tgpedersen wrote:
> What I (or Schmalstieg) claimed was that the original paradigm
> involved two forms: one with e-grade of root and zero-grade of
> suffix (2nd, 3rd sg., 2nd pl.) and one with zero-grade of root and o-
> grade of suffix (1st sg., 1st, 3rd pl.). You can remove the 1st sg.
> from the latter form, and you can remove the 1st pl. but languages
> hang on to having the 3rd pl. with that form. Including Polish.
In Slavic (including Polish), the reflex of the much more frequent
_thematic_ *-oNtI < *-o-nti has been generalised analogically. The same
has happened in several other subgroups of IE, even in Hittite, and
little wonder, since the contrast between *-o-nt(-i) and -ént(-i) (~
-n.t(-i) if unstressed), which arose in pre-PIE, was neither transparent
nor functional in dialectal IE, and speakers were free to confuse the
endings with impunity. There are, however, clear traces of an original
*-é-nt-i both in the 3pl. of 'to be' (*h1s-ént-i) and in other athematic
paradigms (Gk. eisi [Dor. enti], Goth. sind, Umbr. sent, OIr. it, OWel.
hint). These forms can hardly be analogical to anything, so the orthodox
view is that *h1s-enti represents the older, PIE variant.
Piotr