《从修订历史看维基百科翻译的趋势》翻译项目报告.pdf

'．‘＇－＇‘－－．＇－＾＊？■．：．叫中氏＜ｒ、Ｖ．、，—户ｖ先非作一７巧心．产＂）户每占二＇苗吟如－－Ｖ．－一六过产數六扣；、、－＾＂？Ｓ＂：如，心、：－二．－－．－；、？＊、＇节／：：．：：．二？ｉ，Ｉ：Ｃ＊ｒ－，一．ｒ，？．．、心－．：产一一－ｖ．一一一、．，，？入、。气一一一，？一、？？一？．，，？，一，＾ｙ＇．、－’．？＂。？一＇－＾－．心ｃ？？一？．．只Ｖ、Ｖ、一－，心＾．二．．’－－－－？？－－：．－＇吝去Ｇ．；祥户，．知。雜由旅ｙ；賓震？一？一乂一－？／＇运兴给巧巧专化学位硕女论文胃苗．－－、？－．？．，：＞’＇一：：：、：．＇．ｖ．：，．心皆，：；一片一：班．砖ｒ，：．《乂舒々史看＊１１势》ｆｔＡｔ种舶种唉徊种巧《旅告■？．－－二－一－咬；迫．％；＇－号？？，．？－－一一、＊：．＇＇＇’＊－■’？－．＇＇－一＇、．．－、、ｒ子巧‘＞＇＊二ｖ．、—．？—心狂－乂巧．：。．：：定：謀奇、。娜岩拓１－．＇—＇．’产．一－．．‘？：一‘－廿，．、接群择與讓Ｓ；：＾；，苗苗為＞．幕．—．．．＇＇．．、－－．三■－－－Ｌ－．Ｖ‘－－产常兴．苦／Ｖ．ｖｖ；於Ｊ声．＇：：’‘－．＇－＇．．．叫．气．，＇Ｙ，ｃ，Ｃ扣巧＇：六Ｃ吁公瑞听公汽．疗，、＇繁－．，；：指导道洽巧审巧巧巧．；１．茲．＇．＇、１：：：：丈专化巧《巧硕名赛麥卸；一、巧々巧，笔巧々《巧＇＇？．、＇＇＾：、二＾＾＾＾＾＾＾＾破；点Ｖ诘＾＾■■——一—一一＇—－’丰片＇Ｎ兰，＇＇．、，－一、’■＊占＇－人产义２０巧５，论文巧辩时巧年月．二－．一、？－．．：心化文：编号Ｍ６２３ｇｇ心ｒ＇．、一■■－：＿？＇‘■－、ｖ－＼：甲Ｖ．：Ｖ；，、、．一■＇＇．‘．、‘＇．：．Ｖ．Ｗ、：、．Ｖ．－：：一．；？一＇？、－’＞－ｆ三名、、站：：．：？、、、－＇＂＇―．－■■■－＾心’二■？－＂－‘Ｖ甲ＶＡ．＼．＼六气山／；＆一—；Ｖ＾．＾■Ｔ：＇■，＇占－＇■．－——Ｖ．二＾：－．■．‘‘‘‘＂■■＇＇：、一＇一＇＇．：：；．支？二一＇＞’’、户今：乂疋ｒ＇Ｖ冷巧扭 SichuanInternationalStudiesUniversityATranslationProjectReportontheTranslationofRevisionHistory:TranslationTrendsinWikipediabyZhengJiAthesissubmittedtotheGraduateSchoolinpartialfulfillmentoftherequirementsforthedegreeofMasterofTranslationandInterpretingunderthesupervisionofSeniorTranslatorWeiDaopeiChongqing,P.R.ChinaMay2016 i 《从修订历史看维基百科翻译的趋势》翻译报告摘要本翻译报告选自《从修订历史看维基百科翻译的趋势》这篇论文作为翻译及研究文本。原文主要以维基百科这个特殊的众包翻译平台作为研究对象，采用布瑞恩·莫索普的编辑与修订程序分类法为主要研究方法，探讨众包模式下产生的译文中转换错误和语言/排版错误出现的频率以及解决这些错误的方式。报告的主要内容有五部分。第一，翻译项目背景及意义。第二，原文文本介绍，包括作者信息、原文内容及文本特征分析。第三，翻译理论，主要包括纽马克的交际翻译和语义翻译理论。第四，翻译难点与翻译方法。原文中频频出现的专有名词、术语、长难句为翻译过程中的难点，本文就这些难点引发翻译问题作了重点讨论。第五，总结在翻译过程中所得到的启发、教训和仍待解决的问题。作者希望通过撰写此翻译报告，为研究众包翻译的学者提供参考材料，为此类文本翻译研究贡献一份力量。关键词：翻译项目报告；众包翻译；维基百科；修订历史；交际翻译ii ATranslationProjectReportontheTranslationofRevisionHistory:TranslationTrendsinWikipediaAbstractThistranslationprojectreportisbasedontheTranslationofRevisionHistory:TranslationTrendsinWikipedia.ThispaperusesBrianMossop"staxonomyofeditingandrevisingprocedurestostudyacorpusoftranslatedarticlesinWikipediaplatformtodeterminehowoftentransferandlanguage/styleproblemsarepresentinthesetranslationsandassesshowtheseproblemsareaddressed.Thisreportconsistsoffourparts:Partonegivesageneralintroductionofthetranslationproject,includingbackgroundandsignificanceoftheproject.PartTwostatesthebackgroundofthesourcetext,includinganintroductiontotheauthor,thetexttypeandlinguisticfeaturesofthesourcetext.PartThreeinvolvestranslationtheory:Newmark’communicativetranslationandsemantictranslation.PartFourcoversthedifficultiesandmethods.Thedifficultieslieinspecializedtermsandterminology.Partfiveconcludeswiththeenlightenmentandunsolvedproblemsinthetranslationproject.Thisprojectisaimedtoprovideresearchingmaterialsintermsofthecrowdsourcingtranslationstudyandlayafoundationforrelatedpracticaltranslation.Keywords:translationproject;crowdsourcing;Wikipedia;revisionhistory;communicativetranslationiii AcknowledgementsFirstandforemost,Iwouldliketoexpressmyheartfeltgratitudetomysupervisor,ProfessorWeiDaopei,bothforhisintellectualguidanceandforhiswarmandconstantencouragementduringtheprocessofwritingthisthesis.Withpatienceandprudence,helaboredthroughdraftsofthisthesisandpointedoutdefectsinmytheorizing.Therefore,Ioweallthemeritsinthisthesis,ifany,tohim,thoughIamfullyawarethatthethesismightstillcontainsomemistakes,forwhichIbearthewholeresponsibility.IamalsodeeplyindebtedtoalltheothertutorsandteachersinTranslationStudiesfortheirdirectandindirecthelptome.Specialthanksshouldgotomyfriendswhohaveputconsiderabletimeandeffortintotheircommentsonthedraft.Lastbutnottheleast,bigthanksgotomyfamilywhohavesharedwithmemyworries,frustrations,andhopefullymyultimatehappinessineventuallyfinishingthisthesis.iv CONTENTS摘要.........................................................................................................................................iiAbstract.......................................................................................................................................iiiAcknowledgments......................................................................................................................ivChapterOneGeneralIntroduction.............................................................................................１1.1BackgroundoftheReport....................................................................................................１1.2SignificanceoftheReport...................................................................................................１1.3StructureoftheReport.........................................................................................................２ChapterTwoAnalysisoftheSourceText.................................................................................４2.1AbouttheAuthor..................................................................................................................４2.2AbouttheSourceText..........................................................................................................４2.3LinguisticFeaturesandTextTypeoftheSourceText.........................................................５2.3.1LinguisticFeaturesoftheSourcetext..............................................................................５2.3.2TextTypeoftheSourceText............................................................................................６ChapterThreeTranslationTheory.............................................................................................７3.1CommunicativeTranslationandSemanticTranslation.......................................................７3.2TranslationTheoryAppliedintheReport...........................................................................８ChapterFourDifficultiesinTranslationandSolutionstothoseDifficulties........................１０4.1DifficultiesinTranslation................................................................................................１０4.2SolutionstoThoseDifficulties........................................................................................１０4.2.1Terminology..................................................................................................................１０4.2.3Division........................................................................................................................１２4.2.4Shifting.........................................................................................................................１３4.2.5Omission.......................................................................................................................１３ChapterFiveConclusion.......................................................................................................１５5.1Enlightenmentfromtheproject.......................................................................................１５5.2Problemsremaintoberesolved.......................................................................................１５Reference...............................................................................................................................１７AppendixI..............................................................................................................................１８AppendixII............................................................................................................................４６v ChapterOneGeneralIntroduction1.1BackgroundoftheReportCrowdsourcingreferstoapracticethatcompaniesororganizationsappliestoassigntheirtaskstoagroupofunspecificpeopleinsteadofstaffmembers.ThosevolunteerworkersareusuallyunpaidandInternetistheirmainworkplace.“Crowdsourcingisdistinguishedfromoutsourcinginthattheworkcancomefromanundefinedpublic(insteadofbeingcommissionedfromaspecific,namedgroup)andinthatcrowdsourcingincludesamixofbottom-upandtop-downprocesses.”(Wikipedia).Theterm“Crowdsourcing”wasfirstpushedoutbyanAmericanjournalistJeffHowefromtheWiredmagazineinJune,2006.Whiletheterm"crowdsourcing"waspopularizedontheInternettodescribeInternet-basedactivities,thereareexamplesofprojectsthatinretrospectcanbedescribedascrowdsourcingandpeoplearenotstrangetoit.Asasubbranchofcrowdsourcing,crowdsourcingtranslationisdevelopedwiththeinternetandtheneedsoffastconsumabletranslation.Still,thetranslationqualitygeneratedfromthismodeandtheimpactputonthetraditionaltranslationindustryarequestioned,butthenotionthatcrowdsourcingwillbeatrendforthetranslationmarketinthefutureisunarguable.1.2SignificanceoftheReportThebenefitsofcrowdsourcingtranslationareeasytounderstand:itishighlyflexibleandfruitful;theworkforceisalwaysreadytowork,andyoucangetasmuchasyouwant.However,italsohaslimitations.First,theparticipantswithdifferentbackgroundmakeittrickytomanageprojectingoodorder.Second,qualityoftranslationworksgeneratedfromcrowdsourcingisoftencontroverted,formostofthetranslatorsareunqualifiedandresultinenormousimpracticable１ works.What’smore,ethicalconcernsarisewhenmostoftheparticipantsarevoluntarilytocontributeandnoremunerationisguaranteed.However,ifrightsofthoseself-selectedvolunteersorpart-timeworkerswereguaranteed,collaborationbetweenprofessionalandnon-professionaltranslatorswouldachieveagreaterresult,includinghighqualitytranslations,trainingforunprofessionaltranslators,andbetteraccesstoinformationforbothofthem.Inall,somewaysshouldbefoundforfurtherstudyingaboutcrowdsourcingtranslationtosettlethoseproblems.AndthispaperwhichstudiesandanalyzesthequalityoftranslationsfromWikipediaplatformwillmakeadentinthoseproblems.Moreover,thispaperwillbeintroducedtoChinesescholarsandindividualswhoareinterestedincrowdsourcingtranslation.Therefore,Chinesetranslationpractitionerswillbemorefamiliarwithcrowdsourcingtranslationanditslatestdevelopmentsoverseas.ThisprojectcanofferinsightsfortheChinesetranslationindustry.AndifadvancedorganizationpatterncanbeappliedproperlyinChina,thismodelwillproducehigh-qualityworksandchangehowthetranslationindustryproduces,revisesanddisseminatestranslation.Furthermore,thesourcetextcontainsmanytranslationdifficultieswhichneedbothpatienceandendeavorforthetranslatortofigureout.Totranslateitwillpolishthetranslationcompetenceandenrichthecomprehensiveknowledgeregardlessofthepossibilityofimperfectdelivery.Thereforethistranslationprojectreportisofsignificance.1.3StructureoftheReportThisreportconsistsoffourparts:PartOnegivesageneralintroductionofthetranslationproject,includingbackgroundandsignificanceoftheproject.PartTwostatesthebackgroundofthesourcetext,includinganintroductiontotheauthor,maincontentsandlinguisticfeaturesofthesourcetext.PartThreeinvolvestranslationtheory:Newmark’communicativetranslationandsemantictranslation.２ PartFourcoversthedifficultiesandmethods.Thedifficultieslieinthetranslationofotherforeignlanguages(apartfromtheEnglishlanguage),specializedtermsandterminology.Themethodsinclude:literaltranslation,liberaltranslation,zerotranslationandtransliterationandaddition,omission,divisionandshiftarethemaintranslationtechniques.Partfiveconcludeswiththelessons,implicationandunsolvedproblemsinthetranslationproject.３ ChapterTwoAnalysisoftheSourceText2.1AbouttheAuthorJulieMcDonoughDolmaya,theauthorofthesourcetext,isafreelancetranslator,reviserandwritergraduatedfromtheUniversityofOttawawithherPhDdegreeinTranslationStudies.In2008,shelaunchedMcDonoughDolmayaLanguageSolutionstomeettheincreasingdemandforhigh-qualitytranslationsintoEnglish.SheisalsoamemberoftheinternationalpanelofAssociatesandtheARTIS(AdvancingResearchinTranslationandInterpretingStudies)Initiative.Shedevotesherselftothestudyofcrowdsourcingtranslation,whichcanbetoldfromherworksoverthepastfewyears,suchasTheEthicsofCrowdsourcingandAnalyzingtheCrowdsourcingModelandItsImpactonPublicPerceptionsofTranslation.2.2AbouttheSourceTextRevisionhistory:TranslationtrendsinWikipediawasfirstpublishedonRoutledgeon30Jul2014,andthenincludedintheVolume8,Issue1,2015ofTranslationStudies.Thispaperisabout8000wordsandnoChineseversionhasbeenpublished.Asastandardacademicpaper,thesourcetextconsistsofthreemainparts:introduction,corpusanalysisandconclusion.Inthefirstpart,theauthorintroducessomebackgroundinformationofcrowdsourcing,Wikipediaandthemethodologyappliedinthestudy.Bydoingso,heelaboratesthesignificanceofstudyingcrowdsourcingtranslationandthereasonswhyWikipediaandMossop’srevisionparametersoftransferandlanguage/styleproblemswerechosenasthefocusofthisarticle.Moreover,the４ specificdefinitionofediting,revisingandrewritingintheparticularenvironmentarealsoexplainedinthispart.Thesecondpartisthecoreofthispaper.Theauthortriestofindoutthetranslationproblemsinthecorpusandanalyseshowthoseproblemsaresolvedthroughtherevisionhistoryofthosetranslatedarticles.Acorpuscontains94translationsarecarefullystudiedinthispartandmanytablesandinstancesarelistedtosupporttheauthor’sidea.Ultimately,thestudyansweredthethreequestionsmentionedinthebeginningofthispaper.EventhoughtheansweronlyappliestoWikipedia,butthisstudyalsooffersinsightsforthetranslationindustry.Andtheauthorbelievesthisnewmodeloftranslationwillbenefitboththeprofessionalandunprofessionaltranslators.2.3LinguisticFeaturesandTextTypeoftheSourceText2.3.1LinguisticFeaturesoftheSourcetextThesourcetextisanacademicreport,whichbelongstotheacademicofsocialsciences.Socialscientificpapersstudyacademicphenomenonandproblemsfromthesocialrealmtoperceiveandreflectthegeneralprincipleandthetruthofthesociety.Comparetonaturalsciencepapers,thelanguageofthesourcetextismoresentimentalwithintuitivethinking,imaginalthinkingandfiguresofspeech.Tobemorespecific,linguisticfeaturesofthesourcetextareaccurate,abstract,concise,well-structuredandnormalized.Accuracymeanstochoosewordandexpressionspreciselyinanarrowbandofactualcases,butinabroadsensethelanguageappliedinthesourcetextisaimedtoreachclearnessinconcept,accuracyinjudgment,andstrictnessinlogical.Abstractionisalsoobviousinthesourcetextforitslanguageisspecializedofthetranslationfield.Concisenessreflectsinrefinedexpressionandnaturalstyle.What’smore,thestructureofthesourcetextiscoherentnotonlybetweensentences,paragraphs,contextsandparts,butalso５ betweenthetypeofwritingandthelinguisticfeatures.Lastbutnotatleast,thesourcetextisnormalized.Academicpaperbelongstoargumentation,sothesourcetextconformstotheformattingandlogicalstructureofargumentation.2.3.2TextTypeoftheSourceTextAccordingtoPeterNewmark,texttypesaremainlydividedintothreecategories,namelyexpressivetext,informativetextandvocativetext.Literatureworks,authoritativestatements,autobiography,prosesandprivatelattersbelongstoexpressivetext;textsfocusontechnology，economicandindustryandcommerceareinformativetype;vocativetextsaretryingtoinfectreadersandgainalltheinformationtheyneeded.Forthisreason，thesourcetextisantypicalinformativetextwhichaimstotransmitinformationandreflectfactsobjectivelyanditslanguageislessemotionalwithgeneralsentencepatternandcollocation.“Thetargettextofaninformativetextshouldtransmitthefullreferentialorconceptualcontentofthesourcetext.Thetranslationshouldbein‘plainprose’,withoutredundancyandwiththeuseofexplicitationwhenrequired”,saidMunday.６ ChapterThreeTranslationTheory3.1CommunicativeTranslationandSemanticTranslation“Expressivefunction,informativefunctionandvocativefunction”arethethreelinguisticfunctionsputforwardbyBulerandJacobson.Basedonthethreefunctions,PeterNewmarkcameupwithcommunicativetranslationandsemantictranslationinhisworksApproachestoTranslationwhichwasfirstpublishedin1981.Semantictranslation“attemptstoreadersascloselyasthesemanticandsyntacticstructuresofthesecondlanguageallows,theexactcontextualmeaningoftheoriginal.”(Newmark,1981)Sosemantictranslationismoreobjectiveandthetranslationformisclosertotheoriginalform.Communicativetranslation“attemptstoreduceonitsreadersaneffectascloseaspossibletothatobtainedonthereadersoftheoriginal.”(Newmark,1981)Inotherwords,thefocusofcommunicativetranslationistotransferinformationaccordingtolanguage,cultureandpragmaticformsofthetargetlanguage,ratherthantrytofaithfullyreproducetheoriginaltext.Translatorscanexplaintheoriginaltextfreelytosomedegreeundertheguildofthistheorybyadjustingstyle,eliminatingambiguity,evencorrectingmistakesmadebytheoriginalauthor.Toachieveacertaincommunicativepurposewithintargetedreaders,thetranslatorissupposedtobreakthelimitationsoftheoriginaltext.Usuallycommunicativetranslationtheoryisadoptedtoguildthetranslationofnon-literatureworkssuchasnewsreports,textbooksandannouncements.Itisworthnotingthatcommunicativetranslationisnotanextremetranslationstrategy,nordiditfreeorinflexible,butacompromisedway.Translationsgeneratedfromcommunicativetranslationareusuallysmooth,clear,direct,andconformtothespecificregistercategory.Under-translationoftenoccurswithcommunicativetranslation,inotherwords,generalvocabularyaretendtoreplacedifficultdiscourseforthesakeofreader’sunderstanding.７ Althoughdifferencesbetweencommunicativeandsemantictranslationaregreatandeachonehasitspeculiarity,weshouldstilltakethemasawhole.Therefore,in1994Newmarkcameupwithanewtheorywhichcombinesbothcommunicativetranslationandsemantictranslation.Anditiscalledacorrelativeapproachtotranslationwhichwasdefinedas“themoreimportantlanguageoftheoriginalorsourcelanguagetext,themorecloselyitshouldbetranslated.”3.2TranslationTheoryAppliedintheReportAsamacroprinciple,translationstrategyshouldbecloselylinkedtoanyonewhoinvolvedinthetranslationproject,namelythesource-languagewriter,thetranslatorandtargetreaders.Asthetranslatormentionedabove,thesourcetextisatypicalinformativetextandreader-oriented,communicativetranslationshouldbeadoptedasthemaintheoryofthisproject.Communicativetranslationtakesreadersasthecenter,andemphasizeonthereadabilityandacceptabilityofthelanguage.Therefore,inpracticethetranslatorshiftfocustothemessageofthesourcetext.Thatistosay,theoriginalcontentisrepresentedintheprocessoftranslationandthefinalversion.Tomaintainthisobject,thetranslatorcanmakefulluseofthetargetedlanguagetoreachthesameeffectthattheoriginallanguagebringstothereadersandnotboundbytheoriginalform.JustlikeSchleiermachersaid:“Leavethereaderinpeaceasmuchaspossibleandbringthewritertohim”.AccordingtoNewmark,Domesticationshouldbeadoptedasthemainmethodofinformativetextforitsgoalistotransmitinformation.However,inpractice,wecannotseparatedomesticationfromforeignization.Translationprincipledemandsthetranslatortofaithfullyreproducethewriter’sviewandstylewhichareexotic,thereforeforeignizationmustbeapplied.But,ontheotherhand,readers’understandingandthefluencyofthesourcetextshouldalsobetakenintoconsideration,sousingdomesticationisinevitable.Itisunrealistictoselectoneof８ them,anddiscardtheother.Thetranslatortriestomakefulluseoftheiradvantagesbycombinethetwomethodsintheproject.９ ChapterFourDifficultiesinTranslationandSolutionstothoseDifficulties4.1DifficultiesinTranslationAcademicwritingisacomplexandprudentworkandtheauthorneedstoexplainanddescribeintricatephenomenonmeticulouslyfrommanyaspects.Thelongsentencescomposedofseveralclausesbetterelaboratetheauthor’sideaandavoidambiguity.Forthisreason,longsentencesareappliedwidelyinthisarticle.AndwithdifferentsentencelinguistichabitsbetweenEnglishandChinese,longsentencesaretroubleforalltranslatorsallthetime.Excessivemodifiers,parallelcompositionandlanguagehierarchyareusuallywheretroublelies.Anddivision,combination,inversionarefrequentlyusedtosolvethoseproblems.Thesourcetextalsocontainsalotofspecializedterms.Andmostofthemarenotfrequentlyusedinourdailylife.Theymaylookfamiliartous,butalwayshaveunderlyingmeanings.Therefore,thosetermsmaybeconfusionandmisleadthetranslator.But,throughcarefullyverifyingandstudying,thisdifficultymaybeproperlysolved.NowthatthepaperonlychoosearticlesdescribesasbeingoriginalwritteninFrenchorSpanish,andcasesinthosetwolanguagesarepresentedinthispaper.WhetherthoseFrenchandSpanishshouldbetranslatedintoChinesearedifficulttodecide.Circumstancesaltercasesandthisdifficultywillbesolvedinthisproject.4.2SolutionstoThoseDifficulties4.2.1TerminologyExample1:Wikipediaisawell-knownexampleofawebsitewithcontentdevelopedentirelythroughcrowdsourcing.Ithasover4millionarticlesinEnglishalone,andcontentin284otherlanguageversions.１０ 维基百科是典型的众包案例，英语版已收录400多万个条目，此外还有284种语言版本。Thisexampleappearsinthebeginningoftheabstractofthepaper,whichgivesageneralideaoftheWikipedia.Normally,“article”referstoapieceofwritingaboutaparticularsubjectinanewspaperormagazine,butinthisspecialcontextitmeansthebasicunitwhichonlyconcernsonetheme/people/thingoracombinationofaparticulartopic.Therefore,thetranslatortracedtheexactcorresponding“条目”.Andthiswayofexpressionismoreprofessionalandconformstotheprecisenessstandardoftheacademicpaper.Suchcasesarenumerousinthisproject.Usuallythoseterminologiesresemblethesameappearanceofordinaryword,sotheymaybemistranslatedintoageneralsensewithoutnoticingitsspecialmeaning.Hence,thetranslatorpaysmoreattentiontothosesimplewordsandcheckstheirmeaningcarefully,soastomakethetranslationmoreauthenticandaccurate.4.2.2NominalizationExample2:Articleswith15orfewereditshadthelowestpercentageofresolvedproblems,andingeneral,asthenumberofeditsincreased,sodidthepercentageoferrorsthatwereresolved.编辑次数小于或等于15次的文章错误解决率最低。通常来说，编辑次数越多，错误解决率就越高。AsWikipediadefines:“Nominalizationisatypeofwordformationinwhichaverboranadjective(orotherpartofspeech)isusedas(ortransformedinto)anoun,orastheheadofanounphrase,withorwithoutmorphologicaltransformation.”Thissentencedescribessomefactsshowinginthetableand“thepercentageoferrorsthatwereresolved”willbementionedinthefollowingtextfordataanalysis.SothetranslatorconvertsitintoashortnounformwhichismorepracticeforcorpusstudyandmoreaccordwithChineseexpression.NominalizationisnotfrequentlyusedinChineseforthelanguageis１１ parataxis-prominent.Inotherwords,therelationshipamongwordsandsentencesareinvisibleinChinese.However,academicpaperpaysmoreattentiontoconciseness.Onthepremiseofaccuratelytransferthemessage,thetranslatortrieshardtomakethelanguageaccordwiththeacademiclanguage.And,inthisspecialcontext,thetranslatorchoosestonominalizethesentenceintoashortphrase.Evenisnotaconventionalway,butconcreteanalysisshouldbeappliedinconcreteconditions.4.2.3DivisionExample3:Translationperformedonlinebyself-selectinginternetusers,whoareoftennotprofessionaltranslatorsbutwhoarecollaboratingtocompletelargeprojectsfor(virtually)nofinancialremuneration,hasincreasinglybeenatopicofinterestwithintranslationstudies,whetherthispracticehasbeenstudiedunderthenamecollaborativetranslation(DésiletsandvanderMeer2011),communitytranslation(Kelly,Ray,andDePalma2011),crowdsourcing(Cronin2010;McDonoughDolmaya2012;Mesipuu2012),fantranslation/subbing/dubbing(Dwyer2012;O"Hagan2008),user-generatedtranslation(O"Hagan2009;Perrino2009),volunteertranslation(O"BrienandSchäler2010;Olohan2014),wiki-translation(Cronin2010)orsomecombinationoftheseterms.自发从事网络翻译的互联网用户，往往都不是专业翻译，而与人合作完成大型翻译项目也（几乎）没有报酬。由此产生的译文日益成为翻译界探究的话题，并且这类合作方式名目繁多：合作翻译(DésiletsandvanderMeer2011)、社区翻译(Kelly,Ray,andDePalma2011)、众包翻译(Cronin2010;McDonoughDolmaya2012;Mesipuu2012)、粉丝翻译/词幕/译制(Dwyer2012;O"Hagan2008)、用户生成翻译(O"Hagan2009;Perrino2009)、志愿翻译(O"BrienandSchäler2010;Olohan2014)，维基翻译(Cronin2010)等等。诸如此类合作模式下的翻译一直在研究中。Ingeneral,thecommonelementsoflongsentencesareplentyofappositives,attributivesandclauseswhichseparatethesubjectandthepredicate.Thegeneralproceduresaretosimplifythesentencebyfindingoutthebackboneofit,andthenstraightenouttherelationshipwithclausesormodifiers.ThisisthefirstsentenceoftheIntroductionpartandanexplanationofcrowdsourcing.Beingdescriptiveandeasytounderstand,ithastwoclausesqualifyingthe“self-selectinginternetusers”,while“translation”isthesubjectofthemainclause.Consideringthis,１２ literaltranslationanddivisionisadopted.ThetranslatorsplititintotwosentencesinChinese,sothereadercanunderstandthemeaningeasily.AndthisapproachalsoconformstotheChineseexpression:torealizetheconnectionofthewordsorphraseswiththehelpoftheirinnermeaninginsteadoflanguageformsinEnglish.4.2.4ShiftingExample4:ThispaperusesMossop’staxonomyofeditingandrevisingprocedurestoexploreacorpusoftranslatedWikipediaarticlestodeterminehowoftentransferandlanguage/styleproblemsarepresentinthesetranslationsandassesshowtheseproblemsareaddressed.本文采用莫索普的编辑与修订程序分类法，对维基译文汇编而成的语料库进行研究，以确定转换错误和语言/排版错误在译文中出现的频率，并评估这些错误的解决方式。Theselectedsentenceisthepurposeandthefocusofthestudy.Syntactically,“determine”introducestwoadverbialclauseswhichwasconvertedintoanounandaphraserespectivelybythetranslator.Moreover,thetranslatoremployslexical-syntaxshiftasthetranslationtechnique.Shiftingreferstotheprocessofconvertingthelanguageunitsorstructureofthesourcetextintocorrespondinglanguageunitsorstructurewhichresemblesthesimilarorcorrespondingnatureandisinvolvedwithmanyaspectsoflinguistics,suchasrhetoric,semantics,pragmaticsandculture.4.2.5OmissionExample5:“Navahermosa”,SpanishWikipedia(23July2012)AntonioLlorente,alcomentarlafrecuenciadeestetérminoenlazonadeSalamanca,ÁvilaylaparteoccidentaldeSegovia,sitúasuorigenenenlaslenguasdevetonesycarpetanos,aunqueplantealadudadesuorigenvascoenelqueexistelavoznaba“llanura,valle,barranco”.１３ Omissionmeansdonothingtothesourcetextinwhichsomecomponentsaredirectlyintroducedtothetargettext.Inthisexample,thetranslatorchosestoputthesourcelanguagedirectlyinSpanishfortworeasons:first,Spanishisbeyondthetranslator’knowledge;second,themainlyfocusofthisexampleistoshowthatarticlescontainedmanynearlyincomprehensiblepassageswaspossiblygeneratedbyautomatictranslationsoftware.Therefore,therewillnotbeanobstaclefortargetreaders.Therearealsomanyinterlinearnotesaftersomespecialtermssuchas“DésiletsandvanderMeer2011”,“Kelly,Ray,andDePalma2011”andsoon.Inthosenotes,onlytimesandauthorsarestated.Theyarespecialistinthisfilebutnotfamiliartopublic.Forthereasonthetargetreadersofthisprojectarehighlyeducatedorfamiliarwithprofessionalknowledge,thereisnoneedtotranslatethosenotesintoChinese.Duringthisprocess,zerotranslationisputtouseforeasytracetothesource.１４ ChapterFiveConclusion5.1EnlightenmentfromtheprojectTherelationshipbetweentranslationtheoriesandpracticehasbeenunderfiercedebatesincethetranslationstudieshasbeenestablished.Andamongwhichwhethertranslationtheoriescanprovideguidancefortranslationpracticearemostcontroversy.Howeverthisprojectalreadyprovedthattranslationtheoriescannotonlyguidethepractice,butalsotestthetranslation,whichwillhelpthetranslatortounderstandthepracticebetterandimprovetranslationcapability.What’smore,wordslikecrowdsourcingappearsforthefirsttimeinfrontofthetranslator,yetherlifehasalreadybeenfilledwiththispattern,onlyshedoesnotrealize.Thus,thisprojectcallsforattentiontheworldaroundus.Tobeaprofessionaltranslator,thereisstillalongwaytogo,suchasbroadeningoftherangeandscopeofourgeneralknowledge,andTogetabetterunderstandingoftheimportanceingeneralknowledge,thetranslator’sknowledgeofthesubjectmattercannotbeoveremphasized,anditcanalmostbeacceptedasarulethat,giventherequiredlanguageproficiency,thegreaterthetranslator’sknowledgeofthesubjectmatter,thelessarduouslyunderstandingmaycause.5.2ProblemsremaintoberesolvedTherestillexistsomeproblems.Firstly,asalaymanofcrowdsourcingstudy,Iamnotfamiliarwithsomeconceptsandprincipleswithinthisstudy.Hence,sometermsornamesmaybeimproperlytranslated.Secondly,asanacademicreport,thewritingstyleofthisbookisrigorousandprecise.Duetotheinsufficienttranslationpracticesaboutthiskindoftext,sometranslationsinmyworkmightseemawkwardandcolloquial.Thirdly,thelanguageofthefinalversionisnotperfectandstillcontainsmanymistakesandimproperexpression.Sotherehasmuchroom１５ forthetranslatortoimprove.Thetranslatorhopestomastermorepracticaltechniquesconcerningthiskindoftextandbemoreconfidentandwell-preparedwhendealingthesametypenexttime.Thissituationcannotbechangedimmediately,butthetranslatorwillkeeplearningwithanopeneyefornewknowledgeandkeeppracticeintranslation.Andgivenenoughtime,thetranslator’sskilldealingwithsuchproblemswillbeimproved.１６ ReferenceHoward,D.2014.“ThoughtsonWikipediaEditingandDigitalLabour.”Howe,J.2008.Crowdsourcing:WhythePoweroftheCrowdisDrivingtheFutureofBusiness.NewYork:CrownBusiness.HoweJ.Theriseofcrowdsourcing.WiredMagazine,Issue14.06,2006.McDonoughD.J.(2011).TheEthicsofCrowdsourcing.LinguisticaAntverpiensia,10:97-111.McDonoughD.J.(2012).AnalyzingtheCrowdsourcingModelandItsImpactonPublicPerceptionsofTranslation.Translator18(2):167-191.McDonoughD.J.(2015).RevisionHistory:TranslationTrendsinWikipedia.TranslationStudies8(1):16-34.Mossop,B.(2006).RevisingandEditingforTranslators.2nded.Manchester:StJerome.Newmark,P.(1987).ATextbookofTranslation.London:PrenticeHallLongmanELT.Newmark,P.(1988).Approachestotranslation.Oxford:PergamonPress.Wikipedia.2013a.“Wikipedia:ManualofStyle.”AccessedJuly3.Wikipedia.2013b.“Help:MinorEdit”AccessedJuly31.Zaidan,OmarF.,andChrisCallison-Burch.2011.“CrowdsourcingTranslation:ProfessionalthQualityfromNon-Professionals.”InVol.2ofHLT’11Proceedingsofthe49AnnualMeetingoftheAssociationforComputationalLinguistics:HumanLanguagetechnologies,1220_1229.http://www.cs.jhu.edu/ccb/publications/crowdsourcing_traslation.pdf.段建勇,闫启伟,张梅,胡熠.维基百科中翻译对的模板挖掘方法研究.中文信息学报,2015(2):190-200.陆艳.众包翻译模式研究.上海翻译2012(3):74-79.雄兵.翻译研究中的概念混淆.外语与翻译,2013(4):76-79.１７ AppendixIRevisionhistory:TranslationtrendsinWikipediaJulieMcDonoughDolmayaSchoolofTranslation,YorkUniversity(GlendonCampus),Toronto,CanadaWikipediaisawell-knownexampleofawebsitewithcontentdevelopedentirelythroughcrowdsourcing.Ithasover4millionarticlesinEnglishalone,andcontentin284otherlanguageversions.Whilethearticlesinthedifferentversionsareoftenwrittendirectlyintherespectivetarget-language,translationsalsotakeplace.GiventhatapreviousstudysuggestedthatmanyofEnglishWikipedia"stranslatorshadneitherformaltrainingintranslationnorprofessionalworkexperienceastranslators,itisworthexaminingthequalityofthetranslationsproduced.ThispaperusesMossop’staxonomyofeditingandrevisingprocedurestoexploreacorpusoftranslatedWikipediaarticlestodeterminehowoftentransferandlanguage/styleproblemsarepresentinthesetranslationsandassesshowtheseproblemsareaddressed.Keywords:crowdsourcedtranslation;Wikipedia;revision;non-professionaltranslation;user-generatedtranslation;volunteertranslationIntroductionAboutcrowdsourcingTranslationperformedonlinebyself-selectinginternetusers,whoareoftennotprofessionaltranslatorsbutwhoarecollaboratingtocompletelargeprojectsfor(virtually)nofinancialremuneration,hasincreasinglybeenatopicofinterestwithintranslationstudies,whetherthispracticehasbeenstudiedunderthenamecollaborativetranslation(DésiletsandvanderMeer2011),communitytranslation(Kelly,Ray,andDePalma2011),crowdsourcing(Cronin2010;McDonoughDolmaya2012;Mesipuu2012),fantranslation/subbing/dubbing１８ (Dwyer2012;O’Hagan2008),user-generatedtranslation(O’Hagan2009;Perrino2009),volunteertranslation(O’BrienandSchäler2010;Olohan2014),wiki-translation(Cronin2010)orsomecombinationoftheseterms.Inparticular,recentresearchhasexaminedthemotivationsofthosewhoparticipateincrowdsourcedinitiatives(e.g.Olohan2014;McDonoughDolmaya2012;Mesipuu2012;O’BrienandSchäler2010),theethicsoftranslatinginsuchenvironments(e.g.McDonoughDolmaya2011;Drugan2011)andthepedagogicalpotentialofcrowdsourcedtranslationorrevisionprojects(e.g.O’Hagan2008;BrunetteandGagnon2013).However,despitethegenerallypositivediscoursefromthecompaniesthatrelyoncrowdsourcing1andHowe’scontentionthat“thebestpersontodoajobistheonewhomostwantstodothatjob[,]andthebestpeopletoevaluatetheirperformancearetheirfriendsandpeerswho[...]willenthusiasticallypitchintoimprovethefinalproduct”(2008,8),thequalityoftranslationsproducedviathismethodhasnotbeenthesubjectofmuchresearch,particularlywithintranslationstudies,althoughZaidanandCallison-Burch(2011),forinstance,havestudiedtheissuefromacomputationallinguisticsandcomputerscienceperspective.Withintranslationstudies,Jiménez-Crespo(2011)hasexaminedquality-controlmechanismsinFacebook,concludingthatFacebook’smethodofevaluatingtranslations“inwhichalargenumberofusersvotespositivelyornegativelyonproposedtranslations,canhelpguaranteethattheresultingwebsitecompliestotheexpectedfeaturesinadigitalgenrethatanydiscoursecommunitymighthave”(147).Althoughhedidnotevaluatethetranslationsatthattime,Jiménez-Crespo(2013)latercomparedthePeninsularSpanishversionofFacebookwiththeinterfacesof25socialnetworkingwebsitesoriginallyproducedinPeninsularSpanish,lookingspecificallyatverbalvs.nominalforms,imperativevs.infinitiveverbforms,andlexicalunitsusedfortheinteractivefeatures(e.g.log-out,register,editprofile).Interestingly,hefoundthattheFacebooklocalizationmodelusedlanguagethatwassimilartothatofthewebsitesoriginallyproducedinSpanish,unlikeinhispreviousstudiesofwebsitesprofessionallylocalizedintoSpanish,wherethereweresignificantdifferencesbetweenverbtensesandlexicalitemsinthelocalizedandnon-localizedwebsites.Thissuggeststhatinsomecontexts,acrowdsourcingmodelcanproducetranslationsthataresimilartooriginaltexts,thoughofcoursealargerstudy,encompassingseveralplatformsandadditionallanguagepairs,would１９ helptoshowwhetherthisfindingholdstrueforcrowdsourcingmoregenerally.Clearly,morecriticalanalysisofcrowdsourcedtranslationsisneeded.Ifthemodelcanproducehigh-qualitytranslations,itcouldchangehowthetranslationindustryproduces,revisesanddisseminatestranslations.However,crowdsourcedtranslationprojectsusedifferentmethodstorecruit,motivateandrewardusersandtocontrolquality(cf.Mesipuu2012;McDonoughDolmaya2011).Moreover,translation“quality”isarelativeconcept(cf.e.g.Jiménez-Crespo2011,133-135)soitwouldbedifficulttodesignastudyintendedtogenerallyassesseithercrowdsourcedtranslationsortheeffectivenessofthecrowdsourcedtranslationprocess.Thisarticlethereforetacklestheissuebyexaminingonecrowdsourcedtranslationproject-Wikipedia-andfocusingononecriterionoftranslationqualitycontrol-revision-whileacknowledgingthattheresultscannotbegeneralizedbeyondWikipedia’splatform.UsingMossop’s(2006)taxonomyofeditingandrevisingprocedures,thisarticleanalysesacorpusof94translatedWikipediaarticles.ItaskswhetherWikipedia"sunpaid,oftennon-professionaltranslators(cf.McDonoughDolmaya2012)areproducingtranslationsthatrequirerevision,whethertheopen-editingprocessofthewikienvironmenthelpseliminateerrorsinthetranslatedarticles,whethercertaintypesoferrorsaremorecommonlyresolved,andwhetherWikipedia,asatranslatinginstitution,isaviablemodelfortheprofessionaltranslationindustry.AboutWikipediaWikipediaisperhapsthebest-knownexampleofawebsitewhosecontenthasbeendevelopedentirelythroughcrowdsourcing.LaunchedinJanuary2001usingawikiengine,whichallowsanyinternetusertoadd,deleteoreditcontentviaawebbrowser,Wikipediaisnowoneofthemostwidelyknownwebsitesintheworld.InJune2013,accordingtointernetanalyticscompanyAlexa,Wikipediawasrankedtheseventhmostvisitedwebsiteintheworld,behindFacebook,Google,YouTube,Yahoo!,BaiduandAmazon,butaheadofsitessuchasTwitter,LinkedIn,BingandBlogspot.Atthattime,Wikipediaversionsexistedin285languages,8ofwhichcontainedmorethan1millionarticleseach,withEnglish,thelargest,containingover4million.Whilethecontentoftheseversionsisoftenwrittendirectlyintherespectivetargetlanguage(TL),translationsalsotakeplace,asthe２０ encyclopedianotes:Wikipediaisamultilingualproject.Articlesonthesamesubjectindifferentlanguagescanbeeditedindependently;theydonothavetobetranslationsofoneanotherorcorrespondcloselyinform,styleorcontent.Still,translationisoftenusefultospreadinformationbetweenarticlesindifferentlanguages.(Wikipedia2013a)InWikipedia,articletranslation,likearticlecreationandediting,isuser-driven.Articlesarenotchosenfortranslationbytheorganization,butratherattheinitiativeofanyinternetuser.Wikipedianscanselectanarticle,translateallorpartofitintoanotherlanguageandthenpostthetranslationintheappropriatelanguageversionofWikipedia;theycanalsonominatearticlesfortranslationviaoneofthemanyongoingtranslationprojects,including“TranslationoftheWeek”(http://meta.wikimedia.org/wiki/Translation_of_the_week)orthe“TranslationTaskforce”(http://en.wikipedia.org/wiki/Wikipedia:MED/Translation_project).Theopennatureofthetranslationprocess,alongwiththefactthateverychangemadetoeveryexistingarticleisarchivedandeveryusercanjustifythesechangesoneitherthe“RevisionHistory”orthe“Talk”pages,isbothastrengthandaweaknessforresearchpurposes.Ontheonehand,itprovidesawiderangeofarticlesinmanylanguagepairsforanalysis.Ontheother,itmakescomparisonsoftranslationacrosslanguagesdifficult,sincethecontentchosenfortranslationineachlanguagepairmaydifferconsiderably,andsincesomelanguageshavemoreactiveusersandthereforemoretranslationactivity.Moreover,becausesomanydifferenttranslationprojectsaretakingplacewithinthevariousWikipediasatanygiventime,gettingaclearideaofexactlyhowmuchtranslationisoccurring,whatlanguagepairsarethemostcommonandwhattopicsaremostoftenchosenfortranslationisdifficult.ThispaperthereforefocusesontranslationfromFrenchandSpanishintoEnglishinonetranslationproject(“Wikipedia:PagesNeedingTranslationintoEnglish”),knowingthattheconclusionsitreachesmaynotbeapplicabletoalltranslationswithinWikipedia,evenforthesamelanguagepairs.Wikipediawasselectedasthefocusofthisstudyforseveralreasons.First,thelongevityoftheencyclopedia,coupledwiththeplatform’sabilitytotrackeveryeditmadebyeveryuser,meansthattheevolutionofatranslationcanbetracked.Second,itstranslationandstyleguidelinesarefreelyavailableonline.Third,professionaltranslatorsprobablycompriseonlya２１ smallpercentageofWikipedia’stranslators(McDonoughDolmaya2012)andthecollaborativeprocessinWikipediaisentirelyuser-driven,withouttheoversightfromprofessionaltranslatorsadoptedbyothercrowdsourcedprojects,suchasFacebook(Jiménez-Crespo2013,30).ThismeansthatastudyofWikipediatranslationpracticescanprovideinsightintowhetheralargecommunityofgenerallynon-professionaltranslatorscansuccessfullyidentifyandrevisetranslationerrorssolelythroughtheirownefforts.AboutthemethodologyMossop’s(2006)RevisingandEditingforTranslatorsprovidesausefulframeworkforstudyingtherevisioneffortsofWikipediatranslators.First,thebookislargelydescriptive,ratherthanprescriptive(ibid.,4),andwhiletherevisionsectionsinparticularareintendedtobe“universallyvalid”insofarasthatispossible(ibid.,7),Englishisheavilyemphasizedthroughoutthebook,whichishelpful,sincethisarticleaimstodescribeproblemsinEnglishtranslations.Second,Mossopdescribes12“revisionparameters”,groupedinto4broadcategoriesoferrorsthatrevisersareexpectedtoaddress(ibid.,125):transferproblems,whichinvolvecomparingthesourceandtargettextstodeterminehowaccurateandcompletethetranslationis;contentproblems,whichinvolvefindingfactualerrors,nonsense,contradictionsanderrorsinlogic;languageandstyleproblems,whichinvolvecheckingforsmoothness,tailoringthetranslationtotheTLreaders,checkingforunidiomaticlanguage,eliminatinggrammaticalandspellingerrorsandensuringthetranslationconformstothehousestyleguide;andpresentationproblems,whichinvolvecheckingforlayout,formattingandorganizationproblems.Usingjustfourcategoriessimplifiestheanalysis.Spaceandtimeconsiderations,however,meantthatnotallofMossop"srevisioncategoriescouldbestudied.Indeed,asMossopargues(2006,141-142),professionalreviserswillgenerallyneedtoprioritizecertainparameters,dependingonwhowillbereadingthetext(e.g.expertsornon-experts),whythetextwillbereadandforhowlong,whetherthetextwillbeskimmedorreadverycarefully,andwhetheranyoneelsewillbequalitycontrollingthetext.SinceWikipediaisoneofthemostvisitedwebsitesintheworld,articleswillbereadbybothexpertsandnon-experts,contentcouldbeavailableforyears(barringdeletionbyanadminforpolicyviolations)andarticleswillbeskimmedbysomeusersbutstudiedcarefullybyothers.２２ Thismeanscontentproblems-particularlyfactualerrors-areanimportantconsideration,asarelanguageandstyleproblemssuchassmoothnessoftextandtérminologyerrors.ContentproblemsinWikipediahavebeenstudiedelsewhere,mostnotablybyGiles(2005).Transferandlanguage/styleproblemswerechosenasthefocusofthisarticleforthefollowingreasons:transferproblemsaffectaccuracyandcompleteness,arguablythemostimportantfeaturesoftranslation(Mossop2006,126),buttheyhavenotbeenthefocusofotherWikipediastudies.TransferproblemsarealsochallengingfortranslatorstoidentifywithinWikipedia"splatform,sincedoingsoinvolvesmeticulouslycomparingthesourceandtargettextversions,whichtypicallyexistonseparatepages.Bycontrast,languageandstyleproblemsareeasierforWikipedians-particularlythosewithlimitedornoknowledgeofthesourcelanguage-toidentifyandcorrect.Indeed,suchproblemsareasignificantconcernofWikipediaeditors,giventhecreationofgroupssuchastheGuildofCopyEditors2andthedevelopmentofguidelinessuchasWikipedia"s“ManualofStyle”(cf.Wikipedia2013b);noequivalentguidelinescouldbefoundfortransfer-relatedproblems.ContrastingthesetworevisionparametersshouldthereforehelphighlightthestrengthsandweaknessesofWikipedia’splatformforcrowdsourcedtranslation.Tostudytransferandlanguage/styleproblemsintranslatedWikipediaarticles,acorpusofarticlestranslatedinwholeorinpartfromFrenchorSpanishintoEnglishwithinthelasttwoyearswascompiledbyconsultingrecentandpreviousversionsofthe“Wikipedia:PagesNeedingTranslationintoEnglish”page,http://en.wikipedia.org/wiki/Wikipedia:Pages_needing_translation_into_English.Onthatpageisaregularlyupdatedlistsplitintotwocategories:the“translate”section,whichcontainsarticlesthatWikipedianshaveidentifiedashavingcontentinalanguageotherthanEnglishdespitebeingpostedonWikipedia"sEnglishversion,andthe“cleanup”section,whichcontainsarticles(presumably)translatedintoEnglishbutrequiringpost-editing,usuallybecausethetranslationwasdonebymachinetranslationsoftware,anon-nativeEnglishspeakerorsomeonewhodidnothaveagoodgraspofthesourcelanguage.3Althoughthearticleslistedon“Wikipedia:PagesNeedingTranslationintoEnglish”comefromanyofWikipedia"s284otherlanguageversions,onlythosedescribedasbeingoriginallywritteninFrenchorSpanishwerechosenforanalysis.２３ Tocompilethecorpus,the15May2013versionof“Wikipedia:PagesNeedingTranslationintoEnglish,alongwiththearchivedversion(“PageHistory”)onthefifteenthofeverymonthbetween15May2011and15April2013,wasconsulted,bringingthesampleperiodtoatotaloftwoyears.Inthattime,thenumberofarticlesofFrenchoriginlistedineitherthe“translate”or“clean-up”sectionsofthearchivedpagestotalled34,whilethetotalnumberofarticlesofSpanishoriginlistedwas60.Doesthismeanonly94WikipediaarticlesweretranslatedfromFrenchorSpanishintoEnglishbetweenMay2011andMay2013?Unlikely:thearticleslistedonthispagewereidentifiedbyWikipediansasbeing“roughtranslations”orinalanguageotherthanEnglish.Butmanyothertranslatedarticlescouldhavebeencreatedorexpandedduringthistime.Other“rough”translationsmaynothavebeenspottedbyWikipediausers,whiletranslationswithoutthegrammaticalerrorsandincorrectsyntaxassociatedwithnon-nativeEnglishcouldhavepassedunnoticedbyWikipedianswhomighthaveotherwiseaddedthetranslationtothispage.SowhilethissamplegroupofarticlesisprobablynotrepresentativeofalltranslationswithinWikipedia(orevenofFrench/Spanish-to-Englishtranslationinparticular),“Wikipedia:PagesNeedingTranslationintoEnglish”wasstillagoodsourcefromwhichtodrawasampleoftranslatedarticlesthatmayhaveundergonesomesortofrevisionorediting.Eveniftheresultsarenotgeneralizable,theyatleastindicatethekindsofchangesmadetotranslatedarticleswithintheWikipediaenvironment,andtherefore,whetherthisparticularcrowdsourcingmodelisaneffectivewaytotranslate.Aboutediting,revisingandrewritinginWikipediaDiscussingeditingandrevisinginrelationtoWikipediaissomewhatchallenging,sinceWikipediausesthetermseditandeditingdifferentlyfromtranslationstudiesliterature,distinguishingonlybetweenmajorandminoredits,and,somewhatconfusingly,usingtheterm“revisionhistory”asthetitleofthepagethatshowsthehistoryoftheseedits:4[Minoredits]includetypographicalcorrections,formattingandpresentationalchanges,andrearrangementsoftextwithoutmodificationofitscontent.Aminoreditisonethattheeditorbelievesrequiresnoreviewandcouldneverbethesubjectofadispute[...].Bycontrast,amajoreditisonethatshouldbereviewedforitsacceptabilitytoall２４ concernededitors.Therefore,anychangethataffectsthemeaningofanarticleisnotminor,eveniftheeditconcernsasingleword;forexample,theadditionorremovalof“not”,whichcanchangethemeaningofasentence,isamajoredit.(Wikipedia2013c;originalemphasis)Bycontrast,Mossop(2006)distinguishesbetweenrevisingandeditingbynotingthattheformerinvolveschecking“adrafttranslationforerrorsandmakingappropriateamendments”(201),whilethelatterinvolvesdoingthesamewith“anon-transla-tionaltext”(198).Thisdistinctionisalsomadebyothers(e.g.BrunetteandGagnon2013).AccordingtoMossop’staxonomy,editorsdealwithtextsasnon-translationobjectsandfocusoncopyediting,orensuringatextconformstospelling,grammarandmechanicsguidelinescommonlyrecognizedbyalanguagecommunity;stylisticediting,ortailoringlexisandsentencestructuretoreadersbytryingtomakeatextmoreconcise,lessambiguous,andsoon;structuralediting,orreorganizingatexttobetterpresentthematerial;andcontentediting,orsuggestingareaswheretextmaybeaddedordeletedtobettercoverthetopicathand(Mossop2006,27-28).Ofcourse,thetasksofaneditorandrevisercanoverlap:copy-editingandrevisingforlanguageandstyleproblems,forinstance,involvemanyofthesametasks.Moreover,asinglepersoncansometimesberesponsibleforbotheditingandrevising-addressingtransferproblemsononehand,andrestructuringthetextforthetargetaudienceontheother.ThisisthecasewithinWikipedia,asthetranslationguidelinesindicate,andasthepagehistoryofthecorpusarticlesattests.The“Wikipedia:Translation”page,forinstance,providesthefollowingguidelinestowould-betranslators:TranslationbetweenWikipediasneednottransferallcontentfromanygivenarticle.Ifcertainportionsofanarticleappeartobelow-qualityorunverifiable,useyourjudgmentanddonottranslatethiscontent.Onceyouhavefinishedtranslating,youmayaskaproofreadertocheckthetranslation.(Wikipedia2013a)Assuggested,then,translators(andrevisers,sincetheywouldbefollowingthesameguidelines)mayengageincontentediting,sincetheycanomit“low-quality”or“unverifiable”source-text(ST)segmentsfromthetranslation.This,inturn,makesrevisingfortransferproblemschallenging,sincetranslatorsandrevisersmayhavedifferentopinionsaboutwhetherthetranslationiscompletewhenpartoftheSThasbeenomittedfromthetranslation.Notealso２５ thatalthoughrevisionispossibleinWikipedia,sincetranslators“mayaskaproofreadertocheckthetranslation”,itisneitherobligatory,sincetranslatorsdecidewhethertocontactareviser,norisitevenrecommended,sincetranslatorsaretoldthey“may”ratherthan“should”asksomeonetocheckthetranslation.CorpusAnalysisRevisionhistoryofthetranslatedarticlesTohelpdeterminewhethertranslationproblemswerepresent(andlatercorrected)inthecorpusoftranslatedWikipediaarticles,thepagehistoryofall94articleswasconsultedandthefollowingdatawererecorded:(1)thenumberof“edits”(inWikipedia"ssense)thateacharticlehadreceivedbetweenthetimeitwaspostedto“Wikipedia:PagesNeedingTranslationintoEnglish”andthetimethedatawasbeingrecorded,namelyMay2013;and(2)thenumberofeditorswhomadechangestothearticleduringthisperiod.ThisdataisdetailedinTable1.The“unknown”labelinthetablerefersto11articlesthatweredeletedafterbeingpostedto“Wikipedia:PagesNeedingTranslationintoEnglish”.Oncearticlesaredeleted,internetuserscannolongerviewthepagehistory,andsoneitherthenumberofeditsnorthenumberofeditorscouldbedetermined.AsTable1illustrates,morethanhalfthetranslatedarticleswereedited30orfewertimesfromthemomenttheywerepostedto“Wikipedia:PagesNeedingTranslationintoEnglish”untilMay2013-aperiodthatcouldspanuptotwoyears.5Moreover,thisfigureincludesbotheditsbyinternetusersandeditsbybots,theautonomousorsemi-autonomouscomputerscriptsdesignedtocarryoutrepetitivetaskssuchascheckingforcopyrightviolations,correctingspellingerrorsandtypos,removingpagesfrominappropriatecategoriesandrevertingvandalism.Complexrevisiontaskssuchaschangingincorrectorinconsistenttérminologyandmistranslationsaregenerallybeyondthecapabilitiesofabot,sothechancesthatanarticlewithfewerthan30editswasrevisedfortransferproblemsisreduced.２６ Table1.Numberofeditsandeditorsinthe94translatedarticles.No.ofeditorsAverageNo.ofNo.Average(mean)(All(mean)AverageeditsofarticlesMinMax(mean)(Spanishtranslations)(Frenchtranslations)translations)UserUsersBotsBotsUsersBotsUsersBotsUsersBotss15or3210954.912.065.52.14.642.05fewer16-3020601359.252.69.829.072.831-49126120611.923.839.293.2915.64.650-100942>20714.884.5134164.8(approx.)(approx.)>10010164>2016>206>205.17>207.25Unknown11----------Total94Toprovideabetterpictureofhowmanypeoplewereinvolvedintheeditingprocess,the“No.ofeditors”columnsindicatehowmanydifferentinternetusersandbots,onaverage,wereresponsiblefortheeditsmadetothetranslatedarticles.6Thefiguresmarked“approx.”areestimations,ascalculatingthenumberofdistinctusersbeyond20becametoocumbersomeandisnotreallynecessary.Asthetableshows,eacharticlewasgenerallyeditedbyseveralusers,including-inalmostallcases-atleastonebot.Moreover,asthenumberofeditsincreased,sodidthenumberofeditors,indicatingthatinmostcasestherevisionprocesswasacollaborativeoneinvolvingbothhumansandcomputerscripts.TranslationproblemsinthecorpusEventhoughTable1indicatesthatone-thirdofthetranslatedarticleswereeditedfewerthan16timesoncetheywerepostedto“Wikipedia:PagesNeedingTranslationintoEnglish”,thisisnotnecessarilyaproblem.Afterall,asingleeditcanbeasminorasfixingatypoorasmajorasrewritingtheentirearticle.Thus,manytranslationproblemscouldpotentiallybeaddressedinasingleedit.Forthisreason,asub-corpuswascompiledforcloseranalysis.Thetranslatedarticlesweresortedbythenumberofeditseachhadreceivedandeverythirdarticlewasthenselectedforthesub-corpus,foratotalof29articles(11translationsfromFrenchand18fromSpanish).TheAppendixliststheURLsforallarticlesfromthissub-corpusthatare２７ citedintherestofthispaper.Forarticlesinthissub-corpus,thepagehistoriesofthesourceandtargettextswerecomparedtodeterminewhether,onthedaythearticlewaspostedto“Wikipedia:PagesNeedingTranslationintoEnglish”,itcontainedanytransferorlanguageandstyleproblemsthatwouldneedtobeaddressed.Morespecifically,source-andtarget-text(TT)segmentswerecomparedtodeterminewhethertermsandwordswereincorrectlytranslated,mistranslationsexistedorinformationhadbeenomitted(transferproblems)andwhetherthearticleconformedtoWikipedia"s“ManualofStyle”,containedunidiomaticlanguage,hadspelling/grammarerrorsorusedtérminologyinconsistently(language/styleproblems).Threeorfourexamplesoftransferandlanguage/styleproblemswereidentifiedineacharticle,ifpossible,sotheycouldserveassampleerrors.Whenanarticlecontaineduntranslatedtextonthedayitwaspostedto“Wikipedia:PagesNeedingTranslationintoEnglish”,themorerecentversionsofthepagewereconsulteduntilthefirstEnglishtranslationcouldbefound.ArticlesthatstillhaduntranslatedtextbyMay2013wereconsideredtohaveatransferproblem.Table2summarizestheresultsoftheST/TTcomparisons.Sincetworeviserswouldlikelyhavedifferentopinionsaboutexactlyhowmanyerrorseachtranslationcontained,Table2doesnotindicateaprecisetotalnumberoferrors.Instead,itindicateswhetheratleastoneexampleofatransferand/orlanguageandstyleproblemcouldbefoundinthetranslatedarticles.Usually,threeorfourdifferenterrorsfromeachcategorywereidentifiedineveryarticle;thisway,theinitialversionofeachtranslatedarticlecouldbecomparedwiththeMay2013versiontoseewhetherthesesampleerrorshadbeenresolved,aswillbeshowninTable3.AsTable2illustrates,nearlyallarticleshadatleastonetransferandonelanguage/styleproblem,exceptfourofthetranslationsfromSpanish.Ofthesefourarticles,twowerelistedon“Wikipedia:PagesNeedingTranslationintoEnglish”butuponcloserinspectiondidnotappeartoactuallybetranslationsofarticlesfromSpanishWikipedia.ThecontentoftheEnglishandSpanish“FilipeEspinosa”articles,forinstance,differedsubstantially:theSpanishversionwasmorethanfourtimesthelengthoftheEnglisharticle,andbothversionswereorganizeddifferentlyandcontainednumerousdetailsnotpresentintheotherversion.7Inthecaseoftheothernon-translation,“Aviacióndelnoroeste”,noarticlecouldbefoundonthistopicin２８ SpanishWikipedia,norcouldsimilarcontentbefoundinSpanishonotherwebsites.Table2.ProblemsintheinitialversionsofthetranslatedWikipediaarticles.Noofarticles.with:No.ofTraLanguage/stypseudo-nsferleprotranslationsblemsproblemsFrenchtoEnglishtranslations11110(11articles)SpanishtoEnglishtranslations15152(18articles)Alltranslations(29articles)26262TheEnglisharticledidcontainnumerouslanguageandstyleerrorswhenitwaspostedto“Wikipedia:PagesNeedingTranslationintoEnglish”on28July2011,butthesecouldhavebeenduetoanon-nativeEnglishspeakerwritingthearticleinEnglishratherthantranslatingitfromSpanish,ortosomeonetranslatingatextnotcurrentlyavailableonline.LackofaSTforthesetwoarticlesmeanttheycouldnotbeanalysedfortransferandlanguage/styleproblems.8Theothertwocaseswhereanarticlecontainedjusttransferproblemsorjustlanguage/styleproblems(butnotboth)areworthdiscussinginmoredetail.Thefirst,“MocopulliAirport”,waswritteninSpanishdirectlyinEnglishWikipedia.Itwasonly50wordslongandallitsdetailscouldbetransferredintoastandardizedairportinfobox,9whichanotheruserdid4daysafterthearticlewascreated.Althoughthisarticlecontainedalanguageproblem(theSpanishnameoftheairport,leftinparenthesesaftertheEnglishname,ismissinganaccent),ithadnotransfererrors.Similarly,theJoséJoaquínFlírezHernándezarticlewasalsowritteninSpanishdirectlyinEnglishWikipedia.Twoweeksafterthe120-wordarticlewascreatedinSpanish,thecontentwasdeletedandreplacedwitha1-sentencestubbecausenoprogresswasmadeonthetranslation.Thisarticlethereforehadatransfererror(sincethecontentwasnevertranslated),butnolanguageerrors.So,asTable2indicates,apartfromthesefourexceptions,everytranslatedarticleinthesub-corpushadatleastonetransferandonelanguage/styleproblem.Afewexampleswillhelpillustratethetypesofproblemscommonlyfoundinthetranslations.TransferproblemsMistranslations,orpassagesthatcould“misleadthereaderaboutanimportantfeatureof２９ thesourcetext’smessage”(Mossop2006,126),werecommoninthecorpusandtheyrepresentmorethanone-thirdofthe86sampletransferproblems.Somemistranslatedinformationwaspossiblygeneratedbyautomatictranslationsoftware,asinthisarticleabouttheSpanishvillageNavahermosa,whichcontainedmanynearlyincomprehensiblepassages:“Navahermosa”,SpanishWikipedia(23July2012)AntonioLlorente,alcomentarlafrecuenciadeestetérminoenlazonadeSalamanca,ÁvilaylaparteoccidentaldeSegovia,sitúasuorigenenenlaslenguasdevetonesycarpetanos,aunqueplantealadudadesuorigenvascoenelqueexistelavoznaba“llanura,valle,barranco”.[AntonioLlorente,notingthefrequencyofthistermintheareasofSalamanca,ÁvilaandthewesternpartofSegovia,maintainsthatitcamefromthelanguagesusedbytheVetonesandCarpetanos,althoughhedidwonderwhetheritwaspossiblyofBasqueorigin,sincethewordnaba,meaning“plain,valleyorravine”,existsinthatlanguage.]“Navahermosa”,EnglishWikipedia(9September2012)AntonioLlorente,tocommentonthefrequencyofthistermintheSalamanca,ÁvilaandthewesternpartofSegovia,locatedoriginatedinthelanguagesofvetonesandcarpetanos,butraisesthequestionoftheiroriginBasqueinwhichvoicetherenaba‘plain,valley,ravine’.Othermistranslationswereprobablygenerated(oratleastpost-edited)byahuman,asinthisarticleaboutinventor-songwriterRené-LouisBaron,whichmistranslates“laGénérationautomatiquedemélodiesditespopulaires”butwhichaddsinformationtocontextualizetheacronymIRCAM:“René-LouisBaron”,FrenchWikipedia(18April2013)Sestravauxnes’axentpassurlamusiqueexpérimentale,domainelargementexploréparl’IRCAM,maissurlaGénérationautomatiquedemélodiesdites“populaires”.[Hisworkisnotfocusedonexpérimentalmusic,afieldlargelyexplorédbyIRCAM,butratheronpopmusicautomation.]“René-LouisBaron”,EnglishWikipedia(3May2013)Hisworkisnottogetanewexpérimentalmusic(fieldwidelyexplorédbyIRCAMinFrance),butrathermelodiousmusicorpopularmusic.Anotherthirdofthe86sampletransferproblemsinvolvedtermsorwordsthatweretranslatedincorrectly,asinthearticleabouttheColombianbasketballteam,Bukaros,wheretheteam’s“entrenador”wasreferredinEnglishtoasa“trainer”insteadof“coach”,or“manager”,andthearticleabouttheSpanishbandTheZombieKids,whichdescribed３０ SubterfugeRecords,a“compañíadiscográfica”intheoriginalSpanisharticle,asa“recordinghome”ratherthana“recordingstudio”.Otherarticles,suchas“FrenchprésidentialInauguration”,containednumerousincorrectwordsandterms,including“swearing”for“prestationdeserment”,whichwouldhavebeenmoreaccuratelytranslatedas“oathofoffice”or“swearingin[ceremony]”,and“contestant”for“investiture”,whichshouldhavebeentranslatedas“inauguration”,particularlysincethattermwasusedinthetitleofthearticle.Finally,omissionsrepresentedabout20%ofthesampletransfererrorsidentifiedinthecorpus,whileuntranslatedcontentaccountedforabout5%.Forinstance,intheSpanishversionoftheRafaelCorreaarticle,whenCorrea’ssonwasmentionedinthe“Biografía”section,hewasnamed“RafaelMiguel”.InthecorrespondingEnglishsection,“EarlyCareer”,hissonwasjust“Miguel”,althoughtheinfoboxthataccompaniedtheEnglisharticledidlistbothnames.Otherdetailsarealsomissing,suchasthefactthattheschoolwhereCorreacompletedhisprimaryandsecondaryeducationwasCatholic,andthathisworkattheSalesianmissionwasdoneasavolunteer.LanguageandstyleproblemsArangeoflanguageandstyleproblemswereidentifiedinthetranslatedarticles,mostofwhichinvolvedproblemswithmechanics-thatis,notconformingtoWikipedia’sgrammar,spelling,housestyle(Mossop2006,137)-orunidiomaticlanguage(ibid.,136).Someerrorsweregrammatical,suchasverbtenseproblems(e.g.“whocouldnotpronounced‘Señora’”,intheSonoraarticle),incorrectsyntax(e.g.“and[she]diffusedcontinuouslyBoushaame”intheRené-LouisBaronarticle)orpluralizationproblems(e.g.“3000school”intheNavahermosaarticle).Othersinvolvedinconsistentnames.Forinstance,theBukarosarticlereferredtothe“VicenteDiazRomeroColiseum”buttheinfoboxaccompanyingthearticlelistedtheteam’sheadquartersasthe“ColiseoVicenteDiazRomero”.Likewise,theSergeGuinchardarticlereferredto“UniversitéPanthéon-Assas(Paris2)”indifferentways,including“UniversityofPanthéon-Assas-Paris2”,“UniversityofParis2”,“UniversityPantheonParis-Assas-2”and“Panthéon-AssasUniversity”,usuallywithoutalinktotheuniversity"sEnglishWikipediapage,whichcouldhaveclearedupanyambiguities.And,inasimilarvein,variousarticlesusedtérminologyinconsistently:“Ouragan-ClassLandingPlatformDock”,forinstance,referredto３１ theFrenchterm“radier”asbothan“interiorraftorradier”andthenasa“welldeck”,resultinginasub-languageproblem(sincethefield-specifictérminologyisincorrect)andpotentialconfusionforreaderswhomightnotrealizethetwotermsarereferringtothesameobject.RevisingandeditingtechniquesinthecorpusNowthatwehaveseenthattheEnglisharticles,oncetheywerelistedon“Wikipedia:PagesNeedingTranslationintoEnglish”,generallycontainedbothtransferandlanguage/styleproblems,weneedtoexaminewhethertheseerrorswereaddressedinthemonths(oryears)afterthearticleswerelisted.Table3illustrateswhichkindsoferrorsweremostcommonlyaddressedwhenthetranslationswereeditedbyWikipediausers.SinceTable2didnotincludeanexhaustivelistofalltranslationerrorsinthetranslations,butratheranindicationofwhethercertaintypesoferrorsexisted,Table3illustrateswhatpercentageofthesampleerrorsidentifiedwhencompilingthedataforTable2wereaddressedbyWikipediansinsubsequentversionsofanarticle.AsTable3demonstrates,languageerrorsweremoreoftenresolvedthantransfererrors,althoughinbothcases,onaverage,slightlymorethanhalftheerrorshadbeenresolvedbyMay2013.Articleswith15orfewereditshadthelowestpercentageofresolvedproblems,andingeneral,asthenumberofeditsincreased,sodidthepercentageoferrorsthatwereresolved.Interestingly,fewofthetransfererrorsidentifiedinthearticleswithmorethan100editswereresolved;thismaybebecausethenumberofarticlesinthesamplewasquitesmall:alargersamplegroupmighthaveaveragedoutthe1or2articleswhereveryfewofthetransfererrorswereresolved.Table3alsoappearstoshowthatbotheditingandrevisingtookplace:afterall,languageerrorssuchasspellingandgrammarmistakescanoftenbeaddressedbyunilingualWikipedianswithoutconsultingaST,whiletransfererrorssuchasmistranslationsandomissionscanusuallyberesolvedonlybycomparingboththesourceandtargettexts-afindingsupportedbyBrunette,Gagnon,andHine(2005).ButTable4providesabetterpictureofwhethertheproblemswereaddressedbybilingualrevisionorunilingualediting.Itdividestransferproblemsintofourkinds:untranslatedcontentandomissions,representingMossop’s(2006,125)completenessparameterandmistranslatedwords,termsorotherpassages,３２ representingMossop’saccuracyparameter(ibid.).Table3.Translationproblemsresolvedinthesamplearticles.TransferLanguage/styleproblemsproblemsNAArNoofeditsNo.o.ofrticlesNo.Percentticles.ofPercenarticleswoferrorsageoferrorswierrorstageoferrorsithideresolvedbythresolvedbyprntifiedMay2013pridentifieMay2013doblemsoblems15192128.57%91926.32orfewer2%016-3662564.00%613%69.23931-4441566.67%414%75.000050-1331172.73%312%41.67100>441450.00%412%91.67Tota228654.65%267060.00l96%Table4.Typesoftransfererrorsinthesub-corpusandhowoftentheywereresolved.NumberofinstancesNumberPercentagTypeofidentifiedwhenpostedtoofinstanceseofinstancesTransferproblem“Wikipedia:PagesNeedingresolvedbyresolvedbyTranslationintoEnglish”May2013May2013Untranslated33100.00%contentOmissions19631.58%Lexical/té311754.84%rminologicalMistranslatio332163.64%nsTotal864754.65%Omissions,whichwouldonlybenoticedbycarefullycomparingthesourceandtargettexts,wereresolvedleastoften,whileuntranslatedcontent,easilydetectedevenbyanautomatedscript,wereresolvedmostoften.Mistranslationsandlexicalortérminologicalproblemswereresolvedinjustoverhalfthecases,andthesekindsofproblemscansometimesbeaddressedbyediting-thatisbyconsideringthearticlesolelyasanEnglishtext-ratherthanbyrevising-thatisbycomparingtheEnglishwiththecorrespondingFrenchorSpanishtext.Forinstance,intheFrenchprésidentialinaugurationarticle,the“prestationdeserment/swearing”términologyproblemmentionedearlierwaslatercorrectedwhen“swearing”waschangedto“Oathofoffice”,butthisproblemcouldberesolvedwithoutconsultingtheST,sinceanEnglishspeakerwithnoknowledgeofFrenchcanguessfromthecontextthat“swearing”referstotakinganoathofofficeratherthancursing.Infact,the３３ “RevisionHistory”ofthetranslatedarticlesrevealsthattransferandlanguage/styleerrorswereaddressedthroughediting,revising,rewritingoracombinationofallthree.Contenteditingoftenresultedintransfererrorsbeingremovedfromthearticle.Forinstance,“SN1054”,whichhadmorethan100editsbetween4April2011,whenitwaspostedto“Wikipedia:PagesNeedingTranslationintoEnglish”,and30May2013,whenthedataforthispaperwascollected,hadnumeroustransfererrors,someofwhichwereresolvedoverthecourseofseveraldaysinApril2013whenWikipediauserDbachmannremovedmorethan4000wordsofcontentfromthearticledueto“unsourcededitorializing”,“unreferencedexposition”andallegedfactualerrors.Likewise,thetranslatedcontentoftheNavahermosaarticle,whichcontainednumeroustransferandlanguageproblems,wasdeletedandreplacedbyaone-sentencestubtwomonthsafteritwasaddedto“Wikipedia:PagesNeedingTranslationintoEnglish”.Interestingly,thereasongivenfordeletingthecontentwasnotthetransferorlanguageerrors,butratherthelackofsupportingsources.Thiswasalsothecasewhenarticlesweredeleted:11translatedarticlesweredeletedbetweenthetimetheywerepostedto“Wikipedia:PagesNeedingTranslationintoEnglish”andMay2013,butinnocasedidthequalityofthetranslationleadtothearticlebeingdeleted.Rather,articlesweredeletedbecausetheywerepromotional,thesubjectwasnotsignificant,noprogresshadbeenmadeonthetranslation,thearticleviolatedcopyrightorthearticlelackedsourcestosupportitspoints.Othererrorswereresolvedviarewriting,asinthe“LifeandWorksofSEM”sectionofthe“Sem(Artist)”article,whichwas108wordson29April2011,wasconsiderablyrewrittenon3September2011(131words)andwasediteddowntojust39wordson23May2013.TheFrencharticleoriginallyusedasabasisfortheEnglishcontentinApril2011resembledtheEnglisharticleverylittlebyMay2013,apartfromafewmatchingdetails(areferencetoAmedeedeNoeandthefactthatGeorgesGoursatsignedhisworkSEM).Thetransferandlanguageandstyleproblemswereeliminated,butviacontenteditingratherthanrevision.Otherarticlesdo,however,showthatbilingualrevisiontookplace.TheFrenchprésidentialinaugurationarticle,forinstance,waspostedto“Wikipedia:PagesNeedingTranslationintoEnglish”on27June2012,andatthattimecontainednumerousmistranslations,inadditiontothetwolexicalproblemscitedearlier.Thefollowingexampleshowshowtheseerrorswererevised:３４ “InvestitureduprésidentdelaRépubliquefranfaise”,FrenchWikipedia(27June2012)LaConstitutionnementionneaucunerégleétablieconcernantleDéroulementdel’investiture.“FrenchprésidentialInauguration”,EnglishWikipedia(27June2012)TheConstitutiondoesnotmentionanyruleestablishedfortheconductofcontestants.“FrenchprésidentialInauguration”,EnglishWikipedia(15July2012)TheConstitutionofFrancedoesnotmentionanyrequirementsfortheconductofproceedings.Inthisandotherinstances,themeaningofthe27June2012Englishversionisnotveryclear,butthecorrectedversionpublishedon15July2012,whichmoreaccuratelyreflectstheFrencharticle,stronglysuggeststhattheFrenchSTwasconsultedduringtherevisionprocess.Caseswherebilingualrevision-ratherthanunilingualeditingorrewriting-tookplacearemorefrequentinarticlesthat,likethisone,hadmorethan100editsafterbeingpostedto“Wikipedia:PagesNeedingTranslationintoEnglish”.AsTable1demonstrated,though,only10%ofthearticlesinthecorpushadthatmanychanges,andnotalwaysduetothelengthoftimethathadpassedbetweenthedatewhenthearticlewaslistedon“Wikipedia:PagesNeedingTranslationintoEnglish”andthedatewhenthedataforthispaperwascollected:8ofthe10translationsfromFrenchand12ofthe22translationsfromSpanishwithfewerthan16editswerelistedon“Wikipedia:PagesNeedingTranslationintoEnglish”atleast6monthsearlier,andsome(e.g.“LesColoca-taires”,“OuraganClassLandingPlatformDock”,“Aviacióndelnoroeste”)werelistednearly2yearsearlier.ConclusionsLetusreturntothethreequestionsposedintheintroduction,namely:areWikipediansproducingtranslationsthatrequirerevision?Doestheopen-editingprocessofthewikienvironmenthelpeliminateerrorsintranslatedarticles?Andfinally,arecertaintypesoferrorsmorecommonlyresolved?Theanswertothefirstquestionseemsclear:withonly4exceptions,all29translationsinthesub-corpuscontainedbothtransferandlanguage/styleproblems.Becauseerrorslikemistranslationsandomissionsoccurredinmanyofthearticles,translationspostedon“Wikipedia:PagesNeedingTranslationintoEnglish”evidentlyrequirebilingualrevision.This３５ isnotsurprising,sincearticlesarelistedonthispagebecausetheyrequire“clean-up”ortranslation.Articlestranslatedviaotherprojects,orbyWikipediausersactingontheirowninitiative,mightrequirelessrevision.Butthisfindingdoessuggestthatothercrowdsourcedtranslationprojectsarelikelytohavetransfererrors-atleastintheinitialsubmissions-andthatplatformsneedtobedesignedinawaythatwillfacilitateidentifyingandrevisingtheseerrors-somethingthatisnotasstraightforwardinWikipediaasitisinFacebook,forexample,wheresourceandtargetpairsarelinedupwithoneanotheronthescreenduringthetranslationandvotingprocess.Thesecondquestionisnotaseasytoanswer.Sinceoverhalfofthe94translationshad30orfeweredits,andsinceonlyhalfthetransferandlanguage/styleproblemsidentifiedinthesub-corpuswereresolvedbyMay2013,it’sclearthattransferandlanguage/styleproblemspersistformonthsoryearsafterthearticlehasbeenlistedon“Wikipedia:PagesNeedingTranslationintoEnglish”.Buttheseerrorsmaynotexistforever,sinceWikipedia’sopen-editingplatformmeansthatsomeonemayeventuallynoticetheerrorsandcorrect,deleteorotherwiseeliminatethem.Thus,thetranslationscanneverbeconsidereddefinitiveversions,butratherevolvingtextsinwhichsomeproblemswillbeintroducedandpossiblyaddressed,andnewproblemsmaysurfaceasaresult,onlytobecorrectedatalaterdate.Clearly,changesoftenoccurataveryslowpace,sincemanyproblemswereresolvedonlyafterweeksormonths,whileothershadnotbeenresolvedevenaftertwoyears.Despitethisdifficulty,thiscasestudydoesshowthatWikipediaisalucrativesitefortranslationresearch,sinceresearcherscaneasilytrackchangestotranslationsovertimeandexploréissuesbeyondthescopeofthispaper,includingrevisionratesinmajorvs.minorlanguagesandspecializedvs.generaltexts.Asforthethirdquestion,thecorpusanalysisclearlyshowsthatcertaintypesoferrorsaremorelikelytopersistinthetranslatedarticles:omissionswererarelycorrected,whilelanguageerrors,givenenoughedits,werecorrectedveryfrequently.Thisdoessuggestthatediting,ratherthanrevising,ismorecommonintranslationscompletedthrough“Wikipedia:PagesNeedingTranslationintoEnglish”.Yetastheprevioussectionhasshown,somebilingualrevisiondoesoccuraswell,andtransfererrorsaresometimesaddressed,sothisparticularformofcrowdsourcingcansometimesresultintranslationsthataregenerally３６ accurateandfreeofgrammaticalerrors,thoughnotwithindaysorhours,asprofessionaltranslationmightrequire.GiventhetentativelypromisingresultsfromJiménez-Crespo’s(2013)Facebookstudy,andthecautionofBrunette,Gagnon,andHine(2005)againstunilingualeditingratherthanbilingualrevision,thislastfindingshowsthatmoreresearchcomparingthevariousformsofcrowdsourcingisneeded:aretransferandlanguageerrorsfrequentinothercrowdsourcingplatformsand,ifso,howquicklyaresucherrorsrevised?Whatsimilaritiesdotheseplatformsshareregardinguserinterfacedesign,thetypeofcontentbeingtranslated,thesourceandTLcombinations,andparticipantrecruitment,motivationandremuneration?Whateffectmightthesedifferenceshaveonthetranslationsproduced?Thesequestionsareimportanttoanswer,becausealthoughtheinstitutions(corporations,non-profitorganizations,etc.)thatrelyoncrowdsourcingsuggestitisaviablealternativetothetranslation-edit-proofreadmodel(Kelly,Ray,andDePalma2011),researchtoverifythisclaimisstilllacking.WhilethisstudywaslimitedtoasmallcorpusofWikipediatranslations,itcanstillofferinsightsforthetranslationindustry,particularlyregardingtheroleofnon-professionals.Forinstance,thispaperhasdemonstratedthatgivenenoughtime,Wikipediansoftendidresolvetransferandlanguage/styleproblemsintranslatedarticles-eventhoughWikipediadoesnotrequireoversightbyprofessionaltranslatorsorbilingualusers.Thissuggeststhatwhensourceandtargettextsdonothavetocloselycorrespond(asisoftenthecaseinweblocalization)orwhentimeisnotacrucialfactor,thenacrowdsourcingplatformsuchasWikipediacansuccessfullyallowprofessionalsandnon-professionalsaliketowrite,edit,translateandrevisecontent,withoutnecessarilybeingpaid.Buthere,ethicalconcernsarise:justbecauseinternetusersarewillingtovolunteerfornon-profitcauseslikeWikipedia(orevenforfor-profitorganizationslikeFacebook),itisnotnecessarilyethicaltoencouragethemtodosowithoutalsoprovidingawayforthemtobecompensatedfortheirtime,particularlyforuserswhovolunteermanyhourseachweek(cf.e.g.Howard2014).Moreover,findingwaystoallowcrowdsourcedtranslationstobeproducedandrevisedmorequicklyandefficientlyhelpslargelyunpaidprojectscompetemoreefficientlywithprofessionaltranslators,resultingin３７ backlashfromprofessionals(cf.e.g.McDonoughDolmaya2011)when,ifthethornyissueofremunerationwereresolved,collaborationcouldinsteadprovidebenefits,includingtrainingfornovices,visibilityforparticipants,intellectuallystimulatingcontent,andbetteraccesstoinformationfortranslatorsandusersalike.３８ ReferencesBrunette,Louise,andChantalGagnon.2013.“Enseignerlarevisional’eredeswikis:laoul’ontrouvelatechnologiealorsqu’onnel’attendaitplus.”JournalofSpecialisedTranslation19:96-121.http://jostrans.org/issue19/art_brunette_gagnon.php.Brunette,Louise,ChantalGagnon,andJonathanHine.2005.“TheGREVISProject:ReviseorCourtCalamity.”AcrossLanguagesandCultures6(1):29-45.doi:10.1556/Acr.6.2005.1.3.Cronin,Michael.2010.“TheTranslationCrowd.”Revistatradumatica8.http://www.fti.uab.cat/tradumatica/revista/num8/articles/04/04.pdf.Désilets,Alain,andJaapvanderMeer.2011.“Co-CreatingaRepositoryofBestPracticesforCollaborativeTranslation.”LinguisticaAntverpiensia10:27-45.Drugan,Joanna.2011.“TranslationEthicsWikified:HowDoProfessionalCodesofEthicsandPracticeApplytoNon-ProfessionallyProducedTranslation.”LinguisticaAntverpiensia10:111-130.Dwyer,Tessa.2012.“FansubdreamingonViKi:‘Don’tjustwatchbuthelpwhenyouarefree.’”Translator18(2):217-243.doi:10.1080/13556509.2012.10799509.Giles,Jim.2005.“SpecialReportInternetEncyclopaediasGoHeadtoHead.”Nature438:900-901.doi:10.1038/438900a.“HowFansTranslated‘closetoamillionwordsofdialog’inNewBaldur’sGateLocalizations.”PCGamer,November16.http://rn.pcgamer.com/2012/11/16/baldurs-gate-contains-close-to-a-million-words-of-dialog-and-how-fan-translations-helped-the-enhanced-editionHoward,Dorothy.2014.“ThoughtsonWikipediaEditingandDigitalLabour.”http://en.m.wikipedia.org/wiki/Wikipedia:Thoughts_on_Wikipedia_Editing_and_Digital_LaborHowe,Jeff.2008.Crowdsourcing:WhythePoweroftheCrowdisDrivingtheFutureofBusiness.NewYork:CrownBusiness.Jiménez-Crespo,MiguelA.2011.“FromManyOne:NovelApproachestoTranslationQuality３９ inaSocialNetworkEra.”LinguisticaAntverpiensia10:131-152.Jiménez-Crespo,MiguelA.2013.“Crowdsourcing,CorpusUse,andtheSearchforTranslationNaturalness:AComparableCorpusStudyofFacebookandNon-TranslatedSocialNetworkingSites.”TranslationandInterpretingStudies8(1):23-49.doi:10.1075/tis.8.1.02jim.Kelly,Nataly,RebeccaRay,andDonaldA.DePalma.2011.“FromCrawlingtoSprinting:CommunityTranslationGoesMainstream.”LinguisticaAntverpiensia10:75-96.McDonoughDolmaya,Julie.2011.“TheEthicsofCrowdsourcing.”LinguisticaAntverpiensia10:97-110.McDonoughDolmaya,Julie.2012.“AnalyzingtheCrowdsourcingModelandItsImpactonPublicPerceptionsofTranslation.”Translator18(2):167-191.doi:10.1080/13556509.2012.10799507.Mesipuu,Marit.2012.“TranslationCrowdsourcingandUser-TranslatorMotivationatFacebookandSkype.”TranslationSpaces1:33-53.doi:10.1075/ts.1.03mes.Mossop,Brian.2006.RevisingandEditingforTranslators.2nded.Manchester:StJerome.O’Brien,Sharon,andReinhardSchäler.2010.“NextGénérationTranslationandLocaliza¬tion:UsersAreTakingCharge.”ProceedingsofTranslatingandtheComputer32.London:ASLIB(AssociationforInformationManagement).http://doras.dcu.ie/16695/1/Paper_6.pdf.O’Hagan,Minako.2008“FanTranslationNetworks:AnAccidentalTranslatorTrainingEnvironment?’InTranslatorandInterpreterTraining:Issues,MethodsandDebates,editedbyJohnKearns,158-183.London:Continuum.O’Hagan,Minako.2009“EvolutionofUser-generatedTranslation:Fansubs,TranslationHackingandCrowdsourcing.”JournalOfInternationalisationandLocalisation1(1):94-121.Olohan,Maeve.2014.“WhyDoYouTranslate?MotivationtoVolunteerandTEDTranslation.”TranslationStudies7(1):17-33.doi:10.1080/14781700.2013.781952.Perrino,Saverio.2009.“User-GeneratedTranslation:TheFutureofTranslationinaWeb2.0Environment.”JournalofSpecialisedTranslation12:55-78.http://jostrans.org/issue12/art_perrino.pdf.４０ Thicke,Lori.2012.“EnterpriseInnovators:Twitter’s400,000Translators.”MultiLingual23(8):12-13.Toury,Gideon.2012.DescriptiveTranslationStudies-AndBeyond.Reved.Amsterdam:JohnBenjamins.Wikipedia.2013a.“Wikipedia:Translation.”AccessedJune27.http://en.wikipedia.org/wiki/Wikipedia:TranslationWikipedia.2013b.“Wikipedia:ManualofStyle.”AccessedJuly3.http://en.wikipedia.org/wiki/Wikipedia:Manual_of_StyleWikipedia.2013c.“Help:MinorEdit.”AccessedJuly31.http://en.wikipedia.org/wiki/Help:Minor_editZaidan,OmarF.,andChrisCallison-Burch.2011.“CrowdsourcingTranslation:ProfessionalQualityfromNon-Professionals.”InVol.2ofHLT’11Proceedingsofthe49thAnnualMeetingoftheAssociationforComputationalLinguistics:HumanLanguageTechnologies,1220-1229.http://www.cs.jhu.edu/~ccb/publications/crowdsourcing-translation.pdf４１ Appendix1.TranslatedWikipediaarticlescitedinthispaper“ArielNan”English(current):http://en.wikipedia.org/w/index.php?title=Ariel_NanEnglish(13November2011):http://en.wikipedia.org/w/index.php?title=Ariel_Nan&oldid=460464543Spanish(onArielNan’suserpage):http://en.wikipedia.org/wiki/User:Arielnan/Ariel_Nan“Aviacióndelnoroeste”English(current):http://en.wikipedia.org/wiki/Aviaci%C3%B3n_del_NoroesteEnglish(28July2011):http://en.wikipedia.org/w/index.php?title=Aviaci%C3%B3n_del_Noroeste&oldid=441887938NoSpanisharticleonAviacióndelNoroeste“Bukaros”English(current):http://en.wikipedia.org/wiki/BukarosEnglish(6December2011):http://en.wikipedia.org/w/index.php?title=Bukaros&oldid=464354482Spanish(4December2011):http://es.wikipedia.org/w/index.php?title=B%C3%BAcaros_de_Santander&oldid=51886913“FilipeEspinosa”English(current):http://en.wikipedia.org/w/index.php?title=Felipe_EspinosaEnglish(26July2012):http://en.wikipedia.org/w/index.php?title=Felipe_Espinosa&oldid=504191175Spanish(12July2012):http://es.wikipedia.org/w/index.php?title=Felipe_Espinosa&oldid=57847643“FrenchprésidentialInauguration”English(current):http://en.wikipedia.org/wiki/French_présidential_inaugurationEnglish(27June2012):http://en.wikipedia.org/w/index.php?title=French_présidential_inauguration&oldid=499555643French(27June2012):４２ http://fr.wikipedia.org/w/index.php?title=Investiture_du_pr%C3%A9sident_de_la_R%C3%A9publique_fran%C3%A7aise&oldid=80216857“JoséJoaquínFlírezHernández”English(current):http://en.wikipedia.org/wiki/Jos%C3%A9_Joaqu%C3%ADn_Fl%C3%B3rez_Hern%C3%A1ndezEnglish(3December2011):http://en.wikipedia.org/w/index.php?title=Jos%C3%A9_Joaqu%C3%ADn_Fl%C3%B3rez_Hern%C3%A1ndez&oldid=463855088[note:thisversionisinSpanish]“Lescololcataires”English(current):http://en.wikipedia.org/wiki/Les_ColocatairesEnglish(27November2011):http://en.wikipedia.org/w/index.php?title=Les_Colocataires&oldid=462806260French(4November2011):http://fr.wikipedia.org/w/index.php?title=Les_Colocataires&oldid=71774418“MocopulliAirport”English(current):http://en.wikipedia.org/wiki/Mocopulli_AirportEnglish(18January2013):http://en.wikipedia.org/w/index.php?title=Mocopulli_Airport&oldid=533645447English(14January2013):http://en.wikipedia.org/w/index.php?title=Mocopulli_Airport&oldid=533038123(Note:thistextisinSpanish)“NationalUniversityofSanMarcos”English(current):http://en.wikipedia.org/w/index.php?title=National_University_of_San_MarcosEnglish(14May2011):http://en.wikipedia.org/w/index.php?title=National_University_of_San_Marcos&oldid=429032944Spanish(12May2011):http://es.wikipedia.org/w/index.php?title=Universidad_Nacional_Mayor_de_San_Marcos&oldid=46335739“Navahermosa”４３ English(current):http://en.wikipedia.org/w/index.php?title=NavahermosaEnglish(9September2012):http://en.wikipedia.org/w/index.php?title=Navahermosa&oldid=511495698Spanish(23June2012):http://es.wikipedia.org/w/index.php?title=Navahermosa_%28Toledo%29&oldid=58170723“Ouragan-ClassLandingPlatformDock”English(current):http://en.wikipedia.org/wiki/Ouragan-class_landing_platform_dockEnglish(21August2011):http://en.wikipedia.org/w/index.php?title=Ouragan-class_landing_platform_dock&oldid=445909654French(14May2011):http://fr.wikipedia.org/w/index.php?title=Classe_Ouragan&oldid=65350162“RafaelCorrea”English(current):http://en.wikipedia.org/wiki/Rafael_CorreaEnglish(23January2013):http://en.wikipedia.org/w/index.php?title=Rafael_Correa&oldid=534417683Spanish(19January2013):http://es.wikipedia.org/w/index.php?title=Rafael_Correa&offset=&limit=250&action=history“René-LouisBaron”English(current):http://en.wikipedia.org/w/index.php?title=Ren%C3%A9-Louis_BaronEnglish(3May2013):http://en.wikipedia.org/w/index.php?title=Ren%C3%A9-Louis_Baron&oldid=553356601French(current):http://fr.wikipedia.org/wiki/Rene-Louis_Baron“Sem(Artist)”English(current):http://en.wikipedia.org/wiki/Sem_%28artist%29English(29April2011):http://en.wikipedia.org/w/index.php?title=Sem_%28artist%29&oldid=426535046French(29April2011):http://fr.wikipedia.org/w/index.php?title=Sem_%28illustrateur%29&oldid=64820167“SergeGuinchard”English(current):http://en.wikipedia.org/wiki/Serge_GuinchardEnglish(30October2012):http://en.wikipedia.org/w/index.php?title=Serge_Guinchard&oldid=520621660French(22４４ October2012):http://fr.wikipedia.org/w/index.php?title=Serge_Guinchard&oldid=84445670“SN1054”English(current):http://en.wikipedia.org/wiki/SN_1054English(4April2011):http://en.wikipedia.org/w/index.php?title=SN_1054&oldid=422314182French(24March2011):http://fr.wikipedia.org/w/index.php?title=SN_1054&oldid=63635211“Sonora”English(current):http://en.wikipedia.org/wiki/SonoraEnglish(5September2011):http://en.wikipedia.org/w/index.php?title=Sonora&oldid=448622482Spanish(3September2011):http://es.wikipedia.org/w/index.php?title=Sonora&oldid=49494996“TheZombieKids”English(current):http://en.wikipedia.org/wiki/The_Zombie_KidsEnglish(20May2013):http://en.wikipedia.org/w/index.php?title=The_Zombie_Kids&oldid=555948565Spanish(current):http://es.wikipedia.org/wiki/The_Zombie_Kids４５ AppendixII从修订历史看维基百科翻译的趋势朱莉·麦克多诺·多尔玛雅加拿大多伦多约克大学（格伦顿学院）翻译学院维基百科是典型的众包案例，英语版已收录400多万个条目，此外还有284种语言版本。通常，独立运作的语言版本各自生成相应的目标语条目，或借助翻译途径生成。之前的研究表明，许多活跃于英语版的译者既没有接受过正式的培训，也没有专业的工作经验。因此，需对众包生成的译文质量做进一步探究。本文采用莫索普的编辑与修订程序分类法，对维基译文汇编而成的语料库进行研究，以确定转换错误和语言/排版错误在译文中出现的频率，并评估这些错误的解决方式。关键词：众包翻译；维基百科；修订历史；非专业翻译；用户自发翻译；志愿者翻译前言关于众包自发从事网络翻译的互联网用户，往往都不是专业翻译，与人合作完成大型翻译项目也（几乎）没有报酬。由此产生的译文日益成为翻译界探究的话题，并且这类合作方式名目繁多：合作翻译(DésiletsandvanderMeer2011)、社区翻译(Kelly,Ray,andDePalma2011)、众包翻译(Cronin2010;McDonoughDolmaya2012;Mesipuu2012)、粉丝翻译/词幕/译制(Dwyer2012;O"Hagan2008)、用户生成翻译(O"Hagan2009;Perrino2009)、志愿翻译(O"BrienandSchäler2010;Olohan2014)，维基翻译(Cronin2010)等等。诸如此类合作模式下的翻译一直在研究中。依靠众包发展的公司往往对这种模式赞不绝口。杰夫·豪威也指出，“这类工作的最佳人选应是最渴望这份工作的人[,]而评估这些员工表现的最佳人选则为他们的朋友和同事，他们会热心投入，促使结论达到最佳”(2008,8)。扎伊丹和凯里森·伯奇也从计算语言学和计算机科学的视角研究过译文质量的论题，但以这种方式产生的译文质量尚４６ 未成为诸多研究的主题，尤其是翻译研究领域。在翻译研究中，希门尼斯·克雷斯波研究过脸谱网（Facebook）的质量监控机。结论显示，在脸谱网的翻译质量评估方法中，“大量用户对推荐译文进行了正反两方面的投票，这有助于确保生成的网站符合各语种社区所需，并与现存的数字种类的预期特征相吻合”(147)。尽管当时没有评估译文质量，希门尼斯·克雷斯波(2013)之后比较了西班牙语版的脸谱网和西班牙本土25个社交网站的界面，重点关注动词形式和名词形式、祈使句动词和不定式动词的形式，以及用于交互式功能（如注销、登陆和修改个人资料）的词汇单位。有趣的是，他发现脸谱网本地化模型中使用的语言同西班牙本土网站的语言类似。这与他之前研究网站专业本地化为西班牙语的结论有所不同，本地化和非本地化网站中的动词时态和词汇差别显著。这表明，在某些情况下，众包模式能产生与原文相吻合的译文。当然包含多个平台，针对其他语言组合的较大研究更加有助于表明其是否普遍适用于众包。显然，非常有必要对众包翻译进行更多的评判与分析。若这个模型能够产生高质的译文，也许会改变翻译行业的生产、修改和传播方式。然而，众包翻译项目采用不同的方式来招募、激励和奖励用户，控制翻译质量。(cf.Mesipuu2012;McDonoughDolmaya2011)再加上翻译“质量”是一个相对概念(cf.e.g.Jiménez-Crespo2011,133-135)，所以要设计一个旨在以评估众包翻译质量或众包翻译过程有效性的研究非常困难。因此本文只研究维基百科这个众包翻译项目，并只专注修订这个翻译质量控制标准来解决这一问题。必须承认的是，由此得出的结论只适用于维基百科。本文通过使用莫索普的编辑和修订程序分类法来研究由94篇维基文章组成的语料库。旨在探寻维基平台上无偿翻译的业余译者(cf.McDonoughDolmaya2012)产生的译文是否需要修订，维基环境中开放式的编辑处理流程是否有助于消除译文中的错误，是否存在解决率较高的错误类型，维基百科这一翻译项目是否能为专业翻译市场提供一个可行的模式。关于维基百科作为内容完全依靠众包开发的网站，维基百科也许是最为典型的案例。网站成立于2001年，使用维基引擎，允许任何互联网用户通过网络浏览器添加、删除或编辑内容。如今，维基百科已成为世界上知名度最高的网站之一。2013年6月，互联网分析公司Alexa发布报告，称维基百科的访问量位列世界第七，紧跟脸谱网（Facebook）、谷歌（Google）、YouTube视频网站、雅虎（Yahoo!）、百度和亚马逊（Amazon），但领先推特（Twitter）、领英（LinkedIn）、必应（Bing）和Blogspot４７ 部落格平台。当时，维基百科已有285个语言版本，其中有八个语言版本超过100万篇，英语版本内容最多，超过400万篇。然而，这些版本中的内容往往由对应的目标语（TL）直接生成，同时也存在通过翻译生成的内容。维基百科解释如下：维基百科是一个多语言项目。语言不同而主题一致的文章可以独立生成，无需通过翻译，也无需在形式、排版或内容上一一对应。但翻译常在不同语言的文章之间起着信息传播的作用。（维基百科2013a）维基百科中，翻译文章同创作和编辑文章一样，都由用户自发。组织不会选择文章进行翻译，而是由互联网用户自发选择。维基用户可以选择翻译文章的部分或全部，然后将译文发布到相应的语言版本上。他们还可以通过很多正在进行的翻译项目，如“一周翻译”(http://meta.wikimedia.org/wiki/Translation_of_the_week)和“翻译工作坊”(http://en.wikipedia.org/wiki/Wikipedia:MED/Translation_project)等推选文章进行翻译。这种开放的翻译过程对于作者的研究既是优势也是劣势。过程开放意味着每次修改都有存档，每个用户都能通过“修订历史”或“对话”页对这些改变进行整理。这种开放性一方面为研究提供了包含多个语言组合的文章以做分析研究对象。另一方面，让翻译对比分析跨越了语言障碍。因为在不同语言组合中选择的翻译内容大相径庭，并且某些语言版本中用户较为活跃，翻译活动也较多。除此之外，任何时间各个语言版本中都进行着许多不同的翻译项目，若想要明确知道翻译活动的数量，最为常见的语言组合和最受欢迎的翻译主题都是非常困难的。因此本文集中研究从法语和西班牙语译为英语的翻译项目（“维基百科：需译为英语的页面”），深知由此得出的结论不适用于维基百科中的所有译文，甚至无法应用于相同的语言组合之间。本文选择维基百科作为研究重点有几个原因。首先，维基百科的资历深，其开放的平台能够记录用户每次编辑的痕迹，这意味着每篇译文的修订历史都有迹可循。其次，维基百科的翻译和排版指南可以在线免费获取。第三，专业译者只占维基译者的极少部分(McDonoughDolmaya2012)且翻译合作过程完全由用户自发，与脸谱网等其他众包翻译项目采用的专业译者监督机制不同(Jiménez-Crespo2013,30)。这意味着研究维基翻译实践能够让我们有机会观察业余译者群是否能够通过自身努力成功识别并修订译文错误。４８ 关于方法论莫索普（2006）的《翻译修订与编辑指南》为研究维基译者的修订成果提供了一个框架。首先，这本书主要采用描述性语言，而不是规定性语言(ibid.,4)。尽管作者希望此书的修订部分能够“普遍适用”(ibid.,7),但是全文着重强调了英语，这对旨在描述英语翻译错误的研究极为有用。其次，莫索普在书中描述了12种“修订参数”，可划分为修订者需要解决的四个错误类型(ibid.,125)：转换错误即通过比较源语文本和译语文本来判断译文的准确性和完整性；内容错误涉及找出事实错误、多余话语、矛盾内容和逻辑错误；语言/排版错误需检查译文通顺度，使译文贴近目标语读者，甄别不符合语言习惯的表达，消除语法错误和拼写错误并确保译文符合出版风格指南的要求；以及谋篇错误，即检查译文布局、格式错误和结构错误。四种类型极大地简化了分析过程。考虑到时间和空间错误，本文不能对莫索普的修订类别进行一一研究。事实上，莫索普本人认为(2006,141-142)，根据目标读者（如：专业人士或非专业人士），阅读目的和阅读时间，阅读方式，质量控制机制，专业修订者通常需优先考虑某些参数。考虑到维基百科是世界上浏览量最大的网站之一，专业人士和非专业人士都会阅读该网站上的内容，网站内容长期有效（除管理员删除的违规内容），有的用户匆匆浏览，有的用户则会仔细研读，这意味着译者应着重考虑事实错误这类的内容问题。同样，也应着重考虑语言/排版错误，如译文流畅度和术语错误。而之前的研究已探究过维基百科中的内容问题，其中贾尔斯的研究(2005)最为有名。本文选择转换错误和语言/排版错误作为研究重点原因如下：转换错误影响了译文的准确性和完整性，是最重要的特征(Mossop2006,126)，且未曾被重点研究过。在维基平台上识别出转换错误对于翻译来说也是个巨大的挑战，因为需要一丝不苟地比较源语版本与译入语版本，而这些版本通常存在于不同的页面。然而，维基用户却能快速识别和修订语言/排版错误，特别是那些对于源语言一知半解或完全不了解的用户。事实上，因为已经存在如文词编辑协会这样的用户组以及像维基百科“风格指南”(cf.Wikipedia2013b)这类的指导方针，维基编辑对这这两类错误十分重视。但是却没有解决转换错误的指导方针。通过对这两种修订参数进行对比，有助于突出维基众包平台的优势和劣势。为了研究维基译文中的转换错误和语言/排版错误，作者参考了“维基百科：需译为英语的页面”(http://en.wikipedia.org/wiki/Wikipedia:Pages_needing_translation_into_English)现在和以前的版本，汇编了近两年来从法语和西班牙语全文翻译或部分翻译成英语的文章作为语料库。“维基百科：需４９ 译为英语的页面”会定期更新列表，页面内容分为两类：“翻译”板块，包括已经发布到英语版维基百科上却被维基用户识别到包含非英语内容的文章，以及“清理”板块，包含（可能）已经译为英语但需要译后编辑的文章，因为这些译文通常由计算机翻译软件完成，或是由母语为非英语但对源语言掌握较差的译者所完成。尽管“维基百科：需译为英语的页面”上列出的文章来自其他283个语言版本，本文只选取了源语言标注为法语和西班牙语的文章进行分析。为了汇编语料库，作者参考了“维基百科：需译为英语的页面”2013年5月15日的版本，以及2011年5月15日至2013年4月15日期间每月15日的存档版本（页面历史），取样跨时两年。在此期间，存档页面中被归入“翻译”板块或“清理”板块且源语言为法语的文章一共有34篇，而源语言为西班牙语的文章有60篇。难道这意味着2011年5月至2013年5月期间维基百科中只有94篇文章从法语和西班牙语译为英语吗？这当然不可能，因为此页面中所列出的文章被维基用户视为“粗略翻译”或非英语文章。然而，在取样期内可能出现了新的译文或更加完整的译文。维基用户也许还未察觉到其他“粗略”翻译的存在，也没有发现由非英语母语者翻译且没有语法错误和句法错误的译文，否则这些译文早已被发布到此页面。因此尽管本文中的样文不能代表维基百科中的所有译文（甚至不能代表从法语和西班牙语译为英语的文章），但“维基百科：需译为英语的页面”提供了研究所需的样本，因为从中选取的译文都接受过某种程度的修订或编辑。尽管本文研究结论不具普遍适用性，但至少这些结论显示了译文在维基环境中经历的各种变化，因而可以论证是否可以将维基百科这个特殊的众包模式称之为一种高效的翻译模式。关于维基百科中的编辑，修订与改写从某种程度上来说，讨论维基百科中的编辑与修订是一件极具挑战的事情。因为，维基百科中使用的“edit”与“editing”同翻译研究学科中的意义大相径庭，只有主次之分。维基百科中还使用了令人费解的“修订历史”一词作为历史编辑页面的标题：[次要修订]包括排版、格式和表述的修订，在不改变文章内容的同时对文词进行重组。编辑公认，无需复审，且永远不会引起争议。相比之下，主要修订需要复审，且需得到所有相关编辑的认可。因此，任何影响文章意义的修改都不容小觑，即使只改动了一个词。例如，增加或删除“not”会改变句子的５０ 意义，必须视为主要修订。(Wikipedia2013c;originalemphasis)通过对比，莫索普(2006)将修订和编辑进行了区分，前者意为检查“译稿中的错误并进行适当修正”(201)，后者则是在“非译文本”中进行检查和修正(198)。其他学者也对此做出了种种区分（如Brunette和Gagnon2013）。根据莫索普的分类法，编辑将文本视为非译对象，重点关注文词本身，即确保文本符合本语言社区普遍认可的拼写、语法和结构指南；文体编辑，为读者调整词汇和句子结构以求简明扼要、语意明确；结构编辑，将文本进行重新组合以更好地表述材料；内容编辑，即明确建议文中需要增加或删除的区域以更好地覆盖手头的话题(Mossop2006,27-28)。当然，编辑和修订者的工作会重叠：例如，对语言/排版错误进行编辑和修订会涉及许多相同的工作。除此之外，一个人有时可以同时负责编辑和修订工作，一边解决转换错误，一边为目标读者重组文本。翻译指南和译文页面历史表明这个情况在维基百科中普遍存在。例如，“维基百科：翻译”页面为想要成为翻译的人提供了如下指导方针：维基用户无需进行全文翻译。若文章中的某些部分质量较差或无法验证，可自行判断是否翻译这些内容。完成翻译后，可以请修订员对译文进行检查。那么，如上所说，翻译（和修订者都将遵守相同的指导方针）也许会参与内容编辑，因为他们可以省译“低质”或“来源不明”的内容。反之也使得转换错误的修订变得棘手，因为在翻译过程中省略原文中的某些部分也许会让翻译和修订员对翻译的完整性产生歧义。必须注意的是，修订在维基百科中也许可行，因为翻译用户“也许会邀请修订员对译文进行检查”。但这个做法并不是强制性的，也没有得到强力推广。因为用户得到的建议只是“可以”而不是“必须”邀请某人来检查译文，是否联系修订员的决定权完全掌握在用户手里。语料库分析译文修订历史为了弄清语料库中的维基译文是否存在翻译错误（以及这些错误是否得到解决），作者参考了94篇译文的页面历史，记录了以下数据：(1)每篇文章从发布到“维基百科：需译为英语的页面”到本文数据记录截止时的“编辑”（依照维基百科的定义）次数；(2)５１ 期间编辑过文章的人数，详见表1。表1中“未知量”指的是发布到“维基百科：需译为英语的页面”后被删除的11篇文章。文章一旦被删，网络用户就再也无法浏览页面历史，也无法统计编辑次数和编辑人数。表194篇译文中的编辑次数和编辑人数编辑人数平均数（平均值）（西班编辑次数文章数最小数最大数平均数（平均值）平均数（平均值）（法语译文）牙语译文）用户机器用户机器用户机器用户机器用户机器≤153210954.912.065.52.14.642.0516-3020601359.252.69.829.072.831-49126120611.923.839.293.2915.64.650-100942>20714.884.5134164.8(大约.)(大约.)>10010164>2016>206>205.17>207.25未知数11----------总计94如表1所示，自发布当天到2013年5月持续两年时间内，编辑次数小于或等于30次的译文超过总数的1/2。另外，这些数据同时包含了网络用户和机器的编辑次数。全自动或半自动的电脑脚本设计可以运行重复任务，如检查侵权，纠正拼写错误，对网页进行重新分类并修护破坏。通常来说，如修改错误用词或前后矛盾的词语和误译等复杂的修订任务，机器目前还无法胜任。因此修改少于30次的文章因转换错误而得到解决的几率就减小了。为了更加全面的展示编辑过程中所涉及的人数，“编辑人数”一栏标明了用户和机器对译文修改的平均次数。标记为“大约”的数据为估算值，因为统计20个以上的远程用户较繁琐且毫无必要。如表1所示，通常一篇文会由几名用户经手，且至少经过一次机器编辑。此外，编辑次数增长的同时，编辑人数也在增长，这表明在大多数情况下修订工作是一个人类与机器共同合作的过程。５２ 语料库中的翻译错误如表1所示，有1/3的译文在发布到此页面后编辑次数都少于16，但这并不能说明什么。毕竟单次修改可能是订正排版错误之类的次要修订也可能是改写文章之类的主要修订。因此，单次修改可能解决了很多翻译问题。所以，作者在语料库的基础上汇编了子库做进一步研究。根据每篇文章的编辑次数将译文进行分类，每隔两篇汇编入子库，共计29篇（原文为法语的有11篇，原文为西班牙语的有18篇）。所有下文引用的子库文章链接详见附录。子库中的文章发表到此页面当天，作者就对源语文本和目标文本的页面历史进行了对比，以此来确定译文中是否含有需要解决的转换错误或语言/排版错误。更确切地说，即将源语文本片段和译语文本片段进行对比，以识别译文是否存在词汇翻译错误或误译和省译等转换错误，以及是否符合维基百科“风格指南”，是否包含不地道的语言、拼写/语法错误，或措辞不当等语言/排版错误。每篇文章中都存在3至4个典型的转换错误或语言/排版错误，这些错误可以作为错误样本。若一篇文章在发布到此页面当天包含未译文本，作者在发现首个完整翻译版本之前应参考最近的页面版本。若2013年五月之后译文中还存在未译文本即视为存在转换错误。表2总结了对比源语文本和译语文本后得到的结论。由于不同的修订员也许对每篇译文中存在的错误数量持不同看法，表2并未明确表示出译文中错误的总数量。但表2中显示了译文中是否可以至少找出一个转换错误或语言/排版错误的例子。通常来说，每篇文章中都能找出3至4个不同类型的错误；那么，将每篇译文的原始版本同2013年5月的版本相比较就能知道这些样本是否已经得到解决，结论将会在表3中显示。如表2所示，几乎所有文章都至少存在一个转换错误或语言/排版错误，而源语言为西班牙语的4篇译文除外。其中有两篇译文被发布到此页面上，但经过验证这2篇文章的原文并非出自西班牙语版维基百科。例如，“FilipeEspinosa”这篇文章的西班牙语版本和英语版本的内容相差甚远：西班牙语版本的长度是英语版本的4倍多，两个版本中的结构和很多细节都不同。在另一篇非译文章“Aviacióndelnoroeste”中，西班牙语版维基百科中根本找不到相关主题的文章，也无法在其他的网站上找到有关的西班牙语内容。于2011年7月28日发表于此页面的英语译文存在大量的语言/排版错误。这些文章很可能是由母语为非英语的用户直接创作，并非从西班牙语翻译而来，或是因为用户５３ 翻译的文本目前无法在网上获取。这意味着无法对这2篇缺乏源文本的文章进行转换错误和语言/排版错误分析。表2.维基译文原始版本中的错误不同错误类型伪译文转换错误语言/排版错误从法语译为英语的译文(11篇)11110从西班牙语译为英语的译文（18篇）15152全部译文（29篇）26262而另外两个特例中都只存在一种错误类型（非两者兼有），值得详细讨论。一开始英语版维基百科中直接显示了“MocopulliAirport”这篇西班牙语原文。文章只有50词，完全可以翻译成标准的机场信息框。另一名用户于4天后翻译了这篇文章。尽管当中存在一个语言错误（机场的英语名之后用括号附注的西班牙语名缺少重音符号），却没有转换错误。同样地，英语版维基百科中也直接显示了“theJoséJoaquinFlórezHernández”这篇西班牙语原文。文章一开始有120词，两周后全部被删且只剩一条短线，因为翻译工作并未取得任何进展。因此这篇文章存在转换错误（因为没有译文），但是没有语言错误。转换错误误译或“引起读者对于源语文本信息中重要特征误解”的段落(Mossop2006,126)在语料库中随处可见，占86篇转换错误样文的1/3。某些可能是由自动翻译软件生成，例如西班牙村庄Navahermosa这篇文章中包含了许多近乎难以理解的段落：“Navahermosa”，西班牙语版维基百科（2012年7月23日）AntonioLlorente,alcomentarlafrecuenciadeestetérminoenlazonadeSalamanca,ÁvilaylaparteoccidentaldeSegovia,sitúasuorigenenenlaslenguasdevetonesycarpetanos,aunqueplantealadudadesuorigenvascoenelqueexistelavoznaba“llanura,valle,barranco”.５４ 拼写错误［安东尼·奥略伦特，记录了这个词在萨拉曼卡、阿维拉、西塞戈维亚的使用频率，认为它来源于Vetones和Carpetanos两族使用的语言。尽管他认为这个词可能来源于巴斯克，因为naba这个词在巴斯克语中的意思是“平原，山谷或峡谷”］安东尼·奥略伦特评论了这个词在萨拉曼卡、阿维拉和西塞戈维亚的使用频率，将起源锁定于vetones和carpetanos两族所使用的语言，但否认其来自巴斯克，因为在巴斯克语中naba的意思是“平原，山谷，峡谷”。其他的误译可能由人为产生（至少编辑的主体是人），正如创造性作词家勒内路易·巴伦的文章中误译了“laGénérationautomatiquedemélodiesditespopulaires”这句话，但增加了首词母缩略词IRCAM的上下文信息：他将工作侧重于流行音乐自动化，而不是由IRCAM主导的实验音乐。他的工作不是创作新的实验音乐（法国声学与音乐研究中心对此领域进行了广泛地探讨），而是创作美妙的旋律或流行音乐。在86篇存在转换错误的样文中，另外三分之一涉及词汇翻译错误，如哥伦比亚篮球队Bukaros一文中，团队中的“entrenador”相当于英语中的“训练员”，而不是“教练”或者“经理”。以及西班牙乐队僵尸小孩一文中，将西班牙的Subterfuge唱片公司“compañíadiscográfica”错译为“录音棚”而“唱片公司”才是正确译法。而其他的文章中，如“法国总统就职典礼”存在许多词汇错误。包括将“prestationdeserment”译为“咒骂”，而“就职宣誓””或“宣誓就职(仪式)”更为准确；还有将“investiture”译为“竞争者”，而“就职典礼”才是正确译法，尤其适合应用于标题。最后，语料库中存在转换错误的样文有20%为省译，省略内容约占5%。例如，西班牙语版中关于拉斐尔·科雷亚的文章中，“Biografía”部分提到了科雷亚的儿子并将其称为“拉斐尔·米格尔”。而对应的英语版文章中，他的“早期职业生涯”部分把他的儿子叫做“米格尔”。尽管英语译文附注的信息框中将两个名词都列了出来。还丢失了其他的一些细节，如科雷亚的小学和初中都在天主学校就读，以及他曾在慈幼会当义工。语言和排版错误５５ 译文中存在许多语言/排版错误，大都与机制相关，即与维基百科的语法、拼写、排版(Mossop2006,137)或目标语的表达习惯(ibid.,136)不符。有的错误属于语法错误，如动词时态错误（例如，索诺拉省一文中“谁不会念‘夫人’呢”），句法错误（例如，雷内路易斯·巴伦一文中“[她]继续传播Boushadme”）或是复数形式错误（如纳瓦埃尔莫萨一文中的“3000school”）。其他错误还涉及名称前后不一，如Bukaros一文中“韦森特·迪亚兹·罗梅罗·库里森”在文章附注的信息框内将篮球队的总指挥称为“库里森·韦森特·迪亚兹·罗梅罗”。同样在SergeGuinchard一文中对“UniversitéPanthéon-Assas(Paris2)”有不同译法，包括“UniversityofPanthéon-Assas-Paris2”,“UniversityofParis2”,“UniversityPantheonParis-Assas-2”以及“Panthéon-AssasUniversity”。只要参考英语版维基百科就可以消除歧义，但却找不到这所大学的英文版维基百科的链接。同样，很多文章中还存在用词前后不一：如“Ouragan-ClassLandingPlatformDock”一文中将法语词汇“radier”同时用“interiorraftorradier”和“welldeck”表达，从而产生了子语言错误（专业术语错误），读者很可能因为无法理解两个术语代表同一个事物而产生困惑。语料库中的修订和编辑技巧现在我们可以得出英语文章一旦发表到此页面，通常都同时存在两个错误类型。我们需要研究这些错误在几个月（或几年）后是否已经得到解决。表3显示了维基用户编辑译文时解决最多的错误类型。由于表2中并没有列举出译文中所有的错误，只标明了某些类型的错误是否存在，表3则阐明了收集数据时样本占总数的比例，因为表2中的错误在之后已经得到解决。表3中显示语言错误比转换错误得到解决的几率较大。总的来说，这两种错误的解决量占取样截止日错误解决量的1/2。编辑次数小于或等于15次的文章错误解决率最低。通常来说，编辑次数越多，错误解决率就越高。有趣的是，在编辑次数超过100次的文章中转换错误几乎没有得到解决。这也许是因为样文中的文章数量太少，若数量足够就能适用前面的结论，因为只有1-2篇文章中的转换错误没有得到解决。从表3看，编辑和修改同时发生：毕竟单语用户能够独立解决像拼写错误、语法错误等语言错误，无需参考原文。但只有通过比较原文和译文才能解决像误译和省译等转换错误。Brunette,Gagnon,andHine(2005)都支持以上结论。但究竟是通过双语对比还是单语修订解决了这些错误，表4给出了明确答案。表4中将转换错误分为4类：未译和省译代表莫索普(2006,125)的完整参数，词、词、段翻译错误代表莫索普的准确参数(ibid)。５６ 只有通过仔细对比原文与译文才发现省译错误的解决率最低，而未译错误的解决率最高，自动化脚本能够轻易识别。误译和词汇或术语错误的解决率略超50%。这类错误有时不用通过修订只需通过编辑就可以得到解决，即无需对比译文与原文，只需将译文当做原文。例如，前文总统就职一文中存在“prestationdeserment/swearing”这个术语错误，之后通过将“swearing”改为“Oathofoffice”得到解决。但即使不参考原文，这种错误也能够得到解决。因为即使不懂法语，只要懂英语就可以从语境中猜到“swearing”指的是宣誓就职而不是诅咒。其实，从译文的“修订历史”可以看出者两种错误是通过编辑、修改、改写或三者结合得到解决的。表3：样文中得到解决的翻译错误转换错误语言/排版错误s编辑次文存在2013年5月存在存在2013年5已识别数章数错误的文前得到解决的错错误的文章错误的文章月前的错误解决错误数章数误率数数d率≤151292128.57%91926.32%16-30662564.00%61369.23%31-49441566.67%41475.00%50-100331172.73%31241.67%>100441450.00%41291.67%总计29268654.65%267060.00%５７ 表4子库中转换错误的类型和得到解决的几率发布到“维基百科：需译成2013年5月之013年5月之前转换错误类型英语的页面”后识别出的样文数前得到解决的样文数得到解决的样文比率未译内容33100.00%省译19631.58%词汇、术语311754.84%误译332163.64%总计864754.65%在进行内容编辑时，常常会误删转换错误。例如，“SN1054”自2011年4月4日发布到此页面到2013年5月30日本文数据收集结束之间有100多次编辑历史。因“编者按来历不明”、“阐述未附参考”和所谓事实错误，维基百科用户Dbachmann于2013年4月删除了4000多词内容，因而解决了文章中的许多转换错误。同样Navahermosa含有大量转换错误和语言错误，自发布到此页面两个月后被删除，只剩下一条短线。有趣的是，这些内容并非因为含有转换错误或语言错误而被删除，而是因为缺少信息来源。文章被删也是相同原因。自发布日到2013年5月期间有11篇译文被删，都无关译文质量，而是因存在广告成分，主题无意义，译文无进展，侵犯版权或缺少信息来源被删。改写则解决了其他错误。如2011年4月29日“Sem(艺术家)”法语原文中的“LifeandWorksofSEM”部分有108词。2011年9月3日经过改写后扩展为131词。2013年5月23日最后一次编辑后只剩39词，除少量细节（提供了AmedeedeNoe的参考依据和GeorgesGoursat与SEM签订工作的事实）一致之外，与原文相去甚远。无需通过修订，只需编辑就可以消除转换错误和语言/排版错误。而其他文章却证明存在双语修订。例如，于2012年6月27日发布到此页面的法国总统就职典礼一文，除了上文中提到的两处词汇错误之外，还存在许多误译。以下例子可以呈现错误的修订过程。“法国总统就职演讲”英语版维基百科（2012年6月27日）宪法中并无明文约束竞选者的行为。“法国总统就职演讲”英语版维基百科（2012年7月15日）５８ 法国宪法并未对竞选活动做出任何规定。上例中，2012年6月27日的版本表达还不够确切，但于2012年7月5日经过改正后的译文更加准确地传达了法语原文的含义。因此，作者强烈建议在修订过程中参考法语原文。发布后经过100多次编辑的文章出现双语版本的几率较高，单语编辑或改写较少。尽管表1中显示语料库中只有10%的文章经过100多次修订。但修订次数的多少并不取决于采样期内文章发布的时长，如8/10篇法译英和12/22篇西译英的编辑次数都少于16次，其中多数发布日期大于6个月，有些文章(如：“LesColoca-taires”，“OuraganClassLandingPlatformDock”,“Aviacióndelnoroeste”)甚至已超过两年。结论让我们回顾引言中提出的三个问题，分别是：维基用户生成的译文是否需要修订？维基环境中开放式的编辑过程是否有助于消除译文中的错误？以及，哪些错误类型得到解决的几率较高？第一个问题的答案显而易见：子库中有29篇译文都存在转换错误和语言/排版错误，只有4篇例外。因为大多数文章中都存在像误译和省译等错误，毫无疑问都需要对照原文才能解决。这不足为奇，因为发布在此页面的文章都需要“清理”或翻译。通过其他翻译项目或由维基用户自发生成的译文存在的错误较少。但这个结果意味着其他众包翻译项目存在转换错误的几率较大，特别是在第一次提交的版本中；同时也应设计一些平台为识别和修订这些错误提供便利。例如，在翻译过程和投票过程中脸谱网上的源语和目标语交叉排列，而维基百科中更加简洁明了。第二个问题不太好回答。由于94篇译文中有超过一半的编辑次数都小于或等于30，且到2013年5月子库中存在的转换错误和语言/排版错误的解决率只有50%，因此这两类错误可以在发布后存在几个月甚至几年。但这些错误不会永远存在，因为维基平台是开放式的平台，终会有人发现这些错误并改正、删除或消除它们。因此，永远都不能将这些译文视为最终版本，而应视为进化中的版本，可以从中发现和解决一些错误，也可能产生需要日后解决的新错误。毫无疑问，修订的速度非常缓慢，因为很多错误需要几周或几个月才能得到解决，有的甚至在两年后都未能得到解决。尽管过程缓慢，本研究证明了维基平台非常有利于翻译研究，因为研究人员能够便捷地追踪各个时期译文的变更以及探索超越本文范围的议题，如大语种和小语种，专业文本和普通文本的修订率。５９ 至于最后一个问题，通过对语料库的分析可以得出某些特定错误类型的解决率较低：省译的修订率极低，但语言错误经过多次编辑往往能够得到解决。这意味着“维基百科：需译为英语的页面”中的译文通过编辑得到解决的几率比通过修订得到解决的几率更高。然而，上文中已提到存在源语和目标语对照修订且转换错误有时可以得到解决，因此维基百科这个特殊的众包平台有时能够生成较准确且无语法错误的译文，但同专业翻译一样耗时。考虑到希门尼斯·克雷斯波(2013)对脸谱网（Facebook）的研究显示出的良好前景，加之布鲁内特·加尼翁和海恩(2005)支持双语对照修订反对单语编辑，最后结果显示众包翻译研究需要更多形式的对比研究，如转换错误和语言错误是否广泛存在于其他众包平台中，这些错误多久能够得到解决？关于用户界面设计，待译内容类型，源语和译语的组合，参与者的招募、动机、薪酬，这些平台之间有何相似之处？这些差别对译文有何影响？这些问题急需解答。尽管依赖众包的组织（企业、非盈利组织等）认为众包可以代替翻译-编辑-校对模式(Kelly,Ray,andDePalma2011)，但仍缺乏研究来验证这一说法。尽管本文只覆盖了维基译文的一个小型语料库，但仍能为翻译行业提供见解，特别是对非专业翻译的看法。例如，本文已经证实即使专业译者和双语用户不进行监督，维基用户能在时间充足的情况下解决译文中的转换错误和语言/排版错误。这表明当源语和目标语不需要完全对应时（如网站本地化）或者时间不构成障碍时，维基百科这样的众包平台能够同时接纳专业人士和非专业人士写作、编辑、翻译和改写内容，无需给予报酬。但就此引出了一些道德问题。互联网用户自愿为维基百科（即使是为脸谱网这样的盈利机构）等非盈利事业做贡献，但若不能提供一种方式来补偿他们的时间，特别是每周自愿贡献很多时间的用户（参看霍华德2014案例），鼓励他们这样做是不道德的。此外，寻找让众包翻译的生产和修订速度加快的方法可以增强非盈利项目与专业翻译的竞争力，也招致了专业人士的强烈不满（参看麦克多诺·多尔玛雅2011案例）。若薪酬这个棘手的问题得到解决，两者的合作可以提供诸多益处。例如，新手培训、译者显身、创新内容、以及专业翻译和用户都能够更方便地获取信息。６０ ReferencesBrunette,Louise,andChantalGagnon.2013.“Enseignerlarevisional’eredeswikis:laoul’ontrouvelatechnologiealorsqu’onnel’attendaitplus.”JournalofSpecialisedTranslation19:96-121.http://jostrans.org/issue19/art_brunette_gagnon.php.Brunette,Louise,ChantalGagnon,andJonathanHine.2005.“TheGREVISProject:ReviseorCourtCalamity.”AcrossLanguagesandCultures6(1):29-45.doi:10.1556/Acr.6.2005.1.3.Cronin,Michael.2010.“TheTranslationCrowd.”Revistatradumatica8.http://www.fti.uab.cat/tradumatica/revista/num8/articles/04/04.pdf.Désilets,Alain,andJaapvanderMeer.2011.“Co-CreatingaRepositoryofBestPracticesforCollaborativeTranslation.”LinguisticaAntverpiensia10:27-45.Drugan,Joanna.2011.“TranslationEthicsWikified:HowDoProfessionalCodesofEthicsandPracticeApplytoNon-ProfessionallyProducedTranslation.”LinguisticaAntverpiensia10:111-130.Dwyer,Tessa.2012.“FansubdreamingonViKi:‘Don’tjustwatchbuthelpwhenyouarefree.’”Translator18(2):217-243.doi:10.1080/13556509.2012.10799509.Giles,Jim.2005.“SpecialReportInternetEncyclopaediasGoHeadtoHead.”Nature438:900-901.doi:10.1038/438900a.“HowFansTranslated‘closetoamillionwordsofdialog’inNewBaldur’sGateLocalizations.”PCGamer,November16.http://rn.pcgamer.com/2012/11/16/baldurs-gate-contains-close-to-a-million-words-of-dialog-and-how-fan-translations-helped-the-enhanced-editionHoward,Dorothy.2014.“ThoughtsonWikipediaEditingandDigitalLabour.”http://en.m.wikipedia.org/wiki/Wikipedia:Thoughts_on_Wikipedia_Editing_and_Digital_LaborHowe,Jeff.2008.Crowdsourcing:WhythePoweroftheCrowdisDrivingtheFutureofBusiness.NewYork:CrownBusiness.Jiménez-Crespo,MiguelA.2011.“FromManyOne:NovelApproachestoTranslationQuality６１ inaSocialNetworkEra.”LinguisticaAntverpiensia10:131-152.Jiménez-Crespo,MiguelA.2013.“Crowdsourcing,CorpusUse,andtheSearchforTranslationNaturalness:AComparableCorpusStudyofFacebookandNon-TranslatedSocialNetworkingSites.”TranslationandInterpretingStudies8(1):23-49.doi:10.1075/tis.8.1.02jim.Kelly,Nataly,RebeccaRay,andDonaldA.DePalma.2011.“FromCrawlingtoSprinting:CommunityTranslationGoesMainstream.”LinguisticaAntverpiensia10:75-96.McDonoughDolmaya,Julie.2011.“TheEthicsofCrowdsourcing.”LinguisticaAntverpiensia10:97-110.McDonoughDolmaya,Julie.2012.“AnalyzingtheCrowdsourcingModelandItsImpactonPublicPerceptionsofTranslation.”Translator18(2):167-191.doi:10.1080/13556509.2012.10799507.Mesipuu,Marit.2012.“TranslationCrowdsourcingandUser-TranslatorMotivationatFacebookandSkype.”TranslationSpaces1:33-53.doi:10.1075/ts.1.03mes.Mossop,Brian.2006.RevisingandEditingforTranslators.2nded.Manchester:StJerome.O’Brien,Sharon,andReinhardSchäler.2010.“NextGénérationTranslationandLocaliza¬tion:UsersAreTakingCharge.”ProceedingsofTranslatingandtheComputer32.London:ASLIB(AssociationforInformationManagement).http://doras.dcu.ie/16695/1/Paper_6.pdf.O’Hagan,Minako.2008“FanTranslationNetworks:AnAccidentalTranslatorTrainingEnvironment?’InTranslatorandInterpreterTraining:Issues,MethodsandDebates,editedbyJohnKearns,158-183.London:Continuum.O’Hagan,Minako.2009“EvolutionofUser-generatedTranslation:Fansubs,TranslationHackingandCrowdsourcing.”JournalOfInternationalisationandLocalisation1(1):94-121.Olohan,Maeve.2014.“WhyDoYouTranslate?MotivationtoVolunteerandTEDTranslation.”TranslationStudies7(1):17-33.doi:10.1080/14781700.2013.781952.Perrino,Saverio.2009.“User-GeneratedTranslation:TheFutureofTranslationinaWeb2.0Environment.”JournalofSpecialisedTranslation12:55-78.http://jostrans.org/issue12/art_perrino.pdf.６２ Thicke,Lori.2012.“EnterpriseInnovators:Twitter’s400,000Translators.”MultiLingual23(8):12-13.Toury,Gideon.2012.DescriptiveTranslationStudies-AndBeyond.Reved.Amsterdam:JohnBenjamins.Wikipedia.2013a.“Wikipedia:Translation.”AccessedJune27.http://en.wikipedia.org/wiki/Wikipedia:TranslationWikipedia.2013b.“Wikipedia:ManualofStyle.”AccessedJuly3.http://en.wikipedia.org/wiki/Wikipedia:Manual_of_StyleWikipedia.2013c.“Help:MinorEdit.”AccessedJuly31.http://en.wikipedia.org/wiki/Help:Minor_editZaidan,OmarF.,andChrisCallison-Burch.2011.“CrowdsourcingTranslation:ProfessionalQualityfromNon-Professionals.”InVol.2ofHLT’11Proceedingsofthe49thAnnualMeetingoftheAssociationforComputationalLinguistics:HumanLanguageTechnologies,1220-1229.http://www.cs.jhu.edu/~ccb/publications/crowdsourcing-translation.pdf６３ Appendix1.TranslatedWikipediaarticlescitedinthispaper“ArielNan”English(current):http://en.wikipedia.org/w/index.php?title=Ariel_NanEnglish(13November2011):http://en.wikipedia.org/w/index.php?title=Ariel_Nan&oldid=460464543Spanish(onArielNan’suserpage):http://en.wikipedia.org/wiki/User:Arielnan/Ariel_Nan“Aviacióndelnoroeste”English(current):http://en.wikipedia.org/wiki/Aviaci%C3%B3n_del_NoroesteEnglish(28July2011):http://en.wikipedia.org/w/index.php?title=Aviaci%C3%B3n_del_Noroeste&oldid=441887938NoSpanisharticleonAviacióndelNoroeste“Bukaros”English(current):http://en.wikipedia.org/wiki/BukarosEnglish(6December2011):http://en.wikipedia.org/w/index.php?title=Bukaros&oldid=464354482Spanish(4December2011):http://es.wikipedia.org/w/index.php?title=B%C3%BAcaros_de_Santander&oldid=51886913“FilipeEspinosa”English(current):http://en.wikipedia.org/w/index.php?title=Felipe_EspinosaEnglish(26July2012):http://en.wikipedia.org/w/index.php?title=Felipe_Espinosa&oldid=504191175Spanish(12July2012):http://es.wikipedia.org/w/index.php?title=Felipe_Espinosa&oldid=57847643“FrenchprésidentialInauguration”English(current):http://en.wikipedia.org/wiki/French_présidential_inaugurationEnglish(27June2012):http://en.wikipedia.org/w/index.php?title=French_présidential_inauguration&oldid=499555643French(27June2012):６４ http://fr.wikipedia.org/w/index.php?title=Investiture_du_pr%C3%A9sident_de_la_R%C3%A9publique_fran%C3%A7aise&oldid=80216857“JoséJoaquínFlírezHernández”English(current):http://en.wikipedia.org/wiki/Jos%C3%A9_Joaqu%C3%ADn_Fl%C3%B3rez_Hern%C3%A1ndezEnglish(3December2011):http://en.wikipedia.org/w/index.php?title=Jos%C3%A9_Joaqu%C3%ADn_Fl%C3%B3rez_Hern%C3%A1ndez&oldid=463855088[note:thisversionisinSpanish]“Lescololcataires”English(current):http://en.wikipedia.org/wiki/Les_ColocatairesEnglish(27November2011):http://en.wikipedia.org/w/index.php?title=Les_Colocataires&oldid=462806260French(4November2011):http://fr.wikipedia.org/w/index.php?title=Les_Colocataires&oldid=71774418“MocopulliAirport”English(current):http://en.wikipedia.org/wiki/Mocopulli_AirportEnglish(18January2013):http://en.wikipedia.org/w/index.php?title=Mocopulli_Airport&oldid=533645447English(14January2013):http://en.wikipedia.org/w/index.php?title=Mocopulli_Airport&oldid=533038123(Note:thistextisinSpanish)“NationalUniversityofSanMarcos”English(current):http://en.wikipedia.org/w/index.php?title=National_University_of_San_MarcosEnglish(14May2011):http://en.wikipedia.org/w/index.php?title=National_University_of_San_Marcos&oldid=429032944Spanish(12May2011):http://es.wikipedia.org/w/index.php?title=Universidad_Nacional_Mayor_de_San_Marcos&oldid=46335739“Navahermosa”６５ English(current):http://en.wikipedia.org/w/index.php?title=NavahermosaEnglish(9September2012):http://en.wikipedia.org/w/index.php?title=Navahermosa&oldid=511495698Spanish(23June2012):http://es.wikipedia.org/w/index.php?title=Navahermosa_%28Toledo%29&oldid=58170723“Ouragan-ClassLandingPlatformDock”English(current):http://en.wikipedia.org/wiki/Ouragan-class_landing_platform_dockEnglish(21August2011):http://en.wikipedia.org/w/index.php?title=Ouragan-class_landing_platform_dock&oldid=445909654French(14May2011):http://fr.wikipedia.org/w/index.php?title=Classe_Ouragan&oldid=65350162“RafaelCorrea”English(current):http://en.wikipedia.org/wiki/Rafael_CorreaEnglish(23January2013):http://en.wikipedia.org/w/index.php?title=Rafael_Correa&oldid=534417683Spanish(19January2013):http://es.wikipedia.org/w/index.php?title=Rafael_Correa&offset=&limit=250&action=history“René-LouisBaron”English(current):http://en.wikipedia.org/w/index.php?title=Ren%C3%A9-Louis_BaronEnglish(3May2013):http://en.wikipedia.org/w/index.php?title=Ren%C3%A9-Louis_Baron&oldid=553356601French(current):http://fr.wikipedia.org/wiki/Rene-Louis_Baron“Sem(Artist)”English(current):http://en.wikipedia.org/wiki/Sem_%28artist%29English(29April2011):http://en.wikipedia.org/w/index.php?title=Sem_%28artist%29&oldid=426535046French(29April2011):http://fr.wikipedia.org/w/index.php?title=Sem_%28illustrateur%29&oldid=64820167“SergeGuinchard”English(current):http://en.wikipedia.org/wiki/Serge_GuinchardEnglish(30October2012):http://en.wikipedia.org/w/index.php?title=Serge_Guinchard&oldid=520621660French(22６６ October2012):http://fr.wikipedia.org/w/index.php?title=Serge_Guinchard&oldid=84445670“SN1054”English(current):http://en.wikipedia.org/wiki/SN_1054English(4April2011):http://en.wikipedia.org/w/index.php?title=SN_1054&oldid=422314182French(24March2011):http://fr.wikipedia.org/w/index.php?title=SN_1054&oldid=63635211“Sonora”English(current):http://en.wikipedia.org/wiki/SonoraEnglish(5September2011):http://en.wikipedia.org/w/index.php?title=Sonora&oldid=448622482Spanish(3September2011):http://es.wikipedia.org/w/index.php?title=Sonora&oldid=49494996“TheZombieKids”English(current):http://en.wikipedia.org/wiki/The_Zombie_KidsEnglish(20May2013):http://en.wikipedia.org/w/index.php?title=The_Zombie_Kids&oldid=555948565Spanish(current):http://es.wikipedia.org/wiki/The_Zombie_Kids６７'