• Aucun résultat trouvé

BY WENDY P.A. SCAMMELL SUBM

N/A
N/A
Protected

Academic year: 2022

Partager "BY WENDY P.A. SCAMMELL SUBM"

Copied!
158
0
0

Texte intégral

(1)
(2)
(3)
(4)
(5)

TIlE ACOUISITION OF TIlE DATIVE ALTE RNATION

AMU'AKI!CLE MOVEMENT

Dr SECOND I ANGUAGELEARNERS

BY

WENDYP.A.SCAMMELL

SUBMIITEDIN PARTIAL FULJ1ILLMENT OF TIlE REQUIREMENTS FOR TIlE DEGREE OF

MASTEROF ARTS

DEPARnlENT OFLING UISTICS

MEMORIALUNIVERSITY OF NEWFOUNDLAND

MARCH,I991

(6)

1+1

NalionalofCanadaLibrary Bibliothequcnanonarc

ecceoaoa

Canad ianThesesServic e Servicedestheses CJ,I;\(JIcr)'ICS

Theauthorhas gra nted anirrevocab le non- exclusiv e Ilcen ceallowin gthe NationalUbrary ofCanada10reprod uce.loan,disbibut eOfsell copies ofhis/her thesisbyany mean sand in anyform or tomet.makingthlsthesisavailable to interest edpersons.

Th eautho r retainsownership oftheco pyright inhisfhertnests.Neithertilethesisnor subs tantial extrac tsfromitmaybe prin tedor othe rwisereprodu c edwitho uthis/herper- mission .

L'auteuraaccoroeunolicenc e irrcvoca.hlc ot non exclusivepermet tantilIn Btbliolheque nenonere duCan adaderep rocuir o.prC!ler . distribuer ouvenore des copiesde sa these dequetque maruereat sou squetqu c forme Queco soil pour mettre dese)(ern pralre~.II.' ce tte those flIn dispositiondenper :;onlle:;

mteressoes.

L'auteurconservelapm priclo dlJdroitd'nutorrr quiprotegeS,Jthose.Ni IJ1/16s o 111 don oxtrrsts substanuersde ccno-ct no doivc nt euo impr imes auactr cm ontrcproduitsG<'I1~;son cutonsauon.

Canada

(7)

ABSTRACT

Thisthesis looksat the second languageacquisitionof the English dative alterna tion. particlemovement and their interactionby nativespeakers ofMicmac.

The associatedtheoriesofsyntacticMarkedness, whichfollowsfroma theory or Unive rsalGrammarandCasetheoryare assumedas the basisfor Ihisresearch .The dative alternationis argued tohavetheunmarked structure(NP PPJ,aswellusthe marked structure{NP NPj.Theunmarked structure forthe verb-particle construction isassumed10be{V· Prt]and anyotherpositionof the particleinthesentence h marked.

Results of a studyontheacquisition ofthesestructures indicatethatthe unmarkedformsof the dativealternation and the verb-particleconstructionarc acquired first by secondlanguage learners. Agreaternumberof subjectsjUUgL'<I unmarked forms moreacceptablethan markedones accordingto the resultsofan intuitivejudgementrest andemployed morein productionthanmarkedstructures arc.

Resultson theinteractionofthesestructures showthatsentencescontaining an unmarkedcontiguousparticleand a prepositionaldative arejudged most acceptable and are widely employedintheproductiontask.Sentencesinvolving a markedverb- particle constructionand themarked double-objectform of the dativearejudged Jess acceptableandareemployedlessinproduction.The results presented in thisstud y support a continuumofmarkedness forsentencesinvolving both target structures.

(8)

FOREWO RD

Iwould like to express my thanksto manypeopleforassistingmein the completio n of mythesis. First.Jwould like tothank Dr.IreneMazurkewichfo r supervising thisthesis.She initiallystimulated my interestinlanguage acquisitionand suggested my thesis topic.Sheoffered frank criticism of eachdraftofmy thesisand enco uraged me to do thebest job thatIcould.Iwould alsoliketo givethanks to Dr.

John He wson for hishel p with Mic mac, and forhelping me withChapter3 in particular.Also.Mr.BernardFranciswhosehelp and nativeinsightintoMicmac was invaluable. Specialthanksgo10 Anne-MarieWhite for heraid and patience in preparing the final draftof lhisthesis.

I wouldliketothank Mr.Michae lMcleod,andDr.M.A .MacPhersonfor theirpermissiontoconduct researchon students from Whycocom aghFederalSchoo l andthe StraitAreaEducation Recreation CentreinCape Breton,NovaScotia. As we ll.Iam grateful10theAvalonConsolida tedSChoolBoardin51.John's.

Newfoundlandfor grantingpermission 10conductresearchon students at MacDonald DriveJuniorHighSchool.Forhelpon the statisticalanalysis,Iamgrate fulto Mr.

DonMacDona ld,Computing Services.Memorial Universityof Newfoundland.

Specialthanks go to myfamily,especially my hu\band Brianandmyparents Joh n andIreneSheppardfor their mo ral supportandpatience over thepasttwoyears.

Withouttheir guidance,thistaskwould have been allthemore difficult.

Research forthisthesiswas suppo rted bya Memo rialUnivers itySchool of Graduate StudiesFellowshipandanInstituteof Socia l andEconomic Researchtravel grant.Aswell,Jamgratefulfortwo researchassistantshipswith Dr.JohnHewson and Dr.Irene MazurkcwichoftheDepartment ofLinguistics at MemorialUniversity.

iii

(9)

TARLE

or

CONTEN TS

Foreword . iii

1.0 Introduction

1.1 UniversalGrammarandMarkedness.•. 1.2 CaseTheoryandAcquisition

I.J MarkednessTheoryand Acquisition . 1.4 Overviewof Thesis.. ...•..• . • • •...

Footnotes .

2.0 TheoreticalBackground 2.1 Introduction.

10 10

2.2 SyntacticTheory... ..•.. •. • •....

2.2.1the dativealternation... ..•...

2.2.1.1transformationalaccounts... 2.2.1.2 r,O!I·transfonnationalaccounts ..

10 10

10 15

2.2.2particlemovement ... • . ...• • .... .. ... 18 2.2.2.1contiguousanalysis... ... .. III

2.2.2.2non-contiguous analysis .•. 21

2.2.3Stowellanalysis.. 2.3 Summary.•

Footnotes,,.••,•... . .•,,,. ,, ....

. . . .. ,. " 24 27 32

(10)

].0AcquisilionResearch 3.1Englishasa FirstLanguage..

3.1.1acquisitionofdatives .

3.1.2acquisition of panicles•. . . .•• .... ...

3.1.3 acquisitionofinteraction ...

33 33 42 45 3.2 Englishas a SecondLanguage ... . . . ... . .•• . • 45 3.2.1acquisitionof datives ... . . .. .. • . •..• . 45 3.2.2acquisitionof paniclesand interaction. .. .... . ... 5I

4.0

3.3Summary .•.. . ..•. . .. . . ... .... ... .• . . . Footnotes.•. . • . . • . .. ... . • . . . . ... DescriptionofMicmac... , .. . . .•. .•... .

54 5R 59 4.1 DativeStructures...•. . . . ...• •... . . . .. . . 59 4.2Panicles... ... .. . . ... ...•. .• • •. 61 4..1 Summary...

5.0 ExperimentalStudy•. 5.1 Introduction....

5.2 SUbj«IS .

63 64 64

• 64

5.3 ElicitationProcedureUsed

5.3.1clcze lest .... .•• • • ...•...• •. ... . ...

65 65 5.3.2acceptabilityjudgementtask.••. ••• . ..•..• . •. 67 5.3.3sentencecompletiontask.... . . .. .. . . ..• . ... 68

5.4 Test Sentences andTheirClassification 70

5.5 TheoreticalPredictions..• • • • • .. .• . • . . • ..• . . • • • •. 74 Poomores••.• • ••. .• •.•.•••• •..••.•• •. .•.•. .•.• 76

(11)

6.0 Resultsand Discussion 6.1 Dataanalysis .

6.1.1 acceptabilityjudgementtask

n

71 .. .. . ... .." 77

7.0

6. J.2sentencecompteriontask••.••. 6.2 Results.. •... . . ... ....

6.2.1acceptabilityjudgement task•.

6.2.2 senlencecompletiontask ,.

6,3 Discussion Footnotes... . Conclusion., .. •.• • • " ..

1'>

RO su

Itl:.'!

114 121

122 7,1 SummaryofFindings

7,2Implications forAcquisitionResearch AppendixA•.

Appendix8, AppendixC..• . Bibliography

122 124 125 126 DI 138

(12)

CHAPTER I

Introduction

1.1Ilnjver$illGrammar andMarked~

Withinthetheory of UniversalGrammarCUG)outlinedby Chomsky(198Ia.

198 Ib),itispostulatedthai all childrenare endowed withaninnatecapacityfor languageacquisition.He proposesthatthere are principlesassociatedwith Universal Grammarwhicharc consistentfor a1llanguages;however. theindividuallanguages determinewhichprinciples of UGwill be accessed.

Within themorerecent principles-and-parameters modelof UG.language learners startout withanopen setof parameterswhichbecome fixed during the courseof languageacquisition. These parameters whichareset onthe basisof positiveevidence from the environmentmayvary from languageto language. The child's core grammaris determinedwhentheparameters ofUG arefixedinone of thepermittedways (Chomsky.1981b).

UG theoryalsoincorporates anassociatedtheoryof markednesswhich has two functions:Mitimposesapreference structureonthe parameters of UO,anditpermits the extensionof core grammarto amarkedperiphery.Experienceis necessarytofix thevaluesofparametersof coregrammar"(Chomsky.1981b.p.9). Theassumption is thatthe childstar tswiththeunmarked settingfortheparameters and hasto reset Iheparameters forstructures for Illf'languagethey areexposedto. Acquiringa

(13)

language.therefore.involvesappropriatelysetting antheparamcrcrs of UG which pertaintothatlanguage . The predictionthatf~111owSfrom such atheor yisthat acquisition will reflectthestructure of markedness and thisis the positionthatwill be takeninthis thesis.However. it isrecognized that. /ISpointedoutby Chomsky andmany others.therearemany compticetingfactors that may intervene.

1.2 Case Thcoryand Acquisition

In English.abstractcase (Chomsky. 1981b)isassignedby governingclements in a sentenceand involvesconditions relatedtoadjm.:ency. Specifically,an NP receives case atScstructureif itis governedby andadjace nttoatensedinflectional element. a verb or a preposition,Verbs assignobjectivecase and prepositions assign oblique (objectof preposition)case (Chomsky.198 Ib). Thereis a general condit ion of well-formednesswithinCasetheorycalledtheCasefitter whichrequiresthat all lexical noun phrasesbeassignedcase. As first proposed by RouveretandVergnaud (1980).thisCase filter which appliesat surfacestructurecanbeslatedasfollows:

'"NP.whereNPhas a phonological matrixbutno case

(14)

A lexicalNPwhich has no case is noe governed by a tensed inflectionalelement.a verboraprepositioninthesentence.Therefore,any sentencecontainingalexically filled NPwithnocasewillbeblocked by theCaW!filler.

Thisthesislooks atthe acquisitionof thedativealternationandparticle movement.For thedativeconstruction.caseisassignedtothe directObjectand the NPof thedativeprepositionalphrasebywayoffundamentalproperties of Case assignment.Insentence(I):

(I)Johngavethebook 10Bill

~is assigned objectivecase bytheverbM and.Billisassignedobliquecase bytheprepositionlQ. However.thetwoNP'sin thedouble-objectconstruction receivetheir caseinaslightlydifferent way.Insentence-(2) :

(2)JohngaveBillthe book

llillis assignedcaseby theverb&b£;buttheNPl~is saidtobeinherently case-marked(Chomsky. 1981b)asdeterminedbyproperties of itsgovernor.Stowell (1981) offersa somewhatdifferentaccount for casemarkingof thedouble-object constructionwhich will be discussedfurtherinsection 2.2.3.

Withrespect (0[heverb-particleconstructionitwillbearguedin this thesis thai theverb-panicleisa complexverb form thatassignscaseto the adjacentNPand

(15)

thatthisrepresentstheunmarkedfonn . Theassignmenlofeaseinvolvinglhl:

noncontiguousverb -particle formsderived by meansof amovement rule.Move alpha,introduces acomplicationthatwillbe discussed inmore detail in section 2.2.2.

1.3MarkednessTheory andACQujsjtion

Inlookingatthe acquisitionof the dativealternationand particlemovement, whichis thefocus of thisthesis.thereare certainissues that have directbearingon theconsiderationsofmarkedness.Forexample,howdoesa learnerof English set the parametersinvolved inthedative alternationwhenthere aretwooptionsthatare equally grammaticalandavailablefahim'in the linguisticenvironment. Aswell.

how does the learnercome todifferentiate between alternatingandnon-alternating dativeverbs?

Similarlyinthe caseofverb-particle constructions thatallowthepanicleto move,one mustask howit is acquired when therearetwo grammatical optionsfor the positionofthe particle inthe sentence.The verb and particle maybeconsidered as one contiguous lexicalunitwithadistinct meaning.On the other hand, ifthe...crb and particleare assumedtobe underlyinglynon-contiguous,theuniquesemantic propertiesof thecombinationwouldhave tobeexplained. From thepoint ofview of acquisitionthe contiguousanalysisis preferred.

(16)

This studyloo ks atthe acquisitionof the Englishdativealtern ationand particle movement,andtheirinteractio n,byMicmac speakerslearning Englishas-a-second lang uage'. Itis assumedthat eachstructure has botha marked and an unmarked form.For the dative.theunmarked formhas thecomplementstruc ture[NPPPJand the markedform has the complementstructure[NP NPl, Thefollowingisan exampleof each:

(3) Johngavethe book toMary (4) John gaveMary the book

Itisargued thatverbslike~whichalternate,areassignedthe subcatego rizations INPPPJ, as in(3)and [NPNPl. as in (4)inthe lexicon.Ontheother hand, verbs whichdo nOIalterna te.suchas ~,wouldbe assignedonly the subcateg orization INP PPJ.

Itwill be arg ued inChapter 2thatparticle movementis governed by a movementrule whichoptionallymovesthe panicle totherightof thedirectobject NP. Theunmarked form is assumedtohavethe struct ure[VPrtNPj,wherethe particleis contiguous10 the verb, as in(5), and the markedformhas thestructure[V NP Prt], wheretheparticleis noncontiguous,as in (6):

(5) John gave back thebook (6) John gavethebook back

(17)

Whatrenders the noncontiguous forms as mar kedisthefact thatthey requ ire the application of a movement rule.movealpha. As well,thenoncorutguousvcrh- particleformisconsideredmarkedbecausethe direct object iscase-marked byaverb thatallowstheparticle tobeseparatedfrom it.

This study also looks at sentenceswhich involve theinteractio nofthedative alternation and particlemovemen t; forexample:

(7)Johngave back the bookto Mary (8) Johngavethe book back toMary (9)John gaveback Marythe book (10) John gave Mary back the boo k (11)John gave Marythe book back (12)John gavethebooktoMar yback

Viewing the distinctionbetweencore grammarandtheperiphery as a continuum of markedness itis consideredtha t(7) isthe least markedasitco ntains both the contiguous verb-particle andthe[NPPP jcomplement.Sentence8is more marked because itcontains a noncontiguou s verbparticleconstruction anti the prepos itional dativecomplement. Sentence9 ismoremarkedbecauseitcontainsthedouble -object dativeconstructi on and a contiguous verb-pan icleconsuucuo n. Withregard to learning,it isassumedthatthemovement rulethat applies to verb -particle forms is ageneralizedrulein UG, movealpha and,as such,islea rnedvery early.Therefore,

(18)

sentence8which containsone marked structure. a non-contiguousverb -particle construction is consideredless marked than sentence9 which contains onemarked structure,the double-objectconstruction. Sentence10 is more markedthan the previousthree becauseit contains boththe marked doublc-objectdative and a no ncontiguous verb-panicleconstruction.SentencesIIand12 are the mostmarked because theycontain dative constructions and verb-particle constructions inwhichthe particlesare not onlyseparated fromthe verbbut are sentence-final.thefurthest possibleposition fromthe vcrb. As forthe grammaticalstatus ofsentences11 and 12,sentenceIIisdoubtful and sentence 12 is clearly ungram matical.Thesetypes ofsentences areincluded in thetesting inorderto testallofthe logical possibilities ofthe interactionofdatives and verb-pan icles. There seemstobea structural constraint onhow farIhe particleisallowedto be separated from the verb;however, theredoes not seem tobea probleminsentenceswhich involve only one verb complement, asin:

(13) Johnlookedthe information up

The remitsofthisstudy show that in general,the unmarkedformsofboththe dative andverb-particleconstructionsare acquiredbefore the markedforms. For sentencesinvolving the interaction,the resultsshow thatsentencescontainingthe[NP PPldative complementand a contiguousverb-particleare moreacceptable and are produced morethan those involvingthesame dativecomplementand a noncontiguous

(19)

particle,whichin tumare moreacceptable than those containing a double-object dativeand eithera contiguousor noncontiguousparticle. Finally. theresultsof sentencescontainingtheinteraction ofthe target structuresshow thatsentences containing a sentence-final particle and eitherdativeform aretheleast acceptableand are notproduced atall.Itisconcluded that thereisadevelopmentalsequenceinthe acquisitionreflecting a continuumof thedativealternation.particlemovementandthe interactionof both;unmarkedstructuresare acquiredbeforemarkedones' .

1.4Overview of Thesis

Chapter2 looks at thetheoretical background governing boththe dative alternationandthe verb-particleconstruction.Chapter 3 reviews firstand second languageacquisitionstudiesof thedative alternation,part iclemovement.andthe interactionof both. Chapter 4givesa brief descriptionof thestructuresunder consideratio nas theyrelatetoMicmac. Chapter5givesa descriptionof the subjects.

methodology fordata collectionandandescription of sentence types employed inthe analysis of theresults.Chapter6presents the analysisofthedata anda discussion of theresults.Chapter7presentsthe conclusionsdrawnfromthisresearch.and its implicationsfor further research in thefield.

(20)

FOOTNOTES

Forthe sakeof consistency,masculinepronouns willbeused10 represent boIh malesandfemales throughoutthislhesis.

2. There areanumberof studiesofotberconstructionswhichhave beencarried cutwithinthegeneralized parameter-sellingmodelof language. See.for example,Hyams' research onthenull-subjectparameter(Hyams1986).The questionconcerningthedistinctionbetweencoreand periphery,which involve ccnceptcns ofmarkedness,isa complex onethathasimportantimplications forL2theory. Fordiscussionofthese issuessetChomsky1986: Liceras 1988;Gair1988andWhite1989).

3. Pinker(1989)remarksonIhestatistical rarityofmar kedconsuucuonsand arguesthairulesthataccount fortheirderivation"usuallyviolatesome formal principle thatholds ofotherwise similarrulesinagrammar-(p.tJ7).

Furthermore.heargues that thisprovidessupport for the claimthatm.t:'ked structuresare"harderto learn"insomesense.Itison thisbasisthatwewill assume inthis thesis thatconditionsof markednesscanbereflected inthe developmentalsequencebyanunmarkedformbeingacquiredbefore a related markedone.

(21)

CH APTER 1

Theoreti calBack ground

2,1lnJr2<!IlJai<m

As mentioned inChap ter 1,thebasictheo retical approachtakento ac countfor theacquisit ion of theEnglishdative alternation and particle movement is the theo ry ofcoregrammarand markedness,as outlined hyChomsky(198Ia ,b). Various theoreticalaccountsofthe dativealterna tionand particlemovementhave beenposited in theliterat ure.Areview ofsomeof thisre searchis presentedinthis chapter.

2.2 SyntacticTheQry

2.2.1 The DativeAlternation

2.2. 1.1transformationalaccounts

The classicalaccountofthedative alternationinvolvesa transformati onalrule (Fillmore.1965;Jackendoff&Culicover,197 1).Jacke ndoff andCulico v crassume thatthe underlyingorder of objectsin a double-objectcomplementisdirec t-indirect, asfollows:

(22)

"

(I ) John gavethebooktoMary

The dativemovement rulepermutestheobjec tsanddeletesthe prepositionofthe ind irectobject. The rule takesthe general form:

(2) X·V-NP-to-Np·y

I 2)456'">1-2-5-3-0-6(optio nal)

The deletionofthepreposition12isaccounted forbythe rule:

(3 ) X,V-IO-Y

123 4:ll>1-2·0-4 (obli gatory)

Thedeletio n rule(3)appliesaflerthedative movementrule(2).

Emonds(1976),onthe otherhand,accounts(orthe dativealternationbya rule interchangingthe positions oftwoconstituents ofthesamecategory(NP):

(4) X·V·NP·10'Np·Z I 2 3 4 5..>1-4-0-2-5

(23)

12

Emond sclaims that this rulehasthe advantageofbeingstructurepreserving inthai thetwoNP's are moved into positionswhe rethephrasestruc ture rules allowthesame constituents.

He claimsthaihisanalysisgives amore generalaccount ofthe dative alternationthan earlieraccocnrsproposedbyFillmore(1965) andJackcnd o ffand Culico ve r(1971). Theseaccoun ts.which wereno rstructure-prese rving,proposed rules involvingthe permutationof the twoNPobjects.

SmabyandBaldi(1981)accountfordativemovementby abase rulethat gene ra les [VNP PPJstructures.The NPin thePPisthenmo vedbyanoption al rule tothe left ofthe directobject NP,as in:

(5)give a bonetothe dog "">give the dog a bone

However,in anattempt to providea moreunifiedsyntactictheory,Smabyand Baldiarguethat particlemovementand dativemovement,whichareconsidered(0be structure-preservingtransformations.are interconnected.Theyassume thatthe verb +panicleisnoncontiguous in thebasestructureand proposeananalysisthat isbroken downintotwo steps.The firstinvolves arightward NPmovementofthe direct object referredto asCrossParticle Movement(CPM), in whichthe NPmovesacrossthe particle slot.Thesecondstepisdativemovement (DM)proper,whichis theleftward NP movementof the indirectobjectNP tothe positionvacatedby the directobjoct

(24)

13

NP. TIleprepositionisthen deleted asinthe traditionalfo r mulationof dative movement.Their analysis ofthe verb phrase is:

(6) V(NP) (Pn(NP»(toNP)

5maby and Baldi'sanalysisissimilarto Emonds 'inthaiboth claim 10be structurepreserving. However, theydiffer inthat Emonds ' rule involves a simultaneous interchangeof thetwoobj ectNP'swherea sSmaby andBaldi propose [hatthe reare twostepsinvolved.

Larson (1988)alsopresentsa transformaticoalanalysisofthedouble-object constructionwithinthemodemframeworkthatimple mentsa proposal ofdative structure first suggestedbyChomsky(1955/1975).Heclaimsthatthe reareclear reasonswhyonemight wanttoretatepreposdonaldativesanddouble-objectsncrc res transformationatly:

"First.allhoughtherelationbetweenthetwo showsirregularitiesin English,inother languages. therelationship is quite systematic. In particular,in languages with so-called applicativeconstructions(see Marantz (1984),Baker (1985)fordiscussion)obliqueanddoubleobject structuresshowahighly productiverelationstronglysuggestiveof derivationalrelatedness.Thisargues that transformational operations similarto"DativeShift"mustbeavailableinprinciple. Second,a deriva tionalapproachto theda tive-double-object relationisclearly desirableunde r anystrongtheses abouttherelationbetweenstructure andassignmentofthematicroles" (Larson,1988:350).

(25)

14 AttO rdingto this view,adativesenreoce like;

(7)Johnsentaletter[0Mary

is derived froman underlyingform inwhichtheverband itsindirect object makeup

accnsritue nuh a t excludes thedirect objec t.The specificproposal adopted by Larson is thaidative complementconstructionslike(7)invo lveanunderlyingclause-likeVP whose"subjecf'isLkn!:randwhose"obj ect"is!12l..Marx:

(8 )JohnI...,aletter[v-sendto MaryJ)

Thecorrectsurface form is derive dbyan operationofverb raising .1b ismovement leaves atraceintheoriginalsiteandcreates a sequenceofcoindexedv-positions.as in:

(9)Johnsend(VI'aletterLV"110Mary]]

Doeble-objecrformsare thensyntacticallyderivedbydativeshift.Theformerindirect objectM.aabecomes aderivedVP·subj ect" and theforme r directobjecti.kt.1cr assumesadjunctstatuswithin V'.Heclaim sthatthis analysiswouldapply10passives aswell. larson'sapproachdiffersfromthe othertransforma tiooalaccounts in thatinsteadofmovingNP's,theverbmoves.

(26)

IS

Jackc ndoff(1990)preseoustrong argumentsagainst the transformational analysisproposedby larson.JackendoffstalesthatLarsonhasneedlesslycreated a novelstructure forthedouble-ooject construction in orde r10accommodate factson binding,andthatlinearorder, whichplaysanimportant role in the double-object construction , isnottakeniraoaccounr .lackendoff alsocontendsthatLarson'sana.lysis introducesa great dealofstruct urewhichisnotevident from the surfaceof the dative construction. Furth ermore. lackendoff pointsoutthailarson's Detructure representation :

'viola tes twooflhcmostlongstandingand robusthypothesesofsyntactic theory: (I) thataverb's argumentstruc ture isrepresented locally atsome levelofsyntacticstructure,and (2) thaithereisa structural distinction between argumentsandmodifie rs' (Jackendo(f,1990:453).

FurthennorehepointsoutthatLarsondoesnot take lntcconsiderationthe semantic argumentsagainsta dativeshiftanalysisthatmotivated ee proposals for alexical analysis.

2.2. 1.2non-transfo rmationalaccounts

There arethosewhoargue that the dativealternation shouldnotbeaccounted forbya transformationalrule.Forexample,Stowell(198 1).whose'....orkisdiscussed in moredetailinsection2.2.3.argues that throug hthe processof NP-Incorporation.

theneed fOTatransfor mationto accountfor the alternationdisappears.

(27)

16

In a paper dealingwith the"projection problem" which asks howlanguage [earnersacquireadultintuitionsabouttheirlanguagebasedon primarylinguisticdata, Baker(1979) points out problemswiththe transformationalaccount ofthedative alternation.Heargues thatthe classicaltransformational accountincorrectly predicts full grammaticalstatusfortransformed sentencescontainingncnenemartngverbssuch aSSa)'.and~:

(10) "Georgesaid Maxinesomethinguncharitable (II) "We reponedthe police the accidentI

Baker claims thai addingnegative rulefeaturesto exclude nonalternating verbs produces a grammarthatistoo complex with regardtolearning and cites, for example,thewell known factthai childrenlearningEnglishgetlittle negative evidence from their caretakers. Thustheoptionaltransformationrule is too general.Baker further Slatesthat:

"Classicaltransformational theorymakes available forthe descriptionof primarydata from Englisha numberof optionaltransformationalrules that express what appearto bequiteattractivegeneralizations. In manycases. these general izations prove tobefalse. but their falsehoodis not apparentuntil we are provided withthe informationthat certain specific sentencesare ungra mmatical. This isjustthe sort ofinformationto whichchildren learning Englishappear to have no dependableaccess"(Baker.1979:547),

Bakersuggeststhatif wewere to assumethat thereare two phrasestructure ruleswithassociated subcategorization featuresto indicate whichenvironmentsa given

(28)

17

verbappearsin, the problemsassociated with thetransformational accountdisappear. Onlyverbs heardintile primarydata in the double-objec tenvironmentare assumed to occurin thatenvironmentin the grammar.Thisdoesnot, of course,ruleout the possibilityofovergeneralizauonof the rules.

Operatingwithinlexicaltheory,OehrJe(1976)providesa solution10 thedative alternationwhichis very similar to that put forthbyRaker (1979).UnderOehrle's lexicalframework, an alternatingverb is subcategorizedintwo distinctways:[NP PPJ.lNPJ.andthussatisfies Ihe conditions forle xicalinsertionintoboththe prepositionaldative structure and thedouble-object structure. The relationbetween both structuresisstaledbya lexicalredundancyrule.Oehrlestatesthatthis approach has severalattractiveaspects:

"First,rule s ofthis kindarerestric ted to operations onmaterialspecifiedin subcaiegori zancnframes.Thusno ruleof thiskindcouldtakea nounphrase out of a purpose clause, for instance,and make itthesubjectof the sentence.

Second, such rulesaredesigned to accountforcases of syntacticalterations in whichsemantic equivalenceisnot necessarilypreserved.Thus,they are co ncerned withcases in whichnotallpropertiesare invariant.Inview ofthe way suchrules are to be construed,we have an immediateway to build markednessconsiderationsinto therule ttsetr.Finally,rotethat all such rules will of necessitybestructure-preserving,since in everycase the structures related bythe rulesmust meet theconditionsspecified by onephrasestructure rule expansioninorder tobe generated"(Oehrle, 1976:271).

Oehrlealso looksat morphologicaland seman ticconsiderationsof the dative ahcmatlcnand proposesa morphologicalconstraint 10 limit the domain of the

(29)

18

alternation. The lexicalaccount,unlikethetransformationalaccount,hasthe:ability 10incorporate extramorphologicalinformation into theruleitself.

Elabo ratingonOehrle's work,Mazurke wich&While (1984)suggestthat there is evidence oftheexistenceofamorphologicalas well as a semanticconstraintin the acquisitionofthedativealternation. The morphologicalconstraintdictates thatan alternatingverb mustbeofthenativestemclass.Thesemanticconstraintstatesthai the indirectobject mustbethepro spectivepossessor of thedirectobject in ado uble- object construction.Further discussion ofMazurkewichandWhite (1984) can be found insection 2.3.1.1.

2.2.2 ParticleMovement

likedativemovement,particlemovementhastraditionallybeenaccountedfor by atransformational rule(Emonds,1972,1976.1985; Fraser.1976;Smaby and Baldi, 1981).Ho wever,manyresearchers disagreeoverthe deepstructure position oftheparticle;it mayormaynotbecontiguoustothe verb.

2.2.2.1contiguo usanalyses

VanDongen(19 19)lookedatverb-particlecombinatio ns and concluded:(a) that theparticlegenerally precedesthedirect object,(b)thatthe particleusually

(30)

19

qualifiesthe verb andoccurs inclose proximity 10it,and(e)thatthe original meaning oftheparticle is oftenlost.

Live(1965 )arguesthat the contiguity oftheverb andparticleissupportedby la) the retention ofthe:particle alongwiththeverh-ccmponentin thepassive,as in

(12) The disheswerebroken up

(b)the abilityof Ihe particletobesubstitutedbyaone-wordsynonym:

(I] )find out"" discover

(elthefact thattheverb-particlecombinat ionreadily occursinconjunctional parallel witha singleverb,asin:

(14)I sent forandreceivedIhegoods

(d) the grammaticaljuxtapositionofthetwo elementsinaverb-paniclecombination.

c.g.~.~.l

(31)

20

Bolinger(1971)claims that the"phrasalverb"is a lexical unit havinga"sct meaningwhichis not thesum of the meaningsof itspar ts"(Bolinger.1971:xi;).He furthe r arguesthat when a ruleofsyn tax bringstogethe r particularwordsmore freque ntlythan others. as,for example.in the case of phrasal verbs,thehigh- frequencycombinationstendto fossilize. The particlehas a tendencytobefound in the post-verbalposition.as in (15), andalessertendency10befound afterthedirect object(16):

(15) totakeout thegarbage (J6) totake the garbage out

Similarclaimshave been madebyAbsalom (973),who positsatlea stfo ur factors contributing to the treatmentof theverb-part icleco mbinatio n asa unit: (a ) the slrr" ....rsyntacticbehaviour of alarge numberofexamples.(b) thestatisncalco- occurrence of thecomponentsof the combination.(c) an intuitive desirabilityto treat the structureas aunit, (d) thesemanticinterdependence oftheconstituentsofthe combination,

Fraser(1976), indistinguishing particles from prepositions.claimsthat the verb and particleareintroduced into deep structureas a contiguousunitwhich isdominated by the constituentV. He arguesthat particlesaresyntactica lly more closelyassociated

(32)

21

withtheverbwhich precedes them, whereastrueprepositionsare mere closely associatedwiththenoun phrasewhich follows the m. Based onthisevidence,he concludesthat the particlecannotbe categorized asa preposition,asEmonds (1976.

1985;claims.To illustratethe differencesbetween particles and trueprepositions, Fraser givesthefollowingexamples:

(17) a. Harrylooked furtivelyover thefence b. "Harry looked furtively over the client (18) a. Inrhestreet,theman reeledasifdrunk

b. -rnthe line,the manreeledasifdrunk (19) a. He sped up thestreet, andshe, upthe alleyway

b. "He sped uptheprocess, and she,up the distribution]

In (l7a),theadverbial~canprecedethePP,whereas,in(I7b),itcannot precedethe particle.In(18a),Ihe ppcan occur in sentence-initialposition.whereas in(lSb)particles cannot.In(19a), PP's can functionas a syntacticunitin sentence inwhichgappinghas occurred.whereasin (19b) the particle!illcannot.

2.2.2.2non-contiguousanalyses

Emonds(1976)argues that the particleisan intransitive preposition. i.e. an instance of PP. He maintains that theseintransitive prepositions must follow the direct

(33)

22

objectNP indeepstructure,lind thai a movemeOi Iransfonnalion.whichis structure preserving,optionally moves them nest totheverb.

Emon d s presentsevidence of diffe rentidiolect s of Englishinwhichhe found thatthemostfavoured and natura!positionforparticlesin sentenceswithverb• double-obj ec t order is betweenthe twoobjectNP' s as in (20a):

(20) a.JohngalleMary backthecom b b. JohngaveMarythe combhack c.John gavebackMarytile comb

Aswell,hefou nd Ihalthe position of the panicle after thedirectobjec t,as shownin (2Ob),wasusually rejectedandthaitheidiolcc ts differedwhenIhe particle was positioneddirectly aftertheve rb as in(2De).

There is a problemwith Ihis amtlysis:ilislimilat to sentencesthatcontain only alternatingverbs in theirdoublc-object form. Neitheralternatingverbs in their prepo sitional form,nornon-alterna tingverbsappearinthediscussion.

Based on his 1976 study. Emonds (198.5) propose s the follo wing transformation al ruletoaccount for insta ncesof[V.Pnl:

(21)V·NP-Prt->1·3-2

Heclaimsthat insentences containingidiomatic expressio ns .like:

(34)

23 (22)Johntoo k his student to task

thedirect objectobligatorilyintervenes betweenthetwo parts of[heidioml22k....w 1iUk.Emonds arguesthaiifthedeep structure of thisidiomhadthe structureV· PrI.

thenin orderto accountforsentencessuch as (20),its derivationwouldrequire an obligatory transformation tomove thePP part ofthe idiomintothe position following thedirectobject.Basedon evidencesuch as this.Emondsconcludesthat theverband particleina verb-panicle construction are notcontiguousin deepstructure.However, asEmonds states, (22)isan idiomatic expression;theexceptionratherthan the rule.

Asdiscussed in 2.2.1.1.Smaby and Baldi(1981)argue that thebaseposition forthe particle is followingthedirect objectNP.Theyposit a movementrulewhich lheycall"Cross-ParticteMovement-thatisarightwardmovementof thedirect object NPacrossthe panicleslot.Unlike Emonds'analysis.it isnottheparticle which moves,buttheNP whichprecedesit Following Emonds,Smaby andBaldi hypothesize thaithismovementruleis a structurepreservingtransformation.This cross-particlemovementrulewouldtransform (23a) into(2Jb):

(23) a.John threwthegarbage out b.Johnthrewout Ihegarbage

(35)

2.2.3 The Stowell Analysis

Stowell (1981) has developed att;~orywhich is rooted in syntax, but which cannotbeconsidered transformational. Stowell's analysis incorporates Case theory as outlined by Chomsky (l981b).Case theory is a subsystem of UGthat assigns abstract Case to NPswhich indicatesthe grammatical functionsof Nps underthe conditionof adjacency.The adjacency condition on case assignmentstalesthat NP objects must appear adjacent to a governing verb or preposition.Ina sentencesuch

(24)johnsent a leiter to Mary

case assignment followsdirectly from the theory.that is,~is the direct object andMao'.is the indirect object. This structureis assumedto be the unmarked dative structure.The problem. however. arises in the assignment of Case to double-object constructions as in:

(25) John sentMaryaletter

or with verb-particleconstructionswhere the object is not adjacent to the verb, as in:

(26) John gave back the book

(36)

25

Stowellpresentsthefirst unified account ofdativeand particle movement.Hepoints outthaiwe neednot assumethaithegrammarof Englishcontainsalanguage-specific ruleof dative andparticlemovement.Rather, the movementstructuresfollowif one considersthai English has the word-formationrulesof Np-Inccrporation andPanicle Incorporationwhichcanapplyseparatelyorsimultaneouslyto a singleverb.Stowell mainta ins thatthe case assignmentproblemsassociatedwiththedouble-objectand the verb-particle constructionscan betraced tothe assumptionthatbothNP objects in a dative construction,as wellas the particles in a verb-particleconstruction,are complementsof theverb.Under Stowell'sanalysis.thefirst NPinadouble-o bject construction andtheparticleinavcrb+particleconstructionare actuallypart ofa complexverbphrase:

(27) a.[v-NP]

b.Iv-Prtl

TIleverb-internalNP has thestatusofan incorporated objectandtheverb-internal panicle hasthe statusof anincorporated particle.

As slatedearlier,theadjacency conditiononCase assignment requiresthat for anNPtobeassigned case, the NP must be adjacent toits governingverb.This conditionposesaproblemfor thesecond NP ina double-objectconstruction and the

(37)

26

dirtCtobjectNP in averb-particleconstruction.However.IhrollghNp-Incorporntlon and ParticleIncorporationtheadjacencyproblem on case assignmentdisappears.

Considertheexample:

(28) John gaveMary(hebook

The indirectobject NP isincorporatedwithinthecomplexverb[vgave-Mary!andcase is assigned10the directobjectNP[Npt hebook]underadjacency. Theresulting structure is as follows:

(29)John [v·[vgave-Mary](lI1ebooklJ

Similarly, insentences containing the verb-particleconSlruclion:

(30)Iswitchedorr thelight

the particle.which is adjacenllo theverb.isincorporated within the verb to formfbe single complex unit {v!.witchcd-off].Thefollowingis thesymacuc structurefor (30):

(31) I[vp[v'[v-switched-o ff]lheIighllJ~

(38)

27

The direct objectNPlb£.l.ii.hIis adjacentto the complexverbafterapplicationof ParticleIncorporation.' Similarstructuresarise afterthe applicationof NP- Incorporation on double-object structures.

Basedon theprecedingdiscussion,itisassumedthat the dativealternation is betteraccountedforbyalexicalanalysisbased on Oehrle(1976).Ithas beenshown that transfo rmational theory has problemswithconstraining the rule;it can account forverbswhich alternatebutit has no methodother thanadding exira rule features to predict when a verbdoes notalternate. Baker (1979) has pointedoutthat under transformationaltheory,thedative alternationis writtenas an optionalrulewhich automatically fails 10applyto nonaltematingverbs.As well,the transformational analysiscannot accountforthemorphologicaland semantic constraintsproposedby Mazurkewich andWhite (1984) which have been shown to successfullylimitthe domainof the alternation.

On theotherhand,thelexical analysisaccounts for the alternationusinga redundancy rulewhichsubcategorizesalternatingverbsas[NP PP]and(NP NP);this wouldnor applyto nonaltemating verbs [NPPP].The theoryof Case assignment from whichmarkednessconsiderations follow determines the coursetaken in acquisition.Theunmarkedstructurefor thedativealternationis considered to be[NP

(39)

28

PPJwhereasthemarked structure is[NP NPI.Finally. as Oehrte(1970) pointsouI, the lexical accountcan incorporateextramorphologicalinformatio n providedbythe morphologicalconstraintwithin therule itself. Itdoesn'tnCL'tIextra rules as the transformatio nalaccount does.

Particle movementis assumed10beaccountedforbyanoptionaltransformation whichmovestheparticleto the right of the object NP.The unmarkedstructurefor the verb-particleconstructionis considered10 beIv-Prtl with thepaniclecontiguous tothe verb.Itis assumedthat thecontiguousverb-particleforms areunmar kedbased on the semantic and syntacticarguments that treatitas a singlelexical unit(Van Dongen19 19. Live 1965,Bolinger 197 1, Absalom1973. Fraser 1976 andStowell 1981). The evidence thatEmonds(1976,1985) usesto argue fora noncontiguous analysisis weak anditrelies on idiomatic expressions. The hypothesisforwardedby Smaby andBaldi (1981)is alsoweak as theyproposean analysiswhereby theparticle remains stationaryand the directobjectNP movesover it. However, the iranalys is of particlemovementrelies crucially on theiranalysisofthe dative alternation which theyma intain is a movementtransformation.Itis assumedil1!ll:s thesisthat the dative alternation is best accountedforby alexica l analysisasthe transforma tionalaccount has beenshowntobeflawed .

Ithas been arguedthat for the dative alternation,the unmarkedstruc tureisthe prepositiona ldative [NP PP]andthe markedstructureis thedouble-objectstructure [NP NPJ.Like wise forthe verb-particleconstruc tion,theunmarked structureIs a verb

(40)

29

witha contiguousparticle[V'Pr1) and the marked structure isonewith a nonconnguous particle",Returningto sentences5-10in ChapterI which involvethe interaction ofthedative alternationandthe verb-particleconstructionwecanpropose acontinuum of markednessfromleastmarked tomostmarked. Sentence(5) (renumbered sentence (32»:

(32) John gaveback[NplhebookJ[pplOMary)

contains the unmarkeddativestructure[NPPPJ as wellas the unmarked [V-Prt]

structure. Thus, thistype ofsentencewould be theleast marked. Sentence(6) (renumberedsentence (33)):

(33)John gave[Nplhebook)back [wlOMary]

containsthe unmarkeddative structure [NP PPJandthe markedverb-particlestructure in whichtheparticle is separated fromthe verb. Thus, sentence(33)is moremarked thansentence(32).Sentence(7)(renumberedsentence(34)):

(41)

30

is also more markedthansentence(32) becauseitcontains the marked INP NPI

structure andtheunmarked(V·pn lstructure. SentenceIS)(renumberedscmcncc (35»:

(35)Johngave(,...pMary) back[N,t hebook}

contains thetwo markedstructures: double-object dative andnoncontiguousverb- panicleconstruction. Sentence(35) is thusmoremarked than the previousthree.

Sentence(9) (renumberedsentence (36»:

containsthe markeddouble-object dative as wellas a separated verb-particle

construction. However,notonly is the particleseparated.itis also semcnce-rlnal.

the furthestpossibleposition fromtheverbwhich may account forits doubtful grammaticaIity.Sentence(10)(renumberedsentence(37)):

(37)Johngave[NPthebookHpplOMary]back

containsthe unmarkedprepositionaldativeanda sentence-finalparticle.Thissentence isnot grammatical.Althoughsentences(36) and (37) wouldnotbegeneratedby the grammar and,hence,norbeavailable tolearnersin theinput,they wereincluded in

(42)

31

thetesting asthey couldprovide valuabledata aboutthesubjects' reactions to ungrammatical sentences.

Basedon thiscontinuumofmarkedness. itisassumedthat the leastmarked sentencesare easier10acquirethanthemort markedsentences.

(43)

32 rooT~OTES

From Baker(1979) .

2. Allexamplesfrom Live (1%5).

3. From Fraser (1976).

4. From Stowell(1981).

5. Aarts(1989)hu pointedoutthat Stowell'sanalysis failstoaccountfor sentencessuchas:

(i)floo kedthe information up

in which thereisno subject-predicaterelationshipbetweenthedirectobjectNP andthe particle.According10Aerts.Stowellonlylakes into account sentences such as(i)for whichthere isatrue subject-predicaterelationship.

6. ItshouldbenotedIltatthe markedconstruction inthecase ofthedative alternatio n resultsfromthe application of alexical rule. Themarked conscuctionthai results from particlemovement,ontheotherhand,isderived bymeansof a movementrule.However.itis argued herethai thisdoesnot alterthepredictions beingtestedinthisthesis whichconcernthetheor etical assumptionsunderlying thenotionofmarkednessandcoregrammar. Thisdocs raise, nonetheless, thequestion of whetherageneralmovementrulewouldbe acquired before a lexicallyconstrainedrule.

(44)

33 CHA PTER3

Acqu isitionResea rch

The followingare brief summariesof studieson the first and second language acquisitionof the dativealtemation and particle movement,

3,1Foulish asa FirstLanguage

3. 1. 1 Acquisitionof Datives

Fischer(1976)studiedEnglish speakingsubjectswhoseages rangedfrom3;6 to5years .Thetasks consisted(Ifan elicitedimitation taskandapicture choice task.

The resultsshowed tbat in the imitationtask, themost frequenterror was"de- transformation ." That is, Fischer foundthat a sentence like (I) wasoften de- transformed into one suchas (2):

(1) John boughtMarythe book (2) John boughtthebookfor Mary

(45)

3'

Thissupportsthehypo lht:sis thattheunmarked struc turefor tile dativealternationis (NP PP]becausethechildren are not abletorepeatthe,as yet,unacqui red doubleNI' complement.

Fischer alsoreportsthaithere wasamethodologicalproblemintheexperiment.

The resultsshowedthatforfull NP objects.thefour-year-olds responseswere governedbytherecencyeffect; tha t isthey chosethelast item they heard.Five-year- oldswere betterable to overridetherecency effect compared 10theyoungerchildren butitwas found thai theirgrammar was mo relibera l thantherour-ycar-otds'inthat theyacceptedmoreungrammatical sentencesthanthe youngerchildrendid. Five-yo..ar-

oldswe realsoshown10overg ene ralize double-objectforms to co nte xts wherethey arenotgrammatically acceptable.

Asmentioned above,subjectsshowedapref erence forthePPversion rather thanthe double -objectversion orindirectobject construction s.In a moreextensive studycarriedout by Fischer (1971),sheshowedthat doobte-cbjectconstructions:

(3) I gave thegirlthebook

arelearned muchlater andaremore difficult forthechildto processthan their direct objec t plusprepositionalphrase counterparts:

(4) Igave the bookto thegirl

(46)

35

Cook (1976)lookedallheac:quisitionofthedaliveallenlalionbynaliveEJlglish

childrenbetweentheages of 5;0and10;0.Testing consistedofaslcing each childto move toysacoordingto the instructionsof thetest sentences.Sentences containing combinations ofthe verb~,the indirectobjc:ets&iIIand man and the direct objects

grandI2QQkwereused.Half of the sentencescontainedthe(NPlcomplementand theotherhalfcontainedthe[NPPPJ complement.Resonsshowed thatthe subjocu mademany mistakeswiththe [NP NPlcomplementbut very few withthe [NPPPJ co-nptement. Aswell.therewasgreateraccuracywith age forthe double object construction, but not significantly for the prepositionaljg-ccnstruction. This supports thehypothesis thatthel!2'<'onSI11Jctionis acquiredbefore thedouble-object construction.The subjectsalsoshoweda greaterunderstanding of sentences inwhich thedirect objectwas inanimateandtheindirectobjectanimate.This suggestsmat animacy is animportant aidin the acquisitionprocessofthedativealternation.

MazurkewichandWhite(984) arguethat child~initiallyacquiringthedative alternation havea rulebased on positiveevidence whichis more generalthan the adults'rulewhichleadsto overgeneraflzationof thedativealternation.They propose that alternatingverbshavetwosubcategorizauons,[NP PPJ,and[NP NP1.relatedby alexical redundancyrulewhich includes a semantic andamorphologicalconstraint.

The morphologicalconstraint dictatesthat the verbsinvolvedmust befrom thenative- stemclass,notLatinateverbs.The semanticconstraintstales that the indirect object mustbetheprospectivepossessor

oree

directobjectin thedouble-object construction.

(47)

36

Theresultsof thispaper arebasedmainlyonresearchcarriedoutby Mazorkewich.(1982)ofmreegroups ofLIEnglishchildrenwhosemeanageswere 9.0.12 .3and15.6years. The subjects weretestedbymeans of anIntuitive Judge ment Test whichelicited gramrnaticalityjudgementsofpairsofsentencesmade upofeither a verbanda {NPPPIco mple mentoraverb and a[NPNPJcemptemen r.

Someoftheverbs optionallyallow the alternation;forexample,~.whileothers obligatorilypermit onlyPPco mplements,forexempte.~.

The resultsshowedthat allthreeexperimentalgroups wereaccurateinassessing the gra mmaticalityofsentencesco ntaining verbswhich alte rnateaswellassentences withLatinate(non-native)verbs takingonly{N? PPJcomplements . Howe ver, sentencescootaininglatinate verbsand double·NPcomple ments:

(5) "John explainedMarythe answer

weremorewidelyacceptedbythesubjects thanwou ldbeexpected .suggesting·UIC possibi lityof overgeneraJizationby speakersold enoughtoknowthe rele vant verbs·

(Mazurkewich and White. 1984;268).Theysuggest thatthisovergeneralizarionwill belostwhenthe childbecomes aware that an alternatingdativeverbmustbenative and that a certainlypeof semanti c relationship,namely the prospectivepossessor, must bepresent.Thelossoftheovergeneralizationwillbe broughtabout throughpositive evidence in the input,andthechild 's awareness of thesemantic constr aintbefore lhe morphologi cal constraint.

(48)

37

White (1987a)testedher hypolhesisuatcbudrenwho have acquired thesyntax ofthedouble-object construct ion.buthavenotyetlimitedtheindirect object to those NP's which are the prospectivepossessors of the direct object, may be overgeneralizingthesemanticaspectof thedativealternation.The datafrom20 children whose ages ranged from 3;8to5;8 wereused. Testingconsistedof two tasks:an act-out andanimitationwk. For theact-outWk.childrenwereasked, usingtoys,toperformthe action describedinasentencereadaloud bythe experimenter. Forthe imitation task,thechildand theexperimente reach helda puppetand thechild was asked10 make his orher puppetrepea twhatthe experimente r'spuppetsaid.White used verbswhich alternatein the adultgrammar,

suchasll..ulli::.Gt.hu.i.ld.aswellas verbs whichoccuronlywithbenefactivefur·

phrases whichdonot alternateinthe adultgrammar;~.QI2CD..~.

Incomparing the resuhsfrombothtasks,Whiteconcluded that ingeneral,the subjectswerecvergeneralizingthedouble-objectstructure10nonaltemating verbs.

However,theoldersubjectshadatendencytoovergeneraJize muchmore thanthe youngersubjectsdid.Thissuggeststhatthis overgeneralieationisnotindicative of problemsin theinitialstagesof acquisition.Instead,itconstitutesanexample of the type of overgeneralizauoninwhicholderchildren failtolimit syntacticor morphologicalrules10thesemantic class to whichthey apply intheadult grammar.

Thereare seriousproblemswiththisstudy,aspointedoutbyGropen,Pinker, Hollander,Goldbergand Wilson(1989).Adultsare equallyas capable ofacting out andimitating ungrammatical sentencesas childrenart.The factthatthe children in

(49)

38

thisstudy do sowhen instructed10tell s us nothingabouttheir acquisitionofthe double-object construction. As well,in thetest sentences,the directobjectswere alwaysinani mate and theindirect objectswere always ani mate. Thus,thechildre n may havebeenable tocorrectlyactout andimitate theungrammaticalsentencesbased onanimacy, ignoringthesyntax of eachsente nce.The conclusiondrawn by While that Ll learners of Englishovergeneralize thesemanticaspectof thedativealternatio n isnot tenable.

Gropenet al.(1989) reanalysed data fromtheBrowncorpusandconducted a seriesofexperiments totesttwohypot heses: theStrictconservatism hypothesis . whichpredictsthat the doub le-object for m willnot beusedunlessithaspreviousl y bee n heardinthat form;and thecrite ria-governedproductivity hypothe sis, which predicts that both optionsofthe dativealterna tionwill be used productive ly.

InStudy I, thespontaneousspeech of five nativeEnglish childre n and the ir caretakers wasanalyzed. The subjectsfor this study included(hethree child ren studiedlongitud inallyby Brown (1973 ) Ada m,Eve,and Sarah,and two othe r child ren, Ross and Mark, whosetranscri ptwas obtained from the ChilOE.o;projec t of MacWhin ney and Snow(1985).The ages ofthe subjectsrangesfrom 1;5 · 2;7 at the startofrecordi ng to 2;3-6;6 at the end. Utterance swereclassified eitheras double-objectorprepositionaldative.Eleven classes of alternatingverbscompatible with the no tionof causing a change of possessionwereexaminedfrom thecorpus.

(50)

39

The resultsshowtdtha t neitherversionof ue dativecon~istentlyemergedfirs t, contrary toevidence presentedby Fischer (1971. 1976), Cook (1976) and MazurkewichandWhile (1984). Almost all ofthechildren uttered atleast some do u b le-objec t sentencescontai ningverbsnotheardinadultspeech whichrefutes the stro n gest version ofthe conserva tism hy pothesis,However,thesubjects in the Gropen etat.studyare younge rthanthose stud iedbyCook,whose subjects' ages ranged from 5to10yearsandthose ofMaz urkewich andWhite,whosesubjectsrangedinage from 9to15;6.Fischer ' s subjects ' agesrangedfrom3;6to5years,again slightlyolder than thoseofGropcnet01.1-

The first experimentin the stud ybyGropenelal. wasdesigned tolestwhether the moqJho phonolog icaJ(i.e.monosyllabici ty)and semantic(i.e. prospectiv e possession) ccesraints proposed by Maz urkewich and While (1984) are psychological lyrealfor adulls.They suggestedthatjftheseconstraints are not.then theycannotaccountfor how children avoidorrecoverfrom overgeneralizationsof thedative. Theirsubjectswere adultfirstlanguage speakers of EngUshfrom 17-41 years. Using aque stionnaire.eachsubjectwas asked (0 rate the acceptabilityof double-objectsentences containing nove lverbs usinga seven-point ratingscale.The results show that subjects jud geddouble-objec t sentences whichinvolveda change of possessionas beingsignificantly moreacceptablethanthose whichdid not.Aswell , subj ects judged sentenceswith monosyllabicverbsas beingsignificantly more acceptablethanthosewith polysyllabic verb s. Thus. boththe semantic and the

(51)

40

morphologicalconstraintsondativesinEnglishwere shown tobepsychologicallyreal fortheadultsinthisexperiment.

In theirsecondexperiment. Gropen eta). testedwhetherthemorphologica land thesemantic constraints are psychologicallyrealforchildren.NativeEnglish-speaking children between the ages of 5;0and8;6 wert taught fournovelverbs;two monosyllabicandtwotrisyllabic.involving a transferofpossession. Thetasks consisted ofaproductionand a comprehensionlestdesignedtoelicitboth forms of the dative.Results showed thatthe children did prod ucedouble-object forms with verbs theyheardintheprepositionalform aswellasthe dooble-object form.although theypreferredusing verbsinthe constructionsinwhichtheyweretaught. The subjectsalsoshowed a preferenceformonosyllabicverbsoverpolysyllabicverbs which supportslhemorphologicalconstraintproposedbyMazurkew ich andWhilC (1984).

In their third experiment,Gropenetal.attemptedto elicit double-object forms inmore natural settings. Thesubjectswere 32 nativeEnglish-spea king children between the agesof5;8 and 8;11.Tbesamefournovelactionsas inEJ:pcriment 2 were taughtto thesubjectsaswellas fou rnew novel stems:twomono syllab ic and two trisyllabic. Results showed thaichildrenusenovel verbs inthedouble-o bject construction evenif theyhavenever heard such combinationsbefore. The resear chers also found thechildrenweremorelikelytoproduceadouble-obj ec t form if the recipientcouldbeunderstood asanimate than ifit wasunderstood asinanimat e.

(52)

41

Basedon these experiments, Gropen etal.concludethat a weak versionof Strict Conservatismis supportedas theyfoundthatthe productiveusesof verbs in the double-objectformconstitutea tinyminorityof children'susagesandthe vast majority ofthe verbsthey used in bothdative constructionscouldhave beenbased on the languageinput childrenreceivein theirenvironment.

Theyconsiderthatthecriteria-governed productivity,whichholdsthatchildren learn to constrain their rulerc apply 10 monosyllabic verbs denoting possession changes.is consistent withthe data, but foundthat children are not as productive as thishypothesis wouldpredict.Gropenet al.postulate that:

"ifthedative rule changes the semanticstructureof a verb,thenthe interpretationof a sentenceshouldbe ableto changewhenthe verb dativizes:whereas the prepositionalformspecifiesmotion(literalor metaphorical) towardsagoal,thedouble·objcct form specifiesactual causationof possession"(Gropenetal., 1989:242).

They givethe exampleof tileverb

w .

which.initsprepositional formis ambiguous betweensending to alocation ortoaperson asin (6):

(6) Johnsentthepackage to".Mary ,orToronto

whereasin the double-objectform it can only mean sending 10aperson (7), notto alocation(8):

(7) Johnscnt Mary tile package

(53)

42 (8) "'Joh nsentTorontothe package

GropenetaI. view the dative asManoperationon lexicosemanticstructure"

which changes"causeYto goto X"to"cause Zto havev",The)'suggest that this hypothesis solves four problemsat once:

"itexplainswhydifferentargumentsgetmapped ontothesyntactic sur faceobject-positionin the twoconstructions;it explainswhythe interpretationof a singleverbcanchange whenitundergoesthe alternation;itexplainswhyverbswhich takethe prepositionaldative formandare incompatiblewith causation ofchange ofpossession cannot betransformedintolaking the double-objectform; and, symmetrically,itexplains why certain verbscan onlyexistin the double-objectIo-m"(Gropenetat.,1989:242).

3.1.2 Acquisitionof Particles

Thereareveryfewstudies reported in the literaturethatfocusonthe acquisitio nof verb-particleconstructions.One study isthat by Fischer(1976).who loo ked at the acqu isition of verb-particle constructions in LIEnglish childrenwhose ages rangedfrom3;6to 5 years. A sentencechoicetask was administeredwhich tested gram maticalityjudgementsof verb-particleconstructions.Resultsshowed that both particleexte rnalsentencesas in (9' ,andparticle internalsentencesasin (10', are equallygrammatical for these subjects:

(54)

43 (9) Johnpickedthebookup (10) John pic kedup thebook

However, Fische r reportsthat there was a strongrece ncy effectinherexperiment;

thatis.thechildrentendedto choosethe lastitemthe yheard.

Clifton (1977) testedtheacceptability ofparticles in sentenceswhichhave undergonethedative-movement transformation.His subjectswereadultLlEnglish studentsof Spanishas a SecondLanguage.Fourclassificationsof sentenceswere used: (a) post-ve rbalparticle:(b)particlebetweenNP's;(c) sentence-finalparticle; Cd)passiveformedon indirect object. Thirty-two lest sentencesofthesetypeswere presented to thesubjectsinoneof threemodes: oral,written andtimed, orwritten andunnmed. Thesubjectswereasked10indicateOle acceptability ofeachsentence by circling thenumberon asix-pointscalecorresponding to theirjudgement.

The results showedthatthere was considerablevariabilitybetweensubjectsas 10theacceptabilityotme testsentences.However,ingeneral,theacceptablepositions forthe particleseemed todependonueparticle itself. Forexample,&nL2lUand

~werefoundtobeacceptableifthe particlewas post-verbalor between NP's, butnotifthepanicle was sentencefinal;whereas~,wasacceptablebetween NP's or sentencefinal, butnotpost-verbal.Thispatterning suggests that thereis continuumof acceptabilityalongthelinesproposedinthisthesis.

(55)

44

Browman (1986)conductedtwoexperiments wilhadultLIEnglimspeakers 10testteeterswhichshehypolhesiztd10 aff«tseparationandcohesion of the verb and particle inthe verb-panicleconstruction.In Experiment I.Browmantested64 college studentsin california.Taking severalfactorsweh as animateV5 .inanimate directobject into account,a wriuen lest was designedto elicitsentencescontaining liltverb-particlecombination. Resuhsofthisexperimentshowthatalthoughmost subjectsshowedvariabilityintheir responses, afewsubjects consistentlyseparated the verband particle,while others consistentlyplaced the particlenext10 theverb.

The group asawhole showedaslightpreference fornotseparating theverb and particle.

A secondexperiment was designedtotest otherfactorssuch as idiomaticvs.

adverbial usage,whichwerehypothesizedtoaffect contiguityof theverb andparticle.

Resultsfromthis experimentshowed thatthe fac torswhichsignificantly favoured contiguity wereidiomatic usage,andvowel-initial particles.

Basedon theresults ofthesetwoexperimen ts.Browmanconcludes thatthe placementofa particle isnotcompletelyoptional butisaffectedby semanticand phonologicalfactors.As Wt:1I.she concludes that individualsubjectshavetheirown personalpreferencesforseparationorcontiguityofverbs and particles.

(56)

45

3.J.3 AcquisitionoftheInteraction of Dativeand Verb-ParticleConstructions

Asfar ascan be determined,the onlyacquisitionstudies that havebeendone onthe interactionof the dativealternationand verb-particlest ructures on subjects whosefirst languageis EnglishwascarriedoutbyFischer(197 1,(976).However, Fischerteststheeffectsof stressedand nonstressedparticles,andshe also contrasts the effectsof unstressedpronounscomparedto lexicalnouns.Thus.the goalsof her experiments involveissues differentto thosepursuedin thisthesis.

3.2 Englishas aSecondLanguage

3.2.1AcquisitionofDatives

Mazurkewich (l984b)attempted to demonstratethat evidencebasedon the acquisitionofdative structures inEnglishbysecondlang uagelearn ersprovidessupport fora theoryof markedness.In English,adativeNPcan appeareitherin a PP,which isassumedtobeunmarked. oras thefirstNP of a doubleobjec t construction,which isassumedtobemar ked. Mazurkewichhypothesizedthatthe unmarkedversionof the dativewillbeacquiredbeforethe markedone. The datafor thisstudycamefrom previousresearchdonebyMazurkewich (1984a)onnativeFrenchandlnuktitut speakers,whose averageage was 18.0and17.0yearsrespectively.Thesubjects were

(57)

,.

classifiedinto threelevelsofproficiency inEnglish. TIoIo'Onauve Englishcontrol groupswereused,Ihe second of which had anave rag e agecloser 10thaIof lhe experimentalgrou ps.

Testing consistedofintuitivejudgementsofsimple declarativesentences containinga set of 12-dativeverbsanda setoff2I.dativeverbs. Tilesentenceswere classified intofourtypeswhichincludedalternatingverbsintheir double-objectam! p~si tiona1phrase forms,non-alternatingverbs, and distracto r sentences.

Resultsshowedthatbothexperimentalgroupsacquiredthedativeprepositional phrase complementsINPPPJ beforedouble·NP complements [NP NPI'.Aswell, Mazurkewich foundthai FrenchandInuit speakers acquirethe Englishdative alternationinthesamewayasfirst languagelearnersofEnglishdo. However, as mentionedabove,theGropenet al.(1989)study indica testhat bothstructuresappear inthespeech ofve ryyoungchildren.

LeCompag non (1984) looked atthe acquisitionofEnglishdativeverbsby native French speakers. Shepresenteddatabasedon twocasestudies and an experiment.Ineach of thecase studiesshe examinedthespontaneousspeechofan 1I French speakerlearning Englishasasecond lang uage. The first studylasted approximately4 months,andthe second, approximately 2.Inthe experiment,she administeredtwo grammaticalityjudgementtaskscontaininglQ:andfQr.dative verbs tofo uradultsubjec ts whose firstlanguagewas French.The firsttestcontainedfull NPindirectobjectsandthesecond. pronominalindirectobjects.Basedonherfindings

(58)

47

fromthesltxlies.leCompagoon concludedthatthe acquisitionof dativesbysecond langu...gelearnersclosely followsthatofLI learners butthaiLIinterference plays a roleinthe:acquisition of a secondlanguage.Furthermo re, sheconcludesthaiL1

Frenchlearners of English as a secondlanguage usedifferentstrategies inacquiring fullNJ'indirect objecls andpronominalindirectobjectsas aresultof interference from French.

There arcclear problems with thisstudy. Inthe firstcase study,Le Cornpagnonslatesthaterrorssuchas:

(11) "Youexplainedmethe rulemany limes

are a resultofIhe LI Frenchlearnertakingthedouble-objec:t for m ofthedative to betheunmarkedform. Sheattributesthiserror [0interferencefrom French

cliticization.IntheMazurkewi<:hand While (1984)$Iudy,onlylexicaldativenouns wereused. The assumption thatLeCompa-goonseems tubemakingis thatpronouns

and nounsbehave in asimilarmanner,but she presents no evidencetoindicatethai thisis thecase. Oneother problemwithLeCompagnon'sresearchisthatthe experimentalstud~·doesnol containproductiondata.

White (198Th)tested thehypothesisthatlhe secondlanguagelearner'sprior Hngulsttccxperience maypredisposehimor her towards transferring markedstructures from thefirstlanguageto thesecond.The developmentalhypothesis(Mazurk.ewich, 1984,1985)of markednessassumesthatthe learner Starts outwith theunmarked

(59)

48

hypothesis. andwillacquireunmarkedformsas a necessary development",stage beforethe acquisitionof markedforms. Whilesupports thetransfer hypothesiswhich assumes thatthelearner'sL1 playsarole lnthe acquisitionof a secondlanguagein thatthe L2lea rner may transfer marked forms fromthe LIto theintcrfanguage . Whiteproposesthat a situation inwhich the LI marksa structure but theL2 does not would reveal thecorrect hypothesis. Sheclaimsthat inthe developmentalview,such a situation providesno occasionfor thecccurrcnce....fmarked formsin the interlanguag e.The [earner'sinitial hypot hesiswillbethatunmar ked is required,and thiswill remain thehypothesis becausethe L2 in questiondoes notexemplify marked forms.Ifmarked formsshow upinthe interlanguage, this wouldsupportthe transfer view,becausenothing intheL2 evidence couldhave motivatedthese forms, whereas theLIdoes containthem.

In two studies, Whiletested learners of French as a secondlanguage(FSL) usinggrammaticality judgement tasks on thedouble object construction.which is gran-metical in EnglishbutungrammaticalinFrench.

In the first study,shetested 27 adult subjects.Approximately onehalf of the subjects wereEnglish speakersand the other half came from a varietyof different languagebackgroundswhichdo nothavethe double-object constructionas part of their grammar.The test includedfiveu"~rammaticaldouble-object sentences,andthree grammatical dative sentencescontaining the[NP PPJcomp lement.

Results fromthe judgements involving ungrammatical double -object sentences suggest that both experimental groups have difficulty in recognizing the

(60)

49

ungramrna ticaJity ofthese sentences. While suggested thatthisgoes againstthe developmentalhypothesisbecauseinthat view.L2 learners still have accesstothe unmarked case andrevert toit.Therefore,they oughtto reject themarkeddouble- objectconstruction.Theresponsesto thegrammatical sentencesshowthat bothgroups aTC accurate, and that the differencesbetweenthetwo groupsisnotsignificant.

The secondexperimentinvolved childFSLlearnersfromgrades5and6 whose LIwasEnglish aswell as a controlgroupofnative Frenchspeakers.At each grade level, therewere threeexperimentalgroups,eachhavingvaryingdegrees of immersion inFrench.The experimentalsubjects tookthe testsin both Frenchand English10see whethertheytreated the markedstructuresdifferently from theunmarked structures In theL1.Whilepredictedthatif markedstructuresthat have been transferred are easily eradicatedfromthe interlanguage, then the groupwith the most exposure to French shouldshow theleasttendency to transfermarkedstructures. Moreover,the groupwiththe leastexposureto French shouldshowthegreatest tendencytotransfer markedstructures. Two grammaticalityjudgementtaskswere used: alistening comprehension task anda grammatlcalityjudgement taskwhichhad amultiple-choice format.Theresultsfromboth tasks showed that all FSLgroupsaccepteddouble- objectsentencessignificantly more thanthe controlgroupdid.White sees this as a confirmationoftheresultsfromtheadultsandconcludesthatthis transferis nOI just found at thelowest levels.

Results alsoshowedthat givensentence pairs thatInvolveidenticalvocabularyI

whereonesentenceofthepairisthe unmarkedversionandlite other the marked,the

(61)

50

predominant responseis tojudgethesentencesto be thesame;the next mostfrequent response was10prefer the marked version; andtheleastfavouredresponsewas the unmarked version. Whileconcludesthat although the double object construction is formallymarked,itisnot perceivedas psycbclinguisticatlymar kedin theLl.

The rearesome problemswith this study. Inthe first experimentwiththe adults,White doesnotuse a controlgroup tocompare the experimental subjects' resultswith, Aswell, she dividesthe subjectsinto two groups: nativeEnglish speakers andsubjectsof otherlanguagebackgrounds which do nothave the double- object constructionaspart of theirgrammar. Iftransferwerean issue.thenwe would expect the twoexperimentalgroups 10judge thesentencesdifferentlybased on their language backgrounds. Instead,they bothhaveproblems withtheung rammatical French sentences.

As well, inthe second experiment Involvingchild FSLlearners, Whitebases her conclusionsabout thedouble objectconstructionon 3 sentences from each ofthe 2tasks.Infact, Whitestates(p. 272)thatthese sentenceswer eused as dtsrractors for other test

nerr,s.

Herresults cannotbeconclusivebecausethe data are too limih..'(1.

Hawkins(1987)lookedatthe secondlanguage acquisitionoftheEnglishdative alternationbyLlFrenchspeakersranginginagefrom19 to24 years. He used 2 elicitationtasks: agrammaticalityjudgement task,in which subjcctsjudged36 dative verbs in English sentences, andasentence constructiontask,whichtestedthesubjects' knowledg eof42 dative verbs, someof whichappearedin thegrammaticality

(62)

SI

judgement task. Results form both Il'S(S suggested that there is a developmental sequence in the acquisitionof the dative alternation:subjectsfirst acquirethe [NP PPJcomplementand then the[NP NPJcomplement.

3.2.2 Acquisition of Verb-Particle StructuresandStructuresInvolvingtheInteraction of the Dative Alternation and Verb-PanicleCombinations.

To my knowledge, therehave been no studies carried out on the second languageacquisitionof verb-paniclestructures, As well, aside from an unpublished pilot studycarried out by thisresearcher(Sheppard, 1989)and discussedbelow, there have been no secondlanguagestudiesof the interaction of the dativealternationand verb-particlestructures. In that study, 20 firstyear university students for whom English wasa secondlanguage were tested. They camefrom 8 languagebackgrounds and theiraverage age was 23.5 years. As well, a control group of 25 first year University native English speakerswas included in the experiment. The language backgroundsof the ESL groupwere diverse: French, Cantonese,Mandarin, Arabic, Bengali, Amharic, Malaysian,and Ukrainian. Testing consistedof an acceptability judgement task whichtestedlearners'judgements of grammaticaland ungrammatical English sentences. Each subjectwas given a type-writtencopy of the lestsentences along with instructionsand a three-pointjudgement scale for each sentence.In all, there were 44 sentencestestedincludingotsrractors. Six verb-particlestructures were

Références

Documents relatifs

The way we use the Green bundles in our article is different: the two Green bundles will bound the “derivative” below and above (this derivative is in fact the accumulation points

Derive from there that the Hodge decomposition is independent of the choice of the Kähler metric

Aware of the grave consequences of Substance abuse, the United Nations system including the World Health Organization have been deeply involved in many aspects of prevention,

Also use the formulas for det and Trace from the

[r]

Determine all the primitive roots of the prime

[r]

[r]