Re: [tied] Acc. plural of o- and eh2-stems (was: The disappearance

On Tue, 13 Apr 2004, Mate Kapovic wrote:

> > > I don't see why some linguists reconstruct *-o:ns in the A. pl. of
> > o-stems.

[JER:]

> > We do so because we find a long vowel in Sanskrit. Most of the other
> > languages cannot show the difference. Still, I deem it reasonable to
> > derive Lith. -úos- from *-o:ns also, since it is acute.

[MK:]

> Yeah, but Lithuanian would point to *-oh2ns with the inserted laryngeal.

You must be referring to the Leiden doctrine that acute vowels always
reflect laryngeals or Winter's lengthening. I do not accept that. In my
opinion the regular tone on lengthened-grade vowels in disyllables is
acute. The many alleged examples of circumflex have in my opinion all
arisen in monosyllabic forms. But it is a matter of heavy dispute between
Kortlandt and myself.

> It
> is not certain at all that plain long *o: would give an acute in Lith.
> Also, isn't Skr. -a:n < *-ons the same as -i:n < *-ins or -u:n < *uns?

No, Sanskrit masc. -a:n (o-stems) vs. fem. -a:s (a:-stems) was copied as
-i:n/-i:s and -u:n/u:s and even -r.:n/-r.:s which introduced a secondary
gender opposition in i-, u- and r-stems.

>
> > We can if *-o:ms developed differently form original *-aH2ms. The
> > combined evidence of Gothic -o:s and IIr. -a:s points to a PIE form
> > of the acc.pl. of a:-stems without the nasal. It is now commonly
> > assumed to have been lost in the prehistory of IE already, an idea
> > proposed by Stang and now often referred to as his law.
>
> Wouldn't Stang's law be disappearance of the laryngeals, *w and *y before
> final nasal? That's why there's no acute in BSl from PIE A. sg. *-eh2m?

Some say that. I can't know if they are right. There is acute in the Greek
form, so it also causes a problem. I have supposed the form to have sandhi
variants, monosyllabic (acute) *-aH2m before vowels, but disyllabic
(circumflex) *-aH2m. before consonants and zero.

> We can assume that *-n- was dropped already in PIE (and then later
> reappeared analogically) but that doesn't directly influence the question
> of
> A. pl. of o-stems. Also, the *-n- could have been dropped later
> independantly in Germanic and IIr and kept (regularly) in BSl, Lat and
> Greek.

In that case you are not making much comparative linguistics out of the
potential of the languages. Rather you are avoiding it at all costs.

> >The
> > confusion between nom.pl., acc.pl. and even gen.sg. in a:-stems in
> > Balto-Slavic also points in the direction of an acc.pl. without a
> > nasal,
>
> Slavic clearly points to a nasal (palatal ending -e). We need not suppose
> that in A. pl. also *-a:s (as in G. sg. and N. pl) and not *-a:ns. If we
> had
> G. *-a:s, n. *-a:s and a. (regularly) *-a:ns, after *-a:s > *-a: (= N.
> sg)
> that could also result with N. pl. taking the A. pl. ending and then G.
> sg.
> taking N. pl. ending (because they were the same before).

Sure, the nom.pl. could have taken the form of the acc.pl., and the gen.
could have followed in its footsteps. That actually was my own suggestion.
I shocked the audience of the Idg.Gs. at the congress in Berlin in 1983 by
saying that in a discussion, and the speaker promised to include it in the
published version of his paper, but never did. In recent years I have seen
it appearing all over the place, quite possibly mediated by the late
Jochem Schindler. Meanwhile I have myself moved on a little. I now would
also like to integrate the Latvian -as which is used for all three forms.
That is a non-nasalized form. I don't see how a nasal-less acc.pl. can
have been introduced by analogy, so acc.pl. *-a:s will seem to be
inherited. This apparently leaves only the scenario I have chosen.

>
> > We have a few instances of */aHNC/ in IIr. which come out as /aaC/,
> > e.g. váata- 'wind', Av. må /maah/ 'month' (from *meH1-n.s). That
> > does not point to merger of *-eH2ns and *-o:ns in that branch.
>
> OK. That I didn't know. Thanks.
> I am just looking at all possibilities, I am not saying that o-stem A.
> pl.
> *-o:ns is not also possible. It just depends what approach will you take,
> there is more than one way of handling this.

You are so right.

> In BSl maybe we have *-aHs : *-o(:)ns > *-aHns : *-oHns > Lith. -a`s :
> -u`s
> and Slavic *-aHns > -y, -e. The a-stems would take the *-n- from o-stems
> and
> the o-stems would take the *-H- from a-stems. If we assume *-aHns :
> *-o(:)ns
> it would just take analogical spread of *-H- to o-stems.

Interesting thought.

Jens