From: Jens Elmegaard Rasmussen
Message: 36540
Date: 2005-03-01
--- In cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:
[JER:]
> >What is the evidence for a 3pl *-s? The isolated Avestan pluperfect
> ><ciko:it&r&s^> ?
[MCV:]
> Yes, and Vedic -ur.
It is crucial to this argument (but not to the opposition) that -ur can
only continue a form *-r-s containing an *-s. Is that sure? Do we know
that forms like Ved. ú:dhar, áhar reflect *-r. and not a levelled
full-grade stem *-er (or *-or)? I do not exclude the possibility that *-r.
gave Vedic -ar, which is a fine and very intelligent suggestion, but still
only that. If it is true, then the Vedic ending -ur of the 3pl of the
perfect continues a form in *-r.-s. However that cannot be said about the
corresponding Avestan form which is -ar&. So, even with Vedic showing
constant *-r.-s and Avestan showing *-r.-s in the isolated pluperfect
case, we do not get solid evidence that the precursor of the Hittite
hi-conjugation contained an *-s. And even if it did, we do not know that
the hi-conjugation does not continue the perfect.
> >I find no s-marking in the 3pl of the hi-conjugation
>
> Of course there is: *-er-s > *-e:r (I don't need to
> duplicate the argumentation in Jasanoff p. 33).
I cannot see an *-s in this. Sure, it was an *-s that lengthened the vowel
of Hitt. hasterz 'star' from *H2sté:r (+ added -s), but I do not find room
for an *-s in either ker 'heart' or per 'house' which are neuters. There
is also the problem that, if *-e:r is from older *-er-s, one would expect
the unaccented variant to be *-o:r.
> > or in the Toch. prt. III.
> -ar/-är can reflect either -r. or -r.s
Yes, but the s-interpretation is a choice, not a fact. This is not evidence.
> >> >I do not think so. For one
> >> >thing, the two languages disagree fundamentally in the middle
> >voice
> >> >which has a pervasive -s- in Tocharian, but nothing of the sort
> >in
> >> >Anatolian.
> >>
> >> There is no -s- in Anatolian. Tocharian has non-sigmatic
> >> o-grade middles in the _intransitives_ that make an
> >> "s-preterit" (Toch A. nakät, tsakät, etc.). Tocharian B has
> >> sigmatized these, but with retained o-grade. The
> >> _transitives_ are sigmatic in Toch. A & B. Vedic has
> >> s-middles, with e-grade, everywhere. The original situation
> >> is thus preserved in Anatolian.
> >
> >I find this incoherent. It is crucial to your argument that the
> >Tocharian s-less middle forms continue the same category as those
> >having -s-, only in a more original shape. But is that the truth?
> >The verbs concerned all form thematic presents in Tocharian and
> >elsewhere (Ved. nas'a-, daha-, paca-; reflected in the Toch.
> >subjunctive, B nke-, tske-, pke-, identical with Vedic dáha:ti,
> >paca:ti).
>
> As Jasanoff notes, nketär etc. are _middle_ subjunctives, so
> they can't be identical with active thematic forms.
I was talking about the stem; the forms are identical with the middle
forms of the subjunctive appearing in Vedic. I see no good reason to
separate them from Brugmannian tradition.
> >Now, if that represents a displaced root-aorist
> >subjunctive, the original aorists of these verbs were root-aorists.
> >It seems to me to be the easiest solution that these are indeed
> >archaic remains, however not from a putative proto-hi-conjugation,
> >but from the original root aorist. Where the root aorist was not
> >retained it was replaced by the productive sigmatic aorist. Much of
> >that replacement belongs to PIE already. For the verbs concerned
> >Vedic reflects pervasive -s- in páks.at, adha:k, but Tocharian
> >apparently restricted the renewal to the active. Now the decisive
> >question is: Is the vocalism of A nakät, tsakät that of the old
> >middle-voice forms, or has it copied that of the sigmatic forms
> >which is the same? The partly sigmatic structure of preterite III is
> >easily derived from a blend of the perfect and the s-aorist because
> >IE *-o- and *-e:- merged in Tocharian.
>
> Except for palatalization. The preterite ñakäs, ñakär has
> *e:, the midle nakät, nakänt has *o. So much is clear.
There are many cases of analogical palatalization and depalatalization in
Tocharian. A ñak-/nak- does not have to reflect *ne:k^-/*nok^- (altugh
that could be a fine blend of the s-aorist with *-e:- and the perfect with
*-o-), it may also be a levelling of ñak-/*näk- from *ne:k^-/*n.k^- (with
analogical transposition of the new vowel). That would bring the forms
into line with the active and the middle of the s-aorist as in Vedic
áha:rs.am, ahr.s.ata .
> >> >All we need is to find a way for a 3sg ending of the s-
> >> >aorist
> >>
> >> There was no s-aorist.
> >
> >I find that view absurd - I find the s-aorist everywhere.
>
> Not in Anatolian, Tocharian, Germanic and Armenian.
I see their shadows there too.
> >> The s-aorist is derived from the Anatolian/Tocharian
> >> s-preterite by replacing the stative endings *-h2a, *-th2a
> >> by *-m, *-s from the active aorist (root aorist), and adding
> >> *-t to the 3rd. person.
> >
> >The earlier existence of a full s-aorist in a prestage of Tocharian
> >is shown by the present in *-se/o- which are identical with the
> >Vedic s-aorist subjunctives (váks.at);
>
> I don't think this was a normal subjunctive. Subjunctives
> don't normally make imperatives (-sesi > -si); in Tocharian,
> indicatives develop into subjunctives, but only the
> se-subjunctive developed into an indicative present.
> As a working hypothesis, I'm assuming these forms are
> related to the precative. More later.
According to Delbrück's account in Altindische Syntax 308-10 the
imperative-like use is quite a normal function of the second person of the
subjunctive. You are apparently unwilling to consider some of the
evidence: If Tocharian thematic presents (prs. II) quite regularly go with
preterite I, the continuation of the root-aorist, that to me looks like
satisfactory evidence that the subjunctive of the root-aorist could become
a present indicative.
The precative is in my interpretation based on the 2sg optative used as
stem ("pivotal point" Watkins-style) marking the note of begging.
> >these forms have retained
> >their subjunctive value in Vedic, but have been refunctioned into
> >presents in a telling repetition of the genesis of the thematic
> >present from root-aorist subjunctives.
>
> >>
> >> We then have:
> >>
> >> *-m
> >> *-s
> >> *-s-t
> >> ..
> >> *-rs
> >
> >I do not see that. What I see is
> >
> >*-s-m
> >*-s-s
> >*-s-t
> >*-s-me
> >*-s-te
> >*-s-nt
>
> 3pl. *-s-rs, actually. I explained below where these form
> come from.
I still do not see that. And below I do not accept your explanation.
> >The 1.2.3.sg have lengthened grade, and the form of the 3pl (Gk. -
> >san) reflects acrostatic ablaut; so does a middle form like Ved. 1sg
> >ámam.si, Avest. m&:n.ghi: 'I thought' from *mén-s-&2. I find it very
> >hard to account for the last-mentioned form if the middle voice of
> >the s-aorist is supposed not to have existed at the time of the
> >phonetic operation of the ablaut.
>
> The synchronic 1sg. middle ending is -i. I don't know where
> it comes from.
It fits the full-grade form -a like a glove: accented *-H2é > *-H2á,
unaccented *-H2 > *-&2. The full-grade form is seen in the optative -i:yá
reflecting *-iH1-H2á (via *-iyá and analogical langthening of the -i- on
the pattern of the other forms of the athematic optative which have
regular -i:- from *-iH1- before consonants). The form is Avestan also:
tanuiia, diia:. This is just standard stuff.
>
> >> which becomes by a simple analogy:
> >>
> >> *-s-m
> >> *-s-(s)
> >> *-s-t
> >> ..
> >> *-s-rs
> >>
> >> This scenario was played out once, in the core IE languages
> >> that have an s-aorist.
> >
> >This is superfluous if the s-aorist was already there in the
> >protolanguage from which they all come.
>
> It's clear to me that it wasn't.
The clarity with which you see that things I see are not there has yet to
be imparted on me by the use of solid arguments. Until such time I see no
reason to change my opinion.
> The process that produced the s-aorist out of an older
> system where only the 3rd. person had *-s is not only well
> attested in Hittite and Tocharian, but is parallelled
> exactly by the development of the Old Irish t-preterite,
> from 3sg. *-t.
No such process is "attested" in Hittite or Tocharian: it has merely been
suggested, which does not amount to attestation. It is only "parallelled"
by the Old Irish developments if it has a similar background; it may as
well not have, in which case it is not parallelled by it. It is often
forgotten in the euphory that most things are not parallel.
> I'm not aware of anything parallel to a scenario where a
> verbal (aspect) marker present in all personal forms is lost
> everywhere except in the 3sg. Let alone for this to have
> happened twice independently.
That is not a completely fair description of the situation as it would be
seen under the traditional doctrine. The aspect marker is then not lost
everywhere in Tocharian except in the 3sg; it is retained all through the
middle voice, and incidentally no one can see it is not present in the 2sg
also.
If you were willing to separate the two events from each other you could
use any of them as a parallel for the other one, showing that such a
reduced retention is indeed possible. If you cannot find a super-parallel
showing that precisely what is common to the two events has also
co-occurred elsewhere, does it really matter? Can't we just accept it
here? May the way one language reduces its verbal system not have
similarities with the way another one does the same and still be
independent? In my eyes this looks like a pseudoproblem which has been
formulated to justify a novel explanation of a set of facts that is
already perfectly well explained by existing doctrines.
In Modern Greek the imperfect and aorist use the endings -a, -es, -e,
-ame, -ate, -an. In this the 2/3 sg are thematic, while the rest have the
-a- of the aorist. In Old Church Slavic the aorist has thematic endings in
the 2/3 sg, as tec^e, tec^e 'ran, flowed', while the other persons are
sigmatic, either 1sg têxU, pl. têxomU, têste, tês^eN, or 1sg tekoxU, pl
tekoxomU, tekoste, tekos^eN. Here one could say: Where else do we find
retention of an unmarked stem only in the 2/3 sg, while the rest of the
paradigm has acquired a marking? And how on earth could that happen twice?
There must be a connection, and let's begin our rewriting of the
morphology of the IE verb by accepting this Mischparadigma as the original
situation. It seems to me that this is what is being done to IE now.
Jens