Today is the birthday of my friend

sie raʳ

I wish her

mə wəəi niəə bɛɛ reʳ!

That's a Tangutization of Mandarin


shengri kuaile 'birthday happy'

It took me hours to make up that phrase, and I doubt it's right. There are no living Tangut to help me. If the Tangut ever adopted the foreign custom of celebrating birthdays, they'd probably have a totally different expression.

Li Fanwen (1997: 530, 958) and Kychanov (2006: 402) disagree on the Tangut translation of 快樂 kuaile 'happy'. I've chosen

bɛɛ reʳ

following Kychanov instead of

bɛɛ khie

following Li Fanwen, since the former occurs in Tangraphic Sea definitions whereas the latter may only be in Homophones (and without any other attestations, I'm not certain whether it was really a disyllabic word or khie preceded by a semantic clarifier bɛɛ).

Translating 生日 shengri 'birthday' was even harder.

Li Fanwen's (1997: 1198) index contains 12 graphs translated as 生 sheng (which has many meanings besides 'birth'), and I found another graph (2256) listed as a transcription (though it could also mean 生 sheng in the sense of 'raw'):

Tangraph Li Fanwen 1997 number My reconstruction Rhyme number Rhyme.tone Gloss
0009 ʃuo R50 1.48 appear; raise
0354 lhi R11 2.10 give birth to; regret
0487 swụ R62 2.52 life
2256 ʃɛ̃ R42 1.41 raw; beast; monkey
2295 gii R14 1.14 give birth to
2724 tʃhu R2 1.2 life; contain; get into; exist
2823 R28 2.25 give birth to; bear
3281 zị R70 2.60 give birth to
3300 diu ? ? give birth to; bear
4057 dʒɨe R36 2.32 give birth to
5282 bi R11 1.11 grow; health; strong
5339 ʃɛ̃ R42 1.41 give birth to; bear
5435 wəəi R12 1.12 give birth to; bear

Li Fanwen's (1997: 1196) index contains 6 graphs translated as 日 ri 'sun, day':

Tangraph Li Fanwen 1997 number My reconstruction Rhyme number Rhyme.tone Gloss
811 jaaʳ R89 2.75 day
1421 ziə̣ R72 1.69 daytime, day, time
2440 niəə R33 2.29 sun; day
2449 bəi R8 2.7 sun
2532 die R37 1.36 day
3295 riʳ R84 2.72 day

Since I wasn't certain which Tangut translation of 生 was appropriate for 'birthday', I chose the disyllabic word

mə wəəi '誕生 birth'

and added it to

niəə '日 sun, day' (cf. Written Tibetan nyi, Old Chinese *nit, the ancestor of Mandarin 日 ri)

which seems like the most basic word (the others appear to be specialized). Note that the Chinese characters for the translations of mə wəəi and niəə form the Japanese word 誕生日 'birthday'.

I could have taken the easy way out and translated Mandarin 生日 shengri 'birthday' as its near-soundalike

ʃɛ̃ riʳ

using Tangut borrowings from Chinese, but I preferred native Tangut words. QUIET RIGHT, CHEERFUL ALL

When I was trying to translate 樂園/乐园 'paradise' yesterday morning, I discovered two similar tangraphs equivalent to 乐 'joy'. One was derived from the other:


TT3672 bɛɛ R39 1.38 =

Li Fanwen (1997: 530): 'amusement; happy; 娛樂'

Shi et al. (2000: 110): '娱'

Kychanov (2006: 402): 'pleasure; joy; 樂; 娛'

right of TT1134 dziẽ R43 1.42 +

Li Fanwen (1997: 103): 'quiet; 安 peaceful'

Shi et al. (2000: 324): '安居 live in peace'

Kychanov (2006: 417): 'profit; prosperity; 利 ; 安'

a loan from Tangut period Chinese 靜 ?*dzjẽ 'quiet'?

left of TT2511 khie R37 2.33

Li Fanwen (1997: 958): 'happy; cheerful; 娛、樂、快樂'

Shi et al. (2000: 257): ‘庆 congratulate'

did TT2511 merely transcribe 慶/庆, or did it also represent a loanword khie < Tangut period NW Chinese ?*khjẽ?

Kychanov (2006: 402): 'happiness; cheerful; 快樂'

The title is based on a nearly direct translation of the Tangraphic Sea analysis:

dziẽ R43 1.42 biẹ R64 2.54 khie R37 2.33 ŋõʳ R97 2.82

'quiet right cheerful all'

(biẹ is originally 'help' but by extension means 'right side of character' [< 'helper'?]

I ended up choosing TT3672 since it could occur by itself and as the first half of different words whereas TT2511 seems to only represent* the second half of the word (expression?)

bɛɛ khie

Li Fanwen (1997: 958): '快樂 pleasure; 娛樂 pleasure'

Kychanov (2006: 402): 'joy and grief' (!? - but TT2511 doesn't mean 'grief'!)

( In fact, khie can occur by itself. See Li Fanwen [1997: 958] for examples.)

It seems unlikely that TT2511 was devised for the second syllable of bɛɛ khie before TT3671 was devised for the more common morpheme bɛɛ. Why was

Li Fanwen radical 118 'person'

added to distinguish bɛɛ from khie? One might expect

to mean 'happy person' (especially since 'person' + 'happy' is in Tangut noun-adjective order), but it doesn't.

Next: What's between the two 'people' in bɛɛ - and on the left of khie?

*I am excluding the use of TT2511 khie to transcribe Tangut period northwestern Chinese




I am not sure why TT2511 khie was used to transcribed TPNWC *khjẽ. I would expect

TT5644 khiẽ R43 1.42 'give; grant'

which would be a better phonetic match. I wonder if nasalization was lost in Tangut and/or TPNWC when transcriptions with 'mismatched' nasalization were devised.

The choice of TT2511 khie for ?*ki is particularly unexpected because the initial and final do not match. Perhaps the choice was semantically motivated. TT2511 corresponds to 吉 in 維摩吉, a Chinese transcription of Vimalakiirti 'spotless fame'. khie 'cheerful' barely sounds like 吉 *?ki or kiirti 'fame', but like 吉 'good luck', it has a positive meaning unlike the only tangraph pronounced ki R11 1.11


which has a neutral meaning 'record' (borrowed from TPNWC 記 ?*ki 'record'). 古蘭經 THE ANCIENT ORCHID SUTRA

is not a Buddhist text. It's a Chinese name for the Qur'an. Another name is 可蘭經 'possible orchid sutra'. 經 Mandarin jing is sutra (Skt 'string') and 古蘭 Md gulan and 可蘭 Md kelan are phonetic approximations of Qur'an.

The Bible is the 聖經 'sacred sutra' or 聖書 'sacred book' in East Asia. It is not to be confused with the 'Jesus Sutras' about the 彌師訶 (pronounced something like *miʂixa - i.e., messiah - in the early 7th century).

Another spelling of 'messiah' is 迷詩所 *mjejɕiʂə which is probably an error for 迷詩訶 *mjejɕixa.

宋蘭友 Song Lanyou (there's 蘭 'orchid' again!) proposed that 序聽 *zjəthjeŋ is an error for 序數 *zjəɕyə or 序鼠 *zjəɕə ?'Jesus'. These graphs strike me as odd choices. Did the Nestorians pronounce Jesus with [z] instead of IPA [j]? Moreover, 鼠 *ɕə means 'rat' (!). Yet in the body of this sutra, Jesus is called a 移鼠 'moving rat' *jiɕə. Ugh.

I prefer the later transcription 耶蘇 Yesu consisting of

ye (classical Chinese question particle)

su 'resurrect'

One could think of it as 'resurrection?' with a question mark* expressing doubt. There can be no faith without doubt.

*耶 only has an interrogative function in final position, so 耶蘇 cannot mean '(is it) resurrection?'. 'Resurrection?' is a potential mnemonic but not a gloss for 耶蘇. MUSLIM. HALAL. MEAT. SHOP

What's. with. the. periods. after. each. word. but. the. last. in. this. sign (posted on 9.11 - not the best date?)

The name occurs again without spaces:


The Korean under the English says

무슬림 정육점

musUllim cOngyukcOm

精 정 cOng 'refined'

肉 육 yuk 'meat'

店 점 cOm 'shop'

Is English the lingua franca among Muslims in Seoul? "CHICKEN", "BEEF", and "MUTTON" have no translations, though they are accompanied by pictures.

I've known about adult Korean converts to Islam (e.g., Koreans who lived in the Arab world) for years, but I've never heard of a child convert:

"I first went to the mosque out of curiosity and I ended up becoming a Muslim myself," says Ahn Tae-hwan, a 15-year-old middle school student.

I wonder how his parents reacted.

Islam has a long history in Korea. I wonder if Korean mosques were ever called 淸眞寺 (청진사 chOngjinsa) 'pure truth temples' as in China. (Cf. how Allah can be called 眞主 'True Lord' in Chinese.) The Korean Wikipedia article on mosques uses the term 모스크 mosUkhU and Googling chOngjinsa seems to lead to pages about mosques in China.

A character-for-character Tangut translation of 淸眞寺 could be

ɣɛ se tshə

lit. 'truth pure temple' with ((N Adj) N) structure 汉语乐园 CHINESE PARADISE*

I wonder what it's like to be a Korean kid learning Mandarin from an English-language textbook. This reminds me of how I studied Latin in Japan and used a German grammar and a Chinese textbook to study Manchu. (An English textbook only appeared months after I got my PhD.)

*Literally 'Chinese language joy garden'. Where's the

mi ŋwəəu bɛɛ ʔɔ̣

'Tangut language joy garden'? KNOW THE FLOW

se 'count' from the last entry

looks like


Li Fanwen radical 023 'not' +

Li Fanwen radical 132 'head' +

right-hand variant of Li Fanwen radical 025 'water'

which makes no sense. What does counting have to do with negation, heads, or water?

Tangraphic Sea 1.43A31 derives each of the three radicals from three characters:


se 'count' =

'not' from the left side of nwə 'know; realize' +

'head' from the center of nə̣i 'proclaim; publicize' +

'water' from the left side of kiẹ 'flow, pass' (translated by Shi et al. [2000: 142] as 茎 'stalk' and Kychanov [2006: 407] as 'swift current; rapids'!)

Note how the top half 'water' changes shape when it switches sides:


This analysis doesn't make much more sense than a superficial breakdown of the components of 'count'. nwə 'know', nə̣i 'proclaim', and especially kiẹ 'flow' (or 'stalk' or 'swift current') have no obvious semantic or phonetic connection to 'count'. I might expect 'know' or 'proclaim' + 'number' but none of the elements in 'count' match those of

ŋeʳw 'number; numeral'

which looks like


'number' + 'person' (but what do people have to do with numbers?) COUNT ON THE TRUTH

The graph I almost used in my Tangutization of Sarah is analyzed in Tangraphic Sea 1.43A22 as


se 'pure? clear? quiet?' = top of ɣɛ 'truth; true' + all of se 'count'

se 'count' is obviously phonetic, since it is homophonous with se 'pure?'.

The meaning of the top element (Li Fanwen radical 036)

which I've nicknamed 'horned hat' is unknown. If it weren't for the Tangraphic Sea analysis, there would be no way to know whether the 'horned hat' in se 'pure?' came from ɣɛ 'truth' or from 279 other characters.

The structure of the graph for se 'pure?' is roughly analogous to a spelling like tpoor for 'pure' with a silent t- representing truth (out of thousands of words with t-) and the remaining letters providing phonetic information. (Pure has no differently spelled homophone, so I used a near-homophone poor.)

se 'pure?' also appears in the analysis of ɣɛ 'truth' (Tangraphic Sea 1.45A13):


ɣɛ 'truth' = top of se 'pure?'+ bottom of kɪ̣ 'gold'

kɪ̣ 'gold' could be semantic as well as phonetic. Its bottom elements constitute the graph


tswiaʳ 'determine; comprehend; know' = bottom center of tseʳw 'joint; period of time; tally; testimony; relatives' + bottom right of ɣɛ 'truth'

and its top element (Li Fanwen radical 188)

can mean either 'above' (Gong, "Chinese Elements in the Tangut Script") or 'female' (Nishida 1966: 241).



'top/female' + 'determine; comprehend; know' = 'gold'

does not have anything to do with femininity or knowledge. It represents a metal - even though it lacks Li Fanwen radical 293 'metal':

kɪ̣ 'gold' is analyzed in Tangraphic Sea (1.72A31) as


kɪ̣ 'gold' = top of lo 'rich' + bottom of ɣɛ 'truth'

Does gold make one truly rich?

The bottom of lo 'rich' has no independent meaning and may be an abbreviation of a graph like

diu ''have; possess'

Are the rich the 'top possessors'?


'Top' is presumably an abbreviation of a graph like 'gold'. SE RAR, THE PURE FOUNTAIN?

Last night, I briefly considered giving Sarah the Tangut name seraʳ

with a different first character se to transcribe Sa-. There are four reasons I didn't choose it.

First, I preferred sie 'understand' because it corresponded to the grok in the title of her blog.

Second, the Tangut seem to have used sie 'star' (homophonous with 'understand') to transcribe Sanskrit se (Nishida 1964; Grinstead 1972: 194). This suggests that the Tangut syllable sie (or however it was actually pronounced) was closer to Skt se than Tangut se.

Third, there is no consensus on the meaning of se:

Nishida (1966: 316): 淨 'pure; clear'

Grinstead (1972: 96): 'pure'

Li Fanwen (1997: 864):

1. 清淨 'tranquil'

2. 靜 'quiet'; second half of mõʳse 'palace secretary' (equivalent to Chn 中書)

(why doesn't LFW group 'quiet' with 'tranquil'?)

Shi et al. (2000: 98): 净 'pure; clear'

Kychanov (2006: 264): 'чистий; 'pure; clear; 清; 淨'

Fourth, even if se were 'pure', se raʳ cannot mean 'pure fountain' because Tangut normally has noun-adjective order. 'Pure fountain' should be raʳ se. But sie raʳ could mean 'fountain of feelings / knowledge' because Tangut has possessor noun-possessed noun order.

Then again, it's not clear that raʳ 'fountain' occurred as a monosyllabic noun. All attestations I can find are in the disyllabic words

raʳmə 'fountain(s)' (Nishida 1964 treats this as plural but Kychanov 2006 treats it as singular. Since by itself means 'fountain', this could be a redundant compound referring to a single fountain or a collective compound referring to multiple fountains: 'raʳ-s and -s'.)

zɨəəʳraʳ 'blister' (lit. 'water fountain') SIE RAR

Today is the fifth anniversary of my friend Sarah's blog.

Years ago, before I had an HTML editor that could handle Unicode, I posted her name as a Devanagari GIF. Now I can just type it: सेरा. Maybe someday I'll be able to type it in Avestan. But today I can write it in Tangut as sieraʳ:

The characters not only sound like Sarah but symbolize how I see her.

Tangut has no that would be a perfect match for the Sa- of Sarah, so I found a character for a similar syllable sie

meaning 'passion, feeling; know(ledge), understand'. The Tangraphic Sea analysis of this character is unknown, but it appears to consist of a left side of unknown function (Li Fanwen radical 193) plus ʒɨʳ 'wisdom; intelligent' (Li Fanwen radical 266):


A simple ra without a retroflex vowel would have been a better phonetic match for the -rah of Sarah, but Tangut generally does not allow r- to be followed by nonretroflex vowels. Hence I picked a character for a raʳ

meaning 'fountain', which makes me think of one of her favorite books, The Fountainhead.

I was initially reluctant to choose raʳ because Grinstead (1972: 96) glosses it as 'blister?' and Kychanov (2006: 264) glosses it as 'blister' in English but not in Russian or Chinese However, Nevsky (1960 II: 41), Nishida (1966: 435), Li Fanwen (1997: 602), and Shi et al. (2000: 308) do not list such a meaning. Such a meaning, if it exists, must be secondary, as all sources on hand other than Grinstead agree on the meaning 'fountain'. I suspect that raʳ does not mean 'blister' at all, and that this gloss was taken from the compound

zɨəəʳ raʳ 'water fountain'

a euphemism for 'blister' (Nevsky 1960 II: 41) or 'furuncle' (Kychanov 2006: 311). But a part is not necessarily synonymous with the whole.

The Tangraphic Sea analysis of raʳ 'fountain' is unknown, but Nishida (1966: 435) proposed that it consists of the first two-thirds of the character for its homophone raʳ 'flow' (phonetic; I think it's also semantic) and 'water' (which looks like a waterfall to me):


The character for raʳ 'flow' looks like Li Fanwen radical 080 flanked by two 'person' radicals (Li Fanwen radical 118):


I have no idea what the middle part means. I'm still trying to grok the Tangut script. PEOPLE DON'T DIE

Replacing Li Fanwen radical 112 'death' with radical 023 'not'


turns TT3252 xọ R73 2.62 'calamity; disaster; suffering' (from "When People and Grass Die") into TT1435 kɨw R46 2.40 'die in battle; lost; throw off'; discard' (!)


which looks like


Li Fanwen radical 023 'not' +

Li Fanwen radical 118 'person' +

Li Fanwen radical 112 'die'

Why write 'die in battle' as if it were 'people don't die'? It isn't, and presumably the parts of TT1435 are derived from two or more unknown tangraphs.


'not' and 'person'

may be from the left of


TT1433 R18 1.18 'irregular; uneven; fight; quarrel with' =

left of TT1425 nəu R1 1.1 'back; rebel (< turn one's back against); defeat; forget (< put something behind one?)'

right of TT1324 R28 1.27 'seek; look for'

Are fighters searching for defeats?

'Die' could be from any 'die-graph' containing Li Fanwen radical 112. (I'll deal with another kind of 'die-graph' later.) A BOWL OF BUGS

Li Fanwen (1997: 456) translated

TT3193 daa R22 1.22 (from "When People and Grass Die")

as 'magic; demon; 蠱惑 'seduce into wrongdoing; 蠱 'poisonous insect'; 惑 'mislead' (the latter two meanings are implied from his translations in the body of the entry).

'Poisonous insect' is doubtful since the tangraph does not contain either radical for 'insect':

Li Fanwen radical 266

Li Fanwen radical 304

The gloss 蠱惑 'seduce into wrongdoing' may have been extracted from one of the glosses of the disyllabic word

daa R22 1.22 liẽ R65 1.62 蠱惑 'seduce into wrongdoing' (p. 456); 鬼怪 'ghosts and monsters' (p. 286)

The first sinograph in 蠱惑 looks like 蟲 Old Chinese *rluŋ 'insect' (itself composed of three 虫 bugs*) atop 皿 Old Chinese *mraŋ 'vessel, dish bowl'. 蟲 is semantic since 蠱 can mean 'poisonous insect'. But what is the function of 皿, which sounds nothing like 蠱 OC *kaʔ? Karlgren (1957: 34) glossed 蠱 as

noxious vermin in the food, in the stomach; noxious influences, seduction, madness

So 蠱 is a drawing of bugs in a bowl. (I wonder if Boodberg could identify a phonetic in it.)

*Simplified to one bug in the PRC: 蛊 (a variant of 蠱 dating back to the Kangxi dictionary, if not earlier). WHEN PEOPLE AND GRASS DIE

... it's a calamity. TT3744 ŋĩ R16 1.16 'disaster; calamity' looks like


Li Fanwen radical 118 'person' (derived from Chn 人?) +

Li Fanwen radical 207 'grass' (derived from Chn 草?) +

Li Fanwen radical 112 'die' (derived from Chn 死?)

The Tangraphic Sea derives it from


center of TT3140 liee R40 2.35 'devil; demon' +

left of TT3252 xọ R73 2.62 'calamity; disaster; suffering'

No analyses of the component tangraphs are available, so we can only take their parts at face value.

TT3140 looks like a 'person' between two 'deaths':


TT3252 looks like 'grass' plus 'death' + 'demon' (derived from Chn 鬼 'ghost', though it resembles Chn 彳 'left step'?):


Notice how the rightmost stroke of 'death' goes beneath the second diagonal stroke when the radical is in medial position. This variant

'Death' and 'demon' form TT3142 ju R2 2.2 'ghost; spirit':


Taking TT3142 and adding 'grass' to the right instead of the left results in TT3193 daa R22 1.22 'magic; demon; 蠱惑 'seduce into wrongdoing':


Note that 'demon' is elongated to carry 'grass' on its tail. This variant of 'demon' is Li Fanwen radical 303 which only occurs five times in left-hand position. I don't know how frequent it is in medial position.

'Grass' and 'demon' form TT3153 lew R44 2.38 'revenge':


There are no tangraphs with the structures

'demon' + 'grass' or 'death'


or 'grass' + death'


Tangut fonts by Mojikyo.org
Tangut radical font by Andrew West
All other content copyright © 2002-2008 Amritavision