LINGUIST List 12.1855

Thu Jul 19 2001

Disc: Next to Last Issue: Nonobjects/Syntactic Study

Editor for this issue: Lydia Grebenyova <>


  1. Jose-Luis Mendivil Giro, Re: 12.1816, Disc: On Nonobjects of Syntactic Study
  2. Dan Everett, On the Necessary Arbitrariness of UG

Message 1: Re: 12.1816, Disc: On Nonobjects of Syntactic Study

Date: Wed, 18 Jul 2001 19:05:52 +0200
From: Jose-Luis Mendivil Giro <>
Subject: Re: 12.1816, Disc: On Nonobjects of Syntactic Study

Here is a reply to some comments by Dan Everett in several messages:

>A couple of comments on a reply to my posting from:

> Jose-Luis Mendivil Giro <>(...)

>Calling something beyond the senses, which can never be studied directly,
>and which is only an abstraction useful to a particular theory, a 'natural
>object' is a bit beyond my credulity threshhold. But I did that for long
>enough myself to understand the appeal.

But 'an abstraction useful to a particular theory' is more or less the
only definition of reality that scientists consider seriously. For
example, the physicist Weinberg:

"Wave functions are real for the same reason that quarks and
symmetries are --because it is useful to include them in our theories"
(quoted from the book by L. Jenkins (2000): Biolinguistics. Exploring
the Biology of Language. Cambridge University Press, a book you surely
would find very useful).

In a response to Kevin R. Gregg you (Dan Everett) wrote:

>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.

Why? I can see here some 'Cartesian dualism'.

>But UG is quite a different matter. First, it is in principle neither
>spatially nor temporally bounded, as black holes are expected to be.

But if UG would exist as a Language Acquisition Device (LAD) it should
be expressed by the genes and, so, 'spatially and temporally
bounded'. Even for Chomsky a gene is an historical object.

Can you explain why do you think that the existence of some kind of
LAD is impossible? Just because you cannot see it in a CAT scan?

The existence of a LAD in our minds/brains helps when trying to
explain the rapid and stable acquisition of language and the
convergence between human languages. If you can explain these two
facts in a better way and, at the same time, to explain the structure
of every sentence, and to predict the grammaticality of all sentences
in all languages without postulating some kind of LAD or UG,
congratulations, but I would like to see it.

>it claims to be a representation of pure grammar, stripped to 'virtual
>conceptual necessity' (and what is necessary and allowable is nothing more
>than tree structures, by and large). How could one study it? One would have
>to sort through the data and decide that this or that datum is not a result
>of function, history, or sociological factors (among others). Then that
>datum would bear on UG. It is not possible to make that decision, however.

But it is just the opposite!

What can be explained
extra-grammatically/extra-linguistically/pragmatically about the
structure of sentences in a given language (why are they like they are
and have the meaning they have and not other) is not very much by now,
if any. The remainder must be explained grammatically
- linguistically. The only thing you have to do is to construct (to
invent) a theory to explain data, as in Physics, without prejudging if
the construct is real or it is not.

>UG is an essence. Its essence is partially derived from the layers of other
>material constructed around it (LF, PF, etc.) and being in principle more
>and more reifiable until we arrive at the essence of grammar. Not being a
>physicist, I cannot swear that there is nothing like this in physics. But if
>there were, the problem would be the same.

But you are taking (a not very good interpretation of ) the Minimalist
Program as the only possible model of UG, and it is not: even if the
Minimalist Program were wrong, the notion of a UG common to all humans
as part of our nature could be fine (if empirically adequate).

You can say that Chomsky's theory of UG is not correct, or that
grammar is entirely a response to functional pressures, that it is
only the consequence of more general cognitive abilities, etc.,
because these are empirical matters, but you cannot say that we cannot
study the knowledge of language, just because we have been studying it
for centuries. But this is what you are doing when you discredit the
construction of a theoretical model of the knowledge of language.

Generative Linguistics seeks to find out how our knowledge of language
is, and the only way to do it is constructing a(n empirically
testable) theory of this knowledge. Distinguishing what principles or
properties of this system are purely grammatical or derived from other
cognitive systems (phylogenetically or ontogenetically) is a different
task, perhaps a posterior one. Determining how this system is
represented in the brain (if it is) is not a task for the linguist.

>UG makes claims that go far beyond the claims of other disciplines. For
>example, Chomsky claims that language is a perfect biological system. There
>is no other biological system like this and biologists find this claim
>wacko. No other theory makes such claims.

What Chomsky tries with the Minimalist Program is to shave UG with
Occam's razor using reverted engineering (although he has always
distrusted adaptive -neodarwinist- explanations of language, and he
has been condemned for that):

"We may therefore ask to what extent language is a 'good solution' to
the legibility conditions imposed by the external systems with which
it interacts. Until quite recently this question could not seriously
be posed, even formulated sensibly. Now it seems that it can, and
there are even indications that the language faculty may be close to
'perfect' in this sense; if true, this is a surprising conclusion"
(Chomsky, 2000: 9).

So, UG is a system that interacts efficiently with other systems (in a
modular mind), not an essential entity.

If you allow me another quote (the perfect tool for the non English
speaker), Juan Uriagereka addresses this very question quite clearly:

"That lively debate is refueled by one of the basic premises of the
Minimalist Program: that language is, in some sense, optimal. For the
neo-Darwinian this might be good news, if the optimality in question
could be shown to be functional. However, linguistic optimality is not
functional at all, but only structural; in fact, functionally the
linguistic system is definitely suboptimal --support, instead, for the
neo-neo-Darwinian" (Rhyme and Reason. An Introduction to Minimalist
Syntax. The MIT Press, 1998).

When Chomsky uses the terms 'perfect' or 'perfection' he always uses
'quotation marks', because he is referring to a technical sense of the
word. When we question about 'perfection' we are asking about what is
conceptually necessary and optimally economical respect to the meeting
of external conditions (legibility, etc.) on language.

As you can read in Chomsky, or in Uriagereka, the Minimalist program
explores structural elegance and tries to justify it, but always as an
empirical enterprise.

>See John Goldsmith's posting. It seems right. I am not denying science. But
>UG makes claims that go far beyond the claims of other disciplines. For
>example, Chomsky claims that language is a perfect biological system.

When? Where?
What he says, as far as I can remember, is that this level of
'perfection' (in the technical sense) is not obviously reducible to
neo-Darwinian evolution, but that it perhaps demands an explanation
considering 'laws of form' or unknown properties of complex systems
yet to be discovered.

>is no other biological system like this and biologists find this claim
>wacko. No other theory makes such claims. Don't fool yourself with the
>entirely inappropriate, but incredibly common, way of excusing UG by
>comparing it to physics. There is little to compare there at any level.

Now I understand why you don't understand generative grammar.

>Rather, UG should be compared with Freudian constructs.

I see...

Best regards,

Jose-Luis Mendivil
Universidad de Zaragoza
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: On the Necessary Arbitrariness of UG

Date: Wed, 18 Jul 2001 15:26:35 -0500
From: Dan Everett <>
Subject: On the Necessary Arbitrariness of UG


What emerges from the discussion of UG as a nonobject so far is that
the arguments for it are in general old philosophy of science 101
kinds of comparisons with physics and arguments against it are
concerned with the fact that it has no obvious connection to the brain
or that it is not well-definable since UG would have to grammar
without external influences, a substance we cannot expect to find.

It is vital in the overall discussion that we consider yet another
curious aspect of UG - familiar to functionalists for quite some
time. This is its semantic arbitrariness and rather striking distance
from our intuitions. So, for example, Kayne's Antisymmetry of Syntax
book (which I was at one time quite intrigued by, cf. the
acknowledgements to the book) offers the bold conjecture that all
languages are SVO. Minimalist Program papers in general offer analyses
that have movement based on the "greed" of morphological features
rather than, say, the much more obvious (and intuitively appealing)
concept of information structure. I do not wish to argue that such
moves are bizarre, though I think they are. (For one thing, if I said
that, someone would immediately say that the notion of gravity is
bizarre and we'd get going on the "don't linguistics (=Chomskyan) any
worse than physics" line of reasoning.) >>>More importantly, I want to
suggest that UG *requires* such bizarre analyses and places a premium
on removal of the analysis from an intuitive basis. It could not be
any other way.<<< Why is this?

The answer is straightforward. If UG were intuitive and, say,
syntactic "movement" were based on information structure, semantics,
etc., then UG would be learnable (Bob VanValin has several relevant
papers on this very topic at the RRG website
( ). But if UG is learnable
then it is unnecessary. So no theory of it *can* present it in such a
way that it is learnable. Only analyses in which it is unlearnable are
acceptable. Hence the most valued analyses will always of necessity be
detached from semantics and information structure. Let's take an
example, familiar no doubt to all on this list:

(i) The man is in the room.
(ii) Is the man in the room?
(iii) The man who is tall is in the room.
(iv) Is the man who is tall in the room?
(v) *Is the man who tall is in the room?

(v) is supposed to show that children acquire/activate a rule of
question-formation based on an innate notion of main vs. subordinate
clauses. In fact, this argument only goes through if we assume that
children have no access in their rule-formation to either
prosody-marking of information structure OR any notion of what
sentences are about. The sentences in (iii) and (i) are assertions
about the man being in the room, his height being old information
useful to processing. Prosody and other components of information
structure in English, completely apart from a VP or 'main verb'
notion, make this quite clear. So the child need only learn that the
focus, not the topic, is what is questioned.

The literature on UG rarely engages these possibilities seriously and
the syntax literature on UG/syntax can *never allow* semantics or
pragmatics to play a causal role in the statement of any rule or
constraint. Chomsky has traditionally rejected such a role for
semantics. In his articles on the contributions of surface structure
to semantic interpretation, semantic interpretation is allowed to be a
factor in selecting among different grammatical structures, but it may
never be part of the structural description or change of any such
structure, nor could it be if UG is to be maintained.

UG is not only required to be 'pure grammar', this restriction
requires UG to be utterly divorced from semantics and pragmatics,
relegated to the role of a 'Conceptual-Intentional Interface' System.

Pragmatist linguistics, like functionalist linguistics/cognitive
linguistics, is able to take advantage of speaker knowledge across
several domains of knowledge to generate sentence structures as
largely formed by what we are talking about. In this sense, UG for
language is rendered unnecessary, indeed, incompatible with the
research program of these other perspectives.

A further illustration might be to consider the possibility of a UT,
Universal Technology. Most societies, perhaps all, have bows and
arrows. Are bows and arrows innately specified? Of course not. We can
see how, given general cognitive properties of humans, bows and arrows
provide a natural solution to an ever-present problem: eating protein
that can move faster than us featherless bipeds. Once we recognize the
utility, the form follows.

I used to argue with Tom Givon all the time when he said the same
thing. So, it turns out that I was wrong and he was right (though I
don't want to be seen as putting words in his mouth. He has enough
there of his own). This distribution of correctness will not surprise

By the way, let me make it very, very clear that I am not advocating a
function-only approach to syntax, not at all. I do not consider myself
a functionalist. I am proposing no 'either/or' dichotomy between
information structure and serious, formal syntax. We need
both. Although doing both function and form together will obviate the
need for UG, they are not mutually exclusive but complementary. This
is why I have become a fan of Role and Reference Grammar - it works
hard at handling both of these vital aspects of human languages.

- Dan Everett
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue