petusek wrote:
> Why couldn't one imagine devoicing under the influence of the
> unvoiced veral component? I.e. akwa > (akBa >) akFa > affa > afa
>
> As for theorisizing whether kW or k'W, if we find external (i.e.
> non-IE) parallels or internal etym., we can easily decide, otherwise
> we cannot. We have to admit both possibilities until oneof them gets
> completely excluded.
It seems one come back to the lost of velar. Assuming the consonatising of
"W" to any other sound here
one will get a hard to speak cluster "kv, kb, kp, kf," or there where the
"W" does not consonify, one has still the
pronounceable "ku".
Absolutely theoreticaly one should agree the easiest way to speak out a such
cluster should be the reduction of it.
And the reduction got in some language from "kv,kf,kb,kp" to "k", in other
languages to "v,f,b,p". That is:
-some language preserved just the velar of the cluster, some languages
preserved just the labial of the cluster.
Now, what about "gW" versus "kW"? I guess the problem is not here.
We do know about s+voiced consonant > voiced "s" (z). What if the same thing
did happen here? I mean, the velar from "gW" or "kW" determined what kind of
consonant will be this "W". So for "k" since this is a unvoiced one, we have
output just unvoiced (p,f) and for "gW" we have as output just voiced (b,v).
The way appear simple and does not imply any dificult changes since what due
consonation of "w" became hard to speak, was reduced again to speakable way.
What does speak against this idea?
Alex