Miguel Carrasquer wrote:
> I just don't believe a development -oHe > -o:(u) is
> phonetically plausible within the context of PIE. oHe or oe
> (it doesn't matter, same result) give PIE õ, as in Dsg. -õi,
> Npl. -õs, etc.
I'd say that the contraction of *oHe can only be dialect-specific, not
PIE, and that it hasn't been proved that PIE _word-final_ *-o: from
contracted *-o-e was also circumflex; otherwise I agree. The *[w] in
*-o:w is just a hiatus-filling glide, originally non-phonemic and
inserted before a folowing word-initial vowel (that's why I called it
intrusive). The -a:u of the Vedic perfect has the byform -a: (paprau ~
papra: < *pe-ploh1-e). It's clear to me that they are just old sandhi
variants even if the <paprau> type eventually became grammaticalised and
generalised as the normal form of the perfect.
Piotr