(7)Foreign-Language-Teaching-Theories

ForeignLanguageTeachingTheoriesChapter7LanguageTesting7.1Purposesoflanguagetesting•Measurethelanguagebehaviorofindividuals•Evaluateprograms•Summativeevaluation:evaluatingaprogramasawhole;measuringwhatanindividualhaslearntinthewholeprogram•Formativeevaluation:adjustingorguidingtheprogram;todeterminewhatthelearnerhasorhasnotlearnt•Placementtests:testsusedtodeterminetheinitialleveloftheproficiencyoflearnerswhohaveenrolledforaparticularcourseinordertobeabletoplacethematanappropriatelevelinaprogram•Aptitudetests:testsintendedtoprovideinformationonthelikelihoodofsuccessorfailureofalearnerinlanguagelearning•Thesetestsconsistofdifferenttestswhichmeasuresuchabilitiesas•(a)sound-codingability•(b)grammaticalcodingability•(c)inductiveability•(d)memorization(rote-learning)ability•Achievementtests:testsusedtomeasurehowmuchofalanguagealearnerhaslearntwithreferencetoaparticularcourseofinstruction(progresstests,finaltests)•Diagnostictests:testsdesignedtospecifythelearner’sparticularstrengthsandweakness,toshowwhatskillsorknowledgealearnerhasorhasnotmastered,andideallyspeaking,tohelpdiscoverthecausesoffailure.•Achievementtestsanswerthequestionhowmuchthelearnerknows,whilediagnostictestsanswerthequestionwhathe/sheknowsandwhathe/shedoesnotknow,andshouldbeabletoanswerthequestionwhyhe/sheknowsordoesnotknowsomething.•Generalproficiencytests:testsusedtoindicatewhatanindividualiscapableofdoingnowwithanL2,whetheranindividualissufficientlyproficientinanL2.•Suchtestsareusuallystandardizedforworld-wideuse.7.2Somegeneralrequirementsfortests•7.2.1Reliability:thedegreeofconsistency(internalconsistencyandconsistencyovertime)shownbytheresultsofatest.Atestissaidtoreliableifitgivesthesameresultswhenitisgivenondifferentoccasions,orwhenitismarkedbydifferentpeople.•Reliabilityhasmuchtodowiththeobjectivityoftests•Giventhatthetestersandthetestees,andthetestsituationsarenotvaried,wecanmeasurethereliabilityofatestbytestingandretestingthesametesteesattwodifferenttimes,andthencorrelatingtheirscores,--thehigherthecorrelationbetweenthetwoteststhehigherthereliability–calledtest-retestmethodwhichindicatesstabilityreliabilityortest-retestreliability.•Anotherwayistoadministerparallelorequivalentformsofatest(withitemsofthesamedegreeofdifficulty),theparallel-formsmethod,whichindicatestheequivalenceofthetwoformsofthetest,calledtheequivalencereliability.•Inestimatinginternalconsistency,asimpleprocedureisthesplit-halfmethod.Thetestisdividedintotwohalves,usuallyintooddandevennumberitems,andthescoresofthesetwosetsarecorrelated.•Spearman-BrownFormulartt=2rA，B—————1+A，Brtt=reliabilityestimatedbythesplithaltmethodrrA，B=thecorrelationofthescoresfromonehalfofthetestwiththosefromtheotherhalf•Kuder-RichardsonFormula20rtt=n---------(n-1)()st2i2-∑s2t--------------rtt=theKR20reliabilityestimaten=thenumberofitemsinthetestst2=thevarianceofthetestscores∑2=thenumberofthevariancesofallitems,whichmayalsobeallitems,whichmayalsobeexpressedas∑pqi•Kuder-RichardsonFormula21rtt=n---------(n-1)()s2t1--------------rtt=theKR21reliabilityestimaten=thenumberofitemsinthetest=themeanofscoresonthetests2i=thevarianceoftestscoresX__X2/n_X__•Factorsthatcontributetoagreaterdegreeofreliability:theobjectivityofthetest,thelengthofthetest,thequalityofindividualtestitems.•Oneoftheindexesfortestitemsisthedifficultyindex,whichindicatesthedegreeofdifficultyofeachitem.Itisobtainedbydividingthenumberofitemsscoredcorrectlybythetotalnumberoftestees.•∑Cr•P=——•N•P=difficulty,proportionorrect•∑Cr=thesumofcorrectresponses•N=thenumberofexaminees•Anotherindexisthediscriminationindexofeachitem:howwellitsortsoutthebetterstudentsfromtheweakerones.•Discriminationindexisobtainedthisway:todividethetesteesintofourgroups,fourquartiles:thetop25%,thebottom25%,andthetwomiddle25%s;thenfindhowmanyofthetopquartilegotanitemright,andhowmanyofthebottomquartilegottheitemright;theratiobetweenthetwonumbersisthediscriminationindex.Hc•D=——Lc•D=discriminability•Hc=thenumberofcorrectresponsesinthehighgroup•Lc=thenumberofcorrectresponsesinthelowgroupThedegreetowhichatestmeasureswhatitissupposedtomeasure,andnothingelse.Whencompilingatestonemustasktwoquestions:1)Whatpreciselydoesthetestmeasure?2)Howwelldoesitmeasure?•7.2.2Validity•Aretheinstructionsinthetestclear?•Aretheitemsambiguous?•Doesanyitemincludeseveralpossibleanswers?•Istheteststructuredsothatexamineesbeginwitheasyquestionsandproceedtomoredifficultones?•Dotheitemstestuseorknowledgeofthelanguage?•Aretheitemslinguisticallyuseful?•7.2.2.1Contentvalidity–aformofvaliditywhichisbasedonthedegreetowhichatestadequatelyandsufficientlymeasurestheparticularskillsorbehavioritsetsouttomeasure.(skillsorbehaviortobetestedclearlydescribedandeachofthemgivenapercentageweighting)•7.2.2.2constructvalidity•Aformofvaliditywhichisbasedonthedegreetowhichtheitemsinatestreflecttheessentialaspectsofthetheoryonwhichthetestisbased.•7.2.2.3Empiricalvalidity•Ameasureofthevalidityofatestwhichisarrivedatbycomparingthetestwithoneormoreexternalcriteria.Iftheresultsofatestcorrelatewithsomeexternalcriteria,thetestcanbesai