THE SCEPTER SERIES: (LABIO)VELAR, (LABIO)UVULAR, AND/OR (LABIO)GLOTTAL?
In my last entry I mentioned GSR (Grammata Serica Recensa) phonetic series 0879 圭 which Baxter and Sagart currently reconstruct with a mix of velar and uvular initials. What if it was originally a purely labiouvular series?
|Sinographs||Schuessler 2009||Baxter and Sagart 2014||This site||Middle Chinese|
|鮭||(*kw^-), *gw^-||(kʷˤ-, *m-kʷˤ- *)||*Cqʷ-, *m-Cqʷ-||(*kw-), *ɣw-|
|絓||*kw^-, *gw^-||*[k]ʷˤ-||*Cqʷ-, *N-Cqʷ-||*kw-, *ɣw-|
|哇||*gw^-, (*ʔw^-)||*Nqʷˤ-, (*qʷˤ-)||*NCqʷ-, *qʷ-||*ɣw-, (*ʔw-)|
|鼃||*ʔw^-, *w^-||*qʷˤ-, *m-qʷˤ-||*qʷ-, *m-qʷ-||*ʔw-, *ɣw-|
|恚||*ʔw^-, (*w^-)||(*qʷˤ-, *Nqʷˤ-)||*qʷ-, *N-qʷ-||*ʔw-, (*ɣw-)|
|窐||*w^-, *kw^-||(*qʷˤ-, *C.qʷˤ-)||*N-qʷ-, *Cqʷ-||*ɣw-, *kw-|
For simplicity I have ignored Old Chinese medial *-r-.
Schuessler's circumflex on vowels corresponds to Baxter and Sagart's *ˁ.
Parenthesized initials in Schuessler's column are what I would expect in his reconstruction to be the sources of Mandarin initials that are not regular reflexes of Middle Chinese initials: e.g., the w- of Mandarin 哇 wa should be from *ʔw^-, not *gw^-.
I have also placed *kw^- in parentheses since it is my attempt to fill in the missing reconstruction corresponding to 鮭 Middle Chinese *kwej and Mandarin gui.
Parenthesized initials in Baxter and Sagart's column are my attempts to fill in the gaps in their PDF according to my understanding of their system.
Unlike Baxter and Sagart, I do not reconstruct phonemic pharyngealization for uvulars.
Parenthesized Middle Chinese initials are not in the phonological tradition but are the likely sources of unexpected Mandarin initials.
Nearly all Middle Chinese readings of 0879 contain *-w-, so I presume the labiality is original. The exceptions may have lost their labiality to dissimilate from a lost labial preinitial:
*Pqʷ- > *Pq- > *k-
Middle Chinese *ŋ- of 厓崖涯睚 may be a fusion of *m- and a uvular:
*mQʷ- > *mQ- > *ŋ-
(*Qʷ- could be *qʰʷ- or *ɢʷ-; *mqʷ- [= Baxter and Sagart's *mqˁʷ-] became Middle Chinese *ɣ-, not *ŋ-; see section 4.4.2 of Baxter and Sagart 2014.)
Series which have an occasional *-w- in Middle Chinese might have secondary labiality: e.g., *pk- > *kʷ- (cf. *PC- > Cw- in Tangut). If such a change occurred, it might have predated the reductions covered in section 4.4.4 of Baxter and Sagart (2014): e.g., *pk- > *p- (not *kʷ-). Perhaps there was a chain shift:
|Reduction type||Phase 0||Phase 1||Phase 2||Phase 3||Phase 4|
In that scenario, various words with presyllables were reduced at different speeds. Early reducers developed initial *kʷ- (phase 2), middle reducers developed *p- (phase 3), and late reducers may have also developed *p-. No presyllables or preinitials remain in phase 4.
Another possibility is that *pk- > *kʷ- and *pk- > *p- occurred in different dialects. In any case, I expect reduction to have occurred in different ways at different rates in Old Chinese dialects over time.
*1. 1.18.5:54: I think the voicing of the initial of 鮭 'demon' may be from the *m-animal prefix. See Baxter and Sagart (2014: 53-54).
*2. 1.18.6:01: The high vowel of *Cɯ-qʷʰ- was preceded by a nonemphatic (i.e., nonpharyngealized) consonant; that presyllable conditioned the loss of all pharyngealization in the following syllable and the bending of *e to partly match the height of *ɯ:
跬 *Cɯ-qʷʰeʔ [Cɯ-qʷʰˤɛˤʔˤ]? > *Cɯ-kʷʰeʔ > *Cɯ-kʷʰieʔ > Middle Chinese *khwieˀ
*3. 1.18.6:15: I don't know for sure if late reducers underwent the same change as middle reducers. Could the *p-k- remaining in phase 3 have become, say, *ph- in phase 4? (I reconstruct *K- to account for word families with aspiration alternations in Tangut.)
22.214.171.124:54: WHAT DO MOUNTAINS PRODUCE?
Last month I asked if Tangut and Chinese shared a word for 'mountain'. They might if Baxter and Sagart's 2014 reconstruction is correct:
4871 1ngyr1 : 山 Middle Chinese *ʂɤen < Old Chinese *s-ŋrar (B&S 2014: 148)
cf. my *ksan, Schuessler's *sran, Zhengzhang's *sreːn, Starostin's *sraːn, etc.
At that time I didn't have Baxter and Sagart's book, so I didn't know the logic behind *s-ŋrar. But now I have a copy (hence the page number above) and have seen that they think 山 'mountain' belongs to a 'slope, nearly vertical side' word family including
巘 Middle Chinese *ŋɨanˀ ~ *ŋɨenˀ < B&S OC *ŋ(r)ar(ʔ) 'hill'
厓崖涯 Middle Chinese *ŋɤe < B&S OC *ŋˁrar 'river bank, limit'
cf. my *ŋre, Schuessler's *ŋrê, Zhengzhang's *ŋreː, etc.
顏 Middle Chinese *ŋɤan < B&S OC *C.ŋˁrar 'face, forehead'
They quote the 釋名 Shi ming 'Explaining Names' (200 AD; translation theirs):
' 'Mountain' is 'river bank'; it produces things.'
I'm not sure what 'it produces things' means. Does that phrase refer to things growing on a mountain?
I am still skeptical of *s-ŋrar for four reasons:
1. 山 transcribed the [(k)san] of Alexandria in 烏弋山離 from the Records of the Grand Historian (c. 100 BC), three centuries before Shi ming. Obviously *s-ŋrar was not the reading underlying that spelling. Did the Shi ming dialect preserve a *-ŋ- in 山 that was lost in the RGH dialect?
2. Is 涯 a phonetic gloss for 山 in Shi ming? Bodman (1954: 111) regarded
產 Middle Chinese *ʂɤenˀ < Old Chinese *s-ŋrarʔ (B&S 2014: 148)
in that passage as the phonetic gloss of 山.
3. 厓涯崖 belong to GSR (Grammata Serica Recensa) phonetic series 0879 圭 for open syllables, so I would not expect them to end in *-r.
4. Moreover, phonetic series 0879 mostly has stop initials, suggesting that the nasal of 厓涯崖 is from an earlier *NK-cluster. I know of no reason to reconstruct such clusters in 巘 'hill' and 顏 'forehead' which may share a root-initial *ŋ-.
1.17.5:07: At least I finally understand why Baxter and Sagart think the root initial of 山 was nonpharyngealized. They point out on p. 395 that 經典釋文 Jingdian shi wen 'Explaining the Text of the Classics and the Canon' (late 6th c. AD) has another Middle Chinese reading *ʂɨen whose vocalism indicates a nonpharyngealized initial. Jingdian shi wen lists various older readings for *ʂ-words indicating earlier nonpharyngealized initials corresponding to later readings that have what appear to be regular reflexes of pharyngealized initials. Given that trend, it is likely that 山 once had a nonpharyngealized initial, though it is also possible that 山 had coexisting variants with different initials. I would prefer to be agnostic and write the pharyngealization of the initial of 山 in parentheses: *s-ŋ(ˁ)rar.
126.96.36.199:24: SO NEAR, YET SO FAR: A REAP-PRÈS-SAIL
For a long time I assumed that Vietnamese gần [ɣən] 'near' was borrowed from 近 Late Old Chinese (LOC) *gɨənh or Early Middle Chinese (EMC) *gənʰ 'to approach'. The 'softened' Vietnamese initial [ɣ] is not a direct retention of Chinese *g; it reflects a lost presyllable: [ɣən] < *CV-gən. Baxter and Sagart (2014: 118) reconstructed in Old Chinese (OC) as
*s-: 'valence increaser' (here, v.i. > v.t.?; see B&S 2014: 56; how does this differ from *-s?)
*N-: 'intransitivizer'? (implying that *kərʔ was transitive?; see B&S 2014: 54)
also in 近 *N-kərʔ 'near'
Words written with the phonetic 斤 *kər rhymed in the *-r category in the Classic of Poetry; see Starostin (1989: 580-581)
I would expect the root to occur elsewhere as a transitive verb given that 近 'near' has the intransitivizer prefix *N-. But as far as I know, there is no root *kərʔ in Old Chinese words other than 近 'near' / 'approach'.
I would rather reconstruct the root with a simple initial *g-. The complex sequence *s-N-k- seems to be motivated by a problematic comparison with Vietic (see below).
On the other hand, the earliest Sino-Japanese reading kon < *kən for 近 may reflect a dialect with an unprefixed root initial *k-.
*-s: 'outwardly directed action' (here, v.i. > v.t.; see B&S 2014: 59)
> LOC *-h and EMC *-ʰ
I suppose the double marking *s- ... -s of a transitive verb is like English embolden; *s-N-kərʔ-s is to 'en-near-en'.
Baxter and Sagart (2014: 118) compare OC *s-N-kərʔ-s to Rục tŋkɛɲ whose initial tŋk- matches their *s-N-k-. (Rục has no s-preinitials, so t- is the closest possible match of OC *s-.) The finals, however, do not match even if one keeps in mind that Old Chinese *-r had shifted to *-n (but not palatal *-ɲ!) at the time of borrowing. Old Chinese did not have any palatal codas, yet a palatal coda is reconstructible in Proto-Vietic *t-kəɲ (without *-ŋ-; is the velar nasal secondary in that variety of Rục?*) and even in Muong forms such as Hoa Binh kʰəɲ¹.
I conclude that Proto-Vietic *t-kəɲ and Old Chinese 近 *gərʔ(-s) (or *N-kərʔ(-s) if Sino-Japanese k- preserves a root-initial *k-) are unrelated lookalikes unless one can explain why Late OC/EOC *-n was borrowed as Proto-Vietic *-ɲ. Are there other examples of Chinese *-n corresponding to Proto-Vietic *-ɲ?
1.15.23:40: The semantics of Vietic 'near' and Old Chinese *s-N-kərʔ-s 'to approach' do not quite match. I cannot derive the Proto-Vietic form from Old Chinese *N-kərʔ 'near' because the Vietic forms do not reflect an LOC final glottal stop or EMC glottalization, whereas the Vietnamese huyền tone in early loanwords can correspond to Chinese voiced initials followed by reflexes of *-s (B&S 2014: 382).
1.15.23:48: Then again, is it a coincidence that *t-kəɲ is apparently unique to Vietic within Mon-Khmer and similar to a Chinese word? Could it be a borrowing unlike the native word Proto-Vietic *s-ɗəː 'near' which was inherited from Proto-Mon-Khmer *t₂ɗəh (Shorto 2006)? (The absence of *-h in Proto-Vietic is irregular.)
*Other Rục forms at SEAlang lack the velar nasal: təkiɲ¹, ckìːɲ ~ ckɨ̀ːɲ.
188.8.131.52:14: AN EXPENSIVE RANGE REVISITED: GSR 0540
On Monday I proposed that Old Chinese (OC) phonetic series GSR (Grammata Serica Recensa) 540 貴 had uvular initials. That might explain why
貴 Middle Chinese *kujʰ < OC *Cɯ-quj-s 'precious, expensive' (Baxter and Sagart: *kuj-s)
could be phonetic in
隤 Middle Chinese *dwəj < OC *N-rˁuj < *Nʌ-ruj 'exhausted' (Baxter and Sagart: *N-rˁuj)
See Baxter and Sagart (2014: 122) for MC *d- < OC *N-rˁ-
and its homophones 穨 'bald' and 僓 'natural, easy, gentle'. 僓 has another reading:
Middle Chinese *xwɛjʰ < OC *r̥ˁuj-s < *Nʌ-r̥uj-s (Baxter and Sagart: *qʰrˁuj-s?)
See Baxter and Sagart (2014: 116) for MC *x- < OC *r̥ˁ-
I will refer to the two readings of 僓 as 僓a and 僓b.
If OC emphatic *rˁ was uvular *[ʀ] or *[ʁ], it would make sense to write *rˁ-words with uvular phonetics like 貴 for *Quj-syllables. Its voiceless counterpart *r̥ˁ could have been *[ʀ̥] which later became *[χ] in western dialects.
The emphasis of *rˁ and *r̥ˁ was conditioned by a low-vowel presyllable at some point before those words were written with a uvular phonetic:
*Nʌ-r- > *Nˁʌ-r- > *Nˁʌˁ-r- > *Nˁʌˁ-rˁ-
*Nʌ-r̥- > *Nˁʌ-r̥- > *Nˁʌˁ-r̥- > *Nˁʌˁ-r̥ˁ-
Pharyngealized allophones of consonants in the vicinity of low vowels became phonemic when those low vowels were lost:
隤穨僓a *Nʌ-ruj /Nʌruj/ [Nˁʌˁʀˁʊˁjˁ] > *r̥ˁuj /r̥ˁuj/ [ʀ̥ˁʊˁjˁ]
僓b *Nʌ-r̥uj-s /Nʌr̥ujs/ [Nˁʌˁʀ̥ˁʊˁjˁsˁ] > *r̥ˁuj-s /r̥ˁujs/ [ʀ̥ˁʊˁjˁsˁ]
The presyllable of 隤穨僓a was reduced to *N-:
*Nʌ-rˁ- > *N-rˁ- > *N-lˁ- > *lˁ- > *d-
or *N-lˁ- > *nd- > *d-?
The presyllable of 僓b was lost before it could reduce to *N-.
I could also dispense with presyllabically condiitoned emphasis and simply reconstruct primary uvular *ʀ and *ʀ̥:
But do those two added phonemes illuminate anything else about Old Chinese, or are they just ad hoc devices that simplify these reconstructions?
隤 'exhausted' is cognate to 儽 Middle Chinese (MC) *lwəjʰ 'exhausted' which could be from
*Cʌ-ruj-s or *ʀuj-s
If the 'exhausted' word family originally had a uvular initial, I would expect its phonetic series to be purely or at least predominantly uvular. However, series 577 畾 in Schuessler (2009: 294) is about evenly split between words that could be reconstructed with *ʀ- (11) and words that couldn't (9), and it contains no words that could be reconstructed with voiceless *ʀ̥-. I would rather reconstruct 577 畾 (and by extension 隤) with original *r- that became uvular if preceded by a low vowel or if the only vowel in its syllable belonged to the low class (e.g., *o but not *u):
畾 OC *ruj > MC *lwi ~ OC *Cʌ-ruj > *ʀuj > *lwəj 'raised path between fields'
累 (phonetic 畾 abbreviated to 田 'field')
OC *ruj > MC *lwi 'to bind'
OC *Cɯ-rojʔ > MC *lwieˀ 'to accumulate'
OC *Cɯ-roj-s > MC *lwieʰ 'to implicate'
OC *rojʔ > MC *lwaˀ 'naked'
Regardless of whether Old Chinese ever had primary uvular rhotics, I think some sort of uvular rhotic better accounts for the unusual range of the phonetic 貴 than previous solutions:
Karlgren (1957: 145): The reading 僓a arose "through confusion with 544a [i.e., 隤].
but if 隤 is not in the phonetic series of 貴, what is the semantic function of 貴 'expensive' in 隤 'exhausted'?
Schuessler (2009: 294) regarded 貴 as a partial (i.e., rhyme-only phonetic) in 僓a.
Zhengzhang Shangfang reconstructed 貴 as *kluds and 隤 as *l'uːl.
If 貴 had a medial *-l-, it must have disappeared by the time 貴霜 'Kushan' was written in the Book of Han (111 AD).
(1.14.23:23: A medial *-l- would rule out Written Tibetan gus-po 'expensive' as a cognate.)
The *-d and *-l in the codas do not match (the suffix *-s is not a problem), and this mismatch is unnecessary since there is no evidence for *-d (or *-t) in the 貴-series.
184.108.40.206:52: EAGLES SWALLOWING THE CENTER
Baxter and Sagart (2014: 101) reconstructed the Old Chinese phonetic series 0718 央 and 0370 因 with initial glottal stops. That brought to mind two etymologies whose sources I can't remember:
- Siamese กลาง klaːŋ A1 < Proto-Tai *klaːŋ A 'middle' < Chinese 央 'center' (Schuessler 2007: 585 called this a "traditional association" without citing a specific author)
- Siamese กลืน klɯːn A1 'to swallow' < Chinese 咽 'gullet, to swallow'*
There is also a widespread (K)lAŋ-type word for 'eagle' in Mon-Khmer that has been related to Old Chinese 鷹 'eagle' in phonetic series 0890 (Schuessler 2007: 574; also cf. Downer's Proto-Hmong-Mien *klâŋ² - how did Ratliff reconstruct that word?).
We know for certain that 央咽鷹 all had initial glottal stop in Middle Chinese (MC). I once thought the external data above might have pointed to an earlier *q(ɯ-)l- in Old Chinese:
*ɯ later conditioned vowel raising in 央 and 鷹.
Without a preceding *ɯ, *i lowered to *ei and then *e after *q-.
Now I am skeptical for the following reasons:
1. The phonetic series of 央咽鷹 only have glottal stop initials in MC; they have no MC initials that unambiguously point to uvulars in the Baxter-Sagart system.
2. The phonetic series of 央 "lack[s] word-family contacts with velars or uvulars" (Baxter and Sagart 2014: 101).
3. Baxter and Sagart reconstructed 烟 ~ 煙 'smoke' as *[q]ˁi[n], possibly because it might share a root √*q-[n] with 熏 *qʰu[n] 'to smoke'. If it does, then 因 might be a uvular series. However, 煙 also belongs to a pure MC glottal stop series, and I can't think of any other examples of *i ~ *u ablaut.
咽 'to swallow' is cognate to 嚥 'to swallow' which belongs to yet another pure MC glottal stop series (GSR 0243). (Coincidentally, the verb 嚥 'to swallow' is homophonous with the noun 燕 'swallow' as in English!)
Shan oddly has ʔɯn A1 'to swallow' with a glottal stop matching Chinese rather than *k- which regularly corresponds to Siamese kl-.
4. Baxter and Sagart reconstructed the series of 鷹 with initial *[q](r)-, possibly because of the k-words for 'eagle', though that series only has glottal stop initials in MC. I am unaware of any velars or uvular word-family contacts for that series.
5. I know of no Chinese-internal evidence for medial *-l- in the phonetic series of 央咽鷹嚥.
6. Pittayaporn (2009) reconstructed *q- and *kl- but not *ql- in Proto-Tai. Did Proto-Tai speakers hear Chinese *ql- (assuming that cluster is correct!) and approximate it as *kl-?
(1:14.4:01: Pittayaporn reconstructed *qr- and *kr-, so I would expect *ql- in addition to *kl-. Perhaps *ql- could be reconstructed on the basis of cognate sets not in Pittayaporn's dissertation. If Proto-Tai had *ql-, then Chinese *ql- should have been borrowed as *ql- and not as the *kl- of 'middle' [and 'to swallow'?]. I doubt that Siamese kl- could be from *ql- because *q(r)- became Siamese aspirated kʰ-. I predict that *ql- would have also become a Siamese aspirate kʰ- or kʰl-.)
7. Unfortunately, Pittayaporn (2009) did not reconstruct a Proto-Tai word for 'to swallow'. The initial cluster of Wuming klwaŋ A1 may point to Proto-Tai *klw- with a *-w- absent from Chinese, but it may not be related, as I would expect its rhyme to be -ɯn (cf. Wuming xɯn A2 'night' and ʔɯn B1'other' corresponding to Siamese คืน khɯːn A2 and อื่น ʔɯːn B1).
8. Siamese ɯː (and similar vowels in other Tai languages) and Old Chinese *i in 咽 'to swallow' cannot be reconciled, though the former would match Old Chinese *ə if it fronted to *e between coronals in 嚥:
*qlən-s > *qlen-s
However, that is highly improbable since 燕 already belonged to the *-en (not *-ən!) rhyme class in the Classic of Poetry, long before Proto-Tai borrowed from Chinese.
One could claim that Proto-Tai borrowed from an archaic dialect of Chinese preserving *ə, but that wouldn't work either because the fronting of *ə occurred long after the composition of the contents of the Classic of Poetry. The *e of 燕 (and most likely 嚥) is primary, not secondary.
The only way that 嚥 could have a schwa is if GSR 0243 燕 was a mixed *e ~ *ə series like GSR 0227 員 which is sui generis in Schuessler (2009).
In conclusion, I think these are cases of vague similarity among monosyllables rather than truly related words.
*1.14.3:21: Old Chinese 咽 'to swallow' has a *-s suffix absent from 咽 'gullet'. That suffix should correspond to Tai tone B1, but all Tai forms in Hudak (2008: 159) have tone A1 with the exception of Western Nung which has C2. If the Tai word is from Chinese, it must be based on the raw root 'gullet' (which could have become a verb 'to swallow' through zero derivation in the source dialect).
220.127.116.11:36: AN EXPENSIVE RANGE: GSR 0540
As far as I know, in Semitic, uvulars and velars are not mixed in orthography or in morphology. 'Emphatic' and 'nonemphatic' consonants* are also not mixed. Yet Baxter and Sagart (2014)'s reconstruction of Old Chinese (OC) contains such mixtures: e.g.,
忌 *m-k(r)ək-s ~ 誡 *kˁrək-s 'to warn'
禦 *m-qʰ(r)aʔ ~ 戶 *m-qˁaʔ 'to stop'
Here's how I would handle those cases:
phonetic series GSR (Grammata Serica Recensa) 540:貴 *kuj-s 'precious, expensive'
靧 *qʰˁuj-s 'to wash the face' (with 面 'face' on the left)
遺 *[ɢ](r)uj 'to leave, reject' (with 辶 'walking' on the left)
忌 *m-k(r)ək-s ~ 誡 *Cʌ-krək-s 'to warn'
The emphasis of 誡 is secondary and due to harmony with the low presyllabic vowel:*Cʌ-k- > *Cˁʌ-k- > *Cˁʌ-kˁ- = *Cˁʌ-q- > *q-
*Cʌ- may have been *mʌ-, a longer version of the prefix of 忌 *m-k(r)ək-s.
禦 *mɯ-qʰ(r)aʔ ~ 戶 *m-qaʔ 'to stop'
Original *q- normally conditions emphasis.
The high presyllabic vowel blocked emphasis and conditioned the fronting of *qʰ-:*mɯ-qʰ- > *mɯ-kʰ- > *mkʰ- > *ŋkʰ- > *ŋ-
I am not sure that *m-qʰ- became *ŋ-, as I don't know of any firm evidence for aspiration**.
Here is another solution: 禦 *mɯ-ɢ(r)aʔ ~ 戶 *ɢaʔ 'to stop'
I don't know of any voiceless-initial members of the 'stop' family, so I reconstructed the root with a voiced initial *ɢ-.
Once again, the high presyllabic vowel blocked emphasis and conditioned the fronting of a uvular:*mɯ-ɢ- > *mɯ-g- > *mg- > *ŋg- > *ŋ-
See Baxter and Sagart (2014: 132) for the evidence for the shift of *mɢˁ- (= my *mɢ-) to *ŋ-.
One problem with this solution is that 戶 *ɢaʔ no longer has the voiceless initial that would be optimal for a phonetic in 所) *s-qʰ<r>aʔ (= my *sɯ-qʰ<r>aʔ) 'place'. But voiced-initial characters can be phonetics in characters with voiceless onsets: e.g., GSR 0027:
爲 *ɢʷ(r)aj 'to do'
媯 *C.qʷ(r)aj 'a name'
譌 *[m]-qʷʰˤaj 'false'
I don't know why 譌 doesn't have a voiced root initial like 僞 *m-ɢʷ(r)aj-s 'to falsify'; no Middle Chinese reflex for OC *m-qʷʰˤ- is listed in Baxter and Sagart (2014: 130).
I reconstruct GSR 540 as a uvular series with a preinitial conditioning the fronting of *q- to *k- and a presyllable conditioning the lenition of *-ɢ- to *j-:貴 *C.quj-s > MC *kujʰ 'precious, expensive'
靧 *qʰuj-s > MC *xwəjʰ 'to wash the face' (with 面 'face' on the left)
遺 *Cɯ.[ɢ](r)uj > MC *jwi 'to leave, reject' (with 辶 'walking' on the left)
Is there any word-family evidence pointing to an original velar initial for 貴?
The earliest transcription evidence I know of for 貴 is 貴霜 for Kushan in the Book of Han (111 AD). but that only tells us that 貴 had *k- by the second century AD. It could have had a uvular initial in the past.
*I have long considered Arabic uvular q to be the 'emphatic' counterpart of velar k. Jakobson (1957: 515-518, cited in Baxter and Sagart 2014: 383) also considered Arabic pharyngeal ħ to be the 'emphatic' counterpart of velar x. Watson (2002: 44) added a third consonant to this group: Arabic pharyngeal ʕ as the 'emphatic' counterpart of glottal ʔ.
In Baxter and Sagart (2014)'s OC reconstruction, both uvulars and velars can be either emphatic or nonemphatic: e.g., there is a four-way contrast *k : *kˁ : *q : *qˁ.
I know of no language with a similar four-way phonemic contrast. I don't even know of any language with qˁ in addition to q. According to Youssef (2006: 13), Cairene Arabic has [k kˁ q] in its phonetic inventory but not [qˁ]. He regarded [k kˁ] as allophones of /k/ on page 39. On page 40 he did posit qˁ as a "surface realization" which, despite the term, is apparently a intermediate stage between an "underlying representation" and a "phonetic form" in his theory.
The only two pharyngealized velars in UPSID are kˁ in Shilha (probably equivalent to q in the phonemic inventory at Wikipedia) and ŋˁ in !Xu. No languages in UPSID have *kʰˁ, *gˁ, *ŋ̊ˁ, *kʷˁ, *kʷʰˁ, *gʷˁ, *ŋʷˁ, or *ŋ̊ʷˁ which are in Baxter and Sagart's OC reconstruction. Even Cairene Arabic which is full of pharyngealized consonants has only two pharyngealized velars [kˁ xˁ] which are allophones of /k x/ and are hence not phonemic (Youssef 2006: 39).
UPSID only has one language (Rutul) with two (qʰˁ, *ɢˁ) of the six pharyngealized uvulars in Baxter and Sagart's OC reconstruction: *qˁ, *qʰˁ, *ɢˁ, *qʷˁ, *qʷʰˁ, *ɢʷˁ.
Even as I finally read Baxter and Sagart's book, I remain skeptical about reconstructing such a large number of velars and uvulars. I think half of those phonemes are short-lived and secondary, and I'm not even sure they were all phonemes.
**I'm not sure if Baxter and Sagart (2014: 129) intended to say that the 'stop' words are cognate to *qʰ-words for 'place'.
18.104.22.168:59: THE LOST LASO
A year ago I wrote about the Tangut name transcribed in Chinese as 老索 *law so 'old cord'. In his latest post, Andrew West proposed that it may be "a transcription of the otherwise unattested Tangut family name [*la so]":
'Laso' = 5044 1la1 'Tangut surname La' + 2670 2so1 'man'
1la1 may be from *law which in turn may be from *lak. The Chinese transcription may reflect a nonstandard Tangut dialect in which the name was *Lawso.
I think his reconstruction is plausible. Both halves are attested as elements in Tangut family names listed in Miscellaneous Characters:
'Lewla' = 4788 1lew1 'Tangut surname Lew' + 5044 1la1
'Solwo' = 2670 2so1 + 1595 1lwo1 'dim (light), dusky' (a metaphorical adjective?)
Also, the first half transcribed Chinese 老, though that does not guarantee that a Chinese speaker would transcribe 5044 1la1 with 老. (I initially thought the second half transcribed Chinese 索, but I was wrong*.)
I'd like to study Tangut family name structure.
1.12.3:21: ADDENDUM: Although I assumed that 索 was read *so in the Chinese dialect underlying the transcription 老索, there are other possibilities.
In Phags-pa Chinese, 索 had two readings:
ꡛꡓ <saw> < Middle Chinese *sak
ꡚꡗ <shay> < Middle Chinese *ʂɤak ~ *ʂɤek
Jiyun lists a Middle Chinese reading *soʰ (< Old Chinese *sak-s) that would have become *ꡛꡟ <su> in Phags-pa Chinese.
So (pun unintended!) in theory the second syllable might have been standard Tangut sa (< *saw preserved in a dialect?), she (< *shai preserved in a dialect?), or su.
However, I favor *so, as most modern Mandarin readings of 索 including those in the northwest (Coblin 1994: 383) seem to be descended from *so(ʔ) rather than *saw, *ʂaj, or *su.
1.12.3:01: Although Li Fanwen (2008: 439) glossed
2670 2640 2so1 1pho1
in the Tangut translation of the Golden Light Sutra as 索訶 *soxo < *sakxa (the *-k- is inexplicable), a transcription of the first half of Sanskrit sahāpati 'lord of the saha world' (i.e., this world), I think it is actually a transcription of the first syllable of Chinese 娑婆 *sopho < *saba, a transcription of the first half of Sanskrit sabhāpati, a variant of sahāpati. The bh is a hypercorrection by Middle Indic speakers who knew that their h was a lenition of Sanskrit (i.e., Old Indic) bh and who mistakenly assumed that the h of sahā was a lenition of bh.
I do not think Tangut 2so1 1pho1 was a transcription of Sanskrit sahā, as I would expect a combination of the normal Sanskrit transcription characters
Other Chinese transcriptions of Sanskrit sahā are
*1693 0165 1sa4 0ha0 (0 = tone and grade unknown).
娑訶 *soxo < *saxa
沙訶 *ʂɤa xo < *ʂɤa xa (沙 may be a simplification of 娑).
and would have been transcribed in Tangut with a second syllable like ho.