08.12.12.3:36: OC *-oj ~ *-waj > MC *-wie ~ *-wɨɑ
The nonvelar-*ɨɑ syllables in Guangyun all have medial *-w- and two of the velar-initial syllables also have medial *-w-. On Wednesday, I proposed that
MC medial *-ɨ- could be an irregular reflex of OC *-r-
but now I think it's part of a bent *o or *wa.
Codas conditioned the MC reflexes of OC nonemphatic *o:
Zero, glottal, or velar coda
1. OC *-o(K) > *-wa(K) > *-wɨa(K) > MC *-wo(K)
When this bending occurred, OC *-s had backed to glottal *-h.
Dental or palatal coda
2a. OC *-on/r/t > *-wan/t > *-wɨan/t > MC *-wɨen/t (> *-wien/t after dental and palatal initials)
*-r merged with *-n
2b. OC *-oj > *-waj > *-wɨaj > *-wɨej > MC *-wɨe
In all of these shifts, *o broke to *wa whose vowel bent upward to *ɨa and either backed and rounded to *o or fronted to *e depending on the following consonant (if any).
MC *-wɨɑ syllables could be an alternate reflex of OC *-oj or *-waj without any *-r-:
2b'. OC *-oj > *-waj > *-wɨaj > *-wɨa > MC *-wɨɑ
Some *-wɨɑ syllables were written with *-aj phonetics. This may suggest that their graphs were devised after the breaking of *-oj to *-waj (merging with orignal *-waj):
|(骨+山+匕)||*khwɨɑ||*qhoj or *qhwaj||山+匕 a distortion of 它?|
|瘸||*gwɨɑ||*grwaj < ?*Nɯ-qrwaj||加||*qæ||*qraj|
|鞾||*xwɨɑ||*Cɯ-χoj or *Cɯ-χwaj||華||*ʁwæ||*wra|
|(月+日+匕)||*ʔwɨɑ||*ʔɯ-qoj or *ʔɯ-qwaj||none; 日+匕 a distortion of 它?|
|(月+戀)||*lwɨɑ||*Cɯ-roj < ?*Cɯ-ror||戀||*lwienh||*pɯ-ronʔ-s|
Although none of the above graphs seems to be very old, I assume they represent words that existed long before they were written.
(骨+山+匕)/(骨+它) OC *qhoj would normally become uvular-initial MC *qhwɑ, but has a velar initial because of deemphasis that spread from the presyllable of the preceding syllable:
MC *ʔwɨɑ khwɨɑ < OC *ʔɯ-qwaj qhwaj
If (骨+山+匕)/(骨+它) and (月+日+匕) have 它 OC *laj as phonetic, perhaps they represented
OC OC *ʔɯ-qlwaj qhlwaj
with *-l- in both syllables.
鞾/靴 'shoe' and its homophones could have many different OC sources. *χ(w)- could be from clusters like *sqh(w)- or *hŋ(w)- if it is not orignal. The use of uvular-initial phonetics suggests a uvular-initial main syllable preceded by a deemphasizing presyllable.
I have no idea why (亻+蛾) MC *ʔwɨɑ has a nasal-initial phonetic 蛾 MC *ɴɑ. An initial uvular nasal *ɴ- is very unusual. (12.13.18:08: The only language in UPSID with a uvular nasal is Japanese which only permits uvular nasals in codas.)
Last night, I reconstructed (月+戀) MC *lwɨɑ with OC *l because I thought that MC *-ɨ- was a trace of OC *-r- and I wanted to avoid reconstructing OC *rr- as the source of MC *l...ɨ-. (MC *l- is normally a reflex of OC *r-.) But now I think MC *-ɨ- is part of a bent *a rather than an irregular reflex of OC *-r- so I can now reconstruct (月+戀) with *r- in its main syllable.
08.12.11.23:59: GLOTTAL-*ɨɑ SYLLABLES IN GUANGYUNalso theoretically shouldn't exist, but they do:
(月+日+匕) MC *ʔwɨɑ
(亻+蛾) MC *ʔwɨɑ
These graphs represent the first halves of obscure disyllabic words:
(月+日+匕)(骨+山+匕) MC *ʔwɨɑ *khwɨɑ ?'bent limb disorder'
(亻+蛾)(亻+恒 minus last stroke) MC *ʔwɨɑ *khɨɑ 'appearance of idiocy'
Those words look like reduplications. Inspired by Sagart's uvular hypothesis, I could reconstruct their earlier forms as *qoj qhoj and *qoj qhaj, with an initial *q- that became glottal stop and a medial *-qh- that became a velar because it was shielded by *qoj-. (Sagart posited presyllables to shield uvulars, not whole syllables.)
My own system requires something like
OC *ʔɯ-qoj qhoj and OC *ʔɯ-qoj qhaj
with a deemphasizing presyllable.
08.12.10.23:59: ALVEOLAR-*ɨɑ SYLLABLES IN GUANGYUN
theoretically shouldn't exist at all. So where do Middle Chinese *tswɨɑ, *tshwɨɑ, and *lwɨɑ come from? All those syllables were written with Old Chinese *o-phonetics:
脞㛗 MC *tswɨɑ < OC ?*rtsoj
侳(食+坐) MC *tshwɨɑ < OC ?*rtshoj
坐 MC *dzwɑʔ < OC *dzojʔ
(月+戀) MC *lwɨɑ < OC ?*ror
戀 MC *lwien < OC *pɯ-ronʔ-s
䜌 MC *lwɑn < OC *ron
In the first two cases, MC medial *-ɨ- could be an irregular reflex of OC *-r-:
MC *ts(h)wɨɑ < *ts(h)wɨɑj < *ts(h)ɨoj < *ts(h)roj < OC *rts(h)oj
Normally, OC *rts(h)oj would become MC *tʂ(h)wæ: e.g.,
髽 MC *tʂwæ < *tʂwæj < *tʂɔj < *tʂroj < *tsroj < OC *rtsoj
But in the third case, medial *-ɨ- could not be an irregular reflex of OC *-r-, unless OC had sequences like *rr- with *r as initial and medial.
I have not found any instances of (月+戀) 'donkey belly' earlier than Yupian (c. 543 AD), but the word was presumably spoken long before its graph was created. I wonder if it was originally *rʌ-loj, with a prefix shielding *-l- from hardening to late OC *d- (see Sagart 1999: 19). The presyllabic vowel was lost, *r-l- metathesized, and medial *-r- irregularly became *-ɨ-:
(月+戀) MC *lwɨɑ < *lwɨɑj < *lɨoj < *lroj < *rloj < OC ?*rʌ-loj
12.11.0:26: Why was (月+戀) 'donkey belly' written with an *-n phonetic 戀? Why wasn't it written with 累 as in 騾 Late OC *lwɑ < *roj (0:44: or *Cʌ-loj?) OC 'mule' (a cognate word)? Was 'donkey belly' pronounced with a final liquid (*-r or *-l) similar to *-n at the time its graph (月+戀) was devised?
12.11.00:41: I found another word (月+羅) MC *læ 'meat of the bottom of a donkey's stomach' which may be cognate to (月+戀) MC *lwɨɑ 'donkey belly'. MC *læ is unusual since MC *æ < Middle OC *rɑ normally does not follow MC *l-. Perhaps (月+羅) was an *-a- variant of (月+戀) OC ?*rʌ-loj:
(月+羅) MC *læ < *lrɑj < *r-lɑj
(羅 OC *raj was originally an *r-phonetic but was pronounced *lɑ by the time 𦣇 was devised prior to Jiyun [1037 AD].)
08.12.9.23:59: VELAR-*ɑ SYLLABLES IN GUANGYUN
are all listed at the end of the section for the rhyme 戈 *-wɑ, even though most have no labial segment. Graphs for two of those syllables (伽𩨷) appear as final spellers for glottal- and alveolar-initial syllables.
|Middle Chinese 1||Middle Chinese 2||fanqie||sinographs|
|*khɑ||*khɨɑ||丘伽||佉呿(去+欠)(亻+恒 minus last stroke)|
(12.10.00:12: All type B fanqie initial spellers are marked in green.)
There is a major problem with my reconstructions in the first column. The nonvelar-initial syllables in red are homophonous with other *-wɑ syllables: e.g., why create a special fanqie 於靴 for (月+日+匕) and (亻+蛾) if they were homophonous with 倭 which was definitely *ʔwɑ?
The reconstructions in the second column avoid this problem. They contain a medial *-ɨ- which distinguishes these syllables from regular 戈 *-wɑ syllables: e.g.,
(月+日+匕) *ʔwɨɑ : 倭 *ʔwɑ
脞 *tshwɨɑ : 遳 *tshwɑ
(侳 *tswɨɑ has no corresponding *tswɑ)
𦣛 *lwɨɑ : 螺 *lwɑ
This medial *-ɨ-
- sounds like *-w-, explaining why *-(w)ɨɑ syllables were listed under the rhyme 戈 *-wɑ
- normally follows velars and hence is a 'buffer' or 'bridge' preventing the direct combination of velars with *-ɑ which normally follows uvulars. Compare:
|Late Old Chinese||*qɑ||*k(ɨ)ɑ||*kɨa|
|Middle Chinese 2||*qɑ||*kɨɑ||*kɨə|
|Guangyun rhyme||歌 *-ɑ||戈 *-wɑ||魚 *-ɨɑ|
08.12.8.23:59: ADDITIONAL GRASS: WHAT'S *GA-ING ON?
The late Old Chinese word 茄 *gɑ 'eggplant' is anomalous because it combines a velar initial *g- with a low back vowel *-ɑ. In LOC, velar initials normally precede high vowels, and low back vowels normally follow uvulars, not velars. See the table from my last post.
I think there are two possible explanations for the unusual initial-rhyme combination of 茄 *gɑ:
1. The word is of foreign origin and hence not subject to the phonological constraints of native Chinese words. It is little different from the syllables 迦 *kɑ and 佉 *khɑ in Indic borrowings.
2. The word is an archaism partly preserving an earlier *ga which did not shift to the expected *gɨa.
The two explanations are not mutually exclusive. What if 'eggplant' were an early foreign borrowing *ga that shifted to *gɑ instead of *gɨa? The trouble is that I don't know of any plausible foreign source for 'eggplant'. Although eggplants are native to India, *gɑ does not look like Sanskrit mahoṭihaa or mahoṭii 'eggplant' and the resemblance to the -ga- of Hindi baiṃgan (cf. Persian بادنجان and Arabic باذنجان with -j-) is coincidental. On the other hand, I cannot think of any plausible Chinese-internal etymology.
To complicate matters, the Proto-Tai word for 'eggplant' is *khɯa A1 in Li Fang-kuei's reconstruction with a voiceless aspirate initial rather than a voiced initial.
Moreover, the graph 茄 consists of the mostly type A (and hence probably uvular-initial) phonetic 加 *qæ(i) 'add' beneath the semantic element 艹 'grass'. Although I know for sure that 迦 *kɑ and 佉 *khɑ represented Indic syllables with velars (ka and kha), it is possible that 茄 once had a uvular.
Perhaps an early form of 'eggplant' was *Nɯ-qhɑ. (I don't know whether the Chinese word for 'eggplant' predated the loss of *-i after *-æ.)
Proto-Tai *khɯa A1 could reflects a Chinese *khɨa < *Nɯ-khɨa < *Nɯ-qhɑ which had lost a high-vowelled presyllable conditioning a uvular to velar shift and medial *-ɨ-.
Vietnamese cà < *ga 'eggplant' could reflect a Chinese *gɑ < *ŋgɑ < *ŋkhɑ <*Nɯ-kh(ɨ)ɑ < *Nɯ-qhɑ with a voiced initial originating from a nasal presyllable-aspirate sequence with irregular loss / lack of *-ɨ-.
Amoy khe 2 could be from *ghe < *ɴɢhæ < *N-qhæ, a variant with a low front vowel. Cf. the other reading of 茄: *qæ 'lotus stalk'.
(Could Proto-Min voiced aspirates be from *N-Ch nasal-aspirate sequences?)
08.12.7.23:17: UVULARS, VELARS, AND VOWEL BENDING
Old Chinese uvulars and velars conditioned vowel bending in opposite directions. I use uvular symbols to represent the back allophones of velars in late OC and MC.
|OC stage 1||OC stage 2||late OC||MC||bending|
|*(Cʌ-)qa, *Cʌ-ka||*qa||*qɑ||*qo||none; can't go any lower|
|*(Cɯ-)ki, *Cɯ-qi||*ki||*ki||*ki||none; can't go any higher|
OC stage 1 forms are listed in order of probability: e.g., a MC *ki is more likely to be from an early *(Cɯ-)ki than a deemphasized early *Cɯ-qi. I'd like as few emphasis-switchers as possible.
Late OC 迦 *kɑ and 佉 *khɑ are in Indic borrowings, so they have no early OC sources.
I'll discuss late OC 茄 *gɑ 'eggplant' in my next post.