--- In
cybalist@yahoogroups.com, "Glen Gordon" <glengordon01@...>
wrote:
>
> Rob:
> >Given the evidence, one should at least accept that PIE and
> >Proto-Uralic inherited a case form -m from their parent language.
>
> Of course, we should. Since both language groups insist on using
> *-m as an accusative marker... then that means that it's an
> accusative marker until you prove otherwise. Efficiency of thought
> wins, you lose.
Forget about winning and losing. I actually agree with both of you.
IE and Uralic certainly agree on a shared marker *-m of the
topicalized animate object. They apparently also agree on the basic
structure of the verb, in that in the sg. personal markers are added
to the stem, while in the dual and plural, dual and plural morphemes
are added to both stem and person markers. I know this has not
entered the mainstream, but it is quite a widespread analysis, and
one which I believe can be refined to the point of approaching
actual proof. I take the traditional Uralic reconstruction *luke-k-
me-k or *luke-t-me-k 'we read' (Finn. luemme) to represent a stem
put in the plural and followed by the first person marker which is
also put in the plural (or rather dual); the 2pl is *luke-k-te-k or
*luke-t-te-k using the 2nd person marker. Corresponding IE forms
such as 1du *gWhn-wé and 1pl *gWhn-mé, 2pl *gWhn-té I derive from
earlier *gWhen-g-me-g, *gWhen-d-me-g, *gWhen-d-te-g (where -g- and -
d- are meant to be spirants) in which -g- is the dual marker, and -d-
the plural marker. The phonetic reductions needed to clinch the
argument can be parallelled. Now, if a personal marker following a
verbal stem makes it mean 'I am killing' (like *gWen-m, the
injunctive part of Vedic hánam*, ipf. áhanam), then the stem
supposedly means something like 'killing', and lo and behold, that's
what we find in root nouns such as Ved. vrtra-hán- 'Vrtra-killer'.
Then, in the dual and plural, we would expect not only the person to
be dualized or pluralized, but also the stem to be expended by a
number marking, for 'we are killing' should of course properly be
expressed as "we (pl.) are killers (pl.)". Now, I guess the funny -e
of IE *-me, *-te and the equally funny Uralic *-k are to be credited
to markings of the dual, which must mean that a morpheme which must
originally have belonged to the dual has been generalized so as to
sit also on the final part of the plural endings. That act of
paradigmatic levelling will have to be ascribed to IE and Uralic and
can only reasonably be assumed to reflect a change that took place
in a prestage when they were only one language. That must reflect a
closer relationship between IE and Ur. than between IE and any other
branch of Nostratic for which we have this information (I must
remain agnostic about Etruscan & Co.). Certainly Eskimo-Aleut did
not take part in this, for here we find relatively transparent
reflexions of *-m-g, *-t-g, *-m-d, *-t-d, implementing the same
morphemes without any confusion.
As for the morpheme *-m I may now look to Esk.-Al. to see if I can
combine it with anything, and the case ending *-m of the
ergative/genitive immediately comes to mind. That case morpheme is
so deeply entranched in the morphology of Eskimo that it even
precedes possessive enclitics, and also precedes the markers of the
secondary cases which are derived from the ergative/genitive. It is
therefore already in its own right eligible for outside comparison
and a morphological element of the highest antiquity. Now, can that
be identified with the accusative *-m of IE and Uralic? I think it
can. The question is of course how.
It is all in the function of the stem fo the verb. If the IE-Ur.
verbal stems are properly agent nouns, which are of course nouns.
And the object of a verbal noun may be expected to be in the
genitive, so that is what I think the function of the *-m properly
was. There are no functions of the Esk.-Al. "ergative" that cannot
be easily derived from transparent syntactic structures in which it
makes excellent sense as a genitive. In fact the interpretation of
the "ergative" as a genitive completely strips it of the mysticism
generally attached with that case function. It fits even better than
one would perhaps expect, for the ergative is only used in
constructions where the agent/owner and the patient/possessum are
known, the note of topicality seen in the Uralic accusative is
repeated here. I know too little about Kartvelian, but the mere
existence of an /m/ in the ergative/genitive of that branch would
appear to support the antiquity of this function of the morpheme.
I would thus understand an IE sentence such as *H2né:r k^wónm
gWhent 'the man kills/killed the dog' so as to properly mean "the
man (is) the dog's killer".
That most inanimates do not take *-m in IE may be due to their being
too low on a scale of animacy to be topicalized.
I would then suggest that both the extension of the dual marking to
the final part of the 1./2. plural and the reinterpretation of the
old m-genitive to an animate object marker are shared innovations of
Indo-European and Uralic.
Jens