Oxford Scholarship Onlinewhamit.mit.edu/wp-content/uploads/...in...Movement.pdfHead movement is in a way the poor sister in the theoretical typology of movement operations. Of the

Getting Morphemes in Order: Merger, Affixation, and Head Movement

Page 1 of 34

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2015.All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of amonograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: University ofArizona Library; date: 31 August 2015

UniversityPressScholarshipOnline

OxfordScholarshipOnline

DiagnosingSyntaxLisaLai-ShenChengandNorbertCorver

Printpublicationdate:2013

PrintISBN-13:9780199602490

PublishedtoOxfordScholarshipOnline:September2013

DOI:10.1093/acprof:oso/9780199602490.001.0001

GettingMorphemesinOrder:Merger,Affixation,andHeadMovement

HeidiHarley

DOI:10.1093/acprof:oso/9780199602490.003.0003

AbstractandKeywords

Headmovementisusuallydiagnosedbyconsiderationofwhetherthereisevidencefor

displacementofasingle‐worditem.Themultimorphemiccharacterofagivenformoften

beingtakentobearontheissue,particularlywhenthemorphemeordermirrorsthe

orderoftheextendedprojectioninthesyntax.However,justastherecanbehead

movementwithoutaffixation,therecanbeaffixationwithoutheadmovement.Eventhe

issueofwhichmorphemeorderscanbetakenas‘mirroring’thesyntaxissomewhat

morecomplexinimplementationthancommonlyassumed.Additionalmechanismsmust

beatworkinderivingcertaintypesofcomplexforms.AnanalysisoftheCupeñoverbal

complexisarguedtoinvolveanintricateinterplayofindependentlymotivated

possibilities.Finally,someoftheformalproblemsposedbyheadadjunctionanalysisof

headmovementarereviewed,andabriefoverviewofsomealternativetheoretical

approachestoheadmovementisgiven.

Heidi Harley

no


Page 2 of 34


Keywords:Cupeño,theExtensionCondition,mergerunderadjacency,M‐merger,cyclicity

3.1IntroductionHeadmovementisinawaythepoorsisterinthetheoreticaltypologyofmovement

operations.Ofthethreetypesofmovementoperations,headmovementtypicallytravels

quiteashortdistanceinthelinearstring,whichmakesdiagnosingitparticularly

challenging;investigatorsmustoftenrelyheavilyonthepositionofsingle‐word

constituentslikenegationtoconfirmthatdisplacementhastakenplace.Further,its

structuralimplementationhasalwaysbeentheoreticallyproblematic.Inearlier

GovernmentandBindingtheory,thedefinitionofgovernmenthadtobecarefully

formulatedinordertoensurethattracesofheadmovementcouldsatisfytheECP

appropriately.WithinMinimalism,headmovementviolatestheExtendTargetandChain

Uniformityconditions,twointuitivelynaturalrestrictionseasilyderivedfrommore

fundamentalpremisses,andwhichthemselvesimposeempiricallyrobustconstraintson

derivations—theverykindofconstraintthatMinimalistthinkingpredictsshouldapplyin

thesyntaxofnaturallanguage.Finally,headmovementissituatedfirmlyattheinterface

betweenmorphologyandsyntax,whichmeansthatmuchofthecentraldatawhichhead

movementisdesignedtoaccountformustbeconsideredfromamorphological

perspectiveaswellasasyntacticperspective—andtheoriesofthemorphology‐syntax

interfaceareevenmorecontentiousanddisparatethantheoriesofsyntaxproper.

Ontheotherhand,manyoftheempiricalresultsproducedbythetheoryofhead

movementareamongthemostintuitivelysatisfyingideasinmodernsyntactictheory.Itis

gratifying,forexample,toseestudentscometounderstandthefundamentalsofthe

analysisofGermanV2phenomena,andtowatchthe‘aha’momentwhentheygraspthe

explanationoftheclause‐finalpositionofthetensedverbin(p.45) embeddedcontexts

incontrasttoitsV2positioninmatrixcontexts(denBesten1977).1Similarly,head

movementanalyseshaveproducedseveralofthemostplausibleandstraightforward

examplesofparametricvariation:theV‐to‐TparameterthatdifferentiatesFrenchfrom

English(Emonds1976;Pollock1989),theT‐to‐CparameterthatdifferentiatesGerman

fromFrench;theV‐to‐vparameterthatdistinguishesHindiandPersiancomplex

predicates(Folli,Harley,andKarimi2005),theN‐to‐Vparameterthatdifferentiatestrue

incorporationinMohawk(Baker1988)frompseudo‐incorporationinNiuean(Massam

2001)orHindi(Dayal2003).Finally,syntacticheadmovementcanprovideasatisfying

explanationfortheexistenceofthemorphologicalorderinggeneralizationscharacterized

byBaker’s(1985)MirrorPrinciple.Inshort,thetheoryofheadmovementhas

generatedanextremelyfruitfulandempiricallysignificantlineofinquiry,despitethe

difficultiesassociatedwithspecifyingexactlywhatheadmovementis,structurally

speaking

Inthischapter,Iwillfirstbrieflyexemplifysomeofthegenerallyacceptedsymptomsof

headmovement—theempiricalcluesthattendtosuggesttolinguiststhathead

movementmayhaveoccurred(Section3.2).Havingidentifiedthesesymptoms,wecan

thenask,whataretheparticulartheoreticaltoolsavailabletothesyntacticiantomodel

them?Iwillfirstconsidertheanalyticalandtypologicalpossibilitiesofferedbyperhapsthe

moststandardviewoftheheadmovementoperation,headadjunction.Ithendiscuss


Page 3 of 34


casesfromEnglishandCupeñothatpresentclearchallengestothisview,andidentifytwo

independentlymotivatedoperationsonX0terminalsthatareavailableatthemappingto

PF,MergerUnderAdjacency,andAffix‐SpecificLinearization.Theinteractionofthese

threeoperationsexpandsthetypologyofmorphemeordersavailabletothetheory,and

allowsforanaccountofcaseswhichwouldotherwiseposeaninsuperablechallengeto

theunadornedheadadjunctiontheory(Section3.3).Finally,Iwillreviewthetheory‐

internalissueswiththestructuralimplementationofheadmovementincurrent

Minimalistphrase‐structurethinking,andverybrieflydescribethevariedalternative

toolboxfordealingwithheadmovementmadeavailablebyvariousmodernsyntactic

theorists(Section3.4).Thesetheoriesrejectthestandardheadadjunctionanalysis

outlinedinthefirstsectionofthepaperforprincipledtheory‐internalreasons,and

addresstheresultingtheoreticalgapinseveraldifferentways.

3.2Diagnosingheadmovement

3.2.1Position

Aswithanykindofmovement,theprimaryindicationthatmovementhasoccurredisa

reorderingofthelinearstring.Forexample,finiteFrenchverbsappeartotheleft(p.46)

ofthenegativeelementpas(1),whileparticipialversionsofthesameverb,bearingthe

samesemanticrelationshiptonegationandtheclause’sarguments,appeartotherightof

negation(2):

(1)

Jean ne parlait pas français

Jean NEG speak.3P.IMP NEG French

‘Jeanwasn’tspeakingFrench’

(2)

Jean n’a pas parlé français

Jean NEG’has NEG SPEAK.PPL French

‘JeanhasnotspokenFrench’

Similarly,intheVSOlanguageIrish,thefiniteandnon‐finiteverbsappearindifferent

placesinthesentence,thoughinthiscase,thediagnosticelementwithrespecttowhich

theverbisreorderedisthesubjectDP,ratherthanasmallermonomorphemicelement

likenegation:

(3)

a.


Page 4 of 34


Phóg Máire an lucharachán (Carnie2002)

kissed Máire the leprechaun

‘Máirekissedtheleprechaun’

b.

Tá Máire ag pógail an lucharachán

is Mary PROG kiss the leprechaun

‘Maryiskissingtheleprechaun’

SincethesubjectDPcaninvolvearbitrarilylargestructures,thedistancebetweenthe

twopossiblepositionsofthemainverbisinprincipleunbounded.

Inthesecases,themainverbonlysurfacesinthelowerpositionwhentheupperposition

isindependentlyoccupiedbyanovertelement—anauxiliaryintheseexamples.Thiseffect

isreminiscentofboththeWh‐IslandConstraintandthebanonSuperraising.Awh‐

elementcannotmoveintoapositionalreadyoccupiedbyanotherwh‐element,andit

cannotskipsuchpositions,giventheill‐formednessofsentenceslike*WhatdidJohn

wonderwhyBilllikedt.Similarly,aDPcannotA‐moveintoapositionalreadyoccupiedby

anotherDP,andalsocannotskipsuchpositions,asshownbytheill‐formednessof*Two

menseemedtheretobetintheroom.Inthesameway,thehigherauxiliaryintervenes

betweenthemainverbpositionandhigherheadpositions,sothatevenincontextswhere

verb‐frontingwouldbepossibleweretheauxiliarynotpresent,thepresenceofthe

auxiliaryblocksit:*ParléJeanatfrançais?vs.A‐t‐iltparléfrançais?andParlait‐il

français2(Travis1984).

Thisparallelsuggeststhatamovement‐basedaccountofthevariablepositionofthemain

verbin(1)and(2)isappropriate.Thecentralideaisthatthelexicalverbisalwaysbase‐

generatedwithintheVP,whereargument‐structuralrelationsaresatisfied.Whenit

appearsinpositionsmanifestlyoutsidetheVP,ithasmovedthere(p.47) byhead

movement—movementandadjunctiontotheclosestc‐commandingheadposition,

illustratedbelow:

Heidi Harley

EXPLAIN


Page 5 of 34


(4)

Whenthetargetpositionisindependentlyfilled,asin(1)and(2),headmovementis

blocked.Travis(1984)firstmodelledthiseffectwiththeHeadMovementConstraint,

laterarguedbyRizzi(1990)tobeasubcaseofageneralizedlocalityconstraintonall

movementoperations,RelativizedMinimality.Thisparallelismbetweenheadmovement

andothercasesofsyntacticmovementwastakenasstrongevidencefortheproposal

thatamovementoperationwasresponsibleforthevariablepositionoftheverbinpairs

like(1)–l(3).

3.2.2Affixation

Syntacticheadmovementisalsooftenassociatedwithmulti‐morphemicstatus,wherea

head‐moveditemappearsattachedtomorphemesassociatedwiththetargetheadnode.

So,forexample,the‐aitsuffixin(1)indicatestenseandsubjectagreementinformation,

andappearsontheverbonlywhenithasmovedtoT0—imperfective‐markedverbs

alwaysappeartotheleftofnegation.Onenaturalhypothesisistoassumethattheaffixis

thephonologicalcontentassociatedwiththetargetnode.Ifheadmovementcreatesan

adjunctionstructure,wherethemovedheadadjoinstothec‐commandingtargethead,

theneachmorphemeintheverbisassociatedwithasinglesyntacticterminalnode.3

Headadjunctioncreatesthecorrectmorphosyntacticenvironmentfortheaffixtoattach

toitshost(cf.Baker1988:68–74);intheheadadjunctionconfiguration,bothaffixand

hostaredominatedbyasingleword‐levelX0projection:

Heidi Harley

BECAUSE MINIMALISM


Page 6 of 34


(p.48) (5)

Thenotionthatheadmovementisnotjustasyntacticoperation,butisalsoa

morphologicaloperationwhichbuildswordstructures,hasbeenveryinfluentialwithin

generativegrammar(thoughthetwoideasareinprincipleindependentofeachother).

Baker(1988),inastudyofincorporationprocessescross‐linguistically,proposedbuilding

nounincorporationstructuresandmorphologicallycausativeverbsbyheadmovementin

thesyntax—thatis,hearguedthatheadmovementcouldcreatethesemorphologically

complexforms,andthatasyntactictreatmentwasexplanatory:thesyntacticconstraints

onheadmovementaccountfortheattestedandnon‐attestedtypesofnoun

incorporation.So,forexample,Bakerproposedthatincorporationofanobjectnounwahr‐

‘meat’intotheverb‐ake’,‘eat’,inMohawk,wasnotamorphological,derivationalN–V

compoundingoperation,butratherasyntacticheadmovementoperation:4

(6)

a.

Owira’a waha’‐ wahr‐ake’ (Baker1988)

baby AGR‐ meat‐ate

‘Thebabyatemeat’

Heidi Harley

Heidi Harley

EXPLAIN

ARTICULATE WHAT THIS WOULD LOOK LIKE


Page 7 of 34


b.

(p.49) Thisproposalaccountedforboththethematicrelationshipbetweentheverband

thenominalobject,aswellasallowingasyntacticexplanationforthefailureofexternal‐

argumentincorporation,intermsoftheECP.

Thinkingofaffixationasheadmovementalsoallowedthetheorytoaccountforanother

importantobservationofBaker’s:thatmorphemeorderreflectedsemanticscope.Baker

(1985)showedthataverbmarkedwithbothcausativeandreciprocalmorphologycould

beinterpretedasacausativeofareciprocalorareciprocalofacausative,depending

cruciallyontheorderofaffixation.Hedubbedthegeneralizedproposalthatorderof

affixationreflectssyntacticandsemanticscopethe‘MirrorPrinciple’.Inthatpaper,Baker

didnotargueforaheadmovementapproachtoallaffixation,buttheattractionisclear:if

affixationisasyntacticoperation,constrainedbytheHeadMovementPrinciple,the

MirrorPrincipleisderivedasapredictionofthetheory.Thisissobecausethe

hierarchicalstructurecreatedbyseveraliterationsofheadadjunctionupthroughthe

syntactictreewouldnecessarilydirectlymatchthehierarchicalstructureofthetree

itself,giventhatnoheadintheextendedprojectioncouldbeskipped(theHMC),andthat

downwardsyntacticmovementisimpossible.TheMirrorPrincipleisthuspredictedif

affixationistreatedasasyntacticoperation,whileitmustbetreatedasastipulated

correspondencerelationinatheorywheremorphologicaloperationslikeaffixationare

encapsulatedintheirownsubmoduleinthelexicon,separatefromsyntacticstructure‐

building.PatternsofverbalaffixationlikethatofKoreansiph‐ess‐ta,‘want‐Tpast‐Cdecl’,for

example,arethepredictedoutcomeofamodelwhichequates(i)affixationwith

syntacticallyconstrainedheadmovementand(ii)inwhichthehierarchyofprojectionsis

Heidi Harley

Heidi Harley

Heidi Harley

EXPLAIN

DRAW


Page 8 of 34


CP‐TP‐VP;giventhesetwopremisses,theinverseorderoftensemarkingandmood

markingispredictedtobeimpossible.

Insum,twokeypropertieswhichareoftentakentobediagnosticofsyntactichead

movement,especiallyincombination,are(i)variablepositioninthesyntacticstring,

especiallywhenassociatedwithlocalityeffects,and(ii)affixation.

3.2.3Zero‐affixationandstring‐vacuousheadmovement

However,inmanycases,perhapsmost,thetwodiagnosticpropertiesfailtocorrelate.

Onecandetectdisplacementwhichisnotaccompaniedbyaffixation,asinthecaseof

sentence‐initialauxiliariesandmodalsinEnglishyes‐noquestions,whichhaveno

additionalmorphologydespitetheiradjunctiontoC0.Anexampleisgivenin(7a,b)below:

(p.50) (7)

a.Hecantype.

b.Canhetype?

Suchcases,however,areeasilyaccommodatedintheframeworkifoneassumesthat

zeroexponenceisapossibleoutcomeforaterminalnode.Onthataccount,(7b),

involvingheadmovementandadjunctionofT0toC0,doesinvolve‘affixation’,butbya

purelymorphologicalaccident,theaffixrealizingtheC0nodeisanullmorpheme.

Itisalsoeasytoidentifycasesofaffixationwhicharenotobviouslyaccompaniedby

displacement.ThelattersituationisamplyrepresentedinAltaic‐typehead‐finallanguages,

whichtypicallyexhibitaverbinflectedinsequencewithagglutinativemorphemes

respectingtheclause‐structurehierarchy,butwheredisplacement,ifitexists,cannotbe

detectedduetothestrictlyhead‐finalcharacterofthelanguage.Ifinsuchlanguagesall

headpositionsareontheright,themorphemesappearinthecorrecthierarchicalorder

withorwithouttheapplicationofheadmovement,asillustratedinthealternativetreesin

(8b,c):

(8)

a.

John‐wa dono okina piza‐o tabe‐ta‐ka?

John‐TOP which big pizza‐ACC eat‐PAST‐Q?

‘WhichbigpizzadidJohneat?’

Heidi Harley

MENTION GERMAN COMPLEMENTIZERS


Page 9 of 34


b.

(p.51) c.

Again,however,wecaneasilyaccommodatesuchcasesintheframeworkifwearewilling

topositstring‐vacuousheadmovementfortheory‐internalreasons.Mostbroadly

generativetheoriesconsideraffixationtoinvolvebundlingunderasinglesyntactic

terminalnode—thatis,intheunmarkedcase,one‘leaf’ofthesyntactictreeshould

correspondtoasinglephonologicalword.5Giventhisassumption,morphologicalaffixation

istakenasanindicationthat,bythetimeofspell‐out,distinctsyntacticterminalnodes


Page 10 of 34


havebeengroupedtogetherunderasingleX0terminalnode.Ifthatisthecase,theright

analysisfor(8a)is(8c),not(8b),justasthissystemofassumptionsentailsthattheright

analysisfor(7b)involvesazeromorpheme.Eitherdisplacementoraffixation,then,might

bemotivationenoughtoposittheoccurrenceofheadmovementinthesyntactictree.

3.3Affixation,displacement,andtheMirrorPrinciple6

Adoptingthenotionthatheadmovementisheadadjunction,therebyprovidingaunified

theoreticaltreatmentofbothcharacteristicpropertiesoftheprocess(p.52) described

intheprevioussection,wecanderiveaclearpictureofwhat‘ideal’casesofhead

movementshouldlooklike:theyshouldshowevidenceofdisplacementandexhibita

multimorphemic,affixedformwhoseaffixalorderingrespectstheMirrorPrinciple.

However,inmanyveryfamiliarcases,thisidealisnotrealized.Wereviewsomeanalyses

ofsuchdeviationbelow.

3.3.1SupplementalMechanismI:MergerUnderAdjacency

Wehaveseenintheprevioussectionthatdisplacementandaffixationcaneachsurface

independently,withoutclearcorroboratingevidenceforheadmovementprovidedby

theotherdiagnostic.Wenowturntoawell‐knowncaseinwhichthetwodiagnostics

clearlycontradicteachother:Englishtenseinflectiononlexicalverbs.Thediscussionand

analysisbelowisbasedonthatinBobaljik(1994)andHalleandMarantz(1993).

Inauxiliary‐lessEnglishdeclarativeclauses,themainverbshowsaffixationfortenseand

subjectagreement(‐edinpasttense,‐sinpresenttensewitha3sgsubject).Bythe

affixationcriterion,then,theverbmusthavehead‐movedtoT0,formingasinglecomplex

terminalnodewhichcanbespelledoutasasinglephonologicalword.

However,whensuchEnglishclausesaretestedfordisplacementofthemainverbtoT0,

thetestscomeupnegative,asshownbyEmonds(1976).TheEnglishmainverbmust

appeartotherightofVP‐adjoinedadverbssuchasoftenandnever,evenwheninflected

fortense—thatis,itbehaveslikeFrenchnonfinitemainverbs.Incontrast,English

auxiliaryverbsandmodalsappeartotheleftofsuchadverbs,justasFrenchfinitemain

verbsandauxiliaryverbsdo.

(9)

a.Maryoftenwalkedtoschool

b.Marymayoftenwalktoschool

AnothermarkeroftheVPdomain,negation,behavesperfectlynormallyinEnglishclauses

withauxiliaries:itintervenesbetweenthefiniteauxiliaryandthenonfinitemainverb,just

asinFrench.However,whenanauxiliary‐lessclauselike(9a)isnegated,astartling

transformationoccurs:themainverbmaynolongerbeinflectedforTense,andthe

dummyauxiliarydoappearsinthenormalpositionforauxiliaries,indicatingtenseand

agreement:

(10)

Heidi Harley

‘packaged’


Page 11 of 34


a.Marydidnotwalk/*walkedtoschool.

b.Marydoesnotwalk/*walkstoschool.

Similarly,whenT0movestoC0toformayes‐noorwh‐questioninaclausewhichdoes

notinvolveanauxiliary,themainverbmaynotbeinflectedfortense,anddo‐support

applies:

(p.53) (11)

a.DidMarywalk/*walkedtoschool?

b.DoesMarywalk/*walkstoschool?

Thestructuralanalysisofthefactsin(10)and(11)seemsverystraightforward;the

puzzleconcernsthemechanismforattachingtenseinflectiontotheverbincaseslike(9a).

Bobaljikproposesthatapost‐syntacticaffixationoperationoriginallyintroducedby

Marantz(1984)applies:M(orphological)‐Merger,whichhereferstoasMergerUnder

Adjacency.

OnBobaljik’saccount,terminalnodescanbeadjoinedtoeachotherinthepostsyntactic

componentaswellasinthesyntacticcomponent.Post‐syntacticadjunctionispossible

providedthatafterlinearizationthetwonodesarestructurallyadjacenttoeachother.In

anormalEnglishdeclarativeclausewithoutanauxiliary,T0andV0areadjacentinthe

relevantsense,andhencecanundergoMergerUnderAdjacencyinthepost‐syntactic

component.Insuchcases,then,affixationcanoccurintheabsenceofheadmovement,as

akindof‘repair’whenastrayaffixispresent.Thisiseffectivelyareinterpretationofan

AffixHoppinganalysis(Chomsky1957)withinamodernframework.7MergerUnder

Adjacencyisillustratedbelow:

(12)

Interveningheads,suchasnegationin(10),orspecifiers,suchasthesubjectin(11),

disrupttheadjacencyrelation,preventingMergerUnderAdjacency,andrequiringthe

applicationofadifferentrepairoperation,namelytheinsertionofdummydotosupport

strandedT0.8

GiventhelessonofEnglishinflectedmainverbs,wecandrawtwoprimaryconclusions.

Heidi Harley

But adverbs! But late-linearization/late-adjunction? Lack of LF repercussions?


Page 12 of 34


First,displacementisamoreimportantdiagnosticofheadmovementthanaffixation.

Second,headmovementisnottheonlymechanismavailabletoderive(p.54)

morphologicallycomplexforms;rather,thereisatleastoneothersourceforaffixal

behaviour,albeitconstrainedinveryparticularwaysbystructureandlinearorder:

MergerUnderAdjacency.Themodelmustbesupplementedwithsuchanoperationto

accountforthebehaviourofEnglishinflectedmainverbs.

3.3.2SupplementalMechanismII:Affix‐specificlinearization

Turningbacktotheoutcomeofheadmovementproper,wecanaskwhattheoretical

constraintsthereareontheadjunctionoperationitself.InanAntisymmetricapproachto

syntacticstructure(Kayne1994),theresultofheadadjunctionwillalwaysproduce

formsconsistentwiththeRight‐handHeadRuleofWilliams(1981),wherethemovedand

adjoinedlowerhead(e.g.V0)precedestheuppertargethead(e.g.T0).Inan

antisymmetrictheory,then,headmovementwhichproducesstructuresheadedby

prefixes,ratherthansuffixes,shouldbeimpossible.Ahead‐movedverbshouldappear

suffixedwithagglutinativemorphemes,eachofwhichisrealizingtheheadofphrases

dominatingVP.Thesesuffixesshouldappearintheirrespectivehierarchicalorder.

Plentyoflanguages,however,includeprefixationaswellassuffixationinthe

morphologicalmakeupofcomplexinflectedverbs.Consider,forexample,theagreeing

tenseandaspectaffixesoftheUto‐AztecanlanguageCupeño,asdescribedinHill(2005)

andanalysedinBarragan(2003):

(13)

pe‐ ya‐ qál

3SG.PAST‐say‐ IMPF.SG

T/Agr‐ V‐ Asp0

‘Hewassaying’ (Hill2005:ex.2c)

Inanantisymmetricapproach,suchaprefixalpatternhastobetreatedwithoutusing

headadjunction(seediscussioninfootnote11).However,analternativeisavailablein

theorieswhichadoptaparametric‐linearizationviewofmorphologicalheadedness,asin

DistributedMorphology(HalleandMarantz1993).

Letusassume,inaccordancewiththediscussionintheprevioussection,thatthe

complexaffixedformin(13)hasbeenformedbyheadmovement.Further,letusassume

thatUGprovidesatemplatefortheextendedprojectionofVPinwhichTPdominates

AspP,whichinturndominatesVP.(Iwillalsoassumeinthediagramswhichfollowthat

‘VP’isitselfcomposedoftwoprojections,vPandVP,inaccordancewiththetheories

proposedbyHaleandKeyser(1993)andChomsky(1995a,b),amongmanyothers.See

Section3.3forfurtherdiscussion.)

Giventheseassumptions,wecanpositthefollowingkindofunderlyingstructureforthe

Heidi Harley

Heidi Harley

Heidi Harley

Can I explain why? articulate LCA, explain application to head-adjunction structures?

Or semantic composition requirements


Page 13 of 34


complexformin(13):

(p.55) (14)

Withsimpleleft‐adjoiningheadmovement,inanantisymmetryapproach,thepredicted

surfaceformis*ya‐qál‐pe.However,withtheadditionofthestraightforwardassumption

thataffixesthemselvesspecifywhethertheyarepositionedtotheleftortherightoftheir

host,thecorrectformcanbederivedwhilestillrespectingtheMirrorPrinciple.Callthis

assumptionAffix‐specificLinearization.9

Inthisapproach,thesyntacticderivationonlycreateshierarchicalstructure,leaving

linearizationforthemorphologicalcomponentatSpell‐Out.HeadmovementadjoinsV0to

v0,v0toAsp0,andAsp0toT0,creatingafour‐layercomplexT0structure.Therearethen

severalpossiblemorphemeorderswhichrepresentalegitimateoutcomeoflinearization,

fromaMirrorPrincipleperspective,dependingontheprefixalorsuffixalstatusofeach

terminalnodeinthestructure.10Theeightpossibleordersareillustratedbelow:

(p.56) (15)

a.Everythingsuffixestoitssister(theAntisymmetricorder):

b.Everythingprefixal(auniformright‐adjunctionorder)

Heidi Harley

Clothes Hangers


Page 14 of 34


c.EverythingexceptT0suffixal,T0prefixal

d.EverythingexceptAsp0suffixal,Asp0prefixal

(p.57) e.Everythingexceptv0suffixal,v0prefixal

f.BothT0andAsp0prefixal,v0suffixal


Page 15 of 34


g.BothAsp0andv0prefixal,T0suffixal

h.BothT0andv0prefixal,Asp0suffixal

(p.58) Inthiscase,assumethattheTensemorphemepe‐,‘3sg.pst’islistedasaprefix,

whiletheAspectmorpheme‐qál,‘impf.sg’,islistedasasuffix.TheCupeñocomplexT

structure,then,mustlinearizeinconfiguration(15c)or(15h).Sincethev0headinthis

exampleisnotovertlyrealized,forthemomentwecannotdeterminewhichofthesetwo

optionsischosen(thoughperhapslinearizationofnon‐overtmorphemesisirrelevantto

thegrammar,inwhichcaseitremainssimplyunderdetermined).

Supplementedwiththepossibilityofaffix‐specificlinearization,then,thenumberof

Mirror‐Principle‐respectingmorphemeordersisconsiderablylargerthanasimpleleft‐

adjunctionapproachtoheadmovementpermits.TheMirrorPrincipleisnotmade

vacuousbythisadditionalassumption,however.Therearestillmanymorphemeorders

whichareunderivablewiththismechanism.Forexample,anymorphemeorderinwhicha

TensemorphemeintervenesbetweenAspandtheverbstem,orbetweenvandV,is

impossible;similarlyanymorphemeorderinwhichAspectintervenesbetweenvandVis

equallyimpossible.11

3.3.3Exploitingtheanalyticalspace:Cupeñocomplexpredicates


Page 16 of 34


Wehaveseenthatmorphologicallycomplexwordscanbeformedbyheadmovement,

butthattheheadmovementoperationmustbesupplementedwithadditional

assumptionstoaccountfortwofairlybasicpatterns.Thetwoadditionalitemsnowadded

toourtoolboxareaffix‐specificlinearizationandmergerunderadjacency.Wehaveseen

casesinwhicheachofthesetwotoolsisexploitedindependently;itisnaturaltoask

whetherthetypologicalpatternsoftheworld’slanguagesprovideevidencethatthe

effectsofthetwosupplementalmechanismscanbedetectedincombination.

(p.59) Barragan(2003)providesdatafromCupeñowhicharguethatindeed,both

mechanismsmustsometimesbeexploitedinthederivationofasinglecomplexform.

Cupeñocontainsmanycomplexpredicates,constructedfromalexicalrootcorresponding

tothev0nodeinthesplit‐vPstructure,andalightverbelement,correspondingtothev0

node.Theparticularv0usedalternatesdependingonwhetherthecomplexpredicateis

causativeorinchoative,acommonpatterncross‐linguistically(seee.g.Jacobsen1993on

Japanese):

(16)

a.caqe‐inFLAT‐IN‘toflatten’

a′.caqe‐yaxFLAT‐YAX‘tobeoblique’

b.cene‐inROLL‐IN‘rollsomething’

b.′.cene‐yaxROLL‐YAX‘somethingrolls’

c.hiwe‐inLUKEWARM‐IN‘heattolukewarm’

c.′.hiwe‐yaxLUKEWARM‐YAX‘somethingislukewarm’

d.puve‐inROUND‐IN‘makeround’

d′.puve‐yaxROUND‐YAX‘somethingisspherical’

Ineffect,wehavethreekindsofv0inCupeño.Thereareverbswhichoccurwithoutan

overtv0element,likeya,‘lsay’,in(13)above,andtwokindsofovertv0,thecausative‐in,

andtheinchoative‐yax.Iassume,followingHaleandKeyser(1993)andMarantz(1997)

thatthecausativev0selectsanexternalargumentwhichappearsinitsspecifier,andthe

inchoativev0preventstheappearanceofsuchanargument.

Themorphosyntacticproblemposedbysuchverbshastodowiththeirinteractionwith

tenseandaspectmorphology.Recallthatwiththezero‐classverbya‘say’,tensewas

prefixedtotheverbstem,andaspectsuffixed.Withcomplexpredicates,however,tense

morphologyappearsfollowingthelexicalverb,interveningbetweenthelexicalverband

thev0element.Thispatternisillustratedbytheexamplesin(17):

(17)

a.

Heidi Harley


Page 17 of 34


pe‐ ya‐qál (Ø‐classverbs)

PST.3SG‐ say‐IMP.SG (MonomorphemicV)

‘Hewassaying’

b.

mi= wíchax‐ ne‐ n‐ qál (in‐classverbs)

3PL.OB= throw‐ PST.1SG‐ VAGT‐IMP.SG (BipartiteV+v:Agentive)

“Iwasthrowingthem”

c.

nám‐pem‐ yax‐ wen (yax‐classverbs)

cross‐PST.3PL‐ VNONAG‐IMP.PL (BipartiteV+v:Nonagentive)

‘Theyusedtocross’

(p.60) Thisconfiguration—v0‐T0‐v0‐Asp0—ispreciselyruledoutbytheMirrorPrinciple,

onanyinterpretation.InnohierarchicalstructuregeneratedbythesyntaxdoesToccur

betweentheVPprojectionandthevPprojection.

Barraganpointsout,however,thatifweconsiderthev0morphemeonitsown,ignoring

themainverbforamoment,thepatternofaffixationlooksveryfamiliar.Considerthev0

morphemesinandyaxas‘light’verbs—almostlikeauxiliaries.Ifwefollowthenormal

rulesofCupeñogrammarforinflectingthoseelementsasverbs,wederiveexactlythe

patternin(17b)and(c)above:enseprefixedtov0,Aspectsuffixedtov0.Theonlypuzzle

isinthelocationoftheV0morpheme.

InordertoderivetheMirror‐Principle‐violatingmorphemeorderwithbipartiteverbs,

Barraganproposesthatinthecomplexpredicates,itisv0,ratherthanV0,whichhead‐

movestoT0.Inasense,thepatternisexactlylikethatinaV2language.There,amain

verbwillhead‐movetoTexceptwhenthereisanovert,interveningauxiliaryverb,in

whichcasetheauxiliarymovestoT,andthemainverbremainsinsituintheverbphrase.

Intheidenticalway,inCupeño,VmovesuptoT(throughv0)iffthereisnoovertv

morpheme,thatiswhentheverbisamemberoftheØ‐class.Whenthereisanovertv

morpheme,asinthebipartite‐inand‐yaxclasses,v0movestoT,strandingthemainV.

TheVislatersubjecttoMergerUnderAdjacencywiththecomplexT‐v‐Aspform.

LetusconsiderBarragan’sproposalforthederivationoftheform(17c)above.Inthe

syntax,thelightverbyaxhead‐movesfirsttoAsp0(whichsuffixestoit)andthentoT0

(whichprefixestoit).ThenMergerUnderAdjacencyappliesbetweenT0andV0,

prefixingtheheavyverbroottotheT0complex,derivingthesurfacemorpheme

order:12

Heidi Harley

Heidi Harley

Heidi Harley


Page 18 of 34


(18)

(p.61) Thissolutionisconsistentwithwhatweknowabouttheeffectsofovert

interveningheadsonheadmovementinotherlanguages(namelythattheyblockits

applicationtolowerelements),andderivestheunusualmorphemeorderfromtwo

independentpropertiesofCupeñogrammar:theexistenceofverbclasseswithanovert

v0morpheme,andtheprefixalstatusofT0.Theonlyunusualaspectoftheanalysisisthe

applicationofMergerUnderAdjacencytothesyntacticallycomplexT0formcontaining

thelightverbandaspectaswellasT—butsinceMergerUnderAdjacencyisapost‐

syntacticprocess,andheadmovementproperisasyntacticone,thereisnoprincipled

reasonwhythetwocannotbecombinedinthisway.13

3.3.4Affixation:HeadmovementorMergerUnderAdjacency?

Wehaveseenabovethatalthoughaffixationisonecommondiagnosticforhead

movement,itseemsclearthataffixalbehaviourcanalsoarisefromothersources;inthe

above,IhaveendorsedtheproposalthatMergerUnderAdjacencyisonesuchsource.

Thiskindofadjacency‐drivenaffixation?cliticizationoperationhasbeenbroadlyappealed

toinanalysesofapparentnon‐constituentaffixationasinEnglishJohn’sorShe’ll.Wojdak

(2008)arguesthataversionofthisoperationishighlyproductiveintheSalishlanguage

Nuu‐Chah‐Nulth.Similarproposalsconcerningaffixationunderadjacencybetweennon‐

constituents,intheabsenceofheadmovement,aremadebySelkirkandShen(1990)for

ShanghaiChineseandbyMyers(1990)forShonaprepositions.Julien(2002)similarly

proposesthatsuchanoperationisresponsibleforabroadspectrumofaffixationfacts

cross‐linguistically,particularlyincasesofprefixation.

Soaffixationissometimesadiagnosticforheadmovement,andsometimesnot.Affixal

order,too,issometimesadiagnosticforthesyntactichierarchyofprojections,because

ingeneralitrespectstheMirrorPrinciple.Butasshownabove,ifthiskindofMerger

UnderAdjacencyoperationisallowedtointeractwithheadmovementoperations,the

MirrorPrincipleeffectcanbedisrupted,asinCupeñocomplexpredicates.Onecanask

then,whenaffixationtellsusanythingatall—eitheraboutthesyntactichierarchy,or

Heidi Harley

Even better if Cupeño is head-final underlyingly


Page 19 of 34


aboutheadmovement?

Letusconsidertheformerquestionfirst:Whencanapieceofmorpheme‐order‐related

evidencejustifyanargumentaboutfunctionalprojections,ratherthanaboutmorpheme‐

orderingtechnology?Theanswerisfoundinthefactthatwhenaffixalorderreflectsthe

syntactichierarchyoffunctionalprojections,itshouldalsobethecasethataffixalorder

respectssemanticscope,asnotedbyKerenRiceinherdetailedconsiderationof

morphemeorderintheAthapaskanverb(Rice2000).Indeed,thisveryeffectisthe

originalraisond’êtrefortheMirrorPrincipleitself.Baker(1985:395)(p.62) pointsout

thedifferentinterpretationsthatattendthedifferentmorphemeordersinthefollowing

Bembaexamplesinvolvingreciprocalandcausativemorphemes(examplesoriginallyfrom

Givón1976):

(19)

a.

Naa‐mon‐an‐ya MwapenaMutumba

1SGS‐see‐recip‐cause MwapeandMutumba

‘ImadeMwapeandMutumbaseeeachother’

b.

MwapenaChilufya baa‐mon‐eshy‐ana Mutumba

MwapeandChilufya 3PS‐see‐cause‐recip Mutumba.

‘MwapeandChilufyamadeeachotherseeMutumba’

Baker’spoint,ofcourse,isthatthebindingandargumentstructurerelations—

presumablydeterminedbysyntacticconfigurations—arereflectedinlock‐stepbythe

morphemeorderontheverb.Iftheverbiscausativizedfirst,thenreciprocalized,the

subjectofcausativizationbindsthelogicalsubjectoftheembeddedverb.Iftheverbis

reciprocalizedfirst,thencausativized,thelogicalsubjectoftheembeddedverbbinds

theembeddedobject,andthesubjectofcausativizationdoesnotenterintothebinding

relationsdefinedbythereciprocalization.Inshort,thesyntacticandsemanticproperties

oftheseclausesmirrorthemorphologicalorderingontheverb.Inthesecases,then,it

wouldbeamistaketoderivethemorphemeorderviaonemechanismandthesyntactic

andsemantichierarchyviaanothermechanism—thepresenceofMirrorPrincipleeffects

isenoughtosubstantiatetheclaimthatagivenmorphemeorderreflectsthesyntactic

andsemantichierarchy.

Thesecondquestion,whetheraffixationwithoutdisplacementcaneverdiagnosehead

movement,isconsiderablymoredelicate.Ifaparticularaffixalformdoesnotseemto

exhibitdisplacementeffects,arethereotherkindsofevidenceonecanbringtobear

whichcouldargueforaheadmovementanalysis?Ifheadmovementissyntactic,and

henceinvolvessyntacticoperationssuchasCopyandRemerge,thenonemightexpect

Heidi Harley

Heidi Harley

DrawThis


Page 20 of 34


thatheadmovement,likeotherkindsofsyntacticmovement,couldinfactaffectsemantic

interpretation—thatis,thecontentofagivenheadmightbeinterpretedinahigherscopal

positionthanitsFirstMergeposition.Lechner(2006)marshallsanarrayofargumentsin

favourofthisconclusionaboutheadmovement,inoppositiontomanypreviousclaimsto

theeffectthatheadmovementisalwayssemanticallyvacuous—thatis,claimsthatheads

arealwaysinterpretedintheirbase‐generatedFirstMergeposition.Thekeydatahe

adducesinvolvetherelativescopeofinterpretationofamodaloperator,negation,anda

universalquantifier:

(20)Noteverypearlcanbeaboveaveragesize.

Meaning:Itisnotpossibleforeverypearltobeaboveaveragesize.

(p.63) Theinterpretationof(20)involves‘neg‐splitting’.Thenegationcontainedwithin

thesubjectDPtakeswidestscope.Themodalisinterpretedwithinthescopeofnegation

but—crucially—outsidethescopeoftheuniversalquantifier,alsocontainedwithinthe

subjectDP,givingthescoperelations◇〉¬≻∀.Thecrucialproblemposedbythispieceof

datainvolvesestablishingthesyntacticpositioninwhichtheuniversalquantifieris

interpreted.Lechnerbringstogetheracollectionofargumentswhichpointtowardthe

conclusionthattheuniversalquantifiercannotbeinterpretedbelowSpec‐TP.How,then,

canthemodal,inT0,beinterpretedoutsidethescopeoftheuniversalquantifier?

Lechnerproposesthatthemodalhashead‐movedaboveSpec‐TP,toac‐commanding

AgrSPhead;thesubject’ssurfacepositionisinthespecifierofthisAgrSPphrase.The

modal,havingmovedtoAgrS0,thuscantakescopeoverthesubject’suniversal

quantifieratLF,afterthelatterisreconstructedandinterpretedinSpec‐TP.This

proposalrequiresthatheadmovement,likeothersyntacticmovements,issemantically

activeinatleastsomecases,thatis,itproducesinterpretiveeffectsatLF.

ArelatedproposalisadvancedbyKishimoto(2010),whereheadmovementofnegation

toTisarguedtoexpandtheNPI‐licensingdomainoftheclausetoincludethesubject.

SuchheadmovementofnegationcaseswithNPIsubjectsformminimalpairswithcases

wherenegationdoesnotraisetoT,andonlyobjects,butnotsubjects,cancontainan

NPI.Kishimotoarguesthatsinceneg‐movementtoTistheusualcaseinJapanese,there

istypicallynosubject/objectasymmetryforNPIlicensing:

(21)

a.

John‐ga nani‐mo kawa‐nakat‐ta.

John‐NOM anything buy‐NEG‐PAST

‘Johndidnotbuyanything’

b.

Dare‐mo hon‐o kawa‐nakat‐ta.

Heidi Harley

DRAW THIS?


Page 21 of 34


anyone book‐acc buy‐NEG‐PAST

‘Nooneboughtthebook’ (Kishimoto2010)

However,incaseswheredo‐supportseparatesNegationandtheverbstem,as

Kishimotoshowsispossibleinalimitednumberofsituations,weseeatypicalsubject‐

objectasymmetrywithregardtoNPIlicensing:

(22)

a.

John‐ga dare‐mo haire‐nakusi‐ta.

John‐NOM anyone enter.can‐NEGdo‐PAST

‘Johnmadenooneabletoenter’

b.

*Dare‐mo Mary‐o haire‐naku si‐ta.

anyone Mary‐ACC enter.can‐NEG do‐PAST

‘AnyonemadeMaryunabletoenter’ (Kishimoto2010)

(p.64) Kishimototakesthesefactstoshowthatheadmovementdoesaffectscopal

relationssuchasNPIlicensing,andconsequentlythatheadmovementissemantically

active;itisthena‘true’syntacticmovement,resultinginLF‐interpretivedifferences.14

IfLechner’sandKishimoto’sconclusionsprovetohavebroadempiricalapplication,then

inatleastsomecases,scopaleffectsassociatedwithheadmovementwouldbeacrucial

diagnosticforwhetheragivencaseofaffixationshouldbetreatedasacaseoftrue

syntacticheadmovementorasapost‐syntacticMerger.Syntacticheadmovementwould

thenbelikeanyotherkindofmovement,inwhichthestructuresalteredbymovement

havemoreinterpretivepossibilitiesatLFthanstructureswithoutmovement.

Inthenextsection,weturntoasetofalternativeproposalsforheadmovement,someof

whicharedevelopedastheoreticalapproachestotheviewoftheempiricallandscape

takenbyChomsky(2001a,b)amongothers,accordingtowhichheadmovementinfact

hasnointerpretiveramificationswhatever.Itisimportanttonote,however,thatno

advocateofsuchaposition15hasyetofferedanalternativeaccountofLechner’sscope‐

splittingfactsorotherputativesemanticconsequencesofheadmovement.

3.4OtherapproachestoheadmovementLetusfirstquicklyreviewwhydevisingacoherenttechnologytoimplementhead

movementisproblematicforsyntactictheory.Ithastodowiththeparticularproperties

ofthephrasestructurecomponentofmodernMinimalisttheory,BarePhraseStructure

(Chomsky1995a).

Heidi Harley

Draw Tree


Page 22 of 34


3.4.1Theoreticalissues:BarePhraseStructure,ExtendTarget,andChainUniformity

Itisabitironicthatheadmovementshouldposeamajortechnicalproblemforsyntactic

theory.Atthebeginningofthe1990s,itfitbeautifullyintothesystemofassumptionsthat

werecoalescingintotheMinimalistProgram.TheHeadMovementConstraintofTravis

(1984)hadbeenunifiedwithotherconstraintsonmovementasaninstanceofRizzi’s

(1990:11)RelativizedMinimality,thefirstoverarchingvisionofhoweconomy

considerationsmightrestrictMove‐α.AsnotedinSection3.1,theempiricalpicturefor

certainbasiccasesseemedequallyrosy,somuchsothattheywereandarestaplesof

introductorysyntaxclasses.

(p.65) Nonetheless,gettingthestructuralmechanismofheadmovementtointeract

properlywiththeotherfundamentalsofthetheorywasaheadacheevenwithinX‐bar

theory.16WithinChomsky’sBarePhraseStructureformalism,itisessentiallyimpossible.

InBarePhraseStructure,thecrucialnotion‘segmentofX0’becomesincoherent,since

‘head’isequivalentto‘terminalnode’andanX0issimplyaterminalelementwith

somethingadjoinedtoit,sothatitprojects;anythingdominatingabranchingnodeisnot

anX0.Consequently,withinBarePhraseStructure,anadjunction‐to‐X0accountofhead

movementviolatesnotonlycyclicity17(sinceadjunctionistoanon‐rootnode),butalso

ChainUniformity,asoutlinedbyChomsky(1995b):

Wehavesofarsidesteppedaproblemthatarisesinthecaseofordinaryhead

adjunction.Takeα,KtobeX0sin(120)[they’resisters—HH],withαraisingto

targetK,whichprojects,formingL−{〈H(K),H(K)〉,{a,K}}.SinceKprojects,αis

maximal.Thus,αisbothmaximalandminimal.Ifthatistrueoftaswell(e.g.inthe

caseofcliticraising),thenCH[ain]satisfiestheuniformitycondition.Butsupposet

isnonmaximal,asiscommoninthecaseofV‐raisingtoIortoV.Then,undera

naturalinterpretation,[chainuniformity]isviolated;CHisnotalegitimateobjectat

LF,andthederivationcrashes.(Chomsky1995b:321)

BoththecyclicityissueandtheChainUniformityissueareillustratedinthetreein

example(23),repeatedfrom(5)above,shornofitspre‐BPSXPvs.X0annotations:

(23)


Page 23 of 34


BeforeVcanraiseandadjointoT,TmustenterthederivationviaMergewiththe

already‐builtVprojection(theVnodecorrespondingtotheVPnodein(5)).18(p.66)

TheresultofthatMergeoperationwouldrequirealabel,andtheTelement,asthehead

ofthestructure,iscopiedtoprovidethatlabel(producingtheTnodethatcorresponds

totheT′nodein(5)above).19Toimplementtraditionalheadadjunction,atthispointthe

lowestVinthestructure(correspondingtoV0in(5)above)undergoesMove—thatis,

CopyandRe‐Merge.However,inthetraditionalimplementationofheadmovement,the

copyofVmustnowMergewiththenon‐maximalT,ratherthantherootnodeT.Thus,

headadjunctioniscounter‐cyclic.

InordertograsptheviolationofChainUniformityimplicitinthetreeabove,itis

importanttounderstandthatthecategories‘maximalprojection’and‘minimalprojection’

areintendedtobederivedpropertiesinBarePhraseStructure.Followingaproposalof

Speas(1991),Chomskyadoptstheideathat‘minimalprojection’issimplyanynodewhich

doesnotdominateacopyofitself,and‘maximalprojection’isanynodewhichisnot

dominatedbyacopyofitself.ChainUniformity,then,isthenaturalrequirementthat

copiesofagivenconstituentmustmatchtheminimaland/ormaximalstatusofthecopied

element.

In(23),Vinitsbase‐positionisaminimalprojection(itdoesnotdominateacopyofitself)

andnotamaximalprojection(itisdominatedbyacopyofitself).Initshead‐moved

position,however,itisamaximalprojection(sinceitisnotdominatedbyacopyofitself).

ThechainformedbythemovementoftheVparl‐,then,consistsofaheadinoneposition

andamaximalprojectioninanother.Hence,headadjunctionviolatesChainUniformity.

Theproblemisexacerbatedastheprocesscontinues.Atleastbothstepsofhead

movementin(23)operateonandproduceasyntactically‘visible’constituent,thatisa

minimaland/ormaximalprojection.Successive‐cyclicheadmovementposesaneven

greaterproblemforthetheory.Considerthetreein(24),whereVtoTisfollowedbyT

toC:20

(p.67) (24)


Page 24 of 34


Here,thesecondstepofheadmovement,wherethecomplexTmovestoadjointotheC

head,involvesCopyandRe‐mergeofanintermediate‐levelprojection.ThecomplexTat

thepointofCopyisneitherminimalnormaximal.Byhypothesis,syntacticoperationsare

abletoapplyonlytoconstituentstheycan‘see’;non‐maximalconstituentsdonotmeet

thiscriterion,andhenceshouldnotbeabletoundergosyntacticmovementasin(24).

Thesedifficulties,aswellasthedifficultyingettingV2ordertoworkoutcorrectlygiven

certainotherassumptions(Chomsky1995b:368),ledChomskytoconcludethathead

movementisessentiallyphonological—notpartofthesyntacticcomponentatall.Thisview

isreiteratedinChomsky(2001a,n.69).Heprovidesnosuggestionsastohowthis

conclusioncanbeimplementedinsuchawayastoretaintheempiricalgeneralizations

andlocalityeffectsthatmadeasyntactictreatmentofheadmovementsoattractiveinthe

firstplace.

However,discussioninChomsky(2001b:37)doeslayoutanempiricalbasisfor

consideringheadmovementtobeapurelyphonologicaloperation.PriortoLechner’s

proposal,mosttheoreticianshadtakenitasaxiomaticthatheadsarealwaysinterpretedin

theirbaseposition,regardlessofhowmanyiterationsofheadmovementhaveapplied.

SententialnegationinFrenchandEnglishexhibitsthesamescopalrelationshipswith

regardtotheverbinsentenceslikethosein(25),despitethepresenceofverb

movementinonelanguagebutnottheother;thesameistruelanguage‐internally,as

well,inthatFrenchsentenceswithauxiliaries,involvingnoheadmovementofthemain

verb,behavethesamescopallyasthosewithout.

(25)

a. Johndidn’tknowthatMarywasspeakingtohim. ¬;≻know

b. JeannesavaitpasqueMarielui parlait. ¬≻know

J. NEGknewnotthatM. to.himspoke.

(p.68) Similarly,exampleslikethosein(26)havebeentakentoindicatethatmandatory

modalmovementacrosssententialnegationisnotinterptreted.Giventhatthemodal

scopesbelownegation,theassumptionisthatitisinterpretedinitsbaseposition:21

(26)Johncannotleave. ¬≻◇,*◇¬

Inshort,exampleswhereheadmovementfailstoaffectinterpretationarerathereasyto

comeby,incontrasttocaseswhereitdoes.Chomsky(2001b:37)writes‘semantic

effectsofheadmovementinthecoreinflectionalsystemareslightornonexistent,as

contrastedwithXPmovement,witheffectsthataresubstantialandsystematic.That

wouldfollowinsofarasheadraisingisnotpartofnarrowsyntax.’

Insummary,then,implementationofheadmovementassyntacticheadadjunctionposes

significanttechnicalproblems.Theseemingabsenceofsemanticeffectsofheadmovement

(paceLechnerandKishimoto)istakenbyChomskytosuggestthatheadmovementisnot


Page 25 of 34


asyntacticphenomenonatall,butratheraPFoperation.Inwhatfollows,we(very)

brieflydiscussanumberofalternativeproposalsforthetechnicalimplementationofhead

movement.Insomeofthese,headmovementstilloccursinthenarrowsyntax;others

attempttoimplementChomsky’snotionofaPFoperation;stillothersadoptentirelynovel

morphosyntacticstructure‐buildingoperations.

Alternativetheoreticalapproachestoheadmovementaredesignedtosolveeitheror

boththetheory‐internalstructuralproblemandthepotentialissueraisedbythe

apparentlackofinteractionofheadmovementandinterpretation.Iwillnextbriefly

describeafewsuchapproachesintheliterature,finishingwithadiscussionofmyown

approach,itselfanadaptationofaproposalfromHaleandKeyser(2002).These

alternativetechnicalframeworkswouldeachinteractdifferentlywiththemorpheme

orderanddisplacementissuesraisedearlier;Iwillpointoutsomeimplicationsofthat

datawhereIthinkIcanseewhattheymightbe,butextensiveworkwouldstillbe

requiredineachcasetoaddressthesequestionsproperly.

3.4.2Brody(2000):Mirrortheory

Brody(2000)proposesamodelwhichinvertsthemoreusualsyntactico‐centricpriorities

infavourofa‘morphocentric’approach,eliminatingheadmovementfromthetheoryby

viewingsyntacticstructureasaninterpretationofmorphologicalstructure,ratherthan

theotherwayaround.Morphologicallycomplexwords’internalstructureis‘mirrored’in

thesyntacticprojections:morphological(affixal)(p.69) ‘specifiers’areinterpretedand

projectedassyntacticcomplements.Brodyadoptstheideathatspecifiers,morphological

orsyntactic,areuniversallytotheleftoftheconstituentstheyarespecifiersof.Given

thatassumption,thelinearorderofmorphemeswithinawordisadiagnosticfor

morphologicalspecifierhood.AwordliketheItalian3sgfutureformfinira,therefore,

musthaveacomplexstructurelikethatin(27):

(27)

Morphology

fin‐ ir‐ a

finish‐ FUT‐ 3SG

[V [T [AgrS]]]

InBrody’ssystem,theVisamorphologicalspecifierofT,whichisamorphological

specifierofAgr.AfterthisformissubjecttotheoperationMirror,projectingasyntactic

tree,thefamiliarsyntacticprojectionresultsinwhichVisthecomplementofTandTis

thecomplementofAgrS.Thelocusofspell‐outofthecomplexmorphologicalforminthe

projectionlineisdeterminedbyparametricallyvaryingfeaturestrength,asinChomsky

(1995b:195);noheadneedactually‘move’fromanypositiontoanyotherposition

duringthecourseofthesyntacticderivation.22

Heidi Harley


Page 26 of 34


3.4.3Phrasalmovementapproaches

Anotherfamilyofapproaches,derivingfromKayne’sAntisymmetrymodel(Koopmanand

Szabolcsi2000;Mahajan2003,a.o.),treatdisplacementofheadsasadisguisedspeciesof

regularXPmovement.Headmovement,ontheseaccounts,isactuallyremnantphrasal

movement:allmaximalprojectionsinthespecifierandcomplementpositionof,for

example,VP,moveleftwardsandupwardsinthetree,leavingbehindaVPpopulated

onlywiththeVheaditself,togetherwiththetracesofitserstwhilespecifierand

complement.TheVPitselfcanthenmoveleftwardsandupwardsinthetree,toahigher

specifierposition.SincetheVisthesoleremainingoccupantoftheVP,thisproducesthe

appearanceofheadmovementwithoutactuallymovingaheaditself.23Afterseveral

iterationsoftheseprocesses,theresultingderivationsinvolve‘roll‐up’treesona

massivescale.Examples(28)and(29)replicatepartofaderivationofaHungarian

complexpredicatefromKoopmanandSzabolcsi(2000).Therelevantsentenceisgiven

first,thenatreefragmentillustratingthefinalstepofthederivationoftheembedded

complexpredicateisprovided:

(p.70) (28)

Nem akartam szét szedni kezdeni a rádiót

not want.1SG apart take.INF begin.INF the radio.ACC

‘Ididnotwanttobegintotakeaparttheradio’

(29)

Inthisderivation,thepeculiarorderoftheembeddedverbszétszed‐‘takeapart’with

respecttothematrixaspectualverbkezd‘begin’istreatedastheresultofremnant

movementoftheInfPclausecontainedwithinthecomplementofkezd’sVPtokezd‐’s

specifier.(NotethatthemorphemeorderintheembeddedInfPhasalreadybeen

establishedbymovementoftheembeddedVPtospec‐InfP.)Becauseallthearguments


Page 27 of 34


oftheembeddedverbhavealreadymovedoutoftheembeddedInfP(tospecofLP,ina

seriesofroll‐upmovements),theonlyvisibleeffectofmovementoftheembeddedInfPis

toplacetheinfinitiveembeddedverbtotheleftofthematrixverbinwhosespecifieritis

sitting,leavingtheembeddedobjecttoitsright.Morphologicalprocesseswillthenapply

toderiveappropriatephonologicalwordsfromadjacentelementsoftheappropriate

types,asinMergerUnderAdjacency.

3.4.4Headmovementapproaches:Matushansky(2006)(alsoPlatzack(Chapter2,thisvolume))

Inaproposaldirectlyaimedataddressingthedifficulttheoreticalcontradictionsraised

byheadmovement,outlinedinSection3.4.1,Matushansky(2006)proposesaversionof

headmovementwhichobeystheExtensionCondition,adjoiningmoved(p.71) headsto

thenodeattherootofthetreeunderconstruction.Thisisthenfollowedbyaversionof

theMergerUnderAdjacencyoperation,m(orphological)‐merger,wherebythemoved

headisloweredandadjoinedtothenow‐adjacentheadoftheprojectionithasadjoined

to.Thissequenceofoperationsisillustratedin(30):

(30)

a.UpperheadXprobestovalueitsfeaturesagainstlowerheadY:

b.LowerheadYcopiesandre‐mergestoXP,creatingaspecifierposition:


Page 28 of 34


c.M-mergerapplies,loweringY0toX0tosatisfymorphologicalconstraints

SubsequentoperationsmightcreatefurtherspecifiersofXP,ormovethecomplexX0

bundlebythesamesequenceofoperationsdescribedin(30).

TheoutputofMatushansky’sproposedprocessisonthesurfaceidenticaltothe

traditionalheadadjunctionaccount,andcapturesmanyofitsbenefits.Inparticular,head

movementisunifiedwithotherformsofmovementinreflectingafeature‐checking

operation,anditslocalcharacterisaconsequenceofthestandardlocality(p.72)

constraintsonfeature‐checking.Italsosucceedsineliminatingtheformalproblem

involvingcyclicity/ExtendTarget,and(asinthediscussionabove)onlymakesuseof

independentlymotivatedoperationsnecessaryelsewhereinthetheory.Inthese

regards,itisasuccessfuladaptationoftraditionalheadmovementwithinMinimalist

Programassumptions.

However,inoneregard,itfailstobeafullysatisfactorysolutiontotheformalproblems

raisedbyheadmovementinBarePhraseStructure.ItfailstoamelioratetheChain

UniformityviolationidentifiedbyChomsky(1995b),outlinedinthequoteattheendof

Section3.4.1.ThehighercopyoftheheadY0in(30b,c)is,bydefinition,phrasal,sincenot

dominatedbyacopyofitself.Thelowercopy,ontheotherhand,isnecessarilyahead.

Thechain[Y0,tY],then,isnotuniform.


Page 29 of 34


3.4.5Conflation(HaleandKeyser(2002),Harley(2004))

ThefinalapproachtoheadmovementIwilloutlineheregivesitafundamentallydifferent

characterfromtruesyntacticmovement;inasense,thisproposalisaspeciesof‘no‐

movement’approach.Nonetheless,theeffectofheadmovementinthisviewistriggered

byasyntacticoperation,soitdoesnotrequireaviewofmorphologyasseparateor

independentofsyntax,butrathercontinuestoallowthestandard‘interpretive’viewof

themorphological/PFcomponent.

HaleandKeyser(2002)putforwardaformalmechanismforderivingwhattheyterm

‘conflation’phenomena,whichHarley(2004)showscanalsoaccountcleanlyforthecore

casesofheadmovement.Conflationinvolvescopyingthephonologicalfeaturematrixof

thesisterslabelatMergeofanewX0constituent.SincethesisterslabelinBarePhrase

Structureisacopyofitshead,copyingthephonologicalfeaturematrixofthesister

constituentisequivalenttocopyingthephonologicalfeaturematrixoftheheadofthe

sister.ThederivationofanEnglishyes‐noquestionusingconflationtoaccountforthe

apparentmovementofT0toC0,isillustratedbelow.Thederivationisillustratedfromthe

pointatwhichthe[+Q]C0elementisdrawnfromthenumerationandMergedwiththe

existingstructureintheworkspace,aTP.AtMerge,C0triggersconflation:the

phonologicalfeaturematrixoftheelementitismergedwith—thelabelofTP—iscopied

intothephonologicalfeaturematrixoftheC0.Thephonologicalfeaturematrixthatis

associatedwiththelabelofTPis‘can’,sothisiscopiedintothephonologicalmatrixofC0.

C0’sownphonologicalmatrixhappenstobeaØ‐morphemeinEnglish,sotheeffectisthat

theC0terminalnodeispronouncedas‘can’.Thephonologicalfeaturematricesofeach

nodeareindicatedassubscriptstothenode:

(p.73) (31)

a.{C[Ø]}Mergeswith


Page 30 of 34


b.

Thefinalspell‐outofeachterminalnodewillasusualaffectonlythehighestcopyofany

phonologicalfeaturematrixinthetree.

Understandingheadmovementasconflationprovidesaclearexplanationforitslocal

character:conflationisanoperationthatonlyappliestotwosisternodesatMerge,

motivatedbythesyntacticallyaffixalstatusoftheMergedelement.Itwillnaturallybe

successive‐cyclicincharacter,sincethelabelsresultingfromMerge,themselvescopies

oftheMergedhead,willcontainthecopiedphonologicalfeaturematrix.Andsincethereis

noactualmovementinvolved,noformalproblemsconcerningthechainconditionor

cyclicityarise—indeed,theconflationoperationisstrictlycyclicinitsapplication.

Notethatinmanycases(asinthederivationoftheFrenchverbparl‐aitin(5))the

copyingheadwillitselfalsocontainanovertphonologicalfeaturematrix,albeitonewhich

cannotstandasawordonitsown.Insuchcases,conflationcreatesanaffixation

relationshipbetweenthecopiedphonologicalfeaturematrixandthecopyinghead.Itis

worthnotingthatinaLate‐insertionmodellikeDistributedMorphology,thecopied

materialisnotliterallyaphonologicalstring,butratherthePosition‐Of‐Exponencewhichis

associatedwitheveryterminalnodeinthenumeration.Inordertopredictinternal

morphologicalhierarchicaleffects(forexample,theexistenceofmorphophonological

processeswhicharesensitivetoword‐internalstructure,asinlevel‐orderedphonology),

weneedtoassumethatthegrammarkeepsarecordofthehierarchyofpositions‐of‐

exponencewithintheheadwhichtriggered(p.74) conflation.Thisrecordcouldbe

derivationalincharacter—cyclic,phase‐likespell‐outofeachpositionofexponencein

turn,forexample—orrepresentational,endowingthestringofpositionsofexponence

resultingfromconflationwiththeequivalentofbracketedstructure.Itisclear,however,

thatinsertionoperationsandmorphophonologicalreadjustmentoperationsaresensitive

tothisorderedhierarchywithintheword,sotheoutputoftheconflationoperationmust

beadjunction‐like,ratherthansubstitution‐like,incharacter.

Heidi Harley

Heidi Harley


Page 31 of 34


Harley(2004)proposedthatconflationcanprovideaformalmechanismtoaccommodate

Chomsky’s(2000a)assertionthatheadmovementisphonologicalincharacter,not

syntactic,whileretainingnaturalandsyntacticexplanationsformanyofitskeyproperties.

Indecidingbetweenatrulymovement‐basedapproachandtheconflationaccount,the

empiricalquestionofwhetherheadmovementisassociatedwithsemanticeffectswill

provedecisive.Conflationcannot,inprinciple,resultinsemanticeffects,sinceitinvolves

onlyphonologicalfeatures/positionsofexponence;atruemovementaccount,onthe

otherhand,candoso.Consequently,theproposalsanddataadvancedbyLechnerand

Kishimotothatweredescribedabovewillbecentraltothedebatearoundwhichisthe

correctmodel.

3.5ConclusionTheprimarygoalofthispaper,followingalongwiththegoalsofthevolume,hasbeento

addressthequestionofwhatthediagnosticsforsyntacticheadmovementare,and

whethertheseindividualdiagnosticsareinfactreliablecues.Ihopetohaveshownthat

onetemptingpairofdiagnosticsmustbeappliedwithextremecaution,namely,affixation

andmorphemeorder.Itisclearthatwhileaffixationcanresultfromtheapplicationof

syntacticheadmovement,itisnotthecasethataffixationisinvariablydiagnosticofit,as

thereareclearcasesofaffixationwhichcanbeindependentlyshownnottoresultfrom

headmovement.Further,whilesyntacticheadmovementmustproducemorpheme

ordersthatrespectBaker’s(1985)MirrorPrinciple,itisplausiblethattheMirror

Principleitselfislessconstrainingofmorphemeorderthanonemightatfirstimagine,if

weallowforthepossibilityofaffix‐specificlinearization.

Nonetheless,itseemsclearthatwhenaffixation,morphemeorder,andsemanticscope

arecorrelated,itisreasonabletoconcludethatsyntacticheadmovementhasindeed

createdthecomplexform,asinBaker’soriginaldiscussion.Consequently,someformof

headmovement,constrainedbylocalityandcyclicityconsiderations,mustbe

implementedinanyadequatesyntactictheory.Thequestionofhowbesttoaccomplish

this,however,remainsapointofconsiderablecontention.

Notes:

(1)Thisbeing,ofcourse,onecasewhereheadmovementtraversesasignificantlinear

distanceinthestring.

(2)Literallyglossed:*SpokenJohnhasFrench?;HashespokenFrench?;andSpokehe

French?

(3)Insomeformulationsofheadmovement,incontrast,themovingelementis

substitutedforthetarget(e.g.RizziandRoberts1989).

(4)Notethatinthetreediagramin(6b),Iindicateonlythefirststepintheconstruction

ofthecomplexword,namelytheincorporationofthethemeNintotheV.Theresulting

complexVmightormightnotthenhead‐moveintoT0tobeprefixedwithwaha’‐,—the

T/Agrnode—itmightdoso,andexploitaffix‐specificlinearizationtoensureitisrealized

prefixallyratherthansuffixally;ontheotherhand,itmightalsoremaininsituand


Page 32 of 34


undergoMergerUnderAdjacencywiththeverbalcomplexinVonitsright.Seebelow

fordiscussionofthesemechanisms;IdonothaveaccesstodatafromMohawkwhich

mightpotentiallydecidebetweenthem,ifsuchdataareevenpossiblegivenMohawk’s

polysyntheticcharacter.

(5)Though,again,ofcourse,mismatchesinbothdirectionsarenotdifficulttocomeup

with.Therearecertaincasesofapparentmovementthatcanbeperspicuouslytreatedas

headmovementbutdonot(obviously)appeartoresultinaffixation;onecandidatecase

isparticleshift(Johnlookedthenumberupvs.Johnlookedupthenumber),treatedby

Johnson(1991),Koizumi(1993),denDikken(1995,a.o.),asinvolvingheadmovement.

Similarly,thereareplentyofcasesofmorphologicaldependencethatarenotobvious

candidatesforaheadmovementanalysis,forexample‘leaner’cliticslike’llinI’llseeyou

tomorrow.SeealsothediscussioninJulien(2002).

(6)MuchofthediscussioninthissectionalsoappearsinHarley(2011).

(7)EmbickandNoyer(2001)arguethatMergerUnderAdjacencyiseffectivelyapost‐

syntacticLoweringoperation,whichcreatesacomplexterminalnodeundertheV0head

whenT0isadjacenttoV0.Onthatapproach,therelationshipbetweenX0statusandthe

phonologicalwordismaintained.Theydistinguishtwooperations,LoweringandLocal

Dislocation;wewillnotbeconcernedwiththeprecisenatureoftheirdistinctionhere.

(8)Adverbialadjunctslikeoften,duetotheirdistinctstructuralstatus,donotintervene

intherelevantsense,accordingtoBobaljik’sproposal.

(9)Areviewerrightlypointsoutthatinmanycases,treatinglinearizationasdriven

literallybyparticularphonologicalaffixeswillmisslanguage‐widegeneralizations:itisoften

acategory,ratherthanaparticularaffix,whichissubjecttothiskindoflinearization

constraint.InCupeño,aswewillsee,forexample,allTense/Agrmorphemesareprefixes,

notjusttheparticularonepe‐.Consequently,itwouldperhapsbebetterinmanycases

tostatethelinearizationpreferenceatthelevelofthecategory,ratherthantheaffix

itself;thiswouldbeastraightforwardelaborationofthestandardHeadedness

ParameterappliedattheX0levelratherthantheXPlevel.

(10)SeeSpeas(1991)forthisverypoint,althoughnotelaboratedinthedirectiontaken

here.

(11)Julien(2002)alsousesacombinationofheadmovementandmorphophonological

mergertoanalyseaffixationpatternscross‐linguistically,butadoptsamorerestrictive

antisymmetricframeworkinwhichaffix‐specificlinearizationisnotavailable.Sinceheadsin

theprojectionline(TP‐AspP‐VP)willfrequentlybeadjacent(whenevernophrasal

elementsoccupyinterveningspecifierpositions),theyaretypicallygoodcandidatesfor

suchaffixation‐under‐adjacencyprocesses,producingprefixalT‐Asp‐Vorders.Ofcourse,

theyalsoundergoheadmovement,toproduceinverse(suffixal)orders.InJulien’s

antisymmetricframework,however,linearordersofaffixationwhichcannotbederived

fromthecombinationofprojection‐linemergerandheadmovementmustbederivedby


Page 33 of 34


phrasalmovement.So,forexample,themorphemeorderin(15a)(V‐v‐Asp‐T)wouldbe

producedbyheadmovement;theorderin(15b),involvingpreffixation(T‐Asp‐v‐V),

wouldbeproducedbypreffixation‐under‐adjacencydirectlyintheprojectionline,with

nomovementofanykindnecessary.Theorderin(15c)(T‐V‐v‐Asp)wouldbeproduced

byacombinationofheadmovementofVthroughvtoAsp,followedbyprefixation‐under‐

adjacencyoftheTheadtotheleftofAsp,withoutmovementtoT.Orderslikethatin

(15d)(Asp‐V‐v‐T),however,wouldhavetoinvolvephrasalmovement:theycouldbe

producedbyheadmovementofVtov,followedbyremnantphrasalmovementofAspP

totheleftofT,followedbyaffixation‐under‐adjacency.Predictingthatorderings

producedbysuchphrasalmovementareexpectedtoberarerthanorderingsproduced

bybase‐generatedorhead‐movedstructures,Julien(2002)performedananalysisofthe

distributionofV,Asp,andTmorphemeordersin530languages,confirmingthatthe

distributionofthevariousorderingstendtoconformtotheexpectationsofthetheory.

Theaffix‐drivenlinearizationapproachdescribedhere,incontrast,wouldneedtoappeal

toexternalfactorstomotivatedifferentprobabilitiesofoccurrenceofeachofthe

orderingsgivenin(15),sinceformallyallareeqallyallareequallyprobable,beingableto

appearwithoutusingMerger‐UnderAdjacency.SeealsothediscussioninGorrie(2010).

(12)Infact,Cupeñotendstobehead‐final,likemanyUto‐Aztecanlanguages,sothetrees

illustratingthesestructuresshouldverylikelyinfactbemirror‐imagesofwhatis

presentedhere;notethatinthatcase,thecorrectorderbetweenVandtheT‐v‐Asp

complexisderivedsimplybytheusualheadednesspropertiesofthesyntacticstructure.

Inthatcase,MergerUnderAdjacencycouldoperatejustasa‘leaner’cliticization

operationwould.Seealsothediscussioninnote11.

(13)SeeGorrie(2010)fordiscussionandanalysisoftheincreasedtypologicalvariation

introducedbyallowingforthepossibilityofcombiningtheseoperations;remarkably,

morphemeorderevenontheseassumptionsisstillsomewhatconstrainedbythetheory.

(14)Theliteratureconcerningheadmovementinhead‐finallanguagesissubstantial,and

thediscussionofKishimoto’sproposalhereisnotintendedtoimplytheexistenceofa

consensusview.Argumentsagainstaheadmovementanalysisforsuchlanguagesare

presentedinYoon(1994),Koopman(2005),andFukuiandSakai(2003),andcontrasting

argumentsinfavourofheadmovementinOtaniandWhitman(1991),Hanetal.(2007),

andKoizumi(2000).

(15)Includingmyself!

(16)Forexample,Rizzi(1990:117n.19)concludedthatitmustbesubstitution,rather

thanadjunction,asdidRoberts(1991).ChomskyandLasnik(1993:ex.51,58)concluded

theopposite.

(17)AlsoknownasExtendTarget(Chomsky1995b:190)andtheExtensionCondition.

SyntacticMerge,asformulated,withinBPScanonlyoperateonwholephrase‐markers,

notonsubstructureswithinextantphrase‐markers.

Heidi Harley

Heidi Harley

Heidi Harley


Page 34 of 34


(18)AnalternativeistoconsiderthatVcan‘sideways‐Merge’withTbeforeTitselfenters

thederivation,àlaNunes(2001).Thatcouldhelptoresolvethecyclicityviolation,butthe

problemwithChainUniformityremains.

(19)InBarePhraseStructure,thelabellingoperation,perhapsconfusingly,isgenerally

takentobethesamecopyandre‐mergeoperationasmovementis.Mergeitselfcreates

aset,e.g.{V,D}.Inordertolabelthatset,oneofitsmembersiscopiedandre‐merged:

{V{V,D}}.Thisisthestructurecorrespondingtosomethinglike[VD]V′intraditional

bracketednotation;theconfusingpartisthatthelabelitselfisaMergedobjectintheset‐

theoreticnotation.Inthebracketedortreenotation,unlikeinthesetnotation,thefact

thatthelabelitselfisalsotheproductofMergeisnotobviousuponvisualinspection.

(20)ThistoyversionofFrenchfiniteverbinversionisnotrepresentativeofcurrent

thinkingonthederivationofFrenchquestions;see,forexample,PolettoandPollock

(2004)forafullexposition.

(21)WithoutchallengingLechner’sconclusionsconcerningtherelativepositionsofthe

variousinterpretedelementsin(20),itisworthnotingthathisproposalraisesahostof

interestingquestionsconcerningtheavailableinterpretationsofmodals.Thatis,under

whatcircumstancescantheybeinterpretedinamovedposition,andinwhat

circumstancesonlyintheirbaseposition?Whatrulesouttheavailabilityoftwoscopal

interpretationswhenVraisesabovenegation,asin(25)?

(22)AconceptuallyrelatedframeworkisproposedinDiSciullo(2005),inwhichthe

conceptofasymmetryingrammarisarguedtoplayacentralroleinthederivationof

morphemeorder.

(23)Asdescribedinfootnote11,Julien(2002)employsacombinationofsuchremnant‐

movementandmoretraditionalleft‐adjoiningheadmovementinherantisymmetric

proposal.

Accessbroughttoyouby: UniversityofArizonaLibrary

Heidi Harley

Documents

Oxford Scholarship Onlinewhamit.mit.edu/wp-content/uploads/...in...Movement.pdfHead movement is in a way the poor sister in the theoretical typology of movement operations. Of the