On Wed, Apr 22, 2009 at 3:35 PM, Eugene Oh <[log in to unmask]> wrote:

> How would you articulate /gDmes/? It really is such a thorn to me...
> But serious questions/comments:
> 1. Is there a special reason why /tt/ would become [tst] rather than simply
> [tt] or even [t]? It doesn't seem likely for a geminate to differentiate
> like that.

first of all, I don't like geminates ; ). Then, it's the postulated middle
step between the treatment of this sequence in greek and latin... to which
hittite happily agrees:

pot-sum -> possum, pot-es -> potes.

*wid-tos > vi:sus (with compensatory lengthening) in latin, vitt- in
sanskrit... the pure geminate, vista- in avestan, and a-(w)sts in greek.
So sanskrit have -tt-, greek and avestan -st- and latin -ss-. with that one
presupposes *[tst].

> 2. Indeed I would say it is unusual that [T, D] would occur in the
> environments that you have mentioned allophonically, especially when stops
> are easier to manage in precisely those environments. But I do agree that
> [kT] is a nice-sounding cluster ;)

/tk/ [kt] would merge with original [kt], that's not good. The best example
I have, I just remember, is that I say... without anyone near doing the

/ha he hi ho hu/ as [ha he i ho hu], and we doesn't even have // as
phoneme or allophone of anything in portuguese.

> 3. If /t/ became [T] I would expect a more general shift rather than
> localised occurrences.

One descendent of ausonian will make every occlusive before another
occlusive a fricative... but I didn't want to to it all to ausonian itself.
I would have:

/kt/ -> /xt/
/tk/ -> /kT/