Upload
dodung
View
224
Download
0
Embed Size (px)
Citation preview
1
WorkshoponArgumentStructure,LexicalSemanticsandMorphologyUNAM,June24‐July1,2009
HeidiHarley,[email protected]
OverviewInthisworkshop,wewillreviewthechangesinthegenerativetheoryofargumentstructureoverthelastcoupleofdecades,movingfromtheLexicalistviewpointofGBtoapurelysyntacticocentricapproach,asviewedthroughaveryparticularwindowofresearchonargumentstructurealternationsinarangeoflanguages.Firstwewillreviewtheta‐theory,andthenlookatanoverviewofthegenesisofthemoderncomplexVP,involvingavPandpossiblymoreprojections.Wewillthenconsiderargument‐structureanalysesfromseveraldistinctlanguages,includingEnglish,ItalianandJapanese,andwindupwithanoverviewofthemorphosyntaxofargumentstructureintheUto‐AztecanlanguageYaqui(Jiaki,Hiaki).Webpagewithclassreadings:http://dingo.sbs.arizona.edu/~hharley/courses/MexicoCity2009/HarleyWorkshop.htmTopicschedule:Day1:Thetaroles,argumentsforcomplexVPsyntaxDay2:ThelexicalsemanticsofverbaldecompositionDay3:Acasestudy:Causativescross‐linguisticallyDay4:EventstructureandsyntacticstructureDay5:HiakiargumentstructureParticipantPresentations:Anyparticipantwhowishestopresentontheiroriginalresearchiswelcometo,timepermitting!Discussitwithmeafterourfirstclasstoday.Ifyou'reinterestedinpresentingIsuggestthinkingalongthelinesofa20minutesession,likeashortconferencepresentation(10‐15minutes)withaquestionperiod(5‐10minutes),butwecanalsogoforamorefreewheelingformatifthatispreferred.OneortwooftheseperdaywouldIthinkbeverymanageable.Classroomdiscussion:Iwouldverymuchliketoencourageinterruptionsforquestionsanddiscussionatanytime!Thisisintendedtobeaworkshop,notalectureseries,whichmeansparticipationfromallsidesisessential–ifyouwouldliketorequestclarificationonanypoint,providefurtherdiscussion,suggestaconnectiontoanothertopicorpieceofdata,orwhatever,pleasedon'thesitatetodoso.AlthoughIamafraidmyspokenSpanishisutterlyminimal,andIwilllectureanddiscussinEnglish(myapologies!),youshouldfeelfreetoaskquestionsandprovidecommentsinSpanish–Esthelaandotherscanprovidetranslationsforme,andIcanunderstandalittlebit.Itismoreimportanttounderstandanddebatethelinguisticcontent.
2
1. ThematicRolesReading:Harley,H.(2009)ThematicRoles,InTheCambridgeEncyclopediaofthe
LanguageSciences,PatrickHogan,ed.,CambridgeUniversityPress,Cambridge.ModelofthegrammaticalsysteminGovernment‐Bindingtheory(1) GovernmentandBindingmodel PhraseStructureRules Lexicon (X‐bartheory) (Words:Category,pronunciation,θ‐grid) DeepStructure DSConstraints: representation ProjectionPrinciple θ‐Criterion Syntactictransformations HeadednessParameters (Movealpha,coindexation) SurfaceStructure SSConstraints: representation CaseLicensing,Subjacency ECP,BindingConditions EPP,Wh‐parameters PFspecific LFspecific transformations: transformations Intonationcontours Covertmovement (e.g.quantifiers) PF LF representation representationTheconstraintswereconceivedofas'modules'whichappliedindependentlyofeachothertofilteroutill‐formedrepresentationsAlltheconstraintsmadereferencetocertainstructuralrelations—eithergovernmentorbinding,hencethenameofthetheory.Areaofrelevancehere:Whatwereθ‐grids?Howdidtheθ‐criterionapply?Andwhatgeneralizationsremainunaddressedbythisapproach?
1.1 LexicalitemsinGBAlexicalitemhadapronunciation,asyntacticcategory,andaθ‐grid:
"The Base"
3
(2) AlexicaliteminGB: PHON: kiss
SYN:VSEM:[Agent,Patient] (or:[1,2],or[kisser,kissee])+somenotionofwhat‘kiss’means
Whatareθ‐roles?
Grammaticizedrepresentationsofthelexicalsemanticsofthepredicate—theencyclopedia/syntaxinterface,essentially.
Basicfunction:Constrainthenumberofargumentsthatappearwithaverb
Verbs,astheyarepredicates,describesituationswhichnecessarilyinvolveoneormoreentities,i.e.theyrequirearguments
Therunsituationrequiresoneentity Thekisssituationrequirestwoentities Thegivesituationrequiresthreeentities Thetradesituationrequiresfourentities
Thematicrolesgrammaticizetheencyclopedicsemanticsoftheverb,andinGBtheory,
interactedwiththeThetaCriteriontoensurethattheverbwilloccurinasentenceinwhichitcanbeadequatelyinterpreted
Couldjustleaveitat"Numberofarguments"Butwouldmissgeneralizationthatthereseemtobetypesofθ‐roles
1.2 Familiesofθroles
Differenttypesofθ‐rolesseemtoberelevantforsyntacticbehaviorHencenormalusagecategorizesθ‐rolesintobroadsemanticcategories;herearea
sample:(3) Somethematicroles: Agent:Initiatesandcontrolstheexecutionoftheverbalaction Example:MariainMariakissedPedro Theme:Movesorchangesstateasaresultoftheverbalaction Example:ThedoorinGuillermoopenedthedoor
Patient:LikeTheme,butneednotmoveorchangestate—undergoessomethingastheresultofatheverbalaction
Example:PedroinMariakissedPedro
4
Experiencer:Anargumentwhosementalstateisdescribedoraffectedbytheverbalaction.
Example:MercedesinMercedeslikesicecreamGoal:Anargumentexpressingtheendpointofsomemotiondescribedbytheverbal
action Example:ThehouseinAnnaenteredthehouseIncrementaltheme:Anargumentconsumedorcreatedbytheverbalaction,or
whosephysicalextentdelimitstheverbalaction Example:AnappleinFernandoateanapple.Beneficiary/Maleficiary:Anargumentwhichbenefitsfromorisadverselyaffected
bytheverbalaction Example:JoseinSelmabakedJoseacake.Source:Anargumentexpressingtheoriginofsomemotiondescribedbytheverbal
action Example:ThehouseinAnnaleftthehouse.Measure:Anargumentwhichexpressestheextentofanotherverbalargument Example:0.5kginThisbookweighs0.5kg.
Thevariouscategoriesoftheta‐rolesseemtobeimplicatedinseveralgrammaticalprocesses,especiallyinderivaitonalmorphology,butalsoinsyntax
(4) Grammaticalprocessesandtheta‐rolesa. EnglisheraffixationaffectsAgent‐assigningverbsonly; ProducesnominalsreferringtoAgentsorInstruments writer,teacher,sniper,driver,seller,buyer,user filler,chipper,wrapper,dryer,washer,computerb. TagalogpagprefixationappearstoaddanAgentargumenttoanagentless
verb(DatafromTravis2000,inturnfromMaclauchlan1989): Intransitive Transitive t‐um‐umba Xfalldown m‐pag‐tumba YknockXdown s‐um‐abog Xexplode m‐pag‐sabog YscatterX. l‐um‐uwas Xgotothecity m‐pag‐luwas YtakeXtothecity s‐um‐abit Xhang m‐pag‐sabit YhangX s‐um‐ali Xjoin m‐pag‐sali YincludeX (Note:umisaninfixthatappearsaftertheonsetofthefirstsyllableofthe
stem)
5
Amoreinvolvedexample:DifferencesinGoalvs.Beneficiaryθ‐rolesinItalian:FolliandHarley(2006):
(5) a. Beneficiaryargument Ibambinihannomangiatotuttalatortaallamamma thechildrenhaveeatenallthecaketomother
b. Goalargument GiannihadatounlibroaMaria JohnhasgivenabooktoMary
(6) a. Beneficiarydativebadinnominalization: *ilmangiaretuttalatortaallamamma (èstataunabuonaidea) theeatingthewholecaketomom (isbeenagoodidea)
b. Goaldativeokinnominalization
IldonareunlibroaMaria (èstataunabuonaidea) thedonatingabooktoMary (isbeenagoodidea)
(7) BeneficiarydativecliticbadwithA‐movementofobject a. Leèstatamangiatatuttalatorta.
to.herisbeenmadeeatenallthecake
b. *Tuttalatortaleèstatamangiata(daMarco). all the cake to.her is been eaten (by Marc)
(8) GoaldativecliticfinewithA‐movementofobject a. Leèstatodatoillibro(daMarco). to.herisbeengiventhebook(byMarc).
b. Illibroleèstatodato(daMarco). thebookto.herisbeengiven(byMarc).
1.3 θrole/syntaxmapping
InGBtheory,sinceverbswithsimilarsemanticshadsimilartheta‐grids,lexical
operationslikenominalization,couldselectivelyapplytocertainclassesofverbstoadjusttheirtheta‐grids(andsimultaneouslychangetheirmorphology)—themodelincludedagenerativelexicon,inwhichoperationscouldapplytoexpandtheBasepriortosyntacticcomputation.(See,e.g,.Chomsky's1970Remarksonnominalization).
6
Thereareothergeneralizationsconcerningthematicroles,however,whichneededadditionalprinciples/stipulationstocapture,particularlytodowiththelinkingproblem—howtorelatethematicrolestosyntacticpositions
Thereisatightbutnotperfectcorrespondencebetweenthethematicroleandthe
grammaticalfunctionofanargumentE.g.ifanactiveverbhasanAgent,thatargumentwillbethesubjectIfanactiveverbhasaTheme,thatargumentwillbetheobject,etc.Thegeneralinvestigationoftherelationshipbetweenlexicalsemantcisandsyntactic
structureis"LinkingTheory"SpecificLinkingproposalswithinGBincludeBakers(1986)UniformityofTheta
AssignmentHypothesis(UTAH),itselfareworkingofPermutterandPostal'sRelationalGrammarproposalofaUniversalAlignmentHypothesis(UAH).
TheUAHproposedthatthereisaconnectionbetweenahierarchyofthetaroles(Agent>
Patient>Goal…)andgrammaticalfunctions(Subject>Object>IndirectObject)suchthatthehighesttheta‐roleonthehierarchyofagivenverbappearedwiththehighestgrammaticalfunctionavailableintheclause;thenexthighestwiththenexthighest,etc.
TheUTAHmadeasimilarclaim,exceptthatinsteadofassumingthatGFsareprimitives
ofthetheory,Bakerproposedthattheta‐roleswereassignedtosyntacticargumentpositionssothatthethematichierarchycorrespondedwiththec‐commandhierarchyoftheargumentpositions.
1.4 ProblemswithrigidmappinghypothesesThemostrigidversionoftheproposalsrunintotwoproblems:(9) ProblemI:Sametheta‐role,sameverb,butdifferentGFs/syntacticpositions a. JosefinagaveMariathebook. b. JosefinagavethebooktoMaria c. Marcelosprayedthewallwithpaint d. Marcelosprayedpaintonthewall.(10) ProblemII:"Mirror‐Image"verbs—samesituation,samethetaroles,butinverseGFs a. Thecatchasedtherat. b. Theratfledthecat.
7
c. Theratfearsthecat. d. Thecatscarestherat. f. MarcoboughtacarfromPancho. g. PanchosoldacartoMarco.Twokindsofsolutions.Solution1:Dowty(1991)
EachargumentisevaluatedwithrespecttoasetofProto‐Agentcritera(motion,intentionality,animacy,control)andasetofProto‐Patientcriteria(affectedness,inanimacy,lackofcontrol)
ThosescoringhighonProto‐Agentcriterawillmaptosubjectposition;thosehighonProto‐Patientcriteriawillmaptoobjectcondition
Whenatsimilarpointsonthescale,flexibilityinmappingwillbeobserved,asinthesituationsin(9)and(10)above.
Solution2:Doawaywiththematicroles.Theverbisdecomposedintosub‐predicates,
andargumentsareintroducedbydifferentsub‐predicates,eachinitsownsyntacticpositionintheclause,suchasCAUSE,DO,BECOME,BE,GO,orHAVE.Theinterpretationoftheargumentdependsonwhichpredicateitcomposeswith,andsodoesitspositionintheclause.Theencyclopedicverbalsemanticsoftenendupjustfunctioningasamodifierofthebasiclexico‐syntacticstructurethatprovidesthescaffoldingfortheclause.Restrictionsonco‐occurrencewithcertainnumbersofargumentsandsoonbecomealmostepiphenomena…
Thisclass:IntroductiontoonemodernversionofSolution2.Othersomewhatdifferent
versionsavailableoutthere:Ramchand,Borer,Travis,Alexiadou,Pylkkanen,Embick,Marantz…mostlysharethesamecoresetofassumptions,however.
Theoverallapproachalsohasthetheoreticalbenefitofeliminatingawholeclassof
theoreticaldevices:Derivationallexicalmorphologicalprocesses,linkingrules,theta‐roles,theThetatCriterion,UTAH,etc,canallbeeliminatedasstatementsofthetheory—instead,theireffectsarederiveablefromthebasicassumptionsofthenewtheory.Henceit'smoreMinimalist…ournexttopic.
8
2. AMinimalistThetaTheoryReading:Harley,H.(forthcoming).AMinimalistApproachtoArgumentStructure.InC.
Boeckx,(ed.),TheOxfordHandbookofLinguisticMinimalism.Oxford:OUPPress.
2.1 DevicesinpreMinimalismθtheory
(11) AGenerativeLexicon:lexicalentriesarederivedfrom/relatedtoothersby
redundancyrules,e.g.‘passive’,whichcanhavemorphologicalandsemanticeffects a. PHON: kissed
SYN:V (noaccusativecase)SEM:[Patient] (or:[1],or[kissee])+somenotionofwhat‘kissed’means
b. Anotherone:an‘agentivenominalization’rulemightproducethis:
PHON: kisserSYN:NSEM:(indexedθ‐roleoftheV,say;either Agenti) Instrumenti
+somenotionofwhat‘kisser’means(noticethattherearetwopossibilities,bothavailableinEnglish;botharepossible‘causers’ofkissing.Exampleslike‘transmission’aresimilar,onlywithdifferentpossibilitiesfortheindexedθ‐role:Eventi,Themei,orInstrumenti.)
Wheretheactionis:
PrincipleslikeBaker’sUTAHorTenny’sAspectualMappingHypothesisorLevin&Rappoport’slinkingrulesensurethattheappropriateparticipantintheeventendsupintheappropriateplaceinthesyntactictree,accountingfortheta‐role/syntacticstructureregularities.
TheThetaCriterionensuresthatnopredicatecanendupwiththewrong
numberofarguments,andnoargumentcanendupwithoutaninterpretation.
(Some)concerns:
OnaFregeanviewofsyntacticosemanticcomposition,asnotedbyHeimandKratzer1998,theTheta‐Criterondoesn’tdomuchwork;notveryMinimalist.
9
Cross‐linguistically,thereislotsofduplicationofeffort.Englishcausativesare'syntactic';theirtranslationequivalentsinJapaneseandHiakiare'morphological',althoughwithrespecttomanykeysyntacticproperties,theEnglishandJapanese/Hiakicausativesareidentical.InMinimalistterms,it'sverypeculiartohavetwoseparategenerativemechanisms—syntaxandmorphology—eachcapableofgeneratingmuchthesamephenomena,andcompetingforcoveragebothbetweenandwithinlanguages(considerEnglishcomparativeconstructions,e.g.).
Solution:It'sallsyntax.Argumentone:Morphology/syntaxcorrelationscross‐linguistically
2.2 HaleandKeyserandthevPhypothesisAquestion:Nosyntacticprocessseemstorequirereferencetomorethan6‐8theta
roles,mostlyinAnimate/Inanimatepairs—Agent/Causer,Patient/ThemeBenefactee/Goal…
Anobservation:Crosslinguistically,themorphologicalexpressionofunergativeverbs
involvesnominals(12) Jemez:Nounssuffixedwith(incorporatedinto)averbalmorpheme
a. záae‐'a “sing”
song‐do b. hįįl‐'a "laugh" laugh‐do c. se‐'a "speak" speech‐do d. tų‐'a "whistle" whistle‐do e. shil‐'a "cry" cry‐do f. sae‐'a "work" work‐do (13) Basque:Nounsinsyntacticconstructionwithaverb'do'
a. lo egin "sleep"
sleep do b. barre egin "laugh" laugh do c. lan egin "work" work do
10
d. negar egin "cry" cry do e. eztul egin "cough" cough do f. jolas egin "play" play do g. zurrungaegin "snore" snore do(14) English:Asystematiccorrepsondencebetweenunergativeverbsandbareevent
nouns:tolaugh,alaugh;towalk,awalk;torun,arun;towork,work;toswim,aswim;todance,adance;towhistle,awhistle;tosneeze,asneeze;toscream,ascream;toshiver,ashiver…
Theirproposal:Unergativeverbsaresyntacticallycomplex,cross‐linguistically,
involvinga'light'verbheadingitsownprojection,andanominalcomplementcontributingtheencyclopedicsemanticcontent
Theobservedcross‐linguisticdifferencesresultfromdifferencesinthemorphological
realizationofthedifferentheads:incorporation(Jemez,English)vs.noincorporation(Basque,Persian);overtmorphology(Jemez)vs.noovertmorphology(English)
(15) a.UnergativeverbderivationinEnglish vP DPAgent v’ v N ∅ run
b.UnergativeverbderivationinJemez vP DPAgent v’ v N -'a záae DO laugh
11
c.UnergativeverbderivationinBasque(right‐headed) vP DPAgent v’ N v lo egin sleep DOOnlyoneAgentθ‐roleassigner,namely,vDO(Non‐light)verbsaremadeupoftwoprojections,av°andacontentfulelementwhich
incorporatesintoit.Theyproposedasimilarapproachtothepatternofinchoative‐causativealternations1
cross‐linguistically,whichalsoinvolve∅‐morphologyinEnglishbutovertmorphologyinotherlanguages,andevenperiphrasiticcounterparts(seeourdiscussionofPersianinacoupleofdays).
Inchoativeverbsareoftencomposedofanadjectivalpredicateandaverbalizing
morpheme,cross‐linguistically:(16) English Hiaki
Verb Adj Verb Adj toredden red sikisi siki tofatten fat awia awi tosoften soft bwalkote bwalko tosharpen sharp bwawite bwawi warm warm sukawe suka ….Bythesamechainofargumentasabove,wecanconcludethattheyarecomposedofav°
plusacontentfuladjectivalpredicateOneimportantdifference:InchoativeverbsdonothaveAgentarguments,butTheme
ones…Importantdifferencebetweenadjectivesandnouns:Adjectivesarenecessarily
predicative,nounsarenot
1An"inchoative"istheintransitivecounterparttoatransitivecausativeverb,whosesubjectgetsinterpretedliketheobjectofthecorrespondingcausative;theynearlyalwayspassunaccusativitydiagnostics,andhenceweassumetheirsingleargumentisbase‐generatedinobjectposition
12
Conclusion:Thev°involvedininchoativepredicatesisadifferentonefromthatinvolvedinunergativepredicates
TheargumentofaninchoativepredicateisselectedbytheAdjectivalroot,notbythev°Thev°inquestionistheequivalentof'become';itdoesn'tselectanAgent.(17) vP vBECOME AdjP =SC ‐en DP Adj Thesky redSo:Generalizationmaintained—specifierofvPisanAgentCanconcludethatsistertolexicalpredicateposition,belowvP,isPatient/ThemeDifferencebetweeninchoativesandtheircausativealternantsissimplythepresenceof
anAgentargumentSinceweknowthatacertainvarietyofvPselectsforAgents,wecaneasilyexpressthe
mechanismwhichdrivesthecausative/inchoativealternation:InsteadofcomposingwiththevBECOMEhead,theAdjPlexicalcomponentcomposeswiththeagentivev°head
(18) vP DP v'Thesun vDO AdjP =SC ‐en DP Adj thesky redTheAgentargumentisintroducedbythev°,aswithunergatives;thev°contributesthe
AgentivesemanticsSincethey'reboth∅morphemesinEnglish,wecan'tseethechange,but(aswewillsee
inJapaneseinafewdays)inmanylanguages,theinchoative/causativealternationisnecessarilyaccompaniedbyachangeinovertverbalizingmorphology.
13
WehaveseenverbswithNP(intransitive)andAdjP(transitive)downstairsphrases;Naturalextension:PPdownstairs.2Thisultimatelywillbethebasicacountofditransitivestructures,butH&Kcametoitthroughanotherclassofdenominalincorporatedverbs
HaleandKeyser'streatmentofagentiveEnglishLocation/Locatumverbs(19) a. bandage,bar,bell,blindfold,bread,butter,clothe,curtain,dress,fund,gas,
grease,harness,hook,house,ink,oil,paint,pepper,powder,saddle,salt,seed,shoe,spice,water,word.
b.Structure: Thecowboysaddledthehorse=fitthehorsewithasaddle Thecowboybutteredthebread=smearthebreadwithbutter vP DP v Thecowboyv PP =SC Agent ∅ DP P' thehorse P N Location ∅ saddle LocatumWhenP(andv°)areovert:(20) vP DP v Theteacher v PP =SC Agent ‐ate DP P' theconceptP N Locatum en‐ capsule LocationStatusof"Location"and"Locatum"rolesdependsonP°headinvolved.H&KglosstheP
involvedinsaddleas'with'(fitthehorsewithasaddle),whilethePinvolvedine.g.2SeeJaumeMateu'swork,fromBarcelona,forfurtherdiscussionofthisextension
14
boxismorelikeinorat:(puttheshoesinabox).Inbothcases,thesententialobjectisthesubjectofthelowerpredicate('withasaddle'/'inabox'),whilethelower('Goal')Nisincorporated.
So,webegintoseeapattern:(21) θ‐role PositionofDP Agent ≈ Spec‐vP Theme ≈ Spec‐SC("InnerSubject") Goal ≈ Comp‐SCAquestion:Withadjectivalsmallclauses,wecancombineeitherthenon‐Agent‐selecting
v°ortheAgent‐selectingone(givingthecausative/inchoativealternation).CanwedothiswiththePPsmallclauses?orwithincorporatedNs?
Notveryfrequently—butitispossible:(22) a. Manuelalandedtheplane ([vPManuela[v'vCAUS[PPtheplane[P'∅land]]]]) b. Theplanelanded (=[vPvBECOME[PPtheplane[P'∅land]]])Alsogive/getalternationsarelikethisaswell,withoutincorporationofthecomplement
ofP:(23) a. ManuelagaveAnnaabook. (=[vPManuela[v'vCAUS[PPAnna[P'∅abook]]]]) b. Annagotabook (=[vPvBECOME[PPAnna[P'∅abook]]])EvenwithincorporatedNs,whenthesemanticsoftheconstructionareright,wesee
occasionalcompositionwithvBECOME—this,Ithink,isthestructureofEnglishdenominalweatherverbswithexpletivesubjects—thenominalincorporatesintothev°:
(24) a. Itrained (=[vPvBECOME[Nrain]]) b. Itsnowed (=[vPvBECOME[Nsnow]]Nonetheless,itisclearthatamajorityofverbsdonotalternateproductively:*itjumped
(withexpletive'it'),*Thecitydestroyed,*Thehorsesaddled.Inthecurrentframework,thisbecomessomethingofatheoreticalconundrum.There
aretwoapproachestothisproblem:
o Associateverbrootswithlicensingconditions,sothatcertainverbrootscanonlyappearinthecontextofoneortheotherv°(Ramchand2008,Siddiqi2006areexamplesofthisapproach)
o Ascribetherestrictiontothesemantics,notthesyntax:Theproblemwiththehorsesaddledisnotungrammaticality,butuninterpretability,
15
givenwhatweknowaboutsaddling.SeeMarantz1997,HarleyandNoyer2000,Borer2005forvariationsonthisapproach.
2.3 SyntacticmotivationsforthesplitvPapproach:Larson1988Ditransitiveverbsareproblematicforabinary‐branchingsyntax:(25) VP
DP V’AgentJohn V° DP PP gave Theme Goal
abook toBillFirstinternalargumentDPc‐commandssecondone(BarssandLasnik1986)(26) a. MariashowedBillhimself(inthemirror).
a'. *MariashowedhimselfBill.b. MariashowedBilltohimself(inthemirror).b'. *MariashowedhimselftoBill.
Larson1988:SolutionistosplitthevPintotwoprojections,anupper,semantically
emptyAgent‐assigningoneandalowercontentfuloneselectingtheinternalargumentsoftheverb
(27) VP1
DP V’AgentMaria V1 VP2 ∅ DP V’ Theme Bill V2 PP
Goal show tohimselfIfVP1isreallyvP,andVP2isareallyavarietyofPPheadedbyastativeroot√show,this
structureisisomorphictoHaleandKeyser's.
2.4 AdvantagesofthesplitvPapproachwithinMinimalism:DistinguishingunaccusativeandunergativeVsinBarePhraseStructure
16
InMinimalism,X‐bartheoryhasbeeneliminated(Chomsky1995,Ch.4)Instead,wehavethesimpleoperationMerge,whichsimplytakestwoelementsand
combinesthemtoproduceevermorecomplexbinarybranchingstructures:(28) <eat> <the> <eat> <the><boy><eat> <it>
“Theboyateit.”(vPportiononly—noTPrepresented)Crucially,therearenononbranchingprojectionspossibleinthissystem,sinceevery
projectionistheresultofaMergeoperation,involvingatleasttwoelements!However,withtheVP‐internalsubjecthypothesis,GBtheoryreliedcruciallyon
nonbranchingprojectionstomakethespecifier/complementdistinctionthatwascentraltodistinguishingunergativeandunaccusativeverbs.
(29) BeforetheadventofthevPhypothesis:
UnergativeverbsinGBtheory UnaccusativeverbsinGBtheory
VP VP
DP V’ V’ V° V° DP The man laugh arrive the man WithinBPS,thesimplemergerofaverbanditsargumentproducesidenticalstructures,
nomatterwhatlinearorderyouassume:(30) Unergative Unaccusative a. VP<dance> b. VP<fall> DP V<dance> V<fall> DP Jose dance fall JoseUndertheH&Kproposal,thedifferencesbetweentheverbtypesareconsiderablymore
profound;unergativeverbsgeneratetheirexternalargumentinspec‐vP,whileunaccusativeargumentsarebase‐generatedinsistertotheV/Adjprojection,belowtheunaccusativevP.
17
(31) a. vPDO vPBECOME DP vDO' vBECOME VP<fall> Jose vDO N<dance> V<fall> DP dance fall JoseThisprovidedChomskywithasolutionbeforeheevenneededitforhisBPSproposal's
problemwiththeunaccusative/unergativedistinction.
2.5 WrapupSo:Wehaveseenaproposalfordoingwithouttheta‐roles;thestructuralpositionsin
whichDPsaremergedandthecompositionalsemanticsoftheprimitivelexical‐semanticpredicatesassociatedwiththemdetermineDPinterpretation;similarinterpretationsareseeninsimilarstructuralpositions.
Next:Thesemanticmotivationfordecomposingverbs.First:DealingwithFodor's
(1970)argumentsagainstdecomposition,thenKratzer'streatmentofMarantz'sidiomgeneralization.
18
3. Fodor(1970):Threereasonsfornotderiving'kill'from'causetodie'Harley,H.(forthcoming).Lexicaldeompositioninmoderngenerativegrammar.In
HandbookofCompositionality,editedbyWolframHinzen,MarkusWerningandEdouardMachery.Oxford:OUP.
3.1 Fodor'sbasiccontentionArguingagainsttheLexicalSemanticsproposalsofthe1970s,Fodorcameupwiththree
waysinwhichthemonomorphemicchange‐of‐stateverbkilldifferedfromitsmulticlausalparaphrasecausetodie
(32) Paraphraseallowsindependenttemporalmodification(a);verbdoesnot(b) a.JohncausedtheautomaticdoortoopenonSundaybyprogrammingitonSaturday b.#JohnopenedtheautomaticdooronSundaybyprogrammingitonSaturday.(33) ParaphrasehastwopossiblecontrollersforadjunctPRO(b);verbdoesnot(c) a. Billidied[byPROiswallowinghistongue]. b. JohnicausedBilljtodie[byPRO#i/jswallowinghistongue] c. JohnikilledBillj[byPRO#i/*jswallowinghistongue](34) ParaphraseallowstwoantecedentsforVPellipsis(a),verbonlyone(b) a. JohncausedMarytodie,anditsurprisedmethatshedidso[=DIED]. b. *JohnkilledMary,anditsurprisedmethatshedidso[=DIED].Sincethesyntacticpropertiesoftheparaphrasearesodifferentfromthepropertiesof
thesimpleverb,Fodorconcludedthatthesimpleverbshouldnotbederivedby'collapsing'(incorporating)abiclausalstructure—'kill'shouldnotbedecomposedintoCAUSEtoDIE.
3.2 Review:Whydecomposeatall?Wesawpreviouslythattherearemorphosyntacticargumentsfordecomposingtheverb
phraseintoavPandacontentfulpredicativephrase.Therearealsosemanticreasons;theseweretheoriginalgenerativesemanticarguments
infavoroftheproposalConsidertheambiguitypresentinastringsuchasConsuelamadeMarcelohappyagain.
It'sasyntax101exercisetoshowhowthisambiguityisstructural:
19
(35) a. VP b. VP DP V' DP V' V° SC V' Adv° DP AdjP V° SC Adj' Adv° DP AdjP Adj° Adj° ConsuelamadeMarcelohappy againConsuelamadeMarcelo happy againThescopeofmodificationofagaincorrespondsexactlytoitsattachmentsite;because
thetwoattachmentsitesdonotinvolveachangeinlinearorder,thestringisstructurallyambiguous.
Keything:Thesameambiguityispresentwithmonomorphemicchange‐of‐stateverbs
likeopen(36) a. vP b. vP DP v' DP v' v° SC v' Adv DP AdjP v° SC Adj Adv DP Adj° Consuela CAUS thedooropen againConsuelaCAUSthedooropen againVonStechow(1995)arguedthatitwouldbetheheightoftheoreticalprofligacyto
assumethattheambiguityin(35)issyntacticwhilethatin(36)islexical;thetype‐shiftingoperationsneededtoderivetwoscopesforagaininthepurelysemanticdomainwouldascribeanexceptionaldegreeofpowertothesemanticcomponentwhenthenecessarytoolsarealreadybepresentinthesyntax.
SimilarscopalambiguitiesareevidentwithrespecttoothertemporalmodifierssuchasdurativePPslikeforfiveminutes:
20
(37) a. vP DP v' vCAUSE VP DP V' V° AdvP John ∅ thedoor open forfiveminutes b. vP DP v' v' AdvP vCAUSE VP DP V° John ∅ thedoor open forfiveminutes Also:CompositionargumentofKratzer'swhichwe'reabouttoconsidernext.
3.3 Therightwaytoanalyze'kill'as'causetodie'KeytounderstandingtheresponsetoFodor'sobjections:
o Fodor'scounterexamplesinvolveanembeddedinfinitiveTP,completewithto‐marker,itsowntemporallocation,andanembeddedeventiveVP
o Theproposedstructureforchange‐of‐stateverbsinvolveanembeddedstativesmallclause,notanembeddedTP
o AmoreappropriateperiphrasticcomparisontothemodernproposalfordecompositionintoCAUSEandstativesmallclausewouldbeaperiphrasiticcausativeheadedbymakeandtakingastativesmallclausecomplement,likeMarymade[Johnsick].
o Infact,whentestedwithFodor'sowntests,MarymadeJohnsickbehaveslikethemonomorphemicverbs,notlikeMarycausedJohntogetsick
21
o Conclusion:Theparallelbetweentheappropriateperiphrasticconstructionandtheproposedmoderndecompositionisactuallyveryclose
o Fodor'soriginalcomparisonpointsarenotrelevanttothemodernproposal.
First,akeydistinction:eventivevs.stativeEventivepredicatesare[+dynamic]—somethinghappensStativepredicatesare[‐dynanmic]—justanexpressionofthewaythingsare.InEnglish,onegoodtestforstativevs.eventivestatusofagivenpredicateistheir
behaviorinthetruepresenttense,asadescriptionofthewaythingsarerightatthemomentofspeech
Stativesaregrammaticalinthepresenttense;eventivepredicatesrequirethe
progressive.Onegoodwaytoavoidtheconfoundofahabitualinterpretationistoprefacetheexamplewithsomethinglike"Look!"
(38) a. Look!Mariaisswimming!/*swims! eventive b. Look!Mariaissick!/*isbeingsick! stative c. Look!MariaknowsFrench!/*isknowingFrench! stative d. Look!MariatalkingtoJohn!/*talkstoJohn! eventiveEventivepredicates,esp.ifagentive,aregenerallygoodwithimperativemood;Stative
predicatesarenot:(39) a. Swim! b. #KnowFrench! c. LearnFrench! d. #Besick! e. Getsick! f. Die! g. #Bedead!SowhenItalkabouteventivevs.stativepredicatesbelowthesmallclause,that'swhatI
mean.Fodor'sexamplescauseXtodieinvolveaTPwithaneventivepredicatedie.MyparaphrasesmakeXsickinvolveasmallclausewithastativepredicatesick–butthey'restillindubitablysyntactic.
SoifthemonomorphemicchangeofstateverbsbehavelikemakeXsickwithrespectto
Fodor'stests,wecanconcludethatthereisnoreasonnottodecomposethem,aslongastheproposalinvolvesastructuralparalleltomakeXsick—whichitdoes,ofcourse.
22
3.3.1 ControltestSubjectsofstativepredicatescannotcontrolPROinanadjoinedbyphrase(40) a. *MariaiwashappybyPROisinging. b. *MarcelowassickbyPROieatingtoomuch.Comparethesewiththeireventiveinchoativecounterparts:(41) a. MariaibecamehappybyPROisinging. b. MarcelogotsickbyPROieatingtoomuch.Whenmaketakesaneventiveverbalcomplement,theembeddedsubjectcancontrol
PRO,asweexpectfrom(41)(thematrixsubjectcanalsocontrolPRO,whichisexpectedgiventhatmakeiseventive):
(42) JosejmadeMariaidiebyPROi/jswallowingheritongueHowever,whenthecomplementtomakeisstative,theembeddedsubjectcannotcontrol
PRO,onlythesubjectofmakecan:(43) a. JosejmadeMariaisickbyPROj/*ieatingtoomuch b. JosejgotMariaiawakebyPROj/*isettingthealarmclock.Consequently,thefactthatasinglechange‐of‐stateverbbehaveslike(43)isexpectedif
thecomplementtoCAUSEisastativesmalclause,ratherthananeventiveverblikedie:
(44) a. JoseisickenedMariajbyPROi/*jeatingtoomuch b. JoseiwokeMariajbyPROi/*jsettingthealarmclock.Theparallelbetween(43)and(44)suggestthata[CAUSE[Mariasick]SC]v'structurefor
theverbsin(44)isnottoofaroffhere.
3.3.2 VPellipsistestSameremarksapplyhere:Itturnsoutthatstativepredicatesingeneralcannotbe
antecedentsforVP‐ellipsisinEnglish,whilealleventivepredicatescan.(45) a. Mariawassick,*andConsueladidsotoo. b. Pepewasawake,*andMarcelodidsotoo.(46) a. Mariagotsick,andConsueladidsotoo. b. Pepewokeup,andMarcelodidsotoo.
23
SothisagainpredictsacontrastbetweenamakeXgetsick(orcausetodie)periphrasisandamakesickperiphrasis,withmakesickbehavingthesamewayasthecausativeverbsicken.
(47) a. MariamadeConsuelasick,andAnnadidsotoo.(doso=makeXsick) b. Pepegotsomeoneawake,andJosedidsotoo.(doso=getXawake)(48) a. MariasickenedConsuela,andAnnadidsotoo.(doso=sickenX) b. Pepewokesomeoneup,andJosedidsotoo.(doso=wakeXup)(49) a. MariamadeConsuelagetsick,andAnnadidsotoo.(doso=makeorgetsick) b. Pepemadesomeonewakeup,andJosedidsotoo.(doso=makeorwakeup).Sameconclusion:causativeverbsbehavethesameasperiphrasticequivalentswith
stativeembeddedclauseswithrespecttodosoellipsis,withtheresultthatthefailureofparallelismwitheventiveembeddedclausesisnotanargumentagainstdecomposingthecausativeverbintoacausativepredicateandastativeresult.
3.3.3 IndependenttemporallocationsFinally,whatabouttheindependenttemporallocationtest?Judgmentscanbeabitfuzzy,butingeneral,theintuitionisthatwithasingleverb,there
isonlyoneevent,andwhichcanonlyoccupyonetemporallocation.Withtwoverbs,therearetwoevents,whichcanoccupytwotemporallocationswithout
contradiction.Withacausativeverbandastativecomplement,weagaingettheeffectofonlyone
event—the'becoming'partiscocercedbyworldknowledge,notpresentinthesyntacticstructure
(50) a. ConsuelacausedMariatogetsickonTuesdaybypoisoningheronSunday. b. ?ConsuelamadeMariasickonTuesdaybypoisoningheronSunday. c. ?ConsuelasickenedMariaonTuesdaybypoisoningheronSunday.Infact,theeffectismuchstrongerwithkill,whichisperhapsrelatedtothefactthatitis
adenominalevent‐namingverb(akill),ratherthanadeadjectivalchange‐of‐statepredicate,properlyspeaking.
Inanycase,Ithinkthere'sacontrastbetween(50a)and(50b,c)thatsupportsthe
directionofargumentationproposedhere.
24
3.4 Casestudiesindecomposition:have,give,get,wantWe'veseeninchoativevBECOMEandcausativevCAUSE,bothofwhichformeventiveverbs.WealsohavestativevBEinstativeverbsKayne,Freezeonhave:composedofbeplusanincorporatedpreposition(orinKayne's
case,possessivedeterminer):(51) vP vBE PP DP P' PHAVE DP have John abookNote:behasnoexternalargumentinthisformulation—there'snoAgentinthe
structure.InanapproachlikeBaker'sorBower's,bewouldbeaPredheadanditssubjectwouldbeinitsspecifier,notinasmallclausecomplement.
Ifweswapv°relations,changingvBEtovCAUSE,wehavethenucleusofadouble‐object
giveverb:(52) vP DP v' vCAUSE° SC DP PP PHAVE DP Mary give John ∅ abookIfgiveandhavebothcontainthesamesubconstituent,asmallclauseheadedbyPHAVE,
theyshouldbehavesimilarlyinanumberofways,andtheydo:AlienabilityrestrictiononthesubjectofhaveandtheGoalargumentofdouble‐object
give:(53) a. Johnhasabook. (alienablepossession,animatesubject)
25
b. Johnhasabignose. (inalienablepossession,animatesubject) c. *Thebookcasehasabook. (alienablepossession,inanimatesubject) d. Thebookcasehasfiveshelves (inalienablepossession,inanimatesubject)
(54) a. *Marygavethebookcaseabook. b. Marygavethebookcasefiveshelves(e.g.whilebuildingorrepairingit)Ifhaveispresentindoubleobjectgive,itshouldhaveapossessionentailment,andit
seemsto:(55) a. MarytaughtthestudentsFrench. (ThestudentsknowsomeFrench) b. MarytaughtFrenchtothestudents. (Thestudentsmayormaynot)Whataboutthefailureofpossessioninthesedoubleobjectstructures?(56) a. MarysentJohnaletter,butheneverreceivedit.(#Johnhadaletter) b. MarybakedJohnacake,butitburned,soheneversawit.(#Johnhadacake).Possessionwaspresent—itwasJohn'sletterandJohn'scake,thoughhenever'had'
them—vBEaddsdurativityrequirementnotimposedbythesimplepredicationofapossessionrelationintroducedbyPHAVE
NormaladverbialadjunctiontestsrevealpresenceofPHAVEsmallclauseindoubleobject
give:(57) a. MarygaveJohnthecaragain Restitutive:Johnhadhadthecarbefore,andMarycausedhimtohaveitagain. Repetitive:MaryhadgivenJohnthecarbefore,andshedidsoagain b. MarygaveJohnthecarforaweek. =>JohnhadthecarforaweekCertainidiomscarryoverfromdoubleobjectgivetohavesentencestoo(Richards
2001):(58) a. Marygavemethecreeps/willies. b. (It'ssospookyinhere,)Ihavethecreeps/thewillies. c. #Ipossess/own/amexperiencing/feelthecreeps/thewillies.CanalsocomposePHAVEwiththeinchoativev°,givingget:
26
(59) vP vBECOME PP DP P' PHAVE DP get John abookAlltheeffectsabovecarryover—animacyeffects,temporalmodificationofresulthavingstate,idioms:(60) a. Animacy/alienabilityeffects: (i) Johngotthebook. (ii) #Philadelphiagotthebook.3 (iii) Philadelphiagotanewfreeway. b. ScopeofadverbialsonlyoverPHAVEstate,notovergettingevent: (i) Johngothisbalanceagain
Restitutivereading:(i)canbetrueinacasewhereJohnneverlosthisbalancebeforeinhislife,sohehadnevergottenhisbalancebefore,sincehe'dneverlostit.Insuchasituation,(i)expressesthenotionthatwhatishappeningagainisthestateofJohnhavinghisbalance,notthatheisundergoingasecondeventofgettinghisbalance.
(ii) Johngotthecarforaweek Lowscopereading:WhatlastsaweekisthestateofJohnhavingthecar,
nottheeventofhimgettingit. c. Igotthecreeps/willies
3.5 Intermediatesummary:WehaveseenthreekindsofargumentsfordecompositionoftheVPintovPandalowercontentfulpredicativephrase:
Morphological(Basque,Jemez) Syntactic(Larsononditransitiveverbs) Semantic(scopeofagainandindependentmodificaitonof
result)
3TheintendedcontextisoneinwhichthebookwassentinthemailtoPhiladelphia,wherePhiladelphiajustreferstothecityasalocation.Famously,thisimprovesif'Philadelphia'isunderstoodtorefertosomeanimate/intentionalcollectiveentity,likethePhiladelphiaofficeofacorporation;thesameeffectisseenin#JohnsentPhiladelphiathebookvs.JohnsentthebooktoPhiladelphia.Sincefreewaysareinalienablesubconstituentsofacity,(iii)isfine.
27
Themorphologicalrelatednessofcausativeandinchoativepairsistoopervasive,cross‐
linguistically,tobeacoincidence.HerearemoreexamplesoftheJemez/HiakitypefromJapanesechange‐of‐stateverbs:(61) Japaneseinchoative/causativealternatingmorphology Inchoativevariant Causativevariant ag‐ar‐u 'rise' ag‐e‐ru ‘raise' ag‐ aratam‐ar‐u 'improve' aratam‐e‐ru ‘improve’ aratam‐ ama‐r‐u 'remain' ama‐s‐u ‘remain ama‐ hita‐r‐u 'soak' hita‐s‐u ‘soak’ hita ‐ arawa‐re‐ru 'show' arawa‐s‐u ‘show' arawa‐ hana‐re‐ru 'separate' hana‐s‐u ‘separatefrom' hana‐ ka‐ri‐ru 'borrow' ka‐s‐u ‘lend’ ka‐ ta‐ri‐ru 'suffice' ta‐s‐u ‘supplement’ ta‐ bak‐e‐ru 'turninto' bak‐as‐u ‘turninto/bewitch’bak ‐ bar‐e‐ru 'cometolight' bar‐as‐u 'bringtolight’ bar‐ ak‐i‐ru 'tire' ak‐as‐u ‘tire’ ak dek‐i‐ru 'comeintobeing dek‐as‐u 'bringintobeing' dek‐ horob‐i‐ru 'falltoruin' horob‐os‐u ‘ruin’ horob‐ ok‐i‐ru 'getup' ok‐os‐u ‘getup’ ok‐Itisimportant,however,tokeepinmindthepointfromBasque,namely,thatwecansee
thesekindofdual‐projectioneffectsinsyntacticallyindependentwords,notjustinmorphologicallyfusedcaseslikeJemezorinmonomorphemiccaseslikeEnglish.
Hereareexamplesofinchoative/causativepairsfromPersian,which,likeBasque,
realizesthev°headandthedownstairspredicatesasindependentwords,inavisiblysyntacticallycomplexway:
(62) a. sabokshodan sabokkardan lightbecoming lightmaking 'degrade(intr)' 'degrade(tr)' b. pahnshodan pahnkardan widebecoming widemaking 'spread(intr)' 'spread(tr) c. kotakxordan zadan kotak beatingcolliding beatinghitting 'togetbeaten' 'tobeat' d. xarshodan xarkardan donkeybecoming donkeydoing 'togetfooled' 'tofool'
28
TheadvantageofthevPapproachisthatitallowsustotreatthePersianandBasque
casesinauniformwaywiththeEnglish,Jemez,Japanese,andHiakicasesTheywouldotherwisebeapeculiarexceptiontothepatternofmorphologicalverb
formationcross‐linguisticallyAnalyzedthisway,however,theyaresimplyapredictedparametrictype,drivenby
incorporation/non‐incorporationofthelowerelementintothehigherv°.Beforemovingontootherquestionsaboutthisdecompositionalscenario,let'sjust
consideronemoreargumentforit:Kratzer'streatmentofMarantz'sgeneralizationconcerningidiomaticinterpretationsofverbs.
29
4. KratzeronSeveringtheExternalArgumentfromitsVerbReading:Kratzer,Angelika.1996.Severingtheexternalargumentfromitsverb.InJ.
RooryckandL.Zaringeds.,PhraseStructureandtheLexicon,109‐137.Dordrecht:Kluwer.
4.1 Aquickprimerincompositionalsemantics:(63) Marysleeps. S DP V Mary sleeps What'stheessentialmeaningofthetwoelementsofthesentence? (64) a. "Mary"=thepersonnamedMary
b. "sleeps"=afunctionwhichtakesanindividualasanargumentandreturnsthevalue"true"iftheindividualissleeping.
Individualsareentitiesintherealworld:wecancallthemthingsoftype<e>.Sentencesarestatementsaboutsituationsintherealworld,whichareeithertrueorfalse:theyareoftype<t>.Verbsarefunctions.Intransitiveverbstakeanentityasanargumentandreturnatruthvaluewhichdependsonwhatconditionsarelikeintherealworld.Theyareoftype<e,t>,wheretheleftmostmemberofthepairistheinputtothefunction,andtherightmostmemberistheoutput.Wecannowlabelournodes,above,withtheirtypes:(65) S<t>DP<e> V<e,t> Mary sleeps Whatabouttransitiveverbs?
30
(66) S<t> DP<e> VP<e,t> Mary V<?> DP<e> likes Sue Well,lookingatthenodesinthetree,wecanseethatatransitiveverbhastobesomethingthattakesanentityasinputandgivesbackanotherfunction,equivalenttoanintransitiveverb,thattakesanentityasinputandgivesatruthvalue.So,atransitiveverbisoftype<e,<e,t>>.Thelexicalentryforanintransitiveverbwillbesomethinglikethis:(67) sleep(x)=trueiffxisasleepThelexicalentryforatransitiveverbwillbesomethinglikethis:(68) like(y)(x)=trueiffxlikesy.Then,thewayyouinterpretasentenceisyoujustputtheargumentstogetherwiththeverb,oneatatime,andseewhatthetruthvalueisattheend.So,tointerpret"MarylikesSue":(69) a. put"Sue"togetherwith"likes" b. getafunctionthattakeanentityxandgives"true"iffxlikesSue. c. Put"Mary"togetherwiththatfunction("likesSue")andget"true"iff MarylikesSue.
4.2 Davidsonianeventsemantics:(70) “WeboughtyourslippersinMarrakesh ∃e[buy(yourslippers)(we)(e)&inMarrakesh(e)] =”TherewasaneventofusbuyingyourslippersandthateventwasinMarrakesh”Syntacticarguments(subjectsandobjects)aredirectargumentsoftheverb—compose
directlywiththeverb.Adjunctsarenot.(71) Neo‐Davidsonian “WeboughtyourslippersinMarrakesh ∃e[buy(e)&Agent(we)(e)&Theme(yourslippers)(e)&inMarrakesh(e)] =“TherewasaneventofbuyingandtheagentoftheeventwasusandthepatientoftheeventwasyourslippersandtheeventwasinMarrakesh.
31
Nothing(excepttheeventargument)isanargumentoftheverb.Eachargumentisintroducedbyitsownseparatepredicate.
EitherapproachcouldbecapturedinatheoryofLCSwitharuleabout‘ordered
argumentassociation’inthesyntax:(72) a. BUY PHON:'buy' SYN:V SEM:λxλyλe[buy(x)(y)(e)] b. BUY PHON:'buy' SYN:V SEM: λxλyλe[buy(e)&Agent(y)(e)&Patient(x)(e)]eitherlexicalentrygivesthesameresult buyisa3‐placefunction(λxλyλe) buycombinesfirstwithitsPatient(λx),thenitsAgent(λy),thenitseventarg(λe)Kratzer:goingtoarguethatsomeofthisdecompositionactuallyhappensinthesyntax.infact,goingtoarguethatbuy(andalltransitiveverbswithexternalarguments)arenot
three‐placepredicates(Agent,Patient&Event),buttwo‐placepreds(PatientandEvent)
Agentsareaddedbyaseparatepredicate,withitsownlexicalentry,whichprojectsits
ownphraseinthesyntax(avP!)
4.3 Externalargumentsarespecial(73) Williams:buy(1,2) Rappaport&Levin:buy(<Agent>,Theme) Grimshaw: Thematic Aspectual (Agent,Theme) & (Initiator,Delimiter) ExternalArgument Marantz:buy(Theme)Kratzer'squestion:howdoyouimplementthisintuitioninacompositionalsemantics?Ifverbs&theirargumentscombinebyfunctionapplication,andifthelexicalentryfor
buylookedlikethis:
32
(74) buy λxλe[buy(x)(e)] thenyou’dhaveapropositionassoonasyoucombinedbuywithitspatient anditseventargument.Howdoestheagentgetinthere?couldjustadditbybruteforce,withspecialsemanticinterpretationruleforVPsick!But:AcluecomesfromMarantz's'idioms':manycasewhereyougetalotofmeaning
variationdependingonthetypeofobjectaverbtakes(75) killabug = causethebugtodie killaconversation = causetheconversationtoend killanevening = whileawaythetimespanoftheevening killabottle = emptythebottle killanaudience = entertaintheaudiencetoanextremedegree throwabaseball throwsupportbehindacandidate throwaboxingmatch throwaparty throwafit takeabookfromtheshelf takeabustoNewYork takeanaspirin takeanap takealetterinshorthand(76) Important!thesearen'texactly‘idioms’—they’renotfixed: killthebottle/thepeanuts/thecasserole/thewine killanhour/afewminutes/timeEvenmoreimportant:youseethiskindofvariationconditionedbyobjects‐‐not
subjects!Bresnan'sandGrimshaw'ssuggestion:theexternalargumentisstill anargumentoftheverb,it'sjustaspecialargument,inthat itcombineslast.Soyoucanhavespecialmeaningwiththeverb andtheobjectwithoutthesubject,butnotviceversa. (Theirprediction:noidiomsofe.g.verbs+adjunctsexcludingtheobject.) Kratzer'sargumentthatB&G'sreplyisinadequate:
33
a)Semanticinterpretationofanoderesultsfromcombiningthetwo daughternodes. b)Verbsarefunctions.Traditionally,hit,e.g.,isatwo‐placefunction: hit(x)(y),inthesystemwesawabove:(77) VP =hit(Agent)(Patient) 3 Agent V' =λx[hit(x)(Patient)] 3 λyλx[hit(x)(y)]Patient "Hit"isafunctionthattakesanargumentxandturnsout afunctionthattakesanotherargumentyandturnsout truthvalue=TRUEiffyhitsx. sotruthvalueoftreeabove=1iffAgenthitsPatientHowdoMarantz'sspecialinterpretationsworkinasystemlikethis?Coulddoitlikethis,positingazillionhomophonousverbskillwithdifferenttruthconditions:(78) kill1isafunctionthattakesanargumentxandturnsout afunctionthattakesanotherargumentyandturnsout: truthvalue=TRUEiffxisananimatebeingandykillsx. kill2isafunctionthattakesanargumentxandturnsout afunctionthattakesanotherargumentyandturnsout: truthvalue=TRUEiffxiscomestibleandyconsumesthelastofx. kill3isafunctionthattakesanargumentxandturnsout afunctionthattakesanotherargumentyandturnsout: truthvalue=TRUEiffxisatimeperiodandywastesx.Orlikethis,withoneverbkillwithseveralif‐thenstatementsabouttruthconditions (79) killisafunctionthattakesanargumentxandturnsout afunctionthattakesanotherargumentyandturnsout: truthvalue=TRUEiffxisananimatebeingandykillsx. truthvalue=TRUEiffxiscomestibleandyconsumesthelastofx. truthvalue=TRUEiffxisatimeperiodandywastesx.Butwhat'stopreventyoufromdoingthesametrickwiththe"y"argument?Neither
approachpredictsthatitshouldbeimpossible:(80) blickisafunctionthattakesanargumentxandturnsout afunctionthattakesanotherargumentyandturnsout:
34
truthvalue=TRUEiffyisananimatebeingandyblicksx. truthvalue=TRUEiffyisatimeperiodandxexistsduringy. truthvalue=TRUEiffyisafooditemandxismadesickbyy.so"XblickedY"haswhatevermeaning'blick'hasinthecontextofX "Todayblickedthemayfly"sayssomethinglike"Themayflyexistedtoday." "ThesausageblickedMary"sayssomethinglike"ThesausagemadeMarysick"Thisispossibleintheusualsystem,butifMarantz'sgeneralizationiscorrect,simply
doesnotoccur!Howtoruleitout?KratzersaysthattheonlywayshecanseetocaptureMarantz'sgeneralization isifexternalargumentsarenotargumentsoftheirverbsafterall,but argumentsofsomeotherverb—alightverb—thatselectsthem,and thencombineswiththemainverbbycoordinationtogivethewhole meaning:
4.4 Externalargumentsareargumentsofaseparatehead,Voice“Supposequitegenerallythatargumentsareintroducedbyheads”Aha!Hung(1988)reportsthatMalagasyhasexactlysuchahead,representedbyvisiblemorphology(81) Morphologicalevidence:Malagasy'active'prefixan M+an+sasa nylamba (aminnysavony) Rasoa T+v+wash theclothes withthesoap Rasoa "Rasoawashestheclotheswiththesoap."Claim:Allverbswithexternalargumentshaveaseparatelittle"v"(Kratzer's"Voice")
thatselectstheexternalargument:(82) vP 3 Ext.Arg. v' 3 v VP 3 (Int.Arg.#2) V' 3 V Int.Arg#1Thenlongwaffleaboutwhethertheexternal‐argument‐introducingheadislexicalorfunctional
35
(thoughIconcurthatsplittingthevP“allowsustoharvestmanyofthepleasantsyntacticconsequencesof[previous]proposals”alongtheselines)HowdothedenotationsofVP&vPgetcombined?“EventIdentification”(83) vP = λe[Agent(Mittie)(e)&feed(thedog)(e)]byF.A. 3 DP v' =λxλe[Agent(y)(e)&feed(thedog)(e)]byE.I. Mittie v VP = λe[feed(thedog)(e)]byFA 3 Ø V DP λyλe[Agent(y)(e)] feed thedog λxλe[feed(x)(e)] “EventIdentificationisoneofseveraladmissibleconjunctionoperations”withthestipulationthattheeventsthatarebeingidentifiedhavetobecompatible.(Thenconfusingexcursusabouthowtoaddanexternalargumenttoastativeverb)Wheredoeseventargumentcomefromtosatisfytheopenargumentslot?Itdoesn’t;it
getsexistentiallyquantified(bound)byanappropriatequantificationalfunctionalheadhigherup(e.g.Tense)
Backtohowthishelpsuswithvariableinterpretationverbs:JohnkilledBill:
There'sa"causing"anda"killing";Johnistheagentofthecausing,Billisthepatientofthekilling,andthecausingandthekillingwerethesameevent‐‐soJohncausedthekillingofBill.
Johnkilledthewine
There'sacausingandakilling;Johnistheagentofthecausing,thewineisthepatientofthekilling,andwhenkill'spatientiscomestible,killmeans'finish',andthecausingandthekillingarethesameevent‐‐soJohncausedthefinishingofthewine.
But—sinceJohnistheargumentoftheCausingevent,notofthekillingevent,nospecial
truthconditionsspecifiedforthekillingeventcantakehisidentityintoaccount!Hence,nospecialmeaningsforverbsinthecontextoftheirexternalarguments.Shefinisheswithanexcursusaboutaccusativecase,Burzio’sgeneralization,andof‐ing
vs.acc‐inc&poss‐inggerunds—oneofthefirststatementsofthe"high/lowattachmenthypothesis",ofwhichmoreanon.
(84) a)Mary'sreadingofPrideandPrejudice b)MaryreadingPrideandPrejudice
36
5. ApplicationsI:SpecialinterpretationsandtheHigh/LowattachmenthypothesisReadings:Harley,H.(2008)Onthecausativeconstruction,HandbookofJapanese
Linguistics,editedbyShigeruMiyagawaandMamoruSaito.pp.20‐53,Oxford:OUP
5.1 Affixalcausativesandarchitecturesforlinguistictheories
5.1.1 TheEmpiricalBase:ThreekindsofV‐sasecombination:(85) (Asubsetof)Lexicalcausatives Miyagawa1980,1984 Taroo‐ga zisyoku‐o niow‐ase‐ta Jacobsen1981,1992 Taro‐N resignation‐Asmell‐ase‐PST Matsumoto2000 “Tarohintedatresigation.”(Lit:‘Taromaderesignationsmell.’)Specialproperties: monoclausalbyalltests(seebelow) canhaveidiomaticinterpretations exhibitallomorphywithotherlexicalcausativeaffixes strongspeakersenseof‘listedness’,non‐productivity mayfeed(non‐productive)nominalization(86) Productivecausatives
a. Make‐causatives Kuroda1965,Kuno1973 Hanako‐wa Yoshi‐o ik‐ase‐ta Hanako‐T Yoshi‐A go‐ase‐past “HanakomadeYoshigo.” b. Letcausatives Hanako‐wa Yoshi‐ni ik‐ase‐ta Hanako‐T Yoshi‐D go‐ase‐past “HanakoallowedYoshitogo/HanakohadYoshigo.”Specialproperties: Biclausalbytestsinvolvingscope,adverbialcontrol,binding,disjunction Monoclausalbytestsinvolvingnegativepolarity Make‐causativemonoclausalbytestsinvolvingcase. Causeemustbeanimate/Agentive ProductiveInmake‐causatives,theCaseontheCauseealternatesbetweenaccusativeanddative
dependingontransitivityofembeddedverb.Whenitisdativeni,itisCaseni,notPni(Sadakane&Koizumi1995).Inlet‐causatives,itseemstobeP‐ni
37
Propertiesofallsasecausatives (manyfromManning,Sag&Iida1999)(87) a. V+sase=phonologicalwordforstress,otherword‐sizeprocesses
Kitagawa1986 b. V+sasesubjecttophonologicalallomorphydependingoncodaofV (ifit’savowel,thensase,ifit’saconsonant,thenase) c. V+sasemayfeed(productive)nominalizationwithkata,‘wayof’ d. sasebyitselfmaynotbehaveasalexicalverb(stem):4 i. maynotreduplicatedbyitselftoexpressrepetition ii. maynotbearfocusintonationbyitself iii. maynotbeinflectedforsubjecthonorificationbyitself. v. maynotstandaloneasananswertoayes‐noquestionAninterestingacquisitiondifferencebetweenlexicalandsyntacticsase:lexicalsase
occursfirst—butnotasearlyaslexicalcausativeusesofverbsshowup(Murasugi2003)
Forausefulsummaryofmostoftheseproperties,seeKitagawa1994,Manning,Sagand
Iida1999.Forsurveysofmanypreviousanalyses,seeCipollone2001,Kuroda2002.Forusefuldiscussionofthe‘make/let’distinction,seeDubinsky1994andcitationstherein.
5.1.2 TheoreticalapproachesThisconstellationofpropertiesreallymakeonefaceone’stheoreticalpriorities.Some
architecturalissuesposedjustbytheproductive‘make’sases:(88) a. Syntacticallymonoclausalintermsofcase,tense,andnegativepolarity
licensing. b. Syntacticallybiclausalintermsofbinding,scope,disjunction,control c. Morphologicallyandphonologicallyasingleword,intermsofaffixation
possibilitiesandprosody.Resolvingtheseissuesusuallyinvolvesradicalreplumbingofgrammaticalarchitectures:
theinfluenceofJapanesecausativesonlinguistictheorycouldn’tbebigger.
4Kuroda1981,1990(ascitedinKuroda2002)presentssomeexamplesfromnegationandinterveningparticlestosuggestthatsasedoeshaveanindependentexistenceasaverbalmorpheme;Miyagawa1989:115f,andKitagawa1994:184f.,followedbyManning,SagandIida1999:47,arguesthatinfacttheseareexamplesoftheaseallomorphsuffixedtolightverbs,‘do’.Kuroda(2002n.14)disagrees,ascribingKitagawa’spositiontogrammaticalityjudgmentdifferences.
38
Thelexicalist’spriorities:(89) a. Becauseallterminalnodesinthesyntaxmustcorrespondto
morphophonologicalwords,thencausativesmustbemonoclausal: oneverb,oneclause.
b. Multiclausalpropertiesofcausativesmustarisefromthe(productive) operationaffixingthecausativemorphemeinthelexicon,producinga complexsyntacticandsemanticword.
c. Conclusion:binding,scope,adjunctionandadverbialinterpretationand controlarerelationsthatdependonlexicaloperations,notsyntactic structure.
(ActualproposalofManning,SagandIida1999;thereplumbingpartistheinclusionof
adjunctionandquantifierscopeas(separate)lexicaloperations.)BiggestProblem:Disjunction(Kuroda2002)5(90) a.Hanako‐ga[[Masao‐niuti‐osoozisuru]‐ka[heya‐dai‐oharaw]]‐aserukotonisita Hanako‐N[[Masao‐Dhouse‐Aclean]‐OR[room‐rent‐Apay]]‐sasethattodo 'HanakodecidedtomakeMasaocleanthehouseorpayroomrent'. Interpretation:sasescopesoverOR;Masaohasachoice. b. Hanako‐ga[[Masao‐niuti‐osoozis‐aseru]‐ka[heya‐dai‐oharaw‐aseru]]kotonisita H.‐NomM.‐Dat house‐Aclean‐sase‐ORroom‐rent‐Apay‐sase thattodo "HanakodecidedtomakeMasaocleanthehouseorshedecidedtomakehim payroomrent" Interpretation:ORscopesoversase;Masaowon’thaveachoice.Also:Problemswithcapturingsyntacticadjunct/argumentasymmetries(Cipollone
2001)P&P’sLF‐as‐syntaxpriorities(91) a.Ifallscopeandcoindexationrelations(anddisjunction,ofcourse)mustbe
syntacticallyrepresented,thenthecausativemorphememusthead(andbeinterpretedin)aseparatesyntacticprojectionthantheverbstemtowhichitisaffixed
b.Syntacticallymonoclausalpropertiesofcausativesmustarisefrom(deficient)propertiesoftheembeddedclausalstructure.
c.Morphologicalandphonologicalwordsarenotinaone‐to‐onerelationshipwithsyntacticterminalnodes.
BiggestProblems:Wherearewordsmade,beforeoraftersyntax,orboth?Whatisthe
constituentstructureoftheembeddedphrase?5Watanabe(pc)saysthatthesameproblemarisesinthetrue'lightverb'constructionswithsuru.
39
ManyproposalswithinbroadlyP&Plines:(92) a. PredicateRaising(e.g.Kuno1973):BiclausalD‐structurecollapsesto
monoclausalS‐structure;Syntaxfeedsword‐formation. b. Parallelmonoclausalandbiclausaltrees.Word‐formationfeedssyntax(e.g.
Miyagawa1984). c. LF‐excorporationandprojection.Word‐formationfeedssyntax(Kitagawa
1986,1994)(ProposalcouldbeunderstoodasavariantofChomsky’s1993lexicalistchecking‐theory.)
d. Incorporation(Baker1988).Syntaxmanipulatesmorphemes,feedsword‐formation.
NecessaryingredientstomakeanIncorporationaccountwork:(93) a.TheVP‐internalsubjecthypothesis(sothatanembeddedsubjectargumentcanbe
introducedinVP,withoutTenseorwhatevertheNPI&Casefunctionalboundaryis.)
b. AtheoryofabstractCasecheckinginwhichaclausalCasedomainisboundedbyaTPprojection,toallowthetransitivityoftheembeddedVPtoaffectthecaseassignedinthewholeclause;similarlyforNPIlicensing(a‘DependentCase’account,oftheMarantz1991type;see,e.g.Miyagawa1999).
c. AtheoryofscopethatallowsquantifierstoscopeattheVPlevelaswellastheCPlevel.
d. ArejectionoftheLexicalistHypothesisatleastforproductivederivationalmorphology;i.e.havetoallowsyntaxtomanipulatemorphemes.(Note:havetoallowsyntaxtoderivekata‘wayof’nominals,too.)
IntheIncorporationapproach,theNumerationisassumedtocontainactual
morphemes,i.e.Vsidentifiedwithphonologicalmaterial.Productiveinflectionalandderivationalaffixescanbeconsideredtobeinputtothesyntax,aswellasregularwords.
DerivationofHanako‐gaTaroo‐nipizza‐otabe‐sase‐tainsuchatheory:6
6(Note:sinceIhavesaidthatsuchanapproachshouldtreatallproductivemorphology(esp.inflectionalmorphology)assyntacticallyattached,Ihaveadopteda‘KaseP’hypothesisforJapanesecaseparticlesinthistree.Fordiscussionofhowthesecasemorphemescanbelicensedagainstcase‐markingheads,seeMiyagawa1999.)
40
(94) Numeration:{HanakoN,TarooN,pizzaN,‐gaK,‐niK,‐oK,tabeV,‐saseV,‐taI} IP KPi I’ NP K VP I Hanako ga ti V’ ta VP V KP V’ sase NP K KP V Taroo ni NP K tabe pizza oNonproductiveaffixes,however,arenotinputtothesyntaxinthisapproach;theycome
pre‐attachedtotheirstemsinapresyntacticmorphologicalcomponent.Thisexplainsa)theirnonproductivity,sincesyntaxissupposedtobethedomainofproductivity,andb)themonoclausalbehavioroflexicalcausatives;oneVinthenumeration,oneVPinthederivation.
Endresult:atypeofhybridaccount,whereproductivecausativesarecombinedwith
theirverbsinthesyntax,butlexicalcausativesaretreatedinaseparate,pre‐syntacticpartofthegrammar.7
Below:What’swrongwiththispicture,andwhattheimplicationsforlinguistictheory
are.
5.2 LexicalcausativesAsinmanylanguages,Japanesederivesmanysemanticallyrelatedinchoative/causative
pairsofverbswithovertmorphologyattachedtoacommonroot.(EvenEnglishdoesthis,forsomepairs).ThesepairshavebeenextensivelydocumentedbyJacobsen1992;thefirsttwoexamplesofeachclassofpairsheidentifiesaregivenbelow:
7Thisbasicpictureonceestablished,manyquestionsremaintobesolved,concerningthemake/letdistinction,theroleofunergativity,unaccusativityandagentivity,psych‐predicatecausatives,restructuringeffects,andmore.Forsomediscussionofrelevantquestions,see,amongmanymanyothers,Dubinsky1994,Terada1992.
Domainforcase‐marking,negativepolaritylicensing.OnlyoneIP,henceonlyonesuchdomain
Domainforsubject‐orientedreflexivebinding,conditionB,adverbialcontrol,quantifierscope.TwoVPs,hencetwosuchdomains.NoteVP‐internalsubjects.
41
(95) Class/# √ Intr Tr Rough√gloss I:e/Ø hag hag‐e‐ru hag‐ø‐u ‘peeloff’ 30pairs hirak hirak‐e‐ru hirak‐ø‐u ‘open’
II:Ø/e ak ak‐ø‐u ak‐e‐ru ‘open’ 44pairs hikkom hikkom‐ø‐u hikkom‐e‐ru ‘drawback’
III:ar/e ag ag‐ar‐u ag‐e‐ru ‘rise’ 71pairs aratam aratam‐ar‐u aratam‐e‐ru ‘improve’
IV:ar/Ø hasam hasam‐ar‐u hasam‐u ‘catchbetween’ 8pairs husag husagaru husagu ‘obstruct(clog,jam?)’
V:r/s ama ama‐r‐u ama‐s‐u ‘remain’ 27pairs hita hita‐r‐u hita‐s‐u ‘soak’
VI:re/s arawa arawa‐re‐ru arawa‐s‐u ‘show(up)’ 18pairs hana hana‐re‐ru hana‐s‐u ‘separatefrom’
VII:ri/s ka ka‐ri‐ru ka‐s‐u ‘borrow/(lend)’ 2pairs ta ta‐ri‐ru ta‐s‐u ‘suffice/(supplement)’
VIII:ø/as hekom hekom‐ø‐u hekom‐as‐u ‘dent’ 38pairs her her‐ø‐u her‐as‐u ‘decrease’
IX:e/as bak bak‐e‐ru bak‐as‐u ‘turninto/bewitch’ 45pairs bar bar‐e‐ru bar‐as‐u ‘come/bringtolight’
X:i/as ak ak‐i‐ru ak‐as‐u ‘tire’8pairs dek dek‐i‐ru dek‐as‐u ‘come/bringinto existence’
XI:i/os horob horob‐i‐ru horob‐os‐u ‘(fallto)ruin’ 6pairs ok ok‐i‐ru ok‐os‐u ‘getup’
XII:Ø/se abi abi‐ru abi‐se‐ru ‘pourover(self/other)’ 6pairs ki ki‐ru kise‐ru ‘puton(self/other)’
XIII:e/akas obi obi‐e‐ru obi‐(y)akas‐u‘(take)fright(en)’ 4pairs hagur hagur‐e‐ru hagur‐akas‐u ‘stray/evade’
XIV:or/e kom kom‐or‐u kom‐e‐ru ‘befullypresent/fill’ 2pairs nukum nukum‐or‐u nukum‐e‐ru ‘warm’
XV:are/e sut sut‐are‐ru sut‐e‐ru ‘fallintodisuse/discard’
3pairs wak wak‐are‐ru wak‐e‐ru ‘divide’
XVI:Misc nigiwa nigiwa‐ø‐u nigiwa‐s‐u ‘(make)prosper’ 25pairs nob nob‐i‐ru nob‐e‐ru ‘extend’
5.2.1 SyntacticandsemanticpropertiesoflexicalcausativesThecausativememberofsuchpairshasonemoreargumentthanitsintransitive
counterpart,andbearsaroughlycausativereadingwithrespecttoit(sometimes
42
oneortheothermemberofthepairhavingundergonesomesemanticdrift),butshowsnoobvioussymptomsofamulticlausalsyntacticstructure,asnotedabove.
Compare,e.g.,theavailablecontrollersforatephraseinasyntacticvs.lexicalcausative
(Dubinsky1994):(96) Basicintransitiveverbanditssyntacticcausative: a. Taroo‐wa arui‐te it‐ta Taro‐T walk‐te go‐PST “Taro,walking,went.” b. Taroo‐wa arui‐te Hanako‐o ik‐ase‐ta Taroo‐T walk‐te Hanako‐A walk‐sase‐PST "TaromadeHanakogo,walking" “Taro,walking,madeHanakogo” (97) Inchoativeintransitiveanditslexicalcausative: a. Hanako‐wanure‐tehi‐e‐ta Hanako‐Twet‐tecool‐inch‐PST “Hanako(‘sbody),gettingwet,cooled. b. Taroo‐wanure‐teHanako‐ohi‐(y)as‐ita Taro‐Twet‐teHanako‐Acool‐caus‐PST “Taroo,gettingwet,cooledHanako.” Impossible:“TaroocooledHanako,(Hanako)gettingwet.”AsshownbyMiyagawa(1980,1984,1989,1994,1998),Zenno(1985),lexicalcausatives
shareanotherpropertywithunderivedtransitiveverbs:theymayappearaspartofanidiom.Sometimestheirinchoativecounterpartalsoparticipates(i.e.theidiomalternates),sometimesnot.(ExamplesbelowarefromMiyagawa1989:126‐127)
(98) Lexicalcausativesinidiomsbythemselves: a. kama‐o kake‐ (intr.kakarnotinthisidiom) sickle‐A splashon ‘trickintoconfessing’ b. zibara‐o kir‐ (intr.kirenotinthisidiom) my.stomach‐A cut ‘payoutofone’sownpocket’ c. tenoura‐o kaes‐ (intr.kaernotinthisidiom) palm‐A return ‘doallatonce’
43
(99) Lexicalcausativesinalternatingidioms: a. te‐ga kuwawar‐ te‐o kuwae‐ hand‐N join hand‐Aadd ‘bealtered’ ‘alter’ b. hone‐ga ore‐ hone‐o or‐ bone‐N breakintr bone‐A breaktr ‘requirehardwork’ ‘exertoneself’ c. mune‐ga itam‐ mune‐o itame‐ heart‐N ache heart‐A hurt ‘beworried’ ‘worry(oneself)’OerhleandNishio(1981)showthatlexicalcausativescanparticipatein‘adversity’
readings,likesimpletransitiveverbs(examplestakenfromMiyagawa1989:130).(100) a. Simpletransitivewith‘adversity’reading: Taroo‐ga ie‐o yai‐ta. Taro‐N house‐A burn‐PST ‘Taroburnedhishouse.’ ‘Taro’shouseburned,andhewasadverselyaffected(hedidn’tcauseit.)’ b. Lexicalcausativewithadversityreading: Boku‐wa kodomo‐o gakekara ot‐os‐ita I‐T child‐A clifffrom drop‐causPST “Idroppedthechildfromthecliff.” “Thechilddroppedfromthecliff,andIwasadverselyaffected.”
5.2.2 V+sase:Thesamepropertiesaslexicalcausatives?ornot?SomeV+sasepairsbehavelikethelexicalcausativesabove.Theyparticipateinidioms,
sometimeswithandsometimeswithouttheirintransitivecounterpart:(101) LexicalV+sasecausativesinidioms: a. tikara‐o aw‐ase‐ power together‐sase‐ ‘pulltogether’ b. mimi‐o sum‐ase‐ ear‐A clear‐sase ‘listencarefully’ c. hana‐ga saku‐ hana‐o sak‐ase‐ flower‐N bloom flower‐A bloom‐sase ‘bedoneheatedly’ ‘engageinheatedly’
44
d. hara‐ga her‐ hara‐o her‐ase‐ stomach‐N lessen stomach‐A lessen‐sase ‘gethungry’ ‘fast/waitforameal’TheseV+saseformsalsoallowadversitycausativeinterpretations:V+saseformsinadversitycausatives(examplesfromMiyagawa1989:129)(102) a. Taroo‐ga yasai‐o kusar‐ase‐ta Taroo‐N vegetable‐A rot‐sase‐PST “Taroospoiledthevegetables.” “Thevegetablesrotted,andTarowasadverselyaffected.” b. Taroo‐ga kaisya‐o toosans‐ase‐ta Taro‐N company‐A bankrupt‐sase‐PST “Tarobankruptedthecompany.” “Thecompanywentbankrupt,andTarowasadverselyaffected.”Butmany(probablymost)V+sasecombinationsdonotexhibittheseproperties.For
instance,thereisnoadversitycausativeinterpretationavailablefortheV+saseformsbelow(Miyagawa1989:130):
(103) a. Boku‐wakodomo‐o gakekara oti‐sase‐ta I‐T child‐A clifffrom drop‐sase‐PST ‘Icausedthechildtodropfromthecliff.’ Impossible:“Thechilddroppedfromthecliff,andIwasadverselyaffected.” b. Kotosi‐wa dekinaigakusei‐o hue‐sase‐ta This.year‐T poorstudents‐o increase‐sase‐PST “Thisyear,wecaused(thenumberof)poorstudentstoincrease.” Impossible:“Thisyear,thenumberofpoorstudentsincreased,andwe wereadverselyaffected.” c. Taroo‐wa niku‐o koge‐sase‐ta Taro‐T meat‐A scorch‐sase‐PST “Tarocausedthemeattoscorch” Pylkkanen2002 Impossible:“Themeatscorched,andTarowasadverselyaffected.”Similarly,givenanintransitiveverbthatparticipatesinanidiom,aV+sasecombination
formedontheintransitiveisnotguaranteedtoalsoparticipateintheidiom(Miyagawa1989:126):
(104) a. kiai‐ga hair‐ *kiai‐o hair‐ase‐
spirit‐N enter spirit‐A enter‐sase ‘befullofspirit’ *‘inspire/putspiritinto’
45
b. hakusya‐ga kakar‐ *hakusya‐o kakar‐ase‐ spur‐N splash.on spur‐A splash.on.sase ‘spuronintr’ ‘spurontr’
5.3 TheblockingeffectandparadigmaticstructureMiyagawa(1980etseq)andZenno(1985)showthatthereisasimplewaytopredict
whenaV+sasecombinationcanbehavelikeotherlexicalcausativesandwhenitmayonlybehaveasananlalyticcausative,withnononcompositionalinterpretationandnoadversitycausative:Onlyintransitiverootswithnoothertransitiveformcanbehavelexicallywithsase.
Thatis,lexicalinterpretationsofsasearepossibleonlyiftheroottowhichitisattached
doesnothaveatransitiveformderivedinanotherway.Thisisaclassicexampleofmorphologicalblocking,seeninbothderivationaland
inflectionalmorphologycross‐linguistically.AsimplecaseistheEnglishpasttense.Someverbsdonothaveapasttenseformedwithed:*runned,*writed,*feeled,*hitted.Thereasonisthattheyhaveanindependentlyformed,irregularpasttense,whichblockstheregularform:ran,wrote,felt,hit.
Similarly,inderivationalmorphology,thesamephenomenonisarguedtooccur.Many
Englishadjectiveshaveanegativeforminun‐,butsomedonot:*unpossible,*unconsiderate,*uncoherent.Theseareblockedbytheindependentirregularnegativeforms,impossible,inconsiderate,incoherent.
Thegrammaticalmechanismthatisresponsibleforblockingeffects,inmanytheoriesof
morphology(forinstance,Paradigm‐FunctionMorphology,mostrecentlydiscussedinStump2001),isthatn‐dimensionalgrammaticalspace:aparadigm.TheideawouldbethateveryEnglishverbalformisunderstoodtobeattachedtoaparadigmspace,definedbytheinflectionalfeaturesofEnglishverbs:past,presentparticiple,1,2,3,sg,pl.Someverbscomewiththeirparadigmspacepartiallyfilledin—forinstance,thepasttensespaceforwrite,theformwroteisalreadyentered—butemptyslots,suchasfortheprogressiveparticiple,arefilledinbyadefaultaffixforthatslot:write+ing.
(105) Paradigminthelexiconforwrite
V:WRITE writeinfinitive presentppl pastppl written
Beforelexicalitemstothesyntax,emptyparadigmspacesarefilledinbydefault
morphology(underlinedinthetablesbelow).
46
(106) Paradigminthelexiconforwrite
V:WRITE writeinfinitive writepresentppl writingpastppl written
Toapplysuchananalysistoderivationalmorphology,onehastohavethenotionofa
multidimensionalgrammaticalspaceforcertainderivationalfeatures,suchasnegative,fortheimpossible/*unpossiblepairs.Wordswithspecialnegativeformswillhavealreadyfilledintheirrelevantparadigmslots,blockingtheinsertionofthedefaultformun‐.
(107)
A:POSSIBLE possiblenegative impossiblenominal possibility
(108) A:LIKELY likelynegative unlikelynominal likelihood
(109) A:HAPPY happynegative unhappynominal happiness
Miyagawa(1980,1984,1989)arguedthattheblockingeffectinJapanesecausatives
showedthataparadigmaticlevelofstructurewasnecessary;withoutit,theblockingeffectcouldn’tbecaptured.Initsessentialpositionandfunctioninthegrammar,Miyagawa’sParadigmaticStructureisthesamelevelofstructurethatparadigm‐functionmorphologistsworkwith,(althoughitseemsMiyagawacameupwithitindependently).
Hedefinedaparadigmspacemadeupofintransitive,transitive,andditransitiveverbs.
Formanyverbstems,anirregularformalreadyoccupiedthe‘transitive’or‘ditransitive’slotintheparadigm;onlyifonedidnotcouldadefaultsaseformfillupthegap.
(110)
V:AG agar‘rise’Intr agarTr ageDitr
47
(111) V:AG sak‘bloom’Intr sakTr sak‐aseDitr
Thisisallverywell,excepttheextralevelofpre‐syntacticlexicalstructureseemed
perhapsexcessive.Notonlythat,MiyagawasawitcannotbeacoincidencethattheseV‐sasecombinations
aremorphophonologicallyindistinguishablefromsyntacticcausatives.Thatis,surely,thereasonthatsyntacticcausativesarespelledoutassaseisjustbecausesaseistheelsewhere,defaultformforacausativemeaning.Iflexicalcausativeshadnothingtodowithsyntacticcausatives,therewouldbenoreasonforthesamemorphemetobeinvolvedinspellingoutboth.
Consequently,hewasledtotheconclusionthatsyntacticcausativeshadtobecreatedin
thelexiconaswell.Butthenalltheproblemswiththelexicalistanalysesofsyntacticcausativescameupalloveragain,leadingtohisproposalthatcausativesareassociatedwithparallelmonoclausalandbiclausalstructures.Thetheorybecameevermorecomplex.
Possibletheoreticalchoices: A:Treatthelexicalandsyntacticcausativescompletelyseparately.Relegate theV+saselexicalcausativestothelexiconwiththerestofthem.Ignore themorphologicalidentitybetweenthedefaultlexicalcausativemorpheme andthesyntacticcausativemorpheme.Thatis:Jacobsenjustmissedclass XVI:Ø/sase. B:Unifythelexicalandsyntacticcausativesbytreatingthembothinthelexicon.
Somethingotherthan‘inthelexicon’hastodistinguishthesyntacticand lexicalcausatives.Parallelstructuresmaydoit,butit’snotclear(howdoes oneallowtheprojectionofaparallelstructureformostditransitiveV+sase combinationsbutnotforalexical‐causativetransitiveone,e.g.?)C:Unifythelexicalandsyntacticcausativesbytreatingthembothinthesyntax. Needed:atheoryofpost‐syntacticmorphology.Againsomethingotherthan ‘inthesyntax’hastodistinguishthetwotypes.
EnterDistributedMorphology,Hale&Keyserv°,andMinimalism.
5.4 LateInsertion,theElsewherecondition,vPsandphases(MostofthefollowingisamildlyrevisedversionofMiyagawa’s1994,1998analysis,
whichappearedinmythesisin1995.)
48
5.4.1 DistributedMorphologyandLateInsertionInDistributedMorphology,thesyntaxmanipulatesabstractfeaturebundles,selectedby
thegrammarofthelanguagefromaninventoryprovidedbyUG,onthebasisofpositiveinfluence.
Thesefeaturebundlesaretheterminalnodesofasyntacticderivation.Afterthesyntaxhasmerged,copied,remerged,probed,Agreed,etc.,andSpell‐Outis
reached,thebundlesaresentofftoPF/LFforinterpretation.AnearlysteponthePF‐sideisLexicalInsertion.VocabularyItems(VIs),specifiedfor
certainfeatures,racetorealizetheterminalnodesthatthesyntacticderivationhasmadeavailable.
Theonewiththemostcompatiblefeatures,andnoincompatibleones,foragiven
terminalnode,realizesthatnode.Forexample,imagineaNumerationsomethinglikethefollowing(imaginetheta‐
featuresontheappropriateitemsifyoulike):(112) {[D+1,+pl,+NOM],[T+past,+NOM],[D+pl,+ACC],[VKEEP,+ACC]}Afterthe(simplified)syntaxisdonewithit,thefollowingtreeishandedofftoSpell‐Out(113) TP Di T’ +1 +pl T° VP +NOM [+past] [+NOM] Di V’ V° D KEEP +pl +ACC +ACC _ we ed kep them I keep it it+Adjacency: Wekep‐ed them+morphophonologyWekep‐t ‘em
WinningVIs
CompetingbutlosingVIs—eligibleforinsertionbutnotmosthighlyspecified
Spell‐outslotsforterminalnodes
49
Benefits:Mirrorprincipleeffects,comprehensiblerelationshipbetweensyntaxand
morphology,singlegenerativeengine(nogenerativelexicon:noparadigmaticstructure,noword‐formationrules,norulesofreferral…)
5.4.2 (Modified)HaleandKeyser(1993,2002)‐typevPsforcausative/inchoativealternations
(114) a. Unaccusativeverbs b. Causativeverbs. vP vP v° √P DPAgent v’ BECOME DP √ John v° √P CAUS thedoor open DP √ thedoor open c. Anotherpossibilityforcausativeverbs:Inchoativecontainedwithinthem? (Miyagawa1994,1998) vP DPAgent v’ John v° vP (I’llargueagainstthisextralayerofstructure CAUS incausatives) v° √P BECOME DP √ thedoor open(115) Hypotheses: a. Externalargumentsarealwaysintroducedbyseparatev°head
(H&K1993,Kratzer1996) b. Differentvarietiesofv°:minimumunaccusativev°andagentive/causativev°. c. Inlanguageswhichshowcausativizing/inchoativizingmorphology,like Japanese,thatmorphologyisarealizationofav°head.
5.4.3 LateinsertionandlexicalcausativesMorphemescompetingtorealizevCAUSinJapanese
50
(116) ‐Ø‐ ↔ CAUS /[√I+IV___v] (38Jacobsenrootsonthelistfor‐Ø‐) ‐e‐ ↔ CAUS /[√II+III+XIV+XV____v](120rootsonlist) ‐s‐ ↔ CAUS /[√V+VI+VII____v] (47rootsonlist) ‐as‐ ↔ CAUS /[√VII+IX+X____v] (91rootsonlist) ‐os‐ ↔ CAUS /[√XI____v] (6rootsonlist) ‐se‐ ↔ CAUS /[√XII____v] (6rootsonlist) ‐akas‐↔ CAUS /[√XIII____v] (4rootsonlist) ‐sase‐↔ CAUS /Elsewhere (norootsonlist) Blockingeffect!
(117) MorphemescompetingtorealizevBECOMEinJapanese: ‐e‐ ↔ BECOME/[√I+IX+XII___v] (79Jacobsenrootsonthelist) ‐ar‐ ↔ BECOME/[√III+IV___v] (79rootsonlist) ‐r‐ ↔ BECOME/[√V___v] (27rootsonlist)
‐re‐ ↔ BECOME/[√VI___v] (18rootsonlist) ‐ri‐ ↔ BECOME/[√VII___v] (2rootsonlist) ‐i‐ ↔ BECOME/[√X+XI___v] (14rootsonlist) ‐or‐ ↔ BECOME/[√XIV___v] (2rootsonlist) ‐are‐ ↔ BECOME/[√XV___v] (3rootsonlist)Elsewhere(similartosase?) ‐Ø‐ ↔ BECOME/[√II+VII+XII____v] (88rootsonlist)Elsewhere?
5.4.4 ImplicationsforsyntacticcausativesIfsaseissimplyanElsewhereformoftheAgent‐introducingvCAUS,andifallsyntactic
causativesarerealizedwithsase,thensyntacticcausativesaretheAgent‐introducingvCAUS,addedontoaphrasebiggerthanaroot—addedon,infact,toanothervPshell:
(118) vP2 … DP v’ Taroo vP1 v° DP v’ sase Hanako √P v° DP √ Ø pizza tabe (Taroo‐gaHanako‐nipizza‐otabe‐sase‐ta)Withasyntacticcausative,head‐to‐headmovementoftherootupthroughitsownv°
andintothematrixsasev°willcreateacomplexstructureinwhichthematrixCAUS
51
v°willnotmeetthestructuraldescriptionforanyspecialroot‐conditionedallomorphsofCAUS:ThematrixCAUSwillbeinsulatedfromtherootbyonelayerofbracketing,theembeddedv°.(Ifthere’snoHM,butit’sjustadjacencyinJapanese,thesameremarksobtain:a‘syntactic’CAUSv°willneverbeadjacenttoaroot.)
(119) (matrixv°afterhead‐to‐headmovement):[[√TABE____v]_____v]
Definitionof‘lexical’causative:aCAUSv°thatisadjacenttoaroot. Definitionofa‘syntactic’causative:aCAUSv°thatisnotadjacenttoaroot
(embedsavP).Comparethelexicalandsyntacticcausativestructuresbelow:(120) a. vP b. vP DP v’ DP v’ Taro‐ga Taro‐ga √P v° vP v° ‐s ‐ase DP √ DP v’ tenoura‐o kae Hanako‐ni √P v° Ø DP √ hansai‐o tutae (121) a.Taro‐gatenoura‐okae‐s… b. Taroo‐waHanako‐nihanasi‐otutae‐sase‐ta Taro‐Npalm‐Areturn‐CAUS Taro‐THanako‐Dstory‐A convey‐CAUS‐PST “Tarodiditallatonce”(?) "TaromadeHanakoconveyastory"Inthelexicalcausative,there’s1vP,1phase,onedomainforQ‐scope,adverbialcontrol,
binding,andtherest.Inthesyntacticcausative,there’s2vPs,hence2domainsforscope,binding,adverbial
control…NotethateveninthesyntacticcausativetherewillstillonlybeoneTP,soonecase
domain,oneNPIdomain
5.4.5 WhynotvBECOMElayerinlexicalcausatives(114cabove)?Becauseitwouldmakeitimpossibletodistinguishbetweenlexicalcausativesand
syntacticcausativesofinchoatives.Comparethestructures,undertheinchoative‐inside‐lexical‐causativeshypothesis,forthefollowingtwosentences,fromMiyagawa1989:130,ex.43a/b:
52
(122) a. Boku‐wa kodomo‐o gakekara ot‐os‐ita I‐Top child‐A cliff‐from drop‐CAUS‐PST “Idroppedthechildfromthecliff." “Thechilddroppedfromthecliff,andIwasadverselyaffected."
Lexical b. Boku‐wa kodomo‐o gake‐kara ot‐i‐sase‐ta I‐Top child‐A clifffrom drop‐BECOME‐CAUS‐PST “Icausedthechildtodropfromthecliff.” #“Thechilddroppedfromthecliff,andIwasadverselyaffected” Syntactic(123) a. vP b. vP DP v’ DP v’ Boku‐wa Boku‐wa vP vCAUS vP vCAUS ‐sase √P vBECOME ‐os‐ √P vBECOME ‐i‐ DP √’ DP √’ kodomo‐o kodomo‐o PP √ PP √ gakekara ot‐ gakekara ot‐IfthelexicalcausativeotosincludesavBECOMEinitsstructure,thentheonlydifference
betweenthelexicalcausativeandthesyntacticcausativeiswhetherornotFusion(apost‐syntacticoperation)hasappliedtothevBECOMEandvCAUSrootstoensurethattheyarespelledoutbythesingleosmorpheme.Thelexical/syntacticdistinctionshouldbemorecategoricalthanameremorphologicaldiacritic,sinceithassuchstrongconsequencesformeaning.
BetterifthelexicalcausativehasthestructurewithouttheinterveningvPBECOME.(124) Observations: (Agentive)vPdomainforspecialmeaning(Kratzer1996,Marantz1997) LF Immediatecontextof√isthedomainforroot‐conditionedallomorphy PF (seealsoArad2002forsimilarclaimsinHebrew)EvenunaccusativevBECOMElookslikeaphaseedge…(Problem:garmorphemeinlexicalcausativeslikeiyagarase,‘bother‐BECOME??‐CAUS’(Problem:lexicalcausativev°morphemesinsideidiomaticnominalizations?seeVolpe
2005)
53
(Problem:whydoessasealwaysalternatewithØ?inprinciple,arootcouldbeonaspeciallistforanunaccusativemorphemelikerore,butnotforacausativemorpheme,andhencealternatewithsase;seeMiyagawa1998foraproposal.)
5.4.6 ThebeginningoftheHigh/LowAttachmentAnalysisThiswasoneofthefirsthigh/lowattachmentanalyses.Attachmentofamorphemetoa
higherfunctionalprojectionresultsinregularmorphologyandcompositionalmeaning,whileattachmentofthesamemorphemetoalowerprojection(oftenthe√),resultsinsomeallomorphyandpotentialmeaningdrift.
OtherearlyexamplesofsuchananalysisistheapproachtoEnglishofingandacc‐ing
gerundspresentedinKratzer1996,andtheapproachtoChichewastativesandpassivessketchedinMarantz1997.
Since,suchapproacheshavebeenextremelyfruitfulinlookingatallkindsof
morphologyonthederivational/inflectional,unproductive/productivecusp,inallkindsoflanguages:
(125) High/lowanalysesfromvariouslanguages Travis2000onMalagasylexicalandsyntacticcausatives. Embick2004onstative,resultative,andpassiveparticiplesinEnglish Fortin2004onMinnangkabucausatives Jackson2005onstativesandresultativesinPima AlexiadouandAnagnostopoulou2005onadjectivalparticiplesinGreek Svenonius2005oncausativesinseverallanguages
5.5 ConclusionsJapanesecausatives—evenomittingthelexicalones—eitherforceonetodomoresyntax
inthelexicon(Manning,Sag&Iida),ormoremorphologyinthesyntax(Baker).Acarefulexaminationoflexicalcausativesforcesonetofigureoutawaytounify
traditionalidiosyncratic,irregularword‐formationwithregular,compositionalsyntax,andyetmaintainaprincipleddistinctionbetweenthetwo.
Apost‐syntacticmorphology—thelateinsertionapproach—withrecursivevPs,allowsa
simple,unifiedtreatmentofallthreetypesoflexicalcausatives,withaprincipledunderstandingofthenatureofthedistinctionbetweenlexicalandsyntacticcausatives.
AdditionalevidenceforthephasalstatusofvP,andsuccessive‐cyclicQRthroughvP.Next:ItaliancausativesandthevPhypothesis!
54
6. ApplicationsII:Capturingthefairepar/faireinfinitifcausativedistinctionReadings:Folli,R.andH.Harley(2007)Causation,obligationandargumentstructure:On
thenatureoflittlev,LinguisticInquiry38.2,197‐238
6.1 Background:TheFI/FPdistinction(Kayne1975)InFrenchcausatives,therearetwopossiblecase‐markersfortheembeddedCauseeinacausativeofatransitiveverb:theprepositionpar‘by’(FP),ortheprepositionà,‘to’(FI)(126) a. MarieafaitnettoyerlarobeàJean MariehasmadecleanthedresstoJean. "MariemadeJeancleanthedress." b. MarieafaitnettoyerlarobeparJeanlenettoyeur. MariehasmadecleanthedressbyJeanthecleaner. "MariegotthedresscleanedbyJeanthecleaner."NonpassivizableidiomsarenotcompatibleundertheFPconstruction(127) a. Safamille acassé lacroûte.
Hisfamily hasbroken thecrust ‘Hisfamilyhadasnack.’
b. #Lacroûte aétécassé parsafamille Thecrust hasbeenbroken byhisfamily(128) Il a fait casser lacroûte àsafamille
He had made break thecrust tohisfamily‘Hehadhisfamilyhaveasnack.’
(129) #Il afaitcasser lacroûte parsafamille. He hadmadebreak thecrust byhisfamily ‘Hehadhisfamilyhaveasnack.’FrenchinalienablepossessionbetweentheCauseeandtheembeddedobjectisokinFI
butnotFP:(130) a. Elle fera lever lamain àJean.
She will.makeraise thehand toJean ‘ShewillhaveJeanraisehishand.’
b. # Elle fera lever lamain par Jean. She will.makeraise thehand by Jean ‘ShewillhavehishandraisedbyJean.’
55
BindingintotheembeddedobjectispossiblefromtheFIàCausee,butnotfromtheFPpar‐Causee(Burzio,inItalian):(131) a. Giannijhafattotemperare lasuai/jmatita aogniragazzoi.
Giannihasmadesharpen thehispencil toeveryboy. ‘Giannijhadeveryboyisharpenhisi/jpencil.’
b. Gianniihafattotemperare lasua*i/jmatita daogniragazzoi. Giannihasmadesharpen thehispencil byeveryboy ‘Giannijhadhis*i/jpencilsharpenedbyeveryboyi.’IntheFP,theda‐phraseisoptional.BurzioandGuastiarguethatintheFI,theàCauseeisnotoptional.HereisanidiominItalianthatisnotcompatiblewiththeFP:(132) Marcononhafattofare untubo aMaria/*daMaria Marc nothasmademake atube toMaria/byMaria ‘Marcdidn'tletMariaachieveanything.’(Lit:‘..didn’tmakeMariamakeatube.’)IfyouomittheCausee,theidiomaticinterpretationisnotpossible:(133) Marconon hafattofare untubo. Marc not hasmademake atube ‘Marcdidn’thaveatubemade.’ ‘#Marcdidn'tletMariaachieveanything.’IftheembeddedclausecontainsnoCausee,then,itmustbeacaseofFP.Lastly,theFIhasamuchmore'direct'causativefeelthantheFP:(134) Marie feraboire cetteeau parsonchien/àsonchien.
Marie will.makedrink thiswater byherdog/toherdog‘Mariewillhavethiswaterdrunkbyherdog/herdogdrinkthiswater.’
ThisisespeciallyobviousiftheCauseeissomeonewhonormallydoesthecausedjob—it'sveryoddtousetheFIinthatcase,sinceyouwouldn'tnormallyhavetoforcethemtodoit:(135) a.Gianni hafattoriparare lamacchina aMario/daMario.
Gianni hasmaderepair thecar toMario/byMario‘GiannigotMariotorepairthecar.’/‘GiannigotthecarrepairedbyMario.’
b. ??Gianni hafattoriparare lamacchina almeccanico diviaFiume. Gianni hasmaderepair thecar to.themechanicofstreetFiume ‘GiannihadthemechanicinFiumeSt.repairthecar.’ c. Gianni hafattoriparare lamacchina dalmeccanicodiviaFiume. Gianni hasmaderepair thecar by.themechanicofstreetFiume GiannihadthecarrepairedbythemechanicinFiumeSt.’
56
SummaryofkeydifferencesbetweenFIandFP:(136) A.TheCauseeofatransitiveembeddedverbismarkedwithdativecaseintheFI,
andbyaprepositiondainFP(inItalian).B. Non‐passivizableidiomsareavailableintheFIbutnotintheFP.C. Thea‐phraseintheFIcanbindtheembeddedobject,whiletheFPda‐phrase
cannot.D. TheCauseemaybeomittedintheFPbutnottheFI.E. Non‐passivizableverbsareacceptableintheFIbutnottheFP.F. ThereisasenseofobligationontheCauseeintheFIbutnotintheFP.
Generalideaofallapproachestothisphenomenon:ThefullargumentstructureofthecausedpredicateispresentintheFI,includingtheexternalargument;intheFP,thereissomereductionintheargumentstructuresuchthattheexternalargumentisnotpresent.Previousapproacheshadtoappealtoalexicalarity‐reductionoperationtogeneratetheexternal‐argumentlessinfinitiveintheFPTheadvantageofthevPapproach:ThedifferencebetweentheFIandtheFPcanjustfollowfromadifferentsizeofembeddedclauseinthetwo:theFIembedsavPandhenceanexternalargument;theFPembedsavP‐lessprojection,andhencelacksanexternalargument.Theexternalargumentcanthenoptionallybespecifiedbyaby‐phrase,asitcaninotherexternal‐argumentlessconstructionslikethepassive.
6.2 ThevPbasedanalysis(buildingonIppolito2002)(137) a.FI b.FP vP vPGianni v′ Gianni v′ vCAUS vP vDO VPNom fare v' DPDAT fare VPNom PP v VPaMario V DPdaMario Ø V DP ripararelamacchina ripararelamacchinaNoterightwardspecifiersofvPintheembeddedclause.Notethatfareisinthistreatmentalightverb,av°itself,justlikeJapanese–sase,butnotaffixal.WhyassumetheinfinitiveintheFPisanominalization?First,infinitivescanbenominalincharacterinItalian:
57
(138) a. [Questocontinuoparlaredell’‐aviaria]DP infastidisce Marco.
[Thiscontinuoustalkingof.the‐bird.flu]DP bothers Marcob. [TuttoquelleggereDostojevsky]DPharovinato Marco. [AllthatreadingDostojevsky]DP haswrecked Marco
Second,wehaveindependentevidencethatonlycertainkindsofv°allownominalcomplements;ifwecanshowthatFPfareisthatkindofv°,thenthatwilcorroboratethenotionthattheFPinfinitiveisnominalincharacter.Toestablishthis,though,weneedsomebackground.
6.3 Theselectionalpropertiesofdifferentv°flavors:FolliandHarley(2005)HaleandKeyser:UnergativeverbsaremadeupoflightverbvDOandanominalcomplementChange‐of‐stateverbslikereddenseemlikethelightverbmustmean'cause'—vCAUSE,andapredicativeSCcomplement.ArethereotherdifferencesbetweenvDOandvCAUSEthatwecandetect?Considerthefollowingeffectofanimacy:(139) a. Johnatetheapple(up).
b. Theseaatethebeach*(away).c. Gianniha/ si=é mangiato unamela. G. has/ REFL=is eaten anapple. ‘Giannihaseaten/haseatenupanapple.’
d. Il mare *ha/si=è mangiatola spiaggia. Thesea has/REFL=is eaten thebeach. ‘Theseaatethebeach.’InanimatesubjectsofconsumptionverbsrequireapredicativeSCcomplement(133b,d),wherethereisaclearresultstatespecified.Animatesubjectscanhaveoneortheother—anominalcomplement(theapple)orapredicativeone([theappleup]).ThisisbecauseinanimatesubjectscanonlybeCAUSERs,notagentsofvDO—onlytrulyintentionalentities(or'teleologicallycapable'entites)canbesubjectsofvDO.vDO,thenrequiresananimateexternalargumentandcantakeanominalcomplementvCAUSEcanhaveaninanimateexternalargumentandrequiresapredicativecomplementclause.
58
(140)
Flavorofv0 Specifier ComplementvDO Agent Nominal
orSmallClausevCAUSE Causer
orAgentSmallClause
Prediction,then:IfFPtakesanominalizedinfinitiveasitscomplement,itshouldrequireananimatesubjectoffare,whilesinceFItakesasubject+predicatevPcomplement,itshouldbeokwithaninanimatesubject.Seemstobetrue:(141) a. Larabbia fecerompere iltavolo a/*daGianni.
Therage madebreak thetable to/byGianni ‘RagemadeGiannibreakthetable.’
b. Lagenerosità fecedonare lacasa a/*daGianni Thegenerosity madegive thehouse to/byGianni ‘GenerositymadeGiannidonatethehouse.’(142) Lafamine afaitmanger desrats aux/*parleshabitantsdelaville.
Thefamine hasmadeeat of.therats to.the/bytheinhabitantsofthecity.‘Thefaminemadetheinhabitantsofthecityeatrats.’
Somuchfortheanalysisoffareitself—intheFIit'svCAUSEandintheFPit'svDOWhatabouttheembeddedvPintheFI?Isthereanythingspecialaboutit,thatcouldhelpusexplainthe'obligationeffect'?Hypothesis:IfthevPembeddedunderfareintheFIisheadedbyvDO,thatwillmeanthatitssubject(theà‐Causee)mustbeananimateAgent.Theobligationeffect,then,couldarisebecausetheonlywaytogetanAgenttodosomethingistoobligehimto—youcan'tphysicallyforcesomeonetodoatrulyagentiveaction;ifyou'rephysicallyforcingthem,they'renotanAgentatall.IftheFIfareembedsavDO,theselectionalpropertiesofvDOpredictthatitshouldbeimpossibletomakeaFIwithaninanimateCausee,sincetheembeddedvDOwillalwaysselectforananimateCausee…Thisseemstobecorrectaswell:(143) a. Maria/Ilramo harotto lafinestra.
Maria/Thebranch hasbroken thewindow.
59
b. Gianni hafattorompere lafinestra aMaria/*alramo. Gianni hasmadebreak thewindow toMaria/to.thebranch ‘GiannimadeMaria/*thebranchbreakthewindow.’ c. Iltecnico/Ilprogramma hadisinfettato ilcomputer. Thetechnician/Theprogram hasdisinfected thecomputer. d. Gianni hafattodisinfettare ilcomputer altecnico/*alprogramma. Gianni hasmadedisinfect thecomputerto.thetechnician/to.theprogram ‘Giannimadethetechnician/*theprogramdisinfectthecomputer.’CertainverbswhicharepurelycausativecanneveroccurunderaFI—object‐experiencerpsychverbs,forexample:(144) a.La discussione/Giannihaassorbito Maria. The discussion/ GiannihasabsorbedMaria. b. *Lalezione/*Lamaestrahafattoassorbire Maria alladiscussione/aGianni.
Thelesson/Theteacherhasmadeabsorb Maria to.thediscussion/toGianni ‘Thelesson/Theteacherhasmadethediscussion/GianniabsorbMaria.’ c. Laguerra/Gianni hadisturbatoMaria. Thewar/Gianni hasdisturbedMaria. d. *Ilprogrammatelevisivo/*MarcohafattodisturbareMariaallaguerra/aGianni.
Theprogramtelevised /Marcohasmadedisturb Maria to.the war/toGianni ‘Thetelevisionprogram/Marcohasmadethewar/GiannidisturbMaria.’So,weconcludethefollowing:
o FIembedsavPo FPembedsanominalizedVPo Embeddedv°inFIisvDO,whichiswhyCauseemustbeanimate,and
whyembeddedcausativeverbsarebadintheFI.o Matrixv°inFPisvDO,whichiswhyCausermustbeanimateo daphraseisoptionaladjuncttonominalcomplementinFP,justasit
isinapassive.There'sasectiononcaseassignmenttodealwiththedative/accusativealternationwhichwecandiscussifyoulike—thefactsareinterestinglysimilarinJapaneseandItalian,withonekeydifferencetodowiththepassive.ThatdifferenceledtoadiscoveryconcerningtheItalianpassivesofcausatives,whichIwillbrieflyoutline:
6.4 InteractionofpassivewithfarecausativeBasiccaseassumptions:FIconstructionsareECMstructures,wherematrixv°fareischeckingcaseontheCausee,andtheembeddedv°oftheverbischeckingcaseontheembeddedobject(ifany).FPconstructionsarecomplexpredicates,withmatrixv°farecheckingcaseontheembeddedobject.
60
Prediction:IffarelosesitsexternalargumentandcannotassigninternalcasetoitsCausee,becausefarehasbeenpassivized,thentheCauseeshouldbecomethenewderivedsubject,andcheckNominativecase.Infact,thisiswhathappensinJapanese:(145) a. Hanako‐ga Tanako‐ni piza‐o tabe‐sase‐ta
Hanako‐NOM Tanako‐DAT pizza‐ACC eat‐CAUS‐ "HanakomadeTanakoeatpizza."b. Tanako‐ga piza‐o tabe‐sase‐rare‐ta Tanako‐NOM pizza‐ACC eat‐CAUS‐PASS‐PAST “Tanakowasmadetoeatpizza”
…butitdoesnothappeninItalian;passivewhichpromotesthedativeCauseetonominativecaseiscompletelyimpossible;theonlytypeofpassivepossiblepromotestheaccusativeembeddedobjecttonominativeposition:(146) a. embeddedaccusativeobjectsoftransitiveverbspassivize:
Illibro fufattoleggere aMario (daGianni). Thebook wasmaderead toMario (byGianni) ‘Mariowasmadetoreadthebook(byGianni).’ b. embeddedaccusativesubjectsofintransitiveverbspassivize: Ilpacchetto fufattoarrivare (daGianni). Thepacket wasmadearrive (byGianni). ‘Thepacketwasmadetoarrive(byGianni).’ c. embeddeddativesubjectsoftransitivesdonotpassive: *Maria fufattamandare unpacchetto (daGianni). Maria wasmadesend apacket (byGianni). ‘Mariawasmadetosendapacket(byGianni).’But!ThereareinterestingconditionsonthepassivizationofcausativesinItalian,whichanyaccountfocussingpurelyon'accusative'vs'dative'casecannotexplain!CausativesofintransitiveverbsassignaccusativecasetotheirCausee,nomatterwhethertheverbisunergativeorunaccusative:(147) a. Giannihafattotelefonare Marco. GiannihasmadetelephoneMarco "GiannimadeMarcotelephone. b. GiannihafattopartireMarco. GiannihasmadeleaveMarco. "GiannihasmadeMarcoleave."
61
Buttheonlycausativesofintransitiveverbswhichcanbepassivizedarecausativesofunaccusatives—causativesofunergativescannotpassivize!(148) a. *Marco èstatofattotelefonare (da Gianni) Marco isbeenmadetelephone (by Gianni) b. *Marco èstatofattoridere (da Gianni). Marco isbeenmadelaugh (by Gianni) c. ??Marco èstatofattopiangere (da Gianni). Marco isbeenmadecry (by Gianni)(149) a. Marcoèstatofattopartire. Marcoisbeenmadeleave ‘Marcowasgottentoleave.’ b. Marcoèstatofattocadere (da Gianni). Marcoisbeenmadefall (by Gianni) ‘Marcowasgottentofall(byGianni).’ c. Ilpacchetto fufattoarrivare (daGianni). Thepackage wasmadearrive (byGianni) ‘Thepackagewasgottentoarrive(byGianni).’ (English??)Alternatingunergative/unaccusativeverbslikesaltareareparticularlyrevealing;whiletheactivecausativeisgoodwithbothinterpretations,thepassivecausativeonlyallowstheunaccusativereadiing:(150) a. GiannihafattosaltareilponteVecchio. GiannihasmadeexplodethebridgeVecchio "GiannimadetheVecchiobridgeexplode." b. GiannihafattosaltareMarco. GiannihasmadejumpMarco. "GiannimadeMarcojump."(151) a. IlponteVecchio fufattosaltare.
ThebridgeVecchio wasmadeexplode. b. ??Marco fufattosaltare. Marco wasmadejump.OurexplanationisthattheItalianpassive,whichisformeddifferentlythantheJapanesepassive,requiresamainverbtooperateon,andonlyFPfarecanbeamainverb—FIfareisonlyav°.WeclaimtherearenopassivesofFIfare.Thisprovidesanexplanationforthefactsin(143‐146)inthefollowingway:FPfareembedsanominalizedVPwithoutavPattached.Unergativeverbs'agentsareintroducedbythevP,whileunaccusativeverbs'singleargumentsareintroducedbythemainverb,theV.
62
SinceFPfaredoesnotembedavP,weconcludethatitcannotembedanunergativeverbwithitsagentargument.Allcasesoffare+unergativeVmustbeFIfare.PassivesoffareareincompatiblewithembeddedunergativeVTherefore,passivesoffarearepassivesofFPfare,notFIfare(iftheywereFIfare,wewouldexpecttoseethemwithembeddedunergativeV).Threequestions:a)WhydoesItalianpassiveneedamainverbtooperateon?b)WhataboutallthosepassivesoftransitiveverbswithanaCausee,asinin(141a)above?Ifthey'renotpassivesofFI,whereisthea‐Causeecomingfrom?c)HowcanwejustifysayingthatFPfareisa'mainverb'?Answertoa):Becauseitinvolvesalteringv°toeliminatetheexternalargumentandleavingaparticiplebehind.(Illustrate).Answertob):Italianallowsdativeapplicativeargumentstobeintroducedquitefreelyintotransitivestructures;theseareapplicativesofFPstructures.InsupportofclaimthatpassivesoffareareFP,notFI—recallwediscoveredarestrictiononthesubjectsofFPfaretotheeffectthattheymustbeanimate?Thatrestrictioncarriesovertotheby‐phrasesinpassivesofcausatives:(152) Èstatofattorompere iltavolo (aMarco) daMaria/*dallarabbia. Isbeenmadebreak thetable (toMarco) byMaria/by.therage. ‘Atablewasmadetobreak(onMarco)byMaria/byrage.’Thisrestrictionreallycomesfromthefarehere,notfromtheby‐phrase;inanimateCausersareperfectlygoodby‐phrasesinpassiveslikeGiannifuportatoalsuicidiodallarabbia,‘Gianniwasdriventosuicidebyrage’,correspondingtotheactiveLarabbiahaportatoGianniasuicidio,‘RagedroveGiannitosuicide.’Answertoc):Well,onits'mainverb'interpretation,'create',faretakesanominalcomplement(makeacake,etc.);theonlyrealdifferencebetweenFPfareand'mainverb'fareisjustthatthenominalcomplementdenotesanevent,ratherthanathing.
6.5 SummarySowehaveseenthefollowinghere:
o ThevPgivesusanewinsightintotheFP/FIdistinctionintermsofselectionoromissionofCauseearguments
o Therearedifferenttypesofexternal‐argument‐selectingv°,whichhavetheirownselectionaleffectsontheirspecifierandcomplement
Next:EventstructureandthevP
63
7. EventstructureandroottypeinEnglishdenominalverbsReading:Harley,H.(2005)Howdoverbsgettheirnames?Denominalverbs,Manner
IncorporationandtheontologyofverbrootsinEnglish,inNomiErteschik‐ShirandTovaRapoport,eds.,TheSyntaxofAspect,42‐64.Oxford:OxfordUniversityPress.
7.1 Introduction:Adifferentsubdivisionofaspectualclasses(153) Discussionsofaktionsartandverbclassgenerallydivideeventiveverbsintothree
kinds:A incrementalthemeverbs(verbsofcreationandconsumption,ormakingand
unmaking)B change‐of‐stateverbs(bothtransitiveandunaccusative)C otherunergativeandtransitiveverbs,ofalltypes:activities,semelfactives,
andsomeaccomplishments Inmostoftheliterature,AandBhavebeentreatedasanaturalclass.BothAandB
verbsareusuallyAccomplishments,andbothmayhavethemesthatMeasure‐Out,inthesenseofTenny1992.Theyhaveusuallybeentreatedtogetherindiscussionsoftherobustconnectionbetweenobjectboundedness,objectcaseandmeasuring‐out(e.g.Tenny2000;VanHout2000).
Claim:adifferenttypologyofverbclassesisneeded Wecanaccountfortheaktionsartpropertiesofmorepredicatesifweunderstandthe
waysinwhichgroupsAandCformanaturalclass,distinctfromB. Hardtoswallowdistinction:Wemustdistinguishbetweenverbswhosenamesare
derivedviaincorporationofaRootfromwithintheargumentstructureandverbswhosenamesarederivedsomeotherway,let'ssaybyamysterious,parametricallyvarying,magicalprocesswhichI'llcallMannerIncorporation
7.2 Background(154) Objectsandmeasuring‐out a. Suedrank/wrote forhours/#in5minutes. b. Suedrankapintofbeer/wroteastory #forhours/in5minutes c. Suedrankbeer/wrotestories forhours/#in5minutes. d. Suewroteatastory forhours/#in5minutes Muchrecentworkontelicityhasturnedontheimportantconnectionbetweenthe
directobjectpositionandthetelicityoftheVP,showninTenny1992andalsoDowty1991.ThecentralobservationisthatinmanyVPs,theboundednessofthedirectobjectdeterminesthetelicityoftheeventdenotedbythewholeVPcomplex.Aproposalthat
64
hasgainedsubstantialcurrencyisthatthereisafunctionalprojectionwhichchecksthefeaturesofthedirectobjecttoprovideanaspectualinterpretation,e.g.Borer1993;Borer1996;vanHoutandRoeper1998,amongmanyothers.Thisprojectionissometimesconflatedwiththeaccusativecase‐checkingprojection,sometimesindependentofit.
Otherauthorshavecalledtheimportanceofthedirectobjectasadetermineroftelicity
intoquestion,notablyJackendoff1991;Jackendoff1996andalsoLevin2000.Thereareverbswhichtakeanovert,bounded,definitedirectobjectandareyetinherentlyatelic(5a,c);theybecometelicwhenagoalargumentisprovided(5b,d).
(155) Objectswithoutmeasuring‐out: a. Suepushedthecart foranhour/#inanhour. b. Suepushedthecarttothefield #foranhour/inanhour. c. Suekickedtheball foranhour/#inanhour d. Suekickedtheballtothecenter #forasecond/inasecond Thereisasimilarsetofunergativeverbsofmotion:theyareessentiallyatelic,asis
expectedsincetheydon'thaveadirectobject,but,theymaybecometelicwiththeadditionofagoalPP(stillwithoutadirectobject)illustratedin(2).
(156) Measuring‐outwithoutobjects a. Suedanced foranhour/#inanhour. b. Suedancedacrossthestage #forfiveminutes/infiveminutes. c. Suehopped foranhour/#inanhour d. Suehoppedacrossthestage #forfiveminute/infiveminutes Anessentiallysimilarclassofverbsofmotionmaybetransitiveaswellasintransitive,
butdonotbecometelicuntilagoalPPisadded:(157) Objectswithoutmeasuring‐outandmeasuring‐outwithoutobjects: a. Suewalked foranhour/#inanhour. b. Suewalkedthedog foranhour/#inanhour. c. Suewalked(thedog)tothepark #for5minutes/in5minutes. Withrespecttotheseverbsofmotion,whenmotionappearstobespontaneousor
internallycaused,thereisawell‐knownconnectionbetweentestsforunaccusativityandthepresenceofagoalPP:
(158) BuygoalPP,getobjectforfree: a. There‐insertion: Thebulletwhistledasitpassedmyear. *Therewhistledabullet(asitpassedmyear). Therewhistledabulletpastmyear. b. AuxiliaryselectioninDutchBorer1996
65
Janheeft/*isgesprongen Janhasjumped. Janisindeslootgesprongen Janisin(to)theditchjumped. Janheeftindeslootgesprongen Janhasintheditchjumped Athirdclassofatelicactivity/semelfactiveverbswithobjectsbecometeliconlywith
theadditionofaresultphraseRappaportHovavandLevin1998:(159) Buyresultativephrase,getmeasuring‐outforfree a. Suehammeredthemetal for5minutes/#in5minutes. b. Suehammeredthemetalflat #for5minutes/in5minutes. c. #Thismetalhammerseasily. d. Thismetalhammersflateasily. Whyaretheseverbsdifferent? fromVanHout2000:"FollowingDowty,TennyKrifkaandVerkuyl,Itakeitthatitisa
lexicalpropertyofverbsthatdistinguishesthepush‐classfromverbslikedrinkandwrite."
7.3 Apurelysyntacticapproach Syntacticvs.semanticbootstrapping Inthispaper,Iproposetoidentifywhatthatlexicalpropertyis.Iclaimthatitisan
intersectionofvariousindependentpropertiesoftheverbroot:itsstructuralposition,itsontologicalclassanditsinherent(un)boundedness.
Weneedawaytomotivatethesuddenacquisitionofmeasuring‐outabilityincases5‐9,
andexplaintheabsenceofmeasuring‐outabilitywhereit'sabsent.ThedominanttypeofexplanationforthesephenomenahasbeenthatasemanticalterationtotheLCSoftheseverbs(e.g.viatheadditionofaPathargumentoraresultativestate),hastheeffectthatthemappingrulesproducedifferentresultsinthesyntax.I'llcallthisasemanticbootstrappingapproach.Iwishtoargue,withMateuFontanals2000,thatinfact,theadditionofPPorresultativestatematerialin5‐9directlyforcesasyntacticchangewhichgivesthecorrectresults.Ifit'snecessaryatall,theLCS‐typeinformationcanbereadoffthesyntax.I'llcallthisasyntacticbootstrappingapproach.
7.3.1 Anoverlookedclassoftelicverbs Tobegintomaketheargumentforsuchanapproach,let'sfirstconsideraclassof
unergativeverbsthat(unusually!)denoteAccomplishments,HaleandKeyser'sdenominalunergativeverbs.
66
(160) HaleandKeyser'sdenominalunergativeswithThingroots a. Themarefoaled #for2hours/in2hours b. Thedogwhelped #for2hours/in2hours c. Thecowcalved #for2hours/in2hours.(161) AnadaptationofH&K'sproposalforverbsofbirthing: vP 3 Themare v' 3 v √P | √ foal HaleandKeyserproposethatunergativeverbs(ingeneral)areessentiallytransitive,
derivedbyincorporatinganounrootinobjectpositionintothetransitiveverbthatselectsit;thatis,byconflatingatransitivestructure.
(162) Telicityofbothunergativeandtransitiveparaphrase a. Themarefoaled #for2hours/in2hours b. Themarehadafoal #for2hours/in2hours Theaktionsartpropertiesoftheseverbscorrespondtotheaktionsartpropertiesof
theirtransitiveparaphrases.Inbothcases,itshouldbeobviousthatthebabyanimal(s)thatarecontainedinthemother'swomb(hencenecessarilyfiniteinnumber)aretheincrementalthemethatdeterminesthetelicityofthepredicate.
(163) Thedifferencebetweenbabiesandotherbodilyemissions a. Thebabydrooled for2hours/#in2hours b. Theatheletesweated for2hours/#in2minutes c. Thewoundbled for2minutes/#in2minutes d. vP 3 Thebaby v' 3 v √P | √ drool Noticethatalltheseunergativeverbsofbodilyemissionareatelic,unbounded.
67
(164) Atelicparaphraseswithincrementalthemes a. Thebabymadedrool for2hours/#in2hours. b. Theathletemadesweat for2hours/#in2hours. c. Thewoundoozed/madeblood for2minutes/#in2minutes. Conclusion#1:intheparaphrasesin(13b)and(15)weattributetelicityorlackofitto
themassvs.countpropertiesoftheincrementalthemeincomplementposition.Inthecorrespondingunergativeverbs,theverbsarederivedviaincorporationofanominalrootfromcomplementposition—theincrementaltheme—whichhasinherentmassorcountproperties.Theparalleltelicitypropertiesoftheunergativeverbsandtheirtransitiveparaphrasesshouldbeattributedtothesamemechanism.Alexicalsyntacticaccountallowsustodothat.
Consequence#1:inatleastthesecases,theboundednesscannotbecheckedinSpec‐
AgrOPorsimilarfunctionalprojectionasacasefeatureorteliceventfeature(c.f.VanHout2000).Conceivablyit*could*bethecasethatfeaturecheckingintheseunergativeverbsisaccomplishedviaincorporationratherthanspec‐headagreement,ifwewishtomaintainafeature‐checkingaccount.
(165) Somebodilyemissionverbsthatneedextraexplanation a. Theboypeed for5minutes/in5minutes b. Johnspit #for5minutes/#in5minutes Thepeecase:peeisamassnoun,likesweatorblood,butinadditiontotheunbounded
reading,thereisaboundedreadingavailable.ThiscanbeexplainediftheUniversalPackagerhasapplied(thatallowsonetoorder"acoffee");notunreasonableinlightofthefactthatitisparticularlysalientthatpeecomesindiscretequantities,limitedbythesizeofthecontainer.Itdoes,however,entailthatthePackagercanbeapurelyinterpretive/pragmaticmechanism,notrequiringasyntacticreflex,asinterveningstructureorabstractmaterialwouldpresumablyblockincorporationoftheroot.
Thespitcase:spitisanapparentproblem.Initsnominalform,itisdefinitelyamass
noun.However,theverbseemstobeasemelfactiveunergativeinitsbehavior(seebelow).Iwillconsiderittobenaminganevent(theactofspitting)ratherthanathing,andtreatits"thing"meaningassecondary.
7.3.2 DenominalunergativeswithEventroots Sofar,wehaveinvestigatedtwotypesof√s:√sthatdenoteThingsthatareeither
boundedorunbounded.Thebounded√sincomplementpositiongiveustelicpredicates,measuredoutbythebounded√,whileunbounded√sincomplementpositiongiveusatelicpredicates.Wecansumupthetypologyofrootssofarasfollows:
68
(166) TwokindsofThingroots
bounded unbounded Thing foal drool
(167) TwokindsofunergativeverbswithEventroots Activities a. Suedanced for5minutes/#in5minutes b. Suewhistled for5minutes/#in5minutes c. Sueslept for5minutes/#in5minutes Semelfactives d. Suehopped #for5minutes/#in5minutes e. Suetripped #for5minutes/#in5minutes f. Thelightflashed #for5minutes/#in5minutes Notethatdenominalunergativeswithevent‐namingrootscannotbetelic,unlikethe
verbsofbirthingabove.Rather,theyareinstantaneousevents,whichmaybecoercedtoarepetitionreadingifcoocurringwithanatelicframeadverbial.FollowingSmith1991,I'llcallthesesemelfactives.
H&Kproposethesamestructurefortheseverbsasforthedenominalverbsabove:(168) Samestructure: a. vP b. vP 3 3 Sue v' Sue v' 3 3 v √P v √P | | √ √ dance hop(169) Sameaktionsartpossibilitieswithparaphraseandunergative a. Suedanced for5minutes/#in5minutes b. Suedidadance for5minutes/in5minutes c. Suehopped #for5minutes/#in5minutes d. Suedidahop #for5minutes/#in5minutes Notetheonedifferenceintheatelicparaphrase:"dance"initsnominalformisacount
noun,andameasured‐outtelicreadingisavailableforthetransitiveparaphrasein20(b).Aswithpeeabove,though,theimportantthingtonoticeisthatitdoesallowanatelicreading,indicatingthatitmaybeinterpretedunboundedly.
AspeculationaboutthenatureofrootsthatnameEvents:
69
TheboundedEventrootsabovedonot"measure‐out";rather,theynameaneventthatoccursatapointintime,notonethatevolvesovertime.ConsiderthatinthecaseoftheboundedThingroots,themeasuring‐outoccuredoverthephysicalquantityoftheboundedThing(s)inquestion.Ihypothesize,followingPustejovsky1991andJackendoff1991thatwhileboundedThingsmustnecessarilytakeupspace,linguisticEventsarefundamentallyeitherpointlike(instantaneous)orextendarbitrarilylong(activities).
Wherewe'regoing:Mosteventsthatevolveovertimetoaculminationpoint
(accomplishments)mustbeconstructedfromtwosub‐eventualities(againfollowingPustejovsky1991).Moreonthisanon.(Note:Incrementalthemeverbs(foaletc.)willconstitutetheexceptiontothisgeneralizationaboutaccomplishments.)
(170) Fourkindsof√s
bounded unbounded Thing foal drool Event hop dance
Thestorysofar:
o Unergativeverbsarecreatedbyincorporatinganominalrootintoalightverb.
o Thetelicityoftheresultingverbcanbepredictedonthebasisofthe
ontologicalcategoryoftheroot(EventorThing),andwhetherthatrootdenotesaboundedoranunboundedentity.
7.4 Transitiveatelicverbs Recallourclassofproblemverbs:theyhaveanon‐affectedobjectwhichcannot
measureout.Inthepast,thishasbeenattributedtotheAffectednessCondition,whichgovernstheapplicationofmappingrules.
(171) Pushing,hitting,kicking a. Johnpushedthecart for5minutes/#in5minutes b. Suedrovethecar for5minutes/#in5minutes c. Suekickedthewall #for5minutes/#in5minutes d. AbirdpeckedSue #for5minutes/#in5minutes IfEvent‐denotingroots(butnotThing‐denotingroots)canselectforacomplement,we
cangroupthesetogetherwiththeunergativeverbswithEvent‐denotingrootsin(18).Notethattheyhavethesameaktionsartpropertiesandtheyallhavecorrespondingevent‐denotingnominals(apush,apeck,etc.).Thiswouldthenentailthattheyhavethestructurebelow:
70
(172) Aproposal a. vP b. vP 3 3 Sue v' Sue v' 3 3 v √P v √P 3 3 √ DP √ DP push ! kick ! thecar thewall Whyisn'tthereacorrespondinggroupoftransitivedenominalverbswhoseroots
denoteThings,notEvents,andwhosetelicitydependsontheboundednessoftheincorporatedthing??LetussupposethatrootsdenotingThingscannotselectarguments8,whileEventscandoso.Ourinventoryofbasicrootpropertiesnowlookslikethis:
(173) Anotherspeculation
no complement complement bounded unbounded bounded unbounded
Event hop sleep kick push Thing foal drool N/A N/A
The$64,000question:Whycan'ttheseobjectsmeasure‐out? Beforeansweringthat,let'sfirsttakealookatthestructureoftheothermajorclassof
verbswhoseobjectsdomeasureout:notIncrementalThemepredicates,butChangeofStatepredicates.
7.5 ChangeofStateverbs(174) Deadjectivalchange‐of‐stateverbs a. Sueclearedthetable #for5minutes/in5minutes. b. Thearchaeologistopenedthesarcophagus#for5minutes/in5minutes c. Suetamedthelion #for5minutes/in5minutes d. Sueroughenedthetiresurface #foraminute/inaminute Theseare,ofcourse,thecanonicalverbsthatappeartohaveaverystraightforward
semanticanalysisintermsofCAUSE+(BECOME)+STATE,whereSTATE=asmallclauseconsistingoftheadjectivalstatepredicatedoftheobject.Someundergotheinchoative/causativealternation,somedonot.
8Maybe.WhataboutBillfatheredason(?in2years/#for2years).
71
Essentiallypreservingtheanalysisofthegenerativesemanticists,H&K(andmany
others)proposethefollowinglight‐verbstructureforsuchverbs:(175) Thelexicalsyntaxofdeadjectivalchange‐of‐stateverbs vP 3 (agent) v' 3 v √P 3 DP √ ! clear thetable NotethattheincorporationofcleardoesnotviolatetheHMC,astheDPisinthe
specifierof√P,andincorporationishead‐to‐headmovement.TheobjectDPisinwhatH&Kcallthe"innersubject"position,asitisthesubjectofasmallclausepredicate,"thetable(is)clear".
Inthesecases,themeasuring‐outiswithrespecttotheentirestatedenotedbythe
smallclause—theendstate.Whenthatstateisacheived,theaccomplishmentdenotedbythewholeconstructionisover.Notethatthewholeisconstructedfromtwoeventualities:theCAUSEevent(littlev),andtheENDSTATEevent(thesmallclause).ThishasthenicepropertyofcorrespondingtothesemanticdecompositionofaccomplishmentsproposedbyPustejovskyandothers.
Finally,noticethatitmustbeinherenttothenatureoftheserootsthattheyare
predicative—theyselectforasubjectargument,notforanobject.Theyarethenfundamentallystative,andneitherboundednorunbounded,addingtoourinventoryofroots:
(30) Athirdkindofroot
no complement complement bounded unbounded bounded unbounded
Event hop sleep kick push Thing foal drool N/A N/A State clear TBA (prepositions)
72
7.6 DenominalLocation/Locatumverbs(176) Thepiècederesistance:denominallocation/locatumverbs. Location:bag,bank,bottle,box,cage,can,corral,crate,floor(opponent),garage,
jail,kennel,package,pasture,pen,photograph,pocket,pot,shelve,ship(theoars),shoulder,tree.
Locatum:bandage,bar,bell,blindfold,bread,butter,clothe,curtain,dress,fund,gas,grease,harness,hook,house,ink,oil,paint,pepper,powder,saddle,salt,seed,shoe,spice,water,word.
Formoreverbsandsignificantdiscussion,seeKiparsky1997. Noticethattheobjectoftheseverbsmaymeasure‐out:(177) Measuring‐outwhilesaddling: a. Johnsaddledthehorse #for5minutes/in5minutes b. Sueboxedthecomputer #for5minutes/in5minutes c. Momblindfoldeda6‐year‐old #foraminute/inaminute. d. Johnsaddledhorses for5minutes/#in5minutes e. Sueboxedcomputers for5minutes/#in5minutes f. Momblindfoldedchildren for5minutes/#in5minutes.(178) Paraphrasehassameaktionsartproperties: a. Momfitthesix‐yearoldwithablindfold #for5minutes/in5minutes. b. Momfitchildrenwithablindfold for3hours/#in3hours.(179) AHale‐and‐Keyser‐stylestructuralproposal: vP 3 (Agent) v' 3 v PP 5 DP P' ! 3 thehorse P √ saddle Essentially, the proposal is that this, too, is a change of state verb. The PP is a small clause,
predicating something like "WITH SADDLE" of the inner subject, the horse. Little v corresponds to CAUSE, as in the deadjectival case, above.
Thesamestructureisproposedforbothlocationandlocatumverbs—thatis,although
in"saddlethehorse",thesaddleisbeingputonthehorse,butin"boxthecomputer",thecomputerisbeingputinthebox,theincorporatedthing(saddle,box)isalwaysthe
73
sisterofPbelowP'.We'llseebelowthatwhatmattersistheboundednessoftheincorporatedthing,notwhetherit'sthelocationorlocatum.
(180) Anothermeasurer‐outerintheparaphrases: a. Sueputthecomputerinboxes for5minutes/#in5minutes b. Suefitthehorsewithsaddles foranhour/#inanhour. Notethat,althoughpragmaticallyodd,manipulatingtheboundednessofthe
prepositionalobjectaffectstheaktionsartofthepredicate.Selectinganunboundedrootforincorporation,then,oughtequallytoaffecttheaktionsartofthepredicate,inawayparalleltothefoal/droolcontrastabove.
(181) Anunbounded,incorporatedLocatum: a. Susanwateredthegarden foranhour/inanhour b. Billgreasedthechain for5minutes/in5minutes c. Jillpaintedthewall foranhour/inanhour d. Adelaidebutteredthebread for2minutes/in2minutes Whilethetelicreadingisavailable,asexpectedgiventhemeasuring‐outpotentialofthe
definite,singularobjects("innersubjects"ofthechangeofstate),anatelicreadingisalsoavailable!Thisisverysurprising.Contrasttheseexampleswiththenecessarytelicityofaverblikesaddle(cf.31aabove).
Conclusion#2:Again,weattributetheintroducedatelicreadingintheparaphrasesin
(33)totheintroducedunboundednessoftheprepositionalobject.Similarly,wecanexplaintheavailableatelicityoftopaintincontrasttothenecessarytelicityoftosaddlebyattributingittotheunboundednessoftheincorporatedprepositionalobjectinpaint,vs.theboundednessoftheincorporatedprepositionalobjectinsaddle.
7.7 Derivingtelicity(182) Thetypologyofargumentstructures,sofara. vPwithnon‐branchingcomplement vP 3 (Agent) v' 3 v X foal,run,drool,dance,calve....b. vPwithbranchingcomplementlackingaspecifier vP 3 (Agent) v' 3
74
v XP 3 X YP push,kick,hit,kiss,pull...c. vPwithbranchingcomplementlackingacomplement(smallclause) vP 3 (Agent) v' 3 v XP 3 YP X clear,redden,clean,weaken...d. vPwithbranchingcomplementwithbothspecifierandcomplement(smallclause) vP 3 (Agent) v' 3 v XP 3 YP X' 3 X ZP saddle,box,water,paint,butter...(Also,withoutincorporationofZP,thisistheLarsonianframeworkforditransitiveverbs:give,send,put....(see,e.g.Harley1996fordiscussion).) Notethatthedistincitonbetweentype(b)and(c)abovecanbemadeonthebasisofthe
ontologicaltype(Statevs.Event)ofX:ifXisanEvent,itcannotbepredicatedofsomething
Assumption#1:Theaboverepresentalltheargumentstructuresavailableinlanguage:
maximumofthree"direct"arguments.Note:nomultiplespecifiersallowed!(183) Adifferentkindofdenominalverb:instrumentalactivities a. Johnhammeredthemetal for5minutes/in5minutes b. Suebrushedthedog for5minutes/in5minutes c. Jillrakedtheleaves foranhour/inanhour Noticethattheboundednessofthenominalrootherehasnoeffectontheavailable
atelicity.Thisisexpectedifthestructuralsourceofthesenominalrootsisnotoneofthepossiblemeasuring‐outincorporatingpositions(i.e.complementtovorcomplementtoP).Consideringtheincorporatednominalinthematicroleterms,thismakessense:theseincorporatednounsareneitherThemesnorLocation/Locatums,butratherInstruments.
75
Assumption#2:TheseareverbscreatedbyMannerIncorporation:namingaverbof
oneofthefourclassesabove((36b),verbsofcontact—push,kick,kiss,etc.)afterasalientaspectoftheMannerinwhichitisaccomplished.Thisconflatestheseverbswithothermanner‐of‐contactverbssuchaswipe,etc.
(184) Whathappenswhenyoutrytoincludeanendstateintheargumentstructureof
push?a. Johnpushedthecart JohnDO(a)PUSH(of)thecartb. JohnpushedthecarttoNewYork JohnCAUSE[thecarttoNewYork]byPUSH Allofasudden,there'snoroomforthePUSHeventnominalintheargumentstructure,
whichisnowsaturatedwithaStatecomplementtovP,completewithaninternalsubject(thecart)andapredicate(PNewYork).PushingisnowrelegatedtoamereMannerelement,whichgetsintotheverbby(tada!)MannerIncorporationon‐the‐fly.
(185) Sameproblemwithmanner‐of‐motionverbsa. Sueran. SueDO(a)RUNb. SuerantoNewYork SueCAUSE[(self)toNewYork]byRUNc. Thebulletwhistled ThebulletDO(a)WHISTLEd. Thebulletwhistledpastmyear BECOME[thebulletpastmyear]whileWHISTLE Whathappensisthatthe(36d)verbframeisbeingused,buttheverbisnamedaftera
mannerelementthatcanalsooccurasitsownverbrootinthe(36aorb)frames.(186) TheargumentstructureofpushthecarttoNewYork. vP 3 John v' 3 v PP 5 DP P' ! 3 the cart P DP | ! to New York Anotherwayofthinkingaboutit:considerGleitman'sexampleoftheindependent
meaningsuppliedbytheditransitiveframe.Ifyoutakeaverblikethink,whichusuallytakesonlyaCPorDPcomplement,andforceitintoaditransitiveframe—SuethoughtthebooktoMary—whatresultsisnotungrammaticality.Rather,weinterpretthinkingasamannerelementdescribingthewayinwhichthebookwastransferredtoMary(telepathicallyortelekinetically,probably).Cf.alsotheinsightsofconstructiongrammar:Goldberg1995.
pushing v°√pushv° ∅
76
(187) Innersubjectsmeasure‐out9 a. Johnpushedcartstothecloakroom for3hours/#in3hours b. Susanhammeredmetalflat for3hours/#in3hours Also,ofcourse,theauxiliaryselectionchangeinDutchresultsfromtheappearanceof
aninnersubjectandresultingavailabilityofanunaccusativestructurefortheverbofmotionjumpwhentheendstateofthejumperisspecified;similarly,theavailabilityofthere‐insertionwithverbsofmotionresultsfromtheappearanceofaninnersubjectandresultingavailabilityofanunaccusativestructurewhentheendstateisrepresented.
7.7.1 TheproductivityofMannerIncorporationvariesparametrically AsdemonstratedbyTalmy1986,verbsofmannerofmotionarenotmuchavailablein
Romancelanguages:(188) LackoflexicalMannerelementsinRomance: a. Thebottlefloatedawayfromthebank. b. Labotella sefué delaorilla flotando. thebottle REFLmoved‐away fromthebankfloating Similarly,resultativeconstructionsareunavailableinRomancelanguages,andmost
verbsofmotiondonotpermittheadditionofgoalPPsorthecausativeaccompaniedmotionconstruction(seeHarley1999;MateuFontanals2000forfurtherdiscussion):
(189) a.Thehorsejumped/Kayjumpedthehorseoverthefence. b.Elcaballobrincó/*Juanbrincóelcaballosobreelcerco. thehorsejumped/*Johnjumpedthehorseoverthefence. Conclusion#3:Ifweunderstandthatresultativeconstructionsandmotion‐to‐a‐goal
constructionsinvolveareanalysisoftheverbrootasaMannerelement,wecanattributetheabsenceofsuchconstructionsinRomancetothelackofproductivityofMannerIncorporationinthoselanguages.
7.8 Reprise:IncrementalThemes Above,theonlyclassesofverbsthatmeasureoutwiththeirdirectobjectarechange‐of‐
stateverbs,withargumentstructures(36c)and(36d)above,whosedirectobjectisaninnersubject.Verbswhosedirectobjectdoesnotaffecttheirtelicityonewayoranotherhavenoinnersubject(frames36aand36babove),exceptinonecase:verbsofmakingorunmaking.Thiswastheoriginalparallelthatledustowardstheideathatdecomposingverbsinthesyntaxmightbeausefulidea.TheverbsthattheyparalleledweretheveryIncrementalThemeverbsthatgotTennyandDowtygoinginthefirstplace:
9Thisistheanswertothe$64,000question:theobjectsofpushverbsarenotinnersubjects.
77
(190) So:whataboutthetelicityofverbswithincrementalthemes? a. Sueatetheapple #for5minutes/in5minutes b. Billbuiltthehouse #forayear/inayear. Justasditransitiveverbsparallellocation/locatumverbswithoutalltheincorporation,
Iwishtoclaimthatverbsofmakingandunmakingparalleltheverbsofbirthingwithoutalltheincorporation.TheverbrootwillbeanincorporatedMannerelement.Thestructureof,e.g.,writewillthenbe:
(191) Astructureforincrementalthemeverbs vP 3 John v' 3 v DP ! the book JohnMAKEthebookbyWRITE Thereisthenasignificantstructuraldifferencebetweentheobjectsthatmeasure‐outin
change‐of‐stateverbs(includingditransitiveverbs),andtheincrementalthemeobjects.Theformerare"innersubjects"ofasmallclause,thelatteraredirectobjectsofalightverbofcreation(ornegativecreation).
Atestwhichmaydistinguishthetwokindsofdirectobjects(may!)isthemiddle
construction.Certainlylocation/locatumverbstakethemiddleeasily...(192) Middlesandmeasuring‐out a. Thesecomputersboxeasily. b. Shetlandponiessaddleeasily. Butitseemsthatverbsofmakingandunmakingresistthemiddle:(193) a. ??Powerbarseatquickly. b. ??FrankLloydWrighthousesdon'tbuildeasily Tenny2000 c. ??Rodinstatuessculpteasily. d. ??Oxfordshirtsdon'tseweasily. andcertainlyverbsofbirthingdo:(194) a. ***Foalsofthistypehaveeasily. b. ??Foalsofthistypebirtheasily.
writing
78
AsTenny(2000)notes,however,itseemsthatsomeclassmemberscanoccurinthemiddle:
(195) Maybemiddlesaren'tthebesttest.... a. ...thesoupthateatslikeameal b. ?Yourinitialsembroidereasilycomparedtomine. However,considerthedifferencebetweenamiddleformedfrom(a)below,agenuine
verbofcreation,andamiddleformedfromthesameverbin(b)below,inachangeofstatefrane:
(196) Butmaybetheyare a. Mariacarvedatoysoldier. a' ??Toysoldierscarveeasily. b. Mariacarvedthewood. b'. Woodcarveseasily. Conclusion#4:Middleformation(may)onlyapplytoverbswhoseargumentstructure
containsaninnersubject.HaleandKeyser1999cometothesameconclusionlookingataverydifferentsetofdatafrompsychverbs.
7.9 Sowhichlightverbisit? Inmyparaphrases,intendedtoelucidatethelexicalsemanticsandlexicalsyntaxof
thesedifferenttypesofverbs,I'veusedseveraldifferentlightverbstocorrespondtothecontributionoflittlev:
(197) DO,CAUSE,andMAKE a. SusanDO(a)DANCE b. BillDO(a)PUSH(of)thecart. c. ThemareMAKE(a)FOAL. d. JenniferMAKEabook(by)WRITING e. JillCAUSEthetableCLEAR f. MariaCAUSEthehorseWITHSADDLE g. PattyCAUSEthecarttoNewYork(by)PUSHING Infact,Ithinkit'sthesamelittlevinallcases:onethatdenotesthebeginningofan
event,anditsinitiator.It'sjustaweaknessofEnglishthatthebeginningsofdifferentkindsofeventsarereferredtobydifferentverbs.WeMAKEThings,weDOEvents,andweCAUSEstates;theinterpretationiswhollydependentontheontologicaltypeofthecomplementtolittlev.InFrench,allthreeEnglishverbstranslatethesameway:faire.
Ididn'taddressthequestionofwhetherthere'salightverbinunaccusativephrasesor
whatitis;Iassumethereis,thatitdenotesthebeginningofaspontaneouschange‐of‐
79
stateevent,andthatitdiffersfromtheFAIRElittlevonlyinthatitdoesnotselectanexternalargumentinitsspecifier.
(198) BECOME a. BECOME[thedoorOPEN] b. BECOME[thescreenCLEAR] c. BECOME[thebulletpastmyear](while)WHISTLING
7.10 SomeConcludingThoughts Takehomemessages
a)Evidencethatroottypeaffectstelicityofunergativeverbsanddenominallocation/locatumverbsarguesforalexical‐syntaxapproachtoargumentstructure
b)APustejovsky‐stylesemanticsforaccomplishments—CAUSE+ENDSTATE—isdirectlyrepresentedintheirlexicalsyntax10.
c)ThefactthatEnglishallowsproductiveMannerIncorporationaccountsforcertaintransitivityalternationsandthemeasuring‐outeffectsthatgowiththem;itcanalsoexplainwhyRomancedoesn'tshowsuchalternations
Next:TheargumentstructureandderivationalmorphologyofHiaki(I'mgoingtoskip
thePersianpaper;it'squitestraightforwardandyou'veseensomeofthedatabefore
10NotethatthisentailsthatnomonomorphemicrootcannameanAccomplishment.
80
8. YaquiargumentstructureandderivationalmorphologyReadings:Tubino‐Blanco,M.andH.Harley.(Toappear).SobrelaopcionalidaddelCausadoenlas
causativasindirectasenyaquiIntheProceedingsofthe2007FriendsofUtoAztecanMeeting/Tallerdelosamigosdelenguasyutoaztecas,EditorialUnison,Hermosillo,Son.,Mexico
H.Harley.(Inprep).ApplicativesandVoicePinHiaki.Handoutfromtalkpresentedat2007
GLOWmeeting,Tromsø,Norway.Jelinek,E.andH.Harley(inprep)ImpersonalpassivesinYaqui.Ms.,UniversityofArizona.Harley,H.,M.Tubino‐BlancoandJ.Haugen.(2009).Applicativeconstructionsand
suppletiveverbsinHiaki(Yaqui).InL.Lanz,AFranklin,JHoecker,EGentryBrunner,MMorrison,andCPace,editors,RiceWorkingPapersinLinguisticsVolume1,pp.42‐51
M.Tubino‐Blanco,H.HarleyandJ.Haugen.(2009).Thesyntaxofhybridverb/affixlexemes
andclausefusioninHiaki(Yaqui)InL.Lanz,AFranklin,JHoecker,EGentryBrunner,MMorrison,andCPace,editors,RiceWorkingPapersinLinguisticsVolume1,pp.79‐91
8.1 AquickintroductiontoHiakimorphosyntax HiakiisanSOV,agglutinative,Uto‐AztecanlanguagespokeninSonoraandinArizona.It
isrobustlyhead‐final,exceptforitsrichdeterminersystem,whichishead‐initial.Ithastwostructuralcases—nominative(‐∅)andaccusative(‐ta)—andabevyof
postpositionsIthasaLOTofadicity‐affectingverbalsuffixes,includingapassive,twocausatives,two
desideratives,anapplicative,adirective,aquotative,participializers,andrelativizers
Thepassiveisanimpersonalpassive.Itcanapplytoanyverbwithahumansubject
argumenttoeliminatethatargument,providinginsteadanexistentiallyboundunderstood'someone'interpretationforthatrole.
Ifthereareanyaccusativecase‐markedargumentsinthecorrespondingactive,the
structurallyhighestonemovestosubjectpositionandbecomesnominative.
81
(199) a. Hose vachi‐ta bwasa’a Joe corn‐ACC cook.TRANS Joeiscookingcorn. b. Vachi bwasa’a‐wa Corn cook.TRANS‐EXST Thereiscornbeingcooked;(People,they)arecookingcorn.iftherearenosucharguments,theimpersonalpassiveissubjectless.(200) a. Ume yoemia aman yaha The.PLpeople there arrive.PL Thepeoplearearrivingthere. b. Amanyahi‐wa Therearrive.PL‐EXST (People,they)arearrivingthere;Arrivingthereisgoingon.HiakireflexivesarebeautifullyobedienttoConditionA.
8.2 Hiakicausatives:DirectandIndirect(WithMercedesTubino‐Blanco)
8.2.1 Introduction Inthispaper,wecontrastthebehaviorofthedirectcausativeaffixtuawiththatofthe
indirectcausativeaffixtevoinHiaki(Yaqui).Wefirstconsiderthestraightforwardcases,andproposeasyntacticanalysisbasedonthem.Wethengoontooutlinesomemorecomplexbehaviorwithtevoincombinationwithparticularverbstems,whichouraccountdoesnotpredict.
8.2.2 TheFacts TwocausativemorphemesinHiaki(201) Directcausation:thecausativesuffix–tuaisaffixedtotheverbalstem(1). a. Mariahitevi‐tauusi‐tahitto‐tua‐k Mariadoctor‐ACCchild‐ACCtreat‐CAUS.DIR‐PERF ‘Mariamadethedoctortreatthechild’ (‘María(le)mandóalmédicotrataralniño’) b. Aaposiyeeva‐vamih‐tua 3SGverypeopleRED‐in.a.hurry‐CAUS.DIR ‘He’salwaysmakingpeoplerush’ (‘Siemprehacecorreralagente’)
82
c. HosePeo‐talautiyevih‐tua‐k HosePeo‐ACCearlyarrive(SG.SUBJ)‐CAUS.DIR‐PERF ‘HosemadePeoarriveearly’(‘HosehizoaPedrollegartemprano’)(202) Indirectcausation:themorpheme–tevoissuffixedtotheverb(2). a. Mariauusi‐tahitto‐tevo‐k Mariachild‐ACCtreat‐CAUS.INDIR‐PERF ‘Mariahadthechildtreated’(‘Maríamandótrataralniño’) b. Aapohivava‐vamih‐tevo 3SGalwaysRED‐in.a.hurry‐CAUS.INDIR ‘He’salwayshaving(people)rush(‘Siemprehacecorrer’) c. Inepoamanyahi‐tevo‐k 1SGtherearrive(PL.SUBJ)‐CAUS.INDIR‐PERF ‘Ihadpeoplebroughtthere’(‘Hicetraeraunosahí,cf.hicequellegaranallí’) TheCauseeargument:In(1)and(2)acontrastisexhibitedbetweenthetwoaffixes
regardingthepresenceoftheCauseeargument,thesubjectoftheembeddedverb,e.g.hitevita,'doctor'.
(203) Directcausation(tua): TheCauseeargument,hitevita‘thedoctor’(‘elmédico’),isobligatory: *Maria[uusi‐tahitto]‐tua‐k Maria[child‐ACCtreat]‐CAUS.DIR‐PERF ‘Mariamade[(someone)treatthechild] (‘Maríamandó[(a?)trataralniño]) Indirectcausation(tevo):Dependingontheembeddedverb...(204) Causeeforbidden(moretypical): a.Santoh‐taachai[(*aayoemiawa)vachi‐taet]‐tevo Santoh‐ACCfather[(*hischildren‐POSS)corn‐ACCplant]‐CAUS.IND ‘Santos’fatherhad[hischildrenplantthecorn]' (‘ElpadredeSantosmandó[asushijossembrarelelote]’) b.Aapohiva(*yee)va‐vamih‐tevo 3SGalwayspeopleRED‐in.a.hurry‐CAUS.INDIR ‘He’salwayshaving[(*people)rush](‘Siempremandaquelagentecorra’)
83
(205) Causeeoptional(lesstypical;withonlytheseverbsinourdatasofar)a.Maria[hitevitauusi‐tahitto]‐tevo‐k [compare(2b)]Maria[doctor‐ACCchild‐ACCtreat]‐CAUS.INDIR‐PERF‘Mariahad[thechildtreatedbythedoctor]’ (‘Maríamandó[trataralniño]’(atravésdelmédico))b.Maalaaayevih‐tevo‐kukayoemta [compare(2c)]mother3SGarrive(SG.SUBJ)‐CAUS.INDIR‐PERFDET.ACCman‐ACC‘Motherhadthemanbroughttoherhouse’(‘Mamámandótraeralhombre(alacasa)’)
8.2.3 TheSyntaxoftheCausee:isitreallyoptional? Weproposethat,despitethepuzzlingfactsabove,theCauseeisreallysyntactically
suppressedwiththeindirectcausativetevo.Itsbehaviorinpassivesshowsit. Thesubjectofpassivizedcausatives:(206) Passiveofdirectcausation(tua):theCauseeispromotedtosubject(LikeJapanese): Uuhiteviuusi‐tahitto‐tua‐wa‐k DETdoctorchild‐ACCtreat‐CAUS.DIR‐PASS‐PERF ‘Thedoctorwasmadetotreatthechild’(‘Almédicoselemandótrataralniño’) Passiveofindirectcausation(tevo):theembeddedobjectispromotedtosubject,even
withverbswheretheCauseemayoptionallyappear(LikeItalianFP)(207) Uuuusihitto‐tevo‐wa‐k DETchildtreat‐CAUS.INDIR‐PASS‐PERF ‘Somebodyhadthechildtreated’(‘Sehamandadotrataralniño’)
8.2.4 SuppressionofCauseeswithembeddedintransitivenumber‐suppletivepredicates CertainintransitiveverbsexhibitnumbersuppletioninHiaki(8).(208) a. Uuuusiamanvuite DETchild(sg)thererun(SG.SUBJ) ‘Thechildisrunning’(‘Elniñocorre’)
PUZZLEN.1:Thefactsin3.1.showaclearcontrastinthebehaviorof–tuaand–tevocausatives:Onlyinthecaseof–tuadoestheCauseehitevi‘doctor’(médico)becomethesubjectofapassive.In–tevopassives,theembeddedobjectuusi‘child’(niño)istheargumentbecomingthesubject.Whycan–tevocausativesexhibitCauseesatallinsentenceslike(205)?
84
b.Umeuusimamantenne/*vuiteDET.PLchild‐PLthererun(PL.SUBJ)/*run(SG.SUBJ)‘Thechildrenarerunning’(‘Losniñoscorren’) Hiakiallowsimpersonalpassivizationofintransitives11:(209) Pahko‐poyi’i‐wa‐k ceremony‐PPdance‐PASS‐PERF ‘Peopledanced/Therewasdancingattheceremony’ Jelinek1997:181[7b] Ifthepassivizedintransitiveverbexhibitsnumbersuppletion(eg.vuite(sg.)/tenne(pl.)
‘run’(‘correr’)),the‘default’agreementwiththesuppressedimplicitsubjectisALWAYSplural:
(210) a. Amantenni‐wa thererun(PL.SUBJ)‐PASS ‘Peoplearerunning/there’srunningthere’ (‘Allísecorre/Haygentecorriendoallí’)COMPARE: b.*Amanvuiti‐wathererun(SG.SUBJ)‐PASS‘Peoplearerunning/there’srunningthere’ (‘Allísecorre/haygentecorriendoallí’) Incausativeenvironments,thesingularformvuite‘run(sg.subj)’(‘correr(suj.sg)’)may
appearembeddedwiththedirectcausative–tuawhenevertheCauseeissingular:(211) Heidiamanaa=vui‐vuititua Heidithere3SG=RED‐run(SG.SUBJ)‐CAUS.DIR 'Heidimakeshimdorunningthere’(‘Heidilehace(aél)correr/darcarrerasallí’) Incontrast,onlythepluralformtenne‘run(pl.subj)’(‘correr(suj.pl)’)isallowedas
embeddedtotheindirectcausative–tevo(withtheCauseeimplicit–i.e.thesubjectof'run'issuppressed,exactlyaswithwain(10a)above):
(12)a.Heidiamante‐tennitevoheidithereRED‐run(PL.SUBJ)‐CAUS.INDIR‘Heidihaspeopledorunningthere’(‘Heidihacecorrer/darcarrerasallí’)b.*HeidiamanvuititevoHeidithererun(SG.SUBJ)‐CAUS.INDIR‘Heidiishavingpeoplerunthere’(‘Heidiestáhaciendocorrerallí’)1211Itisimportanttonotethat,unlikeinEnglish,thepossibilityofspecifyingtheagentintheformofabyphraseisnotanoptioninHiaki(Escalante1990)
85
Conclusion:ThesupressedCauseeargumentwithtevoislikethesuppressed
Agent/Subjectargumentwithwa:semanticallyunspecified,notsingular(likeanimpersonal),andgenuinelysyntacticallyabsent.
8.2.5 ProposalandAnalysis WeproposeananalysiswithinthevP/VoicePframework,proposedinJelinek(1998)
(seealsoPylkkänen2002,Harley2007).SuchananalysisisalsocomparabletotheanalysisofindirectanddirectcausationinHindiproposedinRamchand(2008).
OneobviousapproachistocarryovertheFI/FPdistinction:tuacausativesbehavelike
FI,and–tevocausativesbehavelikeFP,withoutanembeddedsubject:(212) a.Mariahitevi‐tauusi‐tahitto‐tuaMariadoctor‐ACCchild‐ACCtreat‐CAUS.DIR‘Mariaismakingthedoctortreatthechild’ (‘Maríalehamandadoalmédicotrataralniño’)b.Mariauusi‐tahitto‐tevoMariachild‐ACCtreat‐CAUS.INDIR‘Mariaishavingthechildtreated’(‘Maríahamandadotrataralniño’)(213) a.Mariahitevi‐tauusi‐tahitto‐tuavP DP v' Maria vP vº -tua DP v caus.dir hitevi-ta ‘doctor’ VP vº Ø DP Vº uusi-ta hitto- ‘child’ ‘treat’
12Notethattheungrammaticalityofthissentenceisnotduetothelackofreduplicationoftheembeddedverb,as(i)shows:
(i) neeamantenni‐tevo1SGthererun(PL.SUBJ)‐CAUS.INDIR‘I’mhavingpeoplerun’(‘Estoyhaciendocorrer(aunos)’)
(18a):THECAUSATIVEHEAD–tuaTAKESAvPASITSCOMPLEMENT.THISHEADNTRODUCESTHECAUSEEINITSSPECPOSITION[Spec,vP].
MAINDIFFERENCEBETWEEN(18a)AND(18b):ABSENCEOFAvPINCOMPLEMENTOFTEVO(18b).vPISTHEPROJECTIONTHATINTRODUCESTHEEMBEDDEDEXTERNALARGUMENT
86
b. Maria uusi-ta hitto-tevo vP DP v' Maria VP vº -tevo DP vº caus.indir uusi-ta hitto- ‘child’ ‘treat’
8.2.6 Summaryuptothispoint (214) Proposal:tuaand–tevoarecausativelightverbswhichdifferregardingthetypeof
clauseeachembeds: a.tevoselectsforasubjectless,VPcomplementclause. b.–tuaselectsforasubject‐fulvPcomplementclause. Thiscontrastisresponsibleforthe 1)obligatorypresenceoftheCauseein–tuacausatives,hitevita‘doctor’in(3) 2)theobligatoryabsenceoftheCauseein–tevocausatives(4) TheCauseein–tevocausatives,althoughnotpresentsyntactically,issemantically
understoodtobepresent,itsexistencebeingimpliedbythecontext.Interestingly,theexamplesexhibitingnumbersuppletionin(12)suggestthatthisnotional/semanticCauseeinindirect–tevocausativesismorpho‐syntacticallysalientenoughtoforcenumbersuppletion.Thisisparalleledbythepassiveexamplesin(6)and(7),wherenumbersuppletionistriggeredbytheimpersonal,implicitAgentargument,whichissyntacticallyabsentinpassiveconstructions.
8.2.7 Addenda,counterexamplesandunsolvedquestions Problemsunsolved!Ourdatashowspuzzleswestilldon’thaveananswerfor. OptionalityofCauseewithhittoandyevih(mentionedabove) Weclaimthatthesyntaxof–tevo(cf.itslackofVP)doesnot‘makeroom’foraCausee
argumentwithinitsdomain;hence,theabsenceofanovertCauseeinindirectcausatives.
However:Westillneedtoprovideanexplanationforsentenceslikethese(repeated
fromabove)
(18b):THECAUSATIVEVERB–tevoTAKESVPASITSCOMPLEMENT.THISVPDOESNOT‘MAKEROOM’FORACAUSEEARGUMENT,GIVENITSABSENCEOFASPEC‐VPPOSITION.
87
(215) Maria[hitevitauusi‐tahitto]‐tevo‐k Maria[doctor‐ACCchild‐ACCtreat]‐CAUS.INDIR‐PERF ‘Mariahad[thechildtreatedbythedoctor]’ (‘Maríamandó[trataralniño]’(atravésdelmédico)) WhyistheCauseeabletoappearinthissentence? Indirectcausationandembeddedsubjectofyepsa/yevih/yahi,'arrive':apuzzle:A
similarbutmorecomplexproblemariseswithyepsa,'arrive',incombinationwithtevo. Asnotedabove,incombinationwiththeunaccusativeverbyepsa‘arrive’(‘llegar’),tevo
causativesmayretainthesoleargumentoftheembeddedverbasanobject(justaswithtua).
(216) Maalaaayevih‐tevo‐kukayoemta mother3SGarrive(SG.SUBJ)‐CAUS.INDIR‐PERFDET.ACCman‐ACC ‘Motherhadthemanbroughttoherhouse(eg.bysomebody)’ (‘Mamámandótraeralhombre(alacasa)’)(217) HosePeo‐talautiyevih‐tua‐k HosePeo‐ACCearlyarrive(SG.SUBJ)‐CAUS.DIR‐PERF‘HosemadePeoarriveearly’(‘HosehizoaPedrollegartemprano’) However,thereisasubtlesemanticdifferencebetweentheyevihtevoandyevihtua
combinations:Theformer,butnotthelater,impliestheexistenceofanunspecifiedargumentwhichcausedtheembedded'arriving'event(asreflectedinthegloss).
Thisislikeother–tevosentencesinthattheembedded,caused,eventisunderstoodto
involveanunspecifiedCausee Unlikeothertevosentences,theunspecifiedCauseeisnotasuppressedpartofthe
argumentstructureoftheembeddedverb(ie.,it’smerelyimplicitfromthecontext). Theargumentstructureofthisembeddedverbmaybefullypresentwithtevo. Itdoesn'thavetobepresent,however—itcanbehaveasexpected,wherethesingle
argumentoftheembeddedverbissuppressed(asalsoshownabovein(2c).(20)Inepoamanyahi‐tevo‐k1SGtherearrive(PL.SUBJ)‐CAUS.INDIR‐PERF‘Ihadpeoplebroughtthere’(‘Hicetraeraunosahí,cf.hicequellegaranallí’) Whentheembeddedargumentisabsentwiththisverb(20),weseepluralagreement,
asexpected,cf.12aabove:
88
Thesingularagreementwiththeunusualsentencein(18)((5b)),withtheovertCausee,islikethesingularagreementinthecorrespondingtuaconstruction
Intermsofagreement,then(18)and(19)arebehavingexactlyalike:tevoisactingjust
liketua,morphosyntactically.Semantically,however,it'sdifferent—thetevosentenceinvolvesanimplicitembeddedexternalargument
Justtocomplicatemattersfurther,inanotherexamplewithpeculiarcombinationof
yepsaandtevo,theredoesn'tseemtobeanyimpliedargumentatall—inthiscase,itseemsthattevocanfunctionexactlylikethedirectcausativetua:
(218) a.Kaaro‐mhivasivu‐vu’uria.Kiali’ikunvato’oraatakaalauticar‐PLalwaysveryRED‐multiplythat.is.whyin.baptism‐AVRZ‐ACCNEGearlyyahi‐tevo‐k/yahituak arrive(PL.SUBJ)CAUS.INDIR‐PERF/arrive(PL.SUBJ)‐CAUS.DIR‐PERF‘Thereisalwaysalotoftraffic.Thatcausedpeopletoarrivelate’(‘Comohabíamuchoscoches,sehizollegartarde(alagente)’) Passiveinterpretationwithne’e‘fly’(‘volar’):Someintransitiveswithinthescopeof
–tevo(eg.ne’e‘fly’)both(i)allowanovertCauseeand(ii)receiveapassiveinterpretation.Thefactsareshownin(22).
(219) a.Uuuusiam=ni’i‐tua‐kDETchild3PL=fly‐CAUS.DIR‐PERF‘Thechildmadethemfly’(‘Elniñoloshizovolar’)b.Uuuusiwikichimni’i‐tevo‐kDETchildbird‐PLfly‐CAUS.INDIR‐PERF‘Thechildhadthebirdsfly?’(‘Elniñomandóquevolaraelpájaro’)c.Wikichi‐mni’i‐tevo‐kbird‐PLfly‐CAUS.INDIR‐PERF‘Thebirdswereallowedtofly’(‘Dejaronvolaralospájaros’) Tentativesolution?Theaccusativeargumentsappearingin–tevo+intransitive
sentencessuchas(22b,c)aregeneratedasinternalargumentswithinthedomainoftheintransitiveverb,ratherthanasrealCausees(i.e..,intheSpecpositionofvP,which,weclaim,isabsentwithinthedomainof–tevo).(Thiswouldunifythesene'ecasesandtheyepsa/yevih/yahacasesabove,assumingyepsa,'arrive',isunaccusative.)Findingsontheinteractionbetweenintransitivesuppletiveverbsandtheapplicativemorpheme–ria,inHarleyetal.(2006),providefurthersupportregardingthispossibility
Onwards!Totheinteractionofcausativesandapplicatives,andthemotivationofPylkkanen's(2002)VoiceP+vP+VParchitecture!
THEUSUALSTRUCTURE:tua+intransitive+accusativeCausee
THEPUZZLINGSTRUCTURE:tevo+intransitive+…accusativeCauseee?
PUZZLING!–tevo+passiveinterpretation(butnopassivemorpheme–wa)‐‐again,likethereisanimplicitCauseeargument
89
8.3 Applicatives,causativesandtheMirrorPrinciple:vPexternalsubjectsandVoice
8.3.1 IntroductionFollowingworkbyHaleandKeyser(1993),Chomsky(1995)andKratzer(1994,1996),andfollowinggroundworklaidbyLarson(1988),theVPcametohavetwoparts,anupperexternal‐argumentselectingpart,andalower,internal‐argumentselectingpart:vPandVP.(220) TP T' T° vP DP v' v° VP V DP ‐ed Mary ∅ open it IhavethoughtitwaspossibletogetawaywithjustthismuchVPdecomposition(assumingcertainmodificationstodowiththecategoryoftheVPconstituent)Inthistalk,however,Ipresentevidencethatmorestructureisneeded,namely,atleastaVoicePontopofanexternal‐argument‐lessvP: TP T' T° VoiceP DP Voice' Voice° vP v SC DP √ ‐ed Mary ∅ ∅ it open
90
Structureoftalk:ReviewMcGinnis'sandPylkkanen'sanalysisforapplicativestructureswithinthesplit‐vP,
showinghowitprovidesstrongevidenceforthesplit‐vPproposal.IntroduceHiakiverbalmorphologyandsyntax,includinglexicalcausatives,productive
causativesandapplicativesShowhowinteractionofproductivecausativesandapplicativesarguesforstructurein0,
includingVoiceP.ThekeyargumentfortheseparationofVoiceandvisaminorvariationontheargumentpresentedbyPylkkänen2002:122‐125.
Salientpoint:thecausativev°headdoesnotintroducetheovertexternalargumentinHiakicausatives.Rather,theexternalargumentmustbeintroducedbyaVoice°headwhichselectsforthecausativevP
IntroduceHiakiindirectcausativesandshowhowVoicePisniceforthemtoo.SkipoversimilarargumentfromEnglishnominalizationsduetoBorer(2005)Discussconsequencesforbase‐generationofexternalarguments,stackingpassive
morphology,unaccusatives,etc.
8.3.2 McGinnis(2001etseq,),Pylkkanen(2002):ApplicativesandthesplitVPOnemajorbenefitofthesplit‐vP,inamorphologicallydecompositionalsyntax,isanappropriatespottoputtheapplicativemorphemeanditsargument,inlanguageswhichhaveapplicatives(Pylkkanen2002).ApplicativesintroduceanadditionalbenefacteeormalefacteeargumenttoaclauseVeryproductive,morphologicallyovertinmanylanguages.CanonicalexamplesincludeChichewaandKinyarwanda;Hiaki(Yaqui)alsohasaproductiveapplicativemorpheme:13(221) Voocha‐m woita shoe‐pl untie "Untietheshoes"(222) Usi‐ta voocha‐m woita‐ria Child‐acc shoe‐pl untie‐appl "Untietheshoesforthechild!"(223) InepoHose‐ta livrom hinu‐ria‐k IJose‐accbook bought‐appl‐prf "IboughtJoseabook"13Asidefortheapplicativelyinclined:ThefactthattheHiakiapplicativeriacanattachtounergativeverbsasin(225)meansthatitisa'high'applicativeinPylkkanen(2002)'sterms.Indeed,Harley,HaugenandTubino(2006)showthatitcannotattachtounaccusativeverbs,suggestingitselecstsforanagentivevPcomplement.ThefactthattheHiakiapplicativeisstronglyasymmetricinitsinteractionwiththepassiveandtheotherargumentsoftheclause,however,isevidencethatitisnotaphasehead,inMcGinnis'(2001,2003)approachtosymmetricapplicatives.Thissuggeststhatthehigh/lowapplicativedistinctiondoesnotnecessarilycorrelatewiththesymmetric/asymmetricdistinction,contratheproposalofMcGinnis(2001).
91
(224) InepoHose‐tapueta‐ta eta‐ria‐k I Jose‐accdoor‐acc close‐appl‐prf "IclosedthedoorforJose"(225) U’umaaso uusi‐m yi’i‐ria‐k Thedeer.dancer thechildren‐pl dance‐appl‐prf “Thedeerdancerdancedforthechildren”SalientfeaturesoftheHiakiapplicative: (I) Introducesanewargumentintosentence (II) Newargumentisasyntacticobject(withtaaccusativesuffixine.g.(224)) (III) Newargumentisc‐commandedbysubject,c‐commandsallinternal
arguments(Rude1996)Proving(III):Whencoreferentialwiththematrixsubject,theappliedargumentmustbereflexive:(226) Aapo tu’i mo’ove‐ta au= hinu‐ria‐k He goodhat‐acc 3.refl=bought “Heboughthimselfagoodhat.” Dedrick&Casad1999:343[17]Whentheexternalargumentissuppressed(impersonal)passivesuffix,theapplied
argument,nottheinternalargument,becomesthenewnominativesubjectargument:
(227) Hose maso‐ta me'a‐k Jose deer‐acc kill‐prf "Josekilledthedeer"(228) Maasome'e‐wa‐k Deer kill‐pass‐prf "Thedeerwaskilled"(229) Hose Maria‐ta maso‐ta me’e‐ria‐k Jose Maria‐acc deer‐acc kill‐appl‐prf “JosekilledMaria'sdeeronher."("Maria"ismalefacteeargumentofria)(230) Mariamaso‐ta me’e‐ria‐wa‐k Mariadeer‐acc kill‐appl‐pass‐prf “Mariahadherdeerkilledonher”(231) *MasoMaria‐ta me’e‐ria‐wa‐k deer Maria‐acc kill‐appl‐pass‐prf
92
So:Appliedargumentappears(structurally)betweenexternalargumentandinternalargument.Itisintroducedbytheriamorphemeitself,presumablytheheadofanApplP.AppliedargumentinSpec‐ApplP.Split‐VPallowsustocomposetheVwiththeAppl°riabelowthev°andabovetheVP,therebyintroducingtheappliedargumentbelowtheexternalargumentandabovetheinternalarguments,asrequired.(232) TP DP+nom T' vP T°+nom t v' ApplP v°+acc DP+acc ApplP' √P Appl°+acc √° Uumaaso uuuusim yi'i ‐ria ‐∅ ‐k Thedeer.dancer thechildPL dance ‐APPL ‐v° ‐PRF
"Thedeerdancerdancedforthechildren."
93
(233) TP DP+nom T' vP T°+nom t v' ApplP v°+acc DP+acc ApplP' √P Appl°+acc DP+acc √° Inepo Joseta pueta‐ta eta ‐ria ‐∅ ‐k I Jose‐ACC door‐ACC open ‐APPL ‐v° ‐PRF
"IopenedthedoorforJose."ThisisPylkkänen'sargumentforasplit‐VP(fromapplicativeswiththesesame
propertiesinotherlanguages,notinHiaki.)
8.3.3 Hiakiandthesplit‐VPBipartiteverbmorphologyinHiaki:Hiakiexhibitsafairamountofmorphological
evidence,likethatofJapanese,thatsomeagentiveverbsaremadeupofarootandacausativeverbalizinghead.
(234) Change‐of‐statepredicates: Verb Adj Verb Adj toredden red sikisi siki tofatten fat awia awi tosoften soft bwalkote bwalko tosharpen sharp bwawite bwawi towarm warm sukawe sukaAgentivedenominalunergativepredicates:(235) Noun Verb cho'oko 'salt' cho'okote 'tosalt' heewi 'yes!' hewite 'toagree' hiosia 'paper' hiohte 'towrite' (sh/__C) haawa 'steam' hawassate 'tosteam'
94
hipetam 'mat,bed' hipette 'toweave,makeamat' kari 'house' karite 'buildahouse'Bipartiteverbstructurecopeswiththesemorphologicaldivisionsverynicely:thetein
thingslikekarite'house‐do',forexample,couldbeanalyzedasheadingtheexternal‐argument‐introducingv°,asinJapanese,above:
(236) vP Agent v' √P v° √ Santos kari‐ ‐te house‐ ‐do "Santosbuildsahouse" Now,considerthemorpheme‐orderpredictionmadebythesplit‐VPanalysisofapplicatives,above.TheapplicativeheadriaintroducesanappliedobjectbelowtheexternalAgentargument,betweenv°and√.IfriaobeystheMirrorPrincipleprediction,then,itshouldfitlinearlyinbetweenkariandte,asinthefollowingstructure:(237) vP Agent v' ApplP v° Benef. Appl' √P Appl° *Santos Mariata kari‐ ria ‐te house‐ Appl ‐do Thismorphemeorderisimpossible,though.Theapplicativemorphememustfollowthepurportedv°morpheme!
95
(238) SantosMaria‐ta kari‐te‐ria (> Santoskari‐te) SantosMaria‐acc house‐v°‐appl Santoshouse‐v° "SantosisbuildingahouseforMaria." "Santosisbuildingahouse"Hmm!Possibleconsclusion:apparentv°morphologyliketeisn'tactuallyv°,thoughit
mayhistoricallyhavebeen;hasbecomereanalyzedaspartofroot(hence'lexical').Actualv°ismorphologicallynull,andstill'follows'theAppl°head.
Notaviablesolutionforthenextsetofdata:theinteractionofproductivecausativesand
applicativesinHiaki.
8.3.4 SplittingthevP:Voice,causativesandapplicativesCausativesBesideshavingaproductiveapplicativeconstruction,Hiakihasabeautifullyproductiveaffixalcausativeforexpressinggarden‐varietycausation:(239) Art Heidi‐ta utte’a‐po hipaksia‐tua‐k “Art Heidi‐acc strength‐in do.laundry‐caus‐prf “ArtmadeHeididolaundryagainstherwill.”(240) SimonHose‐ta kari‐te‐tua SimonJose‐acc house‐vrb‐caus "SimonismakingJosebuildahouse."(241) Uumaroma umeyoeme‐m kari‐ta ho'o‐'oota‐ka‐m‐ta Theforeman theman‐pl house‐acc back‐bone‐ppl‐s.rel‐acc hoo‐tua‐taite make‐cause‐start "Theforemanismakingthemenstarttobuildahousewithapeakedroof" (Lit:"…ahousethathasabackbone.")Inlinewithagreatdealofresearchonaffixalcausativescross‐linguistically(Harley
1995,Kural1995,Travis2000,Svenonius2005,amongmanyothers),IanalyzetheaffixalcausativemorphemeasaniterationofvP
ThecausativevPaddsanexternalargument(theCauser)tothesyntaxandacausative
morphemetotheverb.Theformerexternalargument(theCausee)becomesthehighestinternalaccusativeargument,gettingaccusativecaseviaECMfromthecausativev°
96
(242) vP …T° DP v' √P v° √ Hose kari ‐te ‐k Jose house ‐make ‐prf "Hosebuiltahouse."(243) vP T° DP v' vP v° DP v' √P v° √ Simon Hose‐ta kari ‐te ‐tua ‐k Simon Jose‐acc house ‐make ‐cause ‐prf "SimonmadeHosebuildahouse."Categorizingthecausativemorphemeasanotherv°ismotivatedbybothsemanticandmorphologicalfactors: (i) Thecausativemorphememeanssomethinglike‘cause’,andforchange ofstatepredicates(Johnopenedthedooretc.)that’ssupposedtobe themeaningofthev°thatintroducestheexternalargument (ii) Forsomelanguages,certainregularexternal‐argument‐introducingv°sare actuallymorphologicallyrealizedas‘lexicalized’versionsofthe causativemorpheme:JapaneseandalsoHiakihavecertainverbslikethis:(244) Mala pale‐ta amanvittua‐la diachronically: vit‐tua Motheryoung.boy‐acctheresend‐compl see‐cause “Mothersenttheyoungboythere” (245) Itepo am hiapsi‐tua‐ne we them heart‐caus‐fut “Wewillfeedthem/careforthem/givethemstrength.” (lit:"causethemtohaveheart")
97
AlsoinMolinaetal.dictionary:alleatua,‘amuse’,asoatua,‘tobirth’,asukatua,‘sweeten’,chihtua,‘nurse(tr.),eetua,‘bother,annoy’,hioptua,‘confuse’,mahatua,‘frighten’,na’atua‘tellonsb.’,puhtua,‘givesb.theevileye’…
Inthesecases,tuahasbeenlexicalizedastheexternal‐argument‐introducingmorphemeforaverblikesend(asforclarifyinEnglish)—thestructureisnodifferentfromthatofabasicverb.ThereareseveralsimilarexamplesinJapanese,wherealexicalcausativeisformedwiththedefault‘syntactic’causativemorpheme,sase(Miyagawa1984,1998))ThisisjustmotivationforarguingthatthevPinaregularlexicalcausativeverbandthevPheadedbytheproductivecausativemorphemehavesomethingincommon,namelytheirdefaultrealizationastua;ergothey'rebothv°s…Crucially,though,theproductivecausativemorphemevPcanalsonotbepartoftheverb'slexicalentry—it'sproductive,applyingtoessentiallyanyverb,includingnovelandborrowedones…Nowaytoclaimthatit'slexicalized/reanalyzedaspartoftheverbstem!
8.3.5 CausativesWITHapplicativesSo:CausativemorphemesintroduceanewexternalargumentBenefactivemorphemesintroduceanewinternalargument,betweentheexternal
argumentandtheverb—crucially,c‐commandedbytheexternalargument.Howdotheyinteract?Exampleofacausativesentence:(246) a. Uuavion ne'e‐k Theplane fly‐prf "Theplaneflew" b. Nee ukaavion‐ta ni'i‐tua‐k I the.accplane‐acc fly‐caus‐prf "Imadethe(model)airplanefly."
98
(247) vP DP v' vP v° DP v' √P v° √ Nee avion‐ta ni’i ∅ ‐tua ‐k I plane‐acc fly v° ‐cause ‐prf "Imadetheplanefly."WhatifIwantedtosay,“Imadethemodelplaneflyforthechild”—makeabenefactiveoutofthiscausativesentence?Introducedapplicativeargument(‘thechild’)appearsbelowtheexternalargument—henceshouldbebelow‘I’Morphemeordershouldmirrorsyntacticstructure,soiftheapplicativeargumentisbelowthecauserargument,theapplicativemorphemeshoulddefinitelybeinsidethecausativemorpheme.Butit’snot:(248) Neeiliusi‐ta avion‐ta ni'i‐tua‐ria‐k Ilittlechild plane‐acc fly‐caus‐appl‐prf "Imadethe(model)planeflyforthechild."MorphemeorderistellingusthattheriaphraseisabovethetuaphraseArgumenthierarchytestsaretellingusthattheargumentoftuaisabovetheargumentofriaCauserc‐commandsbenefacteeNote:Thisisnotacausativeofabenefactiveoffly.Adirectbenefactiveoftheembeddedclausehere,"theplaneflew"isimpossible,becauseplanesarenotintentional—intentionalityisnecessarytolicensebenefactive:(249) #Uuavion iliusi‐ta ni'i‐ria‐k Theplane littlechild‐acc fly‐appl‐prf "Theplaneflewforthechild."
99
Crucially,causativesofbenefactivesarepossible,withtheoppositemorphemeorder,whenthesemanticsareright.Thisscopally‐motivatedreorderingisimpossibleinotherlanguageswithbothkindsofsuffixes(Hyman2003,BuellandSy2004)(250) Nee iliusi‐ta mala‐ta aa tu'ute‐ria‐tua‐k I littlechild‐acc mother‐acc it clean‐appl‐caus‐prf "Imadethechildcleanitformother."Herewehaveabenefactivesentence—'Thechildcleaneditformother'whichiscausativized(Imade[thechildcleanitformother]).Themorphemeorderreflectsthisderivationalhistory.Theargumentorderdoesnot.
8.3.6 AnalyticaloptionsA:Morphemeorder=syntacticstructure,syntaxmessyOptionA: Morphemeorderreflects‘true’syntacticstructure,BenefacteegeneratedaboveCauser,Causermovesaround/acrossbenefacteetoSpec‐TPsubjectpositionApplicativeofcausativeexampleagain:(251) Nee iliusi‐ta avion‐ta ni'i‐tua‐ria‐k I littlechild plane‐acc fly‐caus‐appl‐prf "Imadetheplaneflyforthechild."Ifexternalargumentisselectedbytua,‘cause’,andbenefactiveargumentisselectedbyria,‘appl’,thenunderlyingstructureislikethis:(252) ApplP DP Appl’ vP Appl° DP v’ vP v° DP v’ √P v° √ iliusita nee avion‐ta ni’i ∅ ‐tua ria ‐k littlechild I plane‐acc fly‐ v°‐ caus‐appl ‐prf
100
Wouldhavetobetransformedintoactualclausestructurebymovementofnee,‘I’,tospec‐TP,acrosstheappliedargumentiliuusita,'littechid'. Problems: Minimalityviolation:WhyisCauserabletomoveacrossBenefactee? (can’tbebecauseBenefacteeisinert/hasinherentcase—inpassive, Benefacteegetsnominative) Actuallycancausativizeapplicativeclauseswithdifferentscope—inthat case,orderappl‐causegivesahigherCauseethanBenefactee.Analysis wouldrequireCauseetomovearoundBenefacteeforaccusativeECM fromthecausativemorpheme,ashereCausermovesaroundB.for nominative. Loseargumentfromapplicativesforseparatingexternalargumentfromverb (sinceapplicativeargumentwouldbebase‐generatedoutside externalargument.)Recall:inimpersonalpassiveofthissamesentence,theBenefacteegetsnominativecase,
andtheCausermaynotappear,showingthattheBenefacteeisnotinertforA‐mvtanddoesnotbearaninherentcase
(253) Iliuusi avion‐ta ni'i‐tua‐ria‐wa‐k littlechild plane‐acc fly‐caus‐appl‐imp‐prf "They/Someonemadetheplaneflyforthelittlechild." [Closesttranslation,structurallyspeaking: “Thelittlechildwasflowntheplane(bysomeone).”]B:Argumentccommandstructure=syntacticstructure,morphologymessy.OptionB: Argumenthierarchyreflects‘true’syntacticstructure,BenefacteegeneratedbelowCauser,benefactivemorphemegeneratedinsidecausativemorpheme.Idiosyncratictemplaticrestrictionscausetheorder‐riatuatoberewrittenastuaria. (cf.Hyman2003)
101
(254) vP DP v’ ApplP v° DP Appl’ vP Appl° DP v’ √P v° √ Nee iliusi‐ta avion‐ta ni’i‐ ‐∅‐ ‐ria‐ ‐tua I littlechild‐accplane‐acc fly v° Appl ‐cause +SomeMorphologicalMessiness (likeinHyman2003) Nee iliusi‐ta avion‐ta ni’i‐‐∅‐ ‐tua‐ ‐ria I littlechild‐acc plane‐acc fly v° cause appl Problems: Orderriatuaisfinewhenriaisanapplicativeoftheembeddedverb— templaticrestrictionswouldhavetotakeintoaccountscopeofria Noevidenceforamorphemictemplatefromelsewhereinthelg—allthe
verbalderivationalaffixesarebeautifullyproductivew/rtoeachotherbasedonlyonsemanticandsyntacticrestrictions…
Morphologicalmessinesswouldhavetobesensitivetoscopeofria,since,asnoted
above,riatuaisawell‐formedsequence—it'sfineasacausativeofanapplicative:(255) Nee iliusi‐ta mala‐ta aa tu'ute‐ria‐tua‐k I littlechild‐acc mother‐acc it clean‐appl‐caus‐prf "Imadethechildcleanitformother."Plusitjustseemswrong:thefirstisanapplicativeofacausative,thesecondacausative
ofanapplicative;surelyonewouldwantthefactthatthemorphemeorderreflectsthistofalloutoftheanalysis,notbeanuglykluge.
OptionC:LexicalismOptionC: Presyntacticwordformation.Theseformsarederivedinthelexicon.Applyingabenefactivetoacausativeaugmentsthetheta‐gridoftheverbinthelexiconwithaBenefactee/Goalargument;generallinkingruleprinciplespredictthatwhenthe
102
complexverbisprojectedtothesyntax,theCauserargumentwillbeprojectedtothehighestposition,theBenefacteetothenext‐highestposition. ‐‐Problems:
Theoretical:Buildingwordsinthesyntaxhasbeenanattractiveandusefulidea,I’mnotreadytogiveuponit.
Empirical:QuestionabouthowtocopewithrelativepositionofBenefacteeandembeddedCausee.LinkingrulesshouldpredicttheembeddedCauseeshouldbehigherthantheBenefacteetoo.Bracketsintheta‐grid?
OptionD:SplittingvP:Voiceforexternalargument,v°forCause,ApplPandBenefacteein
between:OptionD: Theexternalargumentisnotintroducedbythecausativemorphemeatall,
butbyahigherVoicemorpheme.Thecausativemorphemejustintroducestheideaofcausation—itdoesthesemanticandmorphologicaljob,butnotthesyntacticone
(256) VoiceP DP Voice’ ApplP Voice° DP Appl’ vP Appl° VoiceP v° DP Voice’ vP Voice° √ v° Nee iliusi‐ta avion‐ta ni’i∅ ∅ ‐tua ‐ria ‐∅ I littlechildplane‐acc fly‐vDOVoice caus ‐appl Voice Notetheextra'Voice'headintheembeddedvPaswell—giventhemorphemeorder
betweenapplicativesandlexicalcausativeslikekarite,'build.house'andvittua'send',embeddedexternalargumentsarealsointroducedbyVoice°ratherthanbyv°.
Problem:EmbeddedCausersbutnoembeddedpassiveVoicemorphologyallowed?.But:
tevo!
103
VerymuchlikeHaleandKeyser’sdescriptionoftheirorginalidea:theexternal
argumentjustgetsinsertedfor‘free’,inawaycompletelythematicallyunrelatedtoembeddedargumentstructure(thoughnotwhattheywouldhavewantedmorphologically)
8.3.7 AnothermorphologicalreasonforsplittingvPandVoiceP:IndirectcausativesBesidesthetua'direct'causative,wheretheCauseemustbeexpressed(likeRomance
faireinfinitif),Hiakialsohasaproductive'indirect'causative,tevo(likeRomancefairepar),wheretheCauseeisnecessariysurpressed;theembeddedverbgetsa'passive'or'impersonal'reading(withoutanypassiveorimpersonalvoicemorphology)
(257) InepoSantoh‐ta hitto‐tevo‐k I Santos‐acc medical.care‐caus.indir‐prf “IhadSantostreated(bysb.)”Inthesetevocases,whenyoupassivize,theembeddedobjectbecomesthederivedsubject,showingthattheCauseristrulysyntacticallyabsent:(258) Santoshitto‐tevo‐wa‐k Santosmedical.care‐caus.indir‐pass‐prf"(Somebody)got(somebody)totreatSantos"(lit,"Santoswasgottentobetreated.")Thiscannaturallybeaccountedforiftevoselectsaconstituentasacomplementwhich
doesnotcontaintheexternal‐argument‐selectinghead(see,e.g.FolliandHarley(2005)'streatmentoffaireparcausatives,orRamchand(2006)onHindiindirectcausatives).
However,tevoisabletoattachtoaconstituentwhichcontainscausativev°
morphology,eventhoughtheCausee(associatedwiththatcausativemorphology)isnotpresent14
(259) 'Lexical'v°morphology
Nee Santoh‐ta‐u hiohteita vittua‐tevo‐k I Santos‐acc‐toletter‐acc send‐caus.indir‐prf “IhadalettersenttoSantos.”(implied:bysomebody)(260) Productivecausativemorphology,causativeargumentsuppressedbytevo Nee ukaavion‐ta ni'i‐tua‐tevo‐k I theplane‐acc fly‐caus‐caus.indir‐k "Ihad(somebody)flytheplane."14Recall,vittua,'send',ismadeupofrootvit‐'see'plustua'cause';weassumethetuaisarealizationofv°morphology.
104
CausativemorphologycanthusbepresentwithoutintroducingaCausee.ThisisnaturalifvPcanbepresentwithoutVoiceP.Thiswillresultinthedesired'passive'readingfortheembeddedclause,expressinganexternallycausedeventbutwithoutspecifyingwho/whatcausedit.Thestructureisgivenbelow:(261) VoiceP … T° DP Voice’ vPCaus.indir Voice° vPCaus v°Caus.indir VoiceP v°Caus DP vP √P v° √ Nee avion‐ta ni’i ∅ ‐tua ‐tevo ‐∅ ‐k I plane‐acc fly‐ v°‐ ‐caus ‐caus.indir Voice Sotheverbalmorphologyisbuiltup,withinternalargumentadditionsifnecessary,
beforeanyexternalargumentisintroduced.Notethatthisentailsthatthedifferencebetween'lexical'and'syntactic'affixal
causativesisactuallynotaboutpresenceorabsenceofanextravPbetweenthecausativev°andtherestoftheverb,butratherpresenceorabsenceofaninterveningVoicePbetweenthecausativev°andtherestoftheverb.
AnothermorphologicalreasonforsplittingvPandVoiceP: Englishnominalizations(Borer2003)Englisheventnominalizationsdonotcontainatrueexternalargument(Chomsky1970,Marantz1997,HarleyandNoyer2000):(262) a. Johngrewtomatoes b. #John'sgrowthoftomatoes c. Thetelevisionamusedthechildren. d. #Thetelevision'samusementofthechildren e. AdulteryseparatedJimandTammyFayeBakker. f. #Adultery'sseparationofJimandTammyFayeBakker
105
Marantz'sconclusion:Thenominalizingheadselectsforaverb‐phraseconstituentwhichdoesnothavetheexternal‐argument‐selectingheadpresent.In1997,heassumedthiswasvP.
However,morphemeswhichclearlyhaveaverbalizingandapparentlycausativizingeffectexistinEnglish:(263) a.horrify,gratify,justify,certify,specify,vilify,simplify,passify,objectify b.deafen,dishearten,dampen,sadden,neaten,coarsen, c.categorize,terrorize,alphabetize,categorize,customize,digitize,idolize d.complicate,calculate,commemorate,pollinate,decorate,regulate,disambiguateThesewouldbenaturalcandidatesforrealizationsofav°head(especiallyinDM,whererootsareacategorialandv°istheverbalizer).However,thesecanperfectlywellappearinsideaneventnominalization,ofcourse:(264) a. gratificationofdesire,justificationofactions,certificationofresults…. b. deafeningofachild,dishearteningoftheteam,dampeningofthecloth… c. categorizationofnames,alphabetizationofnames,customizationofwheels… d. complicationofmatters,calculationofsums,pollinationofflowers…Conclusion:v°presentwithinnominalizations,butexternal‐argument‐introducingheadisnotv°andexternal‐argument‐introducinghead(VoiceP)areseparate(seeBorer2003andHarley2006forfurtherdiscussionofthemorphologicalimplications.)
8.3.8 AnothermorphologicalreasonforsplittingVoicePandvP: Stackingpassivemorphologyontopofcausativemorphology
Voice°wouldbethelocusoftraditionalvoicemorphology—passivemorphemesandsuch.Allowsniceaccountof'stacking'passivemorphologyoutsidecausativemorphology.Ifexternalargumentsareintroducedbythesameheadthatintroducesthecausativemorphology,hardtoseehowtoimplementthesuppressionoftheexternalargumentwithoutsuppressingthecausativemorphology.Ifexternalargumentintroducedseparately,byVoice,however,thenpassiveVoicemorphologycanembedacausativev°withoutentailingthepresenceofanexternalargument(265) Hiroko pizza‐o tabe‐sase‐rare‐ta (Japanese) H. pizza‐acc eat‐cause‐pass‐past "Hirokowasmadetoeatpizza."(266) Aapo kaa yo'o‐taka kuna‐tua‐wa‐k (Hiaki) 3sg not old‐being marry.f‐caus‐pass‐prf "Shewasmadetomarry/marriedoffwhenshewasn'tveryold."
106
b. Ume yoeme(m) hi’ibwa‐tua‐wa The.plman‐(pl) eat‐caus‐pass “Themenarebeingfed”Noteaswellofcoursethatpassivewaorrareappliesto'lexical'causativesinbothHiakiandJapanesewithoutaffectingtheirovertv°morphology:(267) Aman kari‐te‐wa (Hiaki) There house‐v°‐pass "Housesarebeingbuiltthere."Notealsothatonthisview,passivedoesn'thavetobeconnectedtoanyparticularsemanticsassociatedwithvarious'flavors'ofv°:ExternalargumentsneednotbeAgents;passivizedverbsneednotbeagentive(VoicecapturesWilliams'generalizationasdiscussedbyCollins2005).(268) a. Johnwasseen(byMary). b. Suewasloved(byJohn)Merchant(2007)comestoverysimilarconclusionsbasedongrammaticalityofVP‐
ellipsiswithVoicemismatchinantecedentandelidedVP;conclusionisthatVoicePisnotelided;ratheritscomplementis.Onthepresentaccount,thecomplementofVoicePisidenticalinactiveandpassive(atleastinJapaneseandHiaki),henceonecanbeagoodantecedentfortheother.
(269) a. Thejanitormustremovethetrashwheneveritisapparentthatitshouldbe. b. Thesystemcanbeusedbyanyonewhowantsto. (Merchant2007:3)NotethatonCollins(2005)andMerchant(2007)'svP+VoicePapproach,theexternal
argumentisgeneratedinspec‐vP,andmovesuptoanemptyspec‐VoiceP.Thiscan'tbethecase,giventheseHiakidata:Theexternalargumenthastobebase‐generatedhigherthanthebenefactee,which(accordingtothemorphemeorder)isbase‐generatedoutsidethevP.Pylkkanen'sapproach,wheretheheadintroducingcausationandtheoneintroducingexternalargumentsareindependent,hastobeclosertoright.
8.3.9 VoiceP,passivesandunaccusativesClear(becauseofapplicatives)thatovertexternalargumentcan'tbebase‐generatedinspec‐vP.Intransitives,however,vPintroducescausative/transitivesemantics,includingavariablefortheexternalargumentwhichisboundbytheDPintroducedinspec‐VoiceP
107
Thisvariable,Iassume,iscovert,notsyntacticallypresent,andisboundbytheDPinspec‐VoicePviawhatWurmbrand(2000)calls'semanticcontrol',sinceexhaustive.(SeeWurmbrandfortestsdistinguishingsyntacticcontrolofasyntacticallypresentnullelementlikePROandsemanticcontrol,establishedonlyatLFviavariablebinding.)This'semanticcontrol'approachtoactivetransitivesalsoavoidspotentiallocalityproblemsassociatedwithhavinganullDPpresentinspec‐vP.15ActiveVoicerequiresExternalMergeofaDP;assumingexternalMergeisafeature‐checkingoperation,alongthelinesofAdger(2003),activeVoicemusthaveanexternal‐Merge‐triggeringfeature.Non‐activeVoicewouldthenbecaseswheretheVoiceheadrefusesExternalMergeofanexternalargument—relevantfeaturesabsent.PassiveinterpretationfollowswhenthevariableintroducedbythevPisexistentiallybound—causative/transitivemeaningpresent(thankstovP)butnoDPtobindthevariable.InternalargumentmovesoutofVoicePtosubjectposition.IfVoicePisaphase,itmustdosoviaSpec‐VoiceP—passiveVoicePmusthaveanEPPfeaturetotriggerinternalMerge.ThedifferencebetweenactiveandpassiveVoicewouldthenbethedifferencebetweeninternalandexternalMergefeatures.Whathappensinanunaccusative?InHiakiandJapanese,(andEnglish)VoiceisActive,
yet(bytheusualhypothesis)noexternalargumentisMerged.(270) Yoemem aman koko‐k People there die.pl‐prf "Peoplediedthere"Theusualhypothesis,then,entailsthatinunaccusatives,theexternalMergefeaturesof
Voicemustbeabsent.Again,bytheusualhypothesis,theinternalargumentispromoted,sotheinternal‐MergetriggeringEPPfeaturemustbepresent.Thatis,ontheusualhypothesis,wecannotdistinguishbetweenunaccusativeVoiceandpassiveVoice.
Insomelanguages,orforsomesubsetofunaccusativesinsomelanguages,it'snot
desirabletodistinguish—bothunaccusativeandpassiveVoicearerealizedbythesame'non‐active'Voicemorphology(seeKallulionAlbanian,e.g.)
InHiakiandJapanese(andEnglish)however,unaccusativestakeaclearlyactiveVoice,notpassiveVoice
15Though:Howcanweensurethatthevariablemustbeboundbywhateverisinspec‐VoiceP,orgetexistentiallyboundifVoiceisPassive,evenwhenthere'sacloserpotentialbindersittinginSpec‐ApplPinanapplicative?
108
PossibilityA:PassiveVoicemorphologyintheselanguagesissensitivetothefeaturesof
v°aswellasofVoice—insertedonlywhenVoicehasinternalMergefeatureandv°hasthecausative/external‐argument‐takingsemantics.ActiveVoicemorphologyisthedefault,applyingelsewhere—whenVoicehasanexternalMergefeatureandalsowhenithastheEPPfeaturebutthev°isinchoativev°BECOME.
PossibilityB:VoicePisabsentinunaccusativesintheselanguages.(Thiswillposeno
morphologicalproblems,sincevPwillstillbethere).PossibilityC,maybethemostinterestingone:VoicePisthere,buttheusualhypothesisaboutunaccusativityiswrong.Unaccusativeargumentsarebase‐generatedinspec‐VoicePlikeactivetransitivearguments.NotethatinHiaki,thereissuchathingasapassiveofanunaccusative:(271) Amankoko‐wa‐k Theredie.pl‐pass‐prf “Peoplediedthere/Itwasdiedthere(bypeople)” ThiswouldseemtosuggestthatinactiveunaccusativesinHiaki,theusualprocessof
externalMergeofaDPargumentinspec‐VoicePoccurs,whichcanbepreventedintheusualwaybyanimpersonalpassiveVoice.
Whatabouttheinternalargument,theargumentofthedownstairs√?PerhapsthistakesusbacktoanaccountofunaccusativityinwhichanullDP(behaving
likeaboundvariable)ismergedinobjectpositionandboundbytheexternally‐MergedDPinspec‐VoicePinactiveunaccusatives—thatis,backtoatreatmentinwhichactiveunaccusativesarereflexive.
Thedifferencebetweenthiskindofactive‐voicereflexiveunaccusativeandregular
transitivereflexivesinthiscasewouldresideindifferencesinv°semantics—thedifferencebetween(272)aandbwouldsimplybethatin(a),v°isinchoativeandin(b),v°iscausative.Inthelatter,theexternalargumentinVoicePsemanticallycontrolstheAgentvariableintroducedbyvCAUSaswellasbindingtheobjectDP,whileintheformer,theexternalargumentinVoicePsimplybindstheobjectDP;nosemanticcontrolisestablishedbecausethev°doesn'tintroduceavariable.
(272) a. Jean s'‐est evanoui. John SE‐is fainted "Johnfainted." b. Jean s'‐est tué. John SE‐is killed. "Johnkilledhimself."
109
IfpossibilityCiscorrectforsomecases,suggeststhattherearedifferentkindsof
unaccusativitycross‐lingustically—possibilityAforsome.Alsodifferentkindsofpassives—stackingvs.participle‐forming,e.g.
Voice v° √Transitives Active,Spec
filledbyexternalMerge
introducesvariableboundviasemanticcontrolfromspec‐VoiceP
selectsobjectDP
UnaccusativeA1('externalreflexive'‐Romance;McGinnis1997)
Active,SpecfilledbyexternalMerge
novariable selectsobjectDPcoindexedwithDPinspec‐Voice.Oneoftheseisrealizedasreflexive(external);objDPbecomessubject.
UnaccusativeA2(active‐English,Hiaki)
Active,Specfilledbyexternalmerge
novariable selectsnullobjectDP,coindexedwithspec‐VoiceP
UnaccusativeB(Albanian,Greek)
Nonactive,spec‐VoicePempty
novariable objectDPultimatelybecomessubject
Passive Nonactive,spec‐VoicePempty
variableboundbyexistentialclosure
objectDPundergoesinternalMerge,ultimatelybecomessubject
ConsistentwithprimingdatafromEnglishandSpanishthatsuggeststhatEnglish
passivesbutnotunaccusativesprimeatrace‐typerepresentationinobjectposition,whileSpanishpassivesand(reflexive)unaccusativesbothprimesucharepresentation(SanzandBever1997).
ConclusionsHeadcontainingcausativemorphologyandsemantics(vP)isindependentofexternal‐
argument‐introducinghead(VoiceP).******Repriseofearlierpaperoncausativemorphology:Thecorrectstructures!***** Theinteractionbetweentuaand–tevoItispossibleinHiakiforbothcausative
verbs–tuaand–tevotoappeartogetherinthesamesentence.Ifthisisthecase,thesyntaxoftheindirectcausative–tevosuppressestheCauserargumentof–tua.Thecontrastisshownin(19)andtheanalysisisshownin(20).
(273) a. Neeukaavion‐tani’i‐tua 1SG DET plane-ACC fly-CAUS.DIR ‘I’m making the plane fly’ (‘Estoy haciendo volar al avión’)
110
COMPARE WITH: b. Nee uka avion-ta ni’i-tua-tevo 1SG DET plane-ACC fly-CAUS.DIR-CAUS.INDIR ‘I’m having (somebody) make the plane fly’ (‘He mandado hacer que el avión vuele16) [Adapted from Harley (2007)] (274) a. Neeukaavion‐tani’i‐tua VoiceP DP Voice’ Nee vP Voiceº Ø VoiceP vº Voice -tua DP Voice’ caus.dir uka avion-ta ‘the plane’ vP Voiceº Ø vº Voice ni’i- ‘fly’ (275) Neeukaavion‐tani’i‐tua‐tevo VoiceP DP Voice’ Nee vPcaus.indir Voiceº Ø vPCaus.dir vº Voice -tevo VoiceP vº caus.indir -tua DP Voice’ avion-ta vP Voiceº Ø vº Voice ni’i ‘fly’
16TheinfinitivalversionofthisindirectcausativeisnotpossibletoproduceinMT'sSpanish.
(20a):–tuaTAKESAVoicePASITSCOMPLEMENT,WHICH,INTURN,INTRODUCESTHECAUSEEukaavionta‘theplane’.INADDITION,THEEXTERNALARGUMENTOFtuaISINTRODUCEDBYASEPARATE,MATRIXVOICEP.
(20b):–tuaTAKESVoicePASITSCOMPLEMENT,WHICH,INTURN,INTRODUCESTHEEMBEDDEDEXTERNALARGUMENTukaavionta‘theplane’.HOWEVER,–tevoTAKESONLYAVPASITSCOMPLEMENT.HENCE,THEEXTERNALARGUMENTOFTUAISABSENTFROMTHECOMPLEXCLAUSE
111
9. TakeHomeMessages Significanttheoreticalimprovementoverbaggageoflexicon,linkingoperations,theta
roles,lexicalconceptualstructures,etc. LexicalsemanticprimitiveslikeCAUSE,DO,Appl,etc.arecomposedwitheachotherin
thesyntax Theyareresponsiblefortheapearanceorabsenceofexternalargumentsand
applicativearguments Givesgoodinsightintothesyntacticbehaviorofarangeofphenomenainseveral
languages Noneedformorphologyandsyntaxtocompetewitheachothertodothesamejob—
theyaretreatedbythesamesystem Animacyeffects,hierarchiesofarguments,etc.canbetreatedwellinthiskindofa
system Stilllotsofpuzzlesremaining!