## CONLANG@LISTSERV.BROWN.EDU

#### View:

 Message: [ First | Previous | Next | Last ] By Topic: [ First | Previous | Next | Last ] By Author: [ First | Previous | Next | Last ] Font: Proportional Font

Subject:

Re: MC yod

From:

Date:

Thu, 15 Aug 2019 20:36:43 +0200

Content-Type:

text/plain

Parts/Attachments:

 text/plain (406 lines)
 As a side note to this discussion, there are videos online of a workshop from 2007 presented by Baxter and Sagart, if anyone is interested. The Baxter-Sagart System of Old Chinese Reconstruction (2007-07-02) Webpage: https://hal.campus-aar.fr/medihal-01836384 MP4: https://hal.campus-aar.fr/medihal-01836384/file/42_01_old_chinese_01-lan_720p.mp4 Old Chinese Word Structure (2007-07-02) Webpage: https://hal.campus-aar.fr/medihal-01845046 MP4: https://hal.campus-aar.fr/medihal-01845046/file/42_02_old_chinese_02-lan_720p.mp4 Onsets (2007-07-03) Webpage: https://hal.campus-aar.fr/medihal-01845056 MP4: https://hal.campus-aar.fr/medihal-01845056/file/42_03_old_chinese_03-lan_720p.mp4 Finals (2007-07-03) Webpage: https://hal.campus-aar.fr/medihal-01845064 MP4: https://hal.campus-aar.fr/medihal-01845064/file/42_04_old_chinese_04-lan_720p.mp4 Lines of Evidence (2007-07-04) Webpage: https://hal.campus-aar.fr/medihal-01845078 MP4: https://hal.campus-aar.fr/medihal-01845078/file/42_05_old_chinese_05-lan_720p.mp4 Sino-Tibetan and Beyond (2007-07-04) Webpage: https://hal.campus-aar.fr/medihal-01845098 MP4: https://hal.campus-aar.fr/medihal-01845098/file/42_06_old_chinese_06-lan_720p.mp4 Vítor De Araújo https://elmord.org/ On 15/08/2019 13:35, BPJ wrote: > (NOTE: Apologies for not snipping more. Selecting long chunks one a touch > screen doesn't work all that well for me. Does it for anyone?) > >> Den ons 14 aug. 2019 23:22Alex Fink <[log in to unmask]> skrev: >> >> On Tue, 13 Aug 2019 12:53:17 +0200, BPJ <[log in to unmask]> wrote: > >> >> I hope this isn't insensitive, but I'm impressed that you have memories > with that level of articulatory detail from age 3! > > My memories are a bit later, from age 8–10 when I started to see the point > of speech therapy. Before that I attended an institution for disabled > children where, pervasive palatalization and "spluttering" notwithstanding, > I was among the better speakers. I even acted as interpreter between less > fortunate kids and some less skilled staff. Since everyone who needed to > understand me did I bored myself through group speech therapy sessions > where the focus rarely was on me. When I was 8 I was dumped into a regular > school where I was bullied for my speech; at about the same time I began > individual speech therapy and started to see the point of it in the light > of my new experiences. I continued with speech therapy until age 12, mainly > because the therapist wanted to change my accent. She probably did me a > favor there but at the time I was smart enough to see the difference > between pathology and accent correction and made some resistance to the > latter. I wanted to have an accent like that of my peers! > The reason I know anything about my speech before that age is that I still > have the reel tape recordings my father made of me complete with the device > to play them (although the mike he used wasn't all that good, at least by > modern standards). Mirabile dictu I was able to get a 60's loudspeaker plug > to USB adapter so that I could digitize those tapes. Ironically I will > probably be able to play the original tapes after I lose the ability to > play the CDs I made from them as it isn't certain that I will be able > replace the CD player and CD burner I still have when they break. They made > things sturdier half a century ago! > >> >>>> Contrast maximisation accounts for a preference for Cʲ : Cˠ over Cʲ : C >>> or C : Cˠ, in languages that have a contrast on this axis. But it > doesn't >>> tell us about (in this case, the majority of) languages which lack such a >>> contrast. Among these languages we don't find the single series being Cʲ >>> or Cˠ; it's always plain C. Now that could be merely a bias in the >>> describers, a preference to stick to simple IPA symbols and eschew >>> uninformational diacritics, but I don't think that's all it is. >>> >>> Rather in languages without such a contrast consonants are slightly >>> "colored" by adjacent vowels more than they are in languages with such a >>> contrast, so that consonants next to front vowels are slightly > palatalized >>> (more so the higher the vowel), those next to [u] are slightly velarized > or >>> uvularized, those next to [o] are slightly pharyngealized and those next > to >>> [ɑ] a little less so. >>> (For some reason mid back vowels are normally accompanied by pharyngeal >>> constriction, more than low ones and considerably more than high ones.) >>> However clearly salience does work against such automatic coarticulations >>> becoming as strong as the corresponding distinctive articulations. >> >> Good point. When there's no contrast on this axis, more of it can be > used. And coarticulation effects on this particular axis are pervasive > enough (most consonants are next to a vowel!) that I might be wrong about > the role of a centering force here. >> >> Some axes do have a centering tendency for perceptive salience reasons. > E.g. on the gamut from breathy to creaky voice, modal voice is loudest AIUI. > > Definitely. Also voluntary control of the less modal glottal states may be > somewhat harder to attain than tongue control for diverse PsoA. It's > certainly true for adults whose native languages don't use those states. > May be relevant for L2 learners ancient and modern, though not probably for > (normal) children. In any case glottal coarticulation does occur as > evidenced by intervocalic voicing! > >> >>>> Let me come back to my other prong. OC particles, words of the sort > that >>> cross-linguistically show the greatest reduction, display an even greater >>> bias towards type B. If the A-B contrast was a "balanced" one like Cʲ : >>> Cˠ, we wouldn't expect reduction to favour one side or the other (unless >>> for deeper structural relative-markedness reasons). But if it was >>> privative, if one syllable type had something that the other did not, we >>> would expect reduction to eliminate the extra something. Thus this is >>> evidence that type A had an extra eliminable articulation relative to > type >>> B. >>> >>> Reasonable, but >> >> Did you forget to write something here? > > No, I cut it out, pasted it further down and forgot to remove the "but". > >> >> [snip the distributive suffix stuff] >>> Interesting. OT1H I wonder whether those particles were (originally) > clitic >>> and "unmarked" for the A/B distinction taking up whatever value the host >>> syllable had, only later taking on the B alternant as default. > >> >> I guess so. Though if that were the case it would show that B was the > structurally unmarked member when this assignment of a default happened, > wouldn't it? > > It depends on when the defaulting happened. The original features may have > been lost as such already. Possibly "text" frequency alone was decisive. > > As you probably have understood by now my hypothesis is that A syllables > were velarized while B syllables were palatalized. I don't rule out that A > may have been creaky though. Also I'd like to know what effect breath may > or may not have on vowels. Also IMO it can't be ruled out that vowels in B > syllables were long originally. Diphthongization of long vowels is common > enough! I even think that your suggestion that not all Division III glides > were palatal makes the length+diphthongization hypothesis even more > attractive. The explanation for the B particles may then be that they took > on their long alternant to counteract reduction. > > > >> >>> On the >>> hypothesis that B == palatal it's interesting to note that in Finnish > vowel >>> harmony alternating affixes it is the [+front] alternant which is > neutral, >>> even though front rounded vowels would count as more complex than back >>> rounded ones. NB that /i/ and /e/ do have central(ized) to back > allophones >>> in [+back] words, so they could have them in neutral words. Maybe those >>> allophones are even more complex by some measure, but that's too close to >>> universal markedness for my comfort. There may well be a hierarchy [front >>> unrounded] > [back rounded] > [front rounded] > [back unrounded] but if > so >>> it is one of relative salience. BU : FR vowels are more acoustically >>> similar than FR : BR and BR : FU still less, but I suspect that the > reason >>> that FR is more common than BU is probably just that [i] is the strongest >>> attractor among the vowels for articulatory reasons. OTOH it was the >>> [+back] alternants which "won" when (North) Estonian lost VH! >> >> Front rounded more common than back unrounded? That can't be right... > yeah, in UPSID, there are more languages with a back unrounded vowel than a > front rounded one at each individual height. [ɯ] beats [y], [ɤ] beats [ø], > etc. > > Really? Apparently my perception of things was Euro-biased. Thanks for the > correction! > >> >> It's not clear to me what the evidentiary value of the Finnic harmony > is. In broad strokes, /e/ conditions front vowels because it's front; it's > neutral because, as the continuant of the non-low non-initial vowel of > Proto-Uralic, it failed to split into something like [e ɤ]. > > But /e/ does have a central/back allophone in [+back] words which even > became phonemic in Estonian. In South Estonian (NB that North/Standard > Estonian is more closely related to Finnish than to South Estonian BTW!) > the back allophone of /i/ also became phonemic. > > FWIW I'm familiar with [ᵻ] as the unstressed allophone of /i/ in my > paternal grandmother's accent which my father also could speak as could > and can I (although subject to my childhood speech difficulties!) and my > heritage/native (mother's side) German /ə/ is TTBOMK [ᵻ] most of the time. > Now I hear both /i/ and /e/ in back words as this sound when (trying to) > listening to my in-laws' Finnish, which probably means that they are [ɨ] > and [ɘ] respectively, if not even fully back. > > I do agree however that /i e/ in stems with no back vowels are likely to > (originally have) be(en) front and so triggering front affixes. The fact > that all affixes became back in N Estonian is still interesting. > >> There doesn't seem to be anything to explain in the former. For the > latter, one could appeal to whatever sort of intrinsic preferentiality [e] > has over [ɤ]. But another idea I've run into is supposing this vowel > persisted unsplit as something schwa-like until contact with a language > that had no vowel qualities of that kind (Proto-Germanic?) drove it to > merge into [e]. In any case, the situation that looks like a preference > for [ø] over [o] doesn't emerge till combining both those effects and > stirring. > > I suppose they adduced evidence from East Finnic (non-Baltic) languages? If > not it looks like pure speculation. Also would *the same* have happened In > Hungarian? Its VH isn't that different, although it adds labial harmony. > Turkic and even Mongolic VH is also very similar, so probably North/Central > Eurasian areal. > >> >>> OTOH I wonder what feature similar to A that affix may have carried, and >>> why. Perhaps it was originally a clitic full syllable which was [+A]. > That >>> would seem the most believable explanation IMO, but admittedly it also >>> implies that B was less contagious than A or not contagious at all. >>> Alternatively a same-foot following C had a stronger impact on a V than a >>> preceding C. >> >> Yeah, an erstwhile clitic full syllable would be my first guess too. >> >>>> I'm not familiar with how Classical Tibetan is a controlled language. >>> (More so than other preserved literary languages?) >>> >>> They had committees deciding on lexicon and phraseology, and TTBOMK even > on >>> syntax in some cases, complete with word lists with "official" stamped on >>> them for the purpose of translating Buddhist texts from Sanskrit, mainly >>> through calques. Nyingma ("Old School") texts are a bit less bound by > these >>> but even they would have been concerned about avoiding ambiguity, and > hence >>> homonymy. Versified texts are also a bit freer, but only in the sense of >>> allowing some ellipsis (mainly pars pro toto) rather than introducing >>> non-sanctioned vocabulary. Original texts may also more or less approach >>> vernacular language but even they are under the spell of the official >>> translationese. There are arguably cases where deliberate ambiguity in > the >>> originals, to the extent we have the originals or evidence from Chinese >>> versions, was translated away in the direction of the received exegesis. >>> Also the translation language uses far fewer synonyms than the originals, >>> although TBH literary Sanskrit is extremely synonym happy, not least in >>> order to facilitate versification. It is at least commonly held that the >>> language of Chinese translations was much freer. There certainly was a > more >>> or less strong tradition WRT terminology, but at least no official >>> vocabularies of the kind found in Tibet are preserved AFAIK. >> >> Huh! Did these word lists contain only Buddhist technical terminology, > or also ordinary meanings such as might've been inherited? Do we have any > negative evidence surviving from these committees, i.e. words or usages > they deemed inapt, or only their lists of positive recommendations? > > Mostly but not only purely technical vocabulary — in particular many Skt > particles, prepositions and adverbs are covered — but TTBOMK only positive > recommendations. It is basically just lists of Skt : Tib expressions, > arranged topically starting with "names of the Buddhas" and continuing down > a sort of semantic hierarchy. Alphabetical ordering was notably not > invented, which some actually have taken to point to Chinese originals, but > that is IMO going too far. Alphabetically arranged word lists were probably > only invented once anyway. I wonder by whom. > > See https://en.wikipedia.org/wiki/Mah%C4%81vyutpatti > > It would be interesting to try to count the incidence of homonyms on either > side in the Mahāvyutpatti, let alone actual texts. I do have a facsimile of > Csoma's 19th century Skt : Tib : Eng edition (which is IIRC only partial), > but I likely won't undertake that count myself, not least as to do it > properly you should ideally have a working knowledge of Skt synonymy which > I do not. It might have been a different matter if I didn't suffer from > chronic fatigue and dyscalculia and/or I did it as (paid) degree work. If I > had a set of programmatically accessible dictionaries and text then maybe. > Too bad I didn't get this idea 25 years ago! > > As promised here is part of what B & S 2014 have to say about their > rationale for reconstructing uvulars (with formatting lost in copying and > notes in brackets being mine): > >> As >> pointed out by Pān, words with Middle Chinese fricative and stop initials > tend to >> be kept apart in the script: for instance, Middle Chinese s- and t- > generally do not >> occur in the same phonetic series. But among initials of the ‘laryngeal’ > (hóuyīn >> 喉 音) category, the glottal stop '- (the traditional 影 Yǐng initial) and > the fricatives >> x-, h-, and hj- (the traditional initials 曉 Xiǎo, 匣 Xiá, and 喻 三 Yù sān = > 云 Yún) >> not infrequently share phonetics. >> >> [Examples snipped] >> >> With Pān’s proposal to treat MC '-, x-, and h-/hj- as >> reflexes of Old Chinese uvular stops *q-, *qʰ-, and *ɢ-, respectively (h- > and hj- are >> complementarily distributed in Middle Chinese), contacts between them > turn out to >> be straightforward alternations of voicing and aspiration among > homorganic stops >> in Old Chinese. We follow Pān and reconstruct >> >> [Examples snipped] >> >> However, we have felt it necessary to modify Pān’s original proposal on > several >> points. First, in Pān’s theory the only source of Middle Chinese '- is > *q-; yet the exis- >> tence of relatively long phonetic series having no other Middle Chinese > initial than '-, >> and without a pattern of word-family contacts outside of words with MC > '-, strongly >> suggests that Old Chinese had a glottal stop initial distinct from *q- > and contrasting >> with it [although they merged very early]. > > I note that the only fricative they do reconstruct for OC is a sibilant, > which kind of marrs the argument. The phonetic series (like my Kijieb > conscript BTW) may have distinguished /s/ from /t/ but not distinguished > non-sibilant fricatives from stops. > Also they reconstruct aspirated stops but not any /h/ which is fishy in my > book. I don't know how otherwise to explain their distinct "ʔ" series, > although I note that (Old) Tibetan has a three-way distinction /ʔ/ : /h/ : > /ɦ/ or maybe zero onset for the last. They point out that since they don't > reconstruct any zero onsets their *ʔ may be seen as a realization of zero > onset, which may mean that *ʔ and *∅ merged in MC on a hypothesis with no > uvular phonemes. I guess that means that their MC "x" [x]? and "h" [ɣ]? > were [h] and [ɦ] too. I can't judge the pros and cons of that. And WTH was > their MC "hj"? A /ʝ/ : /j/ contrast? Possible but not very likely. And > under the glottal interpretation? A breathy [j̤]? Possible at most IMO. > Also I wonder how common it is to have velar but no labial fricatives > (although LMC evidently had labial fricatives EMC probably didn't; maybe > that can be seen as a repair.) > > \begin{rant}[long, irrelevant, irreverent] > > BTW I really don't like the way (originally) Baxter's MC notation uses ‹y› > for palatalization, ‹j› for [j] and ‹+› (sic because for whatever reason > they want to restrict themselves to ASCII here!) for [ɨ]. I would rather > have used ‹y› for [ɨ] and ‹š ž›/‹ś ź› and ‹ṣ ẓ› with their obvious values. > If I were to restrict myself to ASCII (although I would do so only as an > input method in 2019!) I'd use ‹tz tzr tzh› for unaspirated, ‹ts tsr tsh› > for aspirated and ‹dz dzr dzh› for voiced affricates, and ‹s sr sh› for > voiceless vs. ‹z zr zh› for voiced fricatives so as to not have to use > non-letters as letters. Heck I might even use ‹q› instead of ‹'› for [ʔ] so > as to free up ‹'› for aspiration and avoid the ‹tz› : ‹ts› hack and extra > ambiguity with ‹h› — or even ‹'› for palatalized although that may be a > non-starter in the Chinese context (even though strictly as an IM so WTH!) > I would also use ‹ạ ẹ› for their ‹ae ea›, so then why not ‹ị› for their > ‹+›?! In that case I would use ‹a+ e+ i+› in my IM and use ‹y› for [j] and > ‹j› not at all. In any case I constantly read their spellings like ‹tsyhi› > disyllabically in my head, so at least I wish they would have swapped ‹y› > and ‹j›. My brain obviously wants to read ‹y› between two consonants as a > vowel! > > I would certainly use acute for "rising" and grave for "departing" tone > (although I wouldn't argue with anyone who would like to put them > before/after the syllable rather than on the vowel!) Their "etymological" > ‹-X -H› are just odious: final caps *and* their ‹x› doesn't mean [ʔ] but > [x]/[h] which makes it confusing anyway (not that ‹-Q› would be any better > in my book, also because their "h" is [ɣ]/[ɦ]!) > > FWIW my Latin script ASCII-to-Unicode Vim keymap uses ASCII punctuation > like {/ \ < !} for (combining) acute, grave, caron, underdot and still > other punctuation (combos) for other combing marks (though for convenience > { [ ] { } * } aren't (yet) used for any marks). In reality I'd rather use > that than my opinionated revised ASCII notation. In case you wonder how I > enter ASCII punctuation without switching keymaps I have mapped {"} + ASCII > character → ASCII character for all printable ASCII characters. (Umlaut is > {:} and double acute is {//} so {"} can be reserved for this. Wackily {^'} > is Vietnamese horn, and {'} and {''} are single and double prime > respectively.) > > BTW the ‹tz› : ‹ts› hack, although I find it reasonable and like it, isn't > my invention. A Tibetan scholar whose name escapes me ATM used it in a > typewritten English-language manual on Tibetan > script/romanization/pronunciation in the 70's (which interestingly and > usefully treats reading pronunciation separately!) His home-grown > typewriter phonetic notation isn't all that beautiful, but ‹tz ts dz› for > the front affricates is just brilliant! > > \end{rant} > > /bpj > > >> >> Alex >> >>