A question about ergative markers

[Wolfgang Schulze]

William McGregor wrote:
>
> Dear typologists,
>
> I am wondering if anyone knows of any language with more than one ergative
> marker?
>
I think the basic question is what you call an "ergative" marker, and
which kind of "ergativity" you refer to. For sake of simplicity, let us
assume that "ergative" refers to a nominal case marker that indicates an
agentive "cluster" in the overt or covert presence of a patient, that is
in strong transitive patterns. This cluster ("A" in traditional terms)
differs from the "S" cluster with intransitives that it (very often)
focuses on the "cause" domain in a "cause->effect" structure (vector),
whereas "S" reflects the "figure" in a "figure->ground" relationship
(which is also present with transitives relating patients ("Figure") and
verbs ("ground") [this is a very rudimentary assumption far from being
valid on a general typological level. But I think it can serve as a
working hypothesis reflecting some prototypical structures of encoding
"events" (or, some aspects of how the linguistically ritualized
construction of events CAN take place)].

The "agentive" cluster "A" itself very often reflects aspects of the
agentivity hierarchy which can be substantiated according to quite
different semantic or pragmatic "features" (I don't enter the discussion
of features and prototypes here, let's take "feature" as a very simple
and pre-theoretical descriptive term). Now, if you have different
ergative markers the first question is what these markers have in common
to justify a label "ergative". Obviously this seems be the parallel
distribution (that is A;S=O and not A=S;O)) [another justification would
be to claim that because there is only one ergative marker in the
„majority" of ergative languages, this should be the „basic" paradigm,
hence multiple ergatives have to be explained as „derivations" of this
paradigm; needless to say that such a claim is methodologically more
than problematic]. But what does this indicate? A categorial co-behavior
that allows us to subsume ALL adequate morphological structures under
the paradigm „ergative"? A common semantic „category" or simply a
technique of syntactic organization? And: Is the common distribution as
described above sufficient? Surely not. In many instances different
„ergative" markers behave different for instance with respect to
causativization (some may occur as embedded „subjects" (causee), others
may not etc.), or to (anti-)passives [naturally there are much more
(well-known) criteria to be mentioned (such as syntactic behavior and
word order, agreement, topicalization and focus procedures,
reflexivization, relativization etc.)].

If we simply refer to „ergative" as a more or less semantic marker in
the sense mentioned above, multiple ergatives seem to do nothing but
specify the polycentric architecture of the „ergative" cluster with
respect to either discourse knowledge or world knowledge. A quite common
parameter seems to be the above-mentioned agentivity hierarchy. The
formal result is what I call DAM „differentiated agent marking" (in
analogy to Bossong’s „DOM" „differentielle Objektmarkierung"), but I
have to confess that DAM is a very problematic term because it
presupposes that all markers that are present in a DAM paradigm are
„agents" (which surely is not true in every case). The prototypical
cluster „ergative" is characterized by a hierarchical downgrading of the
„kernel" itself marked by the highly abstract feature [agentive]
(together with adequate inferential features such as [instigator],
[volitional], [control], [responsible] etc., depending on the single
paradigms). The downgrading via DAM reduces the activation of these
denotational and/or inferential features (or of some of them) either out
of pragmatic reasons (that is to intentionally downgrade the „agentive"
role of  a participant) or because a given participant must not be
related to the semantics of the prototypical kernel of the ergative
cluster. Such selectional restrictions indicate some kind of noun
classification, the most common seems to be AGENTIVE/INTRUMENTAL.
However other instantiations also are possible (such as Speech Act
Participant (SAP) vs. non-Speech Act Participant (nSAP) etc.).  If DAM
is relevant for the whole (pro-)nominal paradigm, it is important to
state the „cut-off point" that may vary from language to language, e.g.
[±animate], [±human], [±male human] etc. In many cases DAM seems to
produce (dichotomic) Aristotelian categories, however, any aspect of
prototypical and/or radial categorization can be decisive. As far as I
know, binary classifications (that is binary DAM) are in the
overwhelming majority. Yet, fluid DAM structures with more than two
elements can also be observed (e.g. in Awar, an East Caucasian
language).

Stefan Georg asked my to list some data from East Caucasian (EC) to
exemplify the given DAM techniques. I’ll do that very briefly and
summarily only, because many things still have to be worked out more
clearly. Most of EC are languages with (covert) noun classification
which can be regarded as a signal that DAM could behave somewhat
analogically. In fact, most instances of DAM in EC are „semantic"
subcategorizing the all nominals in a diptotic (sometimes triptotic)
manner. What is remarkable is that this kind of subcategorization (if
present at all) does not always match the categorization yielded by noun
classification (NC). „Normally" there are (three to) four noun classes
(class I [+male,+human], class I [+female;+human], class III
[-human;+±animate], class IV [-human;±animate] (I won’t go into details
here). DAM, however, splits nouns according to [±male human], [±human],
or [±animate] (depending of the given language). Obviously, NC has a
semantic and syntactic foundation different from DAM in these languages
(multiple categorization is a common feature in EC). To sum up what we
can observe for DAM in EC is:

[±male human]	Ergative markers and „stem augment" (old ergatives)
		Very common in Awaro-Andian, probably also valid for Proto-East
Caucasian
[±human]	Ergative markers (e.g. Tsakhur...)
[±animate]	THE standard DAM via ERG/INSTR
		Chechen, Ingush, Tsezian....
[±SAP]		Nakh languages, some Lezgian languages and Proto-Lezgian
Multiple DAM	Stem augments as old „ergatives" in many (esp. Lezgian)
languages (the classificational effects 		still have to be studied).
Focal DAM	Probable in some Lezgian languages and Dargwa, but still a
matter of research.

This list is by no means complete (with respect to neither the DAM
semantics, nor to the languages mentioned). In case additional data are
wanted I would try to help. Some further notes on DAM in general and in
EC are given in W. Schulze 1998. Person, Klasse, Kongruenz. Fragmente
einer Kategorialtypologie des einfachen Satzes in den ostkaukasischen
Sprachen, Vol. 1: Die Grundlagen. München: Lincom.

Wolfgang


--
Prof. Dr. Wolfgang SCHULZE
Institut für Allgemeine und Indogermanische
Sprachwissenschaft * Universität München
Geschwister-Scholl-Platz 1 * D-80539 München
Tel.: +89-2180 2486
http://www.lrz-muenchen.de/~wschulze/