Deponency in the Mirror A Unied Approach to Deponents and Unaccusatives - PDF document

DeponencyintheMirror:AUniedApproachtoDeponentsandUnaccusativesPhilippWeisserUniversityofLeipzigphilipp.weisser@uni-leipzig.de 0.ClaimDeponencyandunaccusativityexhibitamirroredbehaviourinallmodulesofthegrammarandarethustobeanalysedastwoinstancesofthesamephenomenon.Thisanalysiscapturesthismirrorimageelegantlybyinvokingtwoprinciples:IdentityAvoidance&LexicalOverride1.Introduction(1)Deponencyisamismatchbetweenformandfunction.Giventhatthereisaformalmorphologicaloppositionbetweenactiveandpassivethatisthenormalrealisa-tionofthecorrespondingfunctionalopposition,deponentsarealexicallyspeciedsetofverbswhosepassiveformsfunctionasactives.Thenormalfunctionisnolongeravailable(Baerman2007).(1)capturescasesofcanonicaldeponencylikethemismatchinLatinverbinection.Othercasesofdeponencymaydifferinallofthementionedpropertiesbuttherstone('mismatchbetweenformandfunction').ThepresenttalkwilldealwiththecaseofcanonicaldeponencyinLatin.2.Thedata-PropertiesofdeponentverbsMorphology:passiveThemorphologyofdeponentverbsisalwaysidenticalwithtothemorphologyofregularverbsinpassivevoice.ThisidentityextendstoallpossiblecombinationsofÁ-,tense-,aspect-,andmood-features.(2)amare-'love'(regular)auxiliari-'help'(deponent)PassiveActive1.SG.PRES.INDam-orauxili-or2.SG.PRES.INDam-¯arisauxili-¯aris3.SG.PRES.INDam-¯aturauxili-¯atur1.SG.PERF.INDam-¯atussumauxili-¯atussum2.SG.PERF.INDam-¯atusesauxili-¯atuses3.SG.PERF.INDam-¯atusestauxili-¯atusest3.SG.FUT.INDama-bi-turauxilia-bi-tur3.SG.PRES.SUBJam-e-turauxilietur1 Semantics:activeThesemanticsofdeponentverbsisexactlythesameaswith'normal'transitiveverbsinactivevoice.Manydeponentverbshavesubjectsandobjects(andhencecannotbeunderlyinglypassivi-zed).Deponentverbsfurthermoredonotconstituteaspecialsemanticclassofverbs(andhencecannotbesomekindofunderlyingpsych-verbs)(c.f.Xuetal.2007)(3)CethegusCethegusCiceronisCicero-GENianuamdoor-ACCobsideretbeset-IMPERF-SUBJ-3SGeum=quehim-ACC=andviviolentlyaggredereturattack-IMPERF-SUBJ-3SG'CetheguswastobesetCicero'sdoorandassaulthim'(Embick2000)Syntax:activewithregardto:Case,Agreement,etc.-passivewithrespectto:PeriphrasisThesyntaxofdeponentverbsdoesnotbehaveconsistently.Withregardtocaseassignment,agree-ment,numberofpossiblearguments,etc.,itbehaveslikeanactivetransitiveverb.(4)Puerboy-NOMmilitemsoldier-ACCsequi-turfollow-PASS.3.SG'Theboyisfollowingthesoldier'(Embick2000)Undertheassumptionthatperiphrasisisasyntacticphenomenon(cf.Embick2000),however,thesyntaxbehaveslikeitdoeswithapassivetransitiveverb.(5)a.Via-mway-ACCsecutusfollow.PTCPsum.be.1.SG'Ifollowedtheway.'b.Satisenoughsumbe.1.SGverberatus.beat.PTCP'Iwasbeatenenough(times)'(MacciusPlautus,5.1)c.Domin-usMaster-NOMverbera-v-itbeat-PERF-3.SGserv-um.servant-ACC.'Themasterbeattheservant.'Thedeponentverb'sequi-'follow(Examplea)choosesanalyticverbforminperfectiveaspect,justlikeapassivizedtransitivverb(Exampleb).Anon-deponentverbhoweverchoosesasyntheticforminperfectiveaspect(Ex.c)Exceptions:Ahandfullofdeponentverbsmaynotonlyoccurinclauseswithactivesyntax/semanticsbutalsoinpassives.Thisisalexicalexceptionrestrictedtoafewdeponentverbs.(6)Abbyamicisfriendshorta-re-tururge-IMPERF.SUBJ-PASS.3SG'Hewasurgedbyfriends'(subjunctive)(Embick2000)Therearenomorphologicalexceptions.Anactiveverbformofadeponentverblike'horto'or'auxilio'areungrammatical.Deponentverbscanneverbecombinedwithactivemorphology.2 (7)Propertiesofdeponentverbs: Morphology Passive Syntax Periphrasis Passive Case Active Semantics Active Exceptions Morphology None Syntax Somelexicalexceptions Semantics Somelexicalexceptions PropertiesofunaccusativeverbsMorphology:activeThemorphologyofunaccusativeverbsisundeniablyactive.ThisappliestoGerman,EnglishaswellasLatininallpossiblecombinationsofÁ-,tense-,aspect-,undmoodfeatures.(8)amo-'love'(active)madesco-'getwet'1.SG.PRESam-omadesc-o2.SG.PRESam-¯asmadesc-¯as3.SG.PRESam-¯atmadesc-¯at1.SG.PERFam-¯avimadesc-¯avi2.SG.PERFam-¯avistimadesc-¯avisti3.SG.PERFam-¯avitmadesc-¯avit3.SG.FUT.INDama-bi-tmadesca-bi-t3.SG.PRES.SUBJam-e-tmadesc-e-tSyntax:passivewithrespectto:Case,Agreement,etc.-activewithrespectto:PeriphrasisThesyntaxofunaccusativeverbsisinconsistenteither.WithrespecttoCase,Agreement,etc.itbehavesasifitwaspassive(thecomplementoftheverbisraisedtosubjectpositionreceivingnominativecaseandtriggeringactivemorphology)(c.f.Perlmutter(1978),Levin&Rappaport-Hovav(1995),Alexiadouetal.(2004)).(9)CastorCastor.NOMmadesc-¯atget.wet-PRES.3.SG'Castorisgettingwet'(10)TP NP{case:NOM}T' T{case:NOM}vP vpassiveVP VtNP3 Howeveranunaccusativeverbbehaveslikeatransitiveactiveverbwhenitcomestothequestionofwhethertochooseaanalyticorsyntheticverbforminperfecttense:(11)a.FiliussonquiwhoininMarathoniaMarathonpugnabattlececiditfall.PERF.3.SG...'ThesonwhofellinthebattleofMarathon...'(Cicero,LetterstoAtticus)b.Satisenoughsumbe.1.SGverberatus.beat.PTCP'Iwasbeatenenough(times)'(MacciusPlautus,5.1)c.DominusMaster.NOMverberavitbeat.PERF.3.SGservum.servant.ACC.'Themasterbeattheservant.'Semantics:passive(?)Thesemanticsofunaccusativeverbsisacontroversialtopic.Byandlarge,everyoneagreesthataclausewithanunaccusativepredicateresemblesaclausewithapassivizedtransitiveverb.Itcontainsthesametheta-rolesandthesamedependenciesbetweenitsverbanditsarguments.Basicallyforthesereasonsunaccusativeandpassivizedverbsaresubsumedunderthelabel'non-activevoice-head'byKratzer(1996).Thisvoice-headservesasadefaultchoicewhennoneofthenumerousactivevoiceheadscanbeinsertedintothestructure.However,Embick(2000,2004)arguesthattheremustbesomefurtherdifferencetoaccountforthesyntacticdifferencesbetweenbothtypesofpredicates.(12)a.Theboatsank(*bythecaptain)b.Theboatwassunk(bythecaptain)Apassivizedpredicatecanlicenseanagentinanadjunctphrasewhereasanunaccusativecannot.Furthermore,theimpliciteagentcanlicenseaPRO-argument.(13)a.Theboatsank(*PROtocollecttheinsurancemoney)b.Theboatwassunk(PROtocollecttheinsurancemoney)Exceptions:Therearesomeexceptionswithunaccusativeverbsaswell.Althoughtherearenomorphologicalexceptions,someunaccusativeverbscanundergocausativealternation.Inthatcase,anunaccu-sativeverbbehaveslikea'normal'transitiveverb.However,thisalternationislexicallyrestrictedandisapplicableonlytoaxedsetsofunaccusativeverbs(seee.g.Kalluli(2006)onanticausati-ves).(14)*madesc-or(get.wet-PASS.1.SG.PRES)(15)a.Thevasebroke.b.Johnbrokethevase.(16)a.Johnfell.b.*Johnfellthevase.c.*ThewindfellJohn.4 (17)Propertiesofunaccusativeverbs: Morphology Active Syntax Periphrasis Active Case Passive Semantics Passive(?) Exceptions Morphology None Syntax Somelexicalexceptions Semantics Somelexicalexceptions Comparisonofthepropertiesofunaccusativeanddeponentverbs:Acomparisonbetweenthepropertiesofthesetwoverbtypesillustratesthesimilarities.Whene-veroneofthembehaveslikeatransitiveverbinpassivevoice,theotherbehaveslikeanactiveverbandviceversa.Insyntax,thepatternisinconsistentbutalsothisinconsistenceappliestobothtypesofverbs.Eventheexceptionsseemtopatternalike.(18)Comparison DeponentVerbs UnaccusativeVerbs Morphology Passive Active Syntax Periphrasis Passive Active Case Active Passive Semantics Active Passive(?) Exceptions Morphology None None Syntax lexicalexceptions lexicalexceptions Semantics lexicalexceptions lexicalexceptions 3.Hypothesis(19) Mirror Image Hypothesis:Deponencyandunaccusativityarebothinstantiationsofthesameabstractphenomenon.Thus,agrammaticalanalysismusttreatbothtypesofverbsidenticallyineveryrespect.4.AnalysisAssumptions:1.Lexicalprespecication:Someverbsmaybeinherentlybearafeature[§active]whencomingfromthelexicon.Deponentverbsbear[–active],unaccusativeverbsbear[+active].(cf.Embick2000).'Normal'transitiveverbsremainunspecied.(20)Vunacc:{V,+active,²NP²,}Vdep:{V,–active,²NP²,}Vtrans:{V,²NP²,}5 2.Lightverbs:ThereisanadditionalvP-shellabovetheVP(cf.Kratzer(1994),Chomsky(1995))whichcomesintwotypes:vactiveandvpassive.Thebearthesamefeatures[§active]:(21)vpass:{v,–active,²VP²,}vactive:{v,+active,²VP²,²NP²,*Á*:_,case:acc}3.IdentityAvoidance:Theonlyconstraintthatrestrictsthecombinationofv-headsandV-headsisthefollowing:(22)IdentityAvoidancePrinciple:*[X®active,®active]Itmayapplyafterhead-movementofVtov.IfthecomplexheadV-vcontainsthesamefeaturetwice,thederivationwillcrash.Thisyieldsthefollowingresults:(23)DeponentV{...[–Active]...}+Passivev{...[–Active]...})ruledoutDeponentV{...[–Active]...}+Activev{...[+Active]...})okUnaccusativeV{...[+Active]...}+Passivev{...[–Active]...})okUnaccusativeV{...[+Active]...}+Activev{...[+Active]...})ruledoutRegularV{...[]...}+Passivev{...[–Active]...})okRegularV{...[]...}+Activev{...[+Active]...})okUnaccusativeverbsmustnotoccurwithanactivev-headanddeponentsmustnotoccurwithpassivesyntax.Allothercombinationsareallowed.4.LexicalOverride:Thephonologicalrealisationofvtakesplaceonthebasisofitsfeature[§active].[+active]onvprovidesactivemorphology,[–active]providespassivemorphology.However,ifconictingfeaturesarepresentonthesamehead,itisassumedthatinherentfeatures(thosefeaturesthatcomefromthelexicalV-head)prevail.(24)vP vpassive+Vunacc{...[–active],...,[+active]...}VP tVunaccNP,activerealisation(25)v' vactive+Vdep{...[+active],...,[–active]...}VP tVdepNP,passiverealisationSamplederivationfordeponentverbsa)[VPV[¡active],NP[case:_]]]MergeofV&NPb)[v0v[Åactive],[VPV[¡¡active],NPcase:acc]]MergeofvandVP)CheckingofÁ-featuresandcaseassignmentc)[v0v+V[[Åactive],[¡active]],[VPtVP,NPcase:acc]]HeadmovementofVtovd)[vP[v0NP,v+V[[Åactive],[¡active]],VP]]Mergeofv'andNP6 Samplederivationforunaccusativeverbsa)[VPV[Åactive],NP[case:_]]]MergeofV&NPb)[vPv[¡active],[VPV[Åactive],NP]]MergeofvandVPc)[vPv+V[[¡active],[Åactive]],[VPtVP,NP]]HeadmovementofVtovHowdoesthesystemworkforlanguageswithoutdeponentverbs?Languageswithoutdeponentbutwithunaccusativeverbsmaybederivedeasilyundertheass-umptionthatthefeaturewhichencodesthealternationbetweenpassiveandactivevoiceisnotabinaryfeature[§active]butratheraprivativefeature[active].Allotherassumptionscanbetransferredwithoutfurtheradaptions.Indoingso,verbscannotbespeciedasdeponentinthelexiconbecausethereisnolabelavailableforthesecases.Unaccusativeverbscanstillbelabelled[active]andotherotherverbsremainunspecied:schlagen[V,²NP²]transitivefallen[V,²NP²,active]unaccusativeThev-headsneedtobeadjustedaccordinglysothattheconstraintin(22)onlyexcludesthecom-binationofanunaccusativeverbwithanactivev-head.vpassive:{v,²VP²}vactive:{v,active,²VP²,²NP²,*Á*:_,case:akk}5.Empiricalpredictionsandopenquestions5.1DeponencyandUnaccusativityshouldexcludeeachotherThewholetheoryisbasedonthehypothesisin(19).If,however,deponencyandunaccusativityaretwosidesofthesamecoin,theyshouldexcludeeachother.Averbcouldnotbeunaccusativeanddeponentatthesametime.Thetheorycapturesthisfactbytheassumptionthattherespectiveverbsarelexicallyspeciedas[–active]or[+active].And,ofcourse,averbcannotbespeciedforbothfeaturesatthesametime.Thus,thefollowingpredictionismade:(26)Prediction:Averbcannotbeunaccusativeanddeponentatthesametime.Undernormalcircumstancesthiswouldbeafalsiablehypothesis.Ifadeponentverbwouldpasstestsforunaccusativity,thiswouldbeamajorsetbackforthetheory.Unfortunately,itappearstobevirtuallyimpossibletondsuchtestsforunaccusativityforLatin.Mostoftheclassictests(auxiliaryselection,n-clitisation,impersonalpassives)arenotapplicableinLatin,others(prenominalparticiples)mightbeapplicablebutfailbecauseofthelackofdata.Alexiadou&Anagnostopoulou(1999)trytoestablishnewtestsforGreekbuteventhesetestscannotbeusedinLatin,mainlyforlackofdata.7 5.2Whyaretheresofewlanguageswithdeponency?Theanswertothatquestionisstronglyrelatedtoanindependentfactorofthesyntaxofthelanguage,namelythequestionofwhetherpassivevoiceisexpressedinananalyticorinasyn-theticconstruction.InLatin,apassiveverbis(usually)syntheticwhereasinallofitsderivativelanguagesitisanalytic.Iwillarguethatapassivethatisformedanalyticallythroughoutthewholelanguageisincompa-tiblewiththeconceptofdeponentverbs.Assumption:Analyticpassivesarederivedbyanadditionalpassivephrase(PassP)whereassyntheticpassiveslackthisprojection.(cf.Cinque1999,Adger2003,Collins2005)(27)T' T+Pass{–active}PassP tPass vP v+Vdep{...,[–active]...}VP tVdepNP,PassiverealisationIfthewholerealisationofvoice-featuresisamatterofthepassivephrase,thenthelexicalspe-cicationofaverbplaysnoroleatallbecausethefeaturesoftheverbandthoseofthepassiveheadareneverpartofthesamefeaturebundle.Thus,thelexicalfeaturescanneveroverridethefeaturesofthepassiveheadandthushavenosyntacticorphonologicaleffect.Andfeaturesthathavenoeffectontheoutputareoftensaidtoberemovedforthesakeoflexicon/inputoptimisati-on(Prince&Smolensky(1993)).Thuswecanstateanotherhypothesis:(28)Hypothesis:Alanguagewithanalyticpassivethroughoutallparadigmscannotmaintainaclassofdeponentverbs.1.Totestthesehypothesis,onemayhavealookatvariouslanguageswhichseemtohavecasesofdeponency.IntheRomancelanguages,aswellasGermanandEnglish,nocasesofcanonicaldeponencyareattested.However,wendsomeEuropeanlanguageswhichseemthesupportthehypothesisabove:•Greek(modernaswellasclassical)(Lavidas&Papangeli(2007))(29)itheegkiospregnant.NOM.SGligureftikedesire.PAST.3SG.MPASSpagotoice-cream.ACC.SG'Thepregnantwomandesiredicecream'•Swedish(Ritte(2004))(30)HanHeminna-sremember-PASSmigme.AKKfrånfromnärwhenviweträffade-smeet.PAST-RECIPpåinHultsfredH.'HeremembersmefromwhenwemetinHultsfred.' 1Ofcourse,theterm'deponency'herereferstodeponencyinthecanonicalsense(i.e.deponentw.r.t.thedistinctionbetweenactiveandpassivevoice)8 •Sanskrit(Stump(2007))•Finnish(Buchholz(2005))(31)Me1.PLmennäängo.PASSelokuviin.cinema.ILLATIVE'Wegotothecinema'6.ConclusionThistalkpursuedtwointerrelatedgoals:•Toshowthatthemismatchbetweenmorphologyandsyntaxfoundwithdeponentverbsisnotthatexoticandthatwell-knownphenomenalikeunaccusativitymaybeanalysedasinvolvingasimilarmismatch.•Toestablishamorphosyntacticanalysisfordeponent(andunaccusative)verbs–thatcapturestheobservedmirrorimage–thatmakesuseofasfewstipulationsaspossible–thatoffersexplanationsaboutthephenomenonLiteraturAdger,D.(2003).CoreSyntax:AMinimalistApproach.OxfordUniversityPress.Alexiadou,A.,&Anagnostopoulou,E.(1999).Testsforunaccusativityinalanguagewithouttestsforunaccusativity.EllinikaGrammata,(pp.23–31).Alexiadou,A.,Anagnostopoulou,E.,&Everaert,M.(2004).TheUnaccusativityPuzzle.Oxford:OxfordUniversityPress.Baerman,M.(2007).Morphologicaltypologyofdeponency.InM.Baerman,G.Corbett,D.Brown,&A.Hippisley(Eds.)DeponencyandMorphologicalMismatches.OxfordUniversityPress.Buchholz,E.(2005).GrammatikdernnischenSprache.Bremen,Hempen-Verlag.Chomsky,N.(1995).TheMinimalistProgram.Cambridge,Mass.MITPress.Collins,C.(2005).Asmugglingapproachtothepassiveinenglish.Syntax8:2.Embick,D.(2000).Features,syntaxandcategoriesinlatinperfect.LinguisticInquiry31.Embick,D.(2004).Unaccusativesyntaxandverbalalternations.InA.Alexiadou,E.Anagnostopoulou,&M.Everaert(Eds.)TheUnaccusativityPuzzle.Oxford:OxfordUniversityPress.Kalluli,D.(2006).Auniedanalysisofpassives,anticausativesandreexives.InO.Bonami,&P.Cabredo(Eds.)EmpiricalIssuesonSyntaxandSemantics6..Kratzer,A.(1996).Severingtheexternalargumentfromitsverb.InJ.Rooryck,&L.Zaring(Eds.)PhraseStructureandtheLexicon.Dordrecht:Kluwer.Lavidas,N.,&Papangeli,D.(2007).Deponencyinthediachronicofgreek.InM.Baerman,G.Corbett,D.Brown,&A.Hippisley(Eds.)DeponencyandMorphologicalMismatches.OxfordUniversityPress.Levin,B.,&Rappaport-Hovav,M.(1995).Unaccusativity.Cambridge,MITPress.Perlmutter,D.(1978).Impersonalpassivesandtheunaccusativehypothesis.InJaeger(Ed.)ProceedingsoftheFourthAnnualMeetingoftheBerkeleyLinguisticSociety.Berkeley:UniversityofCalifornia.Prince,A.,&Smolensky,P.(1993).OptimalityTheory.Constraintinteractioningenerativegrammar.Ms.RutgersUniversityandUniversityofColorado.9 Ritte,H.(2004).SchwedischeGrammatik.MaxHueberVerlag.Stump,G.(2007).Anon-canonicalpatternofdeponencyanditsimplications.InM.Baerman,G.Corbett,D.Brown,&A.Hippisley(Eds.)DeponencyandMorphologicalMismatches.OxfordUniversityPress.vanRiemsdijk,H.(2008).Identityavoidance:Ocp-effectsinswissrelatives.InR.Freidin,C.Otero,&M.Zubizarreta(Eds.)FoundationalIssuesinLinguisticTheory.EssaysinHonorofJean-RogerVergnaud.Cambridge,MA:MITPress.Xu,Z.,Aronoff,M.,&Anshen,F.(2007).Deponencyinlatin.InM.Baerman,G.Corbett,D.Brown,&A.Hippisley(Eds.)DeponencyandMorphologicalMismatches.OxfordUniversityPress.10