EXPERIMENTreadnovel

Memory&Cognition1983,11(4),382-389Phonemic-analysistraininghelpschildrenbenefitfromspelling-soundrulesREBECCATREIMANIndianaUniversity,Bloomington,IndianaandJONATHANBARONUniversityofPennsylvania,Philadelphia,PennsylvaniaIthasbeenfrequentlysuggestedthattheabilitytoanalyzespokenwordsintophonemesfacilitateschildren'slearningofspelling-soundrules.
Thisresearchattemptstodemonstratethatlinkbyshowingthatphonemic-analysistraininghelpschildrentakeadvantageofspelling-soundrulesinlearningtoread.
Intwoexperiments,preschoolandkindergartenprereadersparticipatedinananalysisconditionandacontrolconditiononeachof4testdays.
Intheanal-ysiscondition,childrenlearnedtosegment(andinExperiment2,alsotoblend)selectedspokensyllables.
Inthecontrolcondition,theymerelyrepeatedsyllables.
Childrenwerethenintro-ducedtoprinteditemsthatcorrespondedtothespokensyllableswithwhichtheyhadworked.
Thepronunciationofthe"related"itemcouldbededucedfromthoseofotherprinteditemsintheset;thepronunciationofthe"unrelated"itemcouldnotbesodeduced.
Bothexperimentsrevealedasignificantinteractionbetweencondition(analysisvs.
control)anditemtype(relatedvs.
unrelated).
Inthecontrolcondition,childrentendedtomakemoreerrorsontherelateditemthanontheunrelateditem;intheanalysiscondition,theytendedtomakefewererrorsonthere-lateditemthanontheunrelateditem.
Theseresultssuggestacausallinkbetweentheabilitytoanalyzespokensyllablesandtheabilitytobenefitfromspelling-soundrelationsinreading.
Itisimportantforyoungreaderstolearntherulesthatrelatespellingstosounds.
Childrenwhoareabletousetheserules,dubbed"Phoenicians"(Baron,1979),canoftendecodeprintedwordsthattheyhaveneverseenbefore.
Childrenwhoarelessabletousespelling-soundrules,or"Chinese,"areforcedtorelyonword-specificassociations-memorizedassoci-ationsbetweenindividualprintedwordsandtheirpronunciationsand/ormeanings.
Readerswhode-pendonsuchassociationsshouldhavedifficultyasthesizeoftheirwrittenvocabularyincreasesandasnewwordsmustbedeciphered(Rozin&Gleitman,1977).
Empiricalevidencesupportstheviewthat"Phoenicians"tendtobebetterreadersofanalpha-beticwritingsystemthanare"Chinese.
"Firth(1972),WethankAnnFarrell,ShellieHaut-Rogers,IvyKuhn,SharonPacker,andBeckyTaggartfortheirassistanceinvariousphasesofthisresearch.
KathyHirsh-Pasek,DavidPisoni,andLindaSmithgavevaluablecommentsonearlierdraftsofthemanuscript.
ThisresearchwassupportedbyNIHGrantMH-29453toJ.
B.
,GrantBRSGS07RR07031toR.
T.
,andNSFGrantBNS81-09892toR.
T.
PortionsofthesedatawerepresentedatthemeetingoftheSocietyforResearchinChildDevelopment,Boston,April1981.
R.
Treiman'smailingaddressis:DepartmentofPsychology,IndianaUniversity,Bloomington,Indiana47405;J.
Baron'smail-ingaddressis:DepartmentofPsychology,UniversityofPennsyl-vania,Philadelphia,Pennsylvania19104.
forexample,foundthattheabilitytopronouncenonsensewordsaccountedforabout75010ofthevari-anceinreadingabilityamongagroupof916-year-olds.
Furthermore,childrenwithspecificdifficultiesinlearningtoread,ordyslexics,aregenerallypooratusingspelling-soundrules(Boder,1971,1973;Snowling,1980).
Giventhatspelling-soundrulesplayacentralroleinlearningtoread,astheabovestudiesindicate,itisimportanttounderstandwhysomechildrenhavedifficultylearningtheserules.
Onecontributingfac-tormaybetheirlackofphonemic-analysisability.
InalphabeticlanguagessuchasEnglish,writtenlet-terstypicallyrepresentphonemes.
1Manyinvesti-gatorshavearguedthatchildrenhavetroublelearn-ingthecorrespondencesbetweenlettersandphonemesbecausetheylackaccesstoaphonemicrepresenta-tionofspeech(Elkonin,1973;Gleitman&Rozin,1977;Golinkoff,1978;Liberman,Liberman,Mattingly,&Shankweiler,1980;Rozin&Gleitman,1977;Treiman&Baron,1981;Wallach&Wallach,1979).
Indeed,variousstudiesindicatethatchildrenwhohavenotyetlearnedtoread,orchildrenwhoarejuststartingtoread,arepoorinphonemicanalysis.
Forexample,suchchildrenoftenfailtoproduceorappreciaterhymes(e.
g.
,Calfee,Chapman,&Venezky,1972).
Theyhavedifficultyinjudgingthenumberof382Copyright1983PsychonomicSociety,Inc.
phonemesinaspokenword(e.
g.
,Liberman,Shankweiler,Fischer,&Carter,1974;Treiman&Baron,1981),andtheydopoorlyinsegmentationandblendingteststhatrequirethemtopronouncetheseparatephonemesinawordortorecognizeawordwhentheyaregivenastringofphonemes(e.
g.
,Fox&Routh,1975;Hardy,Stennett,&Smythe,1973).
Consistentwiththeviewthatphonemicanalysisiscriticalintheacquisitionofspelling-soundrules,whichareimportantforreadingsuccess,measuresofphonemic-analysisabilitycorrelatewithscoresonstandardreadingtests(e.
g.
,Calfeeetal.
,1972;Calfee,Lindamood,&Lindamood,1973;Firth,1972;Fox&Routh,1975;Helfgott,1976;Jusczyk,1977;Rosner&Simon,1971).
Furthermore,phonemicanalysiscorrelatesmorehighlywithabilitytousespelling-soundrulesthanwithabilitytouseword-specificassociations(Baron&Treiman,1980;Treiman&Baron,1981).
Thatis,Phoenician-stylereadersex-ceedChinese-stylereadersinphonemicawareness.
However,noneoftheabovestudiesconclusivelydemonstratedadirectcausallinkfromphonemicawarenesstothelearningofrules.
Indeed,someevi-dencesuggestsalinkinthereversedirection.
Thatis,peoplemaydowellonphonemic-analysistestspreciselybecausetheyknowhowtoread(Baron&Treiman,1980;Ehri&Wilce,1979;Morais,Cary,Alegria,&Bertelson,1979).
Itisatleastpossiblethatthisreversecausallink-phonemicanalysisasare-sultoflearningtoread-couldexplaintheentirecor-relationbetweenphonemicanalysisandspelling-sound-ruleuse.
Tofindoutifacausallinkexistsfromphonemic-analysisabilitytoabilitytolearnspelling-soundrules,otherexperimentalapproachesarenecessary.
Oneapproachistotrainprereadersinphonemicanalysisandtodeterminewhetherthistrainingim-provestheirlearningofspelling-soundrules.
Inonestudyusingthisapproach,Rosner(Note1)divided16first-graderswhowerenotyetabletoreadintotwogroups.
Childrenintheexperimentalgroupweretrainedfor14weeksintheanalysisofspokenwordsintosyllablesandphonemes.
Childreninthecontrolgroupreceivednospecialinstruction.
Onasubse-quentreadingtest,thetrainedchildrenoutscoredtheuntrainedchildrenbyasubstantialmargin.
Theseresultsarepromising,buttherearetworeasonswhytheyfailtoconclusivelyshowalinkfromphonemicanalysistospelling-sound-ruleacquisition.
Onerea-sonstemsfromtheuseofabetween-groupsdesign.
Theextraattentiongiventheexperimentalgroupmayhavehadgeneraleffects(e.
g.
,increasedmo-tivationorconfidence)thatinturncausedthesu-periorreadingperformance.
Thesecondproblemisthatthereadingtestdidnotdistinguishuseofspelling-soundrulesfromuseofword-specificassociations.
Byhypothesis,phonemic-analysistrainingshouldPHONEMIC-ANALYSISTRAINING383specificallyaffectthelearningofspelling-soundrules.
AstudybyMcNeilandColeman(Note2)partiallysolvedtheseproblemsbyusingtwocontrolgroups.
Theexperimentalgroupof30kindergarteners(GroupE)received3weeksofphonemic-analysistrainingfollowedby3weeksofreadinginstruction.
Onecontrolgroupof30children(GroupC1)re-ceivedthe3-weekreadingprogramtwiceinsucces-sion,foratotalof6weeksofinstruction.
Asecondcontrolgroupof30children(GroupC2)didtheread-ingprogramonce.
Onafinalreadingtest,GroupEperformedsignificantlybetterthandidGroupsC1andC2.
ThesuperiorityofGroupEoverGroupC1couldnothavebeenduetothetimedevotedtospe-cialtraining-thetwogroupswereequalinthisre-gard.
However,thechildreninGroupC1,whodidtheidenticalreadingprogramtwice,mayhavesuf-feredadversemotivationaleffectsduetotherepeti-tionandlackofvarietyintheirreadinginstruction.
Clearly,moreconvergingevidenceisneededtocon-cludestronglythatphonemic-analysistrainingaf-fectsthelearningofspelling-soundrulesratherthan(ortoagreaterdegreethan)someotherreading-relatedskill.
Thepurposeofthepresentexperimentswastopro-videadditionalevidenceonphonemic-analysistrain-ingandthelearningofspelling-soundrules.
TheseexperimentsdifferedfromthoseofRosner(Note1)andMcNeilandColeman(Note2)intwoways.
First,thetrainingmanipulationswerecarriedoutwithinsubjects.
Eachchildreceivedphonemicanalysisin-structionwithcertainsetsofspokensyllablesandnosuchinstructionwithothersetsofsyllables.
Second,thereadingtaskswerespecificallydesignedtomea-surechildren'suseofspelling-soundrules.
Transferofspelling-soundrulesfromonewordtoanotherservedastheindexofruleuse.
'Suchtransferis,foreducationalpurposes,thebestmeasureofchildren'suseofspelling-soundrules.
Wearesatisfiedthatchildrenhavelearnedruleswhentheycantransferthemtonovelitems-whentheycandecodenewwordsonthebasisofpreviouslylearnedwords.
Wepredictedthatchildrenwouldbemorelikelytotrans-ferspelling-soundruleswhentheyreadnovelwordswhosespokenformstheyhadbeentrainedtoanalyzethanwhentheyreadnovelwordswhosespokenformstheyhadnotbeentrainedtoanalyze.
EXPERIMENT1PrereadersparticipatedindividuallyinExperi-ment1on4differentdays.
Eachday,theytookpartinananalysisconditionandthenacontrolcondition.
Inthefirstphaseofeachcondition,thechildrenworkedonlywithspokensyllables.
Intheanalysiscondition,theylearnedtosegmentfoursyllablesintotheirinitialconsonantsandremainingportions.
Forexample,thefoursyllablesmightbe"hem,""lig,"384TREIMANANDBARON"hig,"and"lem.
"Thechildrenlearnedtorespond"h,""em"to"hem,""I,""ig"to"lig,"andsoon,continuinguntiltheyhadreachedacriterion.
Inthecontrolcondition,thechildrenwereexposedtofourdifferentsyllables(forexample"diz,""vok,""dok,""viz")butwerenottaughttosegmentthem.
Instead,theysimplyrepeatedthesyllablesaloudforthesamenumberoftrialsasintheimmediatelypre-cedinganalysiscondition.
Thesecondphaseofeachconditioninvolvedareadingtask.
Thenatureofthetaskwasthesameinboththeanalysisandcontrolconditions.
Thechildrenweretaughtbyapaired-associateproceduretoreadfouritemsthatcorrespondedtothespokensyllableswithwhichtheyhadworkedinthefirstphase.
Inoursampleanalysiscondition,theitemsreadwouldbeH,EM,HEM,andLIG.
HEMiscalledthere-lateditembecauseitisrelatedtothetwosmalleritemsintheset,HandEM.
Itspronunciationcanbededucedfromthoseofthetwosmallitems.
Inotherwords,learningtosay"h"toHand"em"toEMprovidesinformation,givenphonemicanalysis,thattheproperresponsetoHEMis"hem.
"LIGiscalledtheunrelateditembecauseitisnotrelatedtoanyotheritemintheset.
Itspronunciationcannotbededucedfromthoseofotheritems.
Inoursamplecontrolcondition,inwhichthechildrenrepeatedthespokensyllables"diz,""vok,""dok,"and"viz,"theitemsreadwouldbeD,IZ,DlZ(relateditem),andYOK(unrelateditem).
Differencesinperformanceontherelateditemandtheunrelateditemshouldrevealhowchildrenlearntoreadtheitems.
Ifchildrenfailtotakeadvantageoftherepeatedspelling-soundcorrespondences,theyshouldlearntherelateditemnomoreeasilythantheunrelateditem.
Thatis,thefactthatthepronuncia-tionoftherelateditemcanbededucedfromthepro-nunciationsofthetwosmallitemsshouldnotbeofhelp.
Indeed,ifchildrentrulyattempttomemorizetheassociationsbetweenwholeprintedwordsandtheirpronunciations(assuggestedbyGough&Hillinger,1980),theymayhavemoredifficultyontherelateditemthanontheunrelateditemduetothesimilarityoftherelateditemtothetwosmalleritems.
Studiesofpaired-associatelearning(e.
g.
,Horowitz,1962)havedocumentedthatpeopleper-formpoorlywhensimilar-lookingstimuliarepairedwithsimilar-soundingresponses.
Sucheffectshavebeenfoundamongchildrenlearningtoreadnewwords(Baron,1977,pp,211-213;McCutcheon&McDowell,1969;Otto&Pizillo,1970-1971;Samuels&Jeffrey,1966),whichisconsistentwiththeviewthatthesechildrentreatreadingasarote,paired-associatetask.
InBaron's(1977)experiment,forex-ample,5-year-oldsweretaughttoreadpairsofwordssuchasBOXandBUSbyamethodsimilartothatusedhere.
Theywerethentaughttworelatedwords,OXandUS.
Thechildren,apparentlyconfusedbythesimilarityofthenewwordstothepreviouslylearnedwords,actuallytookmoretrialstolearnthesecondpairofwordsthanthefirstpair.
Identicalresultsemergedwhenthetwo-letterwordswerepre-sentedfirst.
Inthepresentstudy,ifthechildrenre-liedonlyonword-specificassociations,theyshouldhavemademoreerrorsontherelateditemthanontheunrelateditem.
Thisisthepatternweexpectedtoobserveinthereadingtaskofthecontrolcondition-theconditioninwhichthechildrenhadnotbeentaughttoanalyzethespokensyllablesthatcorre-spondedtothewrittenitemsofthereadingtask.
Incontrast,ifthechildrenbenefitedfromtherela-tionsbetweenthespellingsandthesoundsofwords,asweexpectedthemtodointheanalysiscondition,theyshouldhavelearnedtherelateditemmoreeasilythantheunrelateditem.
Transferofspelling-soundcorrespondenceslearnedfromthetwosmallitems(e.
g.
,H,EM)shouldhavespeededthelearningoftherelateditem(HEM).
Transfershouldhavebeenofnohelpwiththeunrelateditem(LIG).
Eventhoughchildrenintheanalysisconditionhadbeentaughttosegmentthespokensyllable"lig"into"I"and"ig,"theycouldnotputthisknowledgetouseinthereadingtaskbecausetheyhadnotbeentaughtthewrittenlettersthatcorrespondedtotheconstituentsounds.
Hence,thechildrenwhotookadvantageofspelling-soundrelationsshouldhavelearnedtoreadtherelateditemmoreeasilythantheunrelateditem.
Weexpectedthatthechildrenwouldshowthispatterninthereadingtaskoftheanalysiscondition-theconditioninwhichtheyhadbeentaughttosegmentthespokensyllablesthatcorre-spondedtothewrittenitems.
Tosummarize,thehypothesisthatphonemic-analysistrainingpromotesspelling-sound-ruleusepredictsaninteractionbetweencondition(analysisvs.
control)andtypeofitem(relatedvs.
unrelated).
Inthecontrolcondition,thechildrenshouldhaveperformedmorepoorlyonrelateditemsthanonun-relateditems;intheanalysiscondition,theyshouldhaveperformedbetteronrelateditemsthanonun-relateditems.
MethodProcedure.
Inthefirstphaseoftheanalysiscondition,thechildwasintroducedtoapuppetthat"talkedfunny.
"Theexperi-mentersaideachsyllableherself(e.
g.
,"hem")andthenhadthepuppetsaytheinitialconsonantandtheremainingportion(e.
g.
,"h,""ern").
Whentheinitialconsonantcouldnotbepronouncedinisolation,aschwawasadded.
(Pilotworkhadsuggestedthatuseofthisneutralvowelcausedchildrennoaddeddifficulty.
)Thechildwasaskedtorepeatwhatthepuppethadsaidimme-diatelyafterthepuppethadsaidit.
Afterthechildhadrepeatedeachpuppetformtwice,theexperimenteraskedthechildtoworkthepuppethim-orherself.
Oneachtrial,theexperimentersaidallfoursyllables,inarandomordereachtime,andthechildat-temptedtosayeachsyllablein"puppettalk.
"Ifthechildmadeanerror(hereorinanyotherpartoftheprocedure),theex-perimentercorrectedtheerrorbysaying,"No,yousaid____,butitshouldbe.
"Thisprocedurecon-PHONEMIC-ANALYSISTRAINING385TableIMeanNumberofErrorsonRelatedandUnrelatedItemsinReadingTaskofExperiment1Item=9.
29,p<.
025].
Neithermaineffectwassignif-icant[forcondition,F(I,7)=.
08;foritemtype,F(I,7)=1.
84].
Examinationofthepatternofresultsineachofthetwoconditionsshowedthat,inthecontrolcondition,thereweresignificantlymoreerrorsonrelateditemsthanonunrelateditems(Tukey's,p<.
05).
Thispatternwasshownbysevenoftheeightsubjects.
Intheanalysiscondition,therewasnosta-tisticallysignificantdifferencebetweenerrorsonre-lateditemsanderrorsonunrelateditems.
Sixoftheeightsubjects,however,madefewererrorsonrelateditemsthanonunrelateditemswheninthiscondition.
Withoutphonemic-analysistraining,then,child-renwereactuallyhurtbythespelling-soundrelationsamongthewrittenitems.
Theymademoreerrorsontherelateditemthanontheunrelateditem.
Thisfindingisconsistentwithseveralpreviousreportsthatchildrenhavemoredifficultyinlearningtoreadsimilar-lookingwordsthantheydoinlearningtoreaddissimilarwords(Baron,1977;McCutcheon&McDowell,1969;Otto&Pizillo,1970-1971;Samuels&Jeffrey,1966).
Inthepresentstudy,therelateditemwassimilartothetwosmallitemsbyvirtueofthesharedspelling-soundrules.
Consequently,child-renmademoreerrorsontherelateditemthanontheunrelatediteminthecontrolcondition.
Thephonernic-analysistrainingofferedintheanalysisconditionapparentlyallowedchildrentoovercometheinter-ferencecausedbyrepeatedspelling-soundrules.
Per-formanceontherelateditemsafteranalysistrainingwasreliablybetterthanperformanceonthesametypeofitemswhennosuchtraininghadbeengiven(Tukey's,p<.
05).
Performanceontherelateditemswasnot,however,significantlybetterthanperfor-manceontheunrelateditemsintheanalysiscondi-tion.
Themajorresultofphonemic-analysistrainingwasthereductionofanegativeeffectratherthantheproductionofapositiveone.
Nonetheless,thefindingsareconsistentwiththeviewthatphonemic-analysistrainingenableschildrentobenefitfromspelling-soundrulestoagreaterdegreethanwhentheyhavehadnosuchtraining.
Anexaminationoftheconditionalprobabilityofpronouncingarelateditemcorrectly,givencorrectpronunciationofthetwosmalleritemsonthesametrial,providesadditionalevidenceinfavorofthehy-pothesisthatphonemic-analysistrainingpromotesspelling-sound-ruleuse.
Thisconditionalprobabilitytinueduntilthechildhadachieved2successivecorrecttrialsonallfoursyllablesoruntil10testtrialshadbeencompleted,which-evercamefirst.
Theprocedureforthefirstphaseofthecontrolconditionwasthesame,exceptthatthepuppet(adifferentpuppetfromtheoneusedintheanalysiscondition)repeatedeachsyllablewhole.
Whentheexperimenterhadsaidasyllable,thepuppet,andthechild,merelyrepeatedit.
Thecontrolconditionoccurredaftertheanaly-sisconditiononeachtestday.
Inthisway,thenumberoftesttrialsinthecontrolconditioncouldbematchedtothenumberintheanalysiscondition.
Thechildrenneverproducederrorsinrepeatingthesyllables.
Areadingtaskconstitutedthesecondphaseofeachcondition.
Eachtrialinvolvedallfourprinteditems,forexampleH,EM,HEM,andLIG.
Eachitemwasprintedonaseparatecardinlarge,uppercaseletters.
Theexperimenterfirstpresentedeachcard,gavetheresponse,andaskedthechildtorepeatit.
(Notethattheresponseforasingleletteritem,suchasH,wastheletter'ssoundratherthanitsname.
)Thisprocedurecontinuedfortwotrials.
Thenthechildwasaskedtosupplytheresponses.
Thefourcardswerepresentedinarandomorderoneachtrial,anderrorswerecorrected.
Theprocedurecontinueduntilthechildhadachieved2successivecorrecttrialsonallfouritemsoruntil10testtrialshadbeencompleted,whichevercamefirst.
Stimuli.
ThestimuJiwerebasedoneightpairsofsyllables:"hem,""lig";"diz,""vok";"med,""tap";"pil,tt"wur";"bew,""zus":"fat,""sun";"nip,""rox";and"jaf,""cob".
Eachchildusedalleightpairs,adifferentoneineachoftheeightconditions(i.
e.
,ananalysisconditionandacontrolconditiononeachof4days).
Forthefirstphaseofeachcondition,foursyllableswereused-abasepairplustwoadditionalsyllablesconstructedbyswitchingtheinitialconsonantsofabasepair.
Withthebasepair"hem"and"Jig,"forinstance,thefoursyllableswouldbe"hem,""Jig,""hig,"and"Iem.
"Theanalysisconditiondividedthesyllablesaftertheinitialconsonant(i.
e.
,"hem"'""h,""em")ratherthanafterthevowel(i.
e.
,"hem".
.
.
"he,""rn"),sincetheformerdivisionappearedtobethemorenatural(MacKay,1972;Treiman,inpress).
Forthereadingtask,thefirstmemberofthebasepairservedastherelateditem,andthesecondservedastheunrelateditem.
Therelateditemwasbrokenintoitsinitialconsonantandremaindertocreatetwoadditionalitems(e.
g.
,HandEMfromHEM).
Eachbasepairservedintheanalysisconditionforapprox-imatelyhalfthesubjectsandinthecontrolconditionfortheotherhalf.
Also,theinitialconsonantswithineachbasepairwereswitchedforhalfthesubjects.
Forthesesubjects,theitemsinthereadingtaskwouldbeL,EM,LEM,andHIG,forexample.
Eachbasepairwasusedapproximatelyequallyoftenoneachofthe4testdays.
Thus,differencesbetweentheanalysisandcontrolcondi-tionsortrendsacrosstestdaysshouldnothavereflectedthepar-ticularstimulithatwereused.
Subjects.
Eightpreschoolnonreadersservedasthesubjects.
Therewerefivegirlsandthreeboys;theirmeanagewas4yearsIImonths(range=3years5monthsto6years0months).
Thechildrenattendedlocaldaycarecenters.
Weselectedchildrenwhocouldnotreadbutwhowerewillingandabletoparticipateinourreadingtask.
Teacherschosechildrenwhomtheythoughtreadyfortheexperiment,andthesechildrenwereaskedtoreadtheitemsusedinourreadingtests.
Anychildwhocouldreadanyoftheitemswasdisqualified,aswasanychildwhoperformederror-lesslyonthephonemic-analysistask.
Thiswasdonetoensurethatthesubjectsdidnotalreadyknowtheskillsweweretryingtoteach.
ResultsandDiscussionTable1showsthemeannumberoferrorsonre-latedandunrelateditemsinthereadingtaskofthetwoconditions.
Aspredicted,therewasasignificantinteractionbetweenconditionanditemtype[F(l,7)AnalysisConditionControlConditionRelated4.
888.
38Unrelated6.
133.
38386TREIMANANDBARONwas.
80intheanalysisconditionandonly.
53inthecontrolcondition,asignificantdifference[t(7)=1.
97,p<.
05].
Thatthelinkbetweentherelateditemandthesmallitemswascloserintheanalysiscondi-tionthaninthecontrolconditionsuggeststhatchild-renweremorelikelytodeducetherelateditemonthebasisofitsconstituentsintheanalysiscondition.
Theobserveddifferencesbetweentheanalysisandcontrolconditionssuggestthatthephonemic-analysistraininggivenhereaffectednotthelearningofallspelling-soundrulesbutthelearningofjustthoserulesinvolvingthephonemesthathadbeenusedintraining.
Hadthetrainingeffectsbeenmoregeneral,andhadtheypersistedfromtheanalysisconditionintothesubsequentcontrolconditiononeachtestday,thesubjectswouldhavebeenhelpedinlearningrulesinthecontrolconditionaswell.
Ifsuchhadbeenthecase,wewouldnothaveobserveddiffer-encesinperformancebetweenthetwoconditions.
Althoughwecannotruleoutthepossibilitythatsomegeneralizationfromtheanalysisconditiontothecon-trolconditiondidoccur,itseemsnottohavebeensubstantial.
Inparticular,theinteractionbetweenconditionanditemtypedidnotdiminishoverthe4daysoftheexperiment,aswouldhavebeenex-pectedifchildrenhadincreasinglyappliedphonemic-analysisskillsinthecontrolconditionaswellasintheanalysiscondition.
Whenthedataforthethirdandfourthsessionsoftheexperimentwereanalyzedseparately,theinteractionbetweenconditionanditemtypewasstillsignificant[F(l,7)=5.
85,P<.
05].
EXPERIMENT2Experiment2replicatedthemajorfeaturesofEx-periment1.
However,itdifferedfromExperiment1intwoprimaryways.
First,theinitialphaseoftheanalysisconditionprovidedmoreextensivephonemic-analysistrainingthanhadbeenprovidedinExperi-ment1,includingbothsegmentationandblendingtraining.
Wepredictedthatthismoreextensivetrain-ingwouldallowchildrentoperformsignificantlybetteronrelateditemsthanonunrelateditemsinthereadingtaskoftheanalysiscondition.
Second,thereadingtaskusedinbothconditionswasmodified.
Thechildrenwerefirsttaughtthetwosmallitems(e.
g.
,H,EM)tocriterion.
Onlythenweretherelateditem(HEM)andtheunrelateditem(LIG)intro-duced.
Thisprocedureensuredthatthechildrenwerethoroughlyfamiliarwiththesmalleritemsbeforetheylearnedthelargeritems.
AsinExperiment1,wepredictedaninteractionbetweentrainingcondi-tionanditemtypesuchthatchildrenwouldperformbetterontherelateditemthanontheunrelateditemintheanalysisconditionbutnotinthecontrolcon-dition.
MethodProcedure.
Eachchildparticipatedinananalysisconditionandinacontrolconditiononeachof4testdays.
Inthefirstphaseoftheanalysiscondition,thechildlearnedtosegmentandtoblendfoursyllables(e.
g.
,"hem,""lig,""hig,""Iem").
Thesegmen-tationtrainingwasgivenfirst;itfollowedthesameprocedurethathadbeenusedinExperimentI.
Traininginblendingwasgivennext.
Here,thepuppetfirstsaideachsyllableinsegmentedform(e.
g.
,"h,""em"),andthechildattemptedtosaythewholesyllable("hem").
Twodemonstrationtrialsweregiven.
Testtrialscontinueduntilthechildhadachievedacriterionof2successivecorrecttrialsonallfouritemsoruntil10trialshadbeencom-pleted.
Inthefirstphaseofthecontrolcondition,thechild(andthepuppet)merelyrepeatedspokenitems.
TheprocedurewassimilartothatofExperimentI,buteight,ratherthanfour,items(e.
g.
,"diz,""vok,""dok,""viz,""d,""v,""iz,""ok")werere-peated.
Thisprocedureensuredthatthechildwasexposedtothesmallunitsaswellastothelongersyllables.
Thesecondphaseofeachconditionwasareadingtask.
Itbe-ganwithtwodemonstrationtrialsonthetwosmallitems(e.
g.
,H,EM).
Testtrialsontheseitemscontinueduntilthechildhadreachedacriterionof3successivecorrecttrialsonbothitemsorhadreachedamaximumof10testtrials.
Thenthetwothree-letteritems-therelateditem(e.
g.
,HEM)andtheunrelateditem(e.
g.
,LIG)-wereintroduced.
Onthefirsttrial,theexperimenteraskedthechildtotrytopronouncetheitems.
Shetoldthechildthat,intryingtoreadthenewitems,heorsheshouldthinkoftheitemsthathadbeenlearnedalready.
Onlyafterthechildhadhadachancetoresponddidtheexperimenterprovidethepronunci-ations.
Testtrialscontinueduntilthechildhadachieved3suc-cessivecorrecttrialsonbothitemsoruntil10trialshadbeencom-pleted.
UnlikeinExperimentI,inwhichtheanalysisconditionoc-curredfirstonalltestdays,theanalysisconditionofExperiment2occurredfirston2randomlychosentestdaysforeachchildandthecontrolconditionoccurredfirstontheother2testdays.
Afixednumberofrepetitiontrials-twodemonstrationtrialsandtwotesttrials-wasgiveninthecontrolcondition.
ThischangewasmadetoruleoutapossiblealternativeinterpretationoftheinteractionobtainedinExperimentI.
ThisinterpretationheldthatthechildreninExperimentIweretiredbythetimetheyhadreachedthereadingtaskinthecontrolconditionand,becausetransferofspelling-soundrulesrequiresmentaleffort,didnotusetheserulestotheextentthattheyotherwisewouldhave.
Stimuli.
ThestimuliandcounterbalancinginExperiment2werethesameasinExperimentI.
Subjects.
Thesubjectswere20childrenattendinglocalkinder-gartens.
Therewere9girlsandIIboys;theirmeanagewas5years9months(range=5years3monthsto6years3months).
Poten-tialsubjects,randomlychosenfromchildreninparticipatingclass-rooms,weregivenbothareadingpretestandaphonemic-analysispretest.
Inthereadingpretest,thesubjectswereaskedtoreadthethree-letteritemsusedintheexperiment.
Onlyonepotentialsub-jectcouldreadanyoftheitems,andthissubjectwasdisqualified.
Inthephonemic-analysispretest,thesubjectswereshownhowtoanalyzetwospokensyllablesintoinitialconsonantsandremainders.
Theywerethenaskedtoanalyzefournewsyllablesinthesameway.
Thesubjectswhoperformedcorrectlyonanyofthefourtestsyllablesdidnotparticipateintheexperiment,sincetheyalreadypossessedsomeofthephonemic-analysisskillsweweretryingtoteach.
Sixpotentialsubjectsweredisqualifiedforthisreason.
Anadditionaltwosubjectsweredroppedfromtheexperimentbecausetheydidnotwishtocontinueanddidnotpayattention.
ResultsThetopportionofTable2showsthereading-taskresultsaveragedoverall20subjects.
AnanalysisofTable2MeanNumberofErrorsonRelatedandUnrelatedItemsinReadingTaskofExperiment2AllSubjectsAnalysisCondition9.
2010.
60ControlCondition9.
608.
90SubjectsLessSkilledinPhonemicAnalysisAnalysisCondition10.
4012.
70ControlCondition11.
8010.
10SubjectsMoreSkilledinPhonemicAnalysisAnalysisCondition8.
008.
50ControlCondition7.
407.
70varianceonthesedatashowedthattheinteractionbetweenconditionanditemtypewassignificant[F(l,19)=6.
76,p<.
02],aspredicted.
Neitherthemaineffectofconditionnorthatofitemtypewassignificant[forcondition,F(I,19)=.
39;foritemtype,F(l,19)=.
50].
IncontrastwithExperiment1,nosignificantdifferencewasfoundbetweenrelated-itemerrorsandunrelated-itemerrorsinthecontrolcondition.
Intheanalysiscondition,however,thechildrenmadereliablyfewererrorsonrelateditemsthanonunrelateditems(Tukey's,p<.
05).
Thus,thesubjectsdidbetterontherelateditemsthanontheunrelateditemsintheanalysisconditionbutnotinthecontrolcondition.
Onlyintheanalysiscondi-tionwerethesubjectsabletoderivesignificantposi-tivebenefitfromtherelationbetweenthetwosmallitemsandtherelateditem.
Additionalposthoctestsshowedthatthechildrenmadesignificantlymoreerrorsonunrelateditemsintheanalysisconditionthaninthecontrolcondition(Tukey's,p<.
01).
(Errorsonrelateditemsdidnotdiffersignificantlybetweenthetwoconditions.
)Al-thoughnotspecificallypredicted,thisresultiscon-sistentwiththeviewthatphonemic-analysistrainingpromotesuseofspelling-soundrulesinreadinganddiscouragesuseofword-specificassociations.
Con-sequently,suchtrainingmayhinderperformanceontheunrelateditem,forwhichonlyspecificasso-ciationscanbeused.
Althoughtheaboveresultssupportourhypothe-sis,theyarenotidenticaltothoseofExperiment1,inwhichthetendencytouseword-specificassoci-ationsseemedstronger.
Thisdifferencemaybeduetodifferencesinthesubjectpopulations.
Toassessthispossibility,thedataforExperiment2wereanalyzedseparatelyforthosesubjectswhowerelessskilledinphonemicanalysis(definedasthosewhoseerrorsinsegmentationandblendinginthefirst2daysoftheexperimentexceededthemediannumber)andthosewhoweremoreskilled.
TheseresultsareshowninTable2.
ThesubjectswhowerelessskilledRelatedItemUnrelatedPHONEMIC-ANALYSISTRAINING387inphonemicanalysisshowedahighlysignificantin-·teractionbetweenconditionanditemtype[F(1,9)=13.
09,p<.
006].
Inthecontrolcondition,thesechildrenmademoreerrorsonrelateditemsthanonunrelateditems(Tukey's,p=.
06);intheanalysiscondition,theymadefewererrorsonrelateditemsthanonunrelateditems(Tukey's,p<.
05).
Thatis,thesesubjectsshowedbothpredictedeffects:inter-ferenceduetorepeatedspelling-soundrulesintheabsenceofphonemic-analysistraining,andfacilita-tionafterphonemic-analysistraining.
Thechildrenwhoweremoreskilledinphonemicanalysisattheoutset(i.
e.
,thosewhoseerrorsinsegmentationandblendinginthefirst2daysoftheexperimentwerebelowthemedian)didnotshowasignificantinter-actionbetweenconditionanditemtype[F(1,9)=.
05).
Althoughthemaineffectofitemtypealsowasnonsignificantforthesechildren,theydidtendtomakefewererrorsonrelateditemsthanonunrelateditemsinbothconditions.
(Theresultsaresimilarwhengroupsaredefinedonthebasisofsegmenta-tionandblendingerrorsinall4daysoftheexperi-ment.
)IncontrastwithExperiment1,theinteractionbe-tweenconditionanditemtypedeclinedoverthe4daysoftheexperiment.
Itwasseenwhenthedataforthefirst2daysoftheexperimentwereanalyzed[F(l,19)=4.
27,p<.
055],butnotwhenthedataforthelast2dayswereanalyzed[F(l,19)=.
29].
Theseresultssuggestthat,astheExperiment2subjectsbe-camemoreskilledinphonemicanalysis,theyincreas-inglyusedtheseskillsinthecontrol-conditionread-ingtaskaswellasintheanalysis-conditionreadingtask.
GENERALDISCUSSIONTheresultsofExperiments1and2concurindem-onstratingasignificantinteractionbetweenconditionanditemtypeinthereadingtask.
Inthecontrolcon-dition,inwhichchildrenwereexposedtocertainspokensyllablesbutwerenottaughttoanalyzethem,theyfailedtobenefitfromspelling-soundruleswhentheylearnedtoreadthecorrespondingwrittenitems.
Thatthepronunciationoftherelateditemcouldbededucedfromthepronunciationsofthetwosmallitemsdidnothelpthechildreninthiscondition.
In-deed,theExperiment1subjectsmadesignificantlymoreerrorsontherelateditemthanontheitemwhosepronunciationcouldnotbesodeduced.
Incontrast,intheanalysiscondition,thesubjectstendedtomakefewererrorsontherelateditemthanontherelateditem.
ThistrendwasnotsignificantinExperiment1,butwassignificantinExperiment2.
Thus,explicitinstructioninanalysisofthespokensyllablespermittedthesubjectsinbothexperimentstotakeadvantageofthespelling-soundrelationstoa388TREIMANANDBARONgreaterdegreethantheywereabletodowithoutsuchinstruction.
Thisresultsupportsthehypothesisofadirectlinkfromphonemicanalysistoabilitytobenefitfromspelling-soundcorrespondences.
Itisimportanttonotethatphonemic-analysistrainingdidnotreducethetotalnumberoferrorsinthereadingtaskineitherexperiment.
Nomainef-fectsofconditionwerefound.
However,trainingdidalterthepatternofperformanceonrelatedandunrelateditems.
Wehavesuggestedthatbetterper-formanceonrelateditemsthanonunrelateditemsshowsthatchildrenderivebenefitfromspelling-soundrelationships.
Thus,theinteractionbetweenconditionanditemtypeindicatesthatchildrengainmorebenefitfromspelling-soundruleswithphonemic-analysistrainingthanwithoutit.
Althoughthesubjectsinbothexperimentsper-formedbetteronrelateditemsrelativetounrelateditemsintheanalysisconditionthaninthecontrolcondition,severaldifferencesintheirpatternsofre-sultsmeritdiscussion.
OnedifferenceisthattheEx-periment1subjectswereactuallyhurtbyspelling-soundrelationsintheabsenceofphonemic-analysisinstruction,whereastheExperiment2subjects,asagroup,werenot.
Severalfactorsmayhavecontrib-utedtothisdifference,butwesuggestthattheEx-periment1subjects-younger,lesseducated,andpresumablypoorerinphonemicanalysisthantheEx-periment2subjects-weremorepronetotreatthereadingtaskinarote,paired-associatefashion.
(Forasimilarview,seeGough&Hillinger,1980).
Con-sequently,asdiscussedearlier,theywerehurtbythesimilarityoftherelateditemtothetwosmallitems.
ConsistentwiththisanalysisisthefindingthatthosesubjectsinExperiment2whowerebelowthemedianinphonemic-analysisskillmade-asdidtheExperi-ment1subjects-moreerrorsonrelateditemsthanonunrelateditemsinthecontrolcondition.
Aseconddifferencebetweentheexperimentsliesinthepatternofresultsintheanalysiscondition.
TheExperiment1subjectsdidnotperformsignificantlybetteronre-lateditemsthanonunrelateditemsafterphonemic-analysistraining;theExperiment2subjectsdidmakefewererrorsonrelateditemsthanonunrelateditemsintheanalysiscondition.
Thedifferenceinresultsmaybeduetothemoreintensivephonemic-analysisinstruction,includingbothsegmentationandblend-ingtraining,providedinExperiment2.
Indeed,ithasbeensuggestedthatblendingplaysamoreimportantrolethandoessegmentationinpromotingreadingacquisition(Perfetti,Beck,&Hughes,Note3).
Afinaldifferencebetweentheresultsofthetwoexperi-mentsisthatinExperiment1theinteractiondidnotappeartodeclineoverthe4testdays,whereasinExperiment2itdiddecline.
Thisdifferencemaybeduetodifferencesinthesubjectpopulations:TheolderandmoreexperiencedExperiment2subjectsmayhavebeenmoreabletotransferskillslearnedintheanalysisconditiontothecontrolcondition.
Taskdifferencesmayalsohaveplayedarole.
Thephonemic-analysistrainingprovidedinExperi-ment2,sinceitwasmoreintensive,mayhavebeenmorelikelytogeneralize.
Furtherresearchisneededtoevaluatethesesuggestionsabouttheroleofsub-jectandtaskvariables.
Thepresentresultssupportthetheoreticalview,discussedintheintroduction,thatphonemicaware-nessfacilitatesthelearningofspelling-soundrules.
Theysuggestthatthecorrelationspreviouslyob-servedbetweenphonemic-analysisskillandspelling-sound-ruleuse(Baron&Treiman,1980;Treiman&Baron,1981)reflect,atleastinpart,acausallinkfromphonemicanalysistospelling-sound-rulelearn-ing.
Likewise,ourresultssuggestthatpreviousfind-ingsofimprovedreadingskillfollowingphonemic-analysisinstruction(Rosner,Note1;McNeil&Coleman,Note2)arosebecausethisinstructionhelpedchildrentolearnspelling-soundrelations.
Thelinkbetweenphonemicanalysisandspelling-sound-ruleusepresumablyexistsbecausechildrenwhocanexplicitlyanalyzespokenwordsintotheirphonemiccomponentsareabletolearnthecorrespondencesbetweenphonemeunitsandletterunits.
Sinceabil-itytousethesecorrespondencesappearstobeanim-portantpartofreadingsuccess(e.
g.
,Firth,1972),ourresultssuggestthatphonemic-analysistrainingcanpromotereadingskill.
Thepresentresultsalsohaveimplicationsforedu-cationalpractice.
Ifthegoalofreadinginstructionweresimplytoteachchildrenasmallsetofwords,roteword-specificassociationsmightbesatisfactory.
Inthiscase,ourresultswouldleadustorecommendthatdissimilarwordsbetaughttogether.
Paired-associatelearning,itappears,ismostsuccessfulun-derthesecircumstances.
However,ifthegoalofreadinginstructionis(aswebelieve)toallowchil-drentoreadanywordintheirspokenvocabulary,rotelearningwillnotsuffice.
Childrenmustbeabletodeciphernewprintedwordsonthebasisofpre-viouslylearnedwords.
Theymustbeabletotrans-fertheirknowledgeofspelling-soundrulestorelatedwords.
Certainmethodsofreadinginstruction,suchasthe"linguistic"methodsofBloomfield(1942)andothers,assumethat,ifrelatedwords(e.
g.
,CAT,BAT,andHAT)aretaughttogether,childrenwillinducespelling-soundrules-inthiscasetherulethatATcorrespondsto"at.
"Ourresultssuggestthatmanychildrenwill,instead,behurtbythesimilarityamongitemsthatnecessarilyariseswhenspelling-soundrulesarerepeated.
Theywillmakemoreerrorsonwordsthatembodytherulethanonwordsthatdonot.
Ourresultsalsoshowthatphonemic-analysistrainingwiththecorrespondingspokenwordscanovercomesuchnegativeeffectsandcan,insomecases,produceapositivebenefit.
Indeed,severalsuc-cessfulreadingprograms(e.
g.
,Wallach&Wallach,1979;Williams,1980)includephonemic-analysistrainingasanimportantcomponent.
Researchers(e.
g.
,Lewkowicz,1980)havebeguntostudythewaysinwhichsuchtrainingcanmosteffectivelybedoneinaneducationalsetting.
REFERENCENOTESI.
Rosner,J.
Phonicanalysistrainingandbeginningreadingskills(Publication1971/19).
Pittsburgh:LearningResearchandDevelopmentalCenter,1971.
2.
McNeil,J.
D.
,&Coleman,J.
C.
Auditorydiscriminationtraininginthedevelopmentofwordanalysisskills(FinalReport,U.
S.
OfficeofEducationProjectNo.
5-0503).
LosAngeles:UniversityofCalifornia,1967.
3.
Perfetti,C.
A.
,Beck,I.
L.
,&Hughes,C.
Phonemicknowl-edgeandlearningtoread.
PaperpresentedatthemeetingoftheSocietyforResearchinChildDevelopment,Boston,April1981.
REFERENCESBARON,J.
Mechanismsforpronouncingprintedwords:Useandacquisition.
InD.
LaBerge&S.
J.
Samuels(Eds.
),Basicpro-cessesinreading:Perceptionandcomprehension.
Hillsdale,N.
J:Erlbaum,1977.
BARON,J.
Orthographicandword-specificmechanismsinchil-dren'sreadingofwords.
ChildDevelopment,1979,50,60-72.
BARON,J.
,&HODGE,J.
Usingspelling-soundcorrespondenceswithouttryingtolearnthem.
VisibleLanguage,1978,12,55-70.
BARON,J.
,&TREIMAN,R.
Useoforthographyinreadingandlearningtoread.
InJ.
F.
Kavanagh&R.
L.
Venezky(Eds.
),Orthography,reading,anddyslexia.
Baltimore:UniversityParkPress,1980.
BWOMFIELD,L.
Linguisticsandreading.
ElementaryEnglishRe-view,1942,19(1),125-130;183-186.
BODER,E.
Developmentaldyslexia:Prevailingdiagnosticcon-ceptsandanewdiagnosticapproach.
InH.
R.
Myklebust(Ed.
),Progressinlearningdisabilities(Vol.
2).
NewYork:Grune&Stratton,1971.
BODER,E.
Developmentaldyslexia:Adiagnosticapproachbasedonthreeatypicalreading-spellingpatterns.
DevelopmentalMed-icineandChildNeurology,1973,15,663-687.
BROOKS,L.
R.
Non-analyticcorrespondencesandpatterninwordpronunciation.
InJ.
Requin(Ed.
),AttentionandperformanceVII.
Hillsdale,N.
J:Erlbaum,1978.
CALFEE,R.
C.
,CHAPMAN,R.
S.
,&VENEZKY,R.
Howachildneedstothinktolearntoread.
InL.
W.
Gregg(Ed.
),Cognitioninlearningandmemory.
NewYork:Wiley,1972.
CALFEE,R.
C.
,LINDAMOOD,P.
,&LINDAMOOD,C.
Acoustic-phoneticskillsandreading-Kindergartenthroughtwelfthgrade.
JournalofEducationalPsychology,1973,64,293-298.
EHRI,L.
C.
,&WILCE,L.
S.
Themnemonicvalueoforthographyamongbeginningreaders.
JournalofEducationalPsychology,1979,71,26-40.
ELKONIN,D.
B.
USSR.
InJ.
Downing(Ed.
),Comparativeread-ing.
NewYork:Macmillan,1973.
FIRTH,I.
Componentsofreadingdisability.
Unpublisheddoctoraldissertation,UniversityofNewSouthWales,1972.
Fox,B.
,&ROUTH,D.
K.
Analyzingspokenlanguageintowords,syllablesandphonemes:Adevelopmentalstudy.
JournalofPsycholinguisticResearch,1975,4,331-342.
GLEITMAN,L.
R.
,&ROZIN,P.
ThestructureandacquisitionofreadingI:Relationsbetweenorthographiesandthestructureoflanguage.
InA.
S.
Reber&D.
L.
Scarborough(Eds.
),Towardapsychologyofreading:TheproceedingsoftheCUNYconfer-ences.
Hillsdale,N.
J:Erlbaum,1977.
GOLINKOFF,R.
M.
Phonemicawarenessskillsandreadingachievement.
InF.
Murphy&J.
Pikulski(Eds.
),Theacquisi-tionofreading.
Baltimore:UniversityParkPress,1978.
GOUGH,P.
B.
,&HILLINGER,M.
L.
Learningtoread:Anun-naturalact.
BulletinoftheOrtonSociety,1980,30,179-196.
HARDY,M.
,STENNETT,R.
G.
,&SMYTHE,P.
C.
Auditoryseg-mentationandauditoryblendinginrelationtobeginningread-ing.
AlbertaJournalofEducationalResearch,1973,19,144-158.
PHONEMIC-ANALYSISTRAINING389HELFGOTT,J.
A.
Phonemicsegmentationandblendingskillsofkindergartenchildren:Implicationsforbeginningreadingac-quisition.
ContemporaryEducationalPsychology,1976,I,157-169.
HOROWITZ,L.
M.
Associativematchingandintralistsimilarity.
PsychologicalReports,1962,10,751-757.
JUSCZYK,P.
W.
Rhymesandreasons:Someaspectsofthechild'sappreciationofpoeticform.
DevelopmentalPsychology,1977,13,599-607.
LEWKOWICZ,N.
K.
Phonemicawarenesstraining:Whattoteachandhowtoteachit.
JournalofEducationalPsychology,1980,71,686-700.
LIBERMAN,I.
,LIBERMAN,A.
M.
,MATTINGLY,I.
,&SHANK-WEILER,D.
Orthographyandthebeginningreader.
InJ.
F.
Kavanagh&R.
L.
Venezky(Eds.
),Orthography,reading,anddyslexia.
Baltimore:UniversityParkPress,1980.
LIBERMAN,I.
Y.
,SHANKWEILER,D.
,FISCHER,F.
W.
,&CARTER,B.
Explicitsyllableandphonemesegmentationintheyoungchild.
JournalofExperimentalChildPsychology,1974,18,201-212.
McKAY,D.
G.
Thestructureofwordsandsyllables:Evidencefromerrorsinspeech.
CognitivePsychology,1972,3,210-227.
McCUTCHEON,B.
A.
,&McDoWELL,E.
E.
Intralistsimilarityandacquisitionandgeneralizationofwordrecognition.
TheRead-ingTeacher,1969,13(2),103-107;115.
MORAIS,J.
,CARY,L.
,ALEGRIA,J.
,&BERTELSON,P.
DoesawarenessofspeechasasequenceofphonesarisespontaneouslyCognition,1979,7,323-331.
OTTO,W.
,&PIZILW,C.
Effectofintralistsimilarityonkinder-gartenpupils'rateofwordacquisitionandtransfer.
JournalofReadingBehavior,1970-1971,3,14-19.
RoSNER,J.
,&SIMON,D.
TheAuditoryAnalysisTest:Aninitialreport.
JournalofLearningDisabilities,1971,4,384-392.
RoZIN,P.
,&GLEITMAN,L.
R.
ThestructureandacquisitionofreadingII:Thereadingprocessandtheacquisitionofthealpha-beticprinciple.
InA.
S.
Reber&D.
L.
Scarborough(Eds.
),Towardapsychologyofreading:TheproceedingsoftheCUNYconferences.
Hillsdale,N.
J:Erlbaum,1977.
SAMUELS,S.
J.
,&JEFFREY,W.
E.
Discriminabilityofwordsandlettercuesusedinlearningtoread.
JournalofEducationalPsy-chology,1966,57,337-340.
SNOWLING,M.
J.
Thedevelopmentofgrapheme-phonemecorre-spondenceinnormalanddyslexicreaders.
JournalofExperi-mentalChildPsychology,1980,19,294-305.
TREIMAN,R.
Thestructureofspokensyllables:Evidencefromnovelwordgames.
Cognition,inpress.
TREIMAN,R.
,&BARON,J.
Segmentalanalysisability:Develop-mentandrelationtoreadingability.
InG.
E.
MacKinnon&T.
G.
Waller(Eds.
),Readingresearch:Advancesintheoryandpractice(Vol.
3).
NewYork:AcademicPress,1981.
WALLACH,M.
A.
,&WALLACH,L.
Helpingdisadvantagedchil-drenlearntoreadbyteachingthemphonemeidentificationskills.
InL.
B.
Resnick&P.
A.
Weaver(Eds.
),Theoryandpracticeofearlyreading(V01.
3).
Hillsdale,N.
J:Erlbaum,1979.
WILLIAMS,J.
P.
Teachingdecodingwithanemphasisonpho-nemeanalysisandphonemeblending.
JournalofEducationalPsychology,1980,71,I-IS.
NOTES1.
Althoughtherearedifferentwaysofanalyzingspeechintophonemes,differentanalysesalmostalwaysagreeinrelevantre-spectsforthestimuliweuse.
Thus,weputasidequestionsaboutwhichanalysiscorrespondsmostcloselytospellingortochildren'srepresentationsofspeech.
2.
Severalmechanismsfortransferarepossible.
Forexample,transfermayoccurbydeducingthepronunciationofoneitemfromthepronunciationsofothersorbyanalogy(Baron,1979;Baron&Hodge,1978;Brooks,1978).
(ManuscriptreceivedNovember17,1982;revisionacceptedforpublicationApril4,1983.
)