Here's how I would determine whether a given velar-initial Middle Chinese syllable had an Old Chinese uvular or velar initial.

Probably uvular

If its graph belongs to a pure/mostly type A xiesheng series

type B members of mostly type A series had nonemphatic presyllables

If there is a borrowed form in other languages (e.g., Hmong and Sui) with a uvular initial

If there is a uvular-initial cognate in a Sino-Tibetan language like Zhongu

but not all ST uvulars are necessarily conservative!

Probably velar

If its graph belongs to a pure/mostly type B xiesheng series

type A members of mostly type B series had emphatic presyllables

A velar Hmong or Sui form is no guarantee of an OC velar because such forms could be borrowings from Chinese dialects which had shifted uvulars to velars.

But what if a xiesheng series is not mostly one type or the other? I hypothesize that uvulars were associated with lower vowels (*e *a *o) and velars were associated with higher vowels (*i *ə *u)*: e.g.,

GSR1 可: mixed *-aj series, so could be uvular

GSR651: 今 mixed *-əm series, so could be velar

This does not mean that 可 could not be velar or that 今 could not be uvular; merely that qhaj is more likely than khaj and kəm is more likely than qəm.

*12.7.1:21: These classes are not arbitrary.

Uvulars were 'antihigh' and condition lowering of following vowels:

启 (pure type A GSR588) OC *qhiʔ > MC *khejʔ

亥 (pure type A GSR937) OC *ɢəʔ > MC *ɣʌjʔ

告 (pure type A GSR1040) OC *quk > MC *kouk

In Manchu, uvulars (q, ɢ, χ) were associated with a, o, ʊ whereas velars (k, g, x) were associated with i, ə, u. ka and ko only existed in Manchu borrowings from Chinese. Similarly, in late Old Chinese, *ka and *kha were only in Sanskrit borrowings because a normally followed uvulars: *qa, *qha, *ɢa. (But note that although special graphs 迦佉 were coined for *ka and *kha, there were no special graphs for *ke or *ko. Did uvulars front to velars before *e and *o while remaining intact before *a?) HMONG AND SUI EVIDENCE FOR OLD CHINESE UVULARS?

In "Loanwords as Evidence for Old Chinese Uvular Initials" (1982), EG Pulleyblank listed a number of Hmong and Sui words with uvulars corresponding to Middle Chinese type A velars. I suspect that these words had Old Chinese uvulars, even if their graphs' phonetic series lack Sagart's diagnostic MC initials for uvulars (*ʔ- and *x-):

sinograph White Hmong (Petchabun) Sui GSR series type MC *ʔ- and *x-? my MC my OC
qe 33 pure A no *kæʔ *qraʔ
nqe Ra ?[ʁa] 38 *x- *kæh *Nʌ-qras
qai qai 876 no *kej *qe
qo 675 mixed *x- *kəp *qəp
qhua 766 mostly A no *khæk *rqhlak
qho (Zhangfang: qhaŋ, Guizhu: qhoŋ) 1174 pure A *khoŋ *qhoŋ
nqai 630 *gɛp *Nrqəp
qhe 541 *khəj *qhə
qa 606 *kam *qam
qăm *kəmh *qəms
nqhi 313 mixed *ʔ- and *x- *khat *Nɯ-qhat
(Zhangfang: qau) 1172 mixed no *koŋ *qoŋ
qat 327 pure A *x- *kɛjh *qhrats

(12.6.8:17: I assume that Hmong n- in 價 and 渴 is a trace of an OC nasal prefix which dropped before it could fuse with the root initial. I reconstruct a nonemphatic, high-vowelled prefix *Nɯ- for 渴 to account for the high vowel in Hmong: *Nɯ-qhat >*Nɯ-qhɨat > ?*Nqhɨ. Is Hmong -i due to Hmong-internal raising of *-a [cf. Tangut 'brightening' of *-a to -ie] or is nqhi simply unrelated to 渴?)

There is no correlation between Sagart's diagnostic MC initials for OC uvulars and uvulars in Hmong or Sui. If all of the above words had OC uvulars, Sagart's hypothesis would require Hmong and Sui to have borrowed unprefixed uvular-initial words that had presyllables in mainstream Chinese: e.g.,

雞 Sagart's OC *qqe > southern OC ?*qqai > Hmong and Sui qai

but Sagart's OC *Cə-qqe > MC *kej

whereas my OC *qe is the source of both the Hmong and Sui loanwords (via southern OC *qai) and MC *kej.

Conversely, Hmong and Sui words with nonuvulars correspond to Middle Chinese type B velars from OC velars or uvulars fronted by nonemphatic presyllables:

sinograph White Hmong (Petchabun) Sui GSR series type MC *ʔ- and *x-? my MC my OC
tɕhi 517 mostly B *x- *khɨjh *khəts
tɕho 1129/1138 mixed *gɨaw *Nɯ-q(h)aw*
tɕai tsi, tʃi 1 yes *gɨaj(h) *Nɯ-qhaj(s)**
tʃu 992 mostly B *x- *kuʔ *kuʔ
tʃi 861 pure A no *kɛʔ *qreʔ
kau 1225 *kɔk *qrok

I assume the palatals are due to Hmong and Sui-internal palatalization rather than borrowing from a Chinese language which had undergone palatalization.

Note that 解 and 角 do not have uvulars in Hmong or Sui even though I reconstruct their pure type A (< *emphatic) series with OC *q-. There is no reason to believe that non-Chinese languages ceased to borrow from Chinese after uvulars fronted to velars, so Sui tʃi 'untie' and Hmong kau 'corner' may be borrowings from varieties of Chinese which had undergone fronting. They are not necessarily evidence for reconstructing 解 and 角 with velar initials in OC.

*12.6.8:32: It's not clear whether the aspiration in Hmong tɕho reflects a Chinese dialect with *kh- < *g- or a Chinese dialect with *gh- < *ŋkh- < *Nɯ-qh-.

**12.6.8:15: I reconstruct an aspirated *kh- because I assume that Siamese ขี่ khii < Proto-Tai *khwi 'ride' is a borrowing of a descendant of the unprefixed root. However, PT khwi could also be a borrowing from a later Chinese dialect which had shifted *g- to *kh- (though I can't think of any other cases of Siamese kh- < *kh- : MC *g-). ROUSE-RAISE

The o-e reduplicative pattern in

tʃɨo R53 1.51 tʃɨe R36 1.35 ?'selection' < tʃɨe 'raise'

and especially the mirror image e-o pattern reminds me of the e-o reduplications in Old Chinese (Sagart 1999: 137): e.g.,

邂逅 *ɢreʔ-ɢroʔ 'carefree and happy'

from a root *√ɢr-ʔ; neither half can be an independent word, unlike Tangut tʃɨe 'raise'

I reconstruct 解 GSR861 as a uvular emphatic series because its Middle Chinese readings are all type A (< *emphatic). If this were a nonemphatic series, I would expect every single reading to contain an emphasizing presyllable *-.

If 解 GSR861 was uvular emphatic, then it is not surprising that 邂 861e has an *o-counterpart 逅 in another uvular emphatic series (后 GSR112; see here).

According to Sagart's uvular paper, 解 GSR861 could be reconstructed as velar rather than uvular since its MC readings lack the initials *ʔ- and *x- implying OC uvulars in his system. However, reconstructing 邂 861e with a velar would result in a reduplication *ggreʔ-ɢɢroʔ with mismatching initials in his reconstruction (whose doubled initials correspond to my underlining/emphasis). So perhaps Sagart would reconstruct GSR861 as uvular in spite of the absence of diagnostic initials.

Other reduplicative forms may provide clues to determining whether a phonetic series is velar or uvular. If one half of the form is most likely velar-initial, then the other half must belong to a velar series. Similiarly, if one half of the form is most likely uvular-initial (e.g., 逅), then the other half must belong to a uvular series.

Next: Counterevidence from Sui. VOTING, TANGUT STYLE

Of course, there were no elections in the Tangut Empire. But if the Tangut needed a word for the concept, they might have recycled

tʃɨo R53 1.51 tʃɨe R36 1.35

which Li Fanwen (1997: 772) translated as 選舉 'selection'.

In traditional China, 選舉 'select-raise' referred to selection of officials through examinations but in modern China, the word means 'election'.

However, Kychanov (2006: 468) has a very different set of translations for tʃɨo tʃɨe:

возносить 'raise'

возвеличивать 'glorify'


舉長 ?'elevate and elongate'

尊長 'superiors (< 'the raised'?)'

In any case, the word is a partial reduplication of the root tʃɨe 'raise'.

Other reduplications with o-e patterns are (from Gong 2003: 612, but in my reconstruction):

no ne 'tranquility; stability; safety; peace'

roʳ reʳ 'surround'

kɔ kɛ 'hate; dislike'

syo sye 'to grind; to whet'

diọ diẹ 'to grieve; grievance'

There is also an e-o reduplicative pattern: e.g.,

gɛɛ gɔɔ 'silly; foolish'

ʃɛɛ ʃɔɔ 'collect' HANDS UP!

Can you guess what

TT0205 tʃɨe R36 1.35

means? Hint: 'top' is on 'top' and 'hand' is on the bottom.

Moreover, can you guess what its homophone

TT0206 tʃɨe R36 1.35

with 'birth/feces' on the right means? It's not as gross as you might expect.

No graphic analysis is available for either tangraph. Neither tangraph has an entry in the Tangraphic Sea, even though they appear in Tangraphic Sea definitions and in Precious Rhymes of the Tangraphic Sea. Was their omission accidental?

Scroll over the blank space below to see their meanings according to Li Fanwen 1997:

TT0205: 舉 'raise', 秉 'grasp', 開 'open'

(12.2.0:25: One might expect 'lower', 'release', and/or 'close' to be written as 'hand' over 'down', but no such tangraph exists.)

TT0206: 後 'behind, back' ('raise' is phonetic, but how is 'birth/feces' relevant?) ZERO GRADE IN 序 'SCHOOL'

is impossible, and not just because no school system I know of has a 'zero' grade.

If zero grade ever existed in the history of Chinese, it must have predated writing because there are

- no syllabic sonorant series in sinography

- no syllabic sonorant or nasal vowel (*-ã) rhyme categories in Old Chinese

Since 'schools' usually entail literacy, I doubt* there was a word like *sɯ-lŋ meaning 'school' in pre-Old Chinese. If *sɯ-lŋ existed, it must have meant something else.

Here's what I think happened:

There was a Proto-Sino-Tibetan root √*lŋ 'raise':

Written Tibetan lang 'rise'

Written Burmese langʔ 'high raised frame; stage' (found in Matisoff 2003: 304)

(cf. OC 臺 *lə 'stage' < zero-grade *lŋ?; I'll present an explanation for the unexpected schwa instead of *-a later)

Old Chinese

*Cɯ-laŋ 'lift, raise'

*Cɯ-la < pre-OC *lŋ 'lift'

*kɯ-la-ʔ < pre-OC *lŋ 'lift; raise'

*laʔ lift' then came to mean 'give (originally to superiors?)':

與 OC *Cɯ-laʔ (= ?*kɯ-laʔ) 'give'

予 OC *Cɯ-laʔ (= ?*Tɯ-laʔ) 'give' (with dental or alveolar-initial prefix?)

cf. Japanese ageru 'raise' which has also come to mean 'give'

Zero grade *-ŋ had shifted to *-a by the time sinography was created.

The OC words for 'school' (where sinography was used, if not taught?)



were coined after the zero grade shift by adding affixes to the OC root doublet *laŋ ~ *la 'lift; give'. OC speakers no longer knew that *la was a zero grade of *laŋ.

Without knowing what kind of school a 庠/序 was, I can only guess how to bridge the semantic gap between 'lift; give' and 'school':

'lift' > 'place of elevating (knowledge)' > 'school'?

'lift' > 'stage' > 'stage from which teacher instructs' > 'school'?

''give' > 'place where knowledge is given' > 'school'?

ADDENDUM: OC had another academic sl-word: 習 *s-ləp 'practice; exercise'. Since the word was nonemphatic, its *s- may have once been nonemphatic *sɯ-. 肄** *Cɯ-ləp-s (= *sɯ-ləp-s?) 'practice; exercise' may also have had the *s-presyllable. I could try to link their shared root *ləp to the 'school' words by tracing it back to *lŋ-p. However, there is no known suffix *-p, and I can't think of any plausible bridge between 'lift/give' and 'practice':

'lift' > 'do something to elevate oneself' > 'practice'?

'give' > 'work with what has been given' > 'practice'?

*For similar reasons, I am certain there was no word for 'brush' in Proto-Sino-Tibetan, though Matisoff (2003: 504) implied one by citing OC 筆 'brush' (my *plut) and Written Tibetan pir 'id.' as evidence for PST *u ~ *i variation. WT pir is a loanword from northwestern Middle Chinese *pɨr.

**肄 *Cɯ-ləp-s has a *-t phonetic 聿 *lut.

The graph 肄 must postdate the shift of OC *-ps to *-ts.

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