On 2007-03-15 22:37, tgpedersen wrote:
> Countersuggestion: those cases are caused by 'reduced grade' of that
> vowel, inserted by what Glen used to call 'paradigmatic resistance',
> cf Hittite e:dmi, adanzi, not **danzi.
A prosodically weak vowel could perhaps work for initial *h1, but what
about *h2 and *h3, with their consonantal reflexes in Anatolian? They
all behave in the same way as regards vowel-lengthening in compounds,
etc. Then if Olsen is right (as I think she is) about the
"preaspirating" effect of *h2 and *h1, they belong to the same natural
class (presumably of fricatives, since *h2 was quite clearly a fricative).
Piotr