On 2008-12-03 12:32, Arnaud Fournet wrote:
> This presupposes that the longest form is the oldest. There is no
> clear proof for that. It seems Mycenean already had short forms 'ou'.
> Long forms could also result from expansion of the short one.
True, but other things being equal words get shorter and shorter rather
than longer and longer in the course of time, like a(u)gustus > Fr. /u/.
The form <ouk(H)í> is orthotone, while <ou> is a conditioned
(preconsonantal) sandhi variant of <ouk>. The distribution of proclitic
<ou>, <ouk> and <oukH> is predictable, and there are good reasons to
assume that <ouk> is the basic form from which the other two are derived.
The details can be debated. I personally find *h2oju a little out of
place here if the interptetation is really 'in one's life'. I would
expect a locative, whereas *h2oju is clearly nom./acc. (presumably acc.,
to be precise). Perhaps *ne h2oju kWid should rather be understood,
roughly, as '(for) no lifetime no-matter -how-long' (i.e., never in the
foreseeable future). Anyway, *h2oju-kWid explains also Arm. oc^' and
Alb. s 'not', and an exact parallel (pointed out by Cowgill) is provided
by OIc. eigi 'not' = ei ~ ey 'always' (= Goth aiw, OE a:, cognate to
*h2oju-) + -gi- 'at all' (which may well reflect *kWi-). If Cowgill's
etymology is regarded as brilliant, it's because it reaches way beyond
Greek.
Piotr