28-04-2004 13:59, Miguel Carrasquer wrote:
> On Wed, 28 Apr 2004 03:14:34 +0200, Miguel Carrasquer
> <mcv@...> wrote:
> The archaic Vedic nominative vé:s must come from *HwóiHs.
> If the ending had been simply *-oi, that would have given
> lengthened nom.sg. -o:is > Skt. -a: (e.g. sakha:).
If it's indeed archaic and not back-formed from the allomorph <vi-> and
the genitive (to replace **áva:). I suppose something relatively simple
like *h2áwo:i, *h2áwim, h2wéis, h2wi[j]óm (with branch-specific
levellings) would suffice to generate all the attested reflexes, and to
account for *o-h2wi-om. Why can't the Greek and Celtic forms involve a
suffix not present in the basic form (*h2awi-eto-, related, as you
suggest, to the *-os/*-es- of Gmc. *ajjes- < *h2awjes-)?
...
> ToB eye, pl. awi (EIEC: *h2owe:is, *h2awéies)
>
> Vedic has short a- (not a:-) which must come from the
> oblique forms in *h2a-. Elsewhere, nom. -o- has been
> transferred to the oblique, except in the Toch. plural,
> which shows /a/. The laryngeal in Anatolian is more
> suggestive of *h2 than *h3 (*h3 is usually(?) lost). Arm.
> h- here can reflect either *h2 or *h3.
> The Germanic and Tocharian forms suggest the possibility of
> a heavy *oy-stem paradigm *h2ówo:ys, acc. *h2ówim, gen.
> *h2áwyos, later regularized to *owis, *owim, *owyos (*awis,
> *awim, *awyas in Sanskrit).
Adams's pre-Tocharian forms in the EIEC look monstrous. Ronald Kim
(2000) derives the Tocharian B forms much more plausibly:
*h2ówi-/*h2áwi-
--> nom.sg. *h2áwis > *awi > *awu (with assimilatory rounding, of which
other examples are offered) > PToch. *aw& > Toch.B a:[u]w ([u] =
subscript <u>), a sg. reported by Pinault. The form *eye (unattested
directly but restored from the gen. <ey[e]tse>) must be some kind of
derivative, but Kim admits that the details are not clear to him.
--> nom.pl. *h2aweyes > *aweyes > PToch. aw'&y& > pre-Toch.B ay&y& > awi
The Vedic short /a/ is ambiguous. As Jens has suggested here, it could
be analogical to the closed-syllable oblique variant as in <ávyas>, no
matter if the vowel comes from *a or from *o. Kim also quotes an East
Iranian example (Waxi yobc^, probably reflecting *a:wi-c^a:) that seems
to exhibit the regular operation of Brugmann's Law.
Piotr
Piotr