-shang—le laoshi (Mandarin) I yesterday bump-up-LE teacher ‘Yesterday, I bumped into Teacher.’ b. khruu1 may3 maal (Thai) teacher not come ‘Teacher didn’t come.’ (35) a. guojing shuo ta kandao-le liang-ge hufei (Mandarin) Guojing say he see-LE two-Cl Hufei ‘Guojing said that he saw two Hufei’s.’ 3 What is interesting is that the noun and the classifier in such a construction can be intervened by a verb as in the following examples. (i) a. khaa, sorn3 kinl luuk3 give (me) orange eat Cl ‘Can you give me an orange to eat?’ b. khaas buzrii2 suup2 muan, give (me) cigarette smoke Cl ‘Can you give me a cigarette to smoke?’ This phenomenon is similar to Japanese (Miyakawa 1988). I will not consider it here, though. ' 22 | b. lek. book: Lek tell C ‘Lek said I (36) a nei-ge h” that-Cl HL‘ ‘That Hull” b. dam1kh0n. Dam C 1 ‘That Dam When a proper na irisinite deSpite the f act atcnpanied by a demor 2.43 Pronouns Pronouns in Thai socal hierarchies ii mm of Pronouns user harm, Senior 333mg can be used [0 5M) '~ hILh one to be Us e if; $27, ,ijI; b. lek4 baak2 waa, khaw5 hen, darn1 sarjs khonI (Thai) Lek tell COMP he see Dam two Cl ‘Lek said that he saw two Dam’s.’ (3 6) a. nei-ge hufei zhen bu xianghua! (Mandarin) that-Cl Hufei truly not decent ‘That Hufei is really unreasonable!’ b. darnl khon1nan, may, dii, (Thai) Dam Cl that not good ‘That Dam is not good.’ When a proper name co-occurs with a numeral and a classifier, the sole reading is indefinite despite the fact that a bare proper name is generally definite. A proper name accompanied by a demonstrative is specific. 2.4.3 Pronouns Pronouns in Thai are classified into three major groups according to the three main social hierarchies in Thailand: the Royalty, Buddhist monks and commoners. The choices of pronouns used among the commoners also depend on such factors as social status, familiarity, seniority (in rank and age), and formality. For example, the following pronouns can be used to address the second person: than ,, khun,, thaa,, raw ,, kcew,, miy,, etc. Which one to be used depends on the factors above. To illustrate the point I am going to make, however, I will present a small set of pronouns used by commoners in everyday conversation in the most general situation as below.‘ ‘ The third person singular pronoun khaw, can have a plurality meaning when it is used as an impersonal pronoun as in (i). (i) khaw5 waa, kan1 waa, thii, nan, naaw5 maak, he say RECP COMP place there cold much 23 ‘ \ xi ,. person As shown in (37.). :aiecii'e noun phuak. ‘21 Indifferent form than t neural. (37) Singular mural 1“ person phom, (male) raw, chan5 (female) phuak,-raw, 2nd person khun,(male, female) phuak,-khun, thaa,(male, female) phuak,-thaa, 3'“ person khaw5 (male, female) phuak,-khaw5 man, (=it) phuak,-man, As shown in (37), plural pronouns are usually formed by the addition of a collective noun phuak, ‘group’ to the singular forms. Since the first person plural pronoun has a different form than the singular counterpart, the addition of phuak, ‘group’ is optional. Pronouns in Thai, as in Chinese, behave very similarly to proper names regarding their ability to co-occur with a classifier, a number and/or a demonstrative as in (3 8). This fact leads Cheng and Sybesma to the claim that in Chinese they are also base-generated in N, and then move to Cl for the definite interpretation. (38) a. khaw, hens raw, saarn, khon, he see we three Cl ‘He saw us three/the three of us.’5 b. chan5 may, chaap, khaw, khon, nan4 I not like he Cl that ‘*I don’t like that him.’ ‘They said that it was very cold there.’ 5 The phrase raw, saam, khan, ‘we three Cl’ can also be interpreted as ‘three of us’. However, the phrase with a partial interpretation is pronounced with a comma intonation while that with a definite reading is not interrupted by a pause. 24 However, 'lhai pi an (39,) and (40) respei flab) or definite (39b. (39) a “10“! a" people T1” ‘Thai p60 b. khaws me he no ‘Hemdn (.40) a- dam: kh‘ Dam he ‘Dam <10 b. khun, til you hll ‘“hydm A number of st 11..-: rteoneet. First, an it elements. If the W However, Thai pronouns can also co-occur with a common noun or a proper name as in (39) and (40) respectively. The interpretation of such a construction is either generic (39a,b) or definite (39b, 40). (39) a. khon, thay, raw, chaap, ?i,sa,ra2 people Thai we like freedom ‘Thai people like freedom.’ b. khaw, may, ruu4 waa, maa, man, du2 he not know COMP dog it fierce ‘He did not know that the dog is fierce.’ (40) a. dam, khaw, may, yaak2 maa, Dam he not want come ‘Dam does not want to come.’ b. khun, tii, dam, khaw, tham,may, you hit Dam he why ‘Why did you hit Dam?’ A number of studies claim that such a phrase is a topic construction (Noss 1964, Bandhumedha 1983 among others). Nonetheless, there are reasons to believe that this is not correct. First, an intonation break or comma intonation does not occur between the two elements. If the two are pronounced with a break, the preceding common noun or proper name is focused. But the construction under discussion does not express focus or emphasis. Secondly, such a construction can occur both pre-verbally and post-verbally as in (40).Another characteristic of a topic construction, besides the intonation break, is that the topic particles such as m, nan), rii,, and nan,rii, can be inserted after the topic element (Ekniyom 1977) as in (41). But the insertion of the topic particle in (39a) and (40b) as shown in (423) and (42b) respectively yields ungrammaticality, supporting the idea that the first item is not topicalized or focused. 25 (41)a dami “33 Dam TOP ‘Dam, he ( b. khaws ma} he I101 ‘He did nC olivkhon. th- people Ii bfkhun, Iii you hit More important] 1:43). a fact that point: (43) a. khon, th. people Tl ‘We Thai b. khun, tii. - you hit ‘It is Dam It is also not like 2‘\ {r' . ._-331.1te noun phrase (41) a. dam, na2 khaws may, yaak2 maa, Dam TOP he not want come ‘Dam, he does not want to come.’ b. khaw5 may, ruu4 waa, maaS na2 man, du2 he not know COMP dog TOP it fierce ‘He did not know that the dog is fierce.’ (42) a.*khon, thay, na2 raw, chaap, ?i,sa,ra2 people Thai TOP we like freedom b.*khun, tii, darn, na2 khaw, tham,may, you hit Dam TOP he why More importantly, the topic particle can be inserted afier the whole noun phrase as in (43), a fact that points to the idea that such a construction behaves as a unit. (43) a. khon, thay, raw, na, chaap, ?i,sa,ra2 people Thai we TOP like freedom ‘We Thais like freedom.’ b. khun, tii, dam, khaw5 na2 may4 chay, khaaw, you hit Dam he TOP not true Khaw ‘It is Darn that you hit, not Khaw.’ It is also not likely that the construction above is a kind of apposition because the appositive noun phrase in Thai displays a different word order; that is, a pronoun precedes a common noun or a proper name as in (44). (44) a. raw, khon, thay, rak4 ?i,sa,ra2 we people Thai love freedom ‘We Thais love freedom.’ b.‘phuak,thaa, nak4rian, tan, kha,yan5 group-you student must diligent ‘You, students, must be diligent.’ c. thaa, dam, tar], maa, you Dam must come ‘You, Dam, must come.’ Last but not least, lack of number agreement between the pronoun and the preceding common noun or proper name as in (45) below indicates that such a structure may not be of appositive type. 26 (45) a. mafia m3“ dog it ‘The dogs b.‘maa, phu.; dog grou This piece ofevid apronoun in Thai is not g casidering more data. \\ spelled out. 15 Analysis of Thai bat Hating outlined t Ins in terms of distribi Ins in this section. Th ‘41- in Thai free in diSi Einsinuate) ., '9 and J)a1 (45) a. maa, man, nii, pay, larJ,caak2 kat2 kan, laezew, dog it escape go after bite RECP PRF ‘The dogs ran away afier fighting.’ b."‘maa5 phuak,-man, nii4 pay, lan,caak2 kat2 kan, larew, dog group-it escape go after bite RECP PRF This piece of evidence will be used for my argument in the following section that a pronoun in Thai is not generated in N, but D. Chinese pronouns may also be the same, considering more data. What’s more, the requirement of DP for Thai noun phrases will be spelled out. 2.5 Analysis Of Thai bare nouns Having outlined the similarities and differences between Thai and Chinese bare nouns in terms of distribution and interpretation, I will provide the analyses of Thai bare nouns in this section. The questions to be investigated are 1) why are indefinite bare nouns in Thai free in distribution? 2) is ClP adequate for the analysis of definite bare nouns in Thai?, and 3) are Thai NPs head-initial or head-final? Since nouns are always the first elements in Thai noun phrases, I will first discuss the head-final analysis. However, I will show that the head-initial analysis is preferred because it needs only one syntactic operation, namely NP movement, to account for all the cases to be investigated while the head-final analysis needs three. This finding gives conceptual support to treating all structures in Thai as head initial. 2.5.1 Head-fmal analysis In this section I will show how the head-final analysis can handle the instances of common nouns, proper names and pronouns, but only by positing three distinct syntactic 27 nefarious: covert N ra .2. Num as right adjun Iwill first discu fairiese, and then the it it he distribution restri rec-specific singular in ie‘lt with. iliDefinite bare commt Following Cheng it. mar to Chinese. there «covenlymove to Cl fr H6) a maas “Dar dog $166; operations: covert N raising, overt Cl-to-Num movement and overt Cl-to-D movement via Num as right adjunction. 2.5.1.1 W I will first discuss the Thai definite bare noun which behaves similarly to that in Chinese, and then the indefinite bare noun in Thai which, unlike in Chinese, is not subject to the distribution restriction. The [N+Cl] construction which can express only indefinite non-specific singular interpretation and is restricted to only post-verbal positions is also dealt with. (1) Definite bare common nouns Following Cheng and Sybesma, the definite bare noun is embedded in a ClP. Similar to Chinese, therefore, the Thai definite bare common noun maa, ‘dog’ in (46) has to covertly move to CI for the definite interpretation and the unrestricted distribution. (46) a. maa, naan, yuu2 taay, to?, dog sleep IMPF under table ‘The dog is sleeping under the table.’ b. ClP NP Cl (2) Indefinite bare common nouns Indefinite bare common nouns in Thai are different fi'om those in Chinese in terms of their unrestricted distribution. That is, Thai indefinite bare common nouns can occur in both pre-verbal and post-verbal positions. According to Cheng and Sybesma, the 28 ndefiriite bare n01 5:;9' heads Nurn in a subject positic :5.ch .ite bare nou sane line of argurn fie indefinite bare l EIeipretation is net indefinite bare noun in Chinese is NumP containing the empty head Num ([Cl+N]) or the empty heads Num and Cl (bare noun), thus accounting for the fact that they cannot occur in a subject position due to the lack of lexical governor. The free distribution of the indefinite bare noun in Thai thus indicates that the head Num must be filled. Taking the same line of argument as the N-to-Cl movement, I suggest the N-to-Num movement for the indefinite bare noun in Thai. As the number is not explicitly present, the number interpretation is neutral. The syntactic representation of the indefinite bare noun maa, ‘dog’ in (473) is portrayed in (47b). (47) a. maas kat2 dek2 dog bite child ‘A dog/dogs bit a child/children.’ b. NumP /\ ClP Num /\ NP Cl | N maa, dog (3) [N+Cl] construction Similar to Mandarin, Thai also utilizes the [N+C1] construction to display an indefinite non-specific singular noun. But unlike indefinite bare nouns in Thai, the [N+Cl] construction is restricted in its distribution and the verbs it can be used with. As it can occur only in a post-verbal position, it seems to imply that it is subject to the lexical government restriction and the phrase itself must have an empty head. Since such a construction can carry only the indefinite reading, it should be embedded in NumP. The 29 bm-final analysis can h. piresc burn: muan, ‘cig; (48) a. lthaaS l‘ give (me) .' ‘C an you g VIE head-final analysis can handle this phenomenon very well as shown in (48b) for the phrase buzrii, muan, ‘cigarette C1’ in (48a). (48) a. khaa, bu2rii2 muan, give (me) cigarette Cl ‘Can you give me a cigarette? b. NumP /\ C 1P Num /\ NP C1 | muan , N buzrii, cigarette There is no need fora movement in the head-final analysis as shown in (48b). The fact that Num is empty explains the restricted distribution and the non-specific interpretation of the phrase. Summarizing, the head-final analysis of the definite bare noun in Thai is consistent with Cheng and Sybesma’s analysis of the Chinese definite bare noun; that is, covert N—to-Cl movement is needed. However, unlike the Chinese indefinite bare noun, covert N-to-Num raising is required for the analysis of the indefinite bare noun in Thai. 2.5.1.2 Empernames As suggested by Longobardi (1994) and Cheng and Sybesma (1999), proper names are generated in N. Since the proper name generally denotes definiteness, it is embedded in the ClP. The definite reading is thus acquired when N raises to Cl. Therefore, the prOper name dam , ‘Dam’ is represented by the head-final analysis as in (49) below. 30 (49> Howey .32 say, 1730) (5.0) lel Le ‘L Asthe I C .. V. I H dad for SL‘ T9335 an on tifilore SLY'UC' Tamas TIE '{h “a (49) ClP /\ NP Cl l N dam, However, when a proper name is accompanied by a classifier and a number like dam, say, khan, ‘Darn two C1’ in (50)(=3 5b), an additional movement is needed. (50) lek4 baak2 waa, khaw, hen, dam, sarj, khon, Lek tell COMP he see Darn two Cl ‘Lek said that he saw two Dam’s.’ As the presence of a numeral gives an indefinite interpretation, NumP is to be assumed for such a phrase. To capture the word order fact, the head-final analysis requires an overt raising of C1 to Num as right adjunction. The noun phrase in (50) is therefore structurally represented as below. (51) NumP /\ ClP Num /\ /\ NP C1 Num Cl 1 ti 5905 khonu N two Dam, l T I am assuming here on semantic and syntactic grounds that Num selects ClP, not vice versa. 31 The Cl-tO-Num m the fact that a claf level form a unit.° In brief. to hant N-to—Cl raising like the er hand. are capture 2.5.1.3 P19391335 As discussed ea ces and common no tasters Therefore, t never. that Thai pro iterated in D The Cl-to-Num movement as adjunction in the head-final analysis may result from the fact that a classifier and a numeral are so closely related that they should at one level form a unit.‘5 In brief, to handle bare proper names in Thai, the head-final analysis needs covert N-to-Cl raising like the account of definite bare nouns. Quantified proper names, on the other hand, are captured by overt Cl-to-Num movement as right adjunction. 2.5.1.3 Pronouns As discussed earlier, Cheng and Sybesma treated pronouns on a par with proper names and common nouns owing to their ability to co-occur with numerals and classifiers. Therefore, they claim that a pronoun is base-generated in N. I will argue, however, that Thai pronouns are like those in many other languages in that they are generated in D. First of all, what Cheng and Sybesma seem to miss is that the interpretation of a quantified pronoun can also be definite as in (52)7, unlike the quantified proper name or common noun which expresses only indefinite reading as in (53).8 6 A number of studies consider the close relationship of a classifier and a numeral by proposing that they are one constituent (Crofi 1994:151) or a dual head (Tang 1990:403). 7 It is worth noting that the definite quantified pronoun in Chinese in (52a) exhibits a different word order than the indefinite quantified proper names, common nouns and pronouns as in the example given by Cheng and Sybesma in (29), repeated below. (i) cong nei-ge jing-zi, wo keyi kandao wu-ge wo fiom that-CL mirror I can see five-CL I ‘From that mirror, I can see 5 copies of myself (five I’s/me’s). I was told by Liou Yu-fen (p.c), though, that the pronoun ‘I’ in the example above is used as a common noun, not in the pronoun sense. This is not surprising, given that such an expression cannot be constructed in Thai with a pronoun. The equivalent translation of the above sentence in Thai is as below. 32 (52) a. “'O'm“ we ‘IhC [hrt‘ b. Tawi $38 We thr ‘the thre (53) a. hang—EC two-Cl ‘two HU b. dami 53‘ Dam th: ‘three D Two things that prises and pronouns a 33:2 phrase is C IR hOi mate interpretation? Ied:d for the locus of :33 an indefinite read .isonsiole for definitei neat to handle fight-$61381. - b0( .T Watt ’ “Tic ‘J-‘élf (52) a. women san ge (Mandarin) we three C1 ‘the three of us’ b. raw, saam, khon, (Thai) we three C1 ‘the three of us’ (53) a. liang-ge hufei (Mandarin) two-Cl Hufei ‘two Hufei’s’ b. darn, saarn5 khon, (Thai) Dam three Cl ‘three Darn’s’ Two things that need reconsidering are thus (1) the suggestion that definite noun phrases and pronouns are ClPs, and (2) the proposal that pronouns are Ns. If the definite noun phrase is ClP, how is it possible to account for the quantified pronoun with a definite interpretation? The quantified pronoun cannot simply be ClP because NumP is needed for the locus of the numeral. However, it cannot be NumP either because NumP gives an indefinite reading. Another maximal projection higher than NumP and responsible for definiteness is required. The most likely candidate is DP. Therefore, ClP is insufficient to handle the definite bare noun in classifier languages. DP is still needed. Moreover, if a pronoun is a N like a proper name, the pronoun in a quantified pronoun has to move to D past Cl and Nurn so that the definite interpretation is obtained. The instance of a quantified pronoun thus leads to the assumptions that DP is needed for definite noun phrases and that a pronoun is generated as a D. (ii) chan, hen, tua, chan, haa, tua, nay, lera,cok2 I see body-I five C1 in mirror. The classifier khan, which is used with human beings cannot be used as the classifier here either, indicating that what is talked about is the image or the copy of myself, not me myself. 33 To a arse-step n dlus‘med in (54) To account for the quantified pronoun as in (52b), the head-final analysis requires a two-step movement as right adjunction, namely, Cl-to-Num, and [Cl+Num]-to-D as illustrated in (54). (54) a. DP /\ NumP D /\ raw) ClP Num we /\ /\ NP Cl Num Cl t, saam, khan ,, i three + b DP /\ NumP D /\ /\ ClP Num D Num, /\ t) raw! /\ NP C] we Num Cl t, saam, khan ,, three The movement of the complex head Nurn (Num+Cl) to D as in (54b) may be suggested to result from the feature checking of the pronoun and the number with the classifier as the incompatibility of the three leads to ungrammaticality as in (55).9 (55) afraw, nil], khon, we one Cl b.*raw, saam, tua, we three Cl ' The quantified proper name expresses definiteness when accompanied by a demonstrative. 9 Due to the complex pronoun system in Thai as mentioned in section 2.4.3, the examples in (55) are acceptable if the pronoun raw, ‘we’ is used to mean ‘I’ and the classifier tua, is used with the contempt denotation. 34 Anoth‘ drawn from th discussed in T (56) a. As corr :N cross-lintI 1., Therefore .Eiiough C l is traded the fact tie that the prc ital analysis dt 1565) a shot: if") Another piece of evidence supporting the idea that a pronoun is generated in D is drawn fi'om the fact that pronouns can co-occur with common nouns or proper names as discussed in the previous section with some examples repeated in (56) below. (56) a. khon, thay, raw2 chaap, ?i,sa,ra2 people Thai we like freedom ‘Thai people like freedom.’ b. khun, tii, darn, khawS tham,may, you hit Darn he why ‘Why did you hit Dam?’ As common nouns and proper names are generally assumed to be base-generated in N cross-linguistically, pronouns cannot originate in such a position, considering these data. Therefore, the suggestion that a pronoun in Thai is D seems to be on the right track. Although Cl is claimed to perform the same function as D and the latter may not be needed, the fact that a pronoun can also co-occur with a classifier and a numeral bars the idea that the pronoun is base-generated in C]. With the pronoun generated in D, the head- final analysis does not require any movement at all for the phrase dam, khaw, ‘Dam he’ in (56b) as shown in (57) below. (5 7) DP /\ NP D I khaw, N he dam, Dam It should be noted here that the fact that the interpretation of the phrase consisting of a pronoun and a common noun or a proper noun is either generic or definite can be explained by the filled D. In these instances, however, the pronoun which occupies the D position seems not to really denote the deictic function of pronoun i.e. referring to an 35 mjgy or entities. pm, tho",- thu Ewen-C 581158 am To sum u siience for the In also needs 2 ton and the def define bare nou Allin all. mmmth I Iijunction for W as ad junctic 35.2 Head-in iti: In this sec 4 analysis bec . 1m estigatit 252.1 BTilQL‘; entity or entities. Its function is pretty much like that of a determiner.Io That is why the phrase khan, thay, raw, ‘people Thai we’ in (56a) is interpreted as ‘Thai people’ in generic sense and the meaning of the pronoun ‘we’ is not really included. To sum up, I have shown that Thai can provide another piece of supporting evidence for the claim that a pronoun is generated in D, and that a classifier language like Thai also needs a DP after all. To be consistent, I reanalyze the definite bare common noun and the definite proper name as DPs instead of ClPs. The movement of N in the definite bare noun is thus to D instead of C]. All in all, the head—final analysis requires three syntactic operations, namely N movement to D for definite bare nouns and to Num for indefinite ones, Cl-to-Num raising as adjunction for quantified common nouns and proper names, and Cl-to-D movement via Num as adjunction for quantified pronouns. 2.5.2 Head-initial analysis In this section I will show that the head-initial analysis is preferable to the head- final analysis because it requires only one movement which captures all the instances under investigation. 2.5.2.1 Bmmmgnmmns Having argued in the previous section that ClP is inadequate for the account of the definite noun phrase in Thai, in what follows I will analyze the definite bare noun as DP. ( l) Definite bare common nouns '° I hypothesize that the pronouns used in such constructions are determiner-like elements. They are limited to a small number (raw, ‘we’, khaw, ‘he, she’, man, ‘it’). Therefore, it may be the case that Thai also uses the insertion of a determiner-like pronoun to express definiteness. 36 Like the he movement in the ( unrestricted distril tog in (46) acco ('46) ma, 1 dog s The. (53) 3! lrdefinite bare Recall that TTT that the hea 3331K ”T {”316 h. " L rPleated l (47) dog, t Like the head-final analysis, the head-initial analysis requires covert N-to-D movement in the case of definite bare nouns to account for the interpretation and the unrestricted distribution. The structural representation of Thai bare common noun maa, ‘dog’ in (46) according to the head-initial analysis is as in (58).ll (46) maa, naan, yuu2 taay, to?4 dog sleep IMPF under table ‘The dog is sleeping under the table.’ (58) DP /\ D’ /\ D NP | N maa, dog (2) Indefinite bare common nouns Recall that indefinite bare common nouns in Thai are free in distribution, which means that the head Num must be filled. Therefore, N-to-Num movement is also proposed for the head-initial analysis of Thai indefinite bare nouns as illustrated in (59) for (47), repeated below. (47) maa, kat2 dek2 dog bite child ‘A dog/dogs bit a child/children.’ ” I am assuming here that NumP and ClP are situated between DP and NP. But the 37 :3) 3\'~Cl] construction The head-initial sanction which disp afietinite interpretation ices not seem to work i I: Head Movement C o ?:is over another head “513' here owing to t ahead Num should be ‘T‘ 330%: - ~10 the spent} (5 9) NumP /\ Num’ /\ Num ClP /\ Cl NP | N maa, dog (3) [N+Cl] construction The head-initial analysis seems to face some problems dealing with the [N+Cl] construction which displays only an indefinite non-specific interpretation. Due to the indefinite interpretation, such a construction is embedded in NumP. N-to-Num movement does not seem to work here since the head Cl is on its way to a higher head, thus violating the Head Movement Constraint (Travis 1984) which requires that a moved head may not cross over another head in its cyclic movement. Moreover, N movement to Num is unlikely here owing to the distributional restriction of this construction which implies that the head Num should be empty. A way of coping with this problem is the suggestion that NP moves to the specifier of ClP, leaving Num empty.12 The distribution and interpretation will thus be captured. Although this is not an elegant proposal, everything structure given in (58) is for simplicity. '2 I suggest an NP movement instead of an N movement (probably to Cl as lefi adjunction) here owing to the following example where the whole modified NP seems to raise. (i) khaa, buzrii2 traa, phra4can, muan, give (me) cigarette brand moon Cl ‘Can you give me a cigarette of the brand ‘Moon’?’ 38 ms to be well- felon, is structu r. (48') khaa, give ( ‘Can (60) seems to be well-taken care of. The phrase buzrii, muan, ‘cigarette C1’ in (48), repeated below, is structurally represented as in (60). (48) khaa, buzrii2 muan, give (me) cigarette Cl - ‘Can you give me a cigarette?’ (so) NumP /\ Num /\ Num ClP /\ bu ,rz'i 2, Cl’ cigarette /\ Cl NP muan, t. 9 I If this analysis is on the right track, we can reanalyze the data for the definite and indefinite common nouns: instead of N raising, we can treat all instances of movement as NP raising to the specifier of the relevant functional head. NP moves to the specifier of DP in the definite bare common noun and to the specifier of NumP in the indefinite one. It is not uncommon that an empty head can be ‘strong’ enough to attract an element to either its head or its specifier as long as it has some semantic import (Huang 1988, Alexiadou 1997). We can thus account for the free distribution of the definite and indefinite NPs in Thai due to the fact that although the relevant functional heads are empty, their specifiers are filled. Moreover, I will suggest that NP movement is overt in all the cases. 39 “an .19.: 1.522 W Definite p easement is reqt ..m_ ‘Dam‘ in (6 (61) Although Ttm‘ 0f quanti ‘a'dle [he inSlan T T Pmse da ('62) 2.5.2.2 21929111311125 Definite proper names are like definite conrrnon nouns in that NP-to-DP movement is required, as portrayed by the syntactic representation of the proper name dam, ‘Dam’ in (61). (61) DP /\ dam ,,- D’ Darn /\ D NP Although the head-final analysis needs an additional Cl-to-Num raising for the account of quantified proper names which are indefinite, the head-initial analysis can handle the instance well with simply NP movement to [Spec, Num] as represented in (62) for the phrase dam, say, khan, ‘two Darn’s’. (62) NumP /\ NPi Num’ dam] /\ Dam Num ClP 3905 ‘ /\ two Cl NP khan, t. l Structure (62) confirms the unlikeliness of a head-to-head movement in the head- initial analysis. This is because the heads of ClP and NumP are already occupied. The head N containing a proper name cannot move across these two heads to a higher head position for the movement will violate the Head Movement Constraint. If head-to-head movement as adjunction were to be assumed, N would have to adjoin to Cl and Num successively. The wrong word order *dam , khan, say,‘Dam Cl two’ would be obtained if 40 lei adjunction was per: sequence ‘ say, khan .. . movement operates in t' specifier of C 1? to assur girlie: of NumP for tl 15.2.3 REMARKS Since I have argt atta‘ analysis does not pronoun as illustrated in (63) D ran We “118 makes [h r. lefi adjunction was performed. And right adjunction would also yield the incorrect sequence * say, khan, dam, ‘two Cl Dam’. In sum, it is impossible that head-to-head movement operates in this case. NP may move in successive steps. First, it moves to the specifier of ClP to assure that the classifier agrees with the noun. Then, it raises to the specifier of NumP for the number purpose. 2.5.2.3 Pronouns Since I have argued above that the pronoun in Thai is generated in D, the head- irritial analysis does not need any movement at all when accounting for the quantified pronoun as illustrated in (63). (63) DP /\ D NumP raw, /\ we Num ClP saam, /\ three Cl NP khan, This makes the head-initial analysis preferable over the head-final analysis which requires the two-step movement as adjunction of C1 to D (via Num). In the case of the noun phrase which contains a pronoun and a proper name or a common noun, the NP movement to [Spec, DP] can unproblematically handle it as illustrated in (63) for the phrase dam, khaw, ‘Dam he’. (63) DP /\ dam ,, D’ Darn /\ D NP khaw, t, he | 41 Recall that the l requirement of movem. consistent throughout. Comparing the t- Ind-final one because . 7E the instances of bare 1.6 Conclusion In this chapter, 1 races in the language. Didi as singularizatio: definite noun plume as t Adopting Cheng 75.5.“. mute bare nouns in \' ' um . is net er syntactica wintensely, it can be f Recall that the head-final analysis can handle the latter case without the requirement of movement. However, the raising of NP in the head-initial analysis is consistent throughout. Comparing the two analyses, the head-initial analysis is more attractive than the head-final one because only one syntactic operation, namely NP movementcan capture all the instances of bare nouns under study. 2.6 Conclusion In this chapter, I have shown that Thai also requires the DP despite the lack of articles in the language. Although Cl is claimed to perform the similar deictic function to D such as singularization and individualization, I have argued that the analysis of a definite noun phrase as ClP is inadequate. Adopting Cheng and Sybesma’s (1999) analysis of Chinese, I suggest that indefinite bare nouns in Thai (N, [N+Cl], [N+Num+Cl]) are NumPs. However, the head Num is never syntactically empty. That is, it can be lexically filled by a numeral. Alternatively, it can be filled by a moved N at LF according to the head-final analysis or the movement of NP to [Spec, NumP] according to the head-initial analysis makes the head Num syntactically visible. This explains why the distribution of indefinite bare nouns in Thai is not restricted to lexically governed positions. I further suggest that Thai definite bare nouns are embedded in BPS, not ClPs as proposed by Cheng and Sybesma. The major evidence is from the requirement of D as the base-generated position of a pronoun. The definite reading is acquired by the movement 42 of N to D in the head-final analysis whereas the head-initial analysis requires NP-to-DP movement. Having compared the head-initial and the head-final analyses of Thai bare nouns, I can say that the former is more favorable than the latter because it requires only one syntactic operation of NP movement. The head-final analysis, on the other‘hand, needs three movements to do the same job. In Chapter 4 we will see more evidence that supports the head-initial analysis. 43 ln this chapter 1‘ no positions for adj ect aijoned to M) and the as another type of noun 3.1 Introduction Because Thai w Tm? Vtithout a COpulz CHAPTER 3 NOUN MODIFIERS IN THAI In this chapter I will argue for the existence of Thai adjectives and that there are two positions for adjectives in Thai depending on their interpretations. One is right- adjoined to NP and the other is the specifier of ClP. Nominal modifiers will be introduced as another type of noun modifier, which are also right-adjoined to NP. 3.] Introduction Because Thai words equivalent to English adjectives can be predicated of a noun phrase without a copula as in (1), a number of studies question the existence of Thai adjectives (Bandhumedha 1983, Savetamalya 1989, Hudak 1990, Warotamasikkhadit 1996f. (1) a. dam, suurj, Darn tall ‘Dam is tall.’ b. rritt4 suay, Nit pretty ‘Nit is pretty.’ ' The words that are considered as Thai copulas are pen ,, khr'i,, and chay,. They are generally used with a nominal predicate as in the following examples. (i) khaw, pen, khruu, he be teacher ‘He is a teacher.’ (ii) khaw, khii, naa,yok,rat,tha,mon,trii, he be prime minister ‘He is the Prime Minister.’ (iii) khaw, chay, phuu,chaay, khon, nan4 time, he be man Cl that surely Such words are thus suggested to be stative verbs, and adjectives are claimed to be non-existent in Thai. However, with thorough consideration I will argue that Thai does have a class of words called ‘adjectives’ as these words do not behave exactly the same as stative verbs. Therefore, they should not be grouped into the same category. In terms of distribution, Thai adjectives are unexceptionally post-nominal as shown in (2). Pre-nominal adjectives bring about ungrammaticality as in (3). (2) a. khaw, chaap, phuu,yin, suay, he like woman pretty ‘He likes a/the pretty woman/ pretty women.’ b. dek2 ?uan, suk,kha,phaap, may, dii, child fat health not good ‘A/the fat child/ Fat children is/are not healthy.’ (3) a."‘khaw, chaap, suay, phuu,yirj, he like pretty woman b.*?uan, clek2 suk,kha,phaap, may, dii, fat child health not good However, a few adjectives cannot appear right after nouns. Classifiers must be inserted between nouns and adjectives. Examples of such adjectives are kan, ‘forrner’ and diaw,kan, ‘same’ as illustrated in (4)-(5) respectively. (4) a.*khaw, chaap, naay, kan2 he like boss former ‘He likes the former boss.’ b. khaw, chaap, naay, khon, kan2 he like boss Cl former (5) a.*raw, ?aan2 nan,sii, diaw,kan, we read book same ‘We read the same book.’ b. raw, ?aan2 nan,sii, lem, diaw,kan, we read book Cl same ‘He is surely that man.’ 45 As car adjectives in ( containing the phases ambig parses only t :e: are gener :csitions. Beside “3.535 115 to l I ‘I (it) “her *5?) contains ('6) at s “0 43C}?- rd (Ti ‘T'Tfierl (ii a n d. ‘t b. ,, dc As can be noticed, not only is there a difference in the distribution between the adjectives in (2)~(3) and those in (4)-(5), but the interpretation of the noun phrases containing them are also distinct. While the first group of adjectives makes the noun phrases ambiguous in terms of definiteness and number, the second group gives the noun phrases only the definite singular interpretation. Due to their differences, it is likely that they are generated in different places. Therefore, I will investigate their base-generated positions. Besides adjectives, I will also discuss another type of noun modifiers which enables us to distinguish between ‘true’ classifiers (Cl) and classifier-like elements (C-L) ' as in (6) where (6a) contains a classifier khan, which is used with human beings while (6b) contains a classifier-like word tua, which means ‘body’.2 (6) a. naay, khon, suurj, boss C1 tall ‘the tall boss’ b. naay, tua, suurj, boss C-L tall ‘a/the tall boss(es)’ The interpretations of (6a) and (6b) are different. While (6a) denotes only definiteness and singularity like (4)-(5), (6b) can express either definiteness or 2 The word tua, is considered a classifier-like word because it is homophonous with the classifier used with animals as in (i). (i) a. maa, saam, tua, dog three Cl ‘three dogs’ b. maa, tua, yay2 dog C-L big ‘a/the big dog(s)’ 46 ndehniteness‘ dinently. The ch: existence of ac Tnai has no ad devoted to the aesitions. The 3pc of noun r he chapter, 32 Existence 1 Will ; Ti; and then 3.2.] Pm‘iou Most indefiniteness without number specified like (2)-(3). Therefore, they will be treated differently. The chapter is organized in the following way. In section 2 I will argue for the existence of adjectives in Thai and against some previous studies which have claimed that Thai has no adjectives. Section 3 presents some literature on adjectives. Section 4 is devoted to the discussion of the three types of noun modifiers together with their positions. The interpretation as well as the distribution of noun phrases containing each type of noun modifiers is spelled out in section 5. Section 6 provides the conclusion of the chapter. 3.2 Existence of Thai adjectives I will first present the claims made by several linguists on the lack of adjectives in Thai and then argue for their existence. 3.2.] Previous studies on the lack of Thai adjectives Most of the studies that claim that Thai lacks adjectives consider mainly the fact that adjectives used as predicates do not need copula verbs. Hudak (1990:42), for instance, states that“ (Thai) words considered to be adjectives in English (suay, ‘beautiful’, dii, ‘good’, yaaw, ‘long’) may function as nominal attributes, verbal attributes or as predicates. Because these words behave syntactically as verbs without a copula, they are treated as verbs.” Warotamasikkhadit (1996) also puts words equivalent to English adjectives in the category of verbs. In other words, he does not give a separate category of adjectives when introducing parts of speech in Thai. Likewise, Bandhumedha (1983:29) classifies Thai verbs into five categories and includes adjectives in the category of stative 47 verbs. Their f {ape as dem: (7) a. It shoi Ijxtives do muCed bV’ (3)21.I Anoth, t... . 513T! “0115),; ‘1 T T3181“, Sat verbs. Their function is to modify nouns in various aspects such as size, color, quality and shape as demonstrated in (7). (7) a. mii, kra,thaarj, ma,hi,ma, tan, yuu2 mum, harj, exist vase huge stand stay comer room ‘There is a huge vase in the corner of the room.’ b. khaw, chaap, suam, wman, dam, he like wear glasses black ‘He likes wearing black glasses.’ c. khruu, chaap, nak4rian, kha,yan, teacher like student diligent ‘Teachers like diligent students.’ (I. lan2 suam, siia, khlum, tua, yaaw, she wear clothes cover C—L long ‘She wears a long coat.’ It should be noted that stative verbs other than those which can be classified as adjectives do not normally directly modify nouns as shown in (8) below. A relative clause introduced by the complementizer (COMP) thii, is to be used as in (9). (8) afmii: phuuayios sonzsays nan. yuuz mumi hon. exist woman doubt sit IMPF corner room b.*khruu, chaap, nak4rian, khaw,cay, teacher like student understand (9) a. mii, phuu,yirj, thii, sorjzsay, nan, yuu2 mum, hat], exist woman COMP doubt sit stay comer room ‘There is a woman who is in doubt sitting in the comer of the room.’ b. khruu, chaap, nak,rian, thii, khaw,cay, teacher like student COMP understand ‘Teachers like students who easily understand.’ Another study which argues against the existence of Thai descriptive adjectives is Savetamalya (1989). Utilizing the lexicase analysis which aims to discover and describe the relationships between and among pairs of lexical items or words in constructions with each other, Savetamalya assumes that Thai words translatable as descriptive adjectives in 48 English are in fact a su U as: do not fit the defir an adjective phrase. an edkeidl9971230. ci predicational. She furt native clauses whose words which can be c dependents of nouns Piraie nor be modifi: However, sor Iii”: as former . , same 5 Tffiduced relativ To: - . adjectl‘v'es do n English are in fact a subclass of verbs (so-called stative verbs). The major reason is that they do not fit the definition of ‘adjective’ defined in the lexicase analysis as “the head of an adjective phrase, an endocentric, non-predicational attribute of a noun,” (Savetamalya and Reid 1997: 230, citing Starosta 1988: 51) because such words in Thai are all predicational. She further argues that this kind of words exhibits the same structure as relative clauses whose subjects are omitted or so-called reduced relative clauses. The only words which can be considered adjectives in Thai are numerals because they can only be dependents of nouns without the ability to “function as the sole constituent of a noun phrase nor be modified by determiners or relative clauses.” (Savetamalya 1989:155) However, some Thai words which are equivalent to adjectives in other languages such as former, same, only, etc. cannot be used predicatively and thus cannot be claimed to be reduced relative clauses, as will be elaborated later. Therefore, the assumption that Thai adjectives do not exist seems not to be right. To sum up, the studies presented above claim that descriptive adjectives are missing in Thai and the words equivalent to English adjectives are stative verbs. I will, however, argue against their claim in the following section, by showing that the Thai words translatable as English adjectives do not completely behave as verbs. They are, in fact, different from stative verbs in many respects. The arguments evidently suggest that Thai also possesses adjectives. 3.2.2 Arguments for the existence of Thai adjectives The arguments I will provide for the claim that Thai also has adjectives are that this type of words behaves very much like adjectives in other languages in terms of comparison and that they have their own characteristics not shared by other kinds of 49 nits. Statue y: are as this kind (I) Comparison Thai wo to belong to the can be directly Itorpheme [hit (10) a. words. Stative verbs which Thai adjectives are claimed to be do not behave in the same way as this kind of words. The arguments to be made are as follows. (1) Comparison Thai words that are translatable as adjectives in other languages can be assumed to belong to the adjective category, not verb. The main justification is the fact that they can be directly modified by the comparative morpheme kwaa, and the superlative morpheme thii,-sat, whereas the verbs cannot. (10) a. dam, suurj, kwaa2 khaaw, Dam, tall -er/more Khaaw, ‘Dam is taller than Khaaw.’ b. dam, suurj, thii,-sut2 Dam, tall -est/ most ‘Dam is the tallest.’ c.*dam, ruu,/khaw,chay,/khit4 kwaa2 khaaw, Darn, know/understand/think -er/more Khaaw, ‘Dam khows/understands/thinks more than Khaaw.’ d.*dam, ruu,,/khaw,chay,/khit4 thii,-sut2 Dam, know/understand/think ~est/ most ‘Dam knows/understands/thinks the most.’ The degree word maak, ‘much’ must be introduced before kwaa, ‘-er/more’ and thii,-sat, ‘-est/most’ to indicate the comparative and superlative degrees for Thai verbs respectively. (1 1) a. dam, ruu,,/khaw,chay,/khit4 maak, kwaa2 Khaaw, Dam, know/understand/think much -er/more Khaaw, ‘Dam khows/understands/thinks more than Khaaw. ’ b. dam, ruu,/khaw,chay,/khit, maak, thii,-sut2 Dam, know/understand/think much -est/most ‘Dam khows/understands/thinks the most.’ 50 iiihtroouetion The sec: ofwords. The n denote. which is a noun -cla< taker otideli; m.) a (2) Introduction of classifiers in front of adjectives The second argument is that a classifier can be inserted right in front of this type of words. The noun phrase which contains an adjective preceded by a classifier is always definite, which is consistent with the claim made by Hundius and Kolver (1983:17 7) that, “in a noun-classifier—adjective sequence in Thai, a classifier will be understood as a marker of definiteness.” (12) a. dek2 phaam, khii,rook, child thin sickly ‘A/The thin child is sickly.’ b. dek2 khan, phaam, khii,rook, child Cl thin sickly ‘The thin child is sickly.’ (13) a. baan, yay2 naa,sa,baay, khwaa2 baan, lek, house big comfortable -er/more house small ‘A/The big house/ Big houses is/are more comfortable than a/the small one/ small ones.’ b. baan, larj, yay2 naa,sa,baay, khwaa2 baan, larj, lek, house Cl big comfortable -er/more house Cl small ‘The big house is more comfortable than the small one.” Without a classifier as in (12a) and (13a), the nominal expressions bear an indefinite or a definite interpretation without number specified. On the contrary, the phrases in (12b) and (13b) which contain classifiers can only be definite and singular. A stative verb, on the other hand, can be preceded by a classifier only when it is embedded in a relative clause as in (14) below. (14) a.*khaw, chaap, phuu,yirj, khan, run/khaw,chay, he like woman Cl know/understand ’ (13b) can also be interpreted as (13a). In that case, lag, is treated as a classifier—like word, not a classifier. This will be discussed in what follows. 51 b. khaws Ci he 1‘: ‘HC likC‘r .EiReduplication A Thai adj ectivt irinutive meaning w}: naive verbs to behave if) and of stative verb.- (15) a. khaw, m. he 10. ‘He look b. phuu,cha men ‘It is hard (16M‘mii, khij have per, ‘15 there b-‘khaw, ch he T 1, THC likes. ‘3 th. Tm the f:‘JHCtiOr 1111515 the Onl‘. up. +~ ‘-. in K1540:- E ' - ~; NominaliZation b. khaw, choop3 phuu3yirjs khonl thii,, ruu4/khaw3chay, he like woman Cl COMP know/understand ‘He likes the woman who knows/understands.’ (3) Reduplication A Thai adjective can reduplicate its base without any change to express the diminutive meaning which is equivalent to ‘fairly’ or ‘somewhat’. It is impossible for stative verbs to behave in the same way, as shown by the reduplication of adjectives in ( 15) and of stative verbs in (16) below. (15) a. khaw, moon, tmzez phuu3yirj5 suay,-suay, he look only woman pretty-pretty ‘He looks only at fairly pretty women.’ b. phuu,chaay, diil-diil haaS yaak3 men good-good find hard ‘It is hard to find fairly nice men.’ (16) a."'miil khon, ruu,,-ruu4 sin,tha‘:k2 may4 have person know-khow syntax QP ‘Is there someone who somewhat knows syntax?’ b."'khaw5 choop3 phuu3yi135 khaw,chayl-khaw3chay, ijaay3 he like woman understand-understand easy ‘He likes women who rather easily understand.’ Warotamasikkhadit (1996:76-77) realizes that the reduplication of such words render them the function of noun modification which is a characteristic of adjectives. Yet, he thinks this is the only difference between this type of words and stative verbs. Therefore, he insists on putting them in the same category. (4) Nominalization Thai utilizes two morphemes kaan,- and khwaam,— as prefixes in the nominalization process. The difference between the two is captured by Hudak (19903 8) as follows: 52 \. man l- ‘u\ verbs and so ‘city’, kaan 2. khu'aam; quality or st. It happens that if)" generally denote a Eeused with kaan 1- wh (17) a khaw, p he 1 ‘He talk bflchawS he (18) a. khaw, he - ‘He tall b- khawS he ‘He tal‘ 1. kaan,- ‘the act of, affairs of, matter of” forms abstract nouns from verbs and some nouns: e.g. len3 ‘to play’, kaan,len3 ‘playing’; miianl ‘city’, kaan,miian1 ‘politics’. 2. khwaam,- ‘the condition of’ forms abstract nouns that express a quality or state e. g. ruu,,sik2 ‘to feel’, khwaam,ruu,sik2 ‘feeling’. It happens that Thai adjectives can only be prefixed by khwaam,-, not kaan ,-, as they generally denote a quality or state‘. The stative verb, on the contrary, can sometimes be used with kaan,- when an act is the required interpretation. (17) a. khaw, phuut3 thin, khwaam,dii, he talk about NOM-good ‘He talks about goodness.’ b.*khaw, phuut3 thin, kaan,dii, he talk about NOM-good (18) a. khaw, phuut, thin, khwaaml-khaw3chay1 he talk about NOM-understand ‘He talks about comprehension.’ b. khaw, phuut, thin, kaan,-khaw,chay, he talk about NOM-understand ‘He talks about (the act of) understanding.’ ‘ A very limited number of Thai adjectives can be preceded by the prefix kaan,, but only when it follows the copula pen, in such a structure as below: (i) penl kaan,-diil thii3 khaw, maal be NOM-good COMP he come ‘It is good that he comes.’ (ii) penl kaangyaak3 thii3 rawl ca?2 phop, kanl be NOM-hard COMP we will meet RECP ‘It is hard that we will meet.’ This type of structure, however, is borrowed fiom English. The natural way to express the sentences above is as in (iii)-(iv). (iii) diil thii3 khaw, maal good COMP he come ‘It is good that he comes.’ (iv) yaak3 thii, rawl ca?2 phop4 kanl hard COMP we will meet RECP ‘It is hard that we will meet.’ 53 The arguments dizierent from stative \ 33 Previous literatu r Hating given 1' Ln lhai adjectives are positions. To argue for is: one is Sproat and . categories: direct and l, Sm to be only pania Si" .‘ i 1111' 8‘. .leCtllnes are SEC} 31mm . The arguments above seem to sufficiently prove that this type of words is different from stative verbs. Therefore, they should be considered as adjectives in Thai. 3.3 Previous literature Having given the arguments for the existence of adjectives in Thai, I will suggest that Thai adjectives are of two different types and they are generated in two distinct positions. To argue for my suggestions, three studies on adjectives will be presented. The first one is Sproat and Shih (1988) which claims that Thai adjectives fall into two categories: direct and indirect modifiers. Unfortunately, their account of Thai adjectives seems to be only partially right. The second one is Bernstein (1993) which proposes two syntactic classes of adjectives: regular attributive adjectives and mere-type adjectives. Thai adjectives are seemingly classifiable in the same way. But the requirement of a classifier by a ‘mere-type adjective’ in Thai is reminiscent of the ‘double definiteness’ instance in Mainland Scandinavian languages. Therefore, the study by Kester (1996) on such a phenomenon will also be introduced. 3.3.1 Sproat and Shih (1988) Sproat and Shih make a distinction between ‘direct modification’ and ‘indirect modification’ when considering English and Mandarin adjectival orderings. They suggest that the location of an adjective is cognitively-based; that is, the more comparisons or computations are involved, the further away fi'om the head noun the adjective is. The orderings of various types of adjectives considered as ‘direct modifiers" are thus ‘quality > size > shape > color > material > provenance.’ English adjectives are ‘direct modifiers’ 54 because they are sub). ablation of such an o. The direct moi minions as illustrat (10) a hao l good b.‘hong red (31‘) a. xiao Small b-‘ynan roun 5} 'dE-adjecfivee Cl flame and has as, i' are not Subjec ('22) 3- hag 80o inic b: >112 YOU Cong eniir because they are subject to ordering restrictions above as demonstrated in (19) where the violation of such an ordering (l9b,c) leads to ungrammaticality (p.469). (19) a. large round red apple b.*round large red apple c.*large red round apple The direct modifiers or ‘de-less adjectives’ in Chinese display the same ordering restrictions as illustrated in (20) and (21) below (p.471). (20) a. hao hong pan-zi good red plate b.*hong hao pan-zi red good plate (21) a. xiao yuan pan-zi small round plate b.*yuan xiao pan-zi round small plate Unlike English, however, Chinese also possesses ‘indirect modifiers’ represented by ‘de-adjective’ constructions. According to Sproat and Shih, this type of modifier is order-free and has the structure of relative clauses. The examples in (22) below show that they are not subject to ordering restrictions. (p.465). (22) a. hao-dc yuan-de pan-2i good-DE round-DE plate ‘nice round plate’ b. yuan-dc hao-de pan-zi round-DE good-DE plate Concerning their similarity in structure to relative clauses, Sproat and Shih argue that de-modifiers use the same particle de as the relative clause as shown in (23). Moreover, such an adjective as former which never occurs in the predicate position, and thus cannot appear in relative clauses, is unable to function as indirect modification as in (24). 55 (23)Wo xi-h‘ I like-‘ ‘the book (24) a‘jei-ge this-C l They also Clair irnple modifiers) an, "mitt . “b.4834“, (25) a. ma, dog . ‘big b1 b'imaa dog: (26) a. maas dOg (23) W0 xi-huan-de shu I like-DE book ‘the book which I like’ (24) a.*jei-ge zongtong qian this-Cl president former ‘*This president is former.’ b.*qian-de zongtong former-DE president c. qian zongtong former president They also claim that Thai, like Chinese, has two types of modification - direct (simple modifiers) and indirect (complex modifiers comprising a classifier and an adjective or the word sii, ‘color’ before a color adjective). Their reason is that direct modifiers do observe the ordering restrictions like other languages as shown in (25) while indirect modifiers do not as in (26). The examples (and judgments) are from Sproat and Shih (pp.483-484). (25) a- maas dam. yayz dog black big ‘big black dog’ b."‘maa, yay2 daml dog big black (26) a. maa, sii,-daml tual yay2 dog color-black Cl big ‘a big black dog” b. maa, tual yay2 sii,-daml dog Cl big color-black ‘a big black dog" 5 The phrase in (26a) is ambiguous. Its interpretation can be either ‘A/The black dog is big.’, ‘a dog whose color is black and whose body is big’ or ‘the big [black dog].’ I will argue in section 3.4.3 that the first two readings result from tua, acting as a ‘classifier- like’ word whereas the third one is caused by tua, functioning as a classifier. ° The phrase in (26b) is ambiguous between a phrasal and a sentential readings. The first reading is ‘a dog whose body is big and whose color is black’ and the second is ‘The big 56 As earlier stat; Thai possesses ‘indirc rill argue that they d. in do not seem to o lifol above does not confirm this. (27) 1 phuu,-} Woman ‘tall wh b. phuuB-y Woman As is generallj trier usually results 1 closer an adjective is not other. That is \a laent adjective in (18) a, man dog # As earlier stated, Sproat and Shih’s analysis is not completely right. I agree that Thai possesses ‘indirect modifiers’ as in (26) which I will term ‘nominal modifiers’. But I will argue that they do not consist of a classifier and an adjective. Moreover, adjectives in Thai do not seem to observe the ordering restrictions as they claim in (25). To my ears, (25b) above does not sound had. A similar example to (25) as demonstrated in (27) can confirm this. (27) a. phuu3-yin, suun, khaaw, woman tall white ‘tall white woman’ b. phuu3-yin, khaaw, suun, woman white tall As is generally known, variation in adjective order may occur. An alternation of order usually results from the relationship between a noun and an adjective, that is, the closer an adjective is to a noun, the closer semantic relationship they have with respect to each other. That is why a compound reading is normally derived from the noun and the adjacent adjective as shown in (28). (28) a. maa, dam1 yay2 dog black big ‘a big dog of the black breed’ b. maa, yay2 dam1 dog big black ‘3 black dog of the big breed’ In short, direct modification (simple adjective) and indirect modification (complex modifiers) in Thai as termed by Sproat and Shih seem to display the same behavior with regard to free ordering. I will show in what follows that these two types of dog is black.’ Again, the fact that tua, functioning as a ‘classifier-like’ word and a classifier respectively can account for the ambiguity. 57 modifications share s be: are generated in adjectives which beh. as nailed by Bemsteir 33.2 Bernstein (1993' Bernstein pr0t .\l’ and H5 above Ni position generally co Adjectives in l stall class of adj ectiv 60). In addition, thei nominally but with d (19) el librc the bQQ ‘the red modifications share several other characteristics, which makes it eligible to assume that they are generated in the same position. However, Thai possesses another group of adjectives which behave differently and which are fairly similar to ‘mere-type adjectives’ as called by Bernstein (1993) whose study is presented below. 3.3.2 Bernstein (1993) Bernstein proposes two positions for adjectives in Romance - one is adjunction to NP and XPs above NP, and the other is the head of AP above NumP. The difference in position generally correlates with a difference in interpretation. Adjectives in Romance are post-nominal in the unmarked case (29). However, a small class of adjectives such as mere, other , etc. are required to appear pre-nominally (30). In addition, there are cases where adjectives can occur either pre-nominally or post- nominally but with different interpretations (31). (29) el libro rojo (Spanish) the book red ‘the red book’ (30) a. un mero accidente (Spanish) a mere accident ‘a mere accident’ b.*un accidente mero a accident mere (31) a. el simple hombre (Spanish) a simple man ‘a simple/mere man b. el hombre simple a man simple ‘a simple-minded man’ 9 Bernstein accounts for the examples above by proposing two syntactic types of adjectives. Based on the similarity of adjectives to adverbs which are modifiers adjoined 58 toll) or XPs within ‘ attributive adjectives normal domain. The above NumP because angular attributive a. (33) a. El libr the bool ’The bc b-‘El acc the acv "The HR b0 ’the v: b.‘un ID] a V6 ‘3 Ver ambition that a N fret, ~ "1 ”“0“ 0f the a 535 . to VP or XPs within the verbal domain (Belletti 1990, Chomsky 1991), the regular attributive adjectives as in (29) are suggested to be APs adjoined to NP or XPs within the nominal domain. The mere-type adjective (30), on the other hand, is the head of AP above NumP because it cannot be used predicatively (32b) nor be modified (33b), unlike a regular attributive adjective, as shown below. (32) a. El libro es rojo. (Spanish) the book is red ‘The book is red.’ b.*El accidente es mero the accident is mere ‘*The accident is mere.’ (33) a. el libro muy rojo (Spanish) the book very red ‘the very red book’ b.*un muy mero accidente a very mere accident ‘a very mere accident’ Given the proposal of two distinct positions of adjectives together with the assumption that a N in Romance has to move to Num to pick up number, the different distribution of the adjectives in (29) and (30) is explained by the structures in (34) and (35) respectively. (34) DP /\ D NumP el /\ the Num NP libro,- /\ book AP .5.“ Z—% rojo I red 59 135) The adjective different interpretatit generated in diff ere: born as shoun belc (’36) a E1 is ‘The b. el r the ‘the CF e1 the (35) DP /\ D AP un /\ a A NumP mero /\ mere Num NP accidente,- accident IINI t ‘1 The adjective in (31) which can occur either pre-nominally or post-nominally with different interpretations is thus handled by the suggestion that the adjective is base- generated in different positions. Predication and modification can be used to distinguish them as shown below. (36) a. El hombre es simple. (Spanish) ‘The man is simple-minded.’ (cf. *‘The man is mere.’) b. el hombre muy simple the man very simple-minded ‘the very simple-minded man’ c.*el muy simple hombre the very mere man ‘*the very mere man’ When the adjective simple means ‘simple-minded’, it can be used predicatively (36a) and be modified (36b). Therefore, it is a regular attributive adjective generated as NP adjunct (373). When simple means ‘mere’, it cannot be used in a predicate context (36a) nor be modified (36c). Thus, it is a mere-type adjective, generated as the head of AP above NumP (37b). (37) a. [DP e1 [NumP hombrei [NP simple t,]]] b- [DP 31 [AP SimPIC [NumP hombre, [NP trill] Thai appears to have the two types of adjectives as classified by Bernstein. Most of Thai adjectives can be used predicatively and be modified; hence, they should be 60 mrsidered as regular land behave like the predicative context as on my, v However, the hats, unlike other tiered before then: :ezrdte. & shovm ‘c 59) a.*naa bos: ‘the considered as regular attributive adjectives. A small number of adjectives, on the other hand, behave like the mere-type adjectives. For example, they cannot be used in a predicative context as shown in (38).7 (38) *naay, kon2 boss former ""The boss is former.’ However, the mere-type adjectives in Thai display some different characteristics. That is, unlike other adjectives that can directly modify nouns, classifiers must be inserted before them. Plus, noun phrases containing such adjectives must be interpreted as definite, as shown below. (39) a.*naay, kon2 boss former ‘the former boss’ b. naay, khonl kon2 boss C1 former ‘the former boss’ ( ‘*a former boss’) This phenomenon is somewhat like the ‘double definiteness’ instance in Mainland Scandinavian languages, as discussed by Kester (1996). 3.3.3 Kester (1996) Indefinite and definite noun phrases in Scandinavian languages are different in that the definite article which is morphologically similar to the indefinite one surfaces as an enclitic to the noun as shown below. (40) a. en bil (Swedish) a car ‘a car’ 7 Thai does not have a word equivalent to the adjective mere; therefore, I use the word kan, ‘former’ which belongs to the same type to represent the mere-type adjectives. 61 b. bilen car-th ‘me C3 However, \vhe either definite anicl ('41) a den stoI the big b.*stora bing The above phe esanee. Kester reana Lining the feature [* tithe enclitic err is 2 flinples where the a meme excludes the ('42) a. Den p ‘The pi b. Den sj the se ‘ThEre Moreover, the 'T‘fi .733, . as illustrated ir b. bilen car-the ‘the car’ However, when the definite noun phrase is modified by a definite adjective, another definite article must be inserted before the adjective as in (41).8 (41) a. den stora bilen (Swedish) the big.DEF car-the b.*stora bilen big.DEF car-the The above phenomenon has been called ‘double definiteness’. To account for this instance, Kester reanalyzes the enclitic article en as the head of a functional projection carrying the feature [+familiar] or [+referential]. The assumption is drawn from the fact that the enclitic en is associated with a specific reading as shown in the following examples where the absence of en gives preference to a non-specific reading (42a) and its presence excludes the generic reading (42b). (42) a. Den pilot_ som ska kora vart plan maste vara erfaren (Swedish) the pilot who will drive our plane must be experienced ‘The pilot who is to drive our plane must be experienced.’ b. Den sjuarige pojke (*n) som klarar detta finns inte the seven-year-old boy (*the) who manages this exists not ‘There is no seven-year-old boy who can manage this.’ Moreover, the enclitic article en is claimed to have ‘deictic reference’ (Delsing 1993) as illustrated in (43) where the definite article may be omitted, and also ‘anaphoric reference’ as in (44). Kester thus assumes that the enclitic article en has the feature [+refentiality] and/or [+familiarity]. ' Adjectives in Mainland Scandinavian languages are inflected for number, gender and (in)definiteness. However, the ‘definite’ paradigm is morphologically poor. That is, there is only one definite morpheme in these languages and it does not reflect either number or gender. 62 (43) a (den) f (the) F ‘me Fr b. (den) U (the) ll ‘the th (44) John kép John bOU With the propi [familiar], Kester ex aieetive which is inf. generated in the speci a adjectives as specii :‘fa'niliafl or [+refere thechng with the hea Emite adjective und Im‘ed in (45 ) belo (45) or /\ D do" lb:- . e inset-um] O iii-e35 . uldlng’ as Pmp. (43) a. (den) franska revolutionen (the) French revolution-the ‘the French revolution’ b. (den) tredje gangen (the) third time-the ‘the third time’ (44) John kopte ett vitt hus. Det vita huset var mycket stort John bought a white house. The white house-the was very big. With the proposal that the enclitic en is the head of F P with [+referential] or [+familiar], Kester explains the ‘double definiteness’ phenomenon as follows. The adjective which is inflected with the definite morpheme a is assumed to be base- generated in the specifier of the FP, assuming Cinque’s (1993) proposal of the generation of adjectives as specifiers of various F Ps between DP and NP. The noun with either [+familiar] or [+referential] feature moves to the head of PP as lefi adjunction for feature checking with the head FP under head-head relation and for feature checking with the definite adjective under spec-head agreement. The structural representation of (41a) is as portrayed in (45) below. (45) DP D’ /\ D FP den /\ the AP F’ stora /\ big F NP bib—en | car-the N ti The insertion of another definite article is explained by ‘theta-identification’ and ‘theta-binding’ as proposed by Higginbotharn (1985) who suggests that nouns and 63 aijeetives have theta gaseous of the hour ratification takes p See-head confi gurat to ’be open positior ride is inserted in t As previously Iii an introduction r 333121- Such adj ectiv Refer Moreover, the will explore th' ‘ lS lSSL 3. mp“ and Positir adjectives have theta-roles to be discharged. According to ‘theta-identification’, the open positions of the noun and the adjective must be identified and Kester proposes that the identification takes place via a Spec-head relation. This is why N moves to F to be in a Spec-head configuration with the adjective. ‘Theta-binding’, on the other hand, requires that the open position of the noun be bound by the determiner. Therefore, a definite article is inserted in the D position. As previously mentioned, the mere-type adjectives in Thai are always present with an introduction of classifiers and always give the nominal expressions the definite reading. Such adjectives may, therefore, have the feature [+referential] as claimed by Kester. Moreover, the enclitic article en seems to have similar properties to the classifier. I will explore this issue in detail in the following section. 3.4 Types and positions of noun modifiers I will discuss the three types of noun modifiers in this section. Adjectives will be classified into two types: regular attributive adjectives and ‘referential’ adjectives (derived from mere-type adjectives (Bernstein 1993) and definite adjectives (Kester 1996)). The ‘indirect modification’ as claimed by Sproat and Shih is another type of noun modifier called ‘nominal modifiers’. Their positions will also be spelled out. 3.4.1 Regular attributive adjectives I assume that adjectives appearing right after nouns in Thai nominal expressions are regular attributive adjectives. They behave in the same way as the attributive adjectives discussed by Bernstein (1993) and therefore they are suggested to be NP adjuncts. 3.4.1.1 W Taking into co aijeetives can be used elaborated below. dtPost-copular posit: As mentioned oe ambiguity betwect ads alone without t (46) 1 flakpl‘i'. student ‘a’the d ‘A'The anmg dog ’a’the 1 ‘A-Tht Theonlydirr iIsed 5“} ModifiCati on TMSWPEora mm in (47) belovt. . 7. (4.)1 Phuufl “'Omé; b, baan3 house 6 a hon: 3.4.1.1 WWW Taking into consideration Sproat and Shih (1988) and Bernstein (1993), regular adjectives can be used predicatively, be modified and behave like relative clauses as elaborated below. (1) Post-copular position As mentioned earlier, Thai does not use the copula before an adjective. Therefore, the ambiguity between a sentence and a noun phrase generally arises when an expression stands alone without the context as shown in (46) below. (46) a. nakrrian, kha,-yan, student diligent ‘a/the diligent student(s)’ ‘A/The student(s) is/are diligent.’ b. maa, ?uan, dog fat ‘a/the fat dog’ ‘A/The dog(s) is/are fat.’ The only difference between the nominal and the clausal interpretations is that a pause is required between the noun phrase and the adjective when the sentential reading is targeted. (2) Modification This type of adjective can also be modified by an intensifier and a complement as shown in (47) below. (47) a. phuu,yin3 suay2 cat2 woman pretty very ‘a very pretty woman’ b. baan, yay2 miion, wanl house big like palace ‘a house big as a palace’ 65 t3": Similarity to rela‘ As expected, rersed in a relative t clauses as in (48), (48) a naka'd student ‘arthe s b. maa, I dog C ‘a'the t Having been 5 smein’s criteria, tl Q. - . fall I. I t. Because of the negatives are all adioi (49) a. baan, house ‘a mod b. baa.“3 ]. house ‘I 0- baan3 tl ouse r (3) Similarity to relative clauses As expected, the ability to be used predicatively is compatible with the ability to be used in a relative clause. Therefore, the examples in (46) can be turned into relative clauses as in (48). (48) a. nak4-rianl thii, khal-yan, student COMP diligent ‘a/the student(s) who is/are diligent’ b. maa, thii, ?uan, dog COMP fat ‘a/the dog(s) which is/are fat’ Having been shown to behave like regular attributive adjectives according to Bernstein’s criteria, they will be argued below to be NP adjuncts. 3.4.1.2 Rositionnfregulaaattnhutixeadiectm Because of the lack of ordering restrictions (49), I will propose that Thai adjectives are all adj oined to NP. (49) a. baan3 yay2 khaaw, thanlsalmay, house big white modern ‘a modern big white house’ b. baan3 khaaw, yay2 than,sa,may, house white big modern c. baan, than,sa,may, yay2 khaaw, house modern big white The free ordering indicates that Thai adjectives are adjoined to only one maximal projection which is NP in this case. The switching around of adjuncts of an XP is acceptable. Furthermore, I will assume that they are base-generated as NP right-adjuncts. Left adjunction is excluded here because N movement is to be assumed without motivation other than for the word order fact. 66 To sum up, ti represented in (50). (50) 3.43 Referential ad j This type of a mere-npe adjective ii earn them ‘referentiai and yields a definite r 343-1 Chllmenm Such Thai ad j: firmly the same as th mflmpular position (51) a.*naayl boss S I'el bfluuk3 child “The To sum up, the syntactic position of a regular attributive adjective is as represented in (50). (50) NP /\ NP AP 3.4.2 Referential adjectives This type of adjective behaves differently from a regular adjective. It is like a mere-type adjective in that it cannot be used predicatively nor be modified. However, I term them ‘referential’ owing to the fact that its presence always introduces a classifier and yields a definite reading. 3.4.2.1 Characteristicsnfreferentialndjectiyes Such Thai adjectives as ken2 ‘former’, diaw, ‘only’ and diaw,kan, ‘same’ behave exactly the same as the mere-type adjectives; namely, they cannot be in the predicate or post-copular position (51) and therefore they cannot be in the relative clause (52). (51) a."‘naayl kon2 boss former “The boss is forrner.’ b."‘luuk3 diawl child only “"The child is only.’ c.*phuu,yin, diaw,kan, woman same “The woman is same.’ (52) a‘naay, thii3 kon2 boss COMP former “the boss who is former.’ b."'luuk3 thii3 diaw, child COMP only “the child who is only.’ 67 c‘phuu 'WOHT “the However. tht the definite reading :1 (53) a. naay, boss I ‘The f. bfnaay, boss (54)anit, p Nit t ‘Nit is b.‘ nit, Nit (55h. nit, k Nit vs ‘Nit a h‘nit, 1 Nil v MOre impOr 5’5”“ in (56). A “LIVES may % met the Clan c.*phuu,yin,thii, diaw,kan, woman COMP same ""the woman who is same.’ However, this type of adjective in Thai is always preceded by a classifier and take the definite reading as demonstrated in (53)-(55). (53) a. naay, khon, kon2 cay,-dii, boss C1 former heart-good ‘The former boss is kind.’9 b.*naay, kon2 cay,-dii, boss former heart-good (54) a. nit, pen, luuk, khon, diaw, Nit be child Cl only ‘Nit is the only child."0 b.* nit4 pen, luuk3 diaw, Nit be child only (55) a. nit, kap2 but,sa,ba, khii, phuu,yin, khon, diaw,kan, Nit with Busaba be woman Cl same ‘Nit and Busaba are the same woman.’ b.*nit,, kap2 but,sa,ba, khii, phuu3yin, diaw,kan, Nit with Busaba be woman same More importantly, regular attributive adjectives can occur in such a position too, as shown in (56). As referential adjectives must occur in that position whereas attributive adjectives may occur there and are interpreted referentially, I will call adjectives that appear after the classifier 'referential' adjectives. (56) a. phuu3yin, khon, suun, woman Cl tall ‘the tall woman’ 9 Although ‘former’ can occur in an indefinite noun phrase in English e. g. a former boss, in Thai the noun phrase containing this adjective must be definite. The indefinite reading of ‘former’ is expressed by different words such as khaw, ‘old’ or nay, ?a ,diitz ‘in the past’. '° ‘An only child’ is expressed by a compound noun ‘quk3toon,’. 68 b. naa} boss ‘the Therefore. \\ associated with an a such an adjective ca adjectives for simpl b. naay, khon, ?uan, boss Cl fat ‘the fat boss’ Therefore, what matters here is the notion of definiteness or referentiality which is associated with an adjective rather than the type of an adjective. I thus hypothesize that such an adjective carries the feature [+referential]. Hence, I will call them ‘referential’ adjectives for simplicity. Their position will be justified below. 3.4.2.2 Bositimfreferentiamdjectim Recall the phenomenon of ‘double defrrriteness’ discussed by Kester (1996). The definite adjective is assumed to be in the specifier of an FP headed by the enclitic article en which has the feature [+referential] or [+familiar]. As discussed in Chapter 2, one function of a classifier is referentialization. The presence of a classifier triggers the specific reading (Bisang 1993, Pacioni 1996) and blocks the generic reading (Cheng and Sybesma 1999). (See section 2.2.) The classifier thus seems to have the same property as the enclitic en. I will therefore propose that a referential adjective in Thai is generated in the specifier of ClP. The fact that a referential adjective cannot directly modify the head noun is because the noun in a classifier language is a kind. A classifier is thus needed to individualize and referentialize the noun to make it compatible with the referential adjective. The structural representations of a referential adjective in Thai according to the head-initial and the head-final analyses are portrayed as in (57) and (58) respectively. (57) ClP /\ AP Cl’ /\ Cl NP 69 (53) Another pit“ instance in which t\ I (59) kluay, batman; ‘the bi Sentence (5% regular adjective ge behaves like a com] out of the set of bar Expressed by the re In short, it s hauler of C IP is r ‘\ 3'43 xmutual [n01 (58) ClP /\ AP Cl’ /\ NP Cl Another piece of evidence supporting my proposal comes from the following instance in which two opposite adjectives can co-occur. (59) kluay, lek4 luuk3 yay2 banana small Cl big ‘the big [small banana]’ Sentence (59) can be accounted for as follows. The adjective lek, ‘small’ is a regular adjective generated as NP adjunct so the constituent kluay, lek, ‘banana small’ behaves like a compound and denotes a kind (of small bananas). The classifier then picks out of the set of bananas denoted by NP an individual banana with the property as expressed by the referential adjective yay, ‘big’. In short, it seems that the proposal of a referential adjective being generated in the specifier of ClP is not unreasonable. 3.4.3 Nominal modifiers Thai noun phrases have been claimed to be able to contain more than one classifier, especially when the noun is modified by an adjective (Hundius and Kolver 1983, Hudak 1990) as in the following examples.‘1 (60) a. chan, hen, maa, tua, yay2 saam, tua, I see dog Cl big three Cl ‘1 saw three big dogs’ ” Sproat and Shih also consider this instance as [C1 + Adj]. But as I argued in the last section, an adjective after a ‘true’ classifier is a referential adjective and its presence yields merely definiteness. This type of construction therefore is not formed by a classifier and an adjective, but a classifier-like word and a regular adjective. 70 b. luuk- ballo ‘Thrt Iwill Show tithe above cases. ' togedier with the fo words ‘classifier-lik 3.4.3.1 Charisma The nominal ‘classifier-like’ mor modifies the head m I'Dressions below: (61) a. a ham b. a blue c. a whit AS earlier n: ,3. Consider (62) below (62) phuu3yij Woman ‘21 tall \\ b. luul<3-poon2 luuk3 lek, saam, luuk3 haay, pay, balloon Cl big three Cl lose go ‘Three small balloons were lost.’ I will show that this is not the fact. The true classifier is the one after the ntuneral in the above cases. The word which is homophonous with the classifier is a noun which, together with the following adjective, modifies the head noun. I will call this type of words ‘classifier-like’ morphemes (C-L) and this type of modifiers ‘nominal modifiers’ 3.4.3.1 Characteristicmfnnminalmodifiers The nominal modifiers I will talk about here are constructed from a noun or a ‘classifier-like’ morpheme (C-L) modified by an adjective and the whole unit in turn modifies the head noun. They have the similar construction to English nominal expressions below: (61) a. a hard-cover book b. a blue-collar job 0. a white-tie affair As earlier mentioned, some words are homophonous with classifiers such as tua, which can be a noun meaning ‘body’ or a classifier for inanimate objects and animals. Consider (62) below. (62) Phuusyins mar susz woman body tall ‘a tall woman’ Since tua, is not the classifier for human beings, it is used in (62) as a noun denoting the meaning of ‘body’. Therefore, an adjective which is incompatible with ‘body’ causes unacceptability as in (63). 71 (63)*phuu,yin, tua, cha,laat, woman body smart ‘a smart woman’ If the ‘true’ classifier khan, which is used with human beings is employed, both adjectives in (62)-(63) are acceptable as in (64). (64) a. phuu,yin, khon, suun, woman Cl tall ‘the tall woman’ b. phuu3yin, khon, cha,laat2 woman C1 smart ‘the smart woman’ Another difference between the classifier and the classifier-like word is that only the ‘true’ classifier (i.e. khon,) can be used with a numeral (65a) whereas the classifier- like element (i.e. tua,) cannot (65b). (65) a. phuu,yin, tua, suun, saam, khon, woman body tall three Cl ‘three tall women’ b.*phuu,yin, tua, suun, saam, tua, woman body tall three body Consequently, a noun phrase may be ambiguous between definiteness and indefiniteness when a word that can be either a classifier or a classifier-like morpheme such as tua, shows up. For instance, the noun phrase in (66) is definite when tua, is used as a classifier. On the other hand, it may be definite or indefinite when tua, is used as a noun or a classifier-like word. ' (66) khun, khaay, hen, maa, tua, dam, may4 you EXP see dog Cl/C-L black QP ‘Have you ever seen a/the black dog?’ 72 As stated ea- 1 respects as shown ti (1 t Posttopular p0 Like regular shown in (67). (67) a. phuu wom. ‘That b. luuk,j As stated earlier, nominal modifiers are similar to regular adjectives in many respects as shown below. (1) Post-copular position Like regular attributive adjectives, nominal modifiers can be used predicatively as shown in (67). (67) a. phuu3yin, khon, nan4 tua,-suun, woman Cl that body-tall ‘That woman is tall.’ b. luuk,poon2 bay, nii4 luuk3-too, balloon Cl this C-L-big ‘This balloon is big.’ (2) Similarity to relative clauses Nominal modifiers can also be put into relative clauses without any change in meaning as demonstrated below. (68) a. chan, ruu,,chak2 phuu3yin, tua,-suun, I know woman body-tall ‘I know mm tall woman.’ b. chan, ruu,,chak2 phuu3yin, thii, tua,-suun, I know woman COMP body-tall ‘I know a/the woman who is tall and sexy.’ (3) Free-ordering Nominal modifiers can recur unlimitedly as shown in (69) below and their orders can be switched around. (69) a. phuu3yin, tua,-suun, phiw,-khaaw, hunz-dii, woman body-tall complexion-white figure- good phom,-sii,-blon, khon, nan4 hair-color-blonde Cl that. ‘that sexy, tall, white, blonde woman’ b.1uuk,poon2 sii,-khaaw, luuk,-too, son,-klom, balloon color-white C-L-big shape-round 73 bay,-kaw2 bay, nan, C-L-old Cl that ‘that big round old white balloon’ It is clear from (69b) that two different morphemes quk, and bay, which can both function as the classifiers of luuk3pooy, ‘balloon’ because they denote round objects can occur together within a phrase. They can, moreover, appear anywhere after the head noun, i.e. close to it or be separated from it by some other nominal modifiers. This proves that they are in the class of nominal modifiers. And a nominal expression in Thai can have only one classifier. 3.4.3.2 Bositionnfnominalmodifiers Since nominal modifiers are similar to regular attributive adjectives in all respects concerned, I will assume that they are base-generated in the same position i.e. NP right- adjuncts as illustrated below. (70) NP 3.5 Distribution and interpretation of modified noun phrases in Thai Having provided arguments for the three types of noun modifiers together with their positions, I will discuss the distribution and interpretation of noun phrases containing each of them. 3.5.1 Noun phrases containing regular attributive adjectives A nominal expression containing a regular adjective has free distribution and can express indefiniteness (71) and definiteness (72) without number specified as well as generic reading (73) as demonstrated below. 74 (71) a. nrii, rom, khiiaw, yuu2 bon, to? 4 have umbrella green stay on table ‘There is/are a green umbrella/ green umbrellas on the table.’ b. chaam, lek, may3 poo, say2 kaeaan, rok2 bowl small not enough put curry PRT ‘A small bowl cannot hold the curry.’ (72) a. chan, tham, rom3 hay, son, khan, lasaew4 I make umbrella lose two Cl PRF ‘I have lost two umbrellas.’ chan, rak, rorn3 khiiaw, maak3 sa,,-duay3 I love tunbrella green much PRT ‘I love the green one very much.’ b. chaam, hok2 bay, tl'tii3 chan, sii4 maa, mii, son, kha,naat2 bowl six Cl COMPI buy come have two size ‘The six bowls I bought is of two sizes.’ chaam, lek4 chay,, say2 khaaw, chaam, yay2 chay4 say2 kwmnpiit2 bowl small use put rice bowl big use put soup ‘The small ones are for rice and the big ones for soup.’ (73) a. chan, choop3 chaa, khiiaw, I , like tea green ‘I like green tea.’ b. chaam, yay2 cu?2 maak2 kwaa2 chaam, lek4 bowl big hold much -er bowl small ‘The big bowl has more capacity than the small bowl .’ Only the introduction of a numeral can make the number explicit. And as stated in the previous section, a quantified noun phrase normally displays an indefinite reading as shown in (74). (74) a. mii, rom, khiiaw, nin2 khan, yuu2 bon, to? 4 have umbrella green one Cl stay on table ‘There is one green umbrella on the table.’ b. mii, chaam3 yay2 saam, bay,yuu2 bonl to?4 have bowl big three Cl stay on table ‘There are three big bowls on the table.’ 75 .V t It .. .A’] ‘ if.» («0.) in Similar to a quantified bare noun, it is impossible for a quantified noun phrase with a regular attributive adjective to carry a definite interpretation as demonstrated by the ungrammaticality of the example below. (75)*duu, rom3 khiiaw, saam, khan, si2 suay, can, look umbrella green three Cl PRT pretty very ‘Look at the three green umbrellas. They are very .pretty.’ In the following section, I will show that the assumption that nominal modifiers and regular attributive adjectives in Thai are syntactically equivalent is firrther supported by the interpretation. 3.5.2 Noun phrases containing nominal modifiers Since nominal modifiers are syntactically similar to regular attributive adjectives, they behave in the same way in terms of interpretation and distribution. That is, they are distributionally free and they can exhibit an indefinite (76) and a definite reading (77) without number specified as well as generic (78). (76) a. rrrii, phuu3yin, tua,-suun, maa, haa, khun, have woman C-L tall come see you ‘There is/are a tall woman/tall women conring to see you.’ b. maa, tua, yay2 kat2 dek2 dog C-L big bite child ‘A big dog bit a child.’ (77) a. mii, phuu,yin, maa, sam,phaat3 naan, son, khon, have woman come interview job two Cl ‘Two women came for job interviews.’ phuu,yin, tua, suun, duu, cha,laat, woman C-L tall look smart ‘The tall woman looks smart.’ b. maa, haa3 tua, kat2 kan, rraa3 baan3 dog five Cl bite RECP front house ‘Five dogs fought in front of the house.’ 76 .‘w a» defi maa, tua, yay2 suu3 maa, tua, lek4 may3 day, dog C-L big fight dog C-L small not can ‘The big dogs lost to the small ones.’ (78) a. chan, chaap, phuu,yin, tua, suun, I like woman C-L tall ‘I like tall women.’ b. maa, tua, yay2 mak, duu2 dog C-L big usually fierce ‘The big dog is usually fierce.’ When a noun phrase with a nominal modifier is quantified, the phrase gets an indefinite reading as in (79). (79) a.*duu, phuu,yin, tua,-suun, saam, khon, si2 look woman C-L tall three Cl PRT 'Look at the three tall women.’ b. mii, phuu3yin, tua,-suun, saam, khon, yuu2 nay, ban3 have woman C-L tall three Cl stay in house There are three tall women in the house.’ To sum up, noun phrases containing nominal modifiers are like noun phrases with regular adjectives in both interpretation and distribution. 3.5.3 Noun phrases containing referential adjectives Recall that a noun phrase containing a referential adjective can take only a definite, specific reading. That is why it usually occurs as an answer to the question NP CI which ‘which N’ as illustrated below. (80) A: chan, yaak2 day, muak2 saam, bay, I want get hat three C1 ‘1 want three hats.’ B: saam, bay, nay, three Cl which ‘Which three?’ A: saam, bay, lek, three Cl small ‘The three small ones.’ 77 1’" .4. A’s answer to B saam, bay, Iek, ‘the three small ones’ implies that there are only three small hats in that place. Not Surprisingly, a noun phrase containing a referential adjective cannot display an indefinite or a generic reading as demonstrated below. (81) a."'mii, rom, khan, khiaw, yuu2 bon, to? 4 have umbrella C1 green stay on table “There is the green umbrella on the table.’ b.*phuu,chaay, khon, ?uan, may, naa,duu, man Cl fat not good-looking ‘Fat men are not good-looking.’ Concerning number, such a noun phrase always denotes singularity. This fact can explain why (82) is unacceptable. (82)*naay, khon, kon2 saam, khon, boss Cl former three C1 Since naay, khan, kan, ‘the former boss’ canies a definite singular reading, the attachment of saam , khan, ‘three Cl’ causes contradiction. To quantify such a construction, a number has to be inserted in front of the classifier as in (83). Unlike other quantified noun phrases, however, the quantified noun phrases with referential adjectives are definite. (83) a. rom, saam, khan, khiiaw, suay, thii, sut, umbrella three Cl green pretty -est ‘The three green umbrellas are the prettiest.’ b. chan, may, chaap, naay, saam, khon, khan, I not like boss three Cl former ‘I don’t like the three former bosses.’ And as seen in (83), a noun phrase with a referential adjective is free in distribution like definite bare nouns. 78 In brief, the noun phrase with a referential adjective carries only a definite interpretation. When the numeral is absent, the singularity is the sole reading available. Generic interpretation is impossible for such a construction. 3.6 Conclusion I have shown that Thai possesses a word class of adjectives which are generated in two different positions. Moreover, I have also argued for a third type of noun modifier which has been claimed in the previous literature to be a classifier. I suggest that they are, in fact, a type of noun called ‘classifier-like’ words. These words together with their modifying adjectives modify the head nouns of the phrases. They behave very similarly to regular adjectives in various respects. The two types of adjectives in Thai are classified according to their interpretation and distribution. Regular attributive adjectives can express either a definite or an indefinite reading and they appear right after nouns. Referential adjectives, on the other hand, render noun phrases containing them the definite reading and have to be preceded by a classifier. I, therefore, propose a different position for each type. Regular attributive adjectives are generated as NP right-adjuncts but may also appear as the specifier of ClP whereas referential adjectives must be generated as the specifier of ClP. Since nominal modifiers are similar to regular adjectives in all aspects, they are assumed to be NP right- adjuncts. Modified noun phrases in Thai, like bare nouns, are not restricted in distribution. Their differences in interpretation are summarized in the table below. 79 Definite Indefinite Generic Singular Plural Singular Plural N + Regular APs + + + + + N + Nominal modifier + + + + + N + Referential APs + + - - - Table 1: Summary of interpretations expressed by noun phrases containing different types of noun modifiers. 8O lnthis ties of non analysis is to torrential ad phase in Th 4.1 luttodut In C! had-final a1 “1933- The 1: “Petitions, l w; flabGrated ,‘ adlt‘ttives a llill. lhftefc CHAPTER 4 MODIFIED NOUN PHRASES In this chapter I will provide the analyses of noun phrases modified by the three types of noun modifiers mentioned in the previous chapter and show that the head-final analysis is to be discarded because it fails to account for noun phrases containing referential adjectives and demonstratives. This strongly supports the idea that the noun phrase in Thai is strictly head-initial. 4.1 Introduction In Chapter 2 I argued that the head-initial analysis is more attractive than the head-final analysis because it requires only NP movement to account for all the relevant cases. The head-final analysis, on the other hand, needed three distinct syntactic operations. I will explore in this chapter Thai noun phrases containing the noun modifiers elaborated in Chapter 3. As discussed there, noun phrases containing regular attributive adjectives and nominal modifiers display the same interpretation and distribution. They will, therefore, be discussed together. Noun phrases containing referential adjectives, on the other hand, exhibit different behaviors than the other two; that is, they can take only a definite reading. I will show that the latter type of modified noun phrases causes problems for the head-final approach. 81 Besid As mentione- pinse as illt (Ha. I wil mttnretatio The Besides the three types of noun modifiers, demonstratives will also be discussed. As mentioned earlier, the demonstrative in Thai is the terminal element in the noun phrase as illustrated below. (1) a. phuu,yin, tua,-suun, khon, nan4 woman body-tall Cl that ‘that tall woman’ b. rom, khiaw, khan, nii4 umbrella green Cl this ‘this green umbrella’ I will show that Thai demonstratives display the same distribution and interpretation as referential adjectives in some cases but not in others. The outline of the chapter is as follows. Section 2 concerns the analyses of noun phrases containing each of the three types of noun modifiers mentioned in Chapter 3. Section 3 deals with Thai demonstratives in terms of their distribution, interpretation and suggested syntactic position, and an analysis is given in section 4. Section 5 concludes the chapter. 4.2 Analyses of modified noun phrases in Thai Because the interpretations and distributions of noun phrases containing regular attributive adjectives and those containing nominal modifiers are the same, they will be discussed together. I will show that they can be successfirlly analyzed by both the head- irritial and the head-final approaches. Noun phrases containing referential adjectives, on the contrary, require additional operations. I will show that the head-final approach encounters problems when dealing with these cases. 82 4.2.1 A noun phrase with a regular attributive adjective/a nominal modifier Recall that a regular attributive adjective and a nominal modifier are suggested to be generated as NP right-adjuncts. Noun phrases containing either of them have the same distribution and interpretation as bare nouns. Therefore, it is not surprising that the similar analyses can be provided. 4.2.1.1MQdifiednmmnhrasemithQuLexnlicitnumerals As discussed in Chapter 3, noun phrases containing regular adjectives or nominal modifiers can express indefinite (2), indefinite (3) and generic (4) readings. They can be singular or plural unless a numeral is lexically present (which disarnbiguates the number interpretation and forces an indefinite interpretation). They are also free in terms of distribution. The examples are from Chapter 3, repeated below. (2) a. mii, rom, khiiaw, yuu2 bon, to?4 have umbrella green stay on table ‘There is/are a green umbrella/ green umbrellas on the table.’ b. mii, phuu,yin, tua,-suun, maa, haa, khun, have woman C-L tall come see you ‘There is/are a tall woman/tall women coming to see you.’ (3) a. chan, tham, rom, hay, son, khan, lzeaew4 I make umbrella lose two Cl PRF ‘I have lost two umbrellas.’ chan, rak4 rom, khiiaw, maak, sa4-duay, I love umbrella green much PRT ‘I love the green one very much.’ b. mii, phuu,yin, maa, sam,phaat, naan, son, khon, have woman come interview job two Cl ‘Two women came for job interviews.’ phuu,yin, tua, suun, duu, cha,laat2 woman C-L tall look smart ‘The tall woman looks smart.’ 83 lw’ with the in iitepretati distribution reeds oven rIresentan'. (5n to HIT. NP Pillar, ll'Oma (4) a. chan, choop, rom, khiiaw, I like umbrella green ‘I like green umbrellas.’ b. chan, choop, phuu,yin, tua, suun, I like woman C-L tall ‘I like tall women.’ I will propose along the same line as the analysis of bare nouns that such phrases with the indefinite reading are NumPs while the ones with the definite or generic interpretation are DPs. Following the previous analysis, to account for the free distribution as well as the indefinite and the definite readings, the head-initial approach needs overt NP movement to [Spec, NumP] and [Spec, DP] respectively. The structural representations of (2) and (3-4) are illustrated in (5) and (6) respectively. (5) a. NumP //\ NP, Num’ /\ /\ NP AP Num ClP ram, khiiaw, /\ umbrella green Cl - NP tr b. NumP /\ NP, _ Num’ /\ /\ NP NP Num ClP phuu,yin, tua, suay, /\ woman body tall Cl NP 84 bare noun 1'6 thus it i" l .4) (6) a. DP /\ NP, D’ /\ /\ NP AP D NP ram, khiiaw, t, umbrella green b DP /\ NP, D’ /\ /\ NP NP D NP phuu,yin, tua, suay, t, woman body tall The head-final approach to (2)-(4) is also the same as the head-final analysis of bare nouns; that is, it requires N movement to Num and D at LF respectively. (7) and (8) are thus the structural representations of (2) and (3-4) respectively.‘ (7) a. NumP /\ ClP Num w NP Cl /\ NP AP I khiiaw, N green ram, umlbrella 85 | tua,suun, N body tall phuuayirzs umbrella L (8) a. DP ram, umbrella | tua , snug, N body tall phuutvins woman I In sum, both approaches do not face any problems dealing with these instances because they behave exactly like bare nouns. ' Again although I assume that the internal structure of (8) consists of NumP and ClP, the structure given is a simplified one. 86 OT llOll'. lexical .‘v'umP' 4-2-1-2 Quantifiedrnndifiednnunnhrases Similar to quantified bare nouns, quantified noun phrases with regular adjectives or nominal modifiers can denote only indefiniteness and they are not restricted to lexically-governed positions as shown in (9) below. Therefore, the NP is embedded in NumP with the head Num lexically filled with a numeral. (9) a. rom, khiiaw, saam, khan, yuu2 bon, to?,, umbrella green three Cl stay on table ‘Three green umbrellas are on the table.’ b. chan, hen, phuu,yin, tua, suun, saam, khon, I see woman body tall three Cl ‘1 saw three tall women.’ The noun phrases in (9a, b) are structurally represented according to the head- initial approach as in (10a, b) respectively. As in the other cases, the overt raising of NP is needed. The target of the moved NP in this case is the specifier of NumP. (10) a. NumP /\ NP, Num’ /\ /\ NP AP Num ClP rom, khiiaw, saam, /\ umbrella green three Cl NP khan, t, b NumP /\ NP, Num’ /\ /\ NP NP Num ClP phuu,yin, tua, suay, saam, /\ woman body tall three Cl NP khan, t, 87 Sinil account 0qu (ill: 74 Similarly, the head-final approach requires the overt raising of C1 to Num like the account of quantified bare nouns as illustrated below. (1 1) a. NumP //\ ClP Num /\ /\ NP Cl Num Cl /\ t, saam, khan ,, NP AP three ram, khiiaw, umbrella green b. NumP /\ ClP Num /\ /\ NP Cl Num Cl /\ t, saam, khan ,, NP NP three phuu,yin, tua,.suuy, woman body tall Summarizing, it is apparent that both approaches are able to handle modified noun phrases containing regular attributive adjectives and nominal modifiers. However, the final-headedness of Thai noun phrases is to be rejected when noun phrases containing referential adjectives are discussed. 4.2.2 A noun phrase with a referential adjective As claimed in the previous chapter, a referential adjective is generated in [Spec, ClP] and its presence always gives a noun phrase a definite interpretation. Therefore, a noun phrase containing a referential adjective is always a DP. I will present in this section that besides overt NP raising, the head-initial approach also involves overt Cl-to-Num movement. However, an additional syntactic operation does not hurt the head-initial 88 approachlln ion phrases nodded by : till \jgd Sin: reading. the Elli the p' with) I approach. On the contrary, it can ultimately be concluded that the headedness of Thai noun phrases is initial because the head-final approach fails to account for a noun phrase modified by a referential adjective. 4.2.2.1 Modifiednmmnhrasesmthoumxnlicitnumerals Since a noun phrase containing a referential adjective can have only a definite reading, the phrase is embedded in DP. Given that the referential adjective is in [Spec, ClP], the phrase naay, khan, kan, ‘the former boss’ in (12) has the underlying structure as in (13), according to the head-initial approach. (12) naay, khon, kon2 cay,dii, boss Cl former heart-good ‘The former boss is kind.’ (13) ' DP /\ D NumP /\ Num - ClP /\ AP Cl’ k9nz /\ former Cl NP khan, naay, boss Like all other definite noun phrases, NP-to-DP movement is required by the head- initial approach for the definite interpretation and the free distribution. However, I will propose an additional movement of C1 to Num as adjunction in the syntax of Thai in this case to account for the specific singular reading. As it is widely believed that a classifier and a numeral form a constituent (Tang 1990, Gao 1994, Croft 1994) and that a ClP has a role in expressing grammatical Number (Cheng and Sybesma 1999), the idea of 89 Clio-N sng‘rla tilled .\ tinctio tell. t1 (Chen 0? he "it'll-”v @335 Cl-to-Num movement is not unusual. Without an overt number, the sole interpretation is singular. Hence, I would hypothesize that when Cl is present without a non-lexically filled Num, the default ‘one’ interpretation is taken. In other words, the singularizing function of a classifier takes firll effect when Num is not lexically explicit. I suggest that the movement of C1 to Num suppresses the non-specific numeral reading. Recall that the [N + Cl] construction is always indefinite non-specific singular. The non-specific reading is due to the absence of Cl-to—Num movement as proposed in Chapter 2, section 2.5. As well, the presence of the classifier prevents a noun phrase from taking a generic reading (Cheng and Sybesma 1999). The noun phrase in (12) is thus structurally represented in (14) below. (14) DP naay,, D’ boss /\ ‘ D NumP /\ Num ClP /\ /\ Ntun Cl AP Cl’ 2 khon,, kan, /\ former Cl NP t, t, ' J Turning to the head-final analysis, a problem arises, because the word order facts of the noun phrase containing a referential adjective cannot be accounted for. Let me clarify the point. Taking the noun phrase in (12) into account, its underlying structural representation according to the head-final approach is as in (15). 90 (15) DP /\ D’ /\ NumP D /\ ClP Num /\ AP Cl’ k9": /\ former NP Cl naay, khan , boss Given that Cl-to-Num raising is generally required by the head-final approach when a numeral is present, the movement is suggested to take place here due to the default ‘one’ interpretation of the null Num as well as the specific numeral reading. The unrestricted distribution and the definite interpretation are taken care of by the covert movement of N to D as in other cases of definite noun phrases. However, this analysis fails to get the word order right: instead of naay, khan, kan, ‘boss Cl former’, it predicts kan, naay, khan, ‘former boss Cl’. Even overt NP-to-DP movement (which could, in these cases, replace covert N-to-D raising to capture the interpretation and distribution) cannot save the analysis because then the derived word order would be naay, kan, khon,‘boss former Cl’. Since adjectives are not subject to general movement processes (Bernstein 1993:32), should we discard the head-final account of Thai noun phrases from this moment? Is there a way to rescue the approach? We might save the head-final analysis in this particular case by allowing specifiers to the right of XP in the Thai noun phrase. Therefore, the syntactic representation of (12) with the rightward specifier might be as in (16). 91 (16) DP /\ NumP D /\ ClP Num /\ Cl’ AP /\ k9": NP Cl former naay, khon , boss Although the structure in (16) represents the correct word order, it does so at the expense of allowing right-hand specifiers which are to my knowledge unattested in any language. The head-final approach still requires at LF Cl-to-Num movement for the specific singular reading and N-to-D movement for the definite interpretation and the free distribution. Although covert movement is less costly than overt movement, the head- final approach becomes inconsistent because it requires covert movement in some cases and overt movement in others. That is, C1 has to overtly move to adjoin to Num whenever a numeral is lexically present while the covert raising is required in this particular case where a noun phrase is modified by a referential adjective without an explicit numeral. Moreover, we will see below that the head-final approach cannot still be saved in spite of the rightward specifier when a noun phrase containing a referential adjective is quantified. 4.2.2.2 Qnamifiedmndifiednnunnhtasss Unlike the other quantified noun phrases, a quantified noun phrase modified by a referential adjective carries a definite interpretation as illustrated in (17) below. 92 Bee initial app: for the nor interpreter ll Al to [he heat Imtl9). (17) naay, khon, may2 may, miian, naay, saam, khon, ken2 boss Cl new not like boss three Cl former ‘The new boss is not like the three former ones.’ Because the phrase is definite, it is embedded in a DP. According to the head- initial approach, the NP-to-DP raising together with Cl-to—Num movement as proposed for the non-quantified construction is sufficient to capture the distributional and interpretational facts as well as the word order as shown in (18) below. (18) DP /\ NP D’ naay,, /\ boss D NumP T /\ Num ClP /\ /\ Num C1 AP Cl’ saam, khon,, kan, /\ three ‘ former Cl NP t, I, Adopting the assumption that the specifier of XP in Thai is on the right according to the head—final analysis, the underlying syntactic representation of the phrase in (17) is as in (19). 93 (19) DP NP Cl former naay, khan, boss Recall that the head-final approach accounts for a noun phrase with a referential adjective without an explicit number by covert Cl-to-Num and N-to-D movements. The same operations should have taken place in the analysis of the quantified noun phrases due to the same interpretation and distribution. However, the word order cannot be handled. To conclude, the head-final analysis fails to handle noun phrases containing referential adjectives. I will discard it from this point on. 4.2.3 A noun phrase with a regular attributive adjective! a nominal modifier and a referential adjective I have argued that APs are allowed in two different positions in Thai. This analysis can provide an explanation for instances such as (20) where two adjectives with contradictory meanings can co-occur without causing ungrammaticality or contradiction. (20) a. phuu,yin, suun, khon, tiia, rraa,rals:4 can, woman tall C1 short cute very ‘The [short [tall woman]] is very cute.’ b. chan, yaak2 day, luuk,poon2 luuk, too, bay, lek4 I want get balloon C-L big Cl small ‘I want the [small [big balloon]].’ 94 adje incl: case indit deno are it but ( the at \l’ ac lrelen noun 1 Classil bats t] Isth. I will show that the lack of contradiction is due to the fact that the phrases consist of noun modifiers generated in two different positions. 4.2.3.1 W The phrases in (20) are interpreted as definite and singular. That two contradictory adjectives can co-occur without ungrammaticality results from the fact that an NP including an adjoined AP (regular adjective) or NP (nominal modifier) in this particular case seems to denote a kind or concept or a set of entities and the classifier picks out an individual from the kind denoted by NP. The picked individual must have the property as denoted by the referential adjective. That is, the meanings of the noun phrases in (20a, b) are roughly equivalent to ‘out of the set of all tall women the short one is picked’, and ‘out of the set of all big balloons the small one is picked’ respectively. This indicates that the noun phrases in (20) contain modifiers generated in two different positions- one is an NP adjunct (regular AP/ nominal modifier) and the other is the specifier of ClP (referential AP). It is the latter type of adjective together with the classifier that gives the noun phrase a definite interpretation. And as I have claimed, the mere presence of a classifier without an explicit numeral yields a singular reading; Cl-to-Num movement bans the non-specific numeral reading. The syntactic representations of the above phrases are, therefore, as shown in(21). 95 plm wor hum balloor ofdgffe, Imam, 53¢ ell VP llltlu, NP, D’ /\ /\ NP AP D NumP phuu-yin, suay, /\ woman tall Num ClP /\ /\ Num Cl AP Cl’ 9 khon,, tiia, /\ short Cl NP t, t, b. DP /\ NP, D’ /\ /\ . NP NP D NumP qukpooy, luuk, too, /\ balloon C-L big Num ClP /\ /\ Num Cl AP Cl’ 9 bay,, 16k, /\ small Cl NP t, t, It is likely that the analysis is on the right track as the reverse order of adjectives of differently base-generated positions is impossible as shown below. (22) a.*phuu,yin, khon, tiia, suun, woman Cl short tall b.*phuu,yin, khon, tiia, tua suun, woman Cl short C-L tall It should be noted that (22a,b) cannot be interpreted as phrases, but can be sentences. Their meaning, however, is ‘The short woman is tall.’which is contradictory. These examples also support the proposal of NP movement in Thai. That is, the whole NP including its adjunct has to move together. Assuming that the noun phrases in (22) 96 C01llall adjunc vona iit' Wad .. EOBlal as den adTeen lllOl‘er P11 contain NP adjuncts suay, ‘tall’ and tua snug, ‘C-L tall’, the movement of NP without its adjunct i.e. phuu,yin, ‘woman’, instead of phuu ,yiy, suay, or phuu,yin, tua suay, ‘tall woman’ leads to unacceptability. 4.2.3.2 Quantififlmodifiedmunflrrases Like quantified noun phrases with referential adjectives, the nominal expressions containing modifiers of two positions with explicit numerals always express definiteness as demonstrated in (23). (23) a. phuu,yin, suun, saam, khon, tiia, naa,rak, can, woman tall three Cl short cute very ‘The three [short [tall women]] are very cute.’ b. chan, yaak2 day, luuk,poon2 luuk, too, saam, bay, lek4 I want get balloon C-L big three Cl small ‘I want the three [small [big balloons]].’ Similar to the discussion given for quantified noun phrases containing referential adjectives, the head-initial analysis manages to account for this particular case with NP movement and Cl-to- Num raising as in (24). (24) a. DP /\ NP, D’ /\ /\ NP AP D NumP phuu,yin, suay, /\ woman tall Nurn ClP /\ /\ Num Cl AP Cl’ saam, khan ,, tiia, /\ three short Cl NP t, t, 97 /\ NP, D’ /\ /\ NP NP D NumP qukpooy, luuk, too, /\ balloon C-L big Num ClP /\ /\ Num Cl AP Cl’ saam, bay ,, lek, /\ three small Cl NP 1,- t, In short, it is conclusive by now that Thai noun phrases are head-initial for the following reasons. First of all, the head-initial analysis requires only two syntactic operations to deal with all the cases under examination, namely, NP movement and Cl- to-Num raising with the consistency holding throughout. More importantly, it can handle . all the cases without any problem whereas the head-final analysis completely fails to account for noun phrases containing referential adjectives. 4.3 Demonstratives Having considered noun phrases modified by different types of noun modifiers, I will discuss noun phrases accompanied by demonstratives in Thai. I will first outline the characteristics of Thai demonstratives. Then I will present the study by Bernstein (1996) which argues that DemP is in the specifier of an FP between DP and NP. I will suggest that the Thai demonstrative is generated in the specifier of ClP. 98 4.3.1 Thai demonstratives Thai has four demonstratives nii, ‘this/here’, nan, ‘that/there’, noon, ‘yonder’ and nuun, ‘far yonder”. They differ in the degree of distance fiom the speaker. In terms of distribution, a demonstrative is the last element in a Thai noun phrase. The following examples, which differ only in the positions of the demonstratives exhibit a noun phrase and a clause respectively. (25) a. baan, yay2 lan, nii4 house big C1 this ‘this big house’ b. baan, lan, nii, yay2 house Cl this big ‘This house is big.’ The demonstrative in Thai is usually preceded by a classifier. It is also the case that the presence of a demonstrative with a classifier normally brings about definiteness. For example, a quantified noun phrase which is generally indefinite becomes definite 2 The demonstrative pronouns in Thai are different from the demonstrative adjectives only in tones. They consist of nii, ‘this/here’, nan, ‘that/there’, noon, ‘yonder’ and nuun, ‘far yonder’. This set of words can also be used as locative adverbials. The examples of these two types of words are provided below. (i) a. nii, khii, khruu, chan, this be teacher I ‘This is my teacher.’ b. nan, ?a,ray, that what ‘What is that?’ (ii) a. ma, nii, come here ‘Come here.’ b. khaw, yuu2 nan, he stay there ‘He is there.’ 99 when it is accompanied by a demonstrative as illustrated in (26) below. This is reminiscent of the quantified noun phrase with a referential adjective. (26) a. chan, soon, dek2 saam, khon, I teach kid three Cl ‘I teach three kids.’ b. chan, soon, dek2 saam, khon, nan4 I teach kid three Cl that ‘I teach those three kids.’ Nonetheless, there are cases where a demonstrative is present without a classifier as in (27) where the interpretations of the noun phrases are definite specific with either singular or plural reading (27a) and generic (27b), depending on the context. (27) a. chan, hen, dam, sii4 nan,sii, nii4 miio,waan, I see Dam buy book this yesterday ‘I saw Dam buy this/these book(s) yesterday.’ b. nok, nan, mak4 rrrii, khon, dam, bird that usually have feature black ‘That (kind of) bird usually has black features.’ The interpretation of magic, nii, ‘this/these book(s)’ in (27a) can be either ‘a copy of this particular book’, ‘copies of this particular book’ or ‘copies of these particular books’ (say ‘a copy of Knowledge of Language’, ‘copies of Knowledge of Language’ or ‘a copy of Knowledge of Language and a copy of Minimalist Program’) whereas the phrase nok, nan,‘that bird’ in (27b) is better interpreted as ‘the bird of that particular kind’ (say ‘crows’ or ‘starlings’). Before investigating where the Thai demonstrative is generated, I will present the study on the status and locus of the demonstrative by Bernstein (1996) below. 100 4.3.2 Bernstein (1996) Bernstein regards DemP as a functional phrase separated from DP since there exist some languages where an article and a demonstrative can co-occur (e. g. Spanish, Romanian, Hungarian and Javanese (examples from Bernstein 1996:7, citing Delsing 1988)). (28) ez a haz (Hungarian) this the house ‘this house’ (29) ika n anak (Javanese) this the child ‘this child’ In the Romance and Germanic languages, in which definite articles and demonstratives are in complementary distribution, Dem must move as substitution to D. But in English and some other languages, demonstratives can have two interpretations: deictic (30a) or indefinite specific (30b) (examples fiom Bernstein 1996:10). (30) a. this woman (right here) = this woman b. this woman (from Paris) = a woman Bernstein (1996) claims that the difference in meaning is due to the presence or absence of overt movement of Dem to D. The deictic interpretation results fiom the raising of a demonstrative with the [+definite] feature to D before Spell-out whereas the indefinite specific interpretation does not require a demonstrative containing the [- definite] feature to raise up to D. Having considered some languages which possess such demonstrative reinforcers as here and there like English in (31) and French in (32), she further suggests that DemP lOl is in the specifier position of a functional phrase headed by a reinforcer (whether or not such reinforcers are present in the language) and the FP is immediately below DP. (31) this here guy (32) cette femme-ci this woman-here ‘this woman’ The structure for (31) is illustrated in (33) below. One interpretation or the other (i - e. deictic or indefinite specific) is caused by the overt or covert movement of Dem to D, which does not affect the surface word order in English. (33) DP /\ D FP /\ DemP F’ this /\ F NP here guy Similarly, Dem-to-D movement takes place in syntax in French. In addition, the noun has to move and adjoin to FP. This can account for the word order of (32) where a noun is in between the demonstrative and the reinforcer as demonstrated in the structural representation below: (34) DP /\ D F P cette, /\ femme, F P /\ DemP F’ i /\ Dem F NP t, ci t, l 02 Thor; In French, the only interpretation available is deictic. Therefore, Dem obligatorily raises and substitutes for D. But it is NP, not the head N, that moves up lefiwards and adjoins to PP. The evidence is from a noun phrase comprising a head noun and an adj ectival modifier that appears between a demonstrative and a reinforcer as in (3 5) (Bernstein’s: 1 5). Its syntactic representation which shows that the NP moves up and adjoins to FP is given in (36). (35) cc livre jaune ci this book yellow here ‘this yellow book’ (36) DP /\ D F P cej /\ livrejaune, F P DemP F’ | /\ Dem F XP tj cz t, Bernstein suggests that a ‘strong’ feature of the head F in Romance languages forces the movement of XP containing NP to its specifier. I will take the proposal made by Bernstein that DemP is in the specifier of a F P situated between DP and NP. However, I will suggest that Thai demonstratives are in the specifier of ClP like referential adjectives. 4'33 l’tlsition of Thai demonstratives I will argue in what follows that Thai demonstratives are generated in the specifier 0f ClP - The difference in number interpretation is caused by the presence or absence of the claSSifier. 103 4.3.3.1Afiemonstrati1mmaclassifier There are a number of similarities between demonstratives and referential adjectives which suggest that they are generated in the same positions. (1) Demonstratives and referential adjectives are usually accompanied by the classifier as in (3 7) and therefore rendering the noun phrase definiteness even when the noun phrase is quantified as in (38). (37) a. naay, khonl kon2 boss C1 former ‘the former boss’ b. naay, khonl nan4 boss C1 that ‘that former boss’ (3 8) a. naay, saamS khonl kon2 boss three C1 former ‘the three former bosses’ b. naay, saam, khonl nan4 boss three C1 that ‘those three former boss’ Apart fi'om the interpretation, their distributions are found to be exactly the same too- (2) Like referential adjectives, Thai demonstratives may not take a classifier-like word- They can only be preceded by a ‘true’ classifier’. (39) a."'naayl tual kon2 boss C-L former b."‘naayl tua, nan4 boss C-L that 3 ffs file selection of Thai classifiers is determined by the social status, degrees 0f wh ality, degrees of respect, etc., tua, can be used as a classifier for a human being only n at person is treated as much inferior or with contempt. 104 (3) The co-occurrence of a demonstrative and a referential adjective leads to ungrammaticality as in the following examples. (40) a. *naay, khon1 kon2 nan4 boss Cl former that ¢that former boss b.*phuu3-yirjs khonl suurj5 nan4 woman Cl tall that atthat tall woman . It should be noted that (40a,b) are fine if nan, is regarded as a topic element as shown in (41). The word nan, does not function as a demonstrative in this case. (41) a.naayl khonI kon2 nan4 chanS ruu4 waa.3 cay,-diil boss Cl former TOP I know COMP heart-good ‘The former boss, I know is kind.’ b. phuu,yin, khonl suurj5 nan4 chan, khit3 waa3 khaw, chaap, woman Cl tall TOP I think COMP he like ‘The tall woman, I think he likes.’ To express the concepts of ‘that former boss’ and ‘that tall woman’, however, the phrases in (42a,b) must be used respectively.4 ‘ I consider (i) and (ii) below which contain two classifiers and a possible co-occurrence of a referential adjective and a demonstrative as two separate phrases due to the requirement of an intonation break and their similar usage to appositive. (i) naay, khonl kon2 // khon,nan4 boss Cl former C1 that (ii) phuu,yin, khonl 81111135” khon, nan4 woman Cl tall C1 that The full interpretations of the above are somewhat like the following. (iii) naay, khonl kon2 sir], khii, naay, khonl nan, boss Cl former COMP be boss C1 that ‘the former boss who is that one’ (iV) phuuainJs khon: 81111le sins khiii phuusyins khon. nan. woman Cl tall COMP be woman Cl that That _ ‘the tall woman who is that one’ Is Why this type of construction always carries an emphatic sense. 105 (42) a. naay, khonl nan4 thii3 yuu2 maa, kon2 boss Cl that COMP stay come before ‘that boss who was here before (this one)’ b. phuu3-yi135 8qu], khonl nan4 woman tall C] that These facts clearly show that the locus of Thai DemP is the same as that of referential adjectives, namely the specifier of ClP as represented in (43). (43) ClP /\ DemP/AP (31’ /\ Cl NP 4-3.3.2 Admcnsnanmthomaclassificr As mentioned earlier, a noun phrase containing a demonstrative without a classifier displays a variety of interpretations depending on the context, such as definite specific without number specified (44a) and generic (44b). (44) a. chan5 sii4 narjssiis nii4 maa, khaw,, my4 baat2 I buy book this come only 100 Baht ‘I bought this/these book(s) for only 100 Baht.’ b. naijssiis nii4 mak, naa3biia2 book this usually boring ‘This (kind of) book is usually boring.’ The first impression may be that a demonstrative used in such an instance is like a r egular attributive adjective. However, it happens that these two types of noun modifier d0 DOt possess the same characteristics as presented below. ( 1) Demonstratives cannot be used in a predicative context as in (45). (45)* nanssii, nii4 book this ‘The book is here.’ 106 (2) Unlike regular attributive adjectives, demonstratives cannot be modified. (46) * narjssii5 nii, pe?4 book this right ‘the book right here.’ (3) The position of demonstratives seems to be fixed. That is, demonstratives always follow regular adjectives or nominal modifiers. The reverse order brings about a sentential reading as shown below. (47) a. maaS yay2 nii4 dog big this ‘this big dog’ b. maas nii4 yay2 dog this big ‘This dog is big.’ (48) a. narjssiis sii, dear], nii4 book color red this ‘this red book’ b. narjssiis nii, siis damn, book this color red ‘This book is red.’ The data above shows that Thai demonstratives should not be generated as NP adjuncts like regular adjectives or nominal modifiers. This amounts to saying that they must always appear in [Spec, ClP] and do not have the option of appearing adjoined to NP. 'I‘he differences in interpretation are due to the presence or absence of a classifier. 0‘”ng to its singularizing or individualizing function, the presence of a classifier always yields the number interpretation i.e. either default ‘one’ or plurality as specified by a numeral (Hundius and Kolver 1983). A noun phrase without a classifier is thus aInbigllous between singularity and plurality. This proposal may seem inconsistent with the eaI‘lier claim about referential adjectives which always require classifiers. An 107 explanation can be given by the remark made by Hundius and Kolver (1983: 176) that ‘demonstratives by themselves always serve the very function of referential identification, thus there is no need to employ classifiers in any referential fimction.’ Adjectives, on the contrary, ‘are referentially neutral; their combination with a classifier may serve to compensate this deficiency and thus bring about a definite interpretation of the phrase.’ An alternative way of thinking is that when a modified noun phrase takes a definite reading with number unspecified, the noun modifier which has been assumed to be generated as NP adjunct is in fact generated in the specifier of ClP with the empty Cl. It should also be noted here that the question of whether Thai demonstratives are a Madroot (volition>obligation>pennissionlability) 5 See Ernst (1997) and Shaer (1998), among others, for different approaches to adverb analyses. 126 As observed from (24), the epistemic modals are located higher than the root ones. One piece of evidence supporting his claim is from Scots English which allows double modals; the first and the second of which express the epistemic and the root senses respectively (Brown 1992:77). (25) a. He must can do it; He can surely do it. b. He should can do it; He can likely do it. Unlike the epistemic modals, modals with the root sense are claimed to be of three distinct types which are generated in three different positions of the following ordering: volition>obligation>permission/ability. It should be noted that the modals denoting permission and ability are assumed to occupy the same position. The study of these two modals in Thai will confirm this assumption. However, the distinction between strong and weak obligation modals is not made by Cinque“. Examination of these two types of modals in Thai seems to suggest that they do not occupy the same position. I will start with the root modals and then deal with the epistemic ones. Since the former can appear either before or after VP, the pre-verbal root modals will be examined first, starting fiom the one closest to the VP. The epistemic modals will then be studied. 5.4.1 Root modals The Thai modals with the root sense are naa,, khuan, ‘weak obligation’, ca .7, ‘will’, toy, ‘strong obligation’, day, ‘permission’ and day,, pen ,, way,‘ability’. As already stated, some root modals occur pre-verbally while others, post-verbally. The discussion, ° Obligation is distinguished according to degrees of strength into weak and strong. That is, ‘if a weak obligation is not fulfilled, the consequences are not too serious; but the consequences of not fulfilling a strong obligation are much more severe.” 127 therefore, will start with the pre-verbal modals beginning with the one closest to VP, followed by the post-verbal modals. 5.4.1.l_day,:pennissignj The pre-verbal word day, has several meanings. Hence, it has been treated according to its denotations as a past tense marker (Kanchanawan 1978), a regular verb when it carries the meaning of ‘having an opportunity to do something’ (Sookgasem 1990) or ‘get’, and an auxiliary denoting permission (W arotamasikkhadit 1996). The examples which correspond to the above meanings are provided below. (26) a. khaw, clay3 pay, haa, khun, miia,-waan, he PAST go see you yesterday ‘He went to see you yesterday.’ b. khaw, day, duu, narj5 thuk,-wan, he have an opportunity see movie every day. ‘He has an opportunity to see the movies every day.’ c. khaw, day3 muak2 bay, suay, he get hat Cl beautiful ‘He has got the beautiful hat.’ It should be noted that the meaning of the regular verb day, ‘have an opportunity to do something’ somewhat overlaps with that of day, ‘permission’. Therefore, sentence (26 b) can also mean ‘He is allowed to see the movies every day.’ Consequently, some scholars do not consider day, as a modal of permission, but a regular verb. However, there are cases where day, expresses merely ‘permission’ not the meaning of ‘having the opportunity to do something’ as in (27) below. (27) A: poo, hay, chan, pay, pat,tha,yaa, phuak,-thaa, day, pay, may, father let me go Pattaya group-you may go QP ‘My father allows me to go to Pattaya. Are you allowed to go?’ (Bybee et a1. 1994:186) Strong obligation in English is expressed by ‘must’ and ‘have to’ whereas weak obligation by ‘should’. 128 B: chan, day, pay, I may go. ‘I am allowed to go.’ C: chan5 may, day, payl I not may go. ‘I am not allowed to go.’ The above dialogue clearly shows that day, can be a separate word denoting ‘permission’, which can be regarded as a modal. As a result, I will include day, as one of the Thai modals. Taking into account the predicator test in (28) below, I would propose that the modal day, expressing permission is located above VP. (28) Q: khaws daYs PaYI pat,tha,yaa, maY4 he may go Pattaya QP ‘Is he allowed to go to Pattaya?’ A: a. day, may ‘Yes, he is allowed to go to Pattaya.’ b-*Pay1 go As obviously seen, the interrogative containing the modal day, requires day, to be the predicator, not the verb. The modal day, thus must be located higher than the verb to be ‘closest’ attracted to the 2 head. The suggested structure is thus as below. (29) ModP,,,,,,,,,,,-,m /\ day, . VP may /\ V XP 129 5.4. ‘mus my, ( thee. UlOda, than t] 5.4.1-2 Lehmanhligatiam The next modal to be considered is the one to the left of day, ‘permission’ - toy, ‘must’, which expresses strong obligation or necessity. Like must in English, the modal toy, can express both root and epistemic meanings as shown in (30 a, b) respectively. (30) a. khaw, may, cheep, thee, treat, khaw, tor), pay, haa, thee, he not like her but he must go see her ‘He does not like her but he has to go to see her.’ b. khaw, may, yuu2 baan, khaw, tor], pay, haa, thee, nice-nae, he not stay home he must go see her surely ‘He is not at home. He surely must go to see her.’7 I will examine in this section only the modal toy, in the root sense as the one with the epistemic meaning will be discussed later. In terms of the syntactic position, the predicator test in (31) shows that the root modal toy, is higher than VP, and (32) points to the fact that it is also generated higher than the modal day,. (31) Q: khaw5 tar], ?aan2 nan,sii, may, he must read book QP ‘Must he read?’ A A: a. tor], must ‘Yes, he must.’ b."'?aan2 read (32) Q: khaw, tor], day, pay, roorj,riian, may, he must may go school QP ‘Must he be allowed to go to school?’ 7 It should be noted that tag, as an epistemic modal is usually accompanied by mew,- mere, ‘surely’, probably for emphasis or to make a clear distinction between the root and the epistemic ones. 130 A: a. tor), must ‘Yes, he must.’ b.*day, may C-*P3Y1 go As it is apparent that the modal toy, ‘must’ is situated higher than VP and the modal day, ‘may’, the syntactic representation of the three elements is as illustrated below. (33) ModP obligation t I)” 3 MOdeermission must /\ day, VP may /\ V XP 5.4.1.3 mammal: The word ca .7, is usually treated as the future marker because it is normally used to express the future time as in (34) (Sriphen 1982). (34) khaws ca?3 PaY2 pat,ta,yaa, thU3nii4 he will go Pattaya tomorrow ‘He will go to Pattaya tomorrow.’ However, ca .7, can occur in a sentence representing a past event as below. (35) khaw, ca?, ?aan2 nan,sii, miie,khiin, tree, fay, clap2 he will read book last night but power out ‘He would read last night but the power was out.’ The more accurate interpretation of (35) is ‘He intended/wanted to read last night.’ Therefore, future indication is not the only meaning conveyed by ca ?,. The word ca .7, is thus claimed to be a non-absolute future marker (Kanchanawan 1978) or suggested 131 to b care othi wor $8111 cam, never to be an intentive marker (Sriphen 1982). My proposal is that ca .7, is of two different categories: one is a future tense marker which includes the meaning of prediction and the other is a modal expressing volition or willingness. I will examine in this section the word ca?, which signifies the meaning of volition as exemplified in the following sentences“. (36) a. khaw, may, pay, khaw, ca?, ?aan2 narjssii, seep2 he not go he will read book take-an—exam ‘He won’t go because he will study for the exam.’ b. khaw5 'ca?, pay, maze, pee, may, a,nu,yaat, he will go though father not permit ‘He will go though his father does not allow him to.’ As illustrated in (37) below, the modal ca .7, fails the predicator test; that is, it cannot be a minimal response to a yes/no question. (37) Q: thee, ca?, ?aan2 na135sii, soap2 may, you will read book take-an-exarn QP ‘Will you study for the exarn?’ A: a.*ca?, will b. ?aan2 read The above instance seems to give the first impression that ca .7, is generated lower than VP. However, there is some evidence disproving such a claim. First of all, ca .7, never occurs post-verbally as in (3 8). Furthermore, ca .7, always precedes tan, ‘must’ and ’ Ekniyom (1979) hypothesizes that ca .7, is derived from the verb cak, which means ‘want, desire’ or the verb ka?, which means ‘plan, expect’. The interchangability of ca ?, and cak, in stylish or archaic writing is prevalent. It might, therefore, be the case that ca .7, has undergone phonological change and as will be claimed, syntactic change in terms of category. 132 day, ‘may’ which are earlier proven to be base-generated higher than VP as demonstrated in (39) and (40)’. (38) a."‘khaw5 ?aan2 ca?, nan,sii, he read will book b.*khaw, ?aan2 nan,sii,ca?, he read book will (39) a. khaw, ca?, tor], ?aan2 nag,sii, he will must read book ‘He will have to read.’ b."'khaw5 tor], ca?, ?aan2 narj,sii, he must will read book (40) a. khaw, ca?, day, pay, pat,tha,yaa, he will may go Pattaya ‘He will be allowed to go to Pattaya.’ b.*khaw, day, ca?, pay, pat,tha,yaa, he may will go Pattaya From the examples above, it is reasonable to assume that the volition modal ca .7, is located above the obligation and the permission modals as represented in the structure below. (4 l ) MOdeolition ca ? 2 MOdPobligation will /\ I 9U 3 MOdeemrission must /\ day, VP may 9 As noted in footnote 8, the word ca .7, in (393) and (40a) can be replaced by cak, in bookish style of writing as in (i) and (ii) respectively. (i) khaw, cak2 tor], ?aan2 narj,sii, he want must read book (ii)khaws cakz daY3 PaYI pat,tha,yaa, he want may go Pattaya 133 men poss attra mOV ~ am the? 12: . Sr, Shim Still, the fact that the modal ca .7, behaves differently from the other modals mentioned earlier in terms of the ability to be a predicator has to be accounted for. One possibility is that ca .7, may be the head but lack V-features; consequently, it cannot be attracted to the 2 head. The next highest element with V-features thus moves to 2 to function as the predicator as in (3 7), repeated below. (37) Q: thee, ca?, ?aan2 nan,sii, seep2 may, you will read book take-an-exam QP ‘Will you study for the exam?’ A: a.*ca?, will b. ?aan2 read However, treating ca .7, as the head of Modeouu-on appears problematic. Since (3 7) suggests that the next highest head with V-features moves to 2 past the head ca .72, this movement would violate the Head Movement Constraint (Travis 1984). Alternatively, if ?aan, ‘read’ in (37) adjoins to ca .7, before moving to 22, ca ?, should always accompany the element which moves to the 2 head as a complex head movement as shown in (42)”. (42) a. ZIP /\ Z MOdeolition Mod VP /\ /\ Mod V, V DP ca ?, ?aan, t, nayfiir', will read book '° Similar to the analysis of noun phrases, the right-adjunction is assumed for Thai clausal structures. 134 is t1 the Can ” It 0111:. C0 "2 the: /\ 2 MOdeolition I /\ Mod, Mod VP /\ t,- /\ Mod V, V DP ca .7, ?aan, t, naggii, will read book However, as the most common response to a question containing the modal of volition is simply the bare verb as shown in (43), the analysis above does not seem right”. (43) Q: thee, ca?, ?aan2 narj,sii, seep2 may, you will read book take-an-exarn QP ‘Will you study for the exam?’ A: a.*ca?, ?aan2 will read b. ?aan2 read It seems, then, that ca .7, may not be a head. What I would tentatively hypothesize is that ca .7, is a phrase in the specifier of ModP,0,,,,-o,,”. The possibility of the movement of the head below ca .7, to 22 without violating the HMC as well as without adjoining to ca .7, can thus be analyzed as in (44). ” It should be noted that ca ?, may accompany the verb to function as a predicator, but only when it is emphasized. This does not constitute an argument against my claim that ca?2 is in the specifier instead of a head. An account of this instance is that ca?, moves to the specifier of ZIP which is also claimed to be the locus of emphasis. 135 /\ ?aan,, ModPvolition read ca 7, Mod’ I Mod VP /\ V DP t, naggit‘, book 5.4.1.4 2122,,me I will discuss the modals naa, and khuan, ‘should’ together because both express the meaning of ‘weak obligation’ and they will be assumed to be generated in the same position as they never co-occur as in (45 )‘3 . (45) a.*khaw,khuan, naa, ?aan, narj,sii, he should should read book b.*khaw, naa, khuan, ?aan, narj,sii, he should should read book Moreover, mm, and khuan, behave exactly the same in terms of their ability to be a predicator as shown in (46) and (47) below. (46) Q: khaw, naa, ?aan2 narj,sii, seep2 may, he should read book take-an-exam QP ‘Should he study for the exam?’ A: a. naa, should ‘Yes, he should.’ b."'?aan2 read ‘2 The word ca .7, is thus hypothesized to undergo both phonological change and syntactic change from the head verb cak, ‘want, desire’ to an element in the specifier. '3 It seems that naa, is losing its root meaning of weak obligation and is becoming an epistemic modal denoting probability whereas khuan , resists the change. The meaning of the former is inclined to be epistemic whereas the latter, root. 136 (47) Q: khaw5 khuan, ?aan2 nar],sii, seep2 may, he should read book take-an-ex'arn QP ‘Should he study for the exam?’ A: a. khuan, should ‘Yes, he should read.’ b."'?aan2 read Although naa, and khuan, are certainly syntactically higher than VP, it is not easy to tell where in the clausal structure they are generated with respect to the other modals mentioned earlier. This is because naa, and khuan, usually carry the epistemic interpretation, not the root one, when they precede toy, ‘must’ as in (48) below". (48) a. khaw5 naa,-ca?2 tor), ?aan2 narj,sii, he should must read book ‘He likely has to read.’ b. khaw, khuan, tor], ?aan2 narj,sii, he should must read book ‘He likely has to read.’ The above phenomenon is not surprising, given that naa, , khuan, ‘should’ and toy, ‘must’ are very close in meaning; that is, weak vs. strong obligation. The fact that they cannot co-occur when they have the root reading can thus be justified. Although it does not sound natural when the root modals naa, and khuan, precede the modal of permission day,, it is absolutely acceptable when the former appear before the modal of ability day, which I will assume in the following section to be base- - generated in the same position as that of the permission modal (as also suggested by Cinque 1999). 137 p0: tha (51 (49) a. khaw, naa,-ca?2 day, pay, pat,,tha,yaa, he should may go Pattaya ‘He is likely allowed to go to Pattaya.’ b. khaw5 khuan, day, pay, pat,,tha,yaa, he should may go Pattaya ‘He is likely allowed to go to Pattaya.’ (50) a. khaw, naa, tham, day, he should do can ‘He should be able to do it.’ b. khaw, khuan, tham, day, he should do can ‘He should be able to do it.’ It seems tempting, therefore, to conclude that naa, and khuan, occupy the same position as tag” which is ModP,,,,,,g,,,on right above Modess,Om,,,,,,y. However, the fact that the modal of volition ca ?2 always precedes tag, but follows naa, and khuan, as in (51-53) renders the conclusion suspicious. (51) a. khaw, ca?2 tor], ?aan2 narj,sii, he will must read book ‘He will have to read.’ b.*khaw, tor], ca?2 ?aan2 narj,sii, he must will read book (52) a. khaw, naa, ca?2 ?aan2 nan,sii, he should will read book ‘He should read.’ b.“'khaw, ca?2 naa, ?aan2 nan,sii, he will should read book (53) a. khaw, khuan, ca?2 ?aan2 narj,sii, he should will read book ‘He should read.’ b.*khaw, ca?2 khuan, ?aan2 narj,sii, he will should read book " It should be mentioned that the word ca ?, in (48a) is obligatory. I will discuss in the following section the status of ca .7, which always attaches to most of the epistemic 138 Importantly, it should be made clear that the word ca .7, following mm, and khuan , is the modal of volition, not a future marker or a predictive marker". Evidence for this comes from the following sentences where ca .7, is replaced without any change of meaning by cak, which means ‘want, desire’ and is used in an archaic, bookish or poetic style of writing (see fir.8). (54) a. khaw, cak,z tor], ?aan,, narj,sii, he want must read book ‘He will have to read.’ b. khaw5 naa, cak2 ?aan2 narj,sii, he should want read book ‘He should want to read.’ c. khaw, khuan, cak2 ?aan2 narj,sii, he should want read book ‘He should want to read.’ modals. '5 Also, ca ?, following the modals of weak obligation is not a part of the word, say naa,- ca/ khuan,-ca,, which I will claim epistemic modals of probability to be such. The evidence supporting the claim is that these two modals with the root sense cannot be accompanied by ca ?, when they function as predicators although ca ?, is present in the question as shown in (i)-(ii). (i) Q: khaw, naa, ca?2 ?aarr2 narj,sii, may, he should will read book QP ‘Should he want to read?’ A: a. naa, should ‘Yes, he should.’ b.*naa, ca?2 should will (ii) Q: khaw, khuan, ca?2 ?aan2 nag,sii, may, he should will read book QP ‘Should he read?’ A: a. khuan, should b.*khuan, ca?2 should will As will be discussed, ca 2 has to accompany naa, and khuan, when the latter are epistemic modals behaving as predicators. 139 Summing up, despite the similar meanings of naa, , khuan, and tag, as well as their same position relative to the modals of ability, they cannot be claimed to be in the same projection due to their different word orders with respect to the modal of volition ca ?2. Therefore, I suggest that the syntactic position of naa, and khuan, is as in (55) where ModPweak ob,,g,,,o,, is above ModPvolition which is in turn above ModP 5mg Gwyn-on.” (5 5) MOdeeak obligation "003 MOdeolition khuan, /\ should ca ?2 Mod’ MOd MOdPsuong obligation /\ IOU 3 MOdeermission must /\ day, VP may 5.4.1.5 WWIWM Having discussed the pre-verbal modals, we are now in a position to analyze the post-verbal modals. Given that Thai is an SVO language, the fact that some modals appear post-verbally is both interesting and problematic. The Thai post-verbal modals are pen, and way, which express ability and day, which means either ability or permission. The difference between the three modals of ability is that day, conveys general ability, pen, denotes mental or intellectual ability and way, expresses the physical ability, as demonstrated below. '6 Cinque (1999) does not discuss the modals of weak obligation. He considers only the strong obligation modal ‘must’. It is not explicitly clear whether he includes ‘should’ under the modal of obligation. 140 (56) a. khaw, ?aan2 narj,sii, day,/pen, he read book can ‘He can read.’ b. khaw, yok, narj,sii, day,/way, he lifi book can ‘He can lift books.’ Due to their post-verbal position, they might be thought of as VP right-adjuncts, not verbs. The predicator test, however, shows that the VP adjunct analysis is incorrect. All three words, although post-verbal, are able to be predicators as illustrated in (57)-(60). This clearly shows that they are heads situated above VP. (57) Q: khaw, ?aan2 nag,sii, day, may, he read book may QP ‘May he read?’ A: a. day, may ‘Yes, he may read.’ b."‘?aan2 read (5 8) Q: khaw, ?aan2 narj,sii, day, may, he read book can QP ‘Can he read?’ A: a. day, can ‘Yes, he can read.’ b."‘?aan2 read (59) Q: khaw, ?aan2 narj,sii, pen, may, he read book can QP ‘Can he read?’ A: a. pen, can ‘Yes, he can read.’ b."'?aan2 read 141 (60) Q: khaw, yok, nag,sii, way, may, he lift book can QP ‘Can he lifi the books?’ A: a. way5 can ‘Yes, he can lift the books.’ b.*yok, ~ lift The examples above suggest that the post-verbal modals are in fact the base- generated pre-verbal ones. They also indicate that the scope concerns are determined by base positions, not the surface ones (as Cinque 1999 also argues is true for negation in Italian). In order to derive the surface word order, we must therefore raise the VP to some position higher than the modal. This can account for the word order fact as illustrated below.17 (61) ModP,,,,,,,,,,,m,”-,ity VP: MOdeerrnission/ability ?aan, naggii, /\ read book Mod’ /\ Mod VP daYJ . ti can Although these post-verbal modals are higher than VP, they must be below the pre-verbal modals discussed above. This is shown in (62) and (63) where the modals tag, ‘must’ and khuan , ‘should’, not the post-verbal modals, function as predicators. (62) Q: khaw, tor], phuut, thay, day, may, he must speak Thai can QP ‘Must he be able to speak Thai?’ '7 I will assume without argument here that the VP adjoins to ModP. See Section 5.5 for arguments based on negation that support this assumption. 142 A: a. tor], must ‘Yes, he must.’ b.*phuut, speak c.*day, can (63) Q: khaw5 khuan, phuut, thay, pen, may, he should speak Thai can QP ‘Should he be able to speak Thai?’ A: a. khuan, should ‘Yes, he should.’ b.*phuut, speak c.*pen, can Rather than creating a new node for these modals, I assume that they are in ModP Evidence for this comes fiom the following facts: 1) the modals of permission° permission and ability never co-occur as demonstrated in (64) and 2) the meanings of day, which denotes permission in the pre-verbal and post-verbal positions are not different as shown in (65). (64) a.*khaw, day, ?aan2 narj,sii, day, he may read book can b.*khaw, day, ?aan2 nan,sii, day, he may read book may c.*khaw, day, ?aan2 narj,sii, pen, he may read book can d.*khaw, day, yok, narj,sii, ways he may lift book can (65) a. khaw, day, ?aan2 narj,sii, he may read book ‘He is allowed to read.’ h 143 b. khaw, ?aan2 nan,sii, day, he read book may ‘He is allowed to read.’ Therefore, the syntactic representation of all the modals we have looked at so far is shown in (66). (66) MOdP weak obligation 71003 MOdeolition khuan should ca .7, MOd,volition will /\ Madvolition MOdPstmng obligation /\ ”Us MOdeemrission/ability must /\ day, VP Penl way, may/can Having sketched out the basic analysis, I will now provide further arguments which support the idea that the post-verbal modals are generated in the pre-verbal position. Then I will attempt to give an account of the operation of VP movement. 5.4.1.5.1 Arguments for the claim that post-verbal modals are generated pre-verbally There are several pieces of evidence suggesting that the post-verbal modals behave exactly like the pre-verbal ones and thus support the assumption that they are also pre-verbal. (1) Negatability Similar to pre-verbal modals, the post-verbal modals can be directly negated by the negator may, ‘not’ as demonstrated below. 144 (67) khaw, may, khuan, ?aan2 narj,sii, he not should read book ‘He should not read.’ (68) khaw, may, tor], ?aan2 narj,sii, he not must read book ‘He needs not read.’ (69) khaw, ?aan2 narjssii, may, day, he read book not can/may ‘He cannot/may not read.’ (2) Scope over VP In spite of their post-verbal position, such modals always take scOpe over VP like all other pre-verbal modals as mentioned earlier". Also, in the negative sentence, if the negator may, ‘not’ directly negates the modals, either pre- or post- verbal ones, the negator may, takes the modals under its scope as in (67—69) above. On the other hand, if the negator may, ‘not’ directly negates the main verb, the VP including ‘not’ is under the scope of the modal as in (70-72) below. (70) khaw, khuan, may, ?aan2 narj,sii, he should not read book ‘He should not read.’ (71) khaw, tor), may, ?aan2 narj,sii, he must not read book ‘He must not read.’ (72) khaw, may, ?aan2 narj,sii, day, he not read book can ‘He can [not read].’ ‘8 I follow here the Scope Principle defined by Ernst (1991:753) as below. I] S E . . 1 An operator A has scope over an operator B in case A c-commands a member of the chain containing B. Also, according to Ernst, scope is determined by strict c-command. 145 Better interpretations of the above sentences are ‘It is advisable that he does not read.’, ‘It is necessary that he does not read.’, and ‘He is able to not read.’ respectively. In sum, no matter where the negator may, is, the modals all take scope over the VP. The negation is under the scope of the modals only when VP is negated. It should thus be repeated here that the scope relations are determined by the base positions, not the surface ones. (3) Co-ordination When two VPs are conjoined, the modals can take scope over both VPs independent of the pre- or post-verbal position as shown below. (73) khaw, khuan, ?aan2 narj,sii, lzeae, tham, kaan,-baan, he should read book and do homework ‘He should read and do his homework.’ (74) khaw, tor], ?aan2 narjssii, lea, tham, kaan,-baan, he must read book and do homework ‘He must read and do his homework.’ (75) khaw, ?aan2 narj,sii, lanai, tharn, kaan,-baan, day, he read book and do homework can/may ‘He can/may read and do his homework.’ However, in a co-ordinated construction where the first conjunct contains a pre- verbal modal and the second conjunct, a post-verbal modal, the pre-verbal modal cannot take scope over the second conjunct (76a). The repetition of the pre-verbal modal in the second conjunct is necessary (76b) if the former is to have scope over the former. (76) a.*khaw, tor], tham, lace, tham, day, he must do and do can ‘He must and must be able to do it.’ b. khaw5 tor], tham, lac, tor], tham, day, he must do and must do can 146 The above examples show that the post-verbal modals cannot be VP right- adjuncts. If they were, sentence (76a) would not be ungrammatical. Let me explain this. If the post—verbal modal of ability day, ‘can’ were VP right-adjunct, the two conjoined phrases would be VPs as in (77). Assuming the co-ordination analysis as in Munn (1992, 1993), the fact that the modal of obligation toy, ‘must’ cannot have scope over both conjoined VPs cannot be explained. (77) *MOdPobligation /\ Mod VP ”Us /\ must VP BP tham l /\ do B VP [@634 /\ and VP VP tham, day, do can The ungrammaticality of (76a) coupled with the requirement of the modal of obligation in the second conjunct is thus attributable to the fact that the post-verbal modal of ability day, ‘can’ projects ModP,bility above VP. Since the second conjunct is a ModP, the first conjunct is required to be a ModP too. The structure is thus as in (78). (78) MOdP obligation MOdP obligation BP /\ Mod VP B MOdPability toy, tham, lcece, /\ must do and . VP, MOdPabiliry (ham , do Mod VP daJ’I It can 147 Since two ModPs are conjoined in (78), the modal of obligation in the first conj unct does not have scope over the second conj unct. Therefore, another modal of obligation in the second conjunct is needed as shown in (79). (79) MOdPobligation MOdP obligation BP /\ MOd VP B MOdPobligation tan, tham, lcece, /\ must do and Mod ModP,bility ”U3 /\ mUSt VP.‘ MOdP ability tham, do Mod VP daYl ti can (4) Other pre-verbal modals becoming post-verbal Perhaps the most important piece of evidence supporting the claim that post- verbal modals in Thai trigger VP movement is that the VP can move above the pre-verbal modals when the post-verbal modals are present. The pro-verbal modals never appear post-verbally when they stand alone. This is shown in (80). (80) a.*khaw, ?aan2 narj,sii, ca?2 he read book will b.*khaw, ?aan2 nar],sii, khuan, he read book should c.*khaw, ?aan2 narj,sii, tor], he read book must However, these pre-verbal modals can be preceded by a VP when the post-verbal modals are present. 148 (81) a. khaw, tor], ?aan2 na135sii, day, he must read book can ‘He must be able to read.’ b. khaw, ?aan2 narj,sii, tor], day, he read book must can ‘He must be able to read.’ (82) a. khaw, khuan, ?aan2 narj,sii, day, he should read book can ‘He should be able to read.’ b. khaw, ?aan, narj,sii, khuan, day, he read book should can ‘He should be able to read.’ (83) a. khaw, ca?2 tor), ?aan2 narj,sii, day, he will must read book can ‘He will have to be able to read.’ b. khaw, ?aan2 narj,sii, ca?2 tor], day, he read book will must can ‘He will have to be able to read.’ It is obvious from the translation that each pair has the same meaning. Let’s find out whether the adjunction analysis can account for the above examples. First of all, to capture the word order fact, the pre-verbal modals have to lower to adjoin to the post- verbal modals which are VP right-adjuncts as in (84), represented for sentence (81b). (84) MOdPobligation Mod, VP 2, /\ VP VP /\ /\ V DP Mod , V ?aan, naggir’, toy, day, read book must can 149 The analysis in (84) does not seem right mainly due to lowering which violates ECP; that is, the trace is not properly governed. In other words, the moved element does not c-command its trace. This renders the first possibility unpromising. The second possibility concerns VP movement. If the post-verbal modals are VP adjuncts, it must be the case that only part of VP moves. In other words, the VP right adjunct day, is stranded as illustrated in (85). (85) ModP /\ VP, ModP ?aan, naggii, /\ read book Mod VP ’9”: /\ must VP VP ti days can Such a movement is illegitimate, given the general assumption that the deleted or moved element must be either X° or XP including its adjunct (Ernst 1992). Moreover, what is unclear is why VP moves further away from the post-verbal modals if the latter trigger the movement. Therefore, the idea that the post-verbal modals are right-adjuncts is improbable. I will attempt to give an account of VP movement in the following sub- section. 5.4.1.5.2 Account of VP movement What triggers VP movement needs explication. It must be the case that the post- verbal modals possess a certain feature that attracts VP, most probably [+V] feature. However, as claimed from the beginning, almost all pre-verbal modals have V-features. Therefore, the question that arises is why the other pre-verbal modals do not attract the 150 VP. A feasible explanation is that the post-verbal modals have strong V-features whereas the pre-verbal, weak ones. In addition, my tentative hypothesis regarding the VP movement above the pre- verbal modals with the presence of the post-verbal is as follows. Although the pair of sentences as in (81), repeated below, are not different in meaning (Sookgasem 1990), they are not exactly the same. That is, the post-verbal modal in sentence (b) is more focused than that in (a). (81) a. khaw, tar], ?aan2 narj,sii, day, he must read book can ‘He must be able to read.’ b. khaw, ?aan2 narj,sii, tor), day, he read book must can ‘He must be able to read.’ I will, therefore, speculate that the post-verbal modal moves to a Focus phrase which is high in the structure. (I am not assuming here what exactly that phrase is, but I will stipulate a FocP for simplicity). On its way to F ocP, the post-verbal modal has to be adjoined to the higher pre-verbal modal to avoid HMC as illustrated in (86a, b) for (81b). The VP movement is then triggered by the feature carried by the post-verbal modal which now substitutes into the head of FocP (86c). (86) a. FocP FCC MOdPobligation Mod ModP,,,,,ity /\ Mod Mod, Mod VP toy, day, t, ?aan2 naggii, must , can I read book 151 /\ Floc ModPobligation K‘\ NtIJOd MOdPability Mod Mod, Mod VP toy, day, t, ?aan, namsii, must can read book c FocP /\ VP, FocP ?aan, nay,sii, /\ read book Foc ModP,,,,,,,,,,,on x | Mod,- MOd MOdP ability /\ t, Mod Mod, Mod VP ‘903 day, ti tk must can I Support for this analysis comes fiorn the predicator fact as demonstrated in (87) where the minimal response to the question as in (81b) is the complex head consisting of the pre- and the post-verbal modals. Either modal used alone as the predicator leads to ungrammaticality. (87) Q: khaw, ?aan2 narj,sii, tor], day, may, he read book must can QP ‘Must he be able to read?’ A: a.*day, can b.*torj, must c. tor), day, must can 152 Although my hypothesis is tentative, it seems to be able to take care of the instances under examination. However, it needs more profound studies to discover whether the analysis works in such a way. 5.4.1.6 Summary To conclude, I have argued that all the root modals are basically pre-verbal although some have post-verbal surface word order. VP movement thus operates in Thai syntax, which makes some pre-verbal modals appear post-verbally. Moreover, all the root modals are heads with V-features, except for the modal of volition ca .7, which is suggested to be a phrase in the specifier of Modemm. The syntactic positions of Thai root modals, fm'thermore, conform in principle to the cross-linguistic orderings proposed by Cinque (1999), except for that of the modals of weak obligation which, according to the empirical evidence, are not in the same base position as that of the strong obligation modal. Hence, I tentatively suggest that the former is positioned above ModP,olifion which is in turn above Modeg mm”. 5.4.2 Epistemic modals Having analyzed the root modals, I now turn to the epistemic modals. All epistemic modals occur pro-verbally. As they all concern the judgment of a proposition on the part of the speaker, they are assumed to be in the same position above the root modals (Cinque 1999). However, the epistemic modals in Thai do not behave in a unified way. Therefore, I will divide them into three groups according to their behavior. 153 5.4.2.1 Mrfinffinidflnalml’ The word tag, as an epistemic modal expresses inferred certainty as shown in (88) below. (88) khaw, may, yuu2 baan, khaw5 tor], pay, haa, thee, nee-nee, he not stay home he must go see her surely ‘He is not at home. He surely must go to see her.’ Like the volition modal ca ?2, the epistemic modal toy, cannot be a predicator as demonstrated below. (89) Q:* khaw, terj, pay, haa, thee, may, he must go see her QP ‘Is it certain that he has gone to see her?’ A: *ten, must Not only is the epistemic modal toy, unable to be a predicator, it cannot be questioned either, no matter whether the question particle is rit', or may,.‘9 (90) a."‘khaw5 tor], pay, haa5 thee, near, rii, he must go see her sure QP b.*khaw, tor], pay, haa, thee, new, may, he must go see her sure QP However, the above fact is not unexpected, considering the same phenomenon in English where the epistemic modals must, should and may are odd in questions because they are incompatible with the request for information (J ackendoff 1972:84-85). The three modals in (91) below thus prefer the root meanings. '9 The two question particles are different although both form a yes/no question. rii, questions the truth value of the proposition whereas may, which is a reduced form of 154 (91) Must/Should/May John have left? In contrast to regular questions, tag questions do not solicit information, and are thus acceptable with epistemic modals as shown by the English examples below. (92) a. John must have left, mustn’t he? b. John should have left, shouldn’t he? c. John may have left, may he not? This is also true in Thai. The epistemic modal tag, can be questioned by chay,- may, ‘true or not’, the tag question in Thai. (93) khaw, tor], pay, haa, thee, naeae, leey, chay,-may, he must go see her sure PRT true-or not ‘Is it true that he surely went to see her?’ However, the modal tog, cannot stand alone nor accompany chay, ‘true’ as the response to (93) as shown in (94), unlike verbs as illustrated in (95). (94) a.*torJ, must b.*chay, tor], true must c. chay, true (95) Q: khaw5 pay, haaS thee, chay,-may, he go see her true-or not ‘Is it true that he went to see her?’ A: a. chay, true b. chay, pay, true go c.?pay, go rii, may, ‘or not’ asks for the affirmation or denial of the proposition, or particularly the verb. Therefore, may, is used as a question particle for the predicator test. 155 The above fact is not surprising when an explanation is given on the semantic grounds. As the epistemic toy, involves certain inference or judgment on the part of the speaker, it seems odd that the statement of inferred certainty is questioned by the speaker himself/herself and confirmed or denied on the part of the hearer.2° From what has been discussed above, syntactic information about the epistemic modal toy, cannot be definitively drawn from its inability to function as a predicator or to be questioned. What can simply be speculated is that it must be positioned in ModPepistemic which is above root modals as indicated by the following examples. (96) a. pee, may, waa, khaw, terj, day, pay, baa, mere, father not scold he must may go bar sure ‘His father did not complain. He certainly must be allowed to go to the bar.’ b. khaw5 sii, maay, maa, khaw, tor], ca?2 som, pra,tuu, name-nae, he buy wood come he must will repair door sure ‘He bought some wood. He certainly must plan to repair the door.’ Hence, the syntactic structure of the epistemic toy, is represented as in (97), where its syntactic position (head or phrase) is left open. (97) MOdPepistemic toy, Mode, Mod VP 2° In fact, the epistemic modal tag, can be used as an answer to (93), but only with the verb and adverbial mere, ‘surely’ as in (i). (i) t91.13 paYI “333 must go surely This is not different from English in which the answer to the question containing an epistemic modal must is usually ‘must + AuxN’ as in (ii). (ii) Q: He must have left, mustn’t he? 156 5.4.2.2..Zaat2122, and khozzLiazlinmhahiliDL’ The epistemic modals ?aat2-ca .7, ‘probably’ and khan, -ca ?, ‘likely’ behave in the same way as the epistemic toy, ‘must’ in terms of their inability to be questioned by question particles other than the tag question chay,-may, as shown below. (98) a."‘khaws ?aat,»ca?2 ?aan2 narj,sii, may, he probably read book QP ‘Does he probably read?’ b.*khaw, ?aat.,-ca?2 ?aan2 nan,sii,rii, he probably read book QP c. khaw, ?aat,-ca?2 ?aan2 narj,sii, chay,-may, he probably read book true-or not (99) a.*khaw, khorj,-ca?2 ?aan2 narj,sii, may, he likely read book QP ‘Does he likely read?’ b."'khaws khorj,-ca?2 ?aan2 narj,sii, rii, he likely read book OP 0. khaw, khorJ,-ca?2 ?aan2 nan,sii, chay,-may, he likely read book true-or not Unlike tom, however, when ?aat2-ca .7, and khay,-ca .7, are questioned by the tag question particle chay,-may,, they themselves can be predicators. (100) Q: khaw, ?aat2-ca?2 ?aan2 narj,sii, chay,-may, he probably read book true-or not ‘Is it true that he probably reads.’ A: a. (chay,) ?aat,-ca?, (true) probably b.*(chay,) ?aan2 (true) read (101) Q: khaw, khorJ,-ca?2 ?aan2 narj,sii, chay,-may, he likely read book true-or not ‘Is it true that he likely reads.’ A: Yes, he must have. 157 A: a. (chay,) khorj,-ca?2 (true) likely b.*(chay,) ?aan2 (true) read The fact that the verb cannot be a predicator in the examples above seemingly implies that ?aat2-ca .7, and khan ,-ca ? , function as heads which block the movement of V to the 2 head. However, ?aat2-ca .7, ‘probably’ and khan, -ca .7, ‘likely’ often stand alone without the accompaniment of ca ?2. This casts doubt on the status of ca ?,. That is, is it an intrinsic part of the words ?aat2-ca ?, and khon,-ca .7, which is optional and thus sometimes omitted or is it a separate element which adjoins to ?aat, and khan,? The pairs ?aat2-ca ?, /?aat, and khan ,-ca ? 2/khan , are not different in meaning. The absence of ca .7, is not a problem, given that probability usually implies futurity or prediction which is expressed by the future marker ca .7, .2' Thus, with or without ca ?2, these modals express the same meaning.22 Let’s take as a start the following examples where ?aat, and khan, cannot occur alone as predicators without ca ?2. (102) Q: khaw, ?aat2 ?aan2 nar],sii, chay,-may, he probably read book true-or not ‘Is it true that he probably reads?’ 21Bybee et a1. (1994:248, 207) state that ‘possibility markers are often restricted to future interpretations,’ and ‘the normal or default interpretation for epistemic possibility and probability with dynamic predicates is that the situation is expected to take place in the firture.’ 2’ Sookgasem (1990:90) considers such words as compound-like words because the meaning of the whole word cannot be decomposed. She also mentions that ‘all compound-like single temporal verbs have ca .7, as their parts, and this ca .7, can be omitted without affecting the meaning of the words’. 158 A: a.*?aat, probably b. ?aat1 ca?2 probably will (103) Q: khaw, khorj, ?aan2 nan,sii, chay,-may, he likely read book true-or not ‘Is it true that he likely reads?’ A: a.*khorj, likely b. khorj, ca?2 likely will The above examples can be interpreted in two different ways. First, suppose ?aatz-ca ?, and khan ,-ca ? 2 have the following structure where ca ?, is the head while ?aat, and khan, are in the specifier. (104) ModP epistemic ?aatjkhon, Mod’ /\ Mod VP ca .7, The proposal that ca .7, is the head can handle the case where ca ?2 may not be omitted in the minimal response to yes/no questions. Still, the remaining problem is ‘is ca .7, generated there or does it move from a head down below?’ Since ca .7, usually appears as a tense marker, it is unlikely to be generated in the head of ModP. Instead it should project its own projection which is presumably TPW. Moreover, if the hypothesis is that ca ?, is the head of TPW, then it is dubious why ca .7, itself cannot behave as a predicator as shown in (105). 159 (105) Q: phrurj,nii, khaw, ca?2 maa, roorj,riian, may, ' tomorrow he will come school QP ‘Will he come to school tomorrow?’ A: a. ma, come b."‘ca?2 will The above example also weakens the possibility of ca .7, moving fiom a lower head which is supposed to be TPmm situated between the epistemic modals and the root , ones (Cinque 1999). Alternatively, the obligatoriness of ca ?2 can be looked at in another way. That is, ca .7, is intrinsically incorporated in the words ?aat,-ca .7, and khan ,-ca ? 2 and they together behave as a compound head as in (106). The incomplete head is, as a consequence, unacceptable as the predicator. This sounds too simple but no apparent problem is found. (106) ModP,,,,,,,,,nic ?aat,-ca .7, /khon,-ca ?, VP Summing up, the only convincing possibility is the simplest one; that is, ?aat2—ca .7, and khon,-ca .7, are compound heads. The word ca .7, in the epistemic modals is thus different from ca ?, as a future marker. Since the epistemic modals under investigation can behave as predicators and block the movement of the verb to 2, they are qualified as heads. 160 The unacceptability of their co-occurrence with the epistemic modal ton, as shown in (107) indicates that aside fi'om their semantic incongruity, they are in the same structural position; namely, Mochphm, as in (108)”. (107) a."'khaw5 ?aat2 ten, ?aan, nan,sii, he probably must read book b."‘khaw5 tor], ?aat2 ?aan, narj,sii, he must probably read book c."‘khaw,khorj, ten, ?aan, narj,sii, he likely must read book d.*khaw, tor], khOIJ, ?aan, nan,sii, he must likely read book (108) MOchpistemic ton, Mode, ?aatz-ca .7, /khon,-ca ?, /\ Mod VP 5.4.2.3 mangLandkthm- flipesailzilitx’ As epistemic modals, naa,—ca .7, and khuan,-ca ?, express supposition or probability as shown in the examples below. (109) a. khaw, pay, haa, thee, thuk,wan, khaw, naa,-ca?2 cheep, thee, he go see she everyday he should like she ‘He goes to see her everyday. He possibly likes her.’ b. khaw, duu, dek2 khaw, khuan,-ca?2 ?en2 kwaa2 thee, he look child he should young more she ‘He looks young. He should be younger than her.’ What needs to be clarified is that ca .7, after naa, and khuan, in the epistemic sense is not the modal of volition as suggested in section 5.4.1.4 for ca .7, following naa, and 2" Sentences (107a,c) are acceptable when tan, is the root modal of strong obligation. 161 khuan, in the root sense. The first supporting piece of evidence comes from the following sentences where cak, ‘want’ which is believed to be the source of ca ?2 cannot replace it. (110) a.*khaw, naa, cak2 cheep, thee, he should want like she b."'khaw5 khuan, cak2 ?en2 kwaa2 thee, he should want young more she The word ca .7, in this case is thus different from the volition modal. Like ca ?, in ?aat,—ca ?, and khan ,-ca ? 2, it expresses futurity or prediction. Also, ca ?, is obligatory when these two modals are predicators as in (111)-(112). However, uner ?aatz-ca .7, and khon,-ca ?2, ca ?, in naa,-ca ?, is obligatory whereas ca ?, in khuan,-ca?, can sometimes be omitted as in (113). (111) Q: khaw, naa,-ca?2 cheep, thee, may, he should like. she QP ‘Is it probably that he likes her?’ A: a. naa,-ca?2 should b.*naa, should (112) Q: khaw, khuan,-ca?2 ?en2 kwaa, thee, may, he should young more she QP ‘Is it likely that he is younger than her?’ A: a. khuan,-ca?2 should b."I khuan, should (113) a.*khaw, naa, cheep, thee, he should like she b.??khaw, khuan, ?en2 kwaa2 thee, he should young more she 162 Taking the same line of arguments for ?aat,-ca .7, and khan ,-ca .7 2, I will assume that naa,-ca .7, and khuan,-ca .7, are compound heads as they also block the movement of V across them to the 2 head as in the examples below. (114) Q: khaw, naa,-ca?2 cheep, thee, may, he should like she QP ‘Is it probably that he likes her?’ A: a. naa,-ca?2 should. b.*cheep, like (115) Q: khaw, khuan,-ca?2 ?en2 kwaa2 thee, may, he should young more she QP ‘Is it likely that he is younger than her?’ A: a. khuan,-ca?2 should b."‘?:)n2 young The impossibility of co-occurrence of all epistemic modals as illustrated in (116) below seems to suggest that they occupy the same position as illustrated in (117). (116) a.*khaw, tor], naa,-ca?2 ?aan, nan,sii, he must should read book b.*khaw,terj, khuan,-ca?2 ?aan, narj,sii, he must should read book c.*khaw, khOI], naa, ?aan, nan,sii, he likely should read book d.*khaw, khuan, ?aat2 ?aan, na135sii, he should probably read book (1 1 7) ModP,,,,,,,,,,,,ic tan, Mode, ?aat,-ca?, /khon,-ca .7, /\ naa,-ca .7, /khuan ,-ca ?2 Mod VP 163 Unlike the other epistemic modals which are odd in questions, however, naa,-ca ?, and khuan,-ca .7, can be questioned by may, as in (114-115) above. I will turn to this issue along with some facts about the negatability of naa,-ca .7, and khuan,-ca ?, as distinct from the other epistemic modals in the following section. 5.4.3.4 Snmmm In general, the epistemic modals in Thai have a unified behavior; that is, they are all heads and thus can function as predicators (although the case often, is not conclusively affirrned). However, the fact that naa,-ca ?, and khuan,-ca ?, can be questioned whereas the others cannot has to be explained. As the question particle may, is a shortened form of rii, may, ‘or not’, the ability to be questioned by may, is closely related to the ability to be negated. Negation is then the issue of discussion in the following section. 5.5 Thai negation The most frequently used negator in Thai is may,, which has also been used as a test for verbness (Noss 1964, Needleman 1973, Punyodyana 1976, Sriphen 1982, Indrambarya 1994)“. That is, the word that can be directly negated or be preceded by 1‘ The other two common negators in Thai are may,-day,, and may,-chay,. Their differences and usages have never been thoroughly studied. Although may,-chay, is undisputably assumed to be the negation of copula verbs in Thai; namely, pen ,, khii,, and chay, as they generally precede a noun phrase or a prepositional phrase, the usages of may, and may,-day, are not agreed upon. For instance, Ekniyom (1979: 60-61) makes a claim that may, denotes irrealis negation whereas may,-day,, realis, as illustrated in the examples below: (i) a. khaw, may, pen, khruu, newmwae, he not be teacher certainly 164 may, is treated as a verb. I will examine in this section what syntactic status the negator may, is as well as argue for my analysis of the modals mentioned in the previous section, utilizing the negation fact. 5.5.1 Distribution of may, The Thai negator may, is a free morpheme. It can basically be placed in fi'ont of verbs and adjectives as shown below”: (118) a. khaw5 may, ?aan, nan,sii, he not read book ‘He does not read book.’ ‘He certainly will not be a teacher.’ b. khaw5 may,-day, pen, khruu, naeae,naeze, he not be teacher certainly ‘He certainly is/was not a teacher.’ However, may, does not always convey the irrealis negation as in (ii) below. (ii) a. khaw, may, klap2 baan, miie,khiin, he not return home last night ‘He did not return home last night.’ b. khaw, may,-day, klap2 baan, miie,khiin, he not return home last night ‘He did not return home last night.’ As obvious, the interpretations of (iia,b) are not different. Therefore, the claim by Ekniyom does not sound completely right. Nonetheless, she also points out that may, can precede all verbs, but only some auxiliaries while may,-day, can precede both verbs and auxiliaries. To be solved is whether may, day, is a single unit or it is the combination of the negator may, and the word day, which can be considered either as a modal of permission or a past tense marker. This issue is beyond the scope of my study. 25 In fact, may, can also precede noun phrases quantified by thuk, ‘every’ as in (i) and (ii) below. (i) khaw, ?aan2 nan,sii, may, thuk, lem, he read book not every Cl. ‘He does not read every book.’ (ii) khaw, ?aan2 narj,sii, may, thuk,wan, he read book not everyday ‘He does not read everyday.’ I will assume that may, ‘not’ in the above cases is the constituent negation, not the sentential one. 165 b. khaw, may, dii, he not good ‘He is not good.’ Plus, two or more occurrences of may, in a sentence where the negations cancel each other out are common as in (6), repeated below. (119) a. khaw5 may, ?aan, nan,sii, may, day, he not read book not can ‘He can’t not read.’ b. khaw, may, naa,-ca?2 may, tor], may, tham,rjaan, he not should not must not work ‘It is unlikely that he does not have to not work.’ 5.5.2 Analysis of may, In this section, I will investigate the syntactic status of may,. Although several studies claim that a negator is a head which projects its own maximal projection NegP (Pollock 1989, Ouhalla 1990, Benmarnoun 1991, Zanuttini 1997) or a negator is in the specifier of NegP (Jung 1991, Ernst 1992, 1995, Cinque 1999), I will argue that may,, like Chinese bu, is in the specifier of AuxP or VP (Ernst 1992, 1995). AuxP in this case can be either ModP or AspP because modals and aspect markers are considered two types of Thai auxiliaries. In other words, may, is in the specifier of a phrase whose head has V- features. . In what follows I will present the study on Chinese negation by Ernst (1995) and my proposal of the status of negation in Thai 5.5.2.1 EmsL(l995) Ernst proposes that the negation bu in Chinese is a proclitic in the specifier of AuxP or VP. Modals and aspects are considered two types of Chinese auxiliaries. Therefore, bu can be in the specifier of ModP or AspP. 166 The evidence he gives for his proposal of bu being in the specifier instead of heading NegP is as follows. (1) Assuming that time and epistemic adverbs are left-adjoined to AuxP, the fact that bu never precedes these two adverbs as shown in (120) follows from the proposal that bu is in the specifier of AuxP or VP. (120) a. Ta mingtian hui qu be tomorrow will go ‘He will go tomorrow.’ b.*Baorong bu jintian lai Baorong not today come ‘Baorong isn’t coming today.’ If bu is the head of NegP which takes AuxP as its complement, the ungrammaticality of (120b) cannot be explained. (2) Topicalized direct objects can be preposed to the position before or after adverbs which are adjoined to VP as in (121). (121) Shujuan (yizhi) dianying (yizhi) don bu kan Shujuan (always) movie (always) all not see ‘Shujuan always doesn’t watch movies.’ When such direct objects and bu co-occur, the former must precede the latter as in (122), indicating that the former which is a VP-adjunct is higher than bu which is in the specifier of VP. (122) a. Wo jiu bu he le I liquor not drink ASP ‘I won’t drink liquor any more.’ b.*Wo bu jiu he le I not liquor drink ASP If bu heads NegP, it would precede the topicalized direct objects and (122b) should be acceptable. 167 Evidence for his claim that bu is a proclitic comes fiom the ungrammaticality of *[bu + V + manner adverb] as illustrated in (123). (123) a.*Ta bu jiang de (hen) qingchu he not speak DE very clear ‘He doesn’t speak very clearly.’ b. Ta jiang de bu (hen) qingchu he speak DE not very clear ‘He doesn’t speak very clearly.’ His reason is that the manner adverb which is generated as an adjunct to V’ is postposed and leaves a trace which blocks the cliticization of bu to the following word. Such a construction is, therefore, unacceptable in Chinese. I will show in the following sub—section that the negator may, in Thai is generated in the specifier of AuxP and VP like the negation in Chinese although Emst’s arguments are not applicable to the Thai data due to the post-verbal position of all Thai adverbials. 5.5.2.2 WW; My proposal for the status of may, as NegP in the specifier of XP with V-features will be clarified as follows. (1) Multiple negation As previously stated, a Thai sentence can contain two or three may ,’s which cancel each other out as in (119) above. In fact, more than three occurrences of may, are possible in a Thai sentence, although their actual number may be constrained by processing limitations. Cinque (1999), adopting some ideas of Zanuttini (1997), accounts for sentences with several negations as in (124) below (Cinque 1999:126) by suggesting that a NegP can be base-generated ‘on top of every adverb-related functional projection, even 168 simultaneously, up to a certain height’ (most probably below the projection occupied by the epistemic modal or the speaker-oriented adverb)”5 (124) a. He couldn ’t (possibly) not have accepted. b. He couldnae have no been no working. (Hawick Scots; Brown 1992, 84) ‘It is impossible that he has not been out of work.’ He claims that the first negation of each sentence above (n ’t, nae) is a head of a NegP whereas the rest (not, no) are specifiers of NegPs interspersed among adverb- related firnctional projections.27 If his suggestion is adopted, the syntactic representation of (1 19b) would be as in (125) below. 2" As mentioned earlier, this is due to the scope concerns. That is, the negation never takes scope over the epistemic modal and the speaker-oriented adverb, both of which are assumed to be in the same projection; namely, the latter is in the specifier of the MOdPepistemic‘ 2’ It should be noted here that Zanuttini (1997:23) regards the real sentential negative marker or the one that can negate the sentence by itself as the head of NegP whereas the one(s) that cannot as being ‘left-adjoined to an independently existing functional projection’. To sum up, there is only one NegP preceding the main verb. That is why the negative markers that co-occur in a sentence are not the same (as demonstrated in (124)). 169 (125) NegP may, Mochme not traa,-ca ?, NegP should /\ may, NegP not /\ Neg MOdPobligation tan, NegP must /\ may, NegP not /\ Neg VP tham ,naan , do work However, sentences with multiple negations in Thai (119) and in English (124) (and ahnost all languages studied by Zanuttini 1997 and Cinque 1999) are different in that in English the negator which is considered a head and those as specifiers are different lexical items whereas in Thai the same negator recurs in the same sentence. Therefore, it is hard to tell why in Thai the only first negator is the head whereas the others are specifiers. In other words, what makes them different? Worse, the following coordination data make this an unlikely solution for Thai negation. (2) Co-ordination When two VPs, the first of which contains may,, are conjoined, the sentence is ambiguous in terms of the scope of the negation. That is, the negation can take scope over the first VP only or it can take both VPs under its scope as demonstrated in the interpretations of the following sentence. a; This is different from the sentence with multiple negation in Thai in which the same 170 (126) khaw, may, duu, thii,vii, late, tham, kaan,baan, he not watch TV and do homework ‘a. He will neither watch TV nor do his homework.’ ‘b. He will not watch TV and will do his homework.’ The first interpretation where the negation takes scope over the two VPs does not disprove the claim that the negation is the head or the specifier of NegP because the structure can be as follows.28 (127) NegP /\ may, VP not /\ VP BP duu, thii,vii, //\\ watch TV B VP lcece, tham, kaan,baan, and do homework The second interpretation, on the other hand, seems to pose a problem for the idea that may, is in a NegP above VP because there is no way that the negation could have scope over the first conjunct only, given (127). Therefore, the hypothesis about the negation projecting its own projection above VP does not seem right. The problems above can be solved if the negator in Thai is considered as a NegP situated in the specifier of a maximal projection whose head has V-features, i.e. VP and AuxP. This suggestion can handle the phenomena stated above. The two interpretations of sentence (126) can be taken care of if NegP is in the specifier of AuxP. (AuxP can be either AspP, ModP or TP.) That is, the one where the negation can take both verb phrases in its scope is the outcome of the negative marker can recur. 2" Again I am assuming here the analysis of conjunction as in Munn 1992, 1993. 171 conjunction of VP as in (128). The negation in the specifier of AuxP c-commands both conjoined VPs in (128). On the contrary, the other where the negation has scope over only the first conjunct is the result of AuxP conjunction as in (129). The NegP in the specifier of AuxP of the first conjunct does not have scope over the second conjunct. (128) AuxP /\ NegP Aux’ may, /\ not Aux VP /\ VP BP duu, thii,vii, /\ watch TV B VP lcece, tham, kaan,baan, and do homework (129) AuxP /\ AuxP BP /\ /\ NegP Aux’ B AuxP may3 /\ lcece, /\ not Aux VP and Aux VP /\ /\ V DP V DP duu, thii,vii, tham , kaan ,baan, watch TV do homework Concerning the double or triple negation (or more), it follows automatically if NegP is in the specifier of VP or ModP. That is, it does not matter how many negations appear in a sentence as long as there are enough heads with V-features. The syntactic representation of (1 19b) above is thus as demonstrated below. 172 (130) ModPepistemic /\ may, Mod’ not /\ MOd MOdPobligation naa,-ca .7, should may, Mod’ not /\ Mod VP 1303 /\ must may, V’ not /\ V DP tham, naan, do work (3) Ellipsis Another piece of supporting evidence comes from the following ellipsis sentences which show that the negator may, cannot be stranded. (131) a. khaw5 duu, thii,vii, taeae, chan, may, duu, he watch TV but I not watch ‘He watches TV but I do not.’ b.*khaw, duu, thii,vii, tare, chan5 may, he watch TV but I not (132) a. khaw, tor), tham, naan, tam, chan, may, ten, he must do work but I not must ‘He has to work but I do not.’ b.*khaw, tor], tham, rjaan, tzeze, chan, may, he must do work but I not ‘He has to work but I do not.’ The fact that may, cannot stand alone without a verb or an auxiliary suggests that it is not a head of NegP. If only heads can license an elided constituent (Lobeck 1990, 1995), then may, is not a head. Moreover, the deleted element must be X° or 173 XP (Ernst 1992); consequently, the illegitimacy of the deletion of VP or AuxP without the negator may, shows that may, is a part of VP. In brief, the negator may, in Thai is proven to be a phrase located in the specifiers of various projections the heads of which carry V-features. 5.6 Interaction between modals and negation Assuming that this analysis of Thai negation is on the right track, the ability or inability of the mOdals and the negator may, to co-occur can be explained, and will support the analysis of modals presented in this chapter. 5.6.1 Modals that can be negated by may, All the modals that can be directly preceded by may, are those that can be predicators, namely, naa, / khuan, ‘should’, ten, ‘must’, day, ‘may’, and day, / pen ,/ way, ‘can’ as shown below. (133) a. khaw, may, naa, duu, thii,vii, ‘ he not should watch TV ‘He should not watch TV.’ b. khaw, may, khuan, duu, thii,vii, he not should watch TV ‘He should not watch TV.’ c. khaw, may, tor], duu, thii,vii, he not must watch TV ‘He need not watch TV.’ (1. khaw, may, day, duu, thii,vii, he not may watch TV ‘He is not allowed to watch TV.’ e. khaw, duu, thii,vii, may, day, he watch TV not can ‘He cannot watch TV.’ f. khaw, yok, thii,vii, may, way, he lift TV not can ‘He cannot lift the TV set.’ 174 This confirms my proposal that these modals are heads with V-features. They can move to Z to function as a response to a yes/no question as well as host the negator may, in their specifier. 5.6.2 Modals that cannot be negated by may, The modals that do not allow may, to precede them are the modal of volition ca ?, ‘will’, and the epistemic modals ton, ‘must’, ?aat, -ca ?z‘probably’ and khan , -ca? , ‘likely’ as illustrated below”. (134) a.*khaw, may, ca?2 duu, thii,vii, he not will watch TV b.*khaw, may, tar], duu, thii,vii, he not must watch TV c.*khaw, may, ?aat, duu, thii,vii, he not probably watch TV d.*khaw, may, khon, duu, thii,vii, he not likely watch TV The non-negatability of the volition modal ca ?, can be accounted for in terms of its being an XP, not a head. If ca ?2 is base-generated in the specifier of the Modeo,,,,o,,, then NegP will have no position. An instance confirming the above assumption is the sentence below where only may, is present but the sentence carries the meaning of volition.30 ’9 Example (134b) is fine when the modal ton, has the root sense of obligation. Also, sentence (1346) where ?aat, is negated by may, is acceptable only when the former has the meaning of ability. Thus, the meaning of (134C) can be ‘He cannot watch TV.’ The root interpretation of ?aat,, however, is archaic and becomes rare. Its only use nowadays is after the negator may,.This also supports the analysis by Bybee et a1. (1994) that the root sense of ability gives the epistemic sense of probability. 3° This is similar to Chinese bu which also carries the meaning of volition as below (Zhou 19982). (i) Zhangsan bu kan zhe ni Zhangsan NEG see PROG you 175 (135) khaw, may, duu, thii,vii, phrurj,nii, khaw, nrii, sop, he not watch TV tomorrow he have exam ‘He will not watch TV because he has an exam tomorrow.’ My speculation is that may, in this case is in the specifier of ModPvolition carrying the zero head without the presence of ca ?, in the specifier. Nevertheless, that the negation cannot precede ten, ‘must’, ?aatz-ca .7, ‘probably’ and khan, -ca .7, ‘likely’ does not necessarily falsify the previous hypothesis which considers the epistemic modals as heads. An explanation can be made, following the fact that the negation never has scope over epistemic modals (Cinque 1999:124); it sounds odd when someone negates their own judgment. Therefore, the assumption still holds as there is no evidence strongly against it. However, the fact that the epistemic modals naa,-ca .7, and khuan, -ca .7, can be negated as in (136) needs explanation which I cannot convincingly provide. (136) a. khaw, may, naa,-ca?2 cheep, thee, he not should like she ‘It is unlikely that he likes her.’ b. khaw, may, khuan,-ca?2 ?en2 kwaa2 thee, he not should young more she ‘It is unlikely that he is younger than her.’ My only (blunt) speculation is that these two modals behave differently than the other epistemic modals in terms of scope. That is, they can be under the scope of negation as shown by the English example in (137). Therefore, the similar phenomenon in Thai should not be a surprise. ‘Zhangsan refuses to/won’t look at you.’ Therefore, it is believed to be either a clitic attached to the zero modal of volition (Huang 1988) or a firsion of negation and an empty modal of volition (Zhou 1998). 176 (137) He should not be younger than her. ‘He is unlikely to be younger than her.’ Another puzzle to be solved is the fact that naa,—ca ?, and khuan, —ca ?, , unlike the other epistemic modals, can be questioned by may,. The answer comes naturally when taking into account the fact that may, is the shortened form of rii, may, ‘or not’. The question particle may, is used only in a sentence containing a verb or an element with verbal properties that can be negated. For example, the copula verb khii, in Thai is one of few verbs that cannot be negated as in (138). Therefore, sentence (139) where the question particle may, questions the copula khii, brings about ungrammaticality although khii, can occur in an interrogative sentence containing different question particles as in (140). (138) *khaw, may, khii, phuu,chaay, hua, laan, khon, nan, he not be man head bald Cl that (139) *khaw, khii, phuu,chaay, hua, laan, khon, nan, may, he be man head bald Cl that QP (140) a. khaw, khii, phuu,chaay, hua, laan, khon, nan, rii, he be man head bald Cl that QP ‘Is he that bald man?’ b. khaw, khii, phuu,chaay, hua, laan, khon, nan, chay,-may, he be man head bald Cl that true-or not ‘Is it true that he is that bald man?’ The explanation follows automatically, given that naa,-ca ?, and khuan, -ca .7, can be negated whereas the other epistemic modals cannot. Therefore, the fact that the former can be questioned while the latter cannot is clarified. 177 5.7 naa, and khuan, revisited It should be noted that unlike the other modals discussed in the study, naa, and khuan, can be followed by thii, ca ?2, the element introducing a subordinate clause". The word thii, is a complementizer and ca .7, is likely to be in this case the modal of volition. When naa, and khuan, are followed by thii, ca ?,, the only interpretation is the root sense of weak obligation as illustrated below. (141) a. khaw, naa, thii, ca?2 ?aan, narj,sii, he should COMP will read book ‘He should read.’ b. khaw, khuan, thii, ca?2 ?aan, narj,sii, he should COMP will read book ‘He should read.’ (142) a.* khaw, naa, thii, ca?2 ?aa,yu, naay, he should COMP will age little ‘It is likely that he is young.’ b.* khaw, khuan, thii, ca?2 ?aa,yu, naay, he should COMP will age little ‘It is likely that he is young.’ More interestingly, naa, and khuan, appear to have the similar behavior to the verb ‘seem’ in English which involves the subject DP movement as demonstrated below”. (143) a. naa, thii, khaw, ca?2 ?aan, narj,sii, should COMP he will read book ‘He should read.’ b. khuan, thii, khaw, ca?2 ?aan, narj,sii, should COMP he will read book ‘He should read.’ 3' Sookgasem (1990) suggests that thii, ca .7, is equivalent to ‘to’ in English. It is true that ca? , behaves in many ways like ‘to’. But as seen in (143), thii, ca?, can be separated or intervened by a DP. 1 will leave this for further study. ’2 I would like to thank Alan Munn for pointing this out to me. 178 It may be the case, consequently, that the structure in (141) is derived from that in (143). This also suggests that naa, and khuan, may be verbs or stative verbs (adjectives) that take a clause as a complement like such verbs as yaak, ‘want’, pha,ya,yaam, ‘try’ and tan, cay, ‘intend’ below. (144) a. khaw, yaakztlrii, ca?2 ?aan, narj,sii, he want COMP will read book ‘He wants to read.’ b. khaw, pha,ya,yaam, thii, ca?2 ?aan, narj,sii, he try COMP will read book ‘He tries to read.’ c.khaw, tarj,cay, thii, ca?2 ?aan, narj,sii, he intend COMP will read book ‘He intends to read.’ The omission of thii, or thii, ca .7, is acceptable and the meaning is not affected as shown below. (145) a. khaw, naa, (ca?,) ?aan, narj,sii, he should (will) read book ‘He should read.’ b. khaw, khuan, (ca?,) ?aan, nan,sii, he should (will) read book ‘He should read.’ 0. khaw, yaak2 (ca?,) ?aan, narj,sii, he want (will) read book ‘He wants to read.’ (1. khaw, pha,ya,yaam, (ca?,) ?aan, nan,sii, he try (will) read book ‘He tries to read.’ e. khaw, tar), cay, (ca?,) ?aan, narj,sii, he intend (will) read book ‘He intends to read.’ As noted, the word ca ?, in all the examples above signifies volition or willingness. This supports my analysis which distinguishes between the root modals naa, 179 and khuan,which can be followed by the modal of volition ca .7, and the epistemic ones which incorporates ca ?, denoting futurity or prediction as a part of the words. To sum up, what should be reanalyzed is the status of naa, and khuan, in the root sense. My proposal is that they are not auxiliaries, but true verbs. The root modality in Thai thus does not distinguish between the modals of strong and weak obligation. The structure of mm, and khuan, in the sense of weak obligation is suggested to be as follows. (146) VP /\ nou/khuan , CP should /\ thii, ModPvolition ca ?2 Mod’ Mod VP 5.8 Conclusion In this chapter, I have investigated the two types of modals in Thai: root and epistemic. We have found that their ordering is consistent with the cross-linguistic one proposed by Cinque (1999) where the epistenric modals are situated higher than the root ones. The former are positioned under the same projection whereas the latter are located in distinct projections. Although the modals of weak obligation empirically indicate their separation from that of strong obligation; that is, the former are structurally higher than the latter with the modal of volition between them, the reanalysis of the former as verbs taking a clause as their complement dissolves the unlikely distinction. Moreover, all 180 modals in Thai, both pre-verbal and post-verbal, are generated pre-verbally. VP movement thus operates in Thai syntax. In terms of their syntactic categories, the predicator fact together with the analysis of the negator as a phrase in the specifier of various projections with V-features leads to the conclusion that almost all the modals under investigation, except the modal of volition, are heads. The one that can neither be a predicator nor be negated is a phrase. Those which can be predicators but cannot be negated are heads which are never under the scope of negation. 181 CHAPTER 6 THAI ASPECTUALITY I will discuss Thai aspect markers in this chapter and argue that post-verbal aspect markers are not generated pre-verbally like post-verbal modals, but as right-adj uncts. I will also investigate the relationship between modals and aspect markers in terms of their relative ordering. 6.1 Introduction Similarly to modals, Thai aspect markers can occur pre-verbally and post-verbally as shown in the following examples.1 (1) a. khaw, kam,larj, ?aan2 narj,sii, he PROG read book ‘He is reading.’ b. khaw, ?aan2 narj,sii, laeaew, he read book PRF ‘He read.’ The main objective of this chapter, therefore, is to find out whether the post- verbal aspect markers behave like the post-verbal modals. In other words, are the post- verbal aspectual words base-generated pre-verbally like the post-verbal modals? I will argue that they are not due to their different characteristics. Instead, I will propose that the post-verbal aspects are right-adj uncts. ‘ I will gloss lcecew, as a perfect or a perfective (PRF) until I discuss its nature as a marker of a temporal boundary. Then I will use ‘already’ as the gloss of lcecew, 182 I will discuss each aspect marker in terms of its position in the clausal structure together with its syntactic status as a head or a phrase. Both the predicator and the negatability tests will again be used to discover the relevant facts. The chapter is organized as follows. Section 2 provides a brief overview of aspect. Section 3 spells out the characteristics of Thai aspect markers including their aspect types, surface positions and distributions. My analysis of the aspect markers will be given in section 4. Section 5 explores the co-occurrences of modals and aspect markers with regard to their relative orderings. Section 6 concludes the chapter. 6.2 Aspect Smith (1997) proposes two kinds of aspect: situation and viewpoint. The former is conveyed by the verb and its arguments. It is categorized into five types according to the temporal properties of dynamism, durativity, and telecity as in (2) below (p.3). (2) Situation types State : static, durative (know the answer, love Mary) Activity : dynamic, durative, atelic (laugh, stroll in the park) Accomplishment : dynamic, durative, telic, consisting of process and outcome (build a house, walk to school, learn Greek) Semelfactive : dynamic, atelic, instantaneous (tap, knock) Achievement : dynamic, telic, instantaneous (win a race, reach the top) Viewpoint aspect, on the other hand, is usually signaled by a grammatical morpheme adjacent to the verb and it is of three types: perfective, imperfective and neutral. Their differences are enumerated in (3) below (Smith 199723). (3) Viewpoint types Perfective viewpoints focus a situation in its entirety, including both initial and final endpoints. Imperfective viewpoints focus part of a situation, including neither initial nor final endpoints. 183 Neutral viewpoints are flexible, including the initial endpoint of a situation and at least one internal stage (where applicable). Examples of the above viewpoint types are illustrated below. (4) a. Perfective: Mary walked to school. b. Imperfective: Mary was walking to school. c. Neutral: Zhangsan dao jia de shihou, Mali xie gongzuo baogao (Chinesez79) Zhangsan arrive home DE time, Mali write work report a. When Zhangsan arrived at home, Mali wrote the work report. b. When Zhangsan arrived at home, Mali was writing the work report. I will consider here only the words expressing viewpoint aspect. Since Thai does not seem to have neutral viewpoint aspect, I will not discuss it further. In addition to the imperfective/perfective distinction, it is also important to make a distinction between perfective and perfect since one Thai aspect marker can denote either one. Perfective, according to Smith (1997), indicates that an event is terminated or a situation is temporally bounded; that is, the situation is considered as a single whole with both initial and final endpoints. Perfect, on the other hand, indicates the existence of some past event with some relation to the present. It is classified into three main types: existential, continuative and resultative (Comrie 1976, Demirdache & Uribe-Etxebarria 1997). The existential perfect indicates that an event occurred once or more in the past and the experience holds of the subject. The continuative perfect signifies that an event or a state extends from the past up to the present or the utterance time. And the resultative perfect signals that the result of a past event still holds. The following examples which contain existential perfect, continuative perfect and resultative perfect respectively are from Demirdache & Uribe-Etxebarria (1997:7-8). 184 (5) a. Max was fired (five times/before). b. Henry has lived in Vancouver for three years (now). (= He is still living in Vancouver.) c. Where is John? John has jumped into the ditch. (= John is in the ditch now.) Thai has two words that express perfect. One is khaay, which conveys existential perfect and the other is lcecew, which indicates both continuative and resultative perfect, in addition to perfective. Moreover, I will show that lceccw, also has the characteristics of ‘inchoative phasal adverbial’. All these together with the imperfective aspect markers will be elaborated in what follows. 6.3 Characteristics of Thai aspect markers I will discuss in this section the viewpoint aspect types Thai aspect markers belong to, their surface positions as well as their distributions. 6.3.1 Types of aspect The words which can be regarded as aspect markers in Thai are khaay, ‘having experience of’, kam,lan, ‘in the process of’, yan, ‘still’, yuu, ‘in the process of’, and lcecew, ‘already’. Boonyapatipark (1983) terms them ‘experiential’, ‘progressive’, ‘persistent’, ‘continuative’ and ‘perfective’ markers respectively. However, the aspect khaay, is better termed an ‘existential perfect’ or an ‘experiential perfect’ (EXP) and lcecew,, (for the time being) a ‘perfect/prefective’ (PRF). Moreover, although kam,lan, and yuu, are often referred to as progressive markers (PROG), yuu, can express a variety of meanings including habitual, progressive, and non-progressive. Therefore, the more appropriate term for yuu, is ‘imperfective’ (IMPF) (Meepoe 1997). The term 185 ‘continuative’ will then be reserved for yan, ‘still’ (van der Auwera 1998, Cinque 1999). I will justify my proposal regarding their aspectual types in the next section where each aspect marker is discussed. Summarizing, I suggest that Thai aspect markers belong to different types of aspect as demonstrated below. (6) a. Perfective: Iwaew, ‘PRF ’ b. Perfect: khaay, ‘EXP’, lceaaw, ‘PRF’ c. Irnperfective: kam,lan, ‘PROG’, yan, ‘still’, yuu, ‘IMPF’ 6.3.2 Surface positions of Thai aspect markers As stated in the Introduction, some Thai aspect markers occur pre-verbally and others, post-verbally. As seen in Table 4 below, the pre-verbal aspect markers are the experiential perfect khaay,, the ‘continuative’ marker yam, and the progressive marker kam,lan,. The two post-verbal aspect markers consist of the imperfective yuu, and the perfect/perfective lcecew, A B C D E khaay, yan, kam,lan, VP W“: Imam. ‘EXP’ ‘still’ ‘PROG’ ‘IMPF’ ‘PRF’ Table 4: Positions of Thai aspect markers 6.3.3 Distribution of Thai aspect markers All aspect markers can co-occur in the restricted order illustrated in Table 4 as shown by the examples in (7). 186 (7) 3' khaws yaIJr kamrlaUr ?aanz naIJsSfis Wuz he still PROG read book IMPF ‘He is still reading.’ b. khaw, kam,larj, ?aan2 narj,sii, yuu2 laemw, he PROG read book IMPF PRF ‘He has been reading.’ c. khaw, kheey, kam,lan, ?aan2 narj,sii,yuu2 miie,chan2 maa, thin, he EXP PROG read book IMPF when I come arrive ‘He has an experience of being reading when I arrived.’ Ungrammaticality arises if aspect markers are placed in different orders than in Table 4 as demonstrated below. (8) a.*khaw, kamrlmlryanr ?aanz “31.155555 W112 he PROG still read book IMPF b.*khaw, kam,lan, ?aan2 nan,sii, laeaw, yuu2 he PROG read book PRF IMPF However, the combination of yan , and lcecew, is unacceptable owing to their semantic incompatibility.2 That is, the word yan, indicating persistence or continuation of a situation is not compatible with the word lceww, which expresses the end of an event or a situation as demonstrated in (9).3 2 The co-occurrence of yan , and lcecew, is possible as in (i) below where the meaning of yan, is ‘even’, not ‘still’. Whether the continuative yan,, like those in some other languages, is used in a non-continuative or additive sense in this case awaits further investigation. But it certainly shows the focus meaning of yan ,. (i) khaw, yarj, ?aan2 narj,sii, leeazw, leey, he still read book PRF PRT ‘Even he read the book.’ 3 The incompatibility of yan , and lcecew, also arises when the latter is an ‘inchoative’ phasal adverbial word laeww, ,marking the beginning of an event or a state as in (i). (i) khaWs (Want) 1009 lwawt he (*still) angry already ‘He has started to get angry.’ 187 (9) *khaw, yarj, ?aan2 narj,sii, lezew, he still read book PRF 6.4 Analysis of Thai aspect markers 1 will examine in this section each aspect marker in terms of its syntactic position and category. The two tests discussed in length in the previous chapter; namely, predicator and negation, will be used here. I will divide the analysis into two parts. The first part deals with the pre-verbal aspects and the second with the post-verbal ones. 6.4.1 Pro-verbal aspect markers 1 will show that the three pre-verbal aspect markers are all different in one way or another. The experiential perfect kheey, presents a clear case of being a head above VP. The continuative yan, ‘still’ will be claimed to be in the specifier of an AspP like the modal of volition ca ?,. I will also suggest that the progressive kam,lan, is a head. But unlike kheey,, it does not have V-features. 6.4.1.1 Experien_tial perfect kheey, The word kheey, is considered a past-tense marker by several linguists (Panupong 1970, Uppakitsinlapasarn 1971) because its meaning is translatable as ‘used to’ in English. However, it is better to regard kheey, as an aspectual word as its use is not restricted to past tense as demonstrated below. The four phasal adverbials consist of not yet ‘continuative negative’, already ‘inchoative’, still ‘continuative’ and no longer ‘discontinuative’ (van der Auwera 1993, 188 (10) a. khaw, kheey, ?aan2 narj,sii, lem, nan, he EXP read book Cl that ‘He has an experience of reading that book.’ b. phrurj,nii, khaw, ca?2 kheey, ?aan2 narj,sii, lem, nan, tomorrow he will EXP read book Cl that ‘Tomorrow he will have an experience of reading that book.’ Therefore, the term ‘an experiential marker’ as used by Boonyapatipark (1983) or more specifically, the term ‘existential perfect’ or ‘experiential perfect’ which is defined as ‘the existence of more occurrences of a given past event.’ (Demirdache & Uribe-Etxebarria 199727) better describes the word kheey,. Although the word indicates that a situation occurs or a state holds at least once without an extension to. the present time, the experience of such a situation or state prevails up to the utterance time. Accordingly, the subjects/experiencers in (11a,b) still have an experience of ‘being pretty’ and ‘visiting me’ although the implications of the sentences in (11) are ‘Nit is not pretty any longer’ and ‘he does not visit me now’ respectively. (11) a. nit, kheey, pen, khon, suay, Nit E)? be person pretty ‘Nit used to be a pretty woman.’ b. khaw, kheey, maa, haa, chan, saam, khrarj, he EXP come visit I three time ‘He has visited me three times.’ According to the predicator test as in (12), the fact that kheey, wins over the main verb to become a predicator tells us that it is situated higher than VP as illustrated in (13). (12) Q: khaw, kheey, ?aan2 narj,sii, lem, nan, may, he EXP read book Cl that QP ‘Has he read that book?’ A: a. kheey, EXP 1998) 189 ‘Yes, he has.’ b. * ?aan2 read ( 1 3) AspPexperiential kheey, VP /\ V XP The negatability test confirms the assumption that the experiential perfect khaay, is a head because it can be directly negated by may, which is claimed in the previous chapter (section 5.5) to be in the specifier of a projection whose head has V-features as shown below. (14) khaw, may, kheey, cheep, thee, he not EXP like she ‘He does not have an experience of liking her.’ ‘He never likes her.’ Nonetheless, may, can appear after kheey, when the former is under the scope of the latter as in (15). (15) khaw, kheey, may, cheep, thee, he EXP not like she ‘He has an experience of not liking her.’ ‘He used to dislike her.’ In short, khaay, can undoubtedly be claimed as a head. The different scope relations between the experiential aspect and the negation are determined by their positions relative to each other. That is, the experiential perfect has scope over the negation when the latter is in the specifier of VP as in (16). On the contrary, the experiential perfect is under the scope of the negation when it has the latter in its specifier as in (17). 190 (16) AspP experiential kheey, VP EXP /\ may, V ’ not /\ V DP cheep, thaa, like she (1 7) AspPexperiential /\ may, Asp’ not /\ Asp VP kheey, /\ EXP V DP cheep, thee, like her 6.4.1.2 Continuative van, As an aspect marker expressing ‘continuation’, yan, ‘still’ occurs after the experiential kheey, as shown in (18). The reverse order in (19) leads to ungrammaticality.4 (18) khaw, kheey, yarj, tham, rjaan, teen, thiiarj,khiin, he EXP still do work when midnight ‘He has an experience of still being working at rrridnight.’ (19)*khaw, yarj, kheey, tham, rjaan, teen, thiiarj,khiin, he still EXP do work when midnight ‘ Sentence (20) is acceptable when yan, expresses the meaning of ‘even’, not ‘continuation’. 191 The above examples will be used as evidence for the precedence of kheey, over yan, because yan, cannot be a predicator as illustrated in (20). Therefore, the predicator fact cannot be used here to discover the position of yan , with respect to kheey,. (20) Q: khaw, yan, cheep, thee, may, he still like she QP ‘Does he still like her?’ A: a.*yarj, still b. cheep, like ‘Yes, he does.’ From the predicator fact, is it possible that yan,, like ca ?,, is a phrase in the specifier of a projection? Let’s consider another test - the negatability. As is predicted, yan, cannot be negated by may, as shown below. (21)* khaws may. yao, tham. oaanl yuuz he not still do work IMPF If yan , is to have scope over the negation, the word khan, which also means ‘persist, remain’ is to be inserted’; otherwise, the sentence is unacceptable.‘ 5 I speculate that khan, is a part of the word yan,khon, because the presence or absence of khan, does not make any difference in meaning as shown in (i) below. (i) a. khaw, yarj, khorj, tham, rjaan, he still remain do work ‘He still works.’ b. khaw, yarj, tham, rjaan, he still do work ‘He still works.’ 6 Sentence (22b) is acceptable when its meaning is ‘He has not yet worked.’ The word yan, in this case is a negative counterpart of ‘already’. 192 (22) a. khaw, yarj, khorj, may, tham, rjaan, he still remain not do work ‘He still does not work.’ b.*khaw, yarj, may, tham, naan, he still not do work Yet, it should be mentioned that it is not the case that yon, can never be under the scope of negation’. Taking into account English examples in (23) below, ‘still’ can be either over or under the scope of the negation respectively. (23) a. He still doesn’t love her. b. He doesn’t still love her. Therefore, the fact that yan, cannot be negated is not attributable to scope. To be consistent, ca .7, and yan, which can neither be predicators nor be negated by may, should be of the same syntactic category. Thus, I would propose that the latter is also a phrase in the specifier of a projection lower than the one whose head is kheey, as illustrated below. I will tentatively assume that the projection of yan , is Aschonfinmvc (Cinque 1999). (24) AspPexpen'ential /\ khaayl Asppcontinuativc /\ yaUI ASP, /\ Asp VP ’ If the continuative yan, is to be under the scope of negation, the negation may,day, is to be used as below. (i) khaw, may,day, yarj, cheep, thee, he not still like she ‘He does not still likes her.’ 193 u , . ‘1' When a yes/no question containing the continuative yan, requires a positive minimal response, the verb moves across yan, to the 2 head without violating the HMC as demonstrated in (25)”. (25) 2P /\ z Aschontinuative thamli /\ do yan, Asp’ /\ Asp VP /\ V DP t, naan, work The continuative yan, ‘still’ may be analyzed as a phasal adverbial because it also carries other meanings denoted by the phasal adverbial ‘still’. According to van der Auwera (1993), the continuative ‘still’ can express non-continuative meanings such as additive and iterative.9 The examples given by van der Auwera (1993:647-648) are in (26) while Thai examples are in (27). (26) a. London ist schon, aber Paris ist noch schoner. (German) London is nice, but Paris is still nicer. ‘London is nice, but Paris is still nicer.’ 3 Similar to the modal of volition ca ?, yan, and verb can co-occur as a predicator. I speculate that yan, always moves to [Spec, 2P] but the omission of yan , is possible when it is not focused. 9 van der Auwera (1993) also claims that the continuative phasal adverbial ‘still’ can carry the meaning of perfective as shown in the German example below. However, yan, cannot denote such a sense. (i) Er gewinnt den lauf noch. he wins the race still ‘He will win the race yet.’ 194 b. Ikh volt es nokh geton. (Dutch) I would it still done ‘I would do it again.’ (27) a. chan, nik, waa, nit, suay, laeaew, treae, neey, yarj, suay, khwaa2 I think COMP Nit pretty PRT but Noy still pretty -er/more ‘I think that Nit is pretty, but Noy is prettier.’ b. khaw, yarj, ca?2 tham, mane, phlaat, pay, khrarj, laeaew, He still will do though fail go time PRF ‘He will do it again although he failed once.’ Also, the negative counterpart of yan , is may, lcecew, ‘not already’ (= ‘no longer’ in English). This is exactly what is claimed for the phasal adverbial ‘still’ as shown below. (28) a. khaw, yarj, tham1 rjaan, he still do work ‘He still works.’ b. khaw, may, tham, rjaan, laaew, he not do work already ‘He no longer works.’ Analyzing yan, as a phasal adverbial supports the claim that it is a phrase. As the position of yan , seems to be fixed, I follow Cinque (1999) in proposing that the continuative adverb is in the specifier of Asmedmm. 6.4.1.3 Progressixekamgan, The word kam ,lan, is unarguably considered to be a progressive marker in Thai. It gives the meaning of an event which is going on without the time specification as shown in (29). (29) a. khaw, kam,larj, ?aan2 narj,sii, he PROG read book ‘He is/was reading.’ 195 b. thee, kam,lan, reen, pleerj, she PROG sing song ‘She is/was singing.’ However, unlike progressive markers in many other languages, kam,lan, can be used with states as shown in (30). Therefore, it seems to have both the progressive and the imperfective properties.‘0 (30) a. chan, kam,larj, dii,cay, I PROG happy ‘*I am being happy.’ b. thee, kam,lan, rak, khaw, she PROG love be ‘*She is loving him.’ Relative to the other pre-verbal aspect markers, the progressive kam,lan, occurs after both the experiential perfect kheey, and the continuative yan, as demonstrated below. (31) a. khaw, kheey, kam,lan, ?aan2 narj,sii, teen, fay, dap2 he EXP PROG read book when power out ‘He was reading when the power was out.’ b.*khaw, kam,lan, kheey,?aan2 narj,sii, teen, fay, clap2 he PROG EXP read book when power out (32) a. khaw, yarj, kam,larj, ?aan2 narj,sii, teen, raw, pay, thin], he still PROG read book when we go arrive ‘He was still reading when we arrived.’ b.*khaw, kam,lan, yarj, ?aan2 nan,sii, teen, raw, pay, thiirj, he PROG still read book when we go arrive Only the word order facts can be used to claim that kam,lan, is situated higher than VP but lower than yan, and kheey,. The predicator test, unfortunately, cannot ‘° According to Comrie (1976), the progressive markers in some languages have the properties of both progressive and imperfective. Therefore, their position may be characterized as ProgP/ImpP, not simply ProgP. I will return to this later. 196 confirm this hierarchical order because kam,lan, cannot be a predicator as shown in (33). What’s more, the predicator of a response to the question containing kam,lan, cannot be a bare verb either. The required predicator is a verb together with kam,lan,. (33) Q: khaw, kam,lan, ?aan2 nan,sii,rii, /*may, he PROG read book QP ‘Is he reading?’ A: a.*kam,larj, PROG b."‘?aan2 read c. kam,lan, ?aan2 PROG read ‘Yes, he is.’ Example (33) also shows that a sentence containing kam,lan, cannot be questioned by the question particle may,. As mentioned in Chapter 5, section 5.5.2, may, is the shortened form of rii, may, ‘or not’. Therefore, it should not be used to question an item that cannot be negated by may, ‘not’. And it is true that kam,lan, cannot be directly negated by may, as below“. (34) *khaw, may, kam,larj, ?aan2 narj,sii, he not PROG read book However, from the discussion in section 5.5.2, the words that can neither be negated by may, nor be questioned by may, are those that may never be under the scope of “ Like ca ?, and yan,, if the negation has to take scope over the progressive kam,lan,, the negation may,day, is to be used as illustrated below. (i) khaw, may,day, kam,lan, ?aan2 narj,sii, he not PROG read book ‘He is not reading.’ 197 negation, namely epistemic modals. In contrast, the modal of volition ca ?, and the continuative aspect marker yan, cannot be negated by may, but can be questioned by the question particle may,. Below is the paradigm summarizing this fact. (35) Negatable (mays) Questionable (may) Epistemic modals x x Volition modal ca ?, x ‘1 Continuative yan,‘still’ x ‘1 Progressive kam,lan, x x An explanation for the above puzzle requires a more profound analysis of the question particle may,. I will begin with the assumption that may, is a reduced form of rii, may, ‘or not’, and that when may, is present, rii, can be phonologically null. It is possible that the change in tone from may, to may, is the phonological reflex of the null conjunction. I will tentatively hypothesize that the conjunction rii, ‘or’ conjoins two ZIPS, the first of which carries the feature [Aff] and the second of which carries the feature [Neg]. With the negator may, suggested to be in the specifier of SF in this case, the sketchy structure of may, is as in (36). 198 (36) CP /\ C 2P /\ ' 2P BP /\ B 2P Vii, /\ or may, 2’ I101 /\ 2 XP Consistent with my analysis of may, in Chapter 5, section 5.5, I will assume that may, occupies the specifier of 2P, but because may, needs to be supported by a verbal head, 2° must be filled with such a head. Once this is satisfied, 2’ may delete at PF. Let’s take a simple sentence as in (37) where (37b) is the full form of (37a). The representation in (3 8) shows how it works. (37) a. khaw, maa, may, he come QP ‘Did he come?’ b. khaw, maa, rii, may, maa, he come or not come ‘Did he come?’ (3 8) 2P /\ 2P BP /\ B 2P r ii5 /\ or may, 2’ 1101 /\ 2 VP f I V maa, come 1 199 After the verb maa, ‘come’ has moved to 2° to satisfy the requirement of may,, 2’ (which contains the verb maa, ‘come’) is deleted, thus giving the sentence in (37a) at PF. Now we can return to the paradigm (35). I will attempt to account for each case. First, the epistemic modals are neither negatable by may, nor questionable by may,. This can be explained by the fact that they are never under the scope of negation. Given the suggested analysis of the question particle may,, the closest head to be attracted to 2 is the epistemic modal which is situated higher than V. Although the derivation converges, the incompatibility in terms of soope relation makes the sentence ungrammatical. That is, the epistemic modal under the scope of negation is unacceptable. This is represented below. (39) a.*khaw, khorj, maa, may, he likely come QP ‘Is it likely that he comes?’ b.*khaw, khOI], maa, iii, may, khorj, maa, he likely come or not likely come c. 2P /\ 2P BP /\ B 2P fit, /\ or may, 2’ not /\ 2 ModP,.,,,,,,mic 4 Mod VP khoUli i likely V .__J maa, come 200 For the case of ca .7, which is a phrase in the specifier of ModeHu-on, it cannot co- occur with the negator may, because they compete for the same position. However, the fact that ca ?, can be questioned by may, is due to the fact that the negator may, is in the specifier of 2P. The attracted verb can move past ca ?, to 2 without violating the HMC as illustrated below. (40) a. khaw5 ca?2 maa, may, he will come QP ‘Will he come?’ b. 2P /\ 2P BP /\ B 2P r if, /\ or may, 2’ not /\ 2 MOdeolition ca .7, Mod’ will /\ Mod VP | i V maa, come I The case of yan, ‘still’ is a bit more complicated although it behaves the same as ca ?2. As the negative counterpart of yan , is may, lcecew, ‘not already’ or ‘no longer’ in English as discussed in section 6.4.1.2, the second conjunct consists not of yan , but may, lcecew,, as shown in (41). 201 ( 41) a.*khaw, yarj, maa, rii, yarj, may, maa, he still come or still not come ""Does he still come or doesn’t he still come?’ b. khaw,ya1], maa, rii, may, maa, lcaew, he still come or not come already ‘Does he still come or doesn’t he come any longer?’ Assuming that lcecew, is right-adjoined to some XP higher than VP (as I will argue for in section 6.4.2.3), the fact above can be accounted for by the representation below where V can move to 2 without any problem. The fact that yan, is not negatable by may, but is questionable by may, can thus be explained. (42) a. khaw,yarj,maa, may, he still come QP ‘Does he still come?’ b. khaw, yarj, maa, rii, may, maa, laezew, he still come or not come already ‘Does he still come or doesn’t he come any longer?’ c. 2P /\ 2P BP /\ B 2P r l l 5 /\ or may, 2’ I101 /\ 2 XP A i\ XP lcecew, /\ already X VP L—A 1 V maa I come I 202 Now let’s consider the case of the progressive kam,lan, which can neither be negated by may, nor questioned by may,. Although kam,lan, patterns like the epistemic modals, it is certainly not the case that the progressive aspect cannot be under the scope of negation like epistemic modals. The English examples in (43) show the scope of negation over the progressive. Their interpretations are ‘It is not the case that he is reading’ and ‘It is not the case that she is singing’ respectively. (43) a. He is not reading. b. She is not singing. Therefore, the questions need solving are the following: Q1: Why can’t kam,lan, be negated? Q2: Why can’t it be questioned by may,? Q3: Why does it always accompany the verb when the latter functions as a predicator? Question 1 seems to point to the idea that the progressive kam ,lan, is a phrase in the specifier of a projection like ca ?, and yan, because unlike the epistemic modals, it can be under the scope of negation. However, this assumption cannot explain questions 2 and 3. If kam ,lan, were in the specifier, the verb would unproblematically move to 2 past it in the sentence questioned by may,, resulting in grammaticality, contrary to the fact. Instead, I hypothesize that kam,lan, is the head. However, it is different from other modal or aspectual heads in that it lacks V-features. My proposal comes fi'om the fact that the progressive marker in Thai is derived from the noun which means ‘strength, power’(Ekniyom 1979, Meepoe 1996). This idea is not uncommon because in many 203 languages, the progressive has certain properties that could be taken as nominal or at least not purely verbal.l2 Therefore, the fact that kam,lan, cannot be questioned by may, is because kam,lan, is unable to move to 2 owing to its lack of V-features (44b). However, since kam,lan, is a head, movement of V to 2 will violate the HMC. This is shown in (44c). (44) a."‘khaw5 kam,lan, maa, may, he PROG come QP b. 2P /\ 2P BP /\ B 2P r R 5 /\ or may, 2’ not /\ 2 ProgP /\ Prog VP kam,lan, | V maa , come ‘2 This is pointed out to me by Cristina Schmitt. One example, suggested by Asuncion Martinez, is the French progressive 'en train de' which is a PP. 204 /\ 2P BP /\ B 2P m, /\ or may, 2’ 1101 /\ 2 ProgP A /\ Prog VP kam ,lan, | V maa , come at J The proposal that kam,lan, is the head without V-features can also answer Question 1; that is, it is incompatible with the negator may, which carries V-features. However, Question 3 still needs an explanation. Given that a head with V-features has to move to 2 to function as the minimal response, the verb under kam,lan, must adjoin to kam,lan, on its way to 2. This can explain why a bare verb is unacceptable as a predicator when the progressive marker is present. The predicator, on the other hand, has to consist of both a verb and the progressive marker as illustrated in (45) below. (45) a. 2P /\ 2 ProgP /\ Pro g VP /\ i V Prog V, kam,lan, maa, I, come 205 /\ 2 ProgP , | /\ Prog, Prog VP t, | Prog V, V kam ,lan, maa , t, come Another piece of evidence supporting my proposed analysis of may, and kam,lan, is the following questions containing the progressive marker and either ca ?, ‘will’ (46a) or yan, ‘still’ (46b). (46) a.*khaw, yan, kam,lan, maa, may, he still PROG come QP b.*khaw, ca?2 kam,lan, maa, may, he will PROG come QP Recall that ca ?, and yan, without kam,lan, can be questioned by may,. The ungrammaticality of (46) can thus be explained as follows. Although V can move past ca .7, which is in the specifier of ModP,0,,,,,,, on its way to 2, it cannot skip the head kam,lan,. The same line of argument applies to (46b). That is, although the negative counterpart of yan , - lcecew, ‘already’ is a right adjunct and thus does not block V movement, the head kam,lan, above V does. In short, the progressive marker in Thai is suggested to be a head of ProgP above VP below the experiential perfect kheey, and the continuative yan, ‘still’ as illustrated in (47). 206 (47) Asppexperiential /\ Asp Aschontinuative kheey, /\ yam Asp still /\ Asp ProgP 9 6.4.1.4 Smnmary To conclude, pre—verbal aspects are all different from one another. It is clear that the experiential perfect kheey, possesses all the verbal properties; namely, its ability to be a predicator and to be negated by may,. Therefore, it is undoubtedly a head of an AspP ,Wmm. The continuative marker yan, and the progressive kam,lan, can neither be a predicator nor be negated. However, the former can be questioned by may,. The proposed analysis is that the continuative marker which behaves similarly to the modal of volition is a phrase in the specifier of Asmefimm. The progressive marker, on the other hand, behaves differently in terms of its inability to be questioned by may, as well as its obligatory accompaniment of the verb which moves up to 2 to act as a predicator. Therefore, it is analyzed as a head without V-features. 6.4.2 Post-verbal aspect markers The two aspect markers that occur post-verbally are yuu, and lcecew, The former is considered as an imperfective whereas the latter, perfect as well as perfective. Their co- 207 occurrence is possible where the imperfective precedes the perfective, the order which is opposite to the cross-linguistic order of these two aspect markers. 6.4.2.1 W In the earlier studies, the combination of a verb and a post-verbal aspect marker was considered as a compound verb (Sereechareonsatit 1984) or a complex verb (Thepkanj ana 1986) in which verbs modify others in the sequence as illustrated below: (48) Sereechareonsatit (1984:262) a. chat, ?aan2 narj,sii, yuu2 laeaew, Chart read book IMPF PRF ‘Chart has already been reading books.’ b. S /\ NP VP chaat, /\ Chart VP V /\ 16353194 VP V PRF /\ yuu, V NP IMPF ?aan, nan,sii, read book (49) Thepkanjana (1986:189) a. su2rii, khiian, cot,maay, yuu2 Suri write letter IMPF ‘Smi is writing a letter.’ b. S /\ NP VP suzriil /\ Suri VP V khiian, cotynaay, yuu, write letter IMPF 208 I agree with the above analyses which place the post-verbal aspect markers as right-adjuncts. However, I disagree on their being heads because considering yuu, and lcecew, as heads cannot explain their different behaviors fi'om other heads discussed so far, as will be elaborated in the following section. Moreover, the analyses by which a maximal projection is mixed-headed is not favorable. Plus, the proposal that these two aspect markers are VP right-adjuncts misses the fact that they take scope over post-verbal modals which are assumed in the previous chapter (section 5.4.1.5) to be above VP. In spite of the suggestion that the post-verbal modals are also VP right-adjuncts like these two as in (50), it cannot explain why the order of these three is fixed, given that they are adjuncts. (50) VP /\ VP V /\ lcecew, VP V PRF /\ ya“: VP V IMPF /\ day, V DP can My proposed solution is that the post-verbal aspect markers are right adjuncts to different XPs. That is why their order cannot be switched around. And the fact that they are not heads can explain their different behaviors than the other auxiliaries which are claimed to be heads. 6.4.2.2 Imperfecfixem, The aspectual meaning of yuu , is not totally agreed among the linguists. It has been regarded in the literature as a ‘progressive’ marker (Warotamasikkhadit 1972), a 209 ‘continuative’ marker (Boonyapatipark 1983), and an ‘imperfective’ marker (Meepoe 1996). If the following diagram proposed by Comrie (1976: 25) is taken into consideration, it can be concluded that yuu, is an imperfective as it can represent all the aspectual meanings under ‘Irnperfective.’ (51) Classification of aspectual oppositions Perfective Irnperfective 1 Habilual Continuous I I 1 Nonprogressive Progressive The following examples represent the aspectual meanings yuu, can convey - habitual, nonprogressive and progressive respectively. (52) a. khaw, suup2 bu,rii2 yuu2 serj, pii, he smoke cigarette IMPF two year ‘He smoked for two years.’ b. khaw, ?uan, yuu2 he fat IMPF ‘He is fat.’ c. khaw, ?aan2 narj,sii, yuu2 he read book IMPF ‘He is/was reading.’ In what follows, I will limit myself to the progressive use of yuu2, but I will call it imperfective to distinguish it from the progressive kam,lan,. Let’s look at the behaviors of yuu, with respect to the ability to be a predicator and be negated. It happens that yuu, , like most of the aspect markers discussed so far, cannot behave as a predicator as shown in (53) below. 210 (53) Q: khaw, ?aan2 narj,sii, yuu2 may, he read book IMPF QP ‘Is he reading?’ A: a.* yuu2 IMP b. ?aan2 read Given the fact that yuu, cannot be a predicator, it is not unexpected to find it unable to be negated by may, as in (54). (54) a.*khaw, ?aan2 narj,sii, may, yuu2 he read book not IMPF b. khaw, may, ?aan2 nan,sii, yuu2 he not read book IMPF ‘He is not reading.’ However, a sentence containing yuu, can be questioned by may, as shown in (55). (55) khaw, ?aan2 narj,sii, yuu2 may, he read book IMPF QP ‘Is be reading?’ Given the above facts, yuu, behaves exactly the same as the modal of volition ca .7, and the continuative yan, in all respects concerned. As a result, the most likely claim is that yuu, is not a head, but a phrase. Is it possible then that yuu, is a phrase in the specifier of 11in as in (56)? (56) Irin /\ W“: 11111), /\ Imp VP The structure shown in (56) seems to be able to explain the failure of yuu, in both the predicator and the negatability tests. As a phrase, yuu, cannot raise to 2 to function as a predicator. And owing to its location in the specifier of ImpP, the negator may, cannot 211 be merged in the very position. Moreover, the movement of V past yuu, to 2 in an interrogative containing may, is unproblematic. Unlike ca .7, and yan,, nonetheless, yuu, always occurs post-verbally. Can VP raising account for the word order fact? This seems promising until the following instance where the order of yuu, and prepositional phrases is switched around is taken into account. (57) a. khaw, kin, khaaw, thii, baan, kap2 phee, yuu2 he eat rice at home with father IMPF ‘He is eating with his father at home.’ b. khaw, kin, khaaw, thii, baan, yuu2 kap2 phee, he eat rice at home IMPF with father c. khaw, kin, khaaw, yuu2 thii, baan, kap2 phee, he eat rice IMPF at home with father There is no way the proposal that yuu, is generated pre-verbally can account for this instance no matter where PPs are assumed to be generated. If PPs are considered ‘adverbial PPs’ as claimed by Cinque (1999:30) and generated in the specifier of various VP shells as in (5 8), yuu, has to lower and tuck in between or after these two PPs to get the word order right. This is unlikely because lowering causes ECP violation. Is it possible that different shells of VP move to a position higher than yuu,? According to this proposal of the base-generation of adverbial PPs, the lowest VP has to successively adjoin to the higher VPs to get the order right. To account for (57b), therefore, the lowest VPk has to adjoin to the VP,, yielding kin, khaaw, thii, baan, ‘eat rice at home’. To get the order right, VPj has to adjoin to ImpP, skipping the highest VP, This too is unlikely. 212 (58) lrin Wuz ImP, /\ ImP VPi /\ PP V’ kap, PhD-73 /\ with father V VPj /\ PP V’ thii, baan, /\ at home V VPk kin, khaaw, eat rice Alternatively, if PPs are suggested to be VP right adjuncts as in (59), the lowering and tucking-in of yuu, to PPs is still required, which is not favorable. The movement of VP without all adjuncts is not suggested either. (59) Irin /\ yuu, ImP, /\ Imp VP /\ VP PP /\ kap, Ph933 VP PP with father /\ thii, baan, V XP at home If yuu2-as—right-adjunct is assumed, the problem can be solved. However, it is unlikely that yuu, is VP right-adjunct mainly because yuu, takes the modals of ability under its scope as demonstrated in (60). And the modals of ability are claimed in section 5.4.1.5 to be above VP. 213 |--1 (60) khaw, kin, day, yuu2 he eat can IMPF ‘He is able to eat.’ (or ‘His ability to eat is in process.) What I would suggest then is that yuu2 is adjoined to an AspP which I will later specify. The flexible order as in (57) is captured by the rightward movement of PPs to some focus position above yuu,. The phonological facts seem to confirm this. That is, with yuu, as the last element, no pause is required. When yuu, is placed between or in front of PPs, a pause is needed after yuu,. With the assumption of yuu, as a ri t-adjunct, the fact that yuu, cannot be negated nor be a predicator can be accounted for. As well, the fact that a sentence containing yuu, can be questioned by may, follows from my analysis of may,; that is, V can move to 2 without yuu, blocking the movement. Now consider the projection to which yuu, is adjoined. As mentioned earlier, the progressive marker in Thai seems to have the dual properties of both progressive and imperfective because it can occur with states. Therefore, the co-occurrence of the progressive kam,lan, and the imperfective yuu, is very common and natural as in (61) below. They are therefore considered by some authors to be a unit called a discontinuous temporal verb (Sookgasem 1990) or a discontinuous auxiliary (W arotarnasikkhadit 1996). (61) a. khaw, kam,lan, ?aan2 narj,sii,yuu2 he PROG read book IMPF ‘He is reading.’ b. khaw, kam,larj, reen,pleetj, yullz he PROG sing song IMPF ‘He is singing.’ 214 However, the combination of kam ,lan, and yuu, conveys simply the meaning of progressive, not continuative as shown in (62). Therefore, I will propose that kam,lan, is in ProgP/ImpP to which the imperfective yuu, is right-adjoined. Their structural representation is as in (63). (62) a.*khaw, kam,lan, suup2 bu,rii2 yuu, saam, pii, he PROG smoke cigarette IMPF three year ‘He is smoking for three years.’ (63) ProgP/ImpP ProgP/ImpP yuu, Prog/Imp VP kam ,lan, /\ V XP The proposal above is able to take care of the scope concerns between yuu2 and other pre-verbal aspect markers. Like the progressive kam,lan,, yuu, is generated lower than the experiential perfect kheey, as shown by the predicator test below. (64) Q: khaw, kheey, diirn2 law, yuu2 teen, naay, maa, truat2 may, he EXP drink alcohol IMPF when boss come supervise QP ‘Has he ever been drinking when his boss came for supervision?’ A: a. kheey, EXP ‘Yes, he has.’ b."'diim2 yuu2 drink IMPF However, the predicator test cannot be used to test the position of yuu, with respect to the continuative yan, because neither can be a predicator. However, the fact that the continuative yan, seems to have semantic scope over the imperfective seems to 215 indicate that the former is higher than the latter. This is shown in (65), the meaning of which is the on-going action of ‘reading’ persists at the reference time. (65) khaws yallr ?aanz nalllsSfis yuuz he still read book IMPF ‘He is still reading.’ I will show in the following section that Icecew, has the same behaviors as yuu,. It will, therefore, be claimed to be a right-adjunct too. 6.4.2.3 Perfect/Perfective Iceww, The word Icecew, is considered a perfective marker by Boonyapatipark (1983) although he realizes that it can convey either perfect or perfective. Likewise, Sookgasem (1990) considers Icecew, as two separate aspectual words: perfect and perfective. The differences between the two are that the perfective cannot occur with stative verbs and the perfect ‘involves the overlap of a described interval of eventuality and an interval of utterance.’ (1990:67) The differences between the two aspects are illustrated in the examples below. (66) a. khaw, cheep, thee, saam, pii, laerew, he like she three year PRF ‘He has liked her for three years.’ b. khaw, yaay, baan, pay, laerew, miie, pii, ken, he move house go PRF when year before ‘He moved last year.’ The word Icecew, in sentence (66a) can be considered as a continuative perfect because the action of liking started three years ago and continues to the utterance time while lcecew, in (66b) is perfective because it indicates the termination of the action 216 ‘moving’. However, a sentence containing lcecew, may carry several readings as illustrated below. (67) khaw, kin, khaaw, lasaew, he eat rice PRF ‘He ate.’ ‘He has eaten.’ ‘He has started eating.’ As can be seen, sentence (67) has three interpretations. It can indicate the termination of the action ‘eating’, which is the concept of ‘perfective.’ It can also be an answer to the question ‘Would he like something to eat?’ In this sense, Icecew, is regarded as a resultative perfect because it implies the resulting state of having eaten; that is, the agent is now full. The last meaning indicates the shift or transition of an event or state. The word lcecew, thus signals an arrival or a start of a new situation or state and the termination of a previous state is left understood (Boonyapatipark 1983:166). In fact, lcecew, is commonly used in the last sense. This is shown by additional examples in (68) below, the implications of which are ‘He was not fat previously’ and ‘He didn’t like her earlier’ respectively. (68) a. khaw, ?uan, laeaw, he fat already ‘He is now fat.’ b. khaw, cheep, thee, laeaew, he like she already ‘He has started to like her’ Such a meaning is assigned to a word equivalent to ‘already’ in various languages and the word is termed ‘inchoative phasal adverbial’ (Lobner 1989, van der Auwera 1998). For simplicity, I will reanalyze the word lcecew, ‘already’ in Thai to be a temporal 217 boundary marker. That is, it can mark the beginning of a state or an event, thus carrying the non-terminative and progressive sense. Plus, it can mark the end of an event, thus expressing the terrninative sense. The former usage of lcecew, is roughly equivalent to the perfect marker including the inchoative adverbial whereas the latter usage is the perfective marker alone. I will also assume that lceaew, can be generated in different positions, and distinct loci correspond to various meanings it denotes. I will show that the position of lazcew, is flexible due to its adverbial nature. Therefore, I consider Icecew, as ‘already’ and will gloss it as such. The examples in (69) obviously show the different positions of lcecew, with respect to the experiential perfect kheey,. (69) a. chan, kheey, kin, khaaw, laeaew, tare, terj, ?eek, pay, kin, I EXP eat rice already but must exit go eat khaan,neek, kap, phiien, ?iik, outside with fiiend again ‘I have an experience of having eaten but having to go out to eat with my fiiend again.’ b. chan, kheey, kin, kop2 lamw, 1 EXP eat frog already ‘I have already had an experience of eating frog.’ The word lwww, ‘already’ in (69a) which is under the scope of the experiential perfect kheey, marks the termination of the event ‘eating’ while lcecew, in (69b) which takes kheey, in its scope marks the begimring of the state of ‘having experience’. Therefore, lcecew, may be generated either above or below the experiential perfect. How about its position relative to the other aspect markers then? 218 With respect to the continuative yan,, Iceww, ‘already’ is incompatible with it as shown in (70) because Icecew, marks the beginning or end but yan, displays continuation which excludes the beginning or end of a state or a situation. (70) *khaw, yarj, kin, khaaw, laeaew, he still eat rice already Since laacew, as the marker of the beginning of an event or a state includes the progressive meaning, the co-occurrence of lazcew, and the progressive kam,lan, and/or the imperfective yuu, is perfectly fine as in (71)-(72) below. (71) A: hiw, carj, hungry very ‘I’m very hungry.’ B: mace, kam,lan, tham, ?aa,haan, laeaew, Mom PROG do food already ‘Mom has been cooking.’ (72) khaw, kin, khaaw, yuu, laeaew, he eat rice IMPF already ‘He has been eating.’ From the translations of (71) and (72), it is obvious that Icecew, has scope over the progressive kam,lan, and imperfective yuu,; therefore, it should be structurally higher. As a result, I would assume that lcecew, can be generated from a position above ProgP/Irin up to a position higher than the experiential kheey,. But what syntactic category is laecew,? Is it a head or a phrase? I will suggest that it is a phrase like the imperfective yuu, because they exhibit the same behaviors as will be spelled out below. 219 Like yuu,, lcecew, cannot be a predicator as in (73) nor be negated by may, as in (74). (73) Q: khaw, ?aan, nan,sii, laezew, yan, he read book already yet ‘Has he read?’ A: a.*?aan, read b.*lmaew, already c. ?aan2 laaew, read already ‘Yes, he has.’ (74) a.*khaw, ?aan, narj,sii, may, laeaew, he read book not already b. khaw, may, ?aan2 narj,sii, lteaew, he not read book already ‘He did not read any longer.’ As the interpretation of (74b) shows, the sentence is not the negative of ‘He has already read.’ In section 6.4.1.2, I mentioned that may, lcecew, ‘not already’ is the negative counterpart of yan , ‘still’. Therefore, (74b) is the negative of ‘He is still reading.’ On the other hand, the negative counterpart of lcecew, ‘already’ is yan, may, ‘still not’ (or ‘not yet’ in English) as shown in (75). This is common in any language which has phasal adverbials. (75) khaw, yarj, may, ?aan, nan,sii, he still not read book ‘He has not yet read.’ Consequently, that the sentence containing lcecew, must be questioned by yan, which is the shortened form of rii,yan, ‘or (not) yet’, instead of may,, as in (76), is 220 explainable, given that a yes/no question in Thai is a kind of alternative question as in Chinese. (76) khaw, ?aan, nan,sii, lmaw, yan, he read book already yet ‘Has he already read?’ I would propose that yan, as a question particle is a negative element. Because its full form is rii,yan, ‘or yet’, I would assume that it works like the question particle may,. The movement of V to 2 whose specifier hosts yan, is thus fine for the case of (76). Taking the same line of arguments as given for the right-adjunction status of yuu ,, the following examples where the order of Icecew, and PPs can be switched around show that Icecew, is also a right adjunct. (77) a. khaw, kin, khaaw, thii, baan, kap, phee, lmaew, he eat rice at home with father already ‘He ate/started to eat with his father at home.’ b. khaw, kin, khaaw, thii, baan, lwaew, kap, phee, he eat rice at home already with father c. khaw, kin, khaaw, laeaew, thii, baan, kap2 phee, he eat rice already at home with father In short, the similar behaviors of yuu, and lcecew, indicate that they are of the same syntactic category, namely, phrases right-adjoined to XPs.13 ‘ '3 I have no account here of the fact that lacew, must accompany the verb to function as a predicator. As a phasal adverbial, Icecew, is believed to have a focus property. Whether this has something to do with this issue or what the analysis exactly is awaits further research. 221 6.4.2.4 Summary The two post-verbal aspect markers in Thai are not heads generated pre-verbally with VP raising like the post-verbal modals. In contrast, they are right-adjuncts. The imperfective yuu, is adjoined to ProgP/ImpP whereas lcecew, which is analyzed as a temporal boundary marker is adj oined to several phrases fiom AspPexpcfimw to XP above ProgP/ImpP as in (78) below. (78) AspP experiential \ \\ AspP experiential (lwwwt) khaayl Asppcontinuative /\ Aschontinuative (1‘8wa yan, ProgP/ImpP Pro gP/ImpP yuu, kamlan, VP 6.5 Co-occurrences of modals and aspect markers Having discussed both modals and aspects separately, 1 will now explore their co- occurrences which are normal in a sentence. Although the general claim is that modals are positioned structurally higher than aspect, the empirical data in Thai do not prove so. It seems that they are interspersed, i.e. some modals are higher than some aspects and vice versa.“ What I discover is that the epistemic modals are structurally highest, which " This is not different from Cinque’s account (1999). Also, as pointed out by Cristina Schmitt, there are many languages where some modals are inflected while others are 222 is not uncommon, given that they are like ‘speaker-oriented’ adverbs. The post-verbal modals are found to be the lowest, which is again not surprising, given that they trigger the raising of VP right below them. 6.5.1 Surface positions of modals and aspect markers As discussed earlier, both modals and aspects can occur either pre-verbally or post-verbally. The two tables below represent their co-occurrences in the pre-verbal and the post-verbal positions respectively. A B C D E F ?aat,ca?, ‘probably’ kheey, yan, ca?, 1903 kam,lan, VP khan/ca?, ‘likely’ ‘EXP’ ‘still’ ‘Win’ ‘must’ ‘PROG’ naa, ca .7, ‘should’ khuan , ca ?, ‘should’ ten, ‘must’ Table 5: Positions of pre-verbal modals and aspect markers G H I VP day, ‘can, may’ yuu, lcecew, pen, ‘can’ ‘IMPF’ ‘PRF’ way, ‘can’ Table 6: Positions of post-verbal modals and aspect markers defective or do not have a real perfective form. In the latter case, tense and aspect are assumed to be higher. 223 As can be seen in the tables above, the root modals are intervened by the progressive kam,lan,. The question is whether the progressive marker is higher than the post-verbal modals or vice versa. 6.5.2 Distributions of modals and aspect markers All the pro-verbal modals and aspects in Table 5 can co-occur in the specified order as long as they do not convey incongruous or odd meanings as exemplified below: (79) khaw, khorj, kheey, terj, kam,lan, tham, naan, he likely EXP must PROG do work teen, khon, ?iin, klap2 lzeaew, when person other return already ‘It is likely that he has an experience of obligatorily be working while others lefi.’ All the post-verbal items can also co-occur as in (80) below". (80) a. khaw, ?aan, narj,sii, day, yuu, he read book able IMPF ‘He is able to read.’ ' b. khaw, ?aan, nan,sii, day, yuu2 lamw, he read book able IMPF already ‘He has already been able to read (for some time now).’ However, it is not the case that all pre-verbal and post-verbal items can co-occur in a sentence as shown below: (81)*khaw,khorj, kheey, yan, terj, kam,lan, tham, naan, day, yuu, lzemw, he likely EXP still must PROG do work able IMPF already Some of the possible combinations are illustrated as follows. '5 If the post-verbal modals and lceazw, co-occur without yuu, as in (i) below, the latter is a ‘phasal adverbial’. It signals the shift of an event or a state as discussed earlier. (1) khaw, ?aan, narj,sii,day, lmaew, he read book able already ‘He is able to read now.’ 224 (82) 3- khaws khOI], khaaYr Yam tall; kamrlanr thamr Uaanr yuuz he likely EXP still must PROG do work IMPF ‘It is likely that he has an experience of obligatorily be working...’ b. khaw, khorj, terj, tharn, naan, day, yuu2 laeew, he likely must do work able IMPF already ‘It is likely that he must have been able to work.’ The following section will examine their positions relative to one another. 6.5.3 Analysis I will justify my proposed analysis of modals and aspects in Thai by investigating their co-occurrences in pairs. The epistemic modals and the root modals are discussed as monolithic elements where applicable. The analysis is divided into two main parts. The . first one concerns the relation between the epistemic modals and the aspect markers. The second one is about the relation between the root modals and the aspect markers. 6.5.3.1 Epistemicmodaluansnectmarkers As suggested in the previous chapter (section 5.4.2), the epistemic modals are monolithic. They are situated in the same position above all root modals. I will show in the following that they are also higher than all aspect markers. (1) Epistenric modals > Experiential kheey, It is common that epistemic modals have to be positioned highest in the tree as they involve the speaker’s judgment. Therefore, they are usually considered as equivalent to ‘speaker-oriented adverbs’ which should have scope over the proposition where other 225 modals and aspects are embedded (Ernst 1997). The reverse order of the two items under investigation leads to ungrammaticality as shown below.” (83) a. khaw, terj, kheey, tham, rjaan, nak, he must EXP do work hard ‘It is certain that he has an experience of working hard.’ b. khaw, khorj, ca?, kheey, tham, rjaan, nak, he likely EXP do work hard ‘It is likely that he has an experience of working hard.’ (84) a.*khaw, kheey, ten, tham, rjaan, nak, he EXP must do work hard b.*khaw, kheey, khorj, ca?2 tham, rjaan, nak, he EXP likely do work hard Besides the surface word order, the predicator test in (85) also confirms the suggested ordering above. (85) Q: khaw, khorj, ca?2 kheey, tham, naan, nak, chay, may, he likely EXP do work hard true or not ‘Is it true that he likely has an experience of working hard?’ A: a. (chay,) khorj, ca?2 (true) likely ‘Yes, it is.’ b.*( chay,) kheey, (true) EXP (2) Epistemic modals > Continuative yan, As mentioned earlier, the experiential aspect kheey, always precedes the continuative yan, ‘still’. Therefore, what I have to prove is whether it is the case that the epistemic modals are also situated higher than yan, ‘still’. The following examples seem '6 Sentence (83a) is grammatical when the modal tan, is a root one expressing obligation. The sentence is thus roughly interpreted as ‘He has an experience of having to work hard.’ 226 to prove the point. The ungrammaticality of (87) is the result of the reverse order of the two elements. (86) a. khaw, ?aat2 ca?2 yarj, tham, rjaan, he probably still do work ‘He may still be working.’ b. khaw, naa,ca?2 yarj, tharn, rjaan, he should still do work ‘He should still be working.’ (87) a.*khaw, yarj, ?aat, ca?2 tham, rjaan, he still probably do work b.*khaw, yarj, naa, ca?2 tharn, rjaan, he still should do work Since the continuative yan, can never function as a predicator, the predicator test cannot be used here to confirm or disconfirm the proposed order. However, the fact that the reverse order is never accepted can sufficiently serve as empirical evidence for the claim. (3) Epistemic modals > Progressive kam,lan, The precedence of the epistemic modals over the progressive kam,lan, is evidenced in their surface ordering relative to each other as shown in (88) and (89). (88) a. khaw, ?aat2 ca?, kam,lan, tlrarn, rjaan, he probably PROG do work ‘He may be working.’ b. khaw, naa, ca?2 kam,lan, tham, rjaan, he should PROG do work ‘He should be working.’ (89) a.*khaw, kam,larj, ?aat, ca?,tharn, rjaan, he PROG probably do work b.*khaw, kam,lan, naa, ca?, tham, naan, he PROG should do work 227 Similar to yan,, the progressive kam,lan, cannot be used as a predicator because it is a head without V-features. However, the fact that the minimal response to the question in (90) is simply the epistemic modal without the accompaniment of kam,lan, shows that kam,lan, is not situated higher than the epistemic modal. Otherwise, the modal would have to adjoin to kam,lan, on its way to 2 like V does. (90) Q: khaw, ?aat, ca?, kam,larj, tham, rjaan, chay, may, he probably PROG do work true or-not ‘Is it true that he may be working?’ A: a. (chay,) ?at, ca?, (true) probably ‘Yes, he probably is’ b."‘(chay,) kam,lan, ?aat, ca?, (true) PROG probably (4) Epistemic modals > Irnperfective yuu, Earlier I suggested that the imperfective yuu, is generated as an adjunct to the right of ProgP/ImpP, lower than the experiential perfect kheey, and the continuative yan,. It is therefore likely that yuu, is lower in the structure than the epistemic modals. Again, since yuu, cannot be a predicator, the predicator cannot be directly used here. However, given that kheey, ‘EXP’ is attested to be higher than yuu,, the fact that the epistemic modals are higher than kheey, implies the hierarchical order of the epistemic modals and yuu,. 228 (5) Epistemic modals > lcecew, Like yuu,, Icecew, cannot be a predicator. Since lcecew, may be higher than kheey,, the fact that the sentence containing both an epistemic modal and lcecew, does not require the question particle yan, is evidence for the precedence of the former over the latter. (6) Summary To sum up, the epistemic modals are the highest in the clausal structure. They precede all other modals as well as aspect markers. The study of their co-occurrences does not disprove the suggested ordering of the aspect markers under investigation. 6.5.3.2 Rootmonalmmectmarkers Unlike the monolithic epistemic modals, the root modals have been shown in the previous chapter (Section 5.4.1) to be located in various positions. I will show that the post-verbal modals are closest to VP. They are separated from the other types of root modals by the progressive marker. I will discuss the root modals as a group, where applicable. (1) Experiential kheey, > Root modals A straightforward account of the ordering of root modals relative to the experiential perfect can be made by considering their surface word order as well as the predicator fact. The surface word order as illustrated in (91) indicates that the experiential kheey, is structurally higher than all the root modals. (91) a. khaw, kheey, ca?2 seen, phaa,saa, ciin, he EXP will teach language Chinese ‘He wanted to teach Chinese.’ 229 b. khaw, kheey, tor], seen, phaa,saa, ciin, he EXP must teach language Chinese ‘He had to teach Chinese.’ c. khaw, kheey, day, seen, phaa,saa, ciin, he EXP may teach language Chinese ‘He was allowed to teach Chinese.’ (1. khaw, kheey, seen, phaa,saa, ciin, day, he EXP teach language Chinese can ‘He could teach Chinese.’ When the order of the experiential kheey, and the root modals is reversed, the resulting sentences are unacceptable as shown in (92)”. (92) a.*khaw, ca?2 kheey, seen, phaa,saa, ciin, he will EXP teach language Chinese b.*khaw, terj, kheey, seen, phaa,saa, ciin, he must EXP teach language Chinese c.*khaw, day, kheey, seen, phaa,saa, ciin, he may EXP teach language Chinese d.*khaw, seen, phaa,saa, ciin, day, kheey, he teach language Chinese can EXP Since both the experiential perfect and the root modals can all firnction as predicators, the examples below suggest that the former is situated higher than the latter. (93) Q: khaw, kheey, terj, seen, phaa,saa, ciin, may, he EXP must teach language Chinese QP ‘Did he have to teach Chinese?’ A: a. kheey, EXP. ‘Yes, he did.’ b.*terj, must (94) Q: khaw, kheey, day, soon, phaa,saa, ciin, may, he EXP may teach language Chinese QP ‘Was he allowed to teach Chinese?’ '7 Sentences (92a,b) sound fine if the word ca .7, in (92a) expresses futurity, not volition, and the modal ten, in (92b) is the epistemic one conveying ‘inferred certainty’. 230 A: a. kheey, EXP ‘Yes, he was.’ b.*day, may (95) Q: khaw, kheey, seen, phaa,saa, ciin, day, may, he EXP teach language Chinese can QP ‘Was he able to teach Chinese?’ A: a. kheey, EXP ‘Yes, he was.’ b.*day, can In short, the experiential perfect can be claimed to be below the epistemic modals but above the root modals. (2) Continuative yan, > Root modals The precedence of the continuative yan, ‘still’ over the root modals is justified simply by the word order facts as in (96) and (97).‘8 (96) a. khaw, yarj, ca?2 seen, phaa,saa, ciin, he still will teach language Chinese ‘He still wants to teach Chinese.’ b. khaw, yarj, terj, seen, phaa,saa, ciin, he still must teach language Chinese ‘He still has to teach Chinese.’ c. khaw, yarj, day, soon, phaa,saa, ciin, he still may teach language Chinese ‘He is still allowed to teach Chinese.’ (1. khaw, yarj, seen, phaa,saa, ciin, day, he still teach language Chinese can ‘He still can teach Chinese.’ ‘3 As in the previous footnote, when ca ?, and ten, precede the aspect denoting continuation as in (97a) and (97b), they express futurity and epistemic sense respectively. 231 i (97) a.*khaw, ca?, yarj, seen, phaa,saa, ciin, he will still teach language Chinese b.*khaw, terj, yarj, seen, phaa,saa, ciin, he must still teach language Chinese c.*khaw, day, yarj, seen, phaa,saa, ciin, he may still teach language Chinese d."‘khaw, seen, phaa,saa, ciin, day, yarj, he teach language Chinese can still Once again, the predicator test cannot be used here because yan, can never function as a predicator. (3) Pre-verbal modals > Progressive kam,lan, > Post-verbal modals It happens that the root modals cannot be discussed as a group when considering their positions with respect to the progressive kam,lan,. It is obvious, however, that the modals of volition ca ?, and obligation ten, have scope over the progressive marker and thus are structurally higher as demonstrated in the examples below. (98) a. wee,laa, chan, klap2 maa, thee, ten, kam,lan, ?aan2 narj,sii, when I return come you must PROG read book ‘When I return, you must be studying.’ b. wee,laa, thee, klap2 maa, chan, ca?, kam,lan, ?aan, narj,sii, when you return come I will PROG read book ‘When you return, I will be studying.’ Another piece of supporting evidence comes from the example in (99). That the sentence containing kam ,lan, can be questioned by may, is attributable to the fact that there is another head with V-features above it which can move to 2. Such a head in this case is the modal of obligation ten, Moreover, ten, can move to 2 to be a response to the yes/no question without the accompaniment of kam,lan, , showing that kam,lan, is not situated higher and thus on the way often, to 2. 232 (99) Q: wee,laa, thee, klap, maa, chan, ten, kam,lan, ?aan, narj,sii, may, when you return come I must PROG read book QP ‘Must I be studying when you return?’ A: a. ten, must ‘Yes, you must.’ b.*kam,larj, ?aan, PROG read When it comes to the modals of permission and ability, it seems that they do not normally co-occur with the progressive marker. In English as in (100) and (101), the co- occurrence of both in either ordering yields ungrammaticality. (100) a.*He is being allowed to sing. b.*He is allowed to be singing. (101) a.*He is being able to sing. b.*He is able to be singing. In Thai, likewise, the co-occurrence of the pre-verbal modal of permission day, and the progressive marker is impossible as in (102)”. Nevertheless, it happens that the post-verbal modals of ability can occur with the progressive marker only when the latter has scope over the former as in (103). (102) a.*khaw, kam,lan, day, reel], pleerj, he PROG may sing song ”He is being allowed to sing.’ b.*khaw, day, kam,lan, reen, pleerj, he may PROG sing song ‘*He is allowed to be singing.’ (103) a. yaa, kuan, khaw, khaw, kam,lan, kin, day, not bother he he PROG eat can ‘Don’t bother him. He can eat at the very moment.’ '9 If the word day, in (102a) is considered as a verb meaning ‘have an opportunity to do something’, then the sentence sounds better. 233 b.*khaw, kin, day, kam,lan, he cat can PROG Sentence (103a) can be uttered in a situation where a person was very sick and could not eat for several days. Then he felt better and could eat at the very moment. If someone wanted to visit him, the nurse could say this sentence for fear of the interruption of his ability to eat. Supporting evidence for the higher position of the progressive aspect than the post-verbal modals comes from the ungrammaticality of (104). (104)*khaw, kam,lan, kin, day, may, he PROG eat can QP The fact that the question particle may, cannot be used here indicates that the movement to 2 of the modal of ability day, which is situated higher than VP is blocked by the progressive kam,lan which is the head between ModP,,,,,,,y and 2P. In brief, different types of root modals are not ordered in hierarchically adjacent position. The post-verbal modals are separated fi'om the rest by the progressive marker. (4) Pre-verbal modals > Irnperfective yuu, > Post-verbal modals Similar to the progressive kam ,lan,, the pre-verbal root modals take scope over the imperfective yuu, whereas the post-verbal ones do not, as illustrated in (105) and (106) respectively. (105) a. khaw, ca?2 tham, rjaan, yuu, con, naay, maa, truat, he will do work IMPF until boss come supervise ‘He will keep working until his boss comes to supervise his work.’ b. khaw, terj, pen, khruu, yuu2 saam, pii, he must be teacher IMPF three year ‘He must be a teacher for three years.’ 234 (106) khaw, kin, day, yuu2 he cat can IMPF ‘He is able to eat.’ As discussed earlier (section 6.4.2.2), the sentence in (106) is better interpreted as ‘His ability to eat is in process,’ which more clearly shows the scope relation between the two items. This is consistent with my suggestion that yuu, is right-adjoined to ProgP/ImpP headed by kam ,lan,. (5) Pre-verbal modals > lcecew, > Post-verbal modals The pre-verbal root modals also precede lcecew, ‘already’2°. Once again, the scope relation can indicate their positions with respect to each other as in (107). (107) a. teen, chan, klap2 maa, khaw, terj, pay, lamw, when I return come he must go already ‘When I return, he must have already left.’ b. teen, thee, klap2 maa, chan, ca?2 tham, kaan,baan,, lzeaew, when you return come I will do homework already ‘When you return, I will have already done my homework.’ When lcecew, combines with the post-verbal modals, it only marks the beginning of the state or the ability as illustrated in the interpretations of sentences (108a, b) below. (108) a. dek, deen, day, lzeaw, kid walk can already ‘The infant has started to be able to walk. (It can walk now.)’ b. khaw, waay,nam, day, laeaew, he swim can already ‘He has started to be able to swim. (He can swim now).’ 2° I cannot find examples that contain both pre-verbal root modals and law, that marks the beginning of an event or state. However, as lwcew, is assumed to be able to attach to various positions high in the structure, it is likely that the root modals are lower than the marker of the beginning which is suggested to be adjoined to the experiential perfect which is higher than the modals. 235 Recall the analysis of the question particle may,. The fact that the sentence in (109) takes the question particle yan,, instead of may,, indicates that lcecew, is higher than the modal of ability. In other words, if the modal of ability were situated higher than Icecew,, the question particle may, would be used here, given that the choice of question particles has a lot to do with the element of the widest scope. (109) dek2 deen, day, laeaew, yan, /*may, kid walk can already yet ‘Has the infant been able to walk?’ (6) Summary Root modals are not like epistemic modals in that they are not monolithic. It happens that the post-verbal modals are situated closest to VP, lower than any aspect marker. This is not surprising, in fact, given that this type of words triggers VP movement. It does not attract any projections other than VP. Therefore, its closeness to VP is not puzzling. The exploration of their relative positions also contributes to the affumation of the analyses made earlier. 6.6 Conclusion I have shown in this chapter the syntactic categories as well as the structural positions of aspect markers in Thai. The two aspect markers which are considered as heads are the experiential kheey, and the progressive kam,lan,. The latter, on the other hand, is claimed to lack V-features, which prevents it from being a predicator, being negated by may, and being questioned by may,. The continuative yan, is a phrase in the specifier of an AspP. Both post-verbal aspects are right-adjuncts; the imperfective yuu, is 236 right-adjoined to ProgP/ImpP whereas lcecew, ‘already’ which is analyzed as a temporal boundary marker can be adjoined to several positions high in the structure. I also investigate the co-occurrences of modals and aspect markers in Thai. Their ordering with respect to one another is summarized as below where lcecew, is left out due to its variable position. (110) epistemic modals > experiential > still > pre-verbal root modals > imperfective/progressive > post-verbal root modals To conclude my study on modals and aspect markers, what I have found is the following. (1) Auxiliaries in Thai are head-initial. (2) VP movement occurs in the syntax of clauses in Thai. (3) The clausal structure in Thai allows phrases as right-adjuncts (the imperfective yuu, and the temporal boundary marker lazcew,) and as specifiers of XPs (the modal of volition ca ?, and the continuative yan,). 237 CHAPTER 7 CONCLUSION Having considered both noun phrases and clauses in Thai, I am now in a position to answer the research questions raised in the Introduction. All in all, I have found that noun phrases and clauses in Thai are parallel in all the respects mentioned, namely in their head-directionality, in the syntactic operations that derive the apparent head-final orders, and in the allowance of right-adjunction. Regarding head-directionality, I have shown that noun phrases in Thai are head- initial like all other structures in the language. In spite of the noun being the first element of the noun phrase, the hypothesis that Thai noun phrases are head-final is to be rejected because the head-final analysis requires more syntactic operations than the head-initial (not to mention its inconsistency) and it fails to account for some of the data. With respect to syntactic operations, I have argued that phrasal movements operate in the syntax of both noun phrases and clauses in Thai. In other words, NP raising can capture the distribution, interpretation and word order facts of Thai noun phrases. Similarly, VP raising takes place when a post-verbal modal is present. The analysis of the post-verbal modals as heads generated pre-verbally with VP movement supports the general claim that auxiliaries in an SVO language precede main verbs. Concerning the phrasal categories, noun phrases and clauses allow both right adjuncts and (left) specifiers. Noun modifiers are either NP right-adjuncts (regular attributive adjectives, nominal modifiers) or specifiers of ClP (referential adjectives, demonstratives). Likewise, the post-verbal aspect markers are right-adjoined to different 238 functional projections above VP whereas a modal (modal of volition) and an aspect marker (the continuative marker) are generated in the specifier of ModP and vohuon Asmemmw respectively. To conclude, I have shown that Thai is a well-behaved language although many of the structures in the language do not appear so. I have also shown that Thai is a uniformly head-initial language. In addition, since no study has ever considered the post-verbal modals as pre-verbal auxiliaries, my proposal that they are heads generated pro-verbally makes Thai clausal structures uniform with respect to the headedness of the auxiliaries. Of course, there is much still to be done. I have not dealt with other types of noun modifiers such as possessives and relative clauses, and there is still much to be learned about aspect markers, especially the post-verbal ones which seem to display a variety of meanings. A number of residual problems have also been left unexplained and wait for firrther study. However, I hope my study can be the foundation for further research on the syntax of Thai. 239 REFERENCES Abney, S. (1987). The English noun phrase in its sentential aspect. Ph.D. thesis, MIT. Alexiadou, A. (1997). Adverb placement: A case study in antisymmetric syntax. John Benj amins Publishing Company, Amsterdam, Philadelphia. Bandhumedha, N. (1983). woyyookoenthoy [Thai grammar]. Rungruangsaan, Bangkok. Belletti, A. (1990). Generalized verb movement. Rosenberg & Sellier, Torino. Benmamoun, E. (1991). Negation and V movement. Proceedings of NELS, 21(1): 1 7-3 1. Bernstein, J. (1993). Topics in the syntax of nominal structure across Romance. Ph.D. dissertation. CUNY, New York. Bernstein, J. (1996). Demonstratives and reinforcers in Romance and Germanic languages. Ms., University of Southern Maine. Bisang, W. (1993). Classifiers, quantifiers and class nouns in Hmong. Studies in Language, 17(1):1-51. Boonyapatipark, T. (1983). A study of aspect in Thai. Ph.D. dissertation, University of London. Brown, K. (1992). Double modals in Hawick Scots. In Trudgill P. and Chambers J.K., eds., Dialects of English: Studies in Grammatical Variation, 74-103, Longman, London. Bybee, J ., Perkins, R., and Pagliuca, W. (1994). The evolution of Grammar: Tense, Aspect and Modality in the languages of the world. University of Chicago Press, Chicago. Carlson, G. (1977). Reference to kinds in English. Ph.D. dissertation. University of Massachusetts, Amherst. Cheng, L., Huang, J.-C.T. and Tang, J. (1996). Negative particle questions: A dialectal comparison. Amsterdam studies in the theory and history of linguistic science, Series IV, Current Issues in Linguistic Theory, 139:41-78. Cheng, L. and Sybesma, R. (1999). Bare and not-so-bare nouns and the structure of NP. Linguistic Inquiry, 30(4):509-542 . Chierchia, G. (1998). Reference to kinds across languages. Ms., University of Milan. 240 Chomsky, N. (1986a). Barriers. MIT Press, Cambridge. Chomsky, N. (1986b). Knowledge of language: Its nature, origin and use. Praeger, New York. Chomsky, N. (1991). Some notes on economy of derivation and representation. In Freidin, R., ed., Principles and Parameters in Comparative Grammar. MIT Press, Cambridge. Chomsky, N. (1993). A minimalist program for linguistic theory. In Kenneth Hale and Samuel Jay Keyser, eds., The view fi'om buiding 20: Essays in Linguistics in Honor of Sylvoin Bromberger, 1-52, MIT Press, Cambridge. Chomsky, N. (1995). The minimalist program. MIT Press, Cambridge. Cinque, G. (1993). On the evidence for partial N-movement in the Romance DP. Ms., Universita di Venezia. Cinque, G. (1995). Adverbs and the universal hierarchy of fimctional projections. GLOW Newsletter, 34:14-15. Cinque, G. (1999). Adverbs ondfunctionol heads: A cross~linguistic perspective. Oxford University Press, New York. Comrie, B. (1976). Aspect. Cambridge University Press, Cambridge. Croft, W. (1994). Semantic universals in classifier systems. Ward, 45(2): 145-171. Delsing, L. (1988). The Scandinavian noun phrase. Working Papers in Scandinavian Syntax, 42:57-79. Delsing, L. (1993). The internal structure of noun phrases in the Scandinavian languages. Ph.D. dissertation, University of Lund. Demirdache, H. and Uribe-Etxebarria, M. (1997). The syntactic primitives of temporal relations. Talk presented at Langues et Grammaires 111, Paris. Ekniyom, P. (1977).The topic/comment distinction and passivization in Thai. Working Paper in Linguistics University of Hawaii, 11:55-65. Ekniyom, P. (1979). An internal reconstruction of auxiliaries in Thai. Working Paper in Linguistics University of Hawaii, 9:93-110. Ernst, T. (1991). On the Scope Principle. Linguistic Inquiry, 22(4):750-755. 241 Ernst, T. (1992). The phrase structure of English negation. The Linguistic Review, 9(1):109-144. Ernst, T. (1995). Negation in Mandarin Chinese. NLLT, 13:665-707. Ernst, T. (1997). The scopal basis of adverb licensing. NELS, 28:1-15. Gao, Q. (1994). Chinese NP structure. Linguistics, 32:475-510. Goral, D. (1978). Numeral classifier systems: A Southeast Asian cross-linguistic analysis. Linguistics of the T ibeto-Burmon Area, 4(1):1-72. Grirnshaw, J. (1991 ). Extended projections. Ms., Brandeis University. Higginbotham, J. (1985). On semantics. Linguistic Inquiry, 16:547-593. Huang, C.-T.J. (1982). Logical relations in Chinese and the theory of grammar. Ph.D. dissertation, MIT. Huang, C.-T.J. (1988). Wo pao de kuai and Chinese phrase structure. Language, 64(2):274-31 1. Hudak, T. (1990). Thai. In Comrie, B., ed., The major languages of East and South-east Asia, 29-47, Routledge, London. Hundius, H. and Kolver, U. (1983). Syntax and semantics of numeral classifiers in Thai. Studies in Language, 7(2): 165-214. Indrambarya, K. (1995). Subcotegorizotion of verbs in Thai: A lexicase dependency approach. Ph.D. dissertation. University of Hawaii. J ackendoff, R. (1972). Semantic interpretation in generative grammar. MIT Press, Cambridge, Massachusetts. Kanchanawan, N. (1978). Expressions for time in the Thai verb and its application to Thai-English machine translation. Ph.D. dissertation, University of Texas. Kester, E. (1996). The nature of odjectivol inflection. Ph.D. thesis, Utrecht University. Laka, I. (1990). Negation in the syntax. Ph.D. dissertation, MIT. Lobeck, A. (1990). Functional heads as proper governors. Proceedings of NELS, 20:348- 362. 242 Lobeck, A. (1995). Ellipsis: F unctionol heads, licensing and identification. Oxford University Press, Oxford. Lobner, S. (1989). German schon-erst-noch: An integrated analysis. Linguistics and Philosophy, 12:167-212. Longobardi, G. (1994). Reference and proper names: A theory of N-movement in syntax and logical form. Linguistic Inquiry, 25(4):609-665. Lopez, L. (1999). VP-ellipsis in English and Spanish and the features of auxiliaries. Probus, 11(2):263-297. Martins, A. (1994). Enclisis, VP-deletion and the nature of Sigma. Probus, 6:17 3-206. Meepoe, T. (1996). Imperfectivity in Thai: The distinction between komlan and yuu. Masters thesis, University of California, Los Angeles. Miyagawa, S. (1988). Predication and numeral quantifier. In Poser, W.J., ed., Papers fiam the second international workshop on Japanese syntax, 157-192, Center for the Study of Language and Information, Stanford. Munn, A. (1992). A null operator analysis of ATB gaps. The Linguistic Review, 9: 1-26. Munn, A. (1993). Topics in the syntax and semantics of coordinate structures. Ph.D. dissertation, University of Maryland. Needleman, R. (1973). T oi verbal structures and some implications for current Linguistic theory. Ph.D. dissertation, University of California at Los Angeles. Noss, R. (1964). Thai reference grammar. Foreign Service Institute, Washington, DC. Ouhalla, J. (1990). Sentential negation, relativised mirrimality and the aspectual status of auxiliaries. The Linguistic Review, 7: 1 83—23 1. Pacioni, P. (1996). Classifiers, specificity and typology in Asian languages. Pan-Asiatic Linguistics Proceedings of the Fourth International Symposium on Languages and Linguistics, 4:1987-1997. Palmer, F. (1986). Mood and modality. Cambridge University Press, Cambridge. Panupong, V. (1970). Inter-sentence relations in modern conversation Thai. Siam Society, Bangkok. Pestesky, D. (1998). Phrasal movement and its kin. Ms. MIT. 243 Pollock, J. (1989). Verb movement, universal grammar and the structure of IP. Linguistic Inquiry, 20:365-424. Postal, P. (1969). On so-called ‘pronouns’ in English. In Reibel, D. and Schane, S., eds., Modern studies in English, 201-224, Prentice-Hall, Englewod Cliffs, NJ. Savetamalya, S. (1987). A reanalysis of auxiliaries in Thai. Working Papers in Linguistics University of Hawaii, 19:1-43. Savetamalya, S. (1989). Thai nouns and noun phrases: A lexicase analysis. Ph.D. dissertation, University of Hawaii. Savetamalya, S. and Reid, L. (1997). An explanation for inconsistent word order typologies in some Southeast Asian languages. In Abramson, A., ed., Southeast Asian Linguistic Studies in Honor of Vichin Panupong, 221-236, Chulalongkom University Press, Bangkok. Scovel, T. (1970). A grammar of time in Thai. Ph.D. dissertation, University of Michigan. Sereechareonsatit, T. (1984). Conjunct verbs and verbs-in-series in Thai. Ph.D. dissertation, University of Illinois, Urbana-Champaigrr. Shaer, B. (1998). Adverbials, functional structure, and restrictiveness. NELS, 28:1-15. Sindhvananda, K. (1970). The verb in modern Thai. Ph.D. dissertation, Georgetown University. Smith, C. (1997). The parameter of aspects. Kluwer Academic Publishers, Dordrecht, The Netherlands. Sookgasem, P. (1990). Morphology, syntax and semantics of auxiliaries in Thai. Ph.D. dissertation, University of Arizona Sriphen, S. (1982). The Thai verb phrase. Ph.D. dissertation, University of Michigan. Sproat, R. & Shih, C. (1988). Prenominal adjectival ordering in English and Mandarin. NELS, 18:465-489. Starosta, S. (1988). The case for lexicase. Pinter Publishers, New York. Szabolcsi, A. (1989). Noun phrases and clauses: Is DP analogous to IP or CF? Paper presented at the Colloquium on Noun Phrase Structure, University of Manchester. Tang, C. (1990). A note on the DP analysis of the Chinese noun phrase. Linguistics, 28: 337- 354. 244 Thepkanjana, K. (1986). Serial verb construction in Thai. Ph.D. dissertation, University of Michigan. Travis, L. (1984). Parameters and eflects of word order variation. Ph.D. thesis, MIT. Uppakitsinlapasarn, Phya. (1964). Woyyookonthoy: woociiwiphook (Thai grammar: Morphology). Thai Wattana Pharrit, Bangkok. van der Auwera, J. (1993). ‘Already’ and ‘Still’: Beyond duality. Linguistics and Philosophy, 16:613-653. van der Auwera, J. (1998). Phasal adverbials in the languages of Europe. In van der Auwera, J. and Baoill , D. P., eds., Adverbiol constructions in the languages of Europe, 25-145, Mouton de Gruyter, Berlin. Warotamasikkhadit, U. (1972). Thai syntax: An outline. Mouton, The Hague. Warotamasikkhadit, U. (1996). Thai reference grammar. Rarnkamhaeng University Press, Bangkok. Wilawan, S. (1994). A reanalysis of so-colled serial verb constructions in Thai, Khmer, Mandarin Chinese, and Yoruba. Ph.D. dissertation. University of Hawaii. Zanuttini, R. (1997). Negation and clausal structure. A comparative study of Romance languages. Oxford University Press, New York. Zhou, X. (1998). Mandarin negations as negative aspectual elements. Ms. State University of New York at Stony Brook. 245 "Illlllllllllllllllll