--- In firstname.lastname@example.org, enlil@... wrote:
We might also question why we shouldn't
> have then seen *is-em instead, if Latin speakers knew enough to say
> /id-em/ instead of *id-dem. The explanation above has a few holes
> I can see *iddem reduced, even if irregularly, to /idem/ but the
> way is logically problematic.
Blame that on the speakers of the language. Resegmentation is not
uncommon in documented linguistic change. Why ban it from prehistory?
There is d + d in Latin cre:do: . The regular product then appears
to be *[dzd] > *[zd] > length + /d/. Why is an irregular reduction
of "*iddem [...] to /idem/" less of a problem than an irregular
replacement of -em by -dem?