I left out one more point from part 2. According to Gong (1993)*, Tangut Grade II is derived from pre-Tangut (his 'proto-Tangut') medial *-r-. Hence Tangut rhyme 4 (my -ʏ and Gong's -u) comes from pT *-ru and

1ɣʏ 'head' (Gong: 1ɣu)?

comes from *CV-K/Qru. The frontness of rhyme 4 is a trace of *-r-:

*-ru > *-rʊ > *-jʊ >

However, none of the proposed cognates for 1ɣʏ 'head'

Written Tibetan mgo

gDong-brgyad rGyalrong tɯ-ku

Zbu rGyalrong tə-kuʔ

Pumi qho

Mawo, Taoping, and Ronghong Qiang qə-

have a medial -r- with the possible exception of Old Chinese 后 *ɢ(r)oʔ < ?*N-q(r)oʔ 'ruler'. Although 后 is a phonetic in the spelling of the reduplicative word 邂逅 *N-qreʔ-N-qroʔ 'carefree', that does not necessarily mean that 后 itself had a medial *-r-. The presence of *-r- in one member of a phonetic series does not entail the reconstruction of *-r- in all other members of that series.

I could reconstruct three or four unrelated words for 'head' that happen to sound similar

*QrU > (Old Chinese?, Tangut?)

*QU > (Old Chinese?, Tibetan?), Pumi, Qiang

*KU > (Tibetan?), rGyalrong

Written Tibetan mgo could be from a uvular or a velar root.

*Kru > (Tangut?)

but I prefer a simpler scenario in which *QrU and *KU are derived from a single root *QU via *r-affixation and *Q-fronting. However, I know of no other instance of *Q-fronting in rGyalrong (apart from Somang which has fronted all uvulars to velars and even palatals; see Jacques 2004: 309-310).

Could the uvulars in non-rGyalrong languages be secondary? In other words, does rGyalrong preserve the original velar initial *k- of 'head'?

*I have not seen this (unpublished?) manuscript, but Jacques (2009: 6) refers to it. DANGER A-HEAD FOR THE 2 X 2 HYPOTHESIS (PART 2)

(See part 1 here. Note the different placement of italics in the two titles.)

In my last entry, I stated that GSR 112 后 was a uvular series. Hence I would reconstruct 后 'ruler' as Old Chinese *ɢoʔ. Both Guillaume Jacques and Schuessler (2007: 280) have proposed that 后 is cognate to Written Tibetan mgo 'head' since a ruler is the 'head' of a society.

Could these words, gDong-brgyad rGyalrong tɯ-ku 'head', Pumi qho, and Qiang qə- 'head' be related to Tangut

1ɣʏ 'head' (Gong: 1ɣu)?

The Tangut fricative initial is ambiguous. It could be from a velar or uvular that lenited after the vowel of a lost presyllable: pre-Tangut *CV-K/Q- > Tangut ɣ-.

Written Tibetan g- in mgo is also ambiguous. It could be from a velar or a uvular.

If Sagart's uvular hypothesis is correct, GSR 112 后 has to have a uvular initial *ɢ (in my notation).

The Qiang roots for 'head' that I've seen in Mawo, Taoping, and Ronghong all point to a proto-Qiang *q-.

However, the rGyalrong words for 'head' in Jacques (2004: 300)

gDong-brgyad tɯ-ku

Zbu tə-k

both point to proto-rGyalrong *k- rather than proto-rGyalrong *q- which corresponds to gDb q- and Zbu q- or ʁ-.

(1.25.00:07: Somang ta-kó would be ambiguous in isolation since Somang has merged proto-rGyalrong velars and uvulars. See Jacques 2004: 309-310.)

Were there two similar roots for 'head', one with a uvular (reflected in Old Chinese and Qiang) and another with a velar (reflected in rGyalrong)?

Matisoff (2003) reconstructed Proto-Tibeto-Burman without uvulars. Assuming PTB existed (I don't), one could then hypothesize that

- the Proto-Sino-Tibetan word for 'head' had a uvular

- this uvular was preserved in Old Chinese

- but lost in Proto-Tibeto-Burman (= all of Sino-Tibetan but Chinese)

- and later regained (!) in Pumi and Qiang (and pre-Tangut?)

Proto-Sino-Tibetan *uvular
Old Chinese *uvular Proto-Tibeto-Burman *velar
Pumi and Qiang uvular Tibetan and rGyalrong velar

Although A > B > A shifts are possible, I'd rather not reconstruct them. Here's a scenario without them:

- the Proto-Sino-Tibetan word for 'head' had a uvular

- this uvular was preserved in Old Chinese, Pumi and Qiang, (and pre-Tangut?) but lost in Tibetan

- the velar in rGyalrong is secondary (but what conditioned the fronting of the uvular?) or the rGyalrong root is an innovation unrelated to the uvular root in the other languages

1.25.1:49: I forgot to explain why the above data pose a problem for the 2 x 2 hypothesis of Tangut grades.

1ɣʏ 'head' belongs to Tangut rhyme 4 which I identified as Grade II. (Gong regarded it as Grade I and reconstructed it as -u which was also his reconstruction for the indisputable Grade I rhyme 1. If rhymes 1 and 4 were identical, why were they separated in the Tangraphic Sea?) According to the 2 x 2 hypothesis, Grade II was low and palatal. But none of the proposed cognates for 1ɣʏ have palatal glides or vowels. Is this the kiss of death for a palatal vowel in R4? Not necessarily. Front vowels in one language can correspond to back vowels in another: e.g., English dead : Dutch dood. Could the frontness of the vowel of rhyme 4 in 'head' be secondary? What if 1ɣʏ were from something like *Ce-qu with a nonhigh front presyllabic vowel that conditioned the lowering and fronting of *u?

Tangut fonts by Mojikyo.org
Tangut radical font by Andrew West
All other content copyright © 2002-2011 Amritavision