Positive-and-Negative-Evidence-in-Language-Acquisi

PositiveandNegativeEvidenceinLanguageAcquisition1．IntroductionThenecessityforinputintheprocessofsecondlanguageacquisition(SLA)isawell-acceptedfact,buttheformandtypethatitneedstotakeforlearningtooccurremainsacontroversialissue.Thosesubscribingtoanativistorrationalistpositionofacquisitionsupporttheideathatpositiveevidenceisallthatisrequiredforacquisitiontooccur(Chomsky,1989).Theybelievethathumanknowledgedevelopsfromstructures,processesandideasthatareinthemindatthebirth.Ontheotherhand,thoseworkingwithintheinteractionistparadigmseepositiveevidenceasinsufficientandproposearoleforbothpositiveandnegativeevidence(Labov,1969;Gass,2003).2.PositiveEvidencePositiveevidenceisevidencethataparticularutteranceisgrammaticalinthelanguagethatthelanguagethatthechildislearning.Itconsistsofdescriptiveinformationaboutaformoranutterance.Itconsistsofactuallyoccurringsequences,i.e.sentencesofthelanguage.Variousoptionsexistforpositiveevidenceincludingplentifulexemplarsofthetargetfeaturewithoutanydevicetodrawattentiontoit.Forexample,Trahey(1996)developedmaterialsconsistingofstories,games,andexerciseswiththeaimofsimplyexposinglearnerstothesubject.Inthiscase,acquisitionoccursasaresultoffrequentexposuretoatargetfeature.Itinvolvessomesortofattempttohighlightinstancesofthetargetfeature,thusdrawinglearners'attentiontoit.Positiveevidencecanfunctionentirelybyitself.Learnerscansimplybeaskedtolistentoorreadtextsthathavebeenprovided.Itcanalsobeaccompaniedbysomekindofmeaning-focusedactivitythatincidentallyassistslearnerstofocustheirattentiononthetargetfeature.Forexample,comprehensionquestionsthatcanonlybeansweredcorrectlyifthelearnersprocessthetargetfeature.Therearetasksthataredesignedtoelicitproductionofaspecifictargetfeatureinthecontextofperformingacommunicativetask,andtasksthatareintendedtoresultinlearners'employingsomefeaturethathasbeenspecificallytargeted(White,1987).Severalpatternsinlanguagehavebeenclaimedtobeunlearnablefrompositiveevidencealone.Oneexampleisthehierarchicalnatureoflanguages.Foranygivensetofsentencesgeneratedbyahierarchicalgrammarcapableofinfiniterecursionthereareanindefinitenumberofgrammarsthatcouldhaveproducedthesamedata.Thiswouldmakelearninganysuchlanguageimpossible.Indeed,aproofbyE.MarkGoldshowedthatanyformallanguagethathashierarchicalstructurecapableofinfiniterecursionisunlearnablefrompositiveevidencealone,[inthesensethatitisimpossibletoformulateaprocedurethatwilldiscoverwithcertaintythecorrectgrammargivenanyarbitrarysequenceofpositivedatainwhicheachutteranceoccursatleastonce.However,thisdoesnotprecludearrivingatthecorrectgrammarusingtypicalinputsequencesratherthanparticularlymalicioussequencesorarriveatanalmostperfectapproximationtothecorrectgrammar.Anotherexampleoflanguagepatternclaimedtobeunlearnablefrompositiveevidencealoneissubject-auxiliaryinversioninquestions,i.e.:Youarehappy.Areyouhappy?Therearetwohypothesesthelanguagelearnermightpostulateabouthowtoformquestions:(1)Thefirstauxiliaryverbinthesentence(here:'are')movestothebeginningofthesentence,or(2)the'main'auxiliaryverbinthesentencemovestothefront.Inthesentenceabove,bothrulesyieldthesameresultsincethereisonlyoneauxiliaryverb.But,thedifferenceisapparentinthiscase:Anyonewhoisinterestedcanseemelater.1.Isanyonewhointerestedcanseemelater?2.Cananyonewhoisinterestedseemelater?Ofcourse,theresultofrule(1)isungrammaticalwhiletheresultofrule(2)isgrammatical.So,rule(2)is(approximately)whatweactuallyhaveinEnglish,notrule(1).Theclaim,then,firstisthatchildrendon'tseesentencesascomplicatedasthisoneenoughtowitnessacasewherethetwohypothesesyielddifferentresults,andsecondthatjustbasedonthepositiveevidenceofthesimplesentences,childrencouldnotpossiblydecidebetween(1)and(2).Moreover,evensentencessuchas(1)and(2)arecompatiblewithanumberofincorrectrules(suchasfrontanyauxiliary).Thus,ifrule(2)wasnotinnatelyknowntoinfants,wewouldexpecthalfoftheadultpopulationtouse(1)andhalftouse(2).Sincethatdoesn'toccur,rule(2)mustbeinnatelyknown.3.NegativeEvidenceNegativeevidencereferstoinformationaboutwhichstringsofwordsarenotgrammaticalsentencesinthelanguage,suchascorrectionsorotherformsoffeedbackfromaparentthattellthechildthatoneofhisorherutterancesisungrammatical.Negativeevidenceconsistsofinformationabouttheimpossibilityandungrammaticalityofaformoranutterance.Inotherwords,negativeevidencesuchasexplanations,explicitgrammarteachings,andcorrectionsofwrongsequencesorungrammaticalsentences,showwhatmaynotbedone.Itneedsadditionalevidencefromcorrectionsofimpossiblesequences,readingtherule-books,comprehendingabstractexplanations,andsoon.Therearetimeswhenalearnersuppliesalinguisticallyincorrectresponseinreplytoateacher'sinitiation;theteachertendstoprovidedirect,explicit,overtnegativeevidence.However,Chomsky(1981)holdstheideathatdirectnegativeevidenceisnotnecessaryforlanguageacquisition,butindirectnegativeevidencemayberelevant.It'sveryimportantforustoknowwhetherchildrengetandneednegative,becauseintheabsenceofnegativeevidence,anychildwhohypothesizesarulethatgeneratesasupersetofthelanguagewillhavenowayofknowing