Only one Puxi word with a velarized vowel in Jackson Sun's (2000) article "Stem Alternations in Puxi Verb Inflection" has a Grade II (i.e., velarized) cognate in Tangut:

gloss Puxi Tangut Tangut grade TT
fat tshəɣ tshwu R1 1.1 I 5234
hail lməɣ mur R80 2.69 0599
wear a hat tʃʌɣ tɕeɣj R35 2.31 II 0209
lid spuɣ phjuu R7 1.7 III 1667
melt dʒvəɣ* tɕhjwi R10 1.10 (v.i.) 3011
dʑjwi R10 1.10 (v.t.) 3012

(Underlining in Puxi indicates a low tone instead of pharyngealization.)

The sole instance of matching velarization could be due to chance.

There may be a correlation between Tangut labiality and Puxi velarization. Four of the six Tangut cognates have labial vowels and three cognates have the labial glide -w-. This is not entirely unexpected, since Puxi velarization corresponds to labiality in other rGyalrongic languages (Sun 2000: 215):

gloss Puxi Geshizha (dGe-shi-rtsa) Mu'erzong ('Brong-rdzong) Caodeng (Tsho-bdun) Zhuokeji (lCog-rtse)
fat tshəɣ tshuə tshoʔ tsho tsho
hail lməɣ lmu lmoʔ tə-rmu tə-rmo
melt dʒvəɣ* dʑə dʑə ndʒwiʔ --

Perhaps pre-Puxi labial vowels (*U) developed a preceding *w which conditioned velarization:

*U > wU > *ɰU > *ɰUɣ > Vɣ

A pre-Puxi labial glide became a labiodental fricative in 'melt' which has a nonlabial vowel:

*wi > *wiɣ > ɣ

A transitional stage may have been something like Zbu kɐ-ftɕhɣwiʔ 'cause to melt' with -ɣw-. Unlike Puxi velarization, Tangut velarization has nothing to do with labiality. Tangut preserves earlier labial vowels. See Gong, "The System of Finals in Proto-Sino-Tibetan", p. 44, for examples.

Moreover, the distribution of velarization among vowels in Puxi and Tangut is different:

Puxi vowels (* = no known velarized counterpart):

i * u uɣ
e eɣ ə əɣ o oɣ
ʌ ʌɣ ɔ ɔɣ
æ * a aɣ

Tangut vowels (excluding the tense and retroflex series; * = no known velarized counterpart)

i iɣ u *
e eɣ ə əɣ o oɣ
a aɣ

Puxi has no velarized i, whereas Tangut may have no velarized u. In the past, I have proposed that R4 (Gong's Grade I -u) is Grade II, but now I wonder if pre-Tangut *uɣ became əɣ or oɣ.

Next: Return to R4.

*The Puxi word for 'melt' appears as velarized on p. 219 and dʒvəw as on p. 217. I presume that, superscript w is a typo for a superscript gamma. WAS GRADE II VELARIZED? (PART 2)

If my velarized interpretation of Grades II and III from part 1 is correct, the other two grades of the Middle Chinese were nonvelarized:
nonvelarized velarized
relatively high vowel Grade IV Grade III
relatively low vowel Grade I Grade II

One can view the four grades as a progression:

I > II (nonvelarized to velarized)

II > III (low to high)

III > IV (velarized to nonvelarized)

Here's an example of the four grades based on the Shouwen manuscript (Pulleyblank 1984):

nonvelarized velarized
relatively high vowel kjew kɰeɣw
relatively low vowel kɑw kaɣw

Grades III and IV had merged into a single Grade III in Tangut period northwestern Chinese, and the high/low contrast was reinterpreted as palatal/nonpalatal:

nonvelarized velarized
palatal 嬌澆 kjew (none; 嬌 kɰeɣw had become kjew)
nonpalatal kɑw kaɣw

A similar three-way contrast developed in neighboring Tangut:

Grade I: neutral Grade II: velarized Grade III: palatalized
u (none) ju
i iɣ ji
a aɣ ja
ə əɣ
e eɣ je
o oɣ jo

(For simplicity, I use the same vowel symbols in all three grades, even though I suspect that each vowel had different allophones for each grade.)

Like Chinese Grade II, Tangut Grade II may have originated from ɰV. The absence of a Grade II uɣ is not surprising since pre-Tangut ɰu may have merged wtih u before ɰV became Vɣ.

Unlike late Old Chinese ɰ, pre-Tangut ɰ cannot be from an even earlier r, since r conditioned vowel retroflexion which could occur in all three grades.

Perhaps the three grades of Tangut were conditioned by vowels in lost presyllables:

type of presyllable palatal main vowel nonpalatal main vowel
pre-Tangut Tangut pre-Tangut Tangut
none >
Grade III for palatal vowels and
Grade I for nonpalatal vowels
Ci Cji Ca Ca
high-vowelled > Grade III CɯCi > CiCi Cji CɯCa > CiCja Cja
mid-vowelled > Grade II CɤCi > CɤCɰi Ciɣ CɤCa > CɤCɰa Caɣ
low-vowelled > Grade I CʌCi Ci CʌCa Ca

The vowels of lost presyllables in Old Chinese had somewhat different effects:

type of presyllable high vowel low vowel
early OC late OC early OC late OC
none Ci Ci Ca Cʕɑʕ
high-vowelled CɯCi Ci CɯCa Cɨa
low-vowelled CʌCi Cʕeiʕ CʌCa Cʕɑʕ

Next: Testing the velarized Grade II hypothesis. WAS GRADE II VELARIZED? (PART 1)

Tonight I finally got around to installing Mojikyo and the latest version of Sven Osterkamp's Tangut progream on my new computer. Until now I've been using my old XP laptop to access those goodies. Unfortunately, this post doesn't use them at all. I want to write down my ideas about Grade II in both Chinese and Tangut while they're still fresh in my mind. Here goes ...

In stage 1 Old Chinese, there were six vowels in accented syllables and two reduced vowels (ɯ, ʌ) in unaccented presyllables:

high i ə ɯ (< unaccented high vowels) u
low e a ʌ (< unaccented low vowels) o

(This 6 + 2 arrangement is identical to my early Korean vowel system. Note, however, that Korean had no presyllables, and its ɯ and ʌ were not necessarily minimal vowels.)

In stage 2 OC, each of the accented vowels developed (raised) nonemphatic and (lowered and backed) emphatic allophones:

high nonemphatic i ɘ u
emphatic ɪʕ ʌʕ ʊʕ
low nonemphatic e ɐ o
emphatic ɛʕ ɑʕ ɔʕ

ɯ was the vowel for nonemphatic presyllables and ʌʕ was the vowel for emphatic presyllables.

In stage 3 OC, nonemphatic and emphatic vowels began to 'bend' in opposite directions (cf. Schuessler 2007), becoming diphthongs:

high nonemphatic i ɨɘ u
emphatic eiʕ ʌʕ ouʕ
low nonemphatic ie ɨɐ uo
emphatic ɛʕ ɑʕ ɔʕ

In stage 4 OC, emphasis was lost, and the vowel allophones became phonemic:

high former nonemphatic i ɨɘ u
former emphatic ei ʌ ou
low former nonemphatic ie ɨɐ uo
former emphatic ɛ ɑ ɔ

By this point, presyllables were mostly lost or fused with accented syllables, and those presyllables that did survive lost the ɯ : ʌʕ vowel distinction.

Two changes must predate the next stage:

- Metathesis of rC- clusters

- Retroflexion of coronals next to -r-

e.g., rt- > tr- > ʈr-

or rt- > rʈ- > ʈr-

In stage 5 OC, medial -r- weakened to the velar glide -ɰ- which disappeared before ɨ and u. The back vowels ʌ and ɑ became central ə and a to assimilate to -ɰ-, the consonantal counterpart of the central vowel ɨ.

high former nonemphatic ri > ɰi rɨɘ > ɰɨɘ > ɨɘ ru > ɰu > u
former emphatic rei > ɰei > ɰʌ > ɰə rou > ɰou
low former nonemphatic rie > ɰie rɨɐ > ɰɨɐ > ɨɐ ruo > ɰuo > uo
former emphatic > ɰɛ > ɰɑ > ɰa > ɰɔ

In stage 6 OC, the eight ɰV combinations were reduced to five:

mid nonback ɰei, ɰɛ, and ɰə merged as ɰɛ

mid back ɰou and ɰɔ merged as ɰɔ

Part of future Grade III (original -r- nonemphatics) ɰi ɨɘ u
ɰe ɨɐ uo
Future Grade II (original -r- emphatics) ɰɛ ɰɛ ɰɔ

In southern Middle Chinese, ɰa merged with ɰɛ (possibly via an intermediate stage ɰæ).

Part of Grade III (original -r- nonemphatics) ɰi ɨɘ u
ɰe ɨɐ uo
Grade II (original -r- emphatics) ɰɛ ɰɔ

In northwestern Middle Chinese (and perhaps in parts of the south as well?), the velar glide -ɰ- conditioned velarization of the following vowel before disappearing:

ɰV >ɰVɣ > Vɣ

In northeastern Middle Chinese dialects like the source dialect of Sino-Korean and the ancestor of Mandarin, -ɰ- shifted to -j-:

界 NEMC kɰɛj > kjɛj: borrowed as SK kjəj > now 계 ke; became Md jie

cf. -j-less Kan-on SJ kai borrowed from NWMC kaɣj (< ɣj < kɰɛj)

交 NEMC kɰaw > kjaw, borrowed as SK kjo(w) > now 교 kjo; became Md jiao

cf. -j-less Kan-on SJ koo < kau borrowed from NWMC kaɣw

Sino-Vietnamese is based on a southern MC dialect which had also undergone a -ɰ- to -j- shift:

界 SMC kɰɛj > kjɛɛj, borrowed as SV kjəəj > now 계 giới

交 SMC kɰaw > kjaaw, borrowed as SV kjaaw > now giao

Cantonese lost the -j-:kaaj,kaaw.

Perhaps the SMC Grade II long vowels originated from diphthongs:

ɰV > ɰəV > jVV (schwa assimilated to the following vowel) > Cantonese VV

Such diphthongs might have arisen from a velar fricative-vowel sequence:

rVʕ > ʀVʕ > ʁVʕ > ɣV > ɰəV > jVV

Cf. how the rGyalrong velarized vowel rhyme *-aɣŋ became a fricative-vowel sequence -ɣo in Japhug (Jacques 2004: 232).

Next: The Tangut side of the story. ƯU-MISSION

I just realized why there are no Tangut Grade II rhymes with u. If Grade II was characterized by a velar glide ɰ- or diphthongs beginning with a high achromatic (nonpalatal, nonlabial) vowel -ɨ- or -ɯ-, then it might have been difficult to pronounce -ɰu, -ɨu, or -ɯu. At least some languages with similar vowels or glides do not combine them with labial vowels or glides: e.g.,

- Korean has 의 [ɯi] but no [ɯu]

I think Grade II R9 (Gong's -ie) was something like -ɯi

- Vietnamese -ưu has become [iw] in Hanoi (though it is still [ɨw] in other dialects)

Perhaps some -ju are from earlier *-ɰu.

But why would Tangut have so many medial velar glides or diphthongs with high achromatic vowels? GRADE II IN III, VI, AND IX

After discovering that z- never occurred in Grade II, I used the latest version of Sven Osterkamp's invaluable Tangut software to find that some (but not all) similar initials were either rare or absent from Grade II:

Homophones chapter Number of tangraphs covered in each chapter (Li Fanwen 1986) Gong's reconstructed initial Syllables in Grade II
III (dentals) 869 t- tiaa R23 2.20 x 2
tiọ R74 1.71 x 2, 2.63 x 1
th- none
d- diaa R23 2.20 x 1
diow R58 2.48 x 1
n- many examples
IX (liquids) 1038 l-, lh-, ʑ-
r-, z- none; TT4897 zierj R78 2.67 is a typo for ʑierj
VI (alveolars) 664 ts- none; TT4738 tsịj R64 1.61 is a typo for tsjịj; R64 is a Grade III rhyme
tsh- none; TT2305 tshie R8 2.7 is a typo for tshe; R8 is a Grade I rhyme
dz- none; TT0764 and 4445 dziã R26 2.23 are typos for dʑiã
s- none; TT5706 sie R9 2.8 is a typo for ɕie

If n-syllables are ignored, there are only seven Grade II syllables in chapter III of Homophones. And there are no Grade II syllables in chapter VI.

Until now, I have been assuming that Grade II was characterized by 'emphatic' vowels rather than Gong's medial -i-. The emphatic hypothesis cannot explain why nonsonorant dental and alveolar initials are so rare in Grade II. I would have expected the opposite since modern standard Arabic has emphatic dentals and alveolars (tʕ dʕ sʕ zʕ - and it's lʕ that's rare, whereas in Tangut, l- is common in Grade II!).

Middle Chinese Grade II also lacked dental and alveolar initials, with two exceptions: the common words 打 'hit' (*t-) and 冷 'cold' (*l-). At first I thought that the grades in the two languages were quite parallel, but a closer examination of the distribution of initials proved otherwise:

Language Grade t-, th-, d- n- ts-, tsh-, dz-, s- z- ʑ- l- lh- r- retroflexes palatals
Middle Chinese I y y y n n y not in Chinese n n
II 打 only n n n n 冷 only y n
III n n n n y y n y
IV y y y y n y n n
Tangut I y y y y y y y y not in Tangut* n
II only 7 exceptions y n n y y y n y
III y y y y y y y y y

For ease of comparison, I have regarded the Middle Chinese initial 日 as *ʑ-, an approximation of its northwestern pronunciation, rather than as *ɲ-, its southern pronunciation.

I have organized the differences between the two patterns of distribution below:

In Tangut Grade I but not MC Grade I z- (which patterns like s- and r- but not ʑ-)
In Tangut Grade II but not MC Grade II n- (MC had retroflex *ɳ- instead)
l(h)- is very common
palatals including ʑ- (MC had retroflexes instead)
In Tangut Grade III but not MC Grade III dentals and alveolars (but these were in MC Grade IV, and Tangut grade terminology was based on a Chinese dialect that merged Grades III and IV)

I've been struggling to figure out what Tangut Grade II might be. My current guess - only a few minutes old - is that Grade II might have been characterized by a velar glide -ɰ- or diphthongs beginning with a high central vowel -ɨ-. The palatals in Grade II may come from earlier dentals and alveolars:

*t(s)ɰ- > tɕɰ-

*t(s)hɰ- > tɕhɰ-

*d(z)ɰ- > dʑɰ-

*sɰ- > ɕɰ-

*zɰ- > ʑɰ-

(Is this phonetically plausible?**)

Perhaps the seven cases of nonaffricated t- and d- in Grade II originated from complex clusters.

n- and l(h)- should also have become palatal ɲ- and ʎ(h)-, but Gong's reconstruction lacks palatal nasals and liquids, so they remained unchanged.

3.5.1:00: If Grade II n- and l(h)- were palatal, they should not appear in fanqie initial spellers for Grade I syllables and vice versa, since Grade I had no palatal initials. However, fanqie indicate that Grades I and II both had n- and l-:

TT4033 nẽ R15 1.15 < TT1427 nioow R59 1.57 + TT1299 lẽ R15 1.15

TT0085, 4405 lia R18 1.18 < TT1960 lu 1.1 + TT0124 xia R18 1.18

TT5443 liọ R74 1.71 < TT0573 lẹ R68 1.65 + TT5862 liọ R74 1.71

No lh- appear in initial spellers for Grade I syllables. All extant lh-fanqie contain Grade II and III lh-initial spellers. This does not mean that lh- was palatal [ʎh] in those two grades. First, if Tangut had no [ʎ], it probably would not have a voiceless [ʎh]. Second, Grade I and Grade III lh-tangraphs spell each other in fanqie:

TT2712 lhu R4 1.4 < TT4960 lhji R11 2.10 + TT3706 du R4 1.4

TT4960 lhji R11 2.10 < TT2712 lhu R4 1.4 + TT5226 tji R11 2.10

However, the possibility of palatalized allophones of /n l lh/ in Grades II and III cannot be ruled out. Mixed-grade fanqie spellings may indicate that native speakers perceived /n l lh/ as being (phonemically) identical in different grades even if those initials were subtly phonetically different. (Cf. how English speakers perceive the /k/ in key and coup as the 'same' even though one is palatalized and the other is not.)

*Like all researchers known to me, I don't think Tangut had any retroflex initials, despite a chapter of Homophones devoted to 'tongue top sounds', the Chinese term for retroflexes.

** Some of these palatalizations also occurred before Late Old Chinese:

Middle OC *tɨ- > LOC *tɕ-

Middle OC *thɨ- > LOC *tɕh-

Middle OC *dɨ- > LOC *dʑ-


Middle OC *nɨ- > LOC *ɲ-

had no parallel in Tangut, and MOC alveolars

*ts- *tsh- *dz- *s- *z-

did not palatalize.

Pulleyblank (1984: 179) explained that

Dentals and alveolars are ... already [+front], and [their] palatalization means raising rather than fronting (Bhat 1978: 54).

There are, in fact, well-documented cases of the palatalization of dentals or alveolars before non-high central or back vowels ...

There is, thus, no real obstacle to the assumption that the condition for palatalization of dentals between Old and Middle Chinese was provided by the [+high] feature in the vowels i, ɨ, and u, rather than by a hypothetical j glide for which there is no direct evidence and which never appears in the transcription of foreign words or pre-Tang borrowings of Chinese words in foreign languages.

In Tangut, u did not condition palatalization. But if ɨ or ɰ conditioned palatalization in Grade II, which didn't -j- condition palatalization in Grade III? Why did Tangut have

tj- thj- dj-

tsj- tshj- dzj- sj- zj-

in Grade III in addition to

tɕ- tɕh- dʑ- ɕ- ʑ-?

Perhaps earlier clusters were reduced to simple dental and/or alveolar initials, but there are many instances of such nonpalatalized initials in Grade III, so I suspect that the solution lies partly in the rhymes. More on this later. Z-ERO EVIDENCE

Last night, I proposed that Gong's z- could have been a voiced lateral fricative [ɮ]. If this were true, I would expect Tangut z-words to correspond to lateral-initial words in other languages. I don't have the time to look for cognates of all the z-words, so I looked at Guillaume Jacques' list of gDong-brgyad rGyalrong cognates (now in print!) and found only two correspondences:

Gloss Tangut gDong-brgyad rGyalrong
'long' zjir 2.72 kɯ zri < Proto-rGyalrong *sr- (Jacques 2004: 319)
'leopard' zerw 2.78* kɯ rtsɤɣ

So far, there is zero comparative evidence for a lateral origin of Tangut z-, but the search has barely begun.

I think that Tangut z- could have had up to four types of sources:

1. *z-: e.g., 'child': Tangut zji 1.11

cf. Proto-Tibeto-Burman *za ~ *tsa 'child' (Matisoff 2003: 172)

but see type 3 below

2. *sr-: e.g., 'long' above:

*srji > *zrji > *zrjir > zjir

cf. Proto-Tibeto-Burman *s-riŋ 'long' (Matisoff 2003: 296)

3. *CV-ts-: e.g., 'leopard' above:

*rV-tsek > *rV-dzek > *rV-zek > *rzeɣ > *rzerɰ > *zerw

Possibly also 'child' above:

*CV-tsa > *CV-dzɨa > *CV-ziə > *Czjiə > zji

(The dating of the vowel and coda changes relative to these proposed onset changes is unknown, so the above sequences are highly speculative.)

3a. Tangut z might also have arisen from a lenited *-s-, though I don't have any word family or comparative evidence to support this guess. (Cf. Middle Korean -z- which arose from lenited *-s- as well as *-ts- [Martin 1992: 45].)

4. *sl-: e.g., zjịj 2.54 < ?*s-ljij-H 'collect', if it is cognate to ljij 2.33 'collect'

cf. Old Chinese *sl- > Middle Chinese z-

also cf. Proto-rGyalrong *sl- > Japhug zl- (Jacques 2004: 319)

Any explanation of the origin of z- must account for the curious fact that z- never occurs before Grade II rhymes in Gong's reconstruction**. This implies that Grade II had some phonetic feature that was 'hostile' to z-. Here's a list of all rhymes which could be preceded by z- according to Li Fanwen (1986: 195-199) with Gong's reconstructions of those rhymes:

Grade a ə e i ɨ u o
I R17 -a
R85 -ar
R90 r R8 -e
R15 -ẽ
R82 -er̃
i and ɨ cannot occur without -j- R1 -u
R80 -ur
why not?
(no -aw) (no -əw) R44 -ew
R93 -erw
(no -uw) R56 -ow
R97 -orw
II no z- in Grade II (R41 -iəj) high vowels absent from Gong's Grade II
III R67 -jạ ə and e cannot occur after -j- R11 -ji
R70 -jị
R84 -jir
R31 -jɨ
R72 -jɨ̣
R92 -jir
why not? R75 -jọ
R96 -jor
R21 -jaa R14 -jii R33 -jɨɨ
R92 -jɨɨr
?R7 -juu why not?
(no -jaj) R37 -jij
R64 -jịj
R79 -jirj
why not? (no -juj) (no -joj)

z- did not actually occur before Grade II R41. Li Fanwen's (1986: 479) reconstruction of TT1698 as zĩ 2.36 is an error for zĩ R37 1.36, equivalent to Gong's zjij R37 1.36. Hence I have added R37 to the table.

I have included R21 and R67 since Li (1986) reconstructed z- before those rhymes on p. 477, though he did not include them in his charts on pp. 195 and 198.

Although Li reconstructed Homophones 52B43-45 with z- before R7, Gong reconstructed those syllables with l- (B43, B45) and lh- (B44). Thus I have placed a question mark before R7.

I have written "why not?" in boxes where I would expect z-. z- did not appear before the following non-Grade II rhyme types (ignoring tenseness and retroflexion):

- Grade I -o

- Grade III -jɨj

- Grade III -ju (and -juu?)

- Grade III -joo (a rare rhyme)

These rhymes do not share any common characteristic. Perhaps the absence of z- before these rhymes is due to chance.

*There are two homophonous tangraphs for 'leopard' (TT2384 and 2385) pronounced zerw 2.78.

**TT4897 DEFICIENT zierj R78 2.67 is an error for ʑierj. Homophones lists TT4897 as a homophone of TT2439 DWELL ʑierj R78 2.67. 馬の皮膚を搔く TO SCRATCH A HORSE'S SKIN

is Nishida's (1997: 250) gloss for

TT4767 ljij R37 2.33

He viewed TT4767 as the start of a tangraphic derivation chain:


TT4767 ljij R37 2.33 > TT2071 CHAPTER tjij R37 1.36 > TT2071 IF tjij R37 1.36

There is no doubt that CHAPTER is phonetic in IF, though the function of HORNED-HAT in IF is unknown.

The relationship between SCRATCH and CHAPTER was obscure to Nishida.

I wonder if SCRATCH is phonetic in CHAPTER*. They share the same rhyme (R37) if tone is disregarded. Moreover, I have proposed that l- may sometimes derive from a *t- that lenited before a presyllable that was later lost:

*CV-tjij > *CV-ljij > ljij

If there are other tangraphic phonetic series with lateral/dental stop alternations, this may mean that the creator(s) of tangraphy constructed some characters on the basis of dialects which preserved the earlier dental stops.

*If SCRATCH is a split phonetic like 行 in 衡, what is the function of the vertical line (Nishida's tangrapheme 040)? I presume it is semantic. Unfortunately, Nishida (1966) has no gloss for this element. A wild, very remote possibility is that CHAPTER consists of a Tangut A phonetic (SCRATCH) surrounding a Tangut B phonetic (the vertical line).

3.3.1:09: Shi et al. (2000: 259) defined TT4767 as 跳 JUMP which is very different from SCRATCH.

The Precious Rhymes of the Tangraphic Sea has what appears to be a graphic analysis of TT4767:

(top left of) TT4716 LEAP-OVER R24 gjaa 1.23

the Mojikyo font has an extra dot absent from Nevsky (1960: 195), Nishida (1966: 504), Grinstead (1972: 72), and Li Fanwen (1986: 334)

(left of) TT4779 ENTER R33 bjɨɨ 1.32

They suggest a meaning JUMP-INTO. Their (top) left element is Nishida tangrapheme 230 TILT. I don't see what TILTing has to do with LEAPing, ENTERing, SCRATCHing, or JUMPing.

No analyses are available for CHAPTER or IF.

Li Fanwen (1986: 464) defined TT4767 as 集 COLLECT in addition to 跃 JUMP (not the same sinograph as Shi et al.). I suppose one could view a CHAPTER as a COLLECTion of items.

COLLECT and JUMP could be unrelated homophones written with the same tangraph. COLLECT may have a lateral root initial and be cognate to

TT2810 COLLECT zjịj R64 2.54 < ?*s-ljij-H

Note that the initial that Gong reconstructed as z- was grouped together with l- and other liquids in Homophones. Perhaps z- was a voiced lateral fricative [ɮ].

Nevsky (1960 II: 233) had no gloss for TT4767.

Tangut fonts by Mojikyo.org
All other content copyright © 2002-2008 Amritavision