Gemination in Celtic

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

>
> Originally, I was trying to know if and which
> languages had pre-glottalized in PIE.
>
> Ultimately, I found this :
> - Celtic + Osco-umbrian display
> H2-C (unvoiced) > -CC- unvoiced
> - others
> H2-C (unvoiced) > -C- voiced
>
> Latin is somehow ambiguous
> but Celtic is not.
>
> Examples :
>
> pott- "pottery" < *kwoH2-t-eH2
> k_w_H2 as in Greek kaFiô "to burn"

What Celtic words are you referring to?

>
> bukk- "male"
> Tsigane, avestique buz < *bu-g-

I take it you wouldn't consider Celt. *bukko- as a loan from Germ.
where both *bukka- and *bukkan- are reconstructible, from an n-stem
with suffixal ablaut and gemination (as has recently been sugested by
Schumacher, Keltische Forschungen 2).

Do you then take the Germanic word to be a loan from Celtic? Or does
your gemination rule "H2-C (unvoiced) > -CC- unvoiced" also cover
Germanic?

>
> The point is Eastern PIE is voiced
> when Celtic is geminate.
> This is a LAW not expressive gemination.

Did you look for possible counter-examples to this LAW?

>
> lakk- "slack"
> Greek lag-aros
> One of the clearest examples.
> Cf. peH2-g/k- about the same.

Isn't this ModIr. _lag_ 'weak'. Then it has /g/ from *gg, not *kk,
whatever the etymology.

>
> Breton stuc'h "arrowhead" < stukk-
> German stechen < *st_g < *st_?k-
> skrt tud-ati "to sting" < *tu?-t-
> This example is more complex
> but the voiced -d- has -kk- in Celtic.

Breton /y/ <u> is usually from *ou (*eu) or *oi. So you need to have
this and not simple *u. Anyway, this etymology isn't exactly beyond
doubt, is it?

Anders