Introduction
Thepresentstudy exploresaway toincorporateboth focused and unfocused written corrective feedback (WCF)into an EFL grammar and writing course.Earlier research findingshaveindicatedthatfocusedWCFismoreeffectiveforimprovingL2learners・ability towriteaccurately (Sheen,Wright,& Moldawa,2009;Bitchener,2008;Bitchener& Knoch, 2008,2009a,2009b,2010).However,providedthatlearners・attentionisclearlyfocusedonthe majortargetgrammaticalpoint,itmaybemorepracticalfortheteachertoprovidedirect correction on a few additionalgrammatical,lexical,or stylistic errors.The number of paragraphsor essaysthattheteacher can assign to studentsper semester isnormally limited.Therefore,providing feedback only on oneparticulargrammaticalruledoesnot seem tobethemostefficientoreconomicalapproachtoL2teaching.Thisisparticularlythe casewithEFLcoursesthataredesignedtohelplearnersimprovetheirgeneralgrammatical knowledgeandoverallwritingskillsatthesametime.
Participantsin thepresentprojectwereEFL studentsataJapaneseuniversity with a fairlyadvancedlevelofgrammarknowledge,andthepedagogicalpurposeofthecoursewas nottoteachnovelEnglishgrammaticalrulesbuttorefreshorconsolidatetheirknowledge ofmajorsyntacticrules,providing additionalinformation aboutthedetailed featuresof complex rules.Thecoursewasalsodesignedtoprovideparticipantswith opportunitiesto 学苑英語コミュニケーション紀要 No.894 17～42（20154）
TheEffectsofFocusedandUnfocused
Wri
ttenCorrecti
veFeedbackontheAcqui
si
ti
on
ofEFLGrammarKnowl
edgeandWri
ti
ngSki
l
l
s
Yoshi
masaOgawa
Abstract
ThepresentstudyevaluatedtheeffectsoffocusedWrittenCorrectiveFeedback(WCF)on JapaneseEFL learners・abilitytouseaccurategrammaticalformsinfreewriting,theeffects ofacombinationoffocusedWCFandunfocusedWCF,andlearners・perceptionsofWCF.The analyses of discrete-point grammar tests and essay writings showed that focused WCF contributed to theparticipants・acquisition ofcorrectgrammaticalformsto a statistically significantdegreeand thattheunfocused WCF on avariety ofspontaneouserrorsdid not underminetheeffectivenessoffocused WCF.Furthermore,thequestionnairesurvey results indicated that the participants all preferred to receive error correction, producing counterevidenceagainstTruscott・sproposalthatform-focused WCF wasuselessand should notbeprovided.
practiceusingthemajorgrammaticalformsthroughparagraphoressaywriting.
Onecommon problem with many JapaneseEFL students,including English majorsor high proficiency students,isthatthey haveexplicitly learned grammaticalrulesthrough meta-linguisticexplanationsanddiscrete-pointgrammaranalysisexercisesbutstillcannot consistentlyusethelearnedrulestoassembletarget-likephrasesorsentencesinspeakingor writingactivities.WritingallowsL2learnersmoretimetocarefullyconsiderthestructure of each sentence than speaking which requires instantaneous judgment and language productioninrealtime.However,eveninwriting,manyJapanesestudents,whenconcerned aboutthesemanticorpragmaticcontent,continuetomakegrammaticalmistakesrelatedto suchsyntacticstructuresasthearticles,relativepronouns,hypotheticalconditionals,present perfecttense,andsubject-verbagreement,thepreciseequivalentsofwhichareabsentinthe grammaticalsystem oftheirfirstlanguage.
Thefirstresearchpurposeofthisstudyistoaffirm thepositiveeffectsoffocusedWCF on theparticipants・ability to useaccurategrammaticalformsin new piecesofwriting. Truscott(1996,1999,2007)arguedthaterrorcorrectiondoesnotimproveL2writers・ability to write accurately and that it may even have harmfuleffects.However,this study endeavorstosupporttheoppositeposition,i.e.,grammarcorrectioncontributestoL2l earn-ers・acquisition ofgrammaticalforms in writing (Ferris,1999,2004,2010).The second researchpurposeistoinvestigatethepositiveornegativeeffectsofunfocusedWCF onthe learners・abilitiestowriteaccuratelyinEnglish.Manyofthestudiesinthepastevaluated theeffectivenessofWCF focusedononesyntacticrule(e.g.,definiteandindefinitearticles), butthisisanactionresearchstudyaimedtoevaluate,andimprove,theoveralleffectiveness ofa particular university EFL course.Consequently,severaldifferentgrammaticalrules werecovered during thesemester,whereaseach classsession provided WCF on onerule. Additionally,unfocused WCF wasgiven on participants・spontaneouserrorsrelated to a varietyofgrammaticalforms.
TheteacherprovidedfocusedWCF onthemajortargetgrammaticalform byusingan errorcoding system.ThisindirectWCF wasintended to guidetheparticipantsto notice theirproblemsandfindwaystorepairthem,paying closerattention tothetargetpoint. Thestrugglesthatthelearnersexperienced in theprocessoflanguageproduction were believed to havea positiveeffect(Qi& Lapkin,2001;Sachs& Polio,2007).Asfortheir spontaneous mistakes, the teacher directly corrected ungrammatical or reformulated awkwardforms;thisdirect,unfocusedWCFwasprovidedtohalfofthepartici pantsatodd-numbered classsessionsand totheotherhalfateven-numbered sessions.Thus,although each classsession covered a differentgrammaticalrule,any possibleeffectofthetarget linguistic forms on the effectiveness of the unfocused WCF was controlled for. All participantsweretreatedequallyfrom aneducationalandethicalpointofview.
engagedingrammaranalysisactivitiesbeforeusingtheformsinparagraphwriting.Thus, it is acknowledged that grammar instruction as wellas WCF would account for the participants・ acquisition of target forms. H owever, explicit grammar instruction was administeredtoallparticipantsateveryclasssession,anditseffectsoneitherlearnergroup werecontrolledfor,although theassessmentoftheeffectsofgrammarinstruction perse wasbeyond thescopeand natureofthepresentstudy.Nonetheless,themajorpointof investigation wastodeterminethedegreetowhich WCF contributedtotheiraccurateuse of target forms in new pieces of writing.A within-subjects statistical analysis was conducted to evaluatetheeffectivenessofthetwo treatments:(a)focused WCF and (b)a combinationoffocusedWCF andunfocusedWCF.
At the end of the semester,a questionnaire survey was conducted to probe the participants・perceptionsoffocusedandunfocusedWCF providedduring thesemester.The survey wasintendedtodeterminewhattypesofinstructionaltreatmentlearnerspreferso thatfutureteachingplanscouldbemodifiedtoaccommodatetheirpreferences.Ferris(1995) and H yland (2011)demonstrated thattheparticipantsattended to,and appreciated,their teachers・feedbackongrammaticalerrors,whereasTruscott(1996)arguedthatwhatlearners believe to be the bestfor their learning often differs from whatcan really help their learningandobjectedtotheideaofprovidinggrammaticalerrorcorrectioninaccordwith theirpreference.Thepresentstudystrovetosupporttheformerpositioninthisrespect.
LiteratureReview
Thissectionreviewsthetwoopposingviewsconcerningtheuseoferrorcorrection,the effects of focused and unfocused WCF on L2 learners・accuracy in writing,and the comparisonbetweendirectandindirecttypesofWCF.
FeedbackorNoFeedback
The first major controversy concerning WCF was whether or not correction of grammaticalerrorscan facilitatelanguageacquisition.Truscott(1996,1999)claimed that grammarcorrection doesnotimproveL2learners・ability towriteaccurately in theleast and proposed the policy of abandoning it.Truscott also pointed out that there is a developmentalsequenceforgrammaracquisitionandthatinstructionisnoteffectiveunless learnersarereadyforacertainform;unfortunately,veryfew teacherscandeterminetheir students・currentdevelopmentalstagesandprovidefeedbackontherightgrammaticalitem attherighttiming.Furthermore,even iftheteachercan recognizeandexplain an error, students may not understand the explanation.His objection to provision of grammar correction wasalso based on thefactthatgrammarcorrection overly burdensboth the teacherandstudents.Teacherswasteanenormousamountoftimecorrectingtheirstudents・ errors,whichcouldbe,otherwise,investedforteachingdiscourseconstructionorproviding
content-based feedback.Learners do not feelcomfortable confronting the indication of mistakesand,forfearofmaking mistakes,may avoid writing longerormorecomplex sentences.Truscott・spositionwasthatgrammarcorrectionhasnoroletoplayinlanguage acquisition.
Ferris (1999)protested against Truscott・s proposal,stating that his theorizing was prematureandoverly strong.Sheproposed,instead,thatselective,prioritized,andexplicit types of error correction might be effective.In her opinions,teachers could train to recognizeand correctgrammaticalerrors,and they mightcontinueto providegrammar correction until further studies revealed more decisive evidence for or against error correction.Ferris(2004)furtherarguedthattheexistingresearchbase,althoughinsufficient fordecisiveconclusions,predictedpositiveeffectsoferrorcorrectiononL2writingandthat theteachersshouldmakethebestoftheexistingbodyofknowledgeandcontinuetoprovide WCF.Shealsostatedthatteacherscoulddeepentheirgrammarknowledgeandchoosethe mostappropriateform ofdirectorindirecterrorcorrection depending on theirstudents・ needs,goals,andindividualdifferences.
StudiesinSupportoftheNon-instructionPolicy
In orderto supporthisnon-instruction policy (i.e.,notproviding grammaticalerror correction),Truscott(1996)cited thestudiesby Semke(1984),Robb,Ross,and Shortreed (1986),Kepner(1991),and Sheppard (1992).Semke(1984)conducted a 10-week longitudinal studywith141AmericanuniversitystudentsstudyingGermanasaforeignlanguage(GFL) toevaluatetheeffectsoferrorcorrectiononwritingaccuracy.Theparticipantsweredivided into fourgroupsand received fourdifferenttypesoffeedback on theirfreewritings:(a) directerrorcorrection,(b)commentsandquestionsonthecontent,(c)directcorrectionand positivecomments,and(d)indirectcorrectivefeedbackusingcodes.Theresultsshowedthat therewasnostatisticallysignificantdifferenceamongthefourgroupsintermsofwriting accuracy.Thosewhoreceiveddirecterrorcorrectiondidnotperform anybetterthanthose whoreceived content-based comments.Thecomment-only group expressed amorepositive attitudeconcerningtheirwritingexperiences.
Robb,Ross,and Shortreed (1986)evaluated the effectiveness of direct or indirect feedbackwith134Japanesecollegestudents.Theparticipantsweredividedintofourgroups and received four different types ofWCF:(a)direct correction,(b)coded feedback,(c) highlightedfeedback,or(d)indicationofthenumberoferrorsperline.Theresultsshowed thatdirectcorrection did notresultin greater accuracy,fluency,or complexity in the participants・writing.Instead,allgroupsproduced morecomplex structuresovertimeas theycontinuedtopracticewriting.
Kepner(1991)recruited60intermediateL2SpanishlearnersatanAmericancollegeand evaluated the effects of type of written feedback (error corrections or message-related
comments)andEnglishverbalability(higherorlower)ontheiracquisitionofwritingskills in Spanish.Theparticipantsweredivided intofourgroupsdepending on (a)theirverbal -ability levelsand (b)thetypesoffeedback they received.They engaged in eightjournal writing tasks,and error-correction feedback was administered to two subgroups,and message-related comments were provided to the other two subgroups.One ofthe two subgroupsreceiving eithertreatmentincludedhigherverbal-ability students,andtheother includedlowerverbal-abilitystudents.Thetwo-wayANOVA resultsshowedthatthosewho received message-related feedback produced a significantly greaternumberofhigher-level propositionsintheirjournalwritingandthatthehigher-verbal-abilitywritersconsistently outperformedthelower-verbal-abilitywritersinproductivity.Ontheotherhand,thosewho receivederror-correctionfeedbackdidnotgainsignificantlygreateraccuracythanthosewho receivedmessage-relatedfeedbackregardlessoftheirverbal-ability.
Sheppard (1992)evaluated the effects ofform-focused feedback and meaning-focused feedback on ESL learners・acquisition ofwriting skills.Herecruited26ESL studentswith variousL1backgroundsandprovidedonesubgroup(n＝13)with form-focusedfeedback on grammaticalpointsandtheothersubgroup(n＝13)withholisticfeedbackonmeaning.The resultsshowedthatthemeaning-focusedgroupmadeasignificantprogressfrom pretestto posttest in verb accuracy and punctuation.The form-focused group made a significant progressinaccuracybuttendedtoavoidsubordinationasacomplexanddifficultsentence structure.
Truscott(2007)furtherconductedasmall-scalemeta-analysisbasedontheresultsofsix earlierresearchers・controlledexperiments(includingthefourstudiesabovementioned)and sixadditionalstudiesthatdidnotinvolveacontrolgroupbut,i nstead,measuredpretest-to-posttestgainsthatthegrammar-correctionrecipientsattained.Hecomparedtheeffectsizes forthecitedstudiesandproducedevidencethatgrammarcorrection waslikely tohavea smallharmfuleffecton L2 writers・abilities to write accurately and thatany possible benefitswereminimal.
Summing up,thefindingsfrom thesestudiessuggestedthaterrorcorrection doesnot improveL2learners・accuracy in writing toasignificantdegree,andlearnerswhoreceive content-based feedback tend toproducemorecomplex orideationally bettersentencesand receivea positiveimpression ofwriting activities.Theimplication isthattherolethat grammarcorrectionmightplayforlanguageacquisitionisminimal,ifthereisany. TheoriesandStudiesinSupportoftheProvisionofErrorCorrection
Firstofall,Truscott・stheorizingcontradictsthefunctionoffocus-on-form instruction. Eskey (1983) pointed out that, although communicative language teaching facilitates learners・fluency,itdoesnotautomatically increaseformalaccuracy.With thegrowing popularityofthecommunicativeapproach,therehadbeenageneralshiftfrom theemphasis
on structural accuracy to the development of functional skills that were needed for communicating ideas.However,fluentL2 speakersand writersoften lacked grammatical accuracy,showingthattheacquisitionofcommunicativecompetencedidnotguaranteetheir native-likeuseofthetargetforms.WhengreateraccuracyisdemandedofL2learnersinthe realworld (e.g.writing for professionalpurposes,instead of engaging in daily oral interactions),corrective feedback from the teacher is indispensable.This has also been evidenced by the fact that French immersion students in Canada,who had sufficient opportunitiesforoutput,could notproducecomplex sentencestructuresunlessthey were provided with negative feedback (Swain,1985)or the report that focus-on-form tasks successfully drew immersion students・attention to theirerrorsand consequently improve thequalityoftheirwriting(Swain& Lapkin,1995).
Thereareanumberofempiricalstudiesthatprovideevidencefortheeffectivenessof grammar correction.Sheen (2007)recruited 91 ESL learners in the United States and evaluated the effects of two types of focused WCF (direct error correction with metalinguistic feedback and direct correction only)on their acquisition of the English articles.Theresultsshowed thatthelearnerswho received errorcorrection outperformed thosewhodidnot,andthedirectcorrection with metalinguisticfeedback wasparticularly effectiveforlong-term acquisition.
Bitchener,Young,and Cameron (2005)engaged 53 adult migrant students in New Zealandinwritingtasksoveranextendedperiodandevaluatedtheeffectsoftwodifferent typesofWCF ((a)explicitexplanation ofgrammaticalerrorsandsuggestionsforchanges and (b) explicit explanation and suggestions combined with student-teacher individual conferences) on the participants・ abilities to use three different grammatical rules (prepositions,the simple pasttense,and the definite article).The learners who received eithertypeofWCFoutperformedthosewhoreceivednofeedback.Theresultsalsoindicated thattheexplicitfeedback with individualconferenceshadasignificantly positiveeffecton learners・accuracy with thedefinitearticleand thepasttense,which wererule-governed linguisticfeatures.
Likewise,Bitchenerandhiscolleaguesconductedaseriesofstudiesontheeffectiveness offocused WCF on L2 learners・acquisition ofthe English article system (definite and indefinite articles) and provided positive evidence. Bitchener (2008) engaged 75 l ow-intermediate ESL learners in New Zealand in written picture-description tasks and investigated theeffectsofWCF on theiracquisition ofthearticles.Hedivided thegroup intofourgroupsandprovidedthethreeexperimentalgroupswith(a)directerrorcorrection with written and oralmeta-linguisticexplanation,(b)directerrorcorrection with written explanation,or (c)directerror correction only.The controlgroup did notreceive any feedback.Theparticipantsproducedthreepiecesofwritingdescribingtheprovidedpictures. Theresultsshowed allthreeexperimentalgroupsgained greater accuracy atimmediate
posttestthan the controlgroup and retained the high levelofperformance atdelayed posttest,evidencing thaterrorcorrection waseffectiveand worth theteacher・stimeand effort.
Bitchenerand Knoch (2008)conducted asimilarESL study with 75internationalvisa studentsand 69 migrantstudents.Theexperimentaltreatmentsincluded (a)directerror correction, written meta-cognitive explanation, and oral explanation, (b) direct error correction andwritten meta-linguisticexplanation,and(c)directerrorcorrection only.All threeexperimentalgroupsoutperformedthecontrolgroupattheimmediateposttestandat the delayed posttestadministered seven weeks later.There was no statisticaldifference betweenthemigrantandinternationalstudents・accuracyinwriting.
Bitchener and Knoch (2009a)evaluated the effectiveness of three different types of focusedWCF with52low-intermediateESL studentsinNew Zealand.Thethreetreatments were:(a)directerrorcorrection,written meta-linguisticexplanation,andoralexplanation, (b)directcorrection andwritten meta-linguisticexplanation,and(c)directerrorcorrection only.Allexperimentalgroupsoutperformedthecontrolgroupattheimmediateposttestand thethreedelayedposttestsadministeredtwo,six,or10monthsafterthetreatment.
Bitchener and Knoch (2009b) also conducted a longitudinal study with 52 l ow-intermediateESL learnersinNew Zealandtocomparetheeffectivenessof:(a)wri ttenmeta-linguisticexplanation,(b)written meta-linguisticexplanation followed by an oralform-focusedreview ofthewrittenexplanation,(c)andindicationoferroneouspartsintheform ofcircling.Theparticipantsweredividedintothreeexperimentalgroups,receiving oneof the three forms of feedback, and a control group. All three experimental groups outperformed thecontrolgroup attheimmediateposttestand atthethreedelayed tests. Therewerenosignificantdifferencesamongthethreeexperimentalgroups.
Bitchener and Knoch (2010)investigated the effects of WCF on 63 advanced ESL learners・accurateuseofthedefiniteandindefinitearticles.Theparticipants,ESL learners in theUnited States,received:(a)written meta-linguisticexplanation and an oralform-focusedreview ofthesameexplanation,(b)writtenmeta-linguisticexplanation,or(c)error circling.Thecontrolgroupdidnotreceiveany feedback.Thegroupsthatreceivedwritten meta-linguistic explanation with or without an oral form-focused review achieved significantlygreateraccuracyattheimmediateanddelayedposttestseightweekslater.The groupthatreceivederrorcircling wasabletoincreaseitsaccuracy atimmediateposttest, butitsaccuracyleveldecreasedafterward.Thecontrolgroupdidnotmakeanystatistically significantprogress.
The overall implications of these studies are that L2 learners who receive error correctionarelikelytolearntousegrammaticalformsmoreaccuratelythanthosewhodo not,andexplicitexplanationstendtoreinforcetheeffectivenessofdirecterrorcorrection.
FocusedorUnfocused
On theassumption thatWCF hasa roleto play in languageacquisition,oneofthe controversialissues is whether feedback should be focused on one grammaticalrule or providedonmultiplelinguisticerrorsthatlearnersmakespontaneously.
Sheen (2007),Bitchener(2008),and Bitchenerand Knoch (2008,2009a,2009b,2010),as abovementioned,investigatedtheeffectivenessoffocusedWCF;allofthesestudiesevaluated learners・acquisition oftheEnglish articlesystem.Theresultsindicatedthatexperimental groupsreceivingfocusedWCF invariousforms(e.g.,directerrorcorrection,wri ttenmeta-linguisticinput,ororalmeta-linguisticinput)consistently outperformedthecontrolgroup thatreceivednofeedback.Bitchener,Young,andCameron (2005)alsoevaluatedtheeffects offocusedWCF (explicitwritten feedback andteacher-studentconference,explicitfeedback only,andnofeedback)onthreedifferentgrammaticalforms(thedefinitearticle,thesimple past tense,and prepositions)and revealed that the rule-governed features were more amenabletoWCF.
Studiesthatinvestigatedtheeffectsoffocusedandunfocusedcorrectivefeedback(CF)at thesametimehavebeen scarce.Ellis,Sheen,Murakami,andTakashima(2008)engageda groupof35JapaneseEFL studentsinnarrativewritingtasksandmeasuredtheeffectsof focusedandunfocusedWCFontheiraccurateuseofgrammaticalrulesinEnglish.Thefirst experimentalgroupreceivedWCFonlyonthedefiniteandindefinitearticles,andthesecond experimentalgroup received feedback on severaldifferentgrammaticalforms.Theresults showed that both made a significant improvement from pretest to posttest and also outperformed the controlgroup thatreceived no feedback.However,the study did not indicatewhetherfocusedfeedbackismoreeffectivethanunfocusedfeedbackorviceversa.
Sheen,Wright,andMoldawa(2009)evaluatedtheeffectsofWCF on80ESL learners・ acquisition ofgrammaticalformsataUS college.Theparticipantsweredividedintofour groups:Focused Written CF Group received direct WCF only on the English articles; UnfocusedWrittenCFGroupreceivedfeedbackonthearticlesalongwithothergrammatical structures (i.e.,copular be,regular past tense,irregular past tense,and prepositions); WritingPracticeGroupengagedinwritingpracticebutreceivednofeedback,and;Control Group only took the tests.The results showed that focused WCF could contribute to grammaticalaccuracyinL2writingbutunfocusedWCF wasnotaspedagogicallyusefulas writingpracticeitself.
Truscott and Hsu (2008)provided unfocused WCF to EFL students in Taiwan to demonstratethattheprovidedfeedbackwasineffective.However,theform offeedbackthey provided(simply underlining theincorrectparts)wasattheextremeendofindirectWCF, andthereisapossibilitythatasomewhatmoreexplicittypeofunfocusedWCFmighthave facilitatedL2writers・accuracy.
studiesthathavecomparedtheeffectsoffocusedandunfocusedWCF directly arescarce, and the evidence against unfocused WCF is not sufficient and decisive.The roles and effectivenessofunfocusedWCF needtobefurtherinvestigated.
TypesofFeedback:DirectandIndirectFeedback
Many studieshavedemonstrated theeffectivenessofindirectWCF by experimentally utilizing different forms of indirect feedback,whereas some studies have reported the positivefunctionsofdirecterrorcorrectionaswell.
Lalande (1982)theorized that it is pedagogically more beneficialto guide learners systematically to notice their ow n errors and discover the correct forms,instead of providingdirecterrorcorrections.HedividedagroupofintermediateGerman-as-a-Forei gn-Languagelearners(N＝60)into tw o subgroups:theexperimentalgroup had theirerrors marked w ith codes,and thecomparison group received directerrorcorrecti on.Theerror-codinggroupgainedgreateraccuracyoverallthanthedirect-correctiongroup.Theindi rect-WCF groupalsolearnedtousecasethemostdifficultgrammaticalruleamongalltarget formsaccurately in anew pieceofwriting.Thestudy resultssuggestedthattheerror-awarenessandproblem-solvingtechniquesthatindirectWCFinducedhadapositiveeffecton intermediateGFL students・finalwriting.
Robb,Ross,andShortreed(1986),abovementioned,evaluatedtheeffectsofdirectand indirectWCF onJapanesecollegestudents・writings.Theresultsindicatedthatmoredirect feedback did notresultin greateraccuracy,fluency,orcomplexity in L2writing.Semke (1984)also compared directand indirectWCF,aswellascontent-based feedback,butthe resultsdidnotproduceanyevidencetosuggestthateitheroneismoreeffective.
Chandler(2003)investigated theeffectivenessoffourtypesofWCF:directcorrection, underlining w ith description of an error type,underlining only,and description only. Participantswere36ESL studentsin theUnited States.Theresultsindicated thatdirect correctionandunderliningweresignificantlymoreeffectivethanthedescriptionofanerror type for improving the learners・ accuracy in a subsequent w riting assignment. The participants seemed to receive less discouragement from underlining than from the description oferrortypes.However,they feltthatthey learnedmorefrom self-correction basedontheerror-typedescriptions.OnepossibleinterpretationofChandler・sstudyisthat theindicationofanerroreitherdirectlycorrectingitorsimplyindicatingitslocationcan effectively improve learners・accuracy in new writing tasks,butthe learners mightbe motivatedtolearnthetargetstructurewhenprovidedwithchancestoreflectontheirown errors.
VanBeuningen,DeJong,andKuiken(2012)investigatedtheeffectsofdirectcorrection andindirectWCFonL2Dutchlearners・abilitiestoedittheirdraftsandtowriteaccurately on a new topic(N＝268).ExperimentalGroup 1 received directcorrection,Experimental
Group2receivedindirectcorrectivefeedbackusingtheerrorcodingsystem,ControlGroup 1 received no feedback and self-edited their writings,and ControlGroup 2 received no feedback and engaged in a completely new writing task.Theresultsshowed thatdirect correction resulted in grammaticalaccuracy gains(e.g.,abilitiesto producesyntactically correctformsinvolving articles,inflections,word order,etc.),and indirectCF resulted in improvingnon-grammaticalaccuracygains(e.g.,relatedtolexis,pragmatics,orthography).
Sheen (2007),Bitchener (2008),and Bitchener and Knoch (2008,2009a,2009b),above mentioned, compared the functions of direct correction with or without written meta-linguisticfeedbackand/ororalexplanation.Thegeneraltendencywasthatthewritten meta-linguistic feedback made a major contribution,suggesting that somewhat explicit explanation of a target form reinforces the effects of direct correction on language acquisition.Bitchener,Young,and Cameron (2005)and Bitchener and Knoch (2010)also producedevidencetosupportthisview.
Thefindingsfrom thesestudiesweremixed,reflecting thevarying effectsofdifferent typesofindirectWCFadministeredindifferentteachingenvironments.Oneimportantpoint to note,however,isthattheindirectWCF islikely to involveL2 learnersin cognitive processing,which islikely to facilitatetheirlanguageacquisition.On theotherhand,it mustalsobeacknowledgedthatbothdirecterrorcorrectionandindirectWCFhaverolesto play,helpingtoimprovedifferentaspectsoflearners・writingabilities.
ResearchQuestionsandHypotheses
Allparticipantsreceivedexplicitgrammarinstructionaspartofthecoursework,which mighthavepartiallycontributedtotheiracquisitionofgrammaticalforms.Consideringthis teachingenvironment,thefollowingfourresearchquestionswereputforward.
ResearchQuestion1:DoesacombinationofexplicitgrammarinstructionandfocusedWCF improvetheparticipants・knowledgeofgrammaticalrulesasmeasured in a di screte-pointgrammartest?
Hypothesis1:TheadministrationofgrammarinstructionandfocusedWCF improvestheir knowledgeofgrammaticalrules.
Research Question 2: Does focused WCF facilitate the participants・ accurate use of grammaticalrulesinfreewriting?
Hypothesis2:FocusedWCFfacilitatestheiraccurateuseoftargetgrammaticalrulesinfree writing.
ResearchQuestion3:How doesunfocusedWCF affecttheeffectivenessoffocusedWCF? ThereisnoapriorihypothesisforResearchQuestion3.UnfocusedWCF onavariety
ofgrammaticalformsmay influencetheeffectsoffocused WCF eitherpositively or negatively.
Research Question 4:How do the participants perceive the provision of focused and unfocusedWCF onparagraphwritings?
ResearchQuestion4isexplorative,andthereisnoapriorihypothesisforthisquestion, either.
Method Participants
Participantswere29third-yearstudentsenrolledataJapaneseuniversity:theirexplicit knowledge ofEnglish grammar could be regarded as advanced by Japanese university students・standards.Ninewerefemale,and 20weremale.They had already received two yearsofEFL trainingattheuniversityinadditiontosixyearsoffocus-on-forms-oriented EFL education in juniorhigh schoolandhigh school.Noneofthem hadan experienceof stayingand/orstudyinginanEnglish-speakingcountryforayearorlonger.Theclassmet onceaweekfora90-minutesession.
InstructionalTreatment
Classroom activities.The course in which the participants enrolled was an elective linguisticscourse,designed primarily to deepen theirunderstanding ofEnglish grammar. Thus,eachclasssessionbeganwithgrammaranalysisactivities.Theteacher(theresearcher himself) handed out a worksheet that presented grammar questions and guided the participantstoanswerthem in class.Heoffered meta-linguisticexplanationsforwhathe assumed to bedifficultfeaturesofthetargetgrammaticalrule.Themajorgrammatical rules covered during the experimentalperiod included:definite and indefinite articles, countablenounsand uncountablenouns,presentperfect,passivevoice,theverbsletand make,relative pronouns,hypotheticalconditionals,reported speech,and infinitives and gerunds.Eachclasssessioncoveredoneofthesegrammaticalforms.
Afterthecontrolled grammaranalysisexercises,theparticipantspracticed using the targetgrammaticalform throughparagraphwriting.Theyfirstreceivedthepromptforthe writing task and,asapre-writing activity,engaged in asmall-group discussion toshare interestingideas.Forexample,forthepresentperfecttense,thewritingtaskrequiredthem torefertofiveacademic,athletic,orreal-lifegoalsthatthey hadaccomplishedsincethey enteredtheuniversity(e.g.,Ihaveparticipatedintwointercollegiatetennistournaments;I havegotten my driver・slicenseafterattending a driving schoolforthreemonths;Ihave attainedan 800TOEIC score.).Membersofeach grouprecountedtheirown experiencesso thattheycouldshareideasforparagraphcomposition.Theywereencouragedtocarryout
theirdiscussion in English butwereallowed to switch to Japanesewhen they could not fully expresstheirideasin thetargetlanguage.Then,arepresentativeofeach groupwas calledupontoreportthegeneratedideasbacktotheclassinEnglish.Allparticipantsmade notesfortheirownwritings,tooktheirnoteshometowritetheirparagraphs,andtypedup andsubmittedthetexttotheteacheratthenextclasssession.
Written correctivefeedback.Asforthetargetgrammaticalrule,theteacherprovided participantswithfocusedWCF usingacodingsystem.A tableofcodesusedforWCF was distributed atthebeginning ofthesemester(e.g.,VT fora verb-tenseproblem,WC for wrong word choice,WO for wrong word order,SV-agr for a subject-verb agreement problem);the teacher orally explained some of the codes that were likely to be used frequently.Whenweeklyparagraphsweresubmitted,heunderlinedtheungrammaticalparts and used codesto indicatetheerror types.Regarding thetargetgrammaticalrule,the correctformsthattheparticipantsusedwerecircledaspositivefeedback,aswasindicated atthebeginningofthecourse.Theparticipantswereencouragedtoproduceinformativeand interesting paragraphs while trying to use the target grammaticalforms,instead of mechanically constructing target linguistic structures.They received one grade on the grammarusage(grammargrade)andanotherontheoverallorganization,vocabulary,and content(compositiongrade).
In order to evaluate how the additional direct WCF on a variety of lexical or grammaticalformsmightinfluencetheparticipants・attentiontothemajortargetsyntactic ruleeitherpositivelyornegativelytheteacheralternatelyprovidedhalfoftheparticipants with unfocused WCF.Theclasswasdivided into two subgroups:Group A and Group B. GroupA receivedWCF onallgrammaticalmistakesatodd-numberedclasssessions,i.e.,in addition to focused WCF,whereasGroup B received only focused feedback on themajor targetform.Then,Group B received WCF on allerrorsateven-numbered classsessions whileGroupA receivedonlyfocusedfeedback.
Theinstructionalprocesswascomposedofthreesteps,includinginstruction,submission, andfeedback,andextendedoveraperiodofthreeweeks(i.e.,threeclasssessions).However, every week,theclassproceeded to study a new grammaticalruleso thattheintervals betweendifferenttopicswereefficientlyutilized.
Asmentioned above,many oftheearlierstudieshaveindicated thatfocused WCF is generallymoreeffectivethanunfocusedWCF forthelong-term acquisitionofgrammatical forms. However, there has been no strong evidence that giving feedback on other grammaticalformsreducestheeffectivenessoffocusedWCF.
Testing
Discrete-pointgrammartests.A pretest-posttestdesignwasusedtomeasuretheparti c-ipants・knowledgeofthetargetgrammaticalrules.Atthebeginning ofthesemester,a
pretest, which comprised multiple-choice question items related to all of the major grammaticalrulesto becovered during thesemester,wasadministered to measurethe participants・baseknowledge.Thesametestwasadministeredattheendofthesemesterto measuretheirgains;thequestionitemswerescrambledandmixedwithafew distractorsto preventany possibletesteffect.Becauseofclassadministration constraints,itwasnot possibletoadministeradelayedposttest.
Essaywriting.Theparticipantswroteapreliminaryessayandafinalessay,whichwas intendedtodeterminethedegreetowhich theiroverallwriting skillsandtheirability to usethetargetgrammaticalformsaccurately in freewriting improved.Theparticipants wrotethepreliminaryandfinalessaysonnew topics.AsBitchenerandFerris(2012)stated, L2 writers・ability to editand polish their draftsdoesnotautomatically translateinto languageacquisition and thatitisimportantto evaluatetheaccuracywith which they producenew piecesofwriting.Theextenttowhichtheparticipantsgainedaccuracyovera four-monthperiodwasmeasured.Whereastheparticipantswroteandsubmitted10weekly paragraphsduringthesemester,eightofthem wereusedforanalysis;thefirstassignment was intended as a warm-up assignment and allparticipants received both focused and unfocusedWCF onthelastweeklyessay.
TheeffectsofunfocusedWCF werenotdirectlymeasured.Instead,themeansforthe grammaticalitemsonwhichtheparticipantsreceivedonlyfocusedWCF andthoseforthe itemsonwhichtheyreceivedbothfocusedandunfocusedWCFwerecompared.Ifthemeans forthefocused-WCF-onlyweresignificantlygreater,itwouldmeanthattheunfocusedWCF mighthave interfered with participants・concentration on the main targetgrammatical forms.On theotherhand,ifthemeansforthefocused-and-unfocused-WCF weregreater, therewasapossibilitythatadditionalWCF onavarietyofformshadapositiveeffect.If therewasnosignificantdifference,itmightmean thattheadditionalunfocused feedback wasneitherbeneficialnorharmful.
Questionnairesurvey.Inadditiontotheobjectivegrammartestsandtheessaytests,a questionnairesurveywasconductedattheendofthesemestertofathom theirpositiveor negativeperceptionsoffocusedorunfocusedWCF.
Theparticipantswereasked:
a.whetherornottheypreferredtohavetheirgrammaticalerrorscorrected; b.(ifyeson(a))whattypesoferrorstheywantedtheteachertocorrect; c.(ifyeson(a))towhatextenttheywantedtheteachertocorrecttheerrors;
d.(ifyeson (a))whattypesofcorrectivefeedback they preferred (e.g.,directerror correction,codedfeedback,meta-linguisticexplanation);
e.(ifnoon(a))whytheypreferrednottohavetheirerrorscorrected,and; f.how oftentheyreviewedthewrittencorrectivefeedbackfrom theteacher.
Results t-TestResults
In order to evaluate the effects ofWCF on the participants・acquisition ofexplicit grammar knowledge (Research Question 1),their means for the preliminary and final grammartests(i.e.,pretestand posttest)werecompared.Theparticipants・raw scoresat pretestand posttestwereconverted into Rasch measuresusing thepartialcreditRasch model.Raschmeasuresaremoreusefulforaccuratestatisticalmeasurementthanraw scores becausetheyareequal-intervalmeasuresthatarederivedfrom theprobabilisticrelationships betweenpersonabilitiesanditem difficulties(Bond& Fox,2007).TheRaschpersonmeasures werefurtherconverted to responseprobability units(CHIPS).Thislineartransformation meantthattheaveragepersonmeasurewassetat50,andthehighestandlowestpossible scoreswererespectively 80 and 20.Theitem separation was2.76,which wasabovethe criterion pointof2.00,and the item reliability was 0.88,which was very close to the criterionpointof0.90.AsshowninTable1,thepretestmeanwas51.02(SD＝1.81),andthe posttestmeanwas55.74(SD＝2.74).
Then,a t-testwasconducted to determinethedegreeofstatisticalsignificance.The independentvariablewastest(i.e.,pretestandposttest),andthedependentvariableswere theparticipants・scoreson thediscrete-pointgrammartestsadministeredatthebeginning and theend ofthesemester.Theresultsshowed thattheparticipants・scoresimproved significantlyfrom pretesttoposttest,t(24)＝7.73,p＝0.001.
Table1.DescriptiveStatisticsforDiscrete-PointGrammarTestResults Pretest M 51.02 95％ CI LowerBound 50.27 HigherBound 51.77 SD 1.81 Skewness 0.10 SES 0.46 Kurtosis －0.70 SEK 0.90 Posttest M 55.74 95％ CI LowerBound 54.61 HigherBound 56.87 SD 2.74 Skewness 0.15 SES 0.46 Kurtosis －0.53 SEK 0.90 Note.N＝25.
Repeated-MeasuresANOVA Results
In ordertoevaluatetheoveralleffectsofWCF on theparticipants・abilitiestowrite moreaccurately in freewriting (Research Question 2),theparticipants・gainsin grammar gradesand composition gradeson essay writing weremeasured,and atwo-way repeated-measuresanalysisofvariance(ANOVA)wasconducted.
Obligatorycontextanalysiswasusedtocomputetheparticipants・grammargrades.The number ofcorrectly formed phrases or sentences in each individualwriter・s essay was dividedby thetotalnumberofobligatory contextswheretheuseofthetargetform was required.Regardingtheoverallcompositiongrades,theresearcherandanotherexperienced EFL teacherholistically gradedtheparticipants・preliminary andfinalessayson a5-point scale.Thecriteriaforjudgmentincludedcontent,vocabulary,organization,mechanics,and grammar(relatedtotheuseofallgrammaticalrulesotherthanthemajortargetform),but each ofthetworatersgaveoneholisticscoreon each participant・sessay.Theinter-rater reliabilitywasr＝0.89,p＜0.01,andtheratersnegotiatedover,andadjusted,thegradeson whichtheyhadnotagreed.
Asshown in Table2,thegrammarmean forpreliminary essay writing was3.08(SD ＝0.65)andthegrammarmeanforfinalessaywritingwas3.96(SD＝0.75).Thecomposition meanforpreliminaryessaywritingwas2.46(SD＝0.66),andthecompositionmeanforfinal essay writing was 3.50 (SD＝0.66). There were noticeable gains in both writing-skill categories. Table2.DescriptiveStatisticsfortheGrammarandCompositionGradesfor PreliminaryandFinalEssays Grammar Composition PreliminaryEssay M 3.08 2.46 95％ CI LowerBound 2.81 2.18 HigherBound 3.36 2.74 SD 0.65 0.66 Skewness －0.08 0.17 SES 0.47 0.47 Kurtosis －0.42 0.05 SEK 0.92 0.92 FinalEssay M 3.96 3.50 95％ CI LowerBound 3.64 3.22 HigherBound 4.28 3.78 SD 0.75 0.66 Skewness 0.00 0.07 SES 0.47 0.47 Kurtosis 0.00 －0.81 SEK 0.92 0.92 Note.N＝24.
Subsequently, a two-way repeated-measures ANOVA was performed to determine whetherornotthegainswerestatisticallysignificant.Thewithin-subjectsfactorsweretest with two levels (preliminary essay and finalessay)and skillcategory with two levels (grammarandcomposition).Thedependentvariablesweretheparticipants・gradesforeither writing-skillcategory atpreliminary and finalessay writing.Theunivariatetestresults (Table3)showedthatthetestmaineffectwassignificant,F(1,23)＝59.94,p＜0.05,・2＝0.72,
theskillcategorymain effectwassignificant,F(1,23)＝19.15,p＜0.05,・2＝0.45.Thetestx
skillcategory interaction was not significant,F(1,23)＝0.41,p＞0.05,・2＝0.02.That is,
regardless ofwriting-skillcategory (reflected in grammar and composition grades),the participants・ scores improved to a statistically significant degree at the end of the experimentalperiod.They learned to writemoreaccurately and produceideationally and organizationallybetterwritings.Theparticipants・grammargradeswerehigherthantheir composition grades,which wasnotsurprising becausethetargetgrammaticalformshad been explicitly taughtbefore writing tasks.However,as the grammar and composition gradeswerecomputed on differentscalesindependently from each other,thegrammar-compositiondifferenceperseshouldbeinterpretedcautiously.Nonetheless,itwasimportant to confirm thattheparticipantslearned to writeideationally and organizationally better essaysoverthesemester,notassemblingthetargetsentencestructuresmechanically.
ANCOVA Results
The effectiveness ofthe focused WCF and thatofthe focused and unfocused WCF combined(ResearchQuestion3)wereevaluatedbycomparingtheparticipants・essayscores forthegrammaticalitemsonwhichtheyhadreceivedfocusedandunfocusedWCF ( focused-and-unfocused)duringthesemesterandthoseforthegrammaticalitemsonwhichtheyhad receivedfocused WCF only(focused-only).In otherwords,theparticipants・abilitiestouse thetargetgrammaticalformsinfinalessaywritingwerereanalyzedbydividingthetarget itemsintothosetowhich they mighthavepaid exclusiveattention on theonehand and thosetowhichtheymighthavepaidsomewhatlessattenti
onontheotherhand.Focused-Table3.Repeated-MeasuresANOVA Results(Univariate)
Effect df SS MS F p ・2 Test 1 22.04 22.04 59.94 0.001 0.72 Residual 23 8.46 0.37 Grades 1 7.04 7.04 19.15 0.001 0.45 Residual 23 8.46 0.37 TestxGrades 1 0.17 0.17 0.41 0.53 0.02 Residual 23 9.33 0.41 Note.α＝0.05.
and-unfocused and focused-only scores were computed based on the same participants・ preliminary and finalessays above mentioned butwere,for more accurate assessment, converted into Rasch measures,using thedichotomousmodel.TheRasch measureswere furthertransformedintoresponseprobabilityunits.Theitem separationwas3.12,andthe item reliabilitywas0.91.
Table4displaysthedescriptivestatistics.Thefocused-and-unfocusedmeanforpreliminary essay was48.65(SD＝6.46);thefocused-only mean was49.02(SD＝5.17) .Thefocused-and-unfocused mean forfinalessay was52.47(SD＝6.42);thefocused-onlymean was53.89(SD ＝5.95).Thefocused-onlymeanswereslightlyhigherbothatthebeginningandtheendof thesemester.
Theparticipants・gainsbetween preliminary essay and finalessay wereevaluated by performingaone-wayanalysisofcovariance(ANCOVA).Theindependentvariable,treatment, included two levels: focused-and-unfocused-WCF and focused-WCF-only. The dependent variablewastheparticipants・scoreson thefinalessay,andthecovariatewastheirWCF scores on the preliminary essay.A preliminary analysis evaluating the homogenei ty-of-slopesassumption indicatedthattherelationshipbetween thecovariateandthedependent variabledid notdiffersignificantly asa function oftheindependentvariable,F(1,44),p ＝0.86,・2＝0.001.The ANCOVA was not significant,F(1,45)＝0.58,p＝0.45,・2＝0.013.
Whereastheparticipants・meanafterreceivingfocusedWCF onlywasslightlyhigherthan
Table4.DescriptiveStatisticsfortheTwoTreatments・Effects Focused＋Unfocused FocusedOnly PreliminaryEssay M 48.65 49.02 95％ CI LowerBound 45.93 46.83 HigherBound 51.38 51.20 SD 6.46 5.17 Skewness －0.41 －0.24 SES 0.47 0.47 Kurtosis －0.70 －0.49 SEK 0.918 0.92 FinalEssay M 52.47 53.89 95％ CI LowerBound 49.76 51.38 HigherBound 55.18 56.40 SD 6.42 5.95 Skewness －0.69 0.07 SES 0.47 0.47 Kurtosis －0.02 －0.81 SEK 0.92 0.92 Note.N＝24.
their mean after receiving focused and unfocused WCF,the two means did not differ significantly,suggestingthattheadditionalfeedbackonavarietyofgrammaticalformsdid notinterferewithlearners・concentrationonthemajortargetform.
QuestionnaireSurveyResults
Theresultsofthequestionnairesurvey wereanalyzedtounderstandtheparticipants・ perceptionsofteacherfeedback(ResearchQuestion4).Twostudentswereabsentatthelast classmeeting when thesurvey wasconducted,and theN-sizewasreduced to27.Table5 summarizestheparticipants・responsestothequestionitems.
Table5.QuestionnaireSurveyResults Question NumberofResponses PreferenceforReceivingWCF Prefer 27 DoNotPrefer 0 TypesofErrorsTheyWanttheTeachertoCorrect GrammaticalErrors 27 AwkwardExpressions 26 ParagraphConstructionProblems 11
Spellings 14
Punctuation 8
Japanese-EnglishExpressions 9 ExtentofCorrectionTheyPrefer
AllErrors 21
FocusedGrammaticalForm(s)Only 2
2or3Items 0 4or5Items 0 A LimitedNumberofItemsIfThereAreMany 4 Form ofErrorCorrection DirectCorrection 18 UsingCodes 0 Underlining 6
IndicatingtheNumberofErrors 0 DescribingErrorTypesintheMargin 2
Reformulation 3
TheNumberofTimesTheyReview theFeedback
Never 0
Once 14
TwoorMoreTimes 10
Firstofall,itisnoteworthythatallofthe27participantsansweredthattheywanted toreceiveWCF andthatallofthem indicatedthattheywishedtohavetheirgrammatical errorsto becorrected.Almostallparticipants(26 outof27)also referred to awkward expressions(includingincorrectchoiceofwordsoridioms)asanothermajorerrorcategory on which they preferred to receive teacher feedback.As for the other categories (i.e., paragraphing problems,spellings,punctuation,the use ofJapanese-English expressions), therewerenoticeableindividualdifferences,althougheachcategorywaschosenbyeightto 14people.Thechi-squaretestindicatedthattherewasasignificantdifferenceamongthesix items,χ2＝22.92,p＝0.001.Table6showsthepairwisecomparisonresults.
Regarding the form of feedback,as many as 18 participants answered that they preferreddirectcorrection.Nooneindicatedtheirpreferencefortheuseofacodingsystem. Six studentspreferred underlining orotherformsoforthographicenhancement,and two preferred margin notes explaining the error type, and three preferred provision of alternativeorreformulatedphrasesorsentences.Thechi-squaretestresultindicatedthere was a significant difference among the four types,χ2＝22.45,p＝0.001.The results of
pairwisecomparison areshown in Table7;itisevidentthatthetargetstudentgroup preferreddirecterrorcorrection.
Table6.Chi-squareTestsResults:ItemsforCorrection
PairwiseComparison ObservedFrequency FrequencyExpected χ2 p
Grammarvs.AwkwardExpressions Gr.27 Awk.26 26.5 0.19 0.891 Grammarvs.Paragraphing Gr.27 Par.11 19 6.74 0.009 Grammarvs.Spelling Gr.27 Sp.14 20.5 4.12 0.042 Grammarvs.Punctuation Gr.27 Punc. 8 17.5 10.31 0.001* Grammarvs.JapaneseEnglish Gr.27 Jap. 9 18 9 0.003* Awkwardvs.Paragraphing Awk.26 Par.11 18.5 6.08 0.014 Awkwardvs.Spelling Awk.26 Sp.14 20 3.6 0.058 Awkwardvs.Punctuation Awk.26 Punc. 8 17 9.53 0.002* Awkwardvs.JapaneseEnglish Awk.26 Jap. 9 17.5 8.26 0.004* Paragraphingvs.Spelling Par.11 Sp.14 12.5 0.36 0.549 Paragraphingvs.Punctuation Par.11 Punc. 8 9.5 0.47 0.491 Paragraphingvs.JapaneseEnglish Par.11 Jap. 9 10 0.2 0.655 Spellingvs.Punctuation Sp.14 Punc. 8 11 1.636 0.201 Spellingvs.JapaneseEnglish Sp.14 Jap. 9 11.5 1.087 0.297 Punctuationvs.JapaneseEnglish Punc. 8 Jap. 9 8.5 0.059 0.808 Note.・＝0.05.*＝significantafterHolm・sSequentialBonferroniadjustment.
Asfortheireffortsto pay attention to theteacherfeedback,14answered thatthey wouldreview thefeedbackonce,and10answeredthattheywouldreflectonittwoormore times. The chi-square test results showed that there was no statistically significant differencebetween thesetwo groups,χ2＝0.67,p＝0.41.No onesaid thatthey neverpaid
attentiontothefeedback.Threeparticipantsdidnotrespondtothisquestion,eitherunable to decide or not seriously concerned about the feedback in spite of the fact that all participantsansweredtheypreferredtheteachertocorrecttheirerrors.
Discussion
The firstresearch question was related to the effects ofa combination ofexplicit grammarinstructionandfocusedWCFontheparticipants・acquisitionofgrammaticalrules. Theresultsofdiscrete-pointgrammartestsshowedthattherewasasignificantgaininthe participants・posttestscores,supportingthehypothesisthatpredictedthepositiveeffectsof the grammar teaching,weekly paragraph writing,and WCF on their explicitgrammar knowledge.
The second research question was concerned with the effects of WCF on the participants・ability to use grammaticalrules accurately in free writing.The repeated-measuresANOVA resultsshowedthattheirgrammargradesimprovedsignificantly from preliminary essay to finalessay;thus,the second hypothesis thatWCF would have a positiveeffectwasalsosupported.Itisimportanttonoticethattheircomposition grades also improved from preliminary essay to finalessay,evidencethattheparticipantswere engagedin meaning-focusedcomposition tasks,insteadofmechanically trying toassemble thetargetlinguisticstructuresasrequiredbythetasks.
Truscott(1996,1999,2007)putforwardaverystrongproposalthatgrammarcorrection hasnoroletoplayinL2writinginstructionandthusshouldbeabandoned.Theresultsof thegrammartestsandtheanalysisofparticipants・essaysinthepresentprojectcombined
Table7.Chi-SquareTestResults:TypesofFeedback
PairwiseComparison ObservedFrequency FrequencyExpected χ2 p DirectCorrectionvs.HighlightingDC 18 H 6 12 6 0.14 DirectCorrectionvs.ErrorType
Description DC 18 ETD 2 10 12.8 0.000* DirectCorrectionvs.Reformulation DC 18 Ref.3 10.5 10.71 0.001* Highlightingvs.ErrorType
Description H 6 ETD 2 4 2 0.157 Highlightingvs.Reformulation H 6 Ref.3 4.5 1 0.317 ErrorTypeDescriptionvs.
Reformulation ETD 2 Ref.3 2.5 0.02 0.655 Note.・＝0.05.*＝significantafterHolm・sSequentialBonferroniadjustment.
to producecounterevidenceagainsthisproposaland,instead,to supportFerris・sposition (1999,2004)thatselectiveandprioritizederrorcorrectionfacilitateslearners・accuracyinfree writing.Althoughconductedonaverysmallscale,thepresentstudycanbeaddedtoalist ofstudiesin supportoftheposition thatexplicitly and systematically provided grammar feedbackcanhelpL2learnersuserule-governedformsaccurately(Ferris,1999,2004;Sheen, 2007;Bitchener,2008;Bitchener& Knoch,2008,2009a,2009b,2010;Bitchener,Young,& Cameron,2005).WCFispedagogicallyeffectiveandshouldbeutilizedinL2writingcourses.
Thethirdresearchquestionwasrelatedtotheissueofhow unfocusedWCFonavariety ofgrammaticalitemsmightaffecttheeffectivenessoffocusedWCF.TheANCOVA results indicatedthattheparticipants・focused-onlymeanandtheirfocused-and-unfocusedmeandid notdiffersignificantly,suggestingthatprovisionofWCF onafew additionalgrammatical formsdidnotseriouslyinterferewiththelearners・attentiontothemajortargetform.Of course,theabsenceofameandifferencedoesnotconstituteanyconcreteevidencethatthe combination offocused and unfocused feedback ismoreeffectiveforlanguageacquisition thanthefocusedfeedbackadministeredindependently.However,asstatedatthebeginning ofthispaper,consideringthelimitedclasstime,itispedagogicallymorepracticaltoprovide feedbacknotonlyononegrammaticalrulebutalsoonseveralspontaneousgrammaticaland lexicalerrorsthatEFL learnerscommit.Inthisregard,ithasbeenmeaningfultoconfirm thatunfocusedfeedbackdoesnotunderminetheeffectoffocusedWCF onthemajortarget form.However,prioritizationofthemajortargetform abovetheincidentalerrorsseemsto beimportant.Furtherresearchonthesameissueisabsolutelynecessary.
The fourth research question was how the participants perceived the provision of focusedandunfocusedWCFinparagraphwritingtasks.First,asfarasthisresearchstudy wasconcerned,allparticipantsindicated theirdesireforteacherfeedback.Truscott(1999) stated thatlearnersdonotalwaysunderstand whatbenefitstheirlanguagelearning and thusobjected toFerris・sideaofoffering WCF forthereason thatlearnersexpresstheir desireforit.However,whenallstudentsinaclassprefertoreceivefeedbackandthereis noclearevidenceofaharmfuleffect,theirrequestorpreferencemightbeaccommodatedin ordertoenhancetheirmotivationforfurtherstudying.
In termsofthetypeofWCF,theparticipantsindicated theirstrong preferencefor correction ofgrammaticalerrors,whereasthey alsoappreciatedfeedback on variousother aspectsoftheirwriting.Allparticipantsin thisprojectanswered thatthey wanted the teachertocorrecttheirgrammaticalerrors.ThisisinaccordwithFerris・sreport(1995)that themajorityoftheESLstudentspreferredtoreceivegrammarfeedbackandexpressedtheir satisfaction aboutthefactthatthey could writemoreaccurately and presentideasmore clearly.Ontheotherhand,thestudentsinFerris・sstudyrememberedtheteacher・spositive commentsontheirideasandorganizationmorespecificallyandvividly.Consequently,itis safe to assume that grammar feedback and content feedback have different functions,
insteadofregardingthem asoppositeinstructionalplanstotradeoffwitheachother. It is important to note that learners・ ESL/EFL background and their English proficiencies have effects on the degree to which they preferred grammar or content feedback.HedgcockandLefkowitz(1994,1996)reportedthatforeignlanguagestudentspaid moreattentiontoform,whereasESL students,moreproficientinthetargetlanguage,had astrongerinterestin contentfeedback although they alsopaid attention toform-focused feedback.Likewise,Montgomery and Bakerreported (2007)thatstudentsenrolled in ESL writing coursesweremorelikely to prefergrammarfeedback than studentsworking on discipline-based papers, the latter being more interested in content and organization feedback.Thatis,themoreadvanced and experienced L2writersare,themorethey are concernedaboutcontentandorganization.ThelowerproficiencylearnerstendtowriteinL2 forthepurposeoflanguagepracticeandneedmoregrammarfeedback.Theparticipantsin thisstudywerecategorizedasadvancedbytheJapaneseEFL standards,butmanyofthem stillobserved in the classroom that their grammar was not perfect or that they had difficultiesunderstanding certain grammaticalstructures.Consequently,theirlack ofsel f-confidence,and/ortheirawarenessthatthey werewriting English paragraphsandessays forthepurposeofEnglishpractice,mighthaveinfluencedtheirresponses.
Then,theoverwhelmingmajorityindicatedtheirpreferencetohaveallerrorscorrected by theteacher,and theoverwhelming majority preferred directcorrection.Ferris(2004) statedthatindirectfeedback,which involveslearnersin cognitiveproblem-solving,ismore desirablethan directcorrection.Qiand Lapkin (2001)also proposed thatprovision ofa reformulatedform (i.e.,anotherform ofindirectcorrection)hasan advantageofinducing learners・effortsto noticetheirown errorsand engagein cognitiveprocessing,which,in turn,islikely to resultin long-term acquisition.Thequestionnairesection ofFerrisand Roberts・study(2001),whichevaluatedthelevelofexplicitnessrequiredfortheWCF toL2 learners,showed thatthemostpopularerrorfeedback techniquefortheparticipantswas marking andlabeling errorswith codes.Thefindingsfrom thepresentstudy contradicted thesereportsandtheories.Ontheotherhand,itmustalsobenotedthatChandler・sstudy (2003)indicatedthatdirecterrorcorrectionwasmoreeffectiveforimprovingESL learners・ accuracy in subsequentwritingsand thatVan Beuningen etal.(2012)demonstrated that directcorrection had a positiveeffecton L2 Dutch writers・abilitiesto writeand draft accurately.Furtherresearchisneededtoprovidemoresolidevidencefororagainsttheuse ofdirecterrorcorrection.However,oneinterpretationisthattheparticipantsinthepresent project,many ofwhom were not fully confident about their grammar knowledge and dependentonteacherfeedback,mighthavechosenthequickerandsurerwayofrecognizing theirerrors.
Conclusion
ThepresentstudyexploredawaytointegratebothfocusedandunfocusedWCFintoan EFL writing course.Theresultsshowed thattheparticipants・abilitiesto usethetarget grammaticalrulesinfreewriting,aswellastheirexplicitknowledgeofEnglishgrammar, improved significantly after receiving grammar feedback on their weekly paragraphs or essays.TheunfocusedWCF onavarietyofgrammaticalformsdidnotseem toundermine theeffectsoffocusedWCFforthemaintargetform,either.Intermsoftheirpsychological preference,all of the participants indicated their desire to receive feedback on their grammaticalerrors,andtheyperceiveddirecterrorcorrectiontobethemostuseful.
Theeducationalimplication isthatEFL teachersmay beadvised to providefocused WCF and alsotoprovideunfocused WCF on multiplegrammaticalpointsaslong asthe majortargetform isprioritized and thenumberofauxiliary targetpointsislimited to whatthestudentscan dealwith.Thetypesoffeedback (i.e.,directorindirect,explicitor not explicit) should be carefully considered depending on the students・ L2 learning backgroundsandproficiencylevels,thenatureandpurposeofinstruction(e.g.,composition, grammarandwriting,generalEFL course),andthedegreeoflearners・willingnesstospare timeand effortforwriting tasks.Directcorrection wasappreciated by themajority of participantsinthisprojectbecauseithelpedthem identifyandcorrecttheirerrorsswiftly andaccurately.However,itmustberememberedthatindirectfeedbackcanengagelearners in cognitiveprocessing,which islikely to resultin long-term acquisition ofgrammatical forms.Thisissuehasyetto beverified through moreextensiveand moremeticulously designedresearchstudiesandcontinuousfirst-handclassroom observation.
Thepresentstudy hadvariouslimitations.Although thestudy producedevidencethat unfocusedWCF on avariety ofgrammaticalformsdidnotunderminetheeffectivenessof focusedWCFinaparticularteachingcontext,theabsenceofnegativeeffectsdoesnotmean thatthecombinationoffocusedandunfocusedCFwasmoreeffectiveforL2learners・overall writing performance or language acquisition.In order to find more solid evidence that unfocusedWCF perseisusefulandeffective,afour-way comparison between focusedCF, unfocusedCF,focusedandunfocusedCF,andcontrolconditionneedstobeconducted.
Itisacknowledged thatan action research study with a smallgroup of29students cannot provide any strong evidence for generalization.Replication studies with larger samples,drawnfrom severaldifferentcoursesatdifferentuniversities,needtobeconducted. Thegrammar testsand thecriteria for theevaluation ofessaysshould berevised and polishedforgreaterreliabilitybasedonrepeatedresearchexperiences.
Formoreaccurateassessmentofessay writing,itispreferableto haveparticipants writetheweekly essaysin classsothattheirability towriteonlinecan beevaluated.In thisstudy,theclassmanagementconstraintsforced theresearcherto makeparticipants
writeandtypethescriptsathomeandsubmitthem atthenextclasssession.However,one practicalsolutionmaybetoteachinacomputer-assistedroom sothatstudentscantypeup theirparagraphsandprintandsubmitthem totheteacherattheendofthesamesession.
TheeffectsofdifferenttypesofWCF(e.g.,directcorrection,useofcodes,descriptionof errortypes,andprovision ofreformulatedphrasesandsentences)mustbeevaluatedmore meticulously.The presentstudy・s results suggested thatthe participants were noteven clearlyawareofthefunctionsandcharacteristicsofeachtypeofWCF.Anin-depthinquiry, usingunstructuredinterviewsandnarrativeanalysis,mightbeusefulfortheassessmentof EFL learners・perceptionsofdifferentformsofgrammarfeedback ortheirresponsesto teacherfeedback.
Despitetheselimitations,however,thepresentstudy hasbeen asmallbutimportant stepformoreconcreteandaccurateresearchonmultipleissues:theeffectivenessoffocused andunfocusedWCF,thefunctionsofdirectandindirectfeedback,andlearners・perceptions ofteacherfeedback.Anotherimportantissue,whichthisstudydidnotcover,isthesequence in which contentand grammarfeedback isprovided (seeAshwell(2000)fordetails).The common practice ofproviding content-based feedback on the first draft and grammar-focusedfeedback on thelaterdraftsmay bereevaluated.Itisnecessary todealwith each issueseparatelyinspecificclassroom contextswithclearlydefinedstudentgroups.
References
Ashwell,T.(2000).Patternsofteacherresponseto studentwriting in a multiple-draftcomposition classroom:Is contentfeedback followed by form feedback the bestmethod? JournalofSecond LanguageWriting,9(3),227257.
Bitchener,J.(2008).Evidencein supportofwritten correctivefeedback.JournalofSecond Language Writing,17(2),102118.
Bitchener,J.,& Ferris,D.(2012).Written correctivefeedback in second languageacquisition and writing.London:Routledge.
Bitchener, J., & Knoch, U. (2008). The value of written corrective feedback for migrant and internationalstudents.LanguageTeachingResearch,12(3),409431.
Bitchener,J.,& Knoch,U.(2009a).Thevalueofafocused approach towritten correctivefeedback. ELT Journal,63(3),204211.
Bitchener,J.,& Knoch,U.(2009b).The contribution of written corrective feedback to language development:A tenmonthinvestigation.AppliedLinguistics,31(2),193214.
Bitchener,J.,& Knoch,U.(2010).Raising thelinguisticaccuracy levelofadvancedL2writerswith writtencorrectivefeedback.JournalofSecondLanguageWriting,19(4),207217.
Bitchener,J.,Young,S.,& Cameron,D.(2005).Theeffectofdifferenttypesofcorrectivefeedbackon ESL studentwriting.JournalofSecondLanguageWriting,14(3),191205.
Bond,T.G.,& Fox,C.M.(2007).ApplyingtheRaschmodel:Fundamentalmeasurementinthehuman sciences.Mahwah,NJ:LawrenceErlbaum.
andfluencyofL2studentwriting.JournalofSecondLanguageWriting,12(3),267296.
Ellis,R.,Sheen,Y.,Murakami,M.,& Takashima,H.(2008).Theeffectsoffocused and unfocused writtencorrectivefeedbackinanEnglishasaforeignlanguagecontext.System,36,353371. Eskey,D.E.(1983).Meanwhile,back in therealworld...Accuracy and fluency in second language
teaching.TESOL Quarterly,17(2),315323.
Ferris,D.R.(1995).Studentreactionsto teacherresponsein multiple-draftcomposition classrooms. TESOL Quarterly,29(1),3353.
Ferris,D.R.(1999).Thecaseforgrammarcorrection in L2writing classes:A responsetoTruscott (1996).JournalofSecondLanguageWriting,8(1),111.
Ferris,D.R.(2004).The・grammarcorrection・debateinL2writing:Wherearewe,andwheredowe gofrom here?(andwhatdowedointhemeantime...?・)JournalofSecondLanguageWriting,13(1), 4962.
Ferris,D.R.(2010).SecondlanguagewritingresearchandwrittencorrectivefeedbackinSLA.Studies inSecondLanguageAcquisition,32(2),181201.
Ferris,D.R.,& Roberts,B.(2001).ErrorfeedbackinL2writingclasses:How explicitdoesitneedto be?JournalofSecondLanguageWriting,10(3),161184.
Hedgcock,J.,& Lefkowitz,N.(1994).Feedback on feedback:Assessing learnerreceptivity toteacher responseinL2composing.JournalofSecondLanguageWriting,3(2),141163.
Hedgcock,J.,& Lefkowitz,N.(1996).Someinputoninput:Twoanalysesofstudentresponsetoexpert feedbackinL2writing.TheModernLanguageJournal,80(3),287308.
Hyland,F.(2011).Thelanguagelearningpotentialofform-focusedfeedbackonwriting:Students・and teachers・perceptions.InR.M.Manchon(Ed.),Learning-to-writeandwriting-to-learninanadditional language(pp.159179).Amsterdam:JohnBenjamins.
Kepner,C.G.(1991).Anexperimentintherelationshipoftypesofwrittenfeedbacktothedevelopment ofsecond-languagewritingskills.TheModernLanguageJournal,75(3),305313.
Lalande,J.F.(1982).Reducingcompositionerrors:Anexperiment.TheModernLanguageJournal,66 (2),140149.
Montgomery,J.L.,& Baker,W.(2007).Teacher-written feedback:Studentperceptions,teachersel f-assessment,andactualteacherperformance.JournalofSecondLanguageWriting,16(2),8299. Qi,D.S.,& Lapkin,S.(2001).Exploringtheroleofnoticinginathree-stagesecondlanguagewriting
task.JournalofSecondLanguageWriting,10(4),277303.
Robb,T.,Ross,S.,& Shortreed,I.(1986).SalienceoffeedbackonerroranditseffectonEFL writing quality.TESOL Quarterly,20(1),8395.
Sachs,R.,& Polio,C.(2007).Learners・usesoftwo typesofwritten feedback on a［n］L2writing revisiontask.StudiesinSecondLanguageAcquisition,29(1),67100.
Semke,H.(1984).Effectsoftheredpen.ForeignLanguageAnnals,17(3),195202.
Sheen,Y.(2007).Theeffectoffocused written correctivefeedback and languageaptitudeon ESL learners・acquisitionofarticles.TESOL Quarterly,41(2),255283.
Sheen,Y.,Wright,D.,& Moldawa,A.(2009).Differenteffects offocused and unfocused written correctionontheaccurateuseofgrammaticalformsbyadultESL learners.System,37,556569. Sheppard,K.(1992).Twofeedbacktypes:Dotheymakeadifference?RELC Journal,23(1),103110. Swain,M.(1985).Communicativecompetence:Somerolesofcomprehensibleinputandcomprehensible
(pp.235253).Rowley,MA:NewburyHouse.
Swain,M.,& Lapkin,S.(1995).Problemsinoutputandthecognitiveprocessestheygenerate:A step towardssecondlanguagelearning.AppliedLinguistics,16(3),371391.
Truscott,J.(1996).ThecaseagainstgrammarcorrectioninL2writingclasses.LanguageLearning,46 (2),327369.
Truscott,J.(1999).Thecasefor ・thecaseagainstgrammar correction in L2 writing classes・:A responsetoFerris.JournalofSecondLanguageWriting,8(2),111122.
Truscott,J.(2007).Theeffectoferrorcorrection on learners・ability towriteaccurately.Journalof SecondLanguageWriting,16(4),255272.
Truscott,J.,& Hsu,A.Y.(2008).Errorcorrection,revision,andlearning.JournalofSecondLanguage Writing,17(4),292305.
VanBeuningen,C.G.,DeJong,N.H.,& Kuiken,F.(2012).Evidenceontheeffectivenessofcomprehensive errorcorrectioninsecondlanguagewriting.LanguageLearning,62(1),141.
