Re: Final -oi/-ai in Balto-Slavic

--- In cybalist@yahoogroups.com, Miguel Carrasquer Vidal
<miguelc@...> wrote:

>
> On Sun, 4 Feb 2007 18:41:58 -0800 (PST), Sean Whalen
> <stlatos@...> wrote:
>
> >--- Miguel Carrasquer Vidal <miguelc@...> wrote:
> >
> >> On Sun, 4 Feb 2007 16:42:11 -0800 (PST), Sean Whalen
> >> <stlatos@...> wrote:
> >>
> >> >
> >> >--- Miguel Carrasquer Vidal <miguelc@...>
> >> wrote:
> >> >
> >> >> On Sun, 4 Feb 2007 15:36:33 -0800 (PST), Sean
> >> Whalen
> >> >> <stlatos@...> wrote:
> >
> >> >> > There is no reason to assume oi/ai had merged
> >> >> >at the time final V() were shifting.
> >> >>
> >> >> Yes there is.
> >> >
> >> > They merge. Why before this rule, not after?
> >>
> >> Because the merger a/o is Balto-Slavic.
> >
> > I don't think so. If you believe it how do you
> >explain B-s *swopnos > svapnas and sUnU?
>

Sl. sUnU is from PIE *sup-n-o- (identical with Albanian gjum&)
On the other hand, Lith s~apnas is from PIE *sop-n-o-

Seems that sometimes the Balto-Slavic theory is not good enough...


In Latin w is lost before o or u
(PIE *deiwos > *de:wos > *de:os > Latin deus)
So Latin somnus is more probable from PIE *swop-n-o-

Could somebody explain why we have both *swop-n-o- & *sop-n-o in PIE?

Marius