On 2008-12-03 21:50, Arnaud Fournet wrote:
> Well,
> If this principle were true, we'd be left with only short words like /u/
> after 200 000 years.
I said "other things being equal". Getting shorter is easier than
getting lengthened. Entropy takes care of that.
> Sorry,
> but I've not fully grasped what these "good reasons" are !?
> Can you think of another word that displays that kind of k/kh alternation ?
Have you really had a look at the distribution of the three variants? I
strongly suggest you should do so if you want your opinion to be taken
seriously. Here are some hints: <ouk> is found before vowels; <oukH>
before words with initial /h/. Aspiration in sandhi is common in Greek:
*kata-hodos --> kat'-hodos > katH-odos 'descent'; but _deaspiration_
before vowels is neither attested nor phonetically motivated. It follows
that
(A) <ouk> is the basic form and <oukH> is its conditioned allomorph.
(B) <oukH> is the basic form and <ouk> is its conditioned allomorph.
(Underline the correct answer.)
As regards <ouk> versus <ou>, the latter is found before consonants
(including an etymological digamma). Again: the loss of a final
consonant before another consonant is both natural and common
cross-linguistically (cf. Eng. an ~ a or the French liaison rules). I
leave the final conclusion to you as a homework assignment.
Piotr