--- In
cybalist@yahoogroups.com, enlil@... wrote:
> Jens:
> > I agree completely, except that I do not know that *psd-éye-ti is
> > *not* regular. That depends on how the rule is formulated and how
> > other examples look if there are any.
>
> As I said, if a rule covers a clear majority of cases and you have
> something like *psd-eye- that defies the rule, it's reasonable to
> conclude that *psd-eye- is irregular.
That may not be enough. The rule you suggest can easily be
reformulated so as not to cover the example *psd-eye- which is then
no longer irregular. Still, it is a lone example and not necessarily
a correct one. That question however, should be decided on the basis
of the facts.
> It's not reasonable as you do
> to pick out one irregularity and to question senselessly whether
the
> overwhelming _minority_ is representative of the regular pattern!
Is this grammatically formed as intended? Do you really
mean "question .. whether"? I don't follow.
> You've accepted my account of the forms you posted and yet because
> one root doesn't conform, you will throw out the baby with the
> bathwater.
I will if I have to.
> Why on earth do you persist on siding with statistical
improbabilities?
I will accept anything demanded by the facts. I am not trying to
make IE or its prestage pretty, I'm just trying to get it right.
That however may be a much more complicated matter.
> That would be like concluding: "Well, yeah, that Grimm guy makes
some
> sense with that nifty 'p>f' law he has there, but I got an itching
> feeling that Latin 'pater'<=> English 'pastor' represents the true
> regularity."
That is not at all parallel.
I am not ungrateful for your attempt to make head and tail of the
confusing distribution of -o- and zero in the toga and iterative
structures. That may indeed be what is needed to persuade those who
do not even accept the very existence of the o-element this is all
about. However, I do not feel I can afford to make hasty
generalizations which have known facts against them.
It would indeed be strange if the distribution does not depend on
the root structure, so your suggestions reflect thought I have made
myself in an earlier period. They may be adequate for the limited
material accidentally placed at your disposal, but that is not the
whole story:
As I understand your generalization, a root long enough to contain
an extension (a "suffixe" in Benveniste's theory of the root) has -o-
the full-grade vowel is located in the first part (the minimal
root, Benveniste's "racine", i.e. before Benveniste's first hyphen),
and zero if it is located later. That works fine for the structure
TeRT-, as in Lat. tondeo:, Lith. bandà (roots analyzable as *ten-d-,
*bhen-dh-, which form *tond-éye-, *bhondh-áH2). However, it does not
work for *TReT-: Gk. tropé:, tropéo:, trophé:, strophé:, *prok^-éye-
in OCS prositi, *prok^-áH2 in Avest. p(&)rasa: (*prek^- 'ask').
Therefore, this part of the suggested subrule does not hold.
Jens