Title Reported Speech in Lisu and Burmic

Author(s) Bradley, David

Citation Proceedings of the 51st International Conference on Sino-Tibetan Languages and Linguistics (2018)

Issue Date 2018-09

URL http://hdl.handle.net/2433/235263

Right

Type Conference Paper

Textversion author

Kyoto University

ICSTLL51Kyoto25-28September2018

ReportedSpeechinLisuandBurmicDavidBradley

LaTrobeUniversity,AustraliaAbstractThe embedding of reported speech in discourse is universal; it is often one element in anevidential system. Interestingly, the reported speech formswithin evidential systemsdifferwidelyacrossBurmiclanguages.InLisu,thereisaclause-finalreportedspeechevidentialorhearsaymarkerjo: /dʐo21

/, whichisbyfarthemostfrequently-occurringevidentialmarkerin the language. This has dialect forms including Northern Lisu /dʒo21/ and Southern Lisu/dzo21/and/do21/.

A second frequentwayofembeddingreportedspeech inLisu is tohaveaprecedingverbof speakingor cognition/perception,most frequentlybA /bæ44/ ‘say’, in anominalisednonfinalclauseending in thetopic/conditionalmarkerny /ɲɑ44/as inyi. bA_ m ny [quote]/ji55 bæ44ɑ44 mɑ44 ɲɑ44 [quote]/ ‘What he said was [quote]’. Alternatively, a quote can beprecededbyspeakerand/oraddresseeNPswithoutaverb,followedbythetopic/conditionalmarkerny /ɲɑ44/;orbya simplenon-nominalisedverbclausesuchasyi. bA_ /ji55 bæ44ɑ44/‘Hesaid[quote]’.Nomodificationtopronounorotherformswithinthequotationisrequired,unlikeforexamplewithindirectspeechinEnglish. A third extremely frequent way is to make the quotation into a complement of afollowingshortmainclausewithaspeakingorcognitionverb.InLisu,thecomplementiserbe

/be44/directlyfollowsthequotation,andthenthereisashortclausewithaverbofspeaking,again most often bA, as in [quote] be nuY tA. bA_ lo [quote] /be44 nu33 tæ55 bæ44ɑ44 lo44/‘(Someone)saidtoyouthat[quote]’;manyotherspeakingverbsarepossible.FortheverbbAonly,aslongastherearenoNPsprecedingtheverbinthepostquoteclause,thepresenceofthecomplementiserbe isvariableandinfrequent.

Oftenthesecondandthethirdembeddingstrategiesarecombined,orlessfrequentlyallthree,withtheevidentialreportedspeechmarkerinfinalposition,likeyi. nu tA. bA_ m ny

[quote]be bA_ lo jo: /ji55 nu33 tæ55 bæ44ɑ44 mɑ44 ɲɑ44 [quote] be44 bæ44ɑ44 lo44 dʐo21/ ‘It is saidthatwhathesaidtoyouisthat[quote]’.

A fourth possibility is to embed speechwithin discoursewithout any overtmarkingthat thespeech isquoted; inoralnarrative, thespeakermayadoptadifferentvoice for thequoted passage. This is particularly frequentwhere a long series of quotations of dialogueoccurs,asinnarrative.

Theevidentialjo: is frequent inLisu traditional genres, including archaic songs andproverbs and modern narrative and conversation; unlike other evidential and epistemicmarkers,itcanalsooccurafteranotherevidentialorepistemicformtogivehearsayevidenceaboutthesourceofknowledgeand/ordegreeofcertainty.Thespeakerislesscommittedtotheaccuracyofthereportedspeechwhenusingthisevidentialthanwiththeotherquotation-embeddingstrategies.

The Lisu reported speech evidential has likely cognates in closely-related Ngwilanguages, such as Lahu cev /tʃe54/ and Akha jev /dʑe55/, all related to Proto-Ngwi *jay2(Bradley 1979a etymon 861). Literary Burmese and modern spoken Burmese havecompletelydifferent clause-final reported speechmarkers, literaryhu /hu44/ and spokenlui≥

/lo44/ort≥ /tɛ44/.Thisindicatesthattheseformsmaychangerelativelyrapidly,asisoftenthe

case with evidential and other grammatical marking. Nevertheless, the overall syntacticpatternsaresimilar,withinterestingminordifferences.1.IntroductionItisnecessarytodistinguishbetweenmoreorlessexactdirectquotation–wordforwordorclosely paraphrased, versus hearsay information being reported. In Lisu and some relatedlanguages, for direct quotation there are differences between embedding strategies whichincludeonlynominalelementssuchasspeakerandaddressee,asopposedtothosewhichalsospecifyarelevantspeakingorrelatedverb.Onetypeofembeddingframefordirectquotationprecedesthequotation;another follows it.Likemostevidentialsandepistemics inLisuandrelated languages, the evidential of hearsay or reported speech is clause-final, while theprequoteandpostquoteframessurroundthequotation.Itisalsorelevanthowmuchisbeingquoted:asinglewordorphrase,asingleclause,orlongerstretches.2.LisuIn Lisu, there are various styles and genres with different structural patterns. The mostarchaic,andalsothemoststylised,istraditionalsongs,proverbsandriddles,whichfollowarigidmetricalpatternasdiscussedin2.1.Modernspokennarrativeandconversationalstyleshows considerable dialect differentiation, but all three major dialects follow the samepatternsanduserelatedformsforhearsayandfordirectlyquotedspeech,asdiscussedin2.2.Thehearsayevidentialandotherevidentialsandepistemicsarerelativelyfrequentinspokenlanguage; the quotation strategies are also rather frequent. There is also a third formalChristian literary style, used in Bible translation and genres derived from this such asChristiansermonsandotherChristianwrittenliterature,whichusesprequoteandpostquoteframingdevicesveryfrequently,butuses fewerevidentialsandepistemicsthananyspokendialect; the hearsay evidential is not used in this style. On the other hand, the epistemicswhichareuseinChristianstyleareusedmuchmorefrequentlythaninanyspokendialect,asdiscussedin2.3below.2.4belowprovidesanexampleofanon-Christianwrittentext. Thethreegeneralspeech-embeddingstrategiesareasoutlinedintheabstractabove.Thefurthestfromdirectquotedspeechistheclause-finalhearsaymarkerwiththeformjo:

/dʐo21/inCentralLisuandwrittenLisu,whichis/dʒo21/inNorthernLisu,andzo: /dzo21/ordo: /do21/ in spoken Southern Lisu. However, when writing Lisu, literate speakers of alldialects represent all of thesewith the spelling jo: reflecting the Central Lisu and literaryform.

Theprequoteandpostquote framesembedmoreor lessdirectlyquotedspeech.Themaximal prequote frame is an entire clause specifying a speaker, an addressee (andsometimesothernominalarguments),thenaverbofspeakingplusclause-finalmarkers.Thisusually ends in the topic/conditionalmarkerny /ɲɑ 44/ in spoken language, but not in thesongorChristianstyles.All,anycombinationoftheseornonemaybepresentbeforeadirectquotation. The maximal postquote frame in spoken and Christian literary style has acomplementiser be /be33/ followed by a complete clause which may contain speaker,addressee and other nominal elements, sometimes evenwhen the prequote frame alreadyincludesthesameelements,butalwayscontainsaverbofspeakingorcognitionfollowedbyclause-final markers. Nominal elements are less frequent in the postquote frame, and aremorefrequentwhentheprequoteframedoesnotincludethespeakerand/ortheaddressee,orwhenthequotedpassageislong.WiththeverbbA /bæ44/‘say/speak’,thecomplementiserbe isnot requiredandvery infrequent if therearenonominalelementspreceding it in thepostquote frame; this applies to spoken Lisu as well as to Christian literary language.

Traditional song language lacks the complementizerbe and uses a variety of speaking andcognitionverbsdirectlyafteraquote.ByfarthemostfrequentpostquoteframeinallstylesandgenrescontainsonlytheverbbA plusfinalmarkers.2.1TraditionalSongLanguageStyleTraditional Lisu songs are mainly in seven-syllable paired lines, with the second linerepeatingthegeneralmeaningofthefirstlinebutusingsomealternativelexicalforms,oftenincludingobscureorarchaicwords.Thereisastandardpairwordforeachfrequentnominalandverbalelement,butwithsomevariationbetweenandevenwithinsongs.Mostlinesbreakinto two parts; the first four syllables are usually a nominal element and the last threesyllablesareoftenbutnotalwaysaverbalelement.Thevocalicclitics,vocativeo:/o21/anddiminutiveO: /ɑ21/orO; /ɑʔ21/afteranounandthedeclarativemarkerO /ɑ44/,theperfectivemarkero /o21/andthequestionmarker O: /ɑ21/afteraverborverbalsequence,donotcountas a syllable in the line; in Christian literary style these are written together with theprecedingword.Thishighlyconstrainedmetricalformleadstorelativelynon-complexsyntaxwithin a line.Mostproverbs and riddles followan exactly similarpattern, but containonlyone pair of lines. As amnemonic for these originally oral texts, part of the lexicalmaterialtowardtheendofapairoflinesinasongisoftenpartlyrepeatedintheearlypartofthenextpairoflinesorsoonafterward.

Inlongsongstherearemanyquotations;duetotheabsenceofrelevantnominalformsitissometimesambiguouswhoisspeakingandtowhomtheyarespeaking.Occasionallythisis disambiguatedwith an extra two or occasionally four syllable nominal preface to a linewhichspecifiesthespeakerand/oraddressee;theseaddedelementsdonotcountaspartofthesevensyllablesintheline,occuronlyonthefirstofthetwopairedlines,andtendtobeeditedoutbyLisuwriterswhen transcribing these songs; thus theydonotoftenappear inprintedversionsofthem.WewillconsiderdatafromtwoSouthernLisusongs.

TheLisuNewYearsong(Bradleyetal.2008),atraditionalSouthernLisusongtext,isarelatively short song with only 407 pairs of lines. In this song, the hearsay evidential jo:/dʐo21/occursonceinthefollowingexamplenearthebeginning(p.13lines13-14).

Ti: Ko; FiY nyi: vY VY jo:I

thi21 khoʔ21 tshɿ33 ɲi21 hɑ33 ŋɑ44 dʐo21 one year ten two month be QUOTE

Ti; vY s: FiY vy; VY jo:=

thi21 hɑ33 sɑ21 tshɿ33 hjɑʔ21 ŋɑ33 dʐo21

one month three ten night be QUOTE

‘Itissaidthatoneyearistwelvemonths,Itissaidthatonemonthisthirtynights.’

Inthesametexttherearetenpairsoflineswithaspeechorhearingverbattheendofalineindicatingdirectquotationofspeech:nineinstancesofbAwithitsusualpairwordde:/de21/‘beg’inthesecondlineasinthefollowingexample(p.35lines7-8).OncebAoccursattheendof both lines. Notice that it is sometimes only the context here which indicates who isspeaking.Inthissongthetwopartsaretakenbythehostandtheguest,andthisisfromoneoftheguest’spassages.

viY si PuY fo, i gL:O bAO:I

hi33 sɿ44 phu33 tsho35 ʔi44 gɯ21 ɑ44 bæ44 ɑ21 house mastersilver table set GIVEDEC sayQUESTION

kw si xU fo, t. gL:O de:O:=

kwɑ44 sɿ44 ʂʅ44 tsho35 tɑ55 gɯ21 ɑ44 de21 ɑ21 hearth mastergold table put GIVEDEC begQUESTION

‘Did(you)saythatthehost(you)setthesilvertable,Did(you)askwhetherthehostess(you)putthegoldtable?’

Therearealsofourinstancesoftheperceptionofspeechverbp j: /pɑ44 dʐɑ21/ ‘hear’pairedwith do: j: /do21 dʐɑ21/ which is its song language pair word (not occurring outside songlanguage)andisrelatedtothenoundo: /do21/‘news’,asinthefollowingexample(p.14lines15-16). Not all are reported speech; some report hearing of nonspeech sounds or otheractions.

nyi U Fi; ho: p j: lA:I

ɲi44 ʔy44 tshɿ21 xo21 pɑ44 dʐɑ21 læ21 day count herb put.in hear QUESTION

vy: cy, mA, kL do: j: lA:=

hjɑʔ21 tɕɑ35 mæ35 kɯ44 do21 dʐɑ21 læ21 night reckonherb insert hear.news QUESTION

‘Did(I)hearthat(you)putherbsinonthecountedday,Did(I)hearthenewsthat(you)insertedherbsonthereckonednight?’

Theseexamplesindicatethedifferencebetweentheuseofjo: andaspeechorhearingverbafterthepassage:jo: ishearsayortraditionalknowledge,notattributabletoaspecificpersonwhosaidexactlythistothespeaker;whilethespeechandhearingverbsembedmoredirectquotesorobservations.

Oftheevidentialsandepistemicsinthistext,therearefiveepistemicswhichoccurinthe407pairedlines:eleveninstancesoftheepistemicb.: /bɑ55 ɑ21/ofdoubtorlowlikelihood,inteninstancespairedwithP.: /phɑ55 ɑ21/whichisthecorrespondingepistemicinsomeotherdialectsofLisu;teninstancesoftheepistemicbD, /bɤ35/ofchangeofbelief(Iusedtothinkotherwise,butnowIamcertain;somewouldcallthisan(ad)mirative);fourinstancesofthefuture certainty epistemic (itwill definitely be thisway)nA, /næ35/ (with tonally differentSouthernLisudialect formsseen in2.2below); three instancesof theepistemicPe.o /phe55 o44/ofuncertainty;andtwoinstancesoftheepistemicofabsoluteongoingcertaintylo /lo44/.Thepairforoneinstanceofb.: andoneinstanceofPe.o isdu /du44/whichisaninfrequentevidentialofinference:thespeakerinfersorguessessomething.Apartfromthisandtheonepairofreportedspeechevidentialscitedabove,thereisalsoonepairedlinewiththeSouthernLisuvisualevidentialbo, /bo35/inbothlines(p.32line3-4).

b, b: PuY gL: bi le bo,

bɑ35 bɑ21 phu33 gɯ21 bi44 le44 bo35 father silver box full CHANGEVISUAL

m, m xu gL: lL: le bo,

mɑ35 mɑ44 ʂʅ44 gɯ21 lɯ21 le44 bo35 mother gold box overflow CHANGEVISUAL

‘Iseethatfather’ssilverboxhasbecomefull

Iseethatmother’sgoldboxisoverflowing.’

Thus theSouthernLisuepistemicsandevidentials includingjo: are found in32outof407pairsof linesor9.1%ofall lines,andembeddeddirectspeech is found in fourteenpairsoflinesor3.4%.

Inthefirst80versesor2591pairsoflinesoftheunpublishedSouthernLisumu: gw: d:

m /mu21 gwɑ21 dɑ21 mɑ44/‘BigSong’,theLisuhistoricalandlifecyclesongascollectedbyDavidNgwaza,thereportedspeechevidentialjo: isused22times,mainlyearly inthesongwhenmoreancienttraditionalpracticesarediscussed.Thevisualevidentialbo, isused,herepairedwiththesonglanguageformwo. /wo55/whichisnotattestedinanycurrentspokendialect,26 times;of thevariousepistemics, forvery lowprobabilityapairofb.: andP.: isused19times; there is analternativewithP.: inboth lines three times.Theepistemic formPe.o ofuncertaintyisusedonce.Forfuturecertainty,nA, isused58times.However,theepistemicslo and bD, do not occur. By contrast, direct quotation of speech is embedded with thespeakingverbpairbA andde: in191pairsoflines;threetimesbA isusedinbothlines.Thusthe frequencyofdirectquotewith thespeakingverbbA ismuchhigher than theuseof theevidentialsandepistemics.Thespeakingverbisembeddedwithinthesyntacticstructureofthelineandhasrelevantverbalmarkers,whiletheepistemicsb.:/P.:andPe.oandevidentialsjo: andbo,/wo. onlyoccurasthelastsyllableoftheline.TheepistemicnA, occursattheendofthelinefivetimes,isfollowedbytheperfectiveo 50times,bytheperfectiveandthefinalpoliteness marker m: /mɑ21/ twice, and in the middle of a line once, as in the followingexamples.

m: xUY yi. hw: z: kD nA,I

mɑ21 ʂʅ33 ji55 xwɑ21 dzɑ21 kɤ44 næ35 NEG die NMZR meat eat PAST FUT

m: nY yi. xU: CU; gL nA,=

mɑ21 nɑ21 ji5 ʂʅ21 tʂhʅʔ21 gɯ44 næ35 NEG sick NMZR thing suck PAST FUT

‘(I)willeatmeatwhile(I)amnotdead,(I)willsuckthingswhile(I)amnotnotsick.’

B P: Ti: wu.O: yiY Ko; nA,o

ɣɑ44 pɑ21 thi21 wu55 ɑ21 ji33 khoʔ21 næ35o44 fowl male one topDIM water fetch FUTPERF

B m Ti: t_: wo: hw nA,o

ɣɑ44 mɑ44 thi21 tɑ44 ɑ21 wo21 xwɑ44 næ35o44 fowl female one cackleDIM vegetables find FUTPERF

‘Atthefirstcockscrow,(I)willhavefetchedwater,atthefirstcackleofthehens,(I)willhavefoundvegetables.’

PY bo bo le: z: nA,o m:I

phɑ33 bo44 bo44 le21 dzɑ21 næ35o44 mɑ21 until full full ADV eat FUTPERF POLITE

PY yi; yi; le: doY nA,o m:=

phɑ33 jiʔ21 jiʔ21 le21 do33 næ35o44 mɑ21 until drunk drunk ADV drink FUTPERF POLITE

‘(You)willpleasehaveeatenuntilyouarefull,(you)willpleasehavedrunkuntilyouaredrunk.’

R: CiY CoY nA, dU: j: kDI

zɑ21 tshɿ33 tʂho33 næ35 dy21dʐɑ21 kɤ44son foot follow FUT think PAST

nE lA; bY nA, go; ruY gL=

nø44 læʔ21 bɑ33 næ35 goʔ21 ru33 gɯ44 child hand follow FUT consider PAST

‘(I)thought(I)wouldfollowson’s(your)feet,(I)considered(I)wouldfollowchild’s(your)hands.’

The evidentials and epistemics are less frequent in this long song than in the shorterNewYear song. Speaking verbs embeddingdirect quoted speech are farmore frequent than thehearsayevidential: theyoccur in194versus22pairsof lines;however, this isstillahigherproportionofuseofjo: thanintheshortersong.Overall,embeddedspeechisfoundin194of2591or7.5%oflinepairs,andevidentialsorepistemicsin103or4.0%,including22or0.8%jo:.2.2SpokenStyleThereare substantialdialect and individualdifferences in theuseof thehearsayevidentialandtheembeddingofquotations;heretheusageofaskilledSouthernLisustorytellertellingatraditionalstoryaboutthestupidtigerandthecleverrabbitiscompared. In the 320 clauses of this story, there are 71 direct quotations; of these, 46 areimmediatelyfollowedbybA. do: whichistheSouthernLisuspokenversionofbAO jo:.Oneisfollowedbythesameconstituentsinreverseorder,do: bAandanotherfourarefollowedbyTeY bA. do: ‘thissayQUOTE’.Tenquotationsinthemiddleofastringofquotationsbackandforthbetween the tiger and the rabbit orother interlocutoronlyhavebA do: after the lastquotation; the narrator just changes voices. Seven have bA. with or without additionalfollowingmaterial but do not endwithdo: and three have onlydo: after the quotation. Inaddition, there are twenty instances of do: used as a hearsay marker, not on a directquotation;alsothreeprequotesendingindo:andfourinstancesofnon-quotationsfollowedbybA do: orbA gL: do:. Thus overall there are 78 examples of hearsaydo: in the text, on24.4%ofclauses;thisisthemainmeansofembeddingquotationsandisfrequentlyusedforotherlessdirecthearsay.Thefollowingisanexample.

a. TeYO “a.I VwY a. ti. gL: i; t. m:” bA. do:=

/ɑ55 the33 ɑ44 ɑ55 ŋwɑ33 ɑ55 ti55 gɯ21 iʔ21 tɑ55 mɑ21 bæ55 do21/ then oh, I a.bit give sleep STAY POLITE say.DECQUOTE

‘Then(thetiger)said“Oh,letmesleepabit”.’

Oftheepistemics,SouthernLisunA: /næ21/offuturecertaintyisusedeighttimes,P.:

ofverylowprobabilityisusedonce,andnootherepistemicsorevidentialsareusedatallinthisstory.ThetextalsousesbA. orbA ny orbAO m eighteentimesasdiscoursemarkers,notdirectlyconnectedtothesurroundingsyntacticcontext.

Prequote framesoccurbefore22of71quotations;most frequent is theadverbiala.

TeYO /ɑ55 the33 ɑ44/‘then,’seventimes;orthesamefollowedbype yeI/pe44 je44/‘ithappened,’three times. Five contain only a Speaker nominal, four of which are followed by the topicmarkerny,whileonemorecombinesa. TeYO plusaSpeakernominal;onefurtherprequotehasaSpeakerplustheverbn nyi ‘ask’,andonespecifiesaplaceandtime.ThreeprequotesareclausescontainingbA as theverb followedbyhearsaydo:.Forpostquotes,asdiscussed

above, this text uses a combination of the verb bA and the hearsay marker do: 51 times,nothingtentimes,aclausewithbA astheverbseventimes,andjusthearsaydo: threetimes;this isquiteunlike thepattern in literaryLisuas seen in2.3and2.4. Immediatelybeforeaquotation,theprecedingclauseofthemaintextoftenendswiththeSouthernLisusequentialmarkerve. /he55/‘next’,whoseCentral,Northernandliterarycounterpartissi. /sɿ55/;thisisnotstrictlyspeakingaprequote,asthismarkerindicatesthatthefollowingclauseoccursafteritbutisnotnecessarilyotherwiseconnectedwithit.2.3BibleTranslationandChristianStyleOnly one type of quotation pattern is found in Lisu Bible translations. Of course these aretranslations,andtendtobeformulaic.TakingallofthequotationsfromJesusintheLisuredletterNewTestament(1999),whichofcoursehasallthewordsofJesusprintedinredsotheyareeasytoidentify,thepatternisasfollows.

Of 648 quotations of Jesus in this translation, 89.8% are preceded by at least oneconstituent in the prequote frame, maximally including Time nominal-Speaker-Addressee-Verbplusassociatedclausemarkers,inthisorder.Mostincludeaspeakerandanaddresseeintheprequoteframe.ThespeakerhereisofcourseJesus,andhisnameorapronounyi. /ji55/‘he’ is foundbefore 78.1%of the quotations; this is usually followedby a nominalmarker,mainly the agentive le /le44/ (67.3%), the topicmarkerny /ɲɑ44/ (14.0%), the alternativetopicmarkern:/nɑ21/(6.5%)ornonominalmarker(12.1%).Anaddresseeispresentbefore67.2%ofquotes,alwaysafterthespeakerandallfollowedbytheobjectmarkertA. /tæ55/.29.9% of the quotes with a speaker and/or an addressee also contain a following verbalelementorelementsand their clause-finalelementsbefore thequote; thus in fact themostfrequentprequote framecontainsonlynominalelements.Mostof the10.2%ofquoteswithnoprequoteframearequoteswhichfollowonfromimmediatelyprecedingquotes,separatedonly by a short post-quote speaking verb clause after the preceding quote. Unlike spokenstyle,prequoteframesmostlydonotendinthetopicmarkerny.

After every quote, the postquote frame contains a speaking verb followed byappropriateclausemarkers.Mostquotes (94.8%)are followedbyashortclausecontainingtheverbbA/bæ44/‘say/speak’followedbyvariousclause-markingelements.IfthespeakingverbisanythingotherthanbA thenthecomplementiserbe /be44/alwaysfollowsthequoteandprecedesthatverb;theverbsattestedinthistranslationoftheNewTestamentaren nyi

/nɑ44 ɲi44/ ‘ask’ 45 times,Po, gL: /pho35 gɯ21/ ‘answer’ seven times,TU; w: gL: /thyʔ21 wɑ21 gɯ21/ ‘command’ three times and KuY /khu33/ ‘call’,w: KuY /wɑ21 khu33/ ‘pray’, de: /de21/‘beg’, jo k: gL: /dʐo44 kɑ21 gɯ21/ ‘stop fear’ and bo /bo44/ ‘write’ all once each. Veryoccasionally, an addressee or other nominal element (usually one not present in theprequote) can be added between the complementiser be and the speaking verb. It is alsopossiblebutinfrequent(13of586instancesofbA)tousethecomplementiserbe beforetheverbbA.Thisisobligatorywhenanominalelementsuchasspeakeroraddresseeispresentinapost-quotebeclausebeforebAandotherwiseveryinfrequent.

Many post-quote clauses (29.8%) include the posthead serial verb gL: which ishomophonouswith themainverbgL: /gɯ21/ ‘give’ andmeans that theverbal action is forsomeone’sbenefit.Thisislexicalisedintotheverbs‘answer’and‘command’andisalsooftenfoundwithotherverbsincludingbA (31.6%oftheinstancesofthisverb).

Most of the post-quote clauses (78.8%) endwithlo /lo44/which is the epistemic ofabsolutecertaintyinnormalspokenLisu.Otherpossibilitiesincludenonfinalclausemarkerslikesequentialsi./sɿ55/,temporalTA:/thæ21/,conditionalny /ɲɑ44/andnominaliser_ m /ɑ44

mɑ44/ where a further quote or some immediate consequential action follows. A typicalexamplefollows.

ye-su le yi. tA. [quote] bA gL:_ lo=je44su44le44 ji55 tæ55 bæ44 gɯ21ɑ44 lo44 Jesus AGT he OBJ [quote] say GIVEDEC CERTAIN

‘Jesussaidtothem[quote],hesaid.’(Matthew8:22)

TheBiblestyleofLisudoesnotuseanyoftheevidentialsusedinthevariousspokendialects. This includes the hearsaymarker jo: and related forms,which is a very frequentclause-finalevidential inalldialectsof spokenLisu(Bradley2010).A fewof theepistemicsusedinspokenLisuvarietiesarealsousedinBiblestyle.Theepistemicmarkerloofabsolutecertainty is found in 78.8%of post-quote clauses in theNewTestament translation, and isveryfrequentinotherclausesintheNewTestamentaswell.PresumablythewordsofJesusandmanyof theother clauses in the restof theNewTestamentare followedbylo togivethem absolute certainty and authority; they are not intended to be taken as hearsay orotherwisedoubtful.Thisgreatly increaseduseoflo isnowalso found inreligioussermonsandChristianwrittenstyles.ThisisverydifferentfromotherstylesofLisu, inwhichlo isalow-frequencyepistemicofverystrongcertainty, rarely found in traditional song language;itssemanticstrengthhasbeengreatlyreducedbyitsoveruseinChristianstyle.

2.4SecularWrittenStyleForanexampleofquotationpatternsinnon-ChristianwrittenLisu,wewillconsidersi,-Ju-x(1995),abriefhistoryoftheLisubyaLisucadreinChina,whospeakstheNortherndialectofLisu.ThisbookiswrittenusingtheChristianorthography,butalsousingquotationmarkstoframe some quotations. Its introductions and first four chapters contain 423 clauses,including72quotations;mostquotationsareshortbutsomeare longer thanoneclause.Ofthequotations,16or22.2%haveaprequoteframe,twicewithaSpeaker(oncefollowedbyle

andoncewithoutanominalmarker)andsevenwithanAddresseeplustA. ny andfourwithAddresseeplustA. asexpected;Addresseeisalsooncefollowedbynyi/ɲi44/‘then’. Thepostquote frameoccursaftereverysinglequote;of the72,56containbe plusaverb clause,with six verbs represented; 15have just a verbdirectly after thequotation, ofwhich11arebA asonewouldexpect,threearenegatedm: KuY /mɑ21 khu33/‘notcall’,andoneisbo/bo44/‘write’.Intheoneremaininginstance,thepostquoteisjustanounphrase,buttheclause preceding the quote includes the embedding verb bo ’write’. With the verb bA, 11postquotes are without be, and in 16 be precedes bA. None of these has an interveningnominalbetweenbe andtheverbbA.

ThefollowingexampleshowsaprequoteframewithSpeakerandAddresseefollowedby a quotation, which in turn contains a prequote frame, then another short quotationembeddedwithinit,eachfollowedbyanappropriatepostquoteframewithbe plusverbanditsmarkers.

goY m pL. du suY xU: le ro: tA.

/go33 mɑ44 pɯ55 du44 su33 ʃʅ21 le44 ʒo21 tæ55/that NMZR reason others group AGENT we.INCLOBJ

‘a. xU: pL. du nuY w: li-su TEY Ti: xU: tA. nyI

/ɑ55 ʃʅ 21 pɯ55 du44 nu33 wɑ21 li44 su44 thø33 thi21 ʃʅ21 tæ55 ɲɑ44/what reason youPL Lisu this one group OBJ TOPIC

li-su be KuYO lo?’ be n nyi TA:I

/li44 su44 be44 khu33 44 lo44 be44 nɑ44 ɲi44 thæ21/Lisu COMP callDECFACT COMP ask TIME

‘Forthatreason,whenothergroupsaskus‘WhydoyoucallthisLisugroup“Lisu”,…(Shi1995:9,missingtonemarksadded,NorthernLisupronunciation)

Thereisnouseofthehearsayevidentialjo:oranyotherevidential,whichmaybedueto the influenceof theChristianwrittenstyleand/or to the fact that theauthor intendshisaccount to be authoritative. The absolute certainty epistemic lo is used at the end of 48clausesor11.3%; this is higher than its frequency in anyvarietyof spokenLisu, butmuchlowerthanitsfrequencyinChristiangenres.Thechangedknowledgeepistemicisusedtwice,spelledbB, /bɤɑ35/fusedwithafollowingdeclarativeandthenfollowedbyconditionalny asitsometimesoccursinChristiancontexts.2.ReportedSpeechinOtherNgwiLanguagesTheelementsusedforhearsayandquotedspeechintwootherNgwilanguageswillbebrieflyoutlined:oneotherCentralNgwilanguage,Lahu,andoneSouthernNgwilanguage,Akha.2.1LahuTheCentralNgwilanguageLahuisfairlycloselyrelatedtoLisu.Thusitisnotsurprisingthatitsquotation-embeddingstrategiesaregenerallysimilar,thoughmanyoftheformsusedforembeddingdiffer. InLahuNa(BlackLahu), there isaclause-finalhearsaymarkercev/tʃe54/whichMatisoff(1982:469)describesas“indicatingthatone’sstatementisbasedonsecond-handinformation,ratherthanpersonalexperienceordirectknowledge”; inafootnote(ibid.600-601) he explicitly links this to hearsay evidentiality. There is also a post-quotationmarker tehʌ /tɛʔ21/ or less frequently tehv/tɛ21/ ‘real’ which embeds more or less directquotations,andamore informalalternativehk’e/qhe33/ ‘like’ (ibid.467-468).PlacementofcommasinwrittenLahutextindicatesthatthisquotationmarkerisattachedtotheendofthequotation,nottoafollowingpostquoteframecontainingaverbofspeaking.

Theprequote framing elements in Lahumay include someor all of aTimenominal,Speaker,Addresseefollowedbytheobjectmarker htaʌ/thaʔ21/andverborsequenceofverbspluspostverbalmarkers,mostoftenk’oʌ/qoʔ54/‘say’usuallyfollowedbydeclarativeve/ve33/,oralsoinsteadbyleh/lɛ33/‘also’whentheverbisotherthank’oʌ.InthepostquoteframeinspokenLahu,asinLisu,themainspeakingverbk’oʌveryoftendirectlyfollowsandembedsaquotation (ibid. 465-467); as in Lisu, it is frequently followed by piv /pi54/, the postverbhomophonouswiththeverb‘give’,inLahuindicatingaction(orinthiscasespeech)directedtoathirdperson.UnlikeLisu,wherethecomplementiserbeisobligatoryafterthequotationbeforeotherverbsofspeakingorthinking,variousotherLahuspeakingandcognitionverbsmay alsodirectly followand embedquotations, usuallywith a precedingquotationmarkertehʌorhk’e. Anotherpatternprecedingquotes is tohave the speaker followedby the topicmarkerlehv/lɛ21/,withoutaverb.

UnlikeLisu,Lahuhasanadditionalembeddingdevice,withclause-finalna_/na112/inanonfinalembeddedquestionclause(ibid.470-471),whichisoftenaquotationofone’sownorsomeoneelse’squestion; inLisusuchclausesareembeddedinthenormalwaybybeplusaquestionverbsuchasn nyi/nɑ44 ɲi44/‘ask’.ThisLahuembeddedquestionmarkerisregularlycognatewithLisun: /nɑ21/,Akhanav /na55/,Nisu/no55/andother related formswhicharemarkers of a secondary topic and also of a nonfinal conditional clause, all regularlyreconstructibleas*ʔna2 (Bradley1979aetymon860). InLisu,n: isalsothesongwordpair

fortheprimarytopicandconditionalmarkerny.ThusitappearsthatthisLahuusageofna_isaninnovativedevelopment.

TheLahuNewTestament(1962)andotherLahuscripturesmaximallyuseaprecedingquotation frame containing a Time nominal, Speaker, Addressee and verb or sequence ofverbsofspeakingorcognitionplusassociatedverbmarkers,orSpeakerplustopiclehv/lɛ21/.The postquote frame always has the quotation marker tehʌ,, usually followed by a clausepotentiallyincludingaspeakerandaddresseeandalmostalwaysincludingaverbofspeaking,again most often k’oʌ. Thus the pattern is very similar to that seen in Lisu, and again thereported speech evidential is not used in this literary context where the intention is toindicatethatthequotationsareaccurate,notjusthearsay.

Of133quotationsofJesusintheLahuNewTestamentBookofMatthew,114(85.7%)haveaprecedingquoteframe,and129(97%)haveafollowingquoteframe.Oftheprecedingframes,23startwithaTimenominal,mostoftenovhtav/o54 tha54/‘then’;allhaveaSpeakernominal,71haveanAddresseenominalfollowedbyobjectmarkerhtaʌ,and102endinaverb(ofwhich94arek’oʌ)plusassociatedpostverbalmarkers.Nearlyalloftheprequoteframescontainingk’oʌhavedeclarativeve inclause-finalposition,while fiveoftheeightwithotherverbsinsteadhavefinalleh.Ofthe12prequoteframeswithoutaverb,allcontainaSpeakernominal followed by the topic marker lehv. All quotes in this text end with the quotationmarkertehʌ,;inthepostquoteframe,126of129endinaverb,ofwhich113arek’oʌ;ofthosewithotherverbs,elevenareembeddedquestions followedbyapostquotewith theverbnanyi/na33 ni33/‘ask’;na_isnotusedhere.In24postquotes,aSpeakernominalispresent;in27thereisanAddresseenominalplusobjectmarkerhtaʌ.TheSpeakerandAddresseenominalsaremorefrequentinthepostquoteiftheyarenotpresentintheprecedingprequote;unlikeinLisu, the Speaker is sometimes repeated after a long quote. Unlike Lisu, in Lahu it is alsopossible for the Addressee to precede the Speaker in the prequote or less frequently thepostquote frame. Thequotationmarker tehʌ is alsoused toembedsinglewordquotations,withoutprequoteorpostquoteframe,asinthefollowingexample.

Da_viʌ yawv htaʌ Jawvmawv tehʌ k’oʌ ve k’o,…./da112 viʔ54 jɔ54 thaʔ21 dʒɔ54 mɔ54 tɛʔ21 qoʔ54 ve33 qo33/ David he OBJ lord/master QUOTE say DEC CONDITIONAL

‘IfDavidcallshim“Lord”...’(Matthew22:45)

UnlikespokenLahu,quotationscontainingquestionsareembeddedinexactlythesameway,notwithclause-finalna_. In Lahu Shi (Bradley1979b), thequotation system is fairly similar, but unlikeBlackLahu,thequotemarkerattheendofeveryquoteislehd/lɛ55/orlehx/lɛ13/ratherthantehʌ.Also, the possibility of having just a Speaker nominal plus topic marker in the prequotepositiondoesnotoccur.Forexample,intheLahuShiNewTestament(2015),allofthetwelveinstancesofSpeakerplustopicmarkerin(Black)Lahuinsteadhaveaprequotecontainingaverbof speaking,which isusuallykoz/ko35/, theLahuShi counterpartofk’oʌ.Theexampleparallel to theBlack Lahu example given above is as follows; note that quotations are alsoindicatedbyquotationmarkersinwrittenLahuShi.

Darviq yawd haq “Jawdmawd”lehd koz ve …./da11 viʔ11 jɔ55 haʔ11 dʒɔ55 mɔ55 lɛ55 ko35 ve33 …./ David he OBJ lord/master QUOTE say CONDITIONAL

‘IfDavidcallshim“Lord”...’(Matthew22:45)

2.2AkhaThe SouthernNgwi language Akha has a cognate clause-final hearsay evidential jev /dʑe55/whose function partly resembles those of Lisu jo: and Lahu cev but can also embed actualdirectquotes.Akha,evenmoresothanLisu,hasdevelopedaverylargesystemofclause-finalevidential markers of which this is now just one of many forms (Kya Heh 2003). Exactquotationsaremoreoftenfollowedbylehv/lɛ55/,thenaverbofspeakinginanormalclause.Quotationsareoftenprecededbynyav-eu/ɲa55ɤ33/beforethequote,whetherfollowedbyjevorby lehv.Theprequote framemaycontainonlyaSpeakernominal,ormaybea fullclausewith a speaking verb. Notice the similarity of nyav-eu to the spoken Lisu pre-quotetopic/conditional marker ny which may also be cognate; and functionally to the Lahupostnominal prequote topicmarker lehv. Note also the similarity of form of the Lahu topicmarker, theAkhapostquotemarker lehv and theLahuShipostquotemarkers lehdand lehz.Thefollowingisanexample.

Ngav ma nyav-eu lav awv jev./ŋa21 ma33 ɲa55 ɤ33 la55 ɔ21 dʑe21/ my motherPREQUOTE come IMP POSTQUOTE

‘Mymothersays“Come!”’(KyaHeh2003:179,sentence5.3)

As in Lisu, Akha quotes are often framed by a preceding prequote clause, sometimesspecifyingSpeakerand/orAddresseeaswellasaspeakingverb,withorsometimeswithoutafollowingnyav-euasinthefollowingexample.Directquoteswithafollowinglehvarenormallyframedbyafollowingpostquoteclause,minimallyaspeakingverbwithpostverbalmarkers.Theexampleshowsthatthespeakingverbintheprequoteframeandthespeakingverbinthepostquoteframeareoftenthesame,especiallywhenthequoteislong.UnlikeLisuandLahu,whereasinglespeakingverbpredominatesinprequoteandpostquoteframes,andtheverbintheprequoteandpostquoteneednotbethesame,inAkhavariousverbsoccurfrequentlyinthis position, with the same verb before and after the quote; in Lisu and Lahu, it is veryunusualtohavethesameverbotherthanLisubAorLahu k’oʌorLahuShikozbothbeforeandafterthequote.

MawvHuiv hawv-eu MawvNyiv avhanyaʌ nmv-ahv ehvnehv-eu,/mɔ55 hɯ21 hɔ55 ɤ 33 mɔ55 ɲi55 a21 ha33 ɲaʔ21 nm55 ~~~ø55 ɛ55 nɛ21 ɤ33/

Mawbig and Mawtwo they.dual siblingOBJ tell-DEC

[longquoteoffiveclauses] lehv ehv-nehv-eu meh. /lɛ55 ɛ55 nɛ21 ɤ33 mɛ33/ QUOTE tell-DEC 3.POSITIVE

‘(He)toldthetwobrothers,MawHuiandMawNyi,[quote],hetold(them).’(MahvPoʌGuvLawvGaw&Lewis1999:2)

FormalChristianprose in theAkhaNewTestament (1968)normallyusesprecedingspeaker and/or addressee plus nyav-eu or a normal clause with verb of speaking directlyprecedingaquote,andfollowinglehvplusclauseminimallyincludingaverbofspeakingandfinalmarkersratherthanjevafterquotes.TousethesameexamplefromMatthewagain:

Davbiv avyawʌ-ahv Sahvpav lehv kuv nav,…/da11 bi55 a11 jɔʔ11 ~~~ø55 s ~~~ø11 pa11 lɛ55 ku55 na55/David he OBJ Lord QUOTE call CONDITIONAL

‘IfDavidcallshim“Lord”...’(Matthew22:45)

3.ReportedSpeechinBurmeseModernLiteraryBurmesehasaquote-embeddingmarkerhu /hu44/,asseeninthefollowingexample.

Tiu Sn\ m¥io: kui lMu: t^: Sn\ hu eKÅ qv\

tho22 shan22 mjo41 ko22 loun41 ti41 shan22 hu44 khɔ22 θi22 tho22 shan22 mjo41 go22 loun43 di31 shan22 hu44 khɔ22 ði22 that paddy kind OBJ round beat paddy quote call REALIS

‘Thatkindofriceiscalledloundirice(brownrice).’(Okell&Allott2001:250)

Thishu isderivedfromacontractionofthearchaicverbh /hu22/‘say’followedbyj /jwe44/whichisaclauselinker(Okell&Allott2001:250-251),togethermeaning‘havingsaid’.Itcanembedsinglewordsorentireclauses.Thecontractionhasthesegmentsof the firstsyllableand the tone of the second. The sequence h j /hu22 jwe44/ also occurs in its originaluncontractedforminmoreformalliteraryBurmese.Asafullverb,h wasalreadyinfrequentin the earlynineteenth century (Judson1852 (1966):1058) anddoesnotoccur inmodernspokenBurmeseatall;itisinfrequentandarchaiceveninliteraryBurmeseotherthaninthisconstruction and as first element in a few similar frozen postposition forms (MyanmarLanguageCommission1993:528-529).

Spoken Burmese useslui≥ /lo44/ plus a main verb clause of speaking (Okell & Allott2001:209)asanembeddedquotemarker; sometimesother typesofverbsmayoccur, andsometimesthefollowingclausemaybeomitted,asinthefollowingexample.

s√\: sa: AMu: my\ lui≥ e®pa ty\၊၊

sin41 sa41 oun41 mɛ22 lo44 pjɔ41 tɛ22 sin44 za41 oun41 mɛ22 lo44 pjɔ41 dɛ22 think.over would IRREALIS QUOTE say REALIS

‘(He)saidthathewouldthinkitover.’(Okell&Allott2001:209)

This can perhaps be connectedwith the nominallui /lo22/ ‘way/manner’; but probably notwith the verblui /lo22/ ‘want’. There is also a homophonous clause-final markerlui≥ /lo44/‘because’(usuallybutnotalwaysinanonfinalclause),alsousedasalinkerbetweenverbsinsequence.Forclause-finallui≥ ‘because’innonfinalclauses,thereisanalternativespokenformmui≥ /mo44/whichcanoccuraloneorsometimesinsequencebeforeorafterlui≥‘because’(Okell&Allott 2001: 156-157). Myanmar Language Commission (1993: 439) suggests that lui≥ ofquotation occurs only before four verbs, e®pa /pjɔ41/ ‘speak’,Sui /sho22/ ‘say’, eKÅ /khɔ22/ ‘call’andt∑c\ /twin22/ ‘be known as’. However, themodern use oflui≥ of quotation ismuchmoregeneral:itoccursbeforeamuchwiderrangeofverbs,includingsomewhicharecognitionorperception rather than speaking. TheMyanmar Language Commission dictionary indicatesthat the literarymarkeralsooccurscombinedwithlui≥ inaquotationmarkerhului≥ /hu44 lo44/(ibid.529);thismaybeatransitionalformbetweenliteraryandspokenusage.

TheotherspokenBurmesequotationmarkeristhedirectquotemarkert. /tɛ44/afteraquotationwithoutafollowingspeakingverb,mostoftenembeddingasinglewordorshort

quotation, but also longer direct quotations (Okell & Allott 2001: 76-77). There is also ahomophonous spoken realis relative clause marker t. derived from the realis final clausemarkerty\ /tɛ22/.

eP eP n´≥ m luik\ K¥c\ eta. B¨: t´.၊၊

phe22 phe22 nɛ44 mə laiʔ5 tɕhin22 tɔ44 phu42 tɛ44 phe22 phe22 nɛ44 mə laiʔ5 tɕhin22 dɔ44 bu42 tɛ44/dɛ44 father WITH NEG follow want YET NEG2 QUOTE

‘(He)said“(I)don’twanttogoalongwithDad”.’(Okell&Allott2001:76)

d eAak\ m˙a eta. pui: Ass\ t´.၊၊

di22 auʔ5 hma22 tɔ44 po42 əsiʔ5 tɛ44 di22 auʔ5 hma22 dɔ44 po42 əsiʔ5 tɛ44 this below LOC YET silk real QUOTE

‘Downatthebottomhereitsays“realsilk”.’(Okell&Allott2001:76)

Before either of these spoken Burmese quotation markers, a preceding clause-finalrealis/irrealis marker can be and often is reduced fromty\/my\ /tɛ22/, /mɛ22/ toq/m /θə/,/mə/ as in in questions.When askingwhether someone said something or what someonesaid, thequotationmarker canbe followedby the appropriate yes/noquestionmarkerla:/la42/orthesubstancequestionmarkerl´/lɛ41/.

eragå Ae®K Aen By\ lui Tc\ ®mc\ q t´. l၊၊

jɔ41ga22 ətɕhe22 əne22 bɛ22 lo22 thin22 mjin22 θə tɛ44 lɛ41 jɔ41ga22 ətɕhe22 əne22 bɛ22 lo22 thin22 mjin22 ðə dɛ44 lɛ41 disease progress which way thinksee REALIS QUOTE QUESTION

‘Whatdoeshesayhethinksabouttheprogressofthedisease?’(Okell&Allott2001:76)

Noticethatallthesingle-syllableBurmesequotation-embeddingformshavethecreakytone, which could here be seen as marking grammatical subordination as the creaky toneoftendoes.ThedifferencesbetweenliteraryandspokenBurmeseareindicativeoftherapidchangepossibleinthegrammaticalmarkingofreportedspeech:theliterarylanguage,whichpresumably more or less reflects the spoken language of 700 or so years ago when thelanguage was standardised, has a completely different form from the spoken language.Furthermore, thespoken languagehas twoalternative forms fordirectand indirectquotes,while the literary language has only one, with uncontracted and contracted stylisticalternatives.4.ConclusionAshasbeenshown,thesyntaxofquotationandhearsayiscomplexinLisuandotherBurmiclanguages, and themarkers for embedding thismay show relatively rapid change throughtime,aswehaveseenbycomparingdifferentgenresofLisuanddifferentdialectsofLahuandobservingthedifferencesbetweenliteraryandspokenBurmese.Lisuhasbothaclause-final

hearsayevidentialandadistinctembeddedquotationstructure.Thehearsayevidentialhaspossible hearsay evidential cognates in otherNgwi languages, but as is often the casewithclause-final markers, the phonological correspondences are not regular. Lisu embeddedquotationsareoftenframedbyprecedingand/orfollowingclausesorpartialclauses,astheyalsocanbeinLahu,AkhaandBurmese.Themaximalprecedingclausespecifiessomeothernominal elements such as time and place, the speaker, the addressee, and then gives aspeakingverbwithfollowingverbalmarkers;themaximalfollowingframemayagaincontainsomenominalelementsbutveryoftenisjustaspeakingverbwithfollowingverbalmarkers.The syntactic patterns in Lisu, Lahu, Akha and Burmese differ in subtle ways, but aregenerallysimilar.

ThereareinterestingepistemicnuancesofmeaningwhichdifferentiatetheLisu,Lahuand Akha hearsay markers. In Lisu, the hearsay marker implies less commitment of thespeakertotheaccuracyandtruthvalueofthereportedstatementthanaquotationframedbya prequote and/or postquote sequence. In Lahu, the hearsay marker implies substantialuncertaintyanddoubt,while inAkha, thesemanticdifferencebetweenhearsayand framedquotationisminimal.

TheLisuhearsayevidentialjo:isprobablycognatewithLahucevandAkhajev,thoughthe rhyme and tone correspondence is somewhat irregular: the initial correspondenceregularlyreflectsProto-Ngwi(Proto-Loloish)*j,therhymecouldbefrom*aybutisirregularinLisu, and the tone correspondence reflects*2 inLisuandLahu,but*1 inAkha (Bradley1979a etymon 861). Conversely, the phonetically similar Burmese and Lahu quotationmarkerst. andtehʌareprobablynotrelated;Lahutehʌ‘real’appearstobeaShanloanword,Burmese t´. is a recent spoken form derived from the final realis marker ty\, and thesimilarityoftheformsisprobablyaccidental,thoughcontactcouldalsobeafactor.Additionalcognatesmaybe foundamong theothermarkersused: the subjector topicmarker*C-lay2(Bradley1979aetymon847),thetopicmarker*nya1andthesecondarytopicorconditionalmarker *na2. (Bradley 1979a etymon 860). These are redeployed in different ways indifferentrelatedNgwilanguagestoembedquotations.

ReferencesBradley, David. 1979a. Proto-Loloish. Scandinavian Institute of Asian Studies MonographSeries50.London:CurzonPress.

Bradley,David.1979b.LahuDialects.Canberra:AustralianNationalUniversityPress.Bradley, David. 2010. Evidence and certainty in Lisu.Linguisticsof theTibeto-BurmanArea33/2:63-84.

Bradley, David, Byabe Loto & David Ngwaza. 2008. li-su Ko; xU: mu: gw: [Lisu New Yearsong,inLisuandEnglish].ChiangMai:ActscoPublishing.

G’uiv sha ve Liʌ Hpuv Awv Suh- [God’s New Testament, in Lahu]. 1962. Rangoon: The BibleSocietyofBurma.

JavLiʌYawvShuiʌ[NewTestament,inAkha].1968.Rangoon:TheBibleSocietyofBurma.Judson, Adoniram, 1852. Burmese-English Dictionary. Moulmein: American Baptist Mission,editedbyE.A.Stevens.1893,enlargededition,editedbyRobertC.Stevenson.1918,revisedand enlarged by F. H. Eveleth. 1953, centenary edition, edited by F.G. Dickason. 1966,secondprinting.Rangoon:BaptistBoardofPublications.

KyaHeh,Noel. 2003.ADescriptiveStudyofAkhaSentenceFinalParticles.MAThesis, PayapUniversity,ChiangMai.

LadhurSiLiqPhuAwgSuhx[LahuShiNewTestament]2015.ChiangMai:WycliffeThailand.MahvPoʌGuvLawvGaw&PaulLewis.1999.AvkavDawvmawvdawvoev[Akhalegends,storiesandmyths,inAkha].ChiangRai:DevelopmentandAgriculturalProjectforAkha.

Matisoff, James A. (1973) 1982. The Grammar of Lahu. . Second printing. University ofCalifornia Publications in Linguistics 75. Berkeley, Los Angeles, London: University ofCaliforniaPress.

Myanmar Language Commission. 1993. ®mn\ma-Agçlip\ABiDan\, Myanmar-English Dictionary.Yangon:MinistryofEducation,UnionofMyanmar.

Ngwaza,David.ND.mu: gw: d: m [BigSong,inLisu].Manuscript.Okell, John & Anna Allott. 2001. Burmese/Myanmar Dictionary of Grammatical Forms.Richmond,Surrey:Curzon.

si,-Ju-x (史富相).1995.li-su xU: cH mi: [HistoryoftheLisugroup,inLisu].Fugong:Author. wu-s si: xy To L KL: xU; xU; be mu: gw_ be Vo: jEY [God’sNewTestament, Psalms andProverbs,redletteredition,inLisu].1999.Noeditor,placeorpublishe

rindicated.