From: fournet.arnaud
Message: 55512
Date: 2008-03-19
----- Original Message -----
From: Miguel Carrasquer Vidal
It's no more difficult to believe than that the same
distinction was maintained in Greek, Armenian and Tocharian,
where *ih2/3 and *uh2/3 yield *ya(:) [~ *yo(:)] and *wa(:)
[~ *wo:], but *ih1 and *uh1 yield *i: and *u:. This soundlaw
was discovered in 1970 by E.D. Francis and independently in
1977 by R. Normier.
In Balto-Slavic, we must have had:
1) sjuh1-láh2- > sju:lá: > sjú:la: [Hirt] > CS s^i"la
2) bhuh2-láh2 > bhu&-lá: > bu:lá: [no Hirt] > CS bylá
3) pih3-láh2 > pi&-lá: > pi:lá: [no Hirt] > CS pilá
=======================
Chinese xiu4 < sjiuH "embroider"
provides support for *s_y_w_H1
(it cannot be a cognate of course)
Now the problem is
Why is neighboring Uralic without
any trace of H1
and we got to go as far as China
to find a loanword supporting H1
in PIE ?
Arnaud
=================