LINGUIST List 12.1845

Tue Jul 17 2001

Disc: On Nonobjects of Syntactic Study

Editor for this issue: Lydia Grebenyova <lydialinguistlist.org>

Directory

1. John Goldsmith, RE: On Nonobjects of Syntactic Study
2. Kevin R. Gregg, Re: 12.1823, Disc: On Nonobjects of Syntactic Study
3. Robert R. Ratcliffe, Re: 12.1824, Disc: On Nonobjects of Syntactic Study

Whitney Anne Postman wrote: "In Linguist 12.1819, John Goldsmith
wrote: "it is controversial that that system of knowledge lives in our
brains," where "that system of knowledge" presumably refers to human
natural language.

"My first question is, If language does not originate in our brains,
then where?"

JG: These are very tough questions, harder still to answer coherently
in a few lines. I make the following agreement with myself (which I
commend to others): I won't do linguistics professionally unless I can
provide myself with a coherent account of what I think I'm doing. I
also have some hopes and suspicions about what will some day turn out
to be true (about brains, language, and so on). But! the coherent
account that I owe myself can't _assume_ the truth of what I hope
someday we'll find out. What this means in practical terms, for me, is
that I owe myself an account of linguistics whose coherence is based
on the actual and practical activities of linguists studying people
and their speech (etc.).

Now, I'm very interested in neural networks, and the field of studying
neural networks is more or less a subfield of applied math that tries
tore out how knowledge structures can be embodied with particular
sorts of structures. This is, to me, an extremely exciting area, even
though results worth reporting back to the LSA at this point are
minimal.

So I'm fascinated -- me too -- by brains and their relation to
knowledge, but we know LOTS about language, and relatively little
about how the brain can be directly involved in embodying that
knowledge. You and I may _believe_ the brain is doing all that, but
that belief isn't a sufficient grounding for the field of
linguistics. And I believe that there _is_ a sufficient grounding for
linguistics: it has to do with the character of the research that we
linguists do in exploring how people speak and write.

(This is analogous to the question as to whether we can understand
mathematics by understanding cognition; I find that position untenable
because mathematics has a far stronger valid certainty associated with
it (I mean, we _prove_ things) than anything we know about the brain.
This too is a subject that requires at least a paper, maybe a book!)


"Goldsmith also wrote: "it is (even more) controversial that
linguistics is a scientific theory of something that is in the brain."
As someone who's investigated acquired aphasia (=language impairment
due to brain damage) firsthand, in particular the very real phenomena
of agrammatic production and comprehension of language (Standard
Indonesian, in my studies), I fail to see what is (even more)
controversial here."

JG: Someone who is studying the effects of brain damage on language
production and comprehension is undoubtedly studying brain function --
there's no doubt about that. But that doesn't rub off (so to speak) on
every thing us linguists do: consider two people arguing about whether
there are ternary feet in the inventory of stress patterns, or two
others arguing about whether certain syntactic rules should be allowed
to have access to word-internal morphology -- it's way not obvious
that they're discussing anybody's brain structure or function. That's
all I meant.

Message 2: Re: 12.1823, Disc: On Nonobjects of Syntactic Study

Dan Everett wrote:
>Of the many things that one might
>say in reply to your point, just imagine a cosmology that got all the facts
>about 'wobbly revolutions' of stars without, say, needing to posit black
>holes. Then a theory which posited them would be less attractive - why
>accept something which can in-principle never be directly observed as an
>explanation? In the case of astronomy, one accepts such things because it is
>useful to do so. Likewise in quantum theory. But Linguistics ought not to be
>confused with physics. The similarities of abstraction are utterly
>superficial. 

****If you have two equally successful theories--i.e. two theories with the
same explanatory power--but one posits black holes and the other doesn't,
of course one chooses, ceteris paribus, the one that doesn't. But this
isn't because black holes can't be observed, but rather because they're
otiose. I don't think this example replies to my point (hardly *my* point,
of course).
 My point, once again, was that in general direct observability
is never a criterion for accepting a construct as a part of a
scientific theory.

 Nor, I should add, are nonobservables only allowed in on
condition that they could in principle be directly observed if we had
better instruments, etc. Mental phenomena, naturally, are
unobservable in principle, yet somehow cognitive scientists of all
stripes soldier on. But in general, certainly if Bogen & Woodward (or
Putnam or Cartwright) are correct, phenomena in general are not
observable: you can't see the melting point of lead, or inheritance of
traits, or speciation, etc. etc.


>Chalmers: '...descriptions of observable
>> states of affairs are in general quite inappropriate for constituting the
>> building blocks from which scientific knowledge is constructed...'
>
>This is hardly new with Chalmers. But he has a vested interest in this kind
>of statement, since he studies consciousness. Still, though, it is unlikely
>that he studies it via pure thought. He must use abduction, induction, and
>deduction from observables just like the rest of us. Observables *are* the
>appropriate building blocks. 

****Actually, I think we're talking different Chalmerses; my Alan
vs.your David. In any case, no one's arguing for 'pure thought'; as I
said, we use observations as the foundation for our inferences,
inductive, deductive, or abductive. The inferences are the building
blocks, not the descriptions.
 I'm not about to try to defend UG, even if I wanted to. As I said,
it could easily be as wacko a concept as Dan Everett suggests. But there's
nothing whatever wacko about positing a non-observable UG to explain
various facts about language; and insofar as a UG theory does a better job
of explaining those facts than any competitor, it would be the height of
wackosity to reject it on the grounds of non-observability.


Kevin R. Gregg
Momoyama Gakuin University
(St. Andrew's University)
1-1 Manabino, Izumi
Osaka 594-1198 Japan
tel.no. 0725-54-3131 (ext. 3622)
fax. 0725-54-3202

Message 3: Re: 12.1824, Disc: On Nonobjects of Syntactic Study

> Date: Sun, 15 Jul 2001 10:16:19 -0400
> From: Whitney Anne Postman <wap2cornell.edu>
>
> My first question is, If language does not originate in our brains,
> then where? To head off a particular vein of criticism with which
> I've been confronted before, let me state that I assume that the idea
> that language is a part of human biology is not in the least bit in
> conflict with the idea that language is also part of human culture.
> Why should these two notions (beliefs?) be mutually exclusive?

Why indeed. I found myself pretty much in sympathy with Dan Everett's
and John Goldsmith's postings on this subject. I also am inclined to
reject the Chomskyian paradigm as an adequate basis for scientific
research on language. But I'm not particularly attracted to positivism
or functionalism and of course I see nothing wrong with modelling what
we cannot directly observe. My objection to Chomsky's theory is simply
that it is ambiguous and too full of unstated and hidden assumptions,
that in a truly scientific theory would have to be stated explicity
and, ideally subject to falisfiability by empirical test.

The most important hidden assumption is whatever is behind the "/ " in
the formulation "language is a part of the mind/brain," cited a couple
of times in this thread. Does / mean = ? In that case why refer to the
mind at all. He seems to be making a distinction between knowledge
(mind) and the tissue structures which support knowledge (brain), but
then asserting that sometimes this distinction does not apply and some
kind of knowledge (knowledge of language, knowledge of syntax?) simply
is tissue. At least that is the only sense I can make of the claim
that grammar is an organ or a natural object.

 I mean, I am perfectly willing to be convinced by Chomsky's two
central propostions-- namely that language is in the mind, and that
the human brain is uniquely adapted to acquire language. But accepting
these two propostions does not commit me logically to any particular
theory of how they are related. There are a variety of ways we might
model the relationship between mind and brain, or between knowledge of
an individual language and the neurological capacities which support
that knowledge. Rather than asserting fallaciously that UG is the
only possible theory, we should be engaged in explictly formulating
various models and looking for empirical tests which would help us to
choose between them.

 Getting back to the orginial question of the non-objects of
syntactic study, I've always felt that besides the unresolved
ontological questions, there was a fundamental epistemological paradox
at the heart of the generative enterprise, which goes like this: If
grammar is LEARNED from surface language data, either by a process of
induction or a process of theory formation and experimentation, then
it is reasonable to assume that we linguists could arrive at a model
of that knowledge by the same process of induction or experimention
over surface data. But if grammatical knowledge is structured
according to principles which are not and cannot be learned from
surface language data (UG), then there is no reason for linguists to
think we could learn anything about these principles by analyzing
languages.

 I see many examples of this paradox in action in the generative
literature. There is a particular research cycle that goes like this:
Someone proposes principle X to account for a particular phenomenon in
language A and it is proposed that principle X belongs to UG. It is
found that principle X does not apply in language B. If it was clear
in linguistics, as it is in other sciences, that the goal of theory
formation is to account for a body of data, then everyone would
immediately start looking for a better theory which could cover both A
and B. But in generative linguistics (as often as not) the
non-applicability of X to language B is embraced as 'proof' for the
universality of the principle and the innateness hypothesis itself.
Principle X (it is proposed) in fact does apply in language B at a
deep level, but subsequent derivations erase all surface evidence for
it. Since it can't be learned by speakers of B from the surface
evidence, it must be innate! This is called 'abstractness': the less
applicable a grammatical principle is in the languages of the world,
the more abstract the derivations needed to make it applicable, the
more likely it is to be universal.


 -- -----------------------------------------------------------
Robert R. Ratcliffe
Associate Professor, Arabic and Linguistics,
Dept. of Linguistics and Information Science
Tokyo University of Foreign Studies
Asahi-machi 3-11-1,
Fuchu-shi, Tokyo
183-8534 Japan