I was so tired when I wrote part 2 that I typed "labiovelar" instead of "labiodental" several times. Yes, I do know the difference between the two, and no, there is no labiovelar section of the Homophones dictionary ... though I'd like to keep an open mind about labiovelars in Tangut.

I am a little more closed-minded about labiodentals. I have my own hypothesis which I'll reveal eventually so I can test it in subsequent parts. Before doing that, I should explain the format of Homophones. Below is a modern-day calligraphic reproduction of page 9B of Homophones from Li Fanwen (1986: 672):

Page 9B of Homophones

And here it is again, with color-coding added:

Page 9B of Homophones with color-coding

The title here and elsewhere in Li Fanwen 1986 (i.e., on p. 240) appears to have an error:

TT3584 SMALL tsəy 1.40 (left) instead of

TT3590 LIGHT 'yiy 1.36 (right).

Only two strokes distinguish LIGHT from SMALL.

Following the title are (near-)homophone groups (generally) separated by small circles at the end of each group. I have coded each group with a different color:

Group 1: blue

Group 2: light green (note the lack of a circle at the end)

Group 3: dark green

Group 4: pink (continues on page 10, not reproduced here)

Each group on this page contains both 'level tone' and 'rising tone' syllables (according to the classification of their tangraphs in Tangraphic Sea and Precious Rhymes of the Tangraphic Sea). Tangraphs for level tone syllables are black, whereas tangraphs for rising tone syllables have been painted grey: e.g., the sole tangraph in group 4 on this page was pronounced with a rising tone. The presence of both level and rising tone syllables in groups indicates that

- members of each group were not truly homophonous, and the author's criterion for homophony ignored whatever the Tangut meant by 'tones' (phonation?)

- or that the author spoke a 'toneless' dialect of Tangut

The earliest extant edition of Homophones is dated 1132 (Clauson 1964: 60). Is it possible that one or more Tangut dialects had lost their tones between the compilation of the tone-based rhyme dictionaries and Homophones? Probably not, but the coexistence of tonal and atonal language varieties has a modern-day parallel among the Qiang: the northern Mawo dialect is toneless whereas the southern Taoping dialect has six tones (Sun 1981).

The large numbers of tangraphs in some groups tell us that tangraphs could not have represented spoken words. Group 1, for instance, contains 31 tangraphs. No spoken language could possibly have 31 homophonous words. If one equates tangraphs with words and assumes the author's dialect had tones, group 1 would contain 13 homophonous level tone words and 18 homophonous rising tone words. Some tangraphs represented words, but others represented syllables of polysyllabic words.

(I am speaking entirely from a Tangut A perspective. If Tangut B existed, tangraphic structure may reflect polysyllabic Tangut B words which were roughly semantically equivalent to monosyllabic roots in Tangut A.)

Next: What is the function of the subscript tangraphs in Homophones? Why do some subscript tangraphs appear on the left while others appears on the right? And why do some main tangraphs have two subscript tangraphs? LIP SOUNDS LIGHT (PART 2)

I ended last night's post with the question:

0 labiodentals for 264 labiodental-initial syllables?

Although the number of labiodentals is an open question, there is no question that Homophones' 'light lip sounds' section contained fewer than 264 syllables. Li Fanwen (1986: 928) counted only 248 entries - a figure which I just confirmed for myself. (Nishida [1964: 18] counted only 244 entries.) So where did the figure of 264 come from?

The second section of Homophones ends with these lines:




End of the Second Chapter: Light Lip Sounds

Two Hundred Sixty-Four Big Characters

Two Hundred Seventy-One Commentary Characters

According to Li Fanwen (1986: 928), the number of commentary characters (which I'll call clarifiers) is 271, though I have counted 272. A discrepancy of one is understandable, but the gap between 264 and 248 is harder to explain. The 'big' (i.e., main) character counts in most of the other sections are also inaccurate. Could these erroneous figures have been retained from some earlier edition of Homophones with a different number of tangraphs?

In any case, if Li Fanwen and Gong Hwang-cherng are correct and Tangut had no labiodentals, why would 248 main tangraphs be assigned to a category that did not exist in Tangut? Were tangraphs with, say, labial initials arbitrarily assigned to the 'light lip sounds' section to make Tangut phonology resemble Chinese phonology? ("We have labiodentals just like the Chinese! See?")

It is true that the Tangut did follow Chinese phonological models, but they were hardly slavish copiers. Yes, the arrangement of rhymes in the Tangraphic Sea does resemble that of the Chinese rhyme dictionary Guangyun (1008), but Guangyun lacks two features of Tangraphic Sea: its four 'rhyme cycles' and its Mixed Categories volume arranged by initial classes instead of rhymes.

The Mixed Categories volume of the extant version of Tangraphic Sea is not complete. Its labial (and labiodental?) sections are missing. The Mixed Categories in the Precious Rhymes of the Tangraphic Sea does contain very short labial and labiodental sections:

level tone, labials: 3-4 tangraphs (the higher figure includes a lost tangraph which might have had a labial initial)

level tone, labiodentals: 2 tangraphs

rising tone, labials: 1 tangraph

rising tone, labiodentals: 1 tangraph

Out of 938 tangraphs listed in the labial and labiodental sections of Homophones, only seven (0.7%!) were placed into the PRTS version of Mixed Categories.

Presumably the syllables represented by the tangraphs in Mixed Categories shared some characteristic(s) that

- rarely occurred with labial, labiodental, and velar initials (i.e., grave initials):

The number of velars in the PRTS version of Mixed Categories is also low: 6 level tone tangraphs and 6-8 rising tone tangraphs (the higher figure includes two lost tangraphs which might have had velar initials). The total of 12-14 is only 1.3-1.5% of the 911 tangraphs in the velar section of Homophones.

- prevented them from being organized by rhyme (ablaut?: i.e., ever-shifting root vowels? but why would there be very little ablaut after grave initials?)

It's still not clear to me why a large chunk of the Tangut syllabic inventory was catalogued in Mixed Characters instead of the tonal volumes.

Next: Is the labiodental category of Homophones a mixed category? LIP SOUNDS LIGHT (PART 1)

The title is taken from the second section of the Homophones dictionary:

TT0484 LIPS r 1.84

TT0099 SOUND Giẹ 1.59

TT3590 LIGHT 'yiy 1.36

'Light sounds of lips' are distinguished from the

TT0484 LIPS r 1.84

TT0099 SOUND Giẹ 1.59

TT4305 HEAVY lyïï 1.32

'heavy sounds of lips' in the first section of Homophones.

In traditional Chinese phonetics, 'heavy lip sounds' are labials (e.g., p, ph, b, m) and 'light lip sounds' are labiodentals (e.g., f, v). The Tangut terms are obviously calques of the Chinese terms, but one cannot necessarily assume that they refer to the same classes of consonants.

Fortunately, if one looks at the Tibetan transcriptions of the 'heavy lip sound' tangraphs from Homophones, one can verify that Tangut 'heavy lip sounds' were also labials (which could be preceded by other consonants): e.g. (H = Homophones page/side/column/graph number):

TT0447 CAUSE Tib tr. phi (H 04A24; Nevsky 1926: 46; #181)

TT1664 TREE Tib tr. phu (H 04B58; Nevsky 1926: 68; #267)

TT3709 ABOVE Tib tr. pho(H), phou (sic!; H 05A77; Nevsky 1926: 30; #114)

TT0422 UNITE Tib tr. Hpho (H 07B17; Nevsky 1926: 77; #297)

TT0903 BURN Tib tr. (H)bu (H 07A71; Nevsky 1926: 39; #151)

TT0446 VICTORY Tib tr. Hbu(H) (H 07B51; Nevsky 1926: 46; #182)

TT3905 ARRANGE-IN-ROWS Tib tr. dbuH (H 07A23; Nevsky 1926: 31; #117)

TT1309 HIGH Tib tr. HbheH (H 02B41; Nevsky 1926: 54; #213)

TT1648 BRILLIANT Tib tr. HbhiH, dbri (sic!; H 02A71; Nevsky 1926: 63; #247)

TT1273 HEAVEN Tib tr. mo (H 03A23; Nevsky 1926: 2 #7)

TT1257 HOUSE Tib tr. dmi, dme, mi, me (H 05B67; Nevsky 1926: 2 #5)

TT2934 PERSON Tib tr. rme (H 02A77; Nevsky 1926: 10; #39)

TT0991 FIRE Tib tr. gmuH, dmu, dmi (H 05A76; Nevsky 1926: 38; #149)

Strangely, Nevsky (1926) happened to lack transcriptions with a simple p, though nothing else indicates that Tangut had few or no p-initial words.

All Tangutologists agree that the 'heavy lip sounds' of Tangut were labials, as their name suggests, though they disagree on some details:

- Nevsky (1926) and 李新魁 Li Xinkui thought Tangut might have had a voiced aspirated bh like Indic languages (though Nevsky regarded b and bh as distinct phonemes and Li Xinkui didn't)

- Nishida, Sofronov, and 黃振華 Huang Zhenhua reconstruct prenasalized mb instead of Li Fanwen, Gong, and Arakawa's simple b (or Li Xinkui's bh)

The 'light lip sounds' are more controversial. In the introduction to the Tangut-Chinese Character Dictionary, Li Fanwen (1997: 12) asked,

Was there light labial sound in the Xixia [Tangut] language? Both Nishida and Sofronov think there were. The former cited [= reconstructed] [f], [v], and [the labiodental nasal] [M] [actually a nasalized [v]: [Mv]], while the latter cited [= reconstructed] [v]. Prof. Gong said there was none. Through the study of light labials in the second section of Homophones, I also think there was no light labial sound in the Xixia language.

More recently, Arakawa has reconstructed [f] and [v] (judging from the chart of his consonant system at 小高裕次 Kotaka Yuuji's Tangut site).

So did Tangut have three, two, one, or zero labiodentals? And if it had zero labiodentals, what was the point of the title of the second section of Homophones?

Next: 0 labiodentals for 264 labiodental-initial syllables? "DON'T LET IT BE LOST" (PART 6)

At the end of 1994, Li Fanwen met with Yuan Tianshu of the Founder Group in Beijing to learn how to computerize the Tangut script. But he returned home only to find that it was "impossible" to simply replace Chinese characters with Tangut ones. He was warned that he might not have enough funds to pay for the development of all-new, Tangut-specific software. His latest cost estimate was over twenty times the figure he had initially envisioned. Even if he did have that much money, there was no guarantee of success - and even if he did succeed, few others would want the software. Unable to see any mass market potential for Tangut input, Yuan Tianshu withdrew his support.

Li Fanwen was back to square one. Since computerization was not an option, he turned to a lower-tech method - photocomposition:

Clipping and pasting Xixia [Tangut] characters was much slower than typesetting. The whole work would take a hundred people to work on it for one and a half month[s]. I could not choose but organize a study class in the Ningxia University and recruit students who stayed in school for the winter vacation to do the job.

Somehow they did it, even though I doubt more than a few of them knew anything about Tangut prior to their labor-intensive 'vacation'. They - and many others - had made Li's dream a reality.

It took me more than twenty years to compile this dictionary ... The work was always in my mind. Sometimes I forgot everything around me except my work. Once I put a pot of meat on the stove to cook before starting to work on my dictionary. Soon I forgot all about it until I smelt a burnt odor and found the meat had already become charcoal. Once when I was riding home on my bike, still thinking of my dictionary, I ran into a standing car. No wonder Italian [actually French] linguist Scaliger (1540-1609) said, "An unpardonable wicked criminal should neither be sentenced to death nor be sent to forced labor, but should be condemned to compile a dictionary, because this kind of work consists of all the physical and mental sufferings and pains." Can anyone truly understand the sweet burden experienced in the compilation of a dictionary?

I wouldn't claim to. I can only express gratitude for Li Fanwen's effort and determination. Far from being even remotely comparable to "[a]n unpardonable wicked criminal", Li Fanwen is one of the stars of Tangutology. May his light shine on.

The end? No, just another chapter in the unending Tangut saga ... "DON'T LET IT BE LOST" (PART 5)

A little recap: Li Fanwen had long wanted to publish a Tangut-Chinese dictionary. He overcame hurdle after hurdle only to have his dictionary rejected for publication partly because he didn't have his own reconstruction. To figure out his own take on the Tangut sound system, he studied the Homophones dictionary - partly in bed after a traffic accident on the way to see fellow Tangutologist Nishida Tatsuo.

Li then studied the Timely Pearl in the Palm and wrote yet another book with Tangut reconstruction as his ultimate goal: The Pronunciation of Northwestern Dialects during the Song Dynasty [i.e., in the time of the Tangut Empire].

(The problem of reconstructing Tangut-period northwestern Chinese is fraught with the perils of circularity, since the best evidence for that language is in Tangut transcription.)

When Taiwanese Tangutologist Gong Huang-cherng visited Li in 1992, he saw the nearly completed manuscript of Li's dictionary:

He asked me [in a]very friendly [manner] if there were any problems still existing in the compilation of the dictionary. I told him that I was not sure the Xixia language had only 105 rhymes and that I was not satisfied with the sound transcription system [i.e., reconstructions] worked out by other scholars. He said, "When I was in Japan I wrote an essay in English and worked out a solution to this problem. If you agree, I'll provide your dictionary with my sound transcription system." In July the next year Prof. Gong sent me his Xixia sound transcription system by mail as he had promised. I immediately replaced my sound transcription system with his system, making the phonetic notation of my dictionary perfected.

After the dictionary was completed, I was confronted with two big problems.

The first was funding. The Regional Financial Department of Ningxia gave him the financial support he needed. But it couldn't help him with his second problem.

Next: The computerization crisis. "DON'T LET IT BE LOST" (PART 4)

Putting aside his uncompleted translation of the Tangraphic Sea, Li Fanwen moved on to another major Tangut dictionary,

TT0099 SOUND Giẹ 1.59

TT5087 SAME ləw 2.38

or Homophones*. He hoped that a study of Homophones would help him reconstruct Tangut for his own Chinese-Tangut dictionary. But then ...

In April of 1984 when I was making good progress in the study of Homophones, unfortunately, I had a traffic accident. I was knocked down by a vehicle when I was riding on my bike to a hotel to meet Prof. Nishida Tatsuo, a Japanese Tangutologist. The result was a fracture in my left thighbone. I was kept on bed for half a year, during which time I continued to write my book A Study of Homophones. Having completed my manuscript in 1985, I wrote in the preface to the book:

Lao Zi said: "Good fortune lies within bad." Recuperation at home provided me a quiet, favorable environment for making a concentrated study of Homophones. Homophones kept my company all the time and it witnessed all my suffering and happiness.

A Study of Homophones was published in 1986. I have a partial copy with me here in Xin Zexi and hope to give an online guided tour of Homophones someday. (I have the whole thing back in Hawaii.) The book contains Li's own reconstruction of Tangut.

Next: Whose reconstruction appears in his dictionary?

*Tangut has noun-adjective order: 'sound same' = 'same sounds' = 'homophones'. Contrast with its Chinese translation 同音 Md Tongyin 'same sound' with adjective-noun order.

TT5087 SAME ləw 2.38 looks like Old Chinese 同 long 'same' and may be a loan. 'Same' could be a derived meaning of 同 'together, join, assemble' (Schuessler 2006: 499). If the Tangut word were cognate, I would expect it to mean 'together', etc. (unless a parallel semantic shift occured in both Chinese and Tangut: 'together' > 'same').

Schuessler (2006: 499) linked 同 'together' (his OC dông) to Written Tibetan sdong-ba 'unite, join', but both Chinese-internal and loan evidence point toward an initial l- in 同 rather than an initial d- (Sagart 1999: 199; also note the spelling 筒 with the phonetic element 同 for OC 筩 'tube' which probably had an l-). "DON'T LET IT BE LOST" (PART 3)

After his dictionary was rejected for publication, Li Fanwen bounced back:

When I was beset with difficulties, leaders of the Ningxia Autonomous Region gave me once more help and support. The Regional Party Secretary Chen Bin, who was in charge of cultural and educational affairs of the region, asked Zhang Yuan, director of the propaganda department, to listen to my report and sent people to accompany me to go to Beijing to borrow a copy of A Sea of Characters [Wenhai / Tangraphic Sea]. After I got a photostat copy of A Sea of Characters, I began to translate it from Xixia into Chinese.

Intent on devising his own reconstruction of Tangut for his dictionary, he

went to the original residential areas of the Dangxiang [党項 Tangut*] people in Gansu and Sichuan to make an investigation of the remnants of the Xixia [Tangut] people ... [The 木 雅 Muya] language [of Liuba People's Commune] is different from Tibetan, but the pronunciatiion of their essential words and phrases is similar to that in A Concise Lexicon of Xixia and Chinese Languages written by Xixia people ... I made a record of their language [Greater Muya of Dawu County, which he thought was more Tangut-like than the 'Lesser Muya' of Liuba] and planned to write a book under the title of An Investigation and Study of the Mi-nyag [= Muya**] Language, but I completed only half of it (about 200,000 characters) and had to stop my work for some reason.

which was ...?

By 1982, Li had translated much of Tangraphic Sea into Chinese, only to find out that his

... Beijing colleagues had already finished the translation of the book.

I don't think his efforts were in vain, though. The act of translation gave him a deeper knowledge of the Sea that could not have been acquired through decipherment alone.

It would be interesting to compare his translation with the one that was published as Wenhai yanjiu (Research on the Tangraphic Sea, 1984).

Next: Yet more misfortune.

*Middle Chinese 党項 tang'Gæwng (> Md Dangxiang) was presumably an attempt to write the name underlying Tangut. Such a name didn't exist in Tangut (A) itself. The Tangut called themselves

TT3316 mi 2.10

Cf. Written Tibetan mi 'person': were the Tangut 'the people'?

Also cf. Tangut words for 'person' like

TT2934 PERSON myiry 2.68

TT0029 MAN myir 1.86

(cf. rGyalrong tU-rme 'person')

The latter tangraph looks like something plus HOLY but was analyzed in Tangraphic Sea as

right of TT1081 SON lyịy 1.61

right of TT1669 BROTHERS kụ 2.51

**Tibetan miñag 'Tangut' - a name that has survived the Tangut themselves (and has somehow been Mandarinized as 木雅 Muya) - may have been borrowed into Tangut as

TT3536 TT5745

myï 2.28 nyaa 2.18 'Tangut'

Both TT3316 and TT3536 in the Tangut words for 'Tangut' (and TT0029 MAN above) contain the tangraph

TT3301 HOLY shyïy 2.37 (a loan from Chinese 聖 'holy', pronounced something like shyẽy in the dialect known to the Tangut?)

(The left-hand element is not to be confused with TT3344 PERSON!)

though the Precious Rhymes of the Tangraphic Sea analyze TT3536 as

[the left side of] TT3316 TANGUT mi 2.10

and [the right side of] TT4496 WISDOM syịy 2.54 (Tib. transcription gseH [Nevsky 1926: 54, #215]) which looks like

TT2369 ONE-OF-TWO dzyiy 2.33

No native graphic analysis is available for TT3301 TANGUT mi 2.10 which appears to consist of TT3301 HOLY shyïy 2.37 plus a low-frequency right-hand element of unknown function. Grinstead (1972: 130) lists only eight tangraphs with it.

You can hear recordings of Muya at Guillaume Jacques' site.

Copyright 2002-2007 Amritavision