From: Richard Wordingham
--- In cybalist@..., Piotr Gasiorowski <piotr.gasiorowski@...>
> As for the absence of palatalisation, the reconstruction *gwist-
assumes an initial cluster (*g + *w), NOT a labiovelar, and the
expected reflex of *gwi- in Albanian is <gi->.
Actually I was being sloppy. I hadn't realised that PIE had a
three-way _word-initial_ contrast gW v. gw v. g^w, spelt gw v. gv v.
g'v in http://flaez.ch/cgi-bin/pok.pl
. Pokorny reconstructs gW, but
based only on Celtic and Germanic forms, which don't distinguish the
three initials. If PIE *gwistis > Albanian gisht is correct, you have
refined the reconstruction to *gwistis (Cybalistic) / *gvistis