NOAM NOAM CHOMSKY

PHILOSOPHY $21.95 U . S .

“ T H E C L A R I T Y O F P R E S E N TAT I O N CHOMSKY
AT T I M E S A P P R O A C H E S T H AT

ON
OF BERTRAND RUSSELL IN HIS
POLITICAL AND MORE POPULAR
PHILOSOPHICAL ESSAYS.”
—Contemporary Psychology

LANGUAGE
“LANGUAGE AND RESPONSIBILITY IS A WELL-ORGANIZED,
CLEARLY WRITTEN, AND COMPREHENSIVE
INTRODUCTION TO CHOMSKY’S THOUGHT.”

ON
—The New York Times Book Review
“REFLECTIONS ON LANGUAGE IS PROFOUNDLY SATISFYING
AND IMPRESSIVE. IT IS THE CLEAREST AND MOST
DEVELOPED ACCOUNT OF [CHOMSKY’S] CASE.”
—P AT R I C K F L A N A G A N

FEATURING TWO OF NOAM CHOMKSY’S

most popular and enduring books in one
omnibus volume, On Language contains some
of the noted linguist and political critic’s most
informal and accessible work to date, making
it an ideal introduction to his thought.
In Part I, Language and Responsibility (1979),
Chomsky presents a fascinating self-portrait of
This volume is one of several classic works
by Chomsky reissued by The New Press. The
others are American Power and the New
Mandarins, For Reasons of State, Problems of
Knowledge and Freedom, Objectivity and Liberal
Scholarship, and Towards a New Cold War.

NOAM CHOMKSY is Institute Professor

LANGUAGE
his political, moral, and linguistic thinking emeritus at the Massachusetts Institute of
through a series of interviews with French lin- Technology, a world-renowned linguist and
guist Mitsou Ronat. In Part II, Reflections on political activist, and the author of numerous
Language (1975), he explores the more general books, including 9-11, Hegemony or Survival,
implications of the study of language and offers Failed States, and the collection Understanding
incisive analyses of the controversies among Power (The New Press).
psychologists, philosophers, and linguists over
fundamental questions of language.

www.thenewpress.com
CHOMSKY’S CLASSIC WORKS
THE NEW PRESS
Cover photo by istock
Design by Pollen, New York THE NEW PRESS
LANGUAGE AND RESPONSIBILITY AND REFLECTIONS ON LANGUAGE
ON LANGUAGE
Chomsky’s Classic Works
Language and Responsibility and
Reflections on Language

NOAM CHOMSKY

THE NEW PRESS

NEW YORK
LONDON
Language and Responsibility © 1977 by Flammarion
English translation and revisions of Language and Responsibility © 1979 by Noam
Chomsky. Reflections on Language © 1975 by Noam Chomsky. All rights reserved.

No part of this book may be reproduced, in any form, without written permission
from the publisher.

Requests for permission to reproduce selections from this book should be mailed to:
Permissions Department, The New Press, 38 Greene Street, New York, NY 10013.

Published in the United States by The New Press, New York, 2007
Distributed by Perseus Distribution

ISBN 978-1-56584-475-9 (pbk)

CIP data available

The New Press was established in 1990 as a not-for-profit alternative to the large,
commercial publishing houses currently dominating the book publishing industry.
The New Press operates in the public interest rather than for private gain, and is
committed to publishing, in innovative ways, works of educational, cultural, and
community value that are often deemed insufficiently profitable.

www.thenewpress.com

Printed in the United States of America

10 9 8 7 65 4 3
 Contents

Language and Responsibility (1979)

Contents
Translator’s Note vii
Introductory Comment vii

PartI. Linguistics and Politics 1

1. Politics 3
2. Linguistics and the Human Sciences 43
3. A Philosophy of Language? 63
4. Empiricism and Rationalism 81

Part II. Generative Grammar 101

5. The Birth of Generative Grammar 103
6. Semantics 136
7. The Extended Standard Theory 163
8. Deep Structure 169
9. Universal Grammar and Unresolved Questions 180

Notes 195
Index 199

Reflections on Language (1975)

Preface vii

Part I. The Whidden Lectures 1

_ 1. On Cognitive Capacity 3
2. The Object of Inquiry 36
3. Some General Features of Language’ 78

Part II. 135

4. Problems and Mysteries in the Study of Human Language

Notes 229
Bibliography 255
Index of Names 267
Language and Responsibility
 Politics al

hensible methods in his political struggles, but because he made

a mistake in the choice of adversaries against whom he turned
these methods. He attacked people with power.
The telephone taps? Such practices have existed for a long
time. He had an “enemies list’? But nothing happened to those
who were on that list. I was on that list, nothing happened to
me. No, he simply made a mistake in his choice of enemies: he
had on his list the chairman of IBM, senior government advis-
ers, distinguished pundits of the press, highly placed supporters
of the Democratic Party. He attacked the Washington Post, a
major capitalist enterprise. And these powerful people defended
themselves at once, as would be expected. Watergate? Men of
power against men of power.
Similar crimes, and others much graver, could have been
charged against other people as well as Nixon. But those crimes
were typically directed against minorities or against movements
of social change, and few ever protested. The ideological cen-
sorship kept these matters from the public eye during the Wa-
tergate period, although remarkable documentation concerning
this repression appeared at just this time. It was only when the
dust of Watergate had settled that the press and the political
commentators turned toward some of the real and profound
cases of abuse of state power—still without recognizing or ex-
ploring the gravity of the issue.
For example, the Church Committee has published informa-
tion, the significance of which has not really been made clear.
At the time of its revelations, a great deal of publicity was
focused on the Martin Luther King affair, but still more impor-
tant revelations have hardly been dealt with by the press to this
day (January 1976). For example, the following: In Chicago
there was a street gang called the Blackstone Rangers, which
operated in the ghetto. The Black Panthers were in contact with
them, attempting to politicize them, it appears. As long as the
Rangers remained a ghetto street gang—a criminal gang, as
depicted by the FBI, at least—the FBI were not much con-
cerned; this was also a way of controlling the ghetto. But radi-
 Linguistics and the Human Sciences 61

marians term the specified subject condition, * because the phe-

nomena which are excluded by this condition have no “human
interest.”
For example, the sentence I mentioned earlier, John seems to
the men to like each other, is excluded by the specified subject
condition. But I doubt that any traditional grammar, even the
most comprehensive one, would trouble to note that such sen-
tences must be excluded. And that is quite legitimate, as far as
traditional grammars of English are concerned; these gram-
mars appeal to the intelligence of the reader instead of seeking
explicitly to characterize this “intelligence.” One can suppose
that the specified subject condition—or any other principle
which excludes this phrase—is simply an aspect of the intelli-
gence of the speaker, an aspect of universal grammar; conse-
quently, it does not require explicit instruction to the person
who reads a traditional grammar.
For the linguist, the opposite is true. The linguist is interested
in what the traditional grammars don’t say; he is interested in
the principles—or at least that is what should interest him, in
my Opinion.
M.R.: The typical reaction one encounters in the human
sciences, against idealization, thus seems linked to the fact that
people are not interested in what they have in common, but...
N.C.: ... but in what differentiates them, yes. And in their
normal human lives, this is the right decision. The same thing
must be true of frogs. No doubt, they would not be interested
*This is a condition which forbids both the extraction of an element belonging to an
embedded phrase, and also its association with an element that is outside this phrase,
if the embedded phrase contains a “specified” subject—“specified” in a meaning of the
term which must be defined precisely. For example, in the sentence, We expected John
to like each other, the phrase each other cannot be associated with the antecedent
we, so that the sentence does not express the meaning, “Each of us expected that John
would like the other.” The specified subject condition prevents this association, because
of the presence of the subject John in the embedded clause, John to like each other. The
same condition operates in the example given above: John seems to the men to like each
other. Here the subject of /ike is not phonetically present, but is “understood” to be
John. For a discussion of these questions, see Chomsky, Reflections on Language (New
York: Pantheon, 1975), chapter 3.
 The Birth of Generative Grammar 107

I take for granted that in something as complex as the actual

utilization of language and the judgments about language,
many systems enter into interaction. No matter how careful our
observations, how objective our methods, and how replicable
our “experiments,” the facts presented are, in my opinion, of
little interest in themselves. What is of interest is their bearing
on explanatory theories that seek to formulate the fundamental
principles of the language faculty. Speaking just for myself,
organizing the “facts of language” does not interest me very
much. The notion “facts of language” has little sense outside
of at least an implicit theory of language. One can perfectly well
have different interests; I am simply trying to make clear what
interests me. Frankly, I do not believe that seeking to account
for “all the facts” constitutes a reasonable goal. In contrast,
what seems important to me is the discovery of facts that are
crucial for determining underlying structure and abstract hid-
den principles. If such principles do not exist, the enterprise is
not worth undertaking. If they do exist, then facts are interest-
ing (to me, at least) insofar as they bear on the truth of these
principles. The discovery of such facts is often a creative
achievement in itself, and very much “theory-related.” “The
facts,” in any interesting sense of that notion, are not simply
presented to us, nor is it of great interest, in my opinion, to
present “the facts” in an exact manner, although of course the
pertinent facts (again a notion that is linked to theory) must be
presented in as precise a manner as possible ...
M.R.: ... as in physics.
/N.C.: ... If you like. That is how it seems to me. At each
stage in the development of physics there have been innumera-
ble unexplained “‘facts,” or facts that seemed totally incompati-
ble with the theories being actively pursued. To take a classic
example, consider the “facts” of sorcery or of astrology, which
seemed very well established by the standards of empirical
research in the period when classical Galilean physics became
established scientific doctrine. Or to take a less exotic example,
consider the problems encountered by seventeenth-century
162 LANGUAGE AND RESPONSIBILITY

able generality. For reasons of principle rooted in the general

theory of grammar, there are two quite different constructions
called “‘passive,” probably more. /
I do not see any reason to suppose that there is a universal
rule covering these distinct kinds of construction. To postulate
a universal “rule of passive” would tend to obscure all these
differences and also the general principles involved in their
explication.
M.R.: French seems to be an intermediate case ...
N.C.: French is an interesting case. I think the question
requires more extensive study. For the moment it remains open.
 Deep Structure 175

see the possibility of a new hermeneutics here. This is actually

the same mistake ...
N.C.: Yes, in terms of things that are “hidden,” that are to
be discovered; but there are many aspects of phonology that are
“deep” in this sense of the word.
M.R.: The word surface* is equally misleading . .
N.C.: Surface structure is something quite abstract, involv-
ing properties that do not appear in the physical form . .. It is
by virtue of such properties that language is worth studying.
M.R.: Personally I find quite fascinating an abstraction or
depth that is linked to trace theory—that is, the astonishing
study of the structuring effect of silence: in phonology, the trace
changes the intonation; in semantics, it blocks co-reference . . .
N.C.: That is a really interesting property of surface struc-
tures.
M.R.: In poetry too, in metrics, these structuring silences
are essential ... But don’t you think that by placing so much
stress on surface structure, you invite being accused of return-
ing to structuralism? In general, whenever you have refined a
concept, you have been accused of abandoning your fundamen-
tal hypotheses. I remember reading that your definition of de-
grees of grammaticalness signifies that you have abandoned the
concept of grammaticalness!
N.C.: Well, in fact, the notion of “degree of grammatical-
ness” was developed at the same time as the notion of “gram-
maticalness,” within the theory of generative grammar, that is,
in the early 1950s. A chapter of LSLT is devoted to this ques-
tion, and I also refer to it in Syntactic Structures. But what is
more important is that the kind of criticism you’re referring to
reveals once again the difference between the attitude of the
natural sciences on the one hand, and one often found in the
social sciences and “humanities” on the other. The latter, which
lack the intellectual content of the natural sciences, are to a
great degree involved with personalities rather than ideas. In
science it is self-evident that concepts are going to change; that
**Superficiel” in French, thus even more so—Translator.
 Deep Structure 179

N.C.: Furthermore, if some field is still at the level where

procedural methods can be applied, then it is at a very primitive
level indeed. A purely descriptive level, say, like Babylonian
astronomy, or not even that. There are no “methods” in this
sense in a field having real intellectual content. The goal is
to find the truth. How to do that, nobody knows. There are
no procedures that can be outlined in advance for discover-
ing scientific truth. You cannot train a creative physicist or
biologist by telling him: “Here are the methods, apply them to
a new organism.” That may be the way to train a lab technician,
but not a scientist. You do that when you don’t know how to
find meaningful work for students. It is an admission of failure.
What you expect of a scientist is to discover new principles,
new theories, even new modes of verification ... That won’t
happen by learning a fixed procedure. The same is true of
linguistics today. It is impossible to explain to someone the
procedure he must apply in order to find the generative gram-
mar of some language. What one looks forward to is the discov-
ery of new phenomena that will show that the theories which
have been proposed are false, that they must be changed—new
questions that no one has thought of posing before, at least in
a clear manner, new contributions to understanding, achieved
perhaps with new “methods.” And finally, new ideas and new
principles, which will reveal how limited, false, and superficial
are the assumptions that we hold to be valid today.
M.R.: At times that may disturb students accustomed to
traditional instruction, where it is sufficient to learn passively
what you are taught. In generative grammar, in effect, instruc-
tion consists in the explication of fundamental concepts, and in
the presentation of the history of the discipline in terms of a
detailed account of different hypotheses. One can explain the
manner in which something is demonstrated, how such a dem-
onstration is constructed. But one cannot tell anyone how to
find a new idea, how to invent. Invention is linked to the desire
to understand one’s chosen field better.
200 INDEX

behavior, 142; empirical approach, censorship, 20, 21, 24, 26, 30-31; na-
68-69; technological approaches, tional patterns, 31; Solzhenitsyn
128; verbal, 129 on, 32. See also ideological control
behaviorism and behavioral science, Central Intelligence Agency (CIA),
84, 128, 129; Chomsky’s critique 29 .
of, 46-47, 49, 113-14 Chicago, 25, 26, 31, 158; FBI activity
beliefs: semantic role of, 142~3, 144, in, 21-24, 25
147-8, 152-3, 189 Chile: U.S. intervention in, 34, 35, 40
Bernstein, Basil, 56 China, 15, 17
“Best Theory, The” (Postal), 150 Chomsky, Noam: Aspects of the The-
Bever, Thomas, 53, 151, 154 ory of Syntax, 136, 150, 151, 163,
biological constraints: on cognitive 169; Cartesian Linguistics, 71-
systems, 84, 94, 98-99; on knowl- 78; Counterrevolutionary Violence:
edge, 63-68; on language, 98; Bloodbaths in Fact and Propa-
theory construction role of, 64— ganda, 20, 37-38; Current Issues
69, 76-77. See also universal gram- in Linguistic Theory, 134, 182;
mar Foucault and, 74-80; Language
biology of language, 133 and Mind, 49, lecture to Nieman
Black English, 53-56 Fellows, 30-31; letter to New York
Black Panthers, 21-24 Times on Vietnam War, 37-38; as
Blackstone Rangers, 21-24 linguistics teacher, 134-5; link be-
Bloch, Bernard, 118, 130 tween linguistic and political ac-
Bloomfield, Leonard: Menomini tivities, 3-8; Logical Structure of
Morphophonemics, 112 Linguistic Theory (LSLT), 106,
body, 81, 83, 84; empirical view, 108-11, 113-14, 122-5, 126,
81-82 131-2, 138, 139, 140, 151n, 170,
Bolshevism, 74, 90 175, 182, 183; “Morphophonemics
Bracken, Harry, 92-93, 94 of Modern Hebrew,” 111-12, 130;
brain: empirical view, 81-82; objections to empiricism, 81-83;
genetic code and, 84. See also cog- philosophical influences on, 70-
nitive structures and systems 74, 132; political writings, 30;
Brazil, 40 “Questions on Form and Interpre-
Bresnan, Joan, 172 tation,” 156-7; Reflections on
Grammar, 61n; Reflections on
Language, 61n, 63, 92, 160, 172;
California: FBI activities, 25-26 “Remind,” 154; Syntactic Struc-
Cambodia, 24 tures, 110, 113, 127, 131, 133, 136,
Canada, 7 138, 139, 140, 144, 145, 151n, 175;
capitalism, 9, 12-13, 70; work on history of ideas, 77. See
role of intelligentsia in, 90-91 also generative grammar
Cartesian Linguistics (Chomsky), Church Committee, 21, 24, 26
77-78 Citizens’ Commission to Investigate
Cartesianism, 176; dualism, 92-93, the FBI, 22
96-97; innésism, 94; mechanics, civil rights movement, 14, 27
96; soul concept, 96-97. See also Clark, Mark, 23
Descartes, René class struggle, 80
case grammars, 154-6, 157 Clemens, Diane, 15
210 INDEX

semantic interpretation, 169, 170-1, social and political analysis, 4-5; cre-
173; role of surface structure, 163. dentials for, 6-7
See also semantics and semantic social change, 70, 74, 80
theory social control, 90. See also ideologi-
semantic representation, 141, 171; cal control
and deep structure, 148, 150-2; vs. social interaction: universal gram-
logical form, 145; and surface mar of, 69-70
structure, 150-2, 171, 173; and social sciences, 6, 12; accessibility of,
syntactic structures, 141; T- 4-5; conceptual change in, 175-6,
marker, 170; and truth conditions, 177-8; human nature concepts,
144-8. See also semantics and se- 70-76. See.also anthropology; hu-
mantic theory man sciences; psychology; sociol-
semantics and semantic theory, 106, ogy
136-62, 164, 175, 184; co-refer- Socialist Workers party, 27
ence principles, 146; and deep socialists and socialism: in American
structure, 172; definitions of, 140; mass media, 9; intelligentsia and,
90. See also Marxism
extensional vs. intensional, 144-5;
Fodor-Katz hypothesis, 140-3;
sociolinguistics, 53-56, 57; questions
about, 190-4; view of generative
generative, 148-54, 164, 171;
grammar, 191-2. See also linguis-
grammar role of, 137-40; integra-
tics; sociology
tion of theory of, 154~6; interpre- sociology, 192; Chomsky’s view, 56—
tive, 145; linguistic theory role of,
59; compared to natural sciences,
139; model-theoretic, 72; of natu-
$8~-59; linguistics and, 53-59;
ral languages, 73; pre-Standard methods, 58; resistance to idealiza-
Theory, 137-8; projection rules, tion, 57-58
144; and systems of belief, 142-3, Solzhenitsyn, Alexander, 31-34
144, 147-8; universal, 140-3, 144. soul, 96-97
See also meaning; semantic repre- Soviet Union, 9, 32; American Com-
sentation munists and, 32, 34. See also
semiology, 71 Russia
sentences and sentence grammar, Spain, 11, 32
120, 147; devoid of meaning, 144; spatial intuition, 67
negation and quantification, 151; speech acts, 72
perceptual influences on, 44-45, Standard Theory, 135, 137, 141, 148,
84; relative clauses, 173-4; trans- 150-2, 163; compared to Fill-
formational history of, 170; truth more’s case grammar, 154; criti-
and, 144-8 cized, 173; deep structure in,
Shaw, Nate, 55 169-70, 173; symmetry of, 173.
simplicity and simplicity measure, See also Extended Standard The-
112, 113, 130 ory; generative grammar
Skinner, B. F., 46-47, 129 stimulus-response learning theory,
slavery, 91 126
Smith, Adam, 70 Stone Age, 55, 95
Smith, Gaddis, 19 structuralism, 76-77, 97, 130, 138,
SNCC (Student Non-violent Coordi- 148, 154, 156, 167; Chomsky’s al-
nating Committee), 14 leged return to, 175-7; contribu-
 Preface

Part I of this book is an elaboration of the Whidden Lec-

tures, delivered in January 1975 at McMaster University.
Part II is a revised version of my contribution to a volume
of essays in honor of Yehoshua Bar-Hillel (Kasher, ed., forth-
coming), submitted for publication in June 1974. The latter
essay considers some critical discussion of the general point
of view developed here, as it had been presented in earlier
work, To preserve the internal coherence of the discussion
in part II, I have retained some material that recapitulates
themes that are developed in a somewhat different form in
the Whidden Lectures.
I have presented much of this material in lectures at MIT
and elsewhere and am indebted to many students, col-
leagues, and friends for valuable comments and criticism.
The work reviewed in chapter 3 of part I, in particular,
incorporates suggestions and research to which many people
have contributed, as the. citations only partially serve to
indicate. Among others, Harry Bracken, Donald Hockney,
Ray Jackendoff, Justin Leiber, Julius Moravcsik, and Henry
Rosemont have made helpful comments on an earlier ver-
sion of this manuscript. I have also profited greatly from
lively and extensive discussions with members of the faculty
of McMaster University.

Noam Chomsky
Cambridge, Massachusetts
April 1975
34 REFLECTIONS ON LANGUAGE

Assuming still the legitimacy of the simplifying assump-

tion about instantaneous learning, the “innateness hy-
pothesis” will consist of several elements: principles for the
preliminary, pretheoretic analysis of data as experience,
which serves as input to LT(H,L); properties of UG,
which determine the character of what is learned; other
principles of a sort not discussed in the foregoing sketch.
We might, quite reasonably, formulate the theory of
language so as to reflect this way of looking at LT(H,L).
A theory is a system of principles expressed in terms of
certain concepts. The principles are alleged to be true of
the subject matter of the theory. A particular presentation
of a theory takes some of the concepts as primitive and
some of the principles as axioms. The choice of primitives
and axioms must meet the condition that all concepts are
defined in terms of the primitives and that all principles
derive from the axioms. We might choose to formulate
linguistic theory by taking its primitive concepts to be
those that enter into the preliminary analysis of data as
experience, with the axioms including those principles ex-
pressing relations between the primitive concepts that en-
ter into this preliminary analysis (thus, the primitive
notions are “epistemologically primitive”; they meet an
external empirical condition apart from sufficiency for
definition). The defined terms belong to UG, and the
principles of UG will be theorems of this theory. Lin-
guistic theory, so construed, is a theory of UG incorpo-
rated into LT(H,L) in the manner described.
The “innateness hypothesis,” then, can be formulated
as follows: Linguistic theory, the theory of UG, construed
in the manner just outlined, is an innate property of the
human mind. In principle, we should be able to account
for it in terms of human biology.
To the extent that our simplifying assumption about in-
stantaneous learning must be revised, along lines to which
38 REFLECTIONS ON LANGUAGE

for attaining it.? But the evidence available does not sup-
port conclusions that are blandly presented in the litera-
ture, without argument, as if they were somehow estab-
lished fact.* .
A physical organ, say the heart, may vary from one
person to the next in size or strength, but its basic struc-
ture and its function within human physiology are com-
mon to the species. Analogously, two individuals in the
same speech community may acquire grammars that dif-
fer somewhat in scale and subtlety.‘ What is more, the
products of the language faculty vary depending on trig-
gering experience, ranging over the class of possible human
languages (in principle). These variations in structure are
limited, no doubt sharply, by UG; and the functions of
language in human life are no doubt narrowly constrained
as well, though no one has as yet found a way to go much
beyond a descriptive taxonomy in dealing with this ques-
tion.®
Restricting ourselves now to humans, suppose that we
understand psychology to be the theory of mind, in the
sense outlined earlier. Thus psychology is that part of
human biology that is concerned at its deepest level with
the second-order capacity to construct cognitive structures
that enter into first-order capacities to act and to interpret
experience. Psychology has as its primary concern the fac-
ulties of mind involved in cognitive capacity. Each such
faculty of mind is represented as one of the LT(H,D)’s
of earlier discussion. These faculties enable a person to
attain intricate and uniform cognitive structures that are
vastly underdetermined by triggering experience, and that
need not relate to such experience in any simple way (say,
as generalizations, higher-order generalizations, etc.).
Rather, the relation of a cognitive structure to experience
may be as remote and intricate as the relation of a non-
trivial scientific theory to data; depending on the character
 The Object of Inquiry 41

mates (e.g., lateralization) play a fundamental role.®

We might expect that the procedures used to train apes
in forms of symbolic behavior will succeed as well for hu-
mans with severe damage to the neural structures involved
directly in language. There is some evidence that this is
true.” Efforts to induce symbolic behavior in other species
might illuminate the specific properties of human lan-
guage, just as the study of how birds fly might be ad-
vanced, in principle, by an investigation of how people
jump or fish swim. Some might argue that more is to be
expected in the latter case: after all, flying and jumping
are both forms of locomotion; both involve going up and
coming down; with diligent effort and special training
people can jump higher and farther. Perhaps some hope-
lessly confused observer might argue, on these grounds,
that the distinction between jumping and flying is arbi-
trary, a matter of degree; people can really fly, just like
birds, only less well. Analogous proposals in the case of
language seem to me to have no greater force or signifi-
cance,

Returning to human psychology, consider the question

how the language faculty fits into the system of cognitive
capacity. I have been assuming that UG suffices to deter-
mine particular grammars (where, again, a grammar is a
system of rules and principles that generates an infinite
class of sentences with their formal and semantic proper-
ties), But this might not be the case. It is a coherent and
perhaps correct proposal that the language faculty con-
structs a grammar only in conjunction with other faculties
of mind. If so, the language faculty itself provides only an
abstract framework, an idealization that does not suffice
to determine a grammar.
Suppose that there is no sharp delimitation between
those semantic properties that are “linguistic” and those
that form part of common-sense understanding, that is, the
cognitive system dealing with the nature of things named,
 The Object of Inquiry 43

ponent of mental structure. Rather, the position we are

now considering postulates that this faculty does exist,
with a physical realization yet to be discovered, and places
it within the system of mental faculties in a fixed way.
Some might regard this picture as overly complex, but
the idea that the system of cognitive structures must be
far more simple than the little finger does not have very
much to recommend it.
The place of the language faculty within cognitive ca-
pacity is a matter for discovery, not stipulation. The same
is true of the place of grammar within the system of ac-
quired cognitive structures. My own, quite tentative, be-
lief is that there is an autonomous system of formal
grammar, determined in principle * by the language fac-
ulty and its component UG. This formal grammar gener-
ates abstract structures that are associated with “logical
forms” (in a sense of this term to which I will return)
by further principles of grammar. But beyond this, it may
well be impossible to distinguish sharply between linguis-
tic and nonlinguistic components of knowledge and be-
lief. Thus an actual language may result only from the
interaction of several mental faculties, one being the fac-
ulty of language. There may be no concrete specimens of
which we can say, These are solely the product of the
language faculty; and no specific acts that result solely
from the exercise of linguistic functions.*
Questions of this nature face us no matter what corner
of language we try to investigate. There is no “sublan-
guage” so primitive as to escape these complexities, a fact
that comes as no surprise to someone who is persuaded
of the essential correctness of the rationalist framework
outlined earlier. For in accordance with this view, a gram-
mar is not a structure of higher-order concepts and prin-
ciples constructed from simpler elements by “abstraction”
or “generalization” or “induction.” Rather, it is a rich
structure of predetermined form, compatible with trigger-
46 REFLECTIONS ON LANGUAGE

tems that enter into the act of naming need not be open
(or even accessible) to the conscious mind. It is, again,
an empirical problem to determine the character of the
cognitive structures that are involved in the apparently
simple act of naming.
Names are not associated with objects in some arbitrary
manner. Nor does it seem very illuminating to regard them
as “cluster terms” in the Wittgensteinian sense.'’ Each
name belongs to a linguistic category that enters in a de-
terminate way into the system of grammar, and the ob-
jects named are placed in a cognitive structure of some
complexity. These structures remain operative as names
are “transferred” to new users.'* Noting that an entity is
named such-and-such, the hearer brings to bear a system
of linguistic structure to place the name, and a system of
conceptual relations and conditions, along with factual
beliefs, to place the thing named. To understand “nam-
ing,” we would have to understand these systems and the ©
faculties of mind through which they arise.
I mentioned the notion “essential properties,” referring
it, however, to the systems of language and common-sense
understanding. But it has sometimes been argued that
things have “essential properties” apart from such desig-
nation and categorization. Consider the sentences:
(1) Nixon won the 1968 election
(2) Nixon is an animate object 1°
Surely statement (1) is in no sense a necessary truth.
There is a possible state of the world, or a “possible
world,” in which it is untrue, namely, if Humphrey had
won. What about (2)? It is not true a priori; that is, we
might discover that the entity named “Nixon” is in fact
an automaton. But suppose that in fact Nixon is a human
being. Then, it might be argued, there is no possible world
in which (2) is false; the truth of (2) is a matter of “met-
aphysical necessity.” It is a necessary property of Nixon
54 REFLECTIONS ON LANGUAGE

the like. The existence of “institutional facts,” no less

objective, “presupposes the existence of certain human in-
stitutions.” It is an objective fact, say, that Mr. Smith mar-
ried Miss Jones, but “it is only given the institution of mar-
riage that certain forms of behavior constitute Mr. Smith’s
marrying Miss Jones.” Human institutions are “systems of
constitutive rules” of the form “X counts as Y in context
C.” Searle proposes that “speaking a language is perform-
ing acts according to constitutive rules” which determine
institutional facts. He argues further that institutional facts
cannot be explained in terms of brute facts, but only “in
terms of the constitutive rules which underlie them.”
In the view we have been considering, the statement of
“brute facts” takes place within (at least) a dual frame-
work, involving the interaction of the system of language
and the system of common-sense understanding. Likewise,
the statement of institutional facts presupposes a theory of
human institutions and a related linguistic system. I doubt
that the principles entering into the theory of human in-
stitutions that persons have developed (largely without
awareness) can be reduced simply to the form “X counts
as Y in context C,” as Searle suggests. An analysis of institu-
tional structure appears to require principles of much more
abstract nature. Abandoning empiricist bias, there is little
reason to shy away from this conclusion. Again, it is a mat-
ter for discovery, not stipulation; discovery, in this case, in
the course of investigation of still another faculty of mind
and its operation.
In general, cognitive structures of varied sorts are con-
structed as a person matures, interacting with grammar
and providing conditions for language use. An integrated
study of cognition should try to make these connections
precise, thus leading—we may speculate—to further innate
properties of mind.
Note again that there is no inconsistency between this
view and the thesis of autonomy of formal grammar, that
 Some General Features of Language 79

conventions [that determine] the meanings of the sentences

of a language” (Strawson, 1970), and more important, to
discover the principles of universal grammar (UG) that lie
beyond particular rules or conventions. Some believe that
“the general nature of such rules and conventions can be
ultimately understood only by reference to the concept of
communication-intention” (Strawson, 1970). For reasons al-
ready discussed, I do not believe that the claim has been
substantiated and doubt that it can be. It seems to me to
misconceive the general character of language use and to
ignore an intellectual element that cannot be eliminated
from any adequate account of it. But whatever the future
holds on that score, we can still turn with profit to the com-
mon project. What, then, can plausibly be said about the
rules that determine the formal and semantic properties of
the sentences of a language, that is, its grammar?
In the past few years, a number of approaches to the
question have been developed and fruitfully applied. I will
not be able to survey them here, or to give any compelling
reasons in support of those that seem to me most promising,
or even to deal with objections that have been raised to the
point of view I will present.’ In the present context, these
deficiencies are perhaps less serious than they might seem.
My primary purpose is to give some idea of the kinds of
principles and the degree of complexity of structure that it
seems plausible to assign to the language faculty as a
species-specific, genetically determined property. Alterna-
tive approaches, while differing in a number of respects,’
are comparable, I believe, in their implications concerning
the more general questions that I have in mind.
Let us begin by considering some implications of the
principle of structure-dependence. If it is correct, then the
rules of grammar apply to strings of words analyzed into
abstract phrases, that is, to structures that are called “phrase
markers” in the technical literature. For example, the sen-
tence (1) might be assigned a phrase marker giving the
164 REFLECTIONS ON LANGUAGE

no reason for surprise. Nor do I see any objection to S’s

practice in referring to the cognitive structure attributed
to the subject as a system of beliefs.
Consider now the case of grammar. If-S speaks English,
he will say that some of his subjects have “learned French”
and now “know French.” Furthermore, in many specific
instances, they can articulate their knowledge, as knowl-
edge that so-and-so. Again, the set of such cases is unlikely
to be of much interest in itself. $ will attempt to explain
these instances by showing how they follow from the
grammar of French, interacting with other cognitive struc-
tures. In this way, he will try to account for facts analogous
(for French) to those mentioned earlier (cf. (1)-(13)). The
problems of confirmation and choice of theories are anal-
ogous to those that arise in the case of investigation of
common sense.
Obviously, the grammar is not in itself a theory of per-
formance (behavior). It is, however, proper for S to propose
that the grammar is a component of such a theory and to
proceed at this point to construct a theory of interaction
of structures that would serve as a theory of performance
for his subjects.’® S might refer to the grammar attributed
to the speaker as a representation (or model) of his know]-
edge of his language. S might also want to say that the
subject who knows the language knows the grammar, and
that in his initial state he knew universal grammar. Thus
S’s subject differs from an English speaker, a rock, or a
bird in that he knows the grammar of French (to use the
suggested terminology). He is like an English speaker, and
different from a rock or a bird, in that in his initial state
he knew universal grammar.
Since some might object to this terminology, S might
prefer to invent some technical terms. Let us say that if a
speaker knows the language L then he cognizes L. Further-
more, he cognizes the linguistic facts that he knows (in
any uncontroversial sense of “know”) and he cognizes the
170 REFLECTIONS ON LANGUAGE

plain certain facts in English. On the empirical assumption

of uniformity among humans with respect to language
acquisition, this evidence would serve to counter S’s as-
sumption that the English-speaking subjects. are, in fact,
making use of a cognitive structure that involves a trans-
formational movement rule. Thus the contrary theories of
S’ and perhaps S” receive some indirect, but valuable, em-
pirical confirmation. Many other kinds of evidence can be
sought.
S, S’, and S” have every reason to take their hypotheses
to be working hypotheses concerning the steady state at-
tained by their subject and thus subject to further con-
firmation or disconfirmation. Surely there can be no gen-
eral objection to the normal “realist” assumptions of any
scientist in this case (though, obviously, care must be
taken, etc.).
Perhaps what Schwartz has in mind is something differ-
ent. Perhaps he has in mind a case in which two theories
are compatible with all the evidence that might in prin-
ciple be obtained. If so, S should simply dismiss. this con-
sideration, as any working scientist would. The notion “all
the evidence that might in principle be obtained” surely
requires some explanation. I doubt that any sense can be
made of it. Furthermore, even if we accept it as meaning-
ful, nothing follows with regard to S’s enterprise. In the
real world, he will never have exhausted the evidence, and
with diligence and imagination can seek new evidence to
select between empirically distinguishable theories.
Suppose that S comes up with several theories as the best
ones he can devise without examination of the internal
structure of his subjects. Then he will say, regretfully, that
he cannot determine on the basis of the evidence available
to him which (if any) of these theories correctly charac-
terizes the actual internal structure. In Schwartz's sphere
example, S will not be able to determine whether E or
rather the account in terms of a liquid of density 1 is the
176 REFLECTIONS ON LANGUAGE

used in the context of speech, melodies, etc.); cf. chapter

2, note 8. There also appear to be some differences in the
stage of maturation at which various centers are estab-
lished in their function (recent work on face recognition
reported by Susan Carey is suggestive in this regard).
Whether or not these tentative proposals prove correct on
further investigation, clearly they are of the sort that a
scientist concerned with symbol systems, their character,
interaction, and acquisition, should explore. Schwartz seems
to believe that any departure from the “natural language
mold” in the case of other symbol systems must be acci-
dental—that is that any innate schematism for language
learning must simply be a general schematism for learn-
ing. But he has offered no argument whatsoever for this
contention, and he simply ignores the many obvious prob-
lems it faces, some mentioned earlier.
The concerns about theory and evidence that Schwartz
expresses and his objections to the program outlined earlier
are typical of much recent discussion. But the objections
have no foundation and the concerns, where justified at
all, have no bearing on the issues, so far as I can see, in
that they apply in a comparable way to any variety of
empirical inquiry. They are worth considering in detail
only because of the insight they provide into empiricist
assumptions.
If Schwartz’s claims on these matters had any merit,
they should apply as well to the study of physical organs.
Suppose that S develops a theory T dealing with the struc-
ture and function of the human eye, and postulates the
innate factors F to account for the growth of an organ
satisfying T. Suppose now that he turns his attention to
the liver. Paraphrasing Schwartz, we might argue that it
is “implausible” to postulate distinct innate factors F’
(which, along with F and others, constitute the genetic
coding that determines the nature of the organism) to ac-
count for the growth of the liver. After all, the eye and the
 Problems and Mysteries 181

erature of generative grammar since the outset, and I think

that they are right in principle, though sometimes difficult
to apply in practice.
Suppose that S faces this problem, and does not think
much of Quine’s “unimaginative suggestion”: Ask the na-
tives. Suppose that S has evidence to suggest that intona-
tion patterns are determined by grammatical structure.”
The evidence might come from the language in question
or from other languages, which are relevant for reasons al-
ready discussed. Such evidence might bear on the choice
between the two proposed grammars; thus we might dis-
cover that the rules needed for other cases give the cor-
rect intonation if we take the constituents to be A—BC but
not AB—C. Or suppose that S has reason to postulate that
transformations are structure-dependent in this sense: a
transformation applies to a string partitioned into a se-
quence of strings each of which is either arbitrary or is a
string of a single constant category. Suppose that spe-
cific transformations—say, coordination—observe constituent
structure in this sense. These principles, which might be
supported by all sorts of evidence, might lead to a choice
between the two grammars in this case (say, if we found
that where ABC and ADE are well formed, then so is A—
BC and DE; though where ABC and FGC are well formed,
still AB and FG—C is not). Other evidence might be derived
by application of the principle that contextual features of
lexical items are internal to constituents, or from semantic
considerations of varied sorts. Given a rich general theory
of universal grammar, S might bring many kinds of evi-
dence to bear on the question. Examples abound in the
literature.
Is there anything enigmatic in all of this, apart from the
inescapable problems of empirical uncertainty? I think not.
At least, Quine suggests nothing, here or elsewhere.
Quine’s sole point reduces to the observation that there
will always be distinct theories that are compatible with
 Problems and Mysteries 207

ment that will account for such linguistic universals as there

appear to be.”
No one doubts the importance of searching for conse-
quences of linguistic theory beyond “the language-learning
facts.” But consider Cohen’s argument, which rests on the
assumption that postulating an innate ability to do x to
explain how it is that children are able to do x is a tauto-
logical pretense. Is Cohen’s assumption correct? Suppose
that the scientist S postulates the structure-dependent prop-
erty of rules (SDP) or the principle SSC as an element
of universal grammar; thus he postulates that children do
not learn these principles, but rather construct a linguistic
system observing these principles. Thus S is postulating an
“innate ability to observe these principles” in order to
explain how it is that children observe those principles
“without learning from experience.” By Cohen’s assumption,
S’s hypothesis is a tautological pretense and therefore can-
not be falsified. But in fact it can be falsified all too easily,
say, by further investigation that shows that SDP or SSC
is violated elsewhere in the language or in some other
language. In fact, proposals concerning universal grammar
—hence, on the interpretation suggested earlier, proposals
concerning innate capacity—have repeatedly been revised
on just such grounds. Thus Cohen’s initial assumption is
false, and his argument collapses.*
The theories of universal grammar so far proposed,
though not a tautological pretense as Cohen falsely al-
leges, are still far from sufficiently rich and restrictive to
explain acquisition of language; they do not sufficiently
limit the class of admissible hypotheses. Thus, contrary to
what Cohen asserts, it seems that theoretical progress in
the explanation of language learning should be sought in
the direction of richer theories of innate universals, at
least, until some other approach is suggested that has some
degree of plausibility. Cohen’s suggestion that we seek less
 Notes

Part I THE WHIDDEN LECTURES

Chapter 1. ON COGNITIVE CAPACITY
1. Aristotle, Posterior Analytics 2. 19 (ed. McKeon, 1941), pp.
184-6.
2. Cudworth (1838), p. 75. Except for those otherwise identified,
quotations that follow in this paragraph are from the same
source: respectively, pp. 65, 51, 49, 87, 122-3.
. Leibniz, Discourse on Metaphysics (trans. Montgomery, 1902),
p. 45. For a similar view, see Cudworth (1838), p. 64, For
quotations and further discussion, see Chomsky (1966), § 4.
. Cudworth, True Intellectual System of the Universe, cited by
OM gp

Lovejoy (1908).
. Lovejoy (1908).
Henry More, “Antidote Against Atheism,” cited by Lovejoy
(1908).
. Gregory (1970). Gregory suggests further that the grammar of
I

guage “has its roots in the brain’s rules for ordering retinal
patterns in terms of objects,” that is, “in a take-over operation,
in which man cashed in on” the development of the visual
system in higher animals. This seems questionable. The struc-
ture, use, and acquisition of language seem to involve special
properties that are, so far as is known, not found elsewhere. Lan-
guage is based on properties of the dominant hemisphere that
may also be quite specialized. There seems to be no obvious
relationship to the structure of the visual cortex in relevant re-
spects, though so little is known that one can only speculate.
It is not clear why one should expect to find an evolutionary
explanation of the sort that Gregory suggests. For more on these
matters, see the chapters by R. W. Sperry, A. M. Liberman,
H.-L. Teuber, and B. Milner in Schmitt and Worden (1974).
. This view, popularized in recent years by B. F. Skinner, is for-
eign to science or any rational inquiry. The reasons for its
popularity must be explained on extrascientific grounds. For
 Index 269

Vendler, Zeno, 252-3 n.47 Williams, Edwin S., 239 n.13

Wittgenstein, Ludwig, 45, 46,
Wang, Hao, 244 n.55 60, 237 0.38
Wasow, Thomas, gg, 100 Wood, Ellen, 131
Weiss, Donald D., 234 n.26
Wiesel, T. N., 9 Yolton, John W., 128