Tom's middle ground

[DAVID A. ZUBIN, LINGUISTICS, SUNY BUFFALO, 685 BALDY, BUFFALONY 14260, 716-645-2177, FAX 645-3825]

Dear Tom,

Having paid attention and learned a great deal from your pleas for the
middle ground, and having incorporated a great deal of it into my
thinking, I would still respectfully like to counter you with the
following points:

Good work along extremist lines--"atomism," aka concretist semantics of
the Berkeley school;  pure conceptualist semantics such as Len Talmy's;
abstract semantics such as Roman Jakobson's work, or the Columbia
School--is anything but reductionist in the negative sense.  The
"extremism" of these approaches, the willingness to stretch the envelope,
School--is anything but reductionist in the negative sense.  Theto push
ideas beyond what at first glance seems reasonable, has led to
the development of substantial new ideas, and the testing and refinement
of lideas across 'camp' lines by people willing to listen and worry.  Let
me briefly cite two examples:

1.  The long-term debate between abstractionist and concretist semantics
over whether there is any central "abstract" meaning to a category, and
in the other direction, whether there are conventionalized submeanings
and sus-sub meanings etc.  Abstractions willing to listen have been
troubled by the fact that you can't come anywhere near predicting use
from a single abstract meaning, and have increasing worked on the
conventionalization of inferences run off an abstract meaning + context;
this amounts to a submeaning with some motivation for its association
with the abstract meaning of the category.  On the concretist side, many
have worried about the proliferation of submeanings, and have
increasingly worked on general principles for how submeanings may be
derived and linked.  Both these lines of investigation have been fruitful.
increasingly worked on general principles for how submeanings may be
derived and linked.  Both these lines of investigation have been fruitful.

2.  In recent years increasing attention has been paid to an
exemplar-based theory of category structure.  Now here's a crazy,
extremist idea (from a linguist's point of view) if there ever was one.
No structure, no stable links, just thousands, millions of
representations of individuals that you've encountered in the past and
assigned to that category.  So how does one assign new members to such a
category?  A typical story goes like this:  having gotten into the right
general domain through some heuristic, you grab a bunch of exemplars at
random from each of the candidate categories and assign the new exemplar
whereever it gets the highest resemblance score.  Sounds crazy, but there
is a growing body of evidence in cognitive psychology that something like
this is going on.

Do I believe the extremists?  I believe they should keep stretching their
envelopes to see what they can find.  And I believe they should patiently
listen to eachother.  Its good exercise.  We should all practice a little

                             [END of message text]
  PINE 3.91   MESSAGE TEXT         Folder: sent-mail  Message 10 of 10 END

envelopes to see what they can find.  And I believe they should patiently
listen to eachother.  Its good exercise.  We should all practice a little
exremism in our work.

P.S.  Enjoy Matthew D. while he's on sabbatical at your place.  But you
can't have him.

best,

David