reflection on language testing.docx
《reflection on language testing.docx》由会员分享,可在线阅读,更多相关《reflection on language testing.docx(8页珍藏版)》请在冰豆网上搜索。
reflectiononlanguagetesting
ReflectionsonLanguageTesting
Languagetesting,authoredbyTimMcnamara,isdividedintoeightchapters.Thereisacloserelationshipamongthem.Theyformawholepart.Next,Iwillwriteasummaryforeachoftheminthefollowingparagraphs.
Chapteronepresentsuswiththedefinitionoflanguagetesting,typesoftest,testpurpose,thecriterionandthetest-criterionrelationship.Weallknowthattestingisauniversalfeatureofsociallife.Testingforpurposesofdetectionortoestablishidentityhasbecomeanacceptedpartofsport(drugstesting),thelaw(DNAtests,paternitytests,liedetectiontests),andotherfields.Testtoseehowapersonperformsparticularlyinrelationtoathresholdofperformancehavebecomeimportantsocialinstitutionsandfulfillagatekeepingfunctioninthattheycontrolentrytomanyimportantsocialroles.Whatistrueoftestingingeneralistrueoflanguagetesting.Languagetestingplaysapowerfulroleinmanypeople’slives,actingasgatewaysatimportanttransitionalmomentsineducation,inemployment,andinmovingfromonecountrytoanother.Sincelanguagetestsaredevicesfortheinstitutionalcontrolofindividuals,itisclearlyimportantthattheyshouldbeunderstood,andsubjectedtoscrutiny.Therefore,anunderstandingoflanguagetestingisrelevantbothforthoseactuallyinvolvedincreatinglanguagetests,andalsomoregenerallyforthoseinvolvedinusingtestsorcontexts.Notalllanguagetestsareofthesamekind.Intermsofmethod,wecanbroadlydistinguishtraditionalpaper-and-pencillanguagetestsfromperformancetests.Paper-and-pencillanguageteststaketheformofthefamiliarexaminationquestionpaper.Theyareusuallyusedforassessmenteitherofseparatecomponentsoflanguageknowledge(grammar,vocabulary,etc.)orofthereceptiveunderstanding(listeningandreadingcomprehension).Inperformancebasedtests,languageskillsareassessedinanactofcommunication.Performancetestsaremostcommonlytestsofspeakingandwriting,inwhichamoreorlessextendedsampleofspeechorwritingiselicitedfromthetest-taker,andjudgedbyoneormoretrainedratesusinganagreedratingprocedure.Languagetestsalsodifferaccordingtotheirpurpose.Achievementtestsareassociatedwiththeprocessofinstructionandproficiencytestslookstothefuturesituationoflanguageusewithoutnecessarilyanyreferencetothepreviousprocessofteaching.Testsarebasedontheoriesofthenatureoflanguageuseinthetargetsettingandthewayinwhichthisisunderstoodwillbereflectedintestdesign.
Chaptertwointroducescommunicationandthedesignoflanguagetests.Thetermtestconstructreferstothoseaspectsofknowledgeorskillpassedbythecandidatewhicharemeasured.Therewasatendencytoatomizeanddecontextualizetheknowledgetobetested,andtotestaspectsofknowledgetobetested,andtotestaspectsofknowledgeinisolation.Materialtobetestedwaspresentedwithminimalcontext,forexampleinanisolatedsentence.Thispracticeoftestingseparate,individualpointsofknowledge,knownasdiscretepointtesting,wasreinforcedbytheoryandpracticewithinpsychometrics,theemergingscienceofthemeasurementofcognitiveability.Thediscretepointtraditionoftestingwasseenasfocusingtooexclusivelyonknowledgeoftheformallinguisticsystemforitsownsakeratherthanonthewaysuchknowledgeisusedoachievecommunication.Theneworientationresultedinthedevelopmentoftestswhichintegratedknowledgeofrelevantsystematicfeaturesoflanguage(pronunciation,grammar,vocabulary)withanunderstandingofcontext.FurtherOllerproposedwhatcametotheknownastheUnitaryCompetenceHypothesis,thatis,thatperformanceonthewholerangeoftests(whichhetermedpragmatictests)dependedonthesameunderlyingcapacityinthelearner---theabilitytointegrategrammatical,lexical,contextual,andpragmaticknowledgeintestperformance.Hearguedthatcertainkindsofmoreeconomicalandefficienttests,particularlytheclozetest,measuredthesamekindsofskillsasthosetestedinproductivetestsofthetypeslistedabove.Itseemsthattherearenoshortcutsinthetestingofcommunicativeskills.Fromtheearly1970s,anewtheoryoflanguageandlanguageusebegantoexertasignificantinfluenceonlanguageteachingandpotentiallyonlanguagetesting.ThiswasHymes’stheoryofcommunicativecompetence,whichgreatlyexpandedthescopeofwhatwascoveredbyanunderstandingoflanguageandtheabilitytouselanguageincontext,particularlyintermsofthesocialdemandsofperformance.Variousaspectsofknowledgeofcompetencewerespecifiedintheearly1980sbyMichaelCanaleandMerrillSwaininCanada:
grammaticalorformalcompetence,sociolinguisticcompetencestrategiccompetence,anddiscoursecompetence.Furtheryearsofdiscussionandreflectiononthisframeworkhaveledtoitsmoredetailedreformulation.Thisreworkingoftheideaofstrategiccompetencehasimportantimplicationsforassessment.Anotherdevelopmenthasbeentheattempttocharacterizetherealtasksinthecriterionsituationidentifiedthroughjobanalysisintermsoftheaspectofabilityorskilltheycallupon.
Chapterthreetellsushowtodesignandintroduceatest,whichinvolvesadesignstage,aconstructionstageandatry-outstage.Beforetheybeginthinkingindetailaboutthedesignofatest,testdeveloperswillneedtogetthelayoftheland,thatis,toestablishtheconstraintsunderwhichtheyareworking,andunderwhichthetestwillbeadministered.Fromapracticalpointofviewtestdesignbeginswithdecisionsabouttestcontent,whatwillgointothetest.Establishingtestcontentinvolvescarefulsamplingfromthedomainofthetest,thatis,thetestofthetasksortasksorthekindsofbehaviorsinthecriterionsetting,asinformedbyourunderstandingofthetestconstruct.Thenextthingtoconsiderintestdesignisthewayinwhichcandidateswillberequiredtoresponsetothematerials.Onceagain,testdesigninvolvesasortofprincipledcompromise.Alldesignsabouttestmethodinsuchacontextinevitablyinvolveacompromisebetweenthedesirabilityofanappearanceofauthenticityontheonehandandthepracticalitiesimposedbythetestsituationontheotherhand.Notethatthewayweresolvethiscompromisemayhavetheundesiredeffectofjeopardizingthefairnessoftheconclusionswereachaboutindividualcandidates.Obviously,asassessmentbecomesmoreauthentic,italsobecomesmoreexpensive,complex,andpotentiallyunwieldy.Analternativetograpplingwiththedilemmaofauthenticityofresponseinvolvesacceptingtoagreaterorlesserdegreetheartificialityofthetestsituation,andusingarangeofconventionalandpossiblyinauthentictestformats.Theresultofthedesignprocessintermsoftestcontentandtestmethodisthecreationoftestspecifications.Thefourthstageistrailingortryingoutthetestmaterialsandprocedurespriortotheiruseunderoperationalconditions.Thisstageinvolvescarefuldesignofdatacollectiontoseehowwellthetestisworking.Inaddition,test-takerfeedbackshouldbegatheredfromthetrialsubjects,oftenbyasimplequestionnaire.Materialsandprocedureswillberevisedinthelightofthetrials,inpreparationfortheoperationaluseofthetest.Inthischapterwehaveexaminedtheprocessthroughwhichatestingprocedureisconceptualized,developed,andputintooperation.
Chapterfourdiscussestheratingprocess.Therearethreeaspectsofthevalidationofratingprocedure:
theestablishmentofratingprotocols;exploringdifferencesbetweenindividualraters,andmitigatingtheireffects;andunderstandinginteractionsbetweenratersandotherfeaturesoftheratingprocess.Rater-mediatedassessmentisbecomingmoreandmorecentraltolanguagelearningandteaching.Rater-mediatedassessmentisalsoinlinewithinstitutionaldemandsforaccountabilityineducation,asoutcomesofeducationalprocessesareoftendescribedintermsofdemonstrablepracticalcompetenceinthelearner.Introducingtheraterintotheassessmentprocessisbothnecessaryandproblematic.Itisproblematicbecauseratingsarenecessarilysubjective.Inestablishingaratingprocedure,weneedtoconsiderthecriteriabywhichperformanceatagivenlevelwillberecognized,andthentodecidehowmanydifferentlevelsofperformancewewishtodistinguish.Itisusefultoviewachievementasacontinuum.Theassessmentisessentiallynotcompetitive.Thefunctionoftheassessmentatagivenlevelisnottomakedistinctionsbetweencandidates,otherthanabinarydistinctionbetweenthosewhomeettherequirementsofthelevelandthosewhodonot.Mostoften,frameworksforratingaredesignedasscales,asthisallowsthegreatestflexibilitytotheusers,whomaywanttousethemultipledistinctionsavailablefromascale,orwhomaychoosetofocusononlyonecut-pointorregionofthescale.Thepreparationofthescaleinvolvesdevelopingleveldescriptors.Animportantaspectofascaleisthewayinwhichperformanceatthetopendofthescaleisdefined.Thenumberoflevelsonaratingscaleisalsoanimportantmattertoconsider.Thissinglereportingscalemaymaintainitsanalyticorientationinthattheoverallcharacterizationofaleveldescriptionmayconsistofawavingtogetherofstrandsrelatingtoseparateaspectsofperformance.Animportantwaytoimprovethequalityofrater-mediatedassessmentschemesistoprovideinitialandtrainingtoraters.Thisusuallytakestheformofamoderationmeeting.Moderationmeetinghasthefunctionofbringingaboutbroadagreementontherel