On Sun, 26 Nov 2000 18:00:22 +0100, "Piotr Gasiorowski"
<
gpiotr@...> wrote:
>>> Why is there no lengthening in vocatives like *ph2ter or *k^won?
>
>> The last question is easy: the vocative had no *-s ending
>
>But -r and -n themselves are voiced, aren't they?
Yes, but they are not directly preceded by a consonant.
I'd like to take back the unqualified "voiced": *-m causes no
lengthening in *-VCm.
To come back to the "vrddhi-collective": what about the n.pl. of the
Skt. (a)s-stems, <-a:m.si>? If we remove the intrusive nasalization,
we have *-a:si, with -i < *-@2 and *-a:- most likely from *o: (*o is
another possibility, depending on the details and relative timing of
Brugmann's law and laryngeal vocalization), in any case not
comparabale to the *-es-h2 we find elsewhere (OCS -esa, Lat. -era,
etc.).
My guess would be a "vrddhi collective" *[g^enh1]-o:s, secondarily
regularized by extension with *-(n)i < *-(n)h2 in Sanksrit. Cf.
Avestan -a:h (also in the n-stems: n.pl. *-ma:n (> -ma:m)).
=======================
Miguel Carrasquer Vidal
mcv@...