LINGUIST List 7.467

Wed Mar 27 1996

Disc: Economy, Minimalism, and Formalism

Editor for this issue: T. Daniel Seely <dseelyemunix.emich.edu>

Directory

1. Dan Everett, Re: 7.458, Disc: Economy, Minimalism, and Formalism
2. "David M. W. Powers", Re: 7.458, Disc: Economy, Minimalism, and Formalism
3. John E Limber, Re: 7.458, Disc: Economy, Minimalism, and Formalism

Message 1: Re: 7.458, Disc: Economy, Minimalism, and Formalism

Let me make a couple of comments on recent replies by Dick Hudson
and A. Manaster-Ramer to Y. Falk's posting on economy.

First to AMR, who wonders how we could find a language-specific
principle:

Chomsky pointed out what would constitute language-specific principles
quite explicitly in the debate with Piaget years ago. In any case,
the basic argument would be something like, to take one example, "WH-Island
constraints (or their various reformulations, Subjacency, Tensed-S,
Nominative Island Condition, ECP, etc.) - as formulated in GB, minimalist
theory, etc. - are only statable in terms of language. Therefore, they
are language-specific principles."

Now, this may not convince you. I would be surprised if it did. But the
task faced by one wishing to rebut this argument would be to show that an
equally good account of, say, the ECP is possible for language (i.e. which
gets all and only the relevant extraction facts), but which also accounts
for properties of some other cognitive system, say, vision. This seems
highly unlikely. This is Chomsky's usual strategy.

Of course, there is another kind of reply to this argument, namely, to
compare whole theories (e.g. Cognitive Grammar vs. Minimalist Theory)
and see how well a given theory both handles language facts and vision
facts. But since Chomsky is concerned with grammar (or I-Language) and
many other theories are concerned with Language (his E-language), it is
unlikely that we'll be able to compare theories even on their coverage of
language, much less their transferability to other cognitive domains.
Theories which *are* more or less formally comparable to minimalist theory
(e.g. HPSG) on coverage of grammar, often eschew efforts to find
"cognitive reality" for their proposals as premature. So they cannot be
compared with GB/Minimalism on cognitive validity.

So, the arguments for language-specific properties will probably convince
you if you otherwise are convinced by Chomsky's theory/ies but not
otherwise. I do not think we (the community of linguists) have any way to
proceed at present with a theory-to-theory comparison.


As for Dick Hudson's comments, they seem to me to be
inaccurate. Let me repeat them:

> Yehuda Falk says in reaction to Esa Itkonen's comment on the implications of
> `economy' in Minimalism:
>
> > I think this is a misunderstanding of the formalist position. In the
> > first place, "economy" of some sort has always played a role in formal
> > linguistic theory, back to the notion of an "evaluation metric" in early
> > generative grammar. These have always been economy conditions on formal
> > grammatical processes, and this is no less true of Minimalist economy
> > conditions like Procrastinate and Greed than of earlier approaches. To see
> > them as a radical departure away from formal theory is not justified.
>
> dh: The earlier types of `economy' applied to the activity of the *linguist*
> writing a grammar: assume the minimum of rules, categories, etc. The new
> kind of economy is surely completely different, as it applies to the *user*.
> I think Esa Itkonen is right.
>

DE: No, I think Itkonen & Hudson are wrong. Economy (chosing one analysis
over another based on simplicity, for example) as used in Generative
Grammar through the years is a claim about users of language. A rule with
more symbols, for example, was supposed to be harder to learn than a rule
with fewer symbols. A transformation with more steps was supposed to be a
harder structure to process than a transformation with fewer steps. So the
earlier types of economy did NOT apply merely to the activity of the
linguist, but, crucially, to the user.

- Dan Everett

Message 2: Re: 7.458, Disc: Economy, Minimalism, and Formalism

> From: allegrodns.bluesky.net (Stirling Newberry)
> Subject: Re: vol-7-440, Economy, Minimalism, and Formalism
> 
> In vol-7-440, Yehuda Falk writes:
> > Secondly, the concept of modularity/autonomy/... of the linguistic
> >component, or subcomponents within it, does not contradict the idea that
> >some general cognitive abilities may somehow come into play in language. I
> >don't think there's any question that language is somehow tied to general
> >cognition; however, the existence of formal properties unique to language
> >provides the argument for some sort of modularity of the mind.

In twenty years of searching, I have not found any reasonable 
candidates for such unique universals: common to all languages and 
exclusive to language. I would like to see a list of such putative 
universals. Furthermore, I don't see how it is possible to support 
such a negative statement, since it assumes that we already know all we 
need to know, not just about all languages but about all cognitive 
mechanisms. Moreover, we must ask what does it add to a theory of 
language? It seem to distract us from understanding/discovering the 
mechanisms, whether or not they are unique to language. As Richard 
Hudson notes, the problems and universals on which people tend focus 
arise only in the context of particular theories of language, and 
making the axiomatic or modular simply sidesteps the basic mission of 
science, which is explanation rather than description, and adds 
considerable baggage to the theories in terms of any objective measures 
of economy, minimality or parsimony.

Stirling Newberry comments on Falk:
> It does not seem clear that the premise of this statement is correct - if
> the mind is modularized then there are *no* general cognitive modules or
> abilities, merely combinations or patterns of modes that characterize
> particular modes of thinking. Formal properties unique to language could
> occur not out of sections of the brain unique to language, but of forms or
> patterns of these sections unique to langauge. If we see over time that
> every section of the brain used in language is also used in other thought
> modes then we should expect to see formal properties unique to language
> that are not based on single modes activity but in the result of the
> message passing behavior between two or more of them.

This is an interesting argument. Even the assumption of modularity 
leads to linguistic properties that don't belong to the linguistic 
modules. The big problem with modularity is that it goes beyond 
postulating modules, in the sense of regions focussed on 
distinguishable tasks, to a statement on ontogeny - without considering 
any of the physiological, genetic and developmental factors, let alone 
the implications and predictions which follow from the assumption (and 
which seem to lead to many contradictions). It also flies in the face 
of evidence that the precise area of the brain responsible for a 
particular kind of processing can vary considerably, and that 
neighbouring areas are frequently able to take over such functions. 

It is interesting that Fodor placed his modules in the perceptual, 
motor and lower level processing areas and gave up on the deeper 
coordination/semantic task. But this is to me the only area where 
interesting modules involving non-trivial mechanisms might be worth 
proposing, but their nature clearly extends beyond the domain of the 
purely linguistic to our understanding of and manipulation of the 
world. Linguistic modules thus sell language short, since language 
seems to be intimately tied up with phenomena like thinking, planning, 
consciousness and all of higher level (symbolic) cognition. If 
language is modular and exclusive, then what is left to the rest of 
perception and cognition? Or conversely, we can so explain so much of 
the rest of perception and the basic structural organization of 
language that language (as defined by the language modules) is actually 
redefined to be higher intelligence or something similar.

Alexis Manaster Ramer also makes the point that we need to study some 
OTHER aspect of cognition before we can start making such statements. 
I would go further and note the logical requirement to extend this to 
ALL (or, to start to be convincing to people outside of generative 
linguistics, MOST) other aspects before we can start to make them stick 
as autonomous modules and independent mechanisms. On the other hand, I 
do think it will prove difficult to find other aspects of cognition to 
study which won't lead to the uncovering of mechanisms which have 
explicative power in relation to language.

David Powers 

powersacm.org	 http://www.cs.flinders.edu.au/people/DMWPowers.html
Associate Professor David Powers		David.Powersflinders.edu.au
 ACM SIGART Editor; ACL SIGNLL President	Facsimile: 
+61-8-201-3626
Department of Computer Science			UniOffice: +61-8-201-3663
The Flinders University of South Australia	Secretary: +61-8-201-2662
GPO Box 2100, Adelaide	South Australia 5001	HomePhone: +61-8-357-4220

Message 3: Re: 7.458, Disc: Economy, Minimalism, and Formalism

It is surely NOT the case that the concept of evaluation metric was for "the 
*linguist* writing a grammar." It is a necessary component of any theory 
claiming "explanatory adequacy"--i.e.explaining how (potential) users of 
a language choose among alternative grammars. Discussion of this 
issue can be found in regard to syntax and phonology from the earliest 
days of generative grammar, not only in regard to language acquisition 
but also, I would expect, in regard to language change.

In essence, an evaluation metric might viewed an effort-- along with or
as part of "universal grammar"-- to cope with the the problem of
induction. The actual "work" evaluation metrics have to do may have
changed as the constraints on possible grammars are tightened but it is a
mistake to think these were just tools for linguists.
**********

Prof. Richard Hudson writes:

dh: The earlier types of `economy' applied to the activity of the *linguist*
writing a grammar: assume the minimum of rules, categories, etc. The new
kind of economy is surely completely different, as it applies to the *user*.
I think Esa Itkonen is right.

John Limber
Department of Psychology
University of New Hampshire, Durham NH 03824, USA
email:jelchrista.unh.edu 
http://pubpages.unh.edu/~jel (course information, etc.) 
FAX (603)-862-4986