Li Fanwen (2008: ) glossed
= +

2phi = 'low' (Nishida radical 176*) + 'come' (Nishida radical 115)

as 'chopsticks' though its graphic structure does not seem to be relevant, and glossed

2phi 1tʂɨu

as 'bamboo chopsticks'. 1tʂɨu is a transcription character that presumably represents a borrowing of Chinese 竹 (*tʂɨu in the dialect known to the Tangut) used as an adjective (hence its position after the noun).

The D version of Homophones glossed 2phi as

1vɨi 'to do' and

2dwəu 'chopsticks'.

However, Kychanov and Arakawa (2006: 714) glossed 2phi as 'name of an ancestor; from generation to generation; descendants; change of generation; hereafter'. The structure of the character fits this definition: descendants are those who descend - 'come down'.

K&A also have a different definition for 2phi 1tʂɨu: 'have no posterity'. I would not expect this since I do not know of any other instances in which 1tʂɨu means 'to not have'. The Tangraphic Sea defines 1tʂɨu as 'to do'. I also do not know how 1tʂɨu and 2phi, both 'to do', differ from the common verb 1vɨi 'to do'.

Are Li and K&A both right? Tangut is full of monosyllabic homonyms, but did 2phi 1tʂɨu have two completely different meanings? Did Kychanov and/or Arakawa see 2phi 1tʂɨu in a context where 'have no posterity' fits?

They also include a disyllabic expression

2phi 2riʳ 'not leave descendants'

absent from Li's dictionary. The second character is the first half of the disyllabic word

1riʳ 1giu 'alone without descendants'


2phi 1tʂɨu

be a scribal error or a misreading of a cursive form for

2phi 2riʳ?

The top and bottom left of 1tʂɨu and 1riʳ are identical, and the bottom right elements are vaguely similar.

Does 2phi 2riʳ literally mean 'descendant-lonely': i.e., lonely because of the absence of descendants? (Li 2008: 780) defined 2riʳ by itself as 'lonely, solitary', though he did not provide quotations in which it appears alone without 1giu.

The character for 2riʳ might be from 'relative' and 'not have':


2riʳ 'lonely'? = 2ŋwi 'relative' + 1mie 'not have'?

*Nishida (1966: 244) glossed this radical as 'to rise' presumably because it means 'ladder' by itself,


but it appears in characters for words having to do with lowness:

1lwiu 'low' (with 'earth' on the left)

1biə 1bi 'below' (disyllabic reduplicative word; with 'mouth' and 'person' on the left - why?)

so it may be an abbreviation for those characters in compounds such as 2phi. 'HELP'-FUL EVIDENCE FOR TANGUT ʔw < *ɢʷ?

The Tangraphic Sea analyses of


1tsə̣ = bottom left of 2dwəu 'chopsticks' + top left of 2ʔwəụ 'help' + right of 2bɛ̣ 'branch'

the first half of

1tsə̣ 2dwəu 'chopsticks'

from my previous entry, and


1vɨe 'to go' = top and bottom left and center of dʐɨi (tone unknown*) 'to walk' + right of 2ʔwəụ 'to help'

from an earlier entry both contain 2ʔwəụ 'to help' which resembles Middle Chinese 祐 *wuʔ 'help' (cf. Sino-Korean 우 u). Could it be a loan from Chinese? 2ʔwəụ goes back to a pre-Tangut *S-ʔʌ-wuH. A prefix *S- is needed to account for the tense vowel (indicated with a subscript dot) and has no parallel in Chinese. Another prefix ʔʌ- is needed to condition the lowering of the vowel *u. The Tangut second (= 'rising') tone is from an earlier *-H which could be from an even earlier *-ʔ.

Baxter and Sagart (2011) reconstructed 祐 as *ɢʷəʔ in Old Chinese (OC). I considered the possibility that a Proto-Sino-Tibetan *ɢʷ- that was preserved in OC became ʔw- in Tangut, but I would expect OC to correspond to pre-Tangut *-ə, not *-u.

*Possibly level tone since it's listed between level tone syllables in the Precious Rhymes of the Tangraphic Sea, but I'm not sure. WOODEN BRANCHES HELPING HANDS TO TAKE

The Tangut-Chinese bilingual glossary Pearl in the Palm (233) equated the disyllabic Tangut word

1tsə̣ 2dwəu

with Chinese 筯 'chopsticks'.

The word might go back to a pre-Tangut *s-tsə Pʌ-duH:

*s- conditioned the tenseness of the following vowel

*Pʌ- conditioned medial -w- and the partial lowering of *u

*-H conditioned the 'rising tone' (tone 2) and its absence conditioned the 'level tone' (tone 1)

Could the coda of the first syllable have been reinterpreted as the presyllable initial of the second?

*s-tsəp ʌ-duH > *s-tsə Pʌ-duH?

*tsəp could be the Tibeto-Burman *ts-p root for 'squeeze' (cf. also Old Chinese 挾 *tsep 'grasp', ).

*duH may be a borrowing from Chinese. It resembles Vietnamese đũa < *duəh 'chopsticks' which may be a borrowing from Middle Chinese 箸 *ɖɨəh 'id.'

Pre-Tangut *u can be from an even earlier *a: e.g.,

1dzəu < *Cʌ-dzu 'to love'

is cognate to Written Tibetan mdza'-ba 'to love'.

I initially thought that pre-Tangut *duH < *daH could have been borrowed from Old Chinese 箸 *Nɯ-trak-s 'chopsticks' (derived from 著 *Cɯ-trak 'to apply'?), but *Nɯ-trak-s would have become Tangut *2dʐɨu with an affricate, not 2dəu with a stop (cf. how *k-truk with a *tr-cluster became

Tangut 1tʂʰɨiw 'six'

with an affricate).

Maybe pre-Tangut *u was an attempt to imitate the of Middle Chinese 箸 *ɖɨəh at a time when Tangut did not yet have ɨə from *ə. Tangut did not have retroflex stops, so *d was the closest equivalent of Middle Chinese *ɖ-.

Perhaps *s-tsəp ʌ-duH was 'the grasping *duH' with 'grasp' added to clarify the function of the following foreign word. Unfortunately there is no other evidence for a *(s-)tsəp meaning 'grasp' in Tangut; if such a verb ever existed, it only (?) survives in this compound.

I was surprised that Kychanov and Arakawa's dictionary had no entry for 1tsə̣, though they did gloss 2dwəu as 'chopsticks'. Although there is no doubt 2dwəu is a morpheme meaning 'chopsticks' as in this compound from Pearl in the Palm (241),

1məə 2dwəu 'fire chopsticks'

I have not seen it used as a word by itself.

The Tangraphic Sea analyzed the character for 1tsə̣ as


bottom left of 2dwəu 'chopsticks' + top left of 2ʔwəụ 'help' + right of 2bɛ̣ 'branch'

I know of no other analyses in which 'help' was abbreviated as コ. The choice of 'help' might reflect the phonetic 助'help' of Chinese 筯 'chopsticks'. In any case, the right side of 1tsə̣ is not in any other tangraph.

The analysis of 2dwəu is unknown, but I wouldn't be surprised if it contained 1tsə̣:


top of 1si 'wood' + left of 1tsə̣ (i.e., 'hand') + left of 1lhwi 'to take' (i.e., Nishida's radical 112 'grasp')?

Although it is tempting to analyze 1tsə̣ as


top of 1si 'wood' + all of 2kwị 'to harvest',

'to harvest' makes no semantic sense.

1si 2kwị

is an object-verb phrase for 'collect firewood' (Li Fanwen 2008: 827). EQUATING DOGS AND HORSES: REANALYZING CHHA(-RIOT)

The Tangut character for 1tʂʰɨa 'chariot' (chha in my lay romanization) looks like two 'dogs' (Nishida radical 218) beneath a horizontal line:


However, the Tangraphic Sea analysis is quite different:


1tʂʰɨa = top left of 2dʐɛ 'wheel' + bottom left and right of 1ŋwəiʳ 'to equate'

The Precious Rhymes of the Tangraphic Sea analysis may contain a scribal error for 1ŋwəiʳ:


1tʂʰɨa = top left of 2dʐɛ 'wheel' + bottom left and right of 1ʔiaʳ 'to stand'.

I suspect that these analyses are after-the-fact attempts to explain the character 1tʂʰɨa in terms of other Tangut characters.

The character for 1tʂʰɨa could be of Chinese origin, as it vaguely resembles the two horses (?) on the right side of a seal form of 車 'chariot'.

Alternately, it may be a pictogram of a base (a horizontal line) held up by two wheels (represented by angular lines since no circles are permissible in tangraphy). But if the 'dogs' are really supposed to be wheels, why not simply write the left side of 'wheel' (Nishida radical 262) twice?

The right side of 'wheel' (perhaps derived from the right side of Chinese 輪 'wheel') is the mysterious right-hand element

added to some but not all meaningful elements.

I don't know of any cognates of 2dʐɛ 'wheel' which should be from a pre-Tangut *džeH or *Ntš(ʰ)eH. The first syllable of Namuyi tʂhɿ³¹ li³¹  'wheel' looks similar, but what is the second syllable? DID THE TANGUT HAVE A NATIVE WORD FOR 'CHARIOT'?

According to the Tangraphic Sea, the subject of my previous entry plus the top of 1si 'tree, wood' equals 1ko 'chariot':


The character for 1ko is clearly a semantic compound: something made of wood that goes. The Tangraphic Sea defined it as a

1ko 1tʂʰɨa,

whose elements appear in the opposite order in Homophones:

1tʂʰɨa 1ko

The Tangraphic Sea stated that 1tʂʰɨa is homophonous with Chinese (i.e., with 車 'chariot') and is equivalent to the Tangut word 1ko.

One might think that 1ko is a native Tangut word for 'chariot', but I noticed that it resembles one of the Grade III Middle Chinese readings of 車 which Karlgren reconstructed as *ki̭wo = *kjwo, corresponding to my Early Middle Chinese *kɨə (cf. Sino-Korean [kə]) and Later Middle Chinese *kø. (I later found that Li Fanwen [2008: 677] had independently identified 1ko as a loan from Chinese.) None of the above Chinese reconstructions is a precise match for Tangut Grade I 1ko. If the Chinese form were *kjwo or *kɨə, it could have been borrowed as Grade IV *1kio or *1kiə (or even Grade III *1kɨə). So does 1ko point to *kø as the correct reconstruction for the northwestern late Middle Chinese dialect known to the Tangut? (Here I assume Tangut had no vowel like ø. I could be wrong. I have proposed ø as a possible reconstruction of Grade IV o.) In any case, the grades of the Tangut and Chinese words do not match, even though Tangut borrowings from Chinese usually have identical or similar grades:

Grade of Chinese source word I II III IV
Expected grade of borrowing in Tangut I II III or IV depending on initial IV

Could ko reflect a nonstandard Middle Chinese Grade I *ko? The Grade III word is from an earlier syllable like *kla with raising (and fronting?) conditioned by a lost presyllable with a high vowel:

*Cɯ-kla > *Cɯ-k(l)ɨa > *kɨa > *kɨə > *kø

If that presyllable were lost before it conditioned raising (and fronting)?, *kla would have become *ko.

The initial of the presyllable may have been *k- if the word was a borrowing from a form like Proto-Indo-European *kʷekʷlo- 'wheel'. *kʷe- could have been simplified to a presyllable *ki- and then its vowel lost its palatality, becoming *ɯ. This presyllable might have conditioned the aspiration in the other reading of 車:

*kɯ-klja > *k-k(l)ja > *x-k(l)ja > > *kʰ(l)ja > *tɕʰjæ > *tʂʰɨa

The *-j- accounts for the palatalization of *k. Perhaps *tʂʰɨa is from a dialect which shifted medial *-l- to *-j- whereas *kø was from a dialect which lost medial *-l-.

1ko also vaguely resembles Tibetan 'khor-lo 'wheel', but I would expect Tibetan 'khor- to correspond to Tangut *goʳ (< *Nkh-) or *kʰoʳ with a retroflex vowel, not 1ko. Even if 1ko were borrowed from Tibetan before Tibetan developed kh- from *k-, its lack of retroflexion would remain an issue. The Tibetan word might also be of Indo-European origin, though I doubt that it and the Old Chinese *kla-words were borrowed from the same language at the same point in time. DERI-VE-TION: WHERE DID 'GO' COME FROM?

At some point last weekend, I realized that Tangut

1vɨe 'to go'

could be cognate to Old Chinese 于 *Cɯ-wa (Baxter-Sagart: *ɢʷa) 'to go' and 往 *Cɯ-waŋʔ (Baxter-Sagart: *ɢʷaŋʔ) 'to go'.

Tangut v- could be from *w- and -ɨe could be from *-a (see table below).

Degrees of raising and fronting of *-a in Tangut

(Long, tense, and retroflex variants are excluded. See Matisoff 2004 for examples which to my surprise did not include 'to go'.)

Fronted Fronted and labialized Partly fronted Nonfronted Nonfronted and labialized
Raised to high -i, -ɨi, -əi -iu -iə -ɨə -əu
Raised to near-high

Raised to mid -ie, -ɨe, -e

Raised to lower mid (?)

-ia -ɨa, -a

(There are no known cases of < *-a, but I would expect some since is the mid counterpart of which may come from *-a.)

So perhaps 'to go' was *CV-waH in pre-Tangut. The vowel (and consonant?) of the presyllable may have conditioned the raising and fronting of *-a. *w- either blocked from fronting or conditioned the centralization of *i:

*wa > *wɨa > *wɨe or

*wa > *wɨa > *wie > *wɨe

In either case, there was no *vie in Tangut. The combination of v and i was forbidden.

The presyllable of pre-Tangut *CV-wa might have been similar or identical to the presyllables of Old Chinese 于 *Cɯ-wa and 往 *Cɯ-waŋʔ which also conditioned raising:

*Cɯ-wa > *Cɯ-wɨa > *wɨa > *wua > *wuo > *wu > Mandarin

*Cɯ-waŋʔ > *Cɯ-wɨaŋʔ > *wɨaŋʔ > *wuaŋʔ > *ɔŋ > Mandarin wǎng

Tangut 1vɨe lacks a nasal vowel that would enable me to confidently reconstruct pre-Tangut *CV-waN with an *-N corresponding to Old Chinese *-ŋ. It is possible for Tangut oral vowels to correspond to non-Tangut nasals: e.g., Tangut

2miee 'name'

is cognate to Old Tibetan mying 'name', Burmese အမည် maɲ 'be named', and Old Chinese 名 *Cɯ-meŋ 'name' Borrowings of Chinese nasal syllables (e.g.,

2ʂɨẽ 'sage' < 聖; the character [apparently also a component of 'name' above] could be derived from the 耳 at the upper left of 聖)

have nasal vowels, but it is not clear why some native words retain nasality and others don't.

7.22.20:36: Tangut v- can also be from a lenited labial stop, so 1vɨe could be from an earlier *CV-Pa. However, the closest match I can find in STEDT is Proto-Kuki-Chin *pal 'to walk on', and it is unlikely that a root survived only in Tangut and distant Kuki-Chin.

Could be related to these 'Kamarupan' pay-words for 'to come' and 'to go' or these Kuki-Chin weel-words for 'to go (round)'? Again, I doubt that Tangut was the only non-'Kamarupan' Sino-Tibetan language which contained these words. ('Kamarupan' is an umbrella term encompassing Kuki-Chin.)

Pay reminds me of Siamese ไป pay 'go' and weel reminds me of Vietnamese về < *wel 'return'. There is no reason to assume the 'Kamarupan' languages borrowed their verbs from distant Tai or Mon-Khmer, so I regard these similarities as coincidental. Tangut 1vɨe could be a Sino-Tibetan-internal lookalike like Greek theos and Latin deus.

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