o/e or reduplication

I believe there is an important oversight in the attempts to account
for the presumed o/e alternation in verbs. My own position is that
this reflects a break-down of reduplication. Not to belittle others,
I would say my most prominent opponent is Jasanoff who basically
takes the Hittite hi-conjugation at face value positing it
essentially i its Hittite structure already for PIE. I observe that
Jasanoff is not alone, far from it. To me, the hi-conjugation
reflects the inflection of the IE perfect, which in Anatolian was
used as the preterite of almost all verbs that had the vocalism -o-
or something resembling it closly enough, with subsequent creation
of a present with primary markers on the analogy of the inherited
pattern of prs. : ipf., which was also used to form secondary
presents to aorist-based preterites.

I think there is important insight buried in the Germanic class VII
strong verbs. The reduplicating strong verbs have odd vocalisms,
frequently with Germanic /a/: stautan, prt. Goth staistaut. Now, why
would that not form *stiutan, prt. *staut, if niutan naut did that,
and tiuhan tauh and others do fine with such structures? One may say
that we normally do not ask such questions, for comparative
linguistics can only know things in hindsight. Still, the odd
vocalism of the present is a strange fact which has not, but perhaps
should have, been included in the account. Why accept two
irregularities at the same time without attempting to combine them?

Now, if stautan is in origin instead an Indo-European intensive of
the type of Vedic vár-vart-mi, 3pl vár-vr.t-ati from IE *w(e)r-wórt-
ti, *wér-wr.t-n.ti, then we should posit prs. *st(e)u-stówd-ti,
*stéw-stud-n.ti. To our knowledge there was no perfect form of that
which could be used as the corresponding preterite. The perfect of
the base-verb, if treated the usual way, would end up *staut, i.e.
with the same stem as the present. It would seem that the only
preterite form IE offered was the imperfect (present injunctive).

There is one other case of a Germanic preterite reflecting the
imperfect of a reduplicated verb: OSax do:m, prt. deda, i.e. with de-
reduplication in the present and retention of it in the preterite.
The rationale of this distribution may be that the reduplication had
come to mark the preterite. That would reflect influence from the
perfect in a time when the latter was still reduplicated. Later,
reduplication was apparently given up where it was dispensable. It
could be dispensed with in the perfect which had a different ablaut
grade from the present, but the imperfect-based preterites could not
do that, so here it was retained.

In this account the Gothic preterite staistaut is the indirect
reflex of the imperfect of an intensive: *st(e)u-stóud-m. 'I poked
repeatedly'. The many i-diphthong verbs, haitan, laikan, maitan,
skaidan, etc., would have a reduplication with *Cei- which could
explain the generalized reduplication of the structure haihait by
assimilation *hei-hait- > hai-hait-. In this structure, then, I
would see the special reduplication of the intensive.

This would give the type a preterite structure that is known to have
existed. It would bring two stories of integration of reduplicated
verbs into line with each other. And it would free us of the
embarrassment inherent in the claim that the verbs if class VII are
particularly tough archaisms that have retained the reduplication,
when the even more archaic perfects found as preterito-presents
never have.

There are further observations that must be relevant here. The verbs
of class VII quite generally have a closed syllable. That makes them
ineligible for the treatment seen in class VI, the faran fo:r type.
Now, I would assume intensive origin for most of these as well,
certainly for malan mo:l 'grind'. It is plain that a treatment of
staut- forming *sto:ut- (or *sto:ud- at an earlier stage) would come
out as staut- also and so lose its marking. That must have made the
reduplication necessary. Thus, it will seem that old intensives like
malan, faran did in fact adopt the analogical perfect-based
preterite structure seen in Old Norse aka ók 'drive' (*H2o-H2óg^-
/*H2a-H2g^-, both giving Gmc. *o:k-), but that route was blocked for
stautan, haldan, skaidan etc. due to Osthoff's shortening. Thus,
their reduplication was indispensable and was retained. It may even
have been retained in a shape that could better have been produced
by the intensive than by the perfect. The ê2-forms of the other Gmc.
languages must represent a further development, perhaps based on the
weak alternants, such as *stew-stud- -> *ste:wd- with a new [e:]
created after the old /e:/ had already been lowered. A comparable
process may underlie the Hittite a/e arrangement, but I am not
saying it does. Anyway, I see no sufficent basis for a PIE hi-
conjugation.

Jens