W dniu 2012-02-01 23:25, dgkilday57 pisze:
> Yes. I do not have the LIV, but Osthoff's paper in PBB vol. 15 makes
> clear that the forms with -u- are not ancient. Zehnder's invariable
> vocalism would exclude Doric <ma:khos> etc. from this group. On the
> other hand Skt. <magha-> etc. can hardly contain a zero-grade laryngeal.
>
> As a wild guess, I wonder whether *h2 (or perhaps *h4) could have
> interacted with a following *gH so as to reduce the resulting cluster.
> Verner once suggested that the sounds corresponding to our traditional
> "voiced aspirates" were actually voiced affricates before the Gmc. Lv.,
> and if *h2 was an approximant or fricative with velar articulation,
> perhaps *h2gH was actually realized as *xgG > *GgG > gG, that is, what
> we write as *gH. Then the normal grade *meh2gH- would be colored to
> *mah2gH- and reduced to *magH-. But the zero-grade would retain the
> approximant allophone *m&2gH-, falling together with normal grade in
> non-InIr lgs. and with /o/-grade *moh2gH- > *mogH- > *mag- in Gmc.
>
> Now I have to explain the Greek forms with -a:-, so I have only shifted
> the problem.
Those who accept a marginal ablaut pattern involving a primary PIE *a
usually work with something like *a: (strong grade) vs. *a (weak grade),
as in Gk. (w)ágnumi 'break' vs. (w)é(w)a:ge 'is broken' from *wa(:)g^-
(Jasanoff 2003: 31, 150). Some of the Germanic Class VI strong verbs
could belong here, e.g. pres. *skaB-a/i-, pret. *sko:B- 'scratch, shave'
(< *ska(:)bH-).
*ma:gH-/*magH- could well be a root of this type, the question is only
why the expected pattern of *maG-/*mo:G- was levelled out in favour of
the short vowel. Note, however, that Germanic has a long vowel in both
pret.sg. and pl. in Class VI, as in Goth. sko:b/sko:bun, while one would
in principle expect sko:b/*skabun. It seems that this vowel contrast was
utilised to distinguish the present from the preterite. There was no
such need in the case of preterite-presents, so instead of following the
example of Class VI they generalised the one or the other allomorph
completely; hence *o:G- and *maG-, each with the same vowel throughout.
Another similar case might be *mo:t- 'be allowed to, have to', but I'm
not aware of any extra-Germanic cognates.
I still don't know what to make of the short vowel in *(h)agHes- (in
both Greek and Germanic) if *o:G- has an old *a:. It could reflect the
generalised weak-case stem of an original acrostatic paradigm (like the
putative "Narten" stem *g^e:rh2-s-/*g^erh2-s-), but it's merely a
possibility, not a real solution.
Piotr