I'd say that internal reconstruction is a
legitimate way of speculating about the origin of synchronic alternations, also
those reconstructed on a comparative basis. It isn't part of the comparative
method; to put it crudely, it begins where comparison has exhausted its
means. There is no way to constrain internal reconstruction within PIE --
you can exercise as much intellectual freedom there as you wish -- but good
(i.e. typologically plausible) alternative solutions to most problems can easily
be offered and cannot be ruled out. This is why the results must
remain speculative and will never be trusted to the degree that comparative
reconstructions are trusted.
These "extra explanations" have been
provided by many authors. The problem has always been of interest to Slavicists.
Would you also say that Slavic *-U (as in the Acc.sg.m.) and *-o (as in the
Nom./Acc.sg.n.) cannot both derive from IE *-om? I'd put it more cautiously:
they can't both derive from the same form via regular phonological
change. But if there is morphological conditioning involved as well, a
secondary contrast of this type may develop rather easily. At any event, it's an
inner Slavic problem.
Piotr
----- Original Message -----
Sent: Wednesday, November 22, 2000 8:57 PM
Subject: Re: [tied] IE *-su and the Nostratic "equational" marker
*-n :)
For the rest I think internal reconstruction is a legitimate part of
the comparative method. It's obviously more speculative than comparison
tout sec, which would leave the reconstruction of, say, the loc. pl. undecided
as *-su ~ *-si.
...
(same goes for Slavic *-U and *-o, both supposedly from *-os : that
just can't be right, not without a whole lot of extra explanations, which I
haven't seen).