(1) In a transparent compound one would not expect word-internal
treatment of *dH (see <pontifex> etc.).
============
I agree that ponti-f-ex
makes sacer-d-os looks strange
if related to *dheH1
but sacer is itself not an inherited word.
So ??
Arnaud
==============
(2) Unextended *dHeh1- was lost from Italic.
===========
Fac-ere can be from *dh°H1-
with hardened H1
No obvious need for Erweiterung -k-
Arnaud
=================
(3) By positing *-doh3-t- we escape both problems.
=============
At least one.
Although *doH3- and *deH3 is the same result.
Arnaud
================
(4) This *doh3-t- is parallel e.g. to *pleh1-t- in <locuples>. The
vocalism of *dHoh1-t- would not be parallel to anything.
===========
All the more reason to posit *deH3
maybe
Arnaud
===================
The view that *doh3- is more likely here is not only mine.
> "once a phoneme, always a phoneme"
> *vanishingly* rare is subjective.
I'm not saying that PIE *o wasn't a phoneme. Of course it was, but in
nearly all cases it was morphophonologically related to *e in the same
root, e.g. *wo:kW-s vs. *wekWos (*-es-). "Derived" is not the same as
"allophonic".
===============
What does "derived" mean ?
I have much contempt for "morphophonology" as a concept.
Arnaud
================
> I am still waiting for a clear case of *CaC-vowel
> constrasting with *CeC-vowel or *CoC-vowel.
If the cases I've presented so far are not clear enough, explain why you
think so.
Piotr
===========
wa(:)g is bad because of wra(:)g
same meaning
erroga exist
rôks "breach" exists
wa(:)stu is unclear
because it can be *ba?-t-tu "housing"
ghladh- is so uncontrollably complex
it proves nothing.
You wrote you had *lots of* examples
Why don't you give a clear clean example in
a CAC-vowel structure ?
Where no suffix is polluting the situation.
Arnaud
======================