On 2007-09-23 21:43, fournet.arnaud wrote:
> It is sure that we are dealing with a two-syllable word : *ku-H2on.
>
> PIE internal explanations will have to adjust to this fact.
You can't "adjust" PIE reconstructions arbitrarily so that *h2 becomes
invisible in this context. The problem is not only that *uh2 doesn't
yield short /u/ before a consonant but that the preconsonantal weak form
of the stem reflects *k^wn.- and can't be derived from *k^uh2n.- (which
would have given e.g. Skt. instr.pl. !*s'uvabHis rather than the
attested s'vabHis). Compare *h2ju-h3on-/*h2ju-h3n- 'young (hero)', where
the Skt. strong cases have <yuvan-> while the weak ones have <yu:n,
yuva->, i.e. precisely what the 'dog' word ought to (but fails to)
exhibit. You'd better adjust your extra-IE comparanda or give them up ;-)
> I think it matters. You are back to your letter-game addiction.
It's called the comparative method. So far you haven't proved that the
word should be analysed as *k^u- plus a suffix; you merely offer a
stipulation to that effect. The 'dog' word is unanalysable within
reconstructible PIE and, for all we know, may have been etymologically
obscure to PIE-speakers. If so, it was underlyingly *//k^won-//, with
the disyllabic variant *k^uo:n occurring only in the nom.sg., where
Lindeman's Law is applicable.
Piotr