--- Piotr Gasiorowski <
gpiotr@...> wrote:
> This is completely ad hoc, and implausible. There's
> no *a > GAv. &
> before _secondary_ nasals.
This the the same response you made to my earlier
claim; I directly said that L>N earlier
> Of course you can insist
> that something like
> *svans existed already in proto-Indo-Iranian
I just did.
> > Nx > Nh than h itself. In Ossetian *s>*x then
> velar > uvular; *x is
> > more likely to provide the forms seen there and in
> all Ir languages
> > than PIr *h.
>
> I see no reason to posit Iranian *x as the
> intermediate stage between *s
> and /h/. A coronal sibilant may lose its oral
> component directly. I
> can't see why -axa- > -aNXa- should be more likely
> than -aha- > -aNha-.
It allows every case of >N in Ir to be explained as
velar before velar. Oss. would count w as velar; OP
would have no doubling of x.
There are many pieces of evidence for s>x>h in those
languages with h (it's more complicated than just
this). I've written about this often before.
In Welsh *sw>xw; In Avestan *sw>xw>xW; in Armenian
*sw>xw>kHw>kH. It makes more sense to say s>x first
(also needed for other environments) instead of s>h
and h>x independently before w (and other positions in
Avestan and Ossetian) in all.
Word-finally there would be s>x>other in sandhi.
So: final as>ax>o: in certain positions in Av
(regularized to all); final es>ex and >e: in certain
positions in P-Celtic giving *dex/de: 'yesterday' <
g^hd!es; final as>ax>akH>kH in Arm.
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com