n 2008-02-19 15:28, Patrick Ryan wrote:
> I am a (fanatic) believer in the idea that PIE roots had only two
> permissible forms:
>
> *CV, and
>
> *CVC(V).
This might be true of some pre-stage of the protolanguage, but not of
PIE. There are many roots ending in *-CC. In some cases it can be
demonstrated that the final consonant is an old, obscured suffix, and we
can even recover some features of pre-PIE phonotactics, like a
preference for "balanced" *CCeCC- structures rather than *CeCCC-. Thus,
the addition of an s-extension to a CeCC root changes the position of
the root vowel:
*h2aug^- 'grow' but *h2weg^s-
*h2alk- 'defend' but *h2leks-
*meik^- 'mix' but *mjek^s-
Piotr
===========
The longest root I have found is
z_H1_l_k "drag on the ground"
Arabic zaHlaj
With Vowel in first slot : Greek alkos
In second slot : PIE *selk
In third slot : sledge
A variant is z_H1_r_p "crawl on the ground"
hence serpent.
Phonotactic zH1 > s.
H1 is unvoiced.
Arnaud
================