Sheen (2007)

Embed Size (px)

Citation preview

  • 8/14/2019 Sheen (2007)

    1/30

    Teachers of English to Speakers of Other Languages, Inc. (TESOL)

    The Effect of Focused Written Corrective Feedback and Language Aptitude on ESL Learners'Acquisition of ArticlesAuthor(s): Younghee SheenSource: TESOL Quarterly, Vol. 41, No. 2 (Jun., 2007), pp. 255-283Published by: Teachers of English to Speakers of Other Languages, Inc. (TESOL)Stable URL: http://www.jstor.org/stable/40264353 .Accessed: 10/11/2013 12:40

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

    .JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

    .

    Teachers of English to Speakers of Other Languages, Inc. (TESOL) is collaborating with JSTOR to digitize,preserve and extend access to TESOL Quarterly.

    http://www.jstor.org

    This content downloaded from 3 5.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/action/showPublisher?publisherCode=tesolhttp://www.jstor.org/stable/40264353?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/40264353?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=tesol
  • 8/14/2019 Sheen (2007)

    2/30

  • 8/14/2019 Sheen (2007)

    3/30

    empirical tudiesdemonstrating hat CF can lead to acquisition e.g.,Chandler, 004; Ferris, 999, 2004).The study eported n this rticle

    provides urther vidence efuting ruscott's osition.I beginby xamining owcorrective eedback as been addressed nthe SLAand second anguage L2) writing iteratures, hen discussnumber f methodologicalssues elating o research n written F, therelative ffectiveness f different ypes f CF and the role played bylanguage ptitude n mediating he ffects f CF. then eport heresultsof quasi-experimentaltudy. conclude y onsidering he mplicationsof the results f the tudy or oth SLAtheory nd L2writing edagogy.

    CORRECTIVE FEEDBACK: SLA VS.L2WRITING RESEARCH

    SLA research nto CF has been primarily oncernedwith ral CFinrelation o theoretical laims bout the role of input nd interaction(e.g.,Gass,1997;Mackey, 007)and focus n form Long,1996).xThereis now growing vidence that oral CF, as a focus-on-form echnique,facilitates nterlanguage evelopment, lthough here s less consensusabout the effects f different ypes f oral CF (e.g.,explicit ersus m-

    plicit, nput-providing ersus rompts; ee Ellis, 006)L2writing esearchers an benefit rom xamining hemethodologyof oral CF research n SLA. For example,written F studies e.g.,Polio,Fleck,& Leder, 1998;Robb, Ross, & Shortreed, 986) have typicallycompared utcomes n terms f overall mprovement cross number fdifferent rammatical tructures, hereas tudies f oralCFin SLA(e.g.,Han, 2002;Lyster, 004)havefocused n a specific rammatical eatureand measured cquisition n terms f that feature. he SLA researchsuggests hat ntensive F that epeatedly argets single inguistic ea-ture can have a beneficial ffect n interlanguage evelopment e.g.,Doughty& Varela, 1998;Han, 2002;Iwashita, 003; Long, nagaki, &Ortega, 998;Mackey Philp, 1998).

    There are, however, ome obvious differences etweenwritten ndoral CF. Written F is delayedwhereas ral CF occurs mmediately fteran error asbeen committed. ritten F imposes esscognitive oad onmemory hanoral CF,which ypically emands n immediate ognitivecomparison, husrequiring earners o heavily ely n their hort-termmemory. Written F is also different edagogically.Writing eachersoften ry o improve ontent nd organization hilefocusing n the

    1Focusonform Long, 1996) constitutes type of pedagogical intervention. t refers oattempts o draw earners' ttention o form n the context f communication. ne way nwhich this occurs s reactively hrough CF directed t learners' inguistic rrors.

    256 TESOLQUARTERLY

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    4/30

    overall uality f students' writing, n which case accuracy s oftensecondary ssue. On the other hand, a teacher's provision f oral CF

    typically raws earners' ttention o their rroneous tterances s theyarise n communicative ctivities. hese differences ay xplainwhy, ncontrast o the SLA research hat n general has shown hat ral CF iseffective, 2writing esearchers avenot been ableto convincingly em-onstrate hatwritten F leads to improvement n grammatical ccuracyin newpiecesofwriting. owever, nother eason hat he results f SLAresearch nto CF contrast ith heresults f L2writing esearch nto CFmay ie in the methodologies sed bywriting esearchers.

    Methodological ssues in Written F ResearchA key ssue concerning he efficacy f written F was identified y

    Ferris 2004): "Does written F help students o improve n writtenaccuracy ver time?" p. 56). At present, hisquestion has not beendefinitivelynswered ecausepast research asfailed o provide nam-biguous vidence f the ffectiveness fwritten F (for comprehensiveand influential iscussion f this ssue, eeFerris, 004).The reasons orthis re primarily ethodological.

    Manywritten Fstudies, or xample, acked control roup hat idnot receive F (e.g.,Kepner, 991;Robb et al., 1986). n studies hat idhavea control r comparison roup, hegroup differences ere n factnot statistically ignificant e.g., Lalande, 1982;Semke,1984).Also, hecontrol r comparison roup n the written Fstudies ypicallyeceivedsome kind of feedback e.g., comments n content nd/or organiza-tion) n addition o CFdirected t inguistic ccuracy e.g.,Fazio,2001),whichmakes he effects f CF difficult o isolate. Researchers doptedthis procedure or thical easons.When using ntactwriting lasses nthe study, esearchers elt hey ad to provide ome kind of correctionbecause tudents ada

    righto t and

    probably xpectedt.The absence

    of a true ontrol roup,however, onstitutes major imitation n thesestudies.

    The studies lsovaried n howtheymeasured heeffectivenessf CF.Somemeasured mprovementn terms fwhether earners ncorporatedthe corrections n a revision f their irst raft e.g.,Fathman Whalley,1990;Ferris Roberts, 001);others measured mprovement n home-work ssay ssignments r ournal entries ver a long period of time(e.g.,Chandler, 003;Kepner, 991).A few tudiesmeasured mprove-ment n writing n terms f gains n linguistic ccuracy s well s fluency(Ashwell,000;Robbet al., 1986).Thisvariation n how he ffectivenessof CF wasmeasuredmakes eaching ny definite onclusion ery ifficult.

    Other tudies avemeasured mprovement nlybyexamining earn-

    FOCUSEDWRITTENCORRECTIVEFEEDBACKAND LANGUAGEAPTITUDE 257

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    5/30

    er's revised exts. or example, n an examination f ntermediate SLcollege tudents' riting, athman nd Whalley 1990)found hat eed-

    backon form nd feedback n contentwere qually ffectiven produc-ing mprovement n the students' evisions. shwell 2000)also foundthat providing dult earnerswith grammar orrection mproved heirgrammatical ccuracy n written ompositions. ne could argue, how-ever, s Truscott 1999)did, that mprovement n revisions lone doesnot constitute vidence hat earning as occurred. o claim that rrorcorrection esults n learning, ne must xamine whether he mprove-ment n revisions arries ver o a new pieceofwriting r f he mprove-ment s manifested n posttest r delayedposttestmeasures.

    In sum, hemethodological imitations f past tudies n CF and theinconclusivenessf their indings averesulted n a somewhat onfusingpicture, eadingCFcritics oargue against he effectiveness f CF andeven to claim that CF can be harmful Krashen, 982;Truscott, 996,1999,2004).Criticismsn this ein nclude practical oncerns bout thequality f the CF that eachers rovide nd also the danger f raising hestudents' ffectiveilter i.e., provoking nxiety n the earner which n-hibits earning).However, erris 1999,2004)points ut that most tu-dents do want heirwriting rrors o be corrected nd that t s the obof L2writing eachers o attend o their needs.

    RelativeEfficacy f Different ypes of Written F

    Although he absolute efficacy f written F remains ontroversial,mostwritten F research ascompareddifferent ypes f written eed-back to nvestigate hether ertain ypes f CF produce more positiveeffect han thers e.g.,Fazio,2001;Lalande,1982;Robbet al., 1986). none of the earliest tudies, obb et al. examined he writing f 134 EFLstudents nJapan ver ne academicyear o see iffour ifferent Ftypes

    produceddifferential ffects n the

    mprovementn theirwritten

    ssays.The four methods were (a) direct orrection,ndicating he errors ndproviding he orrect orm; b) codedeedback,ndicating he ype f errorbased on an abbreviated ode system; c) uncodedeedback,ndicatingerrors n the textwith yellow ighlighting en without pecifying heirtype; nd (c) marginal, ndicating he total number of errors n themargins f the tudent's rticle. n alltreatments, evisions ere equiredand the instructor hecked for accuracy.Robb et al. found no majordifferences cross he four reatment ypes.

    Ferris 2002)contends hat ndirectrror orrectioni.e., dentifying r-

    rors without roviding hecorrect orm) s more beneficial handirectcorrection ecause it pushes earners o engage n hypothesis esting,thereby nducing eeper internal rocessing, hichhelps them nter-

    258 TESOLQUARTERLY

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    6/30

    nalize the correct orms. owever, mpirical vidence o date suggeststhat there s no advantage or ndirect F over direct CF (Chandler,

    2003;Lalande,1982;Robb et al., 1986).In fact, Chandler found thatdirect orrection assuperior o other types f indirect orrection nproducingmore accurate writing. handler hypothesized hat teach-er's direct orrection elpsESL students nternalize hecorrect orm na more productive aybecause indirect eedback, hough t demandsgreater ognitive rocessing, elays onfirmation f students' ypoth-eses.She also reported hather ESL students avored irect orrection.These findings uggests hat ontrary opedagogical uggestions n theL2writing iterature e.g., Ferris, 002;Ferris Hedgcock, 005), ndi-rectwritten Fmaynot be superior o direct F. This conclusion s alsosupported y SLAresearch n oral feedback,whichprovides videncethat xplicit eedback acilitatescquisition etter han mplicit eedback(Carroll&Swain, 993; Ellis,Loewen & Erlam, 006).

    However, heresearch n written nd oral CF to date has not xploredthe possibility hat the effectiveness f different ypes f CF will varydepending n the ndividual earner. now turn o consider ne majorindividual ifference actor hatmaymediate he effect f CF.

    LanguageAptitudeAptitudes a complex onstruct, istinct rom eneral ntelligence nd

    achievement. arroll 1981)defines anguageptitude s having he fol-lowing our omponents: a) phonemic oding bility, b) grammaticalsensitivity,c) inductive anguage earning bility, nd (d) rote earningability. kehan 1998)combines b) and (c) in a component e labelslanguage nalytic bility. his study s concerned nly with his ompo-nent, efined s the bility oanalyze anguage y reating nd applyingrules o new entences Sawyer Ranta, 001).It s reasonable o think

    that earnerswith high anguage analytic bility illbe better ble to

    engage n the kind of cognitive omparison hat s required f CFis toresult n learning. t can be further rgued that direct metalinguisticcorrection illbenefit nalyticallytrong anguage earners o a greaterextent han analytically eak earners ecause the former will find teasierto use the metalinguisticnformation.

    To date, only wo tudieshave explored he role of aptitude n theefficacy f CF. DeKeyser1993)found hat earners withhigh previousachievement, igh language aptitude, nd low anxiety enefited hemost from rror orrection. imilarly, avranek nd Cesnik 2001)re-

    ported hat orrective eedback s likely o benefit earnerswho haveapositive ttitude oward rror orrection nd high anguage bility. ow-ever, hese wo tudies id not solate different ypes f CF. Nor did they

    FOCUSEDWRITTENCORRECTIVEFEEDBACK ANDLANGUAGEAPTITUDE 259

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    7/30

    investigate he nterrelated ffects f anguage ptitude nd CF on spe-cific arget inguistic tructures. he complexnature f CF(e.g., mplicitor explicit, nput providing r input prompting) aises the questionwhether ifferent spects of CF are mediated differently y anguageanalytic bility.Moreover, heprevious tudies nly xamined anguageaptitude s a whole.The current tudy, herefore, hose anguage na-lytic bility s the component o investigate ecause it seemed to bepotentially he most ikely o interact with wodirect ypes f CF indetermining earning utcomes n the context f a written ask.

    The relationship etween anguage aptitude nd learning utcomehas often een discussed n terms f the cognitive oad imposed n thelearner e.g., DeKeyser, 993;Snow,1987).In the two ypes f CF com-pared n the current tudy, hedifference etween hedirect orrectionand direct metalinguistic orrection as whether he earners eceivedmetalinguistic omments. chmidt 2001) distinguishes oticing rommetalinguisticwarenessnd argues hatmetalinguistic wareness equiresa deeper evel of learning. hus, t maybe hypothesized hatwhereaswritten irect feedback ncreases noticing, irect metalinguistic eed-back increases not only noticing but also encourages awareness-as-understandingi.e., a deeper evelof cognitive rocessing).

    Research QuestionsThis study onstitutes n attempt o address ome of the perceived

    problems n written F research ydrawing n the theory nd method-ology f oral CFresearch n SLA and by cknowledging hat ndividualdifference actors uch s language nalytic bilitymaymediate he ffectof CF. The study onsiders hefollowing esearch uestions:1. Does focusedwritten orrective eedback avean effect n interme-

    diate ESLlearners' cquisition f English rticles?

    2. Is there ny difference n the effect f direct orrection ith ndwithoutmetalinguistic eedback n ESL learners' cquisition f En-glish rticles?

    3. To what xtent oes the earners' anguage nalytic bility mediatethe effectiveness f CF?

    METHOD

    DesignThisstudy sed a quasi-experimental esearch esignwith pretest-

    treatment-posttest-delayed osttest tructure, sing ntact ESL class-

    260 TESOLQUARTERLY

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    8/30

    rooms.The research lso involved correlational nalysis f the rela-tionship etween ne individual earner ariable i.e., anguage nalyticability) nd criterion est cores n the posttests nd delayedposttestswith view oexamining hemoderating ffect f anguage ptitude nthe effect f the error orrection reatment. woweeks rior o the tartof the corrective eedback reatments, heparticipating tudents ooklanguage nalytic bility est. n the following eek, hey ompleted hepretests. he immediate osttests ere ompleted ollowinghe two CFsessions nd the delayed osttests to 4 weeks ater. During ach testingsession, hree ubtestswere administered: speeded dictation est,writing est, nd an error orrection est.

    SettingThe current tudy as onducted n six ntact lassrooms n theAmeri-

    can LanguageProgram ALP)of a community ollege n the UnitedStates. he ALP offers ntensive nd extensive nglish anguage oursesfor those whowish ospeakand write nglish ccurately nd fluently.Classes re offered ach semester, nd each class asts rom hour and20 minutes o 3 hours ach session.Teachershave either native r na-tivelike

    nglishanguage roficiency. he program as four evels: oun-

    dation nd Levels , II, and III (most dvanced evel)with n averageclass size of 15-22students. rior o the current tudy, he researchervisited he itemany imes nd observed nd piloted number f nstru-ments n several lasses t different evels. he Level I classes onsistingof ntermediate-levelearners hosenfor he tudy amefrom ll areas ofthe program, ot specifically riting lasses.

    Participants

    The participants erefive ative-English-speakingmerican eachersand their 11 ntermediate-level tudents. he students eredrawn romboth nternational nd mmigrant SLpopulations nd represented ari-ous languageand ethnic backgrounds. he three major groups wereKorean,Hispanic, nd Polish.Their agesvariedfrom 1-56 years ndthey ame from different ducationalbackgrounds e.g., high schooldiploma o doctoral egree). Class izesranged rom 5-22students. nthe end, 91 students ompleted he anguage nalytic bility est, re-tests, osttests, nd delayedposttests; tudents ith ncomplete atasets

    were excluded from he sample. Out of six intact lassrooms, hreegroupswere ormed: hedirect-only orrection roup n=31 , the directmetalinguistic roup n = 32), and the control roup n = 28).

    FOCUSEDWRITTENCORRECTIVEFEEDBACKAND LANGUAGEAPTITUDE 261

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    9/30

    Operationalizations

    Direct-onlyorrectiononstitutes traditional rror orrection trategythat onsists f ndicating he ocation f an error n the student's extand the provision f the correct orm ydeleting nd/or replacing heerror r by dding linguistic lement.

    Directmetalinguisticorrections operationalized s indicating he oca-tion of an error, roviding hecorrect orm, nd includingmetalinguis-tic comments hat xplain he correct orm.

    Target Structure

    Articles ere hosen s the target tructure or he current tudy itha view o solating heeffect f error orrection rom nypotential ffectof grammar nstruction n general. his decision wasmade after seriesof discussionswith the participating acultymembers t the college,which evealed hat a) participating tudents re not explicitly aughtarticles uring he emester, nd (b) articles, hough onstituting struc-ture where tudents ommonlymakeerrors, re infrequently orrectedbecausethey re nonsalient nd they equire omplicated uleexplana-

    tions. The nd a belong to the top fivemost requently ccurring ordsin English ccording o the COBUILDcorpus Sinclair, 991).However,the SLA iterature as clearly hown hat nglish rticles re consideredto be a nonsalient eature. hey are not phonologically alient, ndmisuse of articles arely eads to communication reakdown Master,2002).Learners ave been observed o experience ifficultyn learningarticles ecause of their omplexnature; hat s,the choiceof an articleis determined yboth inguistic nd pragmatic actors Butler, 002;Liu&Gleason,2002).

    For this reason, care was taken n the current tudy o focus the

    correction n errors nvolvingust twomajor unctions f ndefinite nddefinite rticles: as first mention nd the s anaphoric eference, s inthe following xample:

    He hid a snake n his bag,but the nake scaped.

    Instruments nd Procedures

    The two experimental roups completed he treatments nd tests.

    The control group completed he tests nly. t did not perform henarrative asks nd did not receive ny feedback ut instead followednormal lasses.

    262 TESOLQUARTERLY

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    10/30

    Narrative Task nstruments

    There were two reatment essions. ach session nvolved narrativestimulus oelicit rticle rrors rom he earners. t shouldbe noted hatbecausethe ontext f this esearch as not n L2writing lass, twasnotconsiderednecessary o ensure that the narrative ask constituted necologicallyvalid writing ask. The first narrative ask nvolved nadaptedAesop'sfable, The Fox and the Crow."

    Therewasonce a crowwho stole pieceof cheese from kitchenwindow. he flew ffwith he heese o a nearby ree.A fox aw what hecrow addone, nd he walked ver o the ree. Oh,Mistress row, ouhave uch

    ovelylackfeathers, uch ittle eet, ucha beautiful

    ellowbeak, nd such fine lack yes oumust ave beautiful oice.Wouldyouplease ing orme?"Thecrow elt ery roud. he openedher beakand sangCAW-CAW-CAW-CAW.f course hecheesefell own, nd thefox te the pieceof cheese.

    To minimize he processing oadon the earners whenreproducing henarrative, he researcher eplaced ome difficult ords n the originalfable with asier words nd simplified few entence tructures. henarrative ontained even ndefinite rticles nd seven definite rticles.

    For the second task, he researcher onstructed n interesting etsimple tory itled The Pet Snake."

    Aboybought snake rom pet hop.He took he nakehome.Hismother creamed hen he aw he nake. he told him o take he nakeback o the pet shopbut the owner efused o take he nake ack.Theboy ut he nake n a box and eft t on a seat n thepark earhishouse.An old woman ound hebox.When he saw he nake he had a heartattack.

    This narrative ontained asyvocabulary ith even ndefinite nd tendefinite rticles. he ALPfaculty onsidered he twotasks uitable or

    their ntermediate-leveltudents, et xpected hat he students ouldoften makearticle rrors.

    Corrective eedback Treatment rocedures

    Each sessionwas udio-recorded sing clip-on microphone ttachedto the teacher. There was a one-week nterval etween he first ndsecondnarrative asks. he specific rocedures f the treatment essionwere s follows:

    1. First, he teacher handed out the story ith n emptywriting heetattached o t and told the tudents hat heyweregoing o read thestory nd then rewrite hestory.

    FOCUSEDWRITTENCORRECTIVEFEEDBACKANDLANGUAGEAPTITUDE 263

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    11/30

    2. Students ere sked to read the story ilently.3. The teacher xplained keywords nd discussed he moral of the

    story ith he class.4. The teacher hen ollected he tories y sking he tudents o tearoff he story art nd keep the writing heet only.

    5. Before sking he tudents o rewrite he tory, heteacher eadthestory loud once while he students oted downkeywords.

    6. Then the students ere sked to rewrite hestory s closely s theycould remember.

    7. The teacher ollected he students' written arratives hichwerethen handed to the researcher.

    8. The researcher orrected he narratives ocusingmainly n articleerrors ased on the correction uidelines.9. In the following lass typically or 4 days ater), he students ook

    part n a corrective eedback ession uringwhich hey eceived heirnarratives ith orrections. he students were asked to look overtheir rrors nd the corrections arefully or t east minutes. ow-ever, he eacher id not comment urther n their rrors r give nyadditional xplanation or were tudents sked to revise heir writ-ing.

    Correction uidelines

    Given he mount f time nd abour nvolved,twasdecidedthat heresearcher ould erve s the corrector. he researcher orrected ll thearticle rrors n the earners' narratives, hich ontained etween neand seven errors. o mask the focus of the study, he researcher or-rected few rrors ther han those nvolving rticles.

    This CF treatment sed two different ypes f written orrection. orthe direct-only orrection roup, ach correction onsisted f ndicatingthe error nd writing hecorrection bove t.For the directmetalinguis-tic orrection roup, ach error wasfirst ndicatedwith number. Notesfor ach numbered rror weregiven t the bottom f a learner's heet.The notes ndicated what waswrong using metalinguistic nformationand also provided he correct orm.

    Testing nstruments nd ScoringGuidelines

    Three tests were used to measure acquisition n this study: a) aspeeded dictation est, b) a narrative riting est, nd (c) an errorcorrection est. n the caseof a) and (c), the ame testwasusedfor he

    264 TESOLQUARTERLY

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    12/30

    pretest, osttest, nd delayedposttest, ut the order of the items wasrandomly hanged from ne testing ession o another. n the case of

    (b), the ame test wasused for hepretest nd posttest, ut the delayedposttest asbased on a different arrative. his changewas n responseto students' egative eactions n a preliminary tudy obeing asked torepeat he ame tory n the posttest. he results f pilot tudy nvolvinga different roup of students rom similar ntermediate-level lassshowed hat he narrative asks sed for he twodifferent osttests erecomparable i.e., a t test howed no statisticallyignificant ifference nthe cores btained rom he two asks)The content f the writing estswasdifferent rom he content f the CF treatment asks nd thus pro-vided measure f the extent o which heeffects f the CF transferredto new pieces of writing. tudents were also administered languageanalytic bility est.

    Speeded ictation est 8 Minutes)This test onsisted f 14 items, ach of which ontained ne or two

    sentences nvolving he use of indefinite nd definite rticles see Ex-amples and 2). Itwas ime-pressuredo imit earners' bility o draw ntheir xplicit rammatical nowledgeEllis, 005).Each item n the testhad one or two timuli nvolving rticle-obligatoryontexts seeAppen-dixA) The total number f rticle timuli n the est was9 indefinite nd12definite rticles. xample1measures earners' eceptive nd produc-tiveknowledge f the ndefinite rticle with he referential-firstentionarticle unction "a very ind doctor" s the stimulus n the tem). Ex-ample 2 measures nowledge f the ndefinite rticle and the definitearticle the with the referential-secondentionrticle function the twostimuli n the tem re "a movie" nd "the movie").

    Example : There was very inddoctor n myhometown.Example : I saw movie astnight. he moviemademe sad.

    In administering his est, ach student wasprovidedwith smallnote-book. The researcher irst xplained the procedures o the students.Then the teacher eadtwo ample entences o that he students ouldfamiliarize hemselves ith he procedure. ach item was read at a nor-mal speedand students ere directed o write ownone item per pageas fast s they ould and exactly s they heard it. Once the studentsturned o the next page for the next tem, heywere not allowed toreturn o the previous age. Thisprocedure revented he tudents romconsciously eworking hat hey adwritten.

    Target-like se (TLU)scoreswerecalculated Pica, 1991). The TLUanalysiswas used to measure earners' knowledge f articles y takingoveruse f the target orm nto consideration. rticles ere first cored

    FOCUSEDWRITTEN CORRECTIVEFEEDBACK AND LANGUAGE APTITUDE 265

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    13/30

    for orrect se in obligatory ontexts. his score then became the nu-merator f a ratio whose denominator asthe sum of the number f

    obligatory ontexts or rticles nd the number f nonobligatory on-texts n which rticles were supplied nappropriately. he scoring or-mula s shown n the following quation:

    n correct uppliance n contexts- - x 100= r ercent ccuracyn obligatory ontexts n suppliance rin nonobligatory ontexts

    Writing est (12 Minutes)

    This test wasadapted

    from ne of Muranoi's 2000)test nstrumentsfor English rticles. t consisted f four equential ictures; he tudentswere sked to write ne coherent tory ased on them. Word promptsnext to each picture were included to elicit noun phrases nvolvingarticle sage.For example,next to the first icture heword promptswere old man, aint, nd picture, hereby ncouraging he students oconstruct sentence uch as An old man ikes opaint picture.

    It should be noted that revised ersion f the writing estwasad-ministered or hedelayed osttest. are wastaken o make the revisedwriting askequivalent o the original n terms f difficulty nd the

    likelihood f eliciting oun phrases ontaining rticles. he results f apilot estwith 5ALPstudents ho completed wo ersions f the writingtest ndicated hat tudents ended o score higher n the original est,but the difference etween he two sets of scores was not significant,t(24) = 0.89,ns.

    Thewriting atawere oded usingTLU(Pica, 1991)scores.Using hesame TLU formula s was used for the dictation ests, tudents' coreswere calculated s percentages.However,whereas he dictation estsafforded learobligatory ontexts ased on the stimuli, he constrainedfree-production riting est id not. t required hefollowing dditional

    scoring uidelines:1. When t was not clear whether noun phrase NP) constituted n

    obligatory ontext or or Abased on the tudent's riting, heNPwas not coded. Only suppliance/nonsuppliance n unambiguouscontextswascoded (i.e., the contexts where he researchers oulddefinitely etermine hat or the as needed) Thismeant hat omepossible rrors ere gnored. xceptions re noted n the points hatfollow.

    2. In the case of the wordprompt ark, oth in the ark" r "in a park"werepossible, o NPscontaining hiswordwere xcluded from od-ing. However,when neither rticle waspresent n the NP, it wascoded as nonsuppliance.

    266 TESOLQUARTERLY

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    14/30

    3. In the ase of the word rompts oynd girl, hen he tudent rote,"Aboy nd girl," nly he first P (i.e.,"a boy")wascoded because

    itwasnot lear whether he tudent sed the lision ule orrectly. na similar ein,when the student sed "the boy and girl" s secondmention, nly he first P was coded. However, f the first P waserroneous, ither s first mention r second mention e.g., "boy nda girl" r "boy nd girl"), ach NP was ncluded n the coding.

    4. Codingexcluded any NP where determiner nd an articlewerecopresent, s in the following xample: Aboy nd girl re ook at thehispicture"

    5. Articles n idiomatic hrases for example, all of a sudden, fewminutes, t the momentwere lso excluded from oding.

    Error Correction est 15Minutes)

    Thistest onsisted f 17 tems seeAppendix ). Each tem ontainedtworelated tatements, ne of whichwasunderlined nd contained nerror, hat he earners were skedto correct n writing. he itemswereadapted from est nstruments sed in Liu and Gleason (2002) andMuranoi 2000).Four distracter temswere ncluded, nvolving he useof past tense, modal choice,and subject-verb greement. he two x-

    amples re taken rom hetest, ollowed ythe correct nswers.Example : I saw n interesting ovie ast night. forgot he name ofmovie.Answer: forgot henameof the movie.

    Example : Isyour ncle ar salesman? 'm looking obuy carAnswer:syour ncle car salesman?

    The error orrection estwasscored n a discrete tembasis.One pointwas

    givenor ach correct

    uppliancef an article n the 14obligatory

    contexts n the underlined entences n the test. xcluding hedistract-ers, 4pointswasthe perfect corefor he test, nd students' inal coreswere alculated s percentages.

    LanguageAnalytic bility est 20Minutes)

    The instrument as based on a language nalysis estdevelopedbyOtto and used previously y Schmitt, rnyei, Adolphs, nd Durow(2003) The test onsisted f 14multiple hoice tems. he learners ere

    given a glossary onsisting f words nd sentences rom n artificiallanguage nd their nglish ranslations see Figure1). Theywere thengiven 4English entences nd for ach sentence were skedto choose

    FOCUSEDWRITTENCORRECTIVEFEEDBACKAND LANGUAGEAPTITUDE 267

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    15/30

    FIGURE 1Language Analysis est Examples

    the correct ranslation rom he four choices provided. o make thecorrect hoice, he earners eeded to analyze rammatical arkers up-plied n the glossary nd apply hese o the multiple-choice ranslations.Figure 1 contains n examplequestion.To choose the correct ransla-tion n Example1, the student ad to first educe the rule that i" s apast progressive) arker nd "o" s a present progressive) arker ndthen apply that rule to the translated hoices. Becausethis test waschallenging o the students, he researcher uided them through he

    first tem n the test. This guidancereduced their nxiety nd ensuredthat heywere familiar ith he procedure hey eeded to follow. helanguage nalysis estwasscoredon a discrete tem basiswith 4pointsbeing the perfect core. Students' inal cores were calculated s per-centages.

    Test ReliabilityIn the dictation nd writing ests, secondresearcher oded a sample

    of25%

    of the total data. Thesample

    cameequally

    from hepretests,posttests, nd delayedposttests. n the dictation est, he percentage

    agreement coreswere89.3%, 87.2%,91.4%,respectively.n the writingtest, he percentage greement coreswere 78.4%,83.3%, 79.2%,re-spectively. lthough igher evelsof reliability or the writing est redesirable, he evels chieved ll exceeded75%,which s satisfactory.tshould also be noted that he complexity f the English rticle ystemmakes high evel of consistency ifficult o achieve.As for the errorcorrection nd anguage nalysis ests, nternal onsistency eliability asestimated singCronbach's lpha. The reliability oefficient or he 14

    items n the error orrection retest asa = 0.84 (M= 5.17,SD= 3.59,n = 111 , and for he anguage nalysis est, =0.92(M=51 0,SD=20.0,n= 111).

    268 TESOLQUARTERLY

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    16/30

    TABLE 1Group Means and Standard Deviationsfor Total Test Scores

    Pretest Posttest PosttestCorrection ype M SD M SD M SD

    Direct-only group (n= 31) 44.1 11.9 58.3 15.3 57.5 14.4Direct metalingustic group (n= 32) 49.6 16.9 65.4 16.3 69.4 15.3Control group (n= 28) 48.3 14.2 52.1 15.6 51.2 16.2

    Analysis

    All scores wereentered nto SPSS (2002)and a range of descriptiveand inferential tatistics ere computed. The following nalyseswereused to answer he three research uestions. irst, escriptive tatisticsfor the anguage nalysis est, peeded dictation est,writing est, nderror correction est were computed. Then one-wayANOVAs withposthoc multiple omparison estsusing Tukey, repeated measuresANOVAs,nd ANCOVAswereperformed, ollowed yPearsonproductmoment orrelation.

    RESULTS

    Tables1 through present hedescriptive tatistics or otal cores orthe tests aken ogether s well as for the scoreson the dictation est,writing est, nd error orrection est eparately ver the three estingperiods: retest, osttest (i.e., mmediate osttest) nd Posttest (i.e.,delayedposttest). one-way NOVA howedno statistically ignificantgroupdifferencesn the pretest otal cores mong he hree roups, (2,88) = 1.23,ns.

    Figure shows ach group's mean speeded dictation est coresforthe three esting eriods. n the peededdictation est, ll three roupsproducedgains t Posttest .However, hegraph ndicates hat nly he

    TABLE2Group Means and Standard Deviationsfor the Speeded DictationTest

    Pretest Posttest Posttest

    Correction ype M SD M SD M SD

    Direct-only roup (n= 31) 50.8 15.7 67.6 14.1 65.2 15.6Direct metalinguistic roup (n= 32) 54.6 18.6 70.7 17.2 72.6 15.6Control group {n= 28) 58.1 12.9 64.1 13.9 62.4 13.7

    FOCUSEDWRITTENCORRECTIVEFEEDBACK ANDLANGUAGE APTITUDE 269

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    17/30

    TABLE 3Group Means and Standard Deviationsfor the Writing est

    Pretest Posttest PosttestCorrection ype M SD M SD M SD

    Direct-only roup (w= 31) 53.1 19.8 66.3 21.4 64.9 19.8Direct metalinguistic roup (w= 32) 57.6 21.0 73.9 19.3 78.8 18.4Control group (n = 28) 57.3 21.4 63.9 24.4 63.9 22.8

    direct metalinguistic henceforth irect eta) roup shows consistentincrease ver time whereas he control roup and the direct-only or-rection

    henceforth irect-only)rouphow

    slightecreasefrom ost-

    test 1 to Posttest . One-way NOVAswith multiple omparisons e-vealed that whereas he group differences n Posttest were not signifi-cant, he differences etween he direct meta group nd control roupin Posttest were ignificant, (2, 88) = 3.74,p < 0.05

    Figure 3 shows he mean scores for the writing est ver time. Thethree roups adconsiderably ifferent cores, specially hedirectmetagroup and the control group, who started t almost he same pretestpoint. At Posttest , the mean score of the direct meta group s consid-erably igher han hat f the control roup.The figure lso shows hat

    the direct-only rouphad the ame pattern s in the dictation est n thatthe direct-only roup barely ustained ts gains ndicated n Posttestinto Posttest , whereas he direct meta group continued o gain inPosttest . The posthoc omparisons evealed significant ifference nPosttest among the three groups, (2, 88) = 5.12,p < 0.01, ndicatingthat he meta group outscored oth the direct-only nd control roup.

    Figure shows he results or heerror orrection est. he pattern fthe three groups n this graph reveals hat he two treatment roups'gains over time were substantial ut the control roup showedno im-provement. oreover, hegains n the metagroup ppear to be greaterthan those of the direct-only roup. Also the difference etween hedirectmeta group nd the control roups s shown o be greater n thistest than n the other tests. One-way NOVAswith he posthoc tests

    TABLE 4Group Means and Standard Deviationsfor the Error Correction est

    Pretest Posttest Posttest

    Correction ype M SD M SD M SD

    Direct-only roup(n= 31) 28.3 17.0 41.2 23.1 42.4 24.1Direct metalinguistic roup (n = 32) 36.6 19.7 51.7 24.5 56.9 21.9Control group(w = 28) 29.5 19.6 28.4 21.4 27.4 19.3

    270 TESOLQUARTERLY

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    18/30

    FIGURE2Group Means on Speeded DictationTest

    showed hat hese ifferences ere ignificantn both Posttest ,F(2, 88)= 7.60,

    p= .001,and Posttest , F(2, 8) = 13.5,

    p< .001. More

    specifi-cally, n Posttest both direct-only nd directmetagroups utperformedthe control, n Posttest the direct-only roup manifested ignificantly

    FIGURE 3Group Means on Writing est

    FOCUSEDWRITTENCORRECTIVEFEEDBACK AND LANGUAGEAPTITUDE 271

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    19/30

    FIGURE 4Group Means on Error Correction est

    greater ains han he control roup, nd the metagroup utperformedboth the

    direct-onlynd control

    roups.It shouldbe noted hat ne-way epeatedmeasures NOVAswith otalscores howed hat ven the relatively oderate ainsobserved n thecontrol roup between retest nd Posttest as well s between retestand Posttest were ignificant: (l, 27) =9.57,p =

  • 8/14/2019 Sheen (2007)

    20/30

    TABLE 6Summary f Significant ontrasts etected by Posthoc Comparisons

    Total Dictation Writing Error CorrectionPosttest **D >C **D> C

    **M > C fM > CPosttest fM>D, C *M > C **M> D, C fD > C

    |M > D, C

    Note.D = Direct-only orrection roup, M = Direct metalinguistic roup, C = Control group;*/>< 01, t/>< 001.

    ables.As can be seen n Table 5, the three roups erformed ifferentlyon total est cores, ndicating significant ffect orCF. Also there wasa significant nteraction etween ime nd CF treatment. o statisticallyexamine hedifferences etween airs f groups, osthocmultiple om-parison estswereperformed. he results howed hat he direct metagroup produced higher otal cores han the direct-only roup nd thecontrol roup.One-way NOVAswith he posthoc omparisons evealedthat he differences n the cores were ignificantn both Posttest ,F (2,88) = 5.40,p < 0.01,and Posttest , F (2, 88) = 11.1,p < 0.001. Morespecifically, oth CF treatment roups utperformed he control roupin Posttest and the directmetagroup performed etter han hedirect-onlygroup n Posttest .Table 6 summarizes here he ignificant ifferences mong he hreegroups xist or he three eparate est cores s well as total cores. ntotal test cores,Posttest results avored oth treatment roups, utPosttest results avored he direct meta group. Looking at the testsseparately, hesignificant ontrasts n Posttest mainly ame from heerror orrection est.However, n Posttest , all three ests avored hedirect meta group.

    Table 7 displays hedescriptive tatistics or he anguage nalysis est.The direct meta group had the highest mean score and the controlgroup the owest.However, one-way NOVA revealed hat hesedif-ferenceswere not significant, {2, 88) = 1.09,ns. The results f therepeated measuresANCOVA howed hat herewas a significant ffectfor ptitude s the covariate, (2,88)=4.95,p

  • 8/14/2019 Sheen (2007)

    21/30

    students' otal scores on the tests ver time were mediated by theirlanguage ptitude, s measured ythe anguage nalysis est.However,

    the ANCOVA lso showed hat here wasstill significant ffect orCFafter ontrolling or he ffect f ptitude, (2, 88) =2.45,p

  • 8/14/2019 Sheen (2007)

    22/30

    Posttest indicate hat he written F had a positive ffect n the earn-ing of English articles. n particular, irect metalinguistic eedback

    proved o be effective n improving tudents' ccuracy n all three ests(and total cores) n the second administration f these ests.The current tudy s different rom revious ritten F studies n that

    only ne linguistic eaturewastargeted or he provision f CF and thetests evelopedmeasured tudents' written ccuracy lone. During he2-month eriod of this tudy, he teacher did not explicitly each orcorrect rticles utside he treatment. he students n all groupswere fthe ame evelof proficiency nd received he ame amount nd type finstruction nvolvingdenticalwriting nd reading materials. hus, CFalone could be seen as responsible or he cquisition f English rticles.

    Therefore, hefindings f this tudy rovide vidence hat he focusedwritten F resulted n improved ccuracy.SLAresearch nvolving ral CF has shown hatCFfacilitates earning

    when t focuses n a single inguistic eature nd makes heerror alient(Han,2002;Nicholas, ightbown, Spada,2001).Also, heprovision fCF targeting specific rror ncreases ts effectiveness Doughty &Varela, 998; Muranoi, 000).The current tudy reated nly wo imple,rule-based unctions f articles, nd students ere ikely o have beenfamiliar ith heform f the rticles t the beginning f the tudy. n thiscontext, ritten Fproved ffectiven enabling he earners o mprovetheir ccuracy n article se in both the short nd longer erm.

    It should be noted that he control roup s well s the experimentalgroups mproved ver ime,which uggests hat here was test racticeeffect. owever, he CF treatment roups outperformed he controlgroup, ndicating hat heCFtreatment ad an effect ver nd abovethetest ractice ffect.

    The second research uestion xamined the relative ffects f twodirect Fstrategies n learning: irect orrection ith nd without meta-linguistic nformation. he findings ndicate hat he two CFtypes addifferential ffects: irect orrection ith

    metalinguisticomments as

    superior o direct orrection ithoutmetalinguistic omments. his re-sult an be explained y chmidt's ccount f the roleof wareness n L2acquisition. chmidt 1995,2001)distinguished wareness t the evel fnoticing nd at the evelof understanding, hich s a higher evelofawareness.Noticingnvolves imply ttending o exemplars f specificforms n the nput e.g.,English as a and them entences); nderstand-ing ntailsknowing rule or principle hatgoverns hat spect of an-guage (e.g.,English ses a before he first mention f a noun and thebefore hesecondmention). Thus, t can be argued that whereas oth

    direct CF with nd withoutmetalinguistic omments re likely o pro-mote wareness snoticing, nly irect F with metalinguistic ommentspromotes wareness ith nderstanding. chmidt urther ontends hat

    FOCUSEDWRITTENCORRECTIVEFEEDBACKAND LANGUAGEAPTITUDE 275

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    23/30

  • 8/14/2019 Sheen (2007)

    24/30

    lar), it follows hat uch analytic killswillbe more strongly elated ogains n the group that receivesmetalinguistic F, especially n a test

    such as an error orrection est,which nvites heuse of metalinguisticknowledge. n the other hand, t s interesting hat, n the case of thedirect metalinguistic roup, he writing estgainswere more stronglycorrelatedwith he anguage nalysis coresthan the error orrectiontest gains.A possibleexplanation s that the writing estmost closelyresembled he treatment askwhere earners nitially eeded to applytheir anguage nalytic bilities o understand he metalinguistic om-ments. Another possibility s simply hat earners used their explicitknowledge o a greater xtent n the writing est.

    The current tudy s limited n severalways. irst, he studywas notcarried ut n the context f L2writing lasses. nvestigating he ffect fwritten F in that ontext would have afforded tronger cological va-lidity. econd, hewriting ask reatment asvery hort. more ubstan-tial CF treatment ight aveproduced venstronger nd more robusteffects. hird, he tudy xamined heeffects f CF on ust two elativelysimplefunctions f English rticles nd clearly he results annot begeneralized o other reas of grammatical ccuracy, r even to otheraspects f rticles. ourth, he tudents n the treatment roupswerenotrequired orevise heir writing. his might e considered limitation.However, ne advantage f excluding evision s that t allowsfor theeffect f the CF treatment y tself o be investigated.

    This study lso examined highly ocused ype f written F. It wasfocused n two enses. irst, t ddressed nly inguistic ccuracy. econd,it was directed t a singlegrammatical rea. Further esearch s neededto examine the effectiveness f the selective pproach to CF recom-mended arlier.Whatbenefits ccrue fCF isdirected t specific ritingskills? oes focused inguistic F help students chieve ong term ainsin accuracy o a greater xtent hanunfocused inguistic F? How fo-cuseddoesthe corrective eedback eed to be? One structure r two rthree?

    2 writingesearchers eed to conduct tudies

    especiallyongi-tudinal) that solate the effects f specific ypes f correction. uturestudies hould lsoexamine he effects f CF on more open-ended ndextensive riting asks.

    CONCLUSIONAs Chandler 2004)noted n her rebuttal f Truscott's 1996,2004)

    criticism f written F, the controversy urrounding he effectiveness f

    written F can onlybe resolved

    hrough arefully esignedtudies:

    I accept his] rgument hat he fficacyf error orrection or ccuracyof ubsequent riting an only e demonstrated y tudies ontaining

    FOCUSED WRITTENCORRECTIVEFEEDBACKAND LANGUAGEAPTITUDE 277

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    25/30

    control roupwhich eceives o correction nd experimental roupswhich orrect heir rrors fter ither eceivingirect orrection r hav-

    ingthe ocation f their rrors

    ointedut. SoI

    hopesomeonewill o

    such well-designedtudy, p. 348)The current tudy asdesignedwith hesepoints n mind o address hecentral uestion, oes error eedback elpL2students' ritten ccuracy?The answer o this uestion sa definite yes." hus, hecurrent indingsdo not support ruscott's laim that written F is ineffective. ruscottbased this laimon the fact hat no studies ad demonstrated hatwrit-ten CF had a positive ffect n acquisition. ruscott's ritique f writtenfeedback esearch resented challenge o researchers odevelopmeth-odologicallyound studies. n this espect, heresearch eported n thisarticle s a start n that direction.

    Written F is complex. t addresses different spects of writingcontent, rganization, hetoric, nd mechanics, s wellas linguistic c-curacy. he question rises,however, hetherwritten F should dealwith ll these spects t the ame time r address ifferent spects elec-tively hen correcting ifferent iecesof writing. 2 learners ave im-ited processing apacity nd asking hem o attend o corrections hataddress range f ssues t the ame timemay axtheir bility oprocessthe feedback. ne reason hat revious tudies fwritten Fhavefailed

    to demonstrate ny effect n students' ccuracy n subsequentwritingmay imply e that he inguistic eedbackwasnot sufficiently ocusedand ntensive. he study eported n this rticle oints o the mportanceof a selective pproach to correcting tudents' written ork.Teacherscan be selective n two enses.First, hey an elect to focusvariably ndifferent spects f a student'swriting t different imes, ometimes o-cusing n content r organization nd on other ccasions n linguisticcorrectness. econd,when he focus s on linguistic orrectness, eachersmay chievebetter esults f they elect specific rammatical roblemthat heyhave observed n their tudents' writing ather han a whole

    range of inguistic rrors. he results f this tudy uggest hat ocusedlinguistic F ismore ffective hen t ncorporates oth provision f thecorrect orm nd metalinguistic xplanation.

    ACKNOWLEDGMENTS

    This researchwouldnot have been possiblewithout he support f the faculty n theAmericanLanguage Program t Bergen Community ollege, Paramus, NewJersey,United States, nd the willing articipation f their tudents. am alsodeeplygrate-ful for the valuable feedback nd support by Zoltan Drnyei on earlier versions fthis rticle. Additional hanks o to George Ganat for his assistance with ata codingand the anonymous TESOLQuarterly eviewers or their constructive riticism ndhelpful uggestions. version f this paper waspresented t the 2005Second Lan-

    278 TESOLQUARTERLY

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    26/30

    guageResearch orum eld October -9at Teachers ollege, olumbia niversity,NewYork, ewYork, nited tates.

    THE AUTHOR

    Younghee heenhasrecently ompleted erdoctorate t the UniversityfNotting-ham,Nottingham, ngland. n fall 007, hewill e an assistant rofessornTESOLat American niversity, ashington, C, United tates. er research nterests n-clude nstructed econd anguage cquisition nd, n particular, ral and writtencorrective eedback.

    REFERENCESAshwell,. (2000).Patterns f eacher esponseo tudent ritingn a multiple-draft

    compositionlassroom:s content eedback ollowed yform eedback he bestmethod? ournal f econd anguageWriting, , 227-258.

    Bitchener, .,Young, ., 8cCameron, . (2005).The effect f different ypes fcorrective eedbackn ESL student riting. ournalf econdanguage riting, 4,191-205.

    Butler, . (2002).Second anguageearners' heories n the use of English rticles.Studies n Second anguageAcquisition,4, 451-480.

    Carroll, .B.(1981).Twenty-fiveears f researchnforeign anguage ptitude. n K.Diller (Ed.), Individual ifferencesnd universals n language earning ptitude pp.83-118).Rowley, A:Newbury ouse.Carroll, .,&Swain,M. 1993).Explicitnd mplicit egativeeedback: nempiricalstudy f the earning f inguistic eneralizations.tudiesn Secondanguage c-quisition,5,357-386.

    Chandler, . 2003).Theefficacyfvarious inds f rror eedbackor mprovementin the accuracy nd fluency f L2 student riting. ournal fSecondanguageWriting,2,267-296.

    Chandler, . 2004).A response o Truscott. ournalf econdanguageWriting,3,345-348.

    DeKeyser, . (1993).Theeffect t error orrection n L2grammar nowledgendoral proficiency. odern anguage ournal, 7,501-514.

    Doughty, , 8cVarela, . (1998).Communicativeocus n form. n C. Doughty J.Williams Eds.),Focus n orm n classroomecondanguage cquisition pp. 114-138).Cambridge: ambridge niversity ress.

    Ellis,R. (2005).Measuringmplicit nd explicit nowledgef second anguage:psychometric tudy. tudies n Second anguageAcquisition,7, 141-172.

    Ellis,R. (2006) Researchingheeffects f form-focusednstruction n L2acquisi-tion. n K. Bartovi-HarligcZDrnyeiEds.),ThemesnSLAResearchAILA eview,Vol. 9,pp. 18-41).Philadelphia:ohn enjamins.

    Ellis,R., Loewen, ., &Erlam, . (2006). mplicit nd explicit orrective eedbackand the acquisition of L2 grammar. tudies n Second anguageAcquisition, 8,339-368.

    Fathman, , 8cWhalley, . (1990).Teacher esponse ostudent riting: ocus nform ersus ontent. n B. Kroll Ed.), Secondanguage vriting: esearchnsightsorthe lassroompp. 178-190).Cambridge: ambridge niversity ress.

    Fazio, . (2001).The effect f correctionsnd commentariesn the ournalwriting

    FOCUSED WRITTENCORRECTIVEFEEDBACKAND LANGUAGE APTITUDE 279

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    27/30

    accuracy f minority nd majority anguage students. ournal f Second anguageWriting, 0, 235-249.

    Ferris, . (1999).The case for grammar orrection n L2writing lasses:A responseto Truscott 1996).Journal f econd anguageWriting, , 1-10.

    Ferris, . (2002). Treatment f rror n second anguage writinglasses. nn Arbor: Uni-versity f MichiganPress.

    Ferris, . (2004).The "grammar orrection" ebate n L2writing:Where re we, ndwhere do we go from here? And what do we do in the meantime . . ?).Journalof econd anguageWriting, 3, 49-62.

    Ferris, ., &Hedgcock,J. 2005). Teaching SLcomposition:urpose, rocess,f practice(2nd d.). Mahwah,NT:LawrenceErlbaum.

    Ferris, ., & Roberts, . (2001). Error feedback n L2 writing lasses:How explicitdoes it need to be? Tournai f econd anguageWriting, 0, 161-184.

    Gass, S. (1997). Input, nteraction, nd the econd anguage earner. Mahwah, NJ:

    LawrenceErlbaum.Han, Z.-H. 2002).Astudy f the mpact f recasts n tense onsistency n L2output.TESOLQuarterly, 6, 543-572.

    Havranek,G., & Cesnik,H. (2001). Factors ffecting he success of corrective eed-back. EUROSLAYearbook,, 99-122.

    Iwashita,N. (2003). Negativefeedback nd positive vidence n task-based nterac-tion. Studies n Second anguageAcquisition,5, 1-36.

    Kepner, C.G. (1991).An experiment n the relationship f types f written eedbackto the development f econd-language riting kills.Modern anguage ournal, 5,305-313.

    Krashen, . D. (1982).Principlesnd practicen second anguage cquisition. ewYork:Pergamon nstitute f English.

    Lalande,J. (1982). Reducing composition rror: An experiment. Modern anguageJournal, 6, 140-149.

    Liu, D., & Gleason,J. 2002). Acquisition f the article heby onnative peakers fEnglish. Studies n Second anguageAcquisition,4, 1-26.

    Long, M. (1996).The role of the inguistic nvironment n second language acqui-sition. n W. Ritchie & T. Bhatia Eds.), Handbook f econdanguage cquisition pp.413-468). San Diego, CA: AcademicPress.

    Long, M., nagaki, ., &Ortega, L. (1998).The role of mplicit egative eedback nSLA: Models and recasts n Japanese and Spanish. Modern anguage ournal, 2,357-371.

    Lys er, R. (2004).Differential ffects f prompts nd recasts n form-focused nstruc-

    tion. Studies n Second anguageAcquisition,6, 399-432.Mackey,A. (2007). Interaction s practice. n R. DeKeyser Ed.), Practice n secondlanguage earning: erspectives rom inguistics nd psychologypp. 85-110). Cam-bridge: Cambridge University ress.

    Mackey,A., 8c Philp, J. (1998). Conversational nteraction nd second languagedevelopment:Recasts, esponses, nd red herrings?Modern anguage ournal, 2,338-356.

    Master, P. (2002). Information tructure nd English article pedagogy. System,0,331-348.

    Muranoi, H. (2000).Focus on form hrough nteraction nhancement: ntegratingformal nstruction nto a communicative ask n EFL classrooms. anguage earn-ing, 0, 617-673.

    Nicholas,H., Lightbown, ., 8cSpada, N. (2001). Recastsas feedback to languagelearners. Language earning, 1, 719-758.

    Pica, T. (1991). Foreign anguage classrooms:Making them research-ready nd re-

    280 TESOLQUARTERLY

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    28/30

    searchable. n B. Freed (Ed.), Foreignanguage cquisition esearchnd the lassroom(pp. 393-412). Lexington, MA: D.C. Heath.

    Polio, C, Fleck, N., 8cLeder, N. (1998). "If only had more time": ESL learner'schanges n linguistic ccuracy n essayrevisions. ournal f econd anguageWrit-ing, 7,43-68.

    Robb,T., Ross, ., &Shortreed, . (1986).Salienceof feedback n error nd its ffecton EFLwriting uality. TESOLQuarterly, 0, 83-93.

    Robinson, P. (2001). Individualdifferences, ognitive bilities, ptitude complexesand learning onditions n second anguage acquisition. econd anguageResearch,17,368-392.

    Sawyer,M., 8cRanta, L. (2001). Aptitude, ndividualdifferences, nd instructionaldesign. n P. Robinson Ed.), Cognitionnd secondanguage cquisition pp. 319-353). New York: CambridgeUniversity ress.

    Schmidt, R. (1995). Attention nd awarenessn foreign anguage earning. onolulu:University f Hawaii Press.

    Schmidt, R. (2001).Attention. n P. Robinson (Ed.), Cognitionnd secondanguageinstructionpp. 3-32). Cambridge:CambridgeUniversity ress.

    Schmitt, ., Drnyei, .,Adolphs, ., 8cDurow,V. (2003).Knowledge nd acquisitionof formulaic equences:A longitudinal tudy. n N. Schmitt Ed.), The cquisition,processing,nd useof ormulaic equencespp. 55-86) Amsterdam: ohn Benjamins.

    Semke,H. (1984).The effects f the red pen. Foreign anguage Annals, 7, 195-202.Sinclair,M. (1991). Corpus, oncordance,ollocation.xford: Oxford University ress.Skehan,P. (1998).Acognitivepproach o anguageearning. xford: Oxford University

    Press.Snow,R. E. (1987).Aptitude omplexes. n R. E. Snow 8cM.J. Farr Eds.), Aptitude,

    learning nd instructionpp. 13-59). Hillsdale,NJ:LawrenceErlbaum.SPSS. (2002). SPSS,version 11.5 [software]. hicago, IL: Author. Availablefromhttp://www.spss.com/spss/.

    Truscott, . (1996).The case against grammar orrection n L2 writing lasses.Lan-guageLearning, 6, 327-369.

    Truscott, . (1999). The case for "the case for grammar orrection n L2 wntingclasses":A response to Ferris. ournal f econd anguageWriting, , 111-122.

    Truscott, . 2004).Evidence nd conjecture n the effects f correction: responseto Chandler. Journal f econd anguageWriting, 3, 337-343.

    APPENDIXA

    SpeededDictation est

    Name: Professor: Date:

    (Direction)

    There are 15 sentences n this est.The professor illread each sentence onlyonce, so pleaselisten arefully. fter istening o each sentence,write own the entence n the mallnotebookprovided.Write hesentence s fast s you can. Try o write hesentence xactly s you hear t.

    After ouhave finished, urn o the next page and get ready for the next entence. Once youfinish ach sentence, you must NOT return o the previouspage.

    (Note: Do not worry bout exact spelling.This is not a spelling est.)

    FOCUSEDWRITTENCORRECTIVEFEEDBACKANDLANGUAGEAPTITUDE 281

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    29/30

    ExampleYouwillhear:

    "Moneyannot uy ove."Thenwrite, Money annot uy ove."And hen, URNTOTHENEXTPAGE.For the rofessorTotal15 tems 9 indefinite rticles, 3 definite rticles)Example : feelgoodwhen speak nglish.Example :Tom peaksmany anguages. e'svery alented.

    1. I know hemanwhoruns his ollege.2. The red car cross heroad ooks uspicious.

    3. Doyouknow hepilotwhoflies his irplane?4. I saw movieastnight. he moviemademesad.5. John's nclewaskilled n a plane rash n NewYork.6. The man met n NewYork ecamemy usband.7. Therewas temple earmy ouse.The temple urned own esterday.8. Canyoumove he ar blocking y riveway?9. Please ellmewho he eader fyour lub s.

    10. I know lawyer. he awyer ants omarry e.11. Yesterday, saw police fficer hasing our og.12. Jenny asa dog.Thedogbit her boyfriend.13. John s a student f biologyt BCC.14. Tombought car.Hecrashed he ar the next ay.15. Therewas very ind octor n my ome own.

    APPENDIXB

    Error Correction estPleaseread eachstatement. achstatement as two entences hat re related. ne of thesentencess underlined. he underlined entence ontains t east ne error. here may emore han ne error n each underlined entence. rite ut heunderlined entence orrect-

    ing ll the rrors. Note:There re no punctuation r spelling rrors.)Example : Gloria ave ived n NewYork uring 001. he really njoysivingnNew ork.Answer: loria as ived n NewYork ince 001.Example :John ot cold.He couldn'twent o schoolyesterday.Answer: e couldn't oto schoolyesterday.

    1. Mary sedto ivingn Chicago. he ivesn NewYork ow.2. I look fter little irl nda little oy n Saturday. ittle irlwas mart ut he oy sn't.3. I took hree ests esterday.estswas odifficult.4. Tomquits mokingastweek.Hestarted moking gainbecausehe is too tressed ut.

    5. Theremight eeasyway oget oJohn's ouse.Canyou howmehishouse n the map?6. I saw man na car cross he treet. realized hat he mandriving arwasmy rother.7. Jen nd Bradusedto being ohappy ogether. couldn't elieve hat hey roke p.

    282 TESOLQUARTERLY

    This content downloaded from 35.8.11.3 on Sun, 10 Nov 2013 12:40:31 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/14/2019 Sheen (2007)

    30/30

    8. I saw a very nteresting ovie ast night. forgot he name of movie.9. Last night read a magazine nd a news rticle. don't know where news rticle s today.

    10. Ayoungwoman nd a tall man were talking utside myhouse. Ten minutes ater, youngwomanwasshouting t tall man.

    11. I read book about New York. The author, however,wasfrom California.12. We rented boat last summer. Unfortunately, oat hit another boat and sank.13. We went to basketball ame on Saturday. he players t the game were all very all.14. When you turn nto Paramus Road, you will ee two houses: a blue one and a yellow ne.

    I live n a blue house.15. Is your uncle car salesman? 'm looking to buy a car.16. Bill was so drunk ast night. He couldn't even recognized his girlfriend.17. Mymother was fired yesterday. he will have to find new ob.

    FOCUSEDWRITTENCORRECTIVEFEEDBACKANDLANGUAGEAPTITUDE 283