英语翻译毕业论文Simultaneous-Listening-and-Speaking.doc

SimultaneousListeningandSpeaking1.INTRODUCTIONConcurrentlisteningtosourcelanguage(SL)andspeakingoftargetlanguage(TL)byinterpretersengaginginsimultaneousinterpretation(SI)constituteoneofthemostcomplexcognitiveprocessesahumanbeingcanexhibit.AswaspointedoutbyMoser(1997),thisisanimportantchallengeforresearchersinthisfield,sinceanin-depthstudyofthisuniquephenomenonwillrevealsomeofthesecretsofSI.Theparticularimportanceofthisisthatthesephenomenaareobservabletooutsideresearchersandcanbeeasilyquantifiedforstudypurposes.Ofcourse,thefactthatlisteningandspeakingprocessesareobservabledoesnotnecessarilymeanthatthesetwoprocessesaretheonlyonestakingplacesimultaneouslyinSI.Nooneknowsexactlywhatisgoingonintheinterpreter#39;scognitiveprocessduringSI,anditwouldbeerroneoustoassumethatnoinformationprocessingisoccurringevenduringtheveryshortmomentswhenbothinterpreterandspeakerremainsilent.Rather,itwouldbereasonabletoimaginethatsomeprocessesthatcannotbeexternallyobservedareunderway.TheseunobservablecognitiveprocessesmightincludeconversionfromSLtoTL,storing,note-taking(Gile1997),receivingsignalsthroughvisualchannels(Bhler1985)andreadingaterminologylistortheinterpreter#39;sownnotes,amongotherthings.Onecouldarguethatmulti-processingisnotararephenomenonandthatitalsohappensindailylife.Duringeverydaysmalltalk,ofcourse,peopleanticipatewhatotherswillsaynextandformtheirownanswerinadvance.Peoplecanalsoreadnewspaperswhiletalkingonthephone.Butspeakingandlisteningatthesametimeisveryrareindailylife(Gerver1976).NorcanthedepthoftheconcurrentcognitiveprocessinginvolvedinSIbecomparedtothatindailylife,sincetimeconstraintsandthemagnitudeofthetaskaretotallydifferent.Besides,concurrentjobsfoundineverydaylifealsoexhibittheirownlimits.Forexample,thehightrafficaccidentrateofdriversusingcellularphoneswhiledrivingclearlydemonstratesthatcognitivelimitexistsinone#39;sprocessingcapacity.SinceKoreanisnotaninternationallanguage,KoreannativesinterpretbothL2(English)intoL1(Korean),andL1intoL2.Intheformercase,understandingEnglishiskeyforqualitySIsincecomprehensionofsourcelanguageisthemostimportantstageinSIfromL2intoL1(Weller1991).Koreaninterpreters#39;capacityforlisteningmustbelimitedduetoreducedmemorycapacityforL2comparedtoL1,weakervocabularyandrulesstoredinlongtermmemory.Inmostcases,theperformanceofEnglishandKoreanSIisjudgedbythequalityoftheKoreanoutputsinceEnglishintoKoreanSItakesupthemosttime.Therefore,choosingelegantKoreanequivalentsratherthanjustconveyingthemeaningintheTLisveryimportant.Thus,thefollowingargumentforinterpretingfromL2intoL1quotedinWeller(1991)doesnotalwaysapplytoEnglishintoKoreanSI.Thebasicargumentinfavorofworkingonlyintoone#39;snativelanguageisthattheinterpreterspendslittletimeontheproductionphaseofinterpretationsinceheismorefluentinthenativetongueandinstinctivelyselectsthecorrectwordchoice,grammaticalconstructionandpronunciation.(Weller1991:391)Theproblemhereisthatconcretewordsaretranslatedmorequickly(andmoreaccurately)thanabstractwords,andhigh-frequencywordsaretranslatedmorequickly(andmoreaccurately)(deGroot1997).Exceptforthetimefactor,retrievalofelegantKoreanequivalentsfromlong-termmemoryconsumessomepartofthecognitivecapacityoftheinterpreter.Anotherdifficultyarisesfromsyntacticdifference,asisthecasewithEnglishintoJapaneseSI(Gile1991).WhatGileassertedcanbeappliedtoEnglishintoKoreanSIsinceKoreaninterpretershavetoholdtheEnglishverbforsometimesoastoputitintotheKoreansyntax,spendingaconsiderableamountofmemorycapacity.Thisstudyexaminesthefollowingquestions:WillthequalityoftheconcurrentlisteningandspeakingportionbytheinterpreterinSIbepoorerthanthatofthelisteningonlyportion?Doestheproportionofconcurrentlisteningandspeakingbytheinterpreteraffectthetotalqualityoftheinterpretationinanyway?DothefeaturesofSLaffectthedurationofconcurrentlisteningandspeakingbytheinterpreter?DointerpretersutilizethepauseinSLleftbythespeakerfortheirSIinordertodecreasetheproportionofconcurrentlisteningandspeaking?Concerningconcurrentlisteningandspeakingtime,Chernov(1979)reportsanaverageof70.5%oftotalspeakingtimeintheSLfortheEnglish-Russiancombination,withamaximumof89%.Gerver(1974b)foundthatsomeinterpreterscouldlistenandspeaksimultaneouslyforover75%ofthetotalinput-outputtime,andwerestillabletointerpretcorrectlyover85%oftheinputtexts.Gerver(1975)obtainedresultsof65%ofthetotaltimeforinterpretersworkingfromFrenchtoEnglish.Noneofthesestudies,however,comparetheaccuracyofconcurrentlisteningandspeakingportions2.MATERIALSANDPROCEDURES ThirtyactualEnglishintoKoreanSIaudiotapeswerechosentoobtainstatisticallymeaningfulresults.AsGile(1994)pointedout,generalizationswillbeprematureifthestudyislimitedtoverysmallsamplesofsubjects.Amongthe30samples,18werepairedSI,thatis,nineSLwereeachinterpretedbytwodifferentinterpreters.ThosewereintentionallyincludedtoobservetheperformanceofinterpretersunderthesameSLconditions.ThematerialincludedaUSpresidentialdebate,USPresidentClinton#39;sinaugurationceremony,livecoverageoftheGulfWarandspeechesfrominternationalconferences.11Theuseofreal-lifematerialsisalsoimportantinviewoftheassertionbyGile#39;s(1994)emphasisonusingexperimentalmaterialsandconditionsascloseaspossibletofieldconditions.Elevenprofessionalconferenceinterpreterswereinvolvedinthestudy.TheirmothertonguewasKoreanandEnglishwastheirstrongestpassivelanguage.Allofthemhadatleastthreeyearsofexperience.Thiseliminatedthedangerofusingstudentinterpreters,asGile(1995)warnedtheremaybeverysignificantdifferencesinthewayprofessionals,ontheonehand,andstudentsoramateurtranslators/interpreters,ontheother,performI/Ttasks.TheoriginalEnglishandKoreaninterpretationsweretranscribed,andtheaudiosignalsweresavedforanalysisbyapersonalcomputerwhichcouldmeasureassmallatimesegmentas1millisecond.Twosimultaneityratiosweresetup:1)asyllablesimultaneityratio(SSR)and,2)atimesimultaneityratio(TSR).TheSSR,theportionofsyllablesinterpretersbothlistenedandspokeconcurrently,wasmeasuredbycountingthenumberofsyllablesintheEnglishtextwithSIonthecomputerplayingit.TheTSRwascalculatedusingthefollowingformula:TSR=TST-LTI-SPIST/TST(TST=totalspeakingtime,LTI=listeningtimebyinterpreter,SPIST=speakers#39;pausewithininterpreter#39;sspeakingtime,TST=totalspeakingtime).Twodifferentevaluationmethodswereusedforqualityassessment.Forthequalityofasampleasawhole,thenumberofmeaningunitsintheoriginalEnglishspeechwascalculatedandthenitwasmatchedagainsttheKoreaninterpretationversion.Forexample,whentherewere10meaningunitsintheSLsampleand5wereinterpretedintotheTL,thesamplewasjudgedas50%accurate.Forthequalityoftimesimultaneityportionandlisteningportion,anevaluationbasedonthenumberofwords,notonthemeaningunit,wasemployed.WhenalltheoriginalEnglishwordsinthelisteningportionofanEnglishspeechwereinterpretedintotheKoreanversion,theportionwasjudgedas100%accurate,whereas50%accuracywasassignedwhenhalfofthewordsintheSLappearedintheTL.Thismethodofevaluationforthesimultaneityportionandthelisteningportionwasnecessarysincetheboundarybetweentheseportionwasnotlocatedbetweensentencesallofthetime;rather,itwasinthemiddleofsentencesmostofthetime.SomearticlesorpossessiveadjectivesthatarenotnecessarilyinterpretedinKoreanwerenotconsideredintheevaluation.Duringthequalityassessment,importancewasgiventomaintainingconsistencyofthecriteriasoastocomparethequalityofeachsampleandportion.3.RESULTSSpeakersexhibited77%ofspeechproportion(SP)whileinterpretersshowed58%andtheSPofinterpretertospeaker(I/S)was0.76.Anegativecorrelationwasfoundbetweenthespeaker#39;sSP(SSP)andaccuracyasawhole(r=-0.47p0.01).Interpreter#39;sSP(ISP)demonstratednochangewithanincreaseoftheSSPandspeaker#39;ssyllableperminute(SSPM).Speechrate,orsyllablesperminute,ofaspeakerwas231spm,whilethatofinterpreterwas243spm,andtheI/Sratiowas1.08.Thearticulationrateofthespeakerwas302spmwhilethatofinterpreterwas400spm,andtheI/Swas1.36.Regardingthenumberofsyllablesutteredbyspeakersandinterpreters,theI/Sratiowas1.09,whichmeansthatinterpretersuttered109%moresyllablesthanspeakers.Interpretersbothlistenedandspokeconcurrentlyfor61%oftotalsyllablesutteredbythespeaker.Thesyllablesimultaneityratio(SSR)wasnotinfluencedbyincreasingSSPandSSPM.TheSSR,however,showedpositivecorrelationswithISP(r=0.85p0.001),interpretersyllablesperminute(ISPM)(r=0.52p0.01),SPMI/Sratio(r=0.33,p0.05),SPI/S(r=0.56,p0.001),andtimesimultaneityratio(TSR)(r=0.73,p0.001).Interpreterslistenedandspokeconcurrentlyfor40%ofthetotalutterancetime.This40%oftheTSRremainedunchangedwiththeincreaseofspeakers#39;speechrateandarticulationrate.TheTSR,however,showedapositivecorrelationwithSSP(r=0.53,p0.01),ISP(r=0.79,p0.001),interpreter#39;sspeechrate(r=0.58,p0.001),SPMI/Sratio(r=0.41,p0.05)andSSR(r=0.73,p0.001).AnegativecorrelationwasobtainedbetweentheTSRandaccuracyasawhole(r=-0.32,p0.05).At-testshowedthattheaccuracyofthetimesimultaneityportionwaslowerthantheaccuracyofthelisteningportion(df=29,t=8.81,p0.001).TheaccuracyofthetimesimultaneityportionshowedastrongernegativecorrelationwithSSP(r=-0.57,p0.001)thantheaccuracyofthelisteningportionwithSSP(r=-0.35p0.05).AccuracyofthetimesimultaneityportionshowedastrongerpositivecorrelationwithSPI/S(r=0.51,p0.001)thantheaccuracyofthelisteningportionwithSPI/S(r=0.32p0.05).AstrongernegativecorrelationwasfoundbetweentheaccuracyoftimesimultaneityportionwithSSPM(r=-0.68,p0.001)thantheaccuracyofthelisteningportionwithSSPM(r=-0.45p0.01).TheaccuracyofthetimesimultaneityportionshowedahigherpositivecorrelationwithSSPMI/S(r=0.48,p0.01)thantheaccuracyofthelisteningportionwithSSP(r=0.36p0.05).AstrongernegativecorrelationwasobtainedbetweentheaccuracyofthetimesimultaneityportionwiththeSSP(r=-0.57,p0.001)thantheaccuracyofthelisteningportionwithSSP(r=-0.35p0.05).4.DISCUSSION 4.1.SLandTLFactorsThegreaterSSPoverISPcontradictsBarik#39;sassertion(1973)thatTgenerallyengagedinspeakingforagreaterproportionofthetimethanS.ButitisinlinewithLee(1999)whofoundsentenceSPforinterpreterswas79.6%,whilethatforspeakerswas86.2%forEnglishintoKoreanSI.ThismightbeattributabletotheuniquefeatureofSIingeneral.ItwouldbereasonabletoassumethatinterpretersengaginginSIrequiremoreprocessingcapacitythanspeakers.ThefactthatISPisnotinfluencedbySSPindicatestheinterpreter#39;sprocessingcapacityisalreadyatitslimitinkeepingupwithSLwhichisfasterthantheoptimalspeedof100-120wpm(Seleskovitch1978),or100wpm(Barik1975).Inthissituation,theinterpretercannotincreaseISPbeyond58%sinceaminimumcapacityshouldbereservedforotherprocessinginvolvedinSI.TheISPat58%meansthatinterpretersfill42%oftotalspeakingtimewithsilence.AlthoughtheaudiencelisteningtotheSIassumessuchsilencereflectsalossofinformation(DejeanLeFal1990),interpretershavenootherchoiceinthissituation.ThegreaterI/Sarticulationrateof1.36comparedtoSSPMI/Sof1.08suggeststhatinterpretersspeakquicklysoastocramtheiroutputoncetheyhaveachancetospeak.AlthoughinterpretersuttermorewordsintheirSIthantheoriginalspeech,thequalityoftheSIisfarfrombeing100%.Itbecomesclearthat,foranomission-freeinterpretation,moreKoreanwordsinTLarenecessary.4.2.SyllablesimultaneityratioSSRinthisstudyissimilartotheTSRinotherrelatedfindings.TheSSRisalsoaboutequaltoISP.TheunchangedSSRwithincreasingSSPandSSPMshouldbeunderstoodtomeanthattheinterpreters#39;capacityisalreadyatmaximumlevelandcannotincreaseSSRfurther.ThepositivecorrelationbetweenSSRandSPI/SaswellasSPMI/SmeanstheoverlappingsyllableportionincreasesastheinterpreterincreasesSP.4.3.AccuracyoftimesimultaneityportionTheoretically,interpretersshouldbeabletospeakwithpausesonlyforbreathingduringSI.ButitwasnotthecaseinSIsinceTSRremainedatonly40%.Thisclearlyshowsthatverycomplexinformationprocessingisgoingonintheinterpreters#39;cognitivesystem.WecannotattributethislowTSRtoalackofcapacityonthepartofinterpreters,sincetheyaretoprankinginterpreterswithgoodreputations.Inoptimalconditionsamples,theyrecordedveryhighaccuracy.Therefore,thereasonforthelowTSRseemstobetheuniquesituationofEnglishintoKoreanSI.TheTSRincreasesasinterpretersandspeakersacceleratetheirutterances.OnenoteworthyobservationisthatTSRwasfreefromthechangeinSSPMorspeakers#39;articulationrate.Inotherwords,TSRwasinfluencedbytheSPfactorratherthanbyspeedofSL.Figure1Figure1answerssomeimportantquestionsabouttherelationsbetweenSSP,ISPandTSR.Thefirstisthatthespeaker#39;spausewithinISPdecreasesasSSPincreasesandvice-versa.ThisisbecausethegapbetweenISPandTSRisthespeaker#39;spausewithinISP.AstheSSPincreases,speaker#39;spausewithinISPdecreasesproportionally,therebyleavingISPunchanged.Thisquitereasonableassumptionisconfirmedbythecorrelation(r=0.87,p0.001)betweenthechangesinspeaker#39;spauseintotalutterancetimeandspeaker#39;spause