On Tue, 23 Sep 2003, P&G wrote:
> >In *pótis there is
> >no way the /o/ could be H3-triggered, so we *do* have Sanskrit
> >unlengthened reflexes of IE /o/ from other sources than /H3e/.
>
> The claim (Brugmann's Law) is that only the ablauting -o- was lengthened.
> So what you say here is not relevant.
What claim is that now? The claim made on the basis of presumed regular
short /o/ in *H3ew-i-s and *H3ep-os will have to be that the coloration
product of /e/ by preceding /H3/ was significantly different from /o/ from
other sources in the protolanguage, and that the difference was retained,
of all places, in the pre-Indo-Iranian language stage in which *e, *a, *o
had not yet coalesced into the later common product *a.
Lengthening by Brugmann's Law does not distinguish between /o/ (1) in
suffix syllables like ra:j-a:n-am, ra:j-a:n-as from *-on-m, *-on-es; (2)
in the thematic vowel as bhar-a:-mas; (3) in the infix-based first part of
causative-iteratives, as ma:n-áya-ti; (4) in the reduplication-triggered
vocalism of the perfect, as ca-ká:r-a; (5) in a root with -o- as the short
counterpart of a nom. -o:-, as pád-am.
Are we now to assume that, while these were all phonetically identical,
both the product of *H3e and "non-ablauting o" fall outside of the rules
applying to /o/ of practically all sources we can think of?
The many o-types that do abide by Brugmann's Law have only one thing in
common, which is the sound o. Then, that would seem to be the way /o/ was
processed in this language. How can a vowel that is *always* /o/ be
excepted from these rules? I find all of this extremely hard to believe.
Do we really, really know that the first syllable of apas-, opus was
closed in PIE? That of anas-, onus is. From a list of several reasons I
can reveal Gk. ónomai, aor. ó:nato 'reproached' (the opposite of
'exonorated', cf. also OIr. on 'a blemish') which points to *H3enH2-, and
Hitt. annia- 'make' with -nn- from -nH-. If "*H3ep-" could likewise in
fact be *H3epH- of some sort, and Skt. ávis- is either *H2áwis or *H2ówis
with analogical short -a- from inflected forms (gen. ávyas), there would
be no safely regular examples of unlengthened o in open syllables in
Indo-Iranian. Why does this very weak aspect of the conditioning have
such a strong appeal?
Jens