When I wrote about the Tangut radical


in "Dir-ivatives", I did not know about its long-tailed variant


which is in only one character used to write the first syllable of the disyllabic verb

1kwæ 1kwʌ < *P-kra P-krə 'to step on; to burst, split'

Since short-tailed dir is the normal form of the radical, the character for the second half with dir might have been created before the character for the first half. But how likely is that?

The Tangraphic Sea analysis of the second half does not mention the first:


left two-thirds of 2tiə̣ 'to damage, break' + right of 2kwə 'convulsion, spasm; lame; bent'

The first was analyzed with the same source graphs in reverse:


right of 2kwə 'convulsion, spasm; lame; bent' + left two-thirds of 2tiə̣ 'to damage, break'

In this case, the long-tailed variant dis seems to have been created to accomodate a different arrangement of radicals. Tangut radicals of the 乚-type cannot occur in the center and inserting dir between the two elements

from 2tiə̣ 'to damage, break' was also unacceptable, so there was no choice but to move dir to the left and elongate its tail by analogy with other left-hand 乚-type radicals:



I'll look into why those radicals lengthened their tails later. BEAUTIFUL ENEMY

Why are

1swi (second syllable of

2ziəʳ 1swi 'beautiful')

2si (second syllable of

2khiã 2si 'enemy, foe')

the only two Tangut characters with the component


The analysis of the second half of 2ziəʳ 1swi 'beautiful' in Tangraphic Sea is


left of 2si (second half of 2khiã 2si 'enemy, foe') + left of 2ʂwɨo 'dignified'

and I wouldn't be surprised if the lost analysis of the second half of 2khiã 2si 'enemy, foe' turned out to include the second half of 2ziəʳ 1swi 'beautiful' (and one of the thirteen characters containing dir):


The only thing the two have in common is si-type readings. But if qai was a phonetic for si, why weren't other si-syllables like

1si 'west'

1si 'wood'

2si 'year'


There are only four Tangut characters with


which is almost always on the left side:

Li Fanwen 2008 number Tangraph Reading Gloss Notes
186 2ləu rope, string derived in Precious Rhymes of the Tangraphic Sea from 1622 which is derived from 186 - circular!
488 second half of 1kə 2ləu 'felt, rug' no known derivation; see previous post for right-hand component

is 1kə 2ləu 'rug' derived from 2ləu 'string', and if so, what is 1kə? is it a coincidence that 1kə 2ləu 'rug' sounds like 2kə 2ləu 'rope' except for the tone of the first syllable?
1622 2riẽ string, rope, belt derived in Combined Homophones and Tangraphic Sea from 186 which is derived from 1622 - circular!
4756 silk surrounded by Nishida radical 31 'cloth'; derived in Combined Homophones and Tangraphic Sea from 1622

One possible chain of derivation is:

186  2ləu 'rope, string'

> 488 second half of 1kə 2ləu 'felt, rug' (phonetic; also semantic?)

> 1622 2riẽ 'rope, string' (semantic)

> 4756 2riẽ 'silk' (phonetic; also semantic?)

The reverse is also possible:

1622 2riẽ 'rope, string'

> 4756 2riẽ 'silk' (phonetic; also semantic?)

> 186  2ləu 'rope, string'

> 488 second half of 1kə 2ləu 'felt, rug' (phonetic; also semantic?)

8.2.20:36: What does his mean? Nishida (1966: 244) already identified his radical 239


as 'string', so maybe his is 'rope', though it's not in the first half of

2kə 2ləu 'rope'

whose left half is, of all things,

cir 'water'!

Nor is it in either half of

1ʂɨa 2ʔiew 'rope' (whose first half can mean 'rope' by itself; Li Fanwen listed no nondictionary attestations for the second half, so I don't know why he glossed it as 'straw rope')

Why aren't all 'rope' or 'string' words written consistently? Are the four words written with his different from the others in some way?

Are his and/or hos derived from the Chinese radical 糸 'string'? DIR-IVATIVES

In my last post, I looked at the fourteen characters containing


the left half of

2si 'year'

Now I'll look at the right half of 2si 'year'


which is also on the right of thirteen other characters (and like other ヒ-type components, is never in any other position).

Li Fanwen 2008 number Tangraph Reading Gloss Notes
485 2swiə felt, rug Nishida radical 109 (Boxenhorn code: ful) 'absence' on left

derived from 5878 in Precious Rhymes of the Tangraphic Sea
488 2ləu second half of 1kə 2ləu 'felt, rug' rare radical his on left (in only three other characters)
609 1zi other, that, it function of left side (Nishida radical 128) unknown;

derived from 2977 tha 'narrowness' (tone unknown) and 5878 in Tangraphic Sea
610 2kwə first half of 2kwə 2le 'to burst, break' left is Nishida radical 123 'equal'; right two-thirds is 5308 'bent'

derived from 5308 (derived from 610 - circular!) in Precious Rhymes of the Tangraphic Sea
1167 1twe lame 5308 'lame' beneath Nishida radical 19 'army' from phonetic 1061 1twe 'pile up'

derived from 1061 and 5308 in Precious Rhymes of the Tangraphic Sea
2453 2swiə the surname Swy Nishida radical 204 'person' + 485
2522 1lɨə first half of 1lɨə 2lɨu 'slow' phonetic 2302 1lɨə 'wind' on left; ironic that 'wind' sounds like part of the reduplicative disyllabic word 'slow'
2530 2si second half of 2khiã 2si 'enemy, foe' the left side (Nishida radical 160 / Boxenhorn code qai) is shared with only one other character that surprised me; more in a later post
3886 2dʐæ lameness Nishida radical 105 'leg' + 5308

derived from 3990 'leg' and 5308 in Mixed Categories of the Tangraphic Sea
5308 2kwə convulsion, spasm; lame; bent Nishida radical 236 'hand' on left

derived from 5611, first half of 1khwiə 1lwɨo 'crooked', and 610 (circular!) in Combined Homophones and Tangraphic Sea
5312 1kwʌ second half of 1kwæ 1kwʌ 'to step on; to burst, split' Nishida radical 236 'hand' on left

Nishida radical 123 'equal' in center

derived from 5381 2tiə̣ 'to damage, break' + 5308 in Tangraphic Sea
5583 1tsə̣ first half of 1tsə̣ 2dwəu 'chopsticks' Nishida radical 236 'hand' on left

right-hand component incorporating dir is unique; function of added ユ unknown

see this post for its Tangraphic Sea analysis
5878 2bɛ̣ classifier for long thin objects; branch Nishida radical 157 'small' (my gloss, not his) on left

derived from 1861 tshiẽ 'small' and 5879 'year' (why?) in Precious Rhymes of the Tangraphic Sea
5879 2si year see previous post on the left-hand component

see this post for its Precious Rhymes of the Tangraphic Sea analysis

My guess as to how these characters are tied together (revised 8.2.0:55):

5878 2bɛ̣ 'classifier for long thin objects'

> 5583 (were chopsticks counted with 2bɛ̣?)

> 485 2swiə 'rug' (woven out of yarn - was a strand of yarn counted with 2bɛ̣?)

> semantic in 488, second half of another word for 'rug'

> phonetic in 609 1zi, 2530 2si, 5879 2si

8.1.23:08: But are there other cases of

- sw-phonetics for z- which might have been a very un-s-like [ɮ] (or even the much more similar initial s-)?

- -iə-phonetics for -i?

> phonetic in 2453 2swiə 'the Swy clan' (known for a disabled member? - this is total speculation and the weak link in the chain)

> semantic in 1167 'lame', 3886 'lameness', 5308 2kwə 'lame' > phonetic in 610 2kwə, 5312 1kwʌ

> semantic in 2522, first half of 1lɨə 2lɨu 'slow' (due to lameness?)

I initially thought dir might have originated as a tamga for the Swy clan, but the character for Swy is probably derived from 485 2swiə 'rug' by adding 'person' rather than the other way around.

Was dir a drawing of a branch? A lame leg? Drawings of those things conflated into one?

Is dir in any way related to


which lacks a dot? THE MYS-TEA-RIOUS FOURTEEN

Two posts ago, I was puzzled by the component

which Nishida (1966) had no gloss for. I'll refer to it by David Boxenhorn's code tea.

There are fourteen characters with tea:

Tangraph Position of tea Reading Gloss Notes Class
left 2ɣa in function of right component unknown 5
2si year
1tsha hollow bag 'waist'/'bird'/'sun' on right; see Andrew West's post on this component 3
right 1ʔiee to exempt; to grant a pardon 'not' on left; function of middle component unknown 1
1bəu cattail 'grass' on left; < Chn 蒲 *bəu 4
1ʔie light (not heavy) 'person' on left; derived from bottom right (not the bottom as a whole!) of 2ʂɨə 'light' 1
1xiu void, empty 'waist'/'bird'/'sun' on left (cf. 1tsha 'hollow bag' above) 2
1tswiu kettle 'fire' on left 3
bottom right 2ʂɨə light (not heavy) 1ʔie 'light' beneath top component of unknown function 1
1pəəu full 'wood' on top of 1xiu 'void' (see above) 2
1gaaʳ to play < Chn 玩 *ŋgwã? (but Tangut is missing -w-, and Chinese nasalization is normally borrowed as Tangut nasalization, not retroflexion); top and left components of unknown function
1tswiu pot to boil tea same word as 'kettle' but 'metal' on top if used for tea; characters derived from each other 3
2tshõ cooking vessel 'metal' on top of 1xiu 'void' (see above)
bottom 1xiõ fragrant < Chn 香 *xiõ; top component of unknown function 4

The words written with tea can be grouped into five classes (see the last column above):

1. 'light'

2. 'empty (cf. light)/full' (to pardon is to lighten [Grinstead's gloss] or empty a load, and play can be seemingly purposeless and hence empty)

3. 'container' (something that is empty; was it originally a pictograph for a vessel?)

4. phonetic loan:

1pəəu 'full' as phonetic in the Chinese loan 1bəu 'cattail', though the Tangraphic Sea analysis of the latter doesn't contain the former:


top of 2dʐæ 'disorderly' + all of 1xiu 'void'

1xiu 'empty' as phonetic in the Chinese loan 1xiõ 'fragrant', though the Tangraphic Sea analysis of the latter doesn't contain the former:


right of 2məəi 'coal, mine, holy mountain' + left of 2ɣa 'in'

7.31.22:18: 1ʔie 'light' may be cognate as well as phonetic in 1ʔiee 'to exempt, lighten', and theTangraphic Sea analysis of the latter does contain the former:


left and center of 1lwəụ 'level' (implying exemption for fairness?) + 1ʔie 'light'

5. miscellaneous (and the most mysterious): 'in', 'year'

The first three could be conflated into a single semantic category.

7.31.22:22: Here's how I think the functions of tea are related:

pictograph of a container

> a container is empty > empty > light (devoid of weight) > phonetic for syllables resembling those of words in the preceding semantic categories

> being in a container > in

Could a year be considered a container? WOOD IN THE WEST, BUT NOT WITH THE WEST

In my previous post, I referred to "the si-group" in Homophones. One of its members is


1si 'west' (< Chn 西 *si 'id.') = top of 1si 'wood' + all of 1niəə 'bask, shine on' (cognate to

2niəə 'day' [source of the left side of 1swie from my previous post]

and Old Chinese 日 *nit 'sun'?)

which contains 1si 'wood' in its Tangraphic Sea graphic analysis. I would expect 1si 'wood' to be in its homophone group as well, but 1si 'wood' in fact belongs to a separate group (Nishida's VI.21: 29B62-29B68) with a different fanqie


1siu + 1khi

and without any rising tone members (i.e., no 2si).

What is the difference between these two si-groups? Nishida (1966: 116) reconstructed the 'wood'-si group as 1stshiɦ and the 'west'-si group as 1siɦ, but the Tibetan transcriptions of 'wood' and 'west' are zi and si (Tai 2008: 208). Other characters in the same fanqie initial chain as 'wood' were transcribed with s-, gs-, and bs- as well as z-. There are only two Tibetan transcriptions of characters in the fanqie chain of 'west': zi 'west' itself and

2si (perfective suffix; nominalizer; definition from Jacques 2007: 149)

which was inexplicably transcribed as swi with -w-. Perhaps this -w- reflects a nonstandard form.

Until now I have been following Gong Hwang-cherng who reconstructed an s- for both fanqie chains and regarded z- as a class IX initial (i.e., a liquid). (I have interpreted his z- as [ɮ].) What if Tangut had a three-way distinction between s-, z-, and ɮ-?

Example (all end in 1-i)
'west' group

'wood' group

Homophones class VI (alveolars) IX (liquids)
Homophones group (Nishida numbering) 24 21 15
Nishida (1964) s- stsh- ʁ-
Gong and this site until now s- z- (my [ɮ])
New proposal s- z- ɮ-
Tibetan transcriptions s- s-, gs-, bs-, z- z-, gz-, Hz-

ɮ is not a liquid but is lateral like the true liquid l. There is no Tibetan letter for ɮ. The Tibetan transcriptions with z may indicate dialect(s) in which ɮ- had shifted to z-. YEAR IN A BRANCH

When looking up

2bɛ̣ 'branch'

for the first of my posts on Tangut names for chopsticks, I found its neighbor

2si 'year'

which Li Fanwen regarded as a loan from Chinese 歲 *swi 'id.' I thought it would be appropriate to write about this word and its character for my birthday - the first day of my 42nd or 43rd year depending on which system of reckoning you favor.

I thought it was odd that Chinese *swi would be borrowed into Tangut as 2si even though 2swi is a possible Tangut syllable. In Homophones, 2si is grouped with characters transcribed in Chinese as *si and in Tibetan as si (Nishida's VI.24: 30A23-30A28), not the swi-characters in a separate group (Nishida's VI.32: 30A72-30B11*). Could 2si be a native word that happens to resemble Chinese *swi?

What is the semantic difference between 2si and these other words for 'year'?

1vɨi (character analyzed as containing the right of the character for the next word)


1ʂwɨi (lookalike of Chinese *swi; Chinese *s- would not have been borrowed as ʂ-; written as 1kiew plus a 'horned hat' from 1tʂhɨiiw 'to say')

1tʂɨẹj (character derived from 1kiew and vice versa!)

1swie (< earlier Chinese 歲 *swiejh which later developed into *swi; unlike the other four, not written with Nishida radical 230 'tilt' - and why would 'year' contain 'tilt'?)

Li Fanwen did not list any attestations of 2si outside dictionaries. Could it belong to the register that has been identified as the 'ritual language'? That register contains no Chinese loanwords.

The character for 2si is even more mysterious. The Precious Rhymes of the Tangraphic Sea analyzed it as

2si 'year' = left of 2ɣa 'in' + right of 2bɛ̣ 'branch'

This makes no semantic or phonetic sense. Neither half of the character (Nishida's radical 158 / Boxenhorn code tea and Boxenhorn code dir) has any known meaning.

*7.29.20:16: The swi-group in Homophones contains 4 1swi and 4 2swi. The fanqie for 1swi is


1sw + 1twi

On the other hand, the si-group in Homophones contains 2 1si and 9 2swi. The fanqie for 1si is


2s + 1ki

The fanqie for 2swi and 2si are unknown.

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2013 Amritavision