14.2.8.22:44: FROM SADSHA TO SOCHI

I should have checked the Russian Wikipedia before writing the previous entry. It states that the first attestation of 'Sochi' was in a 1641 text by Evliya Çelebi referring to "Садша" <Sadsha>. I wonder what his original Arabic script spelling was. I presume it was something like سدش <s ... d ... sh ...>, but how can one be sure of the vowels? Do the "а" in the Cyrillic transcription correspond to alifs in the original: سادشا <sādshā>?

Other variants mentioned in the article are Садча <Sadcha> and Сача <Sacha>. None of these three quite match the Abkhaz, Adyghe and Ubykh names for the Sochi River, though their phonetic resemblance surely is not a coincidence.

I have been wondering if historical phonology could reconcile the differences between these names. There is no guarantee that the Northwest Caucasian (NC) languages had been free from sound changes between 1641 and today.

All three local river names have initial 'shibilants': Abkhaz Шә- [ʃʷ], Adyghe Шъ- [ʂ], and Ubykh Шь- [ʃʲ]. Turkish and Russian also have 'shibilants' (ş [ʃ] and ш [ʂ]), so there was no need to borrow a foreign sh- as s-. Were Sadsha, Sochi, etc. based on a NC language variety in which *sh- had become *s-? Or did s- become sh- throughout the area?

Does o in Sochi reflect a NC language variety in which *wa had become *o? Or did *sho break to [ʃʷa] in Abkhaz and lose its labiality in Adyghe and Ubykh?

Was there an earlier cluster *-dsh- or *-dch- that was later reduced to ch? If Çelebi had heard a simple ch, he could have written it as چ‎ <ç> [tʃ].

I still don't know why Sochi ends in -i. All other versions of the name end in -a.

14.2.7.7:14: WHY ISN'T SOCHI CALLED SHACHA?

According to Wikipedia, the city is named after the river whose non-Russian names have no s, o, or i:

Abkhaz Шәача [ʃʷatʃʰa]

Adyghe Шъачэ [ʂaːtʃa] (э is short [a], not [e]!)

Ubykh Шьача [ʃʲatʃa]

(Sochi was in a "region [that] was dominated by the Abkhaz, Ubykh and Adyghe tribes".)

I would expect 'Sochi' to be *Шача <Shacha> which would decline like a regular -a noun: e.g., *Shachi would be the genitive singular and nominative/accusative plural.

Aha, is this the source of the name (emphasis mine)? It's closer to Sochi than any of river names:

The city itself and its major river were named after the Ubykh tribe Soatshe which was living in the area before 1864.

Is its homophony with Georgian სოჭი <sochi> 'fir' coincidental?

14.2.6.15:09: DĒCLĪNĀTIŌ CAPITIS ET CĒTERŌRUM

I have been using caput 'head' (and by extension 'capital city') as a translation of 京 Gin in the titles of my last four posts. Capitis (not *caputis!) is the genitive singular of caput.

I get irritated when people say Latin is a "logical language" ("latin" "logical language" has 76,700 results in Google). What is "logical" about partly unpredictable stem vowel changes (types C-F below)?

A selection of Latin third declension types

I use the genitive singular to exemplify forms other than the nominative and vocative singular.

Type A is the simplest yet is surprisingly uncommon. Hale and Buck (1903: 38) list type A words as exceptions to the general pattern exemplified by type D. According to Hale and Buck (1903: 19), "the retention of the e [in segetis] is due to the assimilating influence of the e of the first syllable." I suppose the e in the second syllable of interpretis had a similar stabilizing effect.

Type B is also straightforward, but there are type C exceptions.

The e of type E became labial u to assimilate to a following labial p (Hale and Buck 1903: 20). Manceps moved from type E to type D in later Latin, leaving auceps as the sole (?) member of type E. Contrast aucupis and early Latin mancupis with type D prīncipis (not *prīncupis!). Did the ī of prīncipis block assimilation before p?

Caputis (type F) is sui generis. My guess is that its i is by analogy with type D.

14.2.4.14:14: SONI LINGUAE CAPITIS (PART 4)

You may have noticed the following initial correspondence in part 2:

Middle Chinese 冊 *tʂʰɛk > Gin tʰat, Hanoi [sac], Saigon [ʂat] 'book'

Here are more examples (excluding straightforward cases when Gin, Hanoi, and Saigon all have tʰ- from *t(s)ʰ-):

Pattern A: PV *Cr- > G tʰ : H [s] : S [ʂ]

In Gin, *Cr- became *tr- and then tʰ-, whereas it became ʂ- in Vietnamese proper, fronting to s- in Hanoi.

A1. I have reconstructed 'six' with *k-r- as well as *p-r- because of variation within Vietic (below):

Hoà Bình Muong kʰaw

Sơn La Muong faw

Mlengbrou parawʔ

A2. The most common pattern I have in the data on hand.

A3. I suspect this word is a borrowing from Old Chinese and was Proto-Vietic *k-saːŋ rather than *k-raːŋ. There is no modern Vietic language with -r- in 'frost'.

(2.4.15:30: I also cannot find any other Mon-Khmer languages with similar forms for 'frost', so the word is probably a Vietic innovation if not a loan into Vietic.)

The c. 10th century Sino-Vietnamese reading sương (Hanoi [sɨəŋ] / Saigon [ʂɨəŋ]) for 霜 happens to be homophonous with the earlier Old Chinese borrowing following Vietnamese-internal developments.

A4. Like A1 and A2 but with c- instead of p- or k-.

A5. I could not find any Proto-Vietic reconstructions or non-Vietnamese data for these words at SEAlang, but have reconstructed them with *Cr- in Proto-Vietnamese by analogy with other pattern A words. I define Proto-Vietnamese as the ancestor of Gin and modern Vietnamese dialects; it is a sister to the ancestors of the varieties of Muong (which are not descendants of a single Proto-Muong; see Phan 2013).

Proto-Vietic
Proto-Viet-Muong				Non-Viet-Muong branches: Northwest, West, Southeast, Southwest, South
Proto-Vietnamese			Muong varieties
Gin	Hanoi	Saigon

A6. I have no idea why Hanoi and Saigon have [tʰ] like Gin.

Pattern B: MC *tʂʰ- > G tʰ : H [s] : S [ʂ]

My guess is that MC *tʂʰ- was borrowed into Proto-Vietnamese as *ʂ- which hardened to tʰ- in Gin but fronted to [s] in Hanoi. Compare with other MC fricatives and affricates that hardened in Vietnamese:

Middle Chinese	Gin	Hanoi	Saigon
*(t)s-, *(d)z-	t-	[t]	[t]
*tʂ-		[c]	[ʈ]
*tsʰ-, *ɕ-, *(d)ʑ-	tʰ-	[tʰ]	[tʰ]

Pattern C: MC *dʐ- > G tʰ : H [s] : S [ʂ]

Pattern D: MC *ʂ- > G tʰ : H [s] : S [ʂ]

MC *dʐ- weakened and devoiced, merging with *ʂ- in Annamese Middle Chinese. That *ʂ- was then borrowed into Proto-Vietnamese as is.

Just as *Cr-clusters became tʰ-, some *Cl-clusters became t- in Gin, whereas others became j-:

The Gin reflexes are largely predictable except for pattern B1:

*nonlabial + *l- > t- (patterns A1, A2)

*p/b- + *l- > j- (patterns B2, B3; possibly with a *z-like intermediate stage resembling Hanoi [z])

Did 'python' have *p- instead of *k- in the ancestor of Gin?

*ml- > ɲ- (pattern C)

The second (but not third!) change is reminiscent of how Old Chinese *ml- became Middle Chinese ʑ-.

*Mostly taken from SEAlang's Mon-Khmer dictionary. I discuss exceptions above.

14.2.3.13:06: SONI LINGUAE CAPITIS (PART 3)

In the 90s I came up with a neutral (ha!) notation for Sinospheric tonal categories based on the phonetic characteristics of onsets and codas that conditioned tonogenesis. Here's my updated version:

Maybe I should partly revert to my original with q for ʔ. I will, however, use v for 'voiced' instead of ɦ (which can be easily confused with h) as in the above proposal or x (which looks like a voiceless fricative) as in my old scheme. Also, I will continue to make a distinction between -c and -:c absent in the original, substituting a colon for IPA ː. I did know about that distinction in Cantonese and Tai at the time but left it out of my old scheme.)

Gin, Vietnamese, and Chinese do not distinguish between the h- and q- categories, so I will drop the h- category in the comparative table below:

Notes on each category:

qv: Gin and Hanoi identical; distinct from Cantonese.

vv: Gin and Hanoi very close; distinct from Cantonese. Could Gin have breathiness? Descriptions of tones may accidentally exclude phonation.

qq/qc: Did the Gin tone move into the 45 space vacated by vh which merged with qh in Gin?

vq: All nonhigh but also all distinct from each other. Could Gin have glottalization?

qh: Gin has a falling-rising contour like northern Vietnamese dialects (though not Hanoi).

vh: Merged with qh in Gin and in southern (!) Vietnamese dialects. Hence no 4 in the Jingyu jianzhi numbering system.

q:c: 45 in Gin native words and early borrowings but 33 in late borrowings from Cantonese. 33 is also the Gin qv tone (see above).

v(:)c: No tones conditioned by vowel length. Gin is identical to Cantonese (unless the former has glottalization as in Hanoi).

There is a single Gin tone sandhi rule without parallel in Vietnamese proper or Cantonese which lack tone sandhi. The dipping tone loses its rise before a high rising tone:

qh/vh 214 + qq 45 > qh'/vh' 21 + qq 45

This is reminiscent of the 'half-third' (qq) tone of Mandarin which has the same contours:

qq 214 + qv 55 / vv 35 / vq/qh/vh 51 > qq' 21b + qv 55 / vv 35 / vq/qh/vh 51

The coda dissimilation rule in part 2 does not involve tonal dissimilation in the two examples given in

v(:)c 22 + v(:)c 22 > vv 22 v(:)c 22

vat²² vat²² > van²² vat²² 'bright (moon)'

ʔut²² ʔut²² > ʔun²² ʔut²² 'quick'

(A ʔ-initial word with a proto-voiced initial tone is anomalous and presumably of late onomatopoetic origin.)

I don't know what happens to reduplications of q(:)c syllables. My guess is that their tones also remain stable:

q(:)c 45 + q(:)c 45 = qq 45 q(:)c 45

q:c 33 + q:c 33 = qv 33 q:c 33

*I have followed the description of Hanoi tones in Wikipedia.

**I have followed the description of Cantonese tones in Jingyu jianzhi.

14.2.2.22:40: SONI LINGUAE CAPITIS (PART 2)

The vowels of Gin as described in Jingyu jianzhi are identical to those of Vietnamese at a glance, but palatal codas (-c, -ɲ) became dentals as in Saigon far to the south of the Chinese islands where Gin is spoken:

However, the parallels between Gin and Saigon are only superficial.

The loss of final palatals in Gin may be due to the influence of Cantonese which lacks final palatals. As far as I know, no modern Chinese language has final palatals*.

In Saigon, the dentals generally backed and the palatals moved into the space they partly vacated: e.g., /aɲ ac/ became /an at/ after original /an at/ became /aŋ ak/.

(The following section was revised on 2.4.13:23.)

The codas of Gin may alter the onset of a following syllable (cf. the section on gemination in Vietnamese on the last page of this PDF):

-p/t/k + h- > -p/t/k + pʰ/tʰ/kʰ-

-n + ʔ- > -n + n- + nonfront vowels?

-n + ʔ- > -n + ɲ- + ɛ (and also e and i?)

-ŋ + ʔ- > -ŋ + ŋ-

-n + j- > -n + ɲ-

-j + n- > -j + ɲ-

This reminds me of Japanese 連聲 renjō 'joined sounds': e.g.,

觀 kwan 'see' + 音 on 'sound' = Kwannon > now Kannon 'Avalokiteśvara'

雪 set 'snow' + 隱 in 'hide' = settin > now setchin 'outhouse'

The stop coda of a reduplicated syllable dissimilates (cf. section 7.13 of Thompson 1965 on this phenomenon in Vietnamese):

CVp CVp > CVm CVp

CVt CVt > CVn CVt

CVk CVk > CVŋ CVk

*I think the final /ɲ/ in the rhyme tables of the English and Chinese Wikipedia articles on Shanghainese might be mistakes. There is no reference to them in the accompanying text or in other descriptiosn of Shanghainese I ahve seen (e.g., 2003).

In any case, I have never seen any description of a modern Chinese language with final /-c/.

Hashimoto reconstructed final palatals in Middle Chinese. The southern Middle Chinese dialect that is the source of most of Sino-Vietnamese may have had final palatals.