Re: IE *p and *b

On 2007-12-09 22:32, stlatos wrote:

> You're forgetting Armenian, which shows almost exactly the same
> changes as Celtic (and also pH did not > f > xW > h, so asp. was not a
> stage of this change). Along with other changes, this must show these
> changes occurred in an area at a time when Celtic and Arm. were spoken
> next to each other (next to Greek and Albanian).

I hesitated about including Armenian and decided to wait until I find
the time to think about it. Armanian has enough problems of its own.
Olsen has suggested, for example, that there was a secondary *po- from
*kWo- in Armenian which _also_ developed into (h)o-. Anyway, just as the
Armenian consonant shift is reminiscent of Grimm's Law but not
homologous to it, the behaviour of *p may be independent to the loss in
Celtic but not _the same_. The changes in Celtic were still in progress
in Roman times, and the geographical separation of pre-Armenians and
pre-Celts is surely _much_ older.

> As for sp-, I'm almost certain that no p>f in this position took
> place in Proto-Celtic. It fits the evidence best, and a fric.
> preventing stop > fric. is common enough. Comparing Arm. again, no
> change in p occurred in this position (until later C-shifts).

I'm also inclined to accept *sp- as Proto-Celtic.

> Also, sp- would be proven dif. than -sp- (in dif. syllables) if it
> were true that *wekWspero+ > *wesfero+ > *wesxWero+ 'evening'. I'm
> certain, but others just consider it a borrowing.

Interesting. Pokorny has some uncanny-looking derivation for <ucher>,
but does regard it as inherited. I think forms like *weskW(e)ro- have
been proposed to account for it, but I can't check the details at the
moment. The Leiden databases aren't working and I have very few books
with me here.

Piotr



Piotr