<<

ONTHEOFCHINESENOMINALS

by

HUIMINJI

(UndertheDirectionofMarlyseBaptista)

ABSTRACT

This dissertation investigates the use of classifiers in Chinese and more specifically comparesMandarinandtheWudialectintermsofusage.Differentsyntacticstructures areproposedforthesurfaceform[Cl+N]toderiveitsdefiniteandindefiniteinterpretations.

Classifiersarearguedtoencodeonlytheofcountability.TheclassifierinWumovestoD whenthe[Cl+N]formhasadefiniteinterpretation. Xie ‘some’, usually considered to be a markerorapluralclassifierinthelinguisticliterature,isarguedtobeaquantifierwhich canbeusedwithbothcountandmass.

Thisdissertationalsoanalyzesthesocalledcomplexdefinites.areassumedtobe in Mandarin. function like and do not encode the featureinMandarin.Multiplepossessivesrealizingdifferentthetarolescanoccur prenominallyandtheirorderisfixed.isassumedthatdifferentpossessivesoccupythe positions of different functional projections. Proper names in Mandarin are argued to be in varioussyntacticpositions.determinersareassumedtoprojecttheirownphrasal categoryinthisdissertation.

INDEX: , classifier phrase, definiteness, specificity, complex definites

ONTHESYNTAXOFCHINESENOMINALS

by

HUIMINJI

B.A. ZhejiangNormalUniversity,,1986

M.A. ShanghaiInternationalStudiesUniversity,China,1995

ADissertationSubmittedtotheGraduateFacultyofTheUniversityofGeorgiainPartial

FulfillmentoftheRequirementsfortheDegree

DOCTOROFPHILOSOPHY

ATHENS,GEORGIA

2007

©2007

HuiminJi

AllRightsReserved

ONTHESYNTAXOFCHINESENOMINALS

by

HUIMINJI

MajorProfessor: MarlyseBaptista Committee: KeithLangston MichaelA.Covington ElectronicVersionApproved: MaureenGrasso DeanoftheGraduateSchool TheUniversityofGeorgia May2007 iv

ACKNOWLEDGEMENTS

Finally,thisdissertationiscomplete.Itistimeformetoexpressmydeepestgratitudeto themanypeoplefromwhomhavereceivedhelpinthewritingprocessandhavehelped shapemeintothepersonthatIamtoday.

Firstofall,Iamdeeplyindebtedtomycommitteemembers:Dr.MarlyseBaptista,Dr.

Keith Langston and Dr. Michael Covington. Dr. Baptista read the drafts of this dissertation meticulouslyandsupervisedmywritingwithprofessionalguidanceandinsightfulcriticism.I wouldliketoexpressmysincerethankstoDr.Baptistaforherpatiencewithmyprocrastination andherconfidenceinme.Withoutherconstantnudgingandencouragement,Iwouldhavelong givenupwritingthisdissertation.IwouldliketothankDr.LangstonandDr.Covingtonfor servingonmycommitteeandfortheirconstructivecommentsonthisdissertation.Ithasalways beenapleasantandenlighteningexperiencetotalkwithDr.Langston.Also,Iwouldliketo thankDr.BaptistaandDr.Langstonfortheirsyntaxandcourses,fromwhich I havebenefitedmost,ofallthecoursesIhavetakenatUGA.

I am grateful to the Linguistics Program and late director Dr. William Provost for admittingme.IcanstillrecallmyexcitementwhentoldmethatIhadbeenadmittedtothe programandrecommendedforGraduateSchoolassistantship. MythanksalsogotoDr.Don

McCreary,Dr.MiRanKimandMsLynnetteLangforalltheirhelpduringmystudyatUGA.

v

IwouldliketothankmyformerteacherProfessorTiexiongWangatZhejiangNormal

University,China,whointroducedmetoEnglishlinguisticsovertwentyyearsago.Atthattime, linguisticscourseswerenotofferedattheuniversitythatIattended;ProfessorWangtutoredme outofclassandlentmebookswhenstudentsalmosthadno access tolinguisticsworks writteninEnglish.Duringthewritingofthisdissertation,hesentmeabookandseveraljournal articlesfromChina.ProfessorWangisafatherfiguretomeandtreatsmelikehisownchild.He onceadvisedmetowashhandsbeforeeatingafterIgraduatedandwasassignedtoworkina medicalcollege.

Duringthewritingofthisdissertation,manyfriends,formerclassmatesandcolleagues helpedme.Ithankthefollowingpeopleforsharingtheirintuitionsabout(un)grammaticalityof

Chinese nominals and/or sending me journal articles: Fengming Cai, Mingsheng Cao, Jiaoe

Chen,JiayunHan,LichaoHe,ZhonghengJia,ShupengLi,XinghuanLuo,ChongPan,Yugang

Shen,DongWang,YaowuWang,YueWangandYanranZhu.

Last,butnotleast,Ithankmyfamily.Iowespecialthankstomyparentsinlawfor encouragingmetopursuefurtherstudyintheU..Ithankmyparentsforinstillinginmethe importanceofeducationandsendingmetocollege,whenmostchildrenofmyagethenworked inthefield.Myparentshavealwaysbeenmysupport,whetherIwasasmallvillagekidat homeorastudentfarawayfromhome.MywifeCongzhouanddaughterYinyuhavealways beenextremelyandsupportive.Withoutthem,thisdissertationwouldhaveneverbeen completed.

vi

TABLEOFCONTENTS

Page

ACKNOWLEDGEMENTS…………………………………………………………………...iv

LISTOFTABLES…………………………………………………………………………….viii

LISTOF………………………………………………………………….ix

CHAPTER

1 INTRODUCTION………………………………………………………………………1

1.1 DefinitenessandSpecificity…...... 3

1.2 TheOrganizationoftheDissertation……………………...... 10

2 LITERATUREREVIEW……………………………………..…………….....……...... 13

2.1 TheoreticalMotivationsfortheDPHypothesis……………………………...... 13

2.2 EmpiricalSupportfortheDPHypothesis……..…………………………………16

2.3 BareNouns………………………………………………………………….…....26

2.4 ResearchonChineseNominals………………………………………………….33

3 MANDARINCHINESE:ANOVERVIEW…………..…………………………….…43

3.1 LanguageUseinChina…………………………………………………………..43

3.2 SomeGrammaticalFeaturesofMandarinNominals…………………………….47

4 THECLASSIFIERSYSTEMINCHINESE…………..………………………………60

4.1 TheDevelopmentoftheChineseClassifierSystem……………………………..62

vii

4.2 ClassificationofChineseClassifiers…………………………………..…………80

4.3 FunctionsofClassifiers…………………………………………………………..83

4.4 NumericallyQuantifiedNominals……………………………………………….88

5 OFCLASSIFIERUSESINWUANDMANDARIN…….………...105

5.1 UsesofClassifiersinWu………………………….…………………………….105

5.2 DistributionsandInterpretationsof[Cl+N]inWuandMandarin….…………...111

5.3 SyntacticStructuresof[Cl+N]inWuandMandarin…………...……………….115

6 COMPLEXNOMINALSINMANDARIN…..…………..…………………………..129

6.1 ConstructionsinMandarin……..………..………………………….130

6.2 PersonalPronounsandProperNames…………………………………………..148

6.3 DemonstrativeDeterminers……………………………………………………..165

7 CONCLUSION…………………………….…………..……………………………..180

BIBLIOGRAPHY……………………………………………………………………………..184

viii

LISTOFTABLES

Page

Table3.1PersonalPronounsinMandarin……………………………………………………49

Table3.2PersonalPronounsintheWenlingDialect……….…………………………………54

Table4.1DifferencesbetweenClassifiersandMeasureWords………………………………70

Table6.1MeaningsandFunctionsofinMandarin……………….………..174

ix

LISTOFABBREVIATIONS

C

Cl classifier

ClP classifierphrase

Col collectivemarkermen inMandarin

Comp complementizer

CP complementizerphrase

D determiner

DE modificationmarker de inMandarin

Dem demonstrative

DemP demonstrativephrase

DP determinerphrase

Exp experientialaspectmarker

GE modificationmarker ge inWu,equivalentto de inMandarin.

I

IP inflectionalphrase

MW measure

N

x

NP nounphrase

Num number

Nume

NumP numberphrase

NumeP numeralphrase

Poss possessivepronoun

Perf perfectiveaspectmarker

Prog progressiveaspectmarker

Quan quantifier

Ques questionmarker

SFP finalparticle

TP tensephrase

V

VP verbphrase

1

CHAPTER ONE

INTRODUCTION

ThisdissertationexaminesthesyntacticstructureofChinesenominalexpressions,with particular attention paid to the functional category of Cl(assifier). It argues against Cheng &

Sybesma’s(1999,2005)viewthattheclassifierinChinesefunctionslikethedefinitein somewesternlanguagesandthatClPismoreorlessequivalenttoDP.

TheclassifierplaysanimportantroleinChinesenominalexpressions.Everydialectof theChineselanguageusesclassifiers,yettheuseofclassifiersvariesamongdifferentdialects.

Notonlydodifferentdialectsusedifferentclassifiersincountingthesamedescribedbya noun, but they have different grammatical functions. This dissertation compares the uses of classifiersinMandarinwiththoseintheWudialect.TheclassifierinbothMandarinandtheWu dialecthasclassifyingandindividualizingfunctions.However,classifiersinWucanalsohave the referentialization function; a nominal like ben shu (literally ‘Cl book’) in Wu, not in

Mandarin,canmean‘thisbook’.ThisdissertationassumesthattheclassifierinMandarinand

WuisbasegeneratedinClandthatonlyinWutheclassifiercanmovetoD.Whenithasan indefinite interpretation, the surface form [Cl + N] in Mandarin and Wu is argued to be a

Nume(ral)Pwiththenumeral yi ‘one’deleted,Thisexplainswhytheindefinite[Cl+N]phrase can only occur in a lexicallygoverned position in both Mandarin and Wu. In addition to an

2 indefiniteinterpretation,the[Cl+N]phraseinWucanhaveadefiniteinterpretation,inwhich casetheclassifierisarguedtomovetoD.

Anothertopicthatisdealtwithinthisdissertation is complex nominalexpressions in

Mandarin. By complex nominal expressions, I mean here those in which personal pronouns, propernames,possessivepronouns,ordemonstrativesoccurprenominally.Theseexpressionsare calledcomplexdefinitesinLyons(1999).Someoftheseelements(likepossessivepronounsand demonstratives)cancooccurinMandarin.

Assuming with Kayne (1994) that the underlying phrase structure for all languages is identical,followingLi(1998,1999)andTang(1990)andcontraCheng&Sybesma(1999,2005),

IholdthatChinesenominalexpressionshaveaDPprojectionalthoughChinesedoesnothavea dedicated article to signal the (in)definiteness of the nominal. This dissertation proposes that personal pronouns and proper names in Mandarin can be basegenerated in D, while the demonstrativeoccupiesalowerlevelpositionwithinDP.Ianalyzecomplexnominalexpressions suchas(1)withinX'theory.

(1) Zhang San (ta-men) zhe san ge xuesheng ZhangSan (heCol) this three Cl student ‘ZhangSanandtheothertwostudents’

In the following, I define and discuss the two important terms “definiteness” and

“specificity”andthenoutlinetheorganizationofthisdissertation.

3

1.1 Definiteness and Specificity

Definitenessandspecificityaretwopivotalconceptsinthediscussionofthesyntaxand of nominal expressions. In this section, I describe some of the main assumptions underlyingthesetwoconceptsintheliteratureandputforwardaworkingdefinitionofthetwo termsforthisdissertation.

Asasemanticopragmaticconcept,definitenessisuniversal(Lyons,1999).The ofsomenominalsinanylanguageareidentifiableorknowntothehearer,nomatterwhetherthis language has any grammatical device to indicate the (in)definiteness of the nominal. Proper nounsandpersonalpronounsareinherentlydefiniteandeverylanguagealsohasdemonstratives which indicate that the noun phrase in which they occur is definite. Still, syntactically, definitenesscanbeencodeddifferentlyindifferentlanguagesanditmaynotbeencodedatallin some languages. In a language like English, definiteness and indefiniteness are signaled by articlesandotherdeterminers.Itiseasytodeterminewhetheranominalexpressionisdefiniteor indefiniteinEnglish,asdefinitenessisalwaysencodedbydefinitedeterminers.Iwillelaborate onitlateron.InalanguagewithoutarticleslikeChinese,(in)definitenessofthenominalusually isnotencodedsyntacticallyandthe(in)definiteinterpretationofthenominalexpressiondepends oncontextualfactorssuchassharedbackgroundknowledgeorpriordiscourse.Ifwordorderis consideredasyntacticdevice,thendefinitenessisencodedsyntactically toacertainextent,as preverbalbarenounsalwayshaveadefiniteinterpretationinChinese.

4

Although we know that the definite article encodes definiteness in English, the term

‘definiteness’ is difficult to define. Lyons (1999) discusses several perspectives from which definitenessisapproachedinthelinguisticliterature.1

Thefirstperspectiveisthefamiliarityhypothesis.Iftheentitydenotedbythenominalis familiartoboththespeakerandthelistener,thenthenominalisdefinite.Thedefinitearticle the inEnglishsignalssuchfamiliarityand a thelackofsuchfamiliarity.Butincertaincontexts,the definite article is used, but the listener is not familiar with the of the nominal. For example,Asays(2)toBwhenBjustenterstheroom:

(2)“Passmethehammer,please.”

SpeakerBmaynotknowthatthereisahammerintheroom,but the isstillusedinthenominal.

Here,thefamiliarityhypothesiscannotexplaintheappropriateuseofthedefinitearticle.

Anotherperspectiveistheassumptionofidentifiabilityofthereferent.Accordingtothis approach,thedefinitearticlesignalsthatthespeakerassumesthatthelistenerisabletoidentify the referent. The identifiability notion can explain the use of the definite article in (2). Even thoughthelistenermaynotknowthatthereisahammerintheroom,hewouldlookaroundand identifythehammerafterhearing(2).Stillincertaincontexts,neitherthefamiliarityhypothesis northeidentifiablityperspectivecanexplaintheuseofthedefinitearticle. Aspeakermayuse thedefinitearticleevenwhenthelistenerisunabletoidentifythereferent.Forexample,one

1 TheexamplesentencesinthissectionarefromLyons(1999,ChapterOne).

5 mightsay(3),evenwhenthehearermaybeneitherfamiliarwithnorabletoidentifythebride mentionedin(3).

(3)I’vejustbeentoawedding.Thebrideworeblue.

ThethirdperspectivethatLyons(1999)discussesisuniqueness.Wecanusethedefinite articleaslongasthereisauniquereferentthatfitsthedescriptionandthedefinitearticlesignals thatthereisonlyonereferentsatisfyingthedescriptionofthenominal.Thesuccessfuluseofthe definitearticlein(3)canbeexplainedbytheideaofuniqueness.Inawedding,theremustbea brideandonlyonebride.However,thedefinitearticlecanalsobeusedwithamassnounora pluralcountnoun.Thereferentisnotuniquewhenthedefinitearticleisusedwithmassorplural countnouns.

Thefourthperspectiveisinclusiveness.Thismeans the reference of definite nominals containingpluralormassnounsinthecontextistothetotalityoftheobjects/peopleorthemass satisfyingthedescriptionofthereferent(s)ofthenominal.Considerthefollowingexample:

(anurseabouttoentertheoperatingtheatre) (4)Iwonderwho theanesthetistsaretoday.

Thedefinitenounphrase the anesthetists in(4)referstoalltheanesthetistsabouttotake partintheoperation.Uniquenesscanbeviewedas a special case of inclusiveness when the totalityofobjects/peoplesatisfyingthedescriptionisonlyone.

6

Fromtheabove,wecanseethateachofthefourapproachescanexplainsomeoftheuses of the ,butnotnecessarilyallofitsuses. Thefournotionsofdefinitenessdiscussedaboveare basicallysemanticopragmatic.Aslongasthereferenceofanominalinvolveseitherfamiliarity, oridentifiability,oruniquenessorinclusiveness,thenominalcanbeconsideredtobedefinite.

The four notions are applicable to the Chinese nominal. The difference between (in)definite nominalsinEnglishandthoseinChineseisthatthere is no obligatory grammatical means to signalfamiliarity,identifiability,oruniqueness/inclusivenessinChinese.

Inthisdissertation,Iuse‘definite’tomeanthatthelistenerknowswhoorwhichthing(s) adefinitenominalexpressionrefersto.Thismeansthatthelistenerisabletoidentifythereferent of the nominal in discourse, thanks to either the context of discourse or shared background knowledge. Herethe‘know’and‘identify’donotmeanthatthelistenerliterallyknows orisabletoidentifytheperson(s)orthing(s)inreallifethatadefinitenominalrefersto.Itonly meansthatthelistenerhasamentalrepresentationofthereferentofthedefinitenominal,based onthelinguisticcontext,situationalcontextandbackgroundknowledge.

Lyons (1999) classifies definites into simple and complex ones. Simple definites are when definiteness is marked by grammatical or functional (like the in English).

Complex definites are proper names, personal pronouns, noun with possessives or demonstratives.2 As we mentioned above, definiteness is language universal as a semantico pragmatic concept. As a grammatical category, whether definiteness is language universal

2 Insomelanguages,possessivesanddemonstrativesdonotnecessarilyhavethedefinitenessfeature,astheycan cooccurwiththedefinitearticleinanominal.

7 depends upon how definiteness is defined. Lyons argues that it is only when definiteness is marked by grammatical morphemes in a language that the language has definiteness as a grammaticalcategory.Thisisthenarrowsenseofdefiniteness. InLyons’view,languageslike

Chinese, which do not have a grammatical marker of definiteness, lack definiteness as a grammatical category. In this dissertation, I adopt the broad sense of definiteness, that is, definiteness as also encoded by proper names, personal pronouns and demonstratives. As a matteroffact,somescholarsregardthepersonalpronounasanarticleordefinitedeterminer

(Postal,1966;Abney,1987).Letusconsiderthefollowingexamples:

(5)a.Youguysdidagoodjob. b.Theguysdidagoodjob. (6)a.Wericharebecomingricher. b.Thericharebecomingricher.

Ifwecompare(5a)and(6a)with(5b)and(6b),weimmediatelynoticethatthepronounsoccupy the same syntactic position as the definite article. In Mandarin, plural personal pronouns can precedebarenounslike you in(5a),asshownin(7).

(7) ni-men xuesheng yao renzhen xuexi . youCol students should diligently study ‘Youstudentsshouldstudydiligently.’

8

Inaddition,personalpronounsinMandarincanprecedethe[Nume+Cl+N]sequence andthe[Dem+Nume+Cl+N]sequence.Wewillcometotheuseofpersonalpronounsas determinersinMandarininChapterSix.

Closely related to definiteness is specificity. According to Frawley (1992: 80),

“specificityreferstotheuniqueness,individuation,orreferentialaccessibilityofanentityina mentally projected world.” While definiteness is encoded by definite determiners in English, specificity is not marked in English. A nominal in English is not ambiguous with regard to definiteness;itcanbeeitherdefiniteorindefinite,butcannotbebothinEnglish.Thisisnotthe casewithspecificity;anominalcanbebothspecificandnonspecific.Forexample,thenominala man who speaks French in(8)canhavetworeadings:

(8)I’mlookingfor a man who speaks French .3

Theobjectnominalhasaspecificreadingwhen(8)means‘I’mlookingforaparticularmanwho speaks French’ and has a nonspecific reading when it means ‘I’m looking for any man who speaks French’. In some languages, the two readings of the nominal in (8) are formally distinguished. In Spanish, the specific and nonspecific readings of (8) are signaled by using differentmoods:theindicativemoodforthespecificreadingandthesubjunctivemoodforthe nonspecificreading(Frawley,1992:70).

3 TheexamplesinthediscussionofspecificityherearefromFrawley(1992,ChapterThree).Fordetailsabout specificity,pleaserefertoFrawley(1992).

9

AccordingtoFrawley(1992:74),specificity“canbefunctionallyinducedbytense,mood, definiteness,contextofutteranceorassumedknowledgebase.” ThepasttenseinEnglishtends toinduceaspecificreadingofthenominalandthehabitualpresenttenseanonspecificreading.

Thisisbecausepasttenseisconnectedtoactuality,whatreallyhappenedinthepast.Consider thefollowingtwoEnglishexamples:

(9)a.Johnboughtthefood. b.Johnbuysthefood.

Theobjectnominalin(9a)hasaspecificreading,whilethatin(9b)canhavebothspecificand nonspecificreading. (9b)canmean“Johnhabituallybuysparticularfood”or“Johnhabitually buysanyfood”andthelatterinterpretation(i.e.,thenonspecificreading)isarguablyeventhe preferred one (Frawley, 1992: 73). I will come to the inducement of specific reading by past tensewhendiscussingtheinterpretationoftheclassifierphraseinChineseinChapterFive.

Definiteness/indefiniteness and specificity/nonspecificity are closely related. Definite nominals tend to be specific, as they tend to encode old information. Still, not all definite descriptionsarespecific.Forexample,thedefinitenounphrase the lion in(10)canhavegeneric reference,so(10)canmean‘anymemberoftheclassoflionsisnoble’.

(10)Thelionisanobleanimal.

10

Indefinitenounphrases canbebothspecific andnonspecific as (8) illustrates, depending on contextualfactors.

Iuse‘specificity’heretomeanthefeatureofanominalincontexttorefertoaparticular member,notanymemberoftheclassofentitiesdescribedbythenoun.Comparethefeatureof

(non)specificityoftheobjectnounphraseinthefollowingtwosentences.

(11)Iboughtabookyesterday. (12)I’mgoingtobuyabookthisafternoon.

In(11), a book hasaspecificreference,asthespeakerboughtaparticularbook,while a book in

(12)canhave eitheraspecificornonspecific reference.Thespeakerplanstobuyaparticular bookoranyonebookthisafternoon.

In this section, I discussed several aspects of the concepts of ‘definiteness’ and

‘specificity’andwhatismeantbythetwotermsinthisdissertation.

1.2 The Organization of the Dissertation

Theorganizationofthedissertationisasfollows: I first conduct a literature review of studies concerning the structures of nominal expressions in Chapter Two. The theoretical motivations and empirical support for the DP hypothesis are described. I also discuss some studiesaboutChinesenominalsinthelinguisticliterature.

InChapterThree,somebackgroundisprovidedaboutlanguageuseinChinaandabout themainrelevantfeaturesofMandarinnominalexpressions.

11

InChapterFour,IdiscussnumeralclassifierswithparticularreferencetoMandarin.The useofclassifiersisaveryimportantfeatureofChinesenominals.Inanumericallyquantified nominalinChinese,thenumeralcannotbeuseddirectly before the head noun; a classifier or measurewordmustbeusedbetweenthem.Asaresult,muchofChapterFourisdevotedtothe discussion of classifiers. I first make a distinction between classifiers and measure words, althoughmanyscholarsusethesetwotermsinterchangeably.Theclassificationofclassifiersin theliteratureandthefunctionsofclassifiersarediscussedthere.Iwilltracethedevelopmentof theclassifiersysteminChinese,astherewerenoclassifiersinancientChinese.ThenIanalyze thesyntacticstructureofnumericallyquantifiednominalsinMandarin.

Chapter Five compares the functions of classifiers in Mandarin and the Wu dialect.

Relativelyspeaking,theuseofclassifiersinthesoutherndialectsofChineseismorediversethan inMandarin,althoughtheuseofclassifierswithinamajordialectareaisnotalwaysconsistent.

The distributions and interpretations of classifiers are discussed. I propose different syntactic structuresofthesurfaceform[Cl+N],whichcanhavetwodifferentinterpretationsintheWu dialect.ThedifferentfunctionsofclassifiersintheWudialectandMandarinareexplainedwith referencetotheirsyntacticstructures.

Complex nominal expressions in Mandarin are dealt with in Chapter Six. Although demonstratives,definitearticlesandpossessivepronounsareincomplementarydistributionin

English,theycancooccurinotherlanguages(likeSpanish,Bosnian).So,demonstrativesand possessive pronouns do not universally occupy D. They may not be inherently definite syntactically. Proper names/personal pronouns, demonstratives or possessives can cooccur

12 before the head noun in Mandarin. The position of possessives is quite flexible and several possessivescanoccurtogetherinaMandarinnominal.Syntacticstructuralanalysesforcomplex nominals in Mandarin will be proposed. Chapter Seven summarizes the conclusion of the dissertation.

13

CHAPTER TWO

LITERATURE REVIEW

In this chapter, I conduct a literature review of research about nominal expressions in generallinguisticsandinChineselinguistics.InSection2.1,thetheoreticalmotivationsforthe

DPhypothesisareprovided.TheempiricalsupportfortheDPhypothesisisdiscussedinSection

2.2.InSection2.3,thesyntacticandsemanticapproachestobareargumentsarepresented. In

Section2.4,IdiscusssomepreviousstudiesaboutChinesenominals.

2.1 Theoretical Motivations for the DP Hypothesis

Duringthe1970’s,atheorywasputforwardinlinguisticstocapturethesimilaritiesof differentphrases.ThistheoryiscalledX'theory,whereXstandsforalexicalcategory(N,V,Adj,

Prep).AccordingtotheX'theory,alexicalcategoryhasaphrasalprojection,inwhichthereisan optionalspecifierwhichissistertoanintermediateprojectioncalledX'(hencethenameofthe theory). Under the intermediate projection X' are the zerolevel category head X and its whichitselfisaphrase.SoaphrasewithintheX'theoryhasthefollowingstructure

(1):

14

(1)

Beforethemid1980’s, anominalexpressionwasanalyzed as an NP and all prehead elements (traditionally called determiners), such as articles, demonstratives, and possessive pronouns,occupythespecifierposition.Soanominalhadthefollowingsyntacticstructure(2):

(2)

Chomsky (1986) proposes that both lexical categories (like N, V) and functional categories (like Complementizer and Inflection) have phrasal projections, and specifiers and complements are always maximal projections. So a is analyzed as CP or IP, with

ComplementizerorInflectionasheadandwhphraseorNPintheirrespectivespecifier positions.Forexample,‘Hebrokethewindow’hasthefollowingstructure(3):

15

(3)

Unliketheanalysisofclausalstructure,therevisedX'theoryhadnotbeenextendedto theanalysisofnominalexpressionsbeforetheDPhypothesis.Anominalexpressionwasthen stillanalyzedasanNP,withdeterminersinitsspecifierposition.Suchananalysisofnominal expressionsisinconsistentwiththerevisedX'theory.Oneinconsistencyisthatdeterminers,asa functionalcategory,donothavetheirphrasalprojection.Anotheristhatthespecifierpositionof anominalexpressionhostsazerolevelcategoryD,notamaximalprojectionassupposedinthe revisedX'theory.

Inthe1980’s,somescholars(Szabolcsi,1983onHungarian;Hellan,1986onNorwegian;

Abney, 1987) proposed that nominal expressions are not NPs. Abney (1987) argues that determiners project their own phrasal category and that nominal expressions are really DPs, headedbyafunctionalcategoryD.BasedontheEnglishdata,Abneydrawstheconclusionthat variouselementssuchasarticles,demonstrativesandpronounsaregeneratedinD,astheyarein

16 complementarydistributioninEnglish. AnominalexpressionwithintheDPanalysishasthe followingstructure(4): (4)

TheDPanalysissolvestheaforementionedtwoinconsistencieswiththerevisedX'theory.

InthefollowingsectionwedealwiththeempiricalsupportfortheDPhypothesis.

2.2 Empirical Support for the DP Hypothesis

Manylinguisticfactsshowthatclausalandnominalstructuresaresimilar.Thisevidence comesfromdifferentareas:morphology,syntaxandsemantics.Thesimilaritiesbetween and nominalphrases cannotbe accommodated by thepreDP hypothesis analysis of nominal expressionsandthesesimilaritiesjustifytheDPhypothesis.Therefore,theDPhypothesiscannot onlymaketheanalysisofthenominalconformtothe revised X' theory, but also capture the

17 similaritiesbetweentheclauseandthenominalphrase.Inthefollowing,Ipresentsomeofthe factsvalidatingtheDPhypothesis(fordetailspleaserefertoBernstein(2001)).

Morphologically,incase,personandnumberbetweenthepossessorandthe possessed noun in the nominal and that between the subject and the verb in the clause are observedtobesimilarinsomelanguages,althoughtheEnglishnominalandclauselacksuch inflectional morphology. Abney (1987) cites data from Yup'ik, a Central Alaskan Eskimo language, Tzutujil, a Mayan language, and Hungarian to demonstrate the similar agreement morphologywithintheclauseandthenominal.Thepossessorinthenominalandthesubjectin theclauseareassignedthesamecaseinYup'ik,TzutujilandHungarian:ergativecaseinYup'ik andTzutujil;nominativecaseinHungarian(Abney,1987:20).Considerthefollowingdatafrom

Yup'ik(5)andHungarian(6):

(5)a.angutet kiputaat manERG(PL) buyOMSM 4 ‘themenboughtit’ b.angutetkuigat manERG(PL) riverSM ‘themen’sriver’ c.angutemkiputaaø manERG(SG) buyOMSM ‘themanboughtit’ 4 TheYup'ikdatawereoriginallyfromReedetal.(1977).OMstandsfor‘objectagreementmarker’andSMfor ‘subjectagreementmarker’.

18

d.angutem kuigaø manERG(SG) riverSM ‘theman’sriver’ (6)a.ateø vendeged5 theyouNOM guessPOSS2SG ‘yourguest’ b.(a)Mariøvendegeø (the)MaryNOM guestPOSS3SG ‘Mary’sguest’ c.Mariøaludtø MaryNOM sleptPast3SG ‘Maryslept’

Fromtheaboveexamples,wecanseethattheclauseandthenominalinbothYup’ikand

Hungariansharetheagreementandcaseassignment.Thesefactsstronglysuggestthat the clause and the nominal share the similar syntactic structure. Based on Hungarian data,

Szabolcsi(1983)arguesforanominalIPnode.Abney(1987)suggeststhatthecounterpartofthe clausalIPinthenominalisDP.

Syntactically,therearealsoparallelsintheclausalandnominaldomains. Intermsof structure, both the verb and the noun can take internal and external arguments, althoughtheargumentinthenominalisnotobligatoryandthatintheclauseusuallyis.Consider

5 ThedatawereoriginallyfromSzabolcsi(1983).(6a)and(6b)werecitedinbothAbney(1987)andBernstein (2001).(6c)wascitedinBernstein(2001).

19 thefollowingtwoexamplesandnoticethesimilaritybetweentheclausalargument(7a)andthe nominalargument(7b).

(7)a. John destroyed the spaceship . b. John ’sdestructionof the spaceship . (Abney,1987:15)

Intheliterature,theexternalargumentoftheverbisarguedtobebasegeneratedinSpecVP andthenitmovestoSpecIPforcasechecking(Koopman&Sportiche,1991).Thisapproachis calledtheVPInternalSubjectHypothesis.Inthesameway,somescholars(e.g.,Ritter,1988) arguethatthesubjectargumentofthenounisgeneratedinSpecNPandthenmovestoSpecDP tocheckitscase.

Another piece of syntactic evidence for the similarity between the clausal and the nominal domain is word order variation in different languages. Much research about word ordervariationwasconcernedwithRomancelanguagesandGermaniclanguages.Forexample, inFrenchmustfollowfiniteverbswhiletheymustprecedefiniteverbsinEnglish,asthe followingexamplestakenfromPollock(1989:367)illustrate.

(8)a. Jean embrasse souvent Marie . JohnkissesoftenMary b.* Jean souvent embrasse Marie . JohnoftenkissesMary c.*JohnkissesoftenMary. d.JohnoftenkissesMary.

20

Pollock(1989)suggeststhatunderlyinglyadverbsprecedefiniteverbscrosslinguistically.The wordordervariationofthewithrespecttothefiniteverbisduetosyntacticmovement.In

French,theverbmovesacrosstheadverbtotheInflnode,whileinEnglishtheverbremainsin theVnode.ThereasonforverbmovementinFrench,notinEnglish,isthatFrenchhasrich morphologyandEnglishdoesnot.Suchwordordervariationintheclausaldomaincanalsobe found in the nominal domain. The equivalent of the adverb is the in the nominal domain.InEnglishnominalsadjectivesprecedethepropernounwhiletheyfollowinRomance languages,asshownin(9)and(10)(drawnfromLongobardi,1994:624).

(9)a.*E’venuto vecchio Cameresi. (Italian) cameolderCameresi b.E’venuto Cameresi vecchio. cameCameresiolder c.E’venuto il vecchio Cameresi. cametheolderCameresi (10)a.OldJohncamein. e.*Johnoldcamein.

Longobardi(1994)arguesthatpropernamesinItalianmoveovertlyacrossadjectivestoDwhen itisnotoccupiedbytheexpletivedefinitearticle(9b),whilepropernamesinEnglishonlymove covertlytoDatLF. Withregardtocommonnouns,adjectivesarearguedtobeontheleftof thenoununderlyinglyinbothRomanceandGermaniclanguages.Thedifferentsurfaceposition

21 oftheadjectivewasattributedtotheoftheheadnouninRomance(butnotinGermanic) toafunctionalheadintermediatebetweenNandD(Cinque,1994).

Semantically,thereisalsosomesimilaritybetweentheclausalandnominaldomain.In the literature, an independent clause is treated as an IP and a sentential argument as a CP.

ConsiderthefollowingEnglishexamples.

(11)a.Hefailedthetest. b.*Thathefailedthetest. (12)a.Thathefailedthetestsurprisedme. b.*Hefailedthetestsurprisedme.

(11a),whichexpressesaproposition,isanIPandthat he failed in the test in(12a)isanargument and it is the complementizer that that turns he failed in the test into an argument. The tree diagramsfor(11a)and(12a)areasfollows.

22

(11a)

23

(12a)

Similarly,nominalpredicatesareNPsandnominalarguments DPs. It is the determiner

thatturnsanominalintoanargument.Lookatthefollowing Italianexamples(from

Longobardi,1994:619):

(13)*Gianniemedicochesicuradavverodeisuoipazienti. Gianniisdoctorwhoreallycaresforhispatients. (14)Giannieunmedicochesicuradavverodeisuoipazienti. Gianniisadoctorwhoreallycaresforhispatients. (15)Giannie(un)medico. Gianniisadoctor.

24

Theungrammaticalityof(13)andgrammaticalityof(14)showthatanominalpredicatewithout anaccompanyingdeterminercannotfunctionastheheadofarelativeclauseinItalian.Butthe determiner is optional when there is no relative clause following the predicate nominal as indicatedin(15).Actually,inEnglishanominalpredicatesometimesdoesnotneedadeterminer either.

(16)a.Childasheis,hehasmoresensethanhisfather. b.Weelectedhimpresident.

In(16),thesingularcountnouns child and president ,functioningaspredicates,arenotpreceded by articles. From the perspective of (non)argument, the sentential argument and nominal argumentarerespectivelyanCPandanDP;asentenceorapropositionandanominalpredicate arerespectivelyanIPandanNP.

Anumberofscholars(Abney,1987:77;Cheng&Sybesma,2005:227)observethat

TensePhrase(TP)(sometimessubsumedunderInflintheliterature)intheclausaldomainis equivalenttoDPinthenominaldomainandVPintheclausaldomainisequivalenttoNPinthe nominaldomain.BothTPandDPhavedeicticproperty,thepropertytobeabletorefer,whileVP andNPdescribe.Forexample,theNP dog justdescribestheintensionalpropertyof‘dogness’ possessedbyacertaintypeofanimalandtheVP bake cookies describesacertaintypeofevent.

WhentheNPandtheVPareembeddedinDPandTPorIPrespectively,thenthenominalandthe verbalhavethepropertyofreferentiality.Asaresult,thenominal the dog andtheverbal baked

25 cookies in(17a)and(17b)refertoaparticulardogandaparticulareventof‘bakingcookies’in thereallifeworldrespectively.

(17)a.Thedogisbarking. b.Johnbakedcookies.

AboveIdiscussedthesimilaritiesbetweentheclausalandthenominaldomainsinterms of their morphology, syntax and semantics. These similarities can only be captured when a higherfunctionallevelaboveNPisassumed.Abney(1987)suggeststhatthishigherfunctional levelisDP.ThelexicalcategoryVPintheclauseisequivalenttoNPinthenominal;theyboth describe. Intermsof(non)argument,thefunctionalcategoryCPintheclausecorrespondsto

DPinthenominal;theycanbothfunctionasargumentoneisasententialargument,theother anominalargument.Intermsofmovementandcasecheckingandreferentiality,thefunctional categoryIPintheclauseisequivalenttoDPinthenominal;thesententialsubjectandnominal subjectmovetoSpecIPandSpecDPforcasecheckingandbothIPandDPhavethepropertyof referentiality.Inthenextsection,Igiveabriefreviewofresearchonbarenouns.

2.3

Inthissection,Ireviewsomestudiesaboutbarenouns.Itisgenerally agreedthatthe article is the typical lexical item that fills the head D of DP. Abney (1987: 266) argues that personal pronouns are “intransitive” determiners and that they occupy the D position. With regardtothepossessivecontainingnominalinEnglish,thepossessorisconsideredtobeinSpec

26

DP and AGR(eement), which does not cooccur with lexical determiners in English, is in D

(Abney,1987:271).

Insomelanguageswitharticles,suchasFrench,theheadofDPmustbelexicallyfilled, whileinotherarticledlanguages,theheadofDPisnotalwayslexicallyfilled,likebare andmassnounsinEnglish.Stillotherlanguagesdonothaveanyarticleatall,likemostSlavic languages and Chinese, and in such languages bare nouns are widely used. Compare the followingexamples.

(Theintendedreadingofthesubjectnominalsofthefollowingexamplesisgeneric)

(18) a. Wine isexpensive. b. The tiger isamammal. c.Tigers aremammals. (19)a. Le vin coûte cher . (French) The wine cost expensive ‘Wineisexpensive.’ b. Le tigre est un mammifere . thetigerisamammal ‘Thetigerisamammal.’ (20)a. Jiu gui . (Chinese) wine very expensive ‘Wineisveryexpensive.’ b. laohu shi purudongwu . Tigerbemammal ‘Thetigerisamammal.’

27

TheaboveexamplesshowthattheargumentnominalinFrenchmusthaveadeterminer,thatin

Chineseitdoesnotneedadeterminer,andthatadeterminerisusedwiththesingularnoun,but nodeterminerisneededwithamassnounorapluralnouninEnglish.

There are two approaches to bare nouns in the literature: a syntactic one, of which

Longobardi (1994) is representative, and a semantic one, of which Chierchia (1998a, b) is representative. Longobardi(1994)suggeststhatargumentnominalsneedtobeintroducedbyD.

Furthermore,heassumesthatthenominalcanonlyhave an existential interpretation if the D positionisleftemptyandthatanemptyheadmustbelexicallygoverned(p.641).Barenouns, whichcanfunctionasargumentsinbothItalianandEnglish,projectDPs.PropernamesinItalian movetoDovertlywhenDisnotlexicallyfilledbytheexpletivearticle,whileinEnglishproper namesmovetoDcovertlyatLF,toavoidthepropergovernmentrequirementofanemptyhead.

Theevidenceforthisproposalisthewordorderdifferencewhenanadjectivemodifiesaproper nameinItalianandEnglish.WhentheDpositionisempty,thepropernamemustoccurbefore theadjectiveinItalian.WhentheDpositionisfilledbythedefinitearticle,thenthepropername canoccuraftertheadjective.InEnglishapropernamecanonlyfollowanadjective.Werepeat

(9)and(10)fromLongobardi(1994:624)hereforillustration.

(9)a.*E’venuto vecchio Cameresi. (Italian) cameolderCameresi b.E’venuto Cameresi vecchio. cameCameresiolder c.E’venuto il vecchio Cameresi.

28

cametheolderCameresi (10)a.OldJohncamein e.*Johnoldcamein.

BarecommonnounsinItalianprojectDP,buttheydonotmovetoD,sotheheadDis empty.AsaresultofDbeingempty,barecommonnounsinItaliancanonlyhaveexistential interpretationandoccuronlyinalexicallygovernedposition.However,inEnglish,barenouns aremuchfreerindistributionthanthoseinItalian.Also,barenounsinEnglishcanbothhave existential and generic interpretation, as the following examples from Longobardi (1994) indicate.

(21)a.Mangio potate. (p.613) Ieat/ameating potatoes. b.*Castoriconstruisconodighe. (p.630) Beaversbuilddams. c.*Canistavanosedutisulmioprato. Dogsweresittingonmylawn.

In(21a)theargumentnominal potate ,whichhasanexistentialinterpretation,projectsaDP,with the head D empty. (21a) is grammatical since the bare nominal with an empty D head is governed by the lexical verb, and (21b) and (21c) are ungrammatical, because the argument nominals are bare and they occur in a position that is not lexically governed. The English equivalentsof(20b)and(20c)aregrammatical; beavers hasagenericinterpretationand dogs an existentialinterpretation.AccordingtoLongobardi,abarenounhavingagenericinterpretation

29 inEnglishmovestoDatLF,thusavoidingaviolationofthelexicalgovernmentrequirement.

The bare noun which has an existential interpretation does not move to D, yet it occurs in nonlexicallygovernedpositioninEnglish.AccordingtoLongobardi(1994:645),theexistential interpretation of bare nouns in the nonlexically governed position in English is a marked phenomenon;inotherGermaniclanguagesbarenounsinthenonlexicallygovernedpositioncan onlybe genericallyinterpreted. Longobardiofferstwoproposalstoexplaintheof theexistentialinterpretationofbarenounsinanonlexicallygovernedpositioninEnglish.One proposalisthatdefaultexistentialclosuredeterminestheinterpretationofthenominalintroduced bythe empty D andlatertheheadnounmovestoD at LF and D is no longer empty, thus avoidingtheviolationofthegovernmentrequirementofanemptyheadatLF.Anotherproposal isthatthesubjectsofstagelevelpredicatesinEnglishareinaVPinternalpositionatLF,even thoughoccurringinSpec IP atSStructure(whichwasoriginallydevelopedinKratzer(1988) andDiesing(1988))andthelexicalgovernmentrequirementofanemptycategoryissatisfied.

Chierchia (1998a, b) approaches bare nouns from a semantic perspective. Nouns generallycanbepredicatesasrestrictorsofquantifiersandcanbeargumentsasnamesofkinds.

But the ways that they are realized in different languages are different. Chierchia (1998a) proposes that the two features [±arg] and [±pred] constrain the way in which N and NP are mappedontotheirinterpretation.BasedonwhetherbareNandNPinalanguagecanoccurinthe argument or predicate position, languages are classified into three types: NP [+arg, pred] languages,NP[arg,+pred]languagesandNP[+arg,+pred]languages.

30

Eachtypeoflanguagehasaclusterofproperties.InNP[+arg,pred]languages,nouns refertokinds;barenounscanbearguments;allnounsaremassandhavenopluralmarkingsbut ratherhaveaclassifiersystemforcounting.Insuchalanguage,thenominaldoesnotneedtobe aDPforargumenthood.ChineseandJapaneseareexamplesofthistypeoflanguage.Iusethe followingChineseexamplesforillustration.

(22) xuesheng zai jiaoshi kai banhui .

Student in classroom hold classmeeting

‘Thestudentsarehavingaclassmeetingintheclassroom.’

(23) you wu *(ge) xuesheng zai da lanqiu .

Have five Cl student Prog beat basketball

‘Therearefivestudentsplayingbasketball.’

Bare noun xuesheng ‘student(s)’ in (22) functions as an argument in. When the noun is numericallyquantified,aclassifiermustbeused(23).

InaNP[arg,+pred]language,everynounisapredicateandthenominalhastoprojecta

DPinordertobeanargument,althoughtheheadDcanbelexicallyfilledorphoneticallynull.

FrenchandItalianaretwoexamples.TheheadDisovertinargumentpositioninFrench,whileit canbenullinItalian.TheoccurrenceofnullDmustobeytheruleforemptycategory,thatis,it mustbelicensedorgovernedbyalexicalhead;therefore,thebareNasargumentisrestrictedin distributioninItalian.IrepeattheItalianexample(21)(fromLongobardi,1994)forillustration.

31

(21)a.Mangio potate. (p.613) Ieat/ameating potatoes. b.*Castoriconstruisconodighe. (p.630) Beaversbuilddams. c.*Canistavanosedutisulmioprato. Dogsweresittingonmylawn.

In(21a)thebarenoun potate ‘potatoes’isinalexicallygovernedpositionandfunctionsasan argument.Wheninanonlexicallygovernedposition,barenounscannotbeanargumentas(21b, c)show.

InaNP[+arg,+pred]language,NandNPcanbeeitherpredicateorargument.Thatis, theycandenoteeitherkindsorpredicates.Insuchalanguage,thereismass/countdistinction amongnouns.Massnounsandpluralsareabletooccurasbarearguments;singularcountnouns donot.Englishisanexampleofalanguageofthistype.Considerexample(24).

(24)a.Iboughtbooks/waterthisafternoon.

b.Ibought*(a)bookthisafternoon.

c.Weelectedhimpresident.

(24a)illustratesthatbarepluralslike books andmassnounslike water inEnglishcanfunctionas anargument.Baresingularnouns,ontheotherhand,cannotbeanargument;theycanfunction asapredicateasshownin(24b,c).

32

With respect to the nominal in languages without dedicated articles to signal

(in)definiteness, scholars are divided about whether the nominal projects DP or not. Some linguists(Progovac,1998;Leko,1999;Tang,1990;andLi,1998,amongothers)assumethatthe projection of DP is a universal property and argue thatDPisprojectedinthenominal,while others(Lyons,1999;Cheng&Sybesma,1999,2005;andTrenkic,2004,amongothers)holdthat languageswithoutarticlesdonotprojectDP.Progovac(1998)useswordorderdifferencesofthe noun/pronounwithrespecttocertainintensifyingadjectivestoconcludethatpronounsoccupya structurallyhigherpositionthannounsinSerboCroatian,wherepronounsprecede andnouns followcertainintensifyingadjectives.Trenkic(2004),ontheotherhand,followingLyons(1999), argues that definiteness is not grammaticalised in Serbian/Croatian/Bosnian, although definitenessasasemantic/pragmaticconceptisuniversal.Cheng&Sybesma(1999,2005)argue thatthereisnoDcategoryinChinesenominals,thattheclassifierinChineseperformssomeof thefunctionsof articles insomewesternlanguages,andthatdefinitenominalsinChineseare

ClP.

Beforeconcludingthissection,Imentionsomestudiesofthemoreelaboratedanalysesof internalstructuresofthenominalsincetheDPhypothesiswasfirstputforward.Abneyoriginally analyzedthenominalasaDPtakinganNPasitscomplement.Sincethen,researchintoother languageshasshownthatdemonstratives,possessivepronounsandotherelementswhichcannot cooccurwiththedefinitearticlebeforenounsinEnglishdooccurtogetherinotherlanguages.

Therefore,laterstudiesproposedaseriesofotherfunctionalprojectionsbetweenDandNPin ordertoaccommodatetheseprenominalelements.NumberPhrasewasproposedforHebrewby

33

Ritter (1991), and Gender Phrase for Romance languages (Picallo, 1991). Cinque (1994) proposed some unidentified intermediate functional categories whose specifier positions host prenominaladjectives.

2.4 Research on Chinese Nominals

WithregardtoChinesenominalexpressions,somescholars(Li,1998,1999;Tang,1990) holdtheviewthatChinesenominalexpressionshavetheDPprojection,whileCheng&Sybesma

(1999,2005)believethatClPinChineseisequivalenttoDPinlanguageswitharticles. Inthe followingsubsections,IwillreviewTang(1990),Li(1998,1999)andCheng&Sybesma(1999,

2005), which are the representative studies concerning Chinese nominal expressions in the literature.

2.4.1 Tang (1990)

FollowingAbney’sDPanalysis(1987),Tang(1990)putforwardthefollowingstructure toanalyzetheChinesenominalexpressions.Inheranalysis,demonstrativesaregeneratedunder

D and there is a phrasal category (ClP) between DP and NP. Numerals and classifiers are generatedundertheClhead,astheyalwaysoccurtogetherandusuallynootherelementcanbe inserted between them in Mandarin. Tang (1990) proposes that a Mandarin nominal has the followingstructure(25):

34

(25)

‘thesethreebooks’

Tang (1990) does not take into account other more complex nominals involving personal pronounsandpropernames.justconsidersboththenumeralandclassifiertobeunderthe headCl.Iwilldiscusswhethernumeralsandclassifiersprojectseparatephrasalcategoriesoras acomplexheadprojectonephrasalcategorytogetherinChapterFour.

2.4.2 Li (1998, 1999)

Li (1998, 1999) analyzes the different structures of numerically quantified Mandarin nominalsindifferentcontextsandthestructureofMandarinnominalswiththesocalledplural markermen respectively.SheproposesinthetwoarticlesthatMandarinnominalshaveaDP projection.

35

NominalsoccurringbeforeaverbtendtobedefiniteinChinese,aspreverbalnominals aretopicsandtopicstendtobeoldinformation.Numerically quantified nominals are usually indefinite,sotheycannotfunctionasasubjectinChinese.Inreality,theycanoccurasasubject in some contexts, but not in others. Li (1998) proposes that numerically quantified Chinese nominals(the[Nume+Cl+N]sequence)canbeanalyzedeitherasanindefiniteindividual denoting DP or a quantitydenoting NumP. Individualdenoting DPs are referential and quantitydenotingNumPsarenotreferential.Thenominal wu-ge xuesheng (‘five students’) in

(26)isanindividualdenotingDPandin(27)aquantitydenotingNumP.

(26) you wu-ge xuesheng kaoshi de ai . (individualdenoting) havefiveClstudenttestgetA ‘TherearefivestudentswhogotanAonthetest’ (27) wo xuyao wu-ge xuesheng bangmang . (quantitydenoting) I need fiveCl student help ‘Ineedfivestudentstohelpme.’

The nominal wu-ge xuesheng (‘five students’) in (26) has specific reference, referring to five particularstudents,whilethatin(27)haskindofgenericreference,meaning‘anyfivestudents’.

Somelinguisticfactssupportthetwodifferentanalysesofthe[Nume+Cl+N]sequence in Mandarin. For example, an individualdenoting [Nume + Cl + N] sequence, that is, an indefinitereferentialnominalexpression,cannotoccurbeforetheverbinMandarin,whichcan be explained by the syntactic principle that an empty category must be lexically governed

(Longobardi,1994)ifweconsideranindefinite[Nume+Cl+N]asaDPwithanullhead.A

36 quantitydenoting[Nume+Cl+N]nominalcanappearbeforetheverb,astheheadNumis filledwithanumeral.Let’scomparethefollowingMandarinandEnglishexamples.

(28)*San-ge xuesheng zai xuexiao shoushang le . (individualdenotingDP) ThreeClstudentatschoolhurt SFP ‘Threestudentswerehurtatschool.’ (Li,1998:694) (29) San-zhi gunzi gou ni da ta ma? (quantitydenotingNumP) ThreeCl sticks enough you hit him Ques ‘Arethreesticksenoughforyoutohithim(with)?’(Li,1998:695) (30)Threestudentswerehurtatschool. (individualdenoting) (31)Threestudentsareenoughtomovethetable. (quantitydenoting)

(28)and(29)showthatanindividualdenoting [Nume+Cl+N]nominalexpressioncannot occurinthesubjectpositioninMandarin,butthequantitydenoting[Nume+Cl+N]nominal can.

In addition, Li (1998) uses the cooccurrence of the universal operator dou ‘all’ and the existentialexpression you ‘have’with[Nume+Cl+N]sequencesincertaincontextsbutnotin other contexts to support her proposal that [Nume + Cl + N] sequences have two different structures. Dou ‘all’and you ‘have’canonlycooccurwithindividualdenotingDPs((32)and

(33)), because the scope of the universal operator isoveranentiresetofindividualsandthe existential expression asserts the existence of individuals. They can not occur together with nonreferentialquantitydenotingNumP((34)and(35)).

37

(32) San-ge xuesheng dou lei le . (individualdenotingDP) ThreeCl student all come SFP ‘Threestudentsallcame.’ (33) You sange xuesheng lei le . (individualdenotingDP) Have threeCl student come SFP ‘Therearethreestudentsthatcame.’ (34) *San-zhi gunzi dou gou ni da ta ma? (quantitydenotingNumP) ThreeCl sticks all enough you hit himQues (35) *You san-zhi gunzi gou ni da ma? (quantitydenotingNumP) have threeCl sticks enough you hit Ques

Basedonthedifferentusesandinterpretationsofthe[Nume+Cl+N]nominal,Li(1998)argues thatDPcanbeprojectedeventhoughMandarindoesnothavearticles.

Li(1999)investigatesthenominalexpressionswiththesocalledpluralmarkermen .Itis generally understood that men can only be added to singular personal pronouns and human nouns.Itisalsobelievedthatthe[Nume+Cl+N]nominalcannotbepluralized(thatis,[Nume

+ Cl] is incompatible with men ). Li (1999) basically provides an explanation for the

(un)grammaticalityofthefollowingnominalexpressions.In(36)and(37),thepropernounand personalpronounwiththepluralmarkermen canoccurbeforethenumeralclassifier. (38)and

(39)areungrammatical;thatis,commonnounswith the plural marker men cannot cooccur withthenumeralclassifier.

(36) XiaoQiang-men / xiaozhang-men san-ge (ren) XiaoQiangCol/principal Col threeCl (person)

38

‘XiaoQiang/Principalandtwoothers(inthegroup)’ (37) ta-men san-ge (ren) they threeCl (person) (38) *Xuesheng-men san-ge (ren) student Col threeCl(person) (39) *san-ge xuesheng-men threeClstudentCol

Li (1999) hypothesizes that pronouns and proper nouns, being definite, are basegeneratedinDandmen isattachedtoanelementinD(butshedoesnotsaywhere men is generated)andthatthepluralizedpronounandpropernounwithmen canbefollowedbythe

[Nume+Cl+N]sequence.Butcommonnouns(whicharegeneratedunderthenodeN)with

men cannotcooccurwiththe[Nume+Cl+N]sequence,thereasonforwhich,accordingtoLi

(1999),isthatthepluralizednounmustbemovedto D to check the definiteness feature, as pluralizednominalexpressionsmustbedefinite;however,theClheadblockstheNmovementto

D.

Liarguesthatnominalexpressionswithmen havethefollowingsyntacticstructure(40):

39

(40)

‘thethreeofthem’

However,thefollowingexpressionscannotbeexplained by Li’s hypothesis. Li (1999) argues thatthepluralmarkermen shouldbeaddedtoelementsinD.Theheadnounswiththemarker

men in(41),(42)and(43)allfollowaclassifier,sotheycannotbeintheDposition.

(41) zhe/a qun hezhi -men this/aClchildCol ‘this/agroupofchildren’ (42) ji wei dashi -men severalClmasterCol ‘severalmasters’ (43)s an-wu-ge ren -men threefiveClmanCol ‘aboutthree?/five?people’‘severalpeople’

40

In(41),thecollectiveclassifier qun (‘group’)cooccurswithapluralizednoun,whilein(42)and

(43), the individual classifiers wei and ge combine with pluralized nouns when the numeral/quantifier+classifierdoesnotexpresstheexactnumberoftheentitydenotedbythe pluralizednoun.(41),(42)and(43)showthatpluralizednounsdonothavetomovetoD. It canbearguedthatpluralizednounsinMandarinarenotalwaysdefinite,justlikepluralssuffixed withtati inJapanese(Nakanishi&Tomioka,2004). Probably,men isnotapluralmarkerat all,butacollectivemarker,asIljicargues(1994).Iwilldiscusstheusageandthegrammatical statusofmen insomedetailinChapterThree.

2.4.3 Cheng & Sybesma (1999, 2005)

Cheng&Sybesma(1999)dealswithbarenounphrasesinMandarinandand

Cheng & Sybesma (2005) analyzes the syntactic structure of bare noun phrases in Mandarin,

Cantonese,Wenzhou(aWusubdialect)andtheMindialect.BarenounsinChinesecanhave both definite and indefinite reference. To them, bare nouns in Chinese are not really bare.

DefinitebarenounsareClPwithanullheadandindefinitebarenounphrasesareNumePwitha nullhead.

FollowingthespiritofLongobardi(1994),C&S(1999,2005)suggestthattheclassifier, like a determiner, has an individualizing or singularizing function and that a bare noun in

Mandarin is at least a ClP. Classifiers in Mandarin perform some functions of determiners in other languages, for example, classifiers turn Chinese nouns, which are predicates, into arguments.

41

C&SconsiderdefinitebarenounsinMandarinasClPsandproposethatnounsmove covertly to the Cl position in order to receive a definite interpretation. Following Chierchia

(1998),theyassumethatforarticlelesslanguageslikeChinese,thenonovert ιoperator,whichis equivalenttoadefinitearticle,isavailable.NtoClmovementisanecessarystepfortheuseof the ιoperator.AsthereisnoovertclassifierindefinitebarenounsinMandarin,theycanbe interpretedeitherassingularorplural.

The indefinite bare nouns and indefinite [Cl + N] phrases in Mandarin Chinese are believedtobeaNumePwithanemptyClheadandanemptyNumeheadintheformerandonly anemptyNumeheadinthelatter.Astothebarenoun having generic interpretation, C & S argues that they, like definite bare nouns, undergo NtoCl movement. They demonstrate similaritiesbetweendefinitebarenounsandgenericbarenounsintermsoftheirdistributionand semantics.TheydrawananalogybetweenbarenounsinMandarin,whichcanbeinterpretedas eithergenericordefinite,andnounphraseswithadefinitearticleinEnglishwhichcanalsobe interpretedaseithergenericordefinite.

(44) gou hen jilin . dogveryintelligent ‘Thedogisveryintelligent.’

Example (44), just like its English counterpart (‘The dog is very intelligent.’), can make a predicationaboutdogsingeneraloraboutaparticulardog.

42

IftheClPinMandarinfunctionsliketheDPinEnglish,thenitisveryhardtoexplainthe structureofcomplexnominalexpressionslike:

(45) wo zhe san ben shu mythisthreeClbook ‘thesethreebooksofmine’

In example (45), both the demonstrative zhe ‘this’ and the classifier ben are used. It is unattractivefortwodifferentheadstohavethesamefeature.Also,itishardtounderstandhow theindefiniteNumePcanhaveasitscomplementtheClPwithadefinitereferenceandanumeral cancancelitsdefinitefeatureiftheClPisreally definite.Besides,the [Cl+N]sequencein

MandarinChineseisverylimitedindistributionandneverhasadefiniteinterpretation. How canaClPinChinesebeequivalenttoaDPinEnglish?IwillreturntoCheng&Sybesma’sideas abouttheclassifierinChineseinChapterFiveandthereIsuggestthattheclassifierinitselfdoes nothavethefeatureofdefinitenessandthe[Cl+N]sequencecanhavedefiniteinterpretation becausetheclassifiermovestoD.Whenthereisademonstrativeornumeralbeforetheclassifier inanominal,theclassifierremainsinitsbaseposition.

In summary, I presented theoretical motivations and empirical supports for the DP hypothesisinthischapter.Ialsoreviewedthetwoapproaches,thesyntacticandthesemanticone, tobarenounsintheliterature.StudiesaboutChinesenominalexpressionswerealsodiscussed.

43

CHAPTER THREE

MANDARIN CHINESE: AN OVERVIEW

Forinformationalpurposes,IdescribeinthischapterthelanguageusesituationinChina

(Section3.1)andsomelinguisticfeaturesofMandarinChineserelatedtonominalexpressions

(Section3.2).

3.1 Language Use in China

Chinaisamultiethnicandmultilingualcountry.ThemainethnicgroupisHan.TheHan peopleaccountforaboutninetytwopercentofthetotalpopulationinChina.Themajorlanguage inChinaistheChineselanguage(inmainlandChinasometimescalled Hanyu ‘thelanguageof the Han people,’ Zhongwen ‘the language of China,’ or Putonghua ‘the common speech’), spoken by the Han people and some minorities. Besides the Han people, there are fiftyfive ethnicminoritiesinChina.Aboutseventypercentoftheseminoritieshavetheirownlanguages, someofwhichdonothaveawritingsystem.Butsomeminoritylanguagesareonthevergeof extinction, because of the popularization of Mandarin by the Chinese government and the reluctanceofminoritychildrentolearntheirownethniclanguages.Minoritiesareunderpressure to speak Chinese in order to better themselves economically. Scholars have advocated for

44 compulsory bilingual education for minority childreninordertosavethedwindlingminority languages.

Several terms are used in linguistics to describe the language spoken by the Chinese people. The term referring to the official or standard Chinese language in mainland China is

Putonghua, thepronunciationofwhichisbasedonBeijingdialectwhichbelongstotheNorth dialect group. In the West, scholars often use Mandarin or Mandarin Chinese to refer to the standardlanguageinChina.SomescholarsalsouseStandardMandarinforthestandardvariety oftheChineselanguage. Inthisdissertation,IuseMandarintorefertothestandardvarietyand

Chinese or the Chinese language to designate the language in general (i.e. both the standard varietyanddialects)usedbytheChinesepeople.Ifaparticularsubdialectisintended,Iusethe nameoftheplacewhereitisspoken.

NooneknowsforcertainhowmanydialectsofChinesethereare.TheChineselanguage isnotoriousforthemutualunintelligibilitybetweenitsdialects.Sometimespeoplecomparethe differences in pronunciation between Chinese dialects to those between Romance languages.

Chineseisoftencitedasanexampleindialectology andsociolinguisticswhenthediscussion evolves around what criterion should be used in deciding whether two speakers speak two dialectsofthesamelanguageortwodifferentlanguages.Ifthelinguisticcriterionof“mutual intelligibility” is adopted to distinguish dialect from language, then Chinese is a group of languages,asspeakersfromdifferentareascannotunderstandeachotherwhentheyspeaktheir local dialects. Sometimes, people living miles away from each other cannot communicate successfullyusingtheirowndialects,especiallyifthereexistssomegeographicalbarrierlikea

45 riveroramountain.Intruth,somescholars(Cheng&Sybesma,1999,2005)intheWestcall

Cantonese,WuandotherChinesevarietiestheChineselanguages. Ethnologue (1996,citedin

Nettle&Romaine2000:29)listedChinese,Wu(dialect)asoneofthetopfifteenlanguagesin termsofnumberofspeakers.However,Chinesepeopleusuallyperceivethedifferentvarieties thattheyuseasdialectsoftheChineselanguagebecausetheysharethesamepoliticalentity,and auniformwritingsystemwhichislogographic.

Traditionally, the Chinese language is divided into seven dialect groups: Northern

(sometimescalledMandarinintheWest),Wu,Yue(Cantonese),Kejia(Hakka),Min,Xiangand

Gan.ThenativespeakersoftheNortherndialectaccountforseventypercentofthetotalChinese population(Li&Thompson,1990).Theyspreadwidelyintermsofgeographicalareaandyet theycanmoreorlessunderstandeachother.Theothersixmajordialectsarespokeninthesouth and southeast part of China. Cantonese is spoken in some areas in Guangdong and Guangxi provincesandHongKong.ManyoverseasChinesealsospeakCantonese.Mindialectisspoken intheFujianandprovinces.ItisusuallysubdividedintoNorthernandSouthernMin.

XiangdialectisspokenintheHunanprovinceandGandialectintheJiangxiprovince.Speakers of Hakka whose ancestors were migrants from the North are scattered in Guangdong, Fujian,

Hunan,andJiangxiprovinces.Manysubdialectsinthesoutharemutuallyunintelligibleandat thesametimesomedialects,likesomeminoritylanguages,aredisappearing.6

6 WiththepopularizationofMandarinbythegovernmentandthemigrationofpeoplefromeconomicallyunderdevelopedareas tothesoutheastpartsincetheimplementationoftheopendoorpolicyinChinainthelate1970s,manydialectsthereare disappearing.AsurveyconductedabouttheJinhuadialect(belongingtotheWudialectgroup)inZhejiangprovinceshowsthat almosteverychildbetween6and14yearsoldcanspeakMandarin. Fiftytwo(Footnotetobecontinuedonthenextpage.)

46

The Wu dialect, which will be discussed in this dissertation, is mainly spoken in the southernpartoftheJiangsuprovince,ShanghaiandtheZhejiangprovince.Intheliterature,the

WudialectissometimesdividedintotheNorthernWu(spokeninthesouthernpartofJiangsu province,ShanghaiandtheNorthernpartofZhejiangprovince)andtheSouthernWu(spokenin thesouthernpartofZhejiangprovince). Thephonological differences among subdialects of the Southern Wu are greater that those among subdialects of the Northern Wu. Speakers of

SouthernWudialectslivingafewmilesapartsometimescannotunderstandeachother.Inthis dissertation,thedataoftheWudialectismainly from the Wenling dialect a Southern Wu dialectIgrewupspeaking.

The most prominent differences among Chinese dialects are within phonetics and phonology.MandarinhasfourtoneswhilesomesoutherndialectslikeCantonesehavesevenor eighttones.SomesoutherndialectsalsoretainmanywordsofOldChinese.Intermsofsyntax thereisnotmuchdifferenceamongdifferentdialects,especiallyinthewrittenform.Oneofthe maingrammaticaldifferencesbetweenMandarinandtheWudialectandCantoneseliesinthe useofclassifiers.WewillreturntothedifferentusesofclassifiersintheWudialectinChapters

FourandFive.

percentofthechildrencannotspeakthelocaldialectandonly22.65percentofthechildrencancommunicateinthelocaldialect quitewell.AnotherreportisaboutShenzhen,aonetimefishingvillageborderingonHongKong.Ithasbecometheonlycityin GuangdongprovincewhereMandarinismostlyspoken. Sincetheeconomicreformstartedin1980,itspopulationhas increased30timesin25years.PeopletherehavetouseMandarinincommunicationandtheoriginallocaldialectisalmost extinct.

47

3.2 Some Grammatical Features of Mandarin Nominals

Genetically, Mandarin belongs to the SinoTibetan language families. In terms of phonological features, Mandarinisatonelanguage,inwhicheverysyllablehasacontrastive pitchwhichisaninherentpartofthesyllable.Therearenoconsonantclustersandonlytwo consonants([n]and[ŋ])canbeinthecodaofasyllable.ThecanonicalwordorderinaMandarin sentenceisSVO.Inthissubsection,IdescribesomefeaturesrelevanttotheMandarinnominal.

3.2.1. An Isolating Language

Mandarinisusuallyconsideredanisolatinglanguage.Eachmorpheme,exceptforloan words,hasonlyonesyllable.UnlikeOldChinese,inwhichmostwordsweremonomorphemic and monosyllabic, modern Mandarin has many dissyllabic words and compound words.

However, it has very little inflectional morphology. Verbs do not inflect for person or tense.

Nounsarenotinflectedforcase,genderornumber.Asthereisnocasemorphologyoragreement markers, word order becomes a very important device in indicating the grammatical relation between the verb and nominal(s) in a sentence. For example, in (1) and (2), the preverbal nominal is the and the postverbal nominal is the patient and only the word order determineswhichnominalisthesubjectandwhichistheobject.

(1) Wo da-le zhe ge ren . I beatPerf this Cl person ‘Ibeatthisperson.’

48

(2) Zhe ge ren da-le wo . thisClpersonbeatPerfI ‘Thispersonbeatme.’

With regard to number marking, scholars agree (Cheng & Sybesma, 1999, 2005;

Chierchia,1998a,b)thatnumberisnotagrammaticalizedcategoryinChinese,asabarenoun canhavebothasingularandpluralinterpretationindifferentcontexts.Thereisindeedabound grammaticalinMandarinwhichissometimesaddedtopronounsandsomehumannouns.

This grammatical morpheme is men . There is no consensus in Chinese linguistics about the statusofthesuffix -men .Manynativespeakers(especiallythosewhohavelearnedaWestern language)wouldthinkthat -men isapluralmarker,butdifferentlinguistshavedifferentideason theissue.Somelinguists(Li&Thompson,1981;Li,1999)believethatmen isapluralmarker addedtopronounsandnounsreferringtopeople.Chao(1968)holdstheviewthatmen isapure pluralmarkerwhenappliedtopersonalpronouns,butitisacollectivemarkerwhenaddedto humannouns. Otherlinguists(Lü,1944;Iljic,1994,2001a,2001b,2005)arguethatmen isa collectivemarker,butnotapluralmarker,nomatterwhetheritisusedwithpronounsorhuman nouns.Iljic(2005)callsmen “personalcollective”.Stillothers(Chen,1987)thinkthatthereare twowaysofmarkingpluralityinMandarin:oneway istousenumeralstoexpresstheexact number of the referents in counting; the other is to suffix men to the noun to indicate the collectivemeaningofplurality.Now,wewilllookatthecontextsinwhich -men can,mustand mustnotbeused.

49

Thecontextinwhichmen isusedisveryrestricted.Evenwhenitcanbeused,itsuseis veryoftenoptional.Inmanycontexts,theuseofmen isprohibited.So,barenounsareprevalent in Chinese and they can have singular or plural interpretation, as the following example illustrates.

(3) jintian xiawu wo qu mai shu . todayafternoon Igobuybook ‘Iwillgotobuya/thebook(s)thisafternoon.’

In(3),thenumberofbookstobeboughtcanbeoneormorethanone.

Thereareonlytwocontextsinwhichtheuseofthesuffixmen isobligatory. Oneis whenapersonalpronounhasmorethanonereferent.

Firstperson Secondperson Thirdperson

Singular wo ‘I’ ni ‘you’ ta ‘he/she/it’‘they’(nonhuman)

Plural wo-men ‘we’ ni-men ‘youpl.’ ta-men ‘they’

Table3.1PersonalPronounsinMandarin

Thethirdpersonsingularpronoun ta canhavebothhumanandnonhumanreference.Thereare three characters to represent respectively the third person male, female human pronouns and nonhumanpronoun.Whenthesuffixmen is added, ta-men can only have human reference.

50

Whenareferenceismadetomorethanonenonhuman object, ta is used without the suffix

men 7. Example(4)isadaptedfromIljic(2001a.:19).

(4) Zhe-xie zha-zi wo ka-huan le, ni ba ta na-qu ba. These magazines I readfinishPerf you prep.it takeaway SFP. ‘Ihavefinishedreadingthesemagazines,youcantakethemaway.’

In(4),thesingularformofthethirdpersonpronoun ta ‘it’refersto‘thesemagazines’.

Thesecondcontextiswhendirectlyaddressinganaudienceofmorethanoneperson.For example,

(5) tongzi-men, zaoshang hao . ComradeColmorning good ‘Goodmorning,folks!’

The -men can also be suffixed to proper names. In such a case, the nominal usuallydoesnotmean‘agroupofpeoplehavingthesamename’.Itmeans‘agroupofpeople consistingofthepersonreferredtobythepropernameandothers’(6)oritmeans‘agroupof peoplewhohasthequalityofthepersonreferredtobythepropername’(7).

(6) xiao-Li- men little Li –Col ‘LittleLiandhiscompanions’

7 Sometimesmen issuffixedtothenonhumanpronoun ta whenreferenceismadetotwoormoreanimateentities.

51

(7) Lei Feng-men xianzai bu jian le .8 LeiFengCol now not see SFP ‘PeoplelikeLeiFengcannotbefoundanymore.’

In the following contexts, -men cannot be used. First, only nouns referring to human beingscanbesuffixedwith -men and -men cannotbeaffixedtononhumannouns(i.e.nouns referring to animals or objects). So we can say xuesheng-men ‘students’, but we cannot say dongwu-men ‘animals’ or zuozi-men ‘tables’. However, in fables or children’s literature, the suffixmen sometimesisaddedtonounsreferringtoanimalsorinanimateobjects.Insuchcases, theuseofmen isusuallyregardedasarhetoricaldevice:personification.

Second,thesuffixmen canbeusedwithonlysomehumannouns.Monosyllabicwords tendnottobesuffixedwithmen (Li&Thompson1981;Zhang,2001).Wecansay xiaote-men

‘thieves’ but cannot say zai-men ‘thieves’.Somecollectivenounscannotbesuffixed, as their collectivemeaningisalreadyconspicuous.Wecannotsay gongzhong-men ‘thepublic’.

Third,nounsthatarepredicatescannotbesuffixedwithmen ashumannounssuffixed with men are referential and predicate nouns cannot be referential. Consider the following examples.

(8) Wo-men shi xuesheng . ICol bestudent ‘Wearestudents.’

8 Lei Feng usedtobeahouseholdname.Hewasverygenerousandwillingtohelpothersattheexpenseofhimself,evenatthe riskofhislife.Hediedathisjobattheageof22.

52

(9) *Wo-men shi xuesheng-men IColbestudentCol

Fourth,nominalexpressionswithgenericreferencecannotbesuffixedwithmen .

(10)* Xuesheng-men yinggai xuexi . StudentColshouldstudy Intendedreading‘Studentsshouldstudy.’

Sentence(10)is grammaticalifthenominalreferstoaparticulargroupofstudents.Soitcan mean‘Thestudentsshouldstudy’.

Fifth,nounssuffixedwithmen arenotcompatiblewithexactcounting.

(11)* san ge xuesheng-men three Cl student Col

As(11)illustrates,theheadnouncannotbesuffixedwithmen whenthe[Nume+Cl]doesan exactcounting.However,wheneverthe[Nume+Cl]expressesapproximatequantityorwhen“ yi

(‘one’)+collectiveCl”isused,theheadnouncanbesuffixedwithmen .

(12) yi qun xuesheng(-men) aClstudent(Col) ‘agroupofstudents’

53

(13) san wu ge ren(-men) three five Cl person(Col) ‘severalpeople’

Both yi qu (‘onecollectiveclassifier’)in(12)and san wu ge (‘threefiveindividualclassifier’) in(13)expressanapproximatequantity.Whenanumeralotherthan yi isusedwithacollective classifier, then -men cannot be suffixed to the head noun as the ungrammaticality of (14a) indicates.

(14)a.* san qun xuesheng-men threeClstudent(Col) b. san qun xuesheng threeClstudent ‘threegroupsofstudents’

Fromtheaboveexamplesoftheuseofmen innominalexpressions,wecanconclude that men isnotreallyapluralmarker,butacollectivemarker.Another convincingargument againsttheviewthatChinesehasagrammaticalcategoryofnumberisthatChinesebarenouns havepluralreference. Ifthereisagrammaticalcategoryofnumber,thentheuseofaplural markershouldbecategorical.However,theuseofmen inChineseinmostcontextsisoptional andinsomecontextsisprohibitedevenwhenthereferentsofanominalaremorethanoneasthe aboveexamplesdemonstrate.Fromthecontextsoftheuse(andnonuse)of -men ,wecanalso concludethatnounssuffixedwith -men mustbereferentialandcannothavegenericreference.

54

Whataboutmen usedwithpersonalpronouns?In generallinguistics, some scholars

(Wales,1996:10)observethat“numberinpronounscharacteristicallyoperatesratherdifferently, signifying‘X+others’,ratherthan‘X+X,etc.’” Wo-men (ICol,meaning‘we’)inMandarinand we inEnglishrefertoagroupofpeoplewiththespeakerasthecenter. We inEnglishdoesnot meanI+I+I….InEnglish,thepluralpersonalpronounsdonotusetheregularpluralsuffix -s atall.

PersonalpronounsintheWenlingdialectalsoshowthatthesocalledpluralinpronouns is actually not plural, but really a collective marker. There is no marker corresponding to

Mandarinmen intheWudialect.So,evenhumannounsintheWudialectdonothaveasuffix.

ButthereisawaytoindicatepluralitywithpersonalpronounsintheWenlingdialect.Fromthe compositionofthepluralpersonalpronounsinthe Wenling dialect, we can say the socalled pluralpronouns we consistsof‘I+thethirdparty’, you (pl.)‘you(sg.)+thethirdparty’and they

‘he+thethirdparty’.

Firstperson Secondperson Thirdperson ngo ‘I’ n ‘you’ gei ‘he/she’ Singular ngo-deng ‘I + others’ n + yi ‘you + gei+yi ‘he + he’ Plural ‘inclusivewe’; he’‘youpl.’ ‘they’ ngo-yi ‘I + he’ ‘exclusivewe’

Table3.2:PersonalPronounsintheWenlingDialect

55

Basedontheabovedescribedlinguisticfacts,IassumethatChinesedoesnothavethe grammaticalcategoryofnumber.

3.2.2 A Language without Articles

Chineseisusuallyconsideredanarticlelesslanguage(Norman,1988;Chierchia,1998a, b). Recently some scholars view ‘ yi (one) + classifier’ as an indefinite article (Xu, 1997) or indefinitedeterminer(Chen,2003)andHuang(1999)arguesthatthedistaldemonstrative na-ge

(literallymeaning‘that+classifier’)isbeinggrammaticalizedintothedefinitearticleinspoken

Chinese. Without articles, bare nouns are prevalent in Chinese. They can be interpreted as definite,indefinite(specificandnonspecific)orgeneric,dependingoncontext.Thesedifferent interpretations of nominal expressions are encoded by articles in languages like English and

French.

(13) Wo mai shu le . IbuybookSFP ‘Iboughta/thebook(s).’ (14) shu hen gui . book very expensive ‘(The)book(s)is/areveryexpensive.’

Inexample(13),thebarenoun shu (‘book)canhavedefiniteandindefiniteinterpretations,while shu inexample(14)canhavedefiniteandgenericinterpretation,butnotindefiniteduetothe featureofsubjectsinChinesetendingtobedefinite.

56

3.2.3 A Numeral-classifier Language

In Chinese, nouns cannot be modified directly by a numeral. Either a classifier or a measure word must occur between the numeral and the noun. Hence, it is called a numeralclassifier language.9 In English numerals directly precede count nouns (like ‘ seven cows’ ) and only mass nouns need measure words in order to be counted (like 10 gallons of gas ).10 So, on the surface, all nouns in Chinese behave like mass nouns in English. The obligatory use of a classifier/measure word in quantifiednominalsisoneofthemainreasons

Chierchia(1998a,b)givestosubstantiatehisideathatallnounsinChinesearemassnouns.

(15) si ben shu fourClbook ‘fourbooks’ (16) si xiang shu four box(MW) book ‘fourboxesofbooks’ (17) *si shu fourbook

9 ClassifiersalonecannotbeusedwithnounsinMandarin.Theyhavetobeprecededbyanumeralordemonstrative.Whenthey areprecededbyademonstrative,itcanbearguedthatthenumeral yi ‘one’isdeleted,asthereisnomeaningdifferencewhether thenumeral yi occursbetweenthedemonstrativeandtheclassifier.ButtheuseofclassifiersinsomeSoutherndialects(like CantoneseandsomeWudialects)isdifferent;theclassifieralonecanbeusedwiththenoun.Wewilldiscusstheuseof classifiersinChineseinChapters4and5. 10 OneEnglishformveryclosetoChinesenumeralclassifierstructureissomethinglike‘sevenheadofcattle’(Greenberg, 1973).

57

In(15)and(16),theclassifier ben andmeasureword xiang ‘box’areusedrespectively.Without them,thenominalswillbeungrammaticalas(17)indicates.

3.2.4. A Topic-prominent Language

Sentence structure in Chinese is different from that in English. The basic Chinese sentence structure is topiccomment rather than subjectpredicate (Chao, 1968). The topic is somethingaboutwhichthecommentsayssomething.Thesubjecthasa semanticrelationship withthepredicate,butthetopicdoesnothaveto.Inexample(18)(takenfromLi&Thompson

1981:15;theEnglishtranslationismine),thetopicis zhei ke shu ‘thistree’andthecommentis yezi hen da ‘leavesareverybig’,thesubjectis yezi ‘leaves’and hen da ‘verybig’isthepredicate.

Thetopic zhei ke shu ‘thistree’hasnogrammaticalrelationwiththepredicate hen da ‘verybig.’

(18) zhei ke shu yezi hen da . thisCltreeleafverybig ‘Asforthistree,itsleavesareverybig.’

Li & Thompson (1981) classify languages into two types: topicprominent and subjectprominent. Inatopicprominentlanguagethetopic,notthesubject,isacrucialpartof the sentence, while in a subjectprominent language the subject is an essential part of the sentence. Chinese is a topicprominent language and English is an example of the subject prominentlanguage.So,aChinesesentence,unlikeanEnglishsentence,doesnotalwaysneed

58 anovertsubject. Thisisillustratedinexample(19)inwhichtheobjectistopicalizedandthere isnosubject.

(19) wufan hai meiyou chi . lunchstillhaven’teat ‘(I)haven’thadlunchyet.’

AsthesubjectisessentialinEnglish,a‘dummy’ it isusedasaformalsubjectwhenthere isnologicalsubjectbearingasemanticrelationwith the verb in a sentence, as example (20) indicates.

(20)It’sverycold.

TheChinesecounterpartof(20)is hen leng (literally‘verycold’)whichlacksasubject.

Thetopicofasentencealwayscomesfirstandalwaysreferstosomethingaboutwhich thespeakerassumesthelistenerhassomeknowledge(Li&Thompson,1981).Whenthereisno elementtopicalizedinasentence,thepreverbalsubjectbecomesthesentencetopicbydefault.

So the preverbal noun in Chinese is usually definite (Chao, 1968; Li, 1998). The following oftcitedexamplesillustratethispoint.

(20) keren lai le . guest come Perf ‘Theguest(s)has/havecome’.

59

(21)?* yi-ge keren lai le . OneClguestcomePerf. ‘Oneguesthascome’

Thebarenoun keren (‘guest’)in(20)mustbeinterpretedashavingdefinitereference.(21)is usuallyconsideredungrammaticalintheliterature,becauseofitssubjectbeingindefinite. For the same reason, the [Cl + N] phrase occurring in subject position in the Wu dialect and

Cantonesemusthavedefinitereferenceandwhenitoccursinobjectposition,itcanhaveboth definiteandindefinitereference.ConsiderthefollowingexamplesfromtheWenlingdialect.

(22) ben shi youqu . Clbookinteresting ‘Thisbookisinteresting.’ (23) ngo mai le ben shi. IbuyPerfClbook ‘Iboughta/thisbook.’

Ben shi in (22), functioning as the subject and topic, can only have a definite interpretation, while ben shi in (23), occurring after the verb, can have both definite and indefinite interpretation.

In this chapter, I briefly described some linguistic features of Chinese nominals.

ComparedtonominalsinlanguageslikeEnglish,theChinesenominalhasthefollowingfeatures: no plural marker and no articles; the mandatory use of a classifier or a measure word in a numericallyquantifiednominal;anddefinitepreverbalnominals.

60

CHAPTER FOUR

THE CLASSIFIER SYSTEM IN CHINESE

Theclassifiersystemisoneofthemainfeaturesofnumericallyquantifiednominalsin modernChinese.AswementionedinChapterThree,anumeralcannotprecedeanounwithouta classifierorameasurewordinbetween.As(1)indicates,thedeletionoftheclassifierleadsto ungrammaticality.

(1) san *(ben) shu threeClbook ‘threebooks’

Onlyinrarecircumstancescananumeralmodifyanoundirectly,suchasincompound words and idiomatic expressions which are residues of Old Chinese. In such cases, the numericallyquantifiednominalisnotusedreferentially.Considerthefollowingexamples:

(2) Taolun shi dajia qi zui ba she . discuss when people seven mouth eight tongue ‘Everyonetalkedatonceduringthediscussion.’

61

(3) san lun motuo che three wheel motor vehicle ‘threewheeledmotorcycle.

In (2), qi-zui-ba-she is an idiom, functioning as predicate and san-lun in (3) is part of a compoundword.However,inOldChineseanumeralcanprecedeorfollowanoundirectlyand theuseofclassifiersisveryrare.Thefollowingexamplesillustratethenumericalexpressionsin

OldChinese.Thenumeral yi ‘one’precedesthenoun niu ‘cow’in(4),whilein(5)thenumeral wu-bai ‘fivehundred’comesafterthenounc he ‘chariot’.

(4) Wu he ai yi niu ? I why love one cow ‘WhyshouldIbegrudgeoneox?’ ( Mengzi ,11 inPulleyblank1995:59) (5) Sang she wu bai . Lose chariot five hundred ‘lostfivehundredchariots.’ ( Zuozhuan ,12 inPulleyblank1995:59)

Theestablishmentoftheclassifiersystemisoneofthemajorchangesinthehistoryof the Chinese language (Wang, 1989). In this chapter, I first discuss the quantified nominal expressions in Old Chinese and trace the development of the classifier system of Chinese

11 Mengzi (knownas Mencius intheWest)isaclassicworkofConfucianismassembledbyMencius(c.372–289 BC)andhisimmediatedisciples.Mencius,betterknowninChinaas“MasterMeng”(Chinese: Mengzi )wasagreat ChinesethinkerandaforemostrepresentativeoftheConfuciantradition. 12 Zuozhuan aChinesehistoricalworktraditionallybelievedtohavebeenwrittenbyZuoQiuming,ahistorianof theWarringStatesperiod(475221BC).

62

(Section 4.1). In Section 4.2, I review the classification of classifiers in Chinese linguistic literature.ThefunctionsofclassifiersaredealtwithinSection4.3.Inthelastsection(Section

4.4),Idiscussthesyntacticstructureofnumericallyquantifiednominalexpressionsinmodern

MandarinChinese.

4.1 The Development of the Chinese Classifier System

Different scholars working on Chinese linguistics use the two terms “classifier” and

“measureword”differently;bothofthemcanoccurafteranumeralinChinese.So,Iwouldlike to first compare the two terms and clarify the distinctions between the two classes of words occurring after a numeral before dealing with the numericallyquantified nominals in Old

Chineseandtracingthedevelopmentoftheclassifiersystem.

4.1.1 Classifier and

Linguists in China never used the term “classifier” until recently. The term liang-ci

(literally‘measureword’)isusedtodesignatetheelementbetweenthenumeralandthenounina numericallyquantifiednominal.Theyusuallydonotmakeadistinctionbetweenclassifiersand measurewords.Chinesescholars(Lü,1980,amongothers) callrealclassifierslike ben in(6) geti-liang-ci (literally‘individualmeasureword’)andmeasurewordslike ping (‘bottle’)in(7) rongqi-liang-ci (literally‘containermeasureword’)andmeasurewordslike pang (‘pound’)in(8) duoliang-liang-ci (literally‘measuremeasureword’,meaning‘standardmeasureword’).

63

(6) san ben shu threeClbook ‘threebooks’ (7)san ping shui threebottle water ‘threebottlesofwater’ (8) san pang rou threepounds meat ‘threepoundsofmeat’

Thatpeopledonotdifferentiateclassifiersfrommeasurewordsmaybeduetothefactthatthey occupy the same syntactic slot in Chinese and that they both have the function of individualizationandmakenounsquantifiable.

Thecounterpartof“classifier”inChinese, fenlei-ci (literally‘classifyingword’),whichis stillunfamiliartomanylinguistsinChina,wasusedbyChineselinguistLü,Shuxiangin1980, whenhetranslatedChaoYuenren’s A Grammar of Spoken Chinese 13 intoChinese.Butwestern scholarsworkingonChineselinguisticsveryoftenusethetwoterms‘classifier’and‘measure word’interchangeably.Quiteoften,scholars(Matthews&Yip,1994:92;Vuori,2000:72)write thatclassifiersarealternativelycalledmeasurewordsinworksonChinese.Somelinguists(Chao

1968, and linguists in China) consider classifiers as a subcategory of measure words, while others(Li&Thompson,1981;Lyons,1977;Denny,1976) use classifier as an umbrella term encompassingmeasurewords. Chao(1968)considersclassifiersasindividualmeasuresandto

13 ThebookwaspublishedbyUniversityofCaliforniaPressin1968.

64

Li&Thompson(1981),“anymeasurewordcanbeaclassifier”.Lyons(1977:463)callsmeasure wordsas measural classifiers whichindividuateintermsofquantitiesandgenuineclassifiersas sortal classifiers whichindividuateintermsofthekindofentitythey refer to. When Denny

(1976:122)discussesthefunctionsofnominalclassifiers,hementionsthatsomeclassifiers(like ben in san ben shu meaning‘threebooks’)‘qualify’thenounandsome(like bag in‘threebags ofpotatoes’)‘quantify’thenoun. WhatDenny(1976)considersasclassifiersquantifyingthe noun are really measure words. The confusion in using ‘classifier’ and ‘measure word’ also occursinthelinguisticliteratureaboutnumeralclassifierlanguagesinSoutheastAsia(suchas

Thai,Burmese). Scholarsworkingonthoselanguagesusuallysubsumemeasurewordsunder classifiers,sothenumberofclassifiersisoftenexaggerated(Goddard,2005:16).

In general linguistics, classifiers refer to morphemes whose function is to mark the semanticclassthatcertainlexicalitemsbelongto(Crystal,2003:74).Itwouldbedesirableto distinguishclassifiersfrommeasurewordsindiscussingnumeralclassifierlanguages.Otherwise, amajordifferencebetweennumeralclassifierlanguagesandwesternlanguageswithregardto numericallyquantifiednominalexpressionswouldbeblurred,asmeasurewordsarealinguistic universal (Goddard, 2005). All human languages have measure words. For example, a nonclassifier language like English uses measure words to quantify mass nouns. What distinguishesquantifiednounsinChinesefromthoseinEnglishisthatclassifiersmustbeused betweennumeralsandcountnounsintheformerwhilenumeralsprecedecountnounsdirectlyin thelatter.Fromtheglossoftheexamplesinthischapter,wecanseemeasurewordsinChinese canbetranslatedintotheircounterpartsinEnglish,whileclassifierscannot.

65

InEnglish,nounsaredividedintocountnounsandmassnouns 14 .Countnounscanbe modifiedbynumeralsdirectly,butmassnounscannot.Massnounsmustbeindividuatedwiththe helpofameasurewordinordertobecounted.However,everynouninChinese,whetheritis countormass,mustcollocatewithaclassifierormeasure word to be numerically quantified.

BecauseofthisandthelackofthepluralmarkerinChinese,somewesternscholarsholdthe viewthatallnounsinChinesearemassnouns(Chierchia,1998a,b).Suchaviewmayberelated totheconfusionintheuseofthetwoterms‘classifier’and‘measureword’inChineselinguistics.

Actually,thereisstilladifferencebetweencountandmassnounsinChineseandthedifference isreflectedintheuseofclassifiers/measurewords(Cheng&Sybesma,1999,2005).

ClassifiersandmeasurewordsinChinesehavesomedifferentfunctionsandgrammatical featuresalthoughtheyalsosharesomefeatures.Intermsofsimilarities,theybothoccupythe samesyntacticslot,i.e.,afternumeralsandhave the function of individualization. Classifiers individualizecountnounsintermsofsemanticfeaturesandmeasurewordsindividualizemass nouns,andcountnounsinthatmatter,intermsofquantity.Butclassifiershavethefunctionof classification,besidesthatofindividualization.Aclassifierplacesintoonecategorynounswhose referentssharesomefeature.Forexample,theclassifier ba ,15 whoseoriginalmeaningis‘hold’, is used with count nouns (such as dao ‘knife’ , jiandao ‘scissors’ , yaoshi ‘key’ , qiang ‘gun’) whosereferents(usuallytools)haveahandleorareusedwithhands.AccordingtoTaiandWang 14 Somemassnounscanhaveacountreadingvia“UniversalPackager”(seePelletier,1979).Insuchacase,themassnounhasa kindreadingorareadingofaconventionalportionandcanbepluralized,e.g.Thestoresellsdifferentbeers.(=kindsofbeers) Hehadtwobeersbeforegoingtobed.(=twoportionsofbeer,twocupsofbeer). 15 Ba canalsobeusedasameasureword,meaning‘handful’,e.g. yi ba tang (onehandfulsweets,meaning‘ahandfulof sweets’.)

66

(1990: 38), “a classifier categorizes a class of nouns by picking out some salient perceptual properties, either physically or functionally based, which are permanently associated with the entities named by the class of nouns; a measure word does not categorize but denotes the quantityoftheentitynamedbyanoun.”

Intermsof,classifierscanbeusedonlywithcountnouns;measurewordscan beusedwithbothcountandmassnouns.Whenameasurewordisusedwithacountnoun,then theindividualityoftheentitydenotedbythenounisdeemphasizedandtheentityistreatedlikea mass. This is the same with the use of measure words with count nouns in English. For example,inEnglish,ameasurewordlike‘box’canbeusedwithbothamassnounlike sand as in(9)andacountnounlike book asin(10):

(9) aboxofsand (10) aboxofbooks

Inthesameway,themeasureword xiangzi (‘box’)inChinesecanbothbeusedwithcountnouns like shu (‘book’)asin(11)andmassnounslike shazi (‘sand’)asin(12):

(11) yi xiangzi shu oneboxbook ‘aboxofbooks’ (12) yi xiangzi shazi oneboxsand ‘aboxofsand’

67

Butthe classifier ben canonlybeusedwithcertaincountnounslike shu (‘book’)or biejiben

(‘notebook’) asin(13) andcannotbeusedwithany mass nouns like shui (‘water’) or shazi

(’sand’)astheungrammaticalityof(14)indicates.

(13) san ben shu/biejiben threeClbook/notebook ‘threebooks/notebooks’ (14)* san ben shui/shazi threeClwater/sand

Withregardtotheirinventories,measurewordsarelargerinnumber.Althoughmeasure wordsaresaidtoformaclosedclass,manycommonnounscantemporarilyfunctionasmeasure words in context, like wuzhi (meaning ‘room’) in yi wuzhi ren (literally ‘one room person’, meaning ‘a roomful of people’). The number of classifiers is limited and constant; no new classifiersarecreatedwhennewthingsareinventedandnounsarecreatedtodenotethem.For example, the noun dian-nao (literally ‘electric brain’, meaning ‘computer’) has a very short historyandtheclassifierusedwithitis tai ,whichisaclassifiercollocatingwithnounsreferring to machines. Estimates of the number of Chinese classifiers vary greatly, from over sixty

(Mathew&Yip,1994:92)toalmostonehundredandforty(Wang,1994).

Semantically,theoriginallexicalmeaningofmanyclassifiersisnolongertransparentto nativespeakers,althoughoriginallytheyarederivedfromlexicalwords,mostlynouns.Some classifiers do not have any lexical meaning now. For example, the general classifier ge was

68 derived from the noun zhu whichmeant‘’inOldChinese.Itstartedto be used as a classifier for counting . Now it has grammaticalized as a general classifier and its originallexicalmeaninghasbeenbleachedout.

Morphologically,classifiersareboundmorphemesinmodernChinese.Measurewords, on the other hand, have lexical meaning and can be used as an independent word in other contexts.The ping ‘bottle’in(15)isameasureword,butitisanounin(16).

(15) yi ping shui one bottle water ‘abottleofwater’ (16)yi zhi ping oneClbottle ‘abottle’

Ping ‘bottle’in(16)hasitsownclassifier zhi whenusedasanoun.

Syntactically,therearetwodifferencesbetweenclassifiersandmeasurewordsinChinese that are quite often discussed in the literature. The first difference is that the modification marker de canbeinsertedbetweenthemeasurewordandthefollowingnounasshownin(20), butcannotbetweentheclassifierandthenounasshownin(18).

(17) san ben shu threeClbook ‘threebooks’

69

(18) * san ben de shu threeClDEbook (19) san pang rou threepoundmeat ‘threepoundsofmeat’ (20) san pang de rou threepoundDEmeat ‘threepoundsofmeat’

Thesecondsyntacticdifferenceisthatmeasurewordscanbemodifiedbyanadjective, but classifiers cannot. In (22), the adjectives da/xiao (‘big/small’) are inserted between the numeralandthemeasureword ping (‘bottle’),buttheycannotbeinsertedbetweenthenumeral andtheclassifier zhi in(24)(fromCheng&Sybesma,1999).

(21) yi ping shui one bottle water ‘abottleofwater’ (22) yi da/xiao ping shui one big/small bottle water (23) yi zhi gou oneCldog ‘adog’ (24)* yi da/xiao zhi gou one big/small Cl dog

70

Thefollowingtablesummarizestheabovediscussedsimilaritiesanddifferencesbetween classifiersandmeasurewordsinChinese.

Classifiers MeasureWords

Individualizingfunction Yes Yes

Classifyingfunction Yes No

Collocation (with)countnouns (with)massandcountnouns

Inventory small (relatively)big

Semantics no/littlelexicalmeaning fulllexicalmeaning

Morphology boundmorpheme independentmorpheme

Premodificationbyadjectives Impossible 16 possible

Insertionofthemarker de No Yes

Table4.1:DifferencesbetweenClassifiersandMeasureWords

4.1.2 Quantified Nominal Expressions in

In Old Chinese, there is no classifier and numerals can come directly in front of nouns.

AccordingtothedistinguishedChineselinguistLiWang(1958),therearethreewaystoexpress quantitiesofobjectsdenotedbynounsinOldChinese.Thefirstwayistocombinethenumeral andthenounwiththenumeralinfrontofthenounlike(25).Thesecondisthatthenumeral followsthenounlike(26). 16 Actually,someclassifierscanbemodifiedbyadjectives.Forexample, yi da ben shu (literally‘abigClbook’).

71

(25) san ren xing bi you wo shi . three person walk must have I teacher ‘Whenthreepeoplearewalkingalong,onecansurelylearnsomethingfromtheother two.’ (AnalectsofConfucius 17 ) (26) fu shi baiwan . prostrate corpse million ‘Theprostratecorpsesareamillion.’

Thethirdisthatthenumeralcomesafterthenounandthenumeralisfollowedbya danwei-ci

(literally‘unitword’,Wang’s(1958)terminologyforclassifiers/measurewords)like(27).

(27) chang wu you milletwine five vase ‘fivevasesofmilletwine’

Wangusestheterm‘unitword’whichcreatesaunitoftheentitydenotedbythenounsothatit canbequantifiedanddoesnotmakeadistinctionbetweenaclassifierandameasureword. The firststructureisthepredominantoneandtheothertwoareveryrare. Whenameasurewordis used,thethirdmethodisemployed,thatis,the[Nume+MW]combinationcanonlyoccurafter thenoun.

Although many scholars hold the view that numerals or [Nume + MW] combinations occurringafteranounfunctionasamodifierofthenouninOldChinese(Wang,1958;Dobson 17 Compiledbyhisdisciples,theAnalectsofConfuciusisarecordofthegreatChinesethinkerandeducator Confucius’s(BC551BC479)wordsandofthediscussionsamonghimandhisdisciples.

72

1962),itcanbearguedherethattheyfunctionasapredicate. Letusconsidersomelinguistic factsofnominalsinOldChinese.

First,theruleofamodifieroccurringbeforethemodifiedinChinesehasneverchanged throughout its long history (Shi & Li, 2002: 13). Just like in modern Chinese, a possessive pronoun(28)orademonstrative(29)occursbeforeanouninOldChinese(Pulleyblank1995).

(28) wo min 18 Ipeople ‘ourpeople’ (29) zi jiu this wine

Whilealltheothermodifiersarepremodifiersinnominalexpressions,numeralsandthe[Nume

+ MW/Cl] combination would be exceptions if they were considered as modifiers when occurringafternouns.

Second,ifthenumeralandthe[Nume+MW/Cl]combinationarepostmodifiers,then whycannumeralsoccurbothbeforeandafternouns,butnot[Nume+MW/Cl]combinationsin

Old Chinese? It is very difficult to explain why [Nume + MW/Cl] combinations cannot function as premodifiers of nouns. If we consider the numeral and the [Nume + MW/Cl] combinationaspredicates,thenthisdifficultywillnotarise.

18 InOldChinese,amarkerconnectingthepossessivepronounandtheheadnounisoptional;alsothepersonal pronoundidnotinflectforplurality.

73

Third,nounsfollowedbynumeralsor[Nume+MW/Cl]combinationsdidnotoccurin thesubjectpositioninOldChinese.Accordingto LiandShi’s(1998: 49)observationofthe examplesofclassifieruseinOldChineseinWang(1990),thereareseventeenexamplesof[N+

Nume+MW/Cl]occurringinobjectpositionandnoexampleinsubjectpositionfromworks writtenbeforetheNorthernandSoutherndynasties(AD420to589).The[N+Nume+MW/Cl] combinationinOldChineseeitheroccurredindependentlyasasentenceorasanobject.When theyoccurredindependentlyasasentence,the[Nume+MW/Cl]combinationwasapredicate.

When they occurred in object position, the [Nume + MW/Cl] combination functioned as an objectcomplement.InOldChinese,thelexicalmeaningoftheveryfewclassifiershadnotbeen bleachedoutandthe[Nume+MW/Cl]combinationwasreallyanounphrase.

Fourth,whenthe[Nume+MW/Cl]combinationoccursafteranoun,anotherwordcould be inserted between them. For instance, in (30), the adverb ge (meaning ‘respectively’, a differentwordfromthegeneralclassifier ge)comesbetweenthenounandthenumeral.

(30) ma niu ge shi pi . horse cow respectively ten Cl ‘horsesandcowsareteninnumberrespectively’

Ifthe[Nume+Cl]combination shi pi andthenouns ma and niu formedanominalconstituentin

(30),thentheadverbcouldnothavebeeninserted. Here,the[Nume+Cl]combination shi pi shouldbeanalyzedasapredicateandthenouns ma and niu functionasthesubject.

74

Based on the above linguistic facts, numerals or [Nume + MW/Cl] combinations occurringafternounsshouldbebetterregardedaspredicates,notaspostmodifiersofthenoun.

Theydonotformanominalconstituent.Therelationshipbetweennumeralsor[Nume+MW/Cl] combinationsandtheprecedingnounsisthatbetweensubject/objectandpredicate.When[N+

Nume]or[N+Nume+MW/Cl]occurindependently,the numeralor the[Nume+MW/Cl] combinationpredicates aboutthesubjectnoun.Whentheyoccuraftertheverb,i.e.inobject position,thenumeralorthe[Nume+MW/Cl]combinationpredicatesabouttheobjectnoun.

Actually,inmodernChinese,the[Nume+MW/Cl]combinationoccursafterthenounin inventoriesorlistings,andChao(1968)regardsthe[Nume+MW/Cl]combinationasanominal predicatewiththenounassubject. Forexample,

(31) zuozhi wu zhang, yi zhi qi ba tablefiveClchairsevenCl ‘fivetables,sevenchairs’

Tang (1996) also discusses the occurrences of [Nume +MW/Cl]combinationafterobjectsin modernChineselike:

(32) w o mai le bi san zhi . IbuyPerfpen threeCl ‘Iboughtthreepens.’

75 andtreats san zhi in(32)asapredicateoftheobject bi (‘pen’).Ofcourse,thisuseof[Nume+

MW/Cl]followingnounsisveryrareinmodernMandarin.

Inthissubsection,IarguedthattheclassifierinOldChinesewasnotgrammaticalizedintoa functional category. The numeral or the [Nume + MW/Cl] functions as a predicate and the measurewordsandtheveryfewclassifiersstillretainedtheirlexicalmeaninginOldChinese.In thenextsubsection,IwilltracethedevelopmentoftheclassifierinChinese.

4.1.3 The Rise of Classifiers as a Functional Category

TheChineselanguagedoesnothaveclassifiersinitsearlyhistory. WhyhasChinese developedtheclassifiersystem? Erbaugh(1986)attributesphonemicmergersasthetriggerof the historical development of classifiers in Chinese. Because of phonological change, many distinctivespeechsoundsinOldChinesewerenolongerdistinctive.Asaresult,manywords became homophonous and classifiers were derived to facilitate communication. On the other hand,Erbaugh(1986)hypothesizesthatclassifiersmaynotbenativetoChineseandcouldhave been borrowed from Tai languages through language contact. Her hypothesis is based on the followingtwofacts:(1)Chineseclassifieruseisvariableratherthancategorical.Thereisgreat variationamongspeakersinusingclassifierswithparticularnouns,soshebelievesthatChinese classifiers“neverbecomeafullyautomaticpartofthegrammaticallevelofthelanguage.”(2)

UnliketheThailanguage,Chinesehasneverhadtheanimate/inanimate,human/nonhumansplit inclassifiers anddoesnothavethefinely graded distinctions which she thinks are centraltoThaiandmostotherclassifierlanguages.Thishypothesisapparentlycanexplainthe

76 phenomenonthatsouthernChinesedialectshavemoreclassifiersthanMandarinChineseasthere might have existed more close language contact between southern dialects and Tai languages

(Peyraube,1991).

IagreewithErbaugh(1986)thatChineseclassifiersweremotivatedtoavoidambiguity andfacilitatecommunication. Ialsosharetheviewthatthereismoreintensecontactbetween southerndialectsandTailanguages.However,theChineseclassifiersystemwasnotnecessarily borrowedasErbaughandsomeotherscholarsassumed(Erbaugh,1986;Jones,1970).Erbaugh’s firstreasonforherborrowingviewthatChineseclassifiershaveneverbecomeafullyautomatic partofthe grammatical levelbecauseof greatvariationintheuseofclassifiersamongnative speakersisnotveryconvincing.Whileitistruethattheuseofclassifierswithsomenounsvaries fromdialecttodialectandthatspeakersarenotcertainordisagreeamongthemselvesaboutthe properclassifierswithsomenouns,itdoesnotmeanthatclassifiersarenotafullyautomaticpart ofChinesegrammar. Thataclassifiermustbeusedinaquantifiednominaldemonstratesthat theclassifierispartofChinesegrammar.Differentclassifierscanbringoutdifferentaspectsof themeaningofthenounstheycollocatewith.Thatisthereasonwhyseveralclassifierscanbe usedwithonesinglenounandthenspeakersmaydisagreeabouttheproperclassifierswithsome nounswhentheymaypayattentiontodifferentfeaturesoftheobjectsthenounsreferto.

AstothereasonthatChinesedoesnothavetheanimate/inanimate, human/nonhuman splitinclassifiers,itisnotcompletelycorrect.AlthoughChinese,unlikeThai,doesnothavea specific classifier used with animal, inanimate, human or nonhuman nouns, there are some classifiers used exclusively with human or nonhuman nouns while others can be used with

77 differenttypesofnouns.Forexample,theclassifier wei isspecifictohumannounswhilethe generalclassifier ge isusedwithdifferenttypesofnouns.ThefactthatChinesedoesnothavethe finely graded honorific distinctions in classifier use cannot count as an argument for the borrowingoftheChineseclassifiersystemfromThai.Theuseofclassifiersindifferentcultures maybeculturespecific.Forexample,thereisaspecificclassifierusedwithelephantsinThai,as theanimalisaveryimportantpartoftheirlife.InChinese,theearlyclassifierswerefirstused withimportantanimalsorobjectsinthethencircumstances. The classifiers pi (used with ma

‘horses’) and ge (now a general classifier, first used in counting zhu ‘bamboos’, then jian

‘arrows’)wereusedlongbeforetheestablishmentoftheclassifiersystem.Horses,bamboosand arrowswereveryimportantintheearlyChinesehistoryduringwhichwarswerecommonand chariotsandhorseswereveryimportantvehiclesandarrowswereveryimportantweapons.

AsWang(1994)argues,thenounsorverbsfromwhichChineseclassifierswerederived wereamongtheChinesevocabulary.ThesewordswerefoundinChinesehistoricaldocuments writtenbeforetheiruseasclassifiersinChinese.ThereisnosolidevidencetoshowthatChinese classifierswereborrowedfromThai.Theearliestknown Thai writing is the Sukhathai script inventedin1283A.D.(Danvivathana,1987,citedinWang,1994),whileChinesewritingsystem datesbacktomorethanthreethousandyearsago.Inaverylonghistory,theChinesepeopleand language were very influential among East Asian and Southeast Asian countries. Many neighboringlanguageswereinfluencedbytheChineselanguage.EvenwhentheChinesepeople wereruledbynonChinesespeakingnationalitiessuchasMongoliansandManchu,theChinese language did not borrow many words from the rulers’ languages (Wang, 1994). Historically,

78

Chinese is a “monolithic ethnic and linguistic entity, highly resistant to any sort of outside culturalorlinguisticinfluence”(Norman,1988:16).

WhendidtheclassifiercomeintobeinginChinese?Differentscholarshavedifferent answers.Foronething,earlyclassifierswerenotfullygrammaticalizedandtheyhadlexical meaning.So,somescholarsjustconsiderearlyclassifiersasnounsmodifiedbynumerals.For another,somescholarsdonotmakeadistinctionbetweenmeasurewordsandrealclassifiers.

MeasurewordswereusedveryofteninOldChinese,althoughtheywerenotmanyinnumber beforethemeasurementsystemwasfirstestablishedduringQindynasty(221to 207BC).

Although it is generally agreed that protoclassifiers can date back to three thousand years ago, the grammatical category of classifiers was established much later. The vast majorityofquantifiednominalexpressionswereofthe[Nume+N]structureinOldChinese.

ThetwoclassifiersthatwereusedinOldChineseandoftencitedinChineselinguisticliterature onlyoccurredafterthenounduringtheearlydevelopmentofclassifiers.Thesetwoclassifiersare pi appliedtohorsesand ge tobamboosandarrows.

(33) ma shi pi .

horsefourCl ‘fourhorses’ (34) fu shi wu-shi ge . carry arrows fiveten Cl ‘carryfiftyarrows’

79

According to Peyraube (1991), real classifiers probably began to appear in the first centuryB.C.Eventhen,theystillretainedmanyfeaturesofthecommonnounsfromwhichthey derived. When classifiers were first used with numerals, the [Nume + Cl] combination was always after nouns and we have argued in Subsection 4.1.2 that [Nume + Cl] combinations functionedasapredicatewhentheyfollowednouns.The[Nume+Cl+N]pattern,i.e.[Nume+

Cl]precedingnouns,startedtooccurinthefirstcenturyB.C.(Shi&Li,1999). Duringthenext one thousand years [Nume + N] was still the predominant pattern of quantified nominal expressions.Shi(2002:149)observedthat[Nume+N]accountfor88%and[Nume+Cl+N]

12%ofthenominalexpressionsin Shishuo Xinyu (writteninAD425).In Laoqida (writtenin

AD1325),the[Nume+Cl+N]patternaccountsfor98%ofallnominalexpressions.Sowecan say that the classifier system was probably established by the 14 th century. However, the

[Nume+N]patternisstillquitecommoninsomeofthenovelswritteninthe17 th or18 th century.

LiteraryChinesehadlongbeenemulateduntiltheNewCultureMovement(19171923),sothe written language may not reflect accurately the speechoftheperiod whentheliteraryworks werewritten.Wecanonlyspeculatethatclassifiersbecameobligatoryinnumericallyquantified nominalsinthe14 th century.

When classifiers were first used, they occurred after nouns. Then a syntactic change occurred in Chinese in sentence structure and the [Nume + Cl] started to precede nouns.

AccordingtoLi&Shi(1998:40),thesentencestructureinOldChineseis:subject+verb+ object+X. TheXelementcanbeanadjective,intransitiveverb,aprepositionalphraseora

[Nume+Cl]combination. TheXelementfunctionsasasecondarypredicate.Lateron,there

80 occurred a change in the sentence structure and the X position gradually disappeared. The prepositionphrasemovesbeforetheverb;theintransitive verb merges with the main verb to formacompoundverb;theadjectiveandnumeralclassifierbecomesapremodifieroftheobject noun.Thatisthereasonwhythe[Nume+Cl+N]firstoccurredintheobjectposition.When classifiersbecamemorefullygrammaticalizedandobligatoryinquantifiednominalexpressions, then they started to occur in the subject position. Only when classifiers are obligatory in numericallyquantifiednominalshastheclassifiersystembeenfirmlyestablishedintheChinese language.

4.2 Classification of Chinese Classifiers

SincemostscholarsworkinginChineselinguistics donotmakeadistinctionbetween classifiers and measure words, our review here of classifications of classifiers in Chinese linguistic literature concerns both classifiers proper and measure words.19 Classifiers can be classifiedintodifferenttypes,dependingontheirfeaturesandthefeaturesofthewordsthatthey collocatewith.

Classifiers in modern Chinese can collocate with both nouns and verbs. When a classifierisusedwithanoun,itiscalledanominalclassifier.Thoseusedwithverbsarecalled verbalclassifiers.Theyindicatethetimesthataneventtakesplace. Nominalclassifiersoccur 19 Insubsection4.1.1,Imadeadistinctionbetweenclassifiersandmeasurewords.Therearesomeotherwords whichfillthesamepositionasclassifiers,butdifferfromtheclassifierproper.Thesocalledcollectiveclassifiers like qun ‘group’, ban ‘class’andkindclassifierslike zhong ‘kind/type’arealsousedinnonclassifierlanguages. Theyareincludedinthissectionforthesakeofcompleteness,butthedissertationmainlydiscussestheclassifier proper.

81 beforeanounandverbalclassifiersusuallyfollowaverb. Forexample,

(35) wo mai le yi ben shu . IbuyPerfone Clbook ‘I(have)boughtabook.’ (36) zhe ben shu wo kan le liang bian . thisClbookI seePerftwoCl ‘Ihavereadthisbooktwice.’

The classifier ben in(35)and(36)isanominalclassifierandtheclassifier bian in(36)isa verbalclassifier.Sinceverbalclassifiersarenotthetopicofthisdissertation,wewillnotdiscuss them.Wewillonnominalclassifiersinthischapter.

Withregardtotheclassificationofnominalclassifiers,researchersinChineselinguistics dividenominalclassifiersindifferentways,dependingontheirperspectivesandresearchtopics.

Toresolvetheconfusionintheuseofclassifiersandmeasurewordsandtohighlightthereality of a count/mass distinction in Chinese, Cheng & Sybesma (1999) coin two terms: count classifiers and massclassifiers. Countclassifiers (like ben in (35)) name the unit of natural semanticpartitioningandareusedwithcountnouns.Massclassifiers(like ping ‘bottle’in(7)) createaunitofmeasure.Cheng&Sybesma’smassclassifiercorrespondstotheterm‘measure word’usedbysomeotherscholarsandmassclassifiersdonothavetheclassifyingfunction.So, thetermmassclassifierisstillamisnomer.

T’sou (1976) divides nominal classifiers in Chinese into four types, based on the two features: [ ±exact] and [ ±entity]. The feature [ ±exact]isrelatedtowhetherthequantityofthe

82 objectdenotedbythenounisexactlydefinedornot.Thefeatureof[ ±entity]referstothequality oftheobject,thatiswhethertheobjectisadiscretephysicalentityandcountable.Thefourtypes ofclassifiersare:individualclassifiers(havingthefeaturesof[+exact]and[+entity]),measure classifiers (having the features of [+exact] and [entity]), collective classifiers (having the features of [exact] and [+entity]) and kind classifiers (having the features of [exact] and

[entity]). A noun can be used with different types of classifiers. The following examples illustratetheuseofdifferenttypesofclassifierswiththenoun yu (‘fish’).

a.individualclassifier tiao , yi tiao yu (meaning‘afish’)

b. measure classifier gongjin (‘kilogram’), yi-gongjin yu (meaning ‘a kilogram of

fish’)

c.collectiveclassifier qun (‘brood’/’group’), yi-qun yu (meaning‘aschooloffish’)

d.kindclassifier zhong (‘kind’), yi-zhong yu (meaning‘akindoffish’)

The above examples and their glosses illustrate that nonclassifier languages are lacking in individual classifiers and the other three types (i.e. measure, collective, kind classifiers) are sharedbybothclassifierandnonclassifierlanguages.

Huang&Kathleen(2003)makeadistinctionbetweenclassifiersandmeasurewordsand divideclassifiersintothreetypes:individual,kindandevent.Whenanounusedwiththethree typesofclassifiers,theywillcoercethreedifferentreadingsofthenoun(seeexamplesinSection

4.3.2).

83

Wewillusetheterm‘classifier’torefertotheindividualclassifierfromnowoninthis dissertationandthefollowingdiscussionaboutclassifiersmainlyconcernedwiththeindividual classifier.Whenaparticulartypeofclassifierisintended,wewillusethespecificterm.

4.3 Functions of Classifiers

4.3.1 Grammatical Functions

Classifiershavedifferentfunctionsindifferentlanguages.AccordingtoBisang(1999), therearetwooperationsofquantification:countingandmeasuring.Theoperationofmeasuring is applied to the quantification of mass nouns which are semantically unbounded, while the operation of counting quantifies semantically bounded nouns. Classifiers occur with counting andindividuateacountnounbydesignatingitssemanticboundaries(Bisang,1999:120).The maingrammaticalfunctionofclassifiersinChineseistoindividuatecountnounsbeforetheyare quantifiedbynumerals.ClassifiersmakecountnounsinChinesenumericallyquantifiable.

Then,classifierscanhavereferentializationfunction(usingBisang’sterm)ordefiniteness feature in some languages. Matthews & Pacioni (1997) argue that the use of classifier is associated with specificity in Cantonese, while Cheng & Sybesma (1999, 2005) argue that classifiers in Cantonese perform functions similar to those of the definite article in many

IndoEuropean languages. This function of classifiers in the southern Chinese dialects is not consistent.ClassifierscanmakeanominalexpressiondefiniteinCantoneseandsomeofWu dialectswithoutthehelpofdemonstrativeswhileclassifiersinMandarinChinesedonothave thisfunction.

84

(37) Gaan ngu hou leng . (Cantonese) Clhouseverypretty ‘Thehouseverypretty.’ (referentialfunction) (38) diao ng du xie . (Wenlingdialect) Clfishbigdegreeword ‘Thisfishisquitebig.’ (referentialfunction) (39) * diao yu hen da (Mandarin) Clfishverybig

Fromtheaboveexamples,wecanseethatnounsprecededbyclassifiersinCantonese(37) and Wenling dialect (38) have definite reference whentheyoccurbeforeverbs.Butasimilar sentence in Mandarin (39) is ungrammatical, as classifiers do not have the referentialization/ definitenessfunctionandsubjectstendtobedefiniteinMandarin.

Withregardtothereferentializationfunctionofclassifiers,Wusubdialectsvaryfromone to the other. In the case of Shanghaiese, the referentialization function of its classifiers is changing.InQian(1997),abookaboutShanghaiese,itismentionedthatclassifiershavedefinite reference when occurring in the subject position in the speech of the older generation. In

ChapterFive,whereIcomparethedistributionsandinterpretationsofthe[Cl+N]phraseinthe

WudialectandMandarin,IwillreturntothefunctionofclassifiersandusedatafromWenling:a southernWudialect.

Also,classifierscanhavearelationalizationfunction(Bisang1999).Classifiersareused toconnectthemodifierandtheheadinpossessives((40)Cantonese;(42)Wenlingdialect)and

85 adjectival constructions ((41) Cantonese; (43) Wenling dialect) and nominal expressions with relativeclauses((44)Wenglingdialect).

(40) Keoi gaan nguk (Cantonese) s/heClhouse ‘Her/hishouse’ (41) Hou daai gaan nguk ! verybigClhouse ‘(What)abighouse!’ (relationalizationfunction) (42) ngo ben shi youqu. (Wenlingdialect) I Cl book interesting ‘Mybookisveryinteresting.’ (43) Wo zuotian mai ben shu hen youqu . I yesterday buy Cl book very interesting ‘ThebookIboughtyesterdayisveryinteresting.’ (relationalizationfunction)

2.3.2 Semantic Functions

Semantically,thefunctionofclassifiersistoplaceobjectsintodifferent classes(Jenny

1976).Usuallyaclassifiercanbeusedwithseveraldifferentnounswhosereferentssharesome featureandtheclassifierpicksoutthisfeature.Letustaketheclassifier tiao forexample. tiao originallymeans‘branch’.ItisaboundmorphemeinmodernChinese.Asaclassifier, tiao is usedwithnounsreferringtolong,slenderobjects.So,wecansay:

86

(44) yi tiao sheng oneClrope ‘arope’ (45) yi tiao she oneClsnake ‘asnake’ (46) yi tiao malu oneClstreet ‘astreet’ (47) yi tiao heliu one Cl river ‘ariver’

The classifier tiao can even collocate with abstract nouns such as xinwen ‘news’.

Interestingly,theuseof tiao with xinwen ‘news’isrelatedtothehistoricalfactthatintheHan dynasty(206BC–AD220)beforethemethodofmakingpaperwasinvented,documentswere writtenonbamboostrips(Norman,1988).

Ontheotherhand,anounsometimescanbeusedwithdifferentclassifierstohighlight differentfeaturesofthereferentofthenoun.Forexample, shu (‘book’)canbothbeusedwith classifiers ben and bu . Ben referstoindividualbooks(emphasizingthefeatureof‘bound’),but bu emphasizesthecontentofindividualbooks.So san bu shu (‘threebooks’)meansthreebooks ofdifferent,i.e.,theircontentsaredifferent,but san ben shu (‘threebooks’)maymeanboth three books of different titles and three copies of the same (Huang & Kathleen, 2003).

Sometimes,peoplecanmanipulatetheuseofclassifierstoconveysubtledistinctionsofmeaning.

87

Forexample,wecanusetheclassifier zhi (whichisusuallyusedwithanimalnouns)withnouns referringtopeopletoshowourcontemptforthepersonreferredtobythenoun.

Huang & Kathleen (2003) convincingly demonstrated that different types of classifiers usedwiththesamenouncancoercedifferentreadings:individual,kind,andevent.Forexample, dianhua (‘telephone’)canhavedifferentreadingswhenusedwithdifferentclassifiers.

(48)y i bu dianhua one Cl telephone ‘onetelephone’ (49) yi zhong dianhua one Cl telephone ‘onekindoftelephone’ (50) yi tong dianhua one Cl telephone ‘onephonecall’

The individual classifier bu in (48) coerces an individual reading of dianhua (‘telephone’), emphasizingthefeatureof‘machinery’.Thekindclassifier zhong in(49)andtheeventclassifier tong in(50) coerceakindreadingandaneventreadingrespectively.

Differentindividualclassifiersusedwithanouncanalsogiverisetodifferentreadings.

Forexample, hua (‘flower’)canbeusedwithtwoindividualclassifiers duo and zhu .Theuseof the two classifiers can bring out the different meanings of the noun hua . Yi duo hua (one +

88 individualclassifier+flower’)means‘oneflower’,but yi zhu hua means‘oneplant’(Huang&

Kathleen,2003:361;Qian,1997).

Inthissection,wediscussedthefunctionsofclassifiers.Wewilldealwiththesyntactic structuresofnumericallyquantifiednominalsinMandarininthenextsection.

4.4 Numerically-Quantified Nominals

TheclassifierinChineseiscalledanumericalclassifierindiscussionsofthetypologyof classifiersingenerallinguistics(Allen,1977,andothers),becausethenumeralorclassifieralone cannotmodifytheheadnounbuttheyhavetobeusedtogethertoquantifythenoun.However, thisisnotveryaccurate,asaclassifiercanalsofollowademonstrative(51)withoutcollocating withanumeral.Aclassifiercanalsoreduplicatetofunctionindependentlyassubject(52)orasa quantifier(53).Theclassifierben in(52)usuallycollocateswith shu ‘book’.Whentheclassifier isreduplicated,thenominalhasadistributivemeaning.

(51) Zhe ben shu hen hou . thisClbookverythick ‘Thisbookisverythick’ (52) ben-ben dou youqu . ClCl all interesting ‘Everybookisinteresting.’/‘Allthebooksareinteresting.’ (53) ge-ge xuesheng du hen yonggong . ClCl student all very diligent ‘Everystudentisverydiligent.’

89

The term ‘numeralclassifier’ is all the more inaccurate when applied to some Wu dialects or

Cantonese, as classifiers in those dialects can precede nouns alone without numerals. In this section,Idiscussthedistributionandthesyntacticstructureofnumericallyquantifiednominals inMandarinChinese.The[Cl+N]expressioninMandarinandtheWudialectwillbetreatedin thenextchapter.

4.4.1 Distribution and Interpretation of Numerically-Quantified Nominals in Mandarin

WementionedinChapterThreethatsubjectsinMandarin(andotherChinesedialectsfor that matter) tend to be definite. Numerically quantified nominals are inherently indefinite, so theyareusuallylimitedtothepostverbalposition(54).(55),inwhichthesubjectisindefinite,is ungrammaticalinChinese.

(54) zuotian wo pengjian liang ge lao pengyou . yesterday ImeettwoCloldfriend ‘Imettwooldfriendsyesterday.’ (55)* liang ge xuesheng zai zhao ni . twoClstudentProgfindyou ‘Twostudentsarelookingforyou.’

Traditionally,itisdeemedthata[Cl+N]phrasecannotbeusedwithoutbeingpreceded by a numeral or a demonstrative in Mandarin. When the numeral is yi (‘one’) in quantified nominalsandtheyareinobjectposition,thenumeral yi canbedeleted,resultinginthesurface

90 formof[Cl+N](56)).Thesurfaceformof[Cl+N]isconsideredasacaseofphonological deletionofthenumeral yi .

(56) wo xiang mai ben shu . I want buy Cl book ‘Iwanttobuyabook.’

However,Cheng&Sybesma(1999)arguethatthe[ yi +Cl+N]phraseandthe[Cl+N] phraseareoftwodifferentstructuresinbothCantoneseandMandarin.The[Cl+N]patternis nottheresultofthephonologicalreductionofthenumeral yi (‘one’).Theybelievethatthetwo patternshavedifferentmeaningsanddistributions.Theindefinite[Cl+N]phrasecanonlyhave nonspecific reference, while the indefinite [ yi + Cl + N] phrase can have both specific and nonspecific reference. (The [Cl + N] in Cantonese and the Wu dialect can also have the interpretationofdefinitereference;wedisregardthisinterpretationhereandwillcometoitinthe nextchapter).Cheng&Sybesmagivethefollowingexampleforillustration:

(57) Wo xiang kan ben shu . I wouldlike read Cl book ‘Iwouldliketoreadabook.’

Itistruethattheindefinite[Cl+N]phrase ben shu tendtohaveanonspecificreference.Butif we change the verb form to the past tense, the interpretation of specific indefinite reference arises.

91

(58) Zuowan wo kan-le ben shu . Yesterdayevening I readPerf Cl book ‘Ireadabookyesterdayevening.’

The[Cl+N]phrase ben shu canonlyhaveaspecificindefiniteinterpretationinexample(58).

Actually,theythemselvesmentionthat“thepredicate maai-zo ‘sold’in(59)(theirnumberingis

(25b)) which they take from Matthews & Pacioni (1996) could in principle facilitate either a specificindefinitereadingoradefinitereading”(1999:524525):

(59) Keoi maai-zo gaa ce . hesellZOClcar ‘Hesoldthecar.’20

Fromtheabove,wecanseethattheverbtenseplaysaveryimportantroleindeciding whetherthe[Cl+N]phrasehasaspecificornonspecificindefiniteinterpretation.Wecaninsert yi ‘one’beforetheobjectnounsin(57)and(58).So,the[ yi +Cl+N]phrasecanhavethesame twointerpretationsasthe[Cl+N]phraseinMandarin,dependingontheverbform.Ispeculateit isthesamewiththespecificandnonspecificindefiniteinterpretationsofthe[Cl+N]andthe[ yi

+Cl+N]phraseinCantonese.

Cheng & Sybesma (1999) also provide two contexts in which nonspecific indefinite cannotoccur,butspecificindefinitecan.Thatmeansthe[Cl+N]expressioncannotbeused,but

20 Thisistheoriginaltranslation;butitcanalsomean‘Hesoldacar’.

92

[yi +Cl+N]nominalscaninthetwocontextsandaccordingtothemthereisnophonological reasonwhy yi cannotbesuppressed.

One context is that the object of bounded predicates which are bounded for reasons independentoftheobjectcannotbenonspecificindefinite.So,accordingtothem,the[Cl+N] nominalcannotappearasanobjectinsuchacontext. The following two examples are taken fromCheng&Sybesma(1999:526;theirnumberingis(29b)and(30b))

(60) Wo he-wan-le yi-wan tang . I drinkfinishPerf onebowl soup ‘Ifinishedabowlofsoup.’ (61) * Wo he-wan-le wan tang I drinkfinishPerf bowl soup

Itistruethat(61)isungrammaticaloratleastmarginalized;butifwesubstituteacompound noun(like sui-la-tang ‘sourhotsoup’)forthemonosyllabicnoun tang ‘soup’,thenthesentence

(61)becomesgrammaticalas(62)shows.

(62) Wo he-wan-le wan suan-la- tang . I drinkfinishPerf bowl sourhotsoup ‘Ifinishedabowlofsourhotsoup.’

Alsotheungrammaticalsentence(61)canberemediedbychangingthecompoundverb he-wan

(‘drinkfinish’)tothemonosyllabicverb he (‘drink’)as(63)indicates.

93

(63) Wo he--le wan tang . I drinkPerf bowl soup ‘Idrankabowlofsoup.’

Boththe[Cl+N]objectsof(62)and(63)onlyhaveaspecificindefiniteinterpretation.The resultativecompound he-wan (‘drinkfinish’)andtheperfectivemarker le insentences(60)and

(62)indicatethatthepredicatesarebounded.Thepredicateofsentence(63)isalsobounded, which is indicated by the perfective marker le . Native speakers think that (60) and (63) are synonymous.

AlthoughCheng&Sybesma(1999)believethatthereisnophonologicalreasonwhythe yi inthe[ yi +Cl+N]cannotbesuppressedincertaincontexts,inactuality,thereisareasonwhy yi

‘one’cannotbedeleted. IndiscussingandwordlengthinChinese,Duanmu&Lu(2002) andDuanmu(2000,2001)arguethatwordchoicesinChinesearerelatedtostress.Inorderto obeytheNonheadStressRule,whichstatesthatsyntacticnonheadsshouldhavestress,Chinese objectsshouldhavemorestressthantheverb.Awordwithmorestressshouldnotbeshorterthan a word with less stress. So an object should not be shorter than its preceding verb. In the followingexamples,whicharetakenfromDuanmu(2001:124),wecanseethattheonesyllable objectoftheonlyungrammaticalphrase(b)isshorterthanthetwosyllableverb;henceviolating theNonheadStressRule.Inthegrammaticalphrases(a),(c),and(d),theobjecteitherhasthe samenumberofsyllablesormoresyllablesthantheverb.

94

.[VerbObject]

a. zhong-zhi da-suan b.* zhong-zhi suan c. zhong da-suan d. zhong suan plantgarlic ‘toplantgarlic’,

WecanextendDuanmu’sanalysistotheuseof[Cl+N]phrases.ClassifiersinMandarin are usually unstressed and numerals are stressed. The occurrence of [Cl + N] phrases or the deletionornondeletionof yi in[ yi +Cl+N]phrasescanbeexplainedbytheNonheadStress

Rule.Whentheobjectisacompoundnoun,the yi in[ yi +Cl+N]phrasescanbedeleted(62).

Whentheverbisacompound,buttheobjectnounismonosyllabic,thenthe yi in[ yi +Cl+N] phrasescannotbedeleted(60).(62)and(63),inwhichthe[Cl+N]phrasesoccurinobject positionandtheyhavethesamenumberofsyllablesastheverbalelementsormoresyllables,are grammatical,becausetheyconformtotheNonheadStressRule.But(61)isungrammatical,as theobjecthasonlytwosyllablesandtheverbalelementhasthreesyllables,thusviolatingthe

NonheadStressRule.Whenthenumeral yi ‘one’isretained,thentheobjectandthepredicate each have three syllables and the Nonhead Stress Rule is obeyed; so example (60) is grammatical.

95

Another context that Cheng & Sybesma (1999) provide is sentences with secondary predication. The object nominal in such sentences, which functions as the subject of the secondarypredicate,mustbespecificindefinite.

(64) Wo jiao-guo yi-ge xuesheng hen congming . I teachExp oneCl student very intelligent ‘Ioncetaughtastudentwhowasveryintelligent’ (65) *Wo jiao-guo ge xuesheng hen congming I teachExp Cl student very intelligent (p.526)

Manynativespeakersthinksentence(65)isgrammatical.Interestingly,onefriendofminethinks both sentences (64) and (65) are ungrammatical. I guess he misanalyzed wo jiao-guo yi-ge xuesheng (‘I once taught a student’) as a nominal with a relative clause ( wo jiao-gu ‘I once taught’).Butallmyinformantsagreethatthefollowingsentence(66), whichalsoinvolvesa secondarypredicate,butwhichisnotlikelytobemisinterpreted,is grammatical,whetherthe numeral yi isdeletedornot:

(66) wo zhao-le (yi)-ge xue-sheng bang-mang . I findPerf (one)Cl student help ‘Ifoundastudenttohelp(me).’

Inthissubsection,Ihavearguedthatthe[ yi +Cl+N]phraseandthe[Cl+N]phrase havethesametwoindefiniteinterpretations:specific andnonspecific. Whichinterpretationis appropriateinparticularcontextsisrelatedtotheverbform. ContrarytoCheng&Sybesma’s

96

(1999)view,the[Cl+N]combinationresultsfromthedeletionof yi inthe[ yi +Cl+N]phrase.

Yi inthe[ yi + Cl+N]cannotbedeletedincertaincontextsforphonologicalreasons.[Nume+

Cl+N]canonlybeusedpostverbally.

4.4.2 The Syntactic Structure of Numerically-Quantified Nominals in Mandarin

Crosslinguistically,thenumeralandtheclassifierareadjacenttoeachotherinnominal expressions of numeralclassifier languages. As to the syntactic structure of numerically quantifiednominals,therearetwodifferentviews. Some scholars (Tang, 1990; Bhattacharya,

2001)treatthenumeralandtheclassifierasformingacomplexheadwhichselectsanNPasits complement,whileothers(Simpson,2005;Cheng&Sybesma,1999,2005; Li,1998)believe that the numeral and the classifier project their own phrasal categories. We will look at the argumentsproposedforthetwoanalysesandseewhichanalysisisbettersuitedtoChineseand numeralclassifierlanguagesingeneral.

Before generative syntacticians became interested in the internal structure of nominal expressionsinnumeralclassifierlanguages,linguistsgenerallyheldtheviewthatnumeralsand classifiersfirstcombinetogethertoformacomplexconstructionandthentheyasaunitquantify the following noun. Croft (1994:151) notes that “the numeral and the classifier form a constituent,nottheclassifierandthenoun.”Greenberg(1975)calledthecomplexconstruction theclassifierphrase. In termsofwordorderamong the three elements (quantifiers/numerals, classifiersandnouns)inquantifiednominals,onlyfourpatternsoccurinthesampleofaboutone hundrednumeralclassifierlanguagesGreenberg(1972)surveyed:QClN,NQCl,ClQNand

97

NClQ. In no numeralclassifier language is there a dominant word order where quantifiers/ numerals and classifiers are separated by head nouns. Greenberg (1972: 28) also writes that numeralsandclassifiersaresocloselyconnectedthatprosodicallytheymayhaveoneaccentand analystsconsiderthenumeralclassifierconstructiontobeasinglewordinmanylanguages. In somenumeralclassifierlanguageslikeBengali,theclassifierisaboundmorphemeandenclitic to the numeral. Indeed, in the discussion of part of speech in Chinese grammar books, the complex of numerals and classifiers is listed as a separate word class called shu-liang-ci ,

(literallymeaning’numeralmeasureword’).

Generative syntacticians working with nominals in numeralclassifier languages either assumethatthecomplexofthenumeralandtheclassifierinstantiatesasingleheadorthatthey aretwoseparateheads.Researchersseldomdiscussthereasonsbehindtheirassumptions.Tang

(1990:403404)suggeststhattheMandarinChineseDPselectsaClPasitscomplementandboth thenumeralandtheclassifierarecontainedintheheadCl. OnceClisinstantiated,boththe numeral and the classifier are obligatory. So, according to Tang (1990), Mandarin Chinese numericallyquantifiednominalshavethefollowingsyntacticstructure(67).

98

(67)

‘threebooks’

Tang (1990) here follows the traditional view that the numeral and the classifier in

MandarinChinesetogetherfunctionasasyntacticunit.Shedoesnotdiscusswhybothnumerals andclassifiersaregeneratedinthenodeClandwhythephrasalcategoryofthecomplexisClP, notNumP.Tangdoesmentioninpassingthatwhetherthenumeralandtheclassifierhavetheir distinctprojectionsornotdoesnotmakeanydifferencewithregardtoMandarinChinesedata.

Like Tang (1990), Bhattacharya (2001) argues that numerals/quantifiers and classifiers form a complex head in Bengali, an Eastern IndoAryan language. The main reason that she givesisthatinBengali,anumeraloraclassifieralonecannotbeusedwithanoun,asexample

(68)shows,andthemovementofNPwithinDPinBengalileavesthenumeralclassifiercomplex strandedandstrandingestablishestheheadnessofthecomplex(p.197).

99

(68)a.* (du)-To boi twoClbook b.* (ko)-jon chele someCl boy (Bhattacharya,2001:195)

Following Lobël (1989), Bhattacharya assumes that there is a Quantifier Phrase between the headsDandN.Itsheadisthenumeralclassifiercomplex.ThefunctionofthecategoryQisto ensurethecountabilityoftheNP.TheQheadismorphologicallyrealizedasapluralsuffixin languageslikeEnglishandGerman,ifthenounisacountableone.Ifthenounisamassnoun, theQheadisrealizedthroughameasureword.Innumericalclassifierlanguages,whichlackthe pluralmarker,theQheadislexicallyrealizedasNum+Cl.

BothTang’s(1990)andBhattacharya’s(2001)analysescanexplainthedataofnominals in the languages that they are concerned with. If we look at nominal expressions in numeralclassifierlanguagesingeneral,thentheirproposalsofthecomplexheadofthenumeral andtheclassifiercannotdealsuccessfullywithdifferentpatternsrelatedtotheclassifier.Ifwe takeintoaccountthedifferentfunctionsofnumeralsandclassifiers,theviewthatthecomplexof thenumeralandtheclassifierprojectsasinglephrasalcategoryisnotveryconvincing.People interpret the obligatory use of the classifier in numeralclassifier languages from two perspectives. Ontheonehand,itcanbesaidthatthenumeralcannotmodifythenoundirectly; the numeral must first combine with classifier in order to modify the noun (Croft, 1994;

Greenberg,1972).Thisisthetraditionalview.Ontheotherhand,wecansaythatnounsinthe numeralclassifier language are mass nouns and in order to be counted, they must first be

100 individualized via classifiers (Chierchia, 1998a, b; Cheng & Sybesma, 1999). Looking at linguisticfactsindifferentnumeralclassifierlanguages,wethinkthatthesecondinterpretation is more plausible. Semantically, numerals and classifiers have different functions. Classifiers have the functions of individualization and classification as we discussed previously, while numerals have that of number specification (Simpson, 2005). In some classifier languages in

Southeast Asia (like Vietnamese, Hmong), classifiers canbeusedalone withoutany numeral

(examples(69)and(70)).Actually,classifiersinmanydialectsinthesoutherndialectareasin

Chinahavesuchauseasshowninexample(71).

(69) Con bo dang an lua kia! 21 Cl cow Prog eat paddy overthere ‘Look!A/Thecowiseatingyourpaddy!’ (Vietnamese;Nguyen,2004:43) (70) Tus tsov tshaib tshaib plab . Cl tiger hungry hungry stomach ‘Thetigeris/wasveryhungry.’ (Hmong;Jaisser1987,citedinSimpson,2005:808) (71) Go leotsi jiu hou lek sin dak . Cl lawyer need very smart onlyokay ‘Thelawyerhadbetterbeverysmart.’(Cantonese;Cheng&Sybesma,1999:524)

Example (69) shows that the [Cl + N] combination in Vietnamese can have a definite or indefinitereference,dependingonthecontext. Examples(70)and(71)indicatethenominals

[Cl + N] in Hmong and Cantonese can have a definite reference. Ferguson (1962, cited in

21 Thetonediacriticsintheexampleareomitted.

101

Greenberg,1975:37)notesthatclassifiersaregenerallycalleddeterminativeinthegrammarof

Bengali,astheyaresuffixedtonounstoexpressdefinitereference(e.g.boikhana‘bookthe’).

Ontheotherhand,incertainconstructionsinsome classifier languages, numerals can combinewithnounswithoutanyclassifier,especiallyifthenumberspecificationisvagueand individualization is not necessarily implied (Simpson, 2005: 808). For example, in Burmese classifiersareoptionalwithnumbersthatarepowersoftenormultiplesoftenandwithcomplex highernumbers.ThefollowingBurmeseexamplesaretakenfromBurling 22 (1965:252253).

(72) nwa oun kaun cowthreeCl ‘threecows’ (73) nwa hna-she cowtwoten ‘twentycows’

EvenincolloquialMandarinChinese,whenthenumberisverylarge,theomissionof classifiersisacceptableasshowninthefollowingexamples:

(74) wu qian xuesheng baodao . five thousand student register ‘Fivethousandstudentsregistered.’

22 Burlingconsiderswords she ‘ten’, ya ‘hundred’, thaun ‘thousand’asclassifiers

102

(75) san ∗(ge) xuesheng meiyou baodao . fiveClstudentnotregister ‘Fivestudentshaven’tregistered.’

Althoughtheoptionaluseofclassifierswithlargenumbers(like bai ‘hundred’, qian ‘thousand’, wan ‘tenthousand’)inChineseisseldommentioned,thisphenomenonisquitecommonwhen theheadnounreferstoahuman.23

Onemoreargumentagainsttheviewofthesinglecomplexheadisthatsomeadjectives canbeaddedbeforeclassifiersinMandarinChinese.WhenCheng&Sybesma 24 (1999)discuss the syntactic differences between measure words and genuine classifiers, they point out that adjectivescanbeaddedbeforemeasurewords,butnotbeforeclassifiers.Thefollowingaretwo oftheirexamplesforillustration.

(76) na yi xiao xiang shu that one small box book ‘thatonesmallboxofbooks’ (77) yi (*da ) zhi gou one big Cl dog

23 Whentheheadnounreferstoanonhumanorinanimateentity,theclassifierisobligatorywithlargenumbers.I speculatethathumannounsusuallytakethegeneralclassifier ge whichdoesnothavespecificmeaningandits classifyingfunctionisnotevident.Theonlyspecificclassifierusedwithhumannounsis wei whichhasa connotationof‘beingrespectfultothereferent’andwhichisnotveryoftenused.Nonhumannounsusuallyhave theirspecificclassifierswhichhaveaclassifyingfunction. 24 Theyusemassifier(collocatingwithmassnouns)andcountclassifier(withcountnouns).Sotheyglossthe measureword xiang inexample(76)as‘CLbox’.

103

Inexample(76)theadjective xiao ‘small’isinsertedbeforethemeasureword xiang ‘box’,but theinsertionof da ‘big’beforetheclassifier zhi leadstotheungrammaticalityofexample(77).

Inreality,adjectivescanbeinsertedbetweennumeralsandcertaingenuineclassifiers.In example(78),theadjective hou ‘thick’isinsertedbeforetheclassifier ben .

(78) Yi hou ben shu onethick Cl book ‘onethickbook’ 25

So,insomeclassifierlanguages,classifierscanbeusedalonebeforenouns(Vietnamese,

Hmong,Cantonese)andlargenumbersoccurbeforenounswithoutclassifiers(Burmese,Chinese) andsomeadjectivescancomebetweennumeralsandclassifiers (Chinese). Most importantly, numerals and classifiers have their own different functions (number specification vs. individualization). We think the numeral and the classifier both head their separate phrasal projections. The numerically quantified nominal in Mandarin has the following syntactic structure(79):

25 Theadjective da ‘big’canbeusedbeforeclassifierslike zhang (aclassifierformaps,stamps,paper,tables), jian (forrooms), zhi (forboxes,cats,lobsters,feet,legs);theadjective chang ‘long”beforeclassifierslike tiao (forsnake, rope,street,river’).

104

(79)

‘threebooks’

Before concluding this chapter, I should say that in Mandarin the numeral projects a

NumeralPhrase.ThereisnoNumberPhraseinChinese,aswearguedthatMandarindoesnot haveapluralmarkerinChapterThree.

105

CHAPTER FIVE

COMPARISON OF CLASSIFIER USES IN MANDARIN AND WU

InChapter4,IdiscussedtheclassifiersysteminMandarinanditsnumericallyquantified nominals.Inthischapter,IdiscusstheusesofclassifiersinWu,makeacomparisonbetweenWu andMandarinintermsofclassifierusageandanalyze the syntactic structure of the [Cl + N] nominalinthesetwovarietiesofChinese.

5.1 The Uses of the Classifier in the Wu Dialect

TraditionalstudyofChineseclassifiersisconcernedwithwhichclassifiersareusedwith which nouns in different dialects. Not until recently have Chinese linguists started to be interestedinthesyntacticfeaturesandstructuresofnominalsinvolvingclassifiers,especiallyin someSouthernChinesedialects(Cheng&Sybesma,1999,2005;Matthews&Yip,1994;Liu,

2002; Qian, 1997). Many of the comparative studies are focused on Cantonese and Wu in comparisonwithMandarin.ThissectiondealswiththeusesofclassifiersinWuinsomedetail.

First,Idiscusstheusesof ge inWu,whichfunctionsasthegeneralclassifierinMandarin.Inthe

Wu dialect, ge hastwootherfunctionsinadditiontobeingaclassifier. Ge canfunctionasa demonstrativeandasamodificationmarkerinWu.

106

Beforediscussingthedifferentusesof ge inWu,wewilllookattheusesof ge inChinese diachronically. Ge is mainly used as a general classifier in modern Mandarin, but it used to perform several other grammatical functions in its history. The classifier ge was originally derivedfromthenoun zhu whichmeans‘bamboo’andwasusedasaclassifierincounting zhu

‘bamboo’. Later on the use of ge as a classifier was extended to nouns referring to other concreteobjectsandthenabstractnouns.Now ge isageneralclassifier,whichcanbeusedwith mostnounsinMandarin.

While the range of nouns collocating with ge was greatly widened, its grammatical functionsalsooncemultipliedinitshistory. Atthebeginning,itwasonlyused asaspecific classifierfor zhu ‘bamboo’andthendevelopedintoageneralclassifier. Initshistory, ge was onceusedasademonstrativeasthefollowingtwoexamplesillustrate.In(1), ge wasusedasa demonstrativeadverb,andin(2), ge functionsasademonstrativedeterminer.26

(1) bai fa san qian zhang, yuan chou shi ge chang . whitehair three thousand foot because sadness like this long ‘Mywhitehairsreachthreethousandfeet,because I am as sad as these are long.’ (Qiu Fu Ge 27 ,8 thC) (2) ge ren hui di ? that person avoidastaboo what ‘Whatdoesthatpersonavoidastaboo?’ ( Bei Qi Shu, c.AD600,citedinShi&Li, 2002:7) 26 Intraditionalgrammar,demonstrativedetermineriscalleddemonstrativeadjective. 27 Qui Fu Ge ,apoemwrittenbythegreatChinesepoet Li Bei (AD701762,sometimestranslatedas Li Bo ). Qiu Fu isaplacewhereLiBeiwrotethepoem.

107

From the 11 th to the 17 th century, ge was grammaticalizedintoamodificationmarker, competingwith de whichisthemodificationmarkerinmodernMandarin(Shi&Li,2002:7; citingCao,1986).ThefollowingexampleistakenfromShi&Li(2002:7).

(3) Ni ge gu shi qi gu . youGEboneisbasebone ‘You’redespicable.’ ( Zhang Xie Zhuangyuan 28 ,c.AD1200)

The aforementioned two functions (i.e. that of a modification marker and that of a demonstrative) are no longer performed by ge in Mandarin Chinese. But these two syntactic functionsarestillperformedby ge inWudialectsasshowninthefollowingexamples.

(4) ge ben shi youqu . (demonstrative) thisClbookinteresting ‘Thisbookisveryinteresting.’ (5) ngo ge shi youqu . (modificationmarker) IGEbookinteresting ‘Mybook(s)is/areveryinteresting.’ 28 TheworkwaswrittenbyananonymousgroupofwriterswhowerefromWenzhou,Zhejiangprovince.Sothe languageusedinthebookmightbetheWenzhoudialect(whichishardlyintelligibletospeakersoftheotherWu subdialects),althoughthewrittenChineselanguageissharedbyspeakersofdifferentdialects.TheEnglish translationoftheexampleismine.

108

In (4), ge functions as a demonstrative determiner and in (5) it functions as a modification marker. In Shanghaiese, ge can be reduplicated for emphasis when functioning as a demonstrative(Qian,1997)as(6)shows.

(6) ge-ge san ben shi Thisthis three Cl book ‘thesethreebooks’

Inthelinguisticliterature,somescholars(Pan,1991)writethatclassifierscanfunctionas demonstrativesandPan(1991:271)observesthatintheWudialect“theclassifier ge hasbecome aparticlethatissimilarinfunctiontothe de inMandarin,”thatis,amodificationmarker.We consider the three different functions as performed by the homophone ge , not by the general classifier ge intheWudialect,althoughetymologicallythethreefunctionsarerelated. Inthe case of the two functions of ge as a demonstrative and a modification marker, ge no longer performsthetwobasicfunctionsofclassifiers:classificationandindividualization.

Why could the general classifier ge have developed into a demonstrative and a modification marker in the Wu dialect, but not in Mandarin? The development of multiple functionsof ge mightberelatedtotheusesofclassifiersingeneralintheWudialect.Aswe discussed in Chapter Four, according to Bisang (1999), classifiers can have the following functions:classification,individualization(inthecontextofcounting),referentialization(inthe contextofreference)andrelationalization(inthecontextofandrelativeclauses).

109

ClassifiersinMandarinonlyhavethefunctionsof classification and individualization, whileintheWudialect,classifiershaveallfourfunctions.Thefunctionofreferentializationis motivated by the individualizing function of classifier. As classifiers in Wu dialects have referentializationfunctionand ge isthe general classifierwhichismoreoftenusedthanother specific classifiers, ge thenisveryoftenusedinreferentialexpressions.Andoneofthemain featuresofthedemonstrativeisreferentiality,soitisquitenaturalforthegeneralclassifierto developintoademonstrative.

Asforthedevelopmentof ge intoamodificationmarkerintheWudialect,itisrelatedto therelationalizationfunctionofclassifiers.InWudialects(andprobablyotherdialects,forthat matter)andmodernMandarin,theremustbealinking element between the modifier and the modifiedelement. ThelinkingelementinWudialectscanbe ge orotherspecificclassifiers, whilethatinMandarincanbe de ordemonstratives 29 .Considerthefollowingexamples:

Wudialect

(7) ngo ge shi youqu .30 I GEbook interesting Mybook(s)is/areveryinteresting.’ (8) ngo ben shi youqu . 29 Inpossessiveconstructions,thelinkingelementcanbedeletedinthecaseofinalienablepossession,likekinship terms,e.g., wo baba (literally‘Ifather’,meaning‘myfather’.) 30 When ge isthelinkingelement,sometimesthenominalexpressionscanbeambiguous,as ge canbothbea modificationmarkerandageneralclassifier.In(6) ge isamodificationmarker,as ge usuallydoesnotcollocatewith thenoun shu ‘book’.Anominalexpressionlike wo ge erzi ‘myson(s)’isambiguous.Ifthe ge isamodification marker,thespeakerhasoneormorethatoneson.If ge isthegeneralclassifier,whichisveryoftenusedwithhuman nouns,thenominalonlyreferstooneparticularsonofthespeakerincontext.

110

IClbookinteresting ‘Mybookisveryinteresting.’Or‘thisbookofmineisveryinteresting.’ (9)* ngo shu hen youqu Ibookveryinteresting Mandarin (10) wo de shu youqu . IDEbookinteresting ‘Mybook(s)is/areveryinteresting.’ (11) wo zhe shu youqu . Ithisbookinteresting ‘Thisbookofmineisveryinteresting.’ (12)* wo shu youqu Ibookinteresting

In(7),when ge isusedasamodificationmarker,thenumberoftheobjectsreferredtocanbeone ormorethanone.Whenthespecificclassifier ben isusedin(8),thenumberofbooksmustbe one. So, here the individualizing function of the specific classifier ben is obvious; ben only concurrentlyfunctionsasalinkingelement.(9)showsthatthedeletionofthelinkingelement between the modifier and the modified element leads to ungrammaticality. The Mandarin example(11)indicatesthatthedemonstrativecanconcurrently function as a linking element.

The modification marker in (10) de actually is developed from the demonstrative di in Old

Chinese. Because of the concurrent function of classifiers as a linking element in the Wu dialect (or the relationalization function in Bisang’s (1999) term), the general classifier ge

111 developedintoamodificationmarker,whileinMandarin,demonstrativesconcurrentlyfunction asalinkingelementandthedemonstrative di developedintothemodificationmarker de .

5.2 Distributions and Interpretations of the [Cl + N] Phrase in Wu and Mandarin

Of the four functions performed by classifiers discussed in Bisang (1999), Mandarin classifierscanonlyhaveclassifyingandindividualizingfunctions.Theclassifyingfunctionis morerelatedtosemanticsthansyntaxandallclassifiersineveryclassifierlanguagehavesucha function. Classifiers classify objects denoted by nouns into different classes based on their meanings.Thechoiceofclassifiersusedwithnounsdependsoncertainfeaturesoftheentities denotedbythenouns.Forexample,thenoun yizi ‘chair’cancollocatewithclassifiers zhang and ba ,whichbringoutdifferentaspectsofthemeaning oftheobjectdenotedbythenoun.The classifierz hang, derivedfromitsverbalmeaningof‘open’,whichcan alsocollocatewith zhi

‘paper’or ditu ‘map’,showsthemeaningof‘havinganopen,flatplane’,while ba ,derivedfrom itsverbalmeaningof‘hold’,whichcanbeusedwithnounslike yaoshi ‘key’, dao ‘knife’or shan

‘umbrella’,bringoutthemeaningof‘havingahandle,orbeingoperatedbyhands’.

The individualizing function of the classifier is a syntactic requirement. Classifiers individualize the objects denoted by nouns so that they can be counted in numeralclassifier languages(Bisang,1999) 31 .Theyperformtheindividualizingfunctioninthecontextofcounting.

So,mostoftenclassifiersshouldcooccurwithnumeralsincountingcontextinMandarinandthe 31 WhenCheng&Sybesma(1999,2005)writethatclassifiersinChinesehavetheindividualizingorsingularizing functionequivalenttothatofarticlesinEnglishorRomancelanguages,theyseemtoequatetheindividualization functionwiththereferentializationfunctionoftheclassifiermeantbyBisang(1999).

112 distributionofthe[Cl+N]combinationinMandarinisquitelimited.However,intheWudialect andsomeotherSoutherndialectslikeCantonese,classifierscanperformallthefourfunctions.

Thereferentializationfunctionofclassifiersgivesrisetotheprevalenceofthe[Cl+N]phrasein the Wu dialect and some other Southern dialects without being preceded by numerals or demonstratives. The following examples illustrate the distribution of the [Cl + N] phrase in

MandarinandWudialect.

Mandarin (13) wo mai-le ben shu . IbuyPerfCl book ‘Iboughtabook.’ (14)* ben shu wo mai-le . Clbook I buyPerf (15)* ben shu hen youyong . Clbookvery useful

Wudialect

(16) ngo ma-le ben shi. I buyPerf Cl book ‘Iboughta/the/thisbook.’ (17) ben shu ngo ma-le . (topicalizedobject,primarytopic) ClbookIbuyPerf ‘The/thisbookIbought.’ (18) ngo ben shu ma-le . (topicalizedobject,secondarytopic) IClbookbuyPerf

113

‘Iboughtthe/thisbook.’ (19) ben shu hen youyong . Clbookvery useful ‘The/thisbookisveryuseful.’

Distributionally, the [Cl + N] phrase in Mandarin can only occur in the postverbal positionandcannotoccursentenceinitiallyasexamples(13)(15)show.The[Cl+N]phrasein the Wu dialect can both be used as postverbal object (16), preverbal object (17), (18) and subject(19).

In terms of interpretation, there also exist differences between the [Cl + N] phrase in

MandarinandthatintheWudialect. InMandarin,the [Cl + N] can only have an indefinite interpretation.Itcanhaveeitherspecificornonspecificreference,dependingoncontext. 32 In

(13) the object ben shu (‘a book’) has an indefinite specific interpretation. In the following example, ben shu tendstohaveanindefinitenonspecificinterpretation.Here,itseemsthatverbal aspect/tenseaffectstheinterpretationoftheobject ben shu .

(20) wo xiang mai ben shu . (Mandarin) IwantbuyClbook ‘Iwanttobuyabook.’

We can also imagine a situation in which the object ben shu has an indefinite specific interpretation. Forexample,aparentwenttobuyabookforhimselfwithhischildintow.The 32 Cheng&Sybesma(1999,2005)holdthatthe[Cl+N]phraseinMandarincanonlyhavethenonspecific indefiniteinterpretation.IarguedagainstthisviewinChapterFour.

114 childfoundonebookveryinterestingandsaidtotheparent(20)withthatparticularbookin mind;insuchasituation,thereferentof ben shu canbespecific.

IntheWudialect,the[Cl+N]nominalhasmorediverseinterpretations.Inadditionto the indefinite specific/nonspecific interpretationsharedbytheMandarin[Cl+N]nominal,it hasadefiniteinterpretation.Whetherithasadefinite or indefinite interpretation depends on context,bothlinguisticandsituational.Whenthe[Cl+N]occurspreverbally,itcanonlyhave definiteinterpretationas(17)and(18)indicate.Aswementionedpreviously,thisisduetothe characteristicofthetopiccommentsentencestructureandthesubjectandtopicalizedobjecttend tobedefinitenominalsinChinese.Whenthe[Cl+N]occurspostverbally,itcanhavebotha definiteandindefiniteinterpretationintheWudialectas(16)shows.

Whenthe[Cl+N]hasadefiniteinterpretation,thetoneofclassifiers changesinthe

Wenzhoudialect(Cheng&Sybesma,2005). IntheSuzhoudialect,thegeneralclassifier ge is stressedwhenthe ge +Nphrasehasadefiniteinterpretation(Liu,2002).Classifiersdonotcarry stress in Chinese when they are preceded by numerals or when the [Cl + N] phrase has an indefinite interpretation. In the Wenling dialect, the classifier in the definite [Cl + N] expressioncanbestressedwhenseveralentitiesareinsightandthespeakerwantstomakeclear whichentityheisreferringto.Usuallythestressingoftheclassifierisaccompaniedbysome gesture,likepointingtotheentity.Otherwise,theclassifierisnotstressedwhetherthe[Cl+N]is definiteorindefinite. Forexample,the ben shu in(21),theequivalentsentenceof(20)inthe

Wenlingdialect,canhaveadefiniteinterpretationinadditiontoanindefiniteone.

115

(21) wo xiang ma ben shi. (Wenlingdialect) IwantbuyClbook ‘Iwanttobuya/thisbook.’

Imaginethatyouwereinabookstoreandaskedashopassistanttohelpyoutakeabookfromthe topshelfandshetookthewrongbook,youcouldsay(21)toher,stressingtheclassifier ben and pointingatthatparticularbook.Thestressingoftheclassifierhereislikeacontrastivestressin

English,e.g.,

(22) Ilikethe red book,nottheyellowone.

In this section, we discussed the distributions and interpretations of the [Cl + N] expressionintheWudialectandMandarin.Wewilldiscusssyntacticstructuresofthe[Cl+N] expressioninthenextsection.

5.3 Syntactic Structures of the [Cl + N] Phrase in Wu and Mandarin

In this section, we first discuss what syntactic feature is encoded in the classifier in

Chinese (Section 5.3.1) and then put forward the syntactic structures for the different interpretationsofthesurfaceform[Cl+N](Section5.3.2).

116

5.3.1 The Countability Feature Encoded in the Classifier

Cheng & Sybesma (2005: 274276) argue that the classifier has something to do with numberandsuggestthattheclassifieristhelocus for grammatical number in Chinese.Their relevantargumentsarethatitistheclassifierinthe[Cl+N]phrasethatdetermineswhetherthe referent is singular or plural and that reduplication of the classifier gives rise to a universal quantification reading. Consider the following examples which are taken from Cheng &

Sybesma(2005:274275).

(23)a. bun syu (Cantonese) Clbook ‘a/thebook’, not‘(the)books’ b. di syu Cl pl book ‘(the)books’,not‘a/thebook’ (24)a. wo xiang mai ben shu . (Mandarin) IwantbuyClbook ‘Iwouldliketobuyabook.’ b. wo xiang mai xie shu . IwantbuyCl pl book ‘Iwouldliketobuysomebooks.’ (25) ge-ge xuesheng (Mandarin) ClClstudent ‘everystudent’

117

In(23a)and(24a),thereferentofthe[Cl+N]phraseissingularandin(23b)and(24b)the referentisplural.In(25),thereduplicationoftheclassifier ge yieldsauniversalquantification reading.ItisgenerallyagreedthatinallChinesedialectsthereisapluralclassifier.Itis di in

Cantoneseand xie inMandarin.

GrantedthattheclassifieristhelocusforgrammaticalnumberinChineseasCheng&

Sybesma(1999,2005)argue,thenumberencodinginChineseisdifferentfromthatinGermanic or Romance languages. For one thing, the individual classifier does not change no matter whetherthenumberoftheobjectdenotedbytheheadnounisoneormorethanone.InEnglish thepluralmarkerisrealizedinthenounandinFrenchthenumberisalsorealizedinthearticle whenthereferentoftheheadnounismorethanoneinnumber.Foranother,thesocalledplural classifier xie inChinesecanonlybeusedwiththenumeral yi ‘one’.

(26)a. yi ben shu oneClbook ‘onebook’ b. san ben shu threeClbook ‘threebooks’ (27)a. yi xie shu / shui oneQuan book/water ‘somebooks/water’ b.* san xie shu / shui threeQuanbook/water

118

IftheclassifierinChineseisthelocusof grammatical number as Cheng & Sybesma

(2005)argue,thenboth(26a)and(26b)aresingularasthesameindividualclassifier ben isused in the two examples. Considering san ben shu ‘three books’ as singular is contrary to one’s intuition.Inthesameway,if xie isaclassifierencodingplurality,howcomethatnumeralsother than yi ‘one’cannotbeusedwithitandwhymassnounslike shui ‘water’canbeusedwithit?

TheencodingofnumberinChinese(iftheclassifierisconsideredtoencodenumber)doesnot conformtothecommonpracticeinlanguagesthatthemarkerencodingpluralityshouldbeused withcountnounsandthenumeral(ifused)shouldbetwoorabove.

Nowwewillconsiderthecontextsinwhichthesocalledpluralclassifier xie isusedand drawatentativeconclusionaboutitsgrammaticalstatus. Xie isoftenusedwiththenumeral yi

‘one’,butthenumeral yi ‘one’canalwaysbedeleted.Themeaningof yi-xie is‘some’andlike

‘some’inEnglish yi-xie canbeusedwithbothcountandmassnounsas(27a)illustrates.

Yi-xie ‘some’canbemodifiedbydemonstratives(28)andadverbs(29)asthefollowing examplesshow.

(28) zhe/na ( yi)-xie xuesheng this/that some student ‘these/thosestudents’ (29)hao ( yi)-xie xuesheng quite some student ‘quiteafewstudents’

119

As yi ‘one’canoftenbedeletedwhenpreceded byotherelementsasin(28)and(29), zhe/na yi-xie becomes zhe/na xie .Peopletendtomakeacomparisonbetween zhe/na xie and zhe/na + theindividualclassifier. Somescholars(Lü,1980;He,2001)consider xie asameasureword and zhe/na xie as the plural counterpart of zhe/na .33 Shi (2003) treats xie as a plural marker, while others (Cheng & Sybesma, 1999, 2005) consider xie as a plural classifier. Lü (1980) considers hao-xie ‘quiteafew/little’asanadjectiveexpressingquantity.

However, Iljic (1994)arguesthat xie inMandarinisamarkerofplurality,notaplural classifier.OneofIljic’s(1994)argumentsisthatinBeijingdialect,thegeneralclassifier ge can beinsertedbetween xie andthefollowingheadnoun.Inanominalexpression,onlyoneclassifier canbeusedwiththeheadnoun. So,accordingtoIljic(1994), xie in(30)cannotbeaclassifier.

(30) zhe xie ge ren thisQuanClperson ‘thesepeople’

Buttraditionally,somescholars(He,2001)holdtheviewthat xie isapluralclassifierand ge in suchsentencesas(30)functionsasasignalforpauseinordertoemphasizethemeaningofthe demonstrative zhe-xie ‘these’. Iljic (1994) holds that ge is a classifier and that it is the only classifierthatcanbeinsertedafter xie ,althoughhewritesinthenotesthatsomeotherclassifiers canbeusedassuchandthatthegrammaticalityofsuchuseoftheclassifiervariesfromspeaker

33 Zhe/na canmodifynounsdirectly,e.g. zhe/na xuesheng ‘this/thatstudent’.Theycanalsobefollowedby‘( yi) + classifier + N’, e.g., zhe/na( yi)-ge xuesheng ‘this/thatstudent’.

120 tospeaker.LichaoHe(p.c.)saysthatsomeclassifiersotherthanthegeneralclassifier ge canbe usedafter xie . Thefollowingexamplesarefromonlinearticles.

(31)a. zhe xie ben shu thisQuanClbook ‘thesebooks’ b. yi xie zhang zhi one Quan Cl paper ‘somesheetsofpaper’

Whilenoteveryclassifiercanbeusedafter zhe-xie ‘these’or yi-xie ‘some’ , after hao-xie ‘quieta few/little’almosteveryclassifiercanbeused.

(32)a. hao xie tiao xinwen quiteQuanClnews ‘quiteafewpiecesofnews’ b. hao xie zhi gou/mao quiteQuanCldog/cat ‘quiteafewdogs/cats’

Theexamplesin(31)and(32)seemtosuggestthatxie isnotaclassifier,asonlyoneclassifier canbeusedwiththeheadnouninChinese.Ofcourse,the classifiersin(31)and (32) canbe omitted.

If xie isnotapluralclassifier,isitamarkerofpluralityasIljic(1994)suggests?Itdoes not seem so. If xie is a marker of plurality, then why does it not suffix to the head noun?

121

Besides,apluralitymarkershouldonlybeusedwithcountnouns.Infact, xie canbeusedboth withmassnouns(33)andwithcountnouns(34).

(33) pingzi li you (yi) xie shui . bottle in have (one) Quan water ‘Thereissomewaterinthebottle.’ (34)a. yi xie xuesheng oneQuanstudent ‘somestudents’ b. hao xie xuesheng quitesomestudent ‘quiteafewstudents’

Ifitisrighttoname hao-xie (whichistheresultingformafter yi isdeletedfrom hao-yi-xie ‘quite a few/little’) a quantifying adjective as Lü (1980) does, then (yi) xie is really a quantifier expressing‘indeterminatequantity’,whichcanbeusedwithcountnounsandmassnouns. (yi) xie isverymuchlikethequantifier‘some’inEnglish,whichcanalsobeusedwithcountnouns andmassnouns.

(35)somebooks/water

Then,whyistheindividualclassifieroptional when (yi) xie ‘some’ quantifies nouns in

Mandarin? I speculate that this is because some quantifiers like xie ‘some’ and xuduo

‘many/much’donotdoanyexactcountingandMandarinclassifiersareusedtoindividualizethe

122 referentsdenotedbytheheadnouninthecontextofcounting.Somescholars(Simpson,2005;

Aikhenvald,2000;Hopper,1986)notethatclassifiersinsomenumeralclassifierlanguageslike

ThaiandMalayareoptionalwhenthenumeralspecificationisvagueandindividuationisnot implied.“InThaiclassifiersdonotoccurwithlargenumberslike1000unlessindividuationis implied” (Aikhenvald, 2000: 100). In Malay, the classifier can be omitted when approximate numeral reference is made (Hopper, 1986). Another quantifier xuduo ‘many/much’ can also directlyquantifytheheadnounswithoutanyclassifierinbetween.Butalmostevery classifier canbeusedafter xuduo ‘many/much’whentheheadnounisacountnoun. (36)a. xuduo (ben) shu manyClbook ‘manybooks’ b. xuduo (pian) wenzhang manyClarticles ‘manyarticles’ Whenthequantifiersareusedwithouttheaccompanyingclassifier,theindividualityof thereferentsisnothighlighted.Treating (yi) xie asaquantifiercansolvethetwopuzzles:oneis whytheclassifiercansometimesbeinsertedbetween (yi) xie andtheheadnounif xie isaplural classifier; another is why (yi) xie canbeusedbothwithcountnounsandmassnounsif it is treatedasapluralclassifieroramarkerofplurality.

123

Ifthesocalledpluralclassifier di inCantoneseisreallylike xie inMandarinasCheng&

Sybesma(1999,2005)andotherssuggest, di mightalsobeaquantifierasIarguedabovefor

Mandarin xie tobe. Di appliestobothpluralanduncountableitems(Matthew&Yip1994:90).

IagreewithCheng&Sybesma(2005:274)thattheremustbeasyntacticmarkerof countabilityinorderforcountnounstobecountable(thisideaisoriginallyDoetjes’s(1996)).

Classifiers in languages like Chinese and number morphology in languages like English and

French serve as a syntactic countability marker. Theclassifieristhecarrierofthesyntactic featureofcountability.

Inthissubsection,IarguedthattheclassifierinChinesedoesnotencodenumberand the socalled plural classifier or marker of plurality xie is really a quantifier. In the next subsection,Iwillanalyzethesyntacticstructureofthesurfaceform[Cl+N].

5.3.2 The Syntactic Structures of the [Cl + N] Phrase in Wu and Mandarin

OfallthestudiesrelatedtoChinesenominalsintheliterature,Cheng&Sybesma(1999,

2005)providethemostdetailedanalysesofthebarenounphrase,the[Cl+N]phraseandthe

[Nume + Cl + N] phrase. Cheng & Sybesma (1999) deals with bare nouns in Mandarin and

Cantoneseandtheir2005chaptercomparesandcontraststhebarenounphrase,the[Cl+N] phraseandthe[Nume+Cl+N]phraseinMandarin,Cantonese,theWenzhoudialectandthe

Min dialect. We mentioned in section 5.2 that the [Cl + N] phrase can be either definite or indefinite in the Wu dialect, while it can only be indefinite in Mandarin. Since it can have different interpretations, then the same surface form [Cl + N] must have different syntactic

124 structures. According to Cheng & Sybesma (1999, 2005), the indefinite [Cl + N] phrase is actually a Numeral Phrase and the quantification of numerals leads to the indefinite interpretation.Thedefinite[Cl+N]phraseisaClassifier Phrase. The syntactic structures of definiteandindefinite[Cl+N]nominalsarerespectivelyrepresentedas(36)and(37).

(37)

(38)

Cheng & Sybesma (2005) argue that the classifier in Chinese encodes definiteness in additiontogrammaticalnumber.Accordingtothem(2005:277),definitenessandnumberseem to be expressed in all languages. All languages have ways of distinguishing definite from indefinitereferenceandwaysofdistinguishingsingularfromplural.Theyfurtherstatethatthe headDofDPinlanguageslikeEnglishhasanindividualizingfunction,whichpicksoutasingle instanceofwhateverisdescribedbythenounanda‘subordinating’function,whichturnsanNP intoanargument.ThesetwofunctionsarecloselyrelatedtothedeicticpropertyofDP,thatis,

125 the property to refer. So, in the nominal domain, the lexical category NP describes and the functionalcategoryDPrefers.Likewise,intheverbaldomain,thelexicalcategoryVPdescribes andthefunctionalcategoryTensePhraserefers.

TheycomparearticlesinEnglishandFrenchwithclassifiers inChinese and conclude thatthefunctionsperformedbytheDinFrenchareperformedbyclassifiersinChinese.Articles inFrenchcanexpressdefiniteness,numberandnounclassification(whengrammaticalgenderis takenintoaccount)asthefollowingexamples(fromCheng&Sybesma,2005:278)illustrate.

(39)a.legarçon,lesgarçons theMSGboythePLboys b.ungarçon,desgarçons aMSGboyaPLboys c.legarçon,lafille theMSGboytheFSGgirl

They givethefollowingCantonesedatatosubstantiate their proposal that Chinese classifiers functioninasimilarwayasarticlesinFrenchdo.

(40) a. bun syu; di syu Clbook;Cl pl book ‘the/abook’;‘the/øbooks’ b. yat-bun syu; syu oneClbook;book ‘abook’;‘books’

126

c. go jan; zek gau Clperson;Cldog ‘a/theperson’; ‘a/thedog’

IfwetakeamomenttolookattheCantoneseexamples,wewillnoticethesamenominalin

Cantonesecanhavetwointerpretations,e.g. bun syu ‘the/abook’in(40a)canbedefiniteand indefinite. Conversely, the same meaning can be expressed in two or three ways, e.g. an indefinitesingularnominallike‘abook’canbeexpressedinCantoneseas ben syu , yat-bun syu and syu .Theyglossthebarenoun syu in(40b)as‘books’;actually,itcanalsomean‘abook’.So, fromtheFrenchandCantonesedataweseesomedifferencesbetweenthemwithrespecttothe usesofdeterminersinFrenchandclassifiersinCantonese.InFrench,anominaliseitherdefinite orindefinite,eithersingularorplural.InCantonese,ontheotherhand,anominalcanbeboth definiteandindefinite;abarenouncanbesingularandplural.Thespecificinterpretationofa

Cantonesenominaldependsonthecontext.

OnemajordifferencebetweendeterminersinFrenchandclassifiersinCantonesecannot bededucedfromthelimiteddata,asCheng&Sybesma(1999,2005)dealwithonlybarenouns andthe[Cl+N]phrase.DeterminersinFrench(andEnglishforthatmatter)alwaysoccuratthe initialpositionofanominal,whileclassifiersinCantoneseandothervarietiesofChinesecanbe precededbynumeralsanddemonstratives.ConsiderthefollowingCantonesedata((41)and(42) fromMatthew&Yip,1994,thediacriticsfortonesomitted).

127

(41)a. ni go behngyahn b. go jeung toi thisClpatientthat Cltable ‘thispatient’‘thattable’ (42) leuhng jek gau twoCldog ‘twodogs’ (43) ni leuhng jek gau thistwoCldog ‘thesetwodogs’

Positionally,classifiersaretoolowtobeasubordinatortotranslateanominalexpression intoanargumentinChinese. Inlanguageswitharticles, numerals never precede articles in a nominal,butinChinese,thenumeralalwayscomebeforetheclassifierifthereisanumeralin thenominal.Cheng&Sybesma(1999,2005)assumethatthereisnoDPanddefinitesinChinese areClPsandindefinitesareNumePs.Thenumeralundoesthedefinitenessoftheclassifier.As

Simpson(2005:824)arguesthatifthenumeralhasaquantificationalscopeoverthedefiniteness ofits[Cl+N]complement,thenainterpretationwillresult(44).ButtheNumePin

Chinesedoesnothavesuchaninterpretationas(45)illustrates.

(44)[QP[threeofDP[thestudents]]] (45) san zhang zhuozi threeCltable ‘threetables’, not‘threeofthetables’

128

So,Iproposethatindefinite[Cl+N]phrasesareNumePs(46)anddefinite[Cl+N]phrasesare reallyDPs(47).Thetwostructuresarerepresentedas:

(46)

‘abook’

(47)

‘thisbook’ Tosumup,thedefinite[Cl+N]nominalintheWudialectisaDPandtheindefinite[Cl+N] nominalinMandarinandWuisaNumeP.

129

CHAPTER SIX

COMPLEX NOMINAL EXPRESSIONS IN MANDARIN

Inthischapter,IanalyzecomplexnominalexpressionsinMandarininvolvingpossessives, personalpronouns,propernamesanddemonstratives,whicharecalledcomplexdefinitesin

Lyons(1999).Likeadjectives,theseelementscanalloccurprenominallyinMandarinnominal expressions.Atthesametime,theycanthemselvesfunctionindependentlyasanominal.For example,in(1)thepossessive wo-de ‘mine’functionsindependentlyasapredicate.

(1) zhe ben shu shi wo-de . ThisClbookisIDE ‘Thisbookismine.’

Insection6.1,IdiscussthepossessivecontainingnominalinMandarin.Section6.2dealswith nominalexpressionscontainingprenominalpronounsandpropernames.Section6.3focuseson nominalswithdemonstrativesinMandarin.

130

6.1 Possessive Constructions in Mandarin

In subsection 6.1.1, I present examples to illustrate the position of the possessive in a

Mandarinnominalexpression.In6.1.2,theinterpretationsoftheMandarinnominalinvolving possessivesarediscussedandsyntacticstructuresareproposedforthesedifferentinterpretations.

6.1.1 Positions of the Possessive

ThepositionofthepossessiveintheMandarinnominalisratherflexible;itcanoccurinthe initialoramiddlepositionofthematrixnominal.Whenthepossessiveisintheinitialposition,it canbefollowedbytheheadnoun(2),the[Nume+Cl+N]sequence(3)andthe[Dem+(Nume)

+Cl+N]sequence(4)asthefollowingexamplesillustrate.

(2) wo-de shu IDEbook ‘mybook(s)’,‘somebooksofmine’,‘someofmybooks’ (3) wo-de san ben shu IDEthreeClbook ‘mythreebooks’,‘threebooksofmine’,‘threeofmybooks’ (4) wo-de zhe (san) ben shu I DE this (three) Cl book ‘this(thesethree)book(s)ofmine’

Whenthepossessiveoccursinthemiddlepositionofthematrixnominal,itmustbebetween theclassifierandtheheadnoun,asthefollowingexamplesshow.

131

(5) ta na le san ben wo-de shu . hetakePerfthreeCl IDEbook ‘Hetookthreeofmybooks/threebooksofmine.’ (6) tai xiang jie zhe (san) ben wo-de shu . hewantborrowthis(three)ClI DEbook ‘Hewantstoborrowthis(thesethree)book(s)ofmine.’

In(5),the[Nume+Cl]sequenceprecedesthepossessiveandin(6),the[Dem+(Nume)+Cl] precedesthepossessive.

ComparedwiththeEnglishpossessivecontainingnominal,inwhichthepossessiveusually occursattheinitialpositionofthenominalandcanbefollowedbyanounorthe[Nume+N] sequence 34 , the Chinese nominal can have the possessive in the middle position and the possessivecancooccurwithdemonstrativedeterminers,whichareprohibitedinEnglish.Since the possessive cannot occur together with definite determiners like the definite article the or demonstrativeslike this, that inanEnglishnominal,thepossessiveinEnglishisconsideredtobe intheDposition(Lyons,1999)orintheSpecDP(Abney1987),andthepossessivecontaining nominalisadefiniteexpression.

What is or what are the syntactic position(s) that the Mandarin possessive may occupy?

Does the Mandarin possessivecontaining nominal have definite reference like its English counterpart?Iwillattempttoanswerthesetwoquestionsinthenextsubsection.

34 Quantifierslike every , all cancooccurwiththepossessiveinEnglish. All precedes(a)and every (b)followsthe possessiveinthenominal.(a)allmyprofessors,(b)hiseverycommand/move.

132

6.1.2 Interpretations and Structures of Possessive-Containing Nominals

Thissubsectionfocusesontheinterpretationsofnominalsinwhichthepossessiveprecedes thebarenounorthepossessiveisusedtogetherwiththe[Nume+Cl]sequenceinanominal.

Nominalsinwhichthepossessiveandthedemonstrativedeterminercooccurwillbediscussed in section 6.3. At the end of the subsection, I propose the syntactic structures of the possessivecontainingnominalinMandarin.

The possessive behaves differently in different languages in terms of its collocation with otherdeterminerswithinanominalanditsrelatednesswiththe(in)definiteinterpretationofthe matrix nominal 35 . In languages like English, the possessive cannot cooccur with definite determinersinanominalandrendersthematrixnominaldefinite,whileinotherlanguageslike

Italian, the possessive can collocate with (in)definite articles and the (in)definiteness of the matrixnominalisindicatedby(in)definitedeterminers(asshownbytheexamplesin(8)below).

Based on these differences, Lyons (1985, cited in Lyons, 1999) classifies languages into determinergenitive(DG)languages(likeEnglish)andadjectivegenitive(AG)languages(like

Italian).ThepossessiveinDGlanguagesoccupiesthepositionreservedfordefinitedeterminers andinAGlanguagesitoccursinadjectivalorsomeotherposition(Lyons,1999:24).Alexiadou

(2004) refers to possessives in DG languages as possessive determiners and those in AG languages as possessive adjectives. Similarly, to account for the different features of 35 Herewediscussnominalswiththepossessorphrasewhichitselfisdefinite,suchas his books or the teacher’s books .Astowhethernominalslike a teacher’s book aredefiniteornot,scholars’viewsdiffer.Lyons(1999)thinks thatsuchnominalsaredefiniteand a teacher’s book means‘thebookofateacher’.Jackendoff(1977),ontheother hand,claimsthatthe(in)definitenessofapossessivecontainingnominaldependsonwhetherthepossessoris (in)definite.

133 possessives in different languages, Giorgi & Longobardi (1991: 155) propose the following parameter:Possessiveelementsaresyntacticallyspecifiedtoberealizedonthesurfaceeitheras adjectives(asinItalian),orasdeterminers(asinEnglishandFrench).Thefollowingexamples fromEnglishandItalianillustratethedifferencesofthetwotypesofpossessives.

(7) (*the) my books (8)a. il mio libro themybook ‘mybook’ b. un mio libro amybook ‘abookofmine’ c. i miei libri themybooks ‘mybooks’

Another difference between the two types of possessives (Giorgi & Longobardi, 1991,

Schoorlemmer,1998,citedinAlexiadou,2004)isthatpossessivepronounsinAGlanguages,just likeotheradjectives,canbeusedinellipsisandpredicativecontexts,whileinDGlanguages,a distinctpronominalform,differentfromtheprenominalone,isusedinellipsisandpredicative contexts.Thefollowingexamplesillustratethispoint.TheItalianexamplesarefromGiorgi&

Longobardi(1991)andtheEnglishonesarefromAlexiadou(2004).

134

(9)a.JohnwillbringhiscarandIwillbring mine .(ellipsiscontext) b.Thisbookismine.(predicativecontext) (10)a. Metti i tuoi libri vicini ai miei .(ellipsiscontext) puttheyourbooksnexttomy ‘Putyourbooksnexttomine.’ b. Questo libro e mio .(predicativecontext) thisbookismy ‘Thisbookismine’

Alexiadou(2004:33)notesthatOldEnglish,OldFrenchandOldHighGermanwereAG languages. In those languages, possessives could cooccur with definite determiners and then theyunderwentachangeinwhichpossessiveadjectivesbecamepossessive determiners. This change coincided with the emergence of distinct pronominal forms used in ellipsis and predicative contexts inEnglish, due to the breakdown of agreement morphology. Like Lyons

(1999),Alexiadou assumes that the possessivedoes not have an inherent specification of the

(in)definitenessfeatureandthat(in)definitenessofthematrixnominalisrelatedtothestructural positionthatthepossessiveoccupies.

Mandarin,anisolatinglanguage,lacksagreementmorphology,yettypologicallyitseems tobeanAGlanguage,asMandarinpossessivessharesomefeaturesofadjectives.Thepossessive hasthefollowingadjectivelikefeaturesinMandarin.Firstly,thereisamarker(‘ de ’)between the possessor and the head noun/the following elements in the nominal expression, just as a relativeclause(12)oranattributiveadjective(13)needs de toconnectwiththeheadnoun.

135

(11) wo de shu IDEbook ‘mybook(s)’,‘somebooksofmine’ (12) wo gang mai de shu IjustbuyDEbook ‘thebook(s)thatIhavejustbought’ (13) youyong de shu usefulDEbook ‘usefulbook(s)’

Secondly,thepossessive,likeanadjective,canoccurinellipsisandpredicativecontextsas shownin(14)and(15)respectively.

(14)a. ta chi-le da-de pingguo, wo chi-le xiao-de . heeatPerfbigDEappleIeatPerf smallDE ‘Heatethebigapple,Iatethesmallone.’ b. ta chi-le ta-de pingguo, wo chi-le wo-de. heeatPerf heDEapple I eatPerf IDE ‘Heatehisapple,Iatemine.’ (15)a. zhe ben shu youyong . ThisClbookuseful ‘Thisbookisuseful.’ b. zhe ben shu (shi) wo-de . ThisClbook(be)IDE ‘Thisbookismine.’

136

Thirdly, the position of the possessive in a nominal expression is flexible and it can cooccurwithdemonstratives,justlikeadjectivesinMandarin.Comparethefollowingtwopairs ofexamples.

(16)a. wo-de na san ben shu IDEthatthreeClbook ‘thosethreebooksofmine’ b. na san ben wo-de shu thatthreeClIDEbook ‘thosethreebooks,whichbelongtome’ (17)a. youyong-de na san ben shu usefulDEthatthreeClbook ‘thosethreeusefulbooks’ b. na san ben youyong-de shu thatthreeClusefulDEbook ‘thosethreebooks,whichareuseful’

(16)and(17)illustraterespectivelythattheMandarinpossessiveandtheadjectivecanprecede thedemonstrativedeterminerorcomeafterthe[Dem+Nume+Cl]inanominal.

Lastly,severalpossessivescanoccurinanominalexpression,justasseveraladjectivescan modifyoneheadnoun.

137

(18)a. wo-de Zhao Yuanren-de yuyanxue-de shu IDE ZhaoYuanrenDE linguisticsDE book ‘ChaoYuanren’sbookonlinguisticsthatbelongstome’ 36 (Tang,1990:422) b. yi ge weida-de, yonggan-de, qinlao-de minzu 37 oneClgreatDEbraveDE hardworkingDE people ‘agreat,brave,hardworkingpeople’

In(18a),twopossessivesareused,oneexpressingthethetaroleofpossession( wo-de ‘my’) and

theotherofagent( Zhao Yuanren-de ‘ZhaoYuanren’s’)andin(18b)threeadjectivesmodifythe

headnoun.

TheabovementionedcharacteristicsindicatethatthepossessiveinMandarin,likethatin

Italian,isapossessiveadjective,notapossessivedeterminer.However,unlikeItalian,Mandarin

does not have any (in)definite article and the (in)definite interpretation of its possessive

containingnominalsdependsoncontext.

Tothebestofmyknowledge,therehasbeennosystematicstudy aboutthepossessive

construction in Mandarin. With regard to its interpretation, there are conflicting ideas among

differentscholars.Huang(1982,1987)andChen(2004)bothtouchuponthe(in)definitenessof

thepossessivecontainingnominal.Huang(1987)mentionsinpassingthatpossessivecontaining

nominalsinMandarinmaybedefiniteorindefinite,dependingon context.Let’s considerthe

followingtwoexamplestakenfromHuang(1987:239).

36 TheEnglishtranslationwassuggestedbyDr.MarlyseBaptista. 37 Whenseveraladjectivesmodifyanoun,themodificationmarker‘ de ’canbedeletedexceptfortheonefollowing thefinaladjective.

138

(19) ta da-le wo-de pengyou . hehitPerfmyfriend ‘Hehit(one/someof)myfriends.’ (20) you wo-de pengyou zai wuzi li . havemyfriendatroomin ‘Thereis/area/somefriend(s)ofmineintheroom.’

Thepossessivepronounisfollowedbytheheadnounintheabovetwoexamples.Accordingto

Huang,thepossessivecontainingnominalin(19)maybedefiniteorindefiniteandtheonein

(20),whichisanexistentialsentence,canonlyhaveanindefiniteinterpretation.Ontheother hand,Huang(1982:64)alsopointsoutthatthefollowingexample(21)isungrammatical.

(21)* you Zhang San de san ben shu zai zher . existZhangSanDEthreeClbook at here

Both(20)and(21)areexistentialsentencesandtheonlydifferencebetweenthenominalin(20) andthatin(21)isthatthereisthenumeral san ‘three’betweenthepossessorandtheheadnoun in(21).Huang’sjudgmentofthe(un)grammaticalityof(20)and(21)seemstosuggestthatthe

[Poss+Nume+Cl+N]nominalinMandarinonlyhasadefiniteinterpretation,whilethe[Poss

+ N] nominal can be both definite and indefinite. I cannot see any reason why (20) is grammaticaland(21)isungrammatical.

Chen (2004), however, holds just the opposite view. He claims that whether the possessivecontainingnominalistobeinterpretedasidentifiableornonidentifiableinMandarin

139 dependsonthepresenceorabsenceofanindefinitedeterminerbetweenthepossessorandthe headnoun(p.1157).Hegivesthefollowingexamplesforillustration.

(22) zhe shi wo-de yi ge pengyou gaosu wo de . thisbemyoneClfriendtell I DE ‘Afriendofminetoldmethis.’ (23) Wo-de qianbi zenme zhao bu dao le ? mypencilhowfindnotarriveCRS 38 ‘HowcomeIcannotfindmypencil/*apencilofmine?

Chen (2004) argues that the possessor phrase imposes a definite interpretation on the matrix nominal when there is no indefinite determiner in between and that the nominal is indefinite whenanindefinitedeterminerisinsertedbetweenthepossessorphraseandtheheadnoun.So,in

Chen’sview,theMandarin[Poss+N]nominalonlyhasadefiniteinterpretationandthe[Poss+

Nume+Cl+N]nominalisindefinite.AlthoughChen(2004)saidthatthepossessorphrasein

(23)imposesadefiniteinterpretation,thenominalstillhasadefiniteinterpretationevenifthe possessivepronounisdeleted. Theirexamplesentencescited aboveandtheirinterpretations indicatethatHuangandChenhavedifferentoroppositeviewsaboutthe(in)definitenessofthe

Mandarinpossessivecontainingnominal.

Except for the interpretation of (21), I agree with Huang’s and Chen’s respective interpretations of their examples. My intuition tells me that (21) is perfectly grammatical. I 38 CRSstandsforCurrentlyRelevantState.Asasentencefinalparticle,thebasicfunctionof le istosignala ‘CurrentlyRelevantState’,whichmeans“astateofaffairshasspecialcurrentrelevancewithrespecttosome particularsituation”(Li&Thompson,1981:240).

140 assumeherethatthepossessivecontainingnominalinMandarincanhavedefiniteandindefinite interpretations,nomatterwhetherornotthereisanindefinitedeterminerbetweenthepossessor phraseandtheheadnoun,justasbarenounscanhave definite and indefinite interpretations.

Whetherapossessivecontainingnominalhasadefiniteorindefiniteinterpretationdependson both linguistic and situational context. For example, let us examine how different contexts influencetheinterpretationoftwonominals‘ wo-de shu’ (literally‘mybook’)and wo-de liang ben shu (literally‘mytwobooks’).

(24)A. ni weisheme bu gaoxin? your whynothappy ‘Whyareyounothappy?’ B. wo-de shu /wo-de liang ben shu bei ren tou le. IDEbook/IDEtwoClbookbe person steal SFP ‘A(some)book(s)ofmine/twobooksofminehas/havebeenstolen.’

In (24), the nominal usually can only have an indefinite interpretation. The possessive containingnominalinB’sanswerlacksthetwomainsemanticfeaturesofadefiniteexpression: identifiabilityanduniqueness/inclusiveness.Thereferentsofthenominal(i.e.,the(two)book(s) stolen)arenotidentifiabletoA.Inanormalsituation,apersonhasmorethantwo.

However,wecanalsoimagineacontextinwhichthetwopossessivecontainingnominals haveadefiniteinterpretation.Forexample,twoclassmateseachtooktwobookstothelibrary andwerepreparingforatesttogetherthere.Afteronehour,theywentoutforabreakandleft their bookbags on the table.When they returned, one found his bookswere gone. In such a

141 context,thenominalshaveadefiniteinterpretationand(24B)shouldbetranslatedinto‘My(two) bookshavebeenstolen’.Thehearerknowsthatthespeakerhasonlytwobooksandheknows whichtwobookswerestoleninthatsituation.

OnethingthatIneedtopointouthereisthattheEnglishpossessiveconstructionhasthe feature of uniqueness/inclusiveness of a definite expression. So the English possessive

containingnominal‘mythreebooks’hastheimplicationthatthespeakerhasonlythreebooks.

Thepossessivecontainingnominalofthe[Poss+Nume+Cl+N]sequenceinMandarindoes nothavesuchanimplication,soitcanbeinterpretedasdefiniteandindefinite,dependingon contextualfactors. Lyons(1999:278)proposesthat definiteness (or what he informally terms

‘semantic/pragmaticdefiniteness’)inlanguages likeChinesewhereitisnotgrammaticalized normallylackstheimplicationofuniqueness/inclusiveness. Nongrammaticalizeddefiniteness means definiteness that is not indicated by some grammatical means like articles a, the in

English.Asasemanticopragmaticnotion,definitenessisalanguageuniversal;thereferentsof somenominalsineverylanguageareidentifiableorknowntotheparticipantsincommunication.

However, not every language has grammatical devices to indicate the (in)definiteness of a nominalexpressionandbarenounsinsomelanguagescanhavedefiniteinterpretation.Chinese is such a language. Lyons (1999: 27) also notes that neither inclusiveness nor identifiability applies to some definite possessive nominals in English. The referents of those nominals are usuallyinalienableorotherintimatepossessions.Considertheexamples(25a)and(26b)taken fromLyons(1999:26).

142

(25)a.I’mgoingtostaywith my brother forafewdays. b.Joehasbroken his leg .

Thepossessivenominalin(25a,b)doesnotimplythatthespeakerhasonlyonebrotherorJoe hasonlyoneleg.Wecanreplace‘mybrother’with‘abrotherofmine’and‘hisleg’with‘oneof hislegs’in(25a,b).So,thedefinitepossessiveconstructioninEnglishcanhaveanindefinite sense(Lyons,1999:128).

Thepossessivecontainingnominalswediscussedabovearethoseinwhichthepossessive isintheinitialpositionfollowedeitherbytheheadnounorthe[Nume+Cl+N]sequence.

Whenthepossessiveoccursafterthe[Nume+Cl]sequenceoritcooccurswithademonstrative inanominal,the(in)definitenessofthematrixnominaliseasytodetermine.Ifthepossessive cooccurswithademonstrativeinaMandarinnominal,thenominalhasadefiniteinterpretation, nomatterwhetherthepossessiveorthedemonstrativeisintheinitialposition.Irepeat(4)and(6) in6.1.1hereforconvenience.

(4) wo de zhe (san) ben shu I DE this (three) Cl books ‘this(thesethree)book(s)ofmine’ (6) tai xiang jie zhe (san) ben wo de shu . hewantborrowthis(three)Cl IDEbook ‘Hewantstoborrowthis(thesethree)book(s),whichbelong(s)tome.’

143

Speakers have different views about whether the two nominals in (4) and (6), involving a possessiveandademonstrative,areinfreevariationorhavedifferenceinmeaning.Myintuition tells me that there is slight difference in meaning. I will discuss the interpretation and the syntactic structure of nominals in which the possessive and the demonstrative determiner cooccurinsection6.3.

Whenthe[Nume+Cl]sequenceprecedesapossessive, the matrix nominal has an indefiniteinterpretationanditisusuallyconsidered to have a partitive reading (Huang, 1982;

Tang,1990).Irepeat(5)hereforillustration.

(5) ta na le san ben wo de shu . hetakePerfthreeClIDEbook ‘Hetookthreeofmybooks/threebooksofmine.’39

FollowingAlexiadou(2004,2005)andLyons(1999),Iassumethatthepossessivedoes notinherentlyencodethedefinitenessfeatureandthatdefinitenessinthepossessivecontaining nominalisrelatedtothestructuralpositionthatthepossessorphraseoccupies.AsIdiscussed above, the possessive in Mandarin behaves like an adjective and it can occur in different positions in the possessivecontaining nominal. I conclude that the possessive occupies the specifierpositionofdifferentphrasalcategories.Sothedefiniteandindefiniteinterpretationsof thepossessivenominalof wo-de san ben shu (literally‘IDEthreeClbook’,meaningeither‘my threebooks’or‘threebooksofmine’)canberepresentedas(26)and(27)respectively.

39 Tome,thereisnocompellingreasontoconsidertheobjectnominalin(5)tohaveapartitivereading.

144

(26)

‘mythreebooks’

145

(27)

‘threeofmybooks’

Asthetwotrees(26)and(27)show,Iassumethatthepossessiveofthedefinitenominalisinthe

[SpecDP],andthatoftheindefinitenominalisinthe[SpecNumeP].

Besidestheflexibilityofitspositioninanominal,theMandarinpossessivecontaining

nominal has another feature: multiple possessives can occur in one nominal. These multiple

possessivesrealizedifferentthetaroles.Besidesthethetaroleofpossessorproper,thepossessive

inMandarincanalsorealizethethetarolesofagentandtheme.Whenmultiplepossessivesarein

onenominal,theorderofthepossessivesrealizingdifferentthetarolesisfixed.Itconformsto

146 thepossessivizationhierarchythatCinque(1980,citedinDimitrovaVulchanova&Giusti,1999:

170)proposesforItalian:the“possessor>agent>theme”hierarchy.

Whenonlyonepossessiveoccursinanominal,thepossessive can realize any of the aforementionedthreethetaroles.Considerthefollowingnominal.

(28) Lu Xun-de shu LuXunDE book ‘(a/the)book(s)writtenbyLuXun’,‘(a/the)book(s)ownedbyLuXun’‘(a/the)book(s) aboutLuXun’

Example(28)canmean‘(the)book(s)ownedbyLuXun’,inwhichcasethepossessiverealizes thethetaroleofpossessor.When(28)means‘(the)book(s)writtenbyLuXun’,thepossessive realizesthethetaroleofagent.If(28)means‘(the)book(s)onLuXun’,thenthethetaroleof themeisrealized.Similarly,theEnglishpossessive can also express the three theta roles. For example,John’spicturecanmean‘thepictureownedbyJohn’(possessor),or‘thepicturetaken byJohn’(agent)or‘thepictureofJohn’(theme).

Whentwopossessivesareusedinanominal,theordersofthethetarolesrealizedare possessor>agent, possessor>theme, agent>theme, which cannot be reversed. Consider the followingpossessivecontainingnominals.

(29)a. ta-de Zhao Yuanren-de shu (possessor>agent) heDE ZhaoYuanrenDE book ‘a/thebook(s)writtenbyZhaoYuanrenandownedbyhim’

147

b. ni-de Xianglinsao-de xiaoshuo (possessor>theme) youDE XianglinauntDE novel ‘thenovelaboutAuntXianglinownedbyyou’ c. Luxun-de Xianglin-sao-de xiaoshuo (agent>theme) LuxunDE XianglinauntDE novel ‘thenovelaboutAuntXianglinwrittenbyLuxun’

Theinterpretationsofthenominalsin(29)givenabovearetheusualonesinnormalcontext, since Zhao Yuanren (aChineselinguist)and Xianglinsao (acharacterinanovel)arewellknown toeducatedChinesepeople.Ofcourse,thethreenominalscanhaveotherinterpretations.For example,(29a)canmean‘a/thebookaboutZhaoYuanrenwrittenbyhim’,inwhichcasethe agent and the theme theta roles are realized. However the nominals are interpreted, the possessivizationhierarchyisstillobserved.

Whenthreepossessivesareusedinonenominal,theonlyorderthatthethreethetaroles arerealizedispossessive>agent>themeasthefollowingexampleshows.

(30)? Zhang San-de Luxun-de Xianglin-sao-de xiaoshuo40 ZhangSanDE LuxunDE XianglinauntDE novel ‘thenovelaboutAuntXianglin,writtenbyLuxun,ownedbyZhangSan’

In this section, we discussed the nominal with the possessive in the initial position.

Next,wedealwiththenominalwithpersonalpronounsorpropernamesintheinitialposition. 40 Thissentenceseemsnotquitegrammaticaltosomespeakers,althoughitisunderstandable.Theytendtousea relativeclauseinsteadofthepossessiverealizingtheagentthetaroleoruseaprepositionphraseforthepossessive realizingthethemethetarole.

148

6.2 Personal Pronouns and Proper Names

Personal pronouns and proper names are inherently definite. They can function independently as a nominal expression. They can also be part of a nominal expression, functioning as a kind of definite determiners in Mandarin. I analyze Mandarin nominal expressions with personal pronouns in the initial position in 6.2.1 and discuss nominal expressionscontainingpropernamesin6.2.2.

6.2.1 Personal Pronouns as Determiners in Mandarin

PersonalpronounsinMandarincanbefollowedbyanoun(31),a[Nume+Cl+N] sequence(32)ora[Dem+Nume+Cl+N](33).

(31) ni-men laoshi hen hao .41 YouCol teacher very nice ‘Youteachersareverynice.’ (32) tai-men san ge ren qu dianyinyuan le. HeColthreeClpersongocinema SFP ‘Thethreeofthemwenttothecinema.’ (33) wo-men zhe san ge xuesheng zhoumo dai zai xuexiao . IColthisthreeClstudentweekend stay at school. ‘Wethreestudents/Thethreeofusstudentswillstayoncampusattheweekend.’

41 Theexpression ‘ni-men lao shi ’isambiguous.Itcanmeanboth‘youteachers’and‘yourteacher(s)’.Ifitmeans ‘yourteacher(s)’,themodificationmarker‘ de ’canbeinsertedafterthepronoun( ‘ni-men de lao shi ’,whichis unambiguous.)

149

Whatistherelationshipbetweenthepersonalpronounsandthefollowingnoun,the[Nume+Cl

+N]sequenceand[Dem+Nume+Cl+N]sequenceintheabovethreeexamples?Aretheyjust twonominalsinappositionordotheyformonenominalconstituentwiththepersonalpronouns functioningasdeterminers?

InEnglishlinguistics,somescholars(Postal,1969;Abney 1987; Lyons, 1999) argue thatpersonalpronounsaredeterminers.Postalclaimsthatthesocalledpersonalpronounsare definitearticlesandthatpronounslike I, she derivetransformationallyfromtheunderlyingforms

Ione , sheone .Thehypotheticalpronounstem one surfacesinnominalslike you great ones , we religious ones (Postal,1969).Lyons(1999:141)callspersonalpronounsoccurringwithinfull nounphraseslike we men , you professors ‘personaldeterminers’.Lyonsobservesthesimilarity betweenpersonalpronounsandpronounsofothertypeslike this , that (demonstrativepronouns), one , some (indefinite pronouns), in that they all can occur prenominally as determiners and independently as pronouns. On the other hand, other scholars (Hawkins, 1974; Delorme &

Dougherty,1972)considerthepersonalpronounandthenouninnominalexpressionslike we men tobeinanappositionalrelationshipwitheachother.Jespersen(1954,citedinPostal,1969) impliesthatnominalslike we men arederivedfromappositiverelativeclauses;thatis, we men is the reduced form of we, who are men. However, Postal argues that this is not the case, as nominals like we men can occur in contexts where nominals cannot have appositive relative clauses.Forexample,certainnegativecontextsexcludeappositiverelativeclauses,butnominals like you men canoccurinsuchcontexts,asthefollowingexamples(takenfromPostal,1970:70) show.

150

(34)*Theyneverinsultedthemen,whowereDemocrats. (35)Theyneveragreedwithusplanners.

IfPostal’sandLyons’analysesofnominalexpressionslike we men areontherighttrack,can personalpronounsinMandarinfunctionasdeterminerstoo?TheMandarinnominalslike ni-men laoshi (‘youteachers’)cannotbeconsideredasthereducedformofanappositiverelativeclause, as Mandarin relative clauses are all prenominal. Can the personal pronoun and its following elementbeinanappositionalrelationship?

Let’sdiscussverybrieflythelinguisticphenomenonof‘appositive’.Anappositiveisa wordorgroupofwordsthatmeansthesamethingas the element that precedes it (Skwire &

Wiener,2005:630).So,whentwonominalsareinapposition, they should refer to the same entity or entities. Appositives can be either restrictive or nonrestrictive. A nonrestrictive appositivegivesadditionalinformationtotheprecedingnoun,buttheadditionalinformationis notessentialtoidentifythenoun,whiletheinformationsuppliedbyarestrictiveappositiveis necessary to identify the noun it follows (Skwire & Wiener, 2005: 630). In English, nonrestrictiveappositivesareseparatedfromotherelementsinasentencebyinwriting.

In speech, nonrestrictive appositives characteristically have the socalled intonation.

ConsiderthefollowingEnglishsentences.

(36)ThepopularU.S.president John Kennedy wasknownforhiseloquentspeeches. (37)JohnKennedy, the popular U.S. president ,wasknownforhiseloquentspeeches.

151

In(36),theappositive‘JohnKennedy’isrestrictive.Withouttheappositive,wedonotknow which popular U.S. president the speaker is talking about. The appositive ‘the popular U.S. president’in(37),whichissetoffbycommas,ontheotherhand,addsnonessentialdescriptive informationaboutJohnKennedy,withoutwhichthemeaningofthesentenceisstillcomplete.

The nonrestrictive appositive in Mandarin can be separated from other elements by commasinwriting,whiletherestrictiveonecannot.ConsiderthefollowingMandarinexamples, whicharesimilartothesubjectsof(36)and(37)instructure.

(38) weida-de zuojia Luxun yi ta de duipian xiaoshuo wenming . greatDE writer Luxun with heDE short novel wellknown ‘ThegreatwriterLuxunwaswellknownforhisshortstories.’ (39) Luxun, yi ge weida-de zuojia, yi ta de duipian xiaoshuo wenming . Luxun,oneCl greatDEwriter withheDE short novel wellknown ‘Luxun,agreatwriter,waswellknownforhisshortstories.’

In(38) Luxun isarestrictiveappositive,while yi ge weida-de zuojia (‘agreatwriter’)in(39)isa nonrestrictiveappositive,separatedbycommas.

Itseems,however,thatthecommaintonationtypicalofnonrestrictiveappositivedoes notalwaysapplyinspeechandanonrestrictiveappositiveisusuallynotsetoffbycommasin writinginMandariniftheprecedingelementisapersonalpronoun.However,thespeakercan alwaysusethecommaintonationinproducingthenonrestrictiveappositiveandseparateitfrom otherelementsbycommasifhewantsto.Considerthefollowingexample.

152

(40) wo yi ge qiong jiaoshi mei nengli mai fangzi . IaClpoorteachernotabilitybuy house a.‘I,apoorteacher,can’taffordtobuyahouse/anapartment.’ b.‘Iamtheonlypoorteacherwhocan’taffordtobuyahouse/anapartment.’(implying otherpoorteacherscanaffordtobuyahouse/apartment.)

Example (40) is ambiguous. When it has (40a) as its meaning, yi ge qiong jiaoshi ‘a poor teacher’isanonrestrictiveappositive. Thespeakercanhavethecommaintonationinsaying(40) orhavetheappositivesetoffbycommasifhewantsto,buthedoesnothaveto.Theappositive seemstoexplainwhythespeakercannotaffordtobuyahouse. Ifthemeaningis(40b),the pronoun wo ‘I’isusuallystressedandtherecannothaveapauseafteritinspeech.Thepronoun limitsthe reference of its following element. So, I would consider the pronoun wo ‘I’ as a determinerifthe(40b)interpretationistheintendedone.Similarly,inEnglish, we and you are alwaysstressedandwithafullvowelwhenusedas determiners and they can be stressed or unstressedwhenusedaspronouns(Sommerstein,1972;citedinLyons,1999:2728).

Letus examinetheMandarinnominalscontainingpersonal pronouns to see whether they have the typical features of appositives. Firstly, in some nominal expressions with a personalpronoun,thepersonalpronounanditsfollowingelementdonothavethesamereferent.

Forexample:

153

(41) wo zuotian pengdao ta liang kouzi .42 Iyesterdaymeethetwoperson ‘YesterdayImethimandhiswife.’

In (41), the personal pronoun ta ‘he’ is singular and the following element liang kouzi ‘two persons’/‘couple’referstotwoindividuals.

Secondly, if the personal pronoun and its following element are in apposition, the appositive should be nonrestrictive, since the information that it adds to the sentence is not crucialtotheidentificationofthereferentofthepersonalpronoun.However,inspeech,nominal expressionslike ni-men laoshi (‘youteachers’)canneverhavethecommaintonationwhichis possiblethoughnotobligatoryforMandarinnonrestrictiveappositivesandinwritingnocommas areinsertedafterthepersonalpronounsastheungrammaticalityof(42)illustrates.

(42)* ni-men, laoshi, dou hen fuzhe youCol teacher all very responsible

Anotherlinguisticfactrelevanttomyargumenthereisthateitherofthetwonominalsin appositioncanstandassubjectorobjectofasentence,sincetheydesignatethesameentity.Take

(43)forexample.

(43)MybrotherJohnisolderthanyourbrotherDavid. 42 ThesentencewasprovidedbyZhonghengJiainemailcorrespondence.Thepluralmarkermen canbeaddedto ta ‘he’inthesentence.

154

In (43), ‘my brother’ and ‘John’ are in apposition and ‘your brother’ and ‘David’ are in apposition too.So, instead of saying (43), one can say either ‘My brother is older than your brother’or‘JohnisolderthanDavid’.

ThisisnotthecasewithMandarinnominalslike wo-men nanren ‘wemen’and ni-men nüren ‘youwomen’.Onecansay:

(44) wo-men nanren bi ni-men nüren liqi da. IColmanthan youColwomanstrength big ‘Wemenarestrongerthanyouwomen’

But when we delete the personal pronouns wo-men ‘we’ and ni-men ‘you’ in (44), then the nominals nanren ‘man’ and nüren ‘woman’ in the resulting sentence (45) have a generic interpretationinnormalcircumstances:

(45) nanren bi nuren liqi da . manthanwomanstrengthbig ‘Menarestrongerthanwomen.’

Thedifferentinterpretationsofthenominalsin(44)and(45)showthatthepersonalpronouns andthefollowingnounsin(44)havedifferentreferentsandsotheyarenotinapposition.Their relationshipiscomparabletotheonebetween‘this’/’that’and‘book’inthefollowingEnglish sentence(46).

155

(46)Thisbookismoreinterestingthanthatbook.

So,Iconcludethatpersonalpronounsinnominalslike wo-men nanren ‘wemen’aredeterminers.

InMandarin,onlypluralpersonalpronounscanbeusedthiswaywhenthefollowingelementis abarenoun;singularpersonalpronounscannotasexample(47)shows.

(47)a. *ni xuesheng (literally‘you(sg.)student’) b.* ta laoshi (literally‘hestudent’) c.*wo laoshi (literally‘Istudent’)

Icannotcomeupwithanexplanationofwhythisisso.Interestingly,onlythepluralfirstand secondpersonpluralpronounsinEnglishcanfunctionasdeterminers( we students , you teachers ).

ThethirdpersonpluralpronouncannotbeusedthiswayinstandardEnglish( *them scholars ) andthesingularpersonalpronounscannotfunctionasdeterminersatall(* I student ,* he student ).

Whenthepersonalpronounisfollowedbythe[Nume+Cl+N]sequence,thestructure canbeambiguousasexample(40)illustrates.Hereisanotherexample.

(48) ta yi ge xiao haizi bu hui zuo zhe jian shi . heoneClsmallchildnotabledothis Cl thing a. ‘He,asmallchild,isunabletodothisthing.’ b. ‘Heistheonlysmallchildwhoisunabletodothisthing.’

Iftheintendedmeaningis(48a),thentherecanbeapauseafterthepronoun ta ‘he’inspeech.In thatcase,thepronounandthefollowing[Nume+Cl+N]sequenceareinapposition.When

156

(48b)istheintendedmeaning,therecannotbeapauseafterthepronoun ta ‘he’andthepronoun functions as a determiner and is usually stressed. When followed by the [Nume + Cl + N] sequence,boththepluralandsingularpersonalpronounscanfunctionasdeterminers.

In certain contexts, the personal pronoun in Mandarin functions like a demonstrative determiner.Supposethatateacherisassigningstudentstocleantheclassroomandsayingthe followingsentence(49).

(49) ni-men liang ge tongxue tuo diban, ni-men liang tongxue youColtwo Clstudentmopfloor youCol two student ca chuanghu . cleanwindow ‘Youtwostudentsmopthefloorandyoutwostudentscleanthewindows.’

Whensaying(49),theteacherusuallypointstothestudents.Thepersonalpronounsusedin(49) arelikethedemonstrativesinthefollowingEnglishsentence.

(50)Thesetwobooksaremineandthosetwobooksareyours.(thespeakerpointingtothe bookswhensayingthesentence)

WherearepronounsbasegeneratedinMandarin?Cheng&Sybesma(1999:509)claim thatpronounsinMandarinarebasegeneratedinNandthenitmovedtoCl,as,intheirview,

MandarindefinitenominalexpressionsareClPs,notDPs. Insupportoftheirargument,they

157 citethefactthatpronouns,likepropernames,can be preceded by numerals and classifiers in

Mandarin.Let’sconsidertheirexample.

(51) Cong nei-ge jing-zi, wo keyi kandao wu-ge wo . fromthatClmirror Icansee fiveCl I ‘Fromthatmirror,Icanseefivecopiesofmyself(fiveI’s/me’s).’

Apparently,itistruethatthe[Nume+Cl]sequenceprecedesthepronoun wo ‘I’.Intheir

Englishtranslation,theyput‘fiveI’s/me’s’intoparentheses.Ontologicallyspeaking,‘I’canrefer toonlyoneperson,i.e.,thespeaker.Actually,the pronoun wo ‘I’means‘imageofmeinthe mirror’in(51).So,themeaningof wo ‘I’in(51)isdifferentfromthemeaningof wo ‘I’in(52), which cannot be preceded by the [Nume + Cl] sequence. Chao (1968: 628) notices that the nonpronominalnatureofsuchuseofpronounsasin(51)andregarded wo ‘I’asquasiquoted.

(52) wo shi xuesheng . Ibestudent ‘I’mastudent.’

BasedontheusageofMandarinpronounsandfollowingotherscholars(Postal,1967;

Abney, 1987; Lyons, 1999), I assume pronouns are definite determiners and pronouns are generatedinDinMandarin.AlsofollowingAbney(1987),Iassumethatpronounsdonottakea nominal complement when they are used independently. So, nominal expressions like ta-men wu-ge xuesheng (literally‘heColfiveClstudent’)havethefollowingstructure(53).

158

(53)

‘thefivestudentsofthem’

Astothestructureofthenominalinwhichthepersonalpronounisfollowedbythe[Dem

+(Num)+Cl+N]sequence,Idiscussitinsection6.3,wherenominalexpressionsinvolving demonstrativedeterminersaretreated.

Ianalyzenominalsinvolvingpropernamesinthenextsubsection.

159

6.2.2 Mandarin Nominals Containing Proper Names

Thissubsectionisconcernedwiththeuseofpropernamesinafullnominalexpression inMandarin.First,Ipresentsomeexamplestoillustratetheircollocationwithotherelementsin thenominal.ThenIanalyzetheirsyntacticstructures.Thenominalinwhichpropernamesand demonstrativedeterminerscooccurwillbediscussedinsection6.3.

PropernamescancollocatewithseveralelementsinaMandarinnominal.Theycanbe followedbythethirdpersonpronouns

(54) Zhang San ta bu qu . ZhangSanhenotgo ‘ZhangSanwillnotgo.’ (55) Zhang San ta-men bu qu . ZhangSan heCol not go ‘ZhangSanandtheotherswillnotgo.’ orbythe[Nume+Cl+N]sequence

(56)a. Zhang San yi-ge xuesheng hui jiao ni shenme ? ZhangSanoneClstudentcanteachyou what ‘As/beingastudent,whatcanZhangSanteachyou?’ b. quan ban zhiyou Zhang San yi-ge xuesheng jige . whole class only ZhangSan oneCl student pass ‘ZhangSanistheonlystudentintheclassthatpassedthetest.’/ ‘OnlyZhangSanpassedthetestintheclass.’

160 orbythe[Pron+Nume+Cl+N]

(57) Zhang San ta-men san-ge xuesheng bu qu. ZhangSanheColthreeClstudentnot go ‘ZhangSanandthetwootherstudentswillnotgo.’ orbythe[Dem+(Nume)+Cl+N]. (58) Zhang San zhe-ge ren hen hao. ZhangSan thisCl person very good. ‘ZhangSanisaveryniceperson.’ (59) Zai xiao-menkou wo pengdao Zhang San zhe san-ge xuesheng .43 at schoolgate I meet ZhangSan this threeCl student ‘ImetZhangSanandtheothertwostudentsatthegateoftheschool.’

Inaddition,propernamescanalsobeprecededbythe[Dem+Cl]sequence.

(60) Zhe-ge Zhang San zhen bu jiangdaoli. ThisCl ZhangSan really not reasonable. ‘ThisZhangSanisreallyunreasonable.’

Fromtheaboveexamples,wecanseethatpropernames,whicharedefinite,cancooccurwith otherelements,whichareeitherindefinite(56)ordefinite(54),(55),(57),(58),(59),and(60).

43 Imyselfwillnotusesuchnominalsas Zhang San zhe san-ge xuesheng (literally‘ZhangSanthisthreeCl student’).SuchnominalscametomyattentionwhenIheardapersonusingitinaTVprogram.LichaoHe(p.c.), whoisfromBeijing,confirmedthatsuchnominalsaregrammaticaltoher.

161

What is the relationship between the proper name and its following element? Are they in appositionwitheachother?

Itwasmentionedin6.2.1thattwonominalsinanappositionalrelationshiprefertothesame entityorentitiesandthatthereisacommaintonationiftheappositionisnonrestrictive.Asthe propernameitselfisdefinite,theappositivemustbenonrestrictiveiftheyareinapposition.Let’s applythesecriteriatothenominalexpressionsintheaboveexamplestoseeifthepropername anditsfollowingelementsareinapposition.

Zhang San andthethirdpersonsingularpronoun ta ‘he’in (54) refer to the same person named Zhang San . In addition, there can be a pause between them in speech. So, it can be concludedthattheyareinapposition.Inthelinguisticliterature,pronounslike ta ‘he’in(54)are calledresumptivepronouns.Thestructureofthesubjectin(54)isaDP,whichitselfiscomposed oftwoDPs.Itcanberepresentedasfollows(61).

(61)

When Zhang San isfollowedbythethirdpersonpluralpronoun ta-men ‘they’asin(55),

162

Zhang San andthepluralpronouncannothavethesamereferent(s) 44 .Inspeechthereisnopause between them. That means they cannot be in apposition. In 6.2.1, I argue that pronouns in

MandarinareinD,astheycanbefollowedbythe[Nume+Cl+N]sequence.Sotheproper namecanonlybeinthespecifierpositionoftheDP.Thestructureofthenominal Zhang San ta-men (literally,‘ZhangSanthey’)in(55)canberepresentedas(62).

(62)

‘ZhangSanandtheothers’

In (56), Zhang San yi-ge xuesheng (literally ‘Zhang San oneCl student’), just like ta yi-ge xuesheng (literally‘heoneClstudent’)discussedin2.6.1,isambiguous.In(56a),therecan beapausebetween Zhang San and yi-ge xuesheng ,althoughthespeakerdoesnotnecessarily have to use the comma intonation in speech. They are two nominals in apposition. The nonrestrictive appositive yi-ge xuesheng canbeconsideredtofunctionadverbially.Itranslate theappositiveinto‘being/asastudent’.(56a)hastheimplicationthat‘ZhangSan,being/asa student,canteachyounothing.’Ontheotherhand, Zhang San yi-ge xuesheng in(56b)cannot have a pause after the proper name Zhang San and Zhang San specifies the referent of yi-ge

44 YuanrenChao(1968)considersthethirdpersonpluralpronoun ta-men usedafterpropernamesasaplural suffix.

163 xuesheng ‘astudent’.Iassumethat Zhang San functionslikeadeterminerandisbasegenerated inDandtakes yi-ge xuesheng ‘astudent’asitscomplement.Sothestructureofthenominalsin

(56a)and(56b)canberepresentedrespectivelyas(63)and(64). (63)

‘ZhangSan,astudent’ (64)

‘theonestudentZhangSan’

164

The subject of (57) Zhang San ta-men san-ge xuesheng (‘Zhang San heCol threeCl student’) seemstobeacombinationof Zhang San ta-men and ta-men san-ge xuesheng .Since there cannot be a pause after Zhang San or after ta-men ‘they’,itcanbeconcludedthatthey cannotbeinanappositionalrelationship.So,thethirdpersonpluralpronoun ta-men isinDand takesanumeralphrase san-ge xuesheng ‘threestudents’asitscomplement.Thepropernameis inthespecifierofDP.Thestructureofthesubjectnominalin(57)canberepresentedas(65).

(65)

‘ZhangSanandthetwootherstudents’

In summary, when proper names, personal pronouns or/and the [Nume + Cl + N] sequencecooccur,theycanbeeitherinanappositionalrelationshiporinaheadcomplementor specifierheadrelationship.Iftheyareinapposition,therecanalwaysbeacommaintonationin

165 speech.ItisclaimedthatpersonalpronounsareinDandpropernamesareeitherinDorthe specifierpositionofDP.

Inthenextsection,nominalscontainingdemonstrativedeterminersareanalyzed.

6.3 Demonstrative Determiners

ThissectiondealswithdemonstrativedeterminersinMandarin.Ifirstdiscusstheusesof demonstratives in general. Then I present examples of nominals containing demonstrative determinersandanalyzethestructuresofthesenominals.

6.3.1 Uses of Demonstratives

Demonstratives can have pronominal, adjectival and adverbial uses. Consider the followingEnglishexamples.

(66)a. This isyourtextbook. b. This textbookisyours. c.Thetextbookis this thick.

In(66a),thedemonstrative this isapronoun.In(66b) this iscalledademonstrativeadjectivein traditionalgrammar.Weuse‘demonstrativedeterminer’todesignatedemonstrativesinsuchuse asin(66b).In(66c) this functionsasanadverb.

Weareconcernedonlywiththeadjectivalusesofdemonstrativeshere.Demonstrative containingnominalsaresimilartonominalswiththedefinitearticle,propernamesandpersonal

166 pronouns; they are all definite. In addition to being definite, demonstratives are deictic: specifyingthespaciotemporallocationofareferentfromtheperspectiveofthespeakerorthe hearerinaspeechsituationorreferringtoanentityinpriororsubsequentdiscourse.

Everylanguagehasatleasttwodemonstrativeslocatingreferentsattwodifferentpoints on a distance scale; they are the proximal and distal demonstratives (Diessel, 1999). The proximaldemonstrativereferstoanentityclosetothedeicticcenter,whichusuallyisthespeaker, whilethedistaldemonstrativedirectsthehearer’sattentiontoareferentdistantfromthedeictic center.InEnglish,theproximaldemonstrativeis this andthedistaloneis that . The proximal demonstrativeinMandarinis zhe andthedistaloneis na .45

Theusesofdemonstrativesfallintofourtypes:situational,anaphoric,discoursedeictic andrecognitional(Himmelmann,1996;Diessel,1999;Chen,2004).

The situational use of demonstratives directs the hearer’s attention to entities in the situationalcontext.Theyareoftenaccompaniedbysomegesture.Whensaying(67),thespeaker probablywaspointingatthebook.

(67) ba na-ben shu gai wo . prep thatCl book give me ‘Givemethatbook.’ 45 Somescholars(Huang,1999;Lü,1980)alsoconsider zhe-ge ‘thisCl’and na-ge ‘thatCl’astheproximaland dismaldemonstrativerespectively.

167

Intheanaphoricuse,thenominalwiththedemonstrativerefersbacktoanothernominal inthepreviousdiscourse.Thetwonominalsarecoreferential;theyrefertothesameentities.

(68) gangcai you yi-ge xuesheng lei zhao ni, zhe-ge xuesheng justnow have oneCl student come lookfor you thisCl student kanshangqu san-shi shui zuoyou . look threeten year about ‘Astudentwaslookingforyoujustnow.Helooksaboutthirtyyearsold.’

In(68) yi-ge ren ‘astudent’and zhe-ge xuesheng arecoreferential.

In the discourse deictic use, the demonstrative refers to a proposition; it is not coreferentialwithanypreviousnominal.Considerthefollowingexample.

(69)A:Johngotascoreonthetest. B: That ’snottrue.

The demonstrative that in (69) does not refer back to any previous nominal; it refers to the propositionalcontentofthesentencethatAsaid.

Demonstrativesinrecognitionaluseactivatesharedknowledgeanddonothaveareferent inthepreviousdiscourse.Theycanintroduceanewreferentintothediscourse.Considerthe followingexampletakenfromHimmelmann(1996:230).

(70)ItwasfilmedinCalifornia, those dusty kind of hills that they have out here by Stocktonandall.

168

In(70),‘dustyhills…’wasintroducedintothediscourseforthefirsttime,butthedemonstrative those showsthatthehearerisfamiliarwiththe‘dustyhills’.Demonstrativesinrecognitionaluse oftenimplyemotionalcloseness,sympathyandsharedbeliefs(Diessel,1999).

Next, I analyze nominals in which demonstrative determiners cooccur with other elementsprenominallyinMandarin.

6.3.2 Demonstrative-Containing Nominals

This subsection analyzes the uses of the demonstrative determiner in Mandarin. The demonstrativeissaidtobeinherentlydefinite(Lyons,1999),butitsusesindifferentlanguages vary.Forexample,thedemonstrativedeterminercannotcooccurwithotherdefinitedeterminers like the definite article or the possessive pronoun in English, whereas in Spanish the demonstrative can cooccur with the definite article and in Bosnian the demonstrative can cooccurwithpossessivepronouns(Brugè,2002)aswitnessedinthefollowingexamples.

(71)a.*Thethisbookisveryuseful. b. el libro este (Spanish,Brugè,2002:15) thebookthis ‘thisbook’ c. knjiga ova tvoja o sintaksi (Bosnian,Brugè,2002:35) bookthisyourofsyntax ‘thisbookonsyntaxofyours’

169

DemonstrativesinMandarincanoccurtogetherwithpersonalpronouns,propernames, possessive pronouns prenominally. In the following I give some examples to illustrate the distributionsofdemonstrativesinanominal.

Demonstrativescanoccupytheinitialpositionofanominalandbefollowedbyanoun, the[(Nume)+Cl+N]sequence.

(72) Zhe/na liang ben shu shi wo-de . This/thattwoClbookbeIDE ‘These/thosetwobooksaremine.’ (73)a.Zhe/na yi ben shu hen youyong . This/thatoneClbookveryuseful b. Zhe/na ben shu hen youyong . This/thatClbookveryuseful c. Zhe/na shu hen youyong . This/that book very useful ‘This/thatbookisveryuseful.’ (74)Zhe/na (ge) Zhang San zhen yonggong . This/that (Cl) Zhang San really diligent ‘This/thatZhangSanisreallydiligent.’ (75) Wo faxian zhe nüren quan si sixinyan .46 Ifindthiswomanallbestubborn ‘Ihavefoundoutthatwomenareallstubborn.’

46 ThisexampleistakenfromLiu(2002),whocitedFang(2002).

170

Thedemonstrativein(72)and(73a)isfollowedbythe[Nume+Cl+N]sequence.Whenthe numeralis yi ‘one’,itcanbedeleted(73b).Whenthenumeral yi ‘one’isomitted,theclassifier canalsobeomitted(73c).Thedemonstrativesin(72)and(73)havetheirusualdeicticmeaning; theproximaldemonstrativereferstobooksclosetothespeaker,whilethedistaldemonstrative referstobooksatadistancefromthespeaker.Thefullnominalsin(72)and(73)canhaveeither ananaphoricreference,meaningthattheymayrefertosomeentityorentitiesmentionedinthe previousdiscourse,oranexophoricreference–theythenrefertosomeentityorentitiesina situationalcontext.

Thedemonstrativesin(74),followedbythe[(Cl)+propername]sequence,donothave much deictic meaning 47 . There is no difference in meaning whether the proximal or distal demonstrativeisusedbeforeapropername.Thedemonstrativesin(74)conveysomeemotional meaning.Lakoff(1974)andLyons(1999)observethesimilaruseofdemonstrativedeterminers in English. Lakoff (1974) calls such use “emotional ”. Lyons (1999: 122) notes that nominalssuchas‘this/myPeter’canpickoutPeterfromothersofthesamenameandthatin suchcasethepropernameisnolongeruniqueanditisrecategorizedasacommonnoun.Lyons alsowritesthattheuseofdeterminersin‘this/myPeter’canbenonrestrictiveandhaveaffective value.ThefollowingEnglishexampleisfromLakoff(1974).

(76) That Henry Kissinger sure knows his way around in Hollywood . 47 Inrarecases,whenseveralpeoplehavethesamename,thenthedemonstrativedeterminerusedbeforetheproper namehasdeicticmeaningtopickoutonepersonamongthoseofthesamename.

171

Diessel(1999:106)callssuchuseof‘that’in(63)‘recognitionaluse’,whichoftenimpliesthat theinterlocutorssharesimilarvieworsympathizewithoneanother.Thespeakerof(76)share someviewwiththeheareraboutHenryKissinger.

Theuseoftheproximaldemonstrative zhe ‘this’in(75)hasrecentlybeenobservedby scholars(Fang,2002;Liu,2002).Thenominalhasagenericreferenceandnoclassifiercanbe insertedaftertheproximaldemonstrative.Accordingto Fang(2002),thedistaldemonstrative doesnothavesuchuse.Fang(2002)andLiu(2002)consider‘ zhe ’asthedefinitearticlethathas developedfromtheproximaldemonstrativewhosedeicticmeaninghasbeenbleachedout.

From the uses of the demonstratives in (72), (73), (74) and (75), it seems natural to concludethatthedemonstrativeinMandarinoccupiesD,astheyoccurattheinitialpositionof thenominals.ThisistheviewthatLi(1998)andTang(1990)assume.Beforeproposingmyown view,Ipresentotherexamplesinwhichthedemonstrativesareprecededbypossessivepronouns orpersonalpronounsanddemonstrativesdonotnecessarilyoccurintheinitialposition.

(77)a. Wo-de zhe liang ben shu hen youyong . IDEthistwoClbookveryuseful ‘Thesetwobooksofmineareveryuseful.’ b. Zhe liang ben wo-de shu hen youyong . ThistwoClIDEbookveryuseful ‘Thesetwobooks,whicharemine,areveryuseful.’ (78)a. Ta / Zhang San zhe/na ge xuesheng hen yonggong . He/ZhangSanthis/thatClstudentvery diligent ‘He/thisstudentZhangSanisverydiligent.’

172

b. Ni-men zhe/*na xie xuesheng hen yonggong . YouCol this/that Quan student very diligent ‘Youstudentsareverydiligent.’ c. Zhang San ta-men zhe/na xie xuesheng hen yonggong . ZhangSan heCol this/that Quan student very diligent ‘ZhangSanandtheotherstudentsareverydiligent.’

In (77a), the possessive precedes the demonstrative in the nominal, while in (77b) it followsthe[Dem+Nume+Cl]combination.Tosomenativespeakers,thetwonominalshave thesamemeaningandthenominalin(77a)soundsmoreidiomatic.ZhonghengJia(p.c.)thinks thatthereisaslightdifferenceinmeaning.Whenthepossessiveprecedesthedemonstrativeasin

(77a),thenominalhasacontrastiveimplication.Sothenominal(77a)means‘thesetwobooksof mine’ in with ‘books of yours/hers/his’ or ‘these/those two books of yours/hers/his’.

Whenthepossessivefollowsthe[Dem+Nume+Cl]combination,thenominaldoesnothave suchacontrastiveimplication.Inaddition,thepossessiveseemstobe nonrestrictive andthe definitenessofthenominalisindicatedbythedemonstrative.Itranslatethepossessiveinthe subjectof(77b)intoanonrestrictiverelativeclause.

In (78a, b, c), a proper name or personal pronoun precedes the demonstrative. The demonstrative in such cases does not have much deictic meaning, or its deictic meaning is weakened.Thedemonstrativeexpressessomeemotional meaning. Zhe xie ‘this+Quan’can be deletedin(78b)withoutchangingthelexicalmeaningofthenominal. Ni-men zhe xie xuesheng

‘youColthisQuanstudents’and ni-men xuesheng ‘youstudents’havethesamereferents.The

173 onlydifferenceisthatthereissomeemotionaleffectwhenthedemonstrativeisused.Whilethe deicticmeaningofthedemonstrativeisweakenedwhenprecededbyapropernameorapersonal pronoun,itseemsthatthedeicticmeaningstillobtains.Onlytheproximaldemonstrativeisused whentheprecedingpronounisthefirstorsecondpersonpronoun,whereasboththeproximal and distal demonstratives can be used if a proper name precedes the demonstrative or the precedingpronounisthethirdpersonpronounasshownin(78a,b,c)andthefollowingexample

(79).

(79) wo zhe/*na ge ren bu hen yonggong . Ithis/*that Clpersonnotvery diligent ‘I’mnotaverydiligentperson.’

Is the subject in (78a, b, c) one nominal with the personal pronoun or the proper name basegeneratedinDoristhesubjectcomposedoftwonominalsinapposition?Iwillassumethat thesubjectisonlyonenominalforthefollowingreasons.First,thereisnocommaintonationin speech.Ifthesubjectin(78)consistsoftwonominalsinapposition,thentheappositionmustbe nonrestrictive,inwhichcasethecommaintonationistypical.Second,thedemonstrativedoes nothavemuchdeicticmeaningandinsomecaseslike(78b),theomissionofthedemonstrative doesnotgiverisetochangeinlexicalmeaning.Thedemonstrativein(78)usuallydoesnotcarry stress and does not have any contrastive implication. Third, to some people, the following example(80)isgrammatical,inwhichthepronounandthefollowingelementdonothavethe samereferent.Twonominalsinappositionshouldrefertothesameentityorentities.

174

(80) Zhang San zhe san ge xuesheng 48 Zhang san this three Cl student ‘Thethreestudents(includingZhangsan)

Tosumup,demonstrativesinMandarincanhavethreedifferentmeanings.

(a) expressingspaciotemporaldeicticmeaning,

(b) expressingemotionaldeicticmeaning,

(c) theproximaldemonstrative zhe ‘this’usedasadefinitearticle.

The following table summarizes the functions of demonstratives and their in

Mandarin.

Position/collocation Examples

(a)Fulldeictic Preceded by the possessive; zhe/na shu ‘this/thatbook’; meaning Followedbyacommonnoun or[(Num)+Cl+(Poss)+N], wo zhe/na (yi) ben shu ‘Ithis/that (one)Clbook’‘thisbookofmine’

(b) weakened deictic Followedby(Cl)+proper zhe/na (ge) Zhang San ‘this/that meaning (emotional name;precededbyproper ZhangSan’, wo/ta/Zhang San zhe ge meaning) namesorpersonalpronouns ren ‘I/he/ZhangSanthisClperson’

(c) zhe ‘this’asa Followedbyacommonnoun zhe nüren/yan ‘women/cigarettes’ definitearticle,no (genericreference) deicticmeaning

Table6.1:MeaningsandFunctionsofDemonstrativesinMandarin

48 Tosomepeople,thethirdpersonpluralpronoun ta-men ‘they’isomittedafterthepropername Zhang San .

175

Following Brugè (2002), who argues that demonstratives are basegenerated in the specifierpositionofFP(functionalprojection)atalowerlevelwithinDPandmoveto[SpecDP] eitherovertlyorcovertlyatLFtocheckitsreferentialfeature,Iassumethatdemonstrativesin

MandarincanalsobebasegeneratedatalowerlevelwithinDPwhentheyhavedeicticmeaning.

However,IassumethatdemonstrativesarebasegeneratedintheheadpositionofDemP,justas

Nguyen(2004)proposesfortheVietnamesedemonstrative,becausepossessivesand adjective phrasescanprecedethedemonstrativeinMandarinasthefollowingexample(81)indicates.

(81) Ta na-zou le wo-de hen hou-de zhe/na liang ben shu . He takewalk Perf IDE very thickDE this/that two Cl book ‘Hetookthese/thosetwoverythickbooksofmine.’

Thedemonstrative zhe ‘this’followeddirectlybyacommonnounwithoutanyotherelementlike classifiersinbetween,whichhasagenericreferenceandiscalled‘definitearticle’inFang(2004) and Liu (2004), is assumed to be basegenerated in D. I repeat here example (75) for convenience.

(75) Wo faxian zhe nüren quan shi sixinyan . Ifindthiswomanallbestubborn ‘Ihavefoundoutthatwomenareallstubborn.’

Thesyntacticstructuresofthedemonstrativecontainingnominals(75)and(81)arerepresented as(82)and(83)respectively.

176

(82)

‘women’(genericreference)

(83)

‘these/thosetwothickbooksofmine’

InSection6.1,IarguedthatthepossessiveinMandarindoesnotnecessarilyimposea definiteinterpretationonthenominal.Asthedemonstrative,notthepossessive,inthenominal

177

(81)giverisetothedefinitenessofthephrase,Iassumethedemonstrativebasegeneratedinthe headpositionofDemPmovestoDtocheckitsreferentialfeatureasBrugè(2002)proposesfor

Spanishdemonstratives.

When the demonstrative follows the proper name or personal pronoun, it mainly expressesemotionalmeaningor‘emotionaldeixis’.Iassumethepropernameandthepersonal pronounareinDandthedemonstrativestaysinitsbaseposition.Ifboththepropernameandthe personalpronounoccurinonenominal,followedbythedemonstrative,asin(78c),theproper name is in [Spec DP] and the pronoun in D. The structure of the nominal in (78c) can be representedas(84).

(84)

‘ZhangSanandtheotherstudents’

178

Asforthedemonstrativeprecedingapropernameorthe[Cl+propername]sequence,it doesnotcontributetothedefinitenessofthenominal.Thedemonstrativedoesnothaveitsusual deicticmeaningandonlyexpressessomeemotionalmeaning.Sothedemonstrativejuststaysin theheadpositionofDemPandthepropernameinthatofNP. Thenominal zhe/na ge Zhang

San ‘this/thatClZhangSan’in(74)hasthefollowingsyntacticstructure(85).

(85)

‘this/thatZhangSan’

Insummary,IarguedinthischapterthatpersonalpronounsaredeterminersinMandarin andarebasegeneratedinD.Propernamescanbeindifferentpositions;theyarein[SpecDP] whenfollowedbyapersonalpronounandthepropernameandthepronounarenotinapposition, inDwhenfollowedbythe[Dem+(Nume/Quan)+Cl+N]orthe[Nume+Cl+N]sequence, and they are in N when preceded by the [Dem + (Cl)].Astothedemonstrative,itcanhave differentfunctionsinMandarin.Ifthedemonstrative zhe ‘this’functionsasadefinitearticleina generic nominal expression, it is basegenerated in D. When the demonstrative has its usual

179 deicticmeaning(i.e.timetemporaldeicticanddiscoursedeicticmeaning),itisbasegeneratedin the head position of DemP and moves to D to check its referential feature. When the demonstrativeexpressestheemotionaldeicticmeaning,itstaysinitsbaseposition.

180

CHAPTER: SEVEN

CONCLUSION

ThisdissertationinvestigatedtheusesofclassifiersinChineseandcomparedspecifically

MandarinwiththeWudialectintermsofclassifierusage.Italsoanalyzedthecomplexnominal inMandarinwithprenominalpossessives,personalpronouns,propernamesanddemonstrative determiners.Thischaptersummarizesthefindingsofthisdissertation.

It was assumed in this dissertation that the classifier in Chinese only encodes the syntactic feature of countability. I argued against Cheng & Sybesma’s (2005) view that the classifierencodesthefeaturesofdefinitenessandnumber.TheuseofclassifiersinChineseis different from that of articles in some Western languages.IarguedthatClPinChineseisnot equivalenttoDPinFrenchorotherWesternlanguagesintermsofdefiniteness,numberandnoun classificationasCheng&Sybesma(1999,2005)implyorargue.

InWesternlanguages,thestructuralpositionofarticlesishighinanominalexpression andinmostcasesarticlesareintheinitialposition,whereasinChinese,classifiersarelowerin positioninthenominalandmanyelementscanprecedethem.InMandarin,thesurfaceform[Cl

+ N] can never have a definite interpretation. I argued in this dissertation that the [Cl + N] expressioninMandarinisanumeralphrasewiththenumeral yi ‘one’deleted.ContrarytoCheng

&Sybesma’s(1999,2005)viewthattheindefinite[Cl+N]expressioninChinesecanonlyhave

181 anonspecificinterpretation,Iprovidedsubstantialevidencetoshowthatthe[ yi ‘one’ +Cl+N] nominal and the indefinite [Cl + N] nominal both can have specific and nonspecific interpretations.Thetwodifferentinterpretationsarerelatedtotheverbtense;thepasttensetends toinduceaspecificinterpretation.Thefactthatthe[yi ‘one’ +Cl+N]expressioncanbeused andthe[Cl+N]nominalcannotincertaincontextswasarguedtobeduetoprosodicreasons.

Thedefinite[Cl+N]expressioninWuwasassumedtobeaDPinthisdissertation,nota

ClPasCheng&Sybesma(1999,2005)argue.Thestructuralpositionofclassifiersistoolowin the nominal for them to turn an NP into an argument. Classifiers are assumed to be basegeneratedinClandtheymovetoDwhenthe[Cl+N]expressionisdefiniteinWu.

In terms of the number feature, I argued against the view that Cl is the locus of grammaticalnumberasCheng&Sybesma(1999,2005)suggest.Although xie isassumedtobea pluralmarkerorapluralclassifierintheliterature,Irefutedthesetwoviewswithlinguisticfacts oftheusageof xie inthisdissertation. First, xie ,like‘some’inEnglish,canbeusedwithboth count nouns and mass nouns in Mandarin. If xie is really a plural marker as some scholars suggest,itcannotbeusedwithmassnouns.So,xie cannotbeapluralmarker.Second,inthe

Mandarinnominal,someindividualclassifierscanbeinsertedafter xie andanominalcanonly have one classifier, so it cannot be a plural classifier. As to why the individual classifier is optional after xie , I suggested that xie does not do exact counting and the individuality of referents is deemphasized when the classifier is not used. If Cl really encodes the number feature,wecannotexplainwhythesocalledpluralclassifier xie cannotbeusedwithnumerals otherthan yi ‘one’andwhythesameindividualclassifierisusednomatterwhetherthereferent(s)

182 ofthenominalis/areoneormorethanone.Iconcludedthat xie isaquantifierwhichcanbeused withbothcountnounsandmassnounsinMandarin.

TodrawparallelsbetweentheclassifierinChineseandthearticleinFrench,Cheng&

Sybesma (2005) suggest that both the classifier and the article function as devices of noun classification.Ijustmentionherebrieflythattheyaredifferent.ClassifiersinChineseclassify nounsintodifferentcategories,basedonthemeaningoftheirreferents.However,theencoding ofthegenderfeatureinarticlesinFrenchhasnothingtodowiththemeaningofthereferents denoted by nouns. The classification function of the classifier is meaningbased while the encodingofthegenderfeatureinarticlesisagrammaticalrequirement.

IprovidedevidencetoshowthatpossessivesinMandarinarelikeadjectives.Unliketheir counterpartsinEnglish,possessivecontainingnominalsinMandarinarenotnecessarilydefinite.

A[Poss+(Nume+Cl)+N]expressioncanhaveeitheradefiniteoranindefiniteinterpretation, justlikebarenounsinMandarin.Thepossessiveisassumedtobein[SpecDP]whenitoccursin theinitialpositionofthedefinitematrixnominal.Whenthematrixnominalisindefinite,the possessiveisassumedtobein[SpecNumeP].

IalsoarguedinthisdissertationthatpersonalpronounsinMandarinaredeterminersand basegeneratedinD.The[Pron+Nume+Cl+N]sequencecanbeonenominal,inwhichcase thepronounisstressedandthereisnobreakinspeechbetweenthepronounandthefollowing element,ortwonominals,inwhichcasethepronounfunctionsindependentlyasanominalandit isinappositionwiththefollowing[Nume+CL+N]nominal. Therecanalwayshavethe commaintonationinspeechwhentheyareinapposition.

183

Propernamesarearguedtobeindifferentsyntacticpositions.Theycanoccurindifferent positionsinanominal.Theyareinthespecifierposition of DP when followed by pronouns.

Proper names are in the position of N when they are preceded by the demonstrative with affectivevalue.

DemonstrativeshavethreefunctionsinChinese:(1) expressing spaciotemporal deictic meaning; (2) expressing emotional deictic meaning; (3) the proximal demonstrative zhe ‘this’ usedasadefinitearticle.ThedemonstrativeisarguedtoprojectitsownphrasalcategoryDemP whenithasdeicticmeaning.Whenthedemonstrativehasitsusualspaciotemporalmeaningina nominal, then it movesto D to check its referential meaning. If the demonstrative expresses emotionaldeicticmeaning,itstaysinitsbasepositionDem.Whentheproximaldemonstrative zhe ‘this’expressesgenericmeaning,itisassumedtobebasegeneratedinD.

184

BIBLIOGRAPHY

Abney,Steven(1987) The English in Its Sentential Aspect .Ph.D.dissertation,MIT.

Aikhenvald,AlexandraY.(2000) Classifiers: A Typology of Noun Categorization Devices .New

York:OxfordUniversityPress.

Alexiadou,Artemis(2004)“OntheDevelopmentofPossessiveDeterminers:Consequencesfor

DP Structure.” In Fuss, Eric & Carola Trips (eds.) Diachronic Clues to Synchronic

Grammar ,3158.Amsterdam:JohnBenjaminsPublishingCompany.

Alexiadou,Artemis(2005)“Possessorsand(In)definiteness.” Lingua 115:787819.

Allan,K.(1977)“Classifiers.” Language 53:284310.

Baptista,Marlyse(toappear)“OntheSyntaxandSemanticsofDPinCapeVerdeanCreole.”In

Baptista, M & Gueron, J. (eds.) Noun Phrases in Creole Languages: A Multi-Faceted

Approach .Amsterdam:JohnBejaminsPublishingCompany.

Bernstein,JudyB.(2001)“TheDPHypothesis:IdentifyingClausalPropertiesintheNominal

Domain.”InBaltin,Mark&ChrisCollins(eds.) The Handbook of Contemporary Syntactic

Theory ,536561.Malden,MA:Blackwell.

Bhattacharya, Tanmoy (2001) “Numeral/QuantifierClassifier as a Complex Head.” In Corver,

Norbert&HenkvanRiemsdijk(eds.) Semi-lexical Categories: The Function of Content

Words and the Content of Function Words ,191221.Berlin:MoutondeGruyter.

185

Bisang,Walter(1999)“ClassifiersinEastandSoutheastAsianLanguages:Countingand

Beyond.”InGvozdanovic,Jadranka(ed.) Numeral Types and Changes Worldwide,113185.

Berlin:MoutondeGruyter.

Brugè,Laura(2002)“ThePositionsofDemonstrativesintheExtendedNominalProjection.”In

Cinque,Guglielmo(ed.)Functional Structure in DP and IP: The Cartography of Syntactic

Structures, Vol 1 ,1553.Oxford:OxfordUniversityPress.

Burling,Robbins(1965)“HowtoChooseaBurmeseNumeralClassifier.”InSpiro,Melford(ed.)

Context and Meaning in Cultural Anthropology ,243264.NewYork:TheFreePress.

Chao,YuenRen(1968) A Grammar of Spoken Chinese .Berkeley:UniversityofCaliforniaPress.

Chen,Guanglei(1987)“关于们 与 -s.”[On -men inChineseandsinEnglish]. Fudan Journal

(SocialSciences)No.5:3034.

Chen,Ping(2003)“IndefiniteDeterminer Introducing Definite Referent: A Special Use of ‘ yi

‘one’+Classifier’inChinese.” Lingua 113:11691184.

Chen,Ping(2004)“IdentifiabilityandDefinitenessinChinese.” Linguistics 42(6):11291184.

Cheng,Lisa,L.S.&RintSybesma(1998)“YiwanTang,YigeTang:Classifiersand

Massifiers.” The Tsing Hua Journal of Chinese Studies . NewSeries28(3):385412.

Cheng,LisaL.S.&RintSybesma(1999)“BareandNotsobareNounsandtheStructureof

NP.” Linguistic Inquiry 30:509542.

Cheng,Lisa,L.S.&RintSybesma(2005)“ClassifiersinFourVarietiesofChinese.”InCinque,

Guglielmo & Richard S. Kayne (eds.) The Oxford Handbook of Comparative Syntax ,

806838.Oxford:OxfordUniversityPress.

186

Chierchia,Gennaro(1998a)“PluralityofMassNounsandtheNotionofSemanticParameter”.In

Rothstein,Susan(ed.) Events and Grammar ,53104.Dordrecht,theNetherlands:Kluwer

AcademicPublishers.

Chierchia, Gennaro (1998b) “Reference to Kinds across Languages”, Natural Language

Semantics 6:339405.

Chomsky,Noam(1986) Barriers .Cambridge,MA:TheMITPress.

Cinque,Guglielmo(1980)“OnExtractionfromNPinItalian.” Journal of Italian Linguistics 5:

4799.

Cinque,Guglielmo(1993)“ANullTheoryofPhraseandCompoundStress”, Linguistic Inquiry

24(2):239297.

Cinque,Guglielmo(1994)“OntheEvidenceforPartialNMovementintheRomanceDP.”In

Cinque, Guglielmo, Jan Koster, JeanYves Pollock & Luigi Rizzi (eds.) Paths Towards

Universal Grammar: Studies in Honor of Richard S. Kayne , 85110. Washington:

GeorgetownUniversityPress.

Croft,William(1994)“SemanticUniversalsinClassifierSystems.” Word 45(2):145171.

Crystal, David (2003) A Dictionary of Linguistics and Phonetics . Malden, MA: Blackwell

PublishingLtd.

Denny,J.Peter(1976)“WhatAreNounClassifiersGoodFor?” Chicago Linguistic Society 12:

122132.

Diesing, Molly (1988) “Bare Plural Subjects and the Stage/Individual Contrast.” In Krifka,

Manfred(ed.) Genericity in Natural Language ,107154.UniversityofTübingen.

187

Diessel,Holger(1999) Demonstratives: Form, Function, and .Amsterdam:

JohnBenjaminPublishingCompany.

DimitrovaVulchanova, Mila & Giuliana Giusti (1999) “Possessors in the Bulgarian DP.” In

DimitrovaVulchanova, Mila. & Lars Hellan (eds.) Topics in South Slavic Syntax and

Semantics ,163192.Amsterdam:JohnBenjaminPublishingCompany.

Dobson,WilliamA.C.H.(1962) Earlier Archaic Chinese: A Descriptive Grammar .Toronto:

UniversityofTorontoPress.

Doetjes,Jenny(1997) Quantifiers and Selection: On the Distribution of Quantifying Expressions

in French, Dutch and English .PhD.Dissertation,LeidenUniversity.

Duanmu,San(2000) The Phonology of .Oxford:OxfordUniversityPress.

Duanmu,San(2001)“StressinChinese.”InXu,DeBao(ed.) Chinese Phonology in Generative

Grammar ,117138.SanDiego:AcademicPress.

Erbaugh,MaryS.(1986)“TakingStock:TheDevelopmentofChineseNounClassifiers

HistoricallyandinYoungChildren.”InCraig,Colette.G.(ed.) Noun Classes and

Categorization ,399436.Amsterdam:JohnBenjaminsPublishingCompany.

Fang,Mei(2002) “指示词“这”和“那“在北京话中的语法化。” [TheGrammaticalization

ofDemonstratives‘this’and‘that’inBeijingDialect]. 中国语文 [ The ]

No.4:275276.

Frawley,William(1992) Linguistic Semantics .Hilldale,NJ:LawrenceErlbaumAssociates,Inc.

Ferguson,CharlesA.(1964)“TheBasicGrammaticalCategoriesofBengali.”InLunt,HoraceG.

(ed.) Proceedings of the Ninth International Congress of Linguistics ,88190.TheHague:

188

Mouton.

Giorgi,Alessandra&GiuseppeLongobardi(1991) The Syntax of Noun Phrase .Cambridge:

CambridgeUniversityPress.

Gil,David(1994)“Summary:NumeralClassifiers.” Linguist List 5:466.

Goddard, Cliff (2005) The Languages of East and Southeast Asia: An Introduction . Oxford:

OxfordUniversityPress.

Greenberg,Joseph.H.(1972)“NumeralClassifiersandSubstantivalNumber:Problemsinthe

GenesisofaLinguisticType.” Working Papers on Language Universals 9:139.Stanford

University.

Greenberg, Joseph. H. (1973) “Numeral Classifiers and Plural Marking: An Implication

Universal.” Working Papers on Language Universals 11:122.StanfordUniversity.

Greenberg,Joseph.H.(1975)“DynamicAspectsofWordOrderintheNumeralClassifier.”InLi,

Charles (ed.) Word Order and Word Order Change , 2746. Austin: University of Texas

Press.

He,Jie(2001) 现代汉语量词研究 [Research on Measure Words in Modern Chinese ].Beijing:

NationalitiesPublishingHouse.

Hellan,Lars(1986)“TheHeadednessofNPsinNorwegian.”InMuysken,Pieter&Henkvan

Riemsdijk(eds.) Features and Projections ,89123.Dorchecht:Foris.

Himmelmann,NikolausP.(1996)“DemonstrativesinNarrativeDiscourse:ATaxonomyof

UniversalUses.”InFox,Barbara(ed.) Studies in ,205254.Amsterdam:John

BenjaminPublishingCompany.

189

Hopper,Paul(1986)“SomeDiscourseFunctionsofClassifiersinMalay.”InCraig,Colette.G.

(ed.) Noun Classes and Categorization ,309325.Amsterdam:JohnBenjaminsPublishing

Company.

Huang,ChuRenandKathleenAhrens(2003)“Individuals,KindsandEvents:Classifier

CoercionofNouns”, Language Sciences 25 :353373.

Huang, C.T James (1982) Logical Relations in Chinese and the Theory of Grammar . Ph.D.

Dissertation,MIT.

Huang,C.TJames(1987)“ExistentialSentencesinChineseand(In)definiteness.”InReuland,

Eric.&AliceB.Meulen(eds.)The Representation of (In)definiteness ,226253.Cambridge,

MA:MITPress.

Huang,Shuanfan(1999)“TheEmergenceofaGrammaticalCategory Definite Article inSpoken

Chinese.” Journal of 31:7794.

Iljic,Robert(1994)“QuantificationinMandarinChinese:TwoMarkersofPlurality.” Linguistics

32:91116.

Iljic,Robert(2001a)“TheProblemoftheSuffixmen inChineseGrammar.” Journal of Chinese

Linguistics 29(1):1168.

Iljic,Robert(2001b)“TheOriginoftheSuffixmen inChinese”, Bulletin of SOAS 64(1):7497.

Iljic,Robert(2005)“PersonalCollectiveinChinese”, Bulletin of SOAS 68(1):77104.

Jackendoff,Ray(1977) X' Syntax: A Study of Phrase Structure .Cambridge,MA:MITPress.

Jaisser,Annie(1987)“HmongClassifiers.” Linguistics of the Tibeto-Burman Area 10:169176.

Kayne,Richard(1994) The Antisymmetry of Syntax .Cambridge,MA:MITPress.

190

Koopman, Hilda & Dominique Sportiche (1991) “The Position of Subjects.” Lingua , 85:

211258.

Kratzer,Angelika(1988)“StageLevelandIndividualLevelPredicates.”InKrifka,Manfred(ed.)

Genericity in Natural Language ,247284.UniversityofTübingen.

Lakoff,Robin(1974)“Remarkson This and That .”InMichaelW.LaGaly,RobertA.Fox&

Anthony Bruck (eds.) Papers from the 10 th Regional Meeting of the Chicago Linguistic

Society ,345356.Chicago:ChicagoLinguisticSociety.

LaPolla, Randy J. (1995) “Pragmatic Relations and Word Order in Chinese.” In Downing,

Pamela & Michael Noonan (eds.) Word Order in Discourse , 297329. Amsterdam: John

BenjaminsPublishingCompany.

Leko, Nedžad (1999) “Functional Categories and the Structure of the DP in Bosnian.” In

DimitrovaVulchanova, Mila. & Lars Hellan (eds.) Topics in South Slavic Syntax and

Semantics ,229252.Amsterdam:JohnBenjaminPublishingCompany.

Li, Charles N. & Sandra A. Thompson (1981) : A Functional Reference

Grammar .Berkeley:UniversityofCaliforniaPress.

Li,CharlesN.&SandraA.Thompson(1990)“Chinese.”InComrie,Bernard(ed.) The World’s

Major Language , 811833.Oxford:OxfordUniversityPress.

Li,YH.Audrey(1998)“ArgumentDeterminerPhrasesandNumberPhrases.” Linguistic Inquiry

29(4):693702.

Li,YH.Audrey(1999)“PluralityinaClassifierLanguage.” Journal of East Asian Linguistics 8:

7599.

191

Liu,Danqing(2002)“ 汉语类指成分的语义属性和句法属性”[TheSemanticandSyntactic

Features of Elements of Generic Reference in Chinese], 中 国 语 文 [ The Chinese

Language ] No.5:411422.

Liu, Yuehua, Wenyu Pan and Wei Gu (2001) 实用现代汉语语法 [Practical Modern Chinese

Grammar].Beijing:CommercialPress.

Löbel,Elisabeth(2001)“ClassifiersandSemilexicality:FunctionalandSemanticSelection.”In

Corver,Norbert&vanRiemsdijik,Henk(eds.) Semi-Lexical Categories: The Function of

Content Words and the Content of Function Words ,223271.Berlin:MoutondeGruyter.

Longobardi, Giuseppe (1994) “Reference and Proper Names: A Theory of Nmovement in

SyntaxandLogicalForm.” Linguistic Inquiry 25(4):609665.

Longobardi, Giuseppe (2001) “The Structure of DPs: Some Principles, Parameters, and

Problems.” In Baltin, Mark. & Chris Collins (eds.) The Handbook of Contemporary

Syntactic Theory ,562603.Malden,MA:Blackwell.

Lu,Bingfu&SanDuanmu(2002)“RhythmandSyntaxinChinese:ACaseStudy.” Journal of

the Chinese Language Teachers Association 37(2):123136.

Lü,Shuxiang(1944) 中国文法要略 [AnOutlineofChineseGrammar].Shanghai:Commercial

Press.

Lü, Shuxiang (ed.) (1980) 现代汉语八百词 [Eight Hundred Words in Modern Chinese].

Beijing:CommercialPress.

Lyons,Christopher(1985)“APossessiveParameter.” Sheffield Working Papers in Language and

Linguistics 2:98104.

192

Lyons,Christopher(1999) Definiteness .Cambridge:CambridgeUniversityPress.

Lyons,John(1977) Semantics Vols.I&II.Cambridge:CambridgeUniversityPress.

Matthews ,Stephen & Virginia Yip (1994) Cantonese: A Comprehensive Grammar . London:

Routledge.

Matthew, Stephen & Patrizia Pacioni (1997) “Specificity and Genericity in Cantonese and

Mandarin.” In Xu, Liejiong (ed.) The Referential Properties of Chinese Noun Phrases ,

4559.Paris:CentredeRecherchesLinguistiquessurl’AsieOrientale.

Nakanish, Kimiko & Satoshi Tomioka (2004) “Japanese Plurals are Exceptional.” Journal of

East Asian Linguistics 13(2):113140.

Nettle, Daniel & Suzanne Romaine (2000) Vanishing Voices: The Extinction of the World’s

Languages .Oxford:OxfordUniversityPress.

Nguyen, Tuong Hung (2004) The Structure of the Vietnamese Noun Phrase .PhDdissertation,

BostonUniversity,BostonMA.

Norman,Jerry(1988) Chinese .Cambridge:CambridgeUniversityPress.

Pan, Wuyun (1991) “An Introduction to the Wu Dialects.” In Wang, William S.Y. (ed.)

Language and Dialects of China .JournalofChineseLinguisticsMonographSeriesNo.3:

237293.

Pelletier,FrancisJ.(ed.)(1979) Mass Terms: Some Philosophical Problems .Dordrecht,Holland:

D.ReidelPublishingCompany.

Peyraube,Alain(1991)“SomeRemarksontheHistoryofChineseClassifiers.” Santa Barbara

Papers in Linguistics 3:106126.

193

Picallo,M.Carme(1991)“NominalsandNominalizationinCatalan.” Probus 3:279316.

Pollock, JeanYves (1989) “Verb Movement, Universal Grammar, and the Structure of IP.”

Linguistic Inquiry 20(3):365424.

Postal,PaulM.(1970)“OnSocalledPronounsinEnglish.”InJacobs,RoderickA.&PeterS.

Rosenbaum(eds.) Reading in English ,5682.Waltham,Mass:

GinnPress.(OriginallyappearedinDinneen,FrancisP.(ed.)(1969) Monograph Series on

Language and Linguistics 19,177206.Washington:GeorgetownUniversityPress.)

Progovac,Ljiljana(1998)“DeterminerPhraseinaLanguageWithoutDeterminers.” Journal of

Linguistics 34:165179.

Pulleyblank,Edwin(1995) Outline of Classical .Vancouver:Universityof

BritishColumbiaPress.

Qian,Nairong(1997) 上海话语法 [AGrammaroftheShanghaiDialect].Shanghai:Shanghai

People’sPublishingHouse.

Reed,Irene,OsahitoMiyaoka,StevenJacobson,PaschalAfcan&MichaelKrauss(1977) Yup’ik

Eskimo Grammar .UniversityofAlaska.

Ritter, Elisabeth (1988) “A HeadMovement Approach to ConstructState Noun Phrases.”

Linguistics 26:909929.

Ritter,Elisabeth(1991)“TwoFunctionalCategoriesinNounPhrases:EvidencefromModern

Hebrew.”InRothstein,Susan.D(ed.) Syntax and Semantics, 25, Perspectives on Phrase

Structure: Heads and Licensing ,3762.NewYork:AcademicPress.

Schoorlemmer,Maaike(1998)“Possessors,ArticlesandDefiniteness.”InAlexiadou,Artemis&

194

ChrisWilder(eds.) Possessors, Predicates and Movement in the DP ,5586.Amsterdam:

JohnBenjaminsPublishingCompany.

Scott, GaryJohn (2002) “Stacked Adjectival Modification and the Structure of Nominal

Phrases.”InCinque,Guglielmo(ed.)Functional Structure in DP and IP: The Cartography

of Syntactic Structures, Vol.1,91120.Oxford:OxfordUniversityPress.

Shi, Yuzhi (2002) The Establishment of Modern Chinese Grammar: The Formation of the

Resultative Construction and Its Effects . Amsterdam: John Benjamins Publishing

Company.

Shi,Yuzhi(2003)“ 汉语的‘数’范畴与‘有定’范畴之关系”[“TheRelationshipbetweenthe

TwoGrammaticalCategoriesinChinese:NumberandDefiniteness”]. 语言研究 [Studies

in Language and Linguistics ]23(2):4050.

Shi,Yuzhi&CharlesN.Li(2002)“TheEstablishmentoftheClassifierSystemandthe

GrammaticalizationoftheMorphosyntacticParticlede inChinese.” Language Sciences

24(1):115.

Simpson,Andrew(2005)“ClassifiersandDPStructureinSoutheastAsia.”InCinque,Guglielmo

&RichardS.Kayne(eds.)The Oxford Handbook of Comparative Syntax ,806838.Oxford:

OxfordUniversityPress.

Skwire,David&HarveyS.Wiener(2005) Student’s Book of College English (tenthed.).New

York:PearsonEducation,Inc.

Sommerstein,AlanH.(1972)“OntheSoCalledDefiniteArticleinEnglish.” Linguistic Inquiry

3:197209.

195

Stowell,Tim(1989)“Subjects,Specifiers,andXbarTheory.”InBaltin,MarkR.&AnthonyS.

Kroch(eds.) Alternative Conceptions of Phrase Structure ,232262.Chicago:Universityof

ChicagoPress.

Szabolcsi,Anna.(1983)“ThePossessorthatRanawayfromHome.” The Linguistic Review 3:

89102.

Tai,JamesH.YandLianqingWang(1990)“ASemanticStudyoftheClassifier tiao .” Journal of

the Chinese Language Teachers Association 25(1):3556.

Tang,CCJane(1990) Chinese Phrase Structure and the Extended X’-theory.Ph.D.dissertation,

CornellUniversity.

Tong, Shengqiang (2002) “‘ 们’的定指意义.” [The Definite Meaning of “men ”] 中国语文

[The Chinese Language ] No.3 :275276.

Trenkic,Danijela(2004)“DefinitenessinSerbian/Croatian/BosnianandSomeImplicationsfor

theGeneralStructureoftheNominalPhrase.” Lingua 114:14011427.

T’sou,Benjamin.K.(1976)‘TheStructureofNominalClassifierSystems.’ InJenner,PhilipN.,

Lawrence C. Thompson & Stanley Starosta (eds.) Austroasiatic Studies , 2, 1215 1247.

Honolulu:UniversityofHawaiiPress.

Vuori, vesaJussi (2000) Repetitive Structures in the Languages of East and South-east Asia .

Helsinki:theFinnishOrientalSociety.

Wales, Katie (1996) Personal Pronouns in Present-Day English . Cambridge: Cambridge

UniversityPress.

Wang,Li(1958) 汉语史稿 [OutlineoftheHistoryofChinese].Beijing:SciencePress.

196

Wang,Li(1989) 汉语语法史 [HistoryofChineseGrammar].Beijing:CommercialPress.

Wang, Lianqing (1994) Origin and Development of Classifiers in Chinese . Ph.D dissertation:

OhioStateUniversity.

Xu, Ding (1997) Functional Categories in Mandarin Chinese .The Hague: HollandAcademic

Graphics.

Zhang,Yisheng(2001)“’N’+‘ 们’的选择限制与 ‘N 们‘的表意功用.”[TheSelective

Restrictions of N + men and the Expressive Functions of Nmen ]. 中国语文 [The

Chinese Language ]No.3:201211.