On 2006-03-14 10:13, Sean Whalen wrote:
> I don't think h3 had any special voicing effect.
> Since there were few cases of b in PIE sound changes
> creating new b's are more likely.
There is some evidence for the voicing of the other voiceless stops
before the Hoffman suffix as well, e.g. Lat. ru:bi:go: 'rust' <
*[h1roudHi-h3kW]-h3o:n.
> Sporadic p>b
> between vowels seems likely (for example, if "drink"
> peih3 > pih3 / poh3 (depending on accent) the p and h3
> would never touch).
They would in reduplications, where the root is usually "clipped" in one
way or another (likewise in compounds). We have, for example,
simplification of consonant cluster through loss of a final laryngeal in
*g^í-g^n-e-ti (root *g^enh1, cf. *newo-g^n-ó-s 'newborn'), *pi-pl.-més
(root *pleh1-), *kWé-kWl-o-s (root *kWelh1-) rather than **g^ig^n.h1eti
etc. In "long-diphthong" roots (however one analyses them, i.e.
*//peih3-// or *//peh3j-//) it's the glide, not the laryngeal, that
disappears first.
Piotr