--- In
cybalist@yahoogroups.com, "Richard Wordingham"
<richard.wordingham@...> wrote:
> Wasn't the glottal stop (if that is what it was) real enough for
us
> to reconstruct PIE *h1su- 'good', *h1sont- 'being'? I'm open to
> dissussion in the latter case - the argument for *h1sont- rather
> than *sont- seems to lie on patterns and on the Greek participle
> eónt-. What demonstrates the *h1 in *h1senti 'they are'?
>
> Incidentally, need the glottal stop have been automatic? If my
> memory serves me right, Classical Arabic contrasts ?VCC- and VCC-
.
> Examples of the latter include _ibn_ 'son' and _ism_ 'name'. The
> initial vowels in the VCC- words do seem to be examples
> of 'automatic' vowels referred to below.
Yes, I believe *H1- is real enough to be reconstructed for the
ancestor form of Gk. eu- vs. Skt. su-. I agree that some of thee
arguments adduced to prove /H1-/ in 'be' are in fact not conclusive,
but it seems hard to dismiss /H1s-/ for the collocations with
lengthening of a preceding vowel: Skt. á:sat, á:sat- 'non-existent'
from /a-/ + sa(n)t- (besides restored ásan, ásat, ásat-); Skt.
á:san 'they were' from augment a- + secondary-ending form of
sánti 'they are. Why would the preceding element be lengthened if
the zero-grade of the root began with simple /s-/?
I have thought about the other point myself and I think now the
argument was not relevant. Even if there are no roots in IE with
initial vowels followed by a simple consonant /VC-/, there still
could have existed roots with underlying initial clusters which were
automatically realized with an initial prop-vowel as /eCCV-/ or the
like. There still is no evidence for it, and actually the failure of
the infixal o-vocalism to remain initial before such roots is
evidence against it, the only evidence we've got it seems. Perhaps
that is why this analysis of mine is considered so terrible.
Jens