LFG and functionalism

[bresnan at csli.stanford.edu]

The relation of LFG to functionalist linguistics is something that is
widely misunderstood.  One often sees authoritative-sounding remarks
by linguists to the effect that despite its name, lfg has nothing to
do with functionalism, but is just a brand of formal linguistics, an
offshoot of the Chomskyan juggernaut.  (I saw this explicitly stated
in a review in LANGUAGE, among other places.)

I disagree, but would appreciate hearing others' perspectives.
Consider the following:

(1) historically, lfg arose from combining lexicalist syntax (Bresnan,
Mchombo, others in MIT tradition) and functional-unification ideas
modelled by ATNs (Kay and Kaplan).  The latter strand stems in its
LINGUISTIC antecedents directly from Halliday, the distinguished British
functionalist.  (Martin Kay can perhaps elaborate
on this.)

(2) lfg's constraint-based architecture of parallel structures,
including forms of expression (c-structures), grammatical relations
(f-structures),  prosodic structures, etc., embodies a typically
multidimensional functionalist conception of language--except that it
(or at least the simpler parts of it) has been formalized in a
well-understood mathematics.  Moreover, it is open-ended, allowing for
extensibility to discourse, etc., through the projections idea.  
Only the fact that so many of us have done our empirical work on the
forms of expression/grammatical relations parts (the syntax) seems to
obscure the big picture and create a solely generative/formalist impression.
But much work in lfg has drawn on discourse, pragmatics, and texts
(Simpson, Matsumoto, Hong, Sells, Bresnan-Halvorsen-Maling, Bresnan
and Kanerva, Bresnan and Mchombo, Austin and Bresnan, etc.).

(3) lfg's general LINGUISTIC principles, such as WYSIWYG (Zaenen
1989), economy of expression (Bresnan 1995b), and nonuniformity of
categorization (Bresnan 1995a), reflect fundamentally functionalist
ideas (compare Haiman's 1985 NATURAL SYNTAX).  Similarly, ideas of
constraint competition and prominence, unifying the linguistic domain
with wider conceptual domains (Mohanan MS., Tara Mohanan 1993, Culy
1995), are functional through-and-through.

(4) lfg has friendly relations with functionalists such as Peter
Austin, Talmy Givo'n, Doris Payne, Bill Foley, Matt Shibatani, Knud
Lambrecht (among others), in the sense that these linguists have
contributed important empirical and theoretical ideas that CLOSELY
MATCH lfg work in spirit.

There are differences of methodology and foci of interest between lfg and
functionalist linguistic theories.  --But, let's face it, the
wider our range of methods and empirical sources, the better our work
can be.

Joan

-----------------------------------------        ______     __o       __o
Joan Bresnan	bresnan at csli.stanford.edu      ______     _`\<,_    _`\<,_
-----------------------------------------       ______   (*)/ (*)  (*)/ (*)