In my last post, I reconstructed a Tangut root √*qu 'head'. Its initial consonant matches the q- of Qiang proper but not the initials of other (Na-)Qiangic words for 'head' in the STEDT database:

Guiqiong and Lyuzu w-

Ersu v-

Taoba Pumi kʰ-

Yajiang Queyu x-

Qiang proper (Northern: Mawo, Yadu / Southern: Longxi, Mianchi, Taoping) q-

Jiulong, Lanping, and Qinghua Pumi, Xinlong Queyu qʰ-

Lower Xumi, Muya, Namuyi ʁ-

The first two briefly reminded me of the -w- in Tangut

1kwə 'back of the head'

wə 'ahead'

but that -w- is from a prefix *P- rather than a root initial.

Moroever, Dominic Yu (2012) reconstructed Proto-Ersuic *w- as the source of the Ersu and Lyuzu initials and derived his Proto-Ersuic *wilje/*wulje 'head' from Proto-Tibeto-Burman (PTB) *d-bu 'id.' Although I don't believe in a 'Tibeto-Burman' subgroup of Sino-Tibetan, it's likely that the Ersuic words and Guiqiong 33 jɛ̃53 are cognate to Written Tibetan dbu 'head' rather than mgo 'head'. (Is Guiqiong -j- < *-lj-?)

The STEDT database also regards the Qiangic ʁ-words as being from PTB *d-bu.

Are Ersuic W-type initials and Qiangic ʁ- the products of intervocalic lenition? Is there any other language in which *b became ʁ-?

The remaining Qiangic initials are velars (like the -g- of Written Tibetan mgo) or uvulars (like Old Chinese 后 *ɢoʔ 'ruler' < 'head of state')? Which point of articulation is original? Rgyalrongic distinguishes between velars and uvulars, and its words for 'head' have -k- or -w-. The STEDT database derives the -w-words from PTB *d-bu, but I wonder if that -w- is from a lenited *-k-.

Maybe I should reconstruct the pre-Tangut root for 'head' with *k. However, I would then have to reconstruct *Cʌ-kru with a presyllable *Cʌ- and medial *-r- to account for the lenited velar and lowered vowel of

1ɣʊ 'head'

and I don't know of any *kru-type word for 'head' in Sino-Tibetan. No, wait, there is Darang k(u)ru 'head', but that's in Deng, a subgroup far from Qiangic. And the STEDT database derives that word from PTB *k-lu 'head'. I would reconstruct such a root at the Proto-Sino-Tibetan level to account for Old Chinese 首 *l̥uʔ 'head'. The voiceless *l̥- could be from an earlier *ql- if the word is borrowed from (or cognate to?) Proto-Austronesian *qulu 'head'.

I think the true Tangut cognate of Darang k(u)ru and Old Chinese 首 *l̥uʔ 'head' might be

2liụ 'brain, head' < *S-luH

Tangut -iu might have been phonetically [y] like the vowel of the first syllable of Xumi (Upper or Lower unspecified) ly5533 'forehead' (but what is the second syllable?).

The compound

1ɣʊ 2liụ  'head, brain, thoughts, leader' (definition from Kychanov and Arakawa 2006: 548)

is reminiscent of Naxi kv33 ly33 'head', regarded as a disyllabic descendant of PTB *k-lu in the STEDT database, but I doubt that 1ɣʊ 2liụ was originally a disyllabic word that was split into two.

Tangut has another word for 'brain' borrowed from Chinese and written with two slightly different characters:

2no < pre-Tangut *naw < northwestern Middle Chinese 腦 *naw < Old Chinese *Cʌ-nuʔ

The significance of the presence or absence of ㄇ (Boxenhorn code cie) is unknown. The two characters are in the same homophone group in Homophones.

Old Chinese *Cʌ-nuʔ may be borrowed from (or cognate to?) Proto-Austronesian *punuq 'brain', but the first vowels do not match. Could an unstressed labial vowel *u have dissimilated before *p-? The first vowel of Old Chinese *Cʌ-nuʔ has to be low to condition the bending of the following stressed vowel:

*Cʌ-nuʔ > *Cʌ-nouʔ > *(Cʌ-)nauʔ > *naw

(*nau and *naw are phonetically identical, but I regard *au as a diphthong and *aw as a vowel-consonant sequence.) (WIFE + FACE) + (BACK OF THE HEAD + THREE DAYS LATER) = ?

... 'comb'?

In the Tangut-Chinese glossary Pearl in the Palm,

5416 5064 2ɣwə 2ɮa

is glossed as 木梳 'wooden comb'.

The Precious Rhymes of the Tangraphic Sea analyses of its two halves are:


5416 2ɣwə 'before, ahead' =

left of 5079 2ɣwə 'wife' (phonetic) +

right of 2783 1ɬwiẹ 'face'


5064 2ɮa =

left of 5284 1kwə 'back of the head' +

right of 2096 2ɣwə (first half of

2096 2532 2ɣwə 1die 'three days later';

2532 1die is 'day')

The graph 5416 2ɣwə 'before, ahead' contains 'head' (Nishida radical 106 / Boxenhorn code gos) on its left side. The word 5416 2ɣwə 'ahead' may be cognate to

2750 1ɣʊ 'head'

and could mean 'head' in 5416 5064 2ɣwə 2ɮa 'wooden comb'. Kychanov and Arakawa (2006: 573) even gloss 5416 2ɣwə as 'head', though Li Fanwen 2008 lists no examples of 2ɣwə with that meaning.)

2750 1ɣʊ 'head' may go back to a pre-Tangut *ɢu (cf. Tibetan mgo 'id.', Old Chinese 后 *ɢoʔ 'ruler' < 'head of state'). The uvular root consonant *ɢ- conditioned the lowering of *-u to and lenited to ɣ-. (There was no syllable *gʊ in Tangut, so *ɢ- did not become a velar stop g- before -ʊ. It is possible that ɣ- was phonetically uvular [ʁ].)

5416 2ɣwə 'ahead' may have come from something like *p-ɢu-H. The *p-prefix is reflected in Tangut -w- and the *-H suffix is reflected in the Tangut second ('rising') tone. I cannot explain the difference in vowels.

(12.9.0:18: Another possibiliy is that 5416 2ɣwə 'ahead' is from 5284 1kwə 'back of the head' plus a prefix conditioning lenition and a suffix conditioning the second [i.e., 'rising'] tone:

2ɣwə < *Cʌ-kwə-H

Maybe those two words share a root with 2750 1ɣʊ 'head':

*qu 'head' (cf. Jiulong Pumi qhuo55 'id.')

> *CV-qu > 2750 1ɣʊ 'head':

> *p-qu-ə > 5284 1kwə 'back of the head'

> *pV-qu-ə-H > 5416 2ɣwə 'ahead'

Intervocalic *-q- lenited to ɣ-. The shift of *p-q- to kw- is parallel to Pittayaporn's [2009: 164] Proto-Tai *p.q- which became Saek kw-.)

5064 2ɮa may have once been an independent word for 'comb', though Li Fanwen (2008: 800) lists no examples of it that are not preceded by 2ɣwə. Perhaps 2ɮa (reconstructed by Gong with z-) is related to the zia55 of Zhaba a33 zia55 ~ gu33 zia55 'comb'.

The left side of 5064 'comb' is 'head' (Nishida radical 106 / Boxenhorn radical gos). It is interesting that 5064 is derived from 5284 'back of the head' given that Emperor Jingzong ordered the Tangut to shave off most of their hair circa 1034, shortly before the Tangut script was officially adopted about two years later. See Dunnell (1994: 181) on Emperor Jingzong's order and figures 1 and 4 in Kepping (2003: 106, 109) for illustrations of the two types of Tangut men (the 'red-faced' and the 'black-headed') without hair on the backs of their heads. Was the graph for 2ɮa conceived before 1034 when Tangut men might still have had hair on the back of their heads to comb? Or does the deriviation refer to the backs of Tangut women's heads? Or is the derivation simply wrong? (It would be strange for a derivation created decades after Jingzong's decree to refer to a place where Tangut men shouldn't have had hair anymore.)

It is likely that the graph of Li Fanwen graph 2096, the first half of 2096 2532 2ɣwə 1die 'three days later', is derived from 5064 2ɣwə 'comb' rather than the other way around. 5064 is surely phonetic in its homophone 2096, but what is 'person' (Nishida radical 204 / Boxenhorn code dex) doing on the left of 2096? The Precious Rhymes of the Tangraphic Sea analysis of 2096 contains 'head' instead of 'comb':


2096 2ɣwə (first half of 2ɣwə 1die 'three days later') =

left and center of 2750 1ɣʊ 'head' +

right of 2132 2ʔiew 'achievement'

2096 2532 2ɣwə 1die may be literally 'ahead day'. I don't know what the relevance of 'achievement' is or the function of the right-hand component (Boxenhorn code dal) that it shares with 2096.

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2013 Amritavision