topics: |  linguistics | chomsky | philosophy

Defends the skepticist position towards semantics. 

	if semantics is the relation between sound and thing, it may not
	exist. p.73

Chomsky embodies a long tradition in American linguistics, going back to
Leonard Bloomfield: "Linguistic study must always start from the phonetic
form and not from the meaning" (1933:162).  Zellig Harris was to expand on
this thought, by proposing formal models of syntax which were carried
forward by Chomsky.

displays a cavalier attitude, as always, towards critics: 

	the issue of innateness of language is a curious one. There is a huge
	literature arguing against the innateness of language; there's
	nothing defending the thesis. p.72

and makes no attempt to refute any of the criticism.  

What is intensely galling about Chomsky is that eventually he adopts many
of the arguments of these critics (e.g. the abandonment of deep structure
in the minimalist program) without acknowledging any of the arguments of
these earlier scholars, many of whom were his disciples.  In the words of 
Geoffrey Pullum: 

    "Taking [the minimalist] view means abandoning the cherished level of
    deep structure (known as ‘d-structure’ in the last two decades)....
    But the names of linguists like Postal, Ross and McCawley, who in the
    late 1960s tried to argue for the elimination of deep structure, are
    completely absent from Chomsky’s bibliography. There is no belated nod
    in the direction of the literature he resolutely resisted for 25 years
    [1967 to 1992]... but whose central thesis he now adopts.  Nor is
    there any real effort to supply intelligible reasons and arguments for
    his abrupt conversion to the tenets of generative semantics. "
    	  - [Pullum 1996, p. 138]

It is unfortunate to find his arrogance undiminished by this host of
criticism. 

Excerpts

how does [language] relate to other aspects of the world?

One form of the question (what's roughly called the quesion of materialism
or physicalism or the mind-body problem or whatever): how can the properties
of the language faculty be realized in the physical world? The second form
they take is a question which is usually called the question of
representation or intentionality ('aboutness'): the question of how
expressions represent reality, how words refer to things.  This is the
second aspect of the question of the relation between language and the
world.

Now, in my opinion, both these questions are radically misconceived, and
have been for a long time. - p.10-11

there is a ton of empirical evidence to support the opposite conclusion on
every single point that I mentioned.  Furthermore, the core assumption of
highly productive recent work -and its pretty impressive achievements- is
that everything I said is wrong; that is, languages are highly imperfect in
all these respects, as indeed you would expect- they have indices and bar
levels, D- structures, S-structures and all kinds of relations, and so on and
so forth. Nevertheless, I think the contrary could well be true. p. 23

The issue of innateness of language is a curious one.  There is a huge
literature arguing against it; there's nothing defending the thesis.

... the most elementary property of the language faculty is the property of
discrete infinity; you have six-word sentences, seven-word sentences, but
you don't have six-and-a-half-word sentences... This property is virually
unknown in the biological world.  The only other one is the arithmetical
capacity, which could well be some offshoot of the language faculty.
p.50-52

[AM: Is it really discrete? what about morphemes that attach to words?, or
multi-word units? ]

Q: What are the latest trends in semantics? Is it likely to develop into a
   science some day with its own units?

Chomsky: That is a really interesting question. ... We have to ask what
semantics is. If semantics is what is meant by the tradition (say, Peirce or
Frege or somebody like that), that is, if semantics is the relation between
sound and thing, it may not exist.

If semantics is the study of relations like agency, thematization, tense,
event-structures and the place of arguments in them and so on that is a rich
subject but that is syntax; that is, it is all part of mental
representations.  It goes on independently of whether there is a world at
all just like the study of phonological representations.  This is
mislabelled 'semantics'.  [...]

Most of what's called 'semantics' is, in my
opinion, syntax.  It is the part of syntax that is presumably close to the
interface system that involves the use of language.  So there is that part
of syntax and there certainly is pragmatics in some general sense of what
you do with words and so on.  But whether there is semantics in the more
technical sense is an open question.  I don't think there is any reason to
believe that there is.

I think it goes back to the old and probably false assumption that there
is a relation between words and things independently of circumstances of
use. p.73

FOOTNOTE: People use words to refer to things in complex ways, reflecting
interests and circumstances, but words do not refer; there is no
word-thing relation of the Fregean variety, nor a more complex
word-thing-person relation of the kind proposed by Charles Sanders Pierce in
equally classic work on the foundation of semantics.  These approaches may
be quite appropriate for the study of invented symbolic systems (for which
they were initially designed at least in the case of Frege).  But they do
not seem to provide the appropriate concepts for the study of natural
language. (Powers and Prospects, Madhyam Books New Delhi 1996, p. 22-3)

Q: By virtue of knowing the concept 'climb', does the child know that the
concept needs an agent and a theme for its realization? Does the child learn
that the concept of 'die' is alternatively realized in English as 'die' and
'kick the bucket'? The innate conceptual and computational components are
presumably different modules; does linguistic experience trigger some kind of
interaction between them with the result that a predicate-argument structure
is generated which is then converted into familiar lexically-filled syntactic
representation?

Chomsky: These questions may be referring to a book of mine of about ten
years ago in which I said that the child has a repertoire of concepts as part
of its biological endowment and simply has to learn that a particular concept
is realized in a particular way in the language. So the child has a concept,
say, 'climb' in some abstract sense with all its weird properties and has to
learn that it is pronounced 'climb', not some other thing. Jerry Fodor's
important work for many years is relevant here, along with Ray Jackendoff's
and much else. These are all perfectly reasonable questions. You can have
various ideas about them; there isn't a lot of understanding. I could tell
you what my own suspicion is about these questions but they are research
topics.

There is overwhelming reason to believe that concepts like climb, chase,
run, tree and book and so on are fundamentally fixed.  They have
extremely complex properties ... which means that they've got to basically be
there and then they get kind of triggered and you find out what sounds are
associated with them.

You can read some of the Q&A session here.

Other reviews

Review by Adriano Palma, Institut Jean Nicod, Paris & Tsh UTC Compiegne

The transcript of the oral presentation covers the first 40 pages of the
published text. It is followed by extensive discussions. Those, as detailed
by the editors in their preface, have been the outcome of an intense
cooperative enterprise between all parties in the conversations. Questions
were both posed orally and in written form. Chomsky replied in both media
(orally in Delhi and in writing from Cambridge, Mass.)

There is the usually unavoidable amount of repetition and unclear
statements. Mostly those are due to the difficulty of the subjects. As it
well known to those who read him, Chomsky's view is that we do have indeed
two sort of intellectual abilities, or faculties. One is roughly coinciding
with common sense, and one is a science forming faculty (far more difficult
to characterize in simple terms) which we can indeed apply, though it
requires training, financial and cognitive resources devoted to it and so
forth.

The minimalists bring in a very novel idea. It may be possible to see that
the constraints (the "rules" that generate traditional grammar rules for
verbs in German, e.g.)  aren't rules at all. They are "taxonomic artefacts"
(p. 14). What is there are sets of parameters that once fixed, against the
background of purely general principles, generate linguistic expressions. The
language organ interacts (or "interfaces") with sensory-motor systems and
with a conceptual-intentional system. I use the plural for the sensory motor
system since (see p. 9) it is empirically known from the existence of sign
languages that systems other than the sound production can access the
language faculty. The conceptual-intentional system is utterly mysterious in
the simple sense that not much is understood about it. In a slogan, it is
where language gets used to talk about something or other. Chomsky is, by the
way, extremely skeptical about the view that linguistic expressions as such
have intentionality in the philosophers' sense of "aboutness".

The point is "that you now, for the first time ever, have some coherent idea
of what a language might be. " (p.15) The minimalist program comes along and
asks new questions. Two questions, among others, how much of what we
attribute to language is only due to the techniques we adopt and how much is
really motivated by empirical evidence and how good is language as a solution
to boundary conditions imposed by the architecture it is in. The second one
allows an answer: perfection or near perfection. Language may be a perfect,
near-perfect, solution to an engineering problem, namely the problem of
providing something legible at the interface. The question and its possible
answer are daring, if for no other reason than its strangeness. Very little
in nature is perfect in this sense. Evolution, the gods, or your preferred
"engine of creation" appear nearly always to be taking bits and pieces in a
junkyard and come up with something that more or less does the job. If
language is perfect, or even almost perfect in this sense, it would be weird,
very strange indeed. It may come close to the sheer oddness of the fact that
nature likes to write letters following strict mathematical rules. It was
remarked centuries ago by Galileo, and rediscovered constantly in the most
unexpected locations: in one of the replies Chomsky makes the same point by
citing the known fact that Fibonacci series show up all over the place (see
p. 49)

The program is not a nice fellow. It is a program with an attitude. One would
have to show that there are no linguistic levels apart from the
phonetic/articulatory and the semantic ones. The only constraints operative
are the ability to use expressions at the interface: "...  there shouldn't be
any other levels because other levels are not motivated by legibility
conditions." (p. 21) All other devices (surface and deep structures, etc.)
have got to go, they're technical jargon that covers up lack of
understanding. Second thing to go by the board is lexical peculiarities. A
lexical item, a collection of properties, called features, contains no
features other than those that are interpreted at the interfaces: "... [we]
have to show that when we abandon X-bar theory, indices, and other such
devices, we find solutions which are not only as good but even better ones."
(p. 22).  Third no structural relations other than those forced by
legibility, hence no adjacency, theta-structure, scope at the level of
logical form. For the more technically inclined only local relations are
kosher in minimalism, "perhaps nothing else. That means there is no
government, proper government, no Binding theory internal to language, and no
interactions of other kinds. To the extent that language is perfect, all of
this has to go."

What is more interesting in my opinion is the clarity with which certain
issues of general interest are presented. It is often and widely thought that
it is a trait of rational inquiry to be sensitive to evidence. The job of a
theory, we used to be taught in school, is to save phenomena. Chomsky takes
exactly the opposite tack.  Consider the following quotation, from an
interview by Adriana Belletti and Luigi Rizzi with Chomsky, in 1999,
available on WWW:

   The phrase [Galilean style] was used by nuclear physicist Steven Weinberg,
   borrowed from Husserl, but not just with regard to the attempt to improve
   theories. He was referring to the fact that physicists "give a higher
   degree of reality" to the mathematical models of the universe that they
   construct than to "the ordinary world of sensation." [4] What was striking
   about Galileo, and was considered very offensive at that time, was that he
   dismissed a lot of data; he was willing to say "Look, if the data refute
   the theory, the data are probably wrong.

In the same vein, Chomsky says here: "[those familiar with technical
literature] are aware that there is a ton of empirical evidence to support
the opposite conclusion on every single point that I mentioned.  Furthermore,
the core assumption of highly productive recent work -and its pretty
impressive achievements- is that everything I said is wrong; that is,
languages are highly imperfect in all these respects, as indeed you would
expect- they have indices and bar levels, D- structures, S-structures and all
kinds of relations, and so on and so forth. Nevertheless, I think the
contrary could well be true." (p. 23)

The false prophet's promise of apples: Ian Robinson

 / Times Higher Education

In 1975, I published The New Grammarians' Funeral, a book assessing the then
state of the Chomskyan school of grammar, known collectively as
transformational-generative grammar. I warned that Chomsky wanted to reduce
grammar to a genetic mechanism for acquiring language, the Language
Acquisition Device, which he says children are born with. However, I judged
transformational-generative grammar itself, with its emphasis on the syntax
of the sentence, an elegant and then timely restatement of traditional
grammar. I thought at that point I was being complimentary.

He calls the more recent phase of his work on grammar "the minimalist
programme", but the claims he makes for it are not minimal. In On Nature and
Language (2002), he states: "The way of looking at things was totally
different from anything that had come before. In fact, I think it is fair to
say that more has been learnt about language in the past 20 years than the
preceding 2,000 years."

In the The Minimalist Program (1995), Chomsky says of his work that it
"constituted a radical break from the rich tradition of thousands of years of
linguistic inquiry, far more so than early generative grammar, which could be
seen... as a revival of traditional concerns and approaches to them". He adds
that his "principles and parameters" approach (which posits the idea that the
grammatical principles underlying languages are innate and fixed) "maintains
that the basic ideas of the tradition, incorporated without great change in
early generative grammar, are misguided in principle - in particular, the
idea that a language consists of rules for forming grammatical constructions
(relative clauses, passives and so on)".

What I took to be a wrong direction in his earlier work, the pursuit of the
Language Acquisition Device, has become the exclusive concern of his later
work; for now the Language Acquisition Device and "principles" are both
indistinguishable from universal grammar, Chomsky's later preoccupation. In
Architecture, Chomsky first declares that universal grammar and the Language
Acquisition Device are one and the same thing; then he says in answer to the
question "What is the nature of the language acquisition device?", "Well,
whatever the nature of language is, that is what it is. According to one
model, presumably oversimplified, if we understand the principles and the
parameters of language, we will know what the language acquisition device
is."

This Language Acquisition Device is the genetically determined universal
grammar developed in the brain of every normal child. "The child has,"
Chomsky writes in Architecture, "a repertoire of concepts as part of its
biological endowment and simply has to learn that a particular concept is
realised in a particular way in the language. There is overwhelming reason to
believe that concepts like, say, climb, chase, run, tree and book and so on
are fundamentally fixed." The tree concept will be found even in the brains
of children who live in places where there are no trees, and the book concept
where there is no written language. There must surely then be a universal
concept of television, and liberalism and feudalism.

Chomsky continues: "It turns out that the concepts are very complex, which
means that they've got to basically be there and then they get kind of
triggered and you find out what sounds are associated with them." The
concepts in universal grammar are available to all languages, but by "slight
variation", different languages allot different sounds to them.

My first objection to this programme remains what it was in 1975. There is no
biological evidence for the existence of the Language Acquisition
Device. Chomsky thinks it must exist because the acquisition of language from
a child's disconnected and incomplete experiences of language cannot possibly
be explained without it, but that is hardly a scientific demonstration.

He says in Architecture : "To say that 'language is not innate' is to say
that there is no difference between my granddaughter, a rock and a rabbit. In
other words, if you take a rock, a rabbit and my granddaughter and put them
in a community where people are talking English, they'll all learn
English. If people believe that, they believe that language is not innate."

Well, unlike rabbits, we have the capacity to acquire language, but where is
the evidence that that capacity is a ready-made universal grammar?

My second objection - so elementary that I am embarrassed to have to point it
out - is that even if Chomsky's neurology is right, the Language Acquisition
Device could still only be understood by way of grammar, and not vice
versa. Can a bodily organ be meaningfully called principles? Drop all mention
of the Language Acquisition Device and nothing in grammar changes. The
concentration on the Language Acquisition Device, however, has become so
exclusive in Architecture as to replace grammar altogether; which from the
point of view of the grammarian is a reductio ad absurdum. ...

The Language Acquisition Device has led Chomsky to become a sort of
mind-brain-identity prophet (he habitually writes of mind-brain as if the
terms are interchangeable) and to fall into the prevalent fallacy that if we
study brains intently enough we shall somehow understand the mind. When
Chomsky began, there were linguists who thought that by staring
scientifically enough at the sounds of language we might get at last to their
sense. He seems now to be falling himself into a fallacy of the same order.

I think he would have done better to stick to grammar.

Ian Robinson is author of The New Grammarians' Funeral, published by
Cambridge University Press.

Neil Smith: Never lost for words

Times Higher Education

Noam Chomsky has shown that there is really only one human language: that the
immense complexity of the innumerable languages we hear around us must be
variations on a single theme. He has revolutionised linguistics and in so
doing has set a cat among the philosophical pigeons.

He has resurrected the theory of innate ideas, demonstrating that a
substantial part of our knowledge is genetically determined; he has
reinstated rationalist ideas that go back centuries but thathad fallen into
disrepute; and he has provided evidence that "unconscious knowledge" is what
underlies our ability to speak and understand. He has overturned the dominant
school of behaviourism in psychology and has returned the mind to its
position of pre-eminence in the study of humankind. In short, Chomsky has
changed the way we think of ourselves, gaining a position in the history of
ideas on a par with that of Darwin or Descartes. And he has done this while
devoting most of his time to dissident politics and activism...

Chomsky's arguments for the partial innateness of the language faculty are
controversial and have dominated philosophical discussion of his work. Some
of the controversy is resolved in The Minimalist Program (1995), which
suggests that what is both innate and strictly linguistic is extremely
limited, and is imposed by the need for "legibility". The products of the
language faculty, representations of the sounds and meanings of sentences,
derive many of their properties from the interfaces with which they have to
interact: systems of perception and production on the one hand; systems of
thought and memory on the other. A crucial goal is then to determine what is
unique to language and unique to humans. Both strands are important: there
has been much recent discussion of the FOXP2 gene, disruption to which causes
serious linguistic problems. But we share this gene with mice, so it is not
unique to humans. Similarly, the infinite expressive power of language is
something shared with mathematics and music, so "infinity" cannot be unique
to the human language faculty. What is peculiar to human language may,
perhaps, be limited to recursion: the ability to embed one sentence in
another. But this property plausibly derives from whatever system we use to
think with - the language of thought. If so, the only thing unique to human
language would be the form of the mapping that links linguistic expressions
to the interfaces. Chomsky has suggested that language is "perfect": the
mapping is the simplest that is logically possible given the nature of the
requirement of legibility. Determining what is innate is still an important
preoccupation, but the status of what is innate, linguistic or more generally
cognitive, changes as we extend the investigation of the human mind.

Second, it is relevant to stress his role in fomenting the diversity of
modern linguistics. The field is vibrant, there are countless approaches,
rivalries, theories and anti-theories. The monolith of the Sixties (if it
ever existed) has crumbled. Many linguists owe their preoccupations to
Chomsky's work, even when they distance themselves from him. Others owe
little to his work except as a focus for their opposition - bemused,
contemptuous or indifferent. Even here the visibility of linguistics and its
centrality to the study of cognition are part of Chomsky's legacy.

Neil Smith is a professor in the department of phonetics and linguistics at
University College London, author of Chomsky: Ideas and Ideals (Cambridge
University Press, £17.99) and has contributed a chapter to the recently
published Encyclopedia of Language and Linguistics (Elsevier).

---
blurb:
In this book Noam Chomsky reflects on the history of 'generative
enterprise' - his approach to the study of languages that revolutionized our
understanding of human languages and other cognitive systems. In his lively
and engaging style, he presents advances in current grammatical theorycalled
'Minimalist Program', sketches some of the key issues that have characterized
generative grammar in recent years, and charts out the agenda for future
research in language theory. Linguists interested in the internal history of
generative linguistics will find this book insightful as alsostudents andIn
The Architecture of Language (2001), he states: "What there is, it seems,
is just general principles, which are properties of the language faculty as
such and slight options of variation, which are called 'parameters'."
general readers who wish to gain an introductory knowledge of the discipline,
its significance, and Chomsky's contribution.