On Sun, 25 Jan 2004 03:40:23 +0000, Glen Gordon <
glengordon01@...>
wrote:
>gLeN's CHALLENGE:
>---------------------------
>If you can answer why we must reconstruct palatals in IE by using
>not only satem evidence but _centum_ evidence as well,
That's a bit too easy: Luwian is not a satem language.
>or by
>showing some peculiarity of IE phonotactics perhaps that shows
>the palatal nature of IE *k, by all means, speak! I would really like
>to know why we must accept palatal velars in Proto-IE itself.
It's interesting to note what happens in areas where uvular and velar stops
are common, such as the North West Coast area of North America. In
Wakashan, which has /k/ and /q/, /k/ has a tendency to go to /c^/, and this
is an areal feature of other NW Coast languages (Salishan, Chimakuan).
Proto-Athabaskan is reconstructed with labialized consonants: *c^, *c^W,
*k, *q, *qW. The absence of *kW is explained by an unconditional merger
with *c^W (Proto-Athabaskan */k/ is described as "front velar").
In a system with both velar and uvular stops, there is therefore a
potential for the velar series to become palatalized.
In a system with palatal and velar stops, there is to my knowledge no
tendency, nor would one expect there to be, to back the palatals so that
they merge with the velars.
Of course a system with palatal*ized* consonants always has a potential to
lose the palatalization, so it matters whether we consider *k^ to be a
palatal stop (IPA /c/), or a palatalized /k'/.
Considering the satem and the centum evidence, the first solution (*k^ =
/c/, *k = /k/) seems out of the question, because of the centum facts. The
second possibility (*k^ = /k'/, *k = /k/) is possible: centum and satem
both could follow from it. The third possibility (*k^ = /k/, *k = /q/)
also explains satem and centum (as in Wakashan), and has the additional
advantage of explaining /a/-colouring near *k.
=======================
Miguel Carrasquer Vidal
mcv@...