Re: Gemination in Celtic

--- In cybalist@yahoogroups.com, "fournet.arnaud"
<fournet.arnaud@...> wrote:

>
> I never talked about H2-g
> Maybe some inadequate examples made you think
> that we were discussing that kind of examples.

Yes, since I depart from a standard PIE system, that may easily
happen.

>
> Basically, my proposal is that
> roots that display alternations like
> Celtic and Osco-Umbrian -CC- unvoiced
> LAtin -C- unvoiced
> Others -g- voiced
>
> Should be reconstructed as -?-C-
> that is to say : glottal stop + unvoiced consonant.
>

But I think we still have to see one convincing example of this
correspondence, let alone enough to posit a new H phoneme.

> And to be clear, I hope, I state that
> in *ALL* languages H2-g- > H2-g-.
>
> How do you distinguish between h2.1 and h2.2? (or other h2's,
> depending on how many you have)
>
> ========
> H2.1 can precisely be identified
> by this property : H2.1 + unvoiced
> yields divergent results
> when other H2.x do not.
>
> Arnaud

So it impossible to find counter-examples to your law? Or what would
a counter-example look like?

Whenever we have a direct reflex of *-h2+k- in Eastern IE it's a
different kind of *h2, right? Or whenever we have *-h2g- in all
branches, it's just underlying *-h2g-. Aren't you afraid of a
circular argument here?

And isn't the lack of really good positive examples of the law
something of a peculiarity?

>
> Then why use Ir. capall ~ Lat. caballus?
> Anders
>
> ============
>
> I think the latin word is a borrowing from eastern PIE.
> But the Irish word is old enough
> to exhibit the change ?-p- > pp.

So PCelt. had a *pp, but no *p? (this must already have become */p\/,
voiceless bilabial fricative). Note that the lack of a phoneme /p/
continued until very late in Irish, cf. Latin loanwords with p > *kW

> OIr. c.

Anders