15 Jul 2008

Update of my "Diachrony of Pre-IE" document

Here is the updated document as promised for today. I've added a lot of things to it and it's already turning into a monster of a pdf. I've filled out the Indo-Aegean section a bit more and rearranged the order of a few things after pondering a bit more on that early stage. After writing the last post on gemination, I'm satisfied with my new rule that combines vocalic lengthening with this consonantal lengthening which appears to be triggered by the same (i.e. the reduction of former schwa to supershort schwa). Here's the pdf hosted on esnips.com:

DiachronyOfPreIE_2008Jul15_2
DiachronyOfPreIE_2008Jul15
Hosted by eSnips

The mess with early gemination that I've been previously speaking about has oddly enough led me down a new quest: Proto-Kartvelian (PK) loans. I previously used *ḱerd- "heart" as one of my MIE examples undergoing gemination and vowel lengthening. Yet it turns out that while it is often associated with PK *m̥-k’erd- "chest", I've never yet come across an accurate explanation as to how Proto-Indo-European (PIE) either acquired these loans or how they were given to Proto-Kartvelian. Where and when?? Then when I survey the PIE words that are purportedly shared with Proto-Kartvelian, I notice that they appear to be loaned most recently within the Late IE period. If that's true, I can't in good conscience ever use this item again when discussing Mid IE. I might speak more on this interesting issue later.

8 comments:

  1. Ah, finally getting the thing to open. Nice work so far. One tech issue BTW - the font you're using seems to conflate <g> and <g~>. Or is it on my end?

    Probably unsurprizingly, a few content questions that spring to mind:
    - Any plans to sketch out your view of the entire phonological system at some select stages? At least the "initial" form would be informativ.
    - You've probably covered this somewhere before, but, any particular reason you reconstruct *H3 as a labialized glottal, not uvular like *H2?
    - Medial PS emphatics yield MIE mediae (well, in the one example you have), but PS mediae yield MIE glottalics? A strange flip-flop... also, shouldn't you be using pre-Syncope forms such as *sébatam and *xénada?
    - And finally, possibly a whole can of worms here, but do you consider the *a/*o / *ā/*ō oppositions to have been in place in all IE branches and simply merged in most? It seems to me a number of steps could be saved by taking the lack of contrast as original in some places.

    ReplyDelete
  2. Tropylium: "Ah, finally getting the thing to open. Nice work so far."

    Thanks! I'm glad it's working :)

    Tropylium: "[...] the font you're using seems to conflate g and g~. Or is it on my end?"

    I'm noticing it on my computer too. It's an issue with the Code2000 font I guess. Maybe someone knows of a better font I could use for IPA that's "phonation friendly"?

    Tropylium: "Any plans to sketch out your view of the entire phonological system at some select stages?"

    Yes, that makes sense. My phonology isn't much different in Mid IE than in PIE, save for the addition of labialized dental stops and sibilants and the absence of a phonemic plain/uvular contrast (or palatal/plain for you traditionalists out there). It's the vowel system that seems to have changed the most. I'll plan on making a file like that when I get a chance.

    Tropylium: "You've probably covered this somewhere before, but, any particular reason you reconstruct *H3 as a labialized glottal, not uvular like *H2?"

    To me, *h3 was ultimately born from the era of Indo-Aegean Centralization when the feature of labialization was transferred to the neighbouring consonant, including *h. So Proto-Steppe *h therefore diverges into Indo-Aegean *h (when before Proto-Steppe *i, *a and ) and *hʷ (before Proto-Steppe *u). Up until Late IE then, *kʷ was aligned with *h3.

    I think what happened next was that once uvular consonants became phonemicized in Late IE, *h2 was uvularized and became aligned with the new sound *q (traditional plain *k). Then I presume that *h3 became voiced either in PIE itself or in non-Anatolian dialects after the diffusion of the PIE speaking community, pronounced at this point as a rounded ayin (this is in order to explain the later reflexes of *piph3eti such as Sanskrit pibati with voiced b).

    Tropylium: "Medial PS emphatics yield MIE mediae (well, in the one example you have), but PS mediae yield MIE glottalics? A strange flip-flop..."

    However, I've made sure to put Ejective Voicing before Proto-Semitic contact. So without ejectives in Mid IE, these exotic Semitic emphatics would have to be interpreted into Pre-IE another way. In the word "seven", *-bʕ- has little choice but to be interpreted as Mid IE aspirated *p since ayin does not exist at this stage. Also, if we conceive of Semitic voiced stops as "semi-voiced" more like in English, it would be natural for Mid IE speakers to interpret this as a creaky-voiced stop rather than a fully-voiced one.

    Tropylium: also, shouldn't you be using pre-Syncope forms such as *sébatam and *xénada?

    Nope. According to the current version of my theory now, **sébatam and *xénada would have yielded PIE **sḗbʰtm̥ and **xēndʰ-. What's more, stems themselves in Mid IE seem to only allow penultimate accentuation. Antepenultimate accentuation only ever surfaces in words with enclitic extensions like *-sa and *-ta.

    Tropylium: "And finally, possibly a whole can of worms here, but do you consider the *a/*o / *ā/*ō oppositions to have been in place in all IE branches and simply merged in most?"

    Yes. Before you scough, consider for a moment what the implications are of a language that only recently changes *a/ into *o/. If we envision a core PIE emerging out of a dialectal soup that did not undergo Vowel Shift, then there's the potential that many of the budding PIE dialects were affected by neighbouring "para-IE" dialects which lacked *o. And thus, a sound change was reversed (but not without traces showing that the dialects in question originally did contain *o once).

    Tropylium: "It seems to me a number of steps could be saved by taking the lack of contrast as original in some places."

    Nope. I've thought of this but we just can't get rid of Vowel Shift. Indo-Iranian for example shows original *o in Finno-Ugric loans (cf. early Indo-Iranian *pórćas "pig" becoming Finnish porsas). I'm sure Anatolian demonstrates original *o too somehow but I can't think of what the clever reasoning behind this is offhand right now.

    ReplyDelete
  3. Oh yes, I just remembered one reason why early Proto-Anatolian must have had *o. It has to do with the commonly known fact that *h3 rounded any neighbouring original short *e to *o. This rounding however is also evident in Hittite since the coloured vowels in these circumstances yield a (< PIE *o) instead of e. Laryngeal colouring took place then at the PIE stage, not later, and this then means that *o existed at this stage as well. In the case of Anatolian languages however, the merger of *o and *a paints a picture of later areal influence from the phonology of languages in that the general area that also lacked o.

    ReplyDelete
  4. So you ARE pairing *H2&3, OK, that's what I was getting at. The pdf starts right away with *x and *hʷ so I got the impression you're merging non-labialized *h into the postvocalic allophone of *H1.

    I can't make out if the tilde for *g is overlapping or not rendering at all, but if it's the former, switching to a superscript should do the trick.

    Your comment about "´ayin not existing" is a bit unclear: its lack in PIE I can't see pushing *bʕ over as *p, as the cluster is still voiced thruout. Or do you mean this is actually an emphatic that decomposes into two phonemes later on in Semitic? Here the simplest answer to me still seems to be devoicing by contact to the *t, similar to *sweks-.

    As for the PS mediae that yield glottalics, one of them seems to be adjacent to an ´ayin and the other, geminate, so how about this does not need to be the general medial outcome, just a conditioned one?

    ReplyDelete
  5. Tropylium: "So you ARE pairing *H2&3, OK,[...] I got the impression you're merging non-labialized *h into the postvocalic allophone of *H1."

    Oh no. I'm not merging Proto-Steppe *h into PIE *h1 (glottal stop) at all. However, I believe that the glottal stop eventually weakened to /h/ in mediofinal positions within the Mid IE period which remained distinct from velar fricative /x/ (= h2) at the time.

    Tropylium: "[...] switching to a superscript should do the trick."

    I'll have to experiment with that tactic, thanks! Unicode still has its bugs. Grr.

    Tropylium: "Your comment about "´ayin not existing" is a bit unclear: [...]"

    I meant that ayin **in Mid IE** did not exist. Some theorize that PIE *h3 was an ayin so this is why I mentioned it. So without even a voiced uvular fricative in Mid IE, there was nothing approximate to this sound other than options like null, creaky voice, aspiration, or velar fricative *x. Here, since follows PSem *b, it would naturally have the potential of being perceived in this particular case as an aspirated stop. Since aspirated stops are only voiceless in Mid IE, we end up with a reflex of *p (and note that creaky voiced *b~ was absent in the phonology so this wasn't an option either). Voicelessness here was further encouraged by the following voiceless *t in the PSem word and also by the fact that Mid IE was obligated to drop a syllable to naturalize a trisyllabic word with initial accent into a disyllabic word with penultimate accent.

    Tropylium: Here the simplest answer to me still seems to be devoicing by contact to the *t, similar to *sweks-.

    Yes, now you got it. The reflexes are not regular because they're borrowings afterall but the phonetic correspondences are understandable when you analyse each individual case and there are tendencies as I've touched on just above. It's like the crazy case of "Merry Christmas" -> Hawaiian Meli Kalikimaka or "McDonald's" -> Japanese Makudonarudo where again we have the intriguing issue of how two differing phonologies interpret each others phonemes. Sometimes the results can be quite unexpected but always understandable when phonological systems are reckoned for.

    Tropylium: "As for the PS mediae that yield glottalics,[...]"

    In my theory, nothing yields "glottalics" because the Old IE ejectives have already been lenited before contact with Proto-Semitic. Your choices for stops then are only "plain voiceless", "plain voiced" and "creaky voiced".

    Tropylium: "one of them seems to be adjacent to an ´ayin and the other, geminate, so how about this does not need to be the general medial outcome, just a conditioned one?

    Yes, pretty much. There are general tendencies but it all depends on how Mid IE speakers would have perceived the word.

    ReplyDelete
  6. Calling them "glottalics" just sounds better than "creaks" or "decem series" or what have you. :)

    I'm not sure why we got stuck on *septm but since we're here, I could probably mention about having crafted a hypothesis on how to get Finno-Permic *śeŋćim- out of that. First, it turns out that eng-acute is actually distinct from en-acute, I had been taking them both as the PU palatal nasal. But no, the former simply marks a velar nasal that palatalizes by some post-PU stage. Damn books never explaining their transcription systems...

    *ŋ is indeed indirectly deduced here, as aside from *ŋk it is directly supported only by Ugric and Samoyedic data AFAIK. I'm assuming the reflexes of *ej are distinct in some branch.

    The interesting part begins when I found out that according to the current understanding [1], a coda laryngeal was regenerated in derivativs after its initial decay into Finnic vowel length, and it has, as far as I can tell, the same outcomes as coda *ŋ. (They can be hard to tell apart intervocally, too.) It doesn't take much after that. Take an Iranic form such as *seftem; allophonic palatalization before *e is likely there (considering its effects on Aryan velars). If we posit a voiceless spirant value for Uralic *x (by this stage anyway), over in PFP the closest equivalent would be śexćim. Case closed. :)

    ReplyDelete
  7. I am sorry I don't know anything about you so forgive me if this is too basic ... but are you looking at PK by yourself or do you have any knowledge of Kartvelian? Have you read, for example, Klimov's 'drevneishie kartvelizmy indoevropeiskix jazykov'?

    ReplyDelete
  8. It has not been approved, but I fear that in my previous comment I wrote of Klimov's Ancient Kartvelisms of Indo-European languages rather than Ancient Indo-Europeanisms in Kartvelian Languages. My mistake if I remember this correctly.

    Anyway, the point is Klimov dates a good number of IE roots found in Kartvelian to (1) Common Kartvelian, (2) Common Georgian-Zan, and (3) Common Georgian-Svan. Previous etymological work has established an estimate of the quantity of reconstructed lexemes for each of these periods (also for Common Zan-Svan, though it has very few). It is generally thought that Georgian-Zan forms the most recent branch, with Svan the outlying language, due to the quite superior number of common lexical items, though the opposite opinion also exists (as Zan has very many grammatical innovations).

    All in all, if I remember correctly without looking this up, the first dispersion of Kartvelian languages is dated to about 3600 BC, so it is likely a couple millenia younger than PIE. Therefore, the about 29 lexemes considered by Klimov for CK

    Thomas

    ReplyDelete