But both changes are entirely ad hoc.
Germanic did not drop unstressed initial syllables or perform the kind of
dissimilatory-loss-cum-compensatory-lengthening you describe. You can't derive
all Germanic preterites from reduplicated perfects without manipulating the
material in unacceptable ways. What about <was, were> from OE waes,
wae:ron < *was, *we:z-? The plural we:z- cannot possibly derive from
*we-ws-', since that would have given *weuz- > *we:or- in the lineage leading
down to English. What about the most transparent strong-verb paradigms of all,
Classes I-III?
dri:fan, dra:f, drifon, drifen
hle:opan, hle:ap, hlupon,
hlopen
singan, sang, sungon, sungen
<sungon> goes back to *sngWH-,
definitely not *se-sngWH-'. There is no trace of lengthening in these classes.
The reason is plain -- all these preterite plurals derive from _unreduplicated_
nil-grade stems. I'm inclined to think that the pattern <saet/sae:t-> goes
back to *sód-/*sed-' (here *e is the "weak" counterpart of *o in a CVC root),
with some kind of vrddhi affecting the plural stem (the influence of the
lengthened-grade aorist, most likely).
Piotr
----- Original Message -----
Sent: Tuesday, April 24, 2001 10:20
AM
Subject: [tied] Re: The potentially
non-stative nature of *es-?
--- In cybalist@...,
"Piotr Gasiorowski" <gpiotr@...>
wrote:
> But in this Germanic verb class the singular has a _short_
vowel;
and in this particular case the original reduplicated plural was
_not_
*CeCVD- but CeCD- (to wit, *se-sd- plus a stressed inflectional
syllable).
I guessed that in the singular *{sesód} > *[sesát]
the unstressed
initial syllable simply dropped, but in the plural with the
stress
elsewhere *{sezd-} > *[sest-] and then the second occurrence of
the
reduplicated consonant dropped with compensatory vowel lengthening
due to some sort of dissimilation effect. There are instances in other
languages of a `vowel consonant' becoming `long vowel' before another
consonant.