Ef托ctsofPronunciationPracticeoftheEnglish／ae／Sound
UtilizingtheJapaneseContractedSoundlmjae］：TypeFrequencyinrIbrmsofDistributionalVarietyandtheEffbcts
OfPracticewithIncreasedDistributionalTypeFrequency
TbshihikoYAMAOKA
／小・ヾ′・UJJh・t、′中川！月・用心げglか′HJJf川J AbstractItisgenerallyobservedthatwhileJapanese−SpeakinglearnersofEnglishfeeltheycan
pronounce cat and cqp easily，they cannot easily pronounce mat and mqp．This article arguesthatthereasonforthisgeneralobservationmaylieinthedifferencesofapplicability
OftheJapanesecontractedsoundsbJa（lkj車】）andmya（lmjael）inEnglishpronunciation：
WhilelkjEelisreadilyappliedinpronouncingcatandcqp，【mjaelisnoteasilyappliedinmat andmqp・ThisarticlealsoarguesthatthedifferenceofapplicabilitybetweentheseJapanese COntraCtedsoundsiscausedbythedifferencesindistributionalvarietyoftheserespeCtive SOundsinJapanese：Whilelkjee］hasarichvariety，［mj∈可hasonlyalimitedvariety．Based Onthisreasoning，itcanbehypothesizedthatapplicabilityofthelmjaeIIsoundinEnglish pronunciationcanberaisedbypracticingthesoundwithitsincreaseddistributionalvariety． Resultsofthispracticethroughaquasi−eXperimentwithJapaneseuniversity studentsare reportedinthisarticleshowlngthefavorablebutlimitedeffectsofthepractice． 1．Introduction IthasbeenpointedoutthatJapanese−SpeakinglearnersofEnglishcanpronouncesuch WOrdscatandcqprelativelyeasilybuttheycannotpronouncewordssuchasmatandmqp easily（Matsuzawa，2004，P．68）．Actually，thefirsttestofpronunciationofthese words administeredinaquasi−eXperimentwithtwenty−SixJapaneseuniversltyStudents，Whichwill beelaboratedinduepartofthisarticle，Showsthatwhilecatandcaparepronouncedwith the／ae／sound more than88％of the chances，mat and map are pronounced、Vith the Japanese／a／soundinsteadofthe／ae／soundlOO％oftheoppc：tunities．ItmaybereasonabletosupposethatJapanese−SpeakinglearnersofEnglishapplythe Japanesecontractedsoundkya（／kja／）inthepronunciationofcatandcapandthustheyfeel
theycaneasilypronouncethem・Asamatteroffact，JapaneseloanwordsfromEnglish Which begin with4ya are pronounced with the／kja／sound（e・g・kyatto，Ayqppu），and， therefore，thesoundisappliedinpronounclngEnglishwordssuchascatandcap・Itmustbe stressedherethatthe／a／sourldin／kja／tendstobepulledabitforwardafterthe／j／soundand
assumesthequalityofthe／a3／sound（Makino，2005，P・29）・Thus，Phoneticdescriptionof
／kja／islkjae］．ThismeansthatJapaneseTSPeakinglearnersofEnglishusuallypronouncecat andcqplkj恕tland【kjaep］ratherthan／ka∋t／and／ka∋P／・Althoughitmaybeapointofargumenttodecidetowhatextentthelkj車lsoundcanbe
an acceptabJe aIternative to the／k㌍／sound and，Of course，ltis a matter to decide empirically，itisatleastpossibletoclaimthatthelkjaelsoundisafarbetteralternativethan ／ka／：／kat／isnomoreacceptablepronunciationofcatthan／mat／isacceptablepronunciation Ofmat．ItshouldalsobenotedthatveryaccuratepronunciationofEnglishisnotneededfor everylearnerofEngJish．ThisisinharmonywiththeclaimbyJenkins（2000，p．17）that
PrOmOtlngmutualintelligibilityisfarmoreimportantthan“pushinglearnersofEnglishto attempt to approximate an Ll accent too closely，”recognlzlng the growlng nature Of
Englishasaninternationallanguage．ThisisespeciallythecasewithJapaneseelementary
SChoolteacherswhoareexpectedtoengagein“Englishactivities”withtheirstudentssince
theyaretoobusytolearnEnglishprorlunCiationsystematically．
Kya（lkjae］）is notthe only contracted soundinJapanese：italsohas mya（lmjae］）
amongothers・（1）Then aquestion mayarisewhyitis the case thatwhiJe theJapanese COntaCted［kjaelsoundisappliedinEnglishpronunciation，thecontractedsourld［mj車lis not・Itseemspossibletofindananswertothisquestionbyinvestlgatingthedistributional differencesofthesetwocontractedsoundsinJapanesephonologlCalenvironments．Thisis tobediscussedinthesectiontofollow． 2・PhonologicalDistributionalDi恥rencesoflkjae］andlmjeEB］inJapanese Inordertoexaminetowhatextenteachofthecontractedsoundslkj∈eland［mjaBlis usedinvariousphonologlCalcontextsinJapanese，thenumberofdifferentsyllablesounds immediatelyfollowlngeaChcontractedsoundiscountedbylookinguptheentrywordsin NihonKokugoDai−Jiten（JqpaneseNationalLanguageLa7YeDictionary）（2ndversion）．
The result ofthisinvestlgation has revealed a striking differencein phonologlCal distributionalrichnessbetweenthesetwocontractedsounds：Whilelkja∋1takesasmanyas forty−threedifferentsyJJabJesoundsimmedjateJyafterjt，lmjaeltakesonJysjx（Ybmaoka，in
PreSS）・Thesesixwordshavingtheinitialcontractedlmja∋］soundaremyaa，myaO−ZOku，
myaku・myaSukqhLfukii，myanmaa・andmyakkan・Ofthesesixwords，myaaisthecrying SOund of a cat（meott）），myaO−ZOku begins with a proper noun（tribe name：Miao）， myasukqbsukiiisaJsoapropernoun（nameofacomposer：叫′aSkovsky），andmyanmaais
OnCeagainapropernoun（nameofacountry：坤anmar），allofwhicharenotexpectedtobe
usedveryfrequentlyindailyconversation・Oftheremainingtwowords，myakkan（blood
VeSSel）isakindofvariationofmyaku㊥ulse）anditmaybeveryinfrequentinuse．Thus，it
maybesafetosaythattheinitiallImjEelsoundisusedonadailybasisalmostonlyinwords
beginningwith myaku such as myaku−haku bulsation），myakuAou bulsation），etc．
includingmyakuitself．
WhatinfluencedoesthisdifferenceinphonologlCaldistributionalvarietyhaveinthe
learnlngOfEnglishpronunciation？ThisisthethemetobetackledinthefollowlngSeCtions． 3．TypeFrequencyinTbrmsofDistributionalRichness InhisUsage−BasedModeloflanguagelearning，Tomasello（2003）hasintroducedan importantdistinctionbetweentokenfrequencyandtypefrequency．BothofthesetwotermsareconcernedwiththefrequencyofalingulSticinstanceexperiencedbythelearner．This
distinction refers to the different natures of frequency：tOken frequency refers to the
frequency with whichthe same speCificinstance appearsinlnput and type frequency indicatesthefrequencywithwhichdifferentinstancesofthesarneconstructionappearin lnput・Itisclaimedthattoken丘equencycontributestoentrenchmentoftheinstanceand typefiequencybringsaboutabstractionandgeneralizationoftheconstruction． Someexamplesmaybeinorderheretoshowthediffbrentcontributionsoftokenand type丘equencyinlanguageleamlng・Englishpastirregularverbsappearindividuallyand areleameditembyitemearlyinleamingbecausetheyareallrichintoken丘equency．In
COntraSt，Englishpastregularverbsappearfbrmalizedinmanydi飴rentverbsandthusthe
generalpatternOfthepastregularisleamedsincethepatternisrichintype丘equency． Whiletokenfrequencyandtype丘・equenCybothrefbrtothefiequencyofoccurrenceof alinguisticinstanceitself，itmayseempossibletoextendthenotionoftypefrequencytotherichness of environmental contextsin which alinguisticitem can appear・Some
lingulSticitemsmayoccurinavarietyoflingulSticcontextsandothersmayoccuronlyina limitednumberofcontexts・Forinstance，OnthephonologlCallevelofEnglish，theword initialclusterofsounds／sa∋／cantakeatleastsixteendifferentphonemesafterit，Whilethe WOrdinitialclusterofsounds／za∋／canonlytakefourdifferentphonemesafterit・（2）
lfalingulSticitemappearsinavarietyofcontextslike／sae／does，WemaySaythatit has a high distributionaltype frequency・In contrast，ifanitem can appear onlyln a restrictednumberofcontextslike／zee／does，WeCanSaythatithasalowdistributionaltype frequency・ThisextendeddefinitionoftypefrequencylntermSOfdistributionalrichnesshas predictionsthathightypefrequencyguaranteesgeneralizationofitsdistributionalcontexts andthatlowtypefrequencyentrenchesandfreezesitsdistributionalcontexts．
thedistributionaldifferencesofthetwoJapanesecontracted sounds【kjaeland【mjeelin Japaneseasdiscussedintheprevioussection：While rkjeelis richin distributionaltype
frequency，【mj車】islow． 4．DistributionalTypeFrequencyandSecondLanguageLearning Distributionaltypefrequencydefinedabovedescribesthedistributionalrichnessofa lingulSticiteminaparticularlanguage・Then，itisaninterestlngqueStionwhetherornot distributionaltypefrequencyinthenativelanguageofasecondlanguagelearnerexertsany influenceinthelearnlngOfthetargetlanguage・lnthisregard，itseemspossibletoexplain
the asymmetrical phenomena of the twoJapanese contracted soundslkjaeIIand【mjae］
discussedearlierinthisarticleintermsofdistributionaltypefrequencyofthern．Inother
WOrds，WeCanSupPOSethat，Ontheonehand，itiseasyforJapanese−SPeakinglearnersofEnglish to apply thelkjae］soundin pronouncing English words becauseit has a high
distributionaltypefrequencyinJapaneseandgeneralization ofits distributionaIcontexts mayalsobeextendedtoEnglish．Ontheother，itisnoteasyforthemtoapplythelmja31 SOundinlearnlngEnglishbecauseithasalowdistributionaltypefrequencyinJapaneseand itsrestrictiveentrenchmentofdistributionalcontextsmayalsorestricttheapplicabilityof thesoundinEnglish．
Based on the argument so far，it may be conceivable that the applicability ofthe Japaneselmjae］soundin English can beincreased by raisingits distributional type
frequency・Thereasoninggoesasfollows：WhileJapaneseqspeakinglearnersofEnglishhave the contracted【mjael soundinJapanese which can potentially be an alternative to the
Englishlm車I sound，the contracted soundislowin distributional type frequencyin
JapaneseandthusitsappIicabilitylSalsorestrictedinlearnlngEnglish，buttheapplicability maybeheightenedbydeliberatelyraislngltStypefrequency．Inthelatterpartofthisarticle， resultsofaquasi−eXperimentfocuslngOnthispossibilityarereported． 5・ClassincationofEnglishWordsBeginnlngwithInitialSingleConsonants＋／ae／ BeforegolnglntOtheexplanationofthequasi−eXperiment，itisnecessarytoclassify Englishwordshavingthe／ae／soundjustaftertheinitialslngleconsonantsintofourclasses according to whether or not corresponding similar contracted sounds are availablein JapaneseandwhetherornotthecontractedsoundsareapplicabletoEnglish・
Tablel shows the result of classification・Availability of corresponding similar
COntraCtedsoundsisdeterminedbycheckingthearrayoftheJapanesecontractedsounds
andtheirapplicabilityinEnglishisjudgedonobservationalfacts．
Japanese（i・e・kya，gya，Cha，ja，Sha）andtheyareeasilyappliedinEnglish・Exarnplesof JapaneseloanwordsfromEnglishsuchaskyanpu，gyqPPu，Chatto，jamu，Shanpuuprovethe applicabilityofthecontractedsoundsinEnglish・ThefourentriesinClassIIallhavethe correspondingsimilarcontractedsoundsinJapanese（i・e・Pya，bya，mya，rya）buttheyare noteasilyappliedin English・NotethatJapaneseloanwordsbelonglngtOthisclassare pronouncedreplacingthe／ae／soundbytheJapanese／a／sound（e・g・Pakku，batto，mattO・ rakku），andtheJapanese／a／soundisalsolikelytobeappliedinprorl0unCingEnglishwords
suchaspack，bat，matand rack．CIassIIIhas sixentrieswith nocorrespondingsimilar
Japanesecontractedsoundsandthusnoapplicability・（3）ThefiveentriesinClassIVare
unlqueinthatallbeginwithconsonantswhicharenotusedinJapaneseandconsequently
theyhavenocorrespondingsimilarcontractedsoundsandnoapplicability・
Tablel C／α∫∫折CαJio〃qfg〝gJ通用伽め風癖血廟印面日通血は軸両C餓叩矧咄＋！詑／加
古・／・川、′り、l…f！−J！−f！中。！JJ川J／（J′’Jl′ハ川k・・、t・（一川J…・・旭日一川〃h八。〃（／／／′t・JJ∴・り一画。・J／−叫、・
C la s s C o m b in a tio n o f V V o rd In itia l A v a ila b ility o f S imi la r A p p lic a b ility o f S imi la r S in g le C o n s o n a n t a n d ／ae ／ C o n tra c te d S o u n d C o n ta c te d S o u n d I ！k お ／，／g a 〕／，／す詑 ！，／匂 お ／，／J 記 ／ Y e s Y e s II ／p ぉ ／，／b お ！，／m ぉ ／，／lぉ ／ Y e s N o III ／tぉ ／，／d a ）／，／n a ∋／，／S 記 ／，／Z 8 9 ／，／h お ／ N o N o IV ／」お ／，／fぉ ／，／V ぉ ／，／e ぉ ／，伯 a e／ N o N o 6．TheExperiment Inordertoexaminetheeffectsofpracticeofrmjae］withitsincreasedtypefrequency，a
quasi．experiment was conductedwith twenty−SixJapanese university students（mostly SOPhomores）majoringelementaryschooleducationintheireightnormalEnglishclassesin thefirstsemester，2007．Thesestudentswerethosenotabsentfromanyoftheeightclasses．
FourdifferentpackagesofmaterialsforpracticewerepreparedutilizlngthreeJapanese WOrds beginningwithlmjEeIl．The common basicidea ofthese activitiesis to raise the
distributionaltypefrequencyoftImjae］bystartingwithoneofthethreeJapanesewordsand thenextendingitintodifferentdistributionaltypeSmakinguseofEnglishwords． Eachofthefouractivitiesconsistedofthreetofourwordstobepracticedfollowlnga Japanesewordintheorderpresentedbelow． Activityl 【mj詑kul→【mj金k］（肋C）→【mj会t］（mdJ）→［mj金p：】（椚（甲） Activity2 lmjaemma：1→【mj会m］（mam）→［mj金nl（man）−→lmj金rlgOUl（mango） Activity3 lmjaesLuko中LuSLUki：］→lmj摩Sl（mass）−→tmj金Jl（mash）一一◆【mj会す］（match）
Activity4
lmjaeku］→【mj由kl（Mac）一→［mj金gl（mag）
→【mj郵I（〝dJ）→ Lmj会d：】（樅ゆ
ItshouldbenotedthatarrangemeTltOfwordsineachactivJtyWaSmadetomakethe transitionofarticulatorymovementsassmoothaspossible・Thesmoothnesswasattainedby keeping the same marlner Ofarticulation（StOp）and transiting the place of articulation consistentlyfrombacktofront（Velar−→alveolar→bilabial）（Activityl），keepingthe samemannerofarticulation（nasal）andtransitingtheplaceofarticulationconsistentlyfrom
fronttoback（bilabial→alveolar→Velar）（Activity2），tranSitingtheplaceofarticulation
graduallyupwardsalongthealveolarridgeandchanglngthemannerofarticulationfrom fricativetoaffricate（Activity3），andturningunvoicedstopstovoicedones（Activity4）・EachoftheseactivitieswasglVeninthefirstfifteenminutesoftheregularclassfor
ninetyminutes・Ineachactivity，theparticIPantSreCeivedinstructionstoputastressonthe
／ae／soundineachEngJishwordandnottoputanepenthesisatitsendwhilelookingata Sheetofpaperonwhichthesewordswereprintedwiththeirpronunciationrepresentedwith Phoneticalphabets・Theparticipantswereremindedthatthestressmarkonthe／Ee／soundin eachEnglishwordindicatedthatthevowelshouldbepronouncedverystrongly・The particIPantSinitiallylistenedtothe modelpronunciation ofthe words bytheir teacheronceandthentheyproducedeachwordafterthemodelwhilelookingatthesheet．
TheTltheyreceivedadi鮪rentsheetofpaperonwhichthesamewordswerepresentedwjth
PICtureS reSpeCtively describirlg the words and theylistened to the modelpronunciation OnCearldthentheylmitatedthemodelonce．Afterthat，the particIPantSreCeivedathird SheetofpaperonwhichonlyplCtureSdescribingthewordswereprintedandtheyengaged inpalrWOrkoflistenlngandproducingthesewordslookingatthesheetofpaper・Importantly，typefrequency ofthelmjae］sourldincreased fromits originalsix to eleveninthesefouractivitiesandtheparticIPantSeXPeriencedthesourldinthesevaried COnteXtSbothreceptJVelyandproductively・Inordertoexaminetheeffectsoftheactivities， threepronunciationtestswereglVemtOtheparticipants．ThefirsttestwasglVenbeforethe treatmentanditcontainedcat，CaP（ClassI，knowntotheparticipants），mat，maP（ClassII， ma−type）andtwootherwordscanopy，CamaS（ClassI，unknowntotheparticipants）．The SeCOndtestadministeredafterthefouractivitiescontainedmat，mqP，man，matCh（ClassII， ma−type，praCticed），marry（ClassII，ma−type，nOn−PraCticed），PariS，bad（ClassII， non一ma−tyPe，nOn−praCticed），Sad，hqppenings（ClassIII，nOn−PraCticed）andrqp，jhct（Class IV，nOn−PraCticed）aswellascat，CqP（ClassI）．Thethirdtestconductedtwoweeksafterthe SeCOnd testincluded mat，man（ClassII−1，ma−type，praCticed），magic，mannerS，matter （CJassII，ma−tyPe，nOn−PraCticed），andhct（C】assIV，r10n−Pr撒Cticed）．
The three pronunciation tests were of the same format and were glVen tO the Participantsinthesamemanner・TheywereinvitedindividuallytoaquletrOOmarldhanded
asheetofpaper？nWhichfourshortconversationsbetweentwoperSOnSWerePrintedwith respectivepicturesdepictingthescenes・（SeeAppendixA）Theywererequestedtoreadthe COnVerSationsaloud・Eachoftheiroralreadingoftheseconversationswastape−reCOrded．
ThesethreetestswereintendedtoexaminethefollowlngSissues：
（l）WhetherornottheJapanesecontractedsound【kjeeIlisapplicabletoClassIwords
unknowntotheparticipants（canqpy，CamaS）？（2）Whether or not and to what extent practice of rmjael withitsincreased distributionaltype frequencylS effective for ClassII ma−tyPe WOrds actually
PraCticed（mat，mqP，man，matCh）？ （3）Whetherornottheeffectofpracticeoflmjee］isextendedtoClassIIma−type WOrdswhichwerenotpracticed（marry，magic，mannerS，matter）？ （4）Whetherornottheeffectofpracticeoflmja∋］isextendedtoClassIInon−ma−type WOrdsbeginningwithsoundcombinationsotherthanma（Paris，bad）andtoClass IIIwords（sad，hqppenings）bothofwhichwerenotpracticed？ （5）Whetherornottheeffectofpracticeoflmjae］isextendedtoClassIVwordswhich WerenOtPraCticed（rqp，カct）？ Allofthesetestsandactivities（includingFifthActivityforgeneralreview）weregiven inthefollowIngOrderontherespectivedates： FirstTest：19thand26thofApril，2007・ FirstActivity：22ndofMay，2007（Mac→mat→maP） SecondActivity：24thofMay，2007（mam→man→mangO） ThirdActivity：29thofMay，2007（mass→maSh→matCh） FourthActivity：3lStofMay，2007（Mac→mag＆mat→mad） SecondTest：5thofJune，2007 FifthActivity：12thofJune，2007（GerLeralreview） ThirdTest：19thofJune，2007 7．ResultsandConclusions
Theresultsoftheexperimentare showninTable2．FromTable2，We Cangetthe followlnganSWerStOthefiveissuespresentedabove： （1）TheJapanesecontractedsound【kjae］isapplicabletoClassIwordsunknowntothe Participants（canqpy，CamaS）．Notealthoughthepercentageofpronunciationwith ／ae／isnothighforcamas，23％oftheparticIPantSPrOnOunCedtheword／k由mas／． （2）Practiceoflmj記］withitsincreaseddistributionaltypefrequencyiseffectivefor ClassIIma−typeWOrdsactuallypracticed（mat，mqP，man，matCh）． （3）EffectofthepracticeismoderatelyextendedtoClassIIma−tyPeWOrdswhichwere notpracticed（marry，magic，mannerS，matter）．
Table2 PeγCe細物ge〆Pro〃〃乃CαJo乃血C〃d物納e／α／∫0〟れd
Ⅵb rd C lass T arget 1SlT e st 2 nd Tb st 3 rd T est N ote C lass I （K now n） CαJ 100 ％（52 ／5 2） 100 ％（52／52） C呼 8 8 ％（4 6／52） 9 2 ％（24 ／2（i） 澗 離 散 耶 勝 研 磨 棚 潤 糟 瀞 研 聯 雄 C lass I （U nkno w ll） C（間呼 γ 7 9 ％（4 1／52） C（7〝1α∫ 35 ％（18／52） 伯〟 ：2 3 ％（12！52） 宴 翻 轢 断 饉 闇 泡 ほ、1 C lass II m かT y pe （P racticed ） 間〝J 0 ％（0／5 2） 7 3 ％（38／5 2） 6 9 ％（18／2 6） 1乃呼 0 ％（0／5 2） 8 1 ％（42／52） 〃J〃〃 75 ％（39／52） 8 1％（2 1／26 ） 椚αgc力 69 ％（36／52） 紺 闇 路 押 馳 鮮 禰 描 旭猷 滅相血路 親 和摘記 流 C la ss II 椚α−T y pe （N o n−praCticed ） 〝1αrJγ 5 8 ％（15 ／26 ） 〃1αg JC 6 9 ％（36／52 ） m α〝〃er∫ 4 0 ％（2 1／52 ） J〃付けPr 4 2 ％（22／5 2） 濫用漕緒捜箭 舶 載 摘 債 蟻 組 鋸 耕 狛 泥滋胱脱 脂散捕 泌据則餞鱈相磯棚減矧滋諜報馳凱即満露蒲説 瞳路矧溺況隕 柑瑚報謝蒜絹紀闇 経 常纏潤鮮 鼎樹璽 離 C la ss II （N o n，P raCticed ） P t汀f∫ 0 ％（0／26 ） あαd 15 ％（4／26） ほ き 法 相 鶴 派 揖 朔 貼翫 拙 ラ． 滞 m 慢乳拙 く 導．、ミ C lass III （N o n−p raCticed） ∫αd 8 ％（2／26 ） 短 揮 嘲海 が 0 ％（0 ／26 ） ぶ 閻 瀧 鮮 甜 謹岩 瀞 餅 瀧 描 は ぼ C la ss IV （N on−P raCticed） r（甲 8 ％（2／26 ） カ タ 8 ％（2／26） カ cJ 0 ％（0／2 6） （D en o駕 篭 霊 悪 霊 霊 悪 霊 漂 i㌫ 忠 霊 綜 霊 窓 t霊 臨 霊 霊 部 （4）EffectofthepracticeisnotextendedtoClassIInon一ma−typeWOrds（Paris，bad） nortoClassIIIwords（sad，hqppenings）．
（5）EffectofthepracticeisnotextendedtoClassIVwords（rqp，jbt，カct）．
From the results，We Can COnClude that（1）practice oflmjae］withits／increased
distributionaItypefrequencyiseffectiveforClassIIma−typeWOrdsactuallypracticed，（2） the practiceis also moderately effective for ClassII ma−type WOrds which were not practicedand（3）thepracticeisnoteffectiveforeitherClassIInon−ma，typeWOrds，Class IIIwords，OrClassIVwords．
ThesefindingsmeetwiththegeneralclaimsmadebyproponentsoftheUsageLBased ModelthatlanguagelearnlngprOCeedsitembyitemandthatitisonlylaterandgradually
thatgeneralizationofanykindbeginstoappear・ Notes （1）Thereareelevencontactedsoundsendingwiththe／a／soundinJapanese（i．e．kya，Sha， Cha，nya，hya，mya，rya，gya，ja，Pya，bya）．AccordingtoHattori（1979，pp．132−3），all thesecontractedsoundshavetheglide／j／soundbefore／a／．Thus，the／a！soundinthese COntraCtedsoundsassumesthequalityof／Ee／． （2）／S詑／：∫α如晩∫α⊆，∫α∈edoαJ，∫α血J，∫dd∫伊の，∫α＆，∫ααd，∫α辺0控，∫α授d，∫喝も∫呼， ∫αとαわれ4∫β通，∫α壬，∫αとdge；／Zまき／：Zdg，Zα辺如α，Zb生がαれ q臣 （3）Notethat，inJapanese，nyaispronouncedlJlaelratherthanlneelandhyaispronounced l9ae］ratherthanlhae］． Acknowledgment
This paperis a result ofthe research being undertaken as a Collaborative Prqject （PrqjectG）fundedbytheJointGraduateSchool（Ph．D．Program）inScienceofSchooI EdtlCation，HyogoUniversltyOfTeacherEducation．
Refbrences
Hattori，S．（1979）Shin−ban on−in−rOn tO Seisho−hou：Shin nihon，Shikituzurino teishou け！い〃′・いヾ．、・刷！川7ル脚岬／1、、・上＼一川・l・tnJ朝日−′・りい、fJJ朝。／J！7川川・…、一・一！リ．り、潮し・、、し・ SPelling）．Tokyo：TaishukanShoten．
Jenkins，J．（2000）．ThephonologyqFEnglishasaninternationallanguage．Oxford：Oxford UniversityPress．
Makino，T．（2005）．Nihonjin no tameno eigoponsei−gaku ressun（Lessons qfEnglish PhoneticsjbrJqpanese）．Tokyo：TaishukanShoten．
Matsuzawa，K．（2004）．Eigomimi：Hatsuongadekirutorisuningugadekiru（Englishear．・ qyoucanpronounce，thenyoucanlisten）．Tokyo：Ascii．
〃肋間0〟gO血行加〃（J呼α乃eJe乃α血服用明押掛沌明勘娩血胱町）（2ndversion）．（2000）． Tokyo：Shougakkan．
Tomasello，M．（2003）．Constructing alanguage：A usage−based theoryqflanguage acquisition・Cambridge，MA：HarvardUniversityPress．
Yamaoka，T．（in press）・Eigo o manabunihongo−WaSha no／ae／no hatsuon gakushuu ni tsuite：Nan−inosetumeitoshidoushishin（Onthelearningofpronunciationofthe／a3／ SOund byJapanese−Speakinglearners ofEnglish：Explanation ofits difficulty and guidelinesforteaching）．NihonKyoukaKyouikuGakkaiShi，30（4）．
AppendixA SecondPronunciationTest 1． A：Doyouknowtheman？ B：Themaninacap？ A：Yes． B：Ofcourse．I’mgolngtOmarryhim． Helovesrapmusic！ 2． A：Showmethemaponthedesk． B：Sure． A：Oh，thisisamapofParis． 3． A：Lookatthecatonthemat． B：Oh，it’safatcatonabluemat！ A：Yeah，Shemayberemembering thehappenlngSOftheday． 4． A：Howwasthetennismatch？ B：Itwasaclosematch．ButIlostit． A：That’stoobad． B：Yes．I’msad．
