Formy project, I used the brain imaging technique...
Transcript of Formy project, I used the brain imaging technique...
For myproject,Iusedthebrainimagingtechniqueelectroencephalography(EEG)tostudyhowpeoplelearningasecondlanguageprocesssentenceswithpronounssuchas‘he’andreflexivessuchas‘himself’.ThisexperimentstartedaspartofaclassprojectinaseminaronNeuro-linguistics,taughtbymymentorsDr.RobFiorentinoandDr.AlisonGabriele.Lastyear,aftertheclassended,ItookovertheprojectandcompletedittofulfilltherequirementstograduatewithDepartmentalHonorsinLinguistics.
1
Exploringhow peoplelearnasecondlanguagecangiveinsightintohowthebrainworks,beyondjustlinguisticknowledge.Mystudyexaminesifsecondlanguagelearnersareabletoprocesssentencesatthesamelevel,usingthesamebrainmechanisms,asnativespeakers.AddressingthisquestioncanbringinsightstothedomainofNeuroscience,sheddinglightonwhetherornotourbrainsretainplasticity.Ifitistruethatsecondlanguagelearnerscanacquirethesamerulesforsentenceprocessingasnativespeakers,thenthatisevidenceforthebrain’sabilitytoformnewneurologicalpathways,evenafteritisfullydeveloped.
Toexplorethisquestion,Ilookedatonespecificaspectofsyntacticprocessing:Pronominaldependency- whichisuseofpronounsorreflexivestorefertosomeoneelsewithinasentence.Inordertoprocesspronominaldependencies,nativespeakersuseasetofgrammaticalrulescalledBindingTheory.
2
TheBindingtheoryisasetofsyntacticprincipleswhichgovernwhichelementsitemssuchasapronounorreflexivemayreferto.For myexperiment,IfocusedonPrincipleAandPrincipleBoftheBindingTheory. PrincipleAofBinding Theorystatesthatareflexivepronoun,suchas“himself”or“herself”,mustbe‘bound’withinitsminimalsentence.ThismeansthatinsentencessuchasexampleAabove:”JohnthinksthatHenrylikeshimself”,theword”himself”canonlygrammaticallyreferto“Henry”.InsentenceA“Henry”istheaccessibleantecedent,theonewhichBindingTheorystatesthat“himself”mustbereferredwith;and“John”istheinaccessibleantecedent. Conversely,PrincipleBofBindingtheorystatesthatanon-reflexivepronoun,suchas”him”or“her”,mustbeboundoutsideitsminimalsentence.So,inthecaseofexamplesentenceB,“John”istheaccessibleantecedent,and“Henry”isinaccessible.
Itiswidelyacceptedthatadultnativespeakersofalanguagehavethisknowledge,andadheretoBindingtheoryinsentenceswithpronominaldependenciessuchastheexamplesabove.StudieshaveshownthatwhentherulesofBindingtheoryarenotfollowed,forexample,iftheaccessibleantecedentdidnotmatchingenderornumberwiththeco-referencedpronoun,nativespeakersshowsensitivitytotheviolation.
3
Inmystudy, Iusedthebrainimagingtechniqueelectroencephalography(or EEG).EEGdatahasveryhightemporalresolution,sowecantrackbraindatainreal-time,whichallowsustoseedifferencesbetweendifferentparticipantgroupseveniftheymightperformsimilarlyonabehavioraltask.
EventRelatedPotentials,orERPs,areburstsofbrainactivitywhichhappenduringaspecificpointintimeandrepresentanaspectofprocessing.ERPsallowustoseehowandwhenasentencewasprocessed,whereabehavioralmethod,suchasacomprehensionquestion,wouldonlybeabletotelluswhetherornotthesentencewasunderstood.
ThespecificERPthatisseenwhenNativespeakersprocessaviolationtoBindingTheoryistheP600;aposteriorpositivitythatoccursbetween500-900msafteronsetoftheviolation.AnotherimportantERPthatisrelevanttomystudyistheLeftAnteriorNegativity,orLAN,whichoccursbetween300and500ms.
4
Oneoftheonlyrecent studiestoexamineonlineprocessBindingTheoryinSecondLanguagelearnersisLianget.al.Thisstudylooksathowsecondlanguagelearnersprocessaviolationwherethegenderofareflexivepronoundoesnotmatchwiththeantecedentthatisgrammatical.Thisiscalledagendermis-match.
Forexample,insentenceAthereflexive“himself”doesnotmatchingenderwiththeantecedent“MissZhao”.SincethesetwowordsshouldbereferencingthesamepersonaccordingtoBindingTheory,wewouldexpectaresponsetothisviolationattheonsetofthepronoun.Conversely,sentenceBiscorrectinaccordancewithBindingtheory,sowedoexpectaviolation.
TheparticipantsusedinthisstudywereagroupofChinesespeakerswhoweretestedontheprocessingoftheseviolationsinChinese,theirnativelanguageandaseparategroupofnativeChinesespeakerswhoweretestedontheprocessingoftheseviolationsinEnglish,theirsecondlanguage.
5
Lianget.al.foundthatconditionswithagendermis-matchviolation,showedaP600forboththegroupreadingtheirnativelanguageandthegroupreadingtheirsecondlanguage.Thisleadthemtoconcludethatsecondlanguagelearnersdoaccessthesamesyntacticrulestoprocesslanguageasnativespeakers.However,oneinterestingthingtheyfoundwasaLANresponseearlierinprocessingthatwaslargerandmoreconsistentfortheparticipantsreadingintheirnativelanguage.
6
Mystudyexpands ontheconceptstestedinLianget.al.,lookingnotonlyathowgender matching influencespronounreference,butalsolookingspecificallyathowtherulesofBindingTheorygovernwhichnounphraseapronounorreflexivecanreferto.Lianget.al.onlytestedreflexives,andeachsentencetypeonlyhadonlyonepotentialnounphraseprecedingthereflexive.Thisisaproblembecauseitleavesthepossibilitythatsecondlanguagelearnersjustattempttoreferthereflexivetotheonlyantecedentinthesentence(MissZhao)asopposedtolinkingthereflexivewiththenounphrasebecauseiswithinitsminimalsentence,asinaccordancewithBindingTheory.Toaddressthis,inmystudyItestedviolationsforsentenceswithbothreflexiveandnon-reflexivepronounsandallsentencesincludedtwopossibleantecedents.
MystudyaddressesatheoryinthesecondlanguageacquisitionliteraturecalledtheShallowStructurehypothesis.Relatedspecificallytopronominaldependencies,thishypothesisarguesthatsecondlanguagelearnerswilltrytolinkthepronounorreflexivetothenounphraseinsubjectposition,ratherthanthenounphrasethatiscorrectaccordingtoBindingtheory.Theyarguethatthenounphraseinsubjectpositionisprominentinthediscourseandlearnersaremorelikelytorelyondiscourseinformationthansyntacticrules.
7
TheissueIamlooking atinthisstudyiswhetherornotSecondLanguageprocesslanguagethesameasNativespeakers.Iwanttoknowwhetherornotanadultcanbecometrulyfluent,toanative-likelevel,inasecondlanguage.
8
Inmystudy, IwantedtoaccountforthepossibilitythatSecondlanguageLearnerscouldbeusingsomethingotherthantherulesofBindingTheorytoproducesimilarresultstoNativeparticipants.Inordertoaccomplishthis,Ichosestimulithatincludedtwonounphrases,aswellastestingconditionswithbothreflexiveandnon-reflexivepronouns.Myconditionsareoutlinedabove.
Condition(a)hasnoviolation;itisagrammaticalsentencewherethepronoun“her”matchesingendertoitsreferent“Daisy.”Condition(b)hasaviolationoftheBindingTheory,because“her”doesnotmatchingenderwith“Phil,”whichisin theplaceofthegrammaticallycorrectnounphrase.Thesamepatternisshownwithreflexives.Condition(c)showsanungrammaticalsentencewherethereis agendermis-matchbetween “herself”and“Phil,”. AndCondition(d)isgrammaticalaccordingtoBindingtheory.
Eachnameusedinthesentenceswaschosentobeunambiguouslyassociatedwithonegender,sothatthesecondlanguagelearnerswouldknowiftherewasagendermis-matcherror.Theorderofthegenderofnamesineachitemwasrandomized.Theverbsusedinthestimuliwereselectedtobecompatibleinsentenceswithpronounsaswellasreflexivesentences.
9
Asforthefillersentences,therewerefourconditionschosentotargetsubjectverbagreement.Therewerebothgrammaticalandungrammaticalversionsofsentenceswithpluralandsingularsubjects.Thesewerechosenbecausewewereabletocreateasimilarsyntacticviolationandsimilarlylengthofsentence,butwithouthavinganypronounreference.Thesubjectsinthefillerswerechosentobeanagent-subjectofanounphrase,ratherthananameasinthetargetsentences.
Allofthesentenceswerecreatedtosoundnaturalandcheckedsothatnoneofthesubjectswererepeated.Wealsomadesurethatnoneofthefillersusedanypronounsanywhereinthesentence.
10
Myparticipantswere9, righthandedadultsChinesespeakersbetween18-50yearswhoaresecondlanguagelearnersofEnglish.EachparticipanttookantestcalledtheLexTale,usedtoassesEnglishvocabulary.Thescoresonthesetestsvaried,whichmeansfluencylevelofparticipantswerevariedaswell,withameanscoreof69.25percent.
AsImentionedearlier,myexperimentwasdoneonanEEGusingERPanalysis.Thismethodwaschosenbecauseitcanmeasuretimingverypreciselyandallowsustoseeexactlyhowandwhenasentenceswasprocessed.
Duringthemainexperiment,participantswereconnectedtotheEEGcapandsatinfrontofthecomputer.TheirbrainactivitywasrecordedbytheEEGsystemastheyreadsentences.
11
Themaintaskwasadministered onaprogramcalledParadigm.Theintroductorysectionincludedawelcomeslide,instructions forthetask,and6practiceitems.Duringthemaintaskeachparticipantssaw160targetitemsand80fillers,foratotalof240itemsrandomlyvariedinaLatin-squaredesign.Sentenceswerepresentedonewordatatimeinthecenterofthescreen,usingRapidSerialVisualPresentation.Aftereachsentencestheparticipantwasaskedtotoanswerajudgmentquestionbypressingabuttononacontroller.Participantswereoffered4breaksduringthemainexperiment,whichtookabout1hourtocomplete.
TimesforstimulipresentationlengthandthelengthoftheexperimentwerechosenbasedonpreviousL2research.
12
Aftereach iteminthemainERPtask,participantswereaskedwhetherornotthesentencewasacceptable,andaskedtopress”yes”or“no”onaresponsepadbeforemovingontothenextitem.Thismeasurewastakeninordertobesureparticipantswerestayingfocusedandunderstandingthesentences.
Asfortheothertasksadministered,beforetheexperiment,eachparticipantwasgivenaconsentformtolookoverandsignbeforebeginning.Theythentookashortlanguagebackgroundquestionnairewhichincludedahandednessquestionnairetoreferenceandbesuretheywerecompatiblewiththestudy.Lastly,afiveminutelexicaldecisiontaskcalledLexTalewasadministered.Thiswasusedtoassesparticipants’knowledgeofEnglishvocabularyandwasusedasaproficiencymeasure.
Afterthemainexperiment,participantsfilledouta20-questionreferencetaskaskingthemtoexplicitlychoosewhichnounphrasetheythoughtthepronounorreflexivewasreferringtointhestimulisentences.Thiswasgivenafterthemainexperiment,soasnottoprimetheexpectedresponseorinformtheparticipantofexactlywhichitemswewerelookingat.
13
Ifthesecondlanguagelearnersareabletoaccesstherulesofbindingtheory,andprocesssentencesinthesamewayasnativespeakers,thenwewouldexpectalargerP600forconditionswhichhadaviolationinBindingTheory,comparedtotheconditionsthatdidnothaveaviolation.ThismeansthatconditionB,theungrammaticalpronoun,willhavealargerpositivitythanconditionA,thegrammaticalpronouncondition.Thesameistrueforreflexives:conditionC,ungrammatical,willhavealargerpositivitythanconditionD,grammatical.Thisissimilartowhatwewouldexpecttoseefromnativespeakers.DuetoconcernsaboutCOVID-19,IwasnotabletotestmyownstimulionagroupofnativeEnglishspeakers,asacomparisongroup.BecauseofthisIwilluseexpectedresultsderivedfromstudiesusingsimilarstimulitounderstandhownativespeakersprocesstheseerrors.OnestudythatlookedatsentenceswithBindingTheoryviolationswasHarriset.al.(2000)TheirstudysupportedtheideathatnativespeakerswouldyieldaprominentP600forsentenceswithBindingTheoryerrors,regardlessofdiscourseprominence.
IfClahsenandFelseret.al.’sshallowstructurehypothesisiscorrect,andsecondlanguagelearnersonlyattempttorefertothenouninsubjectposition,thenwewouldexpectbenative-likeforthepronounconditions,wheretheaccessible
14
antecedentisinsubjectposition,yieldingagreaterP600forconditionB,wherethegendermis-matchedwiththesubjectantecedent,thanforconditionA.Thisissimilartohowweexpectnativestopattern.WealsoexpectagreaterP600fromconditionDthanC,becauseinDthegenderofthereflexivemis-matcheswiththesubjectantecedent,regardlessofwhetherornotitwascorrectaccordingtoBindingtheory.Thisisdifferentfromwhatwewouldexpectfromnativespeakers.
Additionally,somestudiessuchasLianget.al.thatwasoutlinedearlier,foundaLeftAnteriorNegativity,orLANresponseforthesetypesofviolations.
14
Theraw EEGdatacollectedfromeachparticipantwasprocessedtominimizeartifacts,thenusedtogenerateaveragesforthetwotimewindowsfortheERPsofinterest.Welookedatthethewindow300-500msafteronsetoftheviolation,thisistheusualtimewindowfortheLANresponse,whichisanegativitylocalizedtowardstheanteriorregionofthebrain.Theothertimewindowwelookedatwas500-900msafteronset,whichisthetimewewouldexpectaP600,apositiveresponsefoundonposteriorelectrodes.
Usingthesegrandaverages,Icreatedfigurescomparingtheungrammaticalconditiontothegrammaticalconditionforbothsentenceswithreflexivesandnon-reflexivepronouns.
15
Shown abovearethefigurescomparingthegrandaverageresponsesforallparticipantsfortheungrammaticalpronouncondition,wheretherewasaviolationtoBindingtheory,tothegrammaticalpronouncondition.Theungrammaticalconditionisshowninred.
Around300-500msonthegraphtheungrammaticalconditionwasmorenegativethanthegrammaticalcondition,especiallyinelectrodeFZaround500ms.ItispossiblethatthisisanemergingLANeffect,sinceitisfoundinthetimewindowandespeciallynoticeableinanteriorelectrodes.ThisissimilartowhatwasfoundbyLianget.al.
Around500-900ms,whichisthetimewindowfortheP600effect,theungrammaticalandgrammaticalconditionswerenotverydifferentfromoneanother;neitherconditionshowedagreaterP600effect.
16
These graphsshowthegrandaveragefromallparticipantsoftheresponsescomparingthereflexiveconditions.Theungrammaticalconditionisshowninred.
There wasalargernegativityfortheungrammaticalreflexivesthanforthegrammaticalreflexivesbetween300and500ms.ThiscouldbeanemergingLANeffect,similartowhatwesawinthepronounconditions.
Around500-900mstherewasagreateroverallpositivity,especiallyfortheposteriorelectrodesrepresentedbyPZ.ThiscouldbeanemergingP600effect,whichisusuallyfoundatthistimeinprocessingaroundtheposteriorelectrodes.
17
Thepreviousdatapatterns were similartowhatwewouldexpecttoseefromnativespeakersbasedonpreviousstudiessuchasHarriset.al.andLianget.al.Thesefindingsshowtrendssuggestingthatparticipantsaresensitivetobinding theoryviolations,meaningthattheymaybeabletoaccessthesamebrainmechanismsasnativespeakers.Theseresultslendevidencetothepossibilitythatourbrainsareabletoretainplasticityintoadulthood.
Thesearejustthepreliminaryfindingsinwhatcouldbeamuchlargerexperiment.Fluencylevelofmyparticipantsvariedalot,whichcouldexplainwhytherewasnotadifferenceintheP600effectfoundforthepronounconditions.Inthefuture,itwouldbebeneficialtotestonmorehighproficiency learners,whowouldbemoreattunetothesyntacticerrorsinthestimuli.ItwouldalsohelptohavealargerparticipantgroupoverallandtestacontrolgroupofEnglishspeakersusingmyownstimuli.
18
Iwouldliketothank mymentorsDr.RobFiorentinoandDr.AlisonGabrieleforhelpingmedevelopandcarryoutthisproject,especiallyforalltheydidduringtheCOVID-19quarantinetohelpmefinishthisprojectevenwithoutaccessthelabtoprocessdata.IwouldalsoliketothankXiaoYangforallherhelpteachingmehowtorunparticipants,andeveryoneelseintheNeurolinguisticslabwhohelpedmewithEEGsessionsandrecruitingparticipants.IwouldalsoliketothankthestudentsinmySpring2019Linguisticscourse,“CognitiveNeuroscienceofSecondLanguageAcquisition”,whoworkedonthestimuliandpilotstagesofdatacollectionwhenthiswasaclassproject.
19
Herearemyreferences.Thankyou!
20