Re: Gemination in Celtic

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

>
> So
> we have left with the following words :
>
> -kk- versus -g-
>
> mrekk "speckled"
> marga

I don't think a word mrecc exists in OIr.
We have brecc 'speckled' < *brikko- and mrecht 'variegated,
diversified'. Strange as it may seem, they're not compatible.

>
> bhukk- "bock"
> bhug

Which may be a borrowing from Germanic. The Avestan length is
apparently secondary.

>
> bhelH-k- "beam"
> bhelH2-g-

It would make it much easier to figure out what you mean, if you
could provide some examples from actual Celtic languages.

>
> You don't like
> makk / smag

No, I think the Celticity of mak- is much too uncertain to allow any
conclusions.

>
> What about
> LAtin stercus = sterganos ?

I thought the law didn't apply to Latin?

> O. irl menicc = many ?

This, indeed, is a case of Celtic *kk. How do you reconstruct the PIE
form (and especially the necessary sequence of *-h2g-)?

>
> creicc "buy" < kwriH2-k-
> Greek pri-a-o

This word, along with reicc 'selling' are interesting, and I don't
think Thurneysen's explanation is immeditately convincing (GOI 454:
attraction to _ícc_ 'healing). You posit *kWrih2-k-... does you rule
also work with voiceless stops? If not, what is the motivation for
positing *kWrih2-g-?

>
> -tt- versus -d-
>
> You don't like
> pott / catina
> kass- / kad

I think the two examples might be mutually exclusive...

>
> What about
> O.irl deateach < *dett-
> *dheuH- "smoke"

How do you account for the vocalism, for one thing?

>
> -pp- versus -b-
>
> capp-al "horse"
> caballus / kokyla

This is indeed a difficult word. I have no explanation for the
consonantism, other than that it is a borrowing from an unknown
language, independently into some IE languages. It is not optimal,
admittedly.

Anders