09.12.11.11:55: CAN YOU SOLVE THE EQUINE EQUATION? (PART 2)

Here's the solution to yesterday's problem:

All four Proto-ABC clusters have to have an r- since B always has r-.

Language A has preserved the first consonant in a cluster:

nr- > n-

mr- > m-

dr- > d-

br- > b-

Language B has preserved the last consonant of a cluster:

nr-, mr-, dr-, br- > r-

Language C has preserved the middle consonant of a cluster:

nr- > ndr- > dr- > d-

mr- > mbr- > br- > b-

Unlike language C b-, Taiwanese b- is from mb- < m- < mr-, not mbr- < mr-.

Japhug rGyalrong has mbro from Proto-rGyalrong mr-. Written Burmese mrang preserves the original cluster.

This mr-word for 'horse' may have been borrowed from the source of Mongolian morin and Korean mal < mʌr. I don't know whether this word was borrowed by Proto-Sino-Tibetan or by its later daughter languages (e.g., Old Chinese, the ancestor of Mandarin and Taiwanese).

During my flight from Philadelphia to San Francisco, the following linguistic problem occurred to me. It's not very hard, but I wonder if nonlinguists can solve it.

Suppose there are three hypothetical languages A, B, and C.

In most cases, their initial consonants match perfectly: e.g., if a word begins with n- in Language A, it also begins with n- in B and C. In such cases, it is safe to assume that this shared consonant was inherited intact from Proto-ABC, the common ancestor of all three languages.

Table 1. Perfectly preserved Proto-ABC consonants

However, there are cases in which the consonants don't match: e.g.,

Table 2. Imperfectly preserved Proto-ABC consonants

How would you fill in the last row of Table 2? In other words, what would you reconstruct as the sources of the consonants of Languages A, B, and C listed in that table?

Hint 1: You can't reconstruct any of the consonants in the Proto-ABC row of Table 1. Proto-ABC n- always became n- in the daughter languages, m- always became m-, etc.

Hint 3: r has little in common with m or b (apart from being a voiced consonant).

Last Saturday, I wrote,

I followed Schuessler (2009: 104) by reconstructing 圮 as LOC [Late Old Chinese] *bɨəʔ. However, I now realize this was a mistake. In Middle Chinese [MC], 圮 was *bɨiʔ which has two possible sources, LOC *bɨəʔ and *bɨəjʔ. The second LOC reconstruction must be the correct one since it has the *-j found in 非 *pɨəj and 配 *phəjh [the phonetics of 䤏, a variant of 圮] (and 妃 LOC *pɨəj sharing the same phonetic [as 圮]).

I then traced 圮/䤏 LOC *bɨəjʔ back to Early Old Chinese EOC *brəjʔ (as opposed to Schuessler's EOC *brəʔ without *-j).

That got me thinking about the history of *rə(j)-type rhymes in Chinese. Here's my current reconstruction of their development from LOC to MC. I have arbitrarily assigned letters to each rhyme type and left out some conditioned reflexes: e.g., EOC *-ə > MC *-u after labials.

A. EOC *-i > MC *-i (脂 rhyme class; Grade IV)

B. EOC *-ri > MC *-ɰi (脂 rhyme class; Grade III)

C. EOC *-ə > *-ɨə > MC *-ɨ (之 rhyme class; Grade III)

D. EOC *-rə > *-rɨə > *-ɰə > MC *-ɰi (脂 rhyme class; Grade III)

E. EOC *-əj > *-ɨəj > MC *-ɨj (微 rhyme class; Grade III)

F. EOC *-rəj > *-rɨəj > *-ɰəj > MC *-ɰi (脂 rhyme class; Grade III)

And here's Schuessler's (2009: 91, 275) reconstruction for comparison:

A. EOC *-i > LOC *-i

B. EOC *-ri > LOC *-ɨ

C. EOC *-ə > LOC *-ɨ

D. EOC *-rə > LOC *-ɨ

E. EOC *-əj > LOC *-ɨi

F. EOC *-rəj > LOC *-ɨ

Schuessler's reconstruction is simpler and hence initially more appealling, but it doesn't account for the fact that in Middle Chinese rhyme dictionaries, A, B, D, and F belonged to a rhyme class (脂; my *-[ɰ]i) distinct from those of C (之; my *-ɨ) and E (微; my *-ɨj). Schuessler's reconstruction may, however, account for the rhyme categories of some non-dictionary Middle Chinese dialect(s).

Next: Do my intermediate stages match the rhyme categories of Late Old Chinese poetry?

09.12.7.23:21: HOLE EYE TREE SUN

In my last post, I mentioned

目+窅 Early Old Chinese *ʔwit ~ *kwit 'depth of eyes'

(I discovered the second reading tonight.)

According to Schuessler (2009: 173), its phonetic is

窅 EOC *ʔiw (*ʔiû with a final *-u in Schuessler's 2009 reconstruction) 'retired, despondent'

which looks like 穴 'hole' over 目 'eye'.

窅 'retired, despondent' is an extended usage of 杳 EOC *ʔiw 'dark' with a different spelling. 杳 is written as 木 'tree' over 日 'sun'

窅/杳 are cognate to 幽 EOC *ʔiw 'dark' (sans emphasis).

Schuessler proposed that an *-iu phonetic for a *-uit syllable was parallel to a cognate pair he proposed in his ABC Etymological Dictionary of Old Chinese (2007: 397):

餒 EOC *nûiʔ (*nujʔ in my reconstruction) 'hungry'

惄 EOC *niûk (*niwk in my reconstruction) 'hungry'

But I wonder if 窅 is semantic rather than phonetic in 目+窅. That which is deep may also be dark. Shuowen does not identify 窅 as a phonetic in 目+窅.

目+窅 also has a variant 目+穴 with 穴 EOC *wit 'hole'. Perhaps 目+窅 EOC *ʔwit ~ *kwit 'depth of eyes' is a derivative of 穴 EOC *wit 'hole'. Deep eyes are like eyeholes. 穴 may be phonetic as well as semantic in 目+窅 and the second 目 under 穴 may be superfluous.

Shuowen states that 目+窅 is read like 卹 which has a number of Middle Chinese readings according to Jiyun:

*swit < EOC *sut

*sot < EOC *sut

*ʂit < EOC ?*r-sut (the lack of a *-w- in MC is irregular)

Coblin (1983: 177) reconciled the initials of 目+窅 and 卹 by reconstructing them in Late Old Chinese as *sʔiwət and *sjwət which would be equivalent to *sə-ʔwet and *swit in my reconstruction.

Perhaps 目+窅 had at least two EOC readings with different prefixes:

*sʌ-ʔʌ-wit > *sʌ-ʔwit > LOC *sə-ʔwet > MC ʔwet

*kʌ-(ʔʌ)-wit > *kwit > LOC *kwet > MC *kwet

In my last post, I proposed that 圮/嶏 'collapse' (Early Old Chinese root *phəj) and 破 'break' (EOC root *phaj) were cognates. If they shared a consonantal root *ph-j, could this root have a zero grade derivative *phi < *ph-j?

批 < 扌 + 比

EOC *phi 'beat, slap', written as 扌 'hand' (semantic) + 比 EOC *piʔ 'compare' (phonetic)

is a perfect phonetic match for the *phi I am looking for, but its semantics leave something to be desired. 'beat, slap' does not entail destruction.

Although I doubt 批 has anything to do with 圮/嶏 and 破, it is stil relevant for the section of my last post about *-t-suffixation because it has a probable derivative

EOC *bit < ?*Nʌ-phi-t 'knock against'

written with the same graph. How many other cases of -V ~ -Vt alternation can I find?

I did a quick search for -t syllables written with open syllable phonetics in Schuessler (2009) and found:

There are no cases of open-syllable phonetics representing syllables ending in *-p or *-m. This implies that there were no *-p or *-m suffixes or, conversely, no (irregular) *-p or *-m loss. (I should look at stop and nasal-final phonetics to look for open-syllable readings.)

Yesterday afternoon, I thought that it might be possible to claim that there was a pre-EOC suffix *-k which remained intact except after *i: *-i-k > *-it. This would explain why there are no *-k syllables with *-i phonetics.

However, I don't know how to explain the *-et, *-at, and *-ot syllables with open-vowel phonetics. Are they simply random exceptions?

The phonetic similarity of *-(w)ʔ and *-(w)k may have led scribes to write both *-V(w)ʔ and *-V(w)k syllables with the same phonetics. I predict that a predominantly open-syllable phonetic which has no readings ending in glottal stop never has any readings in *-k. I can't think of any counterexamples offhand.

Zero or *-j ~ *-n alternations may be varying reflexes of an original *-r: e.g.,

哂 *hnirʔ > *hninʔ (whereas its phonetic 西 has a non-*n reflex: *snər > *snəj > *sni)

短 *torʔ > *toiʔ in Proto-Min in the south but *tonʔ in northern Chinese

枅 *ke ~ *ken < *ker

憚 *tajs ~ *dans

頎 *gəj ~ *khənʔ

錞 *dujʔ/s ~ *dun

Note that reading alternations do not imply cognate status. Although 枅 *ke ~ *ken are variants of a single *ker 'crosspiece of wood on top of pillar',

憚 *tajs 'fear'

憚 *dans 'exhausted'

are probably unrelated, though I would reconstruct their earliest forms as *tars and *dars with *-r. (Oddly, their phonetic 單 has derivatives which end in *-m in Cantonese: e.g., 蟬 'cicada' and 禪 'Zen' [< Pali jhaana without m] can be read as sim as well as sin.)

I used to explain *Cə ~ *Cəŋ alternations as reflecting an earlier zero grade-schwa grade alternation: *Cŋ ~ *Cəŋ. But that does not account for zero ~ *ŋ alternations before nonschwa vowels. Perhaps the *-Vŋ forms are original and the nasalless forms are irregular reductions:

*-Vŋ > *-Ṽ > *-V

Since there is nothing that defines *a and *u as a set distinct from all other vowels, I assume the absence of

*-a ~ *-aŋ

*-u ~ *-uŋ

alternations is due to chance (or that word families with such alternations were written with different phonetics).