Re: [tied] Bader's article on *-os(y)o

On Tue, 8 Jun 2004, Rob wrote:

> The muddledness of plurality in IE seems to me evidence that
> plurality was a recent phenomenon in IE history.

My logic would tend to say the opposite. The inherited plural formations
of German, Danish or Welsh are totally unpredictable now. Each word must
be learnt by itself.

> Is there such a thing as "too far back in time"?

Not in the world of my dreams, no. But I'm afraid there very much is such
a thing as too far back for us to be able to understand it at the present
time. We move one step at a time.

[On nom.sg.:9

> True. Assuming that the length can be reconstructed for all of IE,
> what caused it? And why only in the nominative singular? If I'm
> correct, you postulate that it was due to lengthening in the vicinity
> of nominative **-z. So, e.g. *we:kWs must have developed something
> like this: wekW-z > wegWz > we:gWz > we:gWs > we:kWs? Just for
> curiosity's sake, are there any other possible explanations for the
> lengthened-grade nominative forms?

In vo:x, IE *wó:kW-s, I believe the root vowel was long already in the
underlying form. This is one of the canonical root-noun types with a
specifically iterative shade in tis semantics. I credit the length to the
meaning; it's iconic lengthening, already in the input. Szemerényi gives a
rundown of older theories accounting for the nominative lengthening, but
it is all very silly. I can only observe that it looks as if the
nominative marker imparted length on its nearest preceding vowel (but not
when it was contiguous with it). It it were only the nom. *-s (**-z?) one
could take it to be a dose of sonority, but the same effect is shown by
the collective marker *-H2, which was [x]. So it may just be the natural
length of a spirant that is redistributed in the word.

> > I meant diphthong stems of the kind of Vedic sákha:, -a:y-am, -y-e;
> > Avest. kauua:, Greek peithó:. With syllabic semivowel of course we
> > have *-i-s, *-u-s. This is apparently posterior to the ablaut and
> in
> > my view demands a special sibilant which I posit as a voiced *-z.
> We
> > have had a fierce debate over that which I don't want to reopen.
>
> Yes, the debate is still fierce, I'm afraid. I don't want to reopen
> it here; we can pursue it (civilized-ly, I can assure you) via e-mail
> if you'd like.
>
> How common are the "sákha:-stems"? I was honestly not even aware of
> their existence, but then again my knowledge of IE is somewhat
> limited (although always enlargening!).

I was over this with Miguel a while back. There are quite a few examples
in Hittite, and the type may have become productive in Tocharian. Greek
has a handful of items; /e:khó:/ (PGk. *wa:khó:) 'echo' must be the best
known word of that type.

>
> > We are very close to real agreement on that point. I published a
> > comparable rule many years ago for the second person and other
> cases
> > of IE s/t alternation. Only the plural marker and the 2nd person
> > morpheme do not appear to be identical, cf. Eskimo-Aleut pl. *-d,
> > 2sg *-t (surfacing as *-t, *-n, respectively).
>
> Well, let's take a look at this. Two possibilities show themselves
> immediately:
>
> 1. The Eskimo-Aleut plural marker *-d is not related to IE *-es, or
> 2. The final **-t of the IE plural marker and that of the 2sg marker
> were once different (perhaps one was aspirated and the other plain,
> etc).

I'm all for 2. But who am I to tell? The 3rd person marker, by the way,
would seem to have been /nt/ (or a monophonemic consonant containing these
features). That would also avoid the problem of a clash between *-t in 2sg
and 3sg. But this is a can of worms again, good grief.

> > The shortness of ákmones vs. the length of ákmo:n is indeed
> > explained:
> >
> > Nom.sg. *H2ék^-men-z > *H2ék^-mon-z > *H2ék^-mo:nz > *H2ék^mo:n.
> The
> > reduction of unaccented *short* vowels occurred after the
> > lenghtneing caused by the nominative marker, so the lengthened
> vowel
> > was retained.
>
> The form *(x)ákmons seems to be possible only after the earlier
> penultimate stress rule (as Glen and I hypothesize) disappeared.

Yes, that rule was designed to predict the *lexical* accent of stems,
wasn't it? That would fit, for this is younger than the lexical assignment
of the accent position.

> That is, *(x)ákmons was a more recent phenomenon, perhaps. There is
> a word *(x)ákus or *(x)akús meaning 'sharp,' right?

Sure, "*(x)ákmons" is more recent than "*(x)ákmens" of which it will be
the direct continuation.

> > Nom.pl. *H2ék^-men-z-c > *H2ék^-mon-z-c > *H2ék^-mo:n-z-c [no
> change
> > at the time of loss of unaccented *short* vowels] > *H2ék^-mon-z-c
> > (with shortening as in *nó:kWt-z > *nókWts, or ptc. *-o:nt-s > *-
> ont-
> > s) > *H2ék^-mon-ezc (with anaptyxis posited purely ad hoc) > PIE
> > *H2ák^-mon-es. It works of course also with original *-c-z.
>
> Hmm. I'm hardly a professional linguist, but is this "ad hoc"
> anaptyxis realistic?

Why wouldn't it be? If one of the s's was reduced to a simple aspiration,
that would already be a voiceless e. It's very close in fact.

>
> > What is not explained, though, is the acc.sg. *H2ák^-mon-m. which
> > should have been *-m.n-m. ; I explain it by analogy: eend-stressed
> > type *-mé:n, acc. *-mén-m. : recessive-stress type *-mo:n, acc. X;
> X
> > = *-mon-m. .
>
> We do agree (as we've said earlier) in the distinction between
> stressed *-mén- vs. unstressed *-mon-. However, what caused this
> distinction? Differences in root characters? Another thing: the *-
> mon- suffix is apparently (according to Sihler) much more common than
> *-mén-.
>
> > How would I know?
>
> Just curious, not so much as to whether you know, but to what you
> think might be possible explanations.

I do not know. I see no complementary distribution in formal terms. But
it's a bit funny there are *no inanimate oxytones*. There are no neuters
with a stressed suffix *-és or *-mén. So perhaps it is an original
opposition of animacy?? In that case its reflexes are perhaps not in a
very clean distribution in PIE anymore.

Jens