In my last post, I explained why the Tangut word for 'eye'

is an open syllable instead of *1mew.

In my Tangut reconstruction, there are almost no syllables with labial initials and labial codas: i.e., pVw, phVw, bVw, mVw, or vVw. (POw syllables are common in Gong Hwang-cherng's reconstruction; his -w corresponds to my nasalization after labial vowels.)

I reconstruct only two PVw syllables in Tangut. The first is


which Li (2008: 382) glossed as an adjective 'poor, needy, impoverished', 貧窮 'poor', and 饑寒 'hungry and cold' and Kychanov and Arakawa (2006: 642) glossed as a noun 'нужда' (need), 'need, want' as well as 窮蹙 'in dire straits'. Li did not list any attestations in running text (as opposed to dictionaries) apart from its use as a transcription character for Tangut period northwestern Chinese 表 *piew. I think the word might be marginal.

Could it have been borrowed after the loss of -w (either original or from *-k) after labials?

*PIk > *PIɣ > *PIɰ > *PIw > PI

But I don't know of any piew- or piek-like Chinese word for 'poor', and -w syllables are alien to Tibetan. I cannot find any Tibetan words of the shape P(y)Ig(s) meaning 'poor'.

Is 1piew an archaism retaining a -w lost elsewhere?

In Homophones, it is paired with 1vaʳ 'poor'*:

Is this a semantic equation 1vaʳ = 1piew or a ('ritual Tangut'?) disyllabic word 1vaʳ-1piew 'poor'?

Next: Anuttarā samyak sambodhi

*7.29.11:09: The Tangraphic Sea analysis of 1piew derives it from 1waʳ:


1piew 'poor' = left of 1waʳ 'poor' + right of tha 'to force' (tone unknown; cf. Chn 迫 'id.' in 饑寒交迫 'suffer from hunger and cold').

I thought the shared left side of the 'poor' tangraphs might be derived from Chinese 貧 'poor':

分 - 八 = 刀 >

貝 - 口 >

but only one other tangraph shares that left side

2ləi 'fear'

and it does not have any phonetic or semantic resemblance to the other two. Yet 2ləi appears in the analysis of 1vaʳ:


1vaʳ 'poor' = left of 2ləi 'fear' + right of 1phie 'to escape' (< *k-bie < Middle Chinese 避 *bieh 'id.'?)

I think the center and right of 2ləi 'fear'

are phonetic, as they are shared with the homophonous tangraph

2ləi 'tiger'

whose components in turn are shared with the tangraphs for (the 'ritual Tangut' word?)

2zeʳw-2ləi 'east' (whose second half is homophonous with 2ləi 'fear' though its shared parts are in the opposite order!)

 If so, what is the left sideof 2ləi 'fear'?

**7.29.11:41: Old Chinese disyllabic noncompounds tend to have halves matching in terms of 'emphasis'. I called this phenomenon "Avoiding ABBA". (A signifies 'emphatic' syllables with nonhigh or lowered vowels, B 'nonemphatic' syllables with high or raised vowels, and AB and BA mixed-'emphasis' disyllables.)

I wonder if Tangut disyllabic noncompounds have a similar tendency. If they do, then 2zeʳw-2ləi 'east' follows it (the first syllable has a nonhigh vowel and the second has a lowered vowel) and 1vaʳ-1piew 'poor' is an exception: the first syllable has a low vowel and the second syllable has a partly raised vowel implying a lost presyllable with a high vowel: *Cɯ-pew or *Cɯ-pek. The first syllable could be from *var or *rʌ-Pa. (*vra would have become *1væ, so the retroflex vowel must be from a non-medial source: a final *-r or a presyllable *rʌ- that also conditioned the lenition of *P- if it was a stop.)

Disyllabic words that only appear in dictionaries may be 'ritual Tangut' (RT) words only used in odes that have not yet been found. However, RT words normally do not contain nonritual Tangut components like 1vaʳ, so 1vaʳ-1piew could be

- an anomalous RT disyllabic word

- a nonritual Tangut disyllabic word (an extended form of 1vaʳ?)

- an equation of two monosyllables rather than a disyllabic word: 1vaʳ = 1piew

If 1piew is a free morpheme, is it a nonritual synonym of nonritual 1vaʳ, or is it an anomalous RT monosyllabic word?

We are fortunate that we have monolingual dictionaries of Tangut that enable us to have the luxury of discussing synonyms - something we cannot do with either Khitan or Jurchen. But more information just leads to more mysteries. There may never be any shortage of Tangut riddles to solve. WHY DOESN'T TANGUT 1ME 'EYE' END IN -W?

That question came to mind while writing about the Proto-Min words for 'eye'. Various Sino-Tibetan words for 'eye' point to an original final *-k: e.g.,

Sinitic (non-Min): Nanchang Gan muk, Meixian Hakka muk, Cantonese muk

Written Tibetan mig (there is no final -k in WT)

Written Burmese myak

STEDT lists many more examples.

Hence the Proto-Sino-Tibetan word for 'eye' must have ended in *-k or something similar as in these reconstructions at STEDT:

Chou Fa-kao 1972: *(s-)myəːk

Coblin 1986: *myikw (cf. my OC ?*miwk from "Three Min Etymologies Revisited")

LaPolla 1987: *mya[a]k

Pre-Tangut would have inherited this *-k which would normally have become Tangut -w after front vowels*. Yet the Tangut word for 'eye' is an open syllable

1me (which by coincidence looks like modern standard Japanese me 'eye' < *ma-i!)

rather than *1mew. The most straightforward pre-Tangut reconstruction for 'eye' would be *me. The *e may be a monophthongization of an earlier *yə or *ya.

Why did pre-Tangut lose *-k in 'eye'? I propose that *-w was lost in pre-Tangut syllables of the type *PIk(H) and *PIw (*P- = labial initial, *I = front vowel) to avoid two labial consonants (in bold) in the same word: e.g.,

PT *mek > *mew > T 1me

(Syllables without PT final *-H developed tone 1.)

PT *mew with original *-w (as opposed to secondary *-w from *-k) would also have become T 1me.

This hypothesis predicts the absence of Tangut syllables of the type PIw. As far as I know, there are only two such syllables. I'll look at them tomorrow.

*7.24.4:04: Tangut only had -w after front vowels. This implies that *-w (partly from *-k) was lost after nonfront vowels:

After labials

*-Uk > *-Uɣ > *-Uɰ > *-Uw > -U

After neutrals

*-Ak > *-Aɣ > *-Aɰ > -A

The above scenario is too simplistic. So far I know of one problem with it. The Tangut cognate of these words for 'six'

Old Chinese *ruk

Written Tibetan drug

Written Burmese khrok


1tʃhɨiw < *k-truk

with -w, not *1tʃhɨu. Perhaps there was a dissimilation:

-Uk > *-Uɣ > *-Uɰ (back vowel + back glide) > *-Iɰ (front vowel + back glide) > *-Iw
Similarly, the Tangut cognate of Written Tibetan shug-pa 'juniper tree' is

1ʒɨiw < *Cɯ-ʃuk

with -w, not *1ʒɨu. Unfortunately I do not know of any other examples of this correspondence. SINO-MANCHU VOWEL LENGTH IN DAGUR?

I dedicate this entry to the late Jerry Norman since it combines his two major interests: Chinese and Manchu.

Written Manchu has only one long vowel <oo>, but perhaps other long vowels existed even though they were not distinguished in writing. Kara György (1985) listed sixty Manchu loanwords with long vowels in Dagur (D). He concluded (p. 128):

Some of these Dagur words of Manchu and Sino-Manchu origin may have secondary long vowels due to the prosodic differences of these languages (the two vowel systems at least look very similar, but little is known e.g. about the nature and rules of Manchu stress), nevertheless the evidence of Dagur proves to be affirming or revealing in several cases.

Dagur Sino-Manchu (SM) words in Kara's article with long vowels are:

24. D gyaa 'street' : Xibe g'a(a), SM giyai < OM 街 *kyai

The D and Xibe forms are reminscent of Sino-Korean 街 ka (not *kae < *kai). I think the -a and -ai words for 'street' reflect variation within early northern Chinese.

25. D gyaan 'reason' : SM giyan < OM 見 *kyan 'see; opinion (i.e., how one sees something)'

Was 'reason' ever a meaning of 見 in OM or was the shift to 'reason' Manchu (or Jurchen)-internal?

39. D kyaaj 'box', kyaas 'basket' : SM hiyase < OM 匣子 *xyatsɨ

The D initial and final consonants do not match SM. I don't know of any modern Chinese language with an initial stop in 'box' (though its first half was *gæp in Late Old Chinese). D does have xy, so I would have expected *xyaas. Was SM h [x] or OM *x misheard as k, or did *xy- become ky- in some nonstandard Manchu dialect? D -j in 'box' seems to directly reflect OM 子 *-tsɨ, not early SM -se. (Was there a nonstandard late SM *hiya dz?)

40. D kyaanci 'chess' : SM hiyangci < OM 象棋 *syangkhi

Both D and SM do not match OM. I would expect D *hyaangki and Manchu *siyangki. The consonants are puzzling:

Even if D ky- is from *xy- (cf. 39 above), Manchu hiy- is not from *siy-.

D has -ngk-, yet the word has -nc-.

Manchu ci is not from *ki.

Could D and SM incorporate sound changes that occurred in an intermediary language - perhaps some nonstandard dialect(s) of Khitan (these changes are not known in written Khitan)? Or are the D and SM forms very recent borrowings from Mandarin after palatalization: OM *syangkhi > Md *ɕyangchi?

41. D kyenkyeen 'strong' : SM kiyangkiyan < OM 強健 *khyangkyan

D -ye- may reflect a nonstandard Manchu dialect with -iye- < -iya- or a [jæ] pronunciation of standard Manchu -iya-.

Once again (cf. 40 above) D has -n- instead of -ng-. I would expect *kyengkyeen.

SM -k- is irregular; I would expect kiyanggiyan.

60. D yamboo 'silver ingot' : SM yuwan boo < OM 元寶 *yüanpau

If D does not have oo < au, this is a straightforward loan postdating the monophthongization of Jurchen au into Manchu oo.

60a (no number in Kara's article but under 60). D lwe̮e̮s ~ lwees 'mule' : Manchu lorin? (my guess; Kara did not identify a Manchu source); cf. OM 驢子 *lütsɨ, *lwo

I have no idea what the breves and underlining in D lwe̮e̮s ~ lwees represent. As far as I know, D only has one kind of e(e).

The D word looks like it's from 驢子, whereas the Manchu word looks like it's from 騾 - but what is -rin? Was it added by analogy with morin 'horse?

The vowel length in SM loanwords in D does not correspond to anything in Chinese other than the absence of the rising tone (which may be an artifact of the small sample size). That is a shame, because I had been hoping to use the words above to test hypotheses of vowel length in earlier Chinese: e.g., Pulleyblank reconstructed 街 (24) and 匣 (39) with long vowels.

7.23.00:32: Not a SM loanword, but Manchu in D nonetheless and not in Kara (1985): monyoo 'monkey' (Tsumagari 2003: 133) from Manchu monio (not monioo!).

7.23.00:45: Kane (2009: 267) wrote,

The Daurs [= Dagur] have also a long standing myth of their origins, which suggests they are descended at least partially from the Kitans [= Khitan]. The discovery of a well preserved body of a female shaman provided DNA which indicates that the claims of both the benren ['original people'] of Yunnan and the Daurs have some substance [...] The question of the linguistic relationship between Kitan and Daur is a fruitful field for further research.

Perhaps some SM loanwords in D are actually Sino-Khitan. However, none of the D phonological oddities above have parallels in the admittedly limited Sino-Khitan corpus in chapter seven of Kane (2009): e.g.,

Sino-Khitan has <k> and <x> for Chinese *k(h)-, not <c> (see 40)

Sino-Khitan never has <k> for Chinese *x- or *s- (see 39 and 40)

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2012 Amritavision