On 2008-02-11 22:37, fournet.arnaud wrote:
> I believe there is sufficient evidence to reconstruct PIE
> *a as an independent phoneme
> Piotr
> ==========
>
> I definitely disagree
>
> What is your sufficient evidence ?
>
> Arnaud
> =================
Lots of unpredictable *a's that have to be considered phonemic. If *a
were always conditioned by an adjacent laryngeal, it would be best
analysed as an allophone of *e. However, it occurs also as the outcome
of sporadic (and therefore unpredictable) retraction of *e next to
"plain velars" (as in *kan-, *kap-, *bHag-, *magH- vs. *(s)ker-, *sek-,
*legH-) and in other roots, where it isn't conditioned by anything
(*wa:stu-/*wastu- 'settlement', *k^asó- 'grey', *gHla:dH-u-/*gHladH-ro-
'smooth', *hna(:)s- 'nose', *wa(:)g^- 'strike' etc.).
Piotr