vum appears in only four tangraphs. I have already mentioned two of them in my last post:

5319 1tiẽ 'first half of 'sun'' (vumdex)

1772 1tiẽ 'bright, intelligent' (helvum)

Since the two tangraphs have nothing in common but their readings (unless 'intelligent' is an attribute of the sun in Tangut mythology), I initially concluded that vum was a phonetic for 1tiẽ. One might expect the other two vum-tangraphs to also have 1tiẽ-like readings, but they were both read 2tsoʳ:

1654 2tsoʳ 'foal' (hinvum; hin = 'horse')

3681 2tsoʳ 'ancestor; source; west?*' (dexceuvum; dex = 'person' elsewhere)

The analyses of these two 2tsoʳ-tangraphs are unknown.

I cannot see any obvious connection between 'sun' and 'intelligent' on the one hand* and 'foal' and 'ancestor' on the other. I suppose one could say 'ancestors are intelligent' or 'the sun sets in the west', but I'd rather not resort to such strained links. So for now, all I can conclude is that the shape vum is correlated with two distinct readings. There are no other 2tsoʳ-tangraphs, and there are no 1tsoʳ tangraphs at all. There are other 1tiẽ-tangraphs without vum:

1113 1tiẽ 'to placate, appease' (why is 'horse' on the left?)

1301 1tiẽ 'to spread out'

3046 1tiẽ 'nail'

*8.1.2:16: David Boxenhorn pointed out that 'bright' is the link between 'sun' and 'intelligent'. I had thought of that, but rejected it as an English-based metaphor. However, such a metaphorical use of 'bright' does exist in China: Md 聰明 congming 'intelligent' is literally 'hear well' + 'bright' (in the sense of light). Moreover, Chn 智 'wisdom' is composed of 知 'know' plus 日 'sun', and 1772 'intelligent' has a similar structure: hel 'know' + vum 'part of 'sun''.

**(Grinstead (1972: 149) translated 3681 as '(west)' with parentheses. Li Fanwen (2008) defined it as 'ancestor; source' in both English and Chinese but translated it as 西 'west' in this example of usage from Sea of Meanings Established by the Sages***:

2məʳ 2tsoʳ 1riaʳ 2nie

lit. 'source ? end east'

西本東末 'western root, eastern end'

Kepping used the term 'Root West' to refer to the pre-Buddhist religion of the Tangut (e.g., on p. 8 of this PDF). Could that term be based on 2məʳ 2tsoʳ?

***This title has been translated into English as Sea of Meanings Established by the Saints, but I prefer to avoid the term saint because it might imply that the book was religious or even specifically Christian, though it was neither. I'd like to examine this book, but I should finish the Golden Guide first. AN OBVIOUSLY INTELLIGENT YET OBSCURE ANALYSIS

The first tangraph of the Odic Tangut word for 'sun'


has the following analysis in Tangraphic Sea:


5319 1tiẽ 'first half of 'sun'' (vumdex) =

1772 1tiẽ 'bright, intelligent' (helvum; phonetic) +

3576 1swie 'clear, obvious' (dexduupak)

5319 and 1772 are the only two tangraphs with vum that are pronounced 1tie.̃ I'll look at other vum tangraphs next time.

The analysis of 1772 includes 5319. Which was created first?


1772 1tiẽ 'bright, intelligent' (helvum) =

1771 2siẹ 'wisdom' (helhae) +

5319 1tiẽ 'first half of 'sun'' (vumdex)

Could 1772 be semantic as well as phonetic, or is it just a semantically more positive phonetic than its homophones

1113 1tiẽ 'to placate, appease' (with 'horse' on the left - why?)

1301 1tiẽ 'to spread out'

3046 1tiẽ 'nail' (< Chn 釘, also with a T shape; with 'water' instead of 'metal' - why?)

Why do 5319 and 3576 have dex 'person'? According to the derivation of 3576, dexduu (not just dex) is from 'night' - not the time when the sun shines!


3576 1swie 'clear, obvious' (dexduupak) =

2111 1kõ 'night' (dexduudexcok) +

5692 1swew 'to shine, illuminate' (dosboupak)

The dex of 2111 is in turn derived from 2491 'night':


2111 1kõ 'night' (dexduudexcok) =

2491 1na 'night' (dexbeldex) +

0176 1nɨaa 'black' (duudexcok)

The dex of 2491 'night' is in turn derived from 3696:


2491 1na 'night' (dexbeldex) =

3696 1na 'first half of 1na-2raʳ 'tomorrow'' (dexbelpax) +

2111 1kõ 'night' (dexduudexcok)

And the analysis of 3696 takes us back to 2491:


3696 1na 'first half of 1na-2raʳ 'tomorrow'' (dexbelpax) =

2491 1na 'night' (dexbeldex) +

5539 1dzị 'to cross' (paxduu)

I still have no idea why dex is in these black/night tangraphs (or of all tangraphs!). For more on this set of characters, see these three posts. O MOON, O SUN!

The second tangraph of the Tangut ritual language (RL) word for 'moon'


has the following analysis in Tangraphic Sea:


0863 1ʔo 'second half of the RL word for 'moon'' (feadau) =

1668 1ʔo 'the surname O' (feapak; phonetic; no other ʔo tangraphs contain fea) +

2826 1lɨẽ 'second half of 1tiẽ-2lɨẽ 'sun'' (ciadexdau)

Why derive 'moon' from 'sun'?

I suspect the word

1tiẽ-2lɨẽ < *Cɯ-teN-teN-H?

for 'sun' is RL because Li Fanwen (2008: 464, 839) did not list any attestations of the word outside native Tangut dictionaries. Since those dictionaries do not explicitly indicate which words are RL and which words are CL (common language), I classify words according to their attestations:

1. I assume that words appearing exclusively in Tangut dictionaries and Tangut odes are RL. Note that I specify words rather than tangraphs. Tangraphs for RL words can also be used to write CL words.*

2. I assume that words appearing exclusively in Tangut dictionaries are probably RL, especially if they have CL synonyms: e.g., the CL synonym of 'sun' is

2449 2bəi (giigeo; shares no components with 1tiẽ-2lɨẽ; gii = 'waist' by itself and 'bird' elsewhere but not here; geo = 'sage'; also in CL geoboucin 'moon')

These synonym equations can be found in Tangraphic Sea definitions: e.g., 2bəi is one TS definition of 1tiẽ, the first half of RL 1tiẽ-2lɨẽ 'sun'.

3. Words that appear outside Tangut dictionaries and Tangut odes are CL.

I am not comfortable with the term RL because I am not sure that RL was actually used in Tangut rituals. The term 'ritual' originates from Kepping's (2003: 26) observation that the words for 'ode' and 'ceremony' were homophonous:


2412 2dzio 'ode' (cirbiidex: cir 'water' + bii '?' + dex 'person' = words flowing like water?; note lack of 'speech' radical)

2179 2dzio 'ceremony' (dexboajem: dex 'person' + boa 'speech' + jem 'before' = an occasion when archaic speech was used?)

Since there is no doubt that 'RL' is in Tangut odes, I prefer to refer to it using David Boxernhorn's term Odic Tangut. However, as Andrew West pointed out, the abbreviation OT can be misinterpreted as Old Tangut, even though Odic Tangut coexisted alongside normal Tangut. Yesterday, I thought of describing RL as 'poetic', but PT (Poetic Tangut) looks like it should stand for Proto-Tangut or Pre-Tangut, a term I have often used on this site. ST (Special Tangut) looks like Sino-Tibetan.

My terms Tangut A and Tangut B are not synonymous with CL and RL. Tangut A refers to both CL and RL readings as recorded in Tangut dictionaries. Tangut B to refers to a hypothetical second language implied by the structure of both CL and RL tangraphs.

Earlier this evening, I came up with Lyrical Tangut (LT) and Default Tangut (DT). Just now, I came up with Elevated Tangut (ET) and Basic Tangut (BT). CL could be changed to CT (Common Tangut), though I'd rather not use its obvious antonym UT (Uncommon Tangut) since low frequency is not the only feature of RL. I am open to suggestions.

* The first tangraph of the RL word

1tiẽ-2lɨẽ < *Cɯ-teN-teN-H?

for 'sun' can also represent the second syllable of the CL word


for 'bodhi', presumably a loan from Chinese 菩提 *phuthi < *bod(i)ej even though the phonetic match is very weak.

If the Tangut word for 'bodhi' were a direct borrowing from Sanskrit, I would expect it to be something like

2bo-1di (tangraphs used for Sanskrit transcription of bo and dhi; see Grinstead 1972: 184)

which would be a very close match. (Tangut had no dh.) RED-FACED ORAL CAVITY

The first tangraph of the Tangut ritual language (RL) word for 'moon'


has the following analysis in Tangraphic Sea:


1846 1ka 'first half of the RL word for 'moon'' (tuepae) =

0758 1za 'the syllable za' (tuejiucin; cf. 1087 below) +

1847 1ka 'oral cavity' (buipae; bui = 'speech')

Nishida (1966: 245) glossed the left-hand radical tue as 'master'. tue also occurs in tangraphs such as

0998 2dziee 'teacher; master' (tuecun; cun = 'speech'; cognate to 1087)

1087 2dziee 'to teach' (tuejiu; cognate to 0998)

3320 1ɣɛw 'to study' (dextuejiu; < Chn 學; with dex 'person' added to 1087)

0758 is the first half of

1za-1giẹ 'Zage, the red-faced Tangut'

the RL equivalent of CL

2nieʳ 1nie lit. 'face red' = 'Nerne, the red-faced Tangut'

the name of what might be one of two main Tangut tribes.

0758 also occurs in the constellation name

1za 1ŋiẹ

in the Timely Pearl in the Palm which has no RL content as far as I know. I wonder if this instance of 0758 represents a CL morpheme 1za 'chamber' (cf. the translations in Nishida 1964: 192, 195) unrelated to the first half of Zage. No, wait ... the entire constellation name is homophonous with Zage. Could the tribal name and the constellation name be the same word with two different spellings? Was this constellation associated with the Zage? 1ŋiẹ is the CL word for 'constellation', so could the Zage be named after the 'Za constellation'?

1846 and 1847

are the only tangraphs containing the phonetic pae for 1ka. They demonstrate that RL and CL tangraphs can contain the same phonetic pae, just as 1846 and 0758 demonstrate that RL and CL tangraphs can contain the same semantic element tue.

7.29.2:41: 1847 has a circular derivation:


1847 1ka 'oral cavity' (buipae; bui = 'speech') =

0652 2khiə 'oral cavity' (buidoa) =

1846 1ka 'first half of the RL word for 'moon'' (tuepae)

The pae of 1847 had to be derived from 1846 since no other tangraph contains that radical. But it would have been interesting if pae had been derived from the top half of one radical and the bottom half of another.

Li Fanwen (2008: 112) regarded 0652 2khiə 'oral cavity' as a borrowing from Chinese 腔, but the rhyme of 腔 was transcribed into Tangut as -ɔ̃ R57 2.48, not R31 2.28 -iə (Gong 2002: 439).

The radical doa is unique to 0652. It vaguely resembles Chinese 言'speech'. 0652 has an unusual analysis which derives doa from the top of gix and all of bak:


0652 2khiə 'oral cavity' (buidoa) =

2193 2khiə 'fresh; even' (dexbui; phonetic) +

2575 1kwɪɪ 'a kind of bird' (cirgix; cir is 'water' but neither 'bird' radical is present) +

1627 1tew 'hook' (bakpik; is bak a drawing of a hook?) A LUNAR LOOK AT THE OTHER TANGUT

"The Ode on Monthly Pleasures" is written in Tangut ritual language (RL) with loose paraphrases in normal Tangut (CL = common language). Here are the first two lines with Kepping's translations:

RL line Tangraph

Li Fanwen number 1846 0863 5288 2479 2750 2082
My reconstructed pronunciation 1ka 1ʔo 2khie 1niooʳ 1ɣʊ 1ʔiəʳ
Tangraph gloss First half of RL 'moon' Second half of RL 'moon' pleasure origin head to ask
Word moon origin
Kepping's translation [Somebody] asks about the origins of the monthly pleasures.
CL line Tangraph

Li Fanwen number 2814 2814 2480 0856 0491 2511
My reconstructed pronunciation lhị lhị 1bɛɛ 2məʳ 2lɨọ 2riəʳ
Tangraph gloss moon moon pleasure origin where to arise
Kepping's translation Where does the origin of the monthly pleasures come from?

Numbers refer to positions in the lines: e.g., 1 = first tangraph in either the RL or CL line.

1-2. The differences in meaning are partly due to a need to keep the lengths of both lines the same.

The monosyllabic CL word for 'moon' is cognate to Written Tibetan sla, etc. I know of no cognates for the disyllabic RL word.

The CL and RL tangraphs for 'moon' share no components unlike the tangraphs for the following word.

3. The tangraph for the CL word for 'pleasure' looks like the tangraph for the RL word plus dex 'person' (why?). This implies that the RL tangraph was created before the CL tangraph. Was tangraphy originally meant for RL, or are there examples of RL tangraphs based on CL tangraphs?

4-5. The RL word for 'origin' contains the CL word for 'head' (cf. English fountainhead). CL 1niooʳ can mean 'origin' by itself and is defined in Tangraphic Sea as RL 2.

Tangut native sources do not specify whether any tangraph or word is CL or RL. RL words may be defined with CL words, and vice versa. The distinction between the two strata of vocabulary was first observed in modern times by Nishida (1986).

6. 1ʔiəʳ 'to ask' in the RL line can also be found in CL texts and is obviously cognate to

0268 2ʔiəʳ 'to ask' (with tun 'skin' instead of dex 'person' on the left)

Was there no RL word for 'to ask'? If the intent was to make the CL and RL lines rhyme, why not pick 0268 2ʔiəʳ which had the same rising tone (= second tone) as 2511 2riəʳ 'to arise'? HELP! DOES TANGUT RITUAL LANGUAGE HAVE A LIVING DESCENDANT?

Has anyone seen a recent conference abstract that equated Tangut ritual language with some obscure, presumably living language? I vaguely recall seeing such an abstract earlier this year, but never got around to blogging about it. I have been searching for it since Sunday. Did I just imagine it? THE GOLDEN GUIDE: LINE 68: TANGRAPHS 336-340

68: Can you spot the two tangraphs which share a source tangraph in their analyses?

Tangraph number 336 337 338 339 340
Li Fanwen number 2717 5469 0273 4828 1082
My reconstructed pronunciation 1lhɔ̃ɔ̃ 2giu 1le 2ŋiõ 2riʳ
Tangraph gloss grave; syllable in surnames the surname Gu; second syllable of surnames the surname Le; first syllable of surnames the surname Ngon; first half of the surname Ngon the surname Rir; second syllable of surnames
Word the surname Lhon'gu the surname Ngonrir
Translation Lhon'gu, Le, Ngonrir.

336: 2717 is a really unfortunate character for surnames, but perhaps the Tangut didn't take notice, just as few dread the surname Graves.


2717 1lhɔ̃ɔ̃ 'grave' (gaidexcin) =

0464 2so 'high' (feigak; gak is the right-hand variant of gai) +

3231 1dʒɨə 'ten' (dexzol) +

2716 1rieʳ 'skillful' (gaedumcin)

'High' might describe a gravemound, but what do 'ten' and 'skillful' have to do with graves?

Did 2717 really mean 'grave' even in surnames, or was it recycled for a homophonous syllable? If the latter, why not create a new tangraph just for Lhon in surnames?

337: 5469 (analysis unknown) is a straightforward combination of the phonetic giu (tas) and 'surname' (pux, short for 2885?).

338: Were the Le related to the Ve? Was the name of the Le cognate to 'beautiful' (which in turn could be a loan from Chinese 麗)?


0273 1le 'the surname Le' (herbambaeciecin) =

0745 2vɨe 'the surname Ve' (besbaxbilbaecin) +

0272 2le 'beautiful' (bumbamciecin)

339: The analysis of 4828 contains the usual source of 'horned hat':


4828 2ŋiõ 'the surname Ngon' (biocuojoi) =

4940 2ʔiə 'the surname Y' (biotumbee) +

0661 2ŋiõ 'sea' (cuojoi)

Were the Ngon related to the Y? Is the name Ngon derived from 'sea'?

340: Were the Rir related to the Ve like the Le?


1082 2riʳ 'the surname Rir' (besyik) =

0745 2vɨe 'the surname Ve' (besbaxbilbaecin) +

1383 ?1ʃɨe 'to give up' (fiayik; loan from Chn 捨?)

What is the function of 1383? I don't even know for sure what the reading of 1383 is. Sofronov (1968 II: 253, 279) listed this tangraph twice:

0021 rhyme 1.35, liquid initial (but this tangraph doesn't occur in the liquid chapter of Homophones; Sofronov added this tangraph by hand to his reproduction of two versions of this chapter)

0022 rhyme unknown, alveopalatal initial

1ʃɨe is my equivalent of the reading 1śjij from Li Fanwen (2008: 231), a composite of the features in Sofronov's two listings.

The placement of 1383 next to the 1.35 tangraph

3885 1lɨe 'lame'

in Precious Rhymes of the Tangraphic Sea suggests that they form a homophone group*. Hence one reading of 1383 may be 1lɨe. That reading isn't much like 2riʳ, so 1383 can't be phonetic in 1082. If it is semantic, does it mean the Rir were known for giving up? Shi et al. (2000: 104) glossed 1383 as 'to winnow' - an action associated with the Rir?

Moreover, the fact that 1383 is in the isolated characters section of the alveopalatal chapter of Homophones (40A) suggests that the other reading of 1383 cannot be 1ʃɨe, since an isolated character by definition must have a unique reading and there is a group of 1ʃɨe tangraphs earlier in that chapter (35A16-35A23). Could its second reading be closer to Chn 捨: e.g., 1ʃia with a Grade IV rhyme instead of the Grade II or III rhymes that normally follow alveopalatals?

*7.27.0:09: But in Homophones, 3885 is listed as an isolated character (55B51) without any homophones. THE GOLDEN GUIDE: LINE 67: TANGRAPHS 331-335

67: I have now reached the one-third point with tangraph 333:

Tangraph number 331 332 333 334 335
Li Fanwen number 4835 5529 2885 5297 4169
My reconstructed pronunciation 1ʔieʳ 2ʃwɨi 1phõ 1ʃɨõõ 1tshõ
Tangraph gloss the surname Er; first syllable of surnames the surname Shwi; second syllable of surnames the surname Phon; first half of the surname Phonngwi the surname Shon; first syllable of surnames desolate; boundless; syllable of surnames
Word Ershwi Shontshon
Translation Ershwi, Phon, Shontshon,

331: The analysis of 4835 has 4940 from line 66 as the source of its 'horned hat' (bio). Do any surname tangraphs with bio have analyses without 4940?


4835 1ʔieʳ 'the surname Er' (biobaezoo) =

4940 2ʔiə 'the surname Y' (biotumbee) +

3510 1ʔieʳ 'firm' (baezoo; zoo = 4995 'iron')

332: dou, the left side of 5529 (analysis unknown), is found in two other ʃwɨi-tangraphs and in one swi-tangraph:

5530 1ʃwɨi 'in a low voice'

5546 2ʃwɨi 'cooked food'

5441 1swi 'aunt'

dou is also in

0083 1vəi 'dragon'

from my post on Bruce Lee. I'll write more about that tangraph in a future post.

333: 2885 shares a phonetic dexpai with 3163. dexpai is unique to those two tangraphs.


2885 1phõ 'the surname Phon' (dexpaipux) =

3163 1phõ 'front of a garment' (dexpaibaxbelcin) +

2888 2mə 'surname' (dexpux; semantic)

334: 5297 shares a phonetic pam with 5144. No other pam tangraphs have similar readings.


5297 1ʃɨõõ 'the surname Shon' (pamdex) =

5144 1ʃɨõõ 'god' (pamjau; phonetic) +

2888 2mə 'surname' (dexpux; semantic)

Could the surname Shon be derived from 'god'? Was the Shon a clan of shamans? Was Shon- prefixed to such clans: 'the divine ...'?

335: The top radical box is normally semantic, but it's a phonetic according to the analysis of 4169:


4169 1tshõ 'desolate' (boxdikhae)

4134 1tshõ 'thorn bushes' (boxjeufer) +

1225 1vɨuu 'sad' (caibasbilbaedik) +

3424 2ləi 'fear' (balhaefal)

The choice of 'wood' (box) on the top of 4169 'desolate' may be influenced by the similarlly shaped radical艹 'grass' atop Chinese 荒 'wasteland'.

Tangut fonts by Mojikyo.org
Tangut radical font by Andrew West
All other content copyright © 2002-2010 Amritavision