In my last entry, I mentioned that Tangut has two characters for 'finger' that consist of the same elements in opposite orders:

<>

I also mentioned the Tangut character for 'heart':

If its elements are reversed, what does the resulting character mean?

<>

The left element of 'heart'

means 'person'. Grinstead (1972) proposed that it was derived from a four-stroke variant of Chinese 人 'person'. Although I think I've seen that variant, it's not in the largest dictionary of Chinese character variants online, though a five-stroke variant is listed.

The right element of 'heart'

also means 'heart', yet it cannot represent 'heart' as an independent character. Although the right side of this element

resembles Chinese 手 'hand' with an extra horizontal stroke, I suspect it is derived from Chinese 忄~㣺 'heart' with the shorter strokes turned ninety degrees and an added horizontal stroke or two added to distinguish it from手.

I don't know if the left side

has a Chinese origin.

Both halves of the 'heart' element

can appear without the other - e.g., on the left sides of

- though their independent meanings are unknown.

Li Fanwen is a great Tangutologist mentioned in this section of John Man's Genghis Khan: Life, Death, and Resurrection:

Chinese efforts to dominate this arcane field [of Tangut studies] rely heavily on the lifelong labour of one man, Li Fanwen, who was by now too frail to see me. His passion for the little-known [Tangut] language was ignited in 1955, and he has been working on its intricacies ever since. Many of the 6,000 symbols [of the Tangut script] were interpreted by Russian scholars; but symbols were only part of the problem. Li also had to wrestle with the grammar, and then apply his understanding to crack multiple signs - new concepts devised by combining and inverting characters. 'Wood' added to 'carve' to make 'chisel' is a simple one. But who could have guessed that 'heart' plus 'evil' would mean 'harm', that 'knee' plus 'hand' plus 'walk' means 'climb'? Or that 'finger' written backwards means 'toe'? After almost 50 years in the field, compiliing 30,000 note cards and transcribing over 3,000 tombstones, Li saw his massive Xi Xia [= Tangut]-Mandarin Dictionary published in 2001 [actually 1997; I saw a copy at Guillaume Jacques' apartment in 2000].

There are a few errors in Man's account of tangraphy (Tangut writing). The first tangraphic formula is more or less correct:

+=

top of 'wood' + all of 'carve' = 'chisel'

However, the left of 'heart' plus the right of 'evil' means 'hate', not 'harm':

+=

The left of 'knee' plus the left of 'hand' plus the top and a third of the bottom of 'walk' means 'crawl', not 'climb':

++=

And reversing the halves of 'finger' results in a character for another word also meaning 'finger':

<>

In the Tangraphic Sea dictionary, the first 'finger' is defined as the second 'finger'.

Not all tangraphs make as much sense as those compounds. I still do not know why the Tangut numerals for 'one', 'four', 'six', and 'nine' all contain the same component which means 'person' in other graphs:

'one', 'four', 'six', 'nine'

<

'seven' < 'eight'

09.12.16.23:56: JOHN MAN ON TANGUT

John Man's Genghis Khan: Life, Death, and Resurrection has a chapter devoted to the

phɔ̃ bie lhiẹ liẹ

Great State of White and High

a.k.a. the Tangut Empire. You can read it for free at Google Books.

Although it contains some errors, I recommend it to anyone who wants to understand why I've been fascinated by the extinct language and script of the Tangut (see the samples above) since 1996.

I said that Genghis's intended victim is best known by its Chinese name [Xi Xia, roughly pronounced 'she shah']. In fact, Xi Xia is hardly known to anyone beyond a few specialists, because Genghis did his best to wipe state, culture and people from the face of the earth. There is a case to be made that this was the first ever recorded example of attempted genocide. It was certainly very successful ethnocide. Xi Xia's successor cultures, Mongol and Chinese, had no interest in saving its records, reading its [unique] script or preserving its relics. It took the scholars of other nations, mainly Russia [centuries later], to start the work of decipherment and understanding [...] Only now is this ancient culture re-emerging into public gaze on the stage from which it was so violently ejected.

[...]

[The Tangut emperor] Yuan-hao also enforced an order made by a predecessor that Tangut be written down [? - accounts vary], for he knew - as Genghis [Khan] realized two centuries later - that writing would be the formal basis of administration and religion, and thus of a national identity. To match his ambitions, the script would have to be a supreme expression of civilization, yet also unique [...] So the 6,000 Tangut characters look Chinese, but they are not. Even those derived from Chinese are so altered that no Chinese can read them.

[...]

It was this script that was used to record laws and to translate the texts of Buddhism [...] The [Tangut] emperor, seeking to acquire merit by performing good deeds, obtained from the [Chinese] Song [Dynasty] emperor a complete copy of the 6,000-chapter Tripitaka, the corpus of Buddhist canonical writing, and had it translated into Tangut [...] This was more than a project of translation and writing. The Tanguts, like the Song [= Chinese], the Liao [= Khitan] and Koreans, printed their material, carving whole pages out of wood in reverse. The Tripitaka required 130,000 printing blocks, each containing hundreds of characters, and each producing two pages of text. And this was just one of thousands of the Buddhist works that were either produced by or had long been available to the Tanguts.

[...]

If the Tanguts' output was staggering, so too were their skill, organization and techniques. To print the Tripitaka, for instance, imagine producing the Encyclopaedia Britannica by making up your own script, and then, for printing blocks, carving every page of all 31 volumes in wood - in reverse.

And, I would add, translating the entire encyclopedia into your language without a dictionary!

Last night I discovered SW Bushell's 1899 article "The Tangut Script in the Nank'ou Pass". Unfortunately, it doesn't reproduce the Tangut script in the Nankou pass*, but it does contain two transcriptions of Tangut coins covered in this article by Andrew West. Tangutology and related fields have come a long way since then:

The clue of the method on which the inventor [of the Tangut script] worked remains to be discovered.

Clues have been discovered since, but the structure of many Tangut characters remains opaque.

The Kitan Tartars [...] had previously framed a national script for themselves from the Chinese li shu [clerical script], by cutting the characters into parts and arranging the sections in new combinations, which were known as Kitan large characters (ta tzŭ), to distinguish them from the small characters (hsiao tzŭ) which the Kitan had previously copied from their Western neighbours the Ouigour Turks.

The Khitan large script is not specifically based on the clerical script. Some Khitan characters look exactly like regular script characters (and hence do not consist of rearranged Chinese character components): e.g.,

一 'one'

二 'two'

三 'three'

十 'ten'

廿 'twenty'

日 'day'

月 'month'

The Khitan small script bears no resemblance to pre-Islamic Uighur script. Bushell is presumably extrapolating from the origin of the Khitan scripts in HIstory of the Liao Dynasty. Back in 1899, the Khitan script was unknown and the Da Jin huangdi dutong jinglüe langjun xingji inscription in Khitan was thought to be in Jurchen (since Da Jin means 'Great Jin' - Jin being the Jurchen Jin Dynasty). The Khitan script was not correctly identified in modern times until the discovery of the Khitan inscriptions in the Liao Imperial Tombs in 1922.

Their [the Jurchen] script consisted of some three or four thousand symbols pencilled in the lines of the ordinary Chinese writing (k'ai shu).

Jin Qicong's Jurchen dictionary (where'd my copy go!?) lists only 1,256 characters excluding duplicates, and if variants are excluded, the total number of characters is less than 800. I wonder where the three to four thousand figures comes from. Was it a guess extrapolated from the hundreds of Jurchen characters in Grube's Die Sprache und Schrift der Jučen (1896)?

*Bushell argues that the Tangut script in the Nankou Pass is not Jurchen. Although I'm sure he's right, he would be even more convincing if he provided visual examples.

Last night, I discovered that Tiger Woods has a Thai name Tont. I wanted to see its spelling in Thai, so I looked him up in the Thai version of Wikipedia which lists his names as

ไทเกอร์ วูดส์

Thaikəə(r) Wuut(s)

'Tiger Woods'

The parentheses indicate silent letters which are marked in Thai with a superscript thanthakhat. Thai words cannot be pronounced with final -r or final clusters like -ts.

เอลดริค

Eendrik

'Eldrick'

The Thai spelling of Tont surprised me:

ต้น

Ton

also a common noun 'beginning; source; trunk, stalk'

It consists of two letters, t (ต) and n (น) with a 2-shaped tone mark (ไม้โท mai thoo 'stick two' ้) on top indicating a falling tone. (Thoo is a borrowing from Pali and is cognate to English two.) Although there is no written vowel, an -o- is the default vowel of a syllable written with two adjacent written consonants. There is no final -t in the Thai spelling. So why was Ton romanized as Tont?

Schuessler's (2009: 91, 275) Late Old Chinese (LOC) reconstruction has rhyme classes that conflict with the classes attested in the Middle Chinese rhyming dictionary tradition.

(This table has been highly simplified. Final glottal stop and *-s are ignored. Only the MC reflex of EOC *-ri after grave initials is in row II. Only the reflex of EOC *-rə after labials is in row III. I am only interested in the development of those rhymes after those initials.)

The reconstruction of rows I and VI are unproblematic. Early Old Chinese (EOC) *-i stayed intact up into Middle Chinese (MC) and EOC *-əj became LOC and MC *-ɨj, ignoring differences in notation.

Schuessler and I agree that rows II, III, and IV merged, though their phonetic details are open to question. However, he believes that row V merged with II/III/IV, even though V is a distinct MC rhyme category. V could not have reattained independent status in MC. How would MC speakers know which II/III/IV/V syllables used to be V? (Similarly, the masculine/feminine/neuter distinction has been lost in English, and modern English speakers do not know the original genders of nouns.)

This question assumes that MC is the descendant of LOC. This doesn't have to be the case. There were many dialects of both MC and LOC, and it is possible that Schuessler reconstructed a LOC dialect that is not the ancestor of dictionary MC (which may have been a composite of various MC dialects).

In any case, I wonder if any attested LOC dialect has Schuessler's three rhyme categories (I *-i, II-V *-ɨ, VI *-ɨi). Starostin (1989: 625-674) has an analysis of the rhyming of five LOC poets spanning the third to fifth centuries AD:

Bao Zhao (405-466): I//IV : V : VI; no data for II, III; some interrhyming between I, IV, V

Tao Yuanming (365-427): I/III/IV/VI : V; no data for II; some interrhyming between I, IV, V, VI

Xie Lingyun (385-433): V : VI; one case each of interrhyming between I : V, IV : V; otherwise no data for I-IV

Ji Kang (210-263): I//IV/VI : V; no data for II, III; some interrhyming between I, V, VI

Yuan Ji (223-262): I/IV/VI : V; no data for II, III (as separate rhyme groups); some interrhyming between I, III, V, VI

(I treat EOC *-its as if it were EOC *-is, since their LOC/MC reflexes are identical.)

The treatment of VI varies. It is not clear whether this reflects phonological or aesthetic differences.