1.Introduction
Fourorfiveyearsbeforethelaunch oftheInternationalCorpusofLearnerEnglish (ICLE)projectattheCenterforEnglishCorpusLinguistics,UniversityofLouvain,Belgium in 1990,acollection ofEnglish compositionsby Japaneseuniversity studentshad already started atShowa Women・sUniversity.Thatwasthestartofour long commitment,of almost25years,tolearnercorpora.TheICLE JapaneseSub-Corpus(ICLE-J)wasincluded into the written language corpus ICLE version 2 in the year 2009.And the Louvain InternationalDatabaseofSpoken English Interlanguage(LINDSEI)JapaneseSub-Corpus (LINDSEI-J)wasincluded intothespoken languagecorpusLINDSEIin theyear2010.A CD-ROM wasincludedinboth.
During theprocessofcompiling ICLE and LINDSEIcorpora,many studiesbased on thosecorporawerealsoadministered.Inthispaper,Iwouldliketoreview sixofmyown studieswhichshow somecommonfeaturesinJapaneseuniversitystudents・Englishinsuch awaythatIcanpresentacertaindirectionforEnglishlanguageteachinginJapan. 2.StudiesBasedonICLE-J
TheICLE-Jcontains366argumentativeessaysandintotalthereare198,241wordsand eachessaycontains542wordsonaverage.Basedon20random sampleratingsonthebasis oftheCommonEuropeanFrameworkofReferenceforLanguages(CEF),18essayswereB2 andlower,and2essayswereC1levels(Grangeretal.2009).
― ２ ― 学苑英語コミュニケーション紀要 No.846 2～16（20114）
WhatSomeStudi
esonICLE/LINDSEI
JapaneseSub-CorporaShow
TomokoKaneko
AbstractSincemy firsttrialofcollecting learnerlanguagedata,almost25 yearshavepassed. Finallyintheyear2010,bothJapaneseuniversitystudents・essaysandspokeninterview data were included into the InternationalCorpus ofLearner English (ICLE)and the Louvain InternationalDatabaseofSpoken English Interlanguage(LINDSEI),respectively,which have been compiled and published atthe Center for English Corpus Linguistics,University of Louvain,in Belgium.TheICLE andLINDSEIincludehugeamountsofelectronicdatafrom studentsfrom 11to16differentlanguagebackgroundsandmakeitpossibletocomparethe interlanguagefeaturesamong them.ThepresentpaperexploresspecialfeaturesofJapanese universitystudents・EnglishinterlanguageandtriestopresentadirectionofEnglishlanguage teachinginJapan.
21.Articles
Thefirststudy I・d liketo introduceison articleuse.Thefollowing showsa brief summaryofthestudy(Kaneko2007).
ResearchQuestions:
1.Whicharticleerrortype(substitution,redundancy,missing)isthemostfrequentamong Japaneselearners?
2.Arethereany differencesaccording totheirarticleproficiency in theway thelearners makeerrors?
3.Inwhatlanguagecontextarethearticleerrorsmostfrequent?
Participants:Advancedlevel(3rdand4thyear)Japaneseuniversitystudents Data:Error-TaggedICLE-J
Analysis:
1.Frequency and concordanceanalyseswereundertaken toascertain theextenttowhich thelearnersusedthearticlescorrectlyandtodistinguishtheerrortypes.
2.The essays were grouped into 3 levels (・poor・, ・mid・ and ・good・) based on the correctnessofarticleuse.
3.Thelinguisticcontextswherethearticleswereusedweremanuallychecked. Results:
1.Missing errorswerethemostfrequentin generaland theuseofdefinitearticlewas moreaccuratethanthatofindefinitearticles.
2.The・poor・group had extremely frequentmissing errors.The・mid・group had more substitutionerrorsandlessdefinitearticleredundancyerrorsthanthe・poor・group.The ・good・grouphadlessfrequenterrorsinbothkindsofarticles,butstillhadmoreerrors inindefinitearticlesthanindefinitearticles.
3.Mostfrequenterroneoususetookplaceinthebasicgrammaticalfunctionsofarticles. ThecollectionofICLE-JwascompletedbytheICLE Japaneseteam partlybeingfunded bytheJapanSocietyforthePromotionofScienceScientificResearchFund.Theteam also compiled130randomlyselectederror-taggedessaysfrom ICLE-J.Theerror-taggedICLE-J containsjustabout50,000words(with70,507tokensand5,096types).Thisstudyanalyzed thearticleerrortypesandthecontextoftheerrors,forexample,grammaticalfunctionsof thearticles,distancebetweenthearticlesandnounsandsoforth.
Chart1showsthatamongthethreeerrortypes,missingerrorswerethemostfrequent ingeneral.
Chart2furthershowsthat・poor・groupshadmanymissingerrors,butinthecaseof ・mid・group,theyhadlotsofsubstitutionerrors.The・good・grouphadtheleastfrequent errors throughoutallthe error types in both indefinite and definite articles.However, indefinitearticleerrorsweremoreconsistentthandefinitearticleerrorseveninthe・good・ group.
Asfor theerror frequency in differentlanguagecontexts,thestudentsmademore errorswhenthearticleswereusedinbasicgrammaticalfunctions(theindefinitearticlesto introduce a new specific or non-specific entity in discourse,and the definite article to indicateanaphoricand indirectanaphoricreference).Thissuggeststhatthelearnershave difficulty in distinguishing countable and uncountable nouns.In addition,the data is suggesting thattheshorter distancebetween thearticlesand thetargetnounsused in idiomsorlexicalphrasesworkasfavorablefactorforcorrectuse.
Among thetotalgrammarerrorsin theError-TaggedICLE-J,articleerrorswerethe mostfrequentandtheywereabout16％,whileverbandnounerrorswereabout11％ and 9％, respectively. Articles are really difficult for learners with a Japanese language background.In this sense,although the article errors are said to be localerrors and erroneoususesofarticlesdonotusuallyleadtoseriousmiscommunication,morecorrectuse willberequiredin writing by university levelstudents.Especially,Japanesestudentsneed tobeawareofthebasicfunctionsofarticles(i.e.,theisdefiniteanda,anareindefinite). Atthesametime,they needtousemoresetphraseswhich often includearticlesin them
― ４ ―
Chart2.ComparisonofErrorFrequencybytheThreeGroups GRS
IndefiniteArticles(a,an) DefiniteArticle(the) AVERAGE CU/ 10,000 IncorrectUse/10,000 CU/ 10,000 IncorrectUse/10,000 CU/ 10,000 10,IU/000 Sub. Red Miss. Total Sub. Red. Miss. Total
POOR 5,241 422 60 4,278 4,759 6,446 393 1,157 2,004 3,554 6,138 3,862 MID 7,554 1,591 208 647 2,446 7,926 1,244 752 234 2,074 7,794 2,206 GOOD 8,800 320 0 400 720 9,644 57 89 207 355 9,540 460
Notes) Sub.＝substitution Red.＝redundancy Miss.＝missing CU＝CorrectUse IC＝IncorrectUse
Chart1.FrequencyofArticleErrorsper10,000Words Articles CorrectUse IncorrectUse
Substitution Redundancy Missing Total Indefinite a 7,562 544 172 1,723 2,438 an 6,583 955 201 2,261 3,417 Average 7,435 597 175 1,792 2,565 Definite the 7,884 239 626 1,252 2,116 TotalAverage 7,735 358 476 1,431 2,265
sothattheywillhavemorechancestonoticetheexistenceofarticlesinEnglish. 22.BoundPrepositions
ThesecondstudyI・dliketoreview isonboundprepositions,especiallytheonesusedin ・verbplusboundpreposition・combinations.Thefollowing showsabriefsummary ofthe study(Kaneko2008).
ResearchQuestions:
1.Isthereany differencein thefrequency ofpreposition errorsaccording towhetherthe prepositionisusedinfreeorboundcombinations?
2.Inthecaseofboundprepositionswithverbs,wheretheform isrealizedas・verbplusbound preposition,・isthereanyeffectofthephrasemeaningonthecorrectnessofuse?
Participants:AdvancedlevelJapanese,French,GermanandItalianuniversitystudents Data:ICLE Japanese,French,German and Italian Sub-Corpora and Louvain Corpus of
NativeEnglishSpeakers(LOCNESS) Analysis:
1.Thefrequencyofprepositionerrorsinboundandfreecombinationswerecountedandthe errortypeswerecategorizedintothreetypes(missing,redundancy,andsubstitution). 2.Thepreposition errorsin ・verb plusbound preposition・combinationswerecategorized
accordingtothesemanticdomainsdesignatedinBiberetal.(1999).
3.The frequency ofbound prepositions in the combination in ICLE Japanese,French, GermanandItalianSub-CorporawascomparedtothatinLOCNESS.
Results:
1.Thereweremoreerrorsin bound prepositionsthan in freeprepositions.Redundancy errorsonlyemergedinprepositionsin・verbplusboundpreposition・combinations. 2.AlthoughJapanesestudentsusedmorecommunicationdomain(ex.talkto)in・verbplus
bound preposition・combinationscomparedtostudentswithotherlanguagebackgrounds, erroneoususesofarticleswerealsothemostfrequentinthiscombination,especiallyin passivevoicepattern(ex.beexpressedin).
Thefrequency oferrorsin bound and freecombinationswerecounted and theerror typeswerecategorizedastowhethertheyweremissing,redundancyorsubstitutionerrors asshowninChart3.
Then preposition errors in bound combinations were categorized according to the semanticdomainsexpressedbythecombinationbasedonLongmanGrammarofSpokenand WrittenEnglish(1999).ThecategoriesbyBiberetal.(1999)are・activity・,・communication・, ・mental・,・causative・,・occurrence・,and,・existence or relationship.・ Atthe same time, frequencyofboundprepositionsinICLE-Jandintheother3sub-corpora(French,German
andItalian)wascomparedtothatinLOCNESS.Chart4showsthecomparisonoffrequency of・verbplusboundpreposition・combinationsbywritersofvariousL1backgroundsbased onthesemanticdomainsforevery10,000words.Itclearlyshowsthatthefrequencyofuse amongthesixsemanticdomainsistheleastamongJapanesestudentscomparedtostudents with other language backgrounds and Japanese students used the ・verb plus bound preposition・combinationsin communication domain almostasoften asthestudentswith theotherL1backgroundsandnativespeakers(NSs).However,atthesametime,whenthe frequencyoferroneoususewascounted,thefrequencywasthehighestinthecommunication domain,especiallywhenthesentenceswereinthepassivevoice.
In thisstudy,an adhocquestionnairewasadministeredtofindouthow thestudents decided touseacertain verb or・verb plusbound preposition・combinationsin sentences. TheresultwasthatJapanesestudentsuseprepositionsanalytically,basedon theoriginal meaning oftheprepositionseven in thecaseofbound prepositions.They aresofamiliar with the analyticalapproach to learning English that they try to understand allthe prepositionsasfreeforms.TheydonotcarealotaboutusingEnglishphrasesasaset.
― ６ ―
Chart3.FrequencyofBoundvs.FreePrepositionErrorsinError-TaggedICLE-J asObligatoryContext100％
ErrorTypes BoundPrepositionErrors FreePreposition Errors
Total Errors XNPR XADJPR XVPR XPRCO Total
Missing 41 12 98 5 156 92 248 Redundancy 0 0 31 0 31 32 63 Substitution 19 2 38 0 59 22 81 Total 60 14 167 5 246 146 392
Notes) XNPR＝Lexico-Grammar,Nouns,DependentPreposition XADJPR＝Lexico-Grammar,Adjectives,DependentPreposition XVPR＝Lexico-Grammar,Verbs,DependentPreposition XPRCO＝Lexico-Grammar,Prepositions,Complementation
Chart4.ComparisonofFrequencyofXVPR byWritersofVarious L1BackgroundsBasedontheSemanticDomains
23.LexicalPhrases
Thelaststudy on ICLE-J isaboutthecharacteristicsin theuseoflexicalphrasesby Japanese university students compared to native speakers.The following shows a brief summaryofthestudy(Kaneko2005).
ResearchQuestions:
1.How doestheratiooffunction wordsandcontentwordsusedby Japaneselearnersof EnglishcomparetothatbyNSs?
2.How doestheuseoflexicalphrasesbyJapaneselearnersofEnglishcomparetothatby NSs?
3.Whatarethetypicalfeaturesoferrorsseen in theuseoflexicalphrasesby Japanese universitystudents?
Participants:AdvancedlevelJapaneseuniversitystudents
Data:ICLE-J,Error-taggedICLE-J,LOCNESS,FLOB (TheFreiburg-LOB ( TheLancaster-Olso/Bergen)Corpus),FROWN (TheFreiburg-BrownCorpus-TheStandardSampleof Present-Day American English),and WC (TheWellington CorpusofWritten New ZealandEnglish)
Analysis:
1.Thefrequency offunction andcontentwordsseen in thetop50wordsin ICLE-J was comparedtothoseseeninfournativespeakercorpora.
2.Two to six-gramsofthefivecorpora wereanalyzed using a computeranalysistool, CollocatebyBarlow (2004).
Results:
1.ThepercentageoffunctionwordsusedbyJapanesestudentswastheleastcomparedto thatoftheNSs.
2.LOCNESS,FLOB,Frown,andWC show ahighlevelofconformanceamongthem asfor thekindsoflexicalphrasesused,whileJapanesesub-corpusshowsonly alow levelof conformancewiththem.
3.ThefrequencyandvariationoflexicalphrasesusedbyJapanesestudentsweremuchless than thosein theNS corpora,becausethelearners:a)used differentsetsoflexical phrasesfrom NSs(ex.whatIwanttosay,Ithink),b)didn・tuselexicalphrasesin ・p atterns(ex.as～as)・which could beapplied toavariety ofexpressions,c)overused 2-gramsand underused morethan 3-grams,and d)often used directexpressionslikeI wantto,whileNSsusedindirectoneslikeI・dliketo.
First,thefrequencyoffunctionandcontentwordswerecomparedbylookingatthetop 50wordsinthecorpora.Chart5showstheresult.
Although Japanese students learn both function words and content words in the ― ７ ―
― ８ ―
Chart5.FrequencyofFunctionWordsandContentWordsSeenintheTop50Words
CORPUS NNS NS
WordTypes Japanese British American Australian ICLE-J FLOB Brown LOCNESS WSC
Content Words noun 7 0 0 1 0 verb 12 9 9 7 10 adj. 2 1 1 1 1 adv. 4 3 3 4 4 total 25(50％) 13(26％) 13(26％) 13(26％) 15(30％) Function Words prep. 7 9 9 9 9 pron. 6 14 16 12 13 art. 2 3 3 3 3 conj. 8 7 6 7 7 aux. 2 2 2 4 1 inter. 0 2 1 2 2 total 25(50％) 37(74％) 37(74％) 37(74％) 35(70％)
Chart6.ComparisonofLexicalPhrasesintheFiveCorpora Category
Grams
CommoninNS&
NNScorpora Appearimorethan3NScorporangonlyin AppeariJapaneseCorpusngonlyin 2-grams haveto suchas
outof morethan forexample eachother notonly soon Ithink 3-grams alotof inorderto aswellas afactthat anumberof intermsof infrontof acoupleof asaresult aseriesof as＊ as andsoon moreandmore 4-grams atthesametime ontheotherhand inthecaseof foralongtime inthefirstplace forthefirsttime itissaidthat others whatIwanttosay
classroom,theyunderusedfunctionwords.Theyusedfewerfunctionwordscomparedtothe fournativespeakercorpora,too.Thesametendencywasseeninthefrequencyandvariation oflexicalphrasesin theuseof・patterns・(ex.as～as)and morethan 3-grams.Chart6 showsthecomparisonoftheiruse.
As for the use oflexicalphrases,the NSs show a high rank order agreementin frequencyoftheuseoflexicalphrasesamongthegroup,however,Japanesestudentsshow farlessagreementwiththem.ThelexicalphrasesusedbyJapanesestudentsaremainlyin twograms.Inaddition,thelexicalphrasesusedbythem areverydifferentfrom theones seeninNScorpora.Forexample,IthinkandwhatIwanttosaywerefrequentlyusedby them butthephrasewasnotlistedatallinChart6inthenativespeaker・scolumn.
Thelessfrequentuseoffunction wordsby Japanesestudentsstrongly relatestothe lessfrequentuseoflexicalphrasesusedby them.Themorefocuson function wordsand alsoonlexicalphraseswhichcontainthoseinclassroom activitieswilldefinitelybewhatis neededforEnglishteachinginJapan.
3.StudiesonLINDSEI
LINDSEIisacomplementaryprojecttoICLE,whichcompilesadvancedleveluniversity students・spoken English.Thekey objectiveoftheprojectisto collectcomparabledata among thestudentsofvariousL1 backgrounds.Thedata wascollected from about15-minuteinterviews.Atfirst,theparticipantwasrequestedtochooseatopicandtalkabout it.Thentheintervieweraskedquestionsrelatedtowhattheparticipanthadspokenabout, andalsoaboutmoregeneraltopics,forexample,lifeatuniversity,hobbies,futurejobs,etc. Inthesecondhalfoftheinterview,theparticipantwasaskedtolookatfourpictureswhich madeupashortstory.
31.PastTenseForms
Thefirststudy on LINDSEI-J isabouttheuseofpasttenseforms.Thefollowing showsabriefsummaryofthestudy(Kaneko2004).
ResearchQuestions:
1.Towhatextentdolearnersusethepasttenseformscorrectly? 2.Whatkindofqualitativefeaturesareshowninlearners・errors? Participants:AdvancedlevelJapaneseuniversitystudents
Data:Fifteenoutof51LINDSEI-J Analysis:
1.The four types ofverbs (irregular,regular,beverbs,and auxiliary)were coded for correctandincorrectuseandthecorrectnessratioswerecompared.
2.Theconditionsforthecorrectuseofthoseverbsareexplored. ― ９ ―
Results:
1.Irregularverbswereusedmorecorrectlythanregularverbsintokens.
2.The erroneous use was often caused by the meaning ofthe verbs.In addition,the amountofinput,whetherlearned in chunksorby rules,theexistenceofpasttense marker(forexample,conjunctionwhen),andthepositioninthesentenceseem toaffect thecorrectnessratio.
Theaim ofthestudywastofindthecorrectnessratioofthepasttenseformsandsome qualitativefeaturesoftheuse.Thepercentageofthefourtypesofpasttenseverbscoded forcorrectandincorrectuseisshowninChart7.
Irregularverbswereusedmorecorrectlythanregularverbsintokens.Onereasonformore correctuseofirregularverbsintokensbutnotintypescouldbethatthelearnersuseda smallersetofirregularverbsmanytimes.
Chart8 shows the average rank ofcorrectness ofthe three types ofregular and irregularverbs.Inthechart,thelessthenumeralsare,thehighertherankorder.Thus,the irregulareventverbswereusedmostcorrectly.
Itwasalsofoundthatverbsoftentaughtandpracticedinclassroomsandusedwithother markerswhich letthelearnersnoticethe・past-ness・,wereused morecorrectly than the otherverbswithouttheseconditions.
Thefactthatirregularverbswereusedmorecorrectly than regularverbsiswhatis expected according to theformer studies(e.g.Bardovi-Harlig,1992).However,when the correctnessratiowascountedbytypes,itisinterestingtoknow thatsincevarietiesofthe irregularverbsarenotenough,theJapanesestudentsusedregularverbsmorecorrectly. 32.MotherTongueUse
Thesecond study on LINDSEI-J isaboutmothertongueuse.Thefollowing showsa ― 10―
Chart7.PercentageofCorrectUsesoftheFourTypesofVerbs WordCount
Types RegulVerbsar IrregulVerbsar BeVerbs Auxiliaries Total Tokens 64.4％ 65.7％ 50.4％ 53.1％ 59.2％ Types 58.2％ 54.7％ 50.0％ 50.0％ 54.8％
Chart8.ComparisonoftheAverageRankofCorrectness TypesofVerbs States Activities Events RegularVerbs 20.38 18.00 13.53 IrregularVerbs 23.75 14.60 12.60
briefsummaryofthestudy(Kaneko2009). ResearchQuestions:
1.TowhatextentisL1usedinL2speechbyJapaneseuniversitystudents? 2.WhatarethefunctionsoftheL1inL2speech?
3.WhatarethedifferencesintheuseofL1accordingtothespeakers・proficiencylevels? Participants:Japaneseadvancedleveluniversitystudents
Data:LINDSEI-J Analysis:
1.AlltheL1inthedataweretaggedinthreecategories(contentwords,interjections,and speakers・internalmonologues).
2.FrequencyofL1ineachcategorywascalculatedandtherelationshipbetweentheuseof L1andthestudents・Englishproficiencylevelswereexplored.
Results:
1.OneL1wordwasusedin every 267words(about0.4％),andin every 36turns(about 3％).
2.L1 content words were used when the speaker couldn・t find appropriate L2 words. AbundantuseofL1interjectionswasnoticeable.
3.Thelowergroupusedmoreinterjectionsthan theuppergroup.Theuppergroupused morevariousL1contentwordsthanthelowergroup.
Inthisstudy,thefunctionsofL1inL2speechanddifferencesintheL1useaccording tothespeakers・proficiencylevelswereexplored.
First,theL1wordsandphrasesinthedatawerecategorizedintothreegroups(content words,interjections,andspeakers・internalmonologues)accordingtohow theL1wasused in thedata.Then,thefrequency ofL1in each category wascompared to thestudents・ Englishproficiencylevels.
Chart9showsthefrequencyofL1use.
Furtherstudy isnecessary to say whetherornottheL1useisfrequent,butsince studentsknew thattheinterview wouldbeadministeredin English,Ishouldsay thatthe ratioof0.4％ showsthatthestudentsusedalotofL1words.Itwasunexpectedthatthere
― 11―
Chart9.FrequencyofL1UseAccordingtotheThreeFunctions ContentWords
Interjections InternalMonologues Total noun verb adjective adverb
22(14.6％) 1(0.7％) 1(0.7％) 1(0.7％)
111(73.5％) 15(9.9％) 151(100.0％) 25(16.6％)
weresomanyL1interjectionsinthedata.
Chart10showsthatthestudentsinthelowergroupusedmoreJapaneseinterjections thantheadvancedlevelstudentsandgraduallythefrequencyofL1interjectionslessened.
TheabundantuseofL1interjectionsinthelowerproficiencygroupsuggeststhatthey oftensearchfortheappropriateexpressionstocontinuetalking.TheuseofL1interjections likeetto,ummtto,ornanndakkeoftenblocksasmoothcommunicationflow inEnglish.This againsuggeststhatsincethestudentsdidn・thaveampletimetocommunicateinL2,they naturally used L2interjectionsin speeches.They connectL2wordsand phrasesjustlike puttingpuzzlepiecestogether.
33.NegativeEmotionVocabulary
Thelaststudy on LINDSEI-J aimed forfinding outapragmaticaspectoflanguage. Thefollowingshowsabriefsummaryofthestudy(Kaneko2003).
ResearchQuestions:
1.How do English learnerswith differentL1backgroundsuseEnglish negativeemotion vocabularyandthestrategiesaccompanyingtheexpressions,comparedtonativespeakers ofEnglish?
2.IsthereacorrelationbetweentheirEnglishproficiencyandthefrequencyofthestrategy use?
Participants:Japanese,Chineseand French advanced leveluniversity studentsand native speakersofEnglish
Data:LINDSEIJapanese,Chinese,and French Sub-corpora aswellasLLC ( TheLondon-LundCorpus)andWC
Analysis:
1.Vocabulary usedforexpressing ・anger・,・surprise・,・anxiety・,and・grief・weretagged inthedataandthefrequencyoftheuseswerecomparedamongthethreesub-corpora. 2.Thestrategiesusedforexpressingnegativeexpressionsweregroupedbasedonemotional
expressionstrategiesbyRintell(1984). ― 12―
Chart10.FrequenciesofL1UseAccordingtotheThreeFunctionsinUpper& LowerGroups ContentWords
Interjections InternalMonologues Total gs. noun verb adjective adverb
U 21.1％ 1.8％ 1.8％ 1.8％ 57.9％ 15.8％ 100.0％ sub-total26.3％
L 10.6％ 0.0％ 0.0％ 0.0％ 83.0％ 6.4％ 100.0％ sub-total10.6％
Results:
1.Theuseoftheemotionvocabularydifferedaccordingtothelearners・languagebackgrounds. Japanesestudentsshowedatendencytopreferlessface-threateningexpressions.Onthe whole,thelearnersexploitlessvariousexpressionsthanNSs.
2.Strategies for expressing negative emotions also differed according to the learners・ linguisticbackgroundsand Japanesestudentsused minimizersmostfrequently among thethreegroups.
3.Englishproficiencylevelaswellasthefrequencyoftheexpressionandofaccompanying supportingstrategiesdidnotseem tocorrelate.
Theuseofnegativeemotionvocabularyandthestrategiesaccompanyingtheexpressionsby Japanese,Chineseand French studentswerecompared tothatby nativespeakersin this study.Thestudy alsocheckediftherewasany correlation between theJapaneselearners・ Englishproficiencyandthefrequencyofstrategiesused.
Chart11clearlyshowsthatthefrequencyoftheuseofemotionvocabularydifferedaccordingto the learners・L1 background.Japanese students showed a tendency to prefer less face-threatening expressions.Forexample,they used Iam angry orsomeoneisangry,while Frenchstudentsusedyouareawfuloritisawfulmoreofteninexpressinganger.Thefact thatJapaneseculturedoesnotpreferface-threatening expressionsmay havelimited the rangeofexpressionsutilizedinaccordancewiththeirlimitedEnglishproficiency.
Asforthestrategy use,thereseemstobethesametendency asshown in Chart12. Japanesestudentsusedminimizers(forexample,kindoforalittle)mostfrequentlyamong thethreegroups.
Thepercentagesshownwithoutbracketsinthechartindicatetheratiooftheuseofthe targetstrategyamongthetotalfrequencyofthestrategiesusedinthetargetcorpus.The percentagesshownwithbracketsindicatetheratiooftheuseofthetargetstrategiesamong thetotalfrequencyofthetargetemotionvocabulary.
― 13―
Chart11.FrequencyofEmotionVocabularyinFiveCorporaper10,000Words Corpora
Emotions Japanese Chinese French LLC WC Anger 5.16 6.85 6.55 0.32 2.22 Surprise 4.39 1.27 1.02 0.05 0.59 Anxiety 3.61 3.34 2.94 0.13 1.25 Grief 0.28 2.07 2.71 0.07 0.58 Total 13.41 13.53 13.22 0.58 4.65
In general,thelearners・repertoireofnegativeemotionalexpressionsand supporting strategiesareverylimited.Thisphenomenonisnaturalconsideringthefactthatamongthe vocabularylearnedinjuniorandseniorhighschoolsinJapan,thevarietiesofadjectivesand adverbsareextremelylimited,becausethelargerportionofthevocabularylearnedamong thelimitednumberofvocabularyinthecourseguidebytheMinistryofEducation,Culture, Sports,ScienceandTechnologyisallottedmoretonounsandverbs.
Finally,asfortherelationshipbetween English proficiency levelandthefrequency of theemotionvocabularyanditssupportingstrategiesused,thereweren・tanycorrelations. 4.Summary
Asawhole,therangeofvocabulary typesandexpressionsusedby Japanesestudents isverylimitedcomparedtoNSsandstudentswithotherL1backgrounds.Thedistribution offrequentlyusedvocabularyandlexicalphrasesbyJapanesestudentsareoftenuniqueand different.
AnimportantfindingwasthatJapaneseICLE andLINDSEI-JssuggestthatJapanese university students are rule learners. They use English based on their grammatical knowledge just like putting puzzle pieces together.For them to utilize their English knowledgeforcommunicationbothspokenandwritten,moregenuineinputandchancesfor spontaneoususeoflanguageseem tobeindispensable.Inusingspontaneouslanguage,the learnerswillhavemorechancestonoticethegapbetweentheirowninterlanguageandthe targetlanguageforms.Anotherimportantfindingwasthatthelearnersfirststartedusing thetargetlanguagefeatureoccasionally and then theuseincreased with lotsoferrors.
― 14―
Chart12.StrategyUseAttachedtoEachEmotionVocabulary Corpor
Strategies Japanese Chinese French LLC WC Sentence Structure Exclamatory Sentence 6 14.3％ 22 12.9％ Modifiers Intensifier 16 84.2％ 47 90.4％ 39 69.6％ 24 57.1％ 101 59.1％ Minimizer 15.8％3 9.6％5 8.9％5 14.3％6 8.2％14 Repetition 3 5.4％ 3 1.8％ Additional Words Oh,(～) (Ohmy)God 9 16.1％ 6 14.3％ 31 16.1％ FrequencyofStrategy 19 (36.6％) 52 (61.2％) 56 (47.9％) 42 (41.4％) 171 (33.4％) FrequencyoftheTarget Emotion(100.0％) 52 85 117 103 512
Finally when they gotto a moreadvanced level,theiruseevensoutwith fewererrors. Thus,under-use,overuseand leveling outseem to bea common processoflearning any languagefeature.
Itried to use the corpora in a few differentcombinations to conductthe studies. Sometimesthestudy wassolely basedon theICLE orLINDSEI-Js,sometimestheresults gained from thedatafrom Japanesestudentswerecompared with corporacollected from students with other L1 backgrounds or by severalnative speaker groups.ICLE and LINDSEImadethiscomparisoneasybecausetheformatofallthesub-corporaisthesame.
In using severalcorpora for thestudiesincluded in thepresentpaper,error-tagged ICLE-Jhasbeenagreatsourcefordetectingtheinterlanguagefeaturesofthelearners.In thenearfuture,although it・llrequirea considerabledegreeofeffortand time,inserting errortagstoalltheICLE-JandLINDSEI-Jdata,Ibelieve,wouldbethenextenterprisefor theICLE Japaneseteam todoinordertomakethosecorporamoreversatile.
I・dalsoliketoaddthatsincemanystudiessofararedescriptive,explanatorystudies wouldbethenextstep.However,atthesametime,therearestillthousandsofthingsto be explored even by descriptive studies.For example,ifwe introduce the difference of meanings expressed by,say,adverbs or adjectives,suggestions and findings lead by descriptivestudiesstillgrow bigger.Inthissense,Ihopethatmanymoreresearchersand teachers use the ICLE/LINDSEIand their Japanese Sub-Corpora,and find many more interestingfactsonlearnerlanguage.
Onelastcommentisthatlanguagespeopleusechangequickly astimegoesby,and researchersneedtoadddatain ordertokeepthatkindofcorporaupdated.On theother hand,thelanguageacquisitionprocesswillneverchange.Thus,fortunatelyorunfortunately, theyearsthatourcorpuswillbeasusefulastheyarenow isalmostforever.
ReferencesConsulted
Biber,D.etal.(1999).Longman GrammarofSpoken and Written English.Harlow,Essex:Pearson Education.
Bardovi-Harlig,K.(1992).Therelationshipofform andmeaning:A cross-sectionalstudyoftenseand aspectintheinterlanguageoflearnersofEnglishasasecondlanguage.AppliedPsycholinguistics 13:253278.
Barlow,M.(2004).Collocation.Softwareintroduced attheworkshop atShowaWomen・sUniversity, October24,2004.
Gaetanelle,G.,S.DeCock,and S.Granger (eds.)(2010).Louvain InternationalDatabaseofSpoken EnglishInterlanguage(LINDSEI).Belgium:UCL PressesUniversitairesDeLouvain.
Granger,S.,D.Estelle,and M.Funny (eds.)(2002).InternationalCorpusofLearnerEnglish (ICLE). Belgium:UCL PressesUniversitairesDeLouvain.
Granger,S.,D.Estelle,M.Fanny andP.Magali(eds.)(2009).InternationalCorpusofLearnerEnglish (ICLE)(Version2).Belgium:UCL PressesUniversitairesDeLouvain.
Kaneko,T.(2003).TheuseofnegativeemotionalexpressioninEnglishbynon-nativespeakers:A
basedcomparativestudy.Gakuen752,124.
Kaneko,T.(2004).Theuseofpasttenseformsby JapaneselearnersofEnglish.In J.Nakamura,N. Inoue and T.Tabata (Eds.),English Corpora under Japanese Eyes (pp.215230).Amsterdam: Rodopi.
Kaneko,T.(2005).EnglishlexicalphrasesusedbyJapaneseuniversitystudents.Gakuen774,114. Kaneko,T.(2007).Whysomanyerrors?:UseofarticlesbyJapaneselearnersofEnglish.Gakuen798,
116.
Kaneko,T.(2008).UseofEnglishprepositionsbyJapaneseuniversitystudents.Gakuen810,112. Kaneko,T.(2009).UseofmothertongueinEnglish-as-a-foreign-languagespeechbyJapaneseuniversity
students.Gakuen822,2541.
Kaneko,T.,T.KobayashiandM.Takami.(2001).StudiesonLINDSEIJapanesedata.Gakuen730,1121. Rintell,E.M.(1984).Buthow didyouFEELaboutthat?:Thelearner・sperceptionofemotioninspeech.
AppliedLinguistics5(3),25564.
（金子 朝子 英語コミュニケーション学科）
