Okay, there's been an onslaught of discussion on
how IE stops should really be written and a lot of
points have been raised.
First off, nothing now can take away the fact that
traditional *k (however it may have been pronounced)
is _marked_ over-and-above the misnamed "palatovelar"
*k^. So, putting aside exact phonetics for now, we
can't get around the fact that *k-*q-*kW is a more
optimal system.
Second, it's been raised by Miguel that **qW is
necessary because if *q were uvular, then in such a
system, it is a "universal" that /qW/ must also
exists. In short, Miguel has mistaken a _tendency_
as a universal and ignores the possibility that IE
*q developed _after_ the development of labialization.
This is entirely possible and in fact what I theorize.
There is nothing phonetically problematic with a
system that distinguishes *k, *q and *kW and it would
logically be rarer than a system of with *qW for
the simple fact that /q/ doesn't often develop right
_after_ labialization. That premise requires that
two specific things happen (labialization, then
uvularization). Hence it is rarer.
It's also been raised that uvulars aren't preserved
in later IE languages but then neither are laryngeals.
Next!
Finally, it was said that uvularization doesn't
work for a system that includes voiceless *k,
voiced *g and voiced aspirate *gH. I agree. That's
because our understanding of IE voicing _also_
violates universals. Instead, we must look at *g
as semi-voiced and *gH as fully voiced. In other
words, IE has a three-way system distinguish voicing
onset that has nothing to do with aspiration.
Even if one doesn't like "uvular" as the marked
feature of our *q, we can always settle for
velarization: *q = /kX/, *G = /kgG/, *gH = /gG/,
*h2 = /x/. If I understand, some languages show
vowel colouring neighbouring velar fricatives too.
So, the criteria of a-colouring depends on the
language methinks.
Mate:
> But if you insist on only one occurence of *k >
> *c^, it could not have been a very late change
> since we have your *c^ already in Anatolian.
As I said, _part_ of the Anatolian dialect area would
have lied in the satem area. They overlap and areal
influence explains all of this. This doesn't detract
from the fact that there is a single wave of *k>*c^
that spreads across _parts_ of the overall IE
speaking community.
= gLeN
__________________________________
Yahoo! Mail - PC Magazine Editors' Choice 2005
http://mail.yahoo.com