08.3.15.23:59: 東冬鐘江 EASTERN WINTER ON BELL RIVER (PART 5)

The following chart shows the interrhyming patterns of 'east' and 'winter'-class words in several late Old Chinese texts using Starostin's (1989) analysis. All reconstructions are 'generic' early OC. No attempt has been made to adjust them to account for interrhyming.

These patterns differ quite strongly from the rhyme categories of Qieyun (601 AD), the 同用 tongyong (interrhymable; lit. 'identical-use') categories of Kanmiu buque Qieyun (706 AD; see Pulleyblank [1984: 140]), and the 攝 she rhyme categories (used since at least the 9th century AD; see Pulleyblank [1984: 70]):

This suggests that the emphatic-driven vowel bending which altered the OC 'east' and 'winter' rhyme categories beyond recognition occurred at a very late date.

3.16.14:01: I was reluctant to upload this post last night because it was lacking in analysis.

It's dangerous to draw strong conclusions because all this data reflects various varieties of Old Chinese which were not ancestral to each other. If variety X is older than variety Y, it does not meant that Y is actually descended from X. In many cases, Y is the 'niece' rather than the 'daughter' of X. Moreover, there is no guarantee that all these poets had similar rhyming criteria in mind.

However, even if the relationship between the OC varieties in Starostin's survey is unclear, a few things are clear:

- There was a dichotomy between the OC 'east' and 'winter' categories that persisted as late as the 5th century AD. But this dichotomy was not airtight. In Qieyun, MC reflexes of OC *-oŋ and *-uŋ were placed in one rhyme category, and interrhyming between OC *-oŋ and *-(r)uŋ can already be found in the 5th century.

- Vowel allophony conditioned by (non)emphasis had no effect on rhyming within the 'east' catetgory as late as the 5th century AD.

(It's hard to say whether such allophony had any effect on rhyming within the 'winter' category because there are no examples of emphatic 'winter' *-(r)ung rhyme words after Qu Yuan in Starostin's survey.)

- Medial *-r- and its later reflex(es) (e.g., vowel retroflexion) had no effect on rhyming as late as the 5th century AD. But by the 9th century, words which once had emphatic *-r- had their own she. And to this day, the 'river' she is still distinct from the 'penetrate' she:

Old Chinese	*-oŋ, *-uŋ	*-(r)oŋ, *-(r)uŋ	*-roŋ, *-ruŋ
Late Middle Chinese she	'penetrate'		'river'
Middle Chinese grade	I	III	II
Mandarin	-ong	-(i)ong	-iang
Cantonese	-uŋ		-ɔŋ

(For simplicity, I have ignored MC Grade IV.)

08.3.13.23:59: 東冬鐘江 EASTERN WINTER ON BELL RIVER (PART 4)

In Daodejing, 東 *-oŋ 'east' words rhyme with 陽 *-aŋ 'yang (male principle)' words (Starostin 1989: 601). Starostin's reconstruction cannot account for the discrepancy in vowels: e.g., he reconstructed two rhyme sequences as

功 *kooŋ : 長 *draŋ

行 *graaŋ : 重 *dhroŋ

My pre-Daodejing reconstructions are not very different:

功 *koŋ : 長 *Ntraŋ

行 *rgaŋ : 重 *Nrtoŋ

There are at least four ways in which 'east' and 'yang' could have been phonetically more similar in the Daodejing language:

1. *o lowered to *ɔ or even *ɒ

*-ɔŋ / *-ɒŋ : *-aŋ

2. *o broke to *wa

*-waŋ : *-aŋ

3. *o broke to *aw

*-awŋ : *-aŋ

4. *a rounded to *ɒ (and raised to *ɔ?)

*-oŋ : *-ɒŋ / *-ɔŋ

A change like solution 1 occurred in the emphatic 'east' rhyme in Xiang and Min: e.g.,

東 'east': OC *toŋ, Shuangfeng Xiang taŋ, Chaozhou Min taŋ

通 'penetrate': OC *lhoŋ : Shuangfeng Xiang thaŋ, Xiamen Min thaŋ

This does not necessarily mean that Xiang and Min are descended from the Daodejing language, or that they should be subgrouped together.

Norman's Proto-Min reconstruction in Schuessler (2001 ms.) has *-oŋ for the emphatic 'east' rhyme. Norman must believe that Fuzhou -øɥŋ and Chaozhou -aŋ are innovations. Perhaps they developed from *-əwŋ:

*-oŋ > *-əwŋ > *-ewŋ > *-øwŋ > Fuzhou -øɥŋ

*-oŋ > *-əwŋ > *-awŋ > Chaozhou and Xiamen -aŋ

Solution 2 recycles a change known to occur before dental codas:

*-on/t > *-wan/t

*-awŋ in solution 3 is reminscent of Pulleyblank's (1978) reconstruction of the 'east' rhyme as *-aŋ^w.

Solution 4 is reminscent of the literary layer of Xiamen in which descendants of the 'east' and 'yang' categories rhyme (if tones are ignored):

通 'penetrate': OC *lhoŋ : MC *thoŋ : Xiamen lit. thɔŋ1 (colloq. thaŋ)

唐 (a name): OC *laŋ : MC *daŋ : Xiamen lit. thɔŋ2 (colloq. thŋ)

(*Voiceless initial syllables developed tone 1 and *voiced initial syllables developed tone 2.)

The source dialect of the literary borrowings must have raised and rounded *a to *ɔ.

I don't know which solution is correct. All I can say is that

- 'east' and 'yang' interrhyming may have been a feature of the dialect spoken in the territory of the former Chu state during the Former Han (Luo and Zhou 1958: 81-2, cited in Coblin 1983: 113)

- yet such interrhyming is absent from the poetry of Qu Yuan of Chu (see the data in Starostin 1989: 615; Qu interrhymed 'east' with 'winter' and interrhymed 'yang' with *-am and *-əŋ once each), implying that the 'east'/'yang' merger postdated the fall of Chu (and the Daodejing)

- the two categories may have merged in Xu Shen's 2nd century AD dialect and Fu Qian's 3rd century AD dialect (Coblin 1983: 113); Xu "was from the northern part of the Chu area", whereas Fu was from Yingyang (now Zhengzhou), east of Luoyang and north of Chu, so the merger was not exclusively a Chu dialect feature.

08.3.12.23:59: 東冬鐘江 EASTERN WINTER ON BELL RIVER (PART 3)

In this post, I'll start to trace the evolution of the Shijing rhyme categories

東 *-oŋ 'east' (> MC 東 'east A', 鐘 'bell', 江 'river' categories)

冬 *-uŋ 'winter' (> MC 東 'east B', 冬 'winter', 江 'river' categories)

by looking at rhyme patterns in later Chinese texts using the data collected by Starostin (1989).

Others have reconstructed those two categories quite differently. Perhaps the most unusual proposals have been Pulleyblank's:

Although most researchers today reconstruct six vowels for Old Chinese, Pulleyblank had been advocating a two-vowel system for Old Chinese since 1963. He compensated for the lack of vowels by proposing various phonemic final consonants that influenced preceding vowels: e.g., */əŋ^ɥ/ (sic*) was phonetically *[øɥŋ] (1991: 49).

Pulleyblank cited the Fuzhou colloquial form 東 tøɥŋ 'east' as evidence for final palatolabiovelars in OC:

Since this [rhyme -øɥŋ] is found only in the colloquial layer, there is a good chance that it is a genuine archaic survival and not the result of a more recent innovation (1991: 48).

Pulleyblank's 1991 reconstruction cannot neatly account for the interrhyming between 'east' and 'winter' which would have different vowels and final consonants. Perhaps poets who rhymed these categories retained earlier *-əŋ^ɥ and rhymed only *-əŋ^ɥ 'winter' words with *-aŋ^ɥ 'east' words. But interrhyming between *-oŋ and *-uŋ is much simpler, and it even occurred in MC.

However, *-oŋ and *-uŋ-type reconstructions cannot easily account for two other types of interrhyming

- between *-uŋ 'winter' and 蒸 *-əŋ 'steam' (only one instance)

- between *-uŋ 'winter' and 侵 *-əm 'invade'

In Pulleyblank's reconstructions, these rhymes share vowels and labial codas:

- between *-əŋ^w/*-əŋ^ɥ 'winter' and 蒸 *-əŋ 'steam' (only one instance)

- between *-əŋ^w/*-əŋ^ɥ 'winter' and 侵 *-əm 'invade'

(Pulleyblank's reconstructions of the 'steam' and 'invade' categories are identical to mine.)

But workarounds are possible even in a six-vowel system with fewer codas. Perhaps such interrhyming reflected dialects in which

蒸 *-əŋ 'steam' had a high vowel [ɨ] like 'winter'

but this is unlikely since the 'steam' word in question was 弘 *gwəŋ which had an emphatic (and hence lowered) vowel allophone [ʌ]

侵 *-əm 'invade' had a rounded high vowel [u] like 'winter'

and *-uŋ 'winter' was [uŋ͡m] like modern Vietnamese -ung

hence Thompson (1987: 30) warned against confusing Viet -um with -ung

but see Sagart's (1999: 52-54) arguments against OC *-um

In the Daodejing (4th-3th c. BC), *-uŋ 'winter' words rhyme only with each other. *-oŋ 'east' words, on the other hand, interrhyme with words of a surprising category that I'll reveal next time.

*I wonder if Pulleyblank meant to write aŋ^ɥ. His OC *ə-rhymes correspond to high vowels in others' reconstructions. I would expect his OC */əŋ^ɥ/ to have been pronounced as *[yŋ]. He noted that Fuzhou -øɥŋ corresponded to his OC */-aŋ^ɥ/ with a low vowel (1991: 48).

08.3.11.23:59: 東冬鐘江 EASTERN WINTER ON BELL RIVER (PART 2)

These four Middle Chinese rhymes have bothered me for about 15 years. Even now I am not quite sure how to reconstruct them. The evidence is difficult to interpret. Moreover, I don't think there is a single right answer, because 'Middle Chinese' is a generic term for a variety of dialects over a number of centuries. Nonetheless, I would like to know what phonetic values underlie the (composite?) prescriptive dialect in the Qieyun (601), and how those values were interpreted four centuries later when the Qieyun-like Tangraphic Sea was compiled.

What follows is my attempt to try to trace the history of those rhymes. Much of this is based on the work of Pulleyblank, Schuessler, and Norman, so I make no claims to originality. Sometimes I understand events better if I describe them to myself step by step.

In the Old Chinese of the Shijing, there were two rhyme categories:

東 *-oŋ 'east'

冬 *-uŋ 'winter' (also called 中 'center')

(These names and categories were conceived in recent centuries. Unlike the MC categories, they do not represent native speakers' categorizations. Thus in theory, the OC categories could be wrong, though most agree with the basic scheme. However, there is little agreement on how to reconstruct the categories.)

Emphasis and medial *-r- were ignored in rhyming: e.g.,

從 *dzoŋ rhymed with 東 *toŋ

窮 *guŋ rhymed with 冬 *tuŋ

螽*tuŋ (with 冬 *tuŋ 'winter' as phonetic) rhymed with 降*Nkruŋ

Cf. nonemphatic/emphatic interrhyming in this Egyptian Arabic poem (go to p. 263):

...ya : ...qa

...taal : ...taal

(I have added emphasis to the segments after q and t based on my understanding of emphasis spread in Cairene Arabic as described by Islam Youssef.)

There are instances of interrhyming between the *-oŋ and *-uŋ categories: e.g.,

龍 *roŋ : 童 *doŋ : 充 *thuŋ

Such interrhyming may reflect

- a merger of the two categories in a poet's dialect

- phonetic similarity of the two categories in a poet's dialect

- different rhyming conventions in the poet's dialect

In any case, 'east' and 'winter' must have been similar in OC, because nonemphatic *-uŋ 'winter' syllables moved into the 'east' category of MC:

This chart is somewhat misleading for two reasons:

First, the grade system was designed for late MC, not the early MC of the Qieyun. Nonetheless, the conditioning factors for the later grade system must have been present in the Qieyun language. Early MC syllables were not distributed among the four late MC grades at random.

Second, if one examines the fanqie of the entries for 'east' in the MC rhyme dictionary Guangyun, based on Qieyun (whose earliest version has been lost), one sees that the original 'east' words and the former 'winter' words are in separate homophone categories and different final spellers: e.g.,

公 'duke': MC *koŋ < OC *kloŋ (original 'east')

fanqie: 古紅 MC *ko + *ɣoŋ < OC *kaʔ + *goŋ

弓 'bow': MC *kuŋ < OC *kuŋ (original 'winter')

fanqie: 居戎 MC *kɨə + *ɲuŋ < OC *ka + *nuŋ

'Duke' and 'bow' should have been in the same homophone group if there was no split within the MC 'east' category. I will divide MC 'east' into 'east A' (from OC 'east') and 'east B' (from OC 'winter').

This split is not merely an artifact of fanqie, as it can be confirmed by examining MC-period loans into foreign languages. Vietnamese, Korean, and Japanese generally have different rhymes for those two subtypes of MC 'east':

̣(The above chart glosses over complexities in the various Sinoxenic strains.)

In Sinoxenic, 'duke' and 'bow' are not homophonous:

Conversely, in general Sinoxenic treats MC 'east' and 'winter' identically, just like all modern Chinese languages. Was the Qieyun language an aberrant dead end which left no descendants? Do Sinoxenic and modern Chinese languages reflect MC dialects which fused the OC categories differently? The late MC grade system seems to apply more neatly to such dialects:

Next: Where did the MC 'river' category come from?

(3.12.0:21: Added Hakka and Amoy. Introduced terms 'east A' and 'east B'.)

08.3.10.23:59: 東冬鐘江 EASTERN WINTER ON BELL RIVER (PART 1)

Some believe that the Tangraphic Sea was modelled after Middle Chinese rhyme dictionaries like Qieyun and Guangyun. The first four rhymes in those dictionaries are in the title. Nishida (1964: 42-43) made the following equations:

(For simplicity of comparison, I am ignoring MC Grade IV.)

Nishida did not equate the other four members of his first rhyme group with MC rhymes:

Nishida did not reconstruct a -jʊ.

The grades of these MC rhyme are in the order I, III, II. I forgot to mention this last night as an argument for interpreting R4 as the missing Grade II -u^ɣ:

I split 'east' (MC 'R1') in two to fit the Tangut better.

See "Return to Rhyme 4" for arguments against identifying R4 as Grade II.

If R2 was Grade II, the grade order for the first rhyme groups of MC and Tangut would no longer match:

MC: I, III, II

Tangut: I, II, III, I

Currently, I don't think it's possible to directly map MC and Tangut rhymes onto each other. Although I don't think it's a coincidence that both rhyme lists begin with u-type rhymes* followed by i-type rhymes, the Tangut recognized that their phonology did not match Chinese phonology: e.g., the three** cycles of the Tangraphic Sea for plain, tense, and retroflex vowels have no parallel in the Chinese rhyme dictionary tradition.

*Transcriptions of late pre-Tangut northwestern Chinese indicate that the 'east' rhyme had become *-uŋ.

**08.3.11.0:05: Like Arakawa (1999), I don't think Sofronov's (1968 I: 138) fourth cycle is necessary.

08.3.9.23:33: RETURN TO RHYME 4

Earlier this year, I was thinking that R4 was the missing Grade II -u rhyme that would enable me to claim that all Tangut vowels could occur in Grade II. However, if R4 is Grade II, it appears in an unexpected location in Tangraphic Sea. In the Sea, rhymes are generally arranged in sets sharing the same vowel. Within each set, rhymes are arranged according to grade: I, II, III. One would expect the -u set to look like this:

However, according to Gong, the -u set actually contains four rhymes and has no grade II rhyme. The set is the only one that ends in Grade I:

Gong's reconstruction did not distinguish between R1 and R4 on the one hand and R2 and R3 on the other. Thus his reconstruction cannot distinguish between minimal pairs such as

TT4139 THEREFORE ku R1 1.1

TT5761 PHOENIX ku R4 1.4

TT0190 OX/ELEPHANT bju R2 1.2

TT4614 ACCORDING-TO bju R3 1.3

Earlier this year, I thought that R2 could have been Grade II:

R2 is a better candidate than R4 for three reasons:

1. R2 is in the expected position for a Grade II rhyme. Compare the Grade II R2 table with this Grade II R4 table:

2. Dental stop initials are very rare in Grade II. R4 has six du 1.4 and six du 2.4, whereas R2 has no dental stops at all.

3. Many Grade II syllables have palatal initials (which probably originated from earlier dentals and alveolars). R2 has palatal initials whereas R4 has no palatal initials.

However, R2 fails to fit the profile of a Grade II rhyme in at least two ways:

1. R2 was transcribed with a mixture of Grade I and III sinographs. This is not a major obstacle because there probably were no Grade II -u sinographs.

2. Alveolar initials are absent from Grade II, yet there are three instances of dzju (?dzu^ɣ) R2 1.2. Is it a coincidence that R2 and R4 both have d(z)-initials that seemingly disqualify them from being classified as Grade II? I suspect that the total absence of all other dental and alveolar initials in those rhymes is somehow significant Perhaps d(z)- in R2 and R4 (and the rare t- and d- in some Grade II rhymes) originated from clusters with n- and l(h)- which are the only common dental and alveolar initials in Grade II:

pre-Tangut simple initials *t-, *ts-, etc. > Grade II tɕ-, etc.

pre-Tangut complex initials *lC-, Cn-, etc. > Grade II t-, d-, dz-?