If you recall our discussion of the
supposed phonetic realisation of the three dorsal series, the frequency and
distribution of *k^ suggest that it was in fact the "plain" velar of PIE, i.e.
/k/ pure and simple (presumably with some normally expected frontish allophones
before front vowels), while *k must have been "marked" in some way (e.g. a
retracted /q/). Now, as it would be extremely unusual for a phonemic inventory
to contain palatal /c/ (or /ts'/), uvular /q/ and labiovelar /kW/ but no
plain velar, it would seem that either the fronting of /q/, or the
delabialisation of /kW/, or both, were part of the same shift that produced the
satem palatals. My guess is therefore that the shifts (at least the crucial
changes of k > c and kW > k) were virtually simultaneous, perhaps with
subtle dialectal differences as regards the timing of q > k. For example, the
fact that Armenian has satemised reflexes of sounds that have remained velar
elsewhere may mean that a local merger of *k /q/ and *k^ /k/ took
place before satemisation, while putative examples of unpalatalised PIE *k
before front vowels in Albanian suggests that the change of q > k (and its
merger with former *kW) was _later_ there than in the "core satem" group
(Balto-Slavic and Indo-Iranian).
Piotr
----- Original Message -----
Sent: Saturday, April 20, 2002 10:26 AM
Subject: Re: [tied] Re: Agriculture and IE
How late is Satemisation? There are two events,
the collapsing of *kW and *k, and the palatalisation of *k'. These
could be at different times, and the latter could be a wave phenomenon after
dialects spread, while the first occurs before the satem dialects break
up.
Do we have any clues as to a date, or an order of events? I can
think only of the later law of palatals in Sanskrit, which does not distinguish
between *kw and *k, so the collapse of *kW and *k must be before the collapse of
*e and *a and *o.