1 Contrastive Function of the Japanese Particle wa

ContrastiveFunctionoftheJapaneseParticlewa

2

Abstract.TheJapaneseparticlewaisofagreatinterestduetoitscomplexityinvolvingboththetopicandcontrastivefunctions.Althoughtheparticlewawiththecontrastivefunctiondistributescross-categorially,manypreviousanalysesfocusonsimpleNPsatthesubjectposition,facingdifficultydealingwithmorecomplexcases.Forexample,fewpreviousproposalsaccountforthecorrectinterpretationofacomplexwa-markedphrase.Inaddition,theinteractionbetweencontrastivewaandtheuniversalquantifierposesaninterestingchallenge.Generallyspeaking,previousanalysesarenotsufficientlyprecisenoraccuratetodealwiththeseproblems.Thispaperpresentsananalysisofcontrastivewathatcanaddresstheseproblemsinasystematicway.Wepursuethisgoalbyobservingthefollowingtwopoints.First,phonologicalprominenceisexplicitlyassociatedwithcontrastivesemantics.Second,thecontrastivesemanticsofwaisanalyzedasaformofcontraststrongerthanthecontrastwithoutwa.TheanalysisisformulatedwithintheframeworkofalternativesemanticsofRooth(1985).

Areaofstudy:Semantics,Pragmatics

3

1.Introduction.TheJapaneseADVERBIALPARTICLEwaisarguedtohavetopic(orthematic)andcontrastivefunctions(e.g.Kuno,1972),asintheexamplesshownbelow.Intheseandlaterexamples,SMALLCAPITALSareusedtoindicatePHONOLOGICALPROMINENCEorgrammaticallabels:TOP(topic),CONT(contrastive),NOM(nominative),ACC(accusative),DAT(dative),andGEN(genitive).

(1)Topicwa(followingtheutterance‘ThepersonwhocameherewasKen.’)

Ken-waKen-TOP

BANANA-o

TABETA.

banana-ACCate

‘Kenatea/thebanana.’

(2)Contrastivewa(inresponsetothequestion‘Amongthosepeople,whoatethebanana?’)

KEN-waKen-CONT

banana-obanana-ACC

tabeta.ate

‘Kenatethebanana(butsomeoneelsedidn’teatthebanana).’

In1,theutterancecontinuestotalkaboutKenandprovidesnewinformationabouthim.In2,theresponsenotonlyanswersthequestionbutalsocarriesthepresuppositionshownintheparentheses.NotethatalthoughKuno(1972)suggestedthatthethematicandthecontrastivefunctionswereexclusive,amorerecentviewisthatthereisanoverlap(e.g.Noda,1996).

Inthispaper,wemainlydiscusstwophenomenainvolvingcontrastivewa.Thefirstisthecasewherecontrastivewaisattachedtoacomplexphraseasshownbelow.(3)a.Ken-wa

Ken-TOP

[

NAOMI-noNaomi-GEN

bananabanana

]

-wa-CONT

tabeta.ate

‘KenateNaomi’sbanana(butdidn’teatsomeoneelse’banana).’b.Ken-waKen-TOP

[

Naomi-noNaomi-GEN

BANANA

]-wa-CONT

tabeta.ate

banana

‘KenateNaomi’sbanana(butdidn’teatsomethingelseofNaomi).’

Dependingonthepositionofphonologicalprominence,thereisacleardistinctionbetweenthepresuppositionsin3a

4

and3b.Theproblemwithpreviousanalysesisthattheydidnotpaymuchattentiontothecaseofwa-markingoncomplexphrasesandoverlookedtheroleofphonologicalprominence.

Thesecondphenomenonisaboutthecaseofasymmetryinvolvingtheuniversalquantifierinrelationtocontrastivewaasshownbelow.

(4)Inresponsetothequestion‘DidKenpraiseNaomi?’

a.Ken-waKen-TOP

MINNA-o

hometa.praised

everyone-ACC

‘Kenpraisedeveryone.’b.*Ken-wa

Ken-TOP

MINNA-wa

hometa.praised

everyone-CONT

‘Kenpraisedeveryone.’

Inthisexample,whiletheaccusativecasemarkero(attheobjectposition)ispossible,thecontrastivewaisnot.Aswewillseelater,theasymmetryisindependentofthegrammaticalrelations,underlyingcasemarking,andwordorder.Previousanalysesfailtoaccountforthisasymmetry.

Thispaperbeginswiththeideathataphonologicalprominenceisassociatedwiththebasiccontrastivemeaning,andanalyzesthatwainteractswithsuchacontrast,resultinginastrongercontrast.TheideacanbeformulatedwithintheframeworkofALTERNATIVESEMANTICS(Rooth,1985).Thisdevelopsintoapreciseandaccurateanalysisofcontrastivewa,leadingtoasolutiontothetwoproblemsintroducedabove.SinceJapanesewaiscloselyrelatedtotheKoreancounterpartn(un),wewillalsociteanalysesforKorean.

Thispaperisorganizedasfollows:Thenextsectionre-examinestheproblemsmoreindetail.Insection3,wereviewpreviousanalyses.Then,Section4introducesanewanalysisofcontrastivewabasedonalternativesemantics,leadingtooursolutiontotheproblems.

2.TheProblemsRevisited.Wenowreturntotheproblemsandalsoreviewthebackgroundinformationneededfor

5

furtherdiscussion.Thefirstproblemisaboutwa-markingoncomplexphrasesasshownin3,repeatedbelow.(5)a.Ken-wa

Ken-TOP

NAOMI-noNaomi-GEN

banana-wabanana-CONT

tabeta.ate

‘KenateNaomi’sbanana.’

Presupposition:‘Kendidn’teatsomeoneelse’banana.’b.Ken-waKen-TOP

Naomi-noNaomi-GEN

BANANA-wa

tabeta.ate

banana-CONT

‘KenateNaomi’sbanana.’

Presupposition1:‘Kendidn’teatsomethingelseofNaomi.’

Presupposition2:‘Kendidn’teatsomethingotherthanNaomi’sbanana.’

LetusfirstclarifythetermPRESUPPOSITIONasusedinthispaper(Beaver,1997,formoredetails);the

presuppositionambiguityin5bwillbediscussedshortly.Here,presuppositionbroadlyreferstothemeaningofanutterance(asspokeninacertaincontext)thatisnotTRUTH-CONDITIONAL(theglossesintheaboveexamplescorrespondtothetruth-conditionalmeaningsofthesentences).Inthispaper,weareparticularlyinterestedinthepresuppositionstriggeredbyphonologicalprominenceandtheparticlewa.

Todistinguishthepresuppositionfromthetruth-conditionalmeaning,wecanusethePROJECTIONTESTacrossnegation.Thetestisbasedontheideathatnegationonlyaffectsthetruth-conditionalmeaning.Inotherwords,presuppositionssurvivenegation.Letusnowobservethispointwithrespecttocontrastivewainthefollowingexample(aratherunusualformofnegationisusedheretoavoidtheinteractionofwawiththematrix-levelnegativeoperator).

(6)Followingtheutterances‘Therewereabananaandamangoonthetable.Kendidn’teatthemango.’

Sosite,and

[

Ken-gaKen-NOM

BANANA-wa

tabetaate

]-to-iuthat

koto-gafact-NOM

hiteisareta.wasdenied

banana-CONT

‘Thefactthat[Kenatethebanana]wasdenied.’

Here,theembeddedclauseisnegatedandthus‘Kenatethebanana’isfalse.Butthepresuppositionassociatedwith

6

contrastivewa(‘Kendidn’teatsomethingelse’)survivesthenegationandisavailableasapartofthe

non-truth-conditionalmeaningoftheutterance.Thispresuppositionhappenstobeconsistentwiththecontextgivenintheaboveexample,resultinginafelicitousutterance.Onthecontrary,thesameutterancecanbeinfelicitousinthefollowingexamplewithaslightlydifferentcontext.

(7)Followingtheutterances‘Therewereabananaandamangoonthetable.Kenatethemango.’

#Sosite,and

[

Ken-gaKen-NOM

BANANA-wa

tabetaate

]-to-iuthat

koto-gafact-NOM

hiteisareta.wasdenied

banana-CONT

‘Thefactthat[Kenatethebanana]wasdenied.’

Theaboveutteranceisidenticalto6withrespecttoboththetruth-conditionalmeaningandthepresupposition.Butinthiscase,thepresuppositionisinconsistentwiththegivencontext.Weuse‘#’toindicatethattheutteranceisinfelicitousinthegivencontext.Incontrast,weuse‘*’toindicatethatthesentenceisunacceptableinanycontextforgrammatical,semantic,and/orpragmaticreasons.

Inlatersections,wewillneedtodistinguishtwotypesofpresupposition:(i)CONVENTIONALIMPLICATURE,whichisagrammaticalizedformofmeaningandthusnotcancellable,and(ii)CONVERSATIONALIMPLICATURE,whichisavailableasaresultofinferenceandcancellableifexplicitlydoneso(Grice,1975;KarttunenandPeters,1979).Thecontrastiveforceassociatedwithwaisconventionalbecauseitisnotcancellableasshowninthefollowingexample:(8)Followingtheutterance‘Kenateallthefruits.’

#BANANA-wabanana-CONT‘Heatebanana.’

ThisutteranceisinfelicitousinthegivencontextbecausethereisnocontrastiveelementsuchthatKendidn’teat.Althoughnotcentraltothispaper,letusbrieflyreturntothepresuppositionambiguityin5basweneedsomeclarification.SuchambiguityisbasicallythesameasthefollowingphenomenonobservedforEnglish(adaptedfromSelkirk,1984).

tabeta.ate

7

(9)a.Sheonlywatched[KOJAK].

b.Sheonly[watchedKOJAK].

Theutterance9acanbearesponseto‘DidMarywatchbothM*A*S*HandKojak?’.Theutterance9bcanbearesponseto‘DidMarydothedishesandwatchKojak?’.But9aand9barepronouncedidentically.Underthenormalcondition,thepresuppositionambiguityin5bisresolvedbythecontext.Intherestofthispaper,wefocusonthe

NARROW

presupposition1in5b.Sinceweadoptthealternativesemanticsapproach(Rooth,1985),whichisusedto

accountforthistypeofdistinctcontrastdomains,ouranalysiscanbeextendedtodealwiththistypeofambiguity.Wenowturntothesecondproblemexemplifiedin4,repeatedbelow.(10)Inresponsetothequestion‘DidKenpraiseNaomi?’

Ken-waKen-TOP

MINNA-o/*wa

hometa.praised

everyone-ACC/CONT

‘Kenpraisedeveryone(incontrasttojustNaomi).’

Notethattheaboveuseofwais‘*’-edbecauseitisunacceptableinanycontext,notonlythecontextgivenabove.ThiscontrastfortheobjectpositionhasbeenobservedbyHan(1998)fortheKoreancounterpartofcontrastivewa.Butthisphenomenonisnotlimitedtotheobjectposition.Universalquantifiercannotcoexistwithcontrastivewaindependentofthegrammaticalrelation,theunderlyingcaseofthewa-markedphrase,andthepresenceofscramblingascanbeseenbelow:(11)a.MINNA-ga/*wa

everyone-NOM/CONT‘EveryonepraisedKen.’b.MINNA-o/*waeveryone-ACC/CONT

Ken-waKen-TOP

hometa.praised

Ken-oKen-ACC

hometa.praised

‘Kenpraisedeveryone.’

Sincetheincompatibilitybetweencontrastivewaandtheuniversalquantifierisindependentofsyntaxasshownaboveandalsoindependentofthecontext,theeffectmustcomefromthesemanticsofcontrastivewa.Notethatthe

8

followingispossible(withoutprominenceonthewa-markedphrase),butwhatisinvolvedhereisonlythetopicfunction,notthecontrastivefunctionofwa.(12)Minna-wa

everyone-TOP

KEN-oKen-ACC

hometa.praised

‘EveryonepraisedKen.’

Aswehaveseenabove,contrastivewamayinteractwithaquantifieraswellasnouns.Moregenerally,thedistributionofcontrastivewaisCROSSCATEGORIAL,includingpositionsafternouns,quantifiers,otherparticles,verbs,adverbs,andcomplementizers(Aoki,1992;Tateishi,1994;Noda,1996).Thus,thesemanticsofcontrastivewamustbeabletointeractwithawiderangeofsemantictypes.Thefollowingexamplesshowafewsuchpossibilities.(13)a.Ken-wa

Ken-TOP

PERU-ni-waPeru-DAT-CONT

ikanakatta.didn’tgo

‘Kendidn’tgotoPeru(butwentsomewhereelse).’b.Ken-waKen-TOP

KINOU-wayesterday-CONT

nenakatta.didn’tsleep

‘Kendidn’tsleepyesterday(butsleptsomeothertime).’c.Ken-waKen-TOP

TABE-wa

sita-ga,did-but

NOMI-wasinakatta.didn’t

eat-CONTdrink-CONT

‘Kenate(butdidn’tdosomethingelse),butdidn’tdrink(didsomethingelse).’

Withrespecttothedistribution,contrastivewabehavesverymuchlikeotheradverbialparticlessuchasdake‘only’,mo‘too’,sae‘even’,andkoso‘inparticular’.Inaddition,itiscomparabletoonlyinEnglish.Aswewillseeinlatersections,thispaperexploresananalysisapplicabletothecross-categorialdistributionofcontrastivewa.

3.PreviousAnalyses.Contrastivewahasbeenthefocusofseveralworks.ButaswehavementionedinSection1,noneoftheanalysesreviewedhereaccountforthetwoproblemsunderdiscussion.Themainshortcomingsarethe

9

following:(i)thatphonologicalprominenceisignoredand(ii)thattheanalysisofpresuppositionassociatedwithcontrastivewaisinaccurate.Letusbrieflydiscussthefirstpointandthenreviewpreviousanalysesmoreinconjunctionwiththesecondone.

Withoutanalyzingthecontrastassociatedwithaphonologicalprominence,itisimpossibletoderivethecorrectpresuppositionforacomplexphrasesuchas3.Forexample,whileHuruta(1982)analyzesthefunctionofcontrastivewaaffixedtoacomplexNP,thedistinctpresuppositionscorrespondingtodifferentphonologicalprominencearenotdiscussed,andleftasambiguous.

Asforthesecondpoint,noneofthepreviousanalysesexplicitlydiscussthedistinctionbetweenthecontrastivepresuppositionswithandwithoutwa.Naturally,theasymmetryin4isnotdiscussedinthisregardeither(exceptforthebriefdiscussioninHan,1998).Althoughthedistributionofcontrastivewaiscross-categorialasmentionedintheprevioussection,manypreviousanalysesdiscussonlysimpleNPsatthesubjectposition,oftenincontrastwithNPsmarkedwiththenominativecasemarkerga(e.g.Kuno,1972).Butalthoughthewa-markingonthesubjectdoesnotco-occurwiththedefaultga-markingonthesubject,wemaystillconsiderthatthereisanunderlyingnominativemarkerga.Then,itisnotthedistinctionbetweenwaandga.Instead,thedistinctionisaboutthepresenceorabsenceofcontrastivewa.Anothershortcominginrelationtothispointisthatthecontrastiveeffectwithothercaseparticles(suchasoACC)israrelydiscussed.

Oneanalysisofthepresuppositionassociatedwithcontrastivewaisthatthereisanotherelementinthecontextincontrasttotheonemarkedwithwa(Miyagawa,1987;Shibatani,1990;Han,1998).Wemightcallthisapproachthe

MERECONTRAST

analysis.Thisisproblematicbecauseitbecomesdifficulttodistinguishbetweenthecontrastswith

andwithoutwa.Inparticular,merecontrastpresuppositionistooweakforthecontrastwithwaasshowninthefollowingexample:

(14)Followingtheutterance‘Hereareabananaandamango.’

Ken-waKen-TOP

BANANA-o/#wa

tabe,ate(and)

MANGO-motabetaate

banana-ACC/CONTmango-too

‘Kenatethebanana,andatethemangotoo.’

10

Eventhoughthemerecontrastrequirementissatisfied,thewa-markingisinfelicitousinthiscontext.

Choi(1997)(citingSzabolcsi)arguesthatcontrastivefocushasEXCLUSIVEforcesuchthatnothingelseistrueforthecase.Butthisistoostrongforcontrastivewaasshownbelow.(15)Followingtheutterance‘Hereareabanana,amango,andatomato.’

Ken-waKen-TOP

BANANA-wa

tabenakatta-ga,didn’teat-but

MANGO-watabeta.ate

banana-CONTtomato-CONT

‘Kendidn’teata/thebanana,butatea/themango.’Ken-waKen-TOP

motironofcourse

TOMATO-mo

tabeta.ate

tomato-too

‘Ofcourse,Kenatetomato,too.’

Thenextgroupofanalysesassumesapresuppositionthatconsiderscontrastswithandwithoutwaidentically(Teramura,1991;Noda,1996;Choi,1997tosomeextent).Theiranalysessharethebasicideashowninthefollowingexample:(16)Ken-wa

Ken-TOP

PERU-de-waPeru-in-CONT

BANANA-o

tabeta.ate

banana-ACC

‘KenatebananasinPeru.’

Presupposition:‘Kendidn’teatsomethingelsesomewhereelse.’

Crucially,thecontrastrelationbetweenPeruandsomewhereelseandtherelationbetweenbananaandsomethingelseareidentical.WemightcallthisPARALLELCONTRASTanalysis.

Thereareafewproblemswiththisapproach.First,theanalysisassumesthatthereisaplacewhereKendidn’teatsomething.Butaswewillobserveindetailinthenextsection,thecontrastwithoutwadoesnothavethisstrongpresupposition.Second,theanalysishasaninherentdifficultydealingwiththeasymmetryinvolvingtheuniversalquantifierasshownin4.Itpredictsthatthereisnoasymmetry.

Mostofthepreviousanalysesdonotdistinguishbetweenthecancellabilityofpresupposition.OnlyChoi(1997)

11

explicitlystatesthatparallelcontrastpresuppositionisaconversational(cancellable)implicature.Butthisstatementistooweak.Aswewilldiscussinthenextsection,thecontrastwithwadohaveaconventionalimplicature,whichneedstobedistinguishedfromthecontrastwithoutwa.Theparallelcontrastanalysisoverlooksthisaspectandlosesthecrucialdistinctionsbetweenthecontrastswithandwithoutwa.

Huruta(1982)presentsaverydetailedanalysisoftheparticlewa.Here,weobservesomeproblemswithhisanalysis.First,Hurutadoesnotcomparethepresuppositionofcontrastivewawiththepresuppositionwithoutwa.Thustheasymmetryin4remainstobeexplained.Next,letusapplyHuruta’sanalysistothefollowingexample:(17)Inresponsetothequestion‘DidKeneata/themango?’

Ken-waKen-TOP

BANANA-waBanana-CONT

tabeta.ate

‘Kenatea/thebanana.’

Presupposition:‘Kendidn’teata/themango.’

ForthesemanticrepresentationXforthewa-markedphrase(bananainthiscase),letusdenotethecorresponding

semanticrepresentationinthepresuppositionasXc.Huruta’sintuitionisthatXc

(somethingotherthanbanana).

Huruta(1982)attemptstocapturethisrelationbyrepresentingXcasX(thispointissharedbyTeramura,1991).Butfacingvariousphrasetypes,Hurutaneededtoassumedifferentrelationsforthemasshowninthefollowingtable(hisformulasaretranslatedintoastandardlogicallanguage):INSERTTABLE1ABOUTHERE

ButtheproposedrelationbetweenXandXcisratheridiosyncratic.TheanalysisXc

Xdoesnotholdingeneral.

Inaddition,Huruta(1982)proposestwoseparateformulasforthenegativecase,reflectingthenegativescopeambiguity.Wewanttocapturetherelationinamoresystematicmanner.

Asseenabove,thepreviousapproachesareinsufficientasasolutiontothetwoproblemswehavebeenfocusingon.Weneedtointegratecontrastivesemanticsassociatedwithphonologicalprominenceandamorepreciseandaccuratewayofrepresentingthepresuppositionassociatedwithcontrastivewa.

12

4.AnalysisofContrastivewa.First,weassumethatthebasiccomponentofcontrastivesemanticsisassociatedwithprominence(e.g.Ladd,1996).Then,theproblemofderivingthecorrectsemanticsbecomestheprocessofcomposingthesemanticsassociatedwithprominenceandthatofcontrastivewa.Ourmainpointsforthissectionarethefollowing:(i)contrastrelationbetweenXandXc(asdiscussedintheprevioussection)canbecharacterizedintermsofALTERNATIVES(Rooth,1985)and(ii)therearetwotypesofcontrastdependingonthepresence/absenceofthecontrastivewa.

Inthefirstsubsectionbelow,weintroducetheideainformally.Thesecondsubsectionpresentsaformaltreatmentofthesameidea.Then,inthesubsequenttwosections,eachoftheproblemswillbediscussedindetail.

4.1.DifferentTypesofContrasts.Inthissubsection,weinformallyexplorethefunctionofcontrastswithandwithoutwa.Sinceweassociatethebasiccontrastivesemanticswithphonologicalprominence,therecanbeacontrastregardlessofthepresenceofwa.Themainpointisthatthecontrastiveforceisdistinctdependingonthepresenceofwa.

First,letusobservethecasewithoutwa,whichinvolvesbothconventional(non-cancellable)andconversational(cancellable)implicaturesasshownbelow.Bothpositiveandnegativesentencesareshowntodemonstratethepolarityofthepresuppositions.(18)a.Ken-wa

Ken-TOP

BANANA-o

tabeta.ate

banana-ACC

‘Kenatea/thebanana.’Presupposition:(i)

(ii)

b.Ken-waKen-TOP

BANANA-o

‘Somethingelseisinvolved.’(conventional,non-cancellable)‘Kendidn’teatsomethingelse.’(conversational,cancellable)tabenakatta.didn’teat

banana-ACC

‘Kendidn’teata/thebanana.’

13

Presupposition:(i)

(ii)

‘Somethingelseisinvolved.’(conventional,non-cancellable)‘Kenatesomethingelse.’(conversational,cancellable)

Theassumptionhereisthatwhenawordisprominent,itssemanticsisincontrastwithanotherreferent.Aformalrepresentationwillbeintroducedinthenextsubsection.Thepresuppositionisactuallyidenticaltothemerecontrastdiscussedintheprevioussection.Butwemustnotethatthemerecontrastanalysisisappliedtothepresuppositionofcontrastivewa.Theotherpartofpresupposition(ii)ispossibleonlyundertheconditionwherethecontextallowsit.Butthispartiscancellable,whichwillbediscussedincontrastwiththewa-markedcasebelow.Wenowturntothecasewithcontrastivewa.(19)a.Ken-wa

Ken-TOP

BANANA-wa

tabeta.ate

banana-CONT

‘Kenatea/thebanana.’

Presupposition:‘Kendidn’teatsomethingelse.’(conventional)b.Ken-waKen-TOP

BANANA-wa

tabenakatta.didn’teat

banana-CONT

‘Kendidn’teata/thebanana.’

Presupposition:‘Kenatesomethingelse.’(conventional)

Thepresuppositionsaboveisactuallyidenticalto18iiforbothaandb.Butitisnow

conventionalized/grammaticalized.Thisdistinctioncanbeobservedin14.Theutterance14cannotbefelicitousifthecontrastwithoutwahasthesamepresuppositionasthecasewithwa.Thesituationcanbesummarizedasfollows:(20)a.Contrastwithoutwa:Thepresupposition(conventional)isthatthereissomedistinctXc(or,somethingelse

isinvolved).

b.Contrastwithwa:ThepresuppositionisthatthereissomeXcwhichdoesnotholdinthecurrentsituation(i.e.XcisnecessarilydistinctfromX).

Since20bismorespecificthan20a,wesaythatthecontrastwithwaisSTRONGERthanthecontrastwithoutwa.

14

InSection3,wehaveobservedthatparallelcontrastanalysisisnotaccurateforexampleslike16.Wecanderivethecorrectpresuppositionsforthesameexample(anditsnegativecounterpart)byapplyingouranalysisasfollows:(21)a.Ken-wa

Ken-TOP

PERU-de-waPeru-in-CONT

BANANA-o

tabeta.ate

banana-ACC

‘KenatebananasinPeru.’Presupposition:(i)

(ii)

b.Ken-waKen-TOP

‘Kendidn’teatbananassomewhereelse.’‘Somethingotherthanbananaisinvolved.’

BANANA-o

(fromPERU-de-wa)(fromBANANA-o)

PERU-de-waPeru-in-CONT

tabenakatta.didn’teat

banana-ACC

‘Kendidn’teatbananasinPeru.’Presupposition:(i)

(ii)

‘Kenatebananassomewhereelse.’‘Somethingotherthanbananaisinvolved.’

(fromPERU-de-wa)(fromBANANA-o)

Here,distinctpresuppositionscoexist.Itisalsopossiblethat,forexamplein21a,thereisaconversational(cancellable)implicaturesuchas‘Kendidn’teatsomethingelsesomewhereelse’.Thiscorrespondstotheparallelcontrastanalysis,whichistoostrongaswehavearguedintheprevioussection.Althoughitisnoteasytoshowthatthistypeofparallelcontrastisonlyconversational(cancellable),thefollowingexampleseemstoprovideasupport.(22)Followingtheutterance‘KenateneitherbananasnormangoesinIndonesia.’

a.Ken-waKen-TOP

PERU-de-waPeru-in-CONT

BANANA-o

tabeta.ate

banana-ACC

‘KenatebananasinPeru.’Presupposition:(i)

(ii)

b.#Ken-wa

Ken-TOP

‘Kendidn’teatbananassomewhereelse.’‘Somethingotherthanbananaisinvolved.’

BANANA-wa

(fromPERU-de-wa)(fromBANANA-o)

PERU-de-waPeru-in-CONT

tabeta.ate

banana-CONT

‘KenatebananasinPeru.’

15

Presupposition:(i)

(ii)

‘Kendidn’teatbananassomewhereelse.’‘Kendidn’teatsomethingelseinPeru.’

(fromPERU-de-wa)(fromBANANA-wa)

Ifthecombinedpresuppositionin22awasaparallelcontrast‘Kendidn’teatsomethingelsesomewhereelse’,theutterancewouldbeunacceptable.Thecombinedpresuppositionin22bactuallycorrespondstosuchaparallelcontrast,andthus,itisunacceptable(assumingthattheonlythingstoeatunderconsiderationarebananasandmangoes).

Insummary,wehaveshownthatthepresuppositionofanutterancecanbederivedfromthepresuppositionsassociatedwithphonologicalprominenceandcontrastivewa.

4.2.AlternativeSemantics.Inthissection,wedevelopaformalrepresentationoftheanalysispresentedintheprevioussection,byadoptingtheALTERNATIVESEMANTICSframework(Rooth,1985;Rooth,1992;Rooth,1996).Theessenceofthisformulationisthatcontrastivesemanticsanalyzedintheprevioussubsectioncanbederivedfromtwosemanticvalues,theordinarymeaningandthealternativessets.AlthoughalternativesemanticshasbeenappliedtoEuropeanlanguagestoagreatextent,e.g.,theanalysisofonlyinEnglish,therehavebeenfewinstancesofapplicationtocontrastivewa(oneexceptionisWee(1995)whoworksonadifferentsetofproblemsfortheKoreancounterpartofwa).

LetusfirstdiscusstherelationbetweensomesemanticcomponentXinanutteranceandacontrastiveelementXc.Thisrelationcanbeuniformlycapturedbythenotionofalternativesregardlessofthephrasetype.ThisgeneralizesthecaseofHuruta(1982)wheredistinctrelationsareusedfordifferentphrasetypes.Now,letusobservethefollowingexample:(23)Ken-wa

Ken-TOP

BANANA-wa

tabeta.ate

banana-CONT

‘Kenatea/thebanana.’

Presupposition:‘Kendidn’teatsomethingelse.’

16

First,thesentencecanbedenotedasϕ

[SKen[banana]C

wa

ate]wherethecontrastmarkedbythephonological

prominencewithinthewa-markedphraseisrepresentedas[]Cwa.

Theparticlewaisanalyzedasanoperatorandnot

includedinthisrepresentationasalexicalunit.Thetruth-conditionalmeaningoftheutteranceistheordinarymeaningϕ

o

eatbanana

kenusingthefunctor-argumentrepresentationfunctorArg1

Argkand

assumingasimplifiedsemanticsforbanana.

Theideaofalternativessetisthatthecontrastinvolvingacertainelementisrepresentedbyasetofalternatives.Forthepresentexample,thealternativessetwithrespecttothecontrastCisrepresentedas,say,

ϕ

cwa

eatbanana

ken

eatmango

ken

.Ingeneral,weneedamechanismtoproperlyfixtheset,

whichisafocusofdiscussionintheliterature(Rooth,1992).Buttoavoidcomplications,weignorethisaspectinthispaper.

TheadvantageofthepresentapproachoverHuruta’s(1982)isthatthecontrastcanalwaysberepresentedasanalternativessetregardlessofthephrasetype.Intheabove,thealternativecomponentcorrespondstothesemanticsofanounphrase.Butwecanalsoconsideralternativessetscorrespondingtoaquantifier,anadverb,oraverb.

Next,thesemanticsofparticlewacanberepresentedasfollows:p

ϕ

cwa

p

falsewith‘p’standingfor

theextensionofpatthecurrentworldasintheMontaguetradition(Montague,1974).Informally,thenotation‘p’picksupanelementfromthesetdenotedbypandgivesitstruthcondition.ThiscorrespondstotheinformaldescriptionthatthereissomeXcthatdoesnotholdinthecurrentsituation.Forthecaseofcontrastwithoutwa,theutteranceisrepresentedasϕ

[SKen[banana]Cate]where[]Cindicates

thatthecontrastisnotwithinawa-markedphrase.Theonlydifferencehereisthatweconsideranoperatortoderive

presupposition:p

ϕ

c

p

ϕ

o

,where‘

’meansthatneithersideoftherelationlogicallyimplytheother

side.Thisoperatorappliestoboththeordinarymeaningandthealternativesset,andderivesapresuppositionthatthereisanalternativewhichislogicallydistinctfromtheordinarymeaning.Thissituationwasglossedas‘somethingelseisinvolved’intheprevioussubsection.

Letusnowconsiderthenegativecasesuchasthefollowing.

17

(24)Ken-wa

Ken-TOP

BANANA-wa

tabenakatta.didn’teat

banana-CONT

‘Kendidn’teata/thebanana.’

Presupposition:‘Kenatesomethingelse.’Werepresenttheutteranceasfollows:ϕ

[SKen[banana]C

ateneg]wherenegativemorphologyisrepresentedin

wa

ϕasinB¨uring(1997)(butunlikeRooth,1996).Thetruth-conditionalmeaningandthepresuppositioncanthenbeformulatedinexactlythesamewayasthepositivecase.Oneaspectthatintroducecomplicationisthescopeofnegation.Whilethisisaninterestingtopic,itisleftforfuturework.Finally,weconsiderthecasewithtwotypesofcontrasts.(25)Ken-wa

Ken-TOP

PERU-de-waPeru-in-CONT

BANANA-o

tabeta.ate

banana-ACC

‘KenatebananasinPeru.’Presupposition:(i)

(ii)

‘Kendidn’teatbananassomewhereelse.’‘Somethingotherthanbananaisinvolved.’

(fromPERU-de-wa)(fromBANANA-o)

Thecombinedpresuppositionistheconjunctionofthetwopresuppositionswehavealreadyseen.Asarguedearlier,apresuppositionsuchas‘Kendidn’teatmangoesinIndonesia’isconsideredaconversationalimplicature,i.e.aresultofgeneralinference.Thepresuppositionsofcontrastivewaandtheothercontrastoperatesimultaneouslybutnotsymmetrically(unliketheparallelcontrastanalysis).

4.3.SolutiontoProblem1:ComplexPhrase.Assoonasweconsiderphonologicalprominenceand

alternative-semanticsapproach,weobtainasolutiontothefirstproblem.Letusconsiderthemodelcase3repeatedbelow.

(26)a.Ken-wa

Ken-TOP

[

NAOMI-noNaomi-GEN

bananabanana

]

-wa-CONT

tabeta.ate

18

‘KenateNaomi’sbanana.’

Presupposition:‘Kendidn’teatsomeoneelse’banana.’b.Ken-waKen-TOP

[

Naomi-noNaomi-GEN

BANANA

]-wa-CONT

tabeta.ate

banana

‘KenateNaomi’sbanana.’

Presupposition:‘Kendidn’teatsomethingelseofNaomi.’

Reflectingthephonologicalprominence,werepresentthesentencestructureasfollows:(27)a.ϕa

b.ϕb

[SKen[Naomi’s]C

bananaate]

wa

wa

[SKenNaomi’s[banana]C

ate]

Theordinarymeaningfor27aand27bcanberepresentedasfollows:

ϕa

o

ϕb

o

eatbanana

ken

possessbanana

naomi.Thealternativessetsareasfollows:

(28)a.ϕa

c

eatbanana

kenkenken

possessbanana

naomierika

eatbanana

possessbanana

b.ϕb

c

eatbanana

possessbanana

naomi

eatmango

ken

possessmango

naomi

Asshownearlier,thesemanticsofwaistheoperatorp

ϕ

cwa

p

false.Theresultingpresuppositioncan

thenbeobtainedasfollows:

(29)a.

b.

eatbanana

ken

possessbanana

erika

falsefalse

eatmango

ken

possessmango

naomi

Theseexpressionscorrectlyrepresentthepresuppositionsinformallydescribedearlier.

4.4.SolutiontoProblem2:UniversalQuantifier.Wenowtackletheasymmetryproblemexemplifiedby4,repeatedbelow.

19

(30)a.Ken-wa

Ken-TOP

MINNA-o

hometa.praised

everyone-ACC

‘Kenpraisedeveryone.’b.*Ken-wa

Ken-TOP

MINNA-wa

hometa.praised

everyone-CONT

‘Kenpraisedeveryone.’

First,thesentencesfortheabovecasescanberepresentedasfollows:(31)a.ϕa

b.ϕb

[SKen[everyone]Cpraised][SKen[everyone]C

o

wa

praised]

o

Theordinarymeaningisϕa

ϕb

XpraiseXken.TheconstraintsonXtobethetypeofhumanbeing

canbeapplied,butisomittedforpresentationpurposes.Forthealternativesset,weconsideralternativequantifiersinthespiritofGeneralizedQuantifier(BarwiseandCooper,1981).UnlikeHuruta’s(1982)approach,wecandirectlyapplyalternativesemanticstoquantifiers.(32)

ιX

naomi

praiseXken

praisenaomi

ken

Naomi

XpraiseXken

everyone

ϕa

c

ϕb

cwa

XpraiseXken

noteveryone

XpraiseXken

someone

...

XpraiseXken

nobody

Now,letusexaminethecaseinvolvingacontrastwithoutwa.Byapplyingthepresuppositionoperator

p

ϕ

c

p

ϕ

o

(forcontrastwithoutwa)ontheordinarymeaningandthealternativesset,wecanobtainany

ofthealternativesdistinctfromtheordinarymeaning.

Forthecaseinvolvingcontrastivewa,theoperatorp

ϕ

cwa

p

falseappliestothealternativesset.All

alternativesbut

XpraiseXken

contradicttheordinarymeaningthatKenpraisedeveryone.Thus,there

20

alreadyisasignificantasymmetryabouttheavailabilityofgeneralizedquantifiers.Now,letusfocusontheonlypotentialproposition,

XpraiseXken.Logicallyspeaking,itisconsistentwiththepropositionofthe

sentence,XpraiseXken.But

XpraiseXken

isinconsistentwithanycontextwhereatleastone

personispraisedbyKen,wherethesentence‘Kenpraisedeveryone’isinfelicitousinJapanese.Infact,thesituationisanalogousinEnglishasshownbelow.

(33)Kenpraisedeveryone.#Infact,hepraisednobody.

Asaconsequence,noalternativessetcouldderivetheexpectedpresupposition.Thisdemonstratestheasymmetrybetweenthecaseswithandwithoutwa.

5.Conclusion.Thispaperpresentsananalysisofcontrastivewaastheinteractionofphonologicalprominenceandtheparticlewawithintheframeworkofalternativesemantics.TheanalysiscanbeappliedtoaccountfortheproblematiccasesinvolvingcomplexNPsandtheuniversalquantifiers.Theproposedformalizationprovidesamoreexplicitanalysisofcontrastivewathanpreviousones,andbringsthediscussiontothelevelcomparabletotheanalysesoffocusforEnglishandotherlanguages.

Oneadvantageofthepresentproposalisthatitisanintegrationoffairlystandardideasincludingphonologicalprominence,non-truthconditionalmeaning,andalternativesemantics,whichareindependentlymotivatedforvariousphenomenainvariouslanguages.Inaddition,theproposedcontrastivesemanticsdoesnotdependonthephrasetypeasinthepreviousworks,allowingasystematicanalysisofvariousphrasetypesappropriateforthecross-categorialdistributionofcontrastivewa.

Thereareseveralfuturedirectionsalongthecurrentapproach.First,sincetheproposedcontrastiveanalysisappliestoanarbitraryphrasetype,thewholerangeofcontrastivewa,inadditiontotheonesshownhere,canbeexplored.Thepresentapproachcanalsoprovideabasisfortheanalysisofinteractionbetweencontrastivewaandnegationincludingnegativescoping.Next,similarapproachescouldbedevelopedforotheradverbialparticles,e.g.mo‘too’.Finally,itwouldbeinterestingtoseetheconnectiontootherlanguages.Forexample,KoreanandHindihaveparticleswhosefunctionsaresimilartothoseofJapanesewa.

21

Bibliography

Aoki,Reiko.1992.GendaigoJoshi“Ha”-noKoubunron-tekiKenkyu(SyntacticAnalysisof“Wa”inModernJapanese).Tokyo:KasamaShoin.

Barwise,JonandCooper,Robin.1981.GeneralizedQuantifersandNaturalLanguage.LinguisticsandPhilosophy,4:159–219.

Beaver,DavidIan.1997.Presupposition.InvanBenthem,JohanandterMeulen,Alice,editors,HandbookofLogicandLanguage,pages939–1008.Cambridge,MA:MITPress.

B¨uring,Daniel.1997.TheGreatScopeInversionConspiracy.LinguisticsandPhilosophy,20:175–194.

Choi,Hye-Won.1997.TopicandFocusinKorean:TheInformationPartitionbyPhraseStructureandMorphology.InSohn,Ho-minandHaig,John,editors,Japanese/KoreanLinguistics,Vol.6,pages545–561.Stanford,CA:CSLIPublications.

Grice,H.P.1975.LogicandConversation.InCole,PeterandMorgan,Jerry,editors,SyntaxandSemantics,3:SpeechActs,pages305–315.NewYork:AcademicPress.

Han,Chung-hye.1998.AsymmetryintheInterpretationof‘-(n)un’inKorean.InAkatsuka,Norikoetal.,editors,Japanese/KoreanLinguistics7,pages1–15.Stanford,CA:CSLIPublications.

Huruta,Kei.1982.Kokugo-de-noHiteiHyougen-noImi:Zokuchou(Afterthoughtof“MeaningofNegationintheJapaneseLanguage”).MathematicalLinguistics,13(7):296–315.

Karttunen,LauriandPeters,Stanley.1979.Conventionalimplicature.InSyntaxandSemantics,Vol.11,pages1–56.NewYork:AcademicPress.

Kuno,Susumu.1972.FunctionalSentencePerspective:ACaseStudyfromJapaneseandEnglish.LinguisticInquiry,3(3):269–320.

Ladd,Robert.1996.IntonationalPhonology.Cambridge:CambridgeUniversityPress.

Miyagawa,Shigeru.1987.WaandtheWHPhrase.InHinds,JohnandIwasaki,Shoichi,editors,PerspectivesonTopicalization:TheCaseofJapanese‘WA’,pages185–217.Amsterdam:JohnBenjamins.

22

Montague,Richard.1974.TheProperTreatmentofQuantificationinOrdinaryEnglish.InThompson,RichardH.,editor,FormalPhilosophy,pages247–270.NewHaven,CT:YaleUniversityPress.Noda,Hisashi.1996.WA-toGA(WAandGA).Tokyo:KurosioPublishers.

Rooth,MatsE.1985.AssociationwithFocus.PhDthesis,UniversityofMassachusettsatAmherst.Rooth,Mats.1992.ATheoryofFocusInterpretation.NaturalLanguageSemantics,1:75–116.

Rooth,Mats.1996.Focus.InLappin,Shalom,editor,TheHandbookofContemporarySemanticTheory,pages271–297.Oxford:BlackwellPublishers.

Selkirk,Elisabeth.1984.PhonologyandSyntax:TheRelationBetweenSoundandStructure.Cambridge,MA:MITPress.

Shibatani,Masayoshi.1990.TheLanguagesofJapan.Cambridge:CambridgeUniversityPress.Tateishi,Koichi.1994.TheSyntaxof‘Subject’.Stanford,CA:CSLIPublications.

Teramura,Hideo.1991.Nihongo-noSintakusu-toImi(JapaneseSyntaxandMeaning),Vol.3.Tokyo:KurosioPublishers.

Wee,Hae-Kyung.1995.MeaningandIntonationAssociatedwiththeSubjectMarkerandtheTopicMarkerinKorean.InHarvardStudiesinKoreanLinguistics6.Cambridge,MA:HarvardUniversityPress.

Phrasetype

Xc

ken

NP(commonnoun)

λPY

childY

PY

λPYchildY

PY

PredicativeadjectiveλX

prettyX

Table1:Huruta’sAnalysisoftheContrastiveRelation

23