tgpedersen wrote:
>>>By Jens' rule, whatever its phonological cause, that -e- is
>>>preferred before unvoiced sound.
>>
>>But Jens' rule is pre-PIE, not post-PIE.
>>
>
>
> Which would be a problem, again, if the semi-thematic paradigm didn't
> belong to pre-PIE, which is does, by hypothesis.
It IS a problem for you, since you're trying to explain the origin of
the THEMATIC pattern as a "generalisation" from the earlier semithematic
pattern.
Here is the orthodox scenario:
*bHér-e-t-, *bHér-o-nt- (with visible effects of Jens's thematic vowel
rule) > Proto-Italic *feret, *feront > Lat. fert, ferunt (with loss of
*e, but not *o). Most of the other IE branches simply retain the old forms.
Where is the problem alluded to above?
Now if we assume that Schmalstieg is right:
*bHér-t-, *bHr-ónt- > *bHér-e-t-, *bHér-o-nt- (with a whole complex of
analogical levellings including the insertion of -e- in the "new
thematic" forms). The colour of this *e can hardly be explained with
recourse to Jens's rule, which is no longer operative at this stage.
Neither late PIE nor the early IE dialects show any evidence of *e being
preferred to *o before voiceless consonants.
Piotr