Greek question & the pre-history of *nekwt

[Patrick C. Ryan]

Dear Yoel and IEists:

-----Original Message-----
From: Yoel L. Arbeitman <yoel at mindspring.com>
Date: Saturday, March 13, 1999 11:16 PM

> The question of what was the form of Hittite and Anatolian "drink" has
>progressed greatly since Sturtevant's time. In 1980 Morpurgo-Davies showed
>that the cognate in Hieroglyphic Luwian was /u:/ and since *gw disappears
>in HL, but *kw does not, the problem is solved. Proto-Anatolian had the
>verb *e/agw-, known so well in that famed sentence deciphered by Hrozny' nu
>NINDA adanzi nu watar akwanzi "And they eat bread and they drink water" .
>It was long thought that here was the cognate, as a verb, of Latin aqua
>"water". But now it is certain that the iterative-durative form cited
>represents /akw=skizzi/ with regressive voice assimilation.

You seem to be saying, if I understand you correctly, that
ak-k{.}u-uS-ki-iz-zi [akuSkitsi]  is derived from /akw+sk+izzi/,
representing [aguSkitsi]. Do you have other examples of devoicing a stop
before /sk/ in Hittite since, of course, it is, to my knowledge, unknown in
IE?

>And, long
>before the 1980 discovery W. Winter proposed the cognation of Latin ebrius
>"drunk"/ sobrius "sober"  (with -b- < *-gwh-) as well as Greek ne:phalos
>"sober" with -ph- < only *-gwh-. Thus, far from being cognate with Latin
>aqua, Anatolian *a/egw- is cognate with Latin ebrius "drunk", Gk. ne:phalos
>"sober" and the long ago proposed Tocharian yok- "to drink". All argument
>has been closed by the Hiero. Luw. verb which is reduced to a mere /u:/ and
>is not open to argument.

Thomas reconstructs IE *eg{w}- for Tocharian yok-; while IE g{w} normally
corresponds to Latin [v] and Greek [b/d/g], even IE g{w}h does not yield the
required Latin [b] either though it *might* yield Greek [ph]. We are being
asked to accept an IE root, *eg{w}h-, which appears *without* the expected
Latin reflex [v] for IE *g{w}h, and with a *-ri formant; and to equate that
with Greek *ne + *eg{w}h with a *-l(o) formant. While this may be correct, I
do not think that the data is so conclusive as to foreclose argument.

Also, are you asserting that IE g{w}h like g{w} appears as HL /u:/?

Pat