12.2.4.23:59: PRECIOUS PETROS (PART 2)

Right after I posted part 1, I got an insight that made me glad I stopped where I did. Hmmm, maybe this post would be even better if I stopped right here. Nah ...

Let's go through a long chain of reasoning to figure out how Paekche 珍惡 'stone' was read.

1. How do we know that 珍惡 is the Paekche word for 'stone' as opposed to, say, 珍 precious 惡 evil?

Because the Paekche place name 珍惡山 '珍惡 Mountain' was renamed 石山 'Stone Mountain' in the Unified Shilla period.

Assumption: 石 'stone' is a translation of Paekche 珍惡.

However, there is no guarantee that two names for the same place mean the same thing: e.g., 江戸 Edo 'River Door' does not mean 東京 Tokyo 'Eastern Capital' and York in New York is not an English 'translation' of Amsterdam in Nieuw-Amsterdam.

Furthermore, even if 石 'stone'  were intended to be a translation, there is no guarantee that the translation was accurate. Maybe the Paekche name sounded like the Shilla word for 'stone' but meant something else. Or the Shilla translator confused a Paekche word for 'stone' with an unrelated (near-)homophone 珍惡.

An extreme skeptic might say there is no way to reconstruct Paekche at all. But let's assume there is and keep going anyway.

2. How do we know how 珍 was pronounced?

Assumption: 珍 had a single sound value (or similar sound values) in the following variant spellings. (Numbers are from Ryu Ryŏl 1983.)

307. 難珍阿 ~ 難鎭阿 ~ 月良

311. 武珍 ~ 茂珍 (both modern Kwangju) ~ 瑞石 (a county name) ~  無等 (a mountain near Kwangju)

Assumption: These names reflect the same Paekche word, even though they don't refer to the same place.

332. 丘斯珍兮 ~ 貴旦 ~ 珍原 (珍 followed by its translation 原?)

401. 馬珍 ~ 馬突 ~ 馬等良 ~ 馬靈 (also cf. 394. 月奈 ~ 靈巖)

Let's focus on 珍 and its alternants:

The common denominator is T-T: a coronal onset and a coronal coda.

Assumption: Like other Altaic languages, Paekche did not have

initial clusters like *tr- or *rt-

retroflexes like *ʈ-

a three-way distinction between initial *t-, *th-, and *d-

Problem: The vowels after *t- have nothing in common:

These vowels could be grouped into three categories:

I (palatal): i

A (achromatic): ɯ, ə, ʌ, a

U (labial): o

I am reminded of East Slavic correspondences and alternations between all three types of vowels:

Ukrainian i  : Russian o: e.g., U ніч : R ночь

Ukrainian e  : Russian o: e.g., U пес : R пёс

Russian unstressed [ɪ] ~ stressed [o]: e.g., звезда ~ звёзды

Ukrainian o : Russian unstressed [ə] ~ [ɐ] (still spelled о): е.g., U голова [ɦɔlɔˈwa] : R голова [ɡəlɐˈva]

Ukrainian and Russian are closely related languages. Yet imagine how difficult it would be to figure out the history of their vowel systems if we only had Chinese character transcriptions instead of historical spellings. Were Paekche and Korean as close as Ukrainian and Russian or further apart?

I don't know. I do know that

- Middle Chinese had no syllables like *TəT, *TʌT, *TɯT

- Middle Chinese had no syllables ending in *-r

so 珍鎭 MC *ʈin (there was no MC *tin), 旦 MC *tanh, and 突 MC *thot ~ *dot might have been attempts to transcribe Paekche syllables like *tər, *tʌr, and/or *tɯr. I will symbolize this range of possibilities as *tƏr, which nicely matches Starostin's reconstructed Old Chinese reading *tər for 珍.

(2.5.8:32: Although there are indications that the Chinese dialect[s] known to peninsular peoples in this period contained archaisms, I doubt that 珍 'precious' was still *tər in those dialects. If the Paekche could transcribe their native *tƏr with such a close or even perfect match, why did they also transcribe that syllable with graphs for Chinese non-*tər syllables? The diversity of spellings suggests the absence of a good match in Chinese.)

The variety of MC vowels transcribing Paekche achromatic (i.e., neither palatal nor labial) vowels is reminiscent of the five Tibetan vowels (a, i, u, e, o) transcribing Tangut ə. Tibetan i and a transcribed Tangut ɨə (= jɨ in Gong's reconstruction).

Middle Korean (MK) cognates point to a final *-r. If the Paekche syllable ended in *-n or *-t, I would expect only MC *-n or *-t graphs, not a mix of both.

The alternation

石 'stone' ~ 等 'rank'

suggests that the Paekche cognates of MK torh 'stone' and tʌrh 'plural suffix' were (nearly) homophonous.

Assumption: MK achromatic vowels can be projected onto Paekche.

(I could be wrong. It's possible that Paekche had only one or two vowels corresponding to MK ɯ, ə, and ʌ.)

Conclusion: The MC reading *ʈin of 珍 doesn't have a labial vowel, so 珍 transcribed a Paekche *tʌr with a nonlabial vowel rather than a Paekche *tor with a labial vowel.

MC 惡 *ʔak ~ *ʔoh alternates with MC 約 *ʔɨək in Shilla (798). 惡- in the Old Korean (i.e., late Shilla) locative postposition corresponds to a(h)- in later idu locative postpositions postdating *k-lenition in Korean. The Sino-Korean (i.e., late Sino-Shilla) reading of 惡 is ak. Therefore 惡 represented Shilla *ak.

This Shilla *ak is similar to the Sino-Japanese reading aku which was borrowed from Paekche.

Assumption: The Paekche and Shilla readings of 惡 were similar or identical.

(2.5.8:25: I could be wrong. Although I'm certain that 惡 was read as *ak in Sino-Paekche, that does not necessarily mean that 惡 represented *ak in transcriptions of native Paekche words: cf. how 惡 is read as aku in Sino-Japanese but also represents native Japanese waru 'bad'.

惡 could have been read as *o [cf. its other MC reading *ʔoh] or like native Paekche morphemes corresponding to MC 惡 *ʔak 'evil' and/or MC 惡 *ʔoh 'to hate'.

But I choose a reading with a final *-k that corresponds to the final -h of MK torh.)

Conclusion: 惡 was read as *ak in Paekche as well as Shilla.

4. Paekche 珍惡 'stone' was *tʌrak (or to be safer, *tƏrak with an uncertain achromatic mid vowel).

珍惡 is a CC (Chinese-Chinese) phonogram sequence.

Unresolved problems:

If both words go back to a Proto-Koreanic *torak, and if the pre-Paekche form were *torák with an accented second vowel, the unaccented vowel could have been reduced to a schwa-like vowel. But if *a were accented, why was it lost in the Korean line?

2. What was the Paekche word for 'spirit' (or whichever meaning of 靈 was intended)? The alternation

靈 ~ 月

makes me think it was close to MK tʌr 'moon'. But I think it was disyllabic because of the alternation

靈 ~ 等良

良 is a phonogram for Old Japanese ra. Since the Japanese learned writing from the Paekche, it's possible that 良 represented Paekche *ra. If Paekche had a cognate of MK tʌrh 'plural suffix', perhaps 等 had a similar reading in Paekche and 等良 <tʌrh.ra> stood for *tʌra 'spirit'. But why was 'spirit' also spelled as 珍 with final *-n and 突 with final *-t?

2.5.00:05: 3. The alternation

難珍阿 ~ 難鎭阿 ~ 月良

with final 阿 *ʔa and 良 *ra leads me to reconstruct the Paekche word for 'moon' as disyllabic *tʌra, a homophone of 'spirit'. But what is the 難- doing at the beginning of the first two spellings? It corresponds to zero in 月良 <MOON.ra>.

2.5.9:05: The Middle Chinese reading *nan(h) of 難 resembles Korean na-n 'appear-MODIFIER' and Korean na-l 'appear-PROSPECTIVE'. The latter is even closer to Old Chinese 難 *nar(s).

Could 難珍阿 ~ 難鎭阿 have been 'the moon that appeared' or 'the moon that will appear'? But there is no spelling like 出月 or 出珍阿 with 出 'appear' followed by 'moon' that would confirm the meaning of 難. (The Japanese placename 出雲 Izumo 'appear-cloud' comes to mind.)

難 means 'difficult' or 'disaster' in Chinese, so it could represent some Paekche morpheme with either of those meanings.

12.2.3.8:00: PRECIOUS PETROS (PART 1)

In part 3 of "A C-l-as-h of Codas", I briefly considered interpreting the Old Korean graph sequence

海惡 (普皆廻向歌 3.4-5)

as <SEA.ak> = *pa(t)tak with a final -k corresponding to the -h of Middle Korean patah 'sea' before regarding it as a logogram <SEA> followed by a phonogram-semantogram sequence 惡中 <ak.MIDDLE> *akɯy.

珍惡

and presumably corresponds to Middle Korean torh 'id.'

Late Old Chinese/Middle Chinese 惡 *ʔak 'evil' could be a phonogram for Paekche *-ak. The Sino-Paekche reading of this graph is unknown, but it was probably *ak since the Sino-Japanese reading is aku, and the earlier strata of Sino-Japanese were probably based on Sino-Paekche.

But the first character 珍 presents a problem. If one knew no Chinese, one might expect it to have been read *tor-. A Paekche 珍惡 <tor.ak> would nicely match MK torh. But 珍 has never had an *o-reading at any point in the history of Chinese. Starostin reconstructed its Old Chinese reading as *tər and Schuessler (2009: 328) reconstructed its OC reading as *trən and its late OC reading as *ʈɨn. I reconstruct its Middle Chinese reading as *ʈin.

Although OC long predated Paekche, there are indications that archaic OC-like readings were known to the peoples of the Korean peninsula presumably via contact with marginal northeastern Chinese dialect(s) that lacked innovations found in mainstream Chinese dialect(s). So perhaps the Paekche knew of a nonstandard Chinese pronunciation of 珍 that resembled both its OC reading and the first syllable of the Paekche word for 'stone'. But what was the nature of that resemblance? Here are a few unconvincing scenarios:

1. 珍 was *tər in nonstandard Chinese and represented a Paekche *tər- < *tor- with vowel reduction (cf. Russian akan'e in which unstressed *o loses its labiality: e.g., голова [ɡəlɐˈva] < *golova 'head'). But are there any other examples of Paekche *ə corresponding to MK o?

2. 珍 was *ʈin in nonstandard Chinese and represented a Paekche *tin- < *tor- with vowel raising and fronting (cf. Ukrainian Itavismus* in which *o becomes i: e.g., Біг [biɦ] < *Bogŭ 'God'). But are there any other examples of Paekche *i corresponding to MK o or Paekche *-n- corresponding to MK *-r-?

3. 珍 was *trən or *ʈɨn or *ʈin in nonstandard Chinese and represented a Paekche *tVn-. (It is highly unlikely that Paekche had initial consonant clusters or retroflexes which are both very un-Altaic characteristics.)Paekche *tVnak 'stone' happened to share an initial consonant with MK torh 'id.' but was not related.

4. There was a native Paekche word *tor- 'precious' which coincidentally vaguely resembled the Chinese reading of 珍 'precious. Hence the reading <tor.ak> combined native and Chinese readings. This final scenario is highly speculative since there is no other evidence for a Paekche word *tor- 'precious'.

5. It's even remotely possible that the -惡 of 珍惡 had readings resembling the unknown native Paekche words for 'evil' or 'to hate'. (惡 also represented Old Chinese *ʔak-s > Late Old Chinese/Middle Chinese *ʔoh 'to hate'.)

I got frustrated and stopped blogging here on Thursday morning. Then on the way to work, a diagram came to mind:

All four patterns can be found in early Korean peninsular writing: e.g.,

CC: Old Korean 阿希 <a.hɯy> *ahɯy (locative); cf. Late Old Chinese 阿希 *ʔa xɨəy

CP: Old Korean 惡中 <ak.MIDDLE> *akɯy (locative); cf. Late Old Chinese 惡 *ʔak

PC: Old Korean 夜音 <NIGHT.ɯm> *pam 'night'; cf. Late Old Chinese 音 *ʔɨəm

PP: Old Korean 海等 <SEA.PLURAL> ?*pa(t)tʌrh 'sea'

Next: Which pattern is in 珍惡?

*I learned this term from the German Wikipedia. Itavism, as I'd call the phenomenon in English, occurred before *ĭ and *ŭ that were later lost. I discuss itavism in detail here. It is a kind of assimilation. However, Paekche *o becoming high *i before low *a could not have been assimilation, and I have never heard of that kind of dissimilation ('vowel disharmony'?)