[tied] Re: Kabardian antipassives

--- In cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:

> I suppose we can agree that the *-e of the perfect endings,
> at least in the singular, acts in every respect as a
> secondary addition. It does not attract the accent, it does
> not affect the o-grade of the verbal root.

That is very accurate.

> Logically, the
> perfect/middle endings at some point in pre-PIE, before the
> addition of *-e, was therefore:
>
> 1. *-h2
> 2. *-th2
> 3. *-0

Yes, that's what it looks like. But these endings are never seen in
isolation. While the perfect has the added non-ablauting *-e which
does not influence the accent, the middle voice has an ablauting *-e
which does influence the accent in the way a syllabic ending does. I
have no clear idea why this is so, except that I have had some of
the same thoughts as you but without being really convinced.

> It's possible that the Skt. 1sg. middle past ending -i
> directly reflects this, as well as the Hittite 1sg.
> hi-conjugation preterite -hun (< *-h2-m., not *-h2a-m).

Possible, yes, but not very likely. Even for Sanskrit it can be
shown that the full-grade ending is -a, while -i is a zero-grade
alternant (1sg opt. root-aor. as'i:ya : 1sg mid. s-aor. á-mam.s-i),
and for -hun we have a better explanation as a replacement of -ha as
preserved in Luvian by adjustment to the corresponding ending -un of
the mi-conjugation.

>
> If I'm correct that the 1/2 sg. endings derive from a
> Nostratic stative in *-k, *-tk, the development *-k > *-h2
> also requires that the velar stop once stood in the absolute
> Auslaut.

I would suppose this is true, although that involves a much longer
and much less secure perspective than the rest.

> As to the nature of the *-e, my best guess is that it is
> indeed the thematic vowel. Assuming the same element was
> added both to the perfect and the middle, we see that the
> middle forms have *o before *-r, *-m, *-i (*-t-o-r,
> *-dhw-o-m, *-nt-o-i), which is consistent with an
> interpretation as the thematic vowel. Two problems remain:
> the fact that the vowel appears as *a after *h2 (e.g.
> *-h2ai), and the fact that in some middle forms, the vowel
> appears as *-o when nothing follows. I would suggest that
> the first phenomenon is regular: a thematic vowel coloured
> by *h2 does not yield *o before a voiced segment (we see the
> same in archaic forms of the word "woman", such as Greek
> gunaik- and Armenian kanay- < *gWn.h2-a-ih2-). The second
> phenomenon (-o for expected -e in the middle), I would
> explain analogically: the (past) middles in -o were
> secondarily created by deleting the middle marker *-i, when
> that had been re-interpreted as a present marker.

I fail to see the problem, or even the topic: The prefect endings
are never followed by anything voiced (or anything else), so the
thematic vowel is expected to behave just like the phoneme /e/ which
produces /a/ in connection with /H2/. I do not see the necessity of
the structure you posit for 'woman', particularly I do not see a
thematic vowel in it by any serious standard I can think of.

Phonologically the added vowel of the perfect could well be the
thematic vowel, but that of the middle voice cannot. And none of
them is too well in keeping with anything else we know about the
thematic vowel. Basically, the vowels here concerned do not form
stems, so the word "thematic" must have a different meaning, and
then the whole point is hard to see. We do not otherwise
see "thematic" vowels dangling after ready-made wordforms, so there
is really nothing in the grammar that helps us if we identify the
element with one known from the grammar.

>
> If the *-e is the thematic vowel, it needs to be explained
> why in the perfect/middle system it appears _after_ the
> desinences, instead of _before_ them, as in the active
> system. The only solution I can think of requires that the
> thematic vowel in these forms indeed be an object marker
> (somehow connected to the anaphoric pronoun *i/*e-),
> something which is otherwise impossible to prove. It's as
> if in the active system, incorporation proceeded according
> to the model V-O-S, while in the "stative" (perfect/middle)
> system, incorporation followed the model V-S-O. Given the
> fact that the active and the stative systems are
> fundamentally different (e.g. in their desinences), I can
> see no problem with the assumption that they were formed
> according to very different models of agglutination
> (enclitic syntax).

I completely fail to se what an object marker would be doing here. I
also fail to see the principle whereby the thematic vowel could be
used to mark the object. Otherwise it expresses belonging. That
incidentally makes some sense: If the bare endings were passive
("was killed"), the extension by a morpheme of belonging would be
possessive ("has killed"), which comes quite close. Still, I do not
know why the morpheme is placed where it is, and why there is an
ablauting vowel in the mediopassive.

> On the other hand, the thematic vowel marking the
> subjunctive, being a modification of the verbal stem itself
> (turning it into a [thematic] adjective?), must always be
> closely linked to the verbal stem, regardless of whether the
> form belongs to the active or the stative system. In this
> analysis, we would then expect a "stative" conjunctive to
> show the desinences:
>
> 1. *-a-h2(a)
> 2. *-e-th2(a)
> 3. *-e-(e).
>
> It is trivial to derive a form like the Latin future (<
> conjunctive) in -a:m, -e:s, -e:t from such a "stative
> conjunctive".

We have other explanations of the Latin endings that do not need all
these speculations. I also completely fail to see why the
subjunctive would derive from a mediopassive form.

Jens