《裕固族、东乡族、保安族社会历史调査》，甘肃民族出版社，1987年，第33页。
王治来、周锡娟译《世界境域志》，1983年铅印本，第69页；第120页。
埃捷尼舍夫着，陈鹏译《突厥语言研究导论》，中国社会科学出版社1981年版，第395~396页。
西州使程记》称髙昌回鹘王国“所统有南突厥、北突厥、大众熨、小众熨、样磨、割录、黠戛司、末蛮、格哆族、预龙族之名甚众”。其中的“黠戛司”也便是“黠戛斯”。这一支当即是怀信可汗时被移徙至当地的。
0参见《世界境域志》第13章。另，笔者所撰《《世界境域志）所载史实及其相关问题考辨》一文(待刊)对此有考。
在勒柯克于1904-1907年自喀喇和卓运走的古代回鹘文抄本中，有一份摩尼教《二宗经》(Ikijyltyznon)的残卷。该残卷是于“福年神佑月吉日”在“显赫的得天保佑的(突厥哈鲁克——Tyrkiaruqi)的城市阿尔乌怛逻私(AreuTalasulus)”撰写的。撰写的目的是“使十箭国信教”。据此可断定，该部至晚在8世纪上半叶时还游牧于七河流域。另请参看[苏]OT克利亚什托尔内着，李佩娟译《古代突厥鲁尼文碑铭一中亚细亚史原始文献》，黑龙江教育出版社，1991年，第131-132页及相关注释。
《Tyrkitillatdiwani》卷1，新通人民出版社，1980年，第494页。
TwoIdenticalFeaturesintheHexiialectKaharBaratCenterfortheStudyofWorldReligionsofHarvardUniversityThestudyofMiddleChinesedialectsinfactrevolvesaroundthestudyInnerAsiantextswhichweregeographicallyconnectedtotheChineseNorth-WesternorHexidialect-SocalledTangWtidaixibeifangyin唐五代西北方音（Luo)andsongdaixibeifangyin宋代西北方音（Li)arecomposedoftheChineseWordsrecordedintheTibetanandTangutscripts.TheycallitadialectinordertodistinguishitfromstandardMiddleChinese.ThefundamentalreconstructionofMiddleChinesehasbeenestablishedonthebasisofYinyunxue音韵学，atraditionalChinesephonology.ThedescriptionsinitsprincipaldictionariesQieyun切韵orGuangyun广韵basicallyrepresentthenationsTangperiodstandardlanguage.ThiswasthemajorlanguagereformpredatingPutonghuainChina；acommonvernacularandliterarystandarddevelopedwhichwasbasedonthedialectofChang，an，anditaffectedtomostChinesedialects.IthasbeenatraditioninChinatodiscriminateagainstthedialectsandimposeastandardlanguageonthewholenation.Similartothesituationtoday，variousdialectsmusthavealwaysexistedthroughoutthehistory，butthelinguistictraditionalwaysstoodforthedominantcorrect”languagespokenintheNortherncapitalcitiesChanganandLuoyang.Wereadthemasiftherewasonlyonelanguageinhistory.Therefore，resultsofmodemreconstructionhavenotyetgonefarbeyondthisprincipalframe.Lately，however，LiFanggui，CoblinandothershavestartedtopayattentiontotheinconsistenciesintheGuangyunandotherrecords.Theyrecognizedthedialecticaldifferenceswererelatedtothehometownsofthewriters-Thisisasophisticatedandsignificantnewwayofresearch-AncientChineseisawrittenlanguage.Byitsnature，Chinesecharactersdonotdescribesound，itmustbeseparatelydescribed.MosthistoricaldictionariesarecompiledwithanadditionalspellingmethodofFanqie反切，whichdoprovideinformationonpronunciationaswereadLatinscriptsinthePinyin.Chinesecharactersarelpgographic；everydialectisreadinitsownway，buttheydonotreflectontheshapeofcharacters《Indeed，thereareothersources，suchasmodemdialecticalmaterials；Chineseloan-wordsinKorean，JapaneseandVietnamese；andtheChinesetranscriptionofBuddhistSanskritterms.TheyhavebeenoftenusedassupportingmaterialsforthereconstructionofthestandardlanguageSinologylacksthemostimportantevidenceforthereconstructionofthesoundsystemoftheChineselanguage-analphabeticrecordofancientChinese-beforethestageMiddleChinese.uringtheNorth-SouthdivisionandtheTang，theChinesecharactersinvadedthesouthernforesthillsuptoVietnam；theypassedthemountainsandspreadtoKorea；theycrossedtheseaandreachedJapan；buttheyjustcouldnotjumpovertheGreatWallandresideamongthenomads.Bythattime，thelanguagesofthesteppeandInnerAsiaWerealreadybenefitingfromtheirconvenientalphabeticwritingsoriginatingfromtheancientPhoenicianalphabet.TheywerecalledHushu胡书（barbarianscripts)inChina，mostofthetimereferringtotheSog-dianscript-TheChinesetermsbegantooccurinthesetextsandincreasedgraduallyastheydeveloped.WeseeafewChineseloan-wordsintheSogdianAncientLettersof312-313A--Gradually，theInnerAsianpeoplelearnedtowritetheirownWordsinalphabeticscripts.丁heearliestdatableOrkhonTurkicKolTeginInscriptionwaserectedin732，andtheTibetantextistheZholInscriptionof763.Inthe10thcentury，writtenliteraturereacheditspeakinthroughoutInnerAsia，includingmaterialsinTibetan，Uygur，Khotanese，SogdianandTokharianTibetologistsfirstdrewattentiontoanoticeablenumberofChinesewordsfromthebilingualSinoTibetaninscriptions-TheywereimmediatelyusedasvaluableevidenceintheMiddleChinesecomparativestudiesAcollectionof818TibetanChineseglosseswaspublishedbyCsongorinI960.Thelastdecadehasbeenaharvestseason-SeveralinformativearticlesonMiddleChineseglosseswrittenin”barbarian”scriptshavebeenpublishedoneafteranotherIn1986，Shogaitocollected500Chinesecharactersandin1995，hepublishedanother292charactersfromUygurtexts.In1993，EmmerickPulleyblankpublishedapartialKhotanesetranscriptionofChineseVajracchedika(T2732)thataloneprovided419glosses.In1994，Yoshidacollected330ChineseglossesfromSogdiantexts.Ihavefound230glossesfromtheUygurXuanzangBiographyIXandX.InthewholeoftheUygurXuanzangBiography，weestimatetherearemorethan500glosses.Mostofthetextsbelongtotheperiodfromthe8thtothe10thcenturies.SeveralthousandChineseglossescouldbecollectedfromthesetextsandtheyhavethepotentialbringingtheMiddleChinesestudyintoanewera-Sincenon-ChineselanguageswerenotobligedtofollowChinasstandardlanguage，theloansweretakendirectlyfromactualvernacularsounds.IncomparisonwiththeGuangyunforthestandardlanguage，theseglossesaremadeforthedialects-Amoredirectandcomfortablemethodofdialectalstudybecamepossiblewiththediscoveryofthe”barbarian”Chineseglosses.