need for hooks

[Brian MacWhinney]

The idea that restrictive relative clauses are composed of material that
has been cognitively unitized is pretty far from hand-waving.
Psycholinguistic research from the 60s and 70s by Rommetveit and Turner,
Clark, Krauss and Glucksberg showed how restrictive relatives are used to
distinguish members of a contrast set.  Typological work by Givon, Keenan,
Comrie, and others demonstrated asymmetries in relativization types that
matched up well with underlying functional characteristics.  In more recent
psycholinguistic work, Bock and her colleagues have explored processes
which allow previously mentioned material to form the kernel of further
utterances.  Bock has focused on passives and datives.  Earlier, Levelt
looked at question structures.  The message from this work on what Bock
calls syntactic persistence is that the use of a syntactic pattern in
previous discourse tends to make it available as unit for further
processing.

Many of the syntactic phenomena that revolve around constraints on raising
from relatives emerge rather directly from these facts.  I can't remember
ever having thought or said that syntax should be eliminated.  I consider
syntax a wonderful, complex, and fascinating fact of nature.  I simply
believe, like Geertz, Wilcox, and Aske, that it should be explicated.  In
particular, I think that syntacticians have a responsibility to the rest of
the linguistic community to make their analyses more penetrable to
explication.   This can be done by including "hooks" in syntactic theory to
concepts and constructs that match up with what we know about language
processing and use.  The treatment of restrictive relative clauses
discussed in some of the previous messages is a prime example of a
construction for which such a "hook" is needed.

Like computer programs that have hooks, theories that have hooks have to be
designed in a way that supports communication between disciplinary
"modules".  For example, the theory of syntax would need to support hooks
for things from psychology like cognitive unitization, syntactic
persistence, memory strings, construction generalization, and the like.
Including hooks to these objects would markedly alter the shape of
syntactic theory.  It would definitely not make syntax disappear.  However,
it would allow syntacticians, functionalists, and psycholinguists to
communicate and collaborate more effectively.  I'm not sure that it would
bring us to the point of using syntactic theory to make statements about
children as expert witnesses, but it might get closer.

I have been told that the increased role of logical form in minimalist
syntax may represent a movement in this direction.  It would be interesting
if that were the case.

--Brian MacWhinney