Thinking about language: Chomsky – Geoff Poole
Noam Chomsky (1928-), Institute Professor of Linguistics, MIT, USA1. The cognitive revolution: linguistics as a science of mind
In the early 1950s, the dominant research paradigm in the humanities and social sciences was behaviourism. Behaviourists were strongly influenced by a philosophical conception of science known as logical positivism, according to which scientific knowledge is solely concerned with empirical observation: discussion of unobservable entities was dismissed as unscientific. For this reason, behaviourists held that psychology should not attempt to describe the internal structure and properties of the mind, which cannot be observed directly; rather, they argued that psychologists should study observable behaviour.
In particular, behaviourists focused on the relationships between the (observable) stimuli presented to an organism and the organism’s (observable) response. A classical model for the study of stimulus-response relationships was Pavlov’s investigation of conditioning and learning by association: in Pavlov’s experiments, a dog that was exposed to the ring of a bell whenever it was fed learnt to associate the sound of the bell with the expectation of food and would salivate upon hearing the bell. Behaviourists regarded human language as not essentially dissimilar, i.e. as a set of stimulus-response links learnt by association.
Noam Chomsky played a key rôle in the downfall of behaviourism. In 1959 he published a devastating review of Verbal behavior (1957), a treatise on language by the then leading behaviourist psychologist B. F. Skinner. Chomsky’s review conclusively showed that the stimulus-response paradigm was wholly inadequate for the study of human language. Moreover, Chomsky established the principle that language is mentally constituted, i.e. that it exists in the minds of speakers.
Chomsky’s pioneering work led to the birth of contemporary cognitive science (the so-called ‘cognitive revolution’). Under Chomsky’s influence, researchers in the humanities and social sciences now acknowledge the need to study the internal structure and properties of the human mind/brain. In contemporary cognitive science, the study of the mind/brain is guided by an analogy with the computer: the mind/brain is regarded as an information-processing device; it contains representations analogous to computer data structures and carries out computational procedures similar to computer algorithms. This research paradigm is often referred to as the computational theory of mind.
Some of the arguments for regarding human language as esssentially mentally constituted were presented in the first lecture for this module and the first lecture for ELL101. Consider for example the following pair of sentences:
(1) / a. / Flying planes are dangerous.b. / Flying planes is dangerous.
In (1a), the subject of the sentence (the string Flying planes) is a Noun Phrase (NP) headed by the plural noun planes —hence the plural verb are; in (1b), Flying planes is a clause headed by the verb Flying with planes as its object —hence the singular verb is. But the string Flying planes is physically identical in both cases. Accordingly, being a NP or a clause cannot be a property of sound strings; it is a property of representations existing only in the mind of speakers.
Consider also the following sentence:
(2) / Bill asked how old Sam was.(2) is ambiguous between a query about Sam’s age and a query about old Sam’s state of health:
(3) / a. / Bill asked [how old] Sam was.b. / Bill asked how [old Sam] was.
Ignoring intonation, the difference between (3a) and (3b) has no physically observable correlates: these two syntactic structures exist only in the minds of speakers.
2. Linguistic creativity: compositionality, rules, and grammars
One of Chomsky’s key concerns is how to account for the creative character of human language. At any point in his life, a human being has only been exposed to a finite set of utterances; yet, in the absence of pathological conditions, an adult human being has the capacity to produce and understand an infinite variety of sentences, most of which he will never have been exposed to before. The problem is: how is this infinity compatible with the inevitably finite character of the human mind/brain?
One key to the answer lies in the compositional nature of complex linguistic signs. Like Saussure, Chomsky acknowledges that the relationship between sound (signifier) and meaning (signified) in simple signs such as cat is arbitrary, as witnessed by the fact that the same signified is associated with different signifiers in French (chien), German (Hund), Spanish (perro), etc. Any speaker can only store a finite stock of such arbitrary simple signs in his mental lexicon. In the case of complex linguistic signs, however, the relationship between sound and meaning is not arbitrary. Consider, for example, the two sentences in (4):
(4) / a. / A dog bit a man.b. / A man bit a dog.
(4a) is relatively trivial; in contrast, (4b) is newsworthy. The difference in meaning between (4a) and (4b), however, is not arbitrary: it correlates in a predictable way with the differences in word order between the two sentences. This correlation is clearly rule-governed; the same pattern can be found in the following examples:
(5) / a. / John loves Maryb. / Mary loves John.
(6) / a. / The policeman spotted the burglar.
b. / The burglar spotted the policeman.
Thus, the meaning of a complex sign is compositional: it is determined by rules on the basis of the meaning of arbitrary simple signs plus the internal structure of the complex sign.
The rule-governed character of complex linguistic signs holds the key to human linguistic creativity. Chomsky argues that, in the mind/brain of speakers, there is represented a (finite) set of linguistic rules. Given a (finite) stock of arbitrary simple signs, these rules can generate an infinite number of complex signs (sentences), whose meaning is compositional. From this viewpoint, the goal of linguistic analysis is the discovery of the mentally represented set of rules that underlie a speaker’s linguistic abilities. Chomsky uses the term ‘grammar’ to refer to this mentally represented set of rules. In its ordinary use among linguists, the term ‘grammar’ refers both to the set of rules represented in the mind of a speaker and to the linguist’s description of that set of rules.
An important property of grammatical rules is structure-dependency. Sentences do not just consist of series of words like beads on a string. Rather, they consist of hierarchy of constituents such noun phrases (NPs), verb phrases (VPs), etc., each of which can in turn be broken down into parts. Grammatical rules do not refer to the linear position of individual words in a sequence, but rather manipulate sentence constituents. Given (7), for example, you might be tempted to say that the rule for forming yes/no-questions in English involves moving the first verb in the sentence to the front. (8), however, shows that what moves is the verb following the subject, regardless of its linear position:
(7) / a. / The boy will arrive soon.b. / Will the boy ___ arrive soon?
(8) / a. / [The person that you have seen] will come back soon.
b. / *Have the person that you ___ seen will come back soon?
c. / Will [the person that you have seen] ___ come back soon?
3. Narrowing the focus: competence vs performance
The foregoing discussion has shown that, for Chomsky, linguistics is an essentially mentalistic enterprise: the focus of linguistic study is the speaker’s mentally represented knowledge of language (in particularly, the rules comprised within the speaker’s mental grammar). To sharpen this focus, Chomsky establishes a fundamental conceptual distinction between competence and performance: ‘competence’ refers to the knowledge of language represented in the mind of an individual speaker; ‘performance’ refers to the behaviour in which knowledge is put to use. For Chomsky, linguists should focus upon the study of competence.
The competence/performance distinction enables Chomsky to abstract away from a multiplicity of factors that, whilst affecting linguistic behaviour, do not reflect the speaker’s underlying knowledge of language: e.g. memory limitations, distractions, processing errors, hesitations, etc. However, the degree of idealization involved in Chomsky’s characterization of competence has been the source of much controversy:
First, some linguists claim that, by pursuing excessively abstract theories of linguistic competence, Chomsky has made it impossible to explain how linguistic knowledge is put to use. These researchers argue that the study of linguistic competence cannot be altogether divorced from performance: competence models —they argue— must be compatible with plausible accounts of language processing in production, comprehension, etc.
Secondly, in a much debated statement of the competence/performance distinction (Aspects of the theory of syntax, 1965, pp. 3-4) Chomsky suggests that the study of competence should abstract away from variation. Many researchers, particularly within the sociolinguistic tradition, are vigorously opposed to this view: they argue that the knowledge of the patterns of variation accepted within a speech community is an essential component of a speaker’s linguistic competence.
Interestingly, Chomsky’s concept of performance is similar to Saussure’s notion of parole: both are behavioural. The Chomskyan concept of competence is, however, significantly different from the Saussurean notion of langue:
Langue / Competence- finite list of simple signs
- finite set of rules generating an infinity of complex expressions
- public/social
- individual psychology
For Chomsky, the proper object of study for linguistics is the individual’s competence. Chomsky regards social/public entities such as ‘English’ or ‘Spanish’ as not well-defined. E.g. if you attempted to write a ‘grammar of English’, where would you draw its geographical and chronological boundaries? Does the same set of rules underlie the language one hears in, say, Newcastle and Bombay?
4. Plato’s Problem, the innateness hypothesis, and Universal Grammar
We saw above that Chomsky’s ideas about language are crucially geared towards accounting for the fact of linguistic creativity. Another of Chomsky’s key concerns is Plato's problem. In linguistics, this term is used to refer to the tough questions raised by language acquisition. The essential problem is: how is it possible for children to acquire language (i.e. to construct a mental grammar) as easily and as swiftly as they do, given the impoverished and limited character of the evidence to which they have access?
Chomsky formulates several arguments to show that language acquisition cannot depend solely upon the primary linguistic data (i.e. the utterances to which the child is exposed):
First, the primary linguistic data to which the child has access fail to reflect the properties of mental representations of language in crucial ways. For example, the multiplicity of factors affecting performance (memory limitations, distractions, errors, etc) often result in the child being exposed to deviant utterances. More seriously, many aspects of linguistic structure are exclusively mental and have no physical manifestation; they are therefore not reflected in the primary linguistic data. For these reasons, the child’s primary linguistic data are to be regarded as impoverished in relation to the mental representations that need to be acquired; this is the poverty-of-the-stimulus argument.
Secondly, children only have access to positive evidence: namely, to a set of utterances in the ambient language. They are not supplied with explicit information as to what structures are not permitted in the target language. Observe, in this connection, that children acquire language without formal tuition. Notably, they altogether ignore adults’ attempts at correcting their ‘mistakes’. Thus, children succeed in constructing a grammar without access to negative evidence.
The arguments from the poverty of the stimulus and from the absence of negative evidence indicate that Plato’s Problem is a tough one. Chomsky's solution is to assume that a good deal of linguistic knowledge is in fact innate. Notably, he asserts that, by virtue of her genetic endowment, the child is born with a cognitive device that facilitates the acquisition of language: this is known as the Language Acquisition Device (LAD). Thus, the child does not come to language acquisition as a blank slate (tabula rasa); rather, she has a basic blueprint of linguistic structure.
Chomsky assumes that the innate, genetically transmitted LAD supplies the child with knowledge of universally valid principles of linguistic structure. For this reason, the LAD is also known as Universal Grammar (UG). As well as universal principles of language, UG offers the child a limited number of options (parameters) among which she must choose when acquiring her language.
Among the principles which, according to Chomsky, are supplied innately to the child by UG is structure-dependency (see above): thus, innate access to the principle of structure dependency prevents children from attempting to cast the rule for yes/no-question formation in English in purely linear terms.
Many linguists agree in principle with Chomsky’s innateness hypothesis, but, as was the case with the competence/performance distinction, the details remain surrounded by furious controversy. There is thus extensive disagreement as to the precise contents of UG. Nonetheless, there are several arguments that support the idea that humans possess an innate language faculty:
1. Species-specificity. Only human beings possess language. This suggests that there is some genetically encoded property of the species that is essential to the acquisition of language.
2. The speed and ease of first language acquisition.
3. Uniformity across populations in the rate and stages and acquisition. This suggests that the pace of acquisition is not essentially dictated by external stimuli, but rather follows an internal genetic programme.
4. The critical period. The ability to acquire language declines dramatically after puberty. This suggests that acquisition must take place while the internal genetic programme is unfolding.
5. Patterns of mistakes and nonmistakes. During language acquisition, children do not make certain mistakes that one would expect to observe if they were freely formulating hypotheses about the primary linguistic data: e.g. they do not formulate structure-independent rules for forming questions. The absence of such mistakes suggests that they are prevented by innate knowledge of universal principles of language.
References
Dahaene, S. (1997). The number sense: how the mind creates mathematics. London: Penguin Books.
Green, D.W. and others (1996). Cognitive science: an introduction. London: Blackwell.
Lyons, J. (1991). Chomsky. London: Fontana.
Maher, J., J. Groves (1996). Chomsky for beginners. London: Penguin Books.
Newmeyer, J. F. (1996). Generative linguistics: a historical perspective. London: Routledge.
Smith, N. (1989). The twitter machine. London: Blackwell.
Smith, N. (1999). Chomsky: ideas and ideals. Cambridge: Cambridge University Press.
1