When one hears that the Khitan small script was created around 925 and was abolished around 1192, one might look at the set of 369 characters at the back of Kane (2009) or the set of 448 characters in Sun et al. (2010) and assume they all rose and fell at the same time. But that is unlikely. A few may even be modern copyists' errors: e.g.,

121 for 063

in the handwritten transcription of the 仁懿 inscription which has been reburied and lost (so there is currently no way to verify the original). The remaining real characters may have come and gone at different points over three centuries: e.g.,


114/352 <ɨi>

may have become obsolete after *-ɨi and *-i had merged into *-i in Chinese. I wrote "after" rather than "when" because 114/352 may have already been an archaism by the time it was used in the 宣懿 and 道宗 inscriptions of 1101. Was it used in any other inscriptions? My guess is that it may have been obsolete when the 郎君 inscription was composed in 1134 after the fall of the Khitan Empire, as Chinese 期 *khɨi (later *khi) was transcribed as 339 <i> in

340-339 <x.i>.

The earliest known Khitan small script inscription is 耶律宗教 from 1053, over a century after the script was created circa 925. It is possible that there are small script characters that only appeared in the earliest, as yet undiscovered texts. And could new characters have been created after 925?

It would be interesting to keep track of the attestations of each Khitan small script character in dated texts. DID KHITAN SMALL SCRIPT CHARACTER 114/352 REPRESENT A DIPHTHONG?

Kane (2009: 48) interpreted Khitan small script character 114

as a long vowel [iː]. He transliterated it and its variant 352

as <î> with a circumflex reminiscent of the circumflex some use to write Japanese vowel length. (That practice seems to be in decline now that support for vowels with macrons is widespread.)

I think he concluded that 114/352 was a long vowel because it appears in transcriptions of Chinese 騎 which was also transcribed with <i.i>:


334-114 <g.î> (宣懿 2.28) :  334-352 <g.î> (道宗 2.24): 334-339-339 <g.i.i> (蕭仲恭 19.34)

However, there are only three instances of 114/352 in the entire corpus in Qidan xiaozi yanjiu (1985). Besides the two instances of Chinese transcriptions above there is also one native (or at least non-Chinese) word:

352-235-359 <î.ri.?> '?' (道宗 7-21)

Was [iː] really only in one Chinese word and one native word? I think 114/352 represented something a bit more exotic than [iː].

Sino-Korean was borrowed from an 8th century dialect of eastern Middle Chinese that was either ancestral or closely related to Liao Chinese. In premodern Sino-Korean, 騎 was 긔 kɯi, implying eastern Middle Chinese *kɨi. Could the diphthong of *kɨi have survived in Liao Chinese? If so, then 114/352 might have stood for [ɨi] or [ɯi], and 339-339 <i.i> could have represented a more Khitanized pronunciation [iː] of an un-Khitan [ɨi] or [ɯi].

But if [ɨi] or [ɯi] were un-Khitan, why would 352 be in a non-Chinese word 352-235-359? Maybe that instance of 352 was an error for some other character with the top element 火~大. Unfortunately, there are no other words of the type x-235-359 in Qidan xiaozi yanjiu, so I cannot determine what that other character might have been. KHITAN SMALL SCRIPT VOWEL CHARACTER FREQUENCY IN QIDAN XIAOZI YANJIU

Kane (2009) identified nineteen vowel characters in the Khitan small script. I only looked at ten core characters in my previous post. Here is a list of all nineteen in order of frequency in the texts indexed in Qidan xiaozi yanjiu (1985). Colors indicate whether a character is only in native (or at least non-Chinese) Khitan words (red), only in Chinese loanwords (blue), or in both (green).

Rank Character Kane's (2009) transliteration Qidan xiaozi yanjiu number Qidan xiaozi yanjiu frequency Notes
1 a 189 618 Khitan and Chinese
2 i 339 520
3 u 131 490
4 o 186 257
5 û 372 256
6 ii 080 248 Khitan only
7 ó 090 194 11.7.23:27: Mostly Khitan; Kane 2009: 79 mentions an instance of Chinese 部 *pu transcribed in Khitan as <pu.ó>, but this is anomalous
8 e 348 130 Khitan only; 109 is a variant of 348
9 109 76
10 ú 245 75 Khitan and Chinese
11 ô 252 36 Khitan only; variant of 186?
12 y 082 24 Chinese only; medial [ɥ]
13 ï 353 15 Chinese only; 113 is a variant of 353
14 ü 226 10
15 ï 113 8
16 ô 253 5 Khitan only; = 252; variant of 186?
17 î 352 2 Khitan and Chinese
18 î 114 1 Chinese only; variant of 352 miswritten as 339 <i> in Kane's index
19 i'i 338 1 Chinese only; = 懿 / 339-339 <i.i>

I am not surprised that the top four are <a i u o>.

However, I was surprised to see <û>, <ii>, and <ó> above <e> which was probably /ə/, the higher/yin counterpart of lower/yang /a/. Then again, if the phoneme /ə/ is an inherent vowel of CV characters, the frequency of the character <e> should be low. Vowel phoneme frequency is distinct from vowel character frequency.

Nonlow achromatic vowels share the top element 火~大 in common:


348 ~ 109 <e> /ə/


353 ~ 113 <ï> /ɨ/ or /ɯ/


352 ~ 114 <î> which may have been /ɨi/ or /ɯi/

I am now wondering if 080 <ii> was not simply long /iː/ which might have been the value of 339-339 <i.i> and 338 <i'i> in Chinese loans. 080 might have stood for an i-like vowel or diphthong absent from Chinese: e.g., <ɪ> given that it followed 323 which might have had a uvular initial in

323-080-222 <qi.ii.ń> '奚 Xi' (許王 52.4)

Aisin Gioro (2011) read 080 <ii> as <əl> which not only results in an odd reading of the name of the Xi (something like <tə.əl.in> in her reconstruction which sounds nothing like Middle Chinese 奚 *ɣej 'Xi'), but also is unlike its fellow converb suffixes which all end in <i> (Kane 2009: 149-151). Perhaps Aisin Gioro regards 080 as the yin counterpart of the yang converb suffix

098 <al>.

<u>, <ú>, and <û> were all used to transcribe Liao Chinese *u, so I can't figure out how they were different.

<o> and <ô> may be the same vowel since they alternate in the same words (Aisin Gioro 2004: 11 cited in Kane 2009: 65). I doubt that these graphic alternations correspond to phonetic alternations due to ablaut, since ablaut is not an 'Altaic' areal trait. (Then again, neither is grammatical gender, an aspect of Khitan that remains largely uninvestigated.)

Liu Fengzhu et al. (2009) interpreted 090 <ó> as <ʊ> - the back counterpart of the <ɪ> that I proposed for 080. AN <II>-SOLATED EXAMPLE?

The Khitan small script spelling

323-080-222 <qi.ii.ń>

for '奚 Xi' (許王 52.4) could also be transliterated <qi.ii.in>. (See my last post for other interpretations of 322.) Those transliterations brought the following questions to mind:

1. Did Khitan have phonemic vowel length?

2. If it did, how did it correlate with what appear to be sequences of identical vowels in the Khitan small script? Long vowels need not be consistently written. They were not distinguished from short vowels in early Attic, and in later Greek which had special letters η and ω for long lower mid vowels and digraphs ει and ου for long upper mid vowels, "the remaining vowel letters α, ι and υ continued to be ambiguous between long and short phonemes." Estonian today has no written distinction between long and overlong vowels.

(Incidentally, Estonian has ɤV-diphthongs like those I reconstruct for Grade II in my current reconstructions of Tangut and Middle Chinese.)

3. How many degrees of vowel length did Khitan have? Did <qi.ii.in> had an overlong vowel like Estonian, or was its vowel simply long [iː]?

4. Was 080 <ii> disyllabic [iʔi] or even [iji]? <ii> is one of a set of converb suffixes (Kane 2009: 149): <ai>, <ei>, <i>, <oi>, <ui>. Were the suffixes other than <i> pronounced as diphthongs or disyllables or simply as [i]: e.g., was

186-107 <o.oi> 'become-then' = 'became ... and then ...' (context in Kane 2009: 150)

pronounced [oj] or [oji] or [oːj] or [oʔoj] or ...?

5. Why is <ii> the only known double/long vowel symbol in Khitan - the "<ii>-solated example" of my title?

Kane (2009: 29) identified a large number of symbols for back vowels (and front rounded vowels?). This table is based on his table 1.32 with the addition of <ii>. It does not include variants.

189 <a>

348 <e>

339 <i>

186 <o>

131 <u>
090 <ó>

245 <ú>

252 <ô>

372 <û>
080 <ii>

6. Could one or more of the lower frequency <o> and/or <u>-type symbols have stood for long vowels? In other words, should <ó>, <ú>, <ô>, and/or <û> be moved to the bottom row (assuming <ii> was a long vowel [iː])?

7. Why are there no other <a> and <e>-type symbols? Were <a> and <e> ambiguous in terms of length? Could long [aː] and [əː] only be specified in writing with graph sequences like <Ca.a>, <e.eC>, etc.? Was, for instance,

053-051-011 <qa.ɣa.an> 'qaghan(-GEN?)' = '(of a?) qaghan'

read as [qɑʁɑːn] with a long vowel? Or was <an> simply a way of writing [n] before a stem ending in [ɑ]?

8. Was the distribution of long vowels in Khitan unbalanced? In Japanese, aa and ee are rare in native words, ii is usually at the ends of adjectives where it is from -iki and -isi, etc. Did ii get its own small script character because it was the most common long vowel in Khitan?

9. Could some Khitan small script CV symbols really be symbols for syllables with long vowels?

10. Did Liao Chinese have vowel length or at least what the Khitan perceived as vowel length? Many Khitan small script spellings of Chinese words have seemingly redundant vowel marking: e.g.,

225-303 <bi.ing> for 兵 *piŋ (instead of *<b.ing>)

Some Chinese words were spelled with one or two vowels: e.g.,

311-131 <b.u> or 311-372-131 <b.û.u> for 部 *pu

See Kane (2009: 244-251) for more examples. THE QI TO THAT BODY?

I was puzzled by the Khitan large script characters


for qi 'that' almost a year ago. But now I think I've figured out their origin. Tonight I learned that the Chinese character 厥 'his, her, its, their' has a variant 𨈐 dating back to Yupian (c. 543). (TLS lists 43 more variants from 異體字字典 and seven more from 漢語大字典: 𠪏𣅲𣐍橛橜瘚𦪘.) Khitan qi may have been like Japanese sono which can be translated as both a third person possessive and 'that'.

I do not know how the shapes of 厥 and 𨈐 can be related. It may be more accurate to speak of unrelated characters for the same word as 'equivalents' instead of 'variants'.

I also do not know the origin of the Khitan small script character 323

for qi. Nor do I understand why 'that' is

323-151 <qi.ɣ>

with a final consonant in 蕭仲恭. The latter cannot be plural: e.g., both <qi ai> and <qi.ɣ ai> mean 'that [specific] yar' in the examples in Kane (2009: 121). (How I wish he included citations for all his examples!) Kane (2009: 73) reported that <qi.ɣ> "is also found elsewhere", though it is not in the texts that I have on hand.

I briefly thought 'that' might have been qiɣ, and <qi.ɣ> was really <qiɣ.ɣ> with a redundant 151 <ɣ>, but the Khitan small script spelling

323-080-222 <qi.ii.ń>

for '奚 Xi' (許王 52.4) makes more sense if 323 represented an open syllable ending in ii.

The reading <qi> is from Kane (2009) and is close to Liao Chinese 奚 *xi. Kane did not explain why he chose q- instead of x- (cf. Aisin Gioro's 2004 reading hi). Perhaps he thought 奚 *xi stood for a foreign name *qi. However,奚 was read as *ɣej centuries earlier when it first appeared as part of the name 庫莫奚 *khoʰmak ɣej Kumo Xi in the Book of Wei. So perhaps 'that' in Khitan was closer to ɣe or ɣi*, and the Khitan name for the Xi might have been something like ɣei(n) or ɣii(n).

The final <ń> 許王 52.4 in may be a genitive suffix; <qi.ii.ń> appears before <qa.ɣa.an go.e.en>, and the phrase

323-080-222 053-051-011 319-348-140 <qi.ii.ń qa.ɣa.an go.e.en>

may mean 'of the goe of the qaghan of the Xi'. The meaning of goe is unknown.

Aisin Gioro (2009) changed her reading of 323 from hi to (cf. Mongolian and Manchu tere 'that'). However, 323-080-222 would then be <tə.ii.ń> which does not sound like 奚. Did she regard 323-080-222 as something other than 'Xi' or did she think the Khitan had an exonym <tə.ii.ń> for the Xi?

323 looks like Chinese 口 'mouth'. Was the Khitan word for 'mouth' (and all other body parts except for 'head'?) qi or ɣi or the like (which would be very different from Mongolian aman 'mouth')? I doubt it, as no other Khitan small script characters are obvious pictographs.

*There was no syllable *ɣi in the Chinese of that period, so 奚 *ɣej could have been an approximation of a foreign *ɣe or *ɣi.

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2014 Amritavision