Last night, I wrote about how Cantonese -aj corresponded to Sino-Vietnamese -e and concluded that the two have related but distinct sources. However, there are a few instances of Cantonese -aj (and Old an/dor Middle Chinese *e) corresponding to V -ay [aj], -ây [əj], -ai [aaj], or -ươi [ɨəj] < *-aj:

(Not all readings of 洒/洗 'wash' are listed. 洒 combines 氵 'water' [semantic] with 西 'west' [phonetic].)

I briefly considered the possibility that SV tây 'west' and tẩy 'wash' were borrowed from a stage before OC *ə had fronted to *e, but the hỏi tone of 'wash' points to a late borrowing. An earlier borrowing would have been *tấy with a sắc tone.

The two SV readings of 西 'west' (tây and the rare tê from 四書不二字音義撮要 as listed in Mineya 1972) are reminiscent of its two Sino-Japanese readings: sai, se. Similarly, colloquial SV lạy and literary SV lễ are like SJ rai (as in 禮記 Raiki 'Book of Rites') and rei 'bow'. The SJ rhymes reflect two different strata of borrowing: an earlier stratum (呉音 Go-on) from the south (hence 呉 Go) filtered through the Paekche language and a later stratum (漢音) from the northwest. The source of Kan-on is not the descendant of the source of Go-on.

Similarly, could literary Colonial Chinese (the source of literary SV) not be the descendant of colloquial Colonial Chinese (the source of colloquial SV) in which *-e had broken to *-əj ~ *-a(a)j? In other words, was it an imported prestige dialect? Or perhaps LCC was descended from earlier CCC, but had been influenced by another literary dialect (e.g., by changing local *-əj to *-ej; imagine American English speakers imitating some aspects of RP).

*Schuessler (2007: 297) regards 妓 as a loan into Chinese "from early Vietnamese", but I suspect the resemblance between 妓 and V cái is coincidental.

However, I do wonder if V bố (as in bố cái 'father [and] mother') is from a southern Late Old or Middle Chinese *poʔ that was a reversed-type variant of mainstream Late Old / Middle Chinese 父 *buoʔ < *baʔ (< *N-paʔ?) 'father'.

Last night, I wrote about a trait that sets Cantonese and Sino-Vietnamese apart from other Chinese languages and Sinoxenic lexica. Is this trait a shared innovation pointing to a common origin? In other words, are both Cantonese and Sino-Vietnamese derived from an earlier southern Chinese dialect which had lowered *i before labial and dental codas? Or are they derived from two southern Chinese dialects that

a. descended from a common close ancestor?

b. are not closely related and ...

... lowering diffused from one to the other?

... lowering diffused into both from a (lost?) third dialect?

... lowered *i independently?

If Ct and SV shared many other unique characteristics, they probably did not develop that set of traits independenty. However, lowering *i is not an exotic change and could happen more than once, so it is not sufficient evidence for a single ancestor.

If Ct and SV shared a single ancestor, ideally all differences between the two were due to sound changes that occurred in Ct and SV after they split: e.g.,

Late southern literary Middle Chinese 家 *kja 'house'

> Ct ka (via *-j-loss)

> SV gia [za] (via *kj- > *kʑ- > *gʑ- > *ɟʑ- > *ɟ- > *ʑ- > z-)

However, both Ct and SV have conflicting forms that postsplit sound changes cannot account for.

One recurring conflict that comes to mind are the Ct and SV rhymes corresponding to *-(i)ej in dictionary Middle Chinese: e.g.,

雞 Dictionary Middle Chinese *kej 'chicken'

Ct kaj (not kej)

SV kê [ke] (not cay [kaj] or cây [kəj])

At first one might simply regard Ct -aj as the result of a postsplit dissimilation of the palatal vowel *e from the following palatal glide *-j. However, there are at least four problems with this solution..

1. Early Middle Chinese transcriptions indicate that *-ej shifted to something like *-əj in the south centuries before the borrowing of SV.

2. An *-əj-like rhyme probably underlies these forms for 'chicken' in other southern Chinese languages:

Meixian Hakka kaj

Colloquial Amoy kue < *kwaj < *koj < *kəj

Chaozhou koj < *kəj

Note that Ct, Hakka, and the Southern Min languages Amoy belong to three different branches of Chinese. The *-ej > *-əj shift may have diffused among them at an early date since it does not affect literary loans from Middle Chinese: e.g., the literary Amoy reading for 'chicken' is ke without -u-.

3. Proto-Tai *kəj 'chicken' (> Siamese ไก่ kaj) may be older than SV, as other Chinese loans in PT look like they could have come from Late Old Chinese.

4. There is no evidence for a Vietnamese-internal change of *-əj or *-aj to -e. Thus SV -e probably directly reflects a literary Colonial Chinese *-e.

I could try to reconcile all of the above by proposing that *-ej did indeed become *-əj in early southern Chinese, but it then became *-ej again (hence SV kê and maybe even literary Amoy ke) only to shift to -aj in Cantonese:

*-ej > *-əj > *-ej > -aj

- In one dialect of early southern Chinese, *-ej became *-əj. This dialect is the source of the Ct, Hakka, colloquial Amoy, Chaozhou, and Proto-Tai forms listed above .

- Literary Colonial Chinese, though similar to the Proto-Cantonese descendant of that dialect in at least one key respect (e.g., *i-lowering), never underwent this shift. SV proper was generally borrowed from this shiftless dialect, though a few older borrowings persisted as readings. I'll discuss one such old borrowing next time.

- Literary Colonial Chinese became extinct after Vietnam was free from Chinese rule.

One trait that Cantonese and Sino-Vietnamese share is a nonhigh achromatic vowel corresponding to i in other Chinese languages and Sinoxenic lexica before nonback codas: e.g.,

Wenzhou also has a before -ŋ < *-m and -n (though not in 'guest': e.g., 新 saŋ 'new'), but not before -Ø < *-p, *-t.

Why don't Cantonese and Sino-Vietnamese have an i or even an e in the above words? Here's what I think happened:

In early southern Middle Chinese, *i was still intact. Hence *sim 'heart' was borrowed into Vietnamese as *sim which became modern tim.

In one dialect of southern literary Middle Chinese (stage 1), *i lowered to *ɪ before labial and dental codas. (stage 2) *ɪ then lowered and backed to mid central *ə (stage 3). Sino-Vietnamese â [ə] probably reflects stage 3. In stage 4, *ə lowered even further to the a of modern Cantonese. The i of modern Cantonese before labial and dental codas is a long vowel [ii] from Middle Chinese *(i)e.

The evolution of Middle Chinese *i and *(i)e before nonvelar nasal and stop codas

Vietnamese has t- and th- instead of b- (< *p-) and ph- in 'guest' and 'measure word for horses' because the palatal quality of *i left a trace on the preceding consonant in late southern literary Middle Chinese. *i may have lowered to dissimilate it from the adjacent palatal(ized) consonant.

Evolution within southern literary MC:

*pin > *pʲin > *pjin > *pjɪn > *pjən

*phit > *phʲit > *phjit > *phjɪt > *phjət

Evolution within Vietnamese after borrowing:

*pjən > *pɕən > *psən > *sən > tân [tən]

*phjət > *phɕət > *ɕət > thất [thət]

Cantonese lost medial *-j-:

*pjən > *pən > pan

*phjət > *phət > phat

The above implies that Sino-Vietnamese and Cantonese are both offshoots of the same late southern literary Middle Chinese dialect, but that is improbable for reasons I'll present next time.

Distribution of *i before codas in early Old Chinese

Distribution of *i before codas in later Old Chinese

The *-iŋ/*-ik gap persisted until late southern literary Middle Chinese simplified *ie to *i before velar codas:

Older Chinese loans in Vietnamese preserve the diphthong lost in later loans and Cantonese.

Cantonese (but not Vietnamese) -ing and -ik are also partly from Old Chinese *-əŋ and *-ək:

Vietnamese preserves the earlier central vowel *ɨ that merged with *i in Cantonese.

In most Chinese languages and in Sino-Japanese, 得 'get' and 德 'virtue' are homophonous:

However, they are not homophonous in some languages:

(Notes on the non-SK/SV readings here.)

The high vowel in OSV được [ɗɨək] < colloquial CC *dɨək is reminiscent of the high vowel in the Sino-Korean reading tŭk [tɯk] for 得 'get'. [tɯk] would be a possible Koreanization of a foreign [dɨək] since Korean has never had diphthongs like *ɨə. For a moment I thought both OSV and SK pointed toward a colloquial *dɨək in both southern and northern colloquial Middle Chinese distinct from the *tək (southern *tak?) of literary MC. But is it likely that two very distant MC dialects, one in the south and one in the north, would have the same colloquial MC form *dɨək from Old Chinese *N-tək? Moreover, OSV -ược is an irregular rendition of the rhyme class of 得, whereas SK -ŭk is regular:

Note that Sino-Korean ŭ later labialized to u after labial initials.

The above chart could be rewritten in terms of main vowel heights. Matching vowel heights are in bold.

SV and SK vowel heights for this rhyme group do not match very much, implying that their underlying MC dialects were very different.

The source of SK (northeastern Late Middle Chinese?) raised *-əŋ/k to *-ɨŋ/k or *-ɯŋ/k, so SK tŭk [tɯk] is the expected literary reading of 得 'get', not a borrowing from a colloquial MC form like *dɨək (which may not have existed in the north). SK tŏk [tək] for 德 'virtue' must have been borrowed before the raising fo *-ək to *-ɨk or *-ɯk. It stood for "a key concept in Chinese philosophy" that was too embedded in Old Korean to be replaced by a newer reading *tŭk; cf. how SV nghĩa for 義 'justice' was not replaced by a newer reading *nghị (more here*).

The source of SV (late literary Colonial Chinese), on the other hand, lowered *-əŋ/k to *-aŋ/k. SV 層 tầng [təŋ] 'story, stratum, layer' must be an old, established borrowing predating that lowering which was not replaced by *tsaŋ > SV *tầng. từng [tɨŋ] is the reversed-type version of that word with a high vowel implying an Old Chinese type B (nonemphatic / nonunderlined) source.

(4.1.2:36: Vietnamese từng 'ever' is a reversed-type version of 曾 MC *dzəŋ whose SV reading is the expected tằng with a low vowel.)

SV 德 đức 'virtue' combines an unusual high vowel (implying Old Chinese type B) with a stop (implying Old Chinese type A) and is hence a mixed-type borrowing like được 'get'. Was đức an old borrowing like SV nghĩa that was too established to be replaced?

4.1.0:14: A hypothetical 'pure' type B version of 德 'virtue' would have been

OC *tək > late literary CC *tɕɨk > SV *chức

with a type B palatal instead of a type A nonpalatal initial.

*4.1.1:50: MC *ə fronted in Wenzhou (E) and Shuangfeng (-iɛ) whereas it backed to o in Japanese and lowered to a in Cantonese and literary Colonial Chinese.

Shuangfeng has no simple -ɛ, so I presume

*-ək > *-ə > *-ɛ > -iɛ

A similar fronting of *ə occurred in the source of Chaozhou literary readings. Amoy tɪk 'virtue' is probably also from *tek. (There is no rhyme -ek in Amoy.)

I assume that Amoy colloquial tit < *tik and Chaozhou tik for 'get' are native Southern Min words. Their high vowel is reminiscent of the high vowel of Viet được < colloquial Colonial Chinese *dɨək 'get', though their tones point to earlier *t- rather than *d-.

*This term implies that OSV words are older than Sino-Vietnamese readings proper. Although many OSV words contain archaic phonological characteristics absent from SV proper, others like được are ambiguous and could be colloquial Colonial Chinese (CC; see above) borrowings that were coeval with literary Colonial Chinese character readings. I use the term OSV out of habit here, but perhaps I should use a more precise and age-neutral term like Colloquial Sino-Vietnamese (CSV). But even that term is hazardous because it's possible that some OSV words were originally also character readings that became obsolete for the literati who borrowed newer readings but persisted in everyday speech. A few SV readings have the same archaic features as OSV words and are probably old readings that were not (completely) displaced by later borrowings: e.g.,

義 'justice': SV nghĩa (cf. early MC *ŋɨeh); the hypothetical SV *nghị corresponding to late MC *ŋɨì doesn't exist)

基 'base': SV cơ (older reading; cf. early MC *kɨə), ki (rarer newer reading; cf. late MC *kɨi)

Last night, I wrote,

In the Middle Chinese dictionary and rhyme table tradition, 印 'seal' is homophonous with 因 'cause', etc. apart from tone. And in Cantonese and other Chinese languages, the two are still homophonous apart from tones ...

However, according to Hanyu fangyin zihui, they are not homophonous even when tones are disregarded in Wenzhou:

印 iŋ

因 iaŋ

Both are yin in standard Mandarin. Md yin generally corresponds to Wenzhou iaŋ. Why is 印 'seal' exceptional?

My initial guess was that 印 iŋ is a borrowing from another Chinese language. However, the rhyme inventory in Hanyu fangyin zihui (1964: 6) doesn't list -iŋ as a rhyme:

Wenzhou -ŋ rhymes

So I suspect 印 is actually iaŋ in Wenzhou.

Wenzhou does have a lot of unusual readings*, but I can't look into them now because I already have too many other topics I want to blog about. This impression isn't only mine:

It [Wenzhou] is situated in a very mountainous region and, as a result, has been isolated for most of its history from the rest of the country, making the local culture and language very different from those of neighbouring areas.

[...]

Wenzhou natives speak Wu Chinese, which is the spoken language of the people of neighbouring Hangzhou, Suzhou, and Shanghai. However, geographic isolation and an admixture of Southern Min Chinese speakers from nearby Fujian Province, have caused Wenzhou's spoken language to evolve into a dialect that has been described as "notoriously eccentric." As a result, people from all over China, or even other regions of Zhejiang and Fujian both have trouble understanding what is known as Wenzhouhua ("Wenzhou Dialect" Chinese: 温州话).

Wenzhou is not mutually intelligible with other varieties of Wu neighboring it to the north and west, let alone with Min Dong [Eastern Min] to the south or with the official language of China, Mandarin.

[...]

Due to its long history and the geographical features of the region on which it is located, Wenzhou Chinese is so eccentric in its phonology that it has the reputation of being the "least comprehensible dialect" for an average Mandarin speaker.

[...]

Due to its high degree of eccentricity, the language is reputed to have been used during the Second Sino-Japanese War during wartime communication. Due to its unique grammar, vocabulary, and pronunciation, the language is basically impossible for any non-local to understand.

There is a famous rhymed saying in China that goes, "Do not fear the heavens and the earth (i.e. anything), but be afraid of hearing a person from Wenzhou speak in their local tongue." (天不怕，地不怕，就怕温州人说温州话)

The rhyme actually works in Wenzhouhua!

怕 'fear': pho (cf. Md pa [pha])

话 'speech': ɦo (cf. Md hua [xwa])

Can you recognize these city names in Wenzhouhua?

paitɕiaŋ

ɦixe

ɕikɔ

Select the blank space below to see the answers.

paitɕiaŋ: Beijing

ɦixe: Shanghai

ɕikɔ: Hong Kong

*For instance, 温州 Wenzhou in Wenzhou is presumably ytɕiu without a nasal or even a nasal vowel. Suzhou uəntsʏ is closer to Mandarin Wenzhou [wəntʂow].

(I have constructed the names of cities in Wenzhou and Suzhou using Hanyu fangyin zihui. The actual city names in those languages may be different from their readings of their components in isolation.)

10.3.29.23:59: AN A-NH-OMALY EXPLAINED?

Last week I couldn't solve the e-nh-igma of the monocausal nasal: the nh- of Sino-Vietnamese

一/壹 nhất (also nhứt) 'one' (cf. Cantonese yat, Mandarin yi without an initial nasal)

因 nhân 'cause' and its homophones (cf. Ct yan, Mandarin yin without an initial nasal)

洇烟茵姻裀氤駰

垔堙湮

Last night, Guillaume Jacques told me about Michel Ferlus' solution:

Ruc is a small Vietic language that did not nasalize Proto-Vietic *ʔb- and *ʔd-.

I had known about the nasalization of preglottalized stops since the mid-90s, but I didn't realize that *ʔj- had also nasalized.

Ferlus' solution can be summed up as *ʔC_voiced- > N-.

One would expect all Middle Chinese *ʔj- to correspond to SV nh-, but there are exceptions. In the Middle Chinese dictionary and rhyme table tradition, 印 'seal' is homophonous with 因 'cause', etc. apart from tone. And in Cantonese and other Chinese languages, the two are still homophonous apart from tones, but they have different initials in SV:

張群顯 Cheung Kwan-hin (b. 1953) found a Cantonese reading for 印 I've never seen anywhere else: ngan, which regularly corresponds to SV ân. Cantonese ng- alternates with zero, so ngan implies an (extinct?) unrecorded *an corresponding precisely to SV ân. Perhaps southern late Middle Chinese had two morphemes for 'seal', *ʔjə̀n and *ʔə̀n. The second might be from an earlier southern MC *ʔɨinh < Old Chinese *ʔrins < *r-ʔins which once had an *r-prefix.

3.30.2:25: There are other morphemes which could be reconstructed with an SLMC *ʔj- not corresponding to SV nh-: e.g.,

One could reconstruct all of these morphemes with SLMC *ʔi- > SV [ʔ] instead of *ʔj- > SV nh, but then one would also have to reconstruct a dubious distinction between SLMC *ʔjə (as in 一 'one' and 因 'cause') and SLMC *ʔiə (as in 挹 'ladle out' and perhaps also 印 'seal'). SLMC *ʔj- would appear only before *-ən (in 因 'cause' and its homophones) and *-ət (in 一 'one'). Such a highly restricted distribution is highly unlikely.

10.3.28.23:59: NON-EMPHATIC ODDITIES IN (SINO-)VIETNAMESE

Not all unusual Chinese loans in Vietnamese look like type A/B mixtures.

1. Reversed type loans

Some Chinese morphemes once had both type A and B variants: e.g.,

瞢 'darkened, ashamed, despondent'

OC type A *Cʌ-məŋ > *məŋ > MC *məŋ

OC type B *məŋ > MC *muŋ

Mandarin meng could be from either variant.

揄 'pull towards oneself, scoop out'

OC type A *lo > MC *dəu > Md tou

OC type B *Cɯ-lo > *lo > MC *juo > Md yu

Perhaps this variation was more widespread in the past. Variants in spoken southern Chinese which were neither preserved in dictionaries nor in modern Chinese languages could have been borrowed by the Vietnamese: e.g.,

安 'peace'

OC type A *ʔan > MC *ʔan > Sino-Vietnamese an, Cantonese on, Mandarin an, etc.

OC type B *Cɯ-ʔan > MC *ʔien > Sino-Vietnamese yên (but modern Chinese type B forms like Cantonese yin, Mandarin yan, etc. do not exist)

A handful of reversed type variants can also be found in other languages, including Chinese languages: e.g.,

所 'place'

Mandarin type A suo, Cantonese type A so, etc.

(instead of the expected type B forms: Md *shu, Ct *seui, etc.)

Sino-Korean type A so

(instead of the expected type B *sŏ corresponding to Sino-Vietnamese sở and Sino-Japanese sho; 3.29.1:39: even the prescriptive Middle Sino-Korean dictionary Tongguk chŏng'un lists so instead of the expected *syŏ)

六 'six'

Thai type A hok < Late OC *houk < OC *hruk < *Cʌ-ruk

(instead of the expected type B luk < Late OC *luk < OC *ruk)

(I reconstruct LOC *h- instead of *x- because *x- would have been borrowed as Proto-Tai *x- and *x- backed to *h- in southern Chinese.)

2. Ablaut loans

Old Chinese had vowel alternations in roots: e.g.,

瞢 OC *məŋ 'darkened, ashamed, despondent'

冥 OC *meŋ 'dark'

仆 OC *phəks, *N-phək, *phoks, *phoks 'fall prostrate' (with type as well as vowel alternation!)

(cognate to 伏 OC *bək < *N-phək? 'lie down')

These alternations are not yet understood. As with syllable type variation, vowel variation in roots may have been more common in the past, and Vietnamese may preserve vowel variants that were lost elsewhere.

合 'unite' has two Sino-Vietnamese readings:

hợp < southern early MC *ɦəp < OC *gəp

(Southern late MC *ɦap would have been borrowed as SV *hạp.)

hiệp < southern late MC *ɦiep < OC *gep, *gep, or *gap

(The latter two would have become type B due to a lost high vowel prefix: *Cɯ-gep, *Cɯ-gap > *gep, *gap.)

SV hiệp seems to reflect a descendant of an OC *-e- or *-a- variant absent elsewhere. There are no forms like Cantonese *hip, Mandarin *xi, etc. from MC *ɣiep.

3. Loans from different chronological strata

Some vowel alternations in Vietnamese simply reflect changes in vowels in southern Chinese during the centuries of Chinese rule over Vietnam: e.g.,

心 'heart'

tâm < southern late MC *səm (cf. Cantonese sam)

tim < southern early MC *sim

會 'group, society'

hội < southern late MC *ɦòi

hụi < southern post-MC *ɦùi 'group, society' (cf. Cantonese wui in which *ɦ- has been lost before *u)

3.29.1:09: 4. Loans from different dialects

All of these Sino-Vietnamese readings have an initial ph- indicating that they were borrowed from Late Middle Chinese which shifted labial stops to *f- (whose closest early Vietnamese counterpart was an aspirated stop *ph- that later lenited to [f]). So they may be of roughly equal age, even though they have different reflexes of OC *-ə.

福 'happiness'

phúc < southern late MC dialect 1 *fuk < LOC dialect 1 *puk < OC *pək

phước < southern late MC dialect 2 *fɨək < LOC dialect 2 *pɨək < OC *pək

鳳 'phoenix'

phụng < southern late MC dialect 1 *fùŋ < LOC dialect 1 *buŋh < OC *brəms

phượng < southern late MC dialect 2 *fɨə̀ŋ < LOC dialect 2 *bɨəŋh < OC *brəms