On 2006-09-26 20:01, tgpedersen wrote:
> It seems to me that this means that outside of the perfect, the rule
> is that PIE o in a open syllable, *before a consonantal resonant*,
> becomes a: in Skt., but that in the perfect, the latter condition
> does not apply, only the former. Did I misunderstand something?
What Lehmann means is that Brugmann's law operates regularly before in
non-final open syllables before resonants (Holger Pedersen) both in the
perfect and elsewhere. Apparent exceptions are caused by laryngeals that
were still there at the time Brugmann's Law operated in the ancestor of
Indo-Iranian, affecting the syllabification:
1sg. *kWe-kWór-h2a (closed syllable) > Ved. cakára
3.sg. *kWe-kWór-e (open syllable) > Ved. caká:ra
*g^onh1-éje-ti (closed sullable) > janáyati
but
*mon-éje-ti (open syllable) > ma:náyati
Unlike (H.) Pedersen I think Brugmann's Law did not require a resonant
-- it was conditioned solely by syllable structure, not by classes of
segments. So, for example, the vowel of RV á:pas 'waters' is long
because the PIE form was *h2óp-es, with an original *o and an open
syllable, while the weak cases like gen.sg. apás contained *h2ap-
(/h2ep-/), hence no lengthening; the same holds for nom.pl. <pá:das>
(*pod-es) vs. gen. sg. <padás> (*ped-os) or the related noun <padám>
'footstep' < *ped-o-m (Gk. pédon, Hitt. pedan 'floor, ground', etc.). In
<rátha-> there is no lenthening because the root had a final laryngeal
(*róth2-o-), making the initial syllable closed. In earlier discussions
on Cybalist I have suggested that <páti> comes from *pot-h1-i-. Some
minor exceptions are more recalcitrant, but they can all be explained
one way or another.
Piotr