Perf. redupl.

From: tgpedersen
Message: 43159
Date: 2006-01-30

--- In cybalist@yahoogroups.com, Piotr Gasiorowski <gpiotr@...>
wrote:
>
> On 2006-01-27 16:12, tgpedersen wrote:
>
> > As Sihler remarks, a form like *se-sokW-e has three full grade
> > vowels. That can't be original.
>
> The last full vowel is desinential, and perhaps originally clitic.
As
> for the *e..o pattern in the perfect singular, there's much room
for
> speculation: it can go back to *e:..ó (length from original full
> reduplication, *sekW- > *se:-?)

So: sg. *sekWsokW-, pl *sekWskW-, or what?

or to *é..o(:) (from a nominal formation related to the perf. act.
part.).

But why the reduplication then?


> > That's one reason I think
> > reduplication in the perfect singular is mostly created by
analogy
> > with the plural. The other reason is that, as Miguel notes
> > reduplication is mostly to represent the action done several
times,
> > ie either by several subjects (plural) or by one subject several
> > times (iterative) (actually M. believes that PIE perfect was
> > ergative and that congruence was with the object; therefore
> > reduplication in the sg. of the perfect is no problem for him. I
> > think my solution is simpler).

>>So there was (almost) no de-
> > reduplication in the singular.
>



> Gradual dereduplication in Germanic is directly attested in Class
VII,
> and since the definition of that class is based solely on
phonological
> criteria, not e.g. on some shared semantic features, it seems that
> dereduplication generally proceeded depending on the phonetic
> environment and the ablaut pattern of the verbs in question,
leaving
> untouched the root vocalism of *CeRT- verbs and inducing some
changes in
> the weak-grade forms of *CeR- and *CeT- ones, while roots of the
shape
> CoC- were affected later, and the CoRT- and Ceh1T- ones retained
> reduplication down to historical times.
>

That last statement was too hasty. Any way I can propose it
conservatively: all perf. sg.'s were reduplicated in analogy with
the plural, before dereduplication set in.


Torsten