On 2008-02-12 14:01, Patrick Ryan wrote:
> Would it be accurate to say that the evidence is sufficient to warrant a
> belief that
>
> *wa:-g^-, *gla:-dh-u-, and *wa:-s- came into PIE with long vowels in place?
I wouldn't divide them like that. With the possible exception of some
peripheral lexical areas (nursery words, onomatopoeia) I don't think *CV
roots existed in PIE even if the vowel was underlyingly long.
> I presume you can point to analogous roots in *e: and *o:?
As far as I can see, *e:-roots are secondary (*e:/*e ablaut is found in
roots which also display the normal *e/*o/zero alternations). I don't
thing the evidence for fundamental (non-apophonic and non-laryngeal) *o
is conclusive. It's therefore possible that all *o's, both short and
long, are derived in PIE. My impression is that *a: represents the basic
reflex of a pre-PIE open vowel in metrically strong positions, while *a
is the weak allomorph of the same, i.e. *a: is primary and *a is secondary.
One still unsolved problem in PIE morphophonology is the existence of
two acrostatic ablaut patterns: *e:/*e (as in Narten presents, sigmatic
aorists and some root nouns) and *o/e (as in *po(:)d-/*ped-). Jens
Rasmussen has some very interesting ideas concerning the latter
alternation, but I find his solutions only partly satisfying and don't
quite believe they can be considered definitive, especially if there's
any truth in Jasanoff's analysis of o-grade verbs in Hittite and IE
(Jens denies that, but I've been re-reading Jasanoff and can't resist
the feeling that he makes some really good points in his book). Whoever
is right, apophonic *o may behave similarly to *e:, revealing its
"strong vowel" character (Brugmannian length of *o in open syllables in
Indo-Iranian also suggests that it was originally some kind of tense
vowel in opposition to short/lax *e).
To sum up, I'm inclined to believe that pre-PIE had at least two
non-high vowel phonemes: a mid one ([-high] and [-low]; let's symbolise
it *E, though I don't insist that it was front; it may have been
schwa-like), and a [+low] one (let's use the symbol *A). They developed
as follows:
*E > *e (secondarily also *e:, apophonic *o(:), zero; *a and *o when
coloured by adjacent segments -- especially, though not exclusively, *h2
and *h3).
*A > *a: (sevondarily also short *a).
These two vowels never alternated with each other in one and the same root.
Piotr