南雄方言与客家话

方言2001年第2期“2—1511趸(2001年5月24日出版于北京) NANXIONGANDHAKKA(南雄方言与客家话) L．SAGART沙加尔 (法国巴黎高等社会科学院东亚语言研究所) AbstractAlthoughHakkaandNanxiongdifferintheirmodeofdevoicing，thedialectof Nanxiongcitvhasasplittreatmentoftheentirezhuoshangcategoryremarkablysimilar toHakkainitslexicalincidence．Thissuggeststhattheysharearecentcommonances’ tor，fromwhichthezhuoshangsplitwasinherited．Itisarguedthatthesetofzhu- osh彻gwordswhichhavetone1 instandardHakkaandtone1 or2inNanxionghad tone4intheparentlanguage．Itisalsoarguedthatthiscommonancestorhadnotyet mergeditsa“anzhuoinitialswiththevoicelessaspiratedinitials．Theancestraltone4 thenmergedwithtone1 inHakka，andwithtone1 or2 inNanxiong；anddevoicing occurredseparatelyinNanxiongandHakka． KeywordsNanxiongdialect，Hakka，devoicing，tones 0．Introduction TheLanguageAtlasofChina(AustralianAcademyoftheHumanitiesandChineseA。 cademvofSocialSciences，1987，maPB13)characterizesthedialectofNanxiong南雄cityin northernGuangdongaspartoftheSO—called“Shaozhou”韶州土话，asetofChinesedialects awaitingclassification．Someinvestigatorsnowrefertothesedialectsas“northGuangdong patois”粤北土话．Inthispaper，itwillbearguedthatHakkaandNanxiongareveryclosely related． WhilethecityofNanxiongitselfspeaksNanxiongdialect，thecountrysidearoundit speaksmostlyHakka．ItappearsthatthecitydialectisolderintheregionthanHakka：ex— DansionofHakkaspeakersintheNanxiongregiontookplaceonlyafterlateMing(Leong 1997：58)．TheNanxiongdialectwasprobablyalreadyspokeninandaroundNanxiongcity beforeLateMing． TherearetwopreviousdescriptionsofthesoundsystemofthedialectofNanxiong citv：S．Egerod’sarticleof1983，inwhichheconcludesthatNanxiongiscloselyrelatedto Min(aconclusionverydifferentfrommyown)；andaseconddescriptionbyProf．XieZili 谢自立publishedinthedialectsectionoftheNanxiongXianZhi(NanxiongXianDifangZhi BianzuanWeiyuanhui1991：770—779)．Thesearetwooutstandingpiecesofdescriptivework， buttherearesomedifferencesbetweenthem，bothinthephoneticsandthephonology．One differenceconcernstonesandhi．AccordingtoEgerod(1983：126)bothprogressiveandregres一 ·142· 方 言 万方数据 sivesandhiarecommoninNanxiong．Xie(2000)disputesthis，showingthatwhatEgerod tookfortonesandhirepresentsakindofdiminutive(小称)morphologyoftheStandardCan— tonesepin—yam变音type，overlookedbyEgerod．IagreewithXiethatNanxionghasdiminu— tivemorphologyandthatEgerodmistookitfortonesandhi．Healsomistooksomeinstances ofalternationsbetweencolloquialandliteraryformsforinstancesoftonesandhi，as1 will showbelow．Althoughtruetonesandhidoesexistinmydata，itisnotveryspectacular． IhadthegoodfortuneofspendingtwoweeksinNanxiongcity④inMay—June1999． ThereIconductedintensivefieldworkonthecitydialect．Afterfamiliarizingmyselfwiththe soundsystem，guidedbythefindingsofEgerodandXie，Irecordedabout13minutesofim— proviseddiscourseonvarioussubjects．Thedataweretranscribedintosystematicphonetic transcriptionandthepronunciationofeachwordcheckedwiththeinformant．Thesedata weresupplementedwitha200一wordSwadeshlist．Intheinitialstagesofmyinvestigation， I alsocollectedcharacterreadingsforthecharacterslistsonpagesx-xiioftheFangyan DiaochaZibiao方言调查字，soastoobtainsamplesofthemainphonologicalcontrasts． Thesecharacterreadingsmayinsomecasescorrespondtonoactualmorphemeintheliving Nanxiongdialect．Thepresentpaperis basedonallthesematerials，wordselicitedfrom wordlistsorextractedfrominviteddiscourse，andcharacterreadings． 1．ThezhuoshangsplitinNanxiong InEgerod’sdescriptionofNanxiong，thecizhuoshangshengcategorywassplitbetween twomainsets：tone4andtone3．Similarly，inmanyHakkadialects，thecizhuoshangsheng categoryissplitbetweentone1 andtone3．Amongcizhuoshangshengwords，therewasa correspondencebetweenHakkatone1 andNanxiongtone4：马买咬，andbetweenHakka tone3andNanxiongtone3：耳，两两个，卵，瓦，五． Thiswassuggestiveofagoodagreementinthesplitofthecizhuoshangshengcategory inthetwodialects，assilentlynotedinSagart(1998)．Unforunately，Egerod’sarticledoesnot containenoughdatatocovertheentirecizhuoshangshengcategory．Moreover．thequanzhuo shangshengcategoryisalsosplitinbothdialects：itwasnecessarytoexaminewhetherthere isagreementtheretoo．ThefollowingaremyownobservationsonNanxiong，basedonthe fieldworkconductedtherein1999．1willlimitmyselftofactsbearingoninitialsandtones． Unlikein(southern)GanandHakka，theMCquanzhuo全浊(voicedobstruent)initials aredevoicedintovoicelessunaspiratedobstruentsinNanxiong，inalltones．Therearesix phonemictones．Inisolatedsyllables，theywereperceivedasfollows： ① I amverygratefultoProf．LinLifang林立芳ofShaoguanUniversityforhishelpinarrangingmy stayinNanxiong；totheCommunistPartycadresandMunicipalitycadresofNanxiongcity，especiallyMr．Li Shiren李仕仁andMr．HeJian何坚fortheirwarnlwelcomeandtheirsupportduringmystay；andtoallthe personneloftheNanxiongMunicipalGuesthouse南雄市政府招待所formakingmystayaspleasantasitwas fruitful．Lastbutnotleast，mythanksgotoMr．XiaoRongiin肖荣金，myinformant．Hewasthena60一year oldretiredschoolteacher．AnativeofNanxiong，hecouldalsospeakfluentPutonghuaandHakka． 2001年第2期 ·143· 万方数据 Tone1(Yinping，“upperLevel”)islong，[1]44or[1】45incontour．Voicequalityis modaloverall．Vocalfoldvibrationendsgradually． Tone2(Yangping，“lowelLevel”)is[J]21 incontour．Itsvoicequalityappearstocorn— bineelementsofcreakinessandbreathiness． Tone3(Yinshang，“upperRising”)is[1]35or[卅】325incontour．Voicequalityis modal． Tone5(Yinqu，“upperDeparting”)islong，[q]32incontour． Tone6(Yangqu／Yangru，“lowerDeparting／Entering”)isshortish，[、1】53incontour．Itis creakytowardstheend，andvocalfoldvibrationendsneatly． Tone7(Yinru，“upperEntering”)isshort，high[11_5，andendsina glottalstop． Inadditionthereisalighttone(ortone0)，similartotheqingshengofStandardChi— nese．Innaturalspeechsomesyllablesarealwaysunstressed，andtheyarelexicallytoneless． Manyofthemaresuffixesandgrammaticalmorphemes．Theyalwaysfollowa tonalsyllable． Thepitchoftonelesssyllablesdependsonthetoneoftheprecedingtonalsyllable． Theprincipalinstancesofcontextualtonesandhioccurinclosejuncture．Theyareas follows： ①atone1 precedingatone2isalwaysslightlyrising：[1]445insteadof[1】44；thisis reminiscentofHakka(forinstanceSungHimTongKakka，Sagart1982)； ②thefirstoftwoconsecutiveT2islevel：【a]22insteadof【J】21； ③atone2precedinga tone3 alsobecomeslevel，butlower：[．|】11insteadof[J]21； ④andfinally，thefirstoftwoconsecutiveT3isslightlyhigherthanthesecond． Therearesomedifferencesbetweenmyobservationsandthoseofpreviousinvestigators， asshowninTable1： Table1：tonecategoriesandcontoursinNanxionaccordingtodifferentinvestigators． Tone Egerod(1983)XieZili(1991)Sagart(thispaper) T1(Yinping)[]】 55 [1】 44 [1】44。[1】445 T2(Yangping)[J】 21 [J】 1l [J]21 T3(Yinshang)[卅] 24 [／j】24(224)[1】 35，[卅]325 T4(Yangshang)[，J] 12 ==Yangping==Yangping T5(Yinqu)[-j】 11 [J】 22 [_】 33(332) T6(Yangqu)’[、J]42 [、|】42 【1】 53 T7(Yinru)[]】 55“ [1】 452 [1】 52 T8(Yangru)【-|】 222 【1】 432 ==Yangqu Notethatthereisnoindependenttone4(Yangshang)inmydata，unlikeinEgerod’S． Egerods’Stones2(10wfalling[J】21)and4(10wrising[J】12)aremergedintotone2([J]21in mydata．Inthis，XieZiliandI agree． Table2presentsthecorrespondencesbetweenMiddleChineseandNanxiongthatcanbe extractedfrommyowndata． ．144． 方 言 万方数据 Table2：developmentofinitialsandtonesinNanxiongcity． MCinitial 平A 上B 去C 入D 清P P—l P—j P——5 P——7 次清P‘ P‘一1 P‘一3 P‘一5 P‘一7 全浊b P——2 P——2 P一1 P——6 P——6 p——6 次浊m m一2 m一2 (m一1)m—j m一6 m一6 ThetreatmentofthezhuoshangcategoryinNanxionginvitesparticularattention： ①Onesetofwords(henceforth：“set1A”)hastone2inmydataandinXie’S，andtone 4inEgerod’S．Thisisregardlessofinitialtype：theMCinitialsaresonorants(cizhuo)and voicedobstruents(quanzhuo)．Thesewordsareallcolloquial．SeeTable3． ②Asecondsetofwords(set1B)hastone1 inmydataandinXie’S．InEgerod’Sarti— clethesewordsalsohavetone1 phonetically，butEgerodtreatsthemastone6underlyingly． Thissetisverylimitednumericallybuttheformsareverybasicandcolloquial．SeeTable4． ③Alastsetofwords(“set2”)hasmodernreflexesconditionedbyinitialtype：tone6 withMiddleChineseobstruents，andtone3 withMiddleChinesesonorantsinallrecorded varietiesofNanxiong．Someofthesewordsarecolloquialandothersliterary．SeeTable5． ThedatapresentedinXie’SdescriptioncorrespondcloselywithEgerod’Sandmine∞ 2．comparisonwithHakka WenowcomparethedevelopmentoftheentirezhuoshangcategoryinNanxiongand Hakka．OurHakkadataaredrawnfromMcIver’Sdictionary(McIver1926)，Toconserve space1 willonlylistEgerod’SdataandmineinthetablesbelowfNanxiongformselicitedas characterreadingsareinplaintype；formselicitedthroughaword—listorextractedfromdis— courseareinboldtype)． ItisnecessarytodiscussbrieflyEgerod’StreatmentoftheformsinTable4．Egerodre— gards上，下and妇aslexicallytone一6forms，changedtotone1 throughtonesandhi．Howev— erneitherXienorI couldobservethissandhiinNanxiong．Theexistenceofthissandhiis veryproblematic．First，tone一6formscomingfromMiddleChinesequshengsyllablesare neveraffectedbyit．Second，theconditioningfactorforthesandhiisnotclearlystatable： EgerodcitedexamplesofT6>T1occurringbeforea T0(ha6>1pooi2‘hascomedown’， sour36>1pooIm2‘hascomeup’：P．141)；beforea T1(SOUl36>1t9iml‘forenoonmeal’)；beforea T2(fu6>1pir32‘woman’，ha6>1m2‘afternoon’)；buttherearemanycasesinhisdatawherethe sameT6doesnotchangetoT1inthesameenvironments．Apreferableanalysisisthatthere isthatthereisnoT6>T1sandhiinNanxiongandthattheformsinTable4 arelexically intone1．Itistruethatboth上and下havetone一6variantsinNanxiongbuttheyinvari一 ① Althoughinhisdescriptionof1991(NanxiongXiandifangZhiBianzuan onlyrecognizedtones3 and6astheregularreflexesofthecizhuocategory，hedid istenceofa tone一2reflexina laterpaper(Xie2000)． 2001年第2期 Weiyuanhui1991：774)Xie explicitlyrecognizetheex一 · 145· 万方数据 ablyOccurinwordsofnorthernorigin，suchas上午‘morning’，晚上‘evening’．Thusallforms showingtheallegedT6>TIsandhiinEgerod’SdataareinfactlexicallyT1words． Table3：zhuoshangwordsfromset1AinNanxiongandHakka Nanxiong(Egerod1983)Nanxiong(Sagart)Hakka(McIver1926) 淡 t6d2 thaml 坐 tso‘ tsh01 柱 tsu2 chhul 重heavy tsemrf tsorj2 chhun91 近 tcjmr34 tCjmr32 khiunl 衙 tGi2 khil 下屋下 ha2 khal，hal 买 moa4 mWQ。 mail 马 ma4 mal 暖 DU矿 n0121 奶milk nai2 nail 懒 16d2 lanl 冷 lar32 lan91 鲤carp li2 1i1 有have i04 j02 yul 野 ia4 jQ2 yal 养raise(cattle) jor32 yon91 咬 10au4 日Q旷 ngaul Table4：zhuoshangwordsfromset1BinNanxiongandHakka Nanxiong(Egerod1983)Nanxiong(Sagart)Hakka(McIver1926) 上 sou习6>1 对(～昼“lunch”)son91 下 10>1 bal(迭～来“fall”)hal 社～官 sal shal 厚 heiI heu5 妇～媲“wife” fll6>1 ful fu5 尾 1221111(～巴“tail”)muil，mil TwoformsinTable4havetone5inMcIver’S tainotherHakkadialects(LiandZhang1992)． 10ne5：fu5in妇人‘woman’，buttoneI：pulor this，NanxiongagreesperfectlywithHakka． ·146· dictionary．However，theyhavetone1 incer— Concerning妇，Hakkadialectsgenerallyshow fulin妇娘(seeLiandZhang1992：325)．In 方 言 万方数据 Table5：zhuoshangwordsfromset2inNanxiongandHakka Nanxiong(Egerod1983)Nanxiong(Sagart)Hakka(McIver1926) 倍 pm6 phui5 父 fu6(in父母)fu5 兆兆头 t8Qw-6 chhau5 件 t硝量6 khen5 米 mi3 mi3 mi3 网 mol33 mion93 五 m? m? n93 老 lau3 lQ矿 lau3 佬peI∞。 lau4 lQ_舻 lau3 出拿moumain 1j寸 lian91，lian92 卵malegenitals 1时(～t06) lon3 两两个人 loourj3 ljorj3 lion93 脑 nQW3 1 naU。 雨 Y3 jy3 yi3 女 n3 ny3 nyi3 巨 J1iJ． 一(～t01) nyi3 绕 paw3 ' nyau" 远 iorj3 ju旬3 ven3 FinallywelistinTable6thosewordswithother(irregular)tonecorrespondences Table6：zhuoshangwordswithirregularcorrespondences Nanxiong(Egerod1983)Nanxiong(Sagart)Hakka(McIver1926) 抱 paw2 phau5 动 t矿(in动物) thun91 母 meis(in父母)mel，mil 猛fieree mar33 m时 man91 每 mUl3 nluil 以 ji3(in以前) yil 眼eye 科 习6d2 ngan3 TheT5readingfor抱inMcIverisexceptionalinHakka．AlltheHakkadialectslisted inLiandZhang(1992：358)showtone1，correspondingregularlywithNanxiongT2．Itmay bethatMcIverislistinga literaryreading，asthecolloquialwordfor(tocarryinone’S arms)inMeixianinnam①．TheT6readingfor动inNanxiongisevidentlypartofamodern literaryword(animal)，while动inHakkaisaverbmeaningtomove．Theyareineffectdif． ①Xiegivestone2forthisword：rjoan2．Heandl areinagreementagainstEgerod． 2001年第2期 ·147· 万方数据 ferentwords．每and以aregrammaticalmorphemespresentinclassicalChinese。itmaybe thatthetone一3readingsforthesewordsinNanxiongareliterary．Theclearestandmostse— riousdivergencesbetweenNanxiongandHakkaareinthewordsformother，fierceandeye． Fromthedatapresentedabove，wecanseethat，witha smallnumberofexceptions，the correspondencesbetweenHakkaandNanxiongareveryregular-thosewordswhichareinsets 1Aand1BinNanxiongareintone1 inMcIver’SHakka；andthosewordswhichareinset 2inNanxionghavetones3(sonorantinitials)and5(obstruentinitials)inHakka． Table7： correspondencesofzhHosh口ngwordsinNanxiongandHakka Nanxiong(Egerod)Nanxiong(Sagart，Xie)Hakka(Mclver1926) set1A tone4 tone2 tone1 set1B tone1 tone1 tone1 set2 Son． tone3 tone3 tone3 obst． tone6 tone6 tone5 3．discussion Thefirstpointtoexamineiswhetherthereisanyconditionforthesplittreatmentof thezhuoshangcategoryinNanxiong． Lookingattables3and4，weobservethatthequanzhuowordsinset1Aaremostly stopsandaffricates，whilethoseinset1Bareallfricatives．NotethatNanxiong下ha2(in屋 下‘home’)cantheoreticallyhavedevelopedoutofanearlier。kha，sincekh—sporadically changestoh—inNanxiong．IfSO，wecanmakethegeneralizationthatinset1，MiddleChi— nesequanzhuoshangshengwordsgotoNanxiongtone2afterstopsandaffricates，andto tone1 afterfricatives．UnfortunatelywecannotbecertainthatNanxiong下ha2reallycomes fromanearlier+kha．Moreoverwecanfindnotphoneticexplanationforthepresenceofa cizhuoword：尾(tail)inset1B． Thedistinctionbetweensets1 and2islesssimple．Insomecasesetymologicaldoublets exist：theindigenous／colloquialformistheninvariablyinset1， andthenorthern／literary forminset2．Examples： Table8：doubletsinsets1and2inNanxiong Nanxiong(coll．) Nanxiong(1it．) 有 j02‘tohave’ jim3(in所有‘all’) 养 j。矛‘tofeed’animals jara3(in培养‘tofoster，totrain’) 社 sa2(in社官‘godofthedomesticaltars’)sa6(in社会‘society’) Whileallwordsinset1(1Aand1B)arecolloquial，notallwordsinset2areliterary． Especiallywithsonorantinitials，set2includesverycolloquialwordswhicharecertainlynot northernloanwords，likeNanxiong佬law3(person)，卵lur33(malegenitals)，atabooword．The tone——3reflexinthesewordscannotbepredictedonthebasisofanyobservablephonological feature．Ithasbeenproposed(O’Connor1976，Norman1986)，thatinHakkathesewords ·148· 方 言 万方数据 oncehadvoicelesssonorantinitials’hm一，+hn一，。h日一，+hl一，。hj—etc．：accordingtotheseau— thors，thiscausedthemtobehavetonallyliketheqing(voiceless)initials，andexplainswhy theyhavetheupper—seriestone3．Insupportofthisidea，notethatStandardTaihasha< ’hrjafor(five)(anearlyloanfromChinese)，correspondingtoNanxiongandHakkam3(five) (set2)． Tosummarize，thedistributionofzhuoshangformsbetweensets1 and2 inNanxiong andHakkaappearsdueto(a)borrowingfromNorthernChinese，plus(b)anotherfactor， perhapsphonological：voicelesssonorantinitialsor(asproposedinSagart1999)voicelesspre— fixes． 4．NanxiongandHakkashareacommonancestor Nanxiongisnota Hakkadialect：itsspeakersregardthemselvesasnon—Hakka；its quanzhuoinitialsarenotaspirated；itdoesnotusea cognateofHakkamak 7 kai5for “what”；doesnotusea cognateofoilfor“mother”；etc．；andyet，itbehavesexactlylikea Hakkadialectfromthepointofviewofthezhuoshangsplit．Thesplititself，aswehave seen，iscomplex：itresultsfromborrowingandphonologicalconditioning．Thedetailedcorre— spondencebetweenHakkaandNanxiongcannotbeexplainedasacoincidence：neithercanit beexplainedastheresultofcontactandborrowingbetweenHakkaandNanxiong．Itisin theoryconceivablethatNanxiongandHakkacouldhavearrivedatsimilarresultsbyborrow- ingseparatelyfromnorthernChinese：butaswehaveseen，thiswouldnotaccountforthe verycolloquialwordsinset2inbothNanxiongandHakka．And，overall，thecorrespondence istoogoodandtoodetailed．theonlyplausibleexplanationisgenetic：NanxiongandHakka shareacommonancestor，whichalreadydistinguishedbetweensetsl and2． 5．Ahistoricalaccountintheformofanarrative Wehavealreadydiscussedtheoriginofthedistinctionbetweensets1 and2．These questionsnOWremaintobeanswered： ①whydoesset1 gototones1 and2inNanxiongbuttotone1 inHakka? ②whydotheMiddleChinesevoicedstopsandaffricatesbecomeaspiratedinHakka butunaspiratedinNanxiong? Toanswerthesequestions，wepresenthereahistoricalexplanationintheformofanar· rative，intermsofthehypothesispresentedinthelastsection．thedistinctionbetweensets1 and2inNanxiongandHakkaisinheritedfromacommonancestor． Somecenturiesago，anearlyformofHakkawasspokeninsouthJiangxi．Thatearly formofHakkahad： ③threecontrastingrowsofstopsandaffricates，thesameasMiddleChinese：plain voiceless，voicelessaspirated，anda thirdserieswhichwas(probably)voicelesswithbreathy releasepfitfikfitsfitffit口fi，and： ④anindependenttone4(Yangshang)whichreceived neseshangshengwordswithvoicedinitials(sonorants，voiced 2001年第2期 some--butnotall—MiddleChi— stops，voicedaffricatesandvoiced ·149· 万方数据 fricatives)．Set1 inNanxiongandHakkahasitsorigininthetone4ofthisearlvHakka language①． Othershangshengwordswithvoicedinitialsoccurredintones3and6．They∞nsisted of(a)indigenouswordswithsonorantinitialsintone3，(b)northernborrowingswithSono— rantinitialsintone3，and(c)northernborrowingswithobstruentinitialsintone6． Asoundchangeturningtone4intotone1 inwordswithvoicedfricativeinitialsthen occurredinthisearlyHakka．Afterthischangehadoccurred，agroupofpeoplespeakinga varietyofthisearlyHakkamigrated，presumablythoughthe梅岭pass，totheNanxiongarea， bringingwiththemtheirlanguage．Theirmigrationprotectedthelanguagefromfurther changesthatweretakingplaceinHakkaofsouthJiangxi：afterthemigrationoftheNanx． 10ngpeople，Hakkacontinuedchangingtone4totonel，expandingthescopeofthechange toincludeallremaininginitialtypes：sonorants，voicedstopsandvoicedaffricates．This changedidnotoccurinNanxiong②，whichkepta separatetone4 untilveryrecently．Note thatatleastoneHakkadialectofsouth-westernJiangxi：Shangyou上犹，maintainsandinde． pendenttone4(Yah1986)③． AsecondmajorchangeoccurredinHakkaaftertheNanxiongmigration：thethirdseries ofstopsandaffricates(voicelesswithbreathyrelease)mergedwiththecorrespondingvoiceless aspirates．ThischangeisverygeneralinHakkaandisconsidereda verybasiccharacteristic ofthedialect．Certainauthorsbelievethatit isveryoldandthatitoccurredinNorthem ChinaevenbeforetheancestorsoftheHakkamigratedtosouthChina．However。Ihave shownina recentpaper(Sagart，inpress)thatHakkamusthavemergeditsoldvoicedini． rialswiththevoicelessaspiratesatamuchlaterdate，whileitsspeakerswereincontact withthespeakersofShe畲，aMiao-YaolanguageoftheEastJiangxi．WestFujian-North Guangdongarea．ThisisbecauseSheisaloneamongMiao-Yaolanguagesinshowingthe Hakkatypeofdevoicing：evidentlydevoicingcouldnothavespreadtoSheifitwasnottak． ingplaceinHakka．Inotherwords，theMiddleChinesevoicedstopsandaffricatesbecame voicelessaspiratedinHakkawhentheHakkaswereincontactwiththeShe，thatis。when theywerealreadyinsouthernJiangxiandWesternFujian． Eventually，thethirdseriesofstopsandaffricates(voicelesswithbreathyrelease)were alsolostinNanxiong：butinNanxiongtheymergedwiththecorrespondingvoicelessunaspi． ratedstopsandaffricates，acommonpatterninNorthernGuangdong．Between1550and1850 HakkamigrantsfromMeixianestablishedthemselvesinthecountryaroundNanxiong．Final． 1y，inthecourseofthe20恤century，tone4mergedwithtone2inNanxiong． ① 丁heideathatanearliertone ②TheNanxiongmigrationmay tosetI inNanxiong． 4istheoriginofset1 inHakkawas haveoccultedwhenthischangewasin proposedinSagart(1993：178)． itsearlystages。since尾belongs ④ AlsosomesouthernGandialects：Anfu安福，Lianhua莲花andSuichuan遂J ·150· 方 言 万方数据 references [1】AustralianAcademyoftheHumanitiesandChineseAcademyofSocialSciences(1987)LanguageAtlas ofChina．Hongkong：Longman(PacificLinguisticsseriesC，N。102)． 【2】Branner，D．P．(2000)ProblemsinComparativeChinesedialectology．TheClassificationofMiinand Hakka．BerlinandNewYork：MoutondeGruyter． [3】Egerod，S．(1983)TheNanxiongDialect．Fangyan1983，2：123—42． 【4】Hashimoto，M．J．(1973)TheHakkadialect．Cambriage：CambridgeUniversityPress． 【5】HuangXuezhen(1988)Kejiafangyanshengdiaodetedian．Fangyan1988，4：241—6． 【6]Leong，S．T．(1997)MigrationandethnicityinChinesehistory．Stanford：StanfordUniversityPress． 【7】LiRulongandZhangShuangqing(1992)Ke—Gan知ngyandiaochabaogao．Xiamen：XiamenUniversity PublishingHouse． 【8】LoHsiang-lin(1933)Kejiayanjiudaolun．Xingning：XishanShucang． [9]LuoChangpei(1940)Li