Re: [tied] PIE dorsals

From: Miguel Carrasquer Vidal
Message: 4993
Date: 2000-12-09

On Sat, 09 Dec 2000 04:12:51 , "Glen Gordon"
<glengordon01@...> wrote:
>
>Argh. Perhaps I misunderstood your viewpoint, which is possible because I
>always look at IndoEuropean through the angle of the earliest stages of IE
>more than later stages (I think of IE in terms of two genders and not three,
>for example). I am very sure that *k^ is ultimately the result of an
>allophonic variation of *k.
>
>I presume then that you are saying that *k and *k^ are distinct phonemes in
>some late IE stage (possibly postIE stage)

Yes, we were discussing whether in PIE (the common stage where all
attested IE languages can be derived from, including in this case
Anatolian), one can make do with two dorsal series (*K(^) and *Kw) or
that it's necessary to reconstruct three (*K, *K^ and *Kw).

>and that they were originally the
>result of allophonic variation of plain *k? If this is correct, then we do
>agree.
>
>Still, as for IE itself having these two distinct velars, to say that *k^ is
>a distinct phoneme in IE means that one has to be sure that *ku- and *k^u-,
>the only interpretational minimal pairs for "distinctness" that I'm aware
>of, were indeed pronounced */ku/ and */k^u/, not just */k^Iw/- versus
>*/kUw/-, say. Afterall, we both agree that *k^wo:n was pronounced with two
>syllables at some point in IE. If you could find blatant examples of widely
>attested IE *k (as opposed to *k^) within the same syllable as a _front_
>vowel like *e or *a, there'd be a case. AFAIK, there isn't such an example
>and we have *ek^- in *ek^wos and *dek^- in *dek^m, just like we should have
>if *k and *k^ are only allophones.

There are enough such examples. E.g. *kih1eu-e-toi "he moves" > Grk.
<seuetai>, Skt. <cyavate>, with *k before front vowel *i.

The present discussion has made me reconsider my position in at least
one way. I still think *K differing from *K^ is unavoidable in PIE,
and I still believe that there is enough interesting evidence to
suggest a three-way split into plain, palatalized and labio-velarized
consonants at some stage of pre-PIE, which squares well with the
(near) absence of the vowels *i and *u (except as zero-grade variants
of *y and *w) in PIE. However, I had only considered the mere
existence of *K, *K^ and *Kw, not their relative frequency. In that
respect, it is clear that *K^ is just too frequent to be merely the
palatalized variant of *K. And Piotr's argument that (if the *K^'s
had been palatalized phonemes) we would expect to find solutions
involving *Ky besides the satem assibilation, is also food for
thought. The only resolution of this dilemma that occurs to me is the
following (and it involves a new set of pre-PIE consonants, I'm
afraid): the contrast was between velar **K and post-velar (uvular)
**Q. When palatalized, both series gave *K^, and when labialized,
both gave *Kw, but the bulk of them gave *K^ when originally velar, *K
when originally post-velar. This would also explain why two
velars/dorsals are allowed in one PIE root, and it explains the
relative infrequency of *K (post-velar phonemes are more marked than
velar ones, and the number of post-velars had further been reduced by
their development into laryngeals under certain unclear circumstances:
e.g. *qost(H)- ~ *h2ost(H)- "bone", *qu@... ~ *h2u@... "steam", cf. the
development OGeo. /q/ > Geo. /x/). In the satem languages, the
velar/post-velar opposition was transformed into a palatal/velar one,
in the centum languages both series eventually merged (apart from
possible effects on a following vowel as suggested by Peter Schrijver
for Latin).


=======================
Miguel Carrasquer Vidal
mcv@...