On 2006-07-17 15:48, Andrew Jarrette wrote:
> And I suppose you would include here <çaye> "lies". But this has
> full grade, not zero grade like *wid-é. Perhaps <stave> has the
> reduced form of the full lengthened grade of a Narten Root, so
> perhaps for non-Narten roots the equivalent formation would have
> zero grade, but I haven't found any examples of zero grade in such
> middle formations, at least in the aorist in Sanskrit (though the
> present has zero grade in athematic verbs, albeit with the Sanskrit
> ending <-te>, not *<-a> from *-é. But we are talking here about the
> origin of thematic _aorist_ forms in Sanskrit, not present forms).
> So what is this *wid-é you hypothesize?
Sorry. My example indeed involves a Narten present, and as such doesn't
illustrate the accent shift and ablaut alternation I was talking about.
Here's a better one: Skt. 3sg. active dógdHi 'milks' (PIE *dHéugH-ti),
3pl. dugHanti (*dHugH-énti), etc., 3sg. middle duhé (RV) 'gives milk'
(*dHugH-éi), beside innovated dugdHe (as if from *dHugH-tói), 3pl. duhré
~ duhaté ~ duhraté (analogy going on and on). There is no reason why a
root aorist should have behaved differently, except that it would have
had no final tense-marker *-i or *-r. The absence of precise examples in
Sanskrit is due to the recessive character (and therefore general
rarity), of the t-less middle. Aorist middles of this type received an
analogical ending borrowed from their active counterparts: *ávida >
ávidat, more or less as in the <duhr(at)é> example, but earlier (before
the time of the oldest Indic texts).
Piotr