From: elmeras2000
Message: 32026
Date: 2004-04-19
--- In cybalist@yahoogroups.com, "P&G" <petegray@...> wrote:
[JER:]
> >> >Brugmann/Pokorny bracket the r of *p&té:(r) and *k^uo:(n)
[P&G:]
> >> That seems to me a classic case of the prioritising of Sanskrit
[JER:]
> >What prioritising? Bracketing its reflex, or mentioning it at all?
> Suggesting that the form without r or n could be original. A
reconstruction
> is not the place to show the vagaries of a particular dialect.
I am sure they are both trying to show the vagaries of the
protolanguage in the form addressed and do not want to make
unsubstantiated choices. Meillet goes even further stating only that
different branches treat the nom. differently (the relevant section
of the Introduction).
> If PIE
> *p&ter is reconstructed with -r, it must not be shown as p&te:
That would
> be wrong reconstruction - which is what Brugmann and Pokorny both
offer
> here, on the basis of a prioritising of Sanksrit forms.
How can you know they are not prioritising Lithuanian in the case of
r-stems, and the whole series Lithuanian-Germanic-Latin-Celtic in
the case of n-stems?
If there are several IE branches actually reflecting, say, *dhug&2-
té:(:) without /-r/, and several others reflecting *-té:r in the
same word, there is a problem for reconstruction. If the branches
not showing the /-r/ have a phonetic rule deleting it, that's it,
but they do not. And worse, the distribution of the branches is not
the same for r- and n-stems (Latin and Celtic reflect -r, but not -
n). If you do not want to be guilty of making unsubstantiated
choices you must keep the possibility open that both forms belonged
to the protolanguage and were both inherited from it. That raises
the question about their distribution in the protolanguage. The
answer could be lexical, some indiosyncratic r- and n-stems having
retained their sonants while other had lost them, or it could be a
principled alternation pertaining to all r- and n-stems in the
protolanguage and later given up. Brugmann very cautiously reports
the opinion (started by Johannes Schmidt) that sandhi variants are
ultimately at the basis of this, but since this cannot be proved one
way or the other he says no more about it, except remembering the
problem every time he specifies a protoform.
> >> Sandhi variation in IE is at least plausible, even if I don't
> >>believe it,
> >Well, how interesting - any arguments pertinent to the matter?
>
> How sweet that you find my opinion interesting. The arguments
for it are
> currently being thrashed around on this list. I see no reason to
assume
> anything for PIE other than *p&te:r, k'uo:n, but I continue to
follow the
> debate.
How would Lith. dukte:~ and Skt. duhitá: proceed from a form in *-
té:r ? And how could Skt. rá:ja:, Lat. homo:, Celtic *-u: (lost in
OIr. and so not having a final consonant) and Germanic loans in
Finnish with -o (not -on) come from a form in *-o:n ? These are very
good and solid reasons to assume that *-e:r, *-o:n is not the whole
story.
>
> >> the existence of the -r in the nominative in
> >> IE, at least at a morphophonemic level.
> >that is what the reconstructions you are criticising mean.
>
> No they don't. They mean that we must reconstruct forms without
the final r
> or n.
Well, only as alternants. And that makes the matter one of
morphophonemics. Brugmann explains his practice very explicitly and
I have reported his statements. And Meillet is equally undecided.
Sihler makes a´the strange choice *-e:r/*-o:r with r-stems, but *-
e:/*-o: with n-stems, however without giving enlightening reasons. I
fail to see any reason why some branches would extend the lack of
stem-consonant in the nominative from n-stems to r-stems only
leaving all other stems as they were. It is hard to see any better
rationale than original sandhi.
> And that is what I am at the moment rejecting, until you or seomone
> else proves that these final-less forms actually are more likely.
They appear to be the only viable protoforms for the later forms
that lack them.
> > independent and objective reasoning over the facts
> >which were also at the disposal [of Brugmann and Pokorny] lead us
to draw
> the same
> >conclusion today as they did then.
>
> No it doesn't. We do not reconstruct p&te: as the nominative,
nor k'uo:.
Maybe you don't, but you should - as variants alongside forms with
retained sonants. And we are quite many who do, but there is also a
number of scholars that do things wrongly.
For the present purposes the sonantless alternants, if such they
were, may be left out of consideration, for if we can explain the
fuller forms the reduced variants should follow by further
development.
Jens