alexandru_mg3 wrote:
> 1. Is not better to consider that r.,l.,m.,n. appears only in Later
> IE as a vocalization in 'some circumstances' of r,l,m,n ? In this
> later case what could be these 'circumstances'?
It seems there was a _pre-stage_ of PIE, approachable via internal
reconstruction, in which the nasals and liquids were asyllabic in all
positions. However, the comparative evidence is clear: in the most
recent common ancestor of the IE family (= PIE by definition) a nasal or
liquid had a syllabic allophone if there was no vowel adjacent to it,
i.e. if it stood between consonants or at word-edge next to a consonant.
More or less the same is true of glides and laryngeals, and there are
some additional rules governing the behaviour of such consonants when
they stand next to one another. Usually (barring a few special cases)
the second of them becomes syllabic if they are both interconsonantal,
e.g. /-twrt-/ is realised as [-twr.t-] rather than [-turt-]. Note also
that initial *wr-, *wl-, *ml- and *mr- are permissible (the first member
of the cluster does not become syllabic).
Sometimes syllabicity is optional (or perhaps originally governed by
some rhythmic preferences regulating PIE phrasal phonology. This is the
case with the so-called Lindeman forms -- disyllabic variants of
monosyllabic words like *k^wo:n, *dje:us, which could be realised as
*k^u[w]o:n, *di[j]e:us. On the whole, syllabicity was predictable from
the structure of the word, so it was allophonic rather than phonemic.
Some authors make a principle of not distinguishing syllabic consonants
from asyllabic ones. This, however, is didactically unsound. I prefer
reconstructions which show the phonetic details we can recover. It gives
you a better grasp of what PIE was actually like.
Piotr