On 2008-10-31 20:06, Arnaud Fournet wrote:
> This is not far away from a minimal pair,
> anyway a minimal pair between a vowel and a consonant does not make sense
> as they are not supposed to appear in the same environments.
Well, Czech /r/ and /l/ can be syllabic or non-syllabic depending on the
environment in which we find them. No-one would say that it doesn't make
sense to claim that the medial sounds in /krk/ 'throat' and /vlk/ 'wolf'
cannot be allophones of consonantal liquids because they occur in a
different (interconsonantal) environment. Likewise, English syllabic
[l.] in <bottle> is not a separate phoneme but an allophone of /l/. High
vowels have non-syllabic allophones in many languages.
> By the way,
>
> do you have examples of -uH2o- in the standard theory ?
Of course.
> What should we expect from that ?
In most branches, we get *-uwo- or some similar disyllabic reflex.
However, *-uwV- and *-ijV- can also come from other sources (especially
Sievers and Lindeman variants of plain *-wV- and *-jV-), so if you have
*k^u[w]o:(n) as a byform of *k^wo:(n), in doesn't guarantee an internal
laryngeal (it's the same phenomenon as *dje:us ~ *di[j]e:us). It's the
zero grade that shows the difference. There's no single IE branch to
show a long nucleus in what would have to be *kuh2n-ós under your analysis.
Piotr