Ergativity correlations

[Jon Aske]

Since several people asked that we continue this chat in public rather than
in private, I am back with some questions I have about Clancy's latest
message to the list.

On Saturday, February 07, 1998 12:32 PM J. Clancy Clements wrote:
> Here is my reasoning.  First, I'm assuming that frequency determines the
> default position in Spanish, then secondarily the distribution of
different
> types of NPs, i.e. definite, indefinite and mass/bare plural NPs.  As you

I guess where we differ is not in the facts, but in the theoretical
interpretation.  I object to the use of a "formal" category or label such as
"basic word order" or "default position" for grammatical categories--for all
languages in general and for Spanish in particular.  I think that this is an
English-centric view, English being a language which displays a strong
statistical correlation between its core grammatical categories (subject and
object(s)) and position with respect to the verb.  (Still, even for English
the correlations are only statistical and don't apply much beyond subject
and object.)

I guess I think it's a mistake to say something is the "default position"
and then the deviations from the norm are explained by extraneous factors,
such as "discourse factors".  To me this reflects a "transformational
mentality" in linguistics (which precedes the transformational grammar
tradition) which I don't think belongs in very many places in a cognitive
theory of language.  At least not in a functionalist one.  I think that
labels such as "basic word order" may be used informally by linguists
involved in typology and description, but cannot be part of a theory of
language (a theory of the cognitive representation of language).

> pointed out, in Spanish intransitive clauses, one finds comparable
> distribution between pre- and post-verbal subjects.  So, frequency as a
> criterion doesn't help.  Next, if we look at the distribution of NP types
> in pre- and post-verbal position, we find that all types mentioned above
> appear postverbally, but not all types appear preverbally. This, as you
> point out, is discourse related. It suggests, nevertheless, that the
> default position for subjects in Spanish intrans. clauses is postverbal.

But if we know the contexts in which subjects are postverbal vs. postverbal,
then why don't we just set that as the "principle" or "strategy" or "rule"
or what have you?  (A in context B and C in context D).  What do we gain by
saying that subjects with certain pragmatic properties deviate from the
"default position" for subjects?  I don't think this is a matter of personal
taste.  This problem is well known in phonology: how do we deal with
variations from a norm?  Sometimes it seems to makes sense to posit an
"underlying" or basic form and other times this is much more questionable
(eg sandhi vs. word internal alternations).  To me "basic order" in syntax
is a questionable construct.

> In transitive clauses, again because of discourse reasons, the subject
> (which is mostly the topic) is overwhelmingly preverbal.  Here frequency
> tells us that for Spanish transitive clauses, preverbal position is the
> default position for subjects.

I think that the only reason that transitive subjects are more likely
(statistically speaking) to be topics than intransitive ones is that whereas
in intransitive assertions subjects are often foci (and thus postverbal), in
transitive clauses an overt (full nominal) object is more likely to be the
focus, for a variety of pragmatic and semantic reasons, which means that the
subject has no "choice" but to be the topic (and thus typically
"preverbal").  As soon as you add an optional complement to an asserted
intransitive clause (Juan vino ayer 'Juan came yesterday'), you find that
this added complement is much more likely to be the focus (than the subject
is), which means that the subject is much more likely to be the topic.  Thus
the likelihood of a subject being preverbal vs. postverbal, ie of being the
topic vs. the focus, is only *correlated* with whether the verb is
transitive or not, but not at all caused or explained by it.  (Note,
however, that not all inverted subjects are foci; some subjects are inverted
for other reasons, which I won't get into).

> So, you find a pattern, apparent in the default position of the subject in
> transitive and intransitive clauses, that is suggestive of an ergative
> marking pattern in that trans. clause objects and intrans. clause subjects

I guess I object to using the label "ergative" as an abstract, "platonic"
property of languages.  I object to using it as if it was "explanatory" in
nature, when it is nothing but a name, or label, for a grammatical
phenomenon: the coding of (some) "intransitive subjects" the same way as
direct objects.

> The claim is that Spanish
> exhibits an ergative marking pattern, even though this pattern is
> accounted for by discourse-related arguments.

As I said, to me ergativity is a (typically morphological) *coding system*,
a system which has its own logic and its own very interesting diachronic
sources (but not an underlying pattern of linguistic organization).   Now,
word order in Spanish does not code grammatical relations at all but rather
functional (informational) relations.  It is true that there are
*statistical* affinities between S and O in Spanish, and in all languages,
but these affinities have explanations which are quite different from those
which explain analogous grammatical coding (ergativity).

> Postverbal subject word order is not grammaticalized in Spanish. Were it
> to become grammaticalized, then one could speak of ergativity. The
> reanalysis of passive marking in some languages leads to an ergative
> marking.  There are, though, transitional stages in the process of the
> reanalysis.  Spanish could be in a similar process with respect to the
> position of its subject.

I don't think that postverbal order of intransitive subjects could ever
become grammaticalized in Spanish, or in any other language, simply because
of the fact that the phenomenon does not have a grammatical basis, but
rather an informational one.  Ergative coding always comes about indirectly,
by the reanalysis of constructions, given the fact the semantic and
informational correlations between S and O are rather weak.

> PO marking marks monotrans. DOs and ditransitive IOs identically, just as
> intrans. subjects and trans. DOs are marked identically.

This applies to human, or human-like, direct objects only, right?  (*Visito
a Paris* 'I visit Paris' I think is out of the question, as far as I know;
but "Visito a mi perro" 'I visit my dog' is OK).  I think that it's possible
that this "syncretism" is motivated by the "need" to disambiguate objects
from subjects when both have similar semantic characteristics (being human,
as most transitive subjects are), and that it was facilitated by the
semantic similarities between direct and indirect objects (both are
typically human and in some way "recipient" or "benefactives" for the
action).  But I still don't understand what this has to do with ergativity.
I am also highly skeptical of this category "primary object".

About the "crucial test" sentence, *Juan presenta Luisa a Marta*, I admit
that this is a very interesting sentence.  Still, the sentence seems to me
more than a bit odd, and I think that to the extent that it is accepted by
speakers, it probably has a very good explanation without having to resort
to a primary object category.  Leismo (and the personal a) makes sense
because human direct objects are typically affected by actions in ways more
like the way typical indirect objects (recipients, benefactives,
malefactives) are than the way non-human (non-animate? non-empathetic?)
objects are.  To the extent that there can only be one such element, it
makes sense that it would be the dative in this sentence (Marta), which
would block the (diachronic or synchronic) extension of personal a marking
to human objects like Luisa in this example.  This would also explain why
this sentence seems odd and is probably avoided by many speakers (except
perhaps those for whom the construction is fully conventionalized).

Anyway, thanks for your comments, and I encourage others to join in in this
exchange.

Best, Jon

____________________________
Jon Aske
aske at earthlink.net
or Jon.Aske at salem.mass.edu
Department of Foreign Languages
Salem State College
Salem, Massachusetts 01970
http://home.earthlink.net/~aske/