I will conclude this series by looking at a word ending in 084 (Kane 2009's <ra>) that has puzzled me for the past few days:

261-362-084 <l.iau.ra> (or <l.iau.ar>?)

I might have found it when looking through 仁懿 for something else (<x.s.ge.du> at 24.5?) Qidan xiaozi yanjiu (1985: 507) defined it as 遼 'Liao'. However,

261-362 <l.iau>

does not occur in the texts I have on hand, and

261-362-131 <l.iau.u>

is a transcription of the Jin Chinese reading of 略 in 郎君 1.9.

According to Kane (2009: 163) Liu Fengzhu (2006) thought

036-177 <xu.ra> (Kane 2009: 162), <hu.ulji> (Aisin Gioro 2012)

was the Khitan word for 'Liao'. If it was, did the Khitan also use the Chinese word 遼 *liau? And if <l.iau> is a borrowing from Chinese, what is <ra>? Could Chinese 遼 *liau be a borrowing from a non-Chinese name like *leura for the Liao River? (The earliest attestation of 遼 as a geographical term that I can find is in the Shiji [Records of the Grand Historian] from the first century BC when the graph was roughly read as *leu; *e broke to *ia centuries later.)

Here are all the occurrences of <liau.ra> that I could find. Qidan xiaozi yanjiu page numbers are supplied for further context.

1. 仁懿 19.20 (p. 507):

162-366-172 / 261-362-084 / 334-097-140

<c.ul.uɣ l.iau.ra g.úr.en>

'? Liao-ra country-GEN'

<c.ul.uɣ> is a hapax legomenon.

Liao here is presumably the name of a country. One might think <ra> is a genitive ending, but the expected genitive ending after <l.iau> is <un>, a variant of <en> after <u>.

2. 道宗 17.3 (p. 521)

368 / 279-354-140 / 261-362-084 / 133-162-112-140

<FOUR po.os.en l.iau.ra m.c.ge.en>

'four (non-masc.) time-?-GEN Liao-ra ?-GEN'

Qidan xiaozi yanjiu glossed <po.os> as 'time'. <po> can mean 'time' by itself. The reading <os> for 354 is from Aisin Gioro (2012: 17). <os> could be a plural suffix, so <FOUR po.os.en> may mean 'of the four times'. In any case, <FOUR po.os.en> modifies <l.iau.ra>.

Is <ra> an adjectival suffix? Does <l.iau.ra> modify the following word?

<m.c.ge> is a noun; it also occurs twice with the dative-locative suffix <de> (蕭仲恭 11.48, 16.7).

3. 蕭仲恭 31.46 (p. 609)

162-131-361-341 / 261-362-084 / 178-041

<c.u.én.er l.iau.ra ku.su> (or <ku.us>?)

'?-PL-ACC/INST or ?-PERF♂ Liao-ra ?'

<c.u.én> also occurs by itself in 蕭仲恭 (32.11) I do not know whether <er> is the masculine perfective verb suffix or the accusative-instrumental noun suffix. If it is the former, <l.iau.ra> might be the subject of the next sentence.

Is <l.iau.ra> an adjective modifying the following word?

<ku.su> or <ku.us> (the reading of 041 is from Aisin Gioro (2012: 8) is a noun. If it is <ku.us>, could it be the <s>-plural of <ku> 'man'? Do the last two words here mean 'Liao people'?

4. 蕭仲恭 34.27 (p. 610)

178-041 / 261-362-084-341 / 340-019-222 / 378  /097-244-341

<ku.su (ku.us?) l.iau.ra.er x.iu.ń REPEAT úr.s.er>

'(men?) Liao-ra-ACC/INST ?-GEN PL?-GEN-PL? ?-PL-ACC/INST or ?-PERF♂'

Could <l.iau.ra> be an adjective modifying <ku.us> 'men'? What is the significance of the different positions of the adjective? (See "The Characters of Ministers" for more instances of the same elements in different orders.)

Here <l.iau.ra.er> and the following two or three blocks may be arguments of a verb.

<x.iu> is attested by itself (道宗 32.3). If it is a noun, <ń> may be a genitive plural suffix (Kane 2009: 135). The reduplication may also indicate plurality.

The noun possessed may be <úr.s.er>. Unfortunately I cannot find the stem <úr.s> with other suffixes. Could it be an <s>-plural of a noun <úr> (仁懿 31.2, 道宗 26.10)?

It is also possible that <úr.s.er> is a masculine perfective of a verb <úr.s>. If so, then <x.iu.ń REPEAT> possesses something that is implied: cf. the use of English genitives without following nouns. <RA>-CONSTRUCTION 5: ARE RHOTIC INITIALS RIGHT?

In part 4, I was skeptical of 084

being read as <ra> in isolation: i.e., with an initial <r>. At least nine other Khitan small script characters have readings reconstructed with initial <r> by various researchers according to Aisin Gioro 2012:

AS: Aisin Gioro

K: Kane (2009)

P: Pre-2002 readings listed by Aisin Gioro as "previous research" without specifying any authors

Q: Qidan xiaozi yanjiu (1985)

L: Liu Fengzhu et al. (2009) (which I have not yet seen)

WJ: Wu and Janhunen (2010) (which I have not yet seen)

Number Khitan Initial <r>-readings Other readings Notes
038 L: <ɽu> AS 2004: <ha> I have no idea what the basis for these readings is. Not used to transcribe Chinese.
069 P: <ri>
K, WJ: <rí>
Q, AS 1999, L: <li>
AS 2004: <ali>
Corresponds to 里 *li in Chinese transcription but not used to transcribe Chinese. Kane (2009: 43) concluded that it had an un-Chinese *r-.
137 AS 2004, WJ: <ir> P: <rə>
L: <ku>, <tsh>
WJ: <gy>
Corresponds to Late Middle Chinese *-ir in <ś.iú m.137>, a loan from LMC 樞密 *ɕumir 'state secret; head of the privy council'.
235 P, Q, AS 1999, WJ: <ri>
L <r>
AS 2004, K, L: <ir> Corresponds to 夷離 *yili in Chinese transcription.
Not used to transcribe Chinese.
236 P, WJ: <ru>
L: <rɘ>
AS 2003, 2004, K, WJ: <ur> See "How <du> You Pronounce 'Final Fire'?".
Not used to transcribe Chinese.
330 L: <ɽĭ> P, Q: <ʐ>
AS 2002: <j>
AS 2004: <ʒ>
K: <ź>
WJ: <zh>
Used to transcribe the Liao Chinese 日 'sun' initial that did not become r-like until long after fall of the Liao. Hence an <r>-reading is anachronistic.
333 P: <ru> P: <lu>
K: <qatun>
May mean 'queen' (Turkic qatun). May correspond to 禄 *lu in a Chinese transcription of a Tangut name. Not used to transcribe Chinese.
341 WJ: <ri> P: <wei>
AS 2004: <ər>
K: <er>
L: <li>
WJ: <il>, <er>
Perfective suffix and accusative/instrumental suffix. Not used to transcribe Chinese.
P: <ruŋ>
AS 2005: <tʃʊŋ>
K: <źuŋ>
WJ: <zhung>
<r>/<z>-type initials  and <uŋ>-type finals are anachronisms.
[ɦyuŋ] in 13th century Hphags-pa Chinese (Coblin 2006: 111).
Khitan transcription of Liao Chinese 榮 *ywiŋ which rhymed with the *-iŋ words 應聲明清塋城銘 (Kane 2009: 249).
I proposed <ywing>.
The graph is clearly derived from Chinese 榮.
Perhaps a special character for this syllable was created because 榮 'glory' is a common word with an uncommon initial cluster *yw- that would be difficult to write with a character sequence.

Most of these nine characters and 084 can be classified into two groups:

A. Non-Chinese transcription characters: 038, 069, 084, 235, 236, 333, 341.

B. Chinese transcription characters: 330, 367.

137 is in a category of its own (A'): it is not a transcription character, but it does represent the rhyme of a pre-Liao Chinese loanword.

If Khitan had a constraint against initial r- (and there is no guarantee that it did), class A characters would not have r- in initial position. Perhaps <rV> characters were read as <Vr> in initial position: e.g.,

<m.ri> 'horse' (cf. Mongolian morin) and

<ir.g.en> 'tribal leader' (transcribed in Chinese as 夷離堇 *yilikin)

In any case, class B characters never had initial r- because northeastern Chinese did not develop that initial until much later. Although Liao Chinese phonology was remarkably similar to that of modern standard Mandarin, the two were far from identical. <RA>-CONSTRUCTION 4: BONES ALONE

At the end of part 4, I mentioned that

According to Aisin Gioro (2012: 9), Liu Fengzhu et al. (2009) took yet another approach to [Khitan small script character] 084, reading it as

An l-intital reading is attractive because 084

can appear in isolation:


<023.ia 084 k.ɣo.ge.e> (興宗 10.22)


<s.oi.de 084 FOUR po.j> (宣懿 25.20)

If Khitan is like other Altaic languages, it should not have native words with initial r-. And there was no potential source of r-initial loanwords in Khitan (except for Xiongnu). Hence it is unlikely that 084 was read as <ra> (Kane 2009 and Wu and Janhunen 2010) in that context. A reading with another initial such as Liu et al.'s <lü> or a zero initial such as Aisin Gioro's (2012) <ær> is more likely.

If Liu et al. are correct, 084 could be a transcription for Liao Chinese *lü (e.g., 呂) or *lüʔ (e.g., 律), though such a word is doubtful in the second context (宣懿 25.20). (The first context in 興宗 is incomprehensible to me; 084 is flanked by hapax legomena.)

呂 and 律 (the second half of 耶律, the transcription of the Khitan imperial clan name) both happen to be words for pitch pipes with osseous cognates:

呂 Old Chinese *kɯ-raʔ 'spine*, pitch pipe' : Written Tibetan gra in nya gra 'fish bones'

律 Old Chinese *rut 'pitch pipe' : Written Tibetan rus 'bone'

Both terms coexist in Jingpho n-rut n-ra 'ribs, skeleton'. Sagart (2014: 181) hypothesized

that after the language that was to become Chinese individualized out of Sino-Tibetan, the Sino-Tibetan terms #rus and #gra (I am using the Written Tibetan forms preceded by the symbol '#' marking a pre-reconstruction) still existed, referring to different kinds of bones or bone structures, as well as to the tubes made out of them: #rus for longer bones or tubes, and #gra for shorter ones; later on, #rus lost its semantic connections to bone (perhaps it lost them to the new term for ‘bone’: 骨 *kˤut). As a result, the Chinese form 律  *[r]ut, its reflex, could refer to  bamboo pitch-pipes.

However, Sagart (2014: 181) also noted that

There seems to be no material or textual evidence at all linking early Chinese pitch-pipes with bone.

In any case, it is a coincidence that 呂 and 律 happened to become near-homophones in Liao Chinese. I would expect them to be transcribed as

261-289 <l.iú>

in the Khitan small script by analogy with other *-ü syllables transcribed with 289 <iú>. But no such block is listed in Qidan xiaozi yanjiu. Is that because Liao Chinese *lü(ʔ) did not appear in the texts or because that syllable was transcribed in some other manner (e.g., as 084)?

*5.23.1:07: Is there any evidence for 呂 meaning 'spine' before this gloss from Shuowen (121 AD)?


'呂 is the spine bone.'

According to Karlgren (1957: 39), the meaning 'spine' is attested in Zhuangzi (c. 300 BC), though Sagart (2014: 179) was

grateful to an anonymous reviewer for pointing out that the characteris not attested in the meaning 'spine' until the Shuōwén Jiězì: earlier 'spine' is written by means of the homophone 旅 *[r]aʔ or its graphic variant 膂 [with 月 'flesh' added to the bottom]. The main meaning ofin pre-Hàn times is music-related.

Schuessler's (1987: 397) Dictionary of Early Zhou Chinese defined 呂 as 'backbone', though both of the examples from Shijing (11th-7th c. BC) in that entry contain the character 旅 rather than 呂. Schuessler defined 旅 as 'line up in a row' and 律 as 'row'. Could these meanings involving 'row' be descended from 'bone': 'bone' > 'skeleton' = 'arrangement of bones' > 'arrangment' = 'row'? <RA>-CONSTRUCTION 3: IMPERIAL VOWELS

In parts 1 and 2, I forgot to address the issue of the mismatch between the vowels of the name of the Khitan imperial clan

<ei.ra.u.ud> (Kane 2009)

<y.ær.u.(u)d> (pieced together from Aisin Gioro 2012)

<y.ar.u.du> or <y.ar.u.ud> (this site; with and without 'law of alternation')

and its Liao Chinese transcription 耶律 *Yelü(ʔ).

I am not sure what the vowel of the northeast Chinese reading of 耶 was at the time the transcription was devised: i.e., the early-to-mid 10th century AD. I do, however, think I know what that vowel was in the preceding and following periods:

Sinograph 8th century 11th century 13th century 21th century
*yæ *ye *ye
*ʔæ *ʔya *ya
*ŋæ *ŋya *ya

Tones are omitted except in the modern readings. Tone classes are known but premodern tonal characteristics are unknown.

The 8th century reconstructions are based on Sino-Korean readings ending in -a. Northeastern front *-æ and back *-ɑ both correspond to Sino-Korean central ㅏ -a.

The 11th century reconstructions are based on Khitan small script transcriptions. By the 11th century, northeastern *-æ had

- raised to *-e after palatals

- broken to *-ya after back consonants (*k-, *kʰ-, *x-, *ŋ-, *ʔ-)

- backed to *-a elsewhere

The exact timing of the first part of this shift could be determined by studying rhyming and seeing when *palatal-æ syllables no longer rhymed with other *-æ syllables. Unfortunately, rhyming cannot date breaking, since *kæ and *pæ woud still rhyme after they became *kya and *pa.

The 13th century reconstructions are based on Coblin's (2006: 175) interpretation of Hphags-pa Chinese. I have rewritten Coblin's as *e since Hphags-pa Chinese had no phonemic distinction between */e/ and */ɛ/. My 11th century *ye was probably [jɛ] in IPA: i.e., identical to Hphags-pa Chinese *ye and modern standard Mandarin ye.

Until now I have been transcribing Khitan small script character 084

as <ar> or <ra> with an  <a> like Kane's (2009) transcription <ra>.

If the imperial clan name was Yaru(u)d, its first syllable would fit a northeastern Chinese reading anywhere on the spectrum between *yæ and *ye since there was no northeastern Chinese syllable *ya. Characters like 亞 *ʔya might have rejected as possible transcriptions of Ya- because of their un-Khitan initial clusters. (Avoidance of initial clusters is an Altaic areal trait.)

On the other hand, if 084 was <ar>, how was it distinct from 123

which Kane (2009), Wu and Janhunen (2010)*, Aisin Gioro (2012), and I all agree was <ar>?

And if 084 was <er>, how was it distinct from 341

which Kane (2009), Wu and Janhunen (2010), and I agree was <er>?

Perhaps 084 had a vowel distinct from those of 123 and 341 as in Aisin Gioro's solution (or two slight variants that I present below):


123 084 341
A <ar> <ær> <er>
B (Aisin Gioro 2012) <ər>
C <er>

There is another possible way to distinguish these three which I will mention below.

As for the mismatch between Khitan (u)u and the ü of Liao Chinese 律 *lüʔ, perhaps 律 'regulation' was chosen as a transcription because of its positive semantics and in spite of its vowel. A very wild possibility occurred to me tonight: 084 might have ended in a palatalized liquid unlike either 123 or 341, so

was read as YVrʸu(u)d or YVlʸu(u)d. A palatal liquid followed by u would be close to *lü(ʔ). Although I have been following the majority so far in assuming the liquid was r rather than l, I know of no certain evidence for reconstructing r instead of l.

Grinstead (1972: 13) identified 野利榮仁 Yeli Rongren as Iri, "the inventor [of the Tangut script] in the Tangut Ballad", and regarded Yeli as a Ye, but without seeing the original Tangut spelling of Iri, I am uncertain about this equation.

According to Aisin Gioro (2012: 9), Liu Fengzhu et al. (2009) took yet another approach to 084, reading it as (presumably equating it with most of Liao Chinese 律 *lüʔ rather than with the rhyme of 耶 *yæ ~ *ye and the initial of 律 *lüʔ).

*I have not yet seen Wu and Janhunen's book and am relying on Aisin Gioro's (2012) table incorporating their readings. <RA>-CONSTRUCTION 2: A SINIFIED SINGULAR?

In part 1, I asked why the final 火 <ud> or <du> of the name of the Khitan imperial clan

<ei.ra.u.ud> (Kane 2009)

<y.ær.u.(u)d> (pieced together from Aisin Gioro 2012)

<y.ar.u.du> or <y.ar.u.ud> (this site; with and without 'law of alternation')

corresponds to zero in its Liao Chinese transcription 耶律 *Yelü. (As I'll explain shortly, that Chinese reconstruction may be wrong.)

I used to assume that transcription was much older: i.e., that it was devised at a time when those characters were pronounced *yæ lüt or *yæ lür (cf. Sino-Korean 야률 Yaryul which may be based on 8th century northeastern Chinese). However, that cannot be the case if the name was only adopted in the 930s. I don't know of any evidence for a final *-r in northeastern Chinese as late as the 10th century. The earliest Khitan small script transcriptions of Liao Chinese date from the 11th century and show no trace of *-r. I conclude that *-r was lost between the 8th and 11th centuries. I have not yet been able to study "the oldest known Khitan inscription of significant length": the memorial for 耶律延寧 Yelü Yanning in the large script. Does it have transcriptions of Chinese with final <r>?

Now I wonder if Yaru(u)d contained the Khitan plural suffix -d (cf. Mongolian -d 'id.'). I just noticed that Kane (2009: 141) had the same idea. I had forgotten about that passage. However, this next proposal might be original: what if 耶律 *Yelü is a transcription of a singular ending in a vowel: e.g., Yaru(u)?

Perhaps not if 耶律 was *Yelüʔ with a final glottal stop in Liao Chinese corresponding to Khitan -d. However, it is not clear whether a final glottal stop was only in poetry or not (Kane 2009: 252), and there are only sporadic possible attempts to transcribe a final glottal stop in the Khitan small script (Kane 2009: 254). So I am not sure whether 耶律 was chosen with a glottal stop in mind or not. Did the Khitan as a rule disregard final glottal stops when writing Khitan names in Chinese? It would be interesting to see if Chinese transcriptions of Khitan names in Chinese-language Liao texts contained possible glottal stops corresponding to zero in Khitan small script spellings. (At least some Chinese transcriptions of Khitan names in non-Liao texts might have been created by Chinese monolinguals and would be less reliable than transcriptions by bilinguals.)

(律 was ryurʔ with an -r and a glottal stop in the artificial idealized Sino-Korean of the 15th century. This glottal stop has left traces in modern Sino-Korean, but I don't know how far back it goes. Is it a genuine relic of northeastern Tang Chinese before *-r-loss, or was a glottal stop added due to the influence of post-Tang speech?)

In any case, the question of whether Yaru(u)d is a plural remains. Could

(道宗 19.23, 宣懿 19.19, 蕭仲恭 28.14)

<ei.ie.ra.u> (Kane 2009)

<y.yæ.ær.u> (pieced together from Aisin Gioro 2012)

<y.ie.ar.u> (this site)

be the singular in spite of an <ie> absent from what might be the plural? Unfortunately, I don't understand enough about the context of its attestations to determine whether <y.ie.ar.u> even refers to the imperial clan or not.

The vowel combination ea would be unusual in Old Turkic, Mongolian, and Manchu*, and it and many other small script spellings raise questions about how Khitan vowel harmony worked and when it was applicable. There are so many oddities in Khitan vowel sequences from an Altaic perspective that I suspect Khitan was like modern Korean and only had traces of an earlier vowel harmony system in verb endings** that had broken down in other domains. What would have caused the breakdown? Native sound changes (e.g., my proposed *ia > e monophthongization in closed syllables)? Harmony violations would be understandable in Chinese loanwords, but not in loanwords from Altaic languages (though of course both would be subject to native sound changes).

*Manchu is not related to Khitan, but it belongs to the same Altaic linguistic area.

5.21.1:08: Maybe Khitan e was neutral as in Finnish (which is of course outside the Altaic area) or 'low' like a as in my reconstruction of earlier Korean vowel harmony. However, the overview of Khitan verb morphology in Kane (2009: 145-155) indicates that e- and a-suffixes belonged in different classes, though that still does not rule out their coexistence in nouns.

**5.21.0:59: Vowel harmony is now breaking down even in Korean verbs (Lee and Ramsey 2011: 96): e.g., 받아 pad-a 'receive, and ...' may be pronounced as [padə] by analogy with other verbs with the ending [ə]. <RA>-CONSTRUCTION 1: YELÜ AND SALADING

I almost included the name of the Khitan imperial clan

020-084-131-344 (道宗 2.27, etc.)

in my entry on 344 火 'final fire', but I decided it was worthy of its own discussion.

The readings of 020 and 131 can be determined from their use in Khitan transcriptions of Liao Chinese (Kane 2009: 259 and 246-247):


<y> and <u> match the initial and the second vowel of 耶律 *Yelü, the Liao Chinese transcription of the name.

Neither 084 nor 344 appear in Khitan transcriptions of Liao Chinese to the best of my knowledge. So they must represent phonemes or phoneme sequences absent from Liao Chinese.

At first one might think that 084 was el, but that is hard to reconcile with its use in the name of the father of the founder of the Khitan Empire*,


which corresponds to 撒剌汀 *salatiŋ in Liao Chinese transcription (Kane 2009: 129). One might guess that 084 was <al>, but that sound value is already associated with


If 084 is not a variant of 098, perhaps it was <ar>, and the Khitan name was Sarɣa(a)diń. (Liao Chinese had no syllable *diń; *tiŋ was the closest available substitute.) Kane (2009: 45) read 084 as <ra>.

If I am right about 火 344 'final fire' and the 'law of alternation', then the imperial family name

was <y.ar.u.du>. Kane (2009: 76) read 344 as <ud> and similarly Aisin Gioro (2012: 16) read 344 as <(u)d>. But there is nothing corresponding to <du> or <ud> in 耶律 *Yelü. (No, wait, maybe there was!) Was the final syllable or consonant simply ignored in Chinese transcription, or is there another explanation?

Next: A Sinified Singular?

* Unfortunately I don't know the Khitan name of the Taizu Emperor. The Chinese transcription of that name is 阿保機 *Apouki, suggesting a Khitan original like

*<a.b.o.u.g.i> *Abougi.

(Khitan voiced obstruents were transcribed with Liao Chinese voiceless unaspirated stops: e.g., Sarɣa(a)d as 撒剌汀 *salat above.)

The only Khitan term for him that I know of is

<HEAVEN xong di> 'heaven emperor' (耶律迪烈 5).

075-037 <xong di> 'emperor' is a loan from Liao Chinese 皇帝 *xoŋ ti 'id.'

Although 075 does not alternate with


a block which does not exist to the best of my knowledge, it is a transcription of the early 12th century Jin Chinese surname 黄 *xoŋ in 郎君 5.30, a homophone of 皇 *xoŋ 'emperor'.

I have not yet seen 037 alternate with


a block which does exist, but I assume that 037 was <di> since it's most likely that Liao Chinese 皇帝 *xoŋ ti was borrowed as a disyllabic whole. (It is remotely possible that 075-037 mixes <xong> with some non-Liao Chinese, possibly even native Khitan reading for 037.)

I do not know why Kane (2009: 43) read 075 as <hoŋ> (= <ɣong> in this site's transcription) instead of <xong>. Kane (2009: 261) did not reconstruct in Liao Chinese.

皇帝 was *ɣwangtej with in early Tang. I would expect an early Tang loanword for 'emperor' in Khitan to be *ɣuangdei, not ɣongdi. Moreover, I would not expect 075 *<ɣuang> to be a transcription for Jin Chinese 黄 *xoŋ. HOW <DU> YOU PRONOUNCE 'FINAL FIRE'?

In "Philial Phonetics", I wrote about how Khitan small script character 150 <ja> "often combines with <a>-type characters". But here's a case in which it is followed by 344 <ud> (resembling Chinese 火 'fire'):

<ja.ud> (蕭仲恭 13.51)

If this represented jaud, why wasn't that syllable written with 015 <jau> and 254 <d> as


And would the syllable jaud even exist if my hypothesis about Khitan short diphthongs is correct? (*jaud should have become juad.)

I have suspected that the Khitan small script had what I'll call the 'law of alternation':

- Some characters could represent both CV and VC sequences (cf. Old Turkic 𐰸 qo/qu/oq/oq and 𐰜 ko/ku/ök/ük)

- Those ambiguous characters were read as


after a vowel(-final) character: ...V.CV ...

before a consonant(-initial) character: ... CV.C ...


after a consonant(-final) character: ...C.VC ...

before a vowel(-initial) character: VC.V ...

That law assumes Khitan words had a simple structure of alternating consonants and vowels (hence the 'law of alternation'). I suspect that was not the case, and that the reality was much messier: the language was full of consonant and vowel clusters while the script did not consistently indicate vowels.

If 344 was an ambiguous character and the 'law of alternation' applied to it,

would be read as <ja.du>, not <ja.ud>. The reading <du> for 344 is similar to the reading <do> proposed by Ji Shi (as described in Kane 2009: 94) for 105, a variant of 火 344 resembling Chinese 大 'large' in

105-236 'east'

Ji Shi: <do.ru> (cf. Mongolian doruna 'east')

Kane: <ud.ur>

Aisin Gioro (2012: 10, 13): <do.ur>

This site: <du.ru>?

I doubt that 'east' had an initial <u> because I don't know of any other instances of a Khitan initial vowel corresponding to zero in Mongolian. (Is there any language in which apheresis is a regular sound change? Such a language would have no words with initial vowels.)

Here are other examples of what I'll call 'final fire' (i.e., 344 in final position; I have not yet found 150 in final position, so there may be no 'last large'):

After consonants:

<al.ud> (興宗 23.24), <m.uɣ.ud> (仁懿 11.28*)

After vowels:

<xua.du> (興宗 3.20), <c.as.a.du> (仁懿 8.19**), <s.od.na.du> (興宗 3.28), <x.s.ge.du> (仁懿 24.5, 29.15, 31.15, 蕭令公 22.5)

I'm saving a particularly interesting example for my next entry.

*5.19.0:21: <m.uɣ.ud> is not on p. 250 of Qidan xiaozi yanjiu's index of Khitan words, but it is in the handwritten transcription of  仁懿 on p. 505.

**5.19.0:41: The entry for this word on p. 268 of Qidan xiaozi yanjiu states that it appears three times in the corpus, but only one appearance is specified.

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