14.9.13.22:40: A DIP INTO WHITE WATERS (PART 4): LEFT-*O-VERS

One-fifth of 白水村 Baishuicun (BSC) 'White Water Village' dialect forms ending in -o cannot be explained using the sound laws I proposed in parts  2 and 3. I will try to explain these forms which fall into eight categories. I could have covered the first two categories in part 3, but they are more complicated than the majority of the *-at class.

1. 佛 fo < *fat (< *but) 'Buddha'

This appears to be a loan from a language in which *-ut became *-at as in Cantonese 佛 fat (though Cantonese is not spoken in Hunan and therefore is not the source of fo). An earlier loan is pu which is close to prestigious Early Middle Chinese *but, an abbreviation of 佛陀 *but da 'Buddha'.

Could the word simply be a very recent borrowing from Mandarin fo (whose rhyme is irregular)?

I am reluctant to guess glosses for BSC forms since the 小學堂 Xiaoxuetang database does not provide any, but in this case I'm pretty sure 佛 is 'Buddha' (though 佛 may have other meanings in BSC), and it would be odd to mention 佛陀 *but da without also mentioning its Indic source.

2. 蝨 so < *sat (< *ʂɨt < *ʂit < *srit < *srik)

This might be a loan from a language in which *-it became *-at as in Cantonese 蝨 sat (though I must note again that Cantonese cannot be the source of this particular word).

3. -o < *-ai

In part 2, I dealt with BSC -o-forms corresponding to -ai-type rhymes in other Chinese languages. The following words may have had pre-BSC *-ai even though they may not have -ai-like rhymes in other Chinese languages:

Pre-BSC *-ai may have been a reflex of Old Chinese *-aj/-əj/-ɨj-type rhymes; it cannot have been borrowed from a prestigious EMC or LMC dialect. (BSC 衣 i is a loan from an prestige LMC-like form.)

篩 is not attested in Old Chinese. Pre-BSC 篩 *sai is reminiscent of Cantonese 篩 sai, though it cannot be from Cantonese. The word could be borrowed from a form like Mandarin 篩 shai (whose reading is from 籭; the hypothetical regular Mandarin reading would be *shi).

4. 麥 mo < *mai

This has an unexpected yang departing tone. Is this a very late loan from a form like Mandarin 麥 mai which lost its final stop and entering tone? If it were a native word or an early loan, it should have an entering tone as a trace of its original *-k.

5. -o < *-raw

咬 ko is ultimately from Old Chinese *kraw but it could be a loan from *kaw or *ko in some later language. I can't look further into this (or anything else BSC-related) because the Xiaoxuetang site is down.

There are three forms in this category: 灑下拏. 咬 from category 4 may also belong to this category, as pre-BSC or a source language may have been like Taiwanese which has -a from both *-ra and *-raw:

咬 T ka (kau is a borrowing)

灑 T sa (borrowed?; se may be native; cf. 下 below)

下 T literary (i.e., borrowed) ha; the colloquial (i.e., native) form is e

拏 T na (borrowed?; displaced a native *ne?)

"Like" does not entail a close relationship. Taiwanese is both genealogically and geographically distant from BSC.

7. -o < *-ə(ŋ)

Is 扔 no from Old Chinese *nəŋ or an open-syllable variant *nə (cf. Japanese no < *nə for 乃, the phonetic of 扔)?

8. Unrelated synonyms

I initially thought -o of 久 no might be a reflex of Old Chinese *-ə in 久 *kʷəʔ, but I doubt n- is from *kʷ-. Moreover, the yang departing tone of no is not what I would expect for a descendant of 久 *kʷəʔ. I conclude that no is an unrelated synonym.

萎 mo < *mat? has an initial and an entering tone (from an earlier final stop) that rule out any relationship with Old Chinese *Cɯ-ʔoj. Even if *C- were *m-, the tone would still be irregular.

拋 no < *nat? has the same problems as 萎 mo; it cannot be related to EMC 拋 *phræw (the word is not attested in Old Chinese).

鷹 ŋo < *ŋat? is not related to Old Chinese 鷹 ʔəŋ.

硬 ko superficially resembles Sino-Japanese kō but may be from *kat which cannot be related to EMC 硬 *ŋɤeŋʰ (the word is not attested in Old Chinese, and the SJ initial is irregular).

14.9.12.23:36: A DIP INTO WHITE WATERS (PART 3): FINAL ST-*AP-S

The chain shift I proposed in part 2 explains three-fifths of -o syllables in the 白水村 Baishuicun (BSC) 'White Water Village' dialect:

*-ai > *-oi > -o

Another fifth requires further sound laws:

*-ap merged with *-at (cf. the *-p > -t shift in Nanchang which is 500 km to the east and not closely related)

*-at > *-ait > *-oi(t) > *-o

I propose that *-t had a fronting effect on *a similar to the fronting effect of -d in Tibetan:

'eight': Proto-Sino-Tibetan ?*prjat >

Old Chinese *pret > pre-BSC *pat > pait > *poi(t) > BSC po

Earlier Tibetan brgyad > Lhasa cɛʔ

Here are more examples to show the merger of multiple *-t and *-p rhymes. The Early and Late Middle Chinese forms here are composites based on prestige dialects and are not directly ancestral to BSC.

Although BSC no longer has any final stops, they have conditioned tones absent from words without original final stops: high level if the initial consonant was *voiceless and mid level if the initial consonant was voiced. I don't know when the final consonants were lost after *a shifted to *ai before *-t.

發 must be a loanword because it has *f- instead of the expected *p- (see my discussion of 煩 xoi and 佛 pu ~ fo in part 1). Foreign *f- might have been borrowed as *f- after BSC developed its own *f- from *xw- in words such as 穴. Conversely, it is also possible that *xw- became *f- after loanwords introduced *f- into the BSC phonemic inventory.

14.9.11.23:41: A DIP INTO WHITE WATERS (PART 2): A CH-OI-N SHIFT?

In part 1, I proposed the following sound change for the 白水村 Baishuicun (BSC) 'White Water Village' dialect:

*-an > *-oi

*-an > *-ai > *-oi > -o

BSC -o often corresponds to *-e/*-aj-type rhymes in prestigious Early and Late Middle Chinese dialects which were not its ancestors (but might have been sources of loans into BSC). I do not include tones in pre-BSC and BSC forms. I have not heard BSC, but I assume that its -i is [j] after vowels, so there is no real difference between MC *-j and (pre-)BSC (*)-i.

My proposal accounts for 59% (58/99) of the -oi forms in the 小學堂 Xiaoxuetang database. I will deal with the others in part 3.

BSC p- corresponding to LMC *f- (e.g., in 吠) indicates a native word. A hypothetical early loan of 吠 would be *xo and a hypothetical late loan would be *fo (see my discussion of 煩 xoi and 佛 pu ~ fo in part 1).

14.9.10.23:40: A DIP INTO WHITE WATERS (PART 1)

Over the past couple of days I have been intrigued by the dialect of 湘南土話 Xiangnan Tuhua 'local speech of southern Hunan' spoken in 白水村 Baishuicun 'White Water Village' in 江永縣 Jiangyong County. In " 'More' Evidence", I found that 更 Old Chinese (OC) *kraŋ(s) / Middle Chinese (MC) *kɤaŋ(ʰ) 'watch of the night'/'more' corresponded to Baishuicun (BSC) koi. I hypothesized that -oi was from an *-aɲ like that of Sino-Vietnamese canh/cánh [kaɲ]. Looking at other BSC -oi forms, I can make a more general statement: *A/O-type vowels followed by nonback nasals (*ɲ, *n, *m) became -oi. The nasals probably merged into *-n before becoming -i.

Pre-BSC is a very rough guess bridging Late Middle Chinese (LMC)* and BSC. It looks more like a typical Chinese language than BSC does.

煩 xoi is probably a loanword. I think the native BSC reflex of EMC *b- is p-: e.g., 佛 pu 'Buddha' (a later borrowed form is fo). BSC is not descended from generic LMC dialects which lenited labial stops to fricatives before *u. The borrowing of 煩 must predate the shift of *-an to *-oi and the borrowing of *f-. *x- was the closest pre-BSC equivalent of foreign *f-. I conclude there are at least two layers of borrowings that can be distinguished by their treatment of foreign labiodental fricatives: an older x-layer and a newer f-layer.

I think the l- of 單 is due to BSC-internal lenition.

The above scheme accounts for most but not all instances of -oi in BSC. Two requiring further investigation are

吾 OC *ŋa > EMC, LMC *ŋo : BSC ŋoi

崖 OC *ŋre > EMC *ŋɤe > LMC *ŋæj : BSC ŋoi

Neither had nasals in earlier Chinese. The normal BSC reflexes of OC *-a and *-re are -u and -o. 崖 ŋoi could be a borrowing from a dialect that had broken *-ɤe to *-æj or the like. But the -i in 吾 ŋoi remains a mystery.

The above scheme cannot account for cases in which -oi did not develop from *-an: e.g., 肝 BSC kaŋ (not *koi!) <  OC/MC *kan. It seems that velars somehow blocked the *-an to -oi shift.

Next: A ch-oi-n shift?

*9.11.23:03: The LMC reconstruction here is a composite of the prestige dialects underlying borrowed forms in Chinese and non-Chinese languages. It is not a direct ancestor of pre-BSC, though such an ancestor may have been similar and may have borrowed from an LMC prestige dialect.

14.9.9.23:27: 'SECONDAR-Y' ROUNDING IN CANTONESE

The unexpected labiovelar /kʷ/ in  Cantonese 梗 /kʷaːŋ˧˥/ 'stem' (among many other meanings) from my last post brought to mind a Cantonese form that has puzzled me for many years: 乙 /jyːt˧/ 'second Heavenly Stem'. I used to think its rounded vowel /yː/ was unique, as it wasn't in any reconstruction or actual form that I had ever seen: e.g.,

Old Chinese *ʔi̯ɛt (Karlgren), *qrig (Zhengzhang), my *ʔrət or *ʔrit (I cannot find any rhyming evidence favoring one vowel over the other*)

Middle Chinese: *ʔi̯ĕt (Karlgren), *ʔɣiɪt (Zhengzhang), my *ʔɨit

Mandarin yi

Taiwanese it

Sino-Vietnamese ất [ʔət]

Sino-Korean ŭl [ɯl] < idealized ʔɯ́rʔ

Sino-Japanese otsu < *ət

However, I now see that rounded vowels are not only in Yue varieties like Cantonese but also a few southern non-Yue varieties:

Yue: too many /y/-varieties to list; other rounded vowels (or glides?) are in

Xintian Fantian, Dapu Taiheng jɵk

Kaiping (Chikan) zuat

Taishan (Taicheng) zᵘɔt (? - there is no syllable like this in Stephen Li's Taishan syllabary)

Mengshan iut

Huaiji wut

Dongguan (Guancheng) zøt

Bao'an (Shajing) (j)iɔʔ

Ping: Nanning yt (loan from Cantonese?)

Hakka: Huizhou yat

Unclassified:

Zhongshan (Gong'an) iuə

Shaozhou Tuhua:

Xingzi ɵy

西岸 Xi'an oi

Bao'an u

All the non-Yue varieties are within the Yue area, so their rounding may be due to Yue influence.

If rounding is a Yue innovation, why did it happen? Both 梗 and 乙 had medial *-r- in Old Chinese. Did that *-r- sporadically become *-w-? (Cf. Elmer Fudd's "wascally wabbit".) Are there other Old Chinese *-r- words with modern labial reflexes?

I would expect kw-reflexes of Old Chinese 甲 *qrap 'first Heavenly Stem', but the only remotely similar forms are

Yixian kɔɐ̆ʔ

Guilin (Chaoyang) kuo

Jiangyong Chengguan (Baishuicun) kuə

whose diphthongs might be breakings of an *o from *a (cf. o-forms like Lingchuan (Tanxia) ko). None of those three varieties have rounded vowels in 乙, though Baishuicun does have a rounded vowel in 梗.

*My Old Chinese *ʔrət and *ʔrit could both become Middle Chinese *ʔɨit:

I have included two other rhymes for comparison.

There was a chain shift:

*-ət > *-ɨət > *-ɨit

That could be interpreted as a push or pull chain:

Push: When *-ət broke to *-ɨət, it 'pushed' original *-ɨət into the 'space' of *-ɨit.

Pull: When original *-ɨət merged with *-ɨit, it left a gap to be filled by *-ət after it broke to *-ɨət.

I generally prefer pull chains, but mixed reflexes of *-ət and *-rət might point to a push chain.

14.9.8.23:08: 'MORE' EVIDENCE FOR THE LIMITS OF THE MIDDLE CHINESE LEXICOGRAPHICAL TRADITION

Last night I saw this passage in the Wikipedia article on Cantonese phonology:

There are about 630 sounds [i.e., syllables disregarding tones?] in the Cantonese syllabary. Some of these, such as /ɛː˨/ and /ei˨/ (欸), /pʊŋ˨/ (埲), /kʷɪŋ˥/ (扃) are not common any more; some such as /kʷɪk˥/ and /kʷʰɪk˥/ (隙), or /kʷaːŋ˧˥/ and /kɐŋ˧˥/ (梗) which has traditionally had two equally correct pronunciations are beginning to be pronounced with only one particular way uniformly by its speakers (and this usually happens because the unused pronunciation is almost unique to that word alone) thus making the unused sounds effectively disappear from the language [...]

At first I was puzzled by 梗 /kʷaːŋ˧˥/ 'stem' (among many other meanings) which has a labiovelar initial even though it is written with a velar phonetic 更 'watch of the night'/'more'. I have never seen an Old Chinese reconstruction of 梗 with a labiovelar or labial. 梗 had no *-w- according to the Middle Chinese lexicographical tradition based on prestige varieties. But not all modern forms arise from those varieties. Forms in multiple branches of Chinese (written here without tones) may point to *-w-:

Yunhe kuɛ (see here for more Wu forms with -u-)

Nanchang kuaŋ (see here for more Gan forms with -u-; is Leping muaŋ a typo for kuaŋ?)

Fuzhou literary (!) kuaŋ ~ colloquial keiŋ (see here for more Min forms with -u-)

Lechang kuɐn (see here for more Yue forms with -u-)

Lingui kʰyɛn (the only Ping form with a labial; the aspiration is irregular and can be sporadically found in other Ping varieties and in Yue, Hakka and even Mandarin)

Meixian literary (!) kuaŋ ~ colloquial kɛn (see here for more Hakka forms with -u-)

Fengyang kua (see here for more Shaozhou Tuhua forms with -u-)

The -u- and -y- of those forms cannot be derived from Middle Chinese reconstructions for 梗 such as my *kɤaŋˀ or Old Chinese reconstructions derived in turn from those reconstructions: e.g., my *kraŋʔ. (I reconstruct its phonetic 更 as Middle Chinese *kɤaŋ(ʰ) from Old Chinese *kraŋ(s).)

梗 has no labials in Mandarin, Jin, or Xiang. Was labiality lost in the north, or is it a common retention of southern languages that do not form a subgroup? Fuzhou, Meixian, and perhaps other varieties may have borrowed from one or more southern literary Middle Chinese dialects with a labial absent from other prestige dialects.

Shaxian and Sanming kɔ̃ (< *kaŋ?; Sanming also has kɛ̃; other Min forms here)

Yangshuo kyɛ̃ (but Lingui kəŋ; other Ping forms here)

Hezhou kɔ (< *kaŋ?)

Jiangyong Chengguan (Baishuicun) koi (< *kaɲ?; cf. Sino-Vietnamese canh 'watch of the night' ~ cánh 'more' [kaɲ])

The labials of most of these forms do not necessary point to *-w-. The shifts I propose for Shaxian, Sanming, and Hezhou have parallels in northwestern Middle Chinese (in which *-aŋ became *-o; a similar shift occurred in neighboring Tangut and its relative Japhug rGyalrong).

14.9.7.23:57: *PI̵K A CODA

Here's something I don't see every day: a Chinese character (逼) whose readings have three different types of codas:

Velar/glottal: Cantonese bik [pɪk], Suzhou pɪʔ (source)

Alveolar/dental: Sino-Japanese hitsu < *pit

Labial: Sino-Korean phip

Its Middle Chinese reading was *pɨk. I can't explain this diversity.

9.8.0:48: There are Chinese readings with -t as well, but they are regular reflexes of *-k after front vowels: e.g.,

Toisanese pet < *pek (source)

Meixuan Hakka pit < *pik (source)

That is not the case with Sino-Japanese hitsu with a -tsu instead of the expected -ki. There was no such fronting rule in Japanese or in the Chinese source dialects of Sino-Japanese.

Conversely, 匹 Middle Chinese *phit has two Sino-Japanese readings: hiki as well as the expected hitsu < *pit.

Sino-Korean is full of irregularly aspirated labial initials. In fact there is no *pa in Sino-Korean; all syllables that should be *pa are pha: e.g., 波 pha < Middle Chinese *pa. I have long wondered if this was the product of hypercorrection. (Korean never had f, so Chinese *f- was Koreanized as the stops p- and ph-, and words without *f- in Chinese such as 波 might have been read as if they had *f-.)

The idealized Sino-Korean readings of Tongguk chŏngun (1448) lack this excess aspiration of labials. (I would expect the Tongguk chŏngun reading of 逼 to be *pík, but I can't find 逼 in that dictionary. Although Martin 1992: 126 listed pík in a table of Tongguk chŏngun readings, I don't see it in the book itself.)

On the other hand, Sino-Korean is almost completely lacking in kh-readings, though Tongguk chŏngun has them where they are expected: e.g.,

可 Sino-Korean ka, Tongguk chŏngun khǎ < Middle Chinese *khaˀ

This may tell us something about the chronology of the development of Korean aspirates which are either borrowed or secondary.