> I won't exclude there are other avenues, but hardly this one. We are
> talking about the creation of a preterite of the subjunctive
Subjunctives do not have preterites. By definition in IE languages they
lose any time reference. They do retain their aspectual reference.
I agree that an original short vowel *-em *-es *-et is much less likely than
an original long vowel stem *se:-. But I still think some other
analogical source is better than your suggestion of analogy from a form that
is not attested, which includes the reconstruction of a grammatical marker
not found anywhere in the language!
>it cannot matter much what vowel length the subjunctive non-preterite
> has, since its stem is plainly not involved.
I think the stem is irrelevant for analogy. Analogical pressure does not
have to be from forms with the same stem. Since a subjunctive -em -e:s -et
etc already exists, it sits in contrast with every other subjunctive, and
can supply anaological pressure. Any other subjunctive would be under
pressure from the three sources of analogy I mentioned:
(a) no other verb form has short -es (except the present of the verb to
be)
(b) all other subjunctives have a long vowel before the -s.
(c) no other paradigm has a vowel change -em -is -et.
But really we should be looking for a better source of an original element
*-se:
>And the 1sg and the 3pl
> which are invoked as models for a short -e- would have had *-om and
> *-ont if the idea is that the ipf.sbj. was the s-aor.sbj. with
> secondary endings.
Was it? The s-aorist subjunctives (really optatives) seem to have given
the Old Latin forms faxim, axim, ausim, negassim, habessit, ambissit. etc.
Besides, an s-aorist should lengthen the vowel stem, but we have a short
vowel retained in darem < *da-se-m. Rather the imperfect and pluperfect
subjunctives must be seen as a new development within Latin. Besides, the
aorist is an inappropriate aspectual element in an imperfect, which shows
incompetion, continuity, or re-iteration. It belongs better where we find
it, with the completed anterior tense. That the imperfect subjunctive is
felt within the language to be made up of infinitive + -em, -e:s -et is
shown by the fact that even irregular infinitives show this construction.
Peter