Re: [tied] o/e or reduplication

From: elmeras2000
Message: 32929
Date: 2004-05-25

--- In cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:
>
> And what's the evidence for o-grade in the singular? No
> non-Indo-Iranian form with /o/ is given in LIV. I can see
> that I-I intensives starting with a (labio-)velar do not
> palatalize, but that may also be consistent with zero grade
> throughout (followed by analogical gun.ation of the
> singular). Am I correct in thinking that o-grade is assumed
> based on a form like {cit}, int. ce:kit- (where zero-grade
> would have palatalized, and k- is explicable as analogical
> after a singular *kek-koit-)?

No, that is not the basis. First, that it is not zero-grade can be
seen even in Indo-Iranian. Second, the o-grade is based on such
forms as carkarmi, jaNghanti, ája:gar. It is supported by older
investigations such as Hiersche, Gab es ein o-stufiges primäres
Präsens im Idg.? from 1963. Hiersche observes that the verbs of this
class express actions that demand a particular amount of repeated
effort and concludes that they represent the intensive.

I have never seen anything to support an idea of "gun.ation" of the
singular. What would be the motive or even the model for such a
thing? If there was a stable stem throughout the inflection, why not
keep it and be happy with it? And further, is this a theory saying
the intensive has *no* correspondences in other branches? Does it
also say that the "o/e type" has no correspondences in Indo-
Iranian?

> If so, why does LIV not
> reconstruct o-grade for the reduplicated desiderative, which
> is <cikitsat> in Vedic?

Not so, and they quite possibly do not reconstruct *o here because
zero cannot come from *o.

> Or are all the k-'s in these
> reduplicated formations just analogical after the perfect?
> And if so, what evidence remains for /o/ in the intensive?

If they were, it would look this way, but why assume that?

> I obviously can't comment on the rest until I understand the
> shape of the PIE intensive.

But even before you have done that you already *know* there was an
o/e type in IE, and that the Hittite hi-conjugation is not the
intensive? It is not the first time I have made this identifcation
on this list.


> I'd look into two things:
>
> Is the reduplicative vowel a: in the Vedic perfect due to
> the root starting with a laryngeal? Of the seven examples
> given by Macdonell, only one (gr. < *h1ger-) has an initial
> laryngeal, as far as I can determine. Therefore, I don't
> think the pehnomenon can be explained away by initial
> laryngeals.

One may add a:nám.s'a (*H2e-H2nónk^-e), ma:mr.j- : Gk. amélgo:,
va:vr.j- 'twist' : Gk. eérgo: 'close', va:vr.s.s- 'rain' from
*Hwers-. There are many for which it is just not known, and perhaps
one or two for which one really knows there was no laryngeal. One
could think of a further source in assimilated structures like Latin
momordi, cucurri. Then a case like da:dhá:ra could reflect
refashioned *dho-dhór-e.

> Would a locative *udén(i)/*wedén(i) have given widá(:)n in
> Hittite? According to Melchert, it would. But so would
> *wedénas (-> *witanas).

As I understand it the process demands a closed syllable. So I would
expect it to work in the endingless locative, but not in the form
with -i, especially not if that reflects the dative. The two words
that have -a:ni are explained by Melchert as changing *-en to *-an,
then transferring that to the longer form where accented *-an-i was
lengthened to -a:ni. That is of course only needed for words that
have -a:ni. If -e:ni is found there is no problem: ú-i-te-e-ni can
be *wedéni/-énei, which however does not mean that this form is that
of PIE.

> The accent doesn't matter. Now
> that would be an argument in favour of witenas = *wednós, as
> I was arguing.

Well, for your stem analysis it does very much matter, for that is
impossible if the middle -e- is accented as it appparently is. I
have forgotten why you can't have accent and full grade in there,
but you seem to be quite phobic about it.

> For the rest, I'm not completely convinced
> by Melchert's argument (-anzi _can_ be analogical after
> -n.ti and -onti, and what about 1pl. -wén(i), 2pl. -tén(i)
> [besides -wan(i), -tan(i)]?)

I understand the old structure should be -wan, -weni, -tan, -teni.

Jens