A PilotStudy:
TowardsSystematicMethodsofErrorCorrection
OsamuTakai Thispaperexaminestherelationshipbetweenthelearners'targetlanguage proficiency andtwotypesoferrors,i.e.,transfererrorsanddevelopmental errorsby usingmultipleinterlanguageanalyses:contrastiveanalysis,error analysis,andT‑unitanalysisonthebasisofthehypothesisestablishedfromthe reviewofthearticlesinthefirstpartofthestudy(Takai:1989). Thehypothe‑ sistobetestedinthisstudyisasfollows:ifthelearnerisabeginnerwhohas apoorproficiencyinthetargetlanguage,therewillbemorefirstlanguage transfertypeoferrors.Asaresult,thehypothesishasbeensupportedbythe statistict･test.
Ⅰ. lNTRODUCTtON
In recentyearsaconcerted attempthasbeen donetounderstandthe processesofinterlanguagedevelopment. Atthetheoreticallevel,thequestion isposedastowhetherornotthedevelopmentsaresimilartothoseinchildfirst languageacquisitionstudies. Themajorprocessofsecondlanguageacqui‑ sitionisclaimedtorestontheequivalentprocessofchildlanguageacquisition (Dulay,Burt,andKrashen,1982).
Undoubtedlyleamer'serrorsarepredictedasthedevelopmentalprocessof thetargetlanguage,i.e.,developmentalerrors.Itshouldbemoretransparent whenlookingatstatementslikethefollowinginDulay,Burt,Krashen,1982:
"Toalargeextent,controversiesovertheroleofthefirstlanguageinsecond languageacquisitionhaveresultedfrom vagueandvaryingusesoftheterms
〔15〕
̀interference'and̀transfer'. Whenthetermsareclarifiedandwhenempirical dataisassembled,thereappearstobeaconvergenceofopinionontheroleof thefirstlanguageinsecondlanguageacquisition. Despitealonghistoryof assumptiohtothetohtrary,presentresearchresultssuggestthatthemajor impactthefirstlanguagehasonsecondlanguageacquisitionmayhavetodo witbaccent,notwithgrammarorsyntax."(p.96)
Befqrethe1960'S,thelearner'sfirstlanguagewasconsideredasamajor hindranceinhisorheracquisitionofasecondlanguage,whichisexplainedand knownas"interference"accordingtothecontrastivehypothesis(Lado,1957;
Kimizuka,1977)undertheassumptionofthestimulus‑responselearningtheory advocatedbybehavioristsofpsychology(seethenoteinTakai,1989). Ac‑ t､Cordingly,learner'serrorswere̲analyzedasaresultofpreventionbyhisorher
firstlanguagehabitsbycontrastiveanalysts.
AfterChomsky'srevolutioninthefield oflinguistics,Secondlanguage learnlngandacquisitionhasalsobeenviewedas"active","creative"processby usinghypothesistestingwithlearner'sinnatefunctioninhisorherbrain,known as"LanguageAcquisitionDevices". Sincethen,secondlanguagelearningand acquisitionhasbeenconsideredintheframeworkofcognitivelearning,and learner'serrorshavealsobeenanalyzedintherealm ofthese;condlanguagethe learnerislearning,whichisknownaserroranalysis(Cor°er,1975;Richards, 1971).
Asaresu
l
t,theconceptof"interlanguage"grammar,theuniquesystem of grammardifferingfrom hisorhernativelanguage,Orthetargetlanguage system asSelinker(1972)advocated,hasprevailed amongsecondlanguage acquisitionresearchers. Here,learner'serrorsareassumedtobeevidenceof his′̲orherlearnlngprocess.Inreferencetolearner'serrors,Krashen(inKrashenandTerrell,1983) admonishedteachersnottousedirectcorrectionoflearner'serrors,foritwill
APilotStudy:TowardsSystematicMethodsofErrorCorTeCtion 17 enhancethelevelofhisorheranxietyandresultininhibitionforthelearner's secondlanguageacquisition. Accordingtohistheory,the"Monitor"model, errorsareconsideredasdevelopmentalprocessandneednotbecorrectedasthe learnercanmakeself‑correctionswithhisorher"monitor"(Krashen and Te汀e
l
l,1983).Theconceptofdefiningof"developmentalerrors"resultingfrom "error‑ analysis"issometimesconceivedsogenerallyastobeabletoaccountforas manyfactorsaspossible(Takai,1986;1989). Veryrecently,amovementofre‑ evaluatingcontrastiveanalysishasemergedfrom pastoblivion. Forexample, Nickel(1989)says:"Onecannothelpavoidingtheimpressionthat̀develop‑
mental'errorsareo允engivenprecedenceinsomeofthemodem EA(i.e.,error‑ analysis)research,asalsohappensinLanguageTuJo. Buttherearealsoother views. Thisagainisprobablyduetothetrendtofinduniversals. Tobegin with,thefactthate汀OrSOfleamersofanativelanguagecoincidewitherrors madebyFLlearners(i.eリforeignlanguagelearners)doesnotprovenecessarily thattheyareofthedevelopmentaltype"(p.299).
ⅠⅠ. STATEMENTOFTHEPROBLEM
Reviewlngthemostrecentmovementsandtrendsinerroranalysis,iwould liketostatetheproblem dealtwithinthispaperasfollows:
Fromthepedagogicalpointofview,we,teachers,alwaysfacetheappropri‑ atedecisiononmakingerrorcorrections. Weknowthereareatleasttwotypes oferrorsinleamer'serrorsfrom theresultsofcontrastiveanderroranalyses:
transfertypeerrors,anddevelopmentaltypeerrors. Andyet,wedon'tknow systematicguidelinesofobjectivecriteriatodraw thelinebetweenthetwo.
SchachterandCelce･Murcia(1980)statedsix potentialweaknessesinerror‑ analysisresearch;andinoneconcerningadescriptiorlOfsystematicerrors
,
theypointedoutthatalargenumberoflearner'serrorsareambiguousastowhethertheyaretransferordevelopmentalerrors. AsKrashen(inDulay,Burt, andKrashen,1982)alsqstated:"Unfortunately,virtuallynoworkhasbeen donetoprobetheconditionsunderwhichinterlingualerrorsaremade"(p.108)
,
wehaven'thadanyscientifictestresultsonthismatter.Inthispaper,Iwouldliketoinvestigateasystematicmethodoferror correctionswithinthetheoreticalframeworkofsecondlanguagelearningand acquisition,usingthecombinationofmultipleanalysesofinterlanguage:i.e.
,
contrastiveanalysis,erroranalysis,andT‑unitanalysisapplyingtothedata obtainedinmyclassofferingasageneralEnglishcourseatthejuniorcollegein OtaruunivesityofCom erce.TheT‑unitanalysishasbeenwidelyusedtomeasureglobal(i.e.,overall growth)grammaticalproficiencyofspeechandwritingsamplesproducedby languagelearners. Gaies(1980)evaluatedT‑unitanalysisinsecondlanguage researchasaglobalmeasurement(i.e"anoverallgrowthindex)oflearner's proficiencyandstatedthatmanyresearchershadexpressedrelativesatisfaction withT･unitlengthandrelatedmeasuresasindicesofsecondlanguagedevelop‑
ment.
ⅠⅠⅠ. HYPOTHESESTO BETESTED
Thefollowlnghypothesishasbeentestedinthisresearch:ifthelearneris abeginnerwhohasapoorproficiencyinthetargetlanguage,therewillbemore firstlanguagetransfertypeoferrors,i.e.,interlingualerrors. Sinceheorshe doesnothaveknowledgeoftheappropriatesecondlノanguagerules,heorshehas tousethefirstlanguagerulestoproducethesecondlanguage.
Ontheotherhand,ifthelearnerisproficientenoughinthesecondlanguage, heorshewillperform developmentalratherthaninterlingualerrors.
APilotStudy:TowardsSystematicMethodsofErrorCorrection 19
ⅠⅤ. METHODSAND PROCEDURES DataGatheringProcedure
Thedataarewrittencompositionsgatheredfrom secondyearJapanese studentsatmyEnglishclassin1989. Therearethirty‑sixwrittensampleson thetopicof"Myfutureplan." ThestudentswereaskedtowriteinEnglish abouttheirowndreamsinthefutureforforty‑fiveminutesduringclass. They wereadvisedtoconcentrateuponwritingratherthancheckingeverysingle wordinadictionarysincethetimewaslimitedtoforty‑fiveminutes.
A writtensampleonthesametopicwaselicitedfrom lanadultnative speakerofEnglishforreferenceandcomparisonwiththesamplestakenfrom theJapanesestudentsinmyclass.
IdentifyingErrors
Sincetheaim ofthisstudyistotestthehypothesisbycomparingtwomajor errortypes,fiveprominentcategoriesortypesoferrorwerechosenprimarily baseduponthepreviousresultsofcontrastiveanalysisforinterlanguage(or transfer)errortypeandtheerroranalysisfordevelopmentalerrortype. The followingisageneralproceduretoidentifyerrorsingeneral:
Thefirststepintheanalyticprocessistoidentifypossibleerrorsinthe samples. Anerrorisdefinedasanoticeabledeviationfromthegrammarofan adultnativespeakerofEnglishinthisresearch. Themeasurementofthe deviationisamatterofdegree. Someerrorsareobvious,overterrorsas Cor°er(1971)stated;andsomearenotveryobvious.atthefirstsigh
t
. Inthis study,thenorm ofthegrammarofanadultnativespeakerwassetonthebasis ofacomposition written by anativespeakerofEnglish and a reference grammarbook,A Comp71ehensiveGyamnwroftheEnglishLanguage(Quirk,et a1.
,1985).Inordertoidentifyerrors,ⅠfollowedCorder'S(1971)procedureofidentify‑
ingerrorsinsecond‑languageproduction. Accordingtohismodel,anysentence producedbyalearnercanbeanalyzedforitsdeviationfrom anorm 佃hetherit isovertorcovert.
Corder'smodelmakesadistinctionbetweenovertandcoverterrors. The overtonesareunquestionablyungrammatical;coverterrorsaregrammatically wellfom edbutnotinterpretablewithinthecontext. Ifthereisnoovertor coverterror,thesentenceisnot"idiosyncratic"(Cor°er,1971). Though a sentenceiscovertlyorovertlyidiosyncratic,themodelindicatesthatinboth cases,ifaplausibleinterpretationcanbemadeontlleSentence,WeCanCOmpare thewellfom edreconstructedonefrom theidiosyncraticone;inthiswaywe canidentifytherulesofthetargetlanguage. Themodelalsoclarifiesthat whenthenativelanguageofthelearnerisknown(inthisresearch,Japanese),we canreconstructthesentenceinthetargetlanguagetoidentifythedifferences betweenthenativeandthetargetlanguagerules. Andyettheremustbesome residueinwhichwecannotpredicttheerror. InthiscaseIcancontactthe subjectofthecompositiohandaskwhatheorshemeant.
AnalysisofErrors
Errorscan generally bedescribed in categoriesofaddition,omission, substitution,and ordering,asin thewayssurface structures are altered.
Researchershavenoticedthaterrorsonsurfacestructuresofalanguageare basedonsomesystematicwayswiththeirownlogic.
Forexample,theomissionofarticlesbyaJapanesestudentoflearning English,forexample,canbeexplainedin twodifferentways. From thepoint ofcontrastiveanalysi9,itisduetothelackofJapaneseformscorrespondingto Englisharticles(Kimiztlka,1977).Itcan,however,berecognizedamongerrors madebypeoplefrom otherlanguagebackgroundsandalsobychildrenlearning
