Re: Laryngeals: arguments from typology?

W dniu 2010-04-08 04:34, gprosti pisze:

> The second element of dvi:pa- begins with a vowel in its independent
> form (ap-). The second element of dvidhA- begins with a consonant in its
> independent form (dha:-). The two-mora pattern may not have applied to
> the latter type of case.

But why, e.g., <su:nára->, if the second element is nar- < *h2ner-,
never with an initial vowel in Indic? And why is there no "analogical"
lengthening if the root is of the shape *CeR-, with the muta cum liquida
normally syllabified together, as in ágHnat- < *n.-gWHn-n.t-?

Piotr