中原www.diediao.com

HALId:halshs-00537990https://halshs.
archives-ouvertes.
fr/halshs-00537990v2Submittedon2Jan2012HALisamulti-disciplinaryopenaccessarchiveforthedepositanddisseminationofsci-entificresearchdocuments,whethertheyarepub-lishedornot.
ThedocumentsmaycomefromteachingandresearchinstitutionsinFranceorabroad,orfrompublicorprivateresearchcenters.
L'archiveouvertepluridisciplinaireHAL,estdestinéeaudéptetàladiffusiondedocumentsscientifiquesdeniveaurecherche,publiésounon,émanantdesétablissementsd'enseignementetderecherchefranaisouétrangers,deslaboratoirespublicsouprivés.
ApproachingthehistoricalphonologyofthreehighlyerodedSino-Tibetanlanguages:Naxi,NaandLazeGuillaumeJacques,AlexisMichaudTocitethisversion:GuillaumeJacques,AlexisMichaud.
ApproachingthehistoricalphonologyofthreehighlyerodedSino-Tibetanlanguages:Naxi,NaandLaze.
Diachronica,Netherlands:JohnBenjamins,2011,28(4),pp.
468-498.
10.
1075/dia.
28.
4.
02jac.
halshs-00537990v2Authormanuscript.
Publishedin:Diachronica28:4(2011),pp.
468-498.
ApproachingthehistoricalphonologyofthreehighlyerodedSino-Tibetanlanguages:Naxi,NaandLaze.
GuillaumeJACQUES*&AlexisMICHAUD***CRLAO-CNRS**LACITO-CNRSrgyalrongskad@gmail.
comalexis.
michaud@vjf.
cnrs.
frSummary:Naxi,NaandLazearethreelanguageswhosepositionwithinSino-Tibetaniscontroversial.
Weproposethattheselanguagesaredescendedfromacommonancestor("proto-Naish").
Unlikeconservativelanguagesofthefamily,suchasRgyalrongandTibetan,whichhaveconsonantclustersandfinalconsonants,Naxi,NaandLazeshareasimplesyllabicstructure(consonant+glide+vowel+tone)duetophonologicalerosion.
Thisraisestheissueofhowtheregularphonologicalcorrespondencesbetweenthesethreelanguagesshouldbeinterpreted,andwhichphonologicalstructureshouldbereconstructedforproto-Naish.
Theregularitiesrevealedbythecomparisonofthethreelanguagesareinterpretedinlightofpotentiallycognateformsinconservativelanguages.
Thiscomparisonbringsoutnumerouscasesofphoneticconditioningofthevowelbytheplaceofarticulationofaprecedingconsonantorconsonantcluster.
Overall,thesefindingswarrantarelativelyoptimisticconclusionconcerningthefeasibilityofunravelingthephonologicalhistoryofhighlyerodedlanguagesubgroupswithinSino-Tibetan.
Zusammenfassung:Naxi,NaundLazesindSprachen,derengenaueAnbindunginnerhalbdersino-tibetischenSprachenkontroversdiskutiertwird.
WirschlagenindiesemBeitragvor,dassdiedreiSprachennahverwandtsindundauseinergemeinsamen,vonuns"Proto-Naish"genannten,Ursprachehervorgegangensind.
Andersalskonservativesino-tibetischeSprachen,diewieRgyalrongundTibetischüberKonsonantenclusterundfinaleKonsonantenverfügen,sindNaxi,NaundLazeaufgrundphonologischerErosiondurcheineeinfacheSilbenstrukturausKonsonant,Glide,VokalundTongekennzeichnet.
DieserUmstandwirftdieFrageauf,wiediephonologischenGemeinsamkeitenzwischenNaxi,NaundLazezuinterpretierensindundwassiefürdieRekonstruktionderphonologischenStrukturdesProto-Naishbedeuten.
WirinterpretierendieRegularitten,diesichausdemVergleichderdreiSprachenherauskristallisierthaben,vordemHintergrundmglicherverwandterFormeninkonservativensino-tibetischenSprachen.
DurchdiesesVorgehenergebensichzahlreicheFlle,indenenVokaledurchdenArtikulationsortdesvorangehendenKonsonantenbzw.
Konsonantenclustersphonetischkonditioniertwerden.
DieseErgebnisselasseninsgesamtdieoptimistischeSchlussfolgerungzu,dasswesentlicheFortschritteinderRekonstruktionderphonologischenEntwicklungvonselbsthochgradigerodiertenSprachgruppeninnerhalbdersino-tibetischenFamiliemglichsind.
Résumé:Lenaxi,lenaetlelazésonttroislanguesdontlapositionpréciseauseindusino-tibétaindemeureunsujetdecontroverse.
Nousdéfendonsl'hypothèseselonlaquelleellespartagentunancêtrecommun,le"proto-naish".
Aladifférencedelanguesconservatricestellesquelergyalrongetletibétain,quipossèdentdesgroupesdeconsonnesàl'initialeetdesconsonnesfinales,naxi,naetlazépartagentunestructuresyllabiquesimple,conséquenced'uneérosionphonologiquepoussée.
L'interprétationdescorrespondancesrégulièresentreceslanguesrequiertlaformulationd'hypothèsesausujetdelastructurephonologiqueduproto-naish.
L'analyse,enpartieguidépardesformespotentiellementapparentéesdansleslanguesconservatrices,faitressortirdenombreuxcasdeconditionnementphonétiquedelavoyelleparlelieud'articulationdel'élémentconsonantiquequilaprécédait.
Cetteétudeillustrelefaitquedesavancéesimportantessontpossiblesdansl'étudedelaphonologiehistoriquedelanguessino-tibétainesmêmetrèsérodées.
Keywords:phonologicalerosion;consonantalconditioningofvowels;syllablestructure;Sino-Tibetan;Naxi;Na;Lazepage2/52Introduction.
HistoricalphonologyinaSino-Tibetancontext:thepitfallsofcomparisonbetweensimpleformsTheaimofthisstudyistoapproachthehistoricalphonologyofthreehighlyerodedSino-Tibetanlanguages:Naxi,NaandLaze.
1"Sino-Tibetanisinstructiveforthechallengesitofferstoclassificationandtotheapplicationofthecomparativemethod.
Forone,thepredominanceofmonosyllabicrootsmakesitmoredifficulttodistinguishinheritancefromsheerchancewhenmonosyllabicwordsarecomparedamongthelanguages.
Thereisjustahigherchanceofpossibleaccidentalsimilaritywhenitcomestocomparedshortforms(.
.
.
).
AnothercomplicationistheextensiveborrowingamonglanguagesintheareawhereSino-Tibetanlanguagesarefound"(CampbellandPoser2008:112).
Thetargetlanguagesofthisstudyareacaseinpoint.
Somewordsareidenticalinthethreelanguages,raisingtheissuewhetherthesecorrespondencesmaynotbeduetoborrowing,e.
g.
(givingtheformsintheorderNaxi:Na:Laze)/gv:gv:gv/2for"back(bodypart)".
Othersaresodifferentthatitisnotobviouswhethertheyarecognateatall,e.
g.
/k:k:tsi/for"star"and/k:k:tsi/for"gallbladder,gall".
TheLazeformsfor"star"and"gall",beingphoneticallydissimilartotheNaxiandNaforms,werecountedamongnon-cognatesbetweenLazeandNaxiinapreliminaryattempttoassessthedegreeofclosenessofLazewithYi/LoloishandNaxi(HuangBufan2009).
Ontheotherhand,thesearchforsystematicsoundcorrespondences,whichconstitutesthebackboneofcomparativework,leadstotheoppositeconclusion,sincethesyllabiccorrespondence/k:k:tsi/betweenNaxi,NaandLazeisillustratedbythreeexamples(thethirdis"tight";Appendix2listsalltheitemsforwhichcognatesetsandreconstructionsareproposed).
Siftingthroughalltheavailablevocabulary(about3,000entries),thenumberofwordsforwhichthereisareasonablycertaincorrespondencebetweenNaxi,NaandLazeisabout700,leavingasideloanwordsandcompounds(see,again,Appendix2).
ThesheermassofregularcorrespondencesleadstotheconclusionthatthesethreelanguagesbelongwithinonesinglebranchofSino-Tibetan,towhichwereferbelowastheNaishbranch.
"Wemayassertorhypothesizeageneticrelationonthebasisof[regularsoundcorrespondences].
Buttheproofofthelinguisticpuddingremainsinthefollow-up,thesystematicexploitation,thefullimplementationofthecomparativemethod,whichalonecandemonstrate,notjustalinguisticgeneticrelationship,butalinguistichistory"(Watkins1990:295).
Investigating1ThesethreelanguagesarespokeninSouthwestChina;backgrounddata,includinggeographiccoordinatesandanbriefreviewofavailablepublications,areprovidedinAppendix1.
2Throughoutthearticle,tonesareindicatedbymeansofChaotone-letters:forH(igh),forM(id),forL(ow),forMH.
page3/52thislinguistichistory–thehistoricalphonologyofNaish–involvesproposingreconstructionsforphoneticcorrespondences.
Againtakingthewordsfor"star"and"gall"asexamples,anumberofdifferenthypothesescanbeadvancedtoaccountforthe/k:k:ts/correspondence,suchasreconstructingadditionalconsonantsorvowels.
Thechoiceamongthewealthofcompetinghypothesesshouldbeguidedbyconsiderationsofplausibilityoftheevolutionsthatoneneedstopostulatefromtheproto-formstothepresent-dayforms.
Thedegreeofplausibilityistobeassessedinphoneticterms,instructuralterms,andalsoinarealterms.
HowdatafromconservativelanguageshelpinterpretcorrespondencesobservedwithintheNaishbranchNaxi,NaandLazeallhaveasimplesyllabicstructure:(C)(G)V+T,whereCisaconsonant,Ganon-glide,Vavowel,andTatone.
Bracketsindicateoptionalconstituents.
ThereareneitherinitialclustersnorfinalconsonantsinanyoftheNaishlanguagesanddialectsstudiedsofar,andgiventhewidecoverageofthesurveysconductedsincetheearlyyearsofthePeople'sRepublicofChina,itisasafeguessthatnonewillcometolightasmorevarietiescomeunderacademicscrutiny.
Itisstandardpracticeinhistoricallinguisticstoturntoconservativelanguageswithinthelanguagefamilyforanalysingthehistoryoferodedlanguages(onthistopic,seeFox1995:57-91andreferencestherein).
Wehypothesisethat,inthecourseoftheirhistory,theNaishlanguageshaveundergoneasimplificationoftheirsyllablestructure,andthatcertaincharacteristicsoftheearliersegmentsconditionedtheevolutionofformsuptothemodernlanguages.
Thishypothesis,whichiscentraltothepresentstudy,makesevidencefromconservativelanguagesespeciallyusefultointerpretthecorrespondencesbetweenNaxi,NaandLaze.
However,unlikeinthecaseoftheLolo-BurmesebranchofSino-Tibetan,wherethestudyofhistoricalphonologycandrawheavilyonaconservativelanguagewithinthebranch(namelyOldBurmese)3,theNaishlanguagesarenotcloselyrelatedtoanyconservativelanguage,sothatpointsofreferencetoanalysenontransparentcorrespondences(suchas/k:k:ts/)havetobesoughtfurtherout.
Table1proposescomparisonswiththreeconservativelanguagesofthefamily.
3AbouttheroleofOldBurmeseinLolo-Burmesereconstruction,seeMatisoff1968'sreviewofBurling1967.
page4/52Table1.
SomecorrespondencesbetweenNa,LazeandNaxi,withpotentialcognatesinRgyalrong4,BurmeseandTibetan.
meaningNaxiNaLazeproto-Naish5RgyalrongBurmeseTibetanstarkktsi*krigrikray—gallbladderkktsi*kri-krt-changediscussedinthenextsection.
Thereafter,the*-icomingfrom*-aunderwentthesame*-i>/-/changeastheother*-i.
1.
5.
2.
Aboutthechangefrom*itoaftersThechangefrom*ito//afterscallsforsomeexplanations.
Itdoesnotconsistinamovementfromafrontarticulation[i]toabackarticulation[]–whichwouldbeasurprisingevolution.
Itisinfactaninstanceofapicalisation–acommonphenomenoninthearea,asnotedbyBaron1974,inparticular–whichresultsphoneticallyinthecombination[sz](inYuen-renChao'snotation:[s]);thevowelinthissyllableistobeanalysedphonemicallyasanallophoneof//,asnotedbyHeJirenandJiangZhuyi(1985:9).
Thechangefrom*ito/v/afterm-mayresultfromasimilarprocesswherebytherhymelosesanyindependentarticulatorytarget,resultinginasyllabicconsonant*m;14thesyllableisthenreinterpretedashavingtherhyme/v/.
Noscenariocanbeofferedatpresentconcerning14Bradleyreferstothisprocessas'rhyme-gobbling':"InvariousLoloishlanguagessomeorallofthenasalsoccurassyllabics.
Inmostsuchcasesthediachronicsourceissyllableswithanasalinitialandahighvowel;sometimesonedialecthasnasalsyllabicswhereothershavenasalsplusahighvowel.
Thiscouldbecalledrhyme-gobbling"(Bradley1989:150;seealsoBjrverud1998:8).
page17/52thedevelopmentsthatleduptothecorrespondence/::i/illustratedinTable1(e.
g.
/k:k:tsi/for"star").
Tibetanloanscorroboratethereconstructionof*iaftersibilants.
Forinstance,thepropernameinYongningNaclearlycomesfromtheTibetannamebKrashis(IPAinterpretation:OldTibetan[pkrais]);whateverthedonordialectofTibetan,thesecondsyllablemusthavebeeneither[i]or[is],andthiswordlaterunderwenttheregularchangefrom*ito//.
151.
6.
Theproto-Naishvocalicsystem:aprovisionalsummaryTheaboveanalysesleadtothereconstructionofthefollowingvowelsystemforproto-Naish:iuoɑThesymbolsand//areproposedasphoneticvaluesfortheentitiesreconstructedas*a,*aC1and*aC2respectively.
Thecontrastbetween//and/ɑ/wasneutralisedaftervelars.
Thisisasomewhatunbalancedsystem,crowdedwithbackvowels.
Thechangesthatoccurredinmostconsonantalcontextscanbeschematisedasfollows:ivuoɑAchainshiftoccurredamongbackvowels,reducingthefour-wayheightcontrasttoathree-ortwo-waycontrastdependingonthelanguage,whileaheightcontrastbetween/i/and/e/wascreatedforfrontvowels.
15Inturn,thisopensintothequestionofthetimeframeofthissoundchange;thedistinctionamonglayersofTibetanloansremainsaquestionforfutureresearch.
page18/522.
OnsetsTheavailableevidencestronglysuggeststhatproto-Naishhadalreadylostallcodas;ontheotherhand,theevidenceforinitialclusterscanbeconsideredtobestrong,aswasexemplifiedintheIntroductionthroughtheexampleofthecluster*kr-.
Itismorecommonforalanguagetoretainfinalstopswhereasinitialclustersarelost:examplesintheSino-TibetandomainincludeLhasaTibetan,whichretainsfinal-pandnasal-m-n-whereasithaslostallconsonantclusters,andCantonese,whichalsoretainsfinal-tand-k.
Thereversesituationisattested,however:thereexistSino-Tibetanlanguagesthatpreservesomeinitialclusterswhereastheyhavelostallfinalconsonants,suchastheTibetandialectofZhongu(Sun2003).
Thissectionsetsoutahistoricalscenariooftheevolutionofinitialclustersfromproto-Naishtothemodernlanguages.
2.
1.
StopconsonantsThebasiccorrespondencesforvoicingfeaturesacrossthethreelanguagesarestraightforward.
Table10providesexamplesforlabials;similarexamplescanbefoundfordentalandvelarstopsandaffricates.
Table10.
Examplesofcorrespondencesamonginitials,andproposedproto-Naishreconstructions.
Nstandsforanasalpre-initial.
meaningRgyalrongTibetanNaxiNaLazecorresp.
proto-Naishwhite16wrum—ppvpvp:p:p*p(*pru)topluck——ppvpvp:p:p*p(*pru)yakqa-mbrmbribbvbvb:b:b*bvillage——mbefv.
biLbiemb:b:b*mbtosteam——pvbvbvp:b:b*NpThefirstthreereconstructionsinTable10,*p,*pand*b,arestraightforward.
Asfor*mb,theevidenceforaseriesofprenasalisedstopscomesfromNaxi;thefactscanbedescribedsimplyasamergerofproto-prenasalisedstopswithplainvoicedstopsinNaandLaze.
17Finally,nasal+unvoicedclusterssuchas*NparetentativelypostulatedonthebasisofthepresenceinNaxiofunvoicedstopscorrespondingtovoicedstopsinNaandLaze.
DatafromoutsidetheNaishgroupdonotshedlightonthisissue.
ThehypothesisedchangeinNa16Alternatively,thisNaformcouldberelatedtoBurmesephru'white'.
17LookingbeyondNaish,thehistoryofprenasalisedstopsinBurmo-Qiangiciscomplex:theprenasalisedstopsofNaxidonotcorrespondtothoseofLolo-Burmese,asalreadypointedoutbyBradley1975.
page19/52andLazeisamergerof*Npand*mb(to*mb)beforethemergerof*mband*b(tob).
InNaxi,nasalitywaslostin*Npinitials,leadingtoamergerwith*p.
TherenderingofNaxiplacenamesinChineseintheYuanYiTongzhi(元一统志),abookdated1286,providesequivalencesbetweenvoicingfeaturesinYuan-dynastyMandarinandYuan-dynastyNaxi.
Table11setsoutthedata.
Table11.
TranscriptionoftwoNaxiplacenamesina13th-centuryChineserecord,andphoneticinterpretation.
placenametranscriptioninthe元一统志(date:1286;citedfromFangGuoyu2008:89)andPinyinromanisation'Phags-paintransliteration18'Phags-paasreconstructedbyCoblin2007present-dayNaxiLijiang样渠头yàngqútóuyangkutiw*jagydwigvdyYongning楼头lóutóuliwtiw*lwdwlydyTheChinesephoneticequivalentsforpresent-dayLijiangandYongningare样渠头and楼头,respectively.
InthevarietyofChineserecordedinthe14th-centuryrhymetableZhongyuanYinyun(中原音韵),theinitialsof渠and头areunvoiced;however,usingthereconstructionof'Phags-pabyCoblin2007,theyareinterpretedas*gyand*dw,i.
e.
withthesamevoicingfeaturesaspresent-dayNaxi.
ThisobservationprovidesevidenceonadisputedpointofChinesehistoricallinguistics:itiscurrentlyanissuewhetherthestandarddialectofYuandynastyChinese(NorthernMandarin)retainedvoicedobstruentsornot;thefactsinTable11suggestthatitdid(wehavenoreasontosuspectthatthepresent-dayvoicedobstruentsofNa,LazeandNaxiaresecondary).
HadtheChinesedialectofthepersontranscribingthesenamesalreadylostthevoicingcontrastanddevelopedaspirationonpreviouslyvoicedobstruents,thetranscriberwouldhaveusedunvoicedstops,ratherthanaspiratedstops,tomatchtheNaishvoicedinitials.
Table12setsoutcorrespondencesforcomplexinitials,againtakinglabialsasanexample.
18'Phags-paisawritingsystemthatwasinventedattheendofthe13thcenturytotranscribeChineseandMongolian.
ItisthefirstfullyconsistentspellingsystemforChinese.
Exceptforitsdisregardfortonalcontrasts,itisthemostfaithfulsourceaboutearlyMandarinpronunciation.
Thedialecttranscribedin'Phags-pascriptisnotthedirectancestorofanymodernChinesedialect;however,itwasstillveryclosetotheancestorofmostMandarindialects.
page20/52Table12.
Examplesofcorrespondencespointingtopre-initial+initialclustersinproto-Naish.
Sstandsforeither*ror*s,andCstandsforastop;itshouldbekeptinmindthatthephoneticvalueoftheunitsSandCisunrecoverableinthesecontextsEnglishRgyalrongBurmeseTibetanNaxiNaLazecorresp.
proto-Naishnbofex.
(predictedcorrespondence)(p:p:f)*S-pdryspp-,t-,k-whereCstandsforoneof{r,s,b,d,g}.
Second,coalescenceofthetwoconsonantsintoone:forinstance,inLhasaTibetanpr-andkr-simplifytotheaffricatet-(thesyllableacquiringahightoneintheprocess).
Third,lenitionofoneoftheconsonantswithinclustersinvolvingtwoobstruents:seeforinstancethecomparisonofLavenandNhaHeunbyFerlus1971.
UnlikeNaxi,wherepre-initialsdisappearedwithoutleavinganytracesbeforestops,inLazethesepre-initialscausedalenitionofthefollowingstop–aphenomenonakintothatobservedinVietnamese,wheremedialconsonantswerespirantised(Ferlus1982;abouttheterm"spirants",seeMartinet1981,1985).
InNa,wealsofindspirantisationinsomeetyma,andwepositadistinctpreinitial*C-toexplainthesecases.
page21/52ThereisnoevidenceofspirantisationinthecaseofdentalsineitherNaorLaze.
Thisobservationisplacedincross-linguisticperspectiveinTable13.
Table13.
DatafromseveralEastAsianlanguagesconcerningthelenitionofC2stopsinC1C2clusters.
AmdoTibetandatafromHuaKanandLongBojia1993.
RgyalrongdataandanalysisfromJacques2009.
SituRgyalrongdatafromHuangLiangrongandSunHongkai2002.
TheShuiluoPumidataaresetoutinTable14.
labialvelarcoronalLhasaTibetanspirantisationofonesinglecluster:db-;e.
g.
dbang>/wa/"power"preservedpreservedRgyalrongspirantisationof*jband*zb;e.
g.
*zb->zw-(e.
g.
/zwr/"mugwort",cp.
SituRgyalrong/spor22prɑm52/)preservedpreservedAmdoTibetanspirantisationofalllabialstops;db->-orw-,e.
g.
dbu"head">//;sp->hw-,e.
g.
spu"hair">/hw/preserved;e.
g.
dka"difficult">/hka/preserved;e.
g.
gtam"speech">/htam/LazespirantisationofalllabialstopsspirantisationofallvelarstopspreservedTangutspirantisationofalllabialstopsspirantisationofallvelarstopsspirantisationofts-;rarecasesofspirantisationoft-ShuiluoPumipreservedspirantisationofallvelarstopsspirantisationofallcoronalaffricatesandsomedentalstopsVietnamesespirantisationofallobstruentsinmedialpositionThePumidatabeinghithertounpublished,relevantexamplesareprovidedinTable14.
TheShuiluodialectlosts+obstruentclustersstillfoundintheLanpingdialect(LuShaozun2001).
TheclustersfoundinLanpingPumiregularlycorrespondtofricativesinShuiluo,whereassimplestopsalwayscorrespondtostops;thiswarrantstheconclusionthataprocessofspirantisationtookplaceinShuiluoPumi.
Labialstopsareanexception:theyneverundergospirantisation.
page22/52Table14.
CorrespondencesshowingthespirantisationofnonlabialstopsinShuiluoPumi.
Datafrom2009fieldworkinMuli,Sichuan,China.
meaningShuiluoPumiLanpingPumitocookxsqónineisgitochoptstɑtofeedtsttrouserssdbeardasàstiautochooseststiédeafzabosdboleafpǎsspàtopatchpixspicebubosbùsboFromthedatainTable13,itisclearthatthereisnouniversalhierarchyofpropensitytospirantisationaccordingtoplaceofarticulation.
InTibetan,RgyalrongandLaze,dentalstopsappeartoresistspirantisation;conversely,inShuiluoPumi,labialstopsresistspirantisation.
Apartfrom*C-and*S-,thereisevidencetoreconstructathirdpreinitial,*r-,inproto-Naish.
Theevidencedoesnotcomefrominitiallenition,butfromvowelcorrespondences.
Unfortunately,mostofthecasesinvolvesofewexamplesastobeinconclusive;theonlysyllabletypesthatcanbereconstructedwithconfidencewithapreinitial*r-areshowninTable15.
Table15.
Cognatesetsreconstructedbackto*rts()Vinproto-Naish.
meaningRgyalrongBurmeseTibetanNaxiNaLazeproto-Naisharticulationt-rtschactsigsts*rtsiwashχti(Siturti)ts*rtsimedicinechertsiHtsfi*rtsiwaisti.
L+MH#its*rtsitoholdtsv*rtsUlungst-rtss/hy/,and*hy>/y/;thisistheonlycontextinNaxiwhere//iscontrastive(Michaud2006a).
Thissuggeststhepossibilityofavoicelessnasal*n(correspondingtopresent-day/h:h:h/)atanearlierstage.
Toaccountforallthecorrespondences,asmanyasfourdifferentinitialsmustbereconstructed.
Proposingconcretevaluesfortheseentitiesisathornytask,becauseallofthemappeartohavegonethroughaphasewheretheywererealisedasoneof/l/or/n/.
Acomplexevolutionarypathmustbehypothesised.
Table17setsoutamodelofthesequenceofchangesineachofthethreelanguages.
page24/52Table17.
ReconstructedsequenceofchangesineachofthethreelanguagesleadingtothecorrespondencesinTable16.
NaxiNaLaze(i)*n-,*l->*l-(mergerof*n-and*l-)(ii)*l->*h-(iii)*Sl->*l-(mergerwith*l-)*Sn-,*Sm-,*S->*n-,*m-,*-*Sl->*l-*Sn-,*Sm-,*S->*n-,*m-,*-*Sl->*l-*Sn-,*Sm-,*S->*n-,*m-,*-(iv)*nV,*mV,*V>*h>hV*nV,*mV,*V>h*l->-*nV,*mV,*V>h*l->-Sinceproto-Naish*l-correspondsbothtoCn-andCl-clustersinTibetanandRgyalrong,itisreasonabletoassumethatitresultsfromthemergerofstillearlier*n-and*l-.
Thismerger,whichconstitutesacommonNaishinnovation,hasparallelsinChinese.
OldChinese*hn-and*hl-mergeinMiddleChinese.
Theseinitialsyieldt-or-inMiddleChinese,dependingonthetypeofsyllable(Baxter1992:194,197).
2.
3.
Correspondencespointingtoacontrastbetweenuvularsandvelarsinproto-NaishThereexistsasmallsetofexamplesfollowingthecorrespondence/v::o/,asshowninTable18.
Table18.
CognatesetscomprisingauvularinNa.
meaningRgyalrongBurmeseTibetanNaxiNaLazeproto-Naishsleeve———lɑjkoi.
qvLjɑq*qUswallowmqla——kov*NqUcave———gykow.
qvMlq*qUthroat-rqovvtrutoplantu3.
47tvtvvStustraightu3.
48astutvtv.
tvLvvlStuthousandu3.
49294thostotvvStuholeu3.
50dopɑtvtusleeveu4.
01lɑjkoi.
qvLjɑqqUswallowu4.
02mqlakovNqUcaveu4.
03285gykow.
qvMlqqUthroatu4.
04t-rqolkogqv.
MH#qtsqUhornu4.
05182ta-rkhyuirukoqvqqUflyu5.
01255mblbv.
#Hbbur47toholdu6.
01tsv*rtsUlungsu6.
02t-rtsschuttsv*rtsU46TheBurmeseformmeans'totakeout'.
47Wesuspectthattheformsfor'fly'inLazeandNaxiresultfromright-to-leftvowelharmony,asporadicphenomenonindisyllables(themorefrequentaword,themorepropensityithastowardsvowelharmony).
Likewisefor'kidneys'inNaxi.
page51/52<*rtss)tocoughu6.
03tsv*rtsUTable26.
Rhymes*aC1and*aC2meaningRefHTBRgyalrongBurmeseTibetanOtherNaxiNaLazeproto-NaishchestC1.
01Tangutar<*C-r-kaCkɑpvM)Nka/aC1tofell(atree)C1.
02ndɑdɑdɑndaC1drumC1.
03ndɑkdɑ.
kLdɑkndaC1allC1.
04tɑtɑMH#(.
tɑMH#)tɑ(tɑ)taC1slantedC1.
05lɑ.
tɑLMlɑtɑlaC1taC1howmuchC1.
06qɑ.
kvMH#kɑika/aC1infrontofC1.
07u.
dɑMudɑdaC1tocoverC1.
08fkaβgebsbkabkɑqɑqɑka/aC1toweaveC1.
09318tarakbtagsdɑdɑdaC1blackC1.
10317anaknagponɑnɑnaC1sharpC1.
11318-9thaktɑtɑtɑtaC1tohitC1.
12ltlɑlɑlɑlaC1wolf48C1.
13qaparmparpɑkpɑpaC1vat,woodenbasinC2.
01lololulaC2tobeashamed49C2.
01ndo.
doMH#hdundaC248TheTibetanandRgyalrongcognatesactuallymean'dhole(Cyonalpinus)'.
49ArelationwiththeformscitedinMatisoff(2003:317)ispossiblebutrequiresfurtherresearch.
page52/52tastyC2.
02sososusaC2toleanagainstC2.
03tototutaC2slopeC2.
04tototubietaC2tohugC2.
05tototo.
toMtutaC2gruelC2.
06hohuhaC2toseeC2.
07dododaC2valleyC2.
08lololaC2toclimbC2.
09ndodogdundaC2tojumpC2.
10mtsatsotsotsutsaC2needleC2.
11342taqaβapkabkouuNkaC2tostudyC2.
12sososusaC2toworkC2.
13luplobelo.
iMluvielaC2pig318-9pawakpagbubuwSbaChand319t-jalaklaglɑolo.
qwLMlɑpielaC1/laC2breath317saksrogsɑsosɑsaC1/saC2thickjalɑloɑplulaC1/laC2deep317rnanakhoohɑlaC1/laC2/SnaC1tokillqokɑkaC1/aC2