I'll reveal the answer to last night's question in a day or so.

In the meantime, check out this sample of modern Tangut calligraphy by 陳金夏 Kim-Ha Albert. You might have seen it if you clicked on the the Tangut link I provided in my last post. Although the caption says "as used by Tangut people", the piece is not authentic and contains errors (e.g., 'two' and 'big' are miswritten). Still, I'm happy to see the script in use.

Albert's myspace page may be the only one in the world with Tangut, Chinese, Khmer, Korean, and English. His native German is nowhere in sight, which is not surprising, since he wrote, "I really don't care for Germany."

*The Chinese name for Tangut, 西夏, is written with the graphs for 'western summer', and 金夏 Kim-Ha means 'golden summer'. I don't know why Albert chose 陳 Chen for his pseudonym's surname. 15 YEARS LATER, THE MOUNTAIN REMAINS

15 years ago today, my favorite comics creator left this world for another. But he is, in a sense, truly amritas - 'immortal'. Another one of my favorites imagined what he might say:

Well, yeah, [I died,] but only physically ... It ain't the physical stuff that's important. It don't last ... What's important is the idea of a thing. See, all we are is ideas ...

The ideas we have, the ideas other people have about us, the ideas we have about ourselves ... What else is a personality? And the thing about ideas over bodies is ideas don't die, and all we are is ideas! ...

Having a body, that’s what limits a creator! See, it's like brushes, pens, human hands... it all slows you down.

- Alan Moore, Supreme: The Return #6

Tangut is an idea. The Tangut people are long dead, but the Tangut idea lives. An idea lives as long as one person remembers it. Even if an idea is forgotten, it has not permanently perished. If it is rediscovered - as Tangut was - it can live again, and be put to new uses.

For instance, I could give the imperator imaginum - the emperor of images - a Tangut name:

ŋəʳ viẹ khiə swiẽ

lit. 'mountain short heel'*

Can you guess his real name?

*Tangut adjectives follow nouns, so ŋəʳ viẹ 'mountain short' = 'short mountain'. khiə swiẽ 'heel' is a disyllabic word. Tangut surnames precede personal names. HOW SIN BEGINS

Relax ... this has not turned into a religious blog.

Last night I failed to note something odd about the Hebrew and Arabic names for China. Hebrew and Arabic have two types of s, emphatic and nonemphatic. I would expect both languages to have the same type of s in 'China', but they don't:

Hebrew סין sin (nonemphatic s)

Arabic صين siin (emphatic s)

Does this imply that the two names did not come from the same source, or that speakers of the two languages interpreted the same source differently?

Old Chinese also had a distinction between nonemphatic and emphatic s. 秦, a possible source of the above names, might have had a number of variant readings. The reading closest to the Semitic names might have been *tsin with a nonemphatic initial. Goerwitz (1996: 490) reconstructed the earliest pronunciation of the letter samekh as nonemphatic *[ts], though I don't know whether it had become *[s] by the time the name was borrowed. Hebrew (but not Arabic) emphatic s later became [ts], which makes me wonder if Chinese *ts- sounded like s to Arabic speakers.

Still other initials of 'China' are th- in Thinæ / Θιναι (unsure of accentuation) and tz- in Greek Tzinitza / Tzinista (I've only seen these in romanized form; are they Τζινιτζα/Τζινιστα?). tz- obviously looks like my proposed Chinese *ts- and the initials of Thinæ and Sinæ may be attempts to imitate that *ts-. I don't know whether Greek had already shifted from an aspirated stop [tʰ] to a fricative [θ] at the time Θιναι was borrowed. MAGNA SERICA EST ... ROMA?

John Bentley and I have been talking about how the Chinese once called the Roman Empire 大秦, now pronounced Da Qin in Mandarin. The name was something like *dɑɕ dzin when it was coined sometime prior to 97 AD. It is obviously not a sinification of Roma or any other Latin name. 大 is 'great' and 秦 is the name of the Qin Dynasty. One could loosely translate it as 'Great China'. The name and its referent are discussed at the Silk Road Seattle site.

Some believe that 秦 is the source of some names of China. On the surface, this seems plausible, since Qin (pronounced 'chin') sounds like China and Asian names such as Sanskrit Ciina, and the s- in names like Latin Sinæ, Hebrew סין, and Arabic صين was the closest foreign equivalent of 'ch'. However, such a view is anachronistic because 秦 was *dzin with a voiced initial until the last centuries of the first millennium. Why wasn't 秦 borrowed as something like zin?

Other early names for China such as Latin Serica and Greek Σῆρες have been linked to Chinese 絲 'silk', now pronounced si (roughly 'suh', not 'see' or 'sigh') in Mandarin. They can also be compared to Korean 실 shil 'silk (from Middle Korean sǐr). Again this initially appears to be believable, but falls apart because the Old Chinese pronunciation of 絲 'silk' was *sə without any -r. The rising pitch of Middle Korean sǐr implies an earlier *sìrV́ which has no Chinese etymology unless it is from late Early Middle Chinese *si 'silk' plus an otherwise unknown -rV suffix.

Here's my attempt to account for the above words (with the exception of Korean shil):

1. 秦 originally was a phonetic in Chinese characters sharing a common phonetic denominator *tsir.

2. The final *-r is the source of retroflexion in the initials of 榛溱臻蓁 Middle Chinese *tʂin < *tsir (cf. Starostin's 珍 Old Chinese *tər which became Middle Chinese *ʈin.)

3. 秦, the name of the Qin Dynasty, was originally *-tsir (*N- = unknown nasal) which developed in three different ways:

Path 1

*-tsir > *Nʌ-tseir > *tseir > *tser

The main syllable assimilated to the emphatic low-vowelled presyllable. The presyllable was then lost, and *tser became the source of foreign ser-names for China. ser-words for 'silk' could have originally meant 'the Chinese thing'; cf. how Eng china 'porcelain ware' may be from china dishes.

Path 2

*-tsir > *N-tsir > *ntsir > *ndzir > *ndzin > *dzin

The presyllable lost its vowel, and the remaining nasal fused with the *ts- of the main syllable, resulting in Middle Chinese *dzin which later became Mandarin Qin.

Path 3

*-tsir > *tsir > *tsin

The presyllable was lost without a trace, and *tsin became the source of Sanskrit ciina, Latin Sinæ, etc.

There are two big problems with the above scenario:

1. There is no Chinese-internal evidence for a presyllable *- as opposed to *Nɯ- or a preinitial *N-.

2. There is no Chinese-internal evidence for a final *-r.

I am reluctant to reconstruct *-. and *-r simply to make an uncertain etymology work. SUN STANDING IN THE MOUTH OF A SEASHELL

is a Chinese analogy for the problems I face in Tangut character analysis.

Suppose you knew nothing about Chinese except for four facts:

li is 'stand'

ri is 'sun'

kou is 'mouth'

bei is 'seashell'

What do you think


means? It is the Chinese translation for the Tangut word biu in the sense of 'rhyme' (discussed in these three posts). Should we try to figure out a way to get from 'stand + sun + mouth + seashell' to 'rhyme'? Or should we reexamine 韻 from another angle?

Now let me introduce two more facts:

yin is 'sound'

yuan is 'circle' (and in modern Chinese, 'member')

韻 could be analyzed not only as


but also as




or even be regarded as an indivisible unit 韻.

However, the only correct analysis is


yin 'sound' is semantic and 員 yuan 'circle; member' is phonetic: 韻 yun 'rhyme' obviously has something to do with 音 sound and almost sounds like 員 yuan 'circle'.

yun does also sound like 音 yin 'sound' in Mandarin, but the two had different consonants in Old Chinese:

音 Md yin < OC *ʔəm

韻 Md yun < OC *wəns

員 Md yuan < OC *wan

韻 OC *wəns shares two consonants with its phonetic 員 OC *wan, but only shares a vowel with 音*ʔəm.

Even if one did not know the OC readings of those characters, an experienced Mandarin user comes to expect phonetics to appear on the right, so 員 Md yuan is more likely to be phonetic than 音 Md yin. Characters with left-hand phonetics are less frequent.

We know much less about Tangut writing (tangraphy) than Chinese writing (sinography). Can we be certain that what might appear to be a combination of two tangraphic elements (cf. Chn 立+日) is really one tangraphic element (cf. Chn 音), or vice versa? SIFTING GRAPHEMES FROM THE SEA

Since no extant native Tangut work explicitly lists the graphemes (minimal semantic and/or phonetic components) of tangraphs, modern Tangutologists have to infer them from the analyses in the two surviving volumes of Tangraphic Sea. (The second of the three volumes remains lost, so about 40% of tangraphs have no known analysis.)

This is not easy, because the analyses are in four-character formulae excluding explicit references to character parts: e.g., a tangraph ABC could be analyzed as

left [of] AB [plus] right [of] BC

rather than

A + B + C

It would be unclear whether the B part of ABC is from AB, BC, or both. And it would not be clear whether AB and/or BC were two-part compounds or indivisible units (cf. Russian Ы which is not a combination of the letter Ь and the extinct letter І in spite of the space inside it; it is in fact originally a ligature of Ъ and І or И which is now regarded as a single letter).

Can we be absolutely sure that the apparent Tangut grapheme


is not really a combination of two other graphemes


'vexation' + 'person'

in all cases?

I think so, though there are tangraphs in which 'bird' doesn't seem to refer to birds - or to vexations or people: e.g., on Saturday I mentioned

TT0422 phioo R55 2.46 'pair'

(with 'language' on the right; could 'bird' on the left be influenced by Chn 雙 'pair' which consists of two 隹 birds atop a 又 hand?)

TT0420 ʃwii R14 2.12 'correspond to; meet the requirement of something'

(same left and right-hand parts as 'pair' with 'person' inserted between them - why?)

An example of a seemingly meaningless right-hand 'bird' is in the tangraph for the second half of

khiə xõ R56 2.47 'row'

with 'person' and an unknown center element also found in the first half of 'row' and the second halves of

diə kwiəəʳ 'Xiongnu (with 'hand' on the right)


diə kwiəəʳ 'a kind of bird' (with another 'bird' element on the right)

(homophonous with 'Xiongnu'; did the Tangut name the Xiongnu after the bird, or was the bird named after the Tangut name for the Xiongnu?) PART HEADS OF THE WRITING OF THE WESTERN SUMMER

('Western Summer' is a literal translation of the Chinese name for the Tangut: 西夏 Xixia.)

Many Chinese dictionaries are organized by radicals (部首 'part heads'). Characters sharing a common component ('radical') are listed together and are ordered by the number of strokes: e.g.,

radical 木 'tree'

radical 木 + 0 strokes

木 itself

radical 木 + 1 stroke

未 末 本 札 术 etc.

radical 木 + 2 strokes

朱 朴杀 etc.

Radicals are not necessarily semantic units. Some are arbitrarily chosen components: e.g.,

上 'top', 不 'not', 丑 'ox', 世 'world', 丘 'hill'

are all lumped together under the radical 一 'one' simply because they share a horizontal line and can't be pigeonholed among any of the other 213 standard semantic radicals. They were not created as derivatives of 一 'one'. Their association with 一 'one' is solely the byproduct of a classification system dating long after their invention.

The actual number of true semantic and phonetic components in Chinese writing is far greater than 214. Each of the 一 'one' graphs that I cited is a true radical. Nonetheless, the 214 radical system endures, not only out of sheer inertia, but also due to practicality. 214 categories are easier to learn and therefore easier to navigate than over a thousand. (Karlgren's Grammata Serica Recensa arranges characters into 1,260 groups which each share a phonetic component with only a few exceptions.)

The Tangut left behind no list of radicals, true or otherwise. All Tangut radical lists were invented by modern scholars primarily as indexing devices. If two characters are listed under a single Tangut 'radical', there is no guarantee of any semantic or phonetic relationship between them. For example, the first two tangraphs under Li Fanwen's (1997) radical 1 (一) are

TT1075 lew R44 1.43 'one' and TT1100 do R51 1.49 'poison'

which merely happen to share a horizontal stroke on top and nothing else. I doubt the Tangut would consider them as related in any way.

Last night, I found Andrew West's compiliation of Tangut radical lists from six different modern works on Tangut. I visit his index to Li Fanwen's (1997) dictionary every single day and I don't know why I never noticed that page before. Each work that West has examined has a different number of radicals ranging from 151 to 444. Some works index tangraphs according to their top left-hand components and others use bottom right-hand components. I must emphasize that these radical lists do not represent their authors' beliefs about tangraphic structure: e.g., Grinstead would never say that all tangraphs are composed of combinations of his 151 radicals (which are all on the bottom right and hence exclude elements found only on the top or the left). Nonetheless, there must be some degree of overlap between these radical lists and the actual set of tangraphic components: e.g., all agree that

Nishida (1966) radical 204

Li Fanwen (1997) radical 118

Kychanov (2006) radical B210


is a genuine unit meaning 'person' in Tangut writing and not just an artifact of a modern index.

Next: Sifting graphemes from the Sea.

Tangut fonts by Mojikyo.org
Tangut radical font by Andrew West
All other content copyright © 2002-2009 Amritavision