Anda di halaman 1dari 6

Sipes1

PeterSipes LexicalAcquisition

ArticleCritique1
Horst,M.(2005).LearningL2vocabularythroughextensivereading:Ameasurementstudy.Canadian ModernLanguageReview/LaRevueCanadiennedesLanguesVivantes,61(3),355382. Fulltextonlineat:http://bit.ly/12Zszx6

Introduction
ExtensiveReading(ER)isamethodoflanguagelearningthatemployswiderangingreadings gradedtostudentabilitythatindicatesgenerallanguagegains(Horst,2005).Previousstudieshave shownthatERcanboostvocabulary,butHorstsresearchlookstomakebettermeasurementsof vocabularylearning.

ArticleSummary
Theresearchersbuiltalibraryofgradedreadersthattheyscannedin.Sinceittookmorethan threehourstoscansomeofthelongerbooks,theydecidedtosetalowerlimitofscanningtothefirst twentypagesofeachreader.Intheend,theyhadseventytitlesintheirERlibrary.Thirtyreadershad morethantwentypages,andasaresulthadsomewherebetween18%and100%ofthetextscanned. Theythennextgavetwotestsof100wordseachtoESLstudents.Onewasoverwordsinthe 1,0012,000frequencyrange.Theotherwasonlowfrequencywords(generallythosenotinthetop 2,000)toseewhattheirlevelofvocabularywasatthebeginningofthestudy. Studentswerethenallowedtoreadthelibraryofscannedreadersforsixweeks.Theingenious partofthestudywasthatthestudentsreadingwastracked,sotheresearchersknewexactlywhat lowfrequencywordseachstudenthadseeninthecourseoftheirER.Additionally,sincethe researcherswerentdirectingstudentreadingselections,theeachstudentsinterestdirectedtheir

Sipes2

reading.Whiletherewasagooddealofvariation,themeannumberofbooksreadbyanystudentwas 10.52booksoversixweeks.Attheendofthesixweeks,asecondtestof100wordswasgivento eachstudent,butthistimetheywerecustomizedtothestudentsreading.Thestudybeganwith21 studentsandfinishedwith17students. Theyfoundthat,despiteearlierconcernsabouttestlength,studentscompletedthepreand posttestinlittletime.Moreimportantly,theyshowedimpressivegains. Table1(fromHorst2005) 1,0012,000wordsratedYESof50wordstotal(n=17) Pretest M SD t=5.47p<.001 Table2(fromHorst2005) OfflistwordsratedYESof50wordstotal(n=17) Pretest M SD t=5.47p<.001 Ascanbeseen,studentspickedupanaverage6.59wordsinthe1,0012,000rangeand 10.29wordsinthelowfrequencylist.Thisamountisaboutdoubletotriplethenumberofwords learnedinmostpreviousstudies(seeHorsttable2). 33.80 8.18 Posttest 43.59 4.30 Difference 10.29 7.62 41.35 5.38 Posttest 47.94 1.89 Difference 6.59 5.47

Critique Goals
ThegoalofthisstudywastoseehoweffectiveERisatbuildingvocabularyknowledgein Englishlanguagelearners.PreviousstudieshaveindicatedvocabularygainswithER(Ferris,1998Pitts,

Sipes3

White&Krashen,1989interalia),butHorstfeltthatpreviousstudiesneithertestedonenoughwords norincludedenoughreading,whetherstudentselectedornot. Shealsoseemedtowantabetterfixonjusthowtomeasuretherate/amountofvocabulary learning.Previousstudieswerelimitedtoshorttestsorreadingsnotselectedbystudentstwelve wordsandasixparagraphreadinginonepreviousstudy.Andbothofthoseposeproblems.Smaller samples(i.e.shorttests)maynotbestatisticallysignificant.Studentsmayalsonotbeasmotivatedto workwhentheyarenotinchargeoftheirwork.Byworkingonsolutionstobothoftheseflaws,Horst shouldbeabletogetbetterresults.

Assumptions
Oneproblemwithvocabularylearningisthataresearcherneedstobeabletoseparatewhich wordswerelearnedwithinthecontextofthestudyandwhichwerelearnedoutsideofthestudy.Toget atthis,Horsttestedtwolevelsofvocabularyfrequency.Theideaisthatinthe1,0012,000listof words,studentscouldrunintothosewordsoutsideoftheER,whereastheofflistwordswereunlikely tobeencounteredanywherebuttheER.Bydividingwordsintovariouslevelsoffrequency,thenoiseof nonERvocabularylearningcanbeseparatedfromthesignalofERvocabularylearningwhilealso showingoverallvocabularygrowth.Aneattrick. HorststatesthatpreviousresearchwithESLstudentsinspokenlanguage,communicative methodsbasedclassroomsfindsveryfewwordsnotfoundonlistsofthemostfrequentEnglishword families(Meara,Lightbown,&Halter,1997).(2005).Togetaroundthisflaw,classroomssupplement instructionwithER.Theassumptionisthatwrittenlanguagehasmorelowfrequencywordsthanspoken language.Icantdisputethispoint.OnanotherprojectformyLexicalAcquisitionclass,Iamlearning

Sipes4

100newEnglishwords.SofarIhavefoundtwentyone,ofwhichonlythreewereinspoken contextsoneinaconversationalsetting.(Theothertwowereontheradioandinapresentation introducingacurriculumchangewhereIteach.) IfoneweretowonderabouttheefficacyofERitself,theresultsofthestudywouldindicatethat vocabularygainsdidhappen.NotthatIwaseverindoubt.

Otherpoints
Astheauthorpointsout,theformativetestatthebeginningofthestudyandthelimited digitizationofreadersposeproblemswhilebeinganimprovementoverpreviouswork. Thebiggestlimitationoftheinitialtestisthatitcannothopetotestthewordsthatstudentswill runacrossintheirERexceptstatistically.Icantthinkofagoodremedyforthisoffhand.Onepossible wayatthiswouldbetomakethepretestlonger.Testingmoreitemsinthe1,0012,000range increasesthelikelihoodofhittingitemsontheposttesttomakedirectcomparisons.Expandingtesting forofflistwordsisprobablynotfeasibletherearejusttoomany.Butgiventhatstudentscompleteda 100itemtestinabouttenminutes,itcouldbequitefeasibletobumpthenumberof1,0012,000words testedfromfiftytoonehundred. Thelimiteddigitizationofthereaderswasalsoaproblemonseveralfronts.First,itdiscounted thenumberoflemmasthatcouldbetestedafterwards.Afterall,ifyourethrowingoutalemmathathas onlyoneoccurrenceinthefirsttwentypagesofthereader,youcouldwellbethrowingoutaninfrequent butcriticalvocabularyiteminthereader.Thisistosaynothingofthevocabularyitemsthatdidnt appearinthefirsttwentypagesofthereaders.Iagreewiththeauthorsapproachofwritingtothe readerspublisherstoseeiffullydigitizedversionsofthetextexist.Isuspectthateightyearsonthat

Sipes5

morereadercopywouldbeavailabledigitallyprovidedcopyrightconcernsdidnttrumpresearch concerns.Thatsaid,youwouldexpectpublishersofgradedreadersandL2materialstoseecommercial importanceinthissortofresearch.Afterall,abettermousetrapshouldsellbetterespeciallyifyoucan marketitastakingthelatestresearchintoaccount. Finally,isERworththework?Thetestingindicatedthatoversixweeks,studentslearnedan averageof6.57wordsinthe1,0012,000range.Theproblemisthis:isthatgainindicativeofa13% growthinvocabulary?(6.57wordsofa50itemtestisabout13%.)Or,morereasonably,isitasmaller percentage?Afterall,onewouldexpectstudentsatthepointofthoseinthestudytoknowatleastsome wordsinthe1,0012,000range.Thepretestindicatedthatonaveragethestudentsknewsomewherein theneighborhoodof80%ofthelistalready.Ifthatisthecase,woulditnotbeeasiertoteachthe remaining200itemsinclassroominstructionorviavocabularylists?Again,gettingintothedata,the numberofbooksstudentscheckedoutfromthelibraryofdigitizedreadersduringthestudyrangedfrom 3to33.Thatvariationisnosmallthing,thoughImnotawareofanyresearchintowhetherfastbut shallowreadingissuperiortoslowbutdeepreading. (Mindyou,IthinkERisthewaytogo.Ijustwanttobringthesequestionsupsincenot everyonelikestoread.)

Conclusion
ERproducesgainsinstudentvocabulary,particularlyinlowerfrequencywordsnotaslikelyto beheardinspokencontexts.WhileHorstsstudyhadseveralselfacknowledgedlimitations,itshows ERisasolidtoolforvocabularylearningandpointsthewaytobettermeasurementofvocabulary learning.

Sipes6

Anda mungkin juga menyukai