On 2006-05-02 11:08, Mate Kapović wrote:
> Well, there's a same thing in Sanskrit (-a:sas for -a:s), but I think it
> is obviously an innovation in Old Indic, can't be an archaism.
To be precise, it occurs also in Iranian, so if it's a branch-specific
innovation, it's at least Indo-Iranian. But something like *-o:s-es,
even if not of PIE date, could have arisen more than once. One could
imagine that the original *-s was reinterpreted as a nominative ending
requiring the agglutination of an explicit plural marker, cf. acc.pl.
*-o:ns (-o:m-s).
My current opinion about the Germanic masc. them. nom.pl. is that it was
*-o:z-iz in most cases, perhaps with *-o:s-iz as a rarer variant (in
originally oxytone stems). If analogical levelling took place in early
Germanic, it was probably in favour of the dominant allomorph, so we
should start with *-o:z-iz.
In Gothic, we have the normal deletion of suffixal *i, leaving *-o:z(z).
The devoicing of final fricatives, as in *xlaiBaz > hlaifs, produces <-o:s>.
In Old Norse, *-o:ziz > *-o:r(r) > -ar.
In West Germanic (if the acc.pl. ending did not replace the nom.pl., as
in Old High German), *-o:ziz > *-o:z > *-o:s, with Ingvaeonic reflexes,
especially OE -as.
Piotr