Re: Gemination in Celtic

From: fournet.arnaud
Message: 56848
Date: 2008-04-06

----- Original Message -----
From: "Anders R. Joergensen" <ollga_loudec@...>

I don't think I said it proved anything. However, countering with a
word which most authorities (if not all) consider post-PIE doesn't
really seal the deal in my opinion. And the evidence for your -?p- in
kobyla etc. is restricted to late Celtic forms which cannot possibly
be inherited.
A
============
It depends until when Celtic kept glottal stop an independant phoneme.
It's clear that my proposal that -?-C- > -CC- is a dvlpmnt of Celtic
I have no idea when it might have happened,
It may even be post-proto-Celtic.
Arnaud
=========

> LAtin iuba "mane" which I consider a LW from Eastern PIE
> based on *dzo?p- "tuft of hair" is short too.
> Eastern buz "bock" is short too.

What's the relevance? I talked about Winter's Law.
A
============
It shows the sequence v_?_C- > v_C+voice
does not generate a long vowel in eastern PIE.
Arnaud
===========
> Maybe it does not work with -t- ?
> only with -kk-

I guess this means that "Celtic" *potta is out as an example.
A.
=============
Not necessarily.
Because there is a certain number of unknowns :
- Insular Celtic may behave differently as say Gaulish,
- There are eight H2 and I don't have a possible example for all,
so that my intuition is that *? and *(t)s? could cause -CC-
While *&ayin doesn't.

I suggested *potta is from *kwoH2-ta
here I suppose H2 is -z- : k_w_z "to burn, to bake"
Arnaud
==========
> How do you reconstruct proto-Celtic for creicc ?

I don't. On second thought, it probably _is_ formed on ícc 'act of
paying for, compensation, atonement', as suggested by Thurneysen, GOI.
A
===========
What is the proto-form for i:kk- ?

Arnaud
============
I consider it does not work with H1,
> only when H2 precisely is glottalized (? or maybe also s.)
>
> As regards Bereza for example < *bhrH1-g-
> Can we contrast H1 and H2 in that position in Balto-Slavic ?
>
> Arnaud
> =========

No, but Lith. bre:ks^ti etc. seems like a good indication of -h1.
Anders
========
Not so clear.
the -sk- suffix can turn *bhreH2-k- into *bhreH1-k-sk-
Unsuffixed *bre:k- would be clearer.

Isn't there a Letton verb blaz- with a ?

I consider this root is *bh_s. with three states :
unaffixed bh_H2
infixed bh-l-_H2
infixed bh-r-_H2
All meaning to shine, to be bright.

Arnaud

==============