Hallo!

On Thu, 8 Oct 2009 17:56:02 -0400, John Vertical wrote:

> >On Thu, 8 Oct 2009 18:37:50 +0200, Jörg Rhiemeier wrote:
> >>It all works out nicely if one
> >>assumes a shift like this:
> >>
> >>*t  > *th > *t
> >>*t' > *t  > *d
> >>*d  > *d  > *dh
> >
> >Are there solid reasons for positing that middle step?  I had always
> >imagined [t_> d] > [?d) d] > [d d_h\].
> >
> >Alex
> 
> Likewise. And it doesn't leave much room for explaining Winter's Law, Danish
> stød, Lithuanian broken tone, implosivs in Sindhi and whatever Kortlandt may
> have come up with last week by lingering preglottalization in PIE — IMHO the
> best option is *?B vs. *B, with the two series mostly then either merging,
> or the former pushing the latter to *Bh before itself deglottalizing.

Indeed, the solution suggested by Alex Fink is better than mine
(and I think that's also what Beekes has suggested).

> I'm not sure what this implies for Germanic; taking the voiceless reflexes
> of glottalics as retention does not seem possible. It also occurs to me
> Armenian could have re-ejectivized them under Semitic influence. But I don't
> kno how tenable it would be to reconstruct ejectivity for Classical 
> Armenian. 

I think there is quite solid evidence for the Germanic and Armenian
sound shifts being innovations rather than retentions.  I have read
somewhere (forgot where) that the oldest stratum of Iranian loanworlds
in Armenian underwent the shift, and western European substratum words
such as _apple_ did undergo the shift in Germanic, too.

Classical Armenian almost certainly did not have ejectives; they are
found only in eastern dialects, which evolved them under the influence
of neighbouring Caucasian languages such as Georgian.

... brought to you by the Weeping Elf