Using Corpora in Translation Studies Research - The ：翻译研究中使用语料库研究.pptPPT格式课件下载.pptx

Using Corpora in Translation Studies Research - The ：翻译研究中使用语料库研究.pptPPT格式课件下载.pptx

《Using Corpora in Translation Studies Research - The ：翻译研究中使用语料库研究.pptPPT格式课件下载.pptx》由会员分享，可在线阅读，更多相关《Using Corpora in Translation Studies Research - The ：翻译研究中使用语料库研究.pptPPT格式课件下载.pptx（60页珍藏版）》请在冰点文库上搜索。

,Handouts,TellmesomethingIdontknow,4,Notallfactsaboutlanguageareaccessiblethroughnative-speakerintuitionItsoftenimpossibletodescribemeaningsofwordswithoutreferencetothecontextsthosewordsoccurinSometimesyouneedalotofdatatoseepatternsSometimesyouneedtocount,Whatacorpuscanhelpussee,5,causehasnegativesemanticprosodyprovidehaspositivesemanticprosody（seeLouw1993;

Stubbs1995）ingeneral,passiveisusedaround10%ofthe,timeinEnglish,（seeHalliday1991）,Corpus,acollectionofnaturally-occurringtextsthataretheobjectofliteraryorlinguisticstudyheldinelectronicformthusamenableto（semi-）automaticprocessingusuallyassembledinsomeprincipledwayoftenhighlystructured,containingverycarefullyselectedtextscompiledforapurposeeglinguisticanalysis,lexicography,NaturalLanguageProcessing,translationresearchoraby-productofanotheractivity,egparliamentarydebates（andtheirtranslationsin,bilingualormultilingualparliaments）,6,Well-knownmonolingualcorpora,7,BritishNationalCorpus（100mwords）http:

/www.natcorp.ox.ac.uk/TheBankofEnglish（524mwords）http:

/www.collins.co.uk/books.aspx?

group=153freeaccesstoasampleof56mwordsathttp:

/www.collins.co.uk/Corpus/CorpusSearch.aspTheentireWorldWideWebhttp:

/www.webcorp.org.uk/,Corporaandcorpusprocessingsoftware,8,Provide:

a（relatively）objectivebasisforcommentaryonlinguisticphenomena,orlinguisticrealizationsof,eg,socialphenomenaaresourceforquantitativeandqualitativeresearchtheabilitytoaccessandmanipulate（sort,display,annotate）vastquantitiesofdata,thusfacilitatinganalysisbyhumans,Accesstocorpora,9,webaccessBankofEnglish,BNC,IDSfreeaccesstosampleraccesstolargercorpusbysubscription/permissionpurchaseowncopy（egBNC）createyourowncorpus,Webaccesstocorpora:

corpusprocessingsoftware,10,Web-accessiblecorporahavededicatedinterfacesthatallowusers,eg,to:

choosetexts/sub-corporasearchthecorpusforinstancesofaparticularword（resultsinKWICconcordanceformat）displayandsometimessortresults,Samplesearch:

CobuildBankofEnglish,11,Sampleresults:

CobuildBankofEnglish,12,Webaccesstocorpora:

corpusprocessingsoftware,13,Moresophisticatedinterfaces,eg,CosmasI*（IDSinMannheim）,allowedsearchesfor:

allformsofalemmagehengetsgehe,ging,gehst,geht,etc.allcompoundsformedfromasearchwordMbelgetsGartenmbel,Mbellager,etc.allformsderivedfromasearchwordKindgetsKindchen,kindlich,etc.*replacedin2003byCosmasII,Buildingyourowncorpus-basicdesignissues,14,WrittenvsspokenvswrittenandspokenStaticvsdynamic（monitor）Synchronic（timeofproduction?

）vsdiachronicGeneralreferencevsspecialisedMonolingualvsbilingualvsmultilingualDomainstobecoveredTexttypestobecoveredLevelofannotation:

rawvsannotationwithextra-textualinfo（headerstoindicatetexttitle,author,speakers,etc）vsdetailedlinguisticannotation,Rawcorpus/cleantext,15,Example:

Asheweakened,Moranbecameafraidofhisdaughters.ThisoncepowerfulmanwassoimplantedintheirlivesthattheyhadneverreallyleftGreatMeadow,inspiteofjobsandmarriagesandchildrenandhousesoftheirowninDublinandLondon.Nowtheycouldnotlethimslipaway.,fromJohnMcGahernsAmongstWomen,Corpusannotation:

structuralandPOStagging,16,ExtractfromtheBritishNationalCorpus:

Asheweakened,Moranbecameafraidofhisdaughters.ThisoncepowerfulmanwassoimplantedintheirlivesthattheyhadneverreallyleftGreatMeadow,inspiteofjobsandmarriagesandchildrenandhousesoftheirowninDublinandLondon.Nowtheycouldnotlethimslipaway.fromJohnMcGahernsAmongstWomen,Buildingyourowncorpus-GettingtheTexts,17,FulltextsvstextsamplesSamplingrandomSamplingorhandpicking?

CopyrightPermissionRepresentativenessdifficultconcepttoapplytotextualdataonusisonresearchertodocumentcorpuscontentsverycarefully,Conversiontoelectronicform?

18,Textnotavailableinelectronicform:

Scanning+OCRvsKeyboardingTextavailableinelectronicformWebpages（seeKilgarriffetal2006）DownloadsfromwebsitesFull-textdatabasesDonationsfromauthors/translatorsFormattextsshouldbesavedin?

.txt?

.html?

.xml?

（knowwhatyoursoftwarecanhandle!

）Isalignmentnecessary?

DoingResearchUsingCorpora,19,Corporacanhelpusanswerquestions,buttheyalsogeneratequestionsIngeneral,passiveisusedaround10%ofthetimeinEnglish;

butnearly20%oftheusesofcauseasaverbinmyscientificcorpusareinthepassivevoiceDoesthistellmesomethingabouttheverbcause?

scientificEnglish?

mysampleofscientificEnglish?

Hypothesisgenerationandtesting,20,HypothesisatentativeclaimegcausetendstobeusedmoreinthepassivethanotherEnglishverbshypothesessometimesemergefromcorpusdatacorpus-drivenresearchhypothesessometimesformedbeforewelookatcorpusdatacorpus-basedresearchnewlygeneratedhypothesescanbetestedagainstother（oftenbigger）corporacyclesofhypothesisformation,testing,refinement,testingcommonincorpusresearch,Someearlydescriptivehypothesesintranslationstudies,21,Translationstendtobesimpler/moreexplicit/moreconventionalthan:

othertextsinthesamelanguagetheirsourcetexts,Buthowdoyouoperationalizenotionslikesimplification,explicitation,normalizationincorpus-basedresearch?

Eg,whatconcretefeaturesofatextshowittobesimplerthananothertext?

（Seeespecially,Baker1993,1995,1996）,Earlyphase:

Developingmethodologiesandresources,22,EgQuestion:

Aretranslationsreallysimplified?

EgAnswer:

Comparedtowhat?

sourcetexts?

=parallelcorpusmethodologyothertextsinthetargetlanguage?

=comparablecorpusmethodology,ParallelCorpus,setofsourcetextsinlanguageAalongsidetheirtranslationsintolanguageB（andperhapslanguagesC,D,E）canbebilingualormultilingualegEnglish-NorwegianParallelCorpus（ENPC）http:

/www.hf.uio.no/ilos/forskning/forskningsprosjekter/enpc/egEuropeanParliamentProceedingsParallelCorpushttp:

/www.statmt.org/europarl/canbeunidirectionalorbidirectionalegGerman-EnglishParallelCorpusofLiteraryTextsde-,en,egENPCen-noandno-en,23,ComparableCorpus,24,MonolingualComparableCorpussetoftextstranslatedintoalanguageAalongsidetextsoriginallywritteninthatsamelanguageTranslationalEnglishCorpus（TEC）http:

/comparablesubsetoftheBNCFinnishComparableCorpus（Mauranen2004）,EarlyCorpus-basedTS:

QuantitativeBias?

25,AttempttoapproachtranslationobjectivelyRelianceonpropertiesoftextthatcanbemeasuredAveragewordlengthAveragesentencelengthLexicaldensityType-tokenratio,etcFocusonmonolingualcomparablecorporaFocusongeneraltendencies（oruniversals）intranslation（Laviosa2002;

MauranenandKujamki2004）,LexicalDensity,26,theratioofcontentwordstothetotalnumberofwordsinatextImadethechemicalshottersotheychanged.（4/8=50%）Raisingthetemperatureproducedachemicalchange.（5/7=71%）（fromGibbons2003:

20）thelowerthelexicaldensity,thesimplerthetext.,Type-TokenRatio,27,theratiooftypestotokensinacorpus,egtheTTRforThecatsatonthematnear,thelogfireis8/10（or80%）assumedtomeasurethe（lexical）varietyinatext:

thehighertheTTR,themorevariedthetextsvocabulary,TTR:

AnExample,28,InthispieceIrailagainstthetendencyoflinguiststowriteaboutthephilosophyofscienceasappliedtotheirsubjectfieldinsteadofwritingaboutwhatlanguagesarelike,whichiswhatlinguistsaresupposedtobegoodat.UnsympatheticcriticswillnodoubtchargethatbydoingsoIinstantiatetheverykindofbehaviorthatIamrailingagainst.FromPullum1991:

123TTR=49/63=78%,TTRcontinued,29,InthispieceIrailagainstthetendencyoflinguiststowriteaboutthephilosophyofscienceasappliedtotheirsubjectfieldinsteadofwritingaboutwhatlanguagesarelike,whichiswhatlinguistsaresupposedtobegoodat.UnsympatheticcriticswillnodoubtchargethatbydoingsoIinstantiatetheverykindofbehaviorthatIamrailingagainst.Thisisnotso.Iamcomplainingaboutunproductivemetaleveldiscussion,whichconsistsoflinguiststalkingaboutdoinglinguisticsinsteadofdoingit.Byofferingacritiqueofsuchwork,Iamoperatingatameta-metalevel,talkingaboutlinguiststalkingaboutdoinglinguisticsinsteadofdoingit.Thereisadifference.TTR=66/115=57%,Type/TokenRatio,30,isextremelysensitivetotextlengththelongerthetextis,thelowertheTTRis（normally）solution:

calculatetheTTRforsuccessivechunksoftexts（egevery100or1,000words）,andthentakeanaveragecountattheend（standardizedTTR）butevenstandardizedTTRsareproblematicasamarkerofsimplicity/complexityastheycantcapturethedifferencebetweenhardwordsandeasywordsegfirevsconflagration,OperationalizingSimplification,31,Iftranslatedtextsaresomehowsimplerthanoriginaltextsinthesamelanguage,thentheymighthave:

shorteraveragewordlengthshorteraveragesentencelengthlowerlexicaldensitylowerstandardizedtype-tokenratioscomparedtooriginals,32,InitialResults:

InvestigatingSimplification（Laviosa1998a）,33,Corpus:

newspaperarticlestranslatedintoEnglish,andnewspaperarticlesoriginallywritteninEnglishTool:

WordListinWordSmithToolsResultslexicaldensity:

lowerintranslatedarticlesaveragesentencelength:

lowerintranslatedarticlesstandardizedtype/tokenratio:

nosignificantdifference,InitialResults:

InvestigatingSimplification（Laviosa1998b）,34,Corpus:

fictiontranslatedin