Sinitic historical phonology

« previous post | next post »

[Or, as David Prager Branner, who wrote the guest post below, jokingly calls it, "hysterical phrenology".  Note that Branner uses Gwoyeu Romatzyh ( "National Language Romanization"), a type of tonal spelling, for the transcription of Mandarin.]

================

This is on the subject of Carbo Kuo's 郭家寶 performance of Shyjing "Shyi yeou charngchuu 隰有萇楚" ("In the low wet grounds is the carambola tree") in Jenqjang Shanqfang's 鄭張尚芳 various antique reconstructions, sent to me by Victor Mair. It pleased me a lot. The issue is one of art, not scholarship, and it should be judged as art.

[VHM:  must hear]

Native Chinese written representations of sound are fundamentally abstract, befitting a linguistic environment with much variation. That may even be the result of multilingualism in the society that developed and spread the script — who knows?

But recitation and cantillation of texts in the past century, which we can observe today in real life and through recordings, make good use of that abstractness, as license for creative expression. These practices run on much the same principles as Chinese calligraphy.

I see no reason why reconstructions shouldn't be part of the cheese on which the blue mold of artistic performance grows. The only irritation — the cheese-maggot, if you will — is the conceit that a reconstruction is "authentic." But that bug is an exotic colonist, not native to poetic circles; its autochthonous home is in the world of historical linguistics.

For my part, I'd like to see more of this rather than less. Take a look at the rich range of reconstructions we have to pick from! They are far more diverse than the modern Chinese dialects. I listed at few in a review of Schuessler's splendid 2009 reconstruction in Minimal Old Chinese and Later Han Chinese: A Companion to Grammata Serica Recensa.



12 Comments

  1. B.Ma said,

    September 11, 2017 @ 5:40 am

    Hmm, it sounds a bit mechanical to me, perhaps because the older language is read very slowly and then he speeds up as the sounds become more "natural" to his mouth.

    Would like to hear someone try to recite some reconstructed Old Chinese in the "dreamy/airy" style of the HK student (who won some competition?) of whom Victor posted a video a while ago.

  2. David Marjanović said,

    September 11, 2017 @ 5:58 am

    the older language is read very slowly

    Also, with lots of epenthetic vowels in the difficult consonant clusters, with verrrrrrrrry drrrrrramatically lengthened [r], with a few minor mistakes in the pronunciation of the IPA symbols – and with tones. All the reconstructions sound like tone languages, even though most of the Old Chinese ones and arguably Early Middle Chinese aren't supposed to; I note that the whole process of tonogenesis is completely omitted from the reconstructions. (For a general impression of a language that "sounds like Chinese without tones", I recommend modern Korean.)

    So, the idea is nothing short of wonderful, but the execution needs a few tweaks still.

  3. Rodger C said,

    September 11, 2017 @ 6:51 am

    Tangential, but "carambola" strikes me as one of those words only an academic would use in poetry. What's wrong with "star-apple"? Other than that the Chinese contains neither "star" nor "apple" …

  4. Chris Button said,

    September 11, 2017 @ 9:50 am

    "The only irritation — the cheese-maggot, if you will — is the conceit that a reconstruction is "authentic." But that bug is an exotic colonist, not native to poetic circles; its autochthonous home is in the world of historical linguistics.

    I can't resist taking the bait here…

    Even ignoring the fact that it better accounts for connections across the Chinese lexicon (e.g. why 新 is related to 生, or why the Shijing rhyme of 火 does not directly correspond to its MC reflex), the reason I prefer a Pulleyblank-style reconstruction over others is that (at least in my interpretation) it does not force a particular surface representation of the rhymes onto Old Chinese. I commented on LL earlier about this in the discussion here: http://languagelog.ldc.upenn.edu/nll/?p=33963#comments

    The famously "vowelless" language Kabardian is reportedly full of vowels when spoken, it is just that deep down in its underlying structure they are not necessarily needed. To give a comparison from English, whether "cot" and "caught" rhyme as /kɑːt/ or are distinct as /kɑːt/~/kɒt/ and /kɔːt/, they all still come from one source and an account for their surface reflexes can be provided accordingly. Where multi-vowel systems force you to accept their abstract notation as a surface reality, a vowelless (or at least syllabic /ə/ with a lower ablaut variant /a/) system can flex to accommodate whatever the surface manifestation may be.

  5. Chris Button said,

    September 12, 2017 @ 11:28 am

    Where he has /naːl/ for 儺, it's interesting that he seems to pronounce it as a /w/~/u/ like sound as if it were originally a dark velarized /ɫ/ as in /bɾaziw/ for "Brazil" in Brazilian Portuguese or /bɒʔʊw/ for "bottle" in Cockney English. I wonder if that's intentional or subconscious?

  6. Jichang Lulu said,

    September 12, 2017 @ 3:38 pm

    @David Marjanović

    The reading of the first few reconstructions might remind you of a tonal language, but my impression is that for the Zhengzhang OC the reciter is assigning tones predictably based on the syllable coda; to the extent that impression is correct, it follows that he's treating the reconstruction as a non-tonal language. Of course those predictable surface intonations are arbitrary, but if anything they are less 'invasive' and speculative than an 'expressive' intonation copied from some modern non-tonal language or made up from cross-linguistic generalisations.

    (If the tradition is to be believed, this was originally sung.)

    I agree about the need for 'tweaks', specifically concerning those epenthetic vowels in clusters and after [b]. If it's any consolation, what are initial clusters in some reconstructions can be presyllables (with a vowel in them) in others… but I'm afraid that wouldn't help in this poem, since I think none of the characters would be sesquisyllabic in e.g. Baxter-Sagart.

    the whole process of tonogenesis is completely omitted from the reconstructions

    You mean from the proposed phonologies of the reconstructed languages, or from this reading of it? The former obviously involve tonogenesis as a key aspect. In the video they are displayed using different colours. I would say tonogenesis is quite prominent in the video (which doesn't mean I think the tones and their phonetic realisations are unquestionable).

    modern Korean

    Tone helps distinguish Korean stops, which has led some to suggest the language is going through tonogenesis right now.

  7. Jichang Lulu said,

    September 12, 2017 @ 3:49 pm

    Serendipitously enough, the poem (Shī 148) features the character 华 huá, here meaning 'flower', but also a name for China (occurring, in particular, in the names of the two countries currently named China). Its possible relation to 夏 xià was brought up a few days ago on the censored tattoo thread.

  8. Chris Button said,

    September 15, 2017 @ 8:24 am

    @ Jichang Lulu

    The more I think about 夏, the more I think the ideas mentioned on the thread about the conflation of two forms may be correct. A case-in-point for this kind of conflation is 天. Baxter & Sagart have proposed a dialectal account for 天, although I would respectfully challenge that on the following grounds…

    The word 天 *tʰə́ɲ (from pre-Shijing *tʰə́ŋʲ) is often reconstructed with a ʰl- initial rather than a tʰ- initial on the basis of the apparent derivative 祆 *ʰ(l)ə́n (from pre-Shijing *ʰljə́m; the parentheses mark a later shift of ʰl- > h- which is a common alternation for unstable pre-aspirated sonorants attested elsewhere in Tibeto-Burman). The pronunciation of 祆 also explains the later use of 天 to transcribe the first syllable of "Hinduka" (the rhyme *-ə́n had shifted to a higher and fronter articulation as if from *-ə́ɲ under the combined influence of the coronal initial ʰl- and coronal coda -n).

    However, inscriptional evidence shows 天 to derive from a phonetic component 丁 (originally a quadrilateral/rectangular shape now just represented by a line on top) and 大. The word represented by 天 *tʰə́ɲ < *tʰə́ŋʲ is in a ə/a ablaut relationship with 丁 *táŋʲ (which is obscured by 6 vowel reconstructions giving a haphazard alternation of -i- versus -e- in the respective forms). Since the xiesheng (characters written with the same phonetic) series of 丁 cannot support lateral initials, the initial of 天 cannot originally have been a lateral ʰl- as in 祆.

    The reason for the confusion stems from the Zhou conquest of the Shang when the character 天, which is barely attested in the earlier inscriptions, starts to be used to represent instead the principal Zhou deity that ousted the Shang 帝. This borrowing effectively left two different traditions for the character 天:

    1. The original 天 *tʰə́ɲ < *tʰə́ŋʲ belongs to the xiesheng series of the quadrilateral 丁 *táŋʲ (the lower -a- vocalism prevents the shift of the coda to a palatal nasal ). The semantic field is similar to that of Proto-Indo-European *h₃regʲ- "straighten out" with derivatives in English like "realm, rex" and "rectangle, regular" (obviously with no implied IE and OC common origin here).

    2. The later 天 (祆) *ʰ(l)ə́n < *ʰljə́m heads its own xiesheng series with ablaut derivations like 忝 ʰljámʔ "shame" (again with the lower -a- vocalism preventing any change to the -m coda). The semantic field is similar to that of Proto-Indo-European *kʲem- "cover, shroud" with derivatives in English like "heaven" and "shame" (again with no implied IE/OC common origin).

  9. David Branner said,

    September 17, 2017 @ 12:32 pm

    @ Chris Button

    I didn't know about 丁 in 天 — that's quite interesting.

    Are you suggesting that there may be two different words written 天? Can you show me different usage to support them? Why does 忝 mean "to humiliate" — is it related to anything else in the 忝 series, most of which seem to share a sense of extension or protrusion?

    Baxter and Sagart have L-IM forms for some graphs in the 覃 series. Do you see a connection between extension/protrusion for words in the 忝 series and "probing" or "depth" for words in the 覃 series?

    Do you consider it possible that 忝 is not intended to contain 天 as a phonetically exact phonophore?

  10. Chris Button said,

    September 17, 2017 @ 10:25 pm

    @ David Branner

    I didn't know about 丁 in 天 — that's quite interesting.

    It's amazing how few people seem to be aware of this. I would have thought that Schuessler's mention of it under 天 in his etymological dictionary would have drawn more attention to it, but it seems that few people have noticed even there. In my opinion, one of the biggest problems in the reconstruction of Old Chinese is an inadequate use of oracle-bone evidence.

    Are you suggesting that there may be two different words written 天? Can you show me different usage to support them?

    Yes and that is why the Shiming passage focuses on this specific character. The evidence comes from pronunciation alone since the two words had essentially the same heavenly-like meaning.

    Why does 忝 mean "to humiliate"…

    For the same reason that "heaven" and "shame" seem to be related in English.

    … — is it related to anything else in the 忝 series, most of which seem to share a sense of extension or protrusion?

    I assume you are thinking of 舔 here? This has been written with the "wrong" phonetic and should go under this series (the original form for the word now represented by 簟 "mat"). Of the other words, 吞 *ʰlə́n < *ʰljə́m "swallow" is onomatopoeic (compare French lamper "gulp"), and 添 ʰljám is a late coinage.

    Baxter and Sagart have L-IM forms for some graphs in the 覃 series. Do you see a connection between extension/protrusion for words in the 忝 series and "probing" or "depth" for words in the 覃 series?

    The 覃 series has a sense of "straining/stretching" (it depicts a winnowing basket after all) which I do not connect with the 忝 series outside of graphic confusion in cases like 簟 above.

    Do you consider it possible that 忝 is not intended to contain 天 as a phonetically exact phonophore?

    I would say that's a good excuse for avoiding having to account for why a word like 天 *tʰə́ɲ < *tʰə́ŋʲ (with phonetic 丁 *táŋʲ) should appear to be being used as phonetic in a word like 忝 *ʰljámʔ.

  11. Chris Button said,

    September 19, 2017 @ 1:12 pm

    @ David Branner

    On further thought, seeing as it is nigh-on impossible to prove the co-existence of two words (*tʰə́ŋʲ and *ʰljə́m) for 天 in terms of actual usage rather than pronunciation, a more persuasive argument (which is simultaneously more favorable to Baxter & Sagart's dialect treatment), might be to assume that Zhou *ʰljə́m completely ousted Shang *tʰə́ŋʲ.

    Evidence for this perhaps comes from the shift in the shape of 天 away from the Shang phonetic 丁 *táŋʲ (Pankenier/Didier's "celestial square") towards a more anthropomorphic shape with 丁 clearly becoming the head. This most likely represents the Zhou reinterpretation of the earlier broad celestial/heavenly sense into something more like the figure of a deity.

    The word *ʰljə́m could then give Early Middle Chinese xɛn or tʰɛn in accordance with Baxter & Sagart's dialectal proposal and clearly playing the role of phonetic in 忝. Although the form tʰɛn happens to coincide with what would also have been the reflex of *tʰə́ŋʲ (representing phonetic 丁) had it survived, this is actually of no consequence (albeit a little confusing!).

  12. Chris Button said,

    September 19, 2017 @ 2:04 pm

    But then again, *tʰə́ŋʲ > *tʰə́ɲ > tʰɛn just happening to coincide with *ʰljə́m > *ʰ(l)ə́n > tʰɛn, and the Shiming just happening to choose 天 as the representative example of a major dialect divergence – i.e. why does only 天 seem to have two pronunciations when everything else went one way or the other (perhaps due to its use in foreign transcriptions?), all seems a little too coincidental.

RSS feed for comments on this post