When Li Fanwen

was to be sent to Beijing to study the Xixia [Tangut] language from Mr. Luo Fuyi [羅福頤] ... [he] had already made more than 30,000 single morpheme [sic; did he mean 'words'?] cards from materials concerning the Xixia language ...

Those cards already had four-corner numbers* assigned to each of the tangraphs in

preparation for the compilation of a "Xixia-Chinese Dictionary" in the future.

I presume they were written during his years pursuing Tangut studies at Ningxia University and at Tangut mausoleum excavation sites in the early 60s.

After arriving in Beijing in May 1973, Li benefited from the vast generosity of Luo Fuyi:

He let me use all the precious books collected by his family [of three Tangutologists**] ... I felt so happy as if I had found a treasure. I copied all the materials night and day with great eagerness.

And by "copied" I don't think he meant 'used a photocopier'.

He completed the first draft of his Tangut-Chinese Character Dictionary in 1976 (needless to say, in longhand, I assume - no Mojikyo fonts or PCs in the PRC back then!). Alas,

The publishing house made three copies of the draft dictionary and sent them for experts' examination. In the end, the publishing house decided not to publish it because I used the transliteration of different schools [i.e., Li didn't come up with his own reconstruction of Tangut but presumbably cited those of Nishida and Sofronov?] and did not use sufficiently the materials from A Sea of Characters (Wen Hai), a Xixia language lexicon [i.e., the Tangraphic Sea].

I'd love to see a draft if one is still around.

Next: Another disappointment.

*Maybe I could use the four-corner method as the basis of the Tangut alphacode. Or at least convert Li's four-corner numbers for tangraphs (which underlie the Mojikyo font and the Tangut GIFs on this blog) into four-syllable strings that are easier to learn!

** 羅 福成 Luo Fucheng (1885-1960) and 羅福萇 Luo Fuchang (1895-1921) as well as 羅福頤 Luo Fuyi (1905-1981). "DON'T LET IT BE LOST" (PART 1)

Seven years ago, I saw a copy of 李範文 Li Fanwen's 夏漢字典 Tangut-Chinese Character Dictionary in Guillaume Jacques' Paris apartment. I never saw the book again until I visited a library on Wednesday and found a signed copy. I only had a half hour to look at it. I could spend half a lifetime studying it.

Li Fanwen spent most of his life working on it.

Its English-language preface begins with this anecdote (p. 7):

Premier Zhou Enlai made an inspection of the Chinese History Museum in January 1972. When he came across the Xixia [= Tangut] documents, he asked: "At present how many people are there who know the Xixia written language?"

"Only a few old scholars do," answered Wang Yeqiu, director of the State Bureau of Cultural Relics.

Premier Zhou instructed immediately, "You must assign people to study this language and don't let it be lost."

Li Fanwen was one of those people.

Next: The first draft. RETROFLEXION: RIGHT OR WRONG? (PART 4)

Nothing is ever easy when it comes to Tangut.

In part 3, I looked at Guillaume Jacques' proposed Tangut-rGyalrong cognates and found that Tangut retroflex vowels corresponded to rGyalrong r in 35 out of 41 cases. Obviously, rGyalrong preserved an r that had been reduced to a vocalic feature in Tangut. Right? Maybe ... but what about the 48 cases in which rGyalrong r does not correspond to Tangut retroflexion? rGyalrong only corresponds to Tangut retroflexion a little over half the time.

Moreover, what about the five cases in which rGyalrong retroflex consonants (presumably from earlier consonant clusters with r) do not correspond to Tangut retroflexion? E.g.,

rG kU ChëG 'six' : T chhyiw 1.46 'six'

which are from some earlier form like consonant + ruk: cf. Old Chinese 六 (mə?)ruk 'six', Written Tibetan drug, Written Burmese khrauk (all sharing a root ruk plus various prefixes)

The situation is almost certainly too complicated to be resolved in a single short blog post, but I'll give it a try. I'm going to assume that not all rGyalrong r are original: i.e., they are not all inherited from the common ancestor of rGyalrong and Tangut (which I'll call Proto-rGyalrong-Tangut = PrGT 'per gut'). There is no reason to expect rGyalrong and early Tangut to have r-affixes in exactly the same places: cf. how German has verstehen 'understand' with ver- corresponding to English understand; the prefixes are different (ver- [: Eng for-], under- [: Ger unter]) but the roots are cognate.

Here are two cases in which rGyalrong has an r-prefix absent in other languages:


Old Chinese 名 meng

Written Tibetan ming < Old Tibetan mying

Written Burmese mañ 'to name'

Tangut myiiy 2.35, 2.25

but rGyalrong të rmi with r-


(Old Chinese 民 min 'people', considered by some to be mi-n, is from ming and is not cognate; see Sagart [1999: 135])

Written Tibetan mi

Tangut myi 2.10, myï 2.28 'Tangut' (< 'the people'?; TT3316 and TT3536; not in Guillaume's list), myï 1.30 'he' (< '[that] person'?)

but rGyalrong tU rme with r-

Note that there are also two Tangut forms for 'person' with retroflex vowels: myïr 1.86 'person, people' and myiry 2.68 'person' (TT2934, not in Guillaume's list; appears in Tibetan script as rme, rmi with r- [Nevsky 1926 #38])

It seems that there was a proto-Tibeto-Burman root mi 'person' with or without a prefix r-:

Unprefixed mi > WT mi, T myi 2.10, myï 2.28, myï 1.30

Prefixed r-mi > rG tU rme (with an added tU-), T myïr 1.86, myiry 2.68

Similar explanations may apply to some other cases of rGyalrong r.

In other cases, I'm not sure if the rGyalrong and Tangut words are truly cognate: e.g.,

rG kU Grum 'white' : T phiow 1.55 'white'

But I may not be able to see the relationship because of my ignorance of rGyalrong.

As for the lack of correspondence of rGyalrong retroflex consonants to Tangut retroflex vowel rhymes, perhaps vowel retroflexion took place in some clusters but not others:

r + non-coronalr + coronal
(early, pre-)PrGT: r-clustersrmtr
(later) PrGT: r-non-coronal cluster intact;
some r-coronal clusters (e.g., tr-?) fused into retroflex consonant (Ch)
(but other r-coronal clusters are intact in rGyalrong: r-initial clusters, tsr-, ndzr-, sr-, zr-, shr-, zhr-)
retroflex consonant retained in rG but shifted to palatal (ch) in TrG rm, T rmrG Ch,
T ch (no vowel retroflexion in T)
r in syllable triggered vowel retroflexion in TrG rm, T rm...r
r-loss in T (in at least some dialects; the Tibetan transcriptions rme, rmi for TT2934 'person' suggest retention in other dialects)rG rm, T (r)m...r

rGyalrong retroflex consonants (and palatal sh- + r-) correspond to Tangut palatals, indicating that Tangut merged retroflexes (and [some?] coronal-r clusters?) with palatals, eliminating retroflex consonants which would otherwise have triggered vowel retroflexion.

The rGyalrong and Tangut words for 'six' might come from an earlier t-ruk (cf. Written Tibetan drug), though the origin of the aspirated initial of the Tangut form is not clear:

PrGT ?t-ruk > Chuk > rG (kU) ChëG, T chhyiw 1.46 (why not chyiw?; chh- < k-Ch-? < k-t-r-?? - early Tangut k- might correspond to the kU- in rG; also cf. Written Burmese khrauk 'six', also with a velar prefix)

Next: Let's get in-tense. (Coming January 26.)

© Copyright 2002-2006 Amritavision