Prince,E.1981.OntheInferencingofIndenite-thisNPs.In:A.Joshi&B.Webber& - PDF document

292K LAUSVON H EUSINGER Prince,E.1981.OntheInferencingofIndenite-thisNPs.In:A.Joshi&B.Webber&I.Sag(eds.). ElementsofDiscourseUnderstanding. Cambridge:CUP,231–250. Ruys,E.G.1992. TheScopeofIndeÞnites .Ph.D.dissertation,UniversityUtrecht.Utrecht:LEd. Speas,P.&Tenny,C.2003.CongurationalPropertiesofPointofViewRoles.InA.DiSciullo(ed.). AsymmetryinGrammar .Amsterdam:JohnBenjamins.315–344. Reinhart,T.1997.QuantierScope:HowLaborisDividedbetweenQRandChoiceFunctions. Linguis- ticsandPhilosophy 20,335–397. vonHeusinger,K.2002.SpecicityandDenitenessinSentenceandDiscourseStructure. Journalof Semantics 19,245–274. vonHeusinger,K.&Kornlt,J.2005.TheCaseoftheDirectObjectinTurkish:Semantics,Syntaxand Morphology. TurkicLanguages 9,3–44. Winter,Y.1997.ChoiceFunctionsandtheScopalSemanticsofIndenites. LinguisticsandPhilosophy 20,399–467. Winter,Y.2005.OnsomeProblemsof(In)denitenesswithinFlexibleSemantics. Lingua 115,915–936. Wright,S.&Giv ´ on,T.1987.ThepragmaticsofIndeniteReference:QuantiedTtext-basedStudies. StudiesinLanguage 11,1–33. Yeom,J.1998.APresuppositionalAnalysisofSpecicIndenites:CommonGroundsasStructured InformationStates.GarlandPublishing. Zamparelli,R.2005.TheStructureof(In)deniteness. Lingua 115,915–936. R EFERENTIALLY A NCHORED I NDEFINITES 291 Bossong,G.1985.EmpirischeUniversalienforschung. DifferentielleObjektmarkierunginden neuiranischenSprachen .T¨ ubingen:Narr. Chierchia,G.2001.APuzzleaboutIndenites.In:C.Checchetto,G.Chierchia&M.-T.Guasti(eds.). SemanticInterfaces:Reference , AnaphoraandAspect .Stanford:CSLIPublications,51–89. Comorovski,I.1983. OnCliticDoubling .MSCornellU. Dede,M.1986.DenitenessandReferentialityinTurkishVerbalSentences.In:D.Slobin&K.Zimmer (eds.). StudiesinTurkishLinguistics .Amsterdam:Benjamins,147–164. Dobrovie-Sorin,C.1994.TheSyntaxofRomanian. ComparativeStudiesinRomance .Berlin:Mouton deGruyter. Enc ¸,M.1991.TheSemanticsofSpecicity. LinguisticInquiry 22,1–25. Farkas,D.1978. DirectandIndirectObjectReduplicationinRumania.PapersfromtheSeventeenth RegionalMeetingoftheChicagoLinguisticsSociety(CLS) 14:UniversityofChicago,88–97. Farkas,D.[1981]1985. IntensionalDescriptionandtheRomanceSubjunctiveMood .NewYork: Garland. Farkas,D.1995.SpecicityandScope.In:L.Nash&G.Tsoulas(eds.). ActesduPremierColloque Langues&Grammaire .Paris,119–137. Farkas,Donka2002.SpecicityDistinction. JournalofSemantics 19,213–243. Fodor,J.&Sag,I.1982.ReferentialandQuanticationalIndenites. LinguisticsandPhilosophy 5, 355–398. Farkas,D.&vonHeusinger,K.2003. StabilityofReferenceandObjectMarkinginRomanian .Ms. Univerist ¨ atStuttgart. Geurts,B.toappear.SpecicIndenites,Presupposition,andScope.In:R.B ¨ auerle,U.Reyle&T.E. Zimmermann(eds.). PresuppositionandDiscourse .Oxford:Elsevier. Haspelmath,M.1997. IndeÞnitePronouns .Oxford:ClarendonPress. Heim,I.1982. TheSemanticsofDeÞniteandIndeÞniteNounPhrases .Ph.D.dissertation.Universityof Massachusetts,Amherst.AnnArbor:UniversityMicrolms. Higginbotham,J.1987.IndenitesandPredication.In:E.Reuland&A.terMeulen(eds.). TheRepre- sentationof(In)deÞniteness .Cambridge/Mass.:MITPress,43–70. Hintikka,J.1986.TheSemanticsof‘acertain’. LinguisticInquiry 17,331–336. Ionin,T.2006.ThisisDenitelySpecic:SpecicityandDenitenessinArticleSystems. Natural LanguageSemantics 14,175–234. J ¨ ager,G.2004. PartialVariablesandSpeciÞcity. MsUniversityofPotsdam. Karttunen,L.1976.DiscourseReferents.In:J.McCawley(ed.). SyntaxandSemantics7:Notesfrom theLinguisticUnderground .NewYork:AcademicPress,363–385. Kasher,A.&Gabbay,D.1976.OntheSemanticsandPragmaticsofSpecicandNonspecicIndenite Expressions. TheoreticalLinguistics 2,145–188. Kornlt,J.1997. Turkish .Routledge:London. Kratzer,A.1998.ScopeorPseudoscope?ArethereWide-ScopeIndenites.In:S.Rothstein(ed.). EventsandGrammar .Dordrecht:Kluwer,163–196. Krifka,M.2001.Non-novelIndenitesinAdverbialQuantication.In:C.Condoravdi&G.Renardel (eds.). LogicalPerspectivesonLanguageandInformation .Stanford:CSLIPublications. Leonetti,M.2003.SpecicityandObjectMarking:TheCaseofSpanisha.In:K.vonHeusinger &G.A.Kaiser(eds.). ProceedingsoftheWorkshopSemanticandSyntacticAspectsofSpeci- ÞcityinRomanceLanguages .Arbeitspapier113desFachbereichsSprachwissenschaft,Universit ¨ at Konstanz,67–102. Ludlow,P.&Neale,S.1991.IndeniteDescriptions:InDefenseofRussell. LinguisticsandPhilosophy 14,171–202. Maclaran,R.1980.OnTwoAsymmetricalUsesoftheDemonstrativeDeterminers’in CornellWorking PapersinLinguistics 1. Maclaran,R.1982. TheSemanticsandPragmaticsoftheEnglishDemonstratives ,Ph.D.dissertation, CornellUniversity. Milsark,G.1974. ExistentialSentencesinEnglish .Ph.D.dissertation,MIT,Cambridge/Mass. 290K LAUSVON H EUSINGER Theexampledemonstratesthattheindexingmechanismisdependentonthecon- guration,andtheformulationin(41)needssomeaddition.Anopenquestionis, however,whetherthisweak-crossovereffectsholdonlyforspecicindeniteswith aparticularadjectivesuchas certain ,orotherspecicindenites,too. 5.SUMMARY Thegeneralwisdomassumesthatspecicindenitenounphrasessignalthat“the speakerhasaparticularreferentinmind”,whilethehearerdoesnotknowitsiden- tity.Thisintuitionshouldcapturetheprominentpropertiesofspecicindenites: theyhave(often)widescope,theyare(often)referentialexpression,theyare(often) presuppositional,andtheyare(often)noteworthy,i.e.theysignaldiscourseimpor- tance.However,Ihaveshownthatthementionedpropertiesarerathersupercial innatureandnotnecessaryforspecicindenites.Thiswasillustratedwithnar- rowscopespecicindenitesor“relativespecic”indenites,whichshowtypical featuresofspecicindenites(choiceofthelexicalitem certain ,choiceofacase markerinTurkishetc.).Analyzingtheseinstancesofspecicindenites,Ihave demonstratedthatspecicityexpressesananchoringrelationbetweenanindenite nounphraseandanargument.BuildingonEnc ¸’s(1991)analysisandgeneralizing it,theproposedanalysisisformallyreconstructedintermsoflechangeseman- tics.Whileadenitenounphraseindicatesthatthereferentisalreadygiveninthe context,aspecicindeniteintroducesanewdiscourseitemthathasa(pragmati- callysalient)linktoanalreadygivendiscourseitem.Anon-specicindenitejust introducesanewdiscourseitem(whichisnotlinkedtotheestablisheddiscourse). Thisgeneralsemanticformforspecicindenitescanbeunderstoodasanunder- speciedrepresentationthatneedsananchorinthecontext.Dependingonthetype ofanchorandthescopalbehavioroftheanchor,thespecicindeniteshowssome oftheabove-mentionedproperties. KlausvonHeusinger Institutf ¬ urLinguistik/Germanistik Universit ¬ atStuttgart D-70049Stuttgart Germany REFERENCES Aissen,J.2003.DifferentialObjectMarking:Iconicityvs.Economy. NaturalLanguageandLinguistic Theory 21,435–483. Bende-Farkas, ´ A.&Kamp,H.2001.IndenitesandBinding:FromSpecicitytoIncorporation. Lecture Notes ,revisedversion.XIIIESSLLI,Helsinki2001. 288K LAUSVON H EUSINGER establishedsetitispartitive(andspecic–accordingtoEnc ¸).Itisalsoobviousthat thepartitivehaswiderscopewithrespecttootheroperatorsinthesentence,since itisrelatedtoanestablishedset. (40)a.Severalchildren 1 enteredmyroom 2 . Dom ( D ) ={ 1 , 2 } (with1denotingaset) b.Iknewtwogirls 3 .3  1and1  Dom ( D ) 4.3.RelativeSpeciÞcity Inordertoaccountforspecicityintermsofrelativespecicity,weformulate thecondition(41)insimilarterms.AnNPisspecicifitsindex(orlename) canbelinkedtoanalreadyestablishedindex.Anadditionalrestrictionisthatthe alreadyestablishedindexmustbefromthecurrentsentence,ratherthanfromthe wholediscourse. 9 Inthissense,specicityissentence-bound,whiledeniteness isdiscourse-bound.Theformalreconstructionofthisviewofspecicitystates thataspecicNP i signalsthattheassociatedindexiislinkedbyasalient(nat- uralor‘reconstructable’)function(orrelation)toanotherindexjfromthesame sentence  . (41) RelativeSpeciÞcityCondition AnNP i inasentence  withrespecttoaleFandtheDomainofle- namesDom (  ) is[ + specic]ifthereisacontextualsalientfunction fsuchthati = f ( j ) andj  Dom (  ) Letusillustratethedenitiononourexamples(17),repeatedbelowas(42a)and (42b).Thespeakerofthedirectspeechin(42a)introducesanewindex1,suchthat theindex2ofthespecicindenitecanbelinkedtoitbyacontextuallysalient functionf.ThisfunctioncouldbespelledoutbysayingthatGeorgecanidentify thatstudentorthattherewasatemporalpointatwhichbothindividualswereatthe samelocation,etc.Thefunctiononlyindicatesthatoncewehavexedtheidentity oftheanchor(George)wecanalsoidentifytheidentityoftheanchoredindenite. In(42b),wehavetwopotentialanchorssuchthatwecanrelatetheindexofthe specicindenitetoeitheroneofthem,yieldingthetworepresentations(i)and (ii),whichstandforthetwoaccessiblereadings:in(i)Georgeistheanchorand “responsible”forthespecicindenite,whilein(ii)Jamesistheanchor: 9 Areviewernotedthatthisrestrictionisnotpreciseenough,sinceitwouldpredictthatin(i)the specicindenitecouldtakenarrowscope,whichisnotanavailablereadingof(i).Amoreelaborate restrictionseemsnecessary(e.g.intermsofc-command). (i)Ifeverytrainerarrivesontime,acertainathletewillsing. R EFERENTIALLY A NCHORED I NDEFINITES 287 Dom(D)andtheyareaccessiblefortheevaluationofthedeniteNPsin(37b).The twoindicesin(37b)canbelinkedtothealreadyestablishedindicesinthedomain forindices(orthedomainofestablisheddiscourseitems),whichlicensesthedef- initenessofthetwoNPs.Incontrast,in(38b)thetwoNPsareindenite,which meanstheirindicescannotbelinkedtoalreadyestablishedindicesordiscourse items.Therefore,theindeniteNPsintroducenewdiscourseitems. 4.2.Enc üÕsPartitiveSpeciÞcity Enc ¸modiestheFamiliarityConditionofdenitevs.indeniteNPstothepar- titivityconditionforthecontrastbetweenspecic/partitivevs.non-specic/non- partitiveindeniteNPs.LikedeniteNPs,specicNPssignalthattheassociated discoursereferentislinkedtothealreadyestablisheddiscourse.Otherthanwith denites,thislinkisnotdirectbutitisthe partof -relationorthepartitiverelation. Hereshehastodistinguishbetweenthepluralcase(i)andthesingular(ii).Inthe pluralcase( severalchildren...twoofthegirls )theformalreconstruction(i)says thatthepartitivityislicensedbythefactthattheindexi(standingforagroupof entities,suchas twoofthegirls )ispartofanindexjthatstandsforanalreadyestab- lishedgroupofentities( severalchildren ).Inthesingularcase(ii),thepartitivityof theNPislicensedbythefactthatthegroupconsistingofthatonediscoursereferent (thereforeasetwithjustoneindex: { i } )ispartofthealreadyestablishedgroupj. (39)isareconstructionofEnc ¸’s(1991,7ex.(22))conditionforpartitiveNPs. 8 (39) Enc üÕsSpeciÞcity/PartitivityCondition(adaptedversion) AnNP i inasentence  withrespecttoaleDandtheDomainof lenamesDom(D)is (i)forNP i plural:[ + specic]ifthereisajsuchi  jandj  Dom ( D ) or (ii)forNP i singular:[ + specic]ifthereisajsuch { i } jandj  Dom ( D ) Sentence(40a)introducesanewindex(ordiscourseitem),asetofseveralchil- dren.The(implicit)partitive twogirl(s) in(40b)isrelatedtothissetbythesubset relation.Thismeansthattheindexordiscourseitem1,i.e.asetoftwogirls,is asubsetofindex3standingforthesetofseveralgirlsalreadyestablished.Since thissetoftwogirlsisnewitisindenite,butbecauseofitsrelationtoanalready 8 Enc ¸’s(1991:7)ownreconstructionismoredifculttoread:“AllNPscarryapairofindices,therst ofwhichrepresentsthereferentoftheNP.Theindicesthemselvesbearadenitenessfeature.Thefeature ontherstindexdeterminesthedenitenessoftheNP,asusual.Thedenitenessfeatureonthesecond indexdeterminesthespecicityoftheNPbyconstrainingtherelationofthereferentoftheNPtoother discoursereferents. (i)Every[NP  ]  i , j  isinterpretedas ( x i ) and x i  x j ifNP  i , j  isplural { x i } x j ifNP  i , j  issingular 286K LAUSVON H EUSINGER 4.SPECIFICITYASREFERENTIALANCHORING Themainthesisofthispaperisthatspecicityindicatesthatanexpressionisrefer- entiallyanchoredtoanotherargumentexpressioninthediscourse.‘Referentially anchored’meansthatthereferentofthespecicNPisfunctionallydependent onthereferentofanotherexpression.Thisideacanbespelledoutbyextending Heim’s(1982:369f)FamiliarityConditionandmodifyingEnc ¸’spartitivecondition forspecicindenites. 7 4.1.FamiliarityforDeÞniteness Enc ¸formalizesherviewofspecicityintermsofHeim’s(1982)familiarity approachtodiscoursestructure.Heimdenesdenitenessintermsoffamiliarity, ormoreformally,intermsofidentityoftheindicesoflecardsfornounphrases (NPs),asdenedin(36),andillustratedby(37)–(38): (36) HeimÕsFamiliarityCondition AnNP i inasentence  withrespecttoaleDandtheDomainof lenamesDom(D)is (i)[ + denite]ifi  Dom(D),anditis (ii)[ Š denite]ifi /  Dom(D) Heim(1982)reconstructsdenitenesswithrespecttothealreadyestablished discourse.EveryNPcomeswithanindexi,whichrepresentsthediscoursereferent (orHeim’s“lecard”)associatedwiththatNP.Ifthediscoursereferentiisalready introducedinthediscourse–ormoreformallyiftheindexiisanelementofthe setofallestablisheddiscoursereferentsDom(D),thentheNPmustbedenite. If,however,thediscoursereferentiisnotamongthealreadyestablisheddiscourse referents,i.e.ifi /  Dom(D),thentheNPmustbeindenite.Denitenesssignals thefamiliarityofthediscoursereferentassociatedwiththeNP. (37)a.Aman 1 meetsawoman 2 .Dom(D) ={ 1,2 } b.Theman 1 talkstoher 2 .1,2  Dom(D) (38)a.Aman 1 meetsawoman 2 .Dom(D) ={ 1,2 } b.Aman 3 talkstoawoman 4 .3,4 /  Dom(D) In(37a)thetwoindeniteNPsintroducenewlecardsordiscourseitems,which weindicatebythetwoindices1and2.Theseindicesformthedomainoflenames 7 Eventhoughtheapproachisformulatedinlechangesemantics,itadaptsalsoinsightsfrom Farkas’(2002)conceptof“dependentreadings”andBende-FarkasandKamp’s(2001)discussionof functionalreadings. R EFERENTIALLY A NCHORED I NDEFINITES 285 canbedisputedonthegroundsofthegivenexample:in(33b), ikikõzõ refersto twogirlsthatareinthesetofmentionedchildren–however,theidentityofthose girlsisnotgiven(onlyrestricted).Wecannotgivejusticetothewholediscussion ofEnc ¸’sapproach(seevonHeusingerandKornlt2005). 3.4.RelativeSpeciÞcityinTurkish Theproblemofrelativespecicindenites(seesection2.5)isthatitiscontroversial whethertheyareepistemicspecicindenitesorexistentialindeniteswithaddi- tionalscopalproperties.ThedatafromTurkishshowthattheypatternwithother specicindenites(withtheepistemicandscopalindenites).Enc ¸presents(34a) thatshowstworeadingsforthespecicindenite acertainathlete :awide-scope reading(34b)andanarrowscopereading(34c): (34)a.Herantren ¨ or bellibiratlet-i /  atlet c ¸als ¸tracak. everytrainercertainoneathlete-Acc.willtrain ‘Everytrainerwilltrainacertainathlete.’ b.allthesameathlete(specic,widescope) c.eachoneadifferentone(specic,narrowscope) NotethatEnc ¸usesherethemodierbelli‘(a)certain’.Thiscontributestothespeci- cityoftheindeniteexpressionandthustothewell-formednessoftheaccusative marker.Itisinterestingtonotethatifwefronttheindenitetosentenceini- tialposition(whichisatopicposition),thenwereceiveonlythewide-scope specicreading,asin(35).Wewillcomebacktothisweak-crossovereffectin section4.4: (35)a. bellibiratlet-i herantren ¨ orc ¸als ¸tracak. certainoneathlete-Acc.everytrainerwilltrain ‘Everytrainerwilltrainacertainathlete’ b.allthesameathlete(specicwidescope) c.  eachoneadifferentathlete(specicnarrowscope) Enc ¸(1991:19)accountsfortheuseoftheaccusativecasebyassumingthatthe directobjectis“somehowdistinguished.Itisdistinguishedbecauseitstandsinthe contextuallysalientrelevantrelationtosomeotherobject”.Shesketchesaformal- izationofthisideabyusingSkolem-functions(orwhatshecalls“assignmentfunc- tions”)forthespecicindenite,followingaproposalbyHintikka(1986),which willbepresentedinthenextsection. 282K LAUSVON H EUSINGER 3.2.DifferentialObjectMarking(DOM)inRomanian Romanianhasthemarker pe tomarkcertaindirectobjects. Pe isobligatoryfor denitepronounsandpropernames,asin(24);itisobligatoryfordenitehuman nounphrasesandoptionalforspecichumannounphrases;(25)isaninstanceof thelatter. Pe -markingisungrammaticalfornon-specicindenitenounphrases,as in(26),wheretherelativeclauseisinthesubjunctive(seeFarkas1978,Farkasand vonHeusinger2003).Thuswecansummarizetheconditionsforhumanfullnoun phrasesinRomanian,asin(27): 3 (24)Maria  (l)-adesenat  (pe)Matei/el . Maria  (CL)hasdrawn  (PE)Matei/him ‘MariadrewMatei/him.’ (25)Maria(l)-adesenat (pe)unb  aiatdinfat üaei. Maria(CL)hasdrawn(PE)aboyinfrontofher. ‘Mariadrewaboyinfrontofher’ (26)Maria(  o)caut  a (  pe ) ostudent ÿ acares ÿ astierom ö anes ¸ te . Maria(  CL)lookfor(  PE)astudentwhoknow SUBJ Romanian ‘MariaislookingforastudentwhoknowsRomanian.’ (27)ConditionsforDOMinRomanian FullNP[ + SpeciÞc][ Š SpeciÞc] [ + Animate] ±Š [ Š Animate] ŠŠ Themarker pe isoptionalforspecicnounphrasesandungrammaticalfornon- specicones. 4 Thismeansthatthepresenceofthemarkerforcesaspecicinter- pretation.InRomanianwendthereversesituationtotheSpanishcontrastin(23): Thepresenceofthemarker pe accompaniedbythedoublingcliticforcesawide- scopereadingin(5a),whiletheabsenceofthemarkerallowsforbothreadingsin (5b)(Dobrovie-Sorin1994,229–230): 5 3 “CL”indicatesadoublingcliticlinkedtothedirectobject.Theconditionsforcliticdoublingin Romanianaresimilarto,butnotidenticalwith,theonesfor pe -marking.Therefore,cliticdoubling and pe -markingoftenco-occur.SeeComorovski(1983)forananalysisofcliticdoublingasobject agreement. 4 Therearecertainexceptions: nimeni (‘nobody’)and cineva (‘somebody’),whichareobligatorily introducedby pe ,evenifnon-specic. 5 WhilethegivenjudgmentsarefromDobrovie-Sorin,IleanaComorovski(p.c.)informsmethat thereadingsshegetsaredifferent.Both(28a)and(28b)haveonlyanarrowscopereading.Anexplicit partitiveexpressedby dintre in(28a’)makesawidescopereadingpossible.Awide-scopereadingfor (28b)ispossibleif ÔÞecare ’isreplacedby to  i (‘all’). 278K LAUSVON H EUSINGER 2.4.SpeciÞcityasNoteworthiness Specicitycanalsoexpressthediscourseprominenceofanindenitenounphrase. Aspecicindeniteisusedifthespeakerintendstosignalthattheassociateddis- coursereferentisimportantandwillbereferredbackbyanaphoricexpressions inthesubsequentdiscourse.Ionin(2006)callsthis speciÞcityasnoteworthiness . Thisdiscourseeffectisoftenthetriggerforthegrammaticalizationofthenumer- alsexpressing“one”towardsspecicindenitearticles(WrightandGiv ´ on1987) forHebrewandHawaiianCreole).Ionin(2006)usesthisconcepttoaccountfor theEnglishspecicindenitearticle this ,whichhasadifferentsemanticsfromthe homonymousdemonstrative(seeMaclaran1980,Prince1981).Thewide-scope readingwithrespecttotheverb want isshowninthecontrastin(12),fromIonin (2006,180). (12)a.Sarahwantstoread  a/  thisbookaboutbutteries,butshecan’t ndit. b.Sarahwantstoread  a/#thisbookaboutbutteries,butshecan’t ndone. Ionin(2006,181)illustratestheconceptofnoteworthinesswith(13),quotedfrom Maclaran(1982,88).(13b),whichcontainsthespecicindenitearticle this is felicitoussinceitsignalsadiscoursereferentthathasanimportantorprominent property. (13)a.Heputon  a/#this31centstampontheenvelope,sohemustwant ittogoairmail. b.Heputon  a/  this31centstampontheenvelope,andonlyrealized laterthatitwasworthafortunebecauseitwasunperforated. Ionin(2006,187)denesnoteworthinessItermsoffelicityconditions,ratherthan presupposition,whichsheonlyusesfordenitenounphrases.Iparaphraseher denitionin(14): (14)Anspecicindenitenounphraseofthetype[spe  ]isfelicitously usedifthespeakerintendstorefertoexactlyoneindividualxandthere existsaproperty  whichthespeakerconsidersnoteworthyandxis both  and  . 2.5.RelativeSpeciÞcity Theterm‘relativespecic’or‘intermediatescopespecicindenites’orrecently ‘longdistanceindenites’describesspecicindenitesthatdependonother expressions,andthereforeshowexiblescopebehavior.Thisobservationwas R EFERENTIALLY A NCHORED I NDEFINITES 277 Thereis,however,nocleartruth-conditionaldifferencebetweenthetworead- ings.Therefore,thepragmaticapproach(e.g.LudlowandNeale1991,Zamparelli 2005)assumesthatthedifferenceisduetotheamountofinformationthatisavail- abletoidentifyareferent.However,thereisaclearcontrastbetweenthespecic readingin(8b)andthenon-speciconein(8c),whichbecomesstrongerwith ananimatenoun,asin(9a)andthecontinuation(9b)and(9c)fromFodorand Sag(1982). (9)a. Astudent inSyntax1cheatedontheexam. b.HisnameisJohn. c.Wearealltryingtogureoutwhoitwas. 2.3.Partitives Milsark(1974)arguesthatindeniteNPscaneitherreceiveaweak(orexistential) interpretationorastrong(orpresuppositional)interpretation.In(10a)theinde- nite someghosts receivesaweakinterpretation,butitgetsastronginterpretationin (10b),i.e.itpresupposesthatthereareotherghosts.Thereadingin(10b)isgener- allycalled“partitive”. (10)a.Thereare someghosts inthishouse. b. Someghosts liveinthepantry;othersliveinthekitchen. Enc ¸(1991)developstheideaofspecicityaspartitivityandargues,basedonexam- pleslike(11),thattheaccusativecaseinTurkishmarksexactlythistypeofspeci- city.(11a)introducesasetofchildren,andtheaccusativemarkeddirectobject ikikõzõ in(11b)mustrefertoasubsetofthepreviouslyintroducedsetofchil- dren.Theunmarkeddirectobject ikikõz in(11c),however,cannotrefertoasub- setoftheintroducedchildren,butmustrefertoanothernotmentionedsetoftwo children. (11)a.Oda-m-a birkac üc üocuk gir-di room-1.sg.-Dat.severalchildenter-Past ‘Severalchildrenenteredmyroom.’ b.  Ikikõz - õ tan-yor-du-m twogirl-Acc.know-Prog.-Past-1.sg. ‘Iknewtwogirls.’ c.  Ikikõz tan-yor-du-m twogirlknow-Prog.-Past-1.sg. ‘Iknewtwogirls.’