This line begins a section devoted to tangraphs for Chinese surnames. The first two surnames are very common, but the others are not. Could Ren, Zhong, and Ji have been more common in the Tangut Empire than in China as a whole?

Zhang, etc. are modern standard Mandarin readings for the surnames. Reconstructions of the surnames' Tangut period northwestern Chinese readings (e.g, *tʃɨõ) are in parentheses. The translations (e.g., Chon) are lay transcriptions of the surnames' Tangutizations.

Tangraph number 411 412 413 414 415
Li Fanwen number 1030 0403 1148 2649 0576
My reconstructed pronunciation 1tʃɨõ 1võ 1ʒɨĩ 1tʃɨõ 1kwi
Tangraph gloss (transcription of Chinese)
(transcription of Chinese) (transcription of Chinese) (transcription of Chinese) honorable; to honor, respect (transcription of Chinese)
Word the surname 張 Zhang (*tʃɨõ) the surname 王 Wang (*wɨõ) the surname 任 Ren (*ʒɨĩ) the surname 鐘 Zhong (*tʃɨõ) the surname 季 Ji (*kwi)
Translation Chon, Von, Zhin, Chon, Kwi

411: Was there a connection between the Chon and the Lujy or Jyghew?


1030 1tʃɨõ 'the surname 張 Zhang (*tʃɨõ)' (ponbee) =

2105 1tʃɨõ 'first moon' (can't be from Chn 正 *tʃɨe?; dexpondal; phonetic) +

0905 0dʒɨə (part of the Tangut surnames Lujy and Jyghew) (beebeefir)

412: 0403 is Grade I 1võ which doesn't match Chinese Grade III *wɨõ. There is no Grade III 1vɨõ in Tangut, so Grade I 1võ was the closest available approximation. Chinese *w may have been *[v] as in Tangut.

0403 has a circular analysis:


0403 1võ 'the surname 王 Wang (*wɨõ)' (folcok) =

0830 1kĩ 'the surname 金 Jin (*kĩ)' (folcin) +

2340 1võ 'prosperous, flourishing' (< Chn 旺) (geofolcok; geo = 'sage')


2340 1võ 'prosperous, flourishing' (geofolcok) =

5307 1ɣwɪ 'power, force, influence' (pikgeo)

0403 1võ 'the surname 王 Wang (*wɨõ)' (folcok)

413: The function of bix in 1148 is unknown. 2541 is a cryptophonetic. Its Chinese translation was 人 *ʒɨĩ which sounded like 1148 1ʒɨĩ.


1148 1ʒɨĩ 'the surname 任 Ren (*ʒɨĩ)' (bixdex) =

1546 2lɨụ 'body' (cognate to Writ Tib lus) (bixgeo) +

2541 2dzwio 'person' (qaldex)

414: Why do two homophonous Chinese surnames have different tangraphs (1030 and 2649)?


2649 1tʃɨõ 'the surname 鐘 Zhong (*tʃɨõ)' (dexjor) =

2888 2mə 'surname' (dexpux; semantic) +

0724 2niə (plural suffix) (hiajor)

Is 0724 really the source of the right side of 2649? The two do not sound alike.

415: What you want is what is valuable:


0576 1kwi 'honorable; to honor, respect' (< Chn 貴 *kwɨi 'honorable; to honor; valuable') (hergirdex) =

1319 1tshi 'to desire, want' (herfea; semantic)

3926 2nia 'thou' (girdexcin) TANGRAPHIC RADICALS 20: DOUBLE DEATH

One of the tangraphs I listed last night, the second half of

5918 0788 1siə-2mie 'death'

(cognate to Old Chinese 死 *siʔ 'to die' and 亡 *maŋ 'to die'; parallel in structure to Chinese 死亡 'to die; death')

has a top radical dur found in only one other tangraph used to write the second half of

1493 0781 1siə̣-2mie 'death'

(same cognates as the previous word; the tense vowel of the first syllable reflects an earlier prefix of unknown function: 1siə̣- < *Sɯ-sə-)

Should these tangraphs be analyzed as

0788 dur + dex + fok

0781 dur + dex + tos (tos = 'demon')

or as

0788 durdex + fok

0781 durdex + tos

dur never occurs without dex, so it's possible that durdex is a single radical.

Since both tangraphs have identical readings and represent the second halves of words for 'death', dur(dex) may have both phonetic and semantic functions.

tos 'demon' is presumably semantic (demons bring death?) in 0781.

fok in 0788 may be short for

4255 1sia 'to kill'

(cognate to Old Chinese 殺 *ksat 'to kill')

which is of higher frequency than 0788 and therefore presumably devised earlier. Its function in 4255 is unknown. TANGRAPHIC RADICALS 19: THE SELFISH SEVEN

Two of the tangraphs I looked at last night have the radical fok on the right:

5091 2gia (first half of 2gia-2mi 'we' [inclusive])

0261 2mio 'I' (with 'skin' on the left!)

There are five more tangraphs with fok. None have anything to do with first person pronouns and none sound like 2gia or 2mio. Three involve death and the other two involve tying. One has two similar readings. (If a tangraph has two readings, those readings are usually completely different.) Do they have anything in common other than a right-hand shape?

Tangraph Li Fanwen number Reconstruction Gloss Source of fok
0186 2ləu rope, string 5848
0406 1dwi transcription tangraph 5819?
0407 1dwiu to kill lice by pressing 4225 (circular)
0788 2mie second half of 1siə-2mie 'death' (cognate to Old Chinese 死 *siʔ 'to die' and 無 *ma 'to not have', 亡 *maŋ 'to die') unknown; 4225?
4225 1sia to kill (cognate to Old Chinese 殺 *ksat 'to kill') 0407 (circular)
5848 1tshio to tie and strap something tightly 5854 (circular)

10.1.1:10: The analysis of 5848 implies that fok is equivalent to foi:


5848 1tshio 'to tie and strap something tightly' (duefok) =

5852 2ʃɨi (first half of 2ʃɨi-2bi 'shallow, superficial') (haafus; not due!) +

5854 1dzio 'to rein in, to tie and strap something tightly' (duefoifer; not fok!)

5854 is derived from 5848: dz- < *N-tsh-.


5854 1dzio 'to rein in, to tie and strap something tightly' (duefoifer) =

5848 1tshio 'to tie and strap something tightly' (duefok; not foi!)

1993 2kwə 'rope, cord' (cirgumfer; why is cir 'water' on the left?)

foi appears in 30 tangraphs and is more common than fok. 16 of those 30 have foi on the right side, so foi cannot be regarded as a nonright variant of fok which is always on the right.

10.1.3:35: Some transcription tangraphs are derived through fanqie:

graph AB pronounced XY =

graph A ... pronounced X ... +

graph B .... pronounced ... Y

0406 doesn't fit the pattern. Its fanqie tangraphs contain almost no components in common. Almost:


0406 1dwi (transcription tangraph) (folbilfok) =

5451 1dwiu 'to close one's eyes' (yeedilcin) +

5819 1lwi 'seed' (panyox)

yox consists of a fok-like element atop fir 'bottom'. Is that a coincidence?

Why does 0406 have two readings? Why not create a fanqie tangraph for 1dwi with the structure yeedilyox? 5451 is the only tangraph with yeedil, so yeedil would be unambiguous.

This begs the question of why yeedilcin has the right-hand radical cin. Why isn't it just yeedil? It turns out that dilcin is short for girdilcin 'black':


5451 1dwiu 'to close one's eyes' (yeedilcin) =

3662 1kaaʳ 'eye' (dexyee; semantic) +

3925 1muʳ 'black' (girdilcin; semantic; cognate to Old Chinese 黑 *hmək 'black')

I don't know what the difference between 1kaaʳ and its more common synonym 1me 'eye' are. 1me 'eye' is cognate to Written Tibetan mig and Old Chinese 目 *muk. 1kaaʳ is Common Tangut rather than Odic Tangut. WE ARE EXCLUSIVELY UNIQUE

Tangut has three words for 'we'. Here are the first two:

2gia-2mi 'we' (inclusive)

2giə-2mi 'we' (exclusive)

These words seem to be related via a ~ ə vowel alternation.

2gia vaguely resembles northwestern Middle Chinese 我 *ŋga 'I' (cognate to Tangut 2ŋa 'I' below) and -2mi has the same initial as the later Mandarin plural suffix 們 -men, but I doubt these pronouns have a Chinese origin. Mandarin also has an inclusive pronoun with a different root: 咱們 zanmen 'we'.

Of course, the resemblance between -2mi and English me is totally coincidental.

Neither 2g-word is related to the singular pronoun which also has a derived plural, the third word for 'we':


2ŋa 'I' > 2ŋa-2niə 'we' (-2niə is a plural suffix)

There is also an honorific 'I' unrelated to the g- or ŋ-words. Could it be related to -2mi via suffixation?

2mio 'I' (with 'skin' on the left!)

The title of this post refers to the fact that the right side (bamquu) of the first half of 2giə-2mi 'we' (exclusive)

is completely unique. Is it a single radical, or can it be split into bam (top) and quu (bottom) from two other sources?

The first half of 2gia-2mi 'we' (inclusive)


is derived from the bottom right of 2giə plus the right (fok) of 2mio. fok is a rare radical that I'll examine tomorrow.

The second half of the inclusive and exclusive 'we'-words has the analysis:


2mi (2nd half of 'we') = 1və̣ə 'to have, own'* + 2ŋa 'I'
In other words, we 'have myself'?

Was the tangraph for 2mi really a simplification of the tangraph for 2ŋa 'I', or was the latter an elaboration of the former?

*Cognate to Old Chinese 有 *wəʔ 'to have, exist'. 挐精 GRASPING SPIRITS

When looking up

2739 2tshwiəʳ 'vinegar'

in David Boxenhorn's Tangut Search, I found

5903 1ŋa 'yeast to make wine'

with the left side of 5818 'rice' instead of 'person' on the left.

Across from 5903 in Li Fanwen (2008: 931) was

5906 2dʒæ̃ 'living being' (jux 'ladder' plus two radicals guo and cin of unknown function)

which is the first half of

2dʒæ̃ 2tsie 'bodhisattva' (lit. 'aware being'; 2tsie is 'to realize, know'; cf. the structure of the Sanskrit original: bodhi 'enlightenment' + sattva 'being')

a word that appears twice in the Tangut-Chinese Pearl glossary with the phonetic glosses

挐精 *ndʒæ tsie ('grasp' + 'spirit')

盞精 *ndʒitʃæ̃ tsie - i.e., *ndʒæ̃ tsie

Li Fanwen (2008: 931) lists a third gloss 拿精 *ndʒæ tsie that I can't confirm in the editions of the Pearl reprinted in Kwanten (1982). This gloss is also absent from Nishida's (1964) version of the Pearl. 拿 'grasp' is a modern variant of 挐 'grasp'.

5906 is only in one original Tangut dictionary, the Precious Rhymes of the Tangraphic Sea, which lists it among the tangraphs of alveopalatal-initial rhyme 2.23 (-æ̃) syllables. It was placed between the homophones

4419 2dʒæ̃ 'Jan' (a place name) and 3927 2dʒæ̃ 'Jan' (name of a star)

suggesting that it belongs to their homophone group. Normally, tangraphs with dʒ- (and dz-) were listed in the Mixed Categories volume of the Precious Rhymes of the Tangraphic Sea and the Tangraphic Sea itself as opposed to the level or rising tone volumes for some unknown reason. Yet the above three dʒ-tangraphs and their homophone

1210 2dʒæ̃ 'egg, ovum'

were in the rising tone volume. Why? Their near-homophone

0205 1dʒæ̃ (for Sanskrit transcription - even though Sanskrit has no æ?)

with a level tone was in the Mixed Categories volume of the Precious Rhymes of the Tangraphic Sea as expected.

Sofronov (1968: II) doesn't list 5903 at all, but Kychanov (2006: 559) lists it with the reading tshɪ 1.30, equivalent to 1tshiə in my reconstruction. This reading obviously is nothing like the Chinese transcriptions in the Pearl. Is it an error, or is there evidence for a second reading with tsh-?

9.29.00:54: I remain baffled by the tsh-reading of 5903 because I can't find any tshiə tangraphs that resemble 5903.

9.29.1:12: Adding 'water' to

5906 2dʒæ̃ 'living thing'

results in

5907 2dʒæ̃ 'blood'; 'living thing' (like its homophone, lookalike, and phonetic 5903)

Or does it? Subtracting 'ladder' from the left results in

2734 1sie 'blood'

I suspect the analysis of 5907 is


5907 2dʒæ̃ 'blood' = 5903 2dʒæ̃ (phonetic) + 2734 'blood' (semantic)

2734 is the normal word for 'blood'. Could 5907 be an Odic Tangut word for 'blood', even though it is monosyllabic rather than disyllabic like other OT nouns? Are 5903 and 5907 cognates: i.e., is 'living thing' from '(thing with) blood'? A NASAL APPEARING OUT OF NOWHERE: THE ENIGMA OF ANGKASA

The Malay name of the Malaysian National Space Agency is

Agensi Angkasa Negara

'agency space nation'

Malay has modified-modifier word order.

Agensi is obviously from English agency.

Negara [nəgara] 'nation' is from Sanskrit nagara 'city' (!). Sanskrit a [ɐ] is not quite like Malay e [ə] or a [a].

I am puzzled by the -ng- of angkasa 'space' which corresponds to Sanskrit aakaaśa without any nasal. Did a Sanskrit long aa really sound like ang to Malay speakers? Are there any other Indo-Malay words with nasals corresponding to Sanskrit vowel length?

Are there any other languages that borrowed VV as VN or that shifted native VV to VN?

9.28.8:28: I found two more Malay -ng- : Sanskrit zero correspondences in "An Introductory Sketch of the Sanskrit Element in Malay" at the beginning of Maxwell's (1907) A Manual of the Malay Language:

M angsoka : Skt aśoka 'a kind of tree' (lit. 'without sorrow'; the a- is also in Amritas 'immortal')

M angsana : Skt asana 'a kind of tree'

These reminded me of the unusual sound change *s > ŋh between vowels in Avestan:

'mind' (gen sg.):

Av manaŋhoo < *manasoo < *manasas

cf. Skt manasas (manaso [mɐnɐsoo] before a voiced consonant or a-)

Was there an intermediate stage *-ŋs- in pre-Avestan like Malay -ngs-? Why did nasalization develop before a fricative?

There is no Avestan parallel to a > ang before k in angsaka.

Did Malay borrow Sanskrit as pronounced by speakers of a Middle Indic or early Dravidian language which had sporadic nasalization of *a? Or was a > ang a Malay-internal development also found in native words? THE GOLDEN GUIDE: LINE 82: TANGRAPHS 406-410

82. This is the last coherent line for a while.

I have no idea who the Shan-O 'mountain masters' are. The name reminds me of French Montagnard. Are the Shan-O an ethnic group distinct from the previous ones mentioned in the Golden Guide, or are they just people of one or more ethnicities who live in the mountains? Kychanov (2006: 171) defined Shan-O as a 'name of a tribe'. Do the Shan-O appear in other texts that clarify their ethnicity?

Tangraph number 406 407 408 409 410
Li Fanwen number 3763 1794 5922 3803 4616
My reconstructed pronunciation 1ʃæ̃ 1ʔo 1ɣõ 1piẽ 2lɨee
Tangraph gloss mountain host; master buckwheat border; cake tasty, delicious; be addicted to
Word Shan-O
Translation The Shan-O are addicted to buckwheat cakes.

406: 3763 'mountain' (a loanword from Tangut period northwestern Chinese 山 1ʃæ̃)̃ is what it seems to be: 'not low':


3763 1ʃæ̃ 'mountain' (ciajux) =

1918 1mi 'not' (ciadiocok) +

3791 2bi 'below; down' (dexjux)

407: 1794 has a circular derivation:


1794 1ʔo 'host, master' (falpik) =

5670 1ʔo 'to hang, join' (haefal; phonetic) +

4489 1phii ''to send, tell someone (fispikbui; semantic?; masters are those who send others?)


5670 1ʔo 'to hang, join' (haefal) =

5642 'to hang' (haezam; semantic) +

1794 1ʔo 'host, master' (falpik; phonetic)

408: 5922 has an inexact derivation. Are dea and bur really equivalent? They cannot be positional variants since they are in the same right-hand position.


5922 1ɣõ 'buckwheat' (pandea) =

5818 2reʳ 'rice' (pancok) +

1046 1khie 'ashen, pale' (hinbur; bur is not dea!)

409: 3803 1piẽ represents unrelated borrowings from Chinese: 邊 *piẽ 'border', 餅 *piẽ 'cake'.


3803 1piẽ 'border; cake' (dexvos)

4951 2lɨaa 'frontier; border' (biodexgik; native Tangut word corresponding to Chn 邊)

4015 2biee 'post, go' (voavos; phonetic)

410: 4616 (analysis unknown) 'tasty; be addicted to' looks like 'mouth' (buk) + 'sweet' (cuoher):


bukcuoher = dexbuk + cuoher

Could 2lɨee 'tasty; be addicted to' be derived from 2lɨaa 'mouth'?

Tangut fonts by Mojikyo.org
Tangut radical font by Andrew West
All other content copyright © 2002-2010 Amritavision