以语料库为基础对学术写作进行的词性分析的报告

Corpus-basedlexicalanalysisforNNSacademicwriting———readingreportforCorpusTechnologyclassPapersource::AlejandroCuradoFuentesUniversityofExtremaduraReporter:LiulanshenStudentnumber:41004276Abstract:•Thispaperdescribesthefirstphaseinatwo-phasestudyaboutNNS(non-native)writers’choiceoflexical-grammaticalitemsinacademicEnglishwriting.•thecorpus-basedstudymeanstoprobecharacteristicNNSlexicalfeaturesbyworkingwithaverydifferentreferencecorpus:nativeEnglishwritinginotherdisciplines.•Thepurposeatthisstageistodescribepatternsthataretheresultofcharacteristicuse,NNSoveruse,misuse,and/orevenfirstlanguagetransferFewworkshaveselectedspecificcorpusmaterialwiththeaimofanalysingL2writinginthelastphaseprior(学前阶段)topublication.ItisafactthatgrowinginterestinacademicEnglishwritingasperformedbyNNShasledtoextensiveliteratureandresearchoverthepastthreedecades,theoriginsprobablywhentheBritishCouncil(英国文化委员会)firstbegantousethetermEAP(EnglishforAcademicPurposes)toreferto“interdisciplinary(各学科间的)studiesinrelationtoexistingpracticesandinstitutions”(Brumfit,1984:17),orto“exchangeofknowledge[...]accordingtospecificfeaturesofspecialisedsubjectfields”(Baunmann,1994:1).Differentperspectiveseversinceonthematterhavebeenadopted,suchasacademicvocabularyknowledge(e.g.,Martin,1976),orwritinginthegenreconventions(e.g.,Swales,1995).Variousschoolsorassociationshavealsoformedasaresult,forinstance,inBritain,BALEAP,fromwhichdifferentapproacheshaveevolved(e.g.,corpusdata-drivenlearning,academicphraseology,discourseanalysisintheacademicsetting,etc—cf.,Johns,1993;Howarth,1998;Lockett,1999,etc.—).lessinvestigatedseemstobetheaimtoexaminecorpus-basedanalysiswithbothNS(native)andNNSmaterialsothatthespecialisedNNSwritingcorpusanalysiscandetectcharacteristicweaknessesandstrengthsinthewritingprior.ThecorpusexaminationhasbeendonebycomparingthetextswithNSmaterialfromaselectionoftheBNC(BritishNationalCorpus)Sampler.Thechiefobjectiveintheprocesshasbeentoidentifybothsimilarityanddivergenceintermsofthesignificantlexicalitemsused,especiallyacademiclexicalitemsand/orrhetorical-lexicalitems.Presentconditionobservations——NNSwritingframework•NSwriterstendtohavehigherdegreeoflexical-grammaticalknowledgeintermsoftheexactwordtest,andthatL2writerstendtohavebettercontentknowledgethanlinguisticcommand.•ContentknowledgeplaysanessentialroleintheL2mindtoplanandstructurethepaper,whereastheNSfocusonwritingseemstobefirstonstructureandthencontent.•Quantitativeandqualitativeoptionsforlexis,grammarpatterns,andcohesive/structuraldevicesseparateNNSfromNSdiscourse.Intheauthor’sbelief…Acorpus-basedanalysisofNNStextsshouldneverbetakenasawaytoprescribe规定awritingmethod,butratherasafundamentallydescriptiveformofexplainingwhat,how,andevenwhysomechangesoccur.WhythesehappenandcharacteriseagivenwritingtypecouldthenrelatetoNNSstrengthsandweaknessesinthelanguagelearningcontext.corpusfeatures0102030405060Tokens(in10,000s)Types(in1,000s)STTR#oftexts#ofdisciplines#ofgenres/texttypesNNSCorpusBNCselectionNNCCorpus:totalnumberoftokens:25,931,distinctwordsortypes:6,856STTR(StandardisedToken-to-TypeRatio,normalisedrateperevery1,000words):48.03BNCCorpus:Tokens:482,022threedisciplines:12textsfromArtsandSciences,fiveinSocialSciences,andfourinArtsandHumanities,magazinearticles,journalarticlesandreports.•Thecoredata:Thecorpusanalysisstartswiththecomparisonofsomefunction/grammaticalwords,andso,themostfrequentgrammaticalwordsamongthefirst100appearasthecoredata.•Example:Theirfrequency-basedpositionsinthetwocorporalistsshoweitherrankingsimilarityordifferentiation.(e.g.,thepossessiveourinposition50intheNStext,butinposition40intheNNStexts).020406080100onbywithbewecanusefromalsoshouldintobotheachsosomemaysuchIifour020406080100NNStextsBNCtextsFigure2:ContrastiveviewofrelativefrequenciesforselectedfunctionwordsfromthewordlistcomparisonThisfirststepinthemanagementofthecorpusdatashouldserveasthebasisfortheexaminationoffrequencysimilaritiesanddivergence.TheanalysisofmostgrammaticalwordsatsuchhighpositionsonthelistsisboundtoproducesignificantresultsThesecondstepinthecorpusanalysisdealswiththecorecontrastiveexplorationoflexicalitemsasattributesofacademicstance(cf.Biberetal.,2004).Theacademicregistershouldbeatleastpartlycheckedbymeansofcollocationalstrengthdegrees—i.e.,intermsofcontent,grammatical-discursiveposition,semanticspace,andtext-itemrelationships(cf.Hoey,2005)—;thus,theseitemsshouldbecloselyexaminedaccordingtotheirrelativefrequencieswithineachcorpussothattheirusepercentagesmaybepairedupbetweenthetwocorpora.Useoflexicalitems024681012INFORTHATONWITHTHISORHAVEATFROMBUTNOTSHOULDONEEACHSOMESUCHIFnumberofitemsSimilaruseNNSgapNNSuseFigure3graphicallyrendersthenumbersoflexicalitemsderivedfromthewordlistfunctionwords.Theitemsanalysedcompriselexicalcollocations,colligations,semanticassociations,andtextualcollocations(cf.Hoey,2005).Collocationthestatisticallysignificantco-occurrenceoftwoormorewordsinthetextsappear*+tobe,formedasacommoncollocationinbothcorpora,beingusedaround20percent(20forN