W dniu 2012-02-09 21:59, dgkilday57 pisze:
> > Not as per Olsen, who claims that the laryngeal "preaspiration" was a
> > PIE phenomenon. The development of *l.h2 > *la: is of course dialectal.
> > One can therefore imagine something like *tl.h2-tlah2 > *tl.h2tHlah2 >
> > *t&tHlah2 (dissimilatory loss) > ... > tabula.
>
> What are the parade-examples for Olsen's dissimilatory aspiration? If it
> explains the short vowel of InIr for 'hoof', it would seem that the *h2
> was absorbed going into PInIr.
As far as I remember, she does not discuss <tabula>. What she says is
that the non-syllabic allophones of *h1 and *h2 (but not *h3) produced
aspiration in a following stop. The best examples involve *t, especially
in the instrument-noun suffix *-tl/ro-, e.g. Lat. pa:bulum
(*pah2-t[H]lo-), cri:brum (*krih1-t[H]ro-) vs. po:culum (*poh3-tlo-).
She isn't very specific as to what happened to the laryngeal. Sometimes
it seems to have been absorbed into the aspirated stop, but the showcase
examples (including also *tl.h2-tó-) usually keep it. Given the tendency
of the PIE ictus to be shifted in collectives and related formations,
there was, somewhat unfortunately, plenty of room for analogical
levelling, e.g. *stáh2-tHlom/*st&2-tláh2 --> st&2tHlom > stabulum
(possible but not really provable). The above suggestion concerning
<tabula> is my own, not Olsen's, and I'm solely responsible for any of
its defects. It strikes me, though, that if there ever was a real
tongue-twister, *tl.h2tHlah2 would have been one. Its reduction to
*t&tHlah2 would be like a full reduplication being reduced to a partial one.
Piotr