>> Well, if we accept that there were original long vowels then I don't
>> see the need for the laryngeal.
>You're forgetting that laryngeals do not only occur in the
>combination *VH. We also have combinations like *HV and
>*CHC and *HC-.
And also -RH-, which explains Greek alternations (eg -ara / ra:)
There are also alternating patterns, where the usual word combinations
produce sometimes -VH- and sometimes -CH-. The laryngeals give a simpler
explanation of a whole range of phenomena.
(a) Sanskrit class 9 present tenses:
*-neH- / *-nH-
so we get -na:- alternating with -niC- and -nV- (from -nHC-
and -nHV-)
This is harder to explain from an original -ne:-
(b) The alternation of -t- and -th- in the paradigm of Sanskrit pantha:h,
and its Avestan cognate, becomes originally regular if we take it back to a
root *p(o)nt(e)H. You can't do this with long vowels.
(c) Thr root *ghrebh (Pokorny 455) has problems - but a root *ghrebH
explains the phonteic outcome in Germanic, the morphological outcome in the
Sanskrit present, and the Lithuanian accent. Trying do that with a long
vowel!
I could go on - but actually most of this stuff is well known, and the
argument that we don't need laryngeals for PIE is no longer widely held.
Peter