>long `e'" -type Latin perfects and Germanic preterits plurals came
from reduplicated forms CeCVD
As might be expected, the long vowel perfects have more than one origin, and
so do not form a distinct class in PIE. Some come from reduplicated
perfects, some from root aorists. Using Latin as a starting point:
(a) Reduplicated perfects from laryngeal stems.
emi, edi, -epi (in coepi) < he-hm-, he-hd-, he-hp-
Here we should add egi < he-hg- perfect of the stem hg-ye/o > aio, and
later shifted to form a new perfect of ago, with which aio was in some forms
identical. (Confusion between the verbs shows in other ways as well).
Note also that the vowel does not have to be -e-: The word ico (strike)
has perfect i:ci, < he-hik or hi-hik and distinct from the iacio ieci word.
(b) Reduplicated forms from non-laryngeal stems. The best example is
sedi < se-sd- by phonological rules internal to Latin.
(c) Long vowel perfects of CeC roots. This is the pattern found in Gothic
qam / qemum etc, Latin veni, legi, regi, fregi. This pattern might go back
to PIE, but is hard to explain. Possible *gwe:m is for gwe-gwm-.
(d) Straight analogy, eg in pango pegi after frango fregi, perhaps capio
cepi after apio epi.
(e) Roots with medial i u h. The sequences -ei- -eu- eh- developed into
a pure long vowel. The Latin perfect is probably a PIE root aorist, so
fe:ci < *dheh-k-, ieci with -h-;
vidi < *weid- (aorist) & the homonyme *woid- (perfect);
fugi < *bheug-, fudi < *bheud-
These forms could be taken back to an -o- grade perfect, but Latin actually
has very few of these!
Peter