From a proterokinetic u-stem alternation
like *gWoh3u-/*gWh3ou-, we get *gWô:us (originally disyllabic with initial
stress, hence the circumflex), gen.sg. *gWous < *gWh3ou-s, loc.sg.
*gWowi < *gWh3ow-i, nom.pl. *gWo:wes < *gWoh3-ow-es, more or less as
reflected in Old Indic. Slavic derivatives containing *gov- would be based on
the oblique stem *gWow- < *gWh3-ow-. The laryngeal (when it still existed)
caused syllabification problems; this is why the reduced form *gW(h3)u- >
*g(W)w- is rarely attested except in some compounds and derivatives like Toch. A
ki < *gw-ih2- and Gk. (hekatom-)be: < *-gw-ah2.
Piotr
----- Original Message -----
Sent: Wednesday, October 31, 2001 11:18 AM
Subject: Question from the back row (Was Re: [tied] Finnish hevonen
"horse"
--- In cybalist@......, "Piotr
Gasiorowski" <gpiotr@......> wrote:
>
*gWeh3-u- (if this is how *gWou- 'cow' is to be analysed).
If so, why
does Slavic fail to reflect h3 in, eg, *gove,do (**gav-
would be
expected)? *gWeh3-u- > *gWoh3u- > *gWo-h3u- > *gWou- ?