On 2008-02-13 19:17, Rick McCallister wrote:
> And what are those reasons?
For the vast majority of roots that have *a- in the Brugmannian
reconstruction, Hittite shows <h-> if it has a cognate at all:
halkuessar (*h2algWH-), ha:ssa- (*h2ah1s-), ha:nz (h2ant-), harki-
(h2arg^-) etc. It's therefore safe to assume that Brugmannian *a-
_almost_ always represents *h2a-. *o- is more tricky, since *h3o-
accounts only for some of its occurrences, *h2o- being another common
source (both have <h-> in Hittite). Words which have Hitt. <e-> or <a->
may go back to *h1e-, *h1o- or *h1a- or reflect *e-, *o-, *a- without a
consonantal onset. But where the two possibilities can be distinguished,
we usually find some evidence of an initial laryngeal, e.g. *h1d-ónt-
with a prothetic vowel in Gk. and Arm., lengthening in Ved. á:sat- 'not
being' < *n.-h1s-n.t- and ipf. 3pl. á:yan 'they went' < *e h1j-ent. Such
evidence is of course easier to find for frequently used roots,
occurring in diverse morphological environments, but if _they_ have to
be reconstructed as *h1ed-, *h1es- and *h1ei-, at least some if not all
of less common roots surely had *h1- as well and vowel-initial roots
were at best rare in PIE.
Piotr