form versus meaning

[Elizabeth Bates]

In response to Edith Moravcsik's message, I think there is one more
hidden assumption: that items either are, or are not, members of a
syntactic class, and if they are members, they are members to the
same degree.  This is a classic approach to category membership, but
its psychological validity is highly questionable, after more than
two decades of research on prototypicality effects, fuzzy boundaries,
ad hoc categories, context-dependent categorizations, and so on.
As it turns out (vis-a-vis our Martian visitor), native speakers give
highly variable judgments of syntactic well-formedness depending on
the relative degree of "verbiness" of a verb, "nouniness" of a noun,
"really-good-subjects" vs. "not-so-great subjects", and so on.  It has
proven very difficult to explain these variations without invoking
something about the semantic content of the item in question, and/or
its pragmatic history, frequency of use, etc.  Now, one can
always make the classic move (since 1957) of ascribing all those
variations to "performance," salvaging one's faith in the purity
of the underlying competence.  But any research program that sets out
to describe competence "first" and deal with performance "later" is
going to run into trouble, because pure data that give us direct
insights into competence are simply not available.  That is precisely
why we are all still having this argument.


And while we are at it, I am puzzled by the suggestion that we should
describe language "first" before any investigation of its biology can
be carried out.  Should physics be complete before we attempt chemistry?
Must biological research stop until we have discovered all the relevant
facts from chemistry?  Why does linguistic description (field linguistics
or self-induced grammaticality judgments) have priority over any other
approach to the study of language?  Is psycholinguistics a secondary
science?  Should research on aphasia come to a halt until we know
exactly what it is that the aphasic patient has lost?  It seems to
me that we need all the constraints that we can get, and that all
levels of inquiry into the nature of language are valid and
mutually informative.  The key is to be sure we know which level
we are working on.  For example, I believe that claims about innateness
are biological claims, that require biological evidence.  Proof that
a given structure is (or is not) universal may be quite interesting and
useful to someone who is investigating the biological underpinnings
of human language abilities (genetic, or neural), but proofs of
universality do not constitute ipso facto evidence for the innateness
of some domain-specific linguistic structure, because that
structure may be the inevitable but indirect by-product of
some constraint (e.g.from information-processing) that is not,
in itself, linguistic (e.g. as Matt Dryer points out, some kind
of memory constraint).  To untangle such problems, many different
kinds of evidence will be required, and none of them should be
granted priority over the others.  -liz bates