On 2006-06-20 23:41, Richard Wordingham wrote:
> Don't we have (apparent) devoicing in more than two branches? Besides
> Germanic and Armenian, we have Tocharian, Anatolian and possibly
> Thracian. I'm not sure how consistent this is with a notion of
> diffusion in the development of *[k'] to /g/.
I Tocharian, all rows of stops fall together, and traces of Grassmann's
Law in Tocharian (demonstrable omly for *dH) show that after the loss of
aspiration *dH merged with *d. Since PToch. *d was lenited out of
existence before sonorants, it must have been some sort of [d] rather
than an ejective or a plain voiceless stop. There is therefore nothing
exotic about the pre-Tocharian system: it was most likely just *t, *d,
*dH. In Anatolian, we have a two-way contrast between voiced and
voiceless (or lenis and fortis) stops, with a merger of *d and *dH on
the voiced/lenis side. How is that supposed to resemble Germanic or
Armenian?
Piotr