LINGUIST List 2.105

Sunday, 31 Mar 1991

Disc: Functionalism

Editor for this issue: <>

Directory

1. Ellen Prince, Re: Functional Linguistics
2. , formal and functional approaches
3. Brian MacWhinney, Is functionalism Inherently Shaky?

>Date: Sun, 24 Mar 1991 10:53 PST
>From: Scott Delancey <DELANCEYoregon.uoregon.edu>
>Subject: autonomous linguistics
[...]
>Poser and Everett, among others, seem to treat "cognitive" and "functionalist"
,
>in this context, as synonymous. This is both correct and incorrect (ass
>any functionalist or cognitive grammarian would predict, of course; this
>is how catgorization works); the essential point in which it is correct
>is that all of the various research programs (there are at least three
>clearly distinguishable ones) which fall under one or the other of these
>terms share an unwillingness to accept a priori the assumption that
>significant aspects of morphology and syntax (phonology is likely a
>different story) are to be explained only in terms of language-specific
>formal priniciples. 

at the risk of starting yet another long-winded debate, i must point out that
there are those of us who consider ourselves 'functionalists' (in that we
study the discourse and/or processing functions of linguistic form) who indeed
do accept that significant aspects of syntax are to be explained only in terms
of (autonomous-)language-faculty-specific formal principles. (i'm assuming
that's what scott meant, not 'language-specific'.) furthermore, i for one have
never seen any compelling evidence to the contrary, although i have seen a good
deal that supports this position. 

so let's not be too quick to make generalizations about what various 'Xists'
do or do not believe, ok guys?

I don't quite agree with Fritz Newmeyer's view that 'formal' and 'functional'
approaches are compatible, but maybe that is because half the discussion over
these two approaches (trends?) seems to be an attempt to define just what they
mean. I did appreciate Fritz's attempt to say where formalists and
functionalists differ:

> ...As I see it, the task of the 
> formalist is to characterize the structural possibilities of language, 
> both universal and language-particular. The task of the functionalist 
> is to elucidate the principles (largely nonspecific to language) 
> governing how those structures are employed in actual discourses. 
> Thus each will find different data relevant to their concerns...

First of all, I think that this underscores a point I made recently about the
needs of computational linguists. Although NLP research owes a great deal to
the work of formalists, NLP researchers really need to be wary of formal
linguists who want to sell their programs without really understanding what
the customer wants. Being able to enumerate grammatical structures is largely
worthless if you can't say how they are employed in actual discourses. You
have to know what to do with the structures once you have them. Come to think
of it, this point probably applies to most everybody outside the field of
linguistics who are interested in linguistic phenomena. They like their
skeletons covered with flesh. 

Secondly, I would like to point out that "the principles...governing how
...structures are employed in actual discourse" cannot possibly be nonspecific
to language. They make crucial reference to the structures that are specific
to language. The strategy that tells me how to use a relative clause has to
know what a relative clause looks like, doesn't it? In order to produce a
discourse, I have to know everything there is to know about the structural
possibilities of discourses, don't I? It is true that I might understand the
ill-formed sentence "I filed the paper after reading" in the way it was
intended by the linguistic perpetrator, but does this mean that the rule which
tells me that the participle is missing an object is *not* a rule of language
use? All I can glean from that sentence is that the speaker doesn't employ
the same linguistic strategies that I do. It should not compromise anyone's
functionalist ideology to believe that such sentences are low-frequency
because they violate regular strategies governing the production of English.
So I don't agree at all that "the functionalist...will show little interest"
in ungrammatical sentences. The question of grammatical well-formedness is
(or ought to be) as important for functionalists as for formalists. If some
functionalists disagreed with this point, I would not take it as an endorsement
of a formalist approach, but rather as an indictment of their rendering
of the functionalist approach.

 -Rick Wojcik

Fritz Newmeyer correctly notes that functionalist approaches must include
some way of deciding the competition between conflicting motives. For example,
if we say that the noun preceding a transitive verb in English is typically
agential, we need to invoke some competing principle to account for the 
word order found in the passive. It is true that some functionalist
accounts have failed to include such mechanisms. However, there is also a
substantial body of work which makes use of the concepts of cue validity, cue
cost, and cue strength within a mathematically explicit framework to escape
the problem cited by Newmeyer. 

The full solution to the problem requires at least three levels
of analysis. On the first level, one identifies the cues and motives
that compete in sentence processing and language change. This
level of analysis was already fairly solid in the functionalism of the
1970's. On the second level, one makes a psychometric/psycholingustic
commitment to the empirical measurement of cue strength. For an
excellent example of this form of analysis, see the fuzzy logic model in
"Speech Perception by Ear and Eye" 1987 by Dominic Massaro. On the
third level, one makes a commitment to predicting cue strength from the
basic cue validity
properties of the input to the language learner, as they are attenuated
by cue cost factors. The concepts of
cue validity and cue cost are applied to data on sentence processing in
over a dozen languages in "The Crosslinguistic Study of Sentence
Processing" by B. MacWhinney and E. Bates (Eds.) 1989. 

My guess is that, once he has taken a look at the Competition Model
in the book edited by MacWhinney and Bates, Newmeyer will ask for still
further commitments regarding the determinants of cue cost. A primary
goal of psycholingusitics is the elucidation of detailed facts about cue
costs as possible determinants of language universals. However, it
believe that the mathematicization of the functionalist model presented
in the MacWhinney-Bates book goes a long way toward addressing
Newmeyer's concerns and demonstrating that functionalist linguistics
need not rest on an empirically shaky foundation.

--Brian MacWhinney
 Carnegie Mellon University

[End Linguist List, Vol. 2, No. 105]