Torsten:
> Bla-bla-bla. Examples?
As much as "bla-bla-bla" is an annoying response, I admit
to deserving it in this instance, at least in part. The
vowel *a most certainly needs to be reconstructed in IE
on typological grounds as well as simply comparative
grounds. It's not true that all instances of *a are
the product of uvular lowering. A perfect example would
be *kad- (traditional k^ad-). This root does not show
evidence of a laryngeal, yet there's *a!
I just made a long, perilous journey out to the SFU to
check up on some much-needed reading and I specifically
looked for ablaut on *a-grade verbs. Well... it was slim
pickin's, lemme tell ya. However, I did find some tiny
morsels of glosses like Gr /ke-kas-mai/ (< *kad-) which
seems to show that *a alternated with neither *e nor *o.
It didn't alternate at all! With this type of reduplication,
you'd expect *ke-kos-mai.
Then, I started reviewing what I'd already proposed about
Mid IE and that rule makes sense. So far, I view IE *a
as being mostly derived from *e due to colouring, with
a few true instances being resistant to shifting to *o
during the Late IE vowel shift of *a > *o neighbouring
labial phonemes. The rest may have once been coloured
by uvulars but the uvulars were replaced by plain sounds,
as perhaps with *kad-. So in total, regardless of anything,
*a is underlyingly derived from the same vowels that would
produce *e or *o, the ones that participate in ablaut.
One would expect that coloured *a, or that is "phonetic"
rather than "phonemic" *a, does still participate
in ablaut, and lo and behold it does: *stax- vs *ste-stox-
(but only because *stax- is in fact *stex-). Whereas,
*kad- and *mad-, not being coloured by uvulars, are TRUELY
*kad- and *mad-. Thus they don't participate in ablaut
because *a is treated as a "second *o". At least in
regards to *mad-, this is what *a should have been if
it weren't for neighbouring *m. So *mad-, being like **mod-,
is already ablauted and therefore doesn't participate in
ablaut.
= gLeN