SwitchreferenceasmultipleagreementwithcyclicexpansionEmilyClemUniver - PDF document

1 Switch-referenceasmultipleagreementwithc
Switch-referenceasmultipleagreementwithcyclicexpansionEmilyClemUniversityofCalifornia,BerkeleyMAD2018,11/08/181IntroductionSwitch-reference(SR;Jacobsen1967)offersaninterestingpuzzlefromtheperspectiveoflocalityinsyntacticoperations–Anelementindicates(non-)coreferenceofargumentsintwodifferentclauses–ThismarkerthusreectsfeaturesofDPsinitsownclauseandintheclausetowhichitattaches1(1)[jaa=xi3SG=NOMvua= kin]=munsing=SA.WHILE=CMATRIXxano=niwoman=ERGxukicornjova=xo=nucook=3.PST=DECL`Whilesheisings,thewomanicookscorn.'(2)[joniimanvua= hain]=munsing=DS.WHILE=CMATRIXxano=njwoman=ERGxukicornjova=xo=nucook=3.PST=DECL`Whilethemanisings,thewomanjcookscorn'Oneofthesimplestwaystocapturethistypeofpatternistoassumethatthecovaryingheadagreesdirectlywithbothrelevantarguments(ArregiandHanink,2018) IamgratefultomembersoftheAmahuacacommunityfortheircollaborationonthisproject.IalsothankAmyRoseDeal,LineMikkelsen,PeterJenks,DavidPesetsky,NorvinRichards,MarkBaker,andaudiencesatUCBerkeleyandNELS49forhelpfuldiscussionofthedataandanalysis.ThisworkwasmadepossiblebyfourOswaltEndangeredLanguageGrants.Allerrorsareminealone.1Thefollowingabbreviationsareusedinglossing:1=rstperson,3=thirdperson,AM=asso-ciatedmotion,C=complementizer,DECL=declarative,DS=differentsubject,EMPH=emphatic,ERG=ergative,GEN=genitive,IPFV=imperfective,NOM=nominative,OS=objectcoreferentialwithintransitivesubject,PL=plural,PRES=present,PST=past,SA=subjectcoreferentialwithtransitivesubject,SG=singular,SO=subjectcoreferentialwithobject,SS=subjectcoreferentialwithintransitivesubject,TR=transitive.ThisraisesquestionsaboutthedirectionalityandlocalityofAgree–IsAgreeupward,downward,orboth?–HowcananelementinanadjunctclauseAgreeintoanotherclause?äIarguethatwecanprovideanAgree-basedaccountofSRwhilemain-tainingthatAgreeisalwaysunderc-commandandisalwayslocalCyclicAgree(Rezac,2003,2004;BéjarandRezac,2009):–Aproberstprobesitsc-commanddomain–Iftheproberemainsunsatised,whentheheadreprojectstoformanintermediateproject

2 ion,theprobereprojectsaswell–Theprobethe
ion,theprobereprojectsaswell–Theprobethenprobesitsnew,expandedc-commanddomain(thespecierofthehead)–Aclassicexampleofthisiswithagreeingv(3)vmax DmaxSUBJ v vmin DmaxOBJ 2 1UndertheassumptionsofBarePhraseStructure(BPS),thereisnoformaldistinctionbetweenthelabelofintermediateandmaximalprojectionsThepredictionofaCyclicAgreemodelcoupledwithBPSisthatmaximalprojectionsshouldbeabletoserveasprobesIarguethat,notonlyisthispredictionborneout,butitprovidesastraightforwardwaytoaccountforSRinthetypeofstructurein(4)1 Switch-referenceasmultipleagreementEmilyClem (4)X Cmax X Cmin Tmax 1 2Specically,IarguefortheexistenceofthisstructureinAmahuaca(Panoan;Peru)wheretheSRmarkerisanagreeingadjunctC–AdjunctCminprobesDPsinitsc-commanddomain,theadjunctclause–BecausetheprobeonCremainsunsatised,Cmaxalsoprobesitsc-commanddomain,agreeingwithmatrixDPsäThus,theAmahuacadataprovidesupportformultipleAgreeviacyclicexpansionandsuggestthatmaximalprojectionscanserveasprobesRoadmap:–§1:Introduction–§2:AmahuacaagreeingC–§3:Multipleagreementwithcyclicexpansion–§4:Alternativeanalyses–§5:Twotypesofmultipleagreement–§6:Predictionsandtypology2AmahuacaagreeingCAmahuacaisanendangeredPanoanlanguagespokeninthePeruvianandBrazilianAmazon–Mostlyheadnal,withahead-initialmatrixC–SOVwordorderwithscramblingofargumentsandadjuncts–Headanddependentmarking–Tripartitealignmentwithergative,nominative,andaccusativecaseAlldatawerecollectedduringmyeldworkinSepahua,PeruoverfoureldtripsIntemporaladjunctclausesinAmahuacatheelementindicatingthetem-poralrelationshipbetweenclausesisanencliticthattypicallysurfacesontheverb(5)[jaa=xi3SG=NOMvua= xon]=munsing=SA.AFTER=CMATRIXxano=niwoman=ERGxukicornjova=xo=nucook=3.PST=DECL`Aftersheisang,thewomanicookedcorn.'Forsimplicity,thefocusonthistalkwillbeon`after'clauses,but`while'and`before'showsimilarbehavior2.1Internalsyntaxof`after'clausesAmahuaca`after'clausesarefullCPs,largeenoughtocontainallargu-mentsoftheverbs,aswellasadjuncts–Theycancontainergative,(6),and

3 nominative,(7),subjects,aswellasobjects(
nominative,(7),subjects,aswellasobjects(6)[xano=niwoman=ERGchopaclothespatza= xon]wash=SA.AFTER=mun=CMATRIXproihatzamaniocjova=hi=ki=nucook=IPFV=3.PRES=DECL`Afterthewomaniwashedclothes,sheiiscookingmanioc.(7)[kiyoo-vi=nixiall-EMPH=NOMnokoo= xon]arrive=SA.AFTER=mun=CMATRIXproihatzamaniocjova=kan=xo=nucook=3PL=3.PST=DECL`Aftereveryoneiarrived,theyicookedmanioc.'2 Switch-referenceasmultipleagreementEmilyClem –Theycancontainadverbs(8)[proikoshiquicklyka= xon]go=SA.AFTER=mun=CMATRIXxano=niwoman=ERGhatzamaniocvana=xo=nuplant=3.PST=DECL`Aftersheiwentquickly,thewomaniplantedmanioc.'(9)[mohaalreadyxano=xiwoman=NOMnokoo= xon]arrive=SA.AFTER=mun=CMATRIXjato=ni3PL=ERGhatzayucaxoka=kan=xo=nupeel=3PL=3.PST=DECL`Afterthewomenihadalreadyarrived,theyipeeledyuca.'–TheycancontainotheradjunctCPs(10)[[proikariyamchoka= xon]wash=SA.AFTERproihatzamaniocxoka= xon]peel=SA.AFTER=mun=CMATRIXxano=niwoman=ERGxukicornjova=xo=nucook=3.PST=DECL`[Aftersheipeeledmanioc[aftersheiwashedyams]],thewomanicookedcorn.'(alternatively,`Thewomanwashedyams,peeledmanioc,andcookedcorn.')`After'clausesalsoallowclause-internalscrambling(11)`AfterIcookedpaca,Ipeeledmanioc.'a.SOV`after'clause[hiya=n1SG=ERGhanopacajova= xon]cook=SA.AFTER=mun=CMATRIXhun1SGhatzamaniocvuro=ku=nupeel=1.PST=DECLb.OSV`after'clause[hanopacahiya=n1SG=ERGjova= xon]cook=SA.AFTER=mun=CMATRIXhun1SGhatzamaniocvuro=ku=nupeel=1.PST=DECL2.2Externalsyntaxof`after'clauses`After'clausestypicallyappearinhighperipheralpositionsItisungrammaticalfor`after'clausestoappearbelowaspectmarking(Notethatnominalizedinternally-headedrelativeclausescanappearinthisposition)(12)`Aftersheisang,thewomaniiswashingmanioc.'a.[proivua= xon]=munsing=SA.AFTER=CMATRIXxano=niwoman=ERGhatzamaniocchoka=hi=ki=nuwash=IPFV=3.PRES=DECLb.xano=ni=munwoman=ERG=CMATRIXhatzamaniocchoka=hi=ki=nuwash=IPFV=3.PRES=DECL[proivua= xon]sing=SA.AFTERc.*xano=ni=munwoman=ERG=CMATRIXhatzamaniocchoka=hiwash=IPFV[proivua= xon]=ki=nusing=SA.AFTER=3.PRES=DECL`After'clausesdonotr

4 econstructbelowmatrixargumentsforConditi
econstructbelowmatrixargumentsforConditionC:regardlessoftherelativepositionofmatrixandadjunctmaterial,aCon-ditionCviolationisnevertriggered(13)`AfterMariaiwentquickly,sheiwashedclothes.'a.[proikoshiquicklyka= xon]=mungo=SA.AFTER=CMATRIXMaria=niMaria=ERGchopaclothespatza=xo=nuwash=3.PST=DECLb.[MariaiMariakoshiquicklyka= xon]=mungo=SA.AFTER=CMATRIXproichopaclothespatza=xo=nuwash=3.PST=DECLc.jaa=ni=mun3SG=ERG=CMATRIX[MariaiMariakoshiquicklyka= xon]go=SA.AFTERchopaclothespatza=xo=nuwash=3.PST=DECL3 Switch-referenceasmultipleagreementEmilyClem Theproposedsyntaxforthese`after'clausesisasgivenin(14)(14)Tmax Cmax T DmaxSUBJ T ... Tmin Tmax Cmin`after' DmaxSUBJ T ... Tmin DmaxOBJ v DmaxOBJ v 2.3Agreementin`after'clausesThereareseveraldifferentformsoftheencliticusedtomean`after'–Thesemorphemesvarydependingoncoreferencerelationshipsbe-tweenarguments(Sparing-Chávez,1998,2012)–Ifthereiscoreferencebetweenanargumentintheadjunctclauseandoneinthematrixclause,theformofthemorphemeissensitivetotheabstractcaseoftherelevantarguments–In(15),theadjunctclausesubjectiscoreferentialwithamatrixtran-sitivesubject(ERG),andtheagreeingadjunctCtakestheform=xon(15)[jaa=xi3SG=NOMvua= xon]=munsing=SA.AFTER=CMATRIXxano=niwoman=ERGxukicornjova=xo=nucook=3.PST=DECL`Aftersheisang,thewomanicookedcorn.'–In(16),theadjunctclausesubjectiscoreferentialwithamatrixin-transitivesubject(abstractNOM),andtheagreeingadjunctCtakestheform=hax(16)[jaa=xi3SG=NOMvua= hax]=munsing=SS.AFTER=CMATRIXxanoiwomanchirin=xo=nudance=3.PST=DECL`Aftersheisang,thewomanidanced.'–In(17),theadjunctclausesubjectiscoreferentialwithamatrixobject(abstractACC),andtheagreeingadjunctCtakestheform=xo(17)[jaa=xi3SG=NOMvua= xo]=munsing=SO.AFTER=CMATRIXhinandog.ERGxanoiwomanchivan-vo=xo=nuchase-AM=3.PST=DECL`Aftersheisang,thedogchasedthewomani.'–In(18),noadjunctclauseDPiscoreferentialwithanymatrixDP,andadjunctCisspelledoutasthedifferentsubjectmarker=kun(18)[joniimanvua= kun]=munsing=DS.AFTER=CMATRIXxanojwomanchirin=xo=nudance

5 =3.PST=DECL`Afterthemanisang,thewomanjda
=3.PST=DECL`Afterthemanisang,thewomanjdanced.'Thefullparadigmof`after'morphemesisgivenin(19)(19)`After'markers Matrix S A O Adjunct S =hax =xon =xo A O =ha =kun IthasbeennotedthatSRsharesmanysimilaritieswithcomplementizeragreementandcanpotentiallybeanalyzedasinvolvinganagreeingcom-plementizer(Watanabe,2000;ArregiandHanink,2018)TheAmahuacapatternlookslikecomplementizeragreementthatissen-sitivetoreferentialindexandabstractcase4 Switch-referenceasmultipleagreementEmilyClem äInterestingly,theagreeingcomplementizerissensitivetofeaturesofDPsinitsownclauseandtheclausetowhichCmaxisadjoined3MultipleagreementwithcyclicexpansionCyclicAgreecoupledwithBPS(Rezac,2003)predictsthatanunsatisedprobeshouldbeabletoprobethec-commanddomainofitsmaximalpro-jectionäIarguethatthepatternofagreeingadjunctCinAmahuacaisderivedviathistypeofcyclicexpansionoftheprobe'sdomainTheingredients:1.BarePhraseStructure(Chomsky,1995)–Thereisnoformaldistinctionbetweenintermediateandmaxi-malprojections2.Cyclicexpansion(Rezac,2003,2004;BéjarandRezac,2009)–Whenalabelreprojects,anunsatisedprobeassociatedwithitmayreproject–Probereprojectionservestoexpandthec-commanddomainoftheprobeandthustheagreementpossibilities3.Probeinsatiability(Deal,2015)–Aprobe'sinteractionconditionscandifferfromitssatisfactionconditions–Ifaprobelackssatisfactionconditions,itwillcontinueprob-ingallpossiblegoalsinitsc-commanddomainuntilreachingaphaseboundaryAdjunctCinAmahuacaisaninsatiableprobeFirst,Cminprobesitsc-commanddomain,whichcontainsthesubjectandobjectoftheadjunctclause–NotethatevidencefromremnantVP-frontingsuggeststhatobjectsundergoshifttoSpec,vP(Clem,2018b)(20)Agreementinsidetheadjunctclause Tmax Cmin DmaxSUBJ ... Tmin DmaxOBJ ... GiventhatC'sprobeisinsatiable,itremainsunsatisedafterprobingthec-commanddomainofCminWhenCreprojectstoformamaximalprojection,theprobeisreprojectedaswellandcanprobeagainThec-commanddomainofthisnewsegmentofC,Cmax,containsthema-trixsubjectandobject,keepingwiththeevidencefromC

6 onditionC(21)Agreementintothematrixclaus
onditionC(21)AgreementintothematrixclauseTmax Cmax T DmaxSUBJ ... Tmin Tmax Cmin DmaxSUBJ ... Tmin DmaxOBJ ... DmaxOBJ ... TheprobeonCagreesin:–Referentialindices(modeledas-features;Rezac2004)–Abstractcasefeatures5 Switch-referenceasmultipleagreementEmilyClem IftwoDPsthatCagreeswithshareareferentialindex,oneofthecorefer-encemarkerswillbeinserted–Theformofthemarkerwillbedeterminedbythecaseofthecoref-erentialDPsIfnoDPsshareareferentialindex,thedefaultdifferentsubjectmarkerwillbeinsertedSamplevocabularyitemsaregivenin(22)2(22)`After'vocabularyitems[[AFTER,[i,NOM*]][i,NOM]]Ø/hax/[[AFTER,[i,NOM*]][i,ERG]]Ø/xon/[AFTER]Ø/kun/4AlternativeanalysesTheaccountoutlinedherebuildsontheinsightofWatanabe(2000)thatSRsharesmanysimilaritieswithcomplementizeragreement(CA)Oneadvantageofthecurrentaccountisitssimplicity–thereareindepen-dentargumentsforallofthenecessarytechnology–CyclicityinAgree(Rezac,2003;BéjarandRezac,2009)–Probeinsatiability(Deal,2015)–Treatingindicesas-features(Rezac,2004)Additionally,previousaccountsofSRand/orCAfaceempiricalchal-lengesgiventheAmahuacadata4.1Non-reference-trackingaccountsofSRSomerecentanalysesofSRassumethatreferencetrackingisnotinvolved–Georgi(2012)arguesthatsamesubjectmarkingisaspecialcaseofcontrol 2WhatIlabelhereNOM*isafeaturethatiscommontoallembeddedsubjects,nominativeorergative.GivenindependentevidencefromcaseassignmentinAmahuaca(Clem,2018b),agoodcandidateforthisfeatureisa[T]featurethatindicatesagreementwithT.–Keine(2012,2013)arguesthatSRreectscoordinationheight,withsamesubjectclausesbeingVPcoordinationBothoftheseaccountspredictthataclausebearingasamesubjectmarkershouldbeunabletohostanovertsubjectDP(Clem,2018a)InAmahuaca,`after'clausescanhostallargumentsoftheverbovertly,includingcase-markedsubjects(23)[mohaalreadyxano=xiwoman=NOMnokoo= xon]=munarrive=SA.AFTER=CMATRIXjato=ni3PL=ERGhatzayucaxoka=kan=xo=nupeel=3PL=3.PST=DECL`Afterthewomeniarrived,theyipeeledyuca.'4.2AccountsofSRparasiticonagreeingTSomedirectreferenc

7 e-trackingaccountsofSRassumethatSRispara
e-trackingaccountsofSRassumethatSRisparasiticonagreementonT(Finer,1984,1985;Watanabe,2000;Camacho,2010)–TheseaccountspositsubjectagreementonTwhichisinterpretedasSRthroughsomemechanismattheCPlevel–Theseaccounts(sometimesexplicitly)ruleoutobjecttrackingsincetheprobeonTisassumedtoonlyagreewiththesubjectTheseaccountscannotstraightforwardlycapturetheAmahuacapatterninwhichCcanshowagreementwithboththematrixandadjunctobject(24)[jaa=xi3SG=NOMvua= xo]=munsing=SO.AFTER=CMATRIXhinandog.ERGxanoiwomanchivan-vo=xo=nuchase-AM=3.PST=DECL`Aftersheisang,thedogchasedthewomani.'(25)[joni=nman=ERGhinoidoghiin= ha]=munsee=OS.AFTER=CMATRIXproikoshiquicklyka=hi=ki=nugo=IPFV=3.PRES=DECL`Afterthemansawthedogi,itiisrunning.'6 Switch-referenceasmultipleagreementEmilyClem IfweweretoallowtheprobeonTtobeinsatiable,thiscouldaccommo-dateobjecttrackingHowever,thisishardtoreconcilewiththeattestedagreementonAmahuacaT–Amahuacatensemarkersindicatethepersonofthesubject–ThepersonoftheobjectisneverindicatedonTSinceAmahuacaTneverinectsforobjectperson,themorestraightfor-wardassumptionisthatTandCprobeseparately(HaegemanandvanKoppen,2012)–T'sprobeissatisedbyany-features(italwaysagreeswiththehighestDP)–C'sprobehasnosatisfactionconditions(i.e.itisinsatiable;itagreeswithallDPsinitsc-commanddomain)AnadditionalissuewiththeaccountsofFiner(1984,1985)andWatanabe(2000)isthattheyrequireamechanismofbindingbetweenmatrixandadjunctCwhichisotherwiseunnecessary4.3BoundanaphoraccountsofCAPatternsofupward-orientedCAhavebeenarguedtoinvolvelocalagree-mentbetweenCandaboundanaphorinitsspecier(Diercks,2013)WecouldimaginethatSRasatypeofdownward-and-upward-orientedCAmayinvolveagreementwithaDPargumentintheadjunctclauseandaboundanaphorinthespecieroftheadjunctCPHowever,thistypeofaccountisinconsistentwiththeAmahuacadata–ThereisnodistributionalevidencethatsuggestsadjunctCPsbeginlowenoughinthestructuretoallowbindingofananaphor–EvenifadjunctCPsbeganlowandobligatorilymovedhigher,theydonotreconstructforConditi

8 onC,(26)(26)[Floria=niFloria=ERGMariajMa
onC,(26)(26)[Floria=niFloria=ERGMariajMariahiin= xo]=munsee=SO.AFTER=CMATRIXMaria=njMaria=ERGFloriaiFloriachivan-vo=xo=nuchase-AM=3.PST=DECL`AfterFloriaisawMariaj,MariajchasedFloriai.'IfthereisnoreconstructionforConditionC,itisunclearhowtherecouldsimultaneouslybereconstructionforanaphorbindingAnadditionalpotentialissuewiththisstyleofaccountliesinthenatureoftheagreementrelationshipbetweenCandtheanaphor–IftheanaphorisinSpec,CP,purelydownwardAgreerequiressomeadditionalmechanismtoderiveSpec-Headagreement–OneproposalforhowtosubsumeSpec-HeadagreementunderageneraltheoryofdownwardAgreeisthroughcyclicexpansion–IfCyclicAgreeisassumed,wewouldneedsomestipulationtoruleoutthepossibilityofadjunctCmaxcontinuingtoprobe(27)Tmax Cmax T anaphor C Tmax Cmin DmaxSUBJ ... Tmin DmaxOBJ ... 7 Switch-referenceasmultipleagreementEmilyClem 5TwotypesofmultipleagreementThetypeofmultipleagreement(MA)activeinderivingthebasicpatternofSRmarkingiswhatIwillcall'Type1MA'–Type1MA:Asingleprobeinteractswithmultiplegoals–InthecaseofAmahuaca,theprobeonadjunctCinteractswithDPgoalsinitsownclauseandintheclausetowhichitisadjoinedTheAmahuacaSRsystemsalsoprovidesevidenceforasecondtypeofMA,Type2MA–Type2MA:AsinglegoalinteractswithmultipleprobesTherearetwophenomenathatprovideevidenceforType2MAinAmahuaca–-agreementonT–NestedSRclauses5.1MAin-agreementonTAmahuacamatrixTshows-agreementcontrolledbythesubject(28)hiya=x=mun1SG=NOM=CMATRIXhun1SGrakuu=ku =nube.afraid=1.PST=DECL`Iwasafraid.'(29)vaku=x=munchild=NOM=CMATRIXrakuu=xo =nube.afraid=3.PST=DECL`Thechildwasafraid.'Anargumentthatiscross-referencedbySRmorphologycanalsoserveasthecontrollerfor-agreementonT(30)[hiya=n1SG=ERGhanopacajova= xon]=muncook=SA.AFTER=CMATRIXhun1SGhatzamaniocvuro=ku =nupeel=1.PST=DECL`AfterIcookedpaca,Ipeeledmanioc.'(31)[MariaiMariakoshiquicklyka= xon]=mungo=SA.AFTER=CMATRIXproichopaclothespatza=xo =nuwash=3.PST=DECL`AfterMariaiwentquickly,sheiwashedclothes.'äThissuggeststhesameDPcanbetargetedbymultipleprobesforAgre

9 e5.2MAinnestedSRclauseSRadjunctclausesi
e5.2MAinnestedSRclauseSRadjunctclausesinAmahuacacanappearin`clausechains',withnogrammaticallimitonthenumberofclausesWhenmultipleadjunctclausesoccurwithasinglemainclause,theycanexhibitastacked,(32),ornested,(33),structure(32)Tmax Cmax T Cmax T (33)Tmax Cmax T Tmax Cmin Cmax T 8 Switch-referenceasmultipleagreementEmilyClem In(34),thetwoadjunctclausesarestacked,andtheSRmarkercross-referencesthematrixsubjectinbothadjunctclauses(34)[hiya=ni1SG=ERGhatzamaniocvana= kin]=munplant=SA.WHILE=CMATRIX[Maria=njMaria=ERGhun1SG.GENvakuchildjiri= kun]feed=DS.AFTERhiya=ni1SG=ERGrivialsojan3SG.GENvakuchildjiri=hifeed=IPFVhun=ka=nu1SG=1.PRES=DECL`[WhileIplantmanioc]I,too,amfeedingherkids,[afterMariafedmykids].'(alternatively,`Mariafedmykids,sowhileIplantmanioc,I,too,amfeedingherkids.')In(35),thetwoadjunctclausesarenested–TheSRmarkerintheloweradjunctclausecross-referencesthesub-jectofthehigheradjunctclause–TheSRmarkerinthehigheradjunctclausecross-referencesthesub-jectofthematrixclause(35)[[hiya=ni1SG=ERGhun1SGhatzamaniocvana= hain]plant=DS.WHILEMaria=njMaria=ERGhun1SG.GENvakuchildjiri= kun]=munfeed=DS.AFTER=CMATRIXhiya=ni1SG=ERGrivialsojan3SG.GENvakuchildjiri=hifeed=IPFVhun=ka=nu1SG=1.PRES=DECL`[AfterMariafedmykids[whileIplantedmanioc]],I,too,amfeed-ingherkids.'(alternatively,'WhileIplantedmanioc,Mariafedmykids,soI,too,amfeedingherkids.')Innestedsame-subjectstructures,thesubjectofthehigheradjunctclausewillbeagoalofAgreefortheCthatisadjoinedtoitsclauseaswellasfortheCofitsownclause(36)[[proikariyamchoka= xon]wash=SA.AFTERproihatzamaniocxoka= xon]=munpeel=SA.AFTER=CMATRIXxano=niwoman=ERGxukicornjova=xo=nucook=3.PST=DECL`[Aftersheipeeledmanioc[aftersheiwashedyams]],thewomanicookedcorn.'(alternatively,`Thewomanwashedyams,peeledmanioc,andcookedcorn.')(37)Tmax Cmax T Tmax Cmin Cmax T DmaxSUBJ peelmanioc Tmax Cmin DmaxSUBJ washyams äThus,theabilityofSRclausestooccurinnestedstructuresisafurtherpieceofevidenceforType2MA6PredictionsandtypologyäSRcanbeaccountedforwithexis

10 tingAgreetechnologyOnequestionwemightas
tingAgreetechnologyOnequestionwemightaskiswhythemajorityoflanguageswithSRonlyallowtrackingofsubjectsThecurrentaccountsuggestsseveralpossibilitiesforhowsuchsystemscouldarise9 Switch-referenceasmultipleagreementEmilyClem 1.Noobjectshift–InAmahuaca,objectshiftallowstheobjecttoescapethevPphaseandbeaccessibletoC'sprobe–Ifalanguagelacksobjectshift,CwillbeunabletoagreewithobjectDPs,resultinginasubject-onlytrackingpattern2.Casediscriminatingprobe–ItispossiblethatinalanguagewithaccusativealignmenttheprobeonCiscase-discriminating,agreeingonlywithnomina-tiveDPs–Thiswouldallowforsubject-onlytrackingeveninalanguagewithobjectshift3.Syncretism–ItispossiblethatalanguagecouldhaveaprobeonCthatagreeswithobjectsbutlackdedicatedmorphologytospelloutanobjectcoreferencerelationship–Evidencethatmorphologicalsyncretismmaybearelevantfac-torcomesfromcomparingtheparadigmsofdifferenttemporaladjunctCsinAmahuaca–Evenwithinasinglelanguage,differentparadigmshavediffer-ingdegreesofsyncretismwithrespecttothemorphologyavail-abletoindicateobjectcoreference(38)a.`After'series Matrix S A O Adjunct S =hax =xon =xo A O =ha =kun b.`While'series Matrix S A O Adjunct S =hi =kin =haito A O =hain c.`Before'series Matrix S A O Adjunct S =katzi/ =xankin A =xanni O =non äCyclicexpansionallowsmaximalprojectionstoserveasprobesAquestionwemightaskiswhywedon'tseemoreinstancesofmaximalprojectionsservingasprobes–Withmanycommonprobes(v,T,complementC),thec-commanddomainofthemaximalprojectiononlycontainstheheadthatselectsit,whichusuallywillnothavethecorrecttypeoffeatures–WithadjunctC,thispatternmayactuallybequitewellattestedgiventhatSRsystemsarerelativelycommon10 Switch-referenceasmultipleagreementEmilyClem ReferencesArregi,Karlos,andEmilyHanink.2018.SwitchreferenceinWashoasmultiplesubjectagreement.ToappearintheProceedingsofNELS48.Béjar,Susana,andMilanRezac.2009.CyclicAgree.LinguisticInquiry40:35–73.Camacho,José.2010.Oncaseconcord:Thesyntaxofswitch-referenceclauses.NaturalLanguage&LinguisticTheory28:

11 239–274.Chomsky,Noam.1995.Theminimalistp
239–274.Chomsky,Noam.1995.Theminimalistprogram.Cambridge,MA:MITPress.Clem,Emily.2018a.Againstnon-reference-trackingtheoriesofswitch-reference.ProceedingsoftheLinguisticSocietyofAmerica3:29:1–9.Clem,Emily.2018b.AmahuacaergativeasagreementwithmultipleheadsToappear,NaturalLanguage&LinguisticTheory.Deal,AmyRose.2015.Interactionandsatisfactionin-agreement.Talkhand-out.Diercks,Michael.2013.IndirectagreeinLubukusucomplementizeragree-ment.NaturalLanguage&LinguisticTheory31:357–407.Finer,DanielL.1984.Theformalgrammarofswitch-reference.DoctoralDis-sertation,UniversityofMassachusetts,Amherst.Finer,DanielL.1985.Thesyntaxofswitch-reference.LinguisticInquiry16:35–55.Georgi,Doreen.2012.Switch-referencebymovement.InPerspectivesonswitch-reference:Localmodelingandempiricaldistribution,ed.PhilippWeisser,1–40.Leipzig:InstitutefürLinguistik,UniversitätLeipzig.Haegeman,Liliane,andMarjovanKoppen.2012.ComplementizeragreementandtherelationbetweenC0andT0.LinguisticInquiry43:441–454.Jacobsen,WilliamH.1967.Switch-referenceinHokan-Coahuiltecan.InStudiesinSouth-Westernethnolinguistics,ed.DellH.HymesandWilliamE.Bittle,238–263.TheHague:Mouton.Keine,Stefan.2012.Switch-referenceascoordination.InPerspectivesonswitch-reference:Localmodelingandempiricaldistribution,ed.PhilippWeisser,107–164.Leipzig:InstitutefürLinguistik,UniversitätLeipzig.Keine,Stefan.2013.Deconstructingswitch-reference.NaturalLanguage&Lin-guisticTheory31:767–826.Rezac,Milan.2003.ThenestructureofcyclicAgree.Syntax6:156–182.Rezac,Milan.2004.Elementsofcyclicsyntax:Agreeandmerge.DoctoralDissertation,UniversityofToronto.Sparing-Chávez,Margarethe.1998.InterclausalreferenceinAmahuaca.InHandbookofAmazonianlanguages,ed.DesmondC.DerbyshireandGeof-freyK.Pullum,volume4,443–485.Berlin:MoutondeGruyter.Sparing-Chávez,Margarethe.2012.AspectsofAmahuacagrammar:Anendan-geredlanguageoftheAmazonbasin.Dallas:SILInternational.Watanabe,Akira.2000.Featurecopyingandbinding:Evidencefromcomple-mentizeragreementandswitchreference.Syntax3: