If the Khitan large script character 日 'day' was read as 捏咿兒 <neir> (as transcribed in the History of the Liao Dynasty), why did Cong, Liu, and Chi (2005: 55) interpret

in the 多蘿里本郎君墓誌銘 epitaph for Lord Duoluoliben as 莫聖哥 <mo.shing.go> rather than 捏咿兒聖哥<neir.shing.go>? I don't know, but I can guess that they might have been influenced by the use of 日 in the KLS spellings for 'snake'


corresponding to

<mu.ɣo.o> (Kane 2009: 118; Liu Fengzhu 1983: <mehai>, Chinggeltei: <mogo>)

in the small script and Written Mongolian moɣai. Did 日 have two readings?

<neir> as a logogram 'sun/day'

<mV> as a phonogram used to write polysyllabic words that had nothing to do with suns or days

If 日 had two readings, what was the reasoning behind the second reading? Was there some other 'sun/day' word in Khitan with the syllable <mV>?

Perhaps I should interpret 日 as <mVɣ(V)> with a second consonant and maybe even a second vowel. The use of the 牛-like KLS graph for <o> in


<p(o).o> ~ <p(o)o>


written as


<p.o> ~ <p.o.o>

in the small script leads me to analyze the KLS spellings of 'snake' as


<muɣ(o).o> ~ <muɣo> ~ <muɣo> ~ <muɣo>

莫 was read as *mak in Middle Chinese and could have been *moɣ (= moh in Kane's transcription) in early Liao Chinese. So was the name


something like Moɣ(o)shinggo?

Could 日 as a KLS phonogram for <mVɣ(V)> have originated as an abbreviation of early Liao Chinese 莫 *moɣ? I cannot think of any other reason for reading 日 with <m>. Turkic words for 'sun/day' have initial k-/g-, not m-. The native Korean word associated with 日 is 날 nal 'day', perhaps ultimately connected to Khitan <neir> and Written Mongolian nara(n) 'sun'. The Parhae word for 'day' could have been something like Korean nal. The Jurchen reading for


was inenggi 'day' and 'sun' was

shun (whose graph is obviously derived from 日).

11.27.3:09: For completeness, I should add that I can't think of any Japonic word for 'sun' or 'day' with initial m-. It is remotely possible that a Japonic language was still spoken in Koguryo and brought to Parhae. However, Vovin has argued against such a late survival on Japonic on the continent and I have yet to see any instance in which Japonic elucidates the structure of the Khitan or Jurchen scripts. And this one instance of what might be potential Koreanic influence on a Jurchen character could just be a coincidence.

(11.27.20:39: I forgot about this other Koreanic-based explanation for a Jurchen character.)

The vowel of the first Khitan small script graph for 'snake'

is uncertain since I can't find it in any Khitan small script transcriptions of Chinese. I presume Kane didn't read it as <mo> in spite of its correspondence to the first syllable of Written Mongolian moɣai because he had already assigned <mo> to

which is the first graph in

<mo.ri> 'horse'

corresponding to Written Mongolian mori(n) 'horse'. *MO THAN ONE READING A 日 DAY?

Both the Khitan large and small scripts have graphs that have been mostly interpreted as logograms for 'day', 'month', and 'year':

Gloss day month year
Khitan ?*neir (transcribed as 捏咿兒 *nieiri; cf. Written Mongolian nara(n) 'sun') ?*sair (transcribed as 賽離 *saili, 賽咿唲*saiiri; cf. Written Mongolian sara(n) 'moon, month') *ai (used to transcribe Liao Chinese *-ai; no Mongolian cognate?)
Khitan large script
Khitan small script

The Khitan large 'day' and 'month' graphs are identical to their Chinese counterparts. The Khitan small 'year' graph may be derived from a mirror image of Chn 年. The other three graphs have nothing to do with Chinese.

The Khitan large and small 'year' graphs have also been used to write 'father'. This implies that 'year' and 'father' were homophones, though some have reconstructed the two words differently (Kane 2009: 33):

Scholar 'year' 'father'
Toyoda 2000 *ai *awi
Aisin Gioro 2004 *ai *aja

Like Kane, I prefer to assume that Khitan characters have only one reading each. However, Cong, Liu, and Chi (2005: 55) interpreted

in the 多蘿里本郎君墓誌銘 epitaph for Lord Duoluoliben as a name 莫聖哥 <mo.shing.go>. Why didn't they read it as 捏咿兒聖哥<neir.shing.go>? Is Liao Chinese 莫聖哥 *moshinggo attested for a Khitan name in any Chinese text? <mo> is not like Written Mongolian edür 'day', and even if there were a similar Mongolian word for 'sun' and/or 'day', one cannot assume a Khitan word was cognate to its Mongolian translation without external evidence: e.g., the Chinese transcriptions of the Khitan words for 'day' and 'month'. Did the Khitan large script character 日 have two readings <neir> and <mo>?

Next: A Serpentine Solution? RIDDLE OF THE ROOFED DOTS

In "Ten-Point Evenings", one of the 'ten-point' Khitan large script characters I mentioned was



It belongs to a set of KLS characters for 'fifty' through 'eighty' which consist of 仒 plus additional strokes. None bear any resemblance to the KLS characters for 'five', 'six', 'seven', 'eight', or 'ten'. (There are no Chinese characters for 'fifty' through 'ninety'.)

Gloss Khitan large script Sinography Gloss Khitan large script Sinography
five fifty 五十
six ~ six ~ 六十
seven seventy 七十
eight eighty 八十
ten hundred

Could 'fifty' through 'eighty' (readings unknown) have shared a common morpheme symbolized by 仒?

Next: *Mo than One Reading a Day?

11.25.1:45: 仒 is a kugyŏl character for the Korean syllable ŏ. I presume it's simplified from the Chinese character於 (pronounced ŏ in Korean). The Unihan database also lists a reading sya. Could 仒 also have been a simplification of an earlier Sino-Korean 舍 sya? For other made-in-Korea characters, see the Wikipedia page on Korean transcription and kugyŏl characters. Wikipedia and zdic also list a Chinese reading bing without any meaning.

Although the KLS graph for 'sixty' resembles 扵, a variant of Chinese 於 (more variants here), I am not sure that 扵 or 於 have anything to do with KLS 仒-graphs because the KLS has other graphs with two dots beneath components with 人-shaped bottoms: e.g.,

uul 'winter' (identical to Chn 冬 'winter')

also used to write the stem of

uul-? 'married'f

with completely un-Chinese second and third KLS characters.

The 仒 in KLS numeral graphs could be a reduction of some 'roofed' sinograph like 舍 or 冬 rather than a direct borrowing from 於 ~ 扵. TEN-POINT EVENINGS

Last night, I wrote,

I don't know of any other KLS [Khitan large script] character with the components 歹 and 十+丶/卞. Perhaps Viacheslav Zaytsev has seen such KLS characters in the long manuscript he is studying.

I just found three KLS characters with what looks like a dotted derivative of Chinese and KLS 十 'ten':

I don't know the readings or meanings of the first two. I guessed that the third is a variant of

<shing> (transcription of Liao Chinese 聖 *shing 'sage, sacred, holy')

with a left-hand element resembling Chn 夕 'evening' instead of Chn 歹 'bad' and Cong, Liu, and Chi (2005: 55) confirmed this by interpreting

in the 多蘿里本郎君墓誌銘 epitaph for Lord Duoluoliben as a name 莫聖哥 <mo.shing.go>.

I still haven't seen any other KLS 歹-characters, but there are at least three other KLS 夕-characters:

The second looks like Chinese 外 'outside', but there is no guarantee that KLS 外 meant 'outside' or was pronounced anything like Liao Chinese *ngwai 'outside'.

11.24.1:57: The first of the KLS characters with 十+丶 has a left-hand component shared with at least 15 other graphs:

This フ+丨element resembles Chn 扌 'hand' which has an exact lookalike in the KLS character


as written in Kane (2009: 177). Both


occur in numeral contexts, so I assume they are variants. Do 扌-variants exist for at least some of the other フ+丨graphs above?

Not all instances of フ+丨may be variants of 扌.

<siang> 'general'

is a variant of

and both are related to Liao Chinese 将 *tsiang. (The Chinese initial *ts- was Khitanized as <s>. Note how the KLS characters have a bottom right component with a flat top unlike Chn 寸 whose vertical stroke intersects its horizontal stroke.)

has a unique (?) left side and may be a variant of

Coming Soon: Sixty Days (in less than sixty days, I hope) BADLY CAPPED RED SAGES

Chinese-like scripts can be like Rorschach ink blots: look at the characters long enough and you can see almost anything you want.

I've long wondered what the reasoning for the Tangut character

2544 2ʃɨẽ 'sage'

was. Currently I am inclined to derive it from the 耳 'ear' at the top left of Chinese 聖 'sage', but earlier this fall, I thought it vaguely resembled the Khitan large script character

transcribing Liao Chinese 聖 *shing 'sage, sacred, holy' (Kane 2009: 181).

This KLS character resembles Chn 歹 'bad' (earlier 'bone fragment') and 十 'ten' plus a dot 丶 (which is close to 卞 'cap'). Could it also have been read as the native Khitan word for 'sacred' written in the small script as


I don't know of any other KLS character with the components 歹 and 十+丶/卞. (11.24.1:59: I do now!) Perhaps Viacheslav Zaytsev has seen such KLS characters in the long manuscript he is studying.

One could subtract strokes from 歹+卞 to create something not far from メ+丨, the cursive form of 2544 (see Grinstead 1972: 300, 302, 303, 304, 305 for examples). But I don't think メ+丨 is derived from 歹+卞. Seeing characters in other characters can border on pareidolia.

The Jurchen graphs for achiburu 'holy' (no Manchu cognate) bear no resemblance to any of the above:


Jin Qizong (1984: 234) derived <achi> from Jin Chinese 赤 *chi 'red', though it looks more like 寺 *sï 'temple' with a 人 roof over it to me. The earliest form

of <achi> doesn't look much like 赤 either.

Jin listed no instances of <achi> without a following <buru>. Nonetheless, he glossed achi as 聖, presumably because -buru is a known derivational suffix. Neither Jin nor I know the derivation of <buru>.

Is <buru> in Jason Glavy's font? I can't find it, so I made a 'frankengraph' out of


<gushin> 'thirty' + <ma>

11.23.4:30: There is another


in the Jurchen script for the first two-thirds of


I wish to thank Viacheslav Zaytsev for Kychanov's 1964 article "К изучению структуры тангутской письменности". On p. 135, Kychanov wrote (my translation),

The comparison of Khitan characters with Tangut characters indicates that the Tangut could have borrowed some graphic elements of the Khitan script which probably in turn go back to sinography. In our opinion, these [elements] are

I will refer to these eleven elements as I-XI. If one were to prove that tangraphy were derived from khitagraphy* (Khitan writing), one would have to find shared innovations that set the two scripts apart from sinography. How many of these elements are shared innovations?

Element I is in the Khitan small script


but only appears as a cursive variant of the element


in Tangut (Kolokolov & Kychanov 1966: 131). Both elements resemble Chn 及 minus the left stroke.

Element II (乚) is in both Khitan and Tangut, but is also found in sinography (e.g., 礼, 札), so it could have been inherited independently from sinography.

I cannot find elements III-V, VII, IX, and X in Khitan. My unfamiliarity with the Khitan large script may be at fault.

The closest Khitan character to element V that I know of is the small script graph for <g>

which is not an exact match for the Tangut element (alphacodes: cua, cix) only appearing in

Element VI looks almost like a Khitan small script logogram

corresponding to Chn 后 'ruler'. It resembles a surrounding radical 'silk' (alphacode diu)

in Tangut: e.g., in

Both resemble Chn 介 'armor; to lie between', so independent derivation cannot be ruled out.

Element VIII resembles an abbreviated variant of the Tangut element

listed in Kolokolov & Kychanov (1966: 128) but I haven't seen it in Khitan.

Element XI is more or less in both scripts:

Khitan small script <dú> and its Tangut near-lookalike

It is similar to Chinese 几 but the 丿on the right could have been a Khitan innovation emulated by the Tangut.

Although I haven't been able to confirm all eleven elements, I cannot rule out khitagraphic influence on tangraphy. The small number of matches above probably reflects my ignorance of khitagraphy. In any case, tangraphy is not structurally like either Khitan script in spite of some degree of graphic similarity.

*11.22.2:36: I previously preferred the term khitanography but I left out the -no- since 'Khitan' appears in small script texts and in borrowings like Китай and Cathay without -n. A similar term for the Jurchen script(s) is Jurchigraphy with a Jurchi- based on Mongolian Jürchid. On the other hand, these new terms are harder to remember than khitanography or jurchenography, and neither of those terms is any shorter than Khitan script or Jurchen script. EIGHT GREAT RICE HORDES

The Khitan small script version of the reign title from my last post looks almost nothing like its large script or Chinese equivalents:

Khitan small script Khitan large script Chinese



*chung 'repeated'



*hi 'splendor'

Almost, because the small script character for 'heaven' (pronunciation unknown) is a derivative of Chinese 天 'heaven' like its large script equivalent. In , the horizontal top stroke is broken in two, resulting in what looks like 八 'eight' over 大 'great' to Chinese eyes. The Khitan small script characters for 'eight' and 'great' not only do not look much like their Chinese equivalents but are actually more complex:


KSS <EIGHT(♂)>' has four or five more strokes (王 added around 八?) and KSS <GREAT> has one more stroke.

The KSS forms above were taken from the eulogy for Emperor Xingzong written in 1055. In the 耶律迪烈 Yelü Dilie epitaph written 37 years later in 1092, the second half of the reign title was written as


with the logogram  <ordu>* (resembling Chn 米 'rice' - why?) in place of <ú.dû> and no <a>. What might this substitution suggest?

1. <ordu> and <ú.dû> had become homophonous by 1092 in the speech of the scribe.

Presumably <r> was lost, though the (later?) Chinese transcription 斡魯朵 *woludo must be based on a dialect with [r].

2. <ordu> and <ú.dû> were synonyms.

Were they unrelated words or dialectal variants?

The various undated KSS spellings of 'ordu' in Kane (2009: 64. 77) suggest possible dialectal variation and/or vowel changes over time:

<ORDU.u> (does the final <u> indicate a long vowel or just clarify the final vowel of <ORDU>?)


<ORDU.ó> (with a surprising final vowel)

<o.ORDU.u> (does the initial <o> indicate a long vowel or just clarify the initial vowel of <ORDU>?)


11.21.1:13: Note that none of these variants begin with an <u>-like vowel.

The diacritics in Kane's transcription system (which I use here with slight modifications: e.g., <ɣ> instead of <h>) may or may not indicate phonetic distinctions (Kane 2009: 26).

3. <ordu> and <ú.dû> were neither homophones nor synonyms but were phonetically close enough for the scribe to substitute the former for the latter.

I think 3 is most likely to be correct, so my gloss of 'ordu-?' for the Khitan large script character is most likely to be incorrect.

Next: Did the Tangut Borrow Graphemes from the Khitan?

*<ordu> is related to Mongolian ordu, the source of horde (with an inexplicable, nonetymological h-).

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2011 Amritavision