I see it the other way: *-gj- > -z- is the normal historical
development (cf. megas : me(i)zo:n), but -z- and -ss-/-tt- often vary
in present-tense verb stems, because both -z- < *-g(^)j-, *-gWj-, *-
dj-, and -ss-/-tt- < *-k(^)j-, -kWj-, *-g^Hj-, *-gWHj-, *-tj-, *-dHj-
behaved like "suffixoids" or stem extensions, not like the output of
synchronic phonological rules (such massive mergers must have
resulted in morphological opacity). The confusion operates both ways,
e.g. we have Aeol. pta:zo: beside Ion. pte:sso (< *pta:-k-jo:),
whereas ple:sso: occurs beside (Att.) plazo: /plazdO:/ < *plang-jo:
(cf. ple:ge:).
Piotr
--- In cybalist@..., "Richard Wordingham" <richard.wordingham@...>
wrote:
> One thing puzzles me. My Greek grammar implies that -gj- normally
> became -ss-, exceptionally -z-. (It also fails to mention -tj-!)
A
> possible example is plé:ssein 'to strike', which Scott and Liddell
> associate (perhaps incorrectly) with 'dasplêtis' = 'horrible'.
> Obvious forms associated with plé:ssein show -g-, and the Latin
> cognates (e.g. pla:ga 'blow' and plangere 'beat noisily'). It
seems
> one has to look to Old Norse flengja 'flog' to find -k-.