Xixia Qiangic-Gong Festo

7/28/2019 Xixia Qiangic-Gong Festo

1/27

1

""""BBBBrrrriiiigggghhhhtttteeeennnniiiinnnngggg"""" aaaannnndddd tttthhhheeee ppppllllaaaacccceeee ooooff f f XXXXiiiixxxxiiiiaaaa ((((TTTTaaaannnngggguuuutttt))))iiiinnnn tttthhhheeee QQQQiiiiaaaannnnggggiiiicccc bbbbrrrraaaannnncccchhhh ooooff f f TTTTiiiibbbbeeeettttoooo----BBBBuuuurrrrmmmmaaaannnn****

James A. Matisoff University of California, Berkeley

1.0 Introduction

Xixia (Tangut) is an extinct Tibeto-Burman language, once spoken in theQinghai/Gansu/Tibetan border region in far western China. Its complex logographic script,invented around A.D. 1036, was the vehicle for a considerable body of literature until it graduallyfell out of use after the Mongol conquest in 1223 and the destruction of the Xixia kingdom.1

A very large percentage of the 6000+ characters have been semantically deciphered andphonologically reconstructed, thanks to a Xixia/Chinese glossary, Tibetan transcriptions, andmonolingual Xixia dictionaries and rhyme-books. Thefanqi\e method of indicating thepronunciation of Xixia characters was used both via other Xixia characters (in the monolingualdictionaries) and via Chinese characters (in the bilingual glossaryPearl in the Palm , where Chinesecharacters are also glossed by one or more Xixia ones).

Various reconstruction systems have been proposed by scholars, including M.V.Sofronov/K.B. Keping, T. Nishida, Li Fanwen, and Gong Hwang-cherng. This paper reliesentirely on the reconstructions of Gong.2

After some initial speculations that Xixia might have belonged to the Loloish group of Tibeto-Burman languages, scholarly opinion has now coalesced behind the geographicallyplausible opinion that it was a member of the "Qiangic" subgroup of TB. The dozen or so Qiangiclanguages, spoken in Sichuan Province and adjacent parts of Yunnan, were once among the mostobscure in the TB family, loosely lumped together as the languages of the Western Barbarians(Xifan = Hsifan). This has all changed in the past quarter of a century, as the Qiangic group has*Previous versions of this paper were presented at the Tenth Spring Workshop on Theory and Method inLinguistic Reconstruction (Ann Arbor, March 26, 2004); then (in Chinese) at the Third Cross-Straits Workshop onTibeto-Burman Languages and Linguistics, City University of Hong Kong (April 17, 2004); and finally (inFrench) at the University of Paris-3 (June 1, 2004). I am grateful for helpful comments made on these occasions,especially by Terrence S. Kaufman (in Ann Arbor), Gong Hwang-cherng (in Hong Kong), and Guillaume Jacques(in Paris).

1See, e.g. Nishida 1967, Grinstead 1972.2In 1999, Professor Gong kindly provided me with a 29-page list of his Xixia reconstructions, correlated to theQiangic etymological sets I had presented in my Taipei talk (1999). It is a pleasure to dedicate this paper to him.


2/27

2

been subjected to intensive fieldwork. mostly by Chinese scholars (e.g. Sun Hongkai, HuangBufan, Dai Qingxia, Sun Tianshin).

The Qiangic languages are of great interest, both phonologically and grammatically. Theytypically have complex systems of initial consonants and vowels, but rather rudimentary tone

systems.3

Grammatically they have systems of markers in the VP which agree with the subjectand/or object of the clause (although these are not as elaborate as, e.g. in the Kiranti group of E.Nepal). These languages are further characterized by well-developed sets of verbal prefixes whichmark the directionality of the verbal event.

Proto-Qiangic remains to be reconstructed. It is already clear that rGyalrong (= Gyarung =Jiarong) and Ergong (= Daofu = Stau) belong together in a separate subgroup of this family.They have preserved PTB prefixes and thus have especially complicated systems of initialconsonants, and also preserve final consonants better than the other languages. Yet their systemsof directional prefixes seem to indicate that they belong somewhere in the Qiangic group.

This paper attempts to confirm the impressionistic insight that Xixia was a Qiangiclanguage by comparing reconstructed Xixia forms with etyma from the modern Qiangic languages.There is, to be sure, a methodological problem involved in comparing a reconstructed language thatreflects an |etat de langue of some 800 years ago with modern spoken languages. Furthermore, themodern data seems sometimes to be overtranscribed due to incomplete phonological analyses,especially where the vowels are concerned.

Still, I believe that by focussing on a particularly striking phonological tendency that Xixiashares with the modern Qiangic languages it is possible to demonstrate conclusively that Xixia wasindeed a member of the Qiangic group.

22 2 2 ....00 0 0 TT T T hh h h ee e e ff f f aaaatt t t ee e e oo o o ff f f PP P P TT T T BB B B ** * * -- - - aaaa ii i i nnnn QQ Q Q ii i i aaaannnngg g g ii i i cc c c

****----aaaa is by far the best attested rhyme in TB languages. There is a strong tendency for thisrhyme to be raised and fronted to ----iiii, both in Xixia and in modern Qiangic languages. Thisdevelopment is typically impeded by velar/postvelar initials, and may be complicated by a numberof other conditioning factors (including sibilant, nasal, and wawated consonants). Several etymashow exceptional developments that cannot yet be explained, but overall there is a rather satisfyingdegree of regularity.

It was Tatsuo NISHIDA who first observed that Proto-Tibeto-Burman (PTB) ****----aaaa usuallydeveloped into Xixia ----iiii, and who pointed out that the same tendency was to be found in Tosu, alanguage known from a bilingual Chinese 18th century source, that Nishida claimed to be the linealdescendant of Xixia.4

3The Qiang language itself has both non-tonal and tonal dialects. See Evans 2001.4See Nishida 1973, 1976. This work, theTosu Yi-yu , was part of a series of bilingual vocabularies preparedunder the general rubric of Xifan Yi-yu (Vocabularies Translated from Western Barbarian Tongues).


3/27

3

Strange as the development of ****----aaaa > ----iiii may seem in the Tibeto-Burman context, it is by nomeans unparalleled in the world's languages. Terrence Kaufman tells me (p.c. 2004) that in the19th century a similar phenomenon in Indo-European languages like Frisian was calledbrightening Since this is a concise and euphonious term, I have adopted it in this paper.

In the following cognate sets, forms from modern Qiangic languages are taken from Sunand Huang 1991 ("ZMYYC"), except for the Pumi Dayang data, which are from my ownfieldwork. As mentioned above, the Xixia reconstructions are by Gong (1999). PTB forms arefrom Matisoff 2003 ("HPTB").

Qiangic language name abbreviations: QQQQNNNN Northern Qiang (Mawo) QQQQSSSS Southern Qiang (Taoping)PPPPTTTT Pumi Taoba PPPPQQQQ Pumi QinghuaPPPPDDDD Pumi Dayang RRRRGGGG rGyalrongEEEEGGGG Ergong (=Stau = Daofu) MMMMYYYY Muya (=Minyag)QQQQYYYY Queyu GGGGQQQQ GuiqiongEEEESSSS Ersu NNNNMMMM NamuyiSSSSXXXX Shixing XXXXXXXX Xixia

22 2 2 ....1111 EE E E tt t t yy y y mmmmaaaa ww w w hh h h ee e e rr r r ee e e PP P P TT T T BB B B ** * * -- - - aaaa hh h h aaaass s s bb b b ee e e cc c c oo o o mmmmee e e XX X X xx x x -- - - ii i i (( ( ( oo o o rr r r ww w w hh h h ee e e rr r r ee e e ,,,, ii i i nnnn tt t t hh h h ee e e aaaabb b b ss s s ee e e nnnncc c c ee e e oo o o ff f f aaaa XX X X ii i i xx x x ii i i aaaa cc c c oo o o gg g g nnnnaaaatt t t ee e e ,,,, oo o o tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c ll l l aaaannnngg g g uu u u aaaagg g g ee e e ss s s hh h h aaaavv v v ee e e -- - - ii i i )) ) ) :: : : 5

There are dozens of examples:6

PP P P TT T T BB B B XX X X ii i i xx x x ii i i aaaa

(1) AXE ****rrrr----pppp aaaa wwww jj j jiiii 7

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i PQ ssssppppyyyy EG llllvvvviiii GQ ppppiiiizi SX bbbbiiiimi

For similar Xixia reflexes see PATCH (22) and SNOW (30).

5I am using "-i" in a loose way to refer to any front or apical vowel, including [-y, -, -e, -,]. Apical vowelreflexes are especially characteristic of syllables with sibilant initials (see below).6At least 37 of these have good Xixia cognates. Eighteen of them are cited in Matisoff 2003 (HPTB): 172. Tenof them I had not identified appear in the rGyalrong/Xixia comparisons of G. Jacques (2003), and have beenadded to the present version of this paper (see 36-45 below).7The subscript dot represents a hypothetical earlier prefix in Gong's system.


4/27

4

(2) BEE/BIRD ****bbbbyyyyaaaa [[[[llllaaaacccckkkkiiiinnnngggg]]]]

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : PQ bbbbt ES bbbbzzzz NM nnnnddddzzzz

The above forms mean 'bee'. G. Jacques (2003:8) cites rGyalrong (gDong-brgyad) ppppaaaa 'bird', perhaps a loan from Tibetan (WT bbbbyyyyaaaa).

(3) BORROW/LEND ****rrrr////ssss----((((yyyy))))aaaa , , ,nnnn jj j jiiiirrrr 8

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QN uuuu sa QS da iiii EG iiii MY QY t iiii GQ iiii NM nth; iiii SX

ZMYYC #692 claims that the QS, QY, and NM forms (though not the others!) areloanwords, presumably from Zeku Tibetan iiiiaaaarrrr (cf. Written Tibetan ggggyyyyaaaarrrr), but this seemshighly unlikely in view of the much better match of these forms with WT bbbbrrrrnnnnyyyyaaaa'lend; borrow' (see STC #190).

(4) CHILD/SON ****zzzzaaaa &&&&****ttttssssaaaa, ; ; ,

zzzz jj j j 'son', zzzz jj j jiiii ''''boy', zzzz jj j jiiii jj j j'male'

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QN tttt iiiimi QS ttttiiiibi PT tttt ~~~~ GQ elettttssssiiiiES jdddd

(5) CHIN ****mmmm----kkkkaaaa [lacking]

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : PT mmmmaaaakkkk PQ mmmmkkkk MY mmmmkkkkhhhh~~~~ QY mmmmkkkkhhhhiiiiGQ nnnnnnnnkkkk ES mmmmiiiixxxxiiii NM mmmmiiiikkkk SX mmmmkkkk

This is one of the rare roots to show fronting in some Qiangic languages despite the velarinitial. QS mmmm skkkkhhhhaaaaand RG tttt mmmm jj j j aaaa show the typical retention of ----aaaa after velars. Seebelow 2.3(a).

8For similar final ----rrrr's in Gong's reconstruction, see HUNDRED (15), LAUGH (16), GNAW/CHEW (47),RIGHTSIDE (56).


5/27

5

(6) COME ****llllaaaa llll jj j j ----rrrr jj j jaaaarrrr

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QN llllyyyy QS llllyyyy MY rrrriiii SX lllliiiiuuuu

(7) EARTH ****ttttssssaaaa ttttsssshhhh jj j jiiii

Several modern Qiangic languages have reflexes with ----aaaa, e.g. PT tttt aaaa, PQ tttt, RGs tttthhhh, EG s tttt aaaa, MY ttttssssaaaa, NM dddd aaaatsh). For discussion of reflexes of ****----aaaa aftersibilant initials, see EAT (next example).

(8) EAT ****ddddzzzzyyyyaaaa , ( )ddddzzzz jj j jiiii ;;;; ddddzzzz jj j joooo

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QN ddddzzzz QS dddd MY nnnnddddzzzz ES ddddzzzzNM ddddzzzz SX ddddzzzz

There is a strong tendency for apical vowels to appear after sibilant initials, both in Chineseand TB. A plausible path of development from PTB*-a after sibilant initials would besomething li\ke: ****----aaaa >>>> ---- >>>> ---- >>>> ---- >>>> ----iiii....

This etymon displays a morphophonemic variation typical of many Xixia verbs (Gong2001:60). The basic form ("Form 1") has ----iiii vocalism. Form 2, which appears beforecertain suffixes, has ----oooo vocalism. See also GET/OBTAIN (12), PUT/PLACE (23),WEAR CLOTHES (35), DO/CLOSE (39), SEND ON ERRAND (42), andSLAUGHTER (an animal) (43).

(9) EDGE ****mmmm----ddddzzzzyyyyaaaa [lacking]

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QN zzzziiii ka PT zzzziiiipe Pumi Dayang [Matisoff 1996/98] ddddzzzz [dz]MY zzzzyyyyiiii QY zzzziiiikha SX zzzzl

(10) FLESH/MEAT/ ****ssssyyyyaaaa tttt||||sssshhhh jj j jiiiiANIMAL

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : :


6/27

6

QS tttthhhh PT PQ GQ iiiiES NM SX bittttsssshhhh

This is another sibilant-initial root with apical or other central vowel reflexes in Qiangic.

(11) FORGET 9 ****mmmmaaaa----tttt mmmm jj j j

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QN rrrrmmmm QS mmmmiiii PQ th mmmm RG k jj j jmmmm ssssPT n mmmm~~~~ SX mmmm~~~~

Several forms point to a Proto-Qiangic prefix ****rrrr---- oooorrrr ****ssss---- with this root:QN rrrrmmmm QS mmmmiiii PQ th mmmm EG llllmmmmuuuuRG k jj j jmmmm ssss

RG k jj j jmmmm ssss suggests suffixal ****----ssss with this etymon, possibly more ancient than the suffixal----tttt that is set up in STC #425 on the basis of forms like Magari hhhhmmmmaaaa hhhhmmmmaaaatttt 'be lost; lose'.

Initial mmmm---- seems to favor the development of schwa. Lahu has a somewhat similarphenomenon: / / is the rarest of the 9 Lahu vowels, but it occurs with especial frequency insyllables beginning with mmmm----....

(12) GET/OBTAIN ****rrrraaaa ,rrrr jj j jiiiirrrr////,,,, rrrr jj j joooorrrr////

See Gong 2001:60. The tonal instability of this word in Xixia is perhaps related to the factthat this etymon is under the relatively rare Tone *3 of Proto-Lolo-Burmese (PLB ****rrrraaaa).

(13) GOD/SOUL/SPIRIT ****mmmm----hhhhllllaaaa ssss jj j jiiii

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : :

MY llll QY iiii GQ NM ''''god', iiii''''ssssoooouuuullll''''

Xixia nnnn jj j jaaaa 'god' might reflect PTB ****nnnnaaaatttt & ****nnnnaaaannnn 'evil spirit, demon',

(14) HANDSPAN ****mmmm----ttttwwwwaaaa [[[[llllaaaacccckkkkiiiinnnngggg]]]]

9The semantic range of this etymon includess LOSE/DISAPPEAR/EXHAUSTED/SPENT. See STC #425.


7/27

7

The medial ----wwww---- seems to have impeded brightening in this root. Most forms have a backvowel, sometimes raised to ----uuuu:10

QN tttt QS ttttuuuu RG t ttttwwwwaaaa ES ddddzzzzuuuuEG tttt MY ttttt uuuu NM ttttuuuu SX ttttuuuu

However, a few languages do show brightening even here:PQ t ttttsssshhhhyyyyiiii QY t tttt hhhhiiii GQ tetttt

(15) HUNDRED ****bbbb----rrrr----ggggyyyyaaaa jj j jiiiirrrr

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QN kkkkhhhhiiii QS tttthhhhiiii PT rrrreeee PQ iiiiEG iiii MY dddd GQ dddd NM hhhhiiiiSX d i /dddd iiii'one'/

(16) LAUGH ****rrrryyyyaaaa----tttt ,dddd jj j jiiiiiiii jj j j ;;;; rrrr jj j jiiiiiiiirrrr

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : RG ka na rrrriiii GQ iiii ES rrrr NM dddddddd

Either Xixia form is a good match for the other Qiangic forms. dddd jj j jiiiiiiii jj j j is glossed 'laugh'; rrrr jj j jiiiiiiiirrrris glossed 'sneer'

(17) LISTEN ****gggg////rrrr----nnnnaaaa nnnn jj j jiiii'listen' ;;;;mmmm jj j jiiii 'hear'

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QS t hy yyyy PT s iiii PQ th nnnniiii EG wu ssssiiiiES ba iiii NM bhhhh~~~~iiii

The Xixia doublet points up the language's tendency to change the position of articulationof initial nasals. See ILL/ACHE. below (73).

(18) MOON/MONTH ****ssss////gggg----llllaaaa aaaa----llllhhhh jj j jiiii----pppphhhhoooowwww 'a month'llllhhhh jj j jiiii'moon'

10 This is identical to the regular Lahu reflex of ****----wwwwaaaa, as in HANDSPAN ****mmmm----ttttwwwwaaaa > Lh. tttthhhhuuuu; CATTLE ****nnnnwwwwaaaa >Lh. nnnnuuuu; TOOTH/TOOTHLIKE PART ****sssswwwwaaaa > Lh. ----ssssuuuu.


8/27

8

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : 'moon' PT iiii PQ iiii MY lllln_

GQ lllliiiimo NM iiiimi SX iiiimi

'month' QS PT iiii PQ iiiiGQ lllliiii NM iiii

Several forms for 'month' reflect the ****ssss---- prefix. Cf. WT zzzzllllaaaa----bbbbaaaa, Lahu hh h h aaaa----ppppaaaa, etc. The Xixiadevelopment here is different from that in the phonologically similar etymon TONGUE(69), where the ****----aaaa has been preserved (see below).

(19) NEGATIVE ****mmmmaaaa ,mmmm jj j jiiii ;;;; mmmm jj j jiiii jj j j

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QS mmmmiiii EG mmmmiiii MY iiii GQ mmmmeeeeNM mmmm

Muya shows palatalization of the nasal, presumably after the change of ****----aaaa to ----iiii....

(20) NEG. IMPERATIVE ****ttttaaaa ,tttt jj j jiiii

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : PT tttt QY tttt GQ tttthhhh NM tttthhhh

Several forms show palatalization and/or affrication of the initial, which presumablyoccurred after the change of ****----aaaa to ----iiii:QN tttt iiii QS ttttssss EG dddd iiii MY tttt

It is interesting that both 'negative' and 'negative imperative' show the brighteningtendency, despite the fact that functors often show exceptional phonological developments.

(21) NOSE ****ssss----nnnnaaaa nnnn jj j jiiiiiiii

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QS iiiiqop EG ssssnnnniiii NM iiiinga


9/27

9

(22) PATCH ****ssss----pppp aaaa pppp jj j jaaaa;;;; wwww jj j jiiii jj j j ;;;; wwww jj j jiiii jj j j

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QN ppppaaaa QS ppppeeee PT x pppphhhhiiii PQ x sssspppphhhh

RG ka mmmmpppphhhhiiii MY pppphhhheeeel ES pppppppp

All the Xixia allofams are good matches with the modern Qiangic forms. The forms withinitial wwww---- are perfectly analogous to the Xixia reflexes of AXE (1) and SNOW (30), so theyrepresent the preferred allofam for comparative purposes.

(23) PUT/PLACE ****ssss----ttttaaaa , ,tttt jj j jiiii 'put';;;; tttt jj j joooo,tttt jj j jiiii'place'

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QS ku ttttyyyy PT kh ttttiiii PQ kh sssstttt EG ssssttttiiiiMY rtttt yyyyiiii NM t hitttthhhh SX tttt

Several forms show secondary palatalization of the initial after ****----aaaa > ----iiii. It is not clearwhich of the two syllables in the Namuyi form is the reflex of this etymon.

\

(24) RABBIT (calendrical animal) ****----llllaaaa lllliiii----oooo

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QS iiii PT tholllliiii PQ yyyyts EG wa lllleeeeMY iiiiv QY iiiiko ES middddzzzz NM tholllliiiiSX thyllllyyyy

The PT, PQ, NM, and SX binomes are very similar to Loloish forms like Lahu tttthhhh----llll\\\\aaaa, Hanitttthhhh\\\\uuuu----hhhhllll||||. This calendrical dissyllable is also found in Tai (e.g. Tai Nuea tttthhhhuuuu----llllaaaaaaaa but thefirst syllable seems to be a loan < Chinese (Mand. tttt\\\\uuuu). This is evidently an early loaninto Qiangic, since it underwent brightening.11

The development of fricatives from palatalized laterals (as in QS, MY, QY) is paralleled inWT (e.g. PTB ****bbbb----llll yyyy 'four' > WT bbbb||||zzzziiii).

(25) REST ****nnnnaaaa [[[[llllaaaacccckkkkiiiinnnngggg]]]]

11 Mu Shihua (p.c. April 2004) claims that this word is actually of Altaic origin.


10/27

10

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QS b iiii PT kh iiii PQ kh nnnniiii EG n nnnniiiiMY ji iiii ES ba iiii SX l

(26) SALT ****ttttssssaaaa ttttsssshhhh jj j j---- uuuu

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QN ttttsssshhhh QS ttttsssshhhh PT ttttsssshhhhiiii PQ ttttsssshhhhiiiiRG ttttsssshhhh EG ttttsssshhhh MY ttttsssshhhh QY ttttsssshhhhiiiiGQ ttttsssshhhhiiii ES ttttsssshhhh NM ttttsssshhhh SX ttttsssshhhh

The vocalism of all these modern Qiangic reflexes [-i, -, -, - , -] shows raising and/orcentralization, the latter due to the sibilancy of the initial.

(27) SATIATED ****kkkk----wwwwaaaa , jj j jiiiiiiii jj j j

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : PT t kkkkuuuuiiii PQ t kkkkuuuuiiii QY luwwwwiiii

This is a general TB root. Cf. WWWWBBBB wwwwaaaa....

(28) SHINE/BRIGHT PLB ****mmmm----bbbbaaaa mmmmbbbbiiiiThis Xixia reconstruction is by Nishida (1966:447). The prenasalized Xixia initialfits well with the Lolo-Burmese forms: WB ppppaaaa, Lahu bbbbaaaa (Lh. voiced obstruents descendfrom PLB *prenasalized initials).12

(29) SINEW/TENDON ****rrrr----ttttssssaaaa [[[[llllaaaacccckkkkiiiinnnngggg]]]]

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QS dddd PQ ssssttttiiiieeee QY ddddzzzziiiika NM guttttssss

However, the tendency in the modern Qiangic languages is to preserve a non-front vowelin this root:PT dddduuuu EG ttttssssaaaa MY nnnndddduuuu ES hhhhtttt

12See Matisoff 1978:17-18.


11/27

11

SX sttttssss

(30) SNOW ****ssss----pppp aaaa((((llll)))) 13 wwww jj j jiiii

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : PT pppp uuuu PQ ssssppppyyyy Pumi Dayang pppp||||

The Xixia reflex is consistent with those of AXE (1) and PATCH (22). Most modern Qiangiclanguages do not show brightening, however, either retaining the ****----aaaa---- (QS m pppp,,,, RG ti jj j jppppaaaa,EG nkh vvvvaaaa,,,, QY khawwwwaaaa) or backing it to ----uuuu or ---- under the influence of the ----wwww---- (QN t i q uuuu, MY vvvv , GQ khwwwwuuuu).

(31) SPARROW ****mmmm----ttttssssaaaa [lacking]

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QS ittttsssshhhhiiiieeee PT gutttt iiii QY ~oputtttssssiiiiGQ ttttsssshhhhttttsssshhhh NM annnndddd SX dddd yyyymi

RG pa-ttttssssaaaa shows no brightening.14

(32) TAIL ****mmmm----bbbbaaaa mmmmbbbb

This Xixia reconstruction is by Nishida (1966:464). This sparsely attested root occursin Mpi (S. Loloish) mmmmppppaaaa(Matisoff 1978:17-18).

(33) TOOTH ****ssss----wwwwaaaa ||||ssss jj j jwwwwiiii

13This etymon is reconstructed ****wwwwaaaa at the proto-Lolo-Burmese level. See HPTB pp. 46, 171, 428.14G. Jacques has suggested that the second syllable of this rGyalrong form is a diminutive suffix derived from themorpheme for CHILD (see #4 above and Mandarin ), so that the root is really the first syllable.(perhaps cognatewith WT bbbbyyyyaaaa 'bird'). Such a formation is indeed attested in several TB languages (e.g. Thadou Chin ggggiiiiiiiitttt----ttttssssaaaa'sparrow' (where the root is definitely the first syllable), and might plausibly be invoked to explain forms likePumi Taoba gggg tttt iiii'sparrow' (& tttt iiii'child') and Pumi Jiulong wwww ttttssssiiii. However, there is abundant evidencefor the existence of an etymon ****mmmm----ttttssssaaaa 'sparrow', which sometimes appears as an independent monosyllable (e.g.Written Burmese ccccaaaa; Nusu ddddaaaa; Hayu ddddzzzzuuuu; Daofu/Ergong zzzz ); sometimes as the second syllable of a binomewhere the first syllable is definitely a prefix (e.g. Jingpho \\\\uuuu----ttttssssaaaa (\uuuu---- is a Jingpho prefix occurring in manyornithonyms); Hani Liuchun xxxxaaaa ddddzzzzaaaa(xxxxaaaa---- is an animal prefix in Hani/Akha); Muya uuuuzzzz ; Yi Mile ttttoooo,Yi Mojiang ddddzzzzoooo); and sometimes as the first syllable of a binome where the meaning of the second element isclear: Lahu jj j j\\\\aaaa----mmmm (the 2nd syllable means 'tamed; not wild'); similarly, Yi Weishan ddddmmmm and Yi Nanhuadddd ____uuuu mmmmiiii. Several binomes might well combine this independent root with the diminutive suffix: Gazhuo ttttssssaaaazzzzaaaa, Guiqiong ttttsssshhhh ttttssss , Bai Jianchuan ttttssssoooottttssssiiii.


12/27

12

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : PQ yyyy GQ xxxxuuuuiiii ES ma SX

Most modern Qiangic languages retain a back vowel; some even preserve medial ----wwww---- (QSssssuuuu , RG t sssswwwwaaaa). In Xixia, the brightening tendency was not impeded by the labialsemivowel.

(34) TROUSERS ****ssss----llllaaaa ,llll jj j jiiiiiiii (n.)llll jj j jiiiiiiii'put on trousers'

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : PT PQ ssssdddd MY dddd QY dddd iiiiGQ xxxxiiii NM lllliiii SX phllllyyyy

(35) WEAR CLOTHES ****ggggwwwwaaaa ,gggg jj j jwwwwiiii;;;; gggg jj j jwwwwoooo

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : PT n ggggiiii PQ t gggguuuuiiii EG ggggiiii

The medial -w- , directly preserved in such forms as QN gggguuuu , QS gggguuuu , MY gggguuuu, preventedthe shift to ----iiii in several modern Qiangic languages, but evidently did not prevent the brighteningin PQ or Xixia.

In his recent study comparing Xixia rhymes with those of the gDong-brgyad dialect of rGyalrong, G. Jacques (2003:6-9) offers several more examples of rGyalrong ****----aaaa corresponding toXixia ----iiii:

rGyalrong Xixia((((33336666)))) AAAABBBBLLLLEEEE////CCCCAAAANNNN kkkk ssssppppaaaa wwww jj j jiiii

((((33337777)))) BBBBIIIILLLLLLLLHHHHOOOOOOOOKKKK tttt ccccaaaa dddd||||zzzz jj j jiiii

((((33338888)))) CCCCOOOOOOOOKKKK kkkk ssssqqqqaaaa jj j jiiiiiiii

((((33339999)))) DDDDOOOO////CCCCLLLLOOOOSSSSEEEE15 kkkk ppppaaaa wwww jj j jiiii & wwww jj j joooo

15 This word means 'do' in the dialect studied by Jacques, but 'close' in the closely related Japhug dialect. Cf.German machen 'do; make', aufmachen 'open', zumachen 'close'.


13/27

13

Cf. Lahu vvvvaaaa 'do, make, work'. See Gong 2001:60, where the Xixia forms are glossed, , .

((((44440000)))) FFFFIIIINNNNEEEE////TTTTHHHHIIIINNNN kkkk mmmmbbbbaaaa bbbb jj j jiiii

((((44441111)))) LLLLIIIIVVVVEEEE////DDDDWWWWEEEELLLLLLLL kkkk rrrr rrrrmmmmaaaa mmmm jj j jiiii

((((44442222)))) SSSSEEEENNNNDDDD ((((oooonnnn eeeerrrrrrrraaaannnndddd)))) kkkk xxxxpppprrrraaaa pppphhhh jj j jiiii & pppphhhh jj j joooo

((((44443333)))) SSSSLLLLAAAAUUUUGGGGHHHHTTTTEEEERRRR ((((aaaannnn aaaannnniiiimmmmaaaallll))))16 kkkk nnnntttt hhhhaaaa ||||ssss jj j jiiiiiiii & ||||ssss jj j joooooooo

((((44444444)))) WWWWEEEETTTT kkkk llllaaaa llllhhhh jj j jiiii

((((44445555)))) YYYYEEEEAAAARRRR////AAAAGGGGEEEE xxxxppppaaaa wwww jj j jiiii

22 2 2 ....22 2 2 EE E E tt t t yy y y mmmmaaaa ww w w hh h h ee e e rr r r ee e e PP P P TT T T BB B B ** * * -- - - aaaa hh h h aaaass s s bb b b ee e e cc c c oo o o mmmmee e e XX X X xx x x -- - - ee e e

In a rather miscellaneous collection of cases, Xixia has brightened TB ****----aaaa only as far as ----eeee.These etyma mostly have resonant (wwww----, llll----, rrrr----) or velar initials; yet there are several otherunimpeachable examples where velar initials have prevented Xixia brightening altogether (below2.3).

(46) BIRD ****wwwwaaaa ,wwwweeee

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - ii i i :: : : QY b~owwwwiiii GQ wwwwiiiitsi

Here the brightening did not "go all the way" to ----iiii,,,, probably because of the inhibiting effectof the wwww----. See also HOOF (48), CATTLE (53), TONGUE (69), RAIN (75).

(47) GNAW/CHEW PQc ****ssss////mmmm----kkkkaaaa kkkkiiiieeeerrrr

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - aaaa:: : : QN a QS GGGG e PQ xqqqq EG qqqqaaaa leRG (gDong-brgyad) kkkk nnnn kkkkaaaa

A couple of forms show some fronting and raising, but not to the point of becoming /i/:

16This is a distinct root from PTB ****gggg----ssssaaaatttt 'kill (in general)' > gDong-brgyad kkkk ssssaaaatttt, Xixia ssss jj j jaaaa.


14/27

14

MY tqqqql NM qqqqhhhh

The Xixia form (cited in Jacques (2003:8) means 'chew'.

(48) HOOF ****kkkkwwwwaaaa kkkkwwwweeee jj j jThe modern languages are virtually unanimous in preserving a low/back vowel in thisetymon:

QN di kkkkuuuu PT kkkkuuuuaaaaa PQ sssspppp RG ta kkkkaaaaEG kkkkoooo MY qqqquuuutsh GQ nnnngggg ES nnnnkkkkhhhhuuuuNM qqqqhhhhaaaats

The only forms showing some brightening here are SX kkkkhhhh,,,, QY ekkkkhhhh,,,, and Xixiakkkkwwwweeee jj j j, though nowhere did it proceed "all the way" to ----iiii.

(49) STRENGTH ****kkkk----rrrraaaa iiiieeee

QS dqqqq PT kkkkaaaa PQ qqqqaaaa NM maaaaaSX khua ji

No modern Qiangic language shows brightening in this root. The Xixia form descendsfrom the unprefixed root ****rrrraaaa, while in the other Qiangic languages the velar component of the initial consonant sequence predominates. See the discussion below 2.3(a).

(50) TIGER ****kkkk----llllaaaa lllleeee

This root, evidently a borrowing into TB from Mon-Khmer (cf. Mon kkkkllllaaaaaaaa), has so far beenattested mostly in Lolo-Burmese (cf. WB kkkkyyyyaaaa [Inscriptional Burmese kkkkllllaaaaaaaa], Lahu llllaaaa).

It is also attested in Qiangic, where the vowel is usually maintained as ----aaaa:ES llll NM llllaaaa SX llll....

In Ergong iiii, however, brightening has occurred (inducing palatalization/frication of theinitial). A similar fronting and raising led to Xixia lllleeee, although in this case the process didnot go so far as to result in ----iiii. These brightened forms indicate that this root must havebeen borrowed into Qiangic (perhaps from Lolo-Burmese) at an early date.17

17Other modern Qiangic languages have borrowed their word for 'tiger' from other sources:(1) from Tibetan (cf. WT ssssttttaaaagggg):

QS pzdddd; MY tttt; QY ttttaaaa; GQ tttt(2) from Chinese (cf. Mand. hhhhuuuu)


15/27

15

Several more examples of rGyalrong ----aaaa corresponding to Xixia ----eeee are to be found in G.Jacques (2003:8):

rGyalrong Xixia

((((55551111)))) AAAARRRRMMMM tttt llllaaaa kkkkiiiiwwwweeee jj j j

((((55552222)))) BBBBOOOOIIIILLLL ((((vvvv....iiii....)))) kkkk llllaaaa lllleeeeBBBBOOOOIIIILLLL ((((vvvv....tttt....)))) kkkk ssss llllaaaa lllleeee

((((55553333)))) CCCCAAAATTTTTTTTLLLLEEEE nnnn aaaa wwwweeee

((((55554444)))) DDDDIIIIFFFFFFFFIIIICCCCUUUULLLLTTTT kkkk NNNNqqqqaaaa ggggiiiieeee

((((55555555)))) FFFFRRRROOOOGGGG qqqqaaaa ppppaaaa ppppiiiieeeeThis etymon descends from PTB ****ssss----bbbbaaaallll, but evidently the final lateral dropped in Qiangic(as also e.g. in Lolo-Burmese) at an early date.

((((55556666)))) RRRRIIIIGGGGHHHHTTTTSSSSIIIIDDDDEEEE cccchhhhaaaa tttt||||ssssiiiieeeerrrr

((((55557777)))) RRRRUUUUSSSSTTTT ssssaaaa wwwweeee

22 2 2 ....33 3 3 EE E E tt t t yy y y mmmmaaaa ww w w hh h h ee e e rr r r ee e e PP P P TT T T BB B B ** * * -- - - aaaa rr r r ee e e mmmmaaaaii i i nnnnss s s XX X X xx x x .... aaaannnndd d d // / / oo o o rr r r mmmmoo o o ss s s tt t t oo o o tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c -- - - aaaa

((((aaaa)))) AAAAff f f tttteeeerrrr vvvveeeellllaaaarrrr iiiinnnniiiittttiiiiaaaallllssss

In Qiangic the fate of ****----aaaa after velars is not a simple story. Each velar-initial root seems tobehave somewhat differently, though the general tendency is clear - a velar initial impedes thefronting and raising of ****----aaaa. We have just seen a number of cases where a velar initial has permittedonly partial brightening of the Xixia form (GNAW/CHEW [47], HOOF [48], STRENGTH [49],

CATTLE [53], RUST [57]). This section presents several etyma where the velar initial hasprevented Xixia fronting and/or raising completely.

(58) BITTER ****kkkkaaaa kkkkhhhhaaaa'bitter' kkkkhhhhiiiieeee 'bitter; evil; hate'

QN uuuu; PT oooo; PQ oooo


16/27

, , , ,

16

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - aaaa:: : : QN qqqqhhhh QS qqqqhhhh PT kkkkhhhhaaaam PQ qqqqhhhhMY qqqqhhhhaaaa NM qqqqhhhhaaaa SX qqqqhhhh

The only modern Qiangic form to show brightening is GQ kkkkhhhhiiiimu. Gong considersthe Xx. form kkkkhhhhiiiieeee 'bitter; evil; hate' to reflect the independent root ****mmmm----kkkkrrrriiii----((((yyyy))))----tttt----ssss 'bile' (seeHPTB:22, 118, 189, 193, 436, 456).

(59) CROW (n.) ****kkkkaaaa [lacking]

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - aaaa:: : : EG qqqqaaaa e QY kkkkaaaalo GQ kkkkli ES kkkk

NM laqqqqaaaa

A couple of forms show raising and fronting:RG kkkkiiii MY qqqq

This root also has an imitative component.

(60) ENEMY ****dddd----ggggrrrraaaa [lacking]

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - aaaa:: : : PT ddddaaaawwwwuuuu RG tttt nnnnggggrrrr MY ddddaaaawwwwuuuu QY ddddaaaaSX dddd

These forms might all be borrowed from Tibetan (cf. WT ddddggggrrrraaaa). The Xixia binomial, ,synonym wwww jj j jiiii jj j j----zzzz jj j j shows fronting in both syllables, but neither syllable

seems cognate to the modern Qiangic forms.

(61) FIVE ****bbbb////llll----aaaa wwww

OO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - aaaa:: : : QN uuuu QS uuuu PT uuuu PQ uuuu~~~~RG k mmmmoooo EG wwwwuuuu MY QY uuuuaaaat ~aGQ ~~~~ ES uuuuaaaa NM aaaa SX ~~~~(ko)


17/27

17

The modern Qiangic languages mostly preserve PTB ****----aaaa in this root. Xixia behavessimilarly, in that the vowel is not fronted, though it is raised somewhat to schwa. This rootshould probably be set up as **** aaaa at the Proto-Qiangic level.

(62) GOOD ****nnnnaaaa aaaaThis root is not widespread in Qiangic. In one of the three modern languages in which it isattested, PTB ****----aaaa is preserved as such (QN nnnn, QS nnnn). The expected brighteningoccurs in NM nnnn, and most interestingly in EGiiii, where the nasal is backed from dentalto velar. In Xixia a similar backing occurred, but apparently with the opposite effect - i.e.the velarity of the nasal blocked the brightening. For a similar development, seeILL/ACHE (73).

(63) I / ME ****aaaa aaaaOO O O tt t t hh h h ee e e rr r r QQ Q Q ii i i aaaannnngg g g ii i i cc c c rr r r ee e e ff f f ll l l ee e e xx x x ee e e ss s s ww w w ii i i tt t t hh h h -- - - aaaa:: : : QS RG aaaa EG MY QY aaaa NM aaaa SX

The only modern form to show any fronting is GQ.

((((bbbb)))) AAAAff f f tttteeeerrrr nnnnoooonnnn----vvvveeeellllaaaarrrr iiiinnnniiiittttiiiiaaaallllssss

There are very few good examples in this category, the best being HOT and TONGUE(and even these can be explained away).

(64) BOX / CABINET ****ddddaaaa [[[[llllaaaacccckkkkiiiinnnngggg]]]]

This form is sparsely attested in Qiangic, always with a low/back vowel:Pumi Dayang tttt|||| PT ttttaaaa SX tttt

Extra-Qiangic forms include Lahu ttttaaaa----qqqqoooo, Naxi tttt, Tujia tttthhhhoooo, Karen dddd____.

[64a] CARRY ****bbbbaaaa ,wwwwaaaaOnly one modern Qiangic form shows brightening: QN bbbbiiii. Others maintain the original****----aaaa: QS bbbbaaaa,,,, ES bbbbaaaa, SX pppppppp. Still others reflect a different etymon with the samemeaning, PTB ****bbbb wwww: EG mmmmbbbb v, GQ bbbbuuuu, NM bbbbbbbb.


18/27

18

This example is spurious, however, at least as far as Xixia is concerned. As G. Jacquespoints out (2003:10) the Xixia form really means 'shoulder; carry on the shoulder',and seems directly cognate to rGyalrong tttt rrrrppppaaaa'shoulder', kkkk mmmm rrrrppppaaaa'carry on theshoulder' (< PTB ****rrrr----ppppaaaakkkk).

(65) DOG ****nnnnaaaa nnnnaaaa

This is quite a rare root in TB, although it does occur both in Qiangic and Loloish.Loloish reflexes include Lisu nnnn and Yi Nanhua nnnn____uuuu(ZMYYC #119).Qiangic forms include Shixing kkkkhhhhuuuu iiiiand rGyalrong kkkkhhhh nnnnaaaa [ibid .], with theformer showing brightening.

By far the most widespread TB root for DOG is ****kkkk yyyy, which underlies the firstsyllables of the Qiangic forms just cited. Lisu also has another word for DOG re-flecting this root, kkkkhhhh.

(66) HAMMER ****ssss----ttttaaaa [[[[llllaaaacccckkkkiiiinnnngggg]]]]

As with the preceding dental-initial etymon (BOX/TRUNK), there is virtually nobrightening tendency with this root, e.g.:

PT x ttttaaaa PQ x sssstttt MY ttttssssaaaa GQ uttttNM nnnntttthhhhaaaa

However, a couple of forms with affricated initials (perhaps arising from metathesis of theprefix and the root-initial) have developed front vowels:QS tttthhhhiiii EG nnnnddddzzzz

(67777) HOT ****ttttssssaaaa----tttt ttttssss jj j jaaaa ;;;; ttttsssshhhh jj j jaaaaA number of modern Qiangic languages have raised and fronted the vowel in this root:

RG k ssssttttssss EG wu ttttssss MY tsttttssss NM ttttsssshhhh

SX ttttssssiiii

Xixia does not show brightening after this sibilant initial, unlike its behavior inCHILD/SON (4), EARTH (7), EAT (8), FLESH/MEAT/ANIMAL (10), and SALT (26).


19/27

19

Several languages point to a dental suffix with this root (e.g. WT ttttsssshhhhaaaa 'hot, illness', ttttsssshhhhaaaadddd----ppppaaaa 'heat; fever'; Lushai ||||ssssaaaa ||||ssssaaaatttt 'hot'), and it seems very likely that the Xixia form reflectsthe rhyme ****----aaaatttt. See also ****gggg----ssssaaaatttt 'kill' > Xixia ssss jj j jaaaa, VOMIT ****mmmm----ppppaaaatttt > Xixia wwww jj j jaaaa. A similarexplanation works for the superficially irregular form in Pumi Dayang (below 3.0).

(68) MOTHER ****mmmmaaaa mmmm jj j jaaaa

No modern Qiangic language shows raising or fronting with this root, and here Xixiaagrees with them all:

QN a mmmmaaaa QS mmmmaaaa;;;; mmmmmmmm PT mmmmaaaa PQ mmmmRG t mmmmoooo EG a mmmm MY mmmm QY ammmmaaaaGQ mmmm ES ammmmaaaa SX mmmmmmmm

We are here dealing with a linguistic universal, however, so this set need not be taken as aserious exception to the brightening tendency.18 As in HOT, Xixia has developed medial---- jj j j---- in this root.

(69) TONGUE ****ssss----llllyyyyaaaa llllhhhh jj j jwwwwaaaa

Many modern languages do develop brightening in this etymon. In several (QS, EG, GQ,SX) the initial has been palatalized/fricativized, leading to an apical vowel:

QS zzzzq PT iiii PQ iiiieeeeqho EG vvvv GQ dddd NM ji SX

Somehow Xixia developed a medial ----wwww---- here, which might have acted to preserve theoriginal vowel quality of the etymon. This did not happen in the very similar rootsGOD/SPIRIT (13) or MOON/MONTH (18). It is interesting that Pumi Dayang alsohas an exceptional reflex in this root (below 3.0).

This root has many other allofams, and it is very possible that the Xixia form descendsfrom one of the others, including ****llllaaaayyyy, ****lllleeeeyyyy, ****llllyyyyaaaakkkk,,,, or ****llllyyyyaaaammmm. See Matisoff 2003:672.

22 2 2 ....44 4 4 EE E E tt t t yy y y mmmmaaaa ww w w hh h h ee e e rr r r ee e e PP P P TT T T BB B B ** * * -- - - aaaa >> > > XX X X ii i i xx x x ii i i aaaa bb b b aaaacc c c kk k k vv v v oo o o ww w w ee e e ll l l

18See Jakobson 1961, "Whymama and papa ?" Cf. also FATHER (82).


20/27

20

We are left with a residue of etyma where Xixia has developed a non-low back vowel from PTBetyma with ****----aaaa. Most of these also have velar and/or nasal initials:

(70) DITCH ****kkkkaaaa gggg jj j juuuu

This root usually appears as the second element in compounds where the first constituentmeans 'water', e.g. Lahu \\\\ ----qqqqhhhhaaaa (\\\\ 'water').

A couple of modern Qiangic reflexes show raising and fronting (EG qqqquuuu, NM qqqqhhhh),but PQ t qqqqhhhhaaaamaintains the proto-vowel, while QS tsuuuuushows raising but nofronting. QN ts gggguuuuiiii and Xixia gggg jj j juuuu are diphthongal forms with both a high back and ahigh front element, which seem to illustrate well the tension between the opposingtendencies of aaaa----brightening and velar "darkening".

(71) EAR ****gggg////rrrr----nnnnaaaa nnnn jj j juuuu----dddd||||zzzz jj j jwwwwoooo

Most of the modern languages preserve ****----aaaa in this root. The brightened forms are:QS iiiikie MY yyyyiiii QY nnnneeeepoto

The high back vowel in the Xixia form is unexplained. As a wild guess, we might claimthat this abnormal development was due to a desire to avoid homophony with NOSE(21). The second Xixia syllable means 'hole'.

(72) FISH ****yyyyaaaa ||||zzzz jj j juuuu

It is not clear whether the Qiangic forms really descend from PTB ****yyyyaaaa, or whether we aredealing with a separate root reconstructible as Proto-Qiangic ****ddddzzzzaaaa.The modern languages display either back or apical vowels:

QN ddddzzzz QS ddddzzzz PT dddd iiii PQ dddd RG tu jj j joooo EG a jj j j MY uuuu QY ~~~~uuuu

GQ tttt i ES zzzzuuuu NM zzzzuuuu SX oooo

The Xixia form has a high back vowel, and is very similar to the RG, EG, QY, ES, andNM forms. The nasalization in the QY form is unexplained.

(73) ILL/ACHE ****nnnnaaaa ,oooo


21/27

21

Several modern languages show the expected brightening:

PT iiii PQ MY eeee QY eeeeES iiii

This is a most interesting case. The PTB initial is not a velar, but rather a dental nasal.Xixia, however, apparently first changed the initial to the velar nasal, which then acted toprevent the raising and fronting to ----iiii, and in fact induced the opposite effect of backing.This is very similar to what happened with GOOD (62). Muya also changed theposition of the nasal, but developed a front vowel anyway.

(74) LOVE / WANT ****mmmm----ddddzzzzaaaa , ,ddddzzzzuuuu,,,, ddddzzzz jj j juuuuThe Qiangic and Loloish languages, as well as Tibetan, reflect either of two distinct roots

with this meaning:

****mmmm----ddddzzzzaaaa > WT mmmmddddzzzzaaaa----bbbbaaaa 'love'; Xixia ddddzzzzuuuu,,,, ddddzzzz jj j juuuu; GQ tttthhhhgi; NM ddddaaaa; QNtttt iiii;SX ttttsssshhhhiiii(QN and SX show brightening); Yi Dafang nnnndddd uuuu, Yi Nanhua _e dddd ____

****mmmm////dddd----ggggaaaa > WT ddddggggaaaa----bbbbaaaa 'like, be willing, intend, wish'; QY ggggaaaa'love'; ES gggg; PQssssggggiiii; PT iiii; EG ggggiiiiaaaa ; Lahu ggggaaaa 'want; desiderative particle'; Yi Xidegggguuuu'love'; Hani Dazhai ggggaaaa

The interesting Guiqiong binome tttthhhhggggiiiiseems to reflect ****ddddzzzzaaaa----ggggaaaa, a sequence of both

roots, with brightening of the second.

(75) RAIN ****gggg////rrrr----wwwwaaaa ddddzzzz jj j juuuu

The modern Qiangic forms (except for SX) reflect the velar prefix. Of these, PT, PQ, andSX show brightening:

PT gggguuuuiiii PQ gggguuuuiiii PPPPDDDD ggggwwww|||| QY xxxxuuuuES gggguuuu SX uuuuiiiiz

The Xixia reflex is very similar to that in LOVE/WANT, but different from that of thephonologically similar WEAR CLOTHES (35). The cognacy of the Xixia form tothose in the modern languages is not certain.19

19 J.P. Evans (p.c. 2004) cites Mianchi Qiang mmmm\\\\ , analyzing the mmmm---- as a prefixization of the root ****mmmm wwww 'sky;meteorological phenomenon'. The root-initial points to the alternatively prefixed prototype ****rrrr----wwwwaaaa (cf. WrittenBurmese rrrrwwwwaaaa).


22/27

22

3333....0000 TTTThhhheeee ff f f aaaatttteeee ooooff f f PPPPTTTTBBBB ****----aaaa iiiinnnn DDDD\\\\aaaayyyy||||aaaannnngggg PPPPuuuummmmiiii20

That Xixia is definitely to be considered a Qiangic language seems especially obvious when

we consider the fate of ****----aaaa in Dayang Pumi (= Prinmi), a dialect on which I conducted a month'sfieldwork in 1996. Not only is the brightening tendency very much in evidence in Dayang, but theconditioning for the development is strikingly similar to that in Xixia with respect to the inhibitingeffect of velar initials. The most frequent Dayang reflex of PTB ****----aaaa is ----iiii, but in many cases(especially after palatal or labial initials), Dayang has developed ---- . While the developments inevery single etymon are not entirely predictable, at least in the present state of our knowledge, thegeneral tendencies are clear enough.

PP P P TT T T BB B B DD D D \\ \ \ aaaa yy y y || | | aaaannnngg g g PP P P uu u u mmmmii i i ****----aaaa >>>> ----iiii(3) BORROW/LEND ****rrrr----((((yyyy))))aaaa &&&&****ssss----((((yyyy))))aaaa d -nnnn(15) HUNDRED ****rrrr----ggggyyyyaaaa ||||(17) LISTEN ****----nnnnaaaa th -nnnn(18) MOON ****ssss////gggg----llllaaaa ||||(18) MONTH ****ssss////gggg----llllaaaa ||||(25) REST ****nnnnaaaa kh -nnnn||||(26) SALT ****ttttssssaaaa ttttsssshhhh(35) WEAR CLOTHES ****ggggwwwwaaaa ggggwwww(71) EAR ****gggg----nnnnaaaa &&&&****rrrr----nnnnaaaa |||| -d|o

(73) ILL/ACHE ****nnnnaaaa ||||(76) BUCKWHEAT cf. PLB ****gggg----rrrraaaa 21 (Taoba t~otttt iiii, Jinghua t~au tttt )(77) RICE ****ddddzzzzaaaa 22 ddddzzzz|||| cooked rice

****----wwwwaaaa >>>> ----iiii(14) HANDSPAN ****mmmm----ttttwwwwaaaa tttt hhhhwwww||||(27) SATIATED ****kkkk----wwwwaaaa (cf. WB wwwwaaaa) kkkkwwww(75) RAIN ****rrrr----wwwwaaaa &&&&****ssss----wwwwaaaa &&&&****gggg----wwwwaaaa ggggwwww||||

****----wwwwaaaa >>>> (1) AXE ****rrrr----pppp aaaa pppp||||(30) SNOW ****ssss----pppp aaaallll; cf. PLB ****wwwwaaaa pppp||||

20This section is adapted from Matisoff 1996/1998 and 1999.21Cf. Lahu aaaa, Hani aaaa, Lisu gggguuuuaaaa.22Cf. Lahu cccc\\\\aaaa 'paddy', Wancho ttttzzzzaaaa, Newar jj j jaaaa


23/27

23

****----aaaa >>>> [mostly after palatals](2) BEE/BIRD ****bbbbyyyyaaaa bbbb||||(4) CHILD/SON ****zzzzaaaa &&&&****ttttssssaaaa tttt NNNN 23(8) EAT ****ddddzzzzyyyyaaaa ddddzzzz|||| [dz|](9) EDGE/SIDE ****mmmm----ddddzzzzyyyyaaaa 24 ddddzzzz [dz](10) FLESH/MEAT ****ssssyyyyaaaa tttt||||(34) TROUSERS ****ssss----llllaaaa zzzz(56) RIGHTSIDE ****gggg----yyyyaaaa zzzz||||(72) FISH ****ssss----yyyyaaaa dddd||||(78) THIN ****bbbbaaaa bbbb||||(79) MANY ****mmmmyyyyaaaa &&&&****mmmmrrrraaaa zzzz||||

****----aaaa >>>> ---- [mostly after postvelars, labials, dentals](5) CHIN ****mmmm----kkkkaaaa mmmm\\\\ ----qqqq||||(16) LAUGH ****rrrryyyyaaaa

(48) HOOF ****kkkkwwwwaaaa 25 dw\Npppp(49) STRENGTH ****kkkk----rrrraaaa 26 qqqq(53) CATTLE ****wwwwaaaa qqqqwwww||||(55) FROG ****ssss----bbbbaaaallll pppp||||(58] BITTER ****kkkkaaaa qqqqhhhh(61) FIVE ****llll////bbbb----aaaa wwwwNNNN 27(64) BOX/CABINET ****ttttaaaa 28 tttt||||(66) HAMMER ****mmmm----tttt((((wwww))))aaaa &&&&****ssss----ttttaaaa 29 sssstttt(80) OPEN ****kkkkaaaa tttt ----qqqq||||(81) THROW ****mmmm----bbbbaaaa((((yyyy)))) &&&&****ssss----bbbbaaaa((((yyyy)))) bbbb|||| (Jinghua ssssbbbbaaaa)(82) FATHER30 ****ppppaaaa bbbb||||

****----aaaa >>>> ----aaaa(70) DITCH ****kkkkaaaa qqqqhhhh||||aaaa

****----wwwwaaaa >>>> oooouuuu(33) TOOTH ****sssswwwwaaaa ||||oooouuuu

23This form apparently reflects the suffixal ****----nnnn that sometimes appears on kinship terms, as in Dhimal tttt\\\\ssssaaaannnn son,Lepcha aaaa----zzzzoooonnnn 'grandchild' (cf. Benedict 1972: n.86, p. 27; n. 284, p. 100). Cf. FIVE (below) for a different sourceof a nasalized vowel.24Cf. Lahu jj j jaaaa, Akha ddddzzzz\\\\aaaa, Limbu jj j jaaaa.25Cf. Written Burmese kkkkhhhhwwwwaaaa.26Cf. Written Burmese aaaa, Lahu aaaa.27With nasalization ofthe vowel, apparently reflecting the original *nasal root-initial. Cf. CHILD for a differentsource of a Pumi nasalized vowel.28Cf. Lahu ttttaaaa----qqqqoooo 'box', Naxi tttt, Tujia tttthhhhoooo, Karen dddd____, Shixing tttt, PT ttttaaaa.29Cf. Lahu tttthhhhaaaa 'strike with flat hand, slap, strike a sharp blow, tttthhhhaaaa----ttttuuuu 'hammer; also Written Tibetan ((((mmmm))))tttthhhhoooo----bbbbaaaa'large hammer' < ****----ttttwwwwaaaa.30We are dealing here with a linguistic universal favoring ----aaaa vocalism (see MOTHER #68). The Xixia reconstruc-tion is wwww jj j jaaaa; rGyalrong dialects have forms like ttttpppp and aaaawwwwaaaa. See Matisoff 2000:#15.


24/27

24

****----yyyyaaaa >>>> ----(69) TONGUE ****ssss----llllyyyyaaaa

This word is also exceptional in Xixia; see above.

****aaaa----tttt >>>> ----

(67) HOT ****ttttssssaaaa----tttt 31

ttttssss||||

4444....0000 CCCCoooonnnncccclllluuuuddddiiiinnnngggg rrrreeeemmmmaaaarrrrkkkkssss

On the one hand, brightening is highly unusual, both in the TB context and in general. Onthe other hand it is strikingly characteristic both of Xixia and of the modern Qiangic languages.This makes the "shared innovation" of brightening a valuable criterion for membership in theQiangic group.

The modern languages do not display brightening to the same degree,32 nor is thephenomenon completely regular, either within the same language or cross-linguistically. However,the similarity of the conditioning factors across the Qiangic languages - especially the inhibitoryeffect of velar initials - leads us to conclude that we are indeed dealing with a tendency that must beimputed to Proto-Qiangic itself.

A few observations on conditioning factors and exceptionality:

(a) Initial or medial /w/, probably because of its velar component, frequently acts as an inhibitoryfactor, as in HANDSPAN (14), BIRD (46), HOOF (48), FIVE (61), TONGUE (69), and RAIN(75). This blocking of brightening does not occur in TOOTH (33), where ssss---- seems to be the

dominant component of the initial consonant sequence; neither does it affect AXE (1), PATCH(22), or SNOW (30), where the dominant segment in the initial is pppp----, with the [w] merely"extrusional".33

(b) Following a widespread tendency in TB/ST, sibilant initials frequently condition thecentralization of high front vowels to the "apical" position, as in EAT (8),FLESH/MEAT/ANIMAL (10), SALT (26), and SPARROW (31).

(c) Syllable-final ----rrrr in Gong's Xixia reconstructions seems to represent a suprasegmental feature of rhotacism, rather than a true consonantal segment, and is apparently an effect of an rrrr---- in the syllableonset, either prefixal or root-initial, as in BORROW/LEND (3), HUNDRED (15), LAUGH (16),

GNAW/CHEW (47), and RIGHTSIDE (56).

31See HOT (66) above. This ----seems to be the regular Dayang reflex of ****----aaaatttt; cf. VOMIT ****mmmm----ppppaaaatttt >>>> pppphhhh||||; KILL****gggg----ssssaaaatttt >>>> ssssyyyy.32 It would be interesting to count the number of brightened forms in the various Qiangic language, in order todetermine the relative strength of the tendency in each of them. Although I have not yet performed thiscalculation, it seems to me impressionistically that Xixia ranks near the top of the scale.33See Matisoff 2000.


25/27

25

(d) Brightening generally occurs after sibilant initials, as in CHILD/SON (4), EARTH (7), EAT(8), EDGE (9), FLESH/MEAT (10), and SALT (26). However, the Xixia reflex of HOT (67)retains the original ****----aaaa, undoubtedly because it descends from an allofam with dental suffix.

(e) Although velar initials generally block brightening, CHIN (5) and STRENGTH (49) are

exceptions, though in the latter case the Xixia reflex points to initial ****rrrr---- as its immediate prototype.(f) WEAR CLOTHES (35) is doubly exceptional in Xixia, since brightening occurs despite thevelar initial and medial ****----wwww----.

(g) Nishida (1964/66) reconstructs a prenasalized series for Xixia, which is confirmed by severalcomparisons with Lolo-Burmese etyma (see SHINE/BRIGHT (28), TAIL (32) and Matisoff 1978:17-18.

Rare as the development ****----aaaa > ----iiii is, it is by no means unheard of elsewhere in SoutheastAsia. It is characteristic of Far Western Hmongic (p.c., David Mortensen), and Mortensen has just

unearthed a similar phenomenon in the newly described dialect of Tangkhul that he calls "EastTusom". Among his examples are the following:

PTB P-Tangkhul E. Tusom(10) ANIMAL ****ssssyyyyaaaa **** aaaassssaaaa ||||(58) BITTER ****kkkkaaaa ****kkkk kkkkhhhhaaaa kkkk kkkkxxxx|||| 34(5) CHIN ****mmmm----kkkkaaaa ****mmmm kkkkhhhhaaaa mmmm||||kkkkxxxx\\\\(71) EAR ****gggg////rrrr----nnnnaaaa ****kkkk nnnnaaaa kkkk nnnn\\\\(8) EAT ****ddddzzzzyyyyaaaa ****kkkk ttttssssaaaa kkkk zzzz||||(21) NOSE ****ssss----nnnnaaaa ****nnnnaaaa---- nnnniiii \\\\(15) HUNDRED ****rrrr----ggggyyyyaaaa ****aaaa---- mmmm\\\\ |||| hhhh||||

Nevertheless, the fact remains that brightening is incomparably more characteristic of Qiangic/Xixia than it is of any other subgroup of Tibeto-Burman.

RRRReeeeff f f eeeerrrreeeennnncccceeeessssBenedict, Paul K. 1972. Sino-Tibetan: a Conspectus. Contributing Editor, James A. Matisoff.

Cambridge and New York: Cambridge University Press. ("STC")

34Note that velar initials do not at all impede brightening in this dialect.


26/27

26

Evans, Jonathan P. 2001. Introduction to Qiang Phonology and Lexicon. Tokyo: ILCAA, TokyoUniversity of Foreign Studies.

Gong Hwang-cherng. 1999. "Tangut cognates or synonyms." (Correlated with the sets presentedin Matisoff 1999.) 29 pp. MS.

----------. 2001. "Xixia-yu dongci de rencheng huying yu yinyun zhuanhuan." () [Verbal agreement and phonological alternations in Xixia]

Language and Linguistics (Taipei) 2.1:21-67.

Grinstead, Eric. 1972. Analysis of the Tangut Script. Lund: Studentlitteratur.

Ikeda, Takumi. 2004. "Explore the Mu-nya people and their language." 42 pp. MS.

Jacques, Guillaume. 2003. "Essai de comparaison des rimes du tangoute et du rGyalrong." Paperpresented at the Tenth Seminar of the International Association for Tibetan Studies,

Oxford. 24 pp. MS.

Jakobson, Roman. 1961. "Why 'mama' and 'papa'?" In B. Kaplan and S. Wapner, eds.,Perspectives in Psychological Theory , pp. 124-34. New York: John Wiley.

Matisoff, James A. 1972. The Loloish Tonal Split Revisited. Research Monograph #7, Centerfor South and Southeast Asia Studies. Berkeley: University of California.

----------. 1978. "Mpi and Lolo-Burmese microlinguistics." Monumenta Serindica 4:1-36.Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa.

----------. 1996. "D\ay|ang Pumi phonology and adumbrations of comparative Qiangic." Paperpresented at ICSTLL #29, Noordwijkerhout, Netherlands. Published (1998) inMon- Khmer Studies 27:171-213.

----------. 1999. "A modest start towards the sorting of materials for the reconstruction of Proto-Qiangic." Paper presented at the Workshop on Qiangic Languages and Linguistics,Academia Sinica, Taipei.

-----------. 2000. "An extrusional approach to pppp----////wwww---- variation in Sino-Tibetan." Language and Linguistics (Taipei) 1.2:135-86.

-----------. 2003. Handbook of Proto-Tibeto-Burman: system and philosophy of Sino-Tibetanreconstruction. Berkeley and Los Angeles: University of California Press.

Mortensen, David. 2004. "Examples of Proto-Tangkhul ****----aaaa > East Tusom ----iiii." One-pg. MS.

Nishida Tatsuo. 1964/1966.Seika-go no kenkyuu. [A Study of the Xixia Language].2 vols. Tokyo: Zauho.


27/27

27

--------------------------------------------.... 1967. Seika moji. [Xixia Writing]. Tokyo: Kinokuniya Shoten.

--------------------------------------------.... 1973. A Study of the Tosu-Chinese Vocabulary Tosu Yi-yu; the structure and lineage of Tosu, a new language. [In Japanese] Kyoto: Shokado.

--------------------------------------------.... 1976. "Hsi-hsia, Tosu, and Lolo-Burmese languages."Studia Phonologica 1:1-15.Sun Hongkai, Huang Bufan, et al., eds. 1991.Zangmianyu yuyin he cihui. Beijing: Chinese

Academy of Social Sciences. ("ZMYYC")

Documents

Xixia Qiangic-Gong Festo