I should have mentioned the obvious in "Of Dawn and Demons":

+= or

Li Fanwen radical 238 'night' + filler = TT0040/K4027, (no TT number)/K4026 thwo R51 1.49 'nightmare; dream; illusion' (Grinstead: 'demon'! < creature in a nightmare?)

K = Kychanov 2006

was not like


Li Fanwen radical 177 'horse' + filler = 'horse'

in which ヒ added no semantic content. 'night' + ヒ represents something to do with night (nightmares) rather than another word for 'night'. Perhaps ヒ is only semantically vacuous when it accompanies nonindependent elements like radical 177.

Tangraphic Sea analyzed TT0040 as


left of TT0038 miee R40 1.39 'dream' (more on this tangraph tomorrow)

(why not all of TT0037 R31 2.28 'night'?)

right of TT1160 bu R2 2.2 'border'

but I am skeptical about whether anyone would actually recognize ヒ as being from 'border', particularly since ヒ could be from over 500 tangraphs and 'border' has no obvious connection with 'nightmare'.

I don't know whether TT0040 can occur alone or not. I suspect that thwo is not a word but can only be half of the word

K4028.1 swi thwo R11 2.10 R 51 1.49 'terrible dream; shadow'

I also don't know if swi can stand alone. Is swithwo originally a compound or was it always a polysyllabic morpheme?

No analysis for the tangraph of swi (TT3041 'unreal/imagined image') is known. Its resemblance to TT0040 is probably deliberate, and I wonder if its right side is the real source of the ヒ in TT0040. Its left side is Li Fanwen radical 007:

Nishida numbered this radical 191 and had no definition for it. It appears on the left side of four other tangraphs:

Tangraph Tangut Telecode Reconstruction Rhyme Tone.rhyme Gloss (Li Fanwen 1997)
3038 pi R11 1.11 (a surname)

3039 wẹ < ?*sɤ-p- R68 1.65 foolish, stupid

3040 si R11 2.10 (auxiliary); K5051: grammatical word <toil; instrument; means; way; method; mode>; Grinstead: Skt si
3042 pi R11 1.11 (name of an insect)

TT3040 is probably phonetic in

TT3725 ʃɛ̃ R42 2.36 'servant, lackey' (with 'person' on the left)

Could LFW radical 007 be a phonetic element for SI and PI-type syllables, or even syllables that have both types of consonants: e.g., TT3041 swi could be from *psi, and TT3039 wẹ may be from *sɤ-pe? (6:15: But clusters and presyllables were presumably gone by the time the script was designed, unless its creators had conservative dialects in mind.) Perhaps, but it appears that 007 had very different phonetic values in right-hand position, as I will demonstrate in a future post. 007 was not unlike the modern hangul letter ㅇ which is zero in initial position and -ng in final position.

Next: Night sight. OF DAWN AND DEMONS

Given tangraphs like

TT5233 rieʳ R79 1.74 'horse'


TT4579 lu (ly?) R2 2.2 'the Chinese surname 吕 Lü'

which apparently have the structure

meaningful element 'X' + semantically vacuous element (filler)


radical 177 'horse' + filler = 'horse'


radical 96 / calque of '吕 Lü' using Tangut 反 = Chn 口 (both 'mouth') + filler = '吕 Lü'

I might expect to figure out what an element X means by examining tangraphs of the type X + ヒ.

Last night, I asked,

Is the resemblance of Li Fanwen radicals 12 and 238 (= TT0037) purely coincidental?

If the two are somehow related, I might expect their X + ヒ derivatives to have similar meanings, but they don't:


radical 12 + filler = TT2178 ɣa R17 1.17 'dawn; morning'

+= or

radical 238 + filler = TT0040 thwo R51 1.49 'nightmare; dream; illusion' (Grinstead: 'demon'! < creature in a nightmare?)

the version with 二 covering both radical 76 and ヒ is a variant without a TT number

the elements beneath 二 look like TT2816 twụ R61 1.58 'each; place' (Grinstead: 'mutual'; Kychanov 2006 should have a listing for this around 4019, but it doesn't seem to)

It seems that radical 12 is the opposite of radical 238, signifying day instead of night. Here is a list of all tangraphs with radical 12 on the left besides TT2178:

Tangraph Tangut Telecode Reconstruction Rhyme Tone.rhyme Gloss (Li Fanwen 1997)

2174 gɨɨ R33 1.32 explore, go in quest of

2177 notes, commentary
2175 ne R34 2.30 evening, night
2176 xo R51 1.49 his; its, that
2179 kwæ R18 1.18 surplus, remnant (resemblance to Middle Chinese 過 *kwah 'exceed' coincidental?; vowel does not match)

2180 lɛ̣ R63 2.53 evening, night; < *sʌ-le (with 'brightened' [i.e., raised] vowel?); cf. Written Tibetan zla-ba < *sl- 'moon', Old Chinese 夕 *s-lak 'evening'

TT2175 and TT2180 both mean 'evening, night' and the others have nothing to do with day or night. The analyses of both tangraphs are probably in the lost second volume of Tangraphic Sea.

TT2175 looks like 'day' + radical 23刂 'not' (normally on the left - why is it on the right here?)/

TT2180 could be a compound of 'day' with the right side of

TT2890 muʳ R80 1.75 'black'

which is the source of the right side of

Tangraph Tangut Telecode Reconstruction Rhyme Tone.rhyme Gloss (Li Fanwen 1997)
2861 dwiu R3 1.3 shut/close one’s eyes (i.e., see blackness); the right of TT3663 kaaʳ R88 1.83 'eye' is on the left)
3625 ɣʊ R4 1.4 evening (with the left of TT3619 ɣʊ R4 1.4 'head' on the left as phonetic)

4232 muʳ R80 1.75 puzzle (Grinstead: 'ignorance'); homophonous with/cognate to muʳ R80 1.75 'black'; < 'something that is obscure'; cf. Eng in the dark (as a metaphor for ignorance)
5055 swiftly (Grinstead: 'disappear' < 'go into blackness'?)
5334 tshɨ R31 1.30 evening, night (Grinstead: 'dusk') (with all of TT5333 tshɨɨ R33 1.32 'ranks, read aloud, chant' on the left as phonetic)
(could this simply be TT5333 plus ヒ rather than the right side of TT2890?)

I don't understand why radical 238 'night' can stand alone

whereas radical 12, 'horse', and '吕' need ヒ to support them:

Is the paucity of radicals that can be independent tangraphs meaningful, or is it totally arbitrary? TWO-NIGHT

TT0037 R31 2.28 'night'

appears to contain Chn 二 'two'.

Last night, I wondered if 二 in TT0037 is derived from the 亠 of Chn 夜 'night', just as 二 in

Li Fanwen radical 079 'sound'

may be derived from the 亠 of Chn 音 'sound'. But the bottom parts of TT0037 and 夜 do not resemble each other at all. The tangraph that 夜 most closely resembles is

TT1955 ʔwɨʳ R92 2.77 'literature'

which has a completely different reading and meaning.

If TT0037 were based on 夜 (or some other sinograph), I would expect it to be an indivisible grapheme. Yet its homophone

TT0036 R31 2.28 'star'

looks like 二 (not TT0037 as a whole) plus

TT2853 tʃɨɨʳ R100 2.85 'constellation'

Moroever, two near-homophones of TT0036 and TT0037 contain left-hand elements resembling TT0037 with the two diagonal strokes atop instead of beneath 二:

TT2174 gɨɨ R33 1.32 'explore, go in quest of'

TT2177 gɨɨ R33 1.32 'notes, commentary'

There is also a tangraph

TT2180 lɛ̣ R63 2.53 'evening, night'

with the same left side whose meaning overlaps with TT0037.

Is the resemblance of Li Fanwen radicals 12 and 238 (= TT0037) purely coincidental?

Next: Of dawn and demons. LI FANWEN RADICAL 18: TWO

The only tangraph with

Li Fanwen radical 203

that I found which wasn't pronounced giuu R7 was

TT0005 ʒi R11 2.10 (transcription character), 'emerald'

consisting of

Li Fanwen radical 18


Li Fanwen radical 107

plus Li Fanwen radical 203.

Radical 18 resembles Chn 二 'two', pronounced something like *ʒi in the Tangut period northwestern Chinese dialect. (Cf. the pre-Tangut period Tibetan transcriptions of 二: zhi, zhï, Hzhi.) Thus it must be phonetic in TT0005 and in these tangraphs:

Tangraph Tangut Telecode Reconstruction Rhyme Tone.rhyme Gloss (Li Fanwen 1997) Gloss (Shi et al. 2000)
0006 ʒɨ R11 2.10 all 皆 'all'
3351 extreme 最 'most'
4563 (unknown) 钻穿 'drill through'
4562 作 'do' 为 'do'
0052 ʒɨi R10 1.10 (transcription character) (a surname)
0007 2.9 (a place name) (transcription character)

0044 dʒi R10? (but Sofronov has R19 -ɨa!) ? unknown 行 'go', 步 'walk'

The inclusion of TT0123 is debatable, as its initial is an affricate, its rhyme is uncertain, and its top and bottom left may constitute a unit

that also appears in

TT0124 wɨe R36 1.35 'go'

Note, however, that the presence of Li Fanwen radical 18 in a tangraph does not guarantee a ʒi-like reading:

1. as phonetic for ʃwio and ʃwiõ

This second phonetic value for 二 is similar to Tangut period northwestern Chinese 雙 ?*ʃwõ 'double':

Tangraph Tangut Telecode Reconstruction Rhyme Tone.rhyme Li Fanwen (1997) gloss
0001 ʃwio R50 1.48 appear, raise
0043 sweep
0041 night
0042 ʃwiõ R58 1.56 patrol; similarity to Tangut period northwestern Chinese 巡 ?*sỹ 'patrol' probably coincidental since initials and finals do not closely match

Note that the last two tangraphs have the same components in slightly different configurations and that the reading for the second has a nasalized vowel.

The left side of TT0041 is

TT0037 R31 2.28 'night'

Is it a coincidence that TT0041 happens to contain a semantic component which in turn contains the phonetic 二?

2. 二 as semantic element 'two'

The only example I know of is

TT0590 nɨɨ R33 1.32 'two'

Note the resemblance of TT0590 to a mirror-image version of Chn 貳 'two'.

3. Function unknown

These tangraphs may contain radicals which contain 二 as an intrinsic component. See the following section.

Tangraph Tangut Telecode Reconstruction Rhyme Tone.rhyme Li Fanwen (1997) gloss Tangraphic Sea analysis
0053 l(d)ị R70 2.60 brave left is source of right of TT1267
1267 kiʳ R84 1.79 brave, courageous right derived from top and left of TT0053
0051 beʳ R82 2.71 meet center is source of right of TT2332
2332 bẽ R41 1.40 pain, suffering, hardship right derived from center of TT0051

4. False

One should not confuse 二 with other radicals which contain 二 as an intrinsic component: e.g.,

Li Fanwen radical 079 'sound' (possibly derived from the top four strokes of Chn 音 'sound' with a horizontal line replacing the top dot and the third and fourth strokes straightened out)

A number of other radical-like elements appear to contain 二 but I need to investigate whether these components can be broken down into 二 plus other graphemes:

I have already mentioned two of these (see the discussions of TT0044 and TT0041 above). There may be others (e.g., the right side of TT2332 above).


While looking at Sofronov's (1968 II: 365) listings of tangraphs with

Li Fanwen radical 114

(TT3966-3979), I found a trio of homophones

TT3963 giuu R7 2.6 'snake'

the right side is from TT0075 phɔ R52 2.43 'snake'

TT3964 giuu R7 2.6 'lucky'

the right side (Li Fanwen radical 258 'child') resembles the left or right sides of Chn 福 'good fortune'

TT3965 giuu R7 2.6 (insect name)

the right side is TT5086 tõ R56 1.54 'insect'

all sharing

Li Fanwen radical 203

which I call the 'inverted bird' because it resembles

Li Fanwen radical 196

which can also be an independent tangraph TT5649 dʑɨw R46 1.45 'waist' (also in some tangraphs for birds - because of its vague phonetic similarity to TT5717 dʑwiõ R58 1.56 'bird'?).

The 'inverted bird' also appears on the right of

TT0119 giuu R7 1.7 (transcription character)

the left side (Li Fanwen radical 079) is 'sound'

TT0005 ʑi R11 2.10 (transcription character acc. to Li Fanwen 1997)

Shi et al. (2000) defined this as 绿玉 'emerald'

is the version in Sofronov (1968 II: 279) with 'person' on the bottom right legitimate or a mistake?


1. Why does 'inverted bird' have the sound value giuu R7?

2. Nishida (1966: 244) defined 'inverted bird' as the radical having to do with 吉祥 good luck (which sounds nothing like giuu in Chinese). This identification is presumably based on TT3964 'lucky', but is it valid for the other non-transcription tangraphs? I suspect that 'inverted bird' is purely phonetic in TT3963 and TT3965 (unless they are extended uses of giuu 'lucky' referring to lucky creatures), though it may be semantic in TT0005 if it meant 'emerald' and if the Tangut considered emeralds to be auspicious.

3. Why isn't TT0005 pronounced giuu?

Next: 'Two' is the answer to question three. WHAT WOULD A SUMEROLOGIST SAY ABOUT TANGUT?

Maybe tangraphy is easy after all:

Sumerian homonymy and the primacy of semantic association in the mapping of the lexicon onto the sign system are responsible for the notorious homophony and polyphony of cuneiform signs, features that complicated the initial decipherment and were the source of considerable resistance to the decipherment once it was accomplished. [Cf. my own resistance to the semantocentric interpretation of tangraphy. -A] Over twenty different signs can be read /du/, and the sign KA has several times the half-dozen readings mentioned above [ka 'mouth', 'tooth', kir4 'nose', inim 'word', 'voice, sound', dug4 'to say' -A].

- Jerrold S. Cooper, "Sumerian and Akkadian", in Daniels and Bright, eds. (1996: 43)

Then again, the logic behind at least those six readings of KA is transparent. KA originated as the pictogram SAG 'head' with "some lines added where the mouth would be" (p. 41). A nose is above the mouth, a tooth is inside the mouth, and sounds and words come out of the mouth when one speaks. The semantic value of many tangraphic elements is unclear: e.g.,

TT0160 lo R51 2.42 or 1.49 'strainer; skimmer'


TT0162 tɕhɨw R46 1.45 'haze'.

which also contains the very common mystery element

appearing in 524 (#3711-4234; 9%) of the 5803 tangraphs in Arakawa and Kychanov (2006). Such frequency is what I would expect from a Bantu class prefix (i.e., something designating a subclass of a part of speech rather than an entire part of speech). But I have not yet found any consistent semantic function for ヒ. Here's a sample of readings and meanings for ヒ-tangraphs from A&K (2006) which include nouns, verbs, adjectives, and even a grammatical word:

3711. la 2.14 'maternal aunt'

3761. rjạ 1.82 'at a distance; measure of distance; within; repeated action; grammatical word'

3811. nju 1.3 'ear'

3861. njẹ 2.72 'lend; loan; deputy'

3911. ŋi 2.10 'fat (n.)'

3961. 1.27 'nape of neck; back of head'

4011. tɕjõ 1.56 'plateau with water; grass; marshland'

4061. mụ 1.75 'stupid; vulgar; ordinary people'

4111. kêi 2.31 'ankle'

4161. ŋwə 1.27 'die (v.); vessel'

4211. wẹ 2.71 'rain (v.)'

4234. ndje 2.35 'grieve; weep'

I have done this exercise before, but it may be useful to repeat it with a different sample. A 'STRAINED' SEMANTIC SOLUTION

Looking over the meanings of the tangraphs containing

TT0160 = A&K0002 lo R51 2.42 or 1.49 'strainer; skimmer'

I realized that many of them involve separating one thing from another:

Connection with 'strainer' A&K number and reading
physical dig out < separate something from the ground 0008 khu 2.5 'dig; look for; turn over'
0009 lo 1.49 'look for; search; dig' (cognate to 0002 lo 2.42 or 1.49 'strainer'; another spelling of the same etymon if the two share the same tone?)
pull out < separate something from another thing 0003 lhje 1.14 'pull out; pluck'
0004 lhje 1.14 'pull out a root; large piece of meat; weed' (latter two 'things to be pulled out'?) (same etymon as 0003?)
0006 ndʑê ? 'pull out; exempt; save from work'
pull string away from bow; cf. Eng strain which is both 'pull tight' and 'filter' 0007 thu 1.1 'draw a bow'
shake/knock/beat to separate something from another thing; pinch < pinch and pull out? 3156 sjə 2.29 'pinch; shake off; knock off, beat out'
mental interrogate, investigate < strain through facts; quick < quick wit < wisdom < ability to strain through facts 0010 0005 ka-lwa 1.17 1.20 'suddenly; wisdom; interrogate'
0012 0005 tsa-lwa 1.20 1.20 'rapidly'
0011 thə̣ 1.68 'quick witted; intellectual; curiosity; knowledge; investigate' ('to strain through facts'?)

There are only two exceptions:

A&K5788 ndʑê ? 'patterned; silk'

in which A&K0002 is an abbreviation for

A&K0006 ndʑê ? 'pull out; exempt; save from work'


A&K4214 tɕhjeɯ R46 1.45 'haze; misty; fogs; vaporous; steaming'

in which the function of A&K0002 is unknown.

Tangraphic Sea analyzed A&K4214 as


left of A&K0006 ndʑê ? 'pull out; exempt; save from work' +

right of A&K4221 khi R11 2.10 'air; breath; vapor; steam; signal by fume'

obviously a loan from Tangut period northwestern Chinese 氣 *khi 'air', etc.

The right side is semantic but the left side couldn't be: fogs obscure rather than 'pull out' things. And I doubt the Tangut used fogs as an excuse to take a break from work.

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