You are on page 1of 13

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/223959049

Chomsky’s Nativism: A Critical Review

Article  in  The Analysis of Verbal Behavior · April 2000


DOI: 10.1007/BF03392954 · Source: PubMed

CITATIONS READS

22 8,757

1 author:

David C Palmer
Smith College
58 PUBLICATIONS   1,944 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Behavior Analysis and Neuroscience: Complementary Disciplines View project

All content following this page was uploaded by David C Palmer on 06 June 2014.

The user has requested enhancement of the downloaded file.


The Analysis of Verbal Behavior 2000, 17, 39-50

Chomsky's Nativism: A Critical Review


David C. Palmer
Smith College

Is grammar innate? Noam Chomsky completely insoluble. This paper out-


holds that it is, or, more accurately, that lines Chomsky's position, emphasizing
the hypothesis that it is innate is the his argument that the brain of the new-
only coherent and plausible one that born infant must be organized to ex-
has yet been proposed to account for tract rules of grammar from samples of
the acquisition of language. Extrapo- speech. This position is criticized on
lating to broader issues, he has cham- two grounds. First, it places too heavy
pioned a retreat from behaviorism and a burden on evolutionary principles.
empiricism to cognitivism and ratio- Second, the putative innate mecha-
nalism, from approaches that seek to nisms must respond to stimuli, to ac-
determine the relationship between an tual physical events, but it appears to
organism's behavior and the environ- be impossible in Chomsky's system to
ment to those that wish to discover the characterize these events. Finally,
organism's "essential nature," of Chomsky's sophistical arguments
which behavior is an incomplete ex- against behavioral accounts of lan-
pression. His arguments, which are de- guage are rejected on grounds that he
tailed, polemical, and persuasive, are has confused properties of his formal
evidently inspiring to a thriving school system with properties of human be-
of linguists and to many laymen with ings. The notion that language consists
an interest in language and philosophy. of an infinite number of sentences must
It is important to assess his position be abandoned when we move from the
carefully, not only because he con- rarefied atmosphere of formal analysis
cludes that little is to be gained by pur- to the world of stimulus and response
suing the analysis of verbal behavior classes.
with the assumptions and methodology
of radical behaviorism but because he CHOMSKY'S ASSUMPTIONS
claims to have achieved considerable
success with very different assump- Chomsky shares a number of fun-
tions and methodology. Success in ex- damental assumptions with behavior-
plaining complex behavior deserves ists and other experimental psycholo-
our attention whatever the approach. gists. He believes that organisms are a
When we examine Chomsky's posi- joint product of their genetic endow-
tion, however, we find that, not only ment and individual experience and
are his objections to other approaches that the experimental approach of the
weak, but the success of which he natural sciences is appropriate for the
speaks has been achieved by rendering study of language. He is tentatively
other problems more difficult, if not monistic; while he freely uses mental-
istic terminology, he does so for the
sake of convenience, believing these
Reprinted, with permission, from P N. Chase terms to be abstractions of physical
& L. J. Parrott (Eds.). (1986). Psychological as-
pects of language (pp. 44-60). Springfield, IL: structures or processes in the body,
Charles C Thomas. presumably the brain (see Chomsky,
I thank John Donahoe for discussions that 1980b, for a recent review of his po-
contributed to the present paper. sition). His goals, however, are differ-
Correspondence conceming this paper should
be sent to the author at Department of Psychol- ent from those of behaviorists. He is
ogy, Smith College, Northampton, Massachusetts not particularly interested in verbal be-
01063 (E-mail: dcpalmer@science.smith.edu). havior itself, influenced as it is by the
39
40 DAVID C. PALMER

idiosyncratic history and circumstanc- A TERMINOLOGICAL QUIBBLE


es of the speaker; rather, he is interest-
ed in the "essential nature" of human To begin with, we may object to
beings that enables us to acquire a lan- Chomsky's terminology, particularly
guage. Specifically, Chomsky wishes his use of the term "universal gram-
to discover those elements of our ner- mar" to refer to unspecified innate
vous systems implicated in language properties of the nervous system. A
that are genetically coded, hence "uni- writer is free to define his terms as he
versal." Chomsky calls these elements chooses, but as Winograd (1977) has
"universal grammar," a name that sug- pointed out, the reader who has agreed
gests his view of the task accomplished to Chomsky's definition finds, in sub-
by these innate mechanisms: providing sequent discussion, that he has agreed
a set of rules to be used in speech pro- to some kind of innate grammar in the
duction and comprehension. traditional sense of the term, i.e., a set
Chomsky is not dogmatic about the of rules. This lends a spurious cogency
nature of universal grammar, so de- to Chomsky's argument. Additionally,
fined. He concedes that it may prove Chomsky freely uses mentalistic terms
such as intention, belief, purpose, will,
to be some general-purpose reinforce- and mind without defining them. Oc-
ment mechanism, but this strikes him casionally he indicates that he is mere-
as implausible. Just as cells in the vi- ly talking about properties of the ner-
sual cortex are organized in special vous system, but he does not tell us
ways not characteristic of cells con- where in the nervous system we will
trolling, say, digestion, those elements find the mind with its intentions, be-
of the nervous system responsible for liefs, and will. As a consequence, his
our ability to acquire language should discussions remain abstract and meta-
not be expected to be organized in the phorical, apparently awaiting the day
same way as those implicated in learn- that someone can operationalize his
ing to ride a bicycle. The language sys- terms without endangering the formal
tem, the visual system, and no doubt system that has been erected on this
other systems, he asserts, may be mod- terminological quicksand.
ular. Even within the formal system itself,
Chomsky's interest in species differ- Chomsky's terms are not always clear-
ences is not what makes his views con- ly defined. A language, we are told, is
troversial. Species differences, of "a set (finite or infinite) of sentences,
course, are of as much interest to be- each finite in length and constructed
haviorists as to linguists. A child learns out of a finite set of elements" (Chom-
to speak, and the family dog does not. sky, 1957, p. 13). Subsequently, we
No one doubts that this is due to ge- learn that human languages are infinite
netic differences, and any illumination sets of sentences. But what is a sen-
of these differences will contribute to tence? Chomsky uses two definitions
our understanding of the behavior of of sentence, a formal, precise one and
organisms. Direct evidence of the ge- an informal one, and he does not con-
netic contribution to human behavior is sistently use either one. The informal
hard to acquire, however. Programs of definition is simply that which native
genetic engineering, surgical intrusions speakers agree to be a sentence when
of the nervous system, and well-con- they are not encumbered by "irrele-
trolled behavioral experiments are, for vant" problems of memory, motiva-
ethical reasons, seldom possible. We tion, time, or patience. This is a rela-
must usually make do with speculation tively small set owing, apparently, to
and extrapolation from indirect evi- the ubiquity of these encumbrances.
dence. It is Chomsky's particular spec- By the formal definition, a sentence is
ulations and his rejection of alternative any string of symbols characterized by
speculations that are controversial. the grammatical rules devised by the
CHOMSKY'S NATIVISM 41
linguist. At the very least, these gram- about these matters, or even that they are
matical rules must be consistent with provided with relevant experience that in-
forms them that they should not make the
the set of sentences defined by consen- obvious inductive generalization, say, that
sus. Thus, the formal definition is de- "each other" takes some plural antecedent
rived from data provided by using the that precedes it. Children make many errors
informal definition, and both defini- in language learning, but they do not as-
tions depend ultimately upon the gram- sume, until corrected, that "The candidates
wanted me to vote for each other" is a
matical intuitions of native speakers. well-formed sentence meaning that each
However, Chomsky does not provide candidate wanted me to vote for the other.
us with criteria for deciding when a Relevant evidence is never presented for
judgment of grammaticality is to be most speakers of English, just as no peda-
gogic or traditional grammar, however
considered valid, an important omis- compendious, would point out these facts.
sion considering that these judgments Somehow this is information that children
are notoriously variable. themselves bring to the process of language
acquisition as part of their mode of cogni-
tion. (Chomsky, 1980b, pp. 43-44)
ARGUMENTS FOR A
GENETICALLY DETERMINED A similar problem is raised by "ques-
GRAMMAR tion formation":
One could argue that these termi- We select some noun phrase in a sentence,
nological issues are irrelevant to eval- replace it by an appropriate question word,
uate the substance of Chomsky's posi- place the latter at the beginning of the sen-
tion, particularly his argument that tence, and with other mechanical opera-
there is an innate language module. So tions, form a question. Thus on the model
of the sentence, "John saw a man," we can
let us turn to this critical argument. As form "Whom did John see?" Or, to take a
noted above, his unit of analysis is the more complex case, on the model of the
sentence, and his data are his judg- sentence, "The teacher thought that his as-
ments, and presumably the judgments sistant had told the class to study the les-
of others, that particular sentences are son," we can question "the class" and ask:
"Which class did the teacher think that his
or are not "well-formed." (He is not assistant had told to study the lesson?" But
concerned with prescriptive rules of consider the following example of roughly
grammar, such as proper use of the ob- comparable complexity: "The lesson was
jective case, but with regularities in harder than the teacher had told the class it
would be." Here, if we question "the
language that are respected by native class," we derive: "Which class was the
speakers without formal training.) lesson harder than the teacher had told that
Thus, (1) is a well-formed sentence, it would be?" Evidently this is not a well-
while (2) is not: formed question, though its intended sense
(1) Is the man who is hungry here? is clear enough and perfectly intelligible,
with a little thought. It is difficult to imag-
(2) Is the man who hungry is here? ine that people capable of these judgments
Similarly, (3), (4), and (5) are well- have all had relevant training or experience
formed, while (6) is not, though its to block the obvious inductive generaliza-
meaning is reasonably clear: tion. Rather it seems that some specific
property of the human language faculty-
(3) Each of the men likes the others. hence a general property of language-
(4) The men like each other. leads to these consequences, a property that
(5) Each of the men expects John to derives from our modes of cognition.
like the others. (Chomsky, 1980b, p. 42)
(6) The men expect John to like each When we analyze the structure of
other. language at a certain level of abstrac-
What is it, Chomsky asks, that pre- tion, according to Chomsky, we dis-
vents people from uttering sentences cover that there are general principles
such as (6)? Surely no one has been of grammar that are violated in sen-
taught such things in grammar school. tences such as (6), for example, the
It can hardly be maintained that children principle that a reciprocal expression
learning English receive specific instruction such as "each other" may not refer to
42 DAVID C. PALMER
an antecedent outside of the clause in theory, a given sentence is presumed
which "each other" occurs unless the to be "represented in the mind" at sev-
latter happens to be the subject of an eral levels. It begins as a declarative
infinitive (Chomsky, 1980b, p. 174), as sentence, with expressed (as opposed
in (7). Note that (8) is ungrammatical to "understood") subjects, verbs, di-
according to this principle: rect objects, and so on. Elements of the
(7) The candidates expect each other sentence may then be deleted or rear-
to win. ranged subject to various constraints
(8) The candidates expect each other such as those governing reciprocal ex-
will win. pressions like "each other." At this
Because children quickly learn to re- level the sentence includes "traces" of
spect such distinctions with little, if deleted and rearranged elements in
any, formal instructions, and because their original positions as well as the
no one has proposed a satisfactory ex- rearranged elements in their new po-
planation of these facts in terms of a sitions. Finally, the surface structure,
theory of learning, Chomsky assumes the sentence as it appears at the behav-
that fundamental elements of the gram- ioral level, is generated by representing
mar of human languages must be ex- phonetically all elements except trace
pressed somehow in the genetic code. elements. This scheme accounts for in-
He suggests that universal grammar, tuitions about relatedness among sen-
triggered by relatively brief exposure tences, ambiguities, and many fine dis-
to a particular language, is able to ex- tinctions about what is grammatical
tract or construct a grammar for that and what is not.
language. Universal grammar presum- This picture is incomplete, of
ably contains those fundamental prin- course, but Chomsky asserts that it
ciples that are common to all human stands in sharp contrast to alternative
languages and constrains the particular theories which do not even attempt to
grammars that can be acquired. explain the kinds of grammatical judg-
Grammar is seen as fundamental; ments people are capable of, judg-
"language" is an epiphenomenon in- ments which have served as grist for
fluenced by motivational variables, the theories of linguists. Chomsky
memory, nonlinguistic concept learn- writes:
ing, and other things. The task of the
linguist is to characterize abstractly The critic's task is to show some funda-
mental flaw in principle or defect in exe-
grammars of various languages as they cution or to provide a different and pref-
would be spoken by ideal speakers in erable account of how it is that what speak-
a homogeneous verbal community in ers do is in accordance with certain rules
an attempt to discover principles of or is described by these rules, an account
that does not attribute to them mental rep-
grammar of wide generality, if not uni- resentation of a system of rules (rules
versality. Work of this sort has been in which in fact appear to be beyond the level
progress for several decades and, ac- of consciousness). If someone can offer
cording to Chomsky, has met with con- such an account of how it is that we know
siderable success. He concedes that what we do know, e.g. about reciprocals, or
judge as we do judge, etc., there will be
conclusions are tentative and will un- something to discuss. Since no such ac-
doubtedly be refined or replaced, a count has been forthcoming, even in the
state of affairs to be expected in any most primitive or rudimentary form, there
empirical inquiry. Nonetheless, lin- really is nothing to discuss. (1980b, p. 130)
guists are becoming increasingly able
to characterize universal grammar and CHOMSKY'S CHALLENGE
hence to offer hypotheses about the ca-
pacities that newborn infants bring into One need not be disconcerted by this
the world as a product of their genetic challenge. Chomsky is charging his
endowments. critics to provide an alternative expla-
In the current form of Chomsky's nation for hypothetical behavior, the
CHOMSKY'S NATIVISM 43
behavior of judging particular sentenc- other" is a single operant, where X and
es grammatical or ungrammatical. He Y have certain prosodic, temporal,
evidently believes that everyone will functional, and perhaps formal features
respond in the same way and for the but are otherwise free to vary. (We
same reason, that is, that there is an would resist the temptation to call X
independent variable called "grammat- "a plural noun phrase" and Y "a
icality" that controls the behavior of verb," for that merely raises the ques-
anyone asked to judge utterances. tion of what the formal, functional,
Since judgments of grammaticality are prosodic, and temporal characteristics
in fact highly variable, he insists that are of plural noun phrases and verbs.)
we consider only the behavior of an We might find that "each other" is a
ideal speaker in a homogeneous verbal component of half a dozen larger units
community. or that it is under intraverbal control of
Analogous idealizations have been a number of stimuli. In the latter case
adopted in other sciences. When many a number of different operants would
variables interact, it is common prac- be formally identical.
tice to consider each in isolation. Once relevant units of behavior and
Hence the physicist assumes point their controlling variables were identi-
masses, frictionless surfaces, and per- fied it would be possible to speculate
fectly elastic collisions. However, these whether a particular utterance would
idealizations are useful only if the var- sound natural or strange to an individ-
iables being omitted are unimportant ual. Since the value of such a predic-
for an understanding of the phenome- tion by no means justifies the effort to
non under study. If one's verbal behav- gather the relevant data, it is unlikely
ior and judgments about utterances are that anyone will answer Chomsky's
in fact a function of the individual's challenge. (We can, of course, invent
particular experience within a specific contingencies to explain any given ex-
community, then considering the intu- ample, but he would regard this as
itions of an ideal speaker in a hypo- empty.) Nonetheless, there are alter-
thetical community will tell us nothing. natives to Chomsky's account that de-
Faced with disorderly data, Chomsky pend, not on internal representations or
removes to a hypothetical world where underlying competence, but on the in-
order emerges. It is not surprising that dividual's long history with relevant
no one has proposed an alternative ac- verbal operants. It is hard to see how
count, for this is a world of Chomsky's it could be otherwise, for we have no
own making. Order has not been dem- intuitions about strings of grammatical
onstrated; it has been assumed. symbols by themselves or about sen-
However, even if we satisfactorily tence tree diagrams. Despite Chom-
demonstrate that instances like those sky's suggestion that intuitions about a
provided by Chomsky are not universal full range of uses of the term "each
the task remains of explaining why cer- other" follow from learning that it is a
tain novel expressions "sound right" reciprocal expression and not the name
to someone while others do not. From of a tree, surely we have no intuitions
a behavioral point of view the task is about strings of nonsense syllables
formidable, requiring that we know a drawn from bins labeled "noun,"
great deal about the individual's rein- "verb," "reciprocal expression," and
forcement history-more than it is so on. Knowing grammatical catego-
usually possible to know. We would ries is no help in judging utterances
have to determine appropriate units of "by ear."
behavior and the individual's history As for a "flaw in principle" that
with respect to these units. We might Chomsky exhorts his critics to cite,
find, for example, that "each other," there appears to be none. There are no
though two words, is a single operant objections in principle to the notion,
or that the frame "X ... Y ... each however vague, that the nervous sys-
44 DAVID C. PALMER

tem is innately designed to extract a ontogenic environment and hence can-


grammar from a sample of speech, but not be accounted for in terms of com-
we can question the extent to which municative contingencies (Chomsky,
this serves as a parsimonious explana- 1980a, p. 41), then it clearly cannot
tion of verbal behavior. Rather than confer a selective advantage to an or-
shedding light on the problem, it ren- ganism endowed with it. Chomsky ac-
ders it more mysterious. knowledges this problem but points out
Chomsky begins by characterizing that a child has only a few years to
grammar in formal terms. Having construct a grammar while the princi-
achieved some success at this he then ple of natural selection has had many
simply imputes the formal apparatus to thousands of years (Chomsky, 1980c,
the speaker as an innate mechanism. p. 44). This will not suffice, however.
This is a tidy solution to a complex If the rules are arbitrary it doesn't
problem, and it might even be true, but make any difference how long selec-
note that, as it stands, it is a homun- tive forces have been at work. Natural
culus theory, and as such, it is unsat- selection is simply not an appropriate
isfactory until the homunculus has in mechanism to explain universal gram-
turn been explained. The genetic en- mar.
dowment is a convenient source of ho- This conclusion does not trouble
munculi for every behavioral phenom- Chomsky. He writes:
enon we don't understand. If we ask
nothing further of such an explanation It is, in fact, perfectly possible that the in-
nate structure of mind is determined by
then it is a universal solution. There are principles of organization, by physical con-
no limits to our invoking it. Two ques- ditions, even by physical laws that are now
tions immediately arise to which quite unknown, and that such notions as
Chomsky has provided no satisfactory "random mutation" and "natural selec-
tion" are as much a cover for ignorance as
answer. First, how did universal gram- the somewhat analogous notions of "trial
mar get selected, and second, how does and error," and "conditioning," "rein-
universal grammar get "triggered" by forcement," and "association." (Chomsky,
a verbal environment? 1969, p. 262)
Again, Chomsky might be right that
UNIVERSAL GRAMMAR there are additional principles in-
AND EVOLUTION volved, but this hardly offers support
Certain grammatical conventions for his position. Rather, it adds a fur-
serve a communicative function and ther burden of proof. In addition to ex-
might be learned because they do so. plaining the origins of grammar, he
Other principles are arbitrary and seem must now formulate and explain the
to have little point, such as that con- workings of new evolutionary princi-
cerning reciprocal expressions. Of this ples or "physical laws now quite un-
principle, Chomsky (1975, p. 175) known."
writes, "It is often a difficult problem Perhaps not wanting to depend on
even to discover examples that bear on
unknown principles, Chomsky has sug-
the hypothesis in question." It is prin- gested another solution:
ciples such as this that Chomsky ar- [Universal grammar] may well have arisen
gues must be innate: The examples are as a concomitant of structural properties of
few and the grammatical rules are ar- the brain that developed for other reasons.
Suppose that there was selection for bigger
bitrary and unnecessary. But this is an brains, more cortical surface, hemispheric
argument that cuts both ways; the same specialization for analytic processing, or
evidence that he adduces to support his many other structural properties that can be
position can be used in a parallel ar- imagined. The brain that evolved might
have all sorts of special properties that are
gument against it. If a grammatical not individually selected; there would be no
principle is an arbitrary restriction miracle in this, but only the normal work-
without practical consequences in the ings of evolution. We have no idea, at pres-
CHOMSKY'S NATIVISM 45
ent, how physical laws apply when 1010 functional analyses of verbal behavior
neurons are placed in an object the size of (e.g., Skinner, 1957) require few, if
a basketball, under the conditions that arose
during human evolution. (Chomsky, 1980a, any, principles in addition to those al-
p. 321) ready known to apply to nonverbal be-
havior; moreover, these principles are
It is certainly true that not every- clearly adaptive (Skinner, 1966), and
thing coded by the genes must be adap- apply with appropriate qualification
tive. Hair color, eye color, and blood down the evolutionary ladder. Humans
type are all genetically determined and have the necessary vocal musculature
are not obviously adaptive, but neither and are particularly sensitive to sec-
are they universal. When a trait is not ondary reinforcement, social contin-
actively selected for, we expect vari- gencies, and, apparently, private stim-
ability, not universality. Moreover, to uli generated by other behavior. These
the extent that arbitrary structures re- are quantitative differences from other
quire energy and resources, we would organisms that are both adaptive and
expect them to be selected against. easily accommodated by evolutionary
The chances that a universal gram- principles. These differences alone
mar was an accidental by-product of may be sufficient to account for verbal
other properties of the nervous system, behavior in humans.' Chomsky, in at-
an unexpected bonus when the human tributing grammatical competence to
nervous system exploded in size, seem the newborn infant, has not solved the
remote indeed, given how abstract and problems of language acquisition; he
complex the putative innate rules are. has simply transferred them to the
Moreover, in the absence of sugges- shoulders of the evolutionary biologist,
tions about the structure of which where they remain as intractable as be-
grammar is a by-product, and what the fore.
relationship between them is, Chom-
sky's answer is no answer at all. We THE STIMULUS CONTROL
might just as plausibly assert that lan- OF INNATE BEHAVIOR
guage is an accidental by-product of
other behavior acquired in the first few An additional difficulty facing
years of life. Chomsky's position is perhaps more
Chomsky (1980a, 1980b) repeatedly fundamental. Let us assume that he is
asserts that the problem of explaining correct, that humans are innately
the genetic basis of universal grammar equipped with a neurological module
is no different from the problem of ex- that extracts an acceptable grammar
plaining the origin of any physical or- from a small and degenerate sample of
gan, say, the liver. No one ever asserts speech, triggered perhaps by critical
that we learn to have arms rather than experiences and with parameters set by
wings, or that we learn to have a heart,
he argues. No doubt there are many ' The development of vocal musculature sen-
things that we don't know about the sitive to reinforcement contingencies may be es-
origin and development of the physical pecially significant. First, it is a response system
that is free from the demands of locomotion, ori-
organs, but to be confident that the ge- entation, and the manipulation of objects. Pri-
netic endowment exerts considerable mates usually have plenty to do with their hands
control, surely it is sufficient to note other than sign with them. Second, and perhaps
that such structures are adaptive and more importantly, when we speak, we stimulate
ourselves exactly as we stimulate others, and we
that they are, in fact, physical struc- do so essentially instantaneously. This immedi-
tures. ate and faithful stimulation, which is not char-
A hypothesis about behavior need acteristic of, say, sign language, is no doubt im-
not specify physiological correlates or portant both in maintaining somewhat uniform
evolutionary origins to be useful, but a contingencies throughout the community and in
facilitating the acquisition of verbal operants.
complex structure with no adaptive Under some conditions, reinforcement will be
significance is anomalous. In contrast, "automatic."
46 DAVID C. PALMER

developments in "other cognitive do- ment contingencies. This is unsatisfac-


mains." Subsequent performance is a tory, since we still do not end up with
hodgepodge of behavior, an epiphe- a class of symbols that coincides with
nomenon, determined in part by the the concept noun. We do, however, end
grammatical module and in part by up with a repertoire of behavior that
many other factors. The problem that coincides precisely with Skinner's con-
now arises is the relationship between cept of tact. Once grammatical distinc-
"degenerate speech" and the device tions are traced to contingencies of re-
that extracts a grammar from this inforcement, the innate grammar is no
speech. The putative device is an in- longer doing any work. On the other
put-output device. In go samples of hand, if they cannot be traced to rein-
speech, and out comes the grammar, or forcement contingencies, then the child
perhaps a set of candidate grammars, (or the innate mechanism) has no way
most of which will be winnowed out of generating a grammar.
later. Setting aside the improbability Chomsky's allusions to imprinting
that such a device is an accidental by- and fixed-action patterns as examples
product of, say, increased cortical sur- of complex innate behavior (1959, pp.
face, we must determine the functional 41-43) suggest that he fails to appre-
relationship between this input and ciate that these behaviors do not occur
output. This can be considered a kind spontaneously but are elicited by spe-
of problem in stimulus control, since cific stimuli. Herring gull chicks do not
each verbal stimulus controls a partic- "peck at their mothers' bills to get
ular response of the device, as in a re- food"; they peck at red spots. Duck-
flex. However, unlike the reflex, the re- lings do not "follow their mothers";
lationship between stimuli and the they are reinforced by proximity to ob-
grammar is arbitrary. Languages vary jects similar to the particular object
from culture to culture, and within a that was bustling around when they
language there is no relationship be- hatched. If we wish to say that a par-
tween the sound of an utterance and its ticular behavior is genetically deter-
grammatical structure. Clearly there is mined or "wired in," it must be pos-
no physical property of the stimulus sible to specify the environmental
that suffices to identify its part of events that elicit, release, or trigger it.
speech. Nothing about the word house Not only has Chomsky failed to do this
enables us to conclude that it is a noun, for his hypothetical grammar-generat-
or that it might be a "subject." ing device, he apparently thinks it can-
The input to this device, then, must not be done:
be the product of a grammatical anal-
ysis rather than raw stimuli. At the Although one might propose various oper-
ational tests for acceptability, it is unlikely
very least, words must be classified that a necessary and sufficient operational
into their parts of speech. But parts of criterion might be invented for the much
speech have formal definitions; they do more abstract and far more important no-
not have operational ones. Nouns are tion of grammaticalness. (Chomsky, 1965,
p. 11)
often uttered in the presence of Furthermore, there is no reason to expect
"things" and verbs in the presence of that reliable operational criteria for the
activity, but many nouns are not deeper and more important theoretical no-
"things," and many verbs are not per- tions of linguistics (such as grammatical-
ceptible actions. Perhaps when a child ness and paraphrase) will ever be forth-
coming. (Chomsky, 1965, p. 19)
utters a particular word in the presence
of a particular class of objects (or state If there are no stimuli, objective crite-
of affairs) and is reliably reinforced for ria, or even a set of operations by
doing so, that acoustical signal is rep- which we (or our innate language ac-
resented and tagged with an "N." Thus quisition devices) can identify such
every grammatical distinction might be theoretical entities as grammatical sen-
traced to a particular set of reinforce- tence, subject, noun phrase, and so on,
CHOMSKY'S NATIVISM 47

then it is a mystery how we can reflex- fore a noun, or to the number of times
ively generate rules characterizing per- we can repeat the word very for em-
missible relationships among these en- phasis; nor is there any limit to the
tities. Chomsky has been able to for- number of sentences or clauses that we
mulate precisely his theoretical ideas can add or insert in other sentences, as
because they have remained abstract, in (9) and (10):
but useful theories cannot remain ab- (9) The rat the cat the dog chased killed ate
stract forever. If there is no way to use the malt.
them to predict, control, or describe ac- (10) Anyone who feels that if so many stu-
tual events, then they are empty. dents whom we haven't actually admitted
are sitting in on the course than ones we
THE UNIT OF ANALYSIS have that the room had to be changed, then
AND THE NOTION OF probably auditors will have to be excluded,
is likely to agree that the curriculum needs
INFINITY revision. (from Chomsky & Miller, 1963,
p. 286)
The choice of a unit of analysis in
behavior is critical. The orderly rela- Although native speakers gape in dis-
tionship between behavior and its con- may when asked if (10) is a grammat-
trolling variables deteriorates if we ical sentence, the authors assure us that
consider units that are too broad, too this "is a perfectly well-formed sen-
long, or too narrowly specified (Skin- tence with a clear and unambiguous
ner, 1935). If one defines one's units a meaning, and a grammar must be able
priori rather than empirically, it is pos- to account for it if the grammar is to
sible that behavior will appear to be have any psychological relevance" (p.
infinitely variable and to bear little re- 286). However, it is obvious that such
lationship to environmental events. sentences are not behavioral units but
Chomsky commits this error by choos- are strings carefully constructed to be
ing the sentence as a unit of analysis. consistent with grammatical rules. It is
He does not defend this choice; he ap- true that there are an infinite number of
pears to regard it as self-evident, de- such strings, but their relevance to ver-
spite the fact that people often do not bal behavior is doubtful. Nevertheless,
speak in sentences and in appropriate Chomsky uses the notion that there are
contexts regard single words or phrases an infinite number of grammatical sen-
as "well-formed." As noted above, the tences to dismiss the use of the term
sentence is a formal unit, not a behav- probability in discussions of language
ioral one, though Chomsky pays little and particularly to criticize Skinner's
heed to this distinction. Since the analysis of language as a repertoire of
speaking of sentences, however de- verbal operants:
fined, typically does not display the It is unclear what sense there would be to
same dynamic properties as, say, key the assertion that a person has "learned" a
pecking in pigeons, he concludes, not sentence that takes twice as long to say as
that he has erred in his choice of units, his entire lifetime. . . On empirical
but that principles formulated in the grounds, the probability of my producing
experimental analysis of behavior are some given sentence of English-say, this
of only peripheral interest in the study sentence or the sentence "Birds fly" or
"Tuesday follows Monday," or whatever-
of language. is indistinguishable from the probability of
Of special significance to Chomsky my producing a given sentence of Japanese.
is the notion that humans have the ca- Introducing the notion of "probability rel-
pacity to speak and understand an in- ative to a situation" changes nothing, at
finite number of grammatical sentenc- least if "situations" are characterized on
es, though actual performance is lim- any known objective grounds. (Chomsky,
1969, p. 267)
ited by motivation, memory, time, and But what does it mean to say that some
other resources. There is no limit to the sentence of English that I have never heard
number of adjectives we can insert be- or produced belongs to my "repertoire,"
48 DAVID C. PALMER
but not any sentence of Chinese (so that the "sentences," it is likely that bees gen-
former has a higher "probability")? erate or respond appropriately to more
(Chomsky, 1971, p. 20) than a hundred or so. (Note that since
According to Chomsky this follows bees have other ways of locating flow-
from the fact that most sentences are ers, this number would be more than
unique and hence have a probability sufficient to satisfy the contingencies
near zero. of natural selection.) To argue that bees
Chomsky is making an extraordi- have the "competence" to interpret an
nary leap from asserting that a gram- infinite number of patterns is to con-
mar can generate an infinite number of fuse a property of our formulation with
sentences to asserting that humans a property of the organism.
have the competence to generate and We can make a similar argument
understand an infinite number of sen- with respect to human language. Sen-
tences. This is clearly not an empirical tence (1 1) is indiscriminable from (10)
fact. It is not even a valid generaliza- in normal discourse.
tion from the empirical fact that behav- (11) Anyone who feels that if so
ior is variable. Let us suppose that we many more students whom we haven't
have arrived at a definition of sentence actually admitted are sitting in on the
that allows us to determine when a sen- course than ones we have that the
tence has been uttered. We have no jus- room had to be changed, that probably
tification for predicting future variabil- auditors will have to be excluded, is
ity until we analyze the variables of likely to agree that the curriculum
which a sentence is a function. If we needs revision.
can show that these are infinitely var- If the two sentences are in print, we
iable, and that human behavior tracks can detect a physical difference in
the full range of this variability, then them, given a pencil and enough time,
we are perhaps justified in predicting but we do so in a purely mechanical
the infinite variability of sentences. way, analogous to comparing signa-
However, behavior and its controlling tures in a forgery case. We clearly do
variables are not divisible into an infi- not do so on the basis of grammar.
nite number of orderly pairs, a point Once again, talk of competence is mere
made by Skinner as early as 1935. We invention. There is no behavioral jus-
can illustrate the point by considering tification for calling these strings dif-
the "language" of honey bees. ferent stimuli, or, if emitted, different
As is well known, a bee, having re- responses. Nonetheless, they are differ-
turned from successful foraging, will ent sentences as defined by Chomsky.
fly in a pattern with a distinctive ori- Evidently the sentence is an inappro-
entation, depending on the position of priate unit of analysis of verbal behav-
the sun and the location of the food ior. Dropping it in favor of an empiri-
source. Other bees, observing this pat- cally defined unit not only avoids the
tern, will successfully locate the food problem that only an infinitesimal frac-
source. As a circle has an infinite num- tion of all sentences are discriminable,
ber of diameters, so there are an infi- it accommodates the awkward fact that
nite number of orientations of a pattern people often do not speak in sentences
of flight. Undoubtedly no two bee at all. Moreover, it obviates the need to
"'sentences" have ever been identical. find a translation between the formal
However, this variability is irrelevant if apparatus and actual data. That is, we
it is not functionally related to the lo- no longer need to find an operational
cation of the food. Clearly no honey definition of "sentence" to match the
bee can discriminate an infinite number formal one.
of patterns, either as a "speaker" or as But when we have abandoned the
a "'listener." Although an abstract sentence as a unit of analysis and the
characterization of bee communication notion that language consists of an in-
could generate an infinite number of finite number of sentences, the argu-
CHOMSKY'S NATIVISM 49
ment against an analysis of language dous diversity of living organisms, is a
as a repertoire of verbal operants function of selecting contingencies in a
breaks down. It now makes sense to diverse environment.
say of a string of phonemes that Methodological problems remain.
would take twice as long to say as Owing to ethical constraints, it may
one's lifetime that it simply is not a never be possible to account for verbal
unit of behavior, and it now makes behavior to the satisfaction of the most
sense to ask whether a particular unit cautious critic. At the moment perhaps
is in one's repertoire. the best we can do is to continue the
work that Skinner and others have be-
BACK TO THE gun; analyze complex verbal contin-
VERBAL OPERANT gencies informally while attacking ex-
perimentally the more tractable prob-
By choosing the sentence as his unit lems in verbal behavior.
of analysis, Chomsky has been led to
maintain that grammar is central to lan- REFERENCES
guage and that grammar must be ge-
netically determined. Since extracting Chomsky, N. (1957). Syntactic structures.
a completely adequate grammar from The Hague: Mouton.
samples of speech is an achievement Chomsky, N. (1959). Review of Verbal Be-
that has eluded many years of effort by havior by B. F Skinner. Language, 35, 26-
linguists, surely it could not be accom- 58.
plished by every 3-year-old unless the Chomsky, N. (1965). Aspects of a theory of
syntax. Cambridge, MA: MIT Press.
job were, in important respects, genet- Chomsky,
ically coded. N. (1969). Some critical assump-
As we have seen, this offers only the tions in modem philosophy of language. In
illusion of an explanation, since we S. Morgenbesser, P. Suppes, & M. White
(Eds.), Philosophy, science and method
must now explain the origin of the (pp. 260-285). New York: St. Martin's
code in the genes, a task for which Press.
evolutionary principles are ill-suited. Chomsky, N. (1971, December 30). The case
Moreover, any innate device must re- against B. F Skinner. The New York Review
spond to actual physical events, not of Books, 17, 18-24.
metaphors or abstractions; unless Chomsky, N. (1975). Reflections on lan-
grammatical terms can be defined guage. New York: Pantheon.
physically or operationally there is lit- Chomsky, N. (1980a). On cognitive struc-
tle reason to believe that such a device tures and their development: A reply to
is possible. Chomsky and his col- Piaget. In M. Piattelli-Palmarini (Ed.), Lan-
leagues have analyzed formal proper- guage and learning: The debate between
Jean Piaget and Noam Chomsky (pp. 35-
ties of language in commendable detail 52). Cambridge, MA: Harvard University
and have found a wealth of curious Press.
regularities that deserve explanation. Chomsky, N. (1980b). Rules and represen-
However, they have not advanced the tations. New York: Columbia University
functional analysis of verbal behavior Press.
at all. Chomsky, N. (1980c). Rules and represen-
When we turn from the sentence to tations (with author's responses). The Be-
the verbal operant as a unit of analysis, havioral and Brain Sciences, 3, 1-61.
we avoid many of the problems faced Chomsky, N., & Miller, G. A. (1963). Intro-
by a formal analysis. Our terms are duction to the formal analysis of natural
languages. In R. D. Luce, R. R. Bush, &
empirically defined, and the principles E. Galanter (Eds.), Handbook of mathe-
invoked are clearly adaptive. Chom- matical psychology (Vol. 2, pp. 269-321).
sky's arguments notwithstanding, nov- New York: Wiley.
elty and diversity are not problems for Skinner, B. F (1935). The generic nature of
a functional analysis. The tremendous the concepts of stimulus and response.
diversity in language, like the tremen- Journal of General Psychology, 12, 40-65.
50 DAVID C. PALMER
Skinner, B. F (1957). Verbal behavior. New Winograd, T (1977). On some contested
York: Appleton-Century-Crofts. suppositions of generative linguistics about
Skinner, B. F (1966). The phylogeny and on- the scientific study of language. Cognition,
togeny of behavior. Science, 153, 1205-1213. 5, 151-179.

View publication stats

You might also like