Ergativity correlations

[David Parkinson]

Hi Funknetters:
        Some of us here at Cornell are having a seminar on ergativity, and
the question came up about whether anyone has ever carried out research to
see whether ergativity correlates with any other aspects of linguistic
structure (word order, head vs. dependent marking, presence of
antipassive). Something like Dryer's or Nichols' research would be nice,
but even something on a less sweeping scale. Anyone know of anything like
this? There are reports in the literature such as "ergative" languages
tending to have verb-initial or verb-final order, but has this ever been
rigorously tested?
        Also, does anyone know if there has been work done on the somewhat
strange status of ergativity vis-a-vis accusativity? For example, it is
often said that while languages exist whose sole means of marking
grammatical relations is accusative patterning, no language exists whose
sole means is ergative patterning. Also, what are unified subject
properties in "accusative" languages often split in "ergative" languages
between grammatical subject (i.e., final 1, uniting Dixon's S and A: e.g.,
reflexive antecedence, equi deletion) and absolutive case (uniting Dixon's
S and O: e.g., relative extraction). It may be accusativocentric to see the
split from the point of view of a unified category in "accusative"
languages, where it may be more appropriate to see them as a collection of
properties which coincide in "accusative" languages and split for
principled reasons in "ergative" languages (i.e., Manning 1996). But why
should a language not exhibit ergative properties everywhere? And why
should subject properties split in the way they do?
        Any help appreciated.

David Parkinson
dp11 at cornell.edu

P.S. I am using scare quotes on the terms "ergative" and "accusative"
language, since it is often a misnomer to identify a language as one or the
other.