On 2008-11-11 22:05, Arnaud Fournet wrote:
> Example :
> *puH1 > pH1u
> Greek phusa, Armenian phukh, Indic phut-
> (But Basque buhatu, Uralic puj-)
Pff! Don't forget Eng. phew and (Winnie the) Pooh. An obviously
onomatopoeic root is of little use in comparative analyses, even if
something like *p(H)uh1- is vaguely reconstructible for PIE. It is good
linguistic practice to disqualify such forms as evidence if you're
trying to establish a sound correspondence. But if you insist on using
it, the Armenian 'wind' word is really <pHowkH> (< *pHuh1-ko-, according
to Olsen), and the Greek 'breath' word is <pHûsa>, both reflecting
_long_ *u:, and therefore _not_ derivable from *ph1u-.
> I think my new proposal works.
> => Your turn.
You haven't finished your turn. Your proposal "works" with two examples
and a rule designed specifically to explain their different behaviour.
There's nothing to reject here.
Piotr