The Processing of Ambiguous Sentences by First and

TheProcessingofAmbiguousSentencesbyFirstandSecondLanguageLearnersofEnglishClaudiaFelsera,LeahRobertsa,RebeccaGrossb&TheodoreMarinisaaDepartmentofLanguageandLinguisticsUniversityofEssexbDepartmentofLinguisticsUniversityofDüsseldorfAbstractThisstudycomparesthewayEnglish-speakingchildrenandadultsecondlanguagelearnersofEnglishresolverelativeclauseattachmentambiguitiesinsentencessuchasThedeanlikedthesecretaryoftheprofessorwhowasreadingaletter.TwogroupsofadvancedL2learnersofEnglishwithGreekorGermanastheirL1participatedinasetofoff-lineandon-linetasks.Whiletheparticipants'disambiguationpreferenceswereinfluencedbylexical-semanticpropertiesoftheprepositionlinkingthetwopotentialantecedentNPs(ofvs.with),therewasnoevidencethattheywereapplyinganystructure-basedambiguityresolutionstrategiesofthetypethathavebeenclaimedtoinfluencesentenceprocessinginmonolingualadults.Thesefindingsdiffermarkedlyfromthoseobtainedfrom6to7year-oldmonolingualEnglishchildreninaparallelauditorystudy(Felser,Marinis,&Clahsen,submitted)inthatthechildren'sattachmentpreferenceswerenotaffectedbythetypeofprepositionatall.Wearguethatwhereaschildrenprimarilyrelyonstructure-basedparsingprinciplesduringprocessing,adultL2learnersareguidedmainlybynon-structuralinformation.EssexResearchReportsinLinguistics40(2002)21.IntroductionOurunderstandingofhowmaturereadersorlistenersprocesstheirnativelanguageinrealtimehasincreasedconsiderablyoverthelastcoupleofdecades.Resultsfromsentenceprocessingstudiesusingarangeofdifferentpsycholinguisticmethodsandtechniqueshaveshownthattheadultparseriscapableofaccessingandrapidlyintegratingvarioustypesofstructuralandnon-structuralinformationduringcomprehension(seeGibson&Pearlmutter,1998,forreview).Comparativelylittleisknown,bycontrast,aboutthewaylanguagelearnersprocessinputfromthetargetlanguage.Instead,bothfirstandsecondlanguageacquisitionresearchtodatehasfocussedprimarilyonthedevelopmentoflinguisticcompetence(compare,forinstance,thearticlesinRitchieandBhatia,1996,1999).Thatthequestionofhowlanguagelearnersprocessthetargetlanguageshouldhavereceivedrelativelylittleattentioninthepastissurprising,though,giventhatlanguageprocessingandgrammaticaldevelopmentareinextricablylinked:Whereasontheonehand,alanguagelearner'sabilitytoprocessaninputstringisacrucialprerequisiteforgrammarbuilding,successfulparsing,inturn,presupposessufficientgrammaticalknowledgetoenabletheparsertosegmentaninputstringintogrammaticallymeaningfulchunks,toassignappropriatecategorylabelstoeachsegment,andtodeterminehierarchicalrelationshipsandintra-sententialdependenciesamongconstituents(formoredetaileddiscussionsofthis‘acquisitionparadox’,seeFodor,1998,orValian,1990).Withthegrammarandtheparserbeingthusentwinedinlanguagedevelopment,itmaybethatatleastsomeofthedifferencesinlinguisticproficiencythathavebeenobservedbetweenchildrenandadults,andbetweensecondlanguagelearnersandnativespeakers,areattributabletodifferencesinthewaylanguagelearnersandadultnativespeakersprocessthetargetlanguage.Asregardsfirstlanguageacquisition,itisconceivable,forexample,thatchildren'slinguisticdevelopmentisconstrainedbycapacitylimitationsofthedevelopingprocessingsystem(Adams&Gathercole,2000).Foradultsecondlanguageacquisition,itispossiblethatthelackofultimateattainmentintheL2isduetoparsingproblemsratherthantoaninabilitytoacquirenative-likegrammaticalcompetenceperse,asisoftenassumed(comparealsoJuffs&Harrington,1995,Kilborn,1992,orVanPatten,1996).Morespecifically,itmaybethecasethatsecondlanguagelearnerstransferprocessingstrategiesfromtheirnativelanguagetotheL2,andthatthispreventsthemfromacquiringfullnative-likecompetenceand/orEssexResearchReportsinLinguistics40(2002)3fluencyinthesecondlanguage.Inshort,answerstothequestionofwhetherornotchildrenandsecondlanguagelearnersprocessthetargetlanguageinthesamewayasadultnativespeakersdoarelikelytohaveimportantimplicationsfortheoriesofbothfirstandsecondlanguageacquisition.OnepossibleanswertotheaboveacquisitionparadoxhasbeenprovidedbytheContinuityHypothesis(Crain&Thornton,1998,Crain&Wexler,1999,Fodor,1998),accordingtowhichhumanbeingsareinnatelyequippednotonlywithaUniversalGrammar,butalsowithauniversalsetof'leasteffort'basedlanguageprocessingroutines,orUniversalParser.Ifthecontinuityapproachiscorrect,thenwedonotexpecttofindanyqualitativedifferencesinparsingperformancebetweenlanguagelearnersandadultnativespeakers,atleastnotwithrespecttotheiruseofuniversalprocessingstrategies.Withinuniversalistmodelsofsentenceprocessing,cross-linguisticvariationinparsingisattributedeithertofactorsaffectingpost-structuralprocessing(compareFrazier&Clifton,1996),orisassumedtoberestrictedtoasmallsetofparameterisedprocessingroutineswhosesettingsaredirectlyorindirectlylinkedtocertainlanguage-specificpropertiesofthegrammar(Cuetos&Mitchell,1988,Frazier&Rayner,1988,Gibson,Pearlmutter,Canseco-Gonzales,&Hickok,1996,Mazuka,1998,Mazuka&Lust,1990).Experience-basedmodelsoflanguageacquisitionandprocessingsuchastheCompetitionModel(MacWhinney,1987,1997,2001),ontheotherhand,assumethatbothlinguisticknowledgea