1. Lexicographic equality

When I first got into Tangut 15 years ago, I was wholly unaware of the Lhwe-Mi distinction. I'd look at, say, the Tangraphic Sea entry for 0012,

the first half of

0012 5873 1bɨu 2kəụ 'brother(s?)',

see that it was defined as

2447 0605 2lɨo 2tiọ 'older and younger brother'

and assume that 0012 5873 and 2447 0605 were synonyms for 'brothers' in the same language. The Tangraphic Sea and Homophones do not label any entries as 'Lhwe', 'Mi', 'literary', 'vulgar', or 'archaic', though they do indicate some tangraphs are for Sanskrit.

The mixture of Lhwe and Mi words in the same dictionaries indicates that the Tangut thought of them as belonging to a single language. The dictionaries cannot be bilingual because the headwords are both Lhwe and Mi, and Lhwe and Mi are mixed in the definitions.

Were the dictionaries an attempt to unify two languages, to treat them as one? Why? Was there extensive code-switching? If so, it isn't evident from Mi texts that rarely contain Lhwe words. Of course, written language does not need to reflect spoken language. Educated Koreans and Vietnamese traditionally wrote in Classical Chinese (e.g., the 眉巖日記 Miam ilgi) but didn't speak it. Did the Tangut speak in their equivalent of Taglish, a blend of Lhwe and Mi, but generally try to write in 'pure' Mi?

2. Orthographic equality

There is no way to visually distinguish tangraphs for Lhwe and Mi words. Both are written with combinations of the same elements in the same layouts. Positive elements are not unique to Lhwe or Mi: e.g., Mi 'sage' is in the second halves of the Lhwe words for 'Lhweji' and 'brothers':


This particular example might lead one to believe that Lhwe tangraphs were derived from Mi tangraphs, but here is a Tangraphic Sea derivation of a Mi tangraph from a simple Lhwe tangraph:


1tsiəəʳ  'lychee' (Mi) = 'wood' (Mi; semantic) + 1tsiəəʳ 'fifth' (Lhwe; phonetic)

And here is a Tangraphic Sea derivation of a Mi tangraph from a more complex Lhwe tangraph:


1giu 'pig' (Earthly Branch; Mi) = 'pig' (Mi; semantic) + 2giu (second half of Lhwe 2lheʳ 2giu 'three'; phonetic)

The most common tangraph element of all

'person' (in 1,187 tangraphs - one out of five!)

is also an independent tangraph 'hermit' that only appears in dictionaries. Its limited attestation could mean* that it is a Lhwe word ... homophonous with


'person' (= 'heart' + 'hermit')

Derivations in both directions lead me to conclude that the tangraphs for Lhwe and Mi were devised simultaneously. Unlike the Chinese-based Vietnamese nom script, Lhwe graphs were not an outgrowth of a preexisting writing system. All elements I have ever seen in nom are recycled from sinography, whereas Lhwe elements exist and are even used to construct Mi graphs.

3. Engineered equality ... or lost literary equality?

Were the Tangut script and dictionaries politically motivated attempts to impose artificial unity upon a population divided into the 'black-headed' and the 'red-faced'? Did the literate simply refuse to write in Lhwe, in spite of a ready-made inventory of 3,000 graphs for it**? Or, as Andrew West wrote on 9.3 (emphasis mine),

And yet someone consciously and deliberately invented those 2-3,000 missing characters used for writing an unattested language [my 'Lhwe'], and scholars of the time (i.e. members of the Tangut elite) thought these characters important enough to include in all the various rhyme dictionaries and word books that have survived. To me, this does not sound like evidence that the missing common language was an unwritten spoken language, but rather that it was a written language that we have no evidence for simply because of the randomness of what texts have survived and what texts have perished without trace.

Last year, Andrew struggled with "The Mystery of Two Khitan Scripts" for a single Khitan language. Now he and I are struggling with the mystery of one Tangut script for what might be two Tangut languages. And as if that weren't enough, is anyone certain that the Jurchen 'small script' is simply assembled Jurchen large script (as in the example below), as opposed to a completely different, as yet undiscovered script?

'Country' in the TJK (Tangut, Jurchen, and Khitan) scripts

Tangut (Mi) 2lhiẹ
̣(Does a Tangut (Lhwe) word for 'country' exist?)

Jurchen (large script?) gur.un
(cf. Chn 囯 'country' and 土 'earth')

Jurchen (small script?) gurun
(same as large script but assembled into one square)

Khitan (large script) gúr
(identical to Chn 囯 'country')

Khitan (small script) g.úr

*9.11.00:33: But as Andrew West warned on 9.3,

I think one of the problems is that we don't really know which characters belong to which "language". Just because a character is unattested outside dictionary usage does not mean that it necessarily belongs to the mystery language [my 'Lhwe'].

So 'hermit' may be Mi after all.

**9.11.2:51: In spite of the invention of hangul,

Throughout the Late Middle Korean period, Hangul was not considered a primary medium of literacy. That role, after all, was served by Chinese characters and Classical Chinese, and the supremacy of Chinese writing remained unchallenged. As a result, the vernacular writing system was employed only as a practical, linguistic tool.

- Lee and Ramsey (2011: 111)

One could paraphrase this for Tangut:

Throughout the history of the Tangut empire, Lhwe was not considered a primary medium of literacy. That role, after all, was served by Mi, whose supremacy remained unchallenged. As a result, the graphs for Lhwe were mostly employed as a practical, linguistic tool: e.g., to gloss the Ode on Monthly Pleasures.

However, what Andrew wrote about the two Khitan scripts also applies to Tangut:

Both scripts are complex enough to require a considerable investment of time and effort to learn to read and write, so how is it possible that both scripts managed to coexist and flourish for so long ? Did the Khitan education system require students to learn both scripts, or were Khitan scholars only able to read and write one or other of the two scripts ?

In Tangut terms:

The graphs for both Mi and Lhwe require a considerable investment of time and effort to learn to read and write. Did the Tangut education system require students to learn graphs for both Mi and Lhwe, or were Tangut scholars only able to read and write one or other of the two languages?

Literate Tangut would have to be literate in both Mi and Lhwe to read dictionaries ... but how many of them actually read dictionaries? The existence of multiple editions of dictionaries - and the handwritten Precious Rhymes of the Tangraphic Sea - suggests that they were actually used.

I had briefly considered a Cantonese or Taiwanese-type situation in which Mandarin was like Mi and Cantonese or Taiwanese was like Lhwe. But thousands of graphs were never created for Cantonese or Taiwanese in traditional China, and nonstandard graphs for those languages were never used to construct graphs for the standard language. Lhwe and Mi graphs are intertwined in a way that has no parallels I can think of.

No, wait. Shong Lue Yang created one writing system (Pahawh) for two unrelated languages, Hmong and Khmu. Unfortunately,

We know nothing about the Khmu' system, not having succeeded in finding anyone who knows it.

- Smalley, Vang, and Yang (1990: 30)

Was the lost Khmu system intertwined with the surviving Hmong system? Unlike Lhwe and Mi, Hmong and Khmu have very different phonologies, so I assume Pahawh Hmong and Pahawh Khmu were visibly different: i.e., each had characters absent from the other.

I can only conclude with Andrew's final line which suits the mystery of the two Tangut 'languages' as well as the mystery of the two Khitan scripts:


(Konijn is Dutch for 'rabbit'. It's cognate to English coney. See below for kao.)

In part 8, I proposed a potential Lhwe-Mi doublet for 'elephant':

0021 1bɨu 'ox, elephant' (only in dictionaries and a name of a bodhisattva; has an unusual combination of b- + -ɨu - an indication of Lhwe origin?)

1119 2biu 'elephant' (common and hence Mi; could be an assimilated variant of Lhwe 1bɨu - do other unassimilated Lhwe : assimilated Mi doublets exist?)

Today, David Boxenhorn wrote,

My current favorite theory is that "Lhwe" is a warrior argot, with words derived in a variety of ways, something like http://en.wikipedia.org/wiki/Lunfardo or http://en.wikipedia.org/wiki/Klezmer-loshn [...]

And, of course, the same toolkit of derivation strategies would be the logic behind tangraphy.

What I meant is that if, say, kenning is a strategy for creating Warrior vocabulary, it could also be used for creating trangraphs for both Warrior vocabulary and Common vocabulary.

I'll have to get back to David's points about tangraphy some other time. For now I'll only say that graphs for Lhwe words share components with graphs for Mi words: e.g., 0021 was analyzed as a combination of two Mi graphs:


0021 'ox, elephant' (Lhwe) = 0125 2miuu 'ox' (Mi) + 1909 1gəuʳ 'cow' (Mi)

The unusual initial-rhyme combination of 0021 1bɨu suggests that it might be artificial - the product of a word game mixing pieces from different words: one with b- and another with -ɨu. Such an artificial word is what I'd expect to find in an argot.

On the other hand, it would defeat the point of an argot if the argot word (1bɨu) sounded almost like its nonargot translation (2biu). So either 1bɨu isn't an argot word or the argot theory is in trouble.

1bɨu could be an archaic Mi word that survives in a sacred name predating the merger of *-ɨu and *-iu in Tangut.

Near-homophony between Lhwe and Mi need not be a bad thing. It could be evidence for Andrew West's kenning hypothesis. On Wednesday, Andrew wrote,

I'm still doubtful that there were two separate languages. If the ordinary Tangut word for "bird" was glossed with a kenning as "wing clothes" [0673 5598 2thə 2gwi in the Ode on Monthly Pleasures], I wonder whether the ordinary Tangut word for "moon" could not have been glossed with a kenning as something like "silver disc" or "night light", or borrowing from Chinese, "great yin" or "abode of the rabbit". If such a kenning were used as a literary alternative for "moon" and/or "month", over time it might be written with its own distinct characters, and its pronunciation may even have diverged somewhat (cf. "Welsh Rabbit" -> "Welsh Rarebit") so that by the time the odes were glossed, "ka .o" [1ka 1ʔo in the notation on this site] might have become a literary term for "moon" and "month" even though its etymology may have been forgotten. L5786 "rabbit" is an almost homophone (only differs by tone) to the second part of "ka .o" (moon), but I can't find a suitable match for "ka".

Notice that the Lhwe word for 'elephant' and the second half of the Lhwe word for 'moon' have reversed tones:

L 1bɨu : M 2biu

L 1ʔo : M 2ʔo 'rabbit'

Was tonal reversal part of a word game? Could David and Andrew both be correct? Were Mi words, in isolation or in kenning combinations, processed through a word game (akin to rhyming slang?) to become Lhwe words? If so, can we reverse-engineer this lost word game and fill in the blank below?

Mi ___ 2ʔo '? rabbit' > Lhwe 1ka 1ʔo 'moon'

A massive searchable database of Mi texts might allow us to find an existing phrase ___ 2ʔo '? rabbit' that is a suitable candidate. But for now I'd be pleased with a searchable database of Lhwe texts: i.e., the odes. As David wrote,

9.3: If we had access to more of the known corpus of Tangut texts we might be able to solve some of these problems. It's shocking that we don't.

9.6: I have a feeling that we might find the answer, or at least hints about the answer, in the texts themselves. If we could get our hands on them.

If word games and/or kennings were involved, why weren't more words converted and/or paraphrased? Why not create a disyllabic substitute for, say, 1ɣʊ 'head'?

The kenning hypothesis seems to be at odds with Andrew's other hypothesis. I would expect kennings to be glossed in regular language rather than the other way around. What if the Mi lines were the glosses? But then why do the Lhwe lines seem so grammatically barren ... like glosses? Are those lines and dictionary entries really all that's Lhwe-ft?

Next: Enigmatic Equality

9.10.1:05: ADDENDUM: Tangut literati were very familiar with fanqie and could have used fanqie in their word games, just as they used fanqie to create new tangraphs (e.g., the transcriptive tangraphs in the supplement of Li Fanwen 2008).

I wondered if the fanqie for 0021 'ox, elephant' might be revealing:


0021 1bɨu =

3697 1bi 'plate, token' (< MIddle Chinese 牌 *bɛ? - but the rhymes don't match!; Mi) +

2318 ɨu, first half of

2318 3329 1ʒɨu 1khiaʳ 'upright'

known only from dictionaries and therefore a possible Lhwe word

Not unless 'upright plate' could somehow imply 'ox' or 'elephant'.

The graphic analyses of Tangraphic Sea have been described as fanqie. Could some of them contain hints about word games? REVISITING THE TANGUT RITUAL LANGUAGE 8: LHWE-TENT VOCABULARY?

It is difficult to draw a line between what I have been calling the 'Lhwe' and 'Mi' languages (or if you prefer, vocabularies). As Andrew West has shown, the 'Lhwe' text in the Ode on Monthly Pleasures contains Mi words: e.g., 1ɣʊ 'head'. I proposed that 1ɣʊ was a Mi loan into Lhwe. Can 1ɣʊ be said to be both a Mi and a Lhwe word? Are there three kinds of words in Tangut?

1. Only used in Mi and in dictionaries

2. Used in both Mi and Lhwe and in dictionaries

3. Only used in Lhwe and in dictionaries

Note that nearly all words appear in dictionaries.

What does 'only used in Lhwe' mean? Only used in the

2dzio 'odes'?

Can Lhwe be said to be 'Odic Tangut', as proposed by David Boxenhorn? What about words that are known so far solely through dictionaries? Are they Lhwe words that might be in undiscovered odes? Could they be Mi words in undiscovered non-odic texts? Could they be Lhwe and Mi words?

Last month, I mentioned

1885 'hunchback', 'waist'

which I've only seen in the B edition of Homophones (and Nevsky's dictionary which doesn't count). Does not appearing in known Mi texts qualify a word as being Lhwe? Not in this case, as this tangraph appears to be a variant of

3087 1dʒɨw 'waist'

which is widely attested in Mi texts.

But not all dictionary-only tangraphs are variants of more common tangraphs. I looked through the first entries in Li Fanwen 2008 and the first dictionary-only tangraphs I found were

0012 5873 1bɨu* 2kəụ 'brothers'

which the Tangraphic Sea equates with the Mi compound

2447 0605 2lɨo 2tiọ 'older and younger brother'

Should we conclude that 0012 5873 is the Lhwe word for 'brothers'? What if it doesn't appear in the known odes? Would it then be a Mi synonym for 2447 0605? Should we go even further and mechanically equate

0012 with 2447 'older brother'

5873 with 0605 'younger brother'

(the isomorphic fallacy?)

Or could 0012 5873 be a monomorphemic word 'brother' (singular and regardless of age)? I don't know. For now, I just want to make a three-way distinction among words only in dictionaries and other reference material:

1. Variant spellings of Mi words: e.g., 1885 for 3087 'waist'. Are there other examples? Are there variant spellings of Lhwe words as well?

2. Latent vocabulary (hence the title): synonyms of Mi words which could be Lhwe: e.g., 0012 5873 'brother(s?)' and 0704 1260 'camel' (the subject of a long series last month)

3. Tangraphs for transcribing foreign (presumably Chinese or Sanskrit) syllables

Many of these appear in the supplement of Li Fanwen 2008. Those supplemental tangraphs are not in the Mojikyo font I use, so I'd have to create special images for them. Not tonight.

I would expect special transcriptive tangraphs for syllables absent from native Tangut (Lhwe or Mi): e.g., 5995** 1tʃwɨə. But 5996 1tʃwɛ̃ is homophonous with the Mi word

0766 1tʃwɛ̃ 'hunchback'.

(Could 1885 be used to write a Lhwe synonym of 0766 as well as the Mi word for 'waist' normally written as 3087?)

Moreover, some foreign syllables similar to native Tangut syllables are not transcribed with special tangraphs: e.g., Sanskrit le was transcribed as

0046 2lie 'to see' (Grinstead 1972: 184).

Why transcribe some but not all foreign syllables similar to existing native syllables with special tangraphs?

Next: A Hare-y Hypothesis - or a Coney-ing Kao-meback?

*9.9.8:44: 1bɨu is an unusual syllable combining the Grade III rhyme -ɨu with the non-Grade III initial b-. It contrasts with Grade IV 1biu.

Normally there are no such Grade III/IV minimal pairs, which is why Gong reconstructed those two rhymes (R2 and R3) as -ju:

Tangut rhyme Gong This site
2 -ju (Grade III) -ɨu (Grade III)
3 -iu (Grade IV)

Do other Lhwe syllables also have unusual initial-rhyme combinations that distinguish them from the phonological pattern of Mi syllables?

It would be neat if all 1bɨu were (in) Lhwe words - or at least Lhwe-tent vocabulary. According to the Tangraphic Sea, 0012 1bɨu has two homophones which might be Lhwe:

5415 'intelligent' (a Mi word; no Sino-Tibetan etymology unless it's related to Written Tibetan dbu 'head'; in the same semantic domain as various Mi s-words with WT cognates; a loan from Lhwe?; its graph is phonetic in 0012)

0021 'ox, elephant' (no Sino-Tibetan etymology?; in dictionaries and in the name of a bodhisattva - an old partly Lhwe-based Buddhist term like 'Tripitaka'?)

0021 is nearly homophonous with the common Grade IV Mi word

1119 2biu 'elephant'

Are 0021 and 1119 cognates? Do they allow us to propose a sound correspondence

Lhwe -ɨu : Mi -iu?

Are they the same Lhwe word borrowed twice, once with adaptation (bɨu > biu) to fit Mi phonology and once without adaptation?

Or are they both members of a Mi word family?

There is one more bɨu with the other (= second / rising) tone:

0785 2bɨu 'border'

whose graph is phonetic in 5415 (which in turn is phonetic in 0012). This is a high-frequency word for 'border'. If bɨu was originally a Lhwe-only syllable, this word would have to be a borrowing into Mi.

In any case, there are no other combinations of labial initials with rhyme 2 -ɨu: i.e., pɨu, phɨu, mɨu. Why is bɨu the  sole exception to a constraint against Pɨu-syllables? There is nothing about b- that inherently makes it more 'friendly' to a following central vowel.

**9.9.8:45: Not the same 5995 as the 5995 of Li Fanwen (1997: 1086) which was the basis for the Mojikyo Tangut font. REVISITING THE TANGUT RITUAL LANGUAGE 7: ISOLATE OR SINO-TIBETAN?

I am struggling to come up with neutral names for the strata of vocabulary (or languages) in Tangut. The pairs 'common' and 'ritual' and 'superstratum' and 'substratum' entail assumptions that may be false.

Nishida's terms 'I' and 'II' are counterintuitive to me, as his 'II' refers to the high-frequency vocabulary whereas 'I' refers to the vocabulary that appears to have a lower frequency in known texts. I would expect 'I' to refer to the basic vocabulary and 'II' to refer to supplemental vocabulary.

I have already used the terms 'A' and 'B' in a different context. 'Tangut A' is the language whose phonology is recorded in dictionaries whereas 'Tangut B' is a hypothetical language underlying elusive elements of the script which seem to have no semantic or phonetic function.

I have been using the terms

'Mi' and 'Lhwe'

but they are not mutually exclusive because the term Lhwe appears in Mi texts (e.g., the preface of the Pearl which refers to both Mi and Lhwe), though according to Kepping (2003: 103) the opposite is not true. Moreover, as I noted in part 5, the Tangraphic Sea defined Lhwe as Mi, and the second tangraph for

Lhweji, the longer form of Lhwe*

contains the tangraph for Mi:


Ji = 'person' + Mi

The term Mi may be cognate to Written Tibetan mi 'person' and Old Chinese 民 *min 'people'. Perhaps the pre-Tangut word for 'people' became their ethnonym while

2dzwio < *Pɯ-dzoH of unknown origin

became the generic word for 'person'.

On the other hand, I don't know of any Sino-Tibetan cognates for Lhwe(ji). This is not necessarily meaningful, as I can't think of any Chinese cognates for other ST ethnonyms. Nonetheless I'm not surprised because I don't know of any ST cognates for Lhwe words, period. So for a long time I've thought that Lhwe was an isolate. However, as Andrew West wrote to me about part 2 (emphasis mine),

Your theory seems quite plausible, and fits in quite nicely with mine, but I fear both our theories suffer from the same shortcoming: mine resorts to an unattested lost text (or corpus of literature); and yours resorts to an otherwise unattested language isolate. I always feel that when you have to resort to a lost text or a language isolate to explain something you're on very unsteady ground.

I agree. However, as unhappy as I am with the isolate hypothesis, I can't think of any plausible alternative. Did the Tangut simply invent a new vocabulary ex nihilo like the Hawaiians in the European accounts that I mentioned in part 1? Even the Hawaiians didn't go so far as to invent 3,000 characters for invented words (and nothing in the accounts necessarily implies that thousands of new Hawaiian words were created). Or can the Lhwe words be derived from Sino-Tibetan after all? David Boxenhorn asked me,

Does Lhwe have Sino-Tibetan cognates that are NOT shared with Mi? (Note that these would have traveled a different path than the Mi cognates, so [they] might not be obvious at first sight.)

In theory there could be Lhwe-Mi doublets like father ~ paternal: nonhomophonous, even nonsynonymous words that share a common ancestor.

I haven't seen any cognates, but that doesn't mean there are no such cognates. I don't think that question can be answered until someone compiles a list of Lhwe vocabulary. So far, all words in Lhwe belong to three out of four categories:

Non-Sino-Tibetan Sino-Tibetan
Distinct from Mi
e.g., 1ka 1ʔo 'moon'
(none known - yet!)
Shared with Mi
e.g., 1jəʳ 'to ask'?

e.g., 1ɣʊ 'head' (cf. Written Tibetan mgo 'head' and Old Chinese 后 *goʔ 'ruler')

(If anyone knows of any possible ST cognates of 1jəʳ 'to ask', let me know. Many Mi words outside a solid ST core have no known cognates.)

Could there be a fourth category: distinct and ST? Even if such words existed, they might be early loans from Chinese and/or other ST languages, so their presence might not disprove the isolate hypothesis.

Next: Lhwe-tent Vocabulary?

*9.7.1:03: Kepping (2003: 103) regarded Mynya as the longer form of Mi


Mi : Mynya


Lhwe : Lhweji

despite a difference in rhymes as well as tangraphs. The My- of Mynya belongs to the -y rhyme group whereas Mi belongs to the -i rhyme group. (I use y to represent central vowels in my nontechnical Tangut romanization.)

The -i and -y rhyme groups may alternate in word families (Gong 2002: 58-60): e.g.,

1918 1mi 'not' (generic)

5643 1miə 'not' (for auxiliary verbs) = my in nontechnical Tangut romanization

(See "Pseudo-Sino-Tangut" for more on this word family.)

so Mi and the My- of Mynya may be cognate. However, the etymology of -nya < *nyak (judging from Written Tibetan Mi-nyag 'Tangut') is unknown, though there is a remote possibility that it may be cognate to the second syllable of Lhweji which Li Fanwen (1986: 376) once reconstructed as *ɲıe ~ *dʑıe. REVISITING THE TANGUT RITUAL LANGUAGE 6: SUMMER SAGES?

If my hypothesis in parts 2-4 were correct, the status of the elite Mi and the lowly Lhwe might have been reflected in the tangraphs for their names.

The tangraph for 'Mi' (analysis unknown) contains 'sage' on its left side:


So far, it's not surprising that the elite would write their name with 'sage'. As Kwyli Rirphu wrote in the introduction to the Pearl, "there are worthy men among the Mi" (as opposed to the Lhwe?).

The right side of 'Mi' (alphacode: bul; function unknown) appears in 56 other tangraphs. We'll see one of them below.

A note in the D edition of Homophones equates Mi with

'Mynya person'
The first tangraph of 'Mynya' looks like 'sage' plus 'one' (for paired body parts)


but its actual analysis in Precious Rhymes of the Tangraphic Sea is 'Mi' + 'wisdom'*:


The second tangraph of 'Mynya' was analyzed in Precious Rhymes of the Tangraphic Sea is 'great' + the surname On:


Why On? I don't know, but there's 'sage' again!

So apparently the Mi = Mynya considered themselves to be great wise people. One might expect some less flattering components to be in the name of the Lhwe(ji). The tangraphs for 'Chinese' contained

'bug' (alphacode: jio)

in them:

Even a tangraph transcribing syllables that sounded like Tangut period northwestern Chinese 漢 *xã 'Chinese' contained 'bug':

So were the


'bugs' or maybe a step above bugs - like 'beasts'? No.

The tangraph for 'Ji' (analysis unknown) looks like 'person' plus ... 'Mi'!


The most negative interpretation I can think of is that the (Lhwe)ji were 'human Mi' - not up to the level of the Mi who were like


1ɣʊ 'gods' (looks like 'sage' + the right of 'great'/left of -nya in Mynya, but was analyzed as the second half of 2thiə 2ɣʊ 'emperor' plus 'bird')

but certainly above the buggy Chinese.

The Tangraphic Sea analyzed the tangraph for Lhwe as 2lhi 'give birth' (phonetic as well as semantic?) + 'origin' = 'born (at their place of) origin', possibly implying 'indigenous person':


(9.6.2:24: The first source tangraph can also mean 'to regret' and the second can also mean 'ancestor'. 'Regrettable ancestor'? 'Those who gave birth to ancestors'?)

I suspect the left side of Lhwe

(meaning unknown)

may have been intended to resemble 夏 'summer', the Xia of 西夏 Xixia 'Western Xia', a Chinese name for the Tangut. Of course there is nothing indigenous about that name. According to Kepping (2003: 111), the Tangut did not use their native word 2dʒwɨe 'summer'

(resembling 夏 'summer' plus the mysterious final element ヒ)

as an autonym until after the fall of the Tangut Empire.

In short, the tangraphic evidence does not support the 'lowly Lhwe' hypothesis. The characters for Mi, Mynya, and even Lhweji all contain 'sage'.

Next: Was the 'Ritual' Language Sino-Tibetan?

*9.6.2:56: The words

1771 2siẹ < *sɯ-seH 'wisdom'

4993 2siə < *səH 'to know'

3469 2sie < *Cɯ-seH 'knowledge'

are cognate to each other and to Written Tibetan shes-pa 'to know'.

Pre-Tangut *-H may be from an even earlier final *-s. REVISITING THE TANGUT RITUAL LANGUAGE 5: A FALSE DICHOTOMY?

In school, I was taught to write a thesis statement, back it up, and conclude with a victory dance. My teachers didn't put it that way, but close enough.

I'm not in school anymore. Teenagers can convince themselves that they have The Answer, or at least something that could pass as one for their papers. I'm an adult now, and I can freely admit that I don't have The Answer - that the answer I've come up with so far is problematic at best if not outright wrong.

No victory dance here.

In parts 2-4, I have proposed that there were actually two Tangut languages:



Substratum spoken by commoners Superstratum spoken by elite
Isolate Sino-Tibetan
Contains borrowings from Mi: e.g., the Mi word 1ɣʊ 'head' in the Lhwe glosses for the Ode on Monthly Pleasures Contains a few borrowings from Lhwe: e.g., 2lheʳ 'three' in 2lheʳ 2ʔụ 'Tripitaka'
Many polysyllabic roots: e.g., 1ka 1ʔo 'moon' Mostly monosyllabic roots: e.g., 2lhị 'moon'
Similar phonology due to being in the same linguistic area (cf. the convergence of Dravidian and Indo-Aryan phonology)
Unknown syntax which cannot be determined from glosses on a Mi text Tibeto-Burman syntax with exceptions which might reflect Lhwe syntax

But did the Tangut themselves consider themselves to be two peoples instead of one? The textual evidence is ambiguous. The Tangut did distinguish between two groups, the 'red-faced' who were more common than the 'black-headed', but I am not aware of any text equating the Lhwe or Mi with either group. In fact, the Tangraphic Sea defines 'Lhwe' as

1. Lhweji (the longer form of Lhwe; Ji [reading uncertain] may also mean 'Tangut' by itself)

2. Mynya (an indivisible disyllabic term for Tangut; cf. Written Tibetan Mi-nyag 'Tangut', presumably borrowed from an earlier version of the name with a final velar stop)

3. Mi people (!)

Unfortunately, the Tangraphic Sea entries for Ji, My-, -nya, and Mi have been lost, but it would not surprise me if they were all defined as Lhwe. If the Lhwe(ji), Mynya, and Mi were the same people, what was the difference between those three terms?

Next: Summer Sages

9.5.2:10: ADDENDUM: The name of the Lhwe

is homophonous with 'face' in the 'Lhwe' term

2783 5315 2lhwiẹ 2ʃwɨii, lit. 'face red' = 'the red-faced'

consisting of words that also occur in 'Mi' texts. Is this a hint that the Lhwe were the 'red-faced'?


0883 2282 2lhwiẹ ?dʒɨi 'Lhweji'

a kenning, a different phrase containing the word 'face'? Ji has five homophones, but none plus 'face' is obviously plausible as a name for a people:

5173 'to pull up, rescue'

4742 'silk'

5604 'skill, art'

0905 (a surname - a different spelling for the name Ji)

4885 'to stretch'

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