On Sun, 18 Sep 2005 01:15:13 +0200 (CEST),
mkapovic@...
wrote:
>Although, if there indeed was *q > *k and *k > *k' change, I would still
>rather reconstruct it in pre-IE than in regular IE coz we find no trace of
>*q in IE lgs and all the evidence point to *k', not *k. Thus, for the last
>phase of IE, I'd much rather reconstruct unstable *k, *k', *kW.
True. If PIE had had *q, *k, *kW (and then, on typological
grounds, necessarily also *qW), we would not only have to
posit a satem group, but also a centum group, with the
shared innovation *q(W) > *k(W). But there *is* no centum
group, and the development *q(W) > *k(W) is an archaism
retained from PIE.
That *k was [q] in pre-PIE is shown by the fact that *k is
often followed by *a, with the same vowel colouring as
triggered by *h2 = [X]. That *k was not [q] anymore by the
time of the split-up of PIE is shown by the fact that we
also have *ke, without colouring. It follows that the
uvular articulation was lost somewhere before the PIE merger
of most short stressed vowels to *e.
=======================
Miguel Carrasquer Vidal
mcv@...