中英文文献翻译-向自动文本输入错误分析添加上下文参考了解孩子如何输入错误_第1页
中英文文献翻译-向自动文本输入错误分析添加上下文参考了解孩子如何输入错误_第2页
中英文文献翻译-向自动文本输入错误分析添加上下文参考了解孩子如何输入错误_第3页
中英文文献翻译-向自动文本输入错误分析添加上下文参考了解孩子如何输入错误_第4页
中英文文献翻译-向自动文本输入错误分析添加上下文参考了解孩子如何输入错误_第5页
已阅读5页,还剩9页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

0AddingContexttoAutomatedTextInputErrorAnalysiswithReferencetoUnderstandingHowChildrenMakeTypingErrorsAkiyoKanoSchoolofComputing,Engineering&PhysicalSciencesUniversityofCentralLancashirePrestonakanouclan.ac.ukABSTRACTDespitetheenormousbodyofliteraturestudyingthetypingerrorsofadults,childrenstypingerrorsremainanunderstudiedarea.ItiswellknowninthefieldofChild-ComputerInteractionthatchildrenarenotlittleadults.Thismeansfindingsregardinghowadultsmaketypingmistakescannotsimplybetransferredintohowchildrenmaketypingerrors,withoutfirstunderstandingthedifferences.Tounderstandhowchildrendifferfromadultsinthewaytheymaketypingmistakes,typingdataweregatheredfrombothchildrenandadults.Itwasimportantthatthedatacollectedfromthecontrastingparticipantgroupswerecomparable.Variousmethodsofcollectingtypingdatafromadultswerereviewedforsuitabilitywithchildren.Severalissueswereidentifiedthatcouldcreateabiastowardstheadults.Toresolvetheseissues,newtoolsandmethodsweredesigned,suchasanewphraseset,anewdatacollectorandnewcomputerexperiencequestionnaires.Additionally,therewasalackofananalysismethodoftypingdatasuitableforusewithbothchildrenandadults.Anewcategorisationmethodwasdefinedbasedontypingerrorsmadebybothchildrenandadults.ThiscategorisationmethodwasthenadaptedintoaJavaprogram,whichdramaticallyreducedthetimerequiredtocarryouttypingcategorisation.Finally,inalargestudy,typingdatacollectedfrom231primaryschoolchildren,agedbetween7and10years,and229undergraduatecomputingstudentswereanalysed.Groupingthetypingerrorsaccordingtothecontextinwhichtheyoccurredallowedforamuchmoredetailedanalysisthanwaspossiblewitherrorrates.Theanalysisshowedchildrenhaveasetoferrorstheymadefrequentlythatadultsrarelymade.Theseerrorsthatarespecifictochildrensuggestthatdifferencesexistbetweenthewaysthetwogroupsmaketypingerrors.Thisfindingmeansthatchildrenstypingerrorsshouldbestudiedintheirownright.ACKNOWLEDGEMENTLikealltheses,thisthesiswastheproductofcontributionsbymanypeople.Iamgratefulthatthesepeopleareinmylifeandwereallkindenoughtohelpme.Withoutmymother,IwouldhaveneverfoundmyselfinEngland,letalonedoaPhD.Shehasalwaysbelievedinmeandsupportedmeineverything.ShehastaughtmethatIcoulddowhateverIwantedregardlessofanydisabilities.Sheisatrueinspirationto1me,andIwillforeverbethankfulthatIamherdaughter.MyDirectorofStudies,Dr.JanetCReaddeservesaspecialmentionforbeingthemostunderstandingsupervisoronecouldaskfor.Herabilitytoknowwhenastudentrequiresabreakandwhentheyneedapushwasthekeyinthecompletionofthisthesis.Herencouragementtowriteforandattendconferencesdrovethisresearchtoamuchhigherlevel.Additionally,IwasluckyenoughtohavereceivedadvicefromProf.AlanDixofLancasterUniversity,whoisafountainofideasandinspirations.Prof.ScottMacKenzieofYorkUniversity,taughtmethatIhadtoomanyvariablesandruthlesslyadvisedmeonhowtowriteacademicpapers.IwouldalsoliketothankDr.JacobWobbrockofUniversityofWashingtonandDr.PetrosRoussosofUniversityofAthensforkindlysharingtheirworkwithme.Iamveryfortunatetobepartofathrivingresearchgroup.IamindebtedtomyfriendandcolleagueNickyDaninoforbeingmyadvocate,keepingmesane,andingoodhumourthroughouttheprocess.GavinSimforteachingmetolookatthingsfromdifferentperspectives.MatthewHortonforassistingmeinmanyofthedatacollectionsessions.LornaMcKnightandJamesEThompsonfortheirfriendshipsandgivingmethesupportIneededtofinishthethesis.Finally,thisthesiswouldnotbeenpossiblewithouttheparticipationofnearly600childrenandstudents.Ithankeachparticipant,theirteachersandtheirfamiliesforallowingthestudiestotakeplace.DedicationThisthesisisdedicatedtomymother,andforallthosethatkeeponachievinginthefaceofdisabilities.1INTRODUCTIONAtthestartofthe20thCentury,whenthefieldoftypingresearchbegan,researcherswereuninterestedinunderstandinghowchildrentyped.Thiswasbecausechildrenrarelyusedthetypewriter.Inthosetimes,typing(ortypewriting)wasaskilltaughtto(mostlyfemale)adultsasapathwaytogainingemploymentastypistsandsecretaries.However,asthetypewriter,andthenthepersonalcomputer,becamemorewidelyused,childrenbegantotype.Inparticular,theintroductionofcomputersinschoolsinthe1950s(OSheaandShelf,1983)andintroductionofmicroprocessorsinthemid1970s,makingcomputersmoreaffordable,ensuredeveryschoolchildgainedexperienceonthekeyboard.Inthe2000s,childrenwhowerebornintoaworldofubiquitousdigitaldevicesacquiretypingskillswithnoformaltraining.Computers(andthustyping)havebecomeacorepartoftheBritisheducationalsystem,whichstateschildrenmustbetaughtICTfromthefirstyearoftheirprimaryschooleducation(DoES,1989).Vastbodiesofliteratureexistonanalysisoftypingerrorsmadebyadulttypists(Chapter3).Thesestudiesrangefromgatheringthecompletespectrumoftypingerrors,tothosefocusedonaspecificfewinanattempttounderstandthepsychologicalcausesoftheerrors.Withresearchspanningoveracentury,itisfairto2saythattypingerrorsmadebyadultsisawell-studiedarea.However,childrenstypingerrorsremainanunderstudiedarea.Manystudiesonthebenefitofusingcomputersineducation(WoodandFreeman,1932;Rowe,1959;Krevolin,1965;Roussos,1992)arguedthattheuseofcomputersimproveschildrensreading,writingandunderstandingofsubjects.However,thesestudiesdonotlookatthedifficultieschildrenexperiencewhentyping.Thereislittleworkonunderstandingthetypingerrorsmadebychildren.Whattypingerrorsdotheymostfrequentlymake?Dochildrenmakethesametypingerrorsasthosemadebyadults?Arethereanyerrorsspecifictoyoungchildren?Thisthesiswasmotivatedbytheseunansweredquestions.1.1.1StructureTointroducethemainthemesofthisthesis,Section1.2providesabriefbackgroundtotypingerroranalysis,andhighlightsthelackofstudiesinvestigatingchildrenstypingerrors.Section1.3discussesthemainissuessurroundingcomparingchildrenandadultstyping,whichwereaddressedinthisthesis.Section1.4formallyoutlinestheobjectivesofthisthesisandthechapterconcludesinSection1.5withasummaryofthechapterandabriefdescriptionofthestructureoftheremainingthesis.1.2TYPINGERRORANALYSISTypingerroranalysisreferstothestudyoftypingerrorsmadewhilstusingthekeyboardtoproduceabodyoftext.Mostfrequentlyinsuchstudies,participantsareshownabodyoftextandareaskedtotypethetextusingatypewriter(computerkeyboardinthelateryears).Thetextpresentedtotheparticipantandwhatwastypedarecomparedtoseeifanytypingmistakesweremade.Typingerroranalysishasbeenpopularintwofields-psychologyandHCI.Psychologistsfirststartedtoinvestigatetypingerrorsinthe1910s,whentypewritersbecamewidelyused.Sincetimewasmoneyandtypingerrorscosttimetocorrect,theywereinterestedinminimisingtheseerrors.Theysuggestedthattoreduceerrors,onemuststudythetypingerrorsthemselvestounderstandthecauses(Book,1925).Theearlierstudieswerelargelydescriptive,suchascountinghowmanytimeseachletterwassubstitutedforanother(Lessenberry,1928).Thissoondevelopedintoresearchersdefiningtheirownsetoftypingerrors(Chapter3willdiscusstheseindetail),whichwereappliedtorealtypingdatacollectedfromprofessionaltypists.Researchersmadeguessesastotheircause,andprovidedworkbooksfortypistsdesignedtoreduceparticulartypingerrors(Opfer,1932;Schoenleber,1932).Uptothisstage,comparisonsweremadeonlybetweentextthattheparticipantwasaskedtocopytype(presentedtext-PT)andthefinaltypedtextthatwasproduce(transcribedtext-TT).However,inthe1960s,anewsurgeofinterestarosefromtheabilitytorecordtimingsofeachkeypress.Thismeantresearcherswereabletostudyindetailtheinputstream(IS),containingeverykeypressedbytheparticipant.SpearheadedbyDonaldNormanandDavidRumelhartattheLNRResearchGroup,SanDiego,timingbetweenkeypressesbecamethecentralfocusinthemodellingof3cognitivefunctionsinvolvedintyping.Manyoftheseworksaresummarisedin(Salthouse,1986).Athirdsurgeininterestoccurredwiththerisingincreaseinmobilecomputingdevicessincethe1990s(MacKenzieandSoukoreff,2002b).Withtheexplosionofnewtextinputmethodsdesignedforthesedevices,itbecameimportanttobeabletoevaluateonemethodagainstanother(MacKenzie,2007).MacKenzieandSoukoreff(SoukoreffandMacKenzie,2001;MacKenzieandSoukoreff,2002a;SoukoreffandMacKenzie,2003;Soukoreff,2010)haveledthedevelopmentofvariousmeasuresoferrorrates-theratiooftypingerrorstotheamountoftypingdone.Algorithmswerealsodeveloped(SoukoreffandMacKenzie,2001;WobbrockandMyers,2006)toenablecomputationoftheseerrorrateswithease.TheconvenienceofautomatedcalculationoferrortypeshasmadethemincreasinglypopularinHCI.However,Soukoreff(2010)pointsoutthattheseerrorratesonlytellwhetheronemethodorparticipantisbetterthananother,nothoworwherethetwodiffer.Incontrasttothewell-studiedareaofadulttyping,littleworkexistsinstudyingthetypingerrorsofchildren.Roussos(1992)carriedoutseveraltypingstudieswithchildrenforhisthesis.Inthesestudies,hecountedthenumberoferrorscreatedbytheparticipantsasanevaluationmethod,butdidnotdefinehowhecountedthem.Readetal.(2001)studiedshortphrasescopiedbychildrenagedbetween6and10years.Thechildrenenteredtextusingfourdifferenttextinputmethods(mouse,keyboard,speechrecognitionandhandwritingrecognition).Toevaluatethesemethods,theydefinedsixerrortypesandclassifiedthetypingerrorsaccordingly.Theyfoundthatchildrendisplayeddifficultiesusingthechosentextinputmethods,suchasExecutionError(e.g.thechildpressedthekeyfortoolong).Theyalsofoundthatchildrendemonstratedexaggeratedversionsoferrorscommonlymadebyadults.Theypinpointedsourcesoferrortobeattheuser,thehardwareandthesoftware.Inanotherstudy,ReadandHorton(2006)foundthatteenagerscarryingoutasimilarphrase-copyingtaskweremorepronetoerrorsincontrollingtheirhandmovements,suchaspressingthekeynexttotheonetheyhadintendedonpressing.Althoughcomparisonsbetweenchildrenandteenagerswerenotmade,thesetwostudiesbyReadshowthereisadifferenceintherangeoferrorsthetwoagegroupsmake.However,thetwostudieshadasmallnumberofparticipants(12and18respectively).Additionally,thecurrentliteraturelacksdirectcomparisonsbetweenchildrenandolderparticipantgroupsusingexactlythesametask.Itwasimportanttoestablishwhetherchildrenmadethesamerangeoftypingerrorstoadultsornot,sinceiftheydidnot,thefindingsfromadultstudiescouldnotbedirectlyappliedtochildren.Thecontributionofthisthesisintermsofstudyingtypingisthedetailedinvestigationofthetypingerrorsmadebychildren,andhowthisdiffersfromtypingerrorsmadebyadults.ThiscomparisonismadeinChapter10.1.3COMPARINGCHILDRENANDADULTSForthefindingsofacomparisonstudybetweenchildrensandadultstypingtobevalid,itwascrucialtoensureboththedatacollectionandanalysismethodsproduced4comparabledatabetweenthetwoparticipantgroups.1.3.1SeekingaComparableDataCollectionMethodSincechildrenstypingerrorshavenotbeencompareddirectlywithadultstypingbefore,itwasunknownastowhetherornotthemethodsusedtocollecttypingfromadultswereappropriateforusewithchildren.Themostfrequentlyusedmethodingatheringtypingfromadultsistoshowthemabodyoftext(e.g.magazinearticles,businessdocumentation,asetofshortphrases)onpaperforthemtocopy.Thequestionshereforchildrenarewhattexttoshowthem,andhowtoshowthetext.Firstly,inorderforthetexttonotcauseanybiastowardstheadults,itwasimportanttoensurethelanguageusedandthelengthofthetextissuitableforchildren.Secondly,itwasunknownastowhetherornotchildrenwouldhaveanydifficultiesincopyingtextpresentedonpaper.MacKenzieandSoukoreff(2002b)arguedthatpaperto-screencopyinghasahigherFocusofAttention(FOA)thanscreen-to-screencopyingwherethepresentedtextisdisplayedonthesamescreenasthetranscribedtext.ItwasunknownastowhetherornotthisincreaseinFOAwouldcausedifficultiesinchildren.1.3.2SeekingaComparableDataAnalysisMethodTheselectionoftherightmethodtocategorisethetypingerrorswasalsoimportant.Firstly,tocaptureallthetypingerrors,thechosencategorisationmethodmustnotexcludeanyerrortypes.Secondly,togainanaccuratepictureofthetypingerrorsmade,themethodmustnotbreakdownlargererrorsintomanysmallertypingerrors.Anexampleofthisistoensurethatanomittedwordiscapturedasoneomittedwordratherthanseveralomittedletters.Thirdly,sincethecauseofchildrenstypingerrorshasnotbeenstudiedindetail,itwasdecidedthatthecategorisationmethodcouldnotmakeanyassumptionsastothecauses.Somecategorisationmethodsincludederrortypessuchaserrorofdistraction(Clem,1929)anddeviationfromcopy(Book,1925),whichforcedassumptionsastothecauseoftheerror.Finally,themethodmustnotassumethatchildrenhaveformaltypingtraining.Intheearlyhalfofthe20thcentury,whenmosttypistsweretrainedtousethetouch-typingmethod(Clem,1929),theyweretaughttouseparticularfingersforparticularbanksofkeysonthekeyboard.Instudyingthesetypists,itwasfairtoassumethatonecoulddefineerrorsbytheuseofwrongfingers.Examplesofsucherrortypesareadjacentlettersbyusingwrongfinger(Opfer,1932)andhomologouserror(Grudin,1983a)wherethecorrectfingeronthewronghandwasusedtotype(resultinginasubstitution).However,childrennowadaysarenolongertrainedintouch-typing.Thisresultsinmosttypistsusingwhateverfingerismostconvenienttotypetheletter.Itwasthereforenotappropriateforthemethodtoassumethatparticipantshavehadformaltouch-typingtraining.1.4THETHESISTheprimaryresearchquestionofthisthesiswasaretherenotabledifferencesbetweentypingerrorsmadebychildrenandadults.Severalobjectiveshelped5achievethisprimaryaim.Thesewere:Establishingadatacollectionmethodfortypingerrorsthatcausestheleastamountofbiasbetweenchildrenandadults.Establishingatypingerrorcategorisationmethodthatencompassesthewholerangeoftypingerrorsmadebychildren,withoutmakinganyassumptionsastotheircause.Automatingthedatacollectionandanalysisprocessasmuchaspossibletoensureanefficient,consistentandvalidstudy.Findanytypingerrorbehavioursthatarespecifictochildren.Thethesisofthisworkis:thereisasetoftypingerrorbehavioursthatarespecifictochildreninphrase-copyingtyping.Althoughchildrenandadultssharesomeerrortypes,thereareerrortypesthatarealmostexclusivelymadebychildren.Thisdifferencesuggeststhattheoriesonhowadultsmaketypingmistakescannotbeapplieddirectlytochildrenwithoutfirsttestingthetheorieswithrealchildrenstypingdata.1.4.1StructureAninvestigationofthemethodsusedtocollectandanalysetypingerrorsrevealedaplethoraofmethods.Chapter2reviewsthemethodsusedinotherstudiestocollecttypingdata.Thesemethodsweredesignedforusewithadultparticipants.Usingtheseadult-orientedmethodswithbothadultsandchildrenwouldhavecreatedabiastowardstheadults,invalidatingthestudy.Anewdatacollectionmethodwasrequiredthatcouldproduceascomparabledataaspossiblebetweenchildrenandadults.Anewphraseset,designedtobesuitableforparticipantsabovetheageofsixyearsoldisintroducedandevaluatedinChapter5.ThemethodofpresentingthesephrasestotheparticipantsonpaperisevaluatedinChapter6,andrejectedinfavourofpresentingthemonscreen.Twoquestionnaires(oneforadultsandoneforchildren)designedtogatherparticipantspreviouscomputerexperiencesareevaluatedinChapter7.Asecondliteraturereview(Chapter3)surveyedtheanalysismethodsusedintypingerrorstudies.Itrevealedasimilarlackofconsiderationofchildtypistsinthesecategorisationmethods.22categorisationmethodswereevaluatedbutnonefulfilledallthecriteriaforcategorisingtypingerrorsmadebychildren.Theexistingmethodswereeitherthoroughbutmadecausalassumptions,oronlydefinedaselectfewerrortypes.Furthermore,onlyoneofthesemethodshadeverbeenappliedtoyoungchildrenstyping.Therefore,itwasnecessarytodesignanewcategorisationmethodbasedonrealtypingerrorsmadebychildren(Chapter8).Usingthecategorisationmethodmanuallywashugelytimeconsumingandwaspronetoinconsistenciesandambiguities.Toaddresstheseissues,aprogramwasdevelopedtocarryoutthedetectionandcategorisation(Chapter9).Thisprogramisanextensionofexistingautomatedanalysers.Theexistinganalyserswereonlyabletocarryoutletter-levelanalysisandlackedthecontextualinformationsuchaswhetheranerrorwaspartofaconsecutivegroupoferrorsornot.Thenewanalyserisabletousecontextualinformationtoprovidemuchmoremeaningfulcategorisation.Italsocarriesoutsomebasicdisambiguationtaskstoreduceambiguitiesinaconsistent6manner.ThestudydescribedinChapter10usesallthetoolsdesignedandevaluatedinthethesistocomparethetypingerrorsmadebychildrenandadults.Itfoundthatchildrenandadultsdomakeadifferentrangeoferrortypes,andidentifiedtypingerrorbehavioursthatwereuniquetochildren.1.5CONCLUSIONSThischapterhasoutlinedthemotivationforthisthesisandtheresearchobjectivesitsetouttoachieve.Theresearchwasmotivatedbythelackofworkininvestigatingtypingerrorsmadebychildrenandcomparingthemtothosemadebyadulttypists.Thisthesisaimedtoestablishnewmethodsthatallowcomparabletypingdatatobecollectedandanalysedfromchildrenandadultstodiscoverwhetherornotthereareanymajordifferencesbetweenthetwogroupsoftypists.Theremainderofthisthesiscomprisesofthefirstthreechaptersdescribingtheliteraturereviewcarriedoutandconsiderationsmadetotheresearchmethodologyusedinthisresearch.Thethreechaptersfollowingthisdesignandevaluateanewdatacollectionmethodthatismorevalidforusewithbothchildrenandadultsthanpreviousmethods.Thisisfollowedbyachapteroncreatingacategorisationmethodbasedontypingerrorsmadebychildrenaswellasthosemadebyadults,alreadyobservedinliterature.Theprocessofautomatingthedetectionandcategorisationoftypingerrorsisthendiscussed,allowingforamuchfaster,moreaccurateandconsistentanalysisoftypingerrors.Finally,alarge-scalestudyusesthesetoolstoanswertheprimaryresearchquestion.7AkiyoKano,AddingContexttoAutomatedTextInputErrorAnalysiswithReferencetoUnderstandingHowChildrenMakeTypingErrors【J】,Human-ComputerInteraction,2009,5726:137-140向自动文本输入错误分析添加上下文参考了解孩子如何输入错误AkiyoKano中央兰开夏大学计算机学院工程与物理学科普雷斯顿akanouclan.ac.uk摘要尽管有大量的文献研究成年人的打字错误,孩子打字错误仍然是一个研究不足的领域。众所周知,在孩子-计算机交互领域中,孩子不是“小成人”。这意味着没有先了解孩子和成年人的差异,关于成年人如何输入错误的结果不能简单地转移到孩子身上。为了理解孩子与成年人在输入错误的方式上有什么区别,从孩子和成年人那里都收集了打字数据。重要的是,对照参与者组收集的数据,两者是可比的。用各种方法审查从成年人那里收集的打字数据是否适合孩子,确定了一些可能对成年人产生偏见的问题。为了解决这些问题,设计了新的工具和方法,例如新的短语集,新的数据收集器和新的计算机经验问卷。此外,缺乏适合孩子和成年人都能使用的打字数据的分析方法。基于孩子和成年人的打字错误定义了新的分类方法。这种分类方法然后被改编为Java程序,这大大减少了进行打字分类所需的时间。最后,在一项大型研究中,分析了来自231名年龄在7至10岁的小学生和229名本科计算专业学生的打字数据。根据他们发生的上下文对打字错误进行分组允许对错误率可能的更详细的分析。分析表明,孩子经常犯的一系列的错误,成年人却很少犯。这些特定于孩子的错误表明两个组之间的打字错误的方式存在差异。这个发现意味着孩子的打字错误应该自己研究。致谢像所有的论文一样,这篇论文是许多人贡献的产物。我很感激在我的生活中这些善良的、足以帮助我的人。没有我的母亲,我永远不会在英国找到自己,更不用说做8博士。她一直相信我,并支持我做的一切。她告诉我,我可以不管任何障碍做任何我想做的。她是我的真正灵感,我会永远感谢我是她的女儿。特别值得一提的是我的研究主任JanetCRead博士,作为最理解人们需求的监督者。她知道学生何时需要休息,什么时候需要推进是本论文完成的关键。她鼓励写和参加会议,使这项研究达到更高的水平。此外,我很幸运地接到兰开斯特大学的AlanDix教授的建议,他是想法和灵感的源泉。约克大学的斯科特麦肯齐教授告诉我我有太多的变化因素,并无情地建议我如何写学术论文。我还要感谢华盛顿大学的JacobWobbrock博士和雅典大学的Petr

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论