On 2008-12-20 12:34, tgpedersen wrote:
> The alternations going the way they do in these glosses, that would
> mean an alternation between stressed o-grade and pretonic o-grade.
> That is not possible.
It _is_ possible. For example, *tómh1os type nouns (in which the initial
accent was secondary anyway) had fixed root accent, but their
collectives had contrastive accent on the *-ah2 ending. This kind of
accentuation remained productive longer than quantitative ablaut, and so
it didn't cause the reduction of the root vowel. Besides, Proto-Germanic
levelled out the vocalism of noun forms in most instances.
Some of the Gmc. *a's reflect PIE *a's, especially where coloured by
*h2, as in *h2antio-
Piotr