I fell asleep early last night, so you get ten tangraphs - a complete couplet - in this installment instead of five.

Tangraph # 21 22 23 24 25
Li Fanwen number 5932 0981 2218 0113 2620
My reconstructed pronunciation 2mə 2ʔwaʳ 1lɨẹ 1ʃɨẽ 2nwi
Tangraph gloss kind; sort; type; various; multitude thing; wealth; property seedling; second Heavenly Stem to accomplish; to achieve; to become can; be able to
Translation Various things can become seedling[s].

21-22. Tangut normally has noun-adjective order, but here has what appears to be adjective-noun order. Nishida (1966: 274) views cases like 21-22 as compound nouns. Adjectival clauses precede nouns (Gong 2003: 618), so perhaps 2mə can be interpreted as a one-word adjectival clause: 'that are various' modifying 'thing'.

22. Li Fanwen (2008: 165) identified 2vaʳ as a loanword from Chinese 物 (something like *wu in northwestern Tangut period Chinese?) but the vowels don't match and the glottal stop and retroflexion have no sources in Chinese.

23-25. Auxiliary verbs (25) follow verbs (24), and objects (23) precede verbs (24). Hence 'seedling become can' = 'can become seedlings'. The plural of 'seedling' in the translation is inferred from (21) 'various'.

24. A loanword from Chinese 成, which was nearly homophonous with Chinese 聖 'sage' in the northwestern dialect at the time. Hence Tangut

2ʃɨẽ 'sage'

a nearly homophonous borrowing from Chinese 聖, is phonetic at the bottom right.

5.23.1:08: The fourth stroke of 'sage' is vertical in the independent tangraph (left) but tilted \ as a component (right):


5.23.1:00: 25 2nwi could be from pre-Tangut *p-ni-H, which has an *n like Chinese 能 *ndəŋ, transcribed in Tibetan as Hding, Hdïng, Hning, Hneng, ning. But Hd = *nd-, not *n-, and i and e = *ə, not *i. I would expect a Chinese *ndəŋ to be borrowed into Tangut as dẽ (there is no -ə̃ rhyme). However, Chinese 能 *ndəŋ was transcribed with nine (!) rhymes other than rhyme (R) 11 -i in the bilingual Tangut-Chinese glossary Pearl in the Palm (Gong 2002: 318-319; Pearl line numbers in parentheses):

R28 (303)

R30 -ɨə (246)

R31 -iə (291)

R32 -əə (112)

R33 -iəə (354)

R71 -ə̣ (283)

R72 -iə̣ (213)

R82 -eʳ (142)

R90 -əʳ (192)

The numbers of the rhymes are taken from the Tangut monolingual dictionary Tangraphic Sea.

Why was 能 transcribed so inconsistently? Had some of these rhymes merged in the dialect of the Pearl which was written over a century after Tangraphic Sea?

Tangraph # 26 27 28 29 30
Li Fanwen number 2894 4739 5379 0433 3784
My reconstructed pronunciation 2ləu 1tseʳw 1tʃɨə 1biu 1bɔ
Tangraph gloss season joint; division; intimates; layer; number order; sequence in accordance with line; ranks
Translation [The] seasons [and] divisions line [up] in order.

26-27. Referring to

1lɨəəʳ 2ləu 1jaʳ 1tseʳw

'the four seasons and the eight divisions' (of the year) (Pearl 093)

27. Li Fanwen (2008: 751) identified this as a loanword from Chinese 節, but it probably came from pre-Tangut *r-tsek or *rʌ-tsik, whereas the Chinese form had no *r- or *-k during the period of contact with Tangut.

I think *r-tsek or *rʌ-tsik is a native Tangut word cognate to 節 Old Chinese *tsik > *tsit > Middle Chinese *tset.

28-29. Tangut postpositions correspond to English prepositions: 'order in' = 'in order'.

30. Li Fanwen (2008: 610) regarded this as a noun, but I interpret it here as a verb modified by 28-29 with 26-27 as its subject.

5.23.3:00: In Tangraphic Sea, 30 is derived from a 'mirror' tangraph which is turn derived from it:

1bɔ 'line; ranks' = left of 2kiəə 'line; ranks' + left of 1tshiəə 'line; ranks; read; chant; name'


1tshiəə 'line; ranks; read; chant; name' = right of 1bɔ 'line; ranks' + left of 2kiəə 'line; ranks'

Which was created first, 1bɔ or 1tshiəə?

Why do the tangraphs for all three 'line' words have


Because people line up?

Nishida (1966: 242) identified the other shared element of 1bɔ or 1tshiəə

as 轉換 'conversion'. What does that have to do with lines? Is that element short for something else? It is only in eight other tangraphs:

Tangraph LFW # Reconstruction Gloss Notes
4011 2biee line Another synonym for 'line'! The left component is only in three other tangraphs.*
4968 2kʌ - first half of reduplicated word 2kʌ-2kɛ < *krəH-kreH 'cervical vertebra'
5832 1tshiə - first half of reduplicated word 1tshiə-1tshiu 'rainbow' with the same Cə-CV pattern as 'cervical vertebra'
5909 1tshiə evening; night shares a phonetic with 5832 1tshiə and 5870 1tshiəə
5934 2gie dark; gloomy? second half of 2ləi-2gie ''; although 2ləi is 'fog', can 2gie stand by itself?
5937 2ʃʌ to rotate; to alternate 2ʃʌ-2dʒɛ 'saṃsaara' with 'horned hats' added represents 2kʌ-2kɛ 'cervical vertebra' (see 4968 above) which has the same rhymes - a remnant of a Tangut word game?
5938 2gie scripture homophonous with 5934; sharing a phonetic?

from Tangut period northwestern Chinese 經 *kjẽ 'scripture' with a nasal prefix: *N-k- > *ŋg- > g-?; but why no nasal vowel -iẽ?

also first half of 2gie-1niạ 'warp' (textile)
5975 1niạ - second half of 2gie-1niạ 'warp' (textile)

How did Nishida conclude that their shared element meant 'conversion'?

*Here are the three other tangraphs with the left element of 2biee 'line':

Tangraph LFW # Reconstruction Gloss Notes
2625 1vɨi visitor; guest with 'person' on left; probably not loanword from Tangut period northwestern Chinese 位 *wi 'polite measure word for people'?
3889 2biee a surname has elements of 2625 in reverse order but homophonous with 4011 and 4015; their shared left element is a phonetic 2biee
4015 roadside stop; to travel see below

5.23.3:36: The right element of 4015 only occurs in three other tangraphs:

Tangraph LFW # Reconstruction Gloss Notes
0073 1tʃɨə̣ branch of a road the top is from 1tʃɨa 'road' but the function of 5830 (see below) is unknown

derived from 1tʃɨa 'road' via vowel alternation and *S-prefixation conditioning a tense vowel?
3803 1piẽ border borrowing from Chinese 邊; right side is phonetic (cf. 4015, 5830)
5830 1biee disobey phonetic on left shared with 3803, 4015

right from 1dʒɨe 'to go'


Thanks to Andrew West for giving me the Li Fanwen numbers of the entire Golden Guide. He's saved me a lot of time.

Tangraph # 16 17 18 19 20
Li Fanwen number 0618 4052 2612 3791 1507
My reconstructed pronunciation 1tsia 2dạ 2phiu 2bi 2ŋwe
Tangraph gloss hot cold; frigid up; above; over; superior; high below; down harmonious; mild
Translation Hot [and] cold, up [and] down [in] harmony.

16-19 are two pairs of antonyms.

16 'hot' is written as 'fire' + 'sun':


The Tangraphic Sea dictionary derives 'hot' from the 'fire' element at the bottom left of the full tangraph for 'fire' plus all of 'sun':


17 'cold' has the element identified by Nishida (1966: 243) as 'snow; ice' (< Chinese 冫 'ice' + 十?)

18 'up' and 19 'down' are far more complex than their Chinese equivalents 上 (pointing up) and 下 (pointing down).

18-20 contain the most common tangraphic component:


5.21.0:21: 18 'up' consists of


the left of 1ɣʊ 'head' +

the middle of 1nwə 'to know; to realize' +

the phonetic 1phiu 'feast; banquet' (up is 2phiu with a 'rising tone')

I have examined the structure of 19 in "Tangut Search V. 3".

20 'harmony' has an initial ŋ- (pronounced 'ng').

Its graph consists of 'person' flanked by left- and right-hand variants of 'word' (< Chinese 讠'word'):


The bottom V-shape of 'word' becomes フ when at the bottom right of a tangraph. THE GOLDEN GUIDE: LINE 3: TANGRAPHS 11-15

Tangraph # 11 12 13 14 15
Li Fanwen number 5120 2491 2920 2547 0762
My reconstructed pronunciation 1swew 1na 1ʒɨə̣ 1tʃɛʳ 2dʒɛ
Tangraph gloss bright; brilliant; light; celestial body (< 'light thing'?) night; darkness left side right side wheel; revolve; turn
Translation Light [and] dark, left [and] right turn,

I've added Li Fanwen (LFW) numbers to all 15 tangraphs so far. Each of the 6,074 tangraphic entries in Li Fanwen's (2008) dictionary has a number.

11-14 are two pairs of antonyms. The tangraphs for each pair do not contain any common elements.

11 'light' contains an element identified as 'night' (Kychanov 1964: 142) and 'before' (Nishida 1964: 244):

I have examined the structure of 12 'night' in "The Best Article on Tangut Writing Ever".

The ʒ of 13 'left' is pronounced like the -s- of vision.

The left of 13 'left' is

1lạ 'hand'

The left of 14 'right' contains 'sage', last seen in 6 'sun' and 7 'moon':

The of 15 'turn' is English j, so 2dʒɛ is like 'jeh'. I analyzed its tangraph in "What Rotates Skillfully?". THE GOLDEN GUIDE: LINE 2: TANGRAPHS 6-10

Tangraph # 6 7 8 9 10
Li Fanwen number 2449 2814 1374 4861 4184
My reconstructed pronunciation 2bəi 2lhiẹ 1tʃɨə 2ziọ 2ʃɨa
Tangraph gloss sun moon that time to appear; to show; to sparkle; to praise; to exclaim
Translation [The] sun [and the] moon appear [at] that time.

6-10. Each tangraph happens to represent a word. This is not always the case, because Tangut has many polysyllabic words written with more than one tangraph: e.g.,

1riuʳ-2kɛ̣ 'world'

from line 1.

None of the tangraphs in this line are pictures of anything.

6 'sun' and 7 'moon' do not resemble a sun or a moon, whereas the corresponding Chinese characters 日 and 月 are stylized drawings of a sun and a moon. These characters share an element

which appears slightly differently in 'sun' in the Mojikyo font I'm using. Can you guess what it means? Answer: 'sage'.

The left side of 'sun' looks like the tangraph

for 'waist'!

5.19.23:33: The lh- of 7 'moon' may have been like Welsh ll [ɬ].

8 'that' is an abstract concept that is difficult to visualize. The native Tangut dictionary Tangraphic Sea says 'that' is composed of the left side of a near-homophone 1tʃhɨə 'to seek; to encourage oneself' and the center of 1tʃhɨw 'that; other; another':


But it's more likely that 1tʃhɨw 'that; other; another' is 'that' plus an extra element

on the right. The word 1tʃhɨw may be from 1tʃhɨə plus a suffix.

The letter ʃ in 1tʃɨə 'that' represents the sound 'sh'. So t + ʃ = = 't-sh' = 'ch'. A h after represents aspiration. tʃh is like 'ch' plus 'h'.

9. I have no idea what the origin of the character for 'time' is.

10. Tangut verbs appear at the ends of sentences. Hence 1tʃɨə 2zɔ̣ 'that time'cannot appear after the verb:

English: The sun and the moon appear at that time. (verb in the middle)

Tangut: Sun moon that time appear. (verb at the end)

Tangut is terse and has no words for 'the' or 'a(n)'. The 'and' and 'at' in the translation also have no Tangut equivalents.

The tangraph for 'appear' looks like 'wood' + 'water' + ? + 'person':

= + + +

The functions of these elements are unknown.

5.19.1:16: Replacing the last element of 'appear' with

'beast' (resembling 'person' with an extra / stroke)

results in the homophonous tangraph

2ʃɨa 'musk'

representing a borrowing from Chinese 麝 'musk'. Just as 麝 consists of a semantic component 鹿 'deer' and a phonetic component 射, the tangraph for 'musk' consists of a semantic component 'beast' plus 3/4 of 2ʃɨa 'to appear' as a semantic component. THE GOLDEN GUIDE: LINE 1: TANGRAPHS 1-5

Tangut (a.k.a. 西夏 Xixia) was a language spoken in the Tangut Empire (1038-1227) that was destroyed by the Mongols. Tangut is distantly related to Chinese, Tibetan, and Burmese, but its closest living relatives may be the Qiangic languages, possibly including the rGyalrongic languages.

The ancient textbook that I call the Golden Guide was intended to teach one thousand of the approximately six thousand tangraphs (Tangut characters) of tangraphy (the Tangut writing system) to Tangut speakers. Its introduction says that the bright can learn a thousand tangraphs in a month and that even the slow can learn them in less than a year. 小高裕次 Kotaka Yuuji has uploaded an electronic version of the entire Golden Guide at his site. Here's the first line of the Golden Guide:

Tangraph # 1 2 3 4 5
Li Fanwen number 3513 2627 4713 4719 1926
My reconstructed pronunciation 1mə 2lɨə̣ 1riuʳ 2kɛ̣ 2nie
Tangraph gloss heaven; sky earth; land; soil world; capital city world; boundary; tide; court before; long ago
Word world; capital city
Translation Heaven [and] earth, [the] world long ago,

Each line of the GG consists of five tangraphs. Each tangraph represents a single syllable, so each line is five syllables long.

All Tangut syllables have one of two 'tones'. I follow Arakawa Shintarou and indicate these tones with a number before each syllable:

1: the 'level tone'

2: the 'rising tone'

It is not clear whether these tones were really level or rising pitches, or even tones at all. (Details here.*)

Many Tangut words are only one syllable long and are differentiated by tones. For example, the first word in GG,

1mə 'heaven'

sounds like

2mə 'appearance; spirit'

if tones are disregarded.

There are many homophonous tangraphs: e.g., 1mə 'heaven' sounds just like the tangraphs for the syllable 1mə in these disyllabic words:

1ɣʊ-1mə 'supernatural being' (probably cognate to 2mə 'spirit')

1mə-2biu 'midge (a kind of insect; 蠛蠓)'

1tʃɨa-1mə 'guardian spirit of sheep' (probably cognate to 2mə 'spirit')

Similar or identical syllables are written with different tangraphs which may or may not share a possible phonetic component indicating their pronunciation: e.g.,

(a distortion of the Chinese near-homophone 勿?)

the syllable 1mə

which is the left side of 'heaven'.

The aforementioned tangraphs also contain that element

and are also pronounced 1mə.

The right side of 'heaven' is the semantic component

'lightning' (Grinstead 1972: 55)



1mə 'heaven' = 1mə (phonetic) + 'lightning' (semantic)

Not all tangraphs have a structure like 'heaven'. Many have components which have no obvious phonetic or semantic function.

The upside-down e (ə) in 1mə is 'uh'. sounds like the mo- or mother.

The first vowel in tangraph 2

2lɨə̣ 'earth'

sounds like Russian ы, a sound like an 'oo' but without lip rounding. The dot under the second vowel indicates 'tenseness' which may have been creaky phonation.

The left side of 'earth' is the semantic component


which cannot stand by itself.

Each of the first two tangraphs represented a word, but the next two tangraphs form a compound word:

1riuʳ-2kɛ̣ 'world' < 'world' + 'boundary'

1riuʳ and 2kɛ̣ can also mean 'world' by themselves.

The basic Tangut vowels a, e, i, o ,u are pronounced as in Spanish. Hence iu is 'ee-ooh'.

The raised ʳ indicates that the preceding vowel was pronounced with r-quality (retroflexion). ʳ is not a true final consonant. Tangut syllables cannot end in any consonant other than -j (a y-sound as in German ja 'yah', not English j) or -w.

2kɛ̣ 'boundary' is borrowed from Chinese 界.

The letter ɛ represents a vowel between e and a.

The final tangraph in the line

nie 'before; long ago'

has the semantic component

(a distortion of Chinese 前 'before'?)

'before' (Nishida 1966: 243)

on its right side. Reversing its components results in

2ma 'in the past; former times'

The two can combine to form the compound

2ma-2nie 'former times'

(Greatly expanded 5.18.1:39.)

*I suspect 'level tone' and 'rising tone' may actually refer to clear (modal) and breathy phonation. The Tangut terms 'level tone' and 'rising tone' were borrowed from Chinese. These terms are still used in modern Chinese linguistics to describe tone categories regardless of actual tone shapes. The qualities of the Chinese tones that the Tangut heard a thousand years are unknown. By the 11th century, the terms may simply have meant 'tone 1' and 'tone 2'. Hence I prefer to use the neutral notation '1'/'2' instead of using superscript symbols like 'ˉ' and 'ˊ'. A RED FOUR-WHEELED BOX

Eric Grinstead's Analysis of the Tangut Script (1972) lists four Tangut translations of 'box'. I've added a fifth at the bottom.

Tangraph Pronunciation Translations in Li Fanwen (2008)
kha basket
khəu casket; a small box can't find any attestations of 'casket' a note in Homophones text D glosses it as 'book box'
siu cupboard; cabinet
jiẽ bag; sack; envelope; scabbard; cocoon
transcription character for various Chinese words including 匣 'small box'

Our understanding of the meanings of tangraphs is constantly changing. The appendix of Li Fanwen's 2008 Tangut dictionary lists 62 translations from the 1997 edition that were corrected (pp. 961-963), followed by 130 tangraphs whose definitions were discovered between the two editions. Those 192 translations are roughly 3% of all tangraphs.

I was looking for 'box' because I was trying to translate 'van' into Tangut, and the first thing that came to mind was 'four-wheeled box'. I guess I could render that as a compound noun

lɨəəʳ dʒɛ khəu

'four wheel (book) box'

since the van I have in mind is more like a bookmobile. (I'm going to overlook the gloss 'casket' unless it's attested. 'Four wheel casket' sounds like a hearse.)

In Tangut, adjectives follow nouns, so a red van would be a

lɨəəʳ dʒɛ khəu nie

'four wheel (book) box red'

ADDENDUM: The transcription character has the analysis


left of (homophone; phonetic), first syllable of xæ-phɔ̃ 'a kind of grass' +

right of lɨə̣ 'earth' (why?)

Since transcription characters represent sounds, they should have no semantic components other than something like


which is in the transcription character


from this post. (The right two-thirds of vã resemble without a dot. Is that meaningful?) What is 'earth' doing in xæ? WHAT IS THE FOURTH SURNAME?

Can you guess the meaning of the first Tangut character?


lɨəəʳ '?' =

bottom of ŋwəʳ 'fourth' (almost homophonous with ŋwə 'five'!) +

right side of lɨəəʳ, second syllable of the surname

məi-lɨəəʳ (by coincidence I saw the trailer for Molière this afternoon)

The answer is at the end of this post.

The second and third characters are derived from ... the first (in bold)!?


right side of lɨəəʳ, second syllable of a surname =

top of ʃwɨi 'year; age' +

which has a 'horned hat' atop kiw 'year'

all of lɨəəʳ '?' (semantic)


lɨəəʳ, second syllable of a surname =

left of the surname riuʳ (a clan related to the məi-lɨəəʳ?)

right of lɨəəʳ '?' (phonetic)

Note that the right of lɨəəʳ resembles

lew 'one'

The vowel of lɨəəʳ implies pre-Tangut *r-ləə with like

Mawo Qiang gʐə

Ronghong Qiang ɣʐə

rather than a front vowel like

Written Burmese လေး leḥ


gDong-brgyad kɯ-βde

Somang kə-wdî

but Zbu kə-vldaʔ!

< all Proto-rGyalrong *pltej?

Written Tibetan bzhi < *blyi

Old Chinese *sli(t)s

Answer (which is obvious to anyone who recognizes one or more of the above Sino-Tibetan forms): Select the space between the single quotes below.


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