语料库研究基本方法

语料库研究基本方法中国外语教育研究中心梁茂成321主要内容☻语料库语言学的性质☻几个常用术语☻语料库研究的基本方法321语料库语言学的性质1☺理性主义与经验主义☺Rationalism:IthinkthereforeIam.☺Empiricism:Mymindisa‘blankslate’.Seeingisbelieving.321语料库语言学的性质1☺theWaxArgument:Heconsidersapieceofwax;hissensesinformhimthatithascertaincharacteristics,suchasshape,texture,size,color,smell,andsoforth.Whenhebringsthewaxtowardsaflame,thesecharacteristicschangecompletely.☺However,itseemsthatitisstillthesamething:itisstillapieceofwax,eventhoughthedataofthesensesinformhimthatallofitscharacteristicsaredifferent.321语料库语言学的性质☺theWaxArgument:Therefore,inordertoproperlygraspthenatureofthewax,hecannotusethesenses.Hemustusehismind.Descartesconcludes:☺“AndsosomethingwhichIthoughtIwasseeingwithmyeyesisinfactgraspedsolelybythefacultyofjudgmentwhichisinmymind.1321语料库语言学的性质☺Empiricism:Empiricismemphasizesthoseaspectsofscientificknowledgethatarecloselyrelatedtoevidence,especiallyasdiscoveredinexperiments.☺Itisafundamentalpartofthescientificmethodthatallhypothesesandtheoriesmustbetestedagainstobservationsofthenaturalworld,ratherthanrestingsolelyonreasoningandintuition.1321语料库语言学的性质☺Scienceisconsideredtobemethodologicallyempiricalinnature.☺Corpuslinguisticsisempiricalinnature.1321语料库语言学的性质☺语言研究中的数据类型☺内省数据（introspectivedata）:rationalism☺实验数据（experimentaldata）:empiricism☺真实数据（anthenticdata）:empricism1321语料库语言学的性质☺语料库语言学提倡真实数据☺我们不排斥其他数据类型1321语料库语言学的性质☺即便在语料库语言学阵营之中☺Corpus-driven:minimumtheory-reliance.Exclusiverelianceoncorpusdataforalltheories☺Corpus-based:Relianceoncorpusdataforhypothesis-testing☺Corpus-referenced/informed:Occasionallyresortingtocorpusdataforillustrations1321语料库语言学的性质☺我们坚决反对不顾语言事实的任何论断☺Nointrospectioncanclaimcredencewithoutverificationthroughreallanguagedata(Teubert2005).1321几个常用术语2☺Corpus☺Corpuslinguistics321几个常用术语☺Token,type,lemmaThelittleboylookedattheotherboys.2321几个常用术语☺Collocationisdefinedasasequenceofwordswhichco-occurmoreoftenthanwouldbeexpectedbychance.☺abigsmoker☺astrongsmoker☺ahardsmoker☺aheavysmoker☺afurioussmoker2321几个常用术语☺Itisquitepossible,infact,todescribeawomanashandsome.☺However,thisimpliesthatsheisnotbeautifulatallinthetraditionalsenseoffemalebeauty,butratherthatsheismatureinage,haslargefeaturesandacertainstrengthofcharacter.☺Similarly,amancouldbedescribedasbeautiful,butthiswouldusuallyimplythathehadfemininefeatures.2321几个常用术语☺Colligationisdefinedasasequenceofgrammaticalcategorieswhichco-occurmoreoftenthanwouldbeexpectedbychance.2321几个常用术语☺SemanticprosodyisinstantiatedwhenawordsuchasCAUSEco-occursregularlywithwordsthatshareagivenmeaningormeanings,andthenacquiressomeofthemeaning(s)ofthosewordsasaresult.Thisacquiredmeaningisknownassemanticprosody.(Stewart2010)2321语料库研究的基本方法3☺Corpus-basedapproach:ahypothesis-testingapproach☺Corpus-drivenapproach:withas“fewpreconceivedideas”aspossible,“keepingtheamountoftheory-reliancetoaminimuminordernottohindertheprocessofdiscoveringnewphenomena”(Römer2005)321语料库研究的基本方法☺Bothapproachesalmostalwaysinvolveacomparionofsomekind.3321语料库研究的基本方法☺Sizesofcorporaincomparison(Rayson2003)☺Small=big☺Equalsizes3321语料库研究的基本方法☺Typesofcomparison☺Acrossgenres☺Acrossusers☺Acrossdifferenttimes☺Across(varietiesof)language(s)3321语料库研究的基本方法☺Corpuscomparability3321语料库研究的基本方法☺Linguisticfeaturesincorpuscomparison☺Lexical☺Lexico-grammatical☺Syntactic☺Discoursal3321语料库研究的基本方法☺Statistictestsincorpuscomparison☺Simple:☺Relationship(correlation,etc)☺Difference(chi-square,loglikelihood,etc.)☺Complicated:regressionanalysis,factoranalysis,clusteranalysis,correspondenceanalysis3321语料库研究的基本方法语料库研究问题研究设计软件统计检验结论?参照语料库对比结果：词汇短语搭配语义韵类联接句式等数据呈现数据分析、解释与讨论3321内容55Thankyou.