On 2006-01-17 13:26, Piotr Gasiorowski wrote:
> An afterthought: do Greek -i(:)[*h]o:n, Gmc. *-izan- reflect
> *(-i)-h1s-o:n as an alternative to *(-i)-h1o:s? This might explain Gmc.
> *-o:zan as *-o-h1s-o:n (with thematic *-o- restored from the basic form
> of the adjective).
The next question is whether the superlative in *-isto- (but Skt.
-is.t.Ha-) should be understood as the comparative ending in the nil
grade extended with *-to- (an analysis usually taken for granted). The
short *-i- in all branches that show the inherited superlative speaks
against *-i-h1s-to-, but I'd like to offer an alternative: an *-i- stem
(as in comparatives) compounded with *-sth2-o-. Thus, comparatives and
superlatives would be original compounds involving, as their first
member, an abstract noun related to the adjective in question, such as
*krétu-s 'strength', corresponding to *kr.tú- 'strong', with the
compositional form *kréti-, hence *kréti-h1o:s (or *kréti-h1s-o:n)
'stronger' and *kréti-sth2-o-s 'strongest'. The precise semantics of the
'be' and 'stand' elements still eludes me, but I'm still thinking :)
Piotr