Okurigana : An Analysis from the Viewpoint of Decoding

著者 本田 啓輔 雑誌名 筑波応用言語学研究 号 15 ページ 87-100 発行年 2008-11-30 URL http://hdl.handle.net/2241/101989 伽〟r励： An Ana）ysis from the Viewpoint of Decoding

Keisuke Honda

Keywords：TheJapaneSeWritingsystem；Decoding；KdTyi；Hiragana；Okurigana

1．lntroduGtion IntheJapanesewritingsystem，hirqganagraphemesconstituteokurikanawhentheyare postposedtokafyigraphemestospe1loutasectionattheendofwords（1a）orelements ofcompoundwords（1b）・1

（1）WR汀TENFORM a．食 べ た2 b．食 べ 過 ぎ た TRANSuTERÅTION tabe be ta tabe be sugigi ta

GLOSS tabe－ta tabe十Sugl－ta eat－PAST eat＋exceed－PAST

Aswi11beel止oratedonbelow，Okur如naarebasicallyusedfbreaseofkaTdidecoding （Hoshina1949，Hayashi1977，Kond62005，Satake2006）．Putdi銑rently，OnemaySay thattheyprovide certainexternalsupport，without which ka吋idecodingwouldbe difncultorevenimpossible．Whatarethedi餓culties，then，andhowcanWefbrmulate the wayln Which okurigana reducethem？Besides，do okuriiana o蝕r su餌cient linguisticinfbrmationrequiredfbrdecoding？Ifnot，Whatkindsofpriorknowledgeare required，Onthebasisofwhichtheycanbefu11yutilised？ TbanSWerthesequestions，thispaperpresentsadescriptiveaccountofokuriiana 丘omtheperspectiveofdecoding，aimlngatPrOVidingbetterinsightsintothenatureof kaTyi，hirqganaandtheirinteractioninJapaneSe．Thepaperisorganisedasfbllows・ First，SeCtion20utlinessomebasicissues，eSPeCiallywhatwillbereferredtoasthree CaSeS OfkaTdiindeterminacy．Next，SeCtion3presentsananalysis ofthe cⅦTent

1Thispaperusesgrqphemetorefbrtothebasicunitofdting（Sproat2000），Withoutcommitting totheoreticaldiscussionsontheappropriatenessofthistraditionalterm（Daniels199l，1994，Herrick 1994a，b）・ForkaTyi，itisusedfortheso－Calledcharacter，andnotfbritsco叩Onent・ 2Throughoutthispaper，eXamplesarephonemical1ytraJISliteratedusingthe肋mねikiromamisation SyStem（Cabinet1954），with a ftwminor mod摘cations．Upper cases areused toindicate Sino－Japaneseitems．TheencodedlinguisticitemsareglossedonlywhennecessarylWherethese detailsareirrelevaJlt，eXamPlesareshowninasimplerfbrmat（e．g．，食べたtabe－ta’ate’）・

ー87－ COnVentionsconcemlngOkuriganaorthographytoformulateokurigana’scontributionto theresolutionofthe ka吋iindeterminacies．Thensection4discusses someproblems Caused by orthogr叩hically permitted variarrts of okurigana・Finally，SeCtion 5 Summarisesthediscussionanddrawsaconclusion．

2．Settings 2．1伽〝igmaas the［ociof kaq／i・功iragaIWinteraction JapaneSeissubstantiatedbythreesetsofgr叩hemesorscripts，namelykanii，hiragana andkatakana．Ofthese，the丘rsttwoconstitutethemaincomponentsofthegrapheme SyStememPloyedinthecurrentJq）aneSeWdtingsystem・TbxtsaretyPica11ycomposed OfbothtypeSOfgrqphemes，eaChcarrylngOutthespecincfunctionssketchedbelow Ka7digraphemesencodelinguisticelementswhichbelongtothenative（NJ）andthe Sino－JapaneSe（SJ）strataofthelexicon．Characterisationoftheseelementsisanissue Ofongoingdebate（e．g．，Kaiser1995，2007）；forthepresentpurpose，however，itwill Sufncetoregardthemasindividualmorphemes（e．g．，K6no1994／1977，SamPSOn1985， Rogers2005）．3contrastively，hiraganagraphemesencodemoraeor；mOrePreCisely， mora－bearingphonemestrings（i．e．，（C）Vunitsorsegmentsoccupyingthesyllal）1e・COda POSition）（e．g．，Poser1992，Ratcli脆2001，Rogers2005）．Tbday，theyareprimarilyused forrepresentingnon－1exicalelementssuchasafnxes，CaSeParticlesandpostpositions・ Withthiskindofdivisionofhbourbetweenkanfiandhiragana，theJqpaneSeWdting

3 Therearetwopointswhichseemtojustifythispostulation．Thenrstpointisconcernedwiththe Sizeofthelinguisticelementstowhichindividualgraphemescorrespond．FortheSJset，kafyiare usedfbr魚・eefbrms（e．g．，金fbr幻Ⅳ‘gold，）andcombiningfbrms（e．g．，鉱fbrKOO’metalmine’， onlypermittedincompoundssuchas金鉱KLⅣ＋KOO’goldmine），Whichcanbegeneralisedas rootmorphemes．FortheNJset，Ontheotherhand，kaTdiareemployednotonlyfbr舟eefbrms（e・g・， 夢yume‘dream，），butalsofbrwordswhichfbrmanin鮎ctionalparadigm（e・g・，見fbrmi－ru ‘see－NON．PAST，，mi－ia’see－PAST，，mi－rO’see－1MP’，etC．）．The recurrent use of the same graphemesinview，itseemsreasonabletosaythatthesegraphemesareassociatedtotherecurrent linguisticelements，namelythestemmorphemes・Theseobservationsaぽbrdgroundsfbrpositingthat ka7yicorrespondtomorpheme－Sizedelements・ Thesecondpointisconcemedwiththelevelofabstractionatwhichkafyioperate：itisapalpable factthatthegraphemesdonotindicatemorphophonologicalalternationsinagraphicallydistinctive fashion．ForinstanCe，Observethefo1lowlngCaSeOfcompounding，WhichinvoIvesvowelaltemation inthenrstelement（eうa）andconsonantvoicinginthesecond（kうg）：酒sa短‘ricewine，十蔵 kura’storehouse，⇒酒蔵sa垣＋gura’rice wine cellar’．Importantly，these graphemes do not Changetheirfbrmsaccordingtothealternations，SuggeStingthatkaTdirelatetolinguisticelementsat a deeperlevelthan the surface phonemiclevel・These two points taken togetheちit seems descriptivelyadequatetointerprettheseelementsasmorphemes・ Note，however；thatthisgeneralisationisnotwithoutexceptions．In血ct，SOmekaTUiencode polymorphemicitems（e．g．，政maturi－gOtO’politics’）orsub－1exicalphonemestringsthatdenoteno meaningsontheirown（e．g．，葡BUand萄DOOin葡萄Btn）00’grape’）・FurtherresearChis neededfbramoresystematictreatmentoftheseissues・

－88一 SyStemischaracterisedbytheinteractionofdif沌renttypesofgraphemes・ Perh叩Sthemostremarkablecaseofka吋i－hiraganainteractionisinstantiatedbythe useofokur如na・Ashasbeenalreadymentioned，formallyokuriganaarehiragana graphemesemployedtospelloutalatterpartofwrittenwordsorelementsofcompound WOrds・Morespecincal1y，theyarepOStPOSedtoka7digraphemesinawaythattheyspe1l Outinflectionalendings（2a），aSeCtionattheendof丘eefbrmsorstemmorphemes（2b）， orboth（2c）．4Inthis way，ka7diand okuriganaamalgamatewitheachother to CO－COnStruCtWdttenwords．

（2）a．見 た b．且 つ c．食 べ た

椚f ぬ 鬼α蝕 血 ぬ∂e ムe J〟 mi－ta katu tabe－ta See－PAST besides eat－PAST

2．2The current convention＄for okwi8mOrthography Thescopeofthepresentp叩erislimitedtotheconventionsstipulatedbyJ卸∂助吋i 坤∂（Cabinet198la）andOkuriganano乃ukekata（Cabinet1981b）．］卸∂Kd吋i砂∂is anOfncialka吋iinventory，Which containsl，945commonly used graphemes U卸∂ ka吋i）and4，087associatedreadings．Despitethenon－bindingnatureofthisinventory，it isadefactostandardadoptedinlegalandpublicdocumentS，aPPrOVedtextbooks，m叫Or dailiesandvariousothertypesofpublications． Okuriganano乃ukekata，Ontheotherhand，isacollectionofregulationsconcern1ng Okurigana orthography・These regulations are prlmarilyintended as guidelines on applyingokuriganatojqy∂ka吋igr叩hemes；infact，however，theyarealso叩Plicable toothernon－Ofncialgraphemesthatareactuallyinuse・Asahybridoftraditionalusage andmodernattemptsofsystematisation，Okuriganano乃ukekataallowsmanyVariantS and exceptions besidesthe basicforms de魚ned by generalmies・In general，the PermittedvariantSCOmPnSeaSmallernumberofokuriganathanthebasicformsdo． （Thispointisdiscussedinsomedetailinsection4．） Adoptingtraditionalus？ge，Okuriganano乃ukekata（Cabinet1981b）constrain＄the

Adoptingtheconventionalconceptionofokurigana，thispaperassumesthatevenifhfrqgana OCCuPythepost－kaqposition，theydonotconstituteokur＆aiftheyrepresentthefbllowlng： i）Anelementsyntactical1yaqiacenttowhatthegivenka7direpresents（includingparticlesand POStpOSitions）（e．g．，夢が押椚ega’dreamNOM，）；Or ii）AnelementofacompoundwordmorphologicallyaqjacenttowhatthegivenkarUirepresents （e・gり見かた刑卜血ぬ‘modeofs¢ei咽り；Or iii）Thelightverbsuru（e．g．，心するkokoTV巾uru’beonthealert，）．

ー89－ applicabilityofokuriganawithrespecttothetypesofkanfireadings・First，theyare usedonlyfbrNJitems，andnotfbrSJitems（asstatedin台4－2－40fthePrefhceto OkurigtmanoI吉ukekata）．Second，Okur卸naarenotemployedforNJnominals，eXCePt fbrnominalderivativesandasmallsetofexceptionalitems（asstatedin専2’Words Withoutco再ugationT）．Becauseoftheseconstraints，atPreSentOkuriganaareprimarily usedfbrNJnon－nOminalitemssuchasverbs，a句ectives，adverbs，etC．

2．3伽〟ribav7aaS aides to decoding General1yspeaking，Wdtingcanbeviewedasarelation，inwhich“particular（SetSOf） 1inguisticelementslicensetheoccurrenceof（setsof）orthogr叩hicelements”（Sproat 2000：9；italicsintheoriginal）．Onceestablished，thisrelationcanbeimplementedas mappingsinbothdirections：録omlinguisticelementsontoorthographicelements（i・e・， ビガ甜威喝），OrViceversa（i息，血od塘），5 InJapaneSe，kaniigraphemesareknowntOCauSedi疏cultyindecoding，andprevious studies have shown three cases where okurigana may be considered to remove or reducethedi餌culty．6First，Satake（2006）pointstothegeneral1ackofphonological infbrmation onthe part Of ka頑i．Tb be more specinc，ka吋igraphemes are not analytically struCturedinawaythatsma11ergraphicconstituentsindicates豆b－1exical Phonemestrings（e．g．，Syllal）les，mOrae，Phonemes，etC．）regular1yandsystematically （K6no1980／1969）．7Thisisillustratedin（3），WherethephonemestringBNisencoded by a setofkaTUigraphemeswithnocommongraphic constituent・The absenceof graphically expressedphonemicanalysisconstitutesaseriousproblemfordecoding， Sinceitmakesitimpossibletoderivethephonemicformofakafdireadingdirectly fromthegrapheme’sarChitecture．Herea氏er，thisproblemwillbetentativelyrefbrredto asthein（ね（el・minaり｝（！fphonemic．hT・mS．

（3）属八「・う（円，延，沿，炎，宴，…‡

5ThisisnottosaythatencodinglSSimplytheinverseofddcoding，Orviceversa・Foradiscussion Onthedi飴rencesbetweenthesemappingsinJapanese，SeeKawai（1970）・ 6Thispointdeservesexplicitemphasis，SincemuchmoreattentionhasbeenglVentOtheencoding sideofokurigana，O鮎nwithparticularrefヒrencetoissuessurroundingorthographicregularisation （e．g．，KokuritsuKokugoKenkyqio1952，Ono1958，T如kishima1970，Uchida1972／1961，Hayashi 1977，Abe1989），Forareviewoftheseissues，SeeAbe（1989）andlもkebe（1979），interalia． 7Inprinciple，theso－Ca11edphoneticcomponentsofsemanti叩honeticcompounds（keねeika頑）are designatedtoindicatethepronunciationofSJitems・Nevertheless，therearediscrepanciesoverthe actualefEbctivenessofthephoneticcomponentsinpresent－dayJapanese（Nomura＆It61978，Stalph 1989，Saito，Kawakami＆Masuda1995）．Moreover，these components provide virtua11y no infbrTnationatallaboutNJitems（exceptfbrasmallnumberofkok74igraphemes）．

－90－ Satake（2006）maintainsthatthefunctionofokuriganaistomakeupfortheshortageof Phonologicalinfbrmationin ka7digraphemes．Ⅰmitively，this remark cq）tureSthe empirical払ctexperiencedbythewriter－readerofJapaneSe：itisoftenthecasethatone makesre丘rencetothephonologicalinformationofokuriganatoworkoutunCertainor unkn0wnkaQireadings．Thatsaid，Satake（2006）doesnotprovideadetailedaccount and，aSamatterOffact，VeryftwstudieshavetakenpalnStOdiscussthephenomenonat hand・Accordingly，1tremainsunClearwhatkindofcontributionokuriganamake，and howitworksfortheeaseofdecoding． InadditiontotheindeterminacyofphonemicfbrmS，tWOmOreCaSeSOfindeterminacy arisefromthehighlyambiguousnatureofkaTyi．AsⅥnce（2002），Kond6（2005）and manyothersobserve，alargenumberofkaTdigraphemeshaveseveralpossiblereadings・ Thisis a situation knownaS homogr呼妙，Where a single orthographic element COrreSPOndstomultiplelinguisticelements（Ⅵmce2002，Rogers2005）．Homographic ka吋iareinl1erentlyambiguous becausethese graphemes do notincorporateany COnStituentsmarkingoutthesingleintendedreadinglnagr叩hica11ydistinctivemanner・ Tb makethe matters more complicated，many homographic ka頑iare POtential1y ambiguousattwolevels（Kond62005，Honda2007）．Atthemorphemelevel，OnOne hand，aSlngleka吋imayencodeseveralNJmorphemes，SeVeralSJmorphemes，Orboth （4a）．Attheword－fbrmlevel，Ontheother，thesamekaTdimaybeusedfbrasetof VariousinnectedfbmlSrealisingaparticularstemmorpheme（4b）．Asaresult，itis hardlypossibletopredicttheintendedreadingofahomographicka吋ibyjustlookingat itsstruCture・Thepresentpapertentativelylabelsthesetwokindsofambiguityasthe in（h（erminaqlQ［mofl，hemesandthei17（k（erminaq∫Qf－i7dlecfed．わ7・mLY，reSPeCtivcly．

（4）a．食 う 〈ねあβ－，た打伸一，血相W－，ぶ阿片坊罰び‡ b．食 う（ねゐβ－r〟，Jd∂ピー叫ね加一和，．．．）

Itiswidelyacceptedthatokuriganabringaboutane飴ctivesolutiontotheseproblems， too．Attheword－fbrmlevel，theyspelloutsomeorallrelevantelementssothatthe targetin鮎ctedformisexplicitlydistinguishedfromotherpossibleforms（Tbukishima 1970，SamPSOn1985，Abe1989，Vance2002，andmanyothers）．Inaddition，SeVeral researchers have suggestedthatokurigana servetO disambiguatehomographickaQi alsoatthemorphemelevel（Hoshina1949，Hi11Ⅰ967，SamPSOn1985，Ⅵmce2002， Coulmas2003，Honda2005）．TbsumuPtheirdescriptions，Okuriganaspecifythesingle intendedmorphemebyspellingoutasectionattheendofitsrealisedfbrm（thispoint wi11befurtherdiscussedinsection3．3）．However，despitetheoccasionalrefもrenceto

－91－ thesepointsintheliterature，StillwantinglSaSyStematicanddetailedanalysisofthe mechanismbywhichokuriganaservetOreduceambiguity．

3．lndeterminacy resolution Inlightofthelackofelaborateinvestigations，thissectionattemptstoformulatetheway Okur如nacaJlbeconsideredtoresoIvethethreecasesofkafyiindeterminacy．Foreach CaSe，relevantconventionsofOkuriganano乃ukekataareanalysedtoglVeafbrmal accountoftheindeterminacyresohrtion．Theanalysisisalsoaimedatidentifyingthe kinds ofpriorknowledgethatarepresupposedforman1Pulatingtheinformationof Okuriganaforthepurposeofdecoding．

3．1Reso（ving theindeterminacy of phonenic forms Asmentionedearlier，theindeteminacyofphonemicfbrmsisascribedtothelackof graphical1yexpressedanalysisoflinguisticelementsintosub－lexicalphonemicstrings・ OkuriganaplayacruCialroleinresoIvingthisproblembyspellingouttheendofthese elements，Whichine飴ctaddsonpartialphonemicanalysisandthusspeci鮎stheka吋i readings．Logical1y，thisisapplicableinallcaseswhereokuriganaarepresent，Sincethe lackofphonemicanalysisisapropertycommontoal1kaniigraphemes・ WithintherangeOfOkur如nanonLSkekata（Cabinet1981b），thisismostvisiblein 丘ee fbrmS SuCh as（non－derived）advetbs（5わ，COrtiunCtions（5b）and，in some exceptionalcases，nOminals（5c）．Here，Okuriganaareappliedtokaniigraphemesin SuChawaythattheyrepeatthe重nalmora（e）oftheintendedreadings．Inthisway，they SuggeStthephonemicfbrmoftheentirewordandthusresoIvetheindeterminacy．

（5）a．再 び b．且 つ C．辺 り

、舟ねねゐ仁鋸 肋〟 血 〟ぬr～ r～ futatabi katu atari again besides su汀0undings

Similarly；Okurikanaindica士e the realised phonemicformS Ofstem morphemes by repeatingasectionatthestem－Bnalposition・8InJapaneSe，Verbstemscanbedivided intotwoclasses，namelythoseendingwithavowel（VJverbs）andthoseendingwitha COnSOnant（C－Verbs）．Followingthetraditionalusage，Okuriganano乃ukekatastipulates

80kur卸nano乃ukekatapermitsmanyexceptions，makingthematterfhrmorecomplicatedthaJlis describedhere．Ytt，thisdoesnota飽ctthepointbeingmade，aSinanycaseokurな訊arepeata Sma11erorl町gerSeCtionatthestem一缶nalposition．

ー92－ thatokuriganaspelloutoneormorestem一員nalmora（e）inVLverbs（6a），andtheunit comprisingthestem戎nalconsonantandthesu餌x－initialvowelinC－Verbs（6b）．9For a4jectives，thebasicru1eistospellouttheinnectionalsu疏x（i．e．，－ior－na）；inreality， however，theyalsorepresentstem－nnalmora（e）invariousclassesofitems（6c）．10

（6）a．食 べ る b．飲 む c．楽 し い

Jα占β 占g r〟 〝0椚 椚〟 Jα乃0∫～∫f J

tabe－ru nOm－u tanOSl・1 eat－NON．PAST drink－NON．PAST pleasurable－NON．PAST

Atthispolnt，itisimportanttOStreSSthatokurなanaprovideonlypartialphonological infbrmation；thatis，theydonotrepeatthewholephonemicfbrmOfakaniireading（e．g．， ＊再ふたたび，＊食たべる，etC．）．‖Accordingly，thephonemicfbrmcannOtbesimply dbrived丘omthewittenfbrm；rather，itmustbeinP？rred録omthepartialinfbrmation provided．Suchinftrencenecessarilypresupposestwokindsofpriorknowledge，namely the knowledge aboutthephonemic fbrmS Ofthe glVenlingulStic element，and the knowledgeaboutgrqphemepto－Phonemecorrespondencerules．Withoutthem，thereis nowayofutilisingtheinformationofokur如nainanymeaningfulway．Tbsumup，One maysaythatokuriganaresoIvetheindeterminacyofphonemicfbrmsonthebasisthat thereaderhasaccesstotheknowledgeofbothtypes・12

3．2ResoJving theindeterminacy ofinf］ected forms Insection2．3，itwasshownthattheindeterminacyofin皿ectedfbrmsarises丘omthe

9ThisdescriptionisbasedonamorphologlCalanalysISOfJapaneseverbs，Whichdiffbrs＆omthe traditionalanalysIS Ofverbcolt）ugationin severalrespects．Forexample，什omthestandpointof morphology，thetwoverbsin（6a）and（6b）canbeanalysed astabe－ruandnom叫reSPeCtively， Wheretabe－aJldnom－COnStitutetheverbstems．Intraditionalschemes，thesameverbs maybe

endwith TSi（e・g・，itUirusi－i’remarkal）le’），一ka（e．g．，atataka－na’wam’），ヅaka（e．g．，0両脚kaTna’calm’），Or －raka（akiT・aka－na’obvious’）．The sameis also true fbr several1exica11y speci鮎ditems（e．g．， αrαJα一刀α‘new’）． 11This standsin marked contrasttojhrigana，Orhirqganagraphemes sup叩OSed onka ゆめ graphemestospellouttheka4ireadingin餌1（e．g，，夢yume’dream，；ゆうyu，めうme）．In thiscase，thephonemicfbrmcanbederiveddirectlyfromthephonologicalinfbTmationPrigma 駅諾 ape，猥S皿eS血hen。db，maybeahmmuse，。ftheJa，弧eSeWritingsystem（nadve。T non－native，leamedorleaming），OramaChinedesignedtoconvertw血ingintospeech（SuChasa text－tO－SpeeChsynthesissystem）．ThepresentdiscusSionisconcernedwithmattersintemaltothe Wdting system per se，and notwiththe psycholinguistic process ofreading or the system arChitecture．

ー93－ PrOblemofhomography－inducedambiguityattheword－fbrmlevel・Therecurrentuseof thesameka吋iforvariousin瓜ectedfbrmsmakesitimpossibletodeterminetheintended fbrmbylookingatthegraphemeitself二MeanWhile，itwasalsoshownthatthereisa Clear consensusamOng reSearChers that okurigana specifytheintendedform by indicatingtheinnectionalending． ThebasicprlnCipleofOkurikanano乃ukekataisthatokurなanaspelloutinnectional endings，Whichconsistofsu餌ⅩeSdenotlngtenSe，aSPeCt，mOdality，etC・13Byextension， Okur如na are c叩able of specifyingany POSSibleinflectionalfbrm，eVenifitis morphologicailyhighlycomplexasmi－SaSe－rare－tagar－anaka－at－ta’didnotwanttObe madetosee，（7）．14

（7）見 さ せ ら れ た が ら な か っ た 邦吉一 ∫α ∫g rJ7 作 ぬ 脚 r（7 乃α 血 9 ね

mi－SaSe－rare－tagar－anak－at－ta See－CAUS－PASS－WanトNEG－be－PAST

Interestingly，thespellingoutofinflectionalendingsisnotonlyexhaustivebutalso Surface－Orientedin terms ofthelevelof abstraction．InJapanese，there are morphophonologicalprocessescollectivelyknownaSOnbin，Whereincertainclassesof Stem－nnalconsonantSalternatewith0ther segments when concatenated to some innectionalsufnxes．ForinstanCe，Whenthestemskak－‘write’，yOm－‘read’and ‘take’are fbllowed bythe pasttense sufnx－ta，they become kai－，yONland toQ－， respectively（8a－C）．15Notethatokuriganarepresentthesurfacephonemicfbrmsrather thanthemoreabstractmorphophonemicforms．

（8）a．書 い た b．読 ん だ c．取 っ た

如か f ね γ0椚－〃 血 わr－ e ぬ

ka［i］－ta yO【n］－ta tO［t］－ta Wdte－PAST read－PAST take－PAST

WiththeexhaustiverepresentationofthesurfhcephonemicformS，Okuriganaspecify

13 TherearevaryingviewsastowhatconstituteinnectionalsufnxesinJapaneSe（e．g．，NaJTOg1998，

i。，di。n，thesymb。1。d。n。t。SC。nS。n狛．1engh，Whichism。，ai。in，apm配 15 InthetranSliteration，theelementsshowninsquarebracketsdenotethesu血cephonemicfbrms・ ThesymboINdenotesthemoraicnasal，Whichmayberealisedby［n］，［m］orb］asaresultof regressiveassimilation．

ー94－ theinnectedfbrmSeXplicitlyandinvariably・Thisisobviouslyane飴ctivesupportfbr decoding，Whichcanbeexploitedonthebasisofknowledgeaboutinnectionaland derivationalmorphologyoftheJapaneSelanguage．

3，3ResoIving theindeterminacy of morphemes Theindeterminacyofmorphemesresults録omtheproblem ofhomogr叩hy－induced ambigultyatthemorphmelevel，WhereinaslnglekaTdiencodesseveralNJmorphemes， SeVeralSJmorphemes，Orboth・Themorphemesmaybemoreorlesslooselyrelated WitheachotherintermSOfmeaning，butusuallythereislittleornosimilaritybetween therealisedphonemicforms．ThisisillustratedbytheexamPle（6a）（repeatedhereas （9）），Whichencodes重vedistinctmorphemes，allrelatedtothemeaning’toeat’・Asis ClearfromthisexamPle，neitherthegraphemestructuresnorthemorphemes，meanlngS provideasolidbasisfbrdetermlnlngtheintendedmorpheme．

（9）a．食 う 一匝払漣抑，血相W－，ぶW拭Z庶丹

Asfarasfhctorsinternaltotheorthographicrepresentationareconcerned，Okurigana O飽re銑ctivesolutionstothisproblem・Oneformofsuchsolutionscomes舟omthe Orthographic conventions．AsKaiser（1995）points out，thepresence ofokurigana implicatesthatthegiven kaTdisteLnds fbr a NJmorpheme（10）・Thisis because okuriganaareconventionallylimitedtoNJnon－nOminalreadings（SeeSeCtion2・2）・16 WiththeknowledgeofthisorthographicconstraintandofthedistinctionofNJandSJ items，itispossibletodistinguishNJnon－nOminals＆omSJitemsandNJnominal＄・17

（10）食 べ 物 C£ 食 物 ゐe 椚0〝0 ぶWぴAのrU’

The phonologlCalinformation of okurigana enables a more explicitform of indeterminacyresolution．AsHoshina（1949）andHi11（1967）mentionit，Okurigana

沌TheabsenceofokuriganadoesnotnecessarilyimplicatethattheintendedmorphemeisSJ，Since 三 ‥ざ aqacentgraphemes．Inatwo－kaTyistring，fbroneinstance，thereadingofthe伽stgraphemetendsto beSJ（e．g．，垂水迎＋StH‘seawater，）．Nonetheless，SuCh‘rulesofthumb’donota恥rdasound basisfbrde＞COdingbecauseitiseasyto触dcounterexamples（e．gり海猫zm2i＋neko’black－tailed gull’）．Amorereliablealternative，prObd）ly，istocheckthesyntacticrelationbymakingrefbrenceto a句acenthiraganagraphemes，Whicharelikelytorepresentcaseparticlesandothergrammatical elements．Combinedwith one，slexicalknowledge，itmightleadtothe corTeCtdecodingofa homogmphic画才．

－95－ WhichspecifytheintendedinnectedfbrmalsospeCifytheintendedmorpheme・More SPeC泊cally，SamPSOn（1985）andCoulmas（2003）arguethatwhenokuriganaspellout the丘nalconsonantofaC－Verbstem，theyareine飴ctru1ingouta1lotherpossible morphemeswhichdonotendwiththisparticularconsonant・TocitetheirexamPle，the grapheme歩encodes（amOngOthers）twonear－SynOnymOuSNJverbs，namelyarukl ‘walk’andLD｝um－’walk’．Theokuriganaく kuspec摘estheformerbyrepeatingthe Stem一重nalk（11a），andむmuspecinesthelatterbyrepeatingthestem一重nalm（11b）・

（11）a．歩 ⊥ b．歩 迦

（〟用か，鱒‡ 血 （惑，q職朋－） 血 †∴∵」

Honda（2007）extendsthisnotiontodrawawidergeneralisation．Thebasicideaisthat thesameloglCaPpliestoallclassesofmorphemeswi10Seendingpartisrepeatedby Okurigana．WithintherangeOfOkuriganano乃〟kekata，theseincludestemsofVverbs andmanyaqjectives，aSWe11as録eefbrmssuchasnon－derivedadverbs，COqugations andasmallsetofnominals（seesection3．1）．Again，Okurなanaspecifytheintended morpheme by spelling outits ending portion，While discardingallother possible Candidateswhichdonot丘ttheindicatedphonemicform． Additional1y，Okurikanano乃ukekatahasbeenfine－tunedsothatokuriganacanbe utilisedfbrdistinguishingbetweenwordsinthesameWOrdfhmily．Thisisillustratedby thetranSitiveandintranSitiveformsoftheverb’tomix’（12a－C）．Morphologically，these WOrdsmaybeviewedasrealisingthreedistinctverbstems（i．e．，thetranSitivemaze－， andtheintranSitivemamr－andmazir－），Whichareinturnderive録omthesingleroot maz－．Here，Okurigana spe1loutnot onlytheinflectionalsu餌Ⅹ（－r）u，but also the Stem一銭nalsectionswherecontrastsoccur（i．e．，－Ze靡，－Zar－andrzir－）．Thismeasure，且rst formulatedbyUchida（1972／1961），istakensolelyfbrthesakeofdecoding．Withoutit， thetwointranSitivefbrmSWOuldbewrittenidentically（i．e．，混る），andthismighteven beconfusedwiththetranSitiveform，eSPeCia11y・Whenitistakenoutofcontext．

（12）a．混 ぜ る b．混 ざ る c．混 じ る

〝‡αZe－ Zβ r〟 〝7αZαr一之α r〟 〝7αZJr一之ヱ r‡J

Asintheresolutionoftheindeteminacyofphonemicfbrms（section3．1），Okur卸na PrOVideonlypartialphonologicalinformation．FromthisitfbllowsthatresoIvingthe indeterminacyofmorphemesalsorequlreSthepnorknowledgeofphonemicformS，On Onehand，andofgrapheme－tO－Phonemecorrespondences，Ontheother．Providedthat

－96－ thesetwokindsofknowledgeareavailable，Okurなαnamaybeutilisedtospecifythe intendedNJnonqnominalmorphemephonographical1y・18

4．lndeterminacy resoIutionin the permjtted varjant＄ OkurigananoThkekatapermitsarangeofvariants，Whichgenera11ycomprlSeaSmaller numberofokuriianathan the basicforms do．According tothe regulations，these VariantSCanbeused”aslongastherearenorisksofmisreading”．Thejudgementasto Whethertheymaycausemisreadingisle氏toindividualwdters．Thus，eVenthough Pr10rityisglVentOthebasicformS，thereisvirtuallynolimittotheuseofthevariants． lnreal1ifb，theyareoRenprefbrredoverthebasicformsbecausetheyconsumeless SpaCe，ink，energy，etC． Becausethepermitted variantS COnSist ofa smalleramOunt Ofokurigana，they PrOVideasmal1eramountofinfbrmationthatcanbeexploitedfbrdecoding．Forone example，thegrapheme祭encodestwomorphemes denoting’fbstival，，namelyNJ maturiandSJSALInitsbasicfbrm，thefbrmerisdttenas祭り，Withtheokurigana り rirepeatingthelastmora．Inthisway，itisdi飴rentiated舟omthelatter，WhichisSJ andthustakesnookurな・ana．However，itispermittedtoomittheokur如nainquestion， andthisresultsinahomographic representationthatisinherentlyambiguous．This VariantisactuallyattestedinvariousfもstivalnameSaCrOSSJapan，andwhetheritshould bedecodedasmaturiorSAIislexicallyspec摘edaccordingtoindividualfbstivals（13）． Fromtheviewpointofthepresentstudy，thisisclearlyasetbackfbrdecoding．

（13）葵祭aoi＋ma（uri‘AShinto蝕tivalcelebratedinKyoto， C£？aoi＋SAI

An0therexamPlecanbefbundintheNJcompounduke十tuke’reception’．Thebasic fbrmforthiswordis受け付け，Whereintheokuriganaけkespellsoutthe丘nalmora in bothelements（14a）．Althoughthe two ka7digraphemes encodethree distinct morphemesrespectively（i．e．，受う（uke－，ukar－，ZYU）；付う（tuke－，tukar－，FU））， the okuriganaindicatestherealisedphonemicfbrmSandthusspeci鮎stheintended morphemes．Nevertheless，thisokuriganaisoftendroppedinoneelement（14b）orboth （14c）．Tb decodethese variantS，itis necessary to make refbrenceto one’slexical

柑 Tbbe触thisconclusionholdstrueOnlylnCaSeSWhereokur如nadoaccompanyka7di graphemes・Inreality，theapplicabilityofokzq如naisheavilyrestrictedbyseveralconditions・Tb makemattersworse，thereaLreSOmekaTyithatarenotsufncientlydisaJnbiguatedevenifokurなαna arepresent．Forafu11discussion，SeeHonda（2007）． Nonetheless，thesefactsbynomeaJISreducetheimportanceofokurなロna’scontributiontothe resolutionoftheindeterminacyofmorphemes．Withoutthem，decodingwouldbeentirelydependent OngueSSingand，atbest，infbrence舟omtheorthographicandsyntacticcontexts（Seefbotnote17）・

－97－ knowledgeand checkifthere are suchlexicalitems as＊uke＋tukar（－i），＊uke＋FU， ＊ZYU＋FUandsoforth，befbredetemlnlngthatuke＋tukeistheonlypossibleand existingword・Again，thisisam印Ordrawbackfbrdecoding・

け一触 け一鹿 （14）a． 受 b．受 付 止 c．受 付

〟鹿J〟お 鹿 以鹿 抽鬼e

ItisclearffomtheaboveexamPlesthatthesmallertheamountOfokuriganaemployed fbrencoding，thelargertheamOuntOflexicalknowledgerequiredfordecoding・A recent survey Of70JapaneSe magaZines conducted bythe NationalInstitute fbr JapaneSe Language SuggeStS a generaltendency to use the basicformS，atleastin non－SPeCialist publications（Kokuritsu Kokugo Kenkyq502006），Still，the sporadic OCCurrenCeSOfthevariantSarebynomeanSnegligible．Inviewofthisfact，itwouldbe Safbtosaythatlexicalknowledgeisindispensableforutilisingokuriganafbrdecoding・

5．ConcJuding rernarks This paper has presented a descriptive account ofJapanese okurなana触）m the VleWPOintofdecoding，withparticularreftrencetothecurrentconventionsconcernlng Okurなanaorthography．Previousstudieshavebeenreviewedtooutlinethreecasesof kanfiindeterminacy，andananalysis has been conducted toformulate the way Okurigana canbe saidto resoIvetheseindeterminacies．Some problems caused by OrthographicallypermittedvariantSOfokuriganaapplicationhavebeenalsodiscussed． Fromthediscussionabove，itseemsreasonabletoconcludethatthefbllowlngtyPeS OfprlOrknowledge are requiredto exploit okurigana fbrthe purpose ofdecoding． Furtherresearchisneededtoinvestigatetheutilityofthesetypesofknowledgeinother aspeCtSOftheJapanesewdtingsystem．

＞ LingulStic －Lexicalknowledge，eSpeCiallyaboutphonemicfbrmSandthedistinctionofNJand SJitems；and －Morphological knowledge，eSPeCial1y aboutinnection，derivation and morphophonemicaltemations． ＞ Orthographic －Knowledgeaboutgrapheme－tO－Phonemecorrespondences；and －Knowledgeal）OutOrthographicconstraints，eSpeCiallythoserestrictingtheway Okuriianacanbeappliedtoka吋igraphemes．

ー98－ 【R曲remces】 Abe，Seiya（1989）J6y6ka一再inookurigana．InKiy可iSat6（ed・），KbTditokok喝OmOndbi，65－98・ MetiiShoin． Cabinet（1954）R6m¢inotsuzurikata．CabinetNot浦cationNo．1，December9，1954．Ka7P∂ G6gai，December9edition，1954．Okurash6Insatsukyoku． （1981a）］卸6ka吋i－hy∂．Cabinet notincation No．1，Octoberl，1981．KaTP∂G∂gai， Octoberledition，1981．Okurash6Insatsukyoku． （1981b）Okur卸nano乃ukekata．Cabinetnot捕cationNo．3，Octoberl，1981．Kd甲∂ G6gai，Octoberledition，19＄LOkurash6Insatsukyoku． Coulmas，Florian（2003）析・血g甲Jg椚βJ加加加血cJわ〃わ鋤か肋g血沈1肌ゆ血． CambridgeUniversityPress． Daniels，PeterT．（1991）1sastructuralgraphemicspossible？LACuSFbrum18：528－537． （1994）ReplytoHerrick．LAC【岱Forum21：425N3l． Hayashi，Oki（1977）Gendaikanazukaitookurigana．InSusumuOno＆TbkeshiShibata（eds．）， Kok堀Okoktdimonddi，135－166．IwanamiShoten． Herrick，EarlM．（1994a）Ofcourseastructuralgraphemicsispossible！LACuSFbrum2l：

413－424． （1994b）ReplytoDaniels’sreply．LAC乙∬Fbrum2l：432－440． Hill，ArchibaJdA．（1967）Thetypologyofwritingsystems．IrlWi川amM．Austin（ed．），PqperT i／l［iT7gtJjsticsinhollOl・qfLe（）1tDosterL，り2－99．Mouton． Hoshina，Koichi（1949）KbkzJgOmOn血igQiiinen．San’y6Shuppan． Honda，Keisuke（2007）Homography－induced ambiguityinJapanese kaTUiand thelexical disambigua山ng凡InCtionofβ払r如乃乱Gビ乃g¢脚払助搾ぶ∂26：17－42． Kaiser，Stefhn（1995）Sekainomqii，ChGgokunomqii，Nihonnomqji：Kart3inoichizukesaik6． 蝕血如拙穴Ⅵ血沼gO勒∂肋5：155－167． （2007）Nihongogakunoshomondai：M可inoruikeirontoNihongonohy6ki．Kbktqakuin

肋ぶ鋸10＄（11）：13－27． Kawai，Yoshifumi（1970）Ka頑niokeruyOmitokakinoparadokkusu．KbiTy∂Kokugogaku55：

23－29．

KokuritsuKokugoKenky句0（1952）Okur如ah∂shiTy6．KokuritsuKokugoKenky句0． （2006）Gg乃血fg∬∫鋸〃0卸∂放Jタタイー〝g〝厄舶虐7β－∫鉱KokuritsuK．ok喝OKenky句0． Kond6，rrbkako（2005）Kanjito okurigana．InTbmiyoshiMaeda＆MasaakiNomura（eds．）， RbqiitoNiho脚，146－163．AsakuraShoteI】． K6no，Rokur6（1980／1969）The Chinese writing aJlditsinfluences on the scripts ofthe neighbouringpeoples：WithspeciaIre托rencetoKoreaandJapan．InRokur6K6no，K∂no R（）k7Jr∂ぐhosak7L＝enShG（VbI．3），15－102．Heibonsha．

ー99－ （1994／1977）M可inohonshitsu．InRokur6K6no，〟毎iron，2q24．Sanseid6． Miyaoka，Osahito（2002）‘Go’towananika：Eskim∂TgOkara購hongoomiru・Sanseid6・ Narrog，Heiko（1998）Nihongod6shinokatsuy6taikei．mho7聯）KLqaku4：7－30． Nomura，Masaaki＆KikukoIt6（1978）Kaqiinohy60ndo．KbiTy∂Kok喝Ogakull（7）：306－311・ e）no，％hoko（1958）Okurigananoshomondai．GengoSeikatsu80：16－23． Poser，Wil］iamJ・（1992）Thestructuraltypologyofphonologicalwriting．L吻LgticSbcieo｝qf America：Sixo）－S加hannualmeetinghandbook，46．LinguisticSocietyofAmerica・ Ratcli恥Robert R．（2001）Whatdo一▼phonemic writing systems represent？Arabichuruuf； Japanesekana，andtheMoraicPrinciple．耶・ilienLang7LageandLiieraqy4（1）：トⅠ4・ Rogers，Henry（2005）析■it加g耶iems：Al卸Lsticqproach．BlackwellPublishing・ Saito，Hirofumi，Masahiro Kawakami＆HisashiMasuda（1995）Kartiik6seiniokeru buhin （bushu）－On’intai6．．席h6BunkaKen砂ii2：89－115． Sampson，GeofnTey（1985）WitわgりげtemS．Stan払rdUniversityPress． Satake，Hideo（2006）Ka頑tohy6ki．InTbmiyoshiMaeda＆MasaakiNomura（eds．），助TUino hataraki44－64．AsakuraShoten． Sproat，Richard（2000）Aco〝甲utationaltheoTyqfwriting即∫temS．CambridgeUniversityPress・ Stalph，Jurgen（1989）Grun‘批噂eneinerGrammatikdbrSinQi呼anLschenSthr終Harrassowitz・ Thkebe，Ybshiaki（1979）mhongono妙∂ki．KadokawaShoten． rrhkishima，Hiroshi（1970）Gendaigonoseishoh6nitsuite：“T6y6KaniiKaiteiOn－KunHy6” OyObi“KaiteiOkurigarlanO一触kekata”oc旭shinni．Ge〝gOSeikaLsu228：18－26・ Uchida，Hyakken（1972／1961）D6shinofuhenkagobi：Shin okuriganah6nikanrenshite・In HyakkenUchida，tkhidb砂akkenzenshi2（Vbl．8），552－554．K6dansha． Vhnce，TimothyJ．（2002）The exception that proves theru1e：Ideography andJapanese kun’yomi．ln＝Mary S・Erbaugh（ed，），D炉Cult Characters：htefdLgc＊linaTyStudies ChineseandJ如anese折■iiing，177－193．NationalEastAsianLanguageResourceCenter， OhioStateUniversity．

－100－