Contemporary Linguistics and Cognitive Science

Noam Chomsky
Sobel S.P. The Cognitive Sciences:
An Interdisciplinary Approach. – London; Toronto:
Mayfield Publishing Company, 2001. – pp. 159-167.
The impetus that set the field of linguistics on its current path came from the publication of Chomsky's Syntactic Structures (1957) and Aspects of the Theory of Syntax (1965). These works ignited a revolution in linguistics, placing it squarely back into the domain of the mind and determining the direction it has followed ever since.
What Chomsky contributed initially was a shift of focus to the (vast and largely unconscious) set of rules he hypothesized must exist in the minds of speakers and hearers in order for them to produce and understand their native language or languages. Like Pвnini, he was concerned with discovering, isolating, and pinpointing these rules, to make their formulation precise and predictive. But, as a 20th-century researcher, he was working within the contemporary framework of science. Scientific effort requires abandoning vagueness in favor of focusing on the observable specifics, which alone lead to productive hypotheses. But unlike the behaviorists, Chomsky based his hypothesis on the assumption of a capacity in the brain that functions without the conscious awareness of the person in whom this functioning is taking place, and which it is indeed possible and profitable to study. The data provided by language permit us to infer what must be taking place as language is produced. In the process, Chomsky proposed a method of formalizing the rules of the components of language. In view of the impact on and pervasiveness of this approach in linguistic research in the second half of the 20th century, a brief introduction is in order.
The first component of language Chomsky addressed was the syntactic component—the portion of one's linguistic competence that handles the arrangement of words into sentences. A simple sentence serves as an example of what formal rules must contain if they are to be capable of generating such a sentence:
The cat chased a mouse.
This sentence contains five separate words, some of which—the cat, a mouse— "feel" as though, when taken together, they form a somewhat larger unit. The words in each grouping must occur in this order *cat the and *mouse a are not permissible English combinations. (The asterisk preceding each such formulation is, by convention, a sign that what follows is not grammatical in the language.) It is also true that in English one or the other of these combinations may come first and the verb, in this case chased, must come between them. The following ordering would also be fine for English, though it expresses a somewhat unusual situation:
A mouse chased the cat.
Also perfectly good sentences of English are these two:
A cat chased the mouse.
The mouse chased a cat.
A rule that would specify that these four orderings are just those that are permitted for this set of words would have to refer to the part of speech each word represents. These sentences demonstrate that nouns may occur both before verbs and after verbs and that articles, when present, must be placed before the nouns they refer to. But the rules would also make clear that not all sentences contain nouns that are preceded by an article:
Babies cry
is a perfectly good English sentence, yet there is no the or a before babies. Nor, for that matter, is there a noun after the verb. So the rules would specify that a verb need not be followed by a noun.
The rules Chomsky formulated making all of this explicit are written, in their most basic form for the simplest of sentences, as follows: Letting S stand for the sentence, N for nouns, V for verbs, and Art for articles, and an arrow, > , for the way in which S can be expanded to include its elements,
S > (Art) N V (Art) (N).
S can be rewritten or expanded as (i.e., the sentence contains) an article followed by a noun followed by a verb followed by an article followed by a noun—in that order. Articles and the noun following the verb are placed in parentheses to indicate that they may or may not be present in the sentence. The first noun, which serves as the subject of the sentence, must be present, as must the verb.
A slightly more complicated sentence might contain another element:
The white cat chased a frightened mouse.
A brave mouse chased the small cat.
Now we must accommodate adjectives. In English, when an adjective is associated with a noun, it occurs before the noun. Modifying our rule to allow for this, we can write
S> (Art) (Adj) N V (Art) (Adj) (N).
But we know that a noun may have more than one adjective associated with it. Therefore, we need a symbol to indicate that indefinitely many adjectives may occur before a noun. To make this clear we place an asterisk after Adj: Adj*. (A moment's thought will suffice to convince you that there can be only one article preceding a noun.) Our rule now looks like this:
S > (Art) (Adj*) N V (Art) (Adj*) (N).
This is not a very economical formulation, because it repeats so many of the elements. It may be condensed, because in principle a noun will always have the possibility of a preceding article and indefinitely many adjectives. If we call this combination a noun phrase, and abbreviate it NP, we can write our rule this way:
S > NPV(NP).
Now we must write a rule that expands NP:
NP > (Art) (Adj*) N.
A sentence can be divided up intuitively much as a noun phrase can, into components that seem to "go together." The white cat forms one part of the sentence, that which is being spoken about—the subject. Chased a frightened mouse forms the other part of the sentence, that which is being said about the subject— the predicate.
The white cat chased a frightened mouse at high speed into the grassy yard.
To capture the intuition that the sentence breaks into two major parts, we can recognize the status of the second part by calling it a verb phrase, or VP. The rule that produces, or generates, sentences can now be stated in the condensed form
S > NP VP.
Our latest sentence now contains two additional phrases, at high speed and into the grassy yard. Each of these, as you can see, contains a preposition, abbreviated Prep (at, into), followed by a noun phrase. This we can categorize as a prepositional phrase, abbreviated PP which we formulate as follows and add to our list of rules:
PP > Prep NP.
As the sentence indicates, there is the possibility for an indefinite number of prepositional phrases following the verb.
Each part of our rule for generating sentences can be expanded by writing the rules we have formulated for each element, giving us the following set:
S > NPVP.
NP > (Art) (Adj*) N.
VP > V(NP) (PP*).
PP > Prep NP.
The NP in this last rule can of course be expanded by means of the already stated NP rule.
It must be understood, of course, that these particular rules apply to English sentences only. The rules for generating the sentences of other languages would require a different formulation. German, for example, would require the verb to occur as the last element in the verb phrase.
So far we have formulated only four rules. Now, having formulated the rules, we must include the words themselves in order to generate the sentence. Each category they involve—nouns, verbs, articles, prepositions—can be further expanded to include the words that constitute that category. Thus:
N > [cat, mouse, speed, yard . . .]
V > [chased . . .]
Art > [a(n), tfie]
Adj > [white, frightened, high, small, grassy . . .]
Prep > [at, into . . .]
You can easily see that a very great number of English sentences can be generated by means of this small set of rules, using the many other words that fit into the categories N, V, Adj, and Prep. (The category Art is different; there are only two articles in English—an being a variant of a.) However, if you consider sentences such as the following, you will also see that there is much more that must be accounted for in English sentences:
Whose mittens are these?
I disagree entirely.
I don't want you to go out tonight.
Get lost!
Let's see the crossword puzzle you have just finished.
Kenny told the girl who came to pick up the books that she couldn 't have them because he wasn 't finished reading them yet.
This will give you some idea of the vastness of the task of formulating the syntactic rules of a language—and we haven't even mentioned the rules for making the words sound right for the language or for constructing the words or for deriving "Jeet jet?' and 'No, joo?' from "Did you eat yet?" and "No, did you?"!
Let us look briefly at an example of unconscious rules of the sort Chomsky sought to formalize, drawn from the phonological component of your linguistic competence, that portion that deals with the sound system of your language. One example will suffice to indicate that the phonological component is also vast and complex. Sounds of a language that are identified by its speakers as being "the same," such as the two instances of the sound p in the English word paper, are in fact pronounced in a somewhat different manner and thus sound somewhat different. You can demonstrate this yourself quite easily. Hold a sheet of paper up to your lips as you pronounce the word paper; you will find that the first p carries with it a puff of air that blows the paper away from your lips, whereas the second p is said without this puff of air. Linguists refer to the puff of air as aspiration, calling the first p "aspirated" and the second one "unaspirated." The variation in their articulation depends on their position in the word: The sound p becomes aspirated when it occurs at the beginning of a syllable and is immediately followed by a stressed vowel (as in the first syllable of paper). When it is in any other position, it does not. Native English speakers never make the mistake of using one version where the other belongs. Yet they are generally unaware that they are using two versions of the sound. Thus we can speak of an underlying notion of the sound, which is stored in the speaker's brain as part of the pattern of speech sounds of his or her language. Just as water can exist as liquid (water), solid (ice), or gas (steam), so too are many of the sounds of language manifested differently in different environ-ments, as the two versions of p in paper clearly show. The variants of a given sound, taken together, constitute a phoneme. A phoneme, then, is a kind of abstraction.
The sounds t and k are articulated in very much the same way as p—that is, without voice and by closing off the passageway in the mouth that allows the sound to be uttered on the expelled breath. (P does this at the lips, t with the tip of the tongue behind the top teeth, and k with the back part of the tongue touching the roof of the mouth. This can easily be demonstrated by simply saying the sounds.) All three of these sounds behave the same way, following the same rule with respect to aspiration. Following a convention that indicates aspiration by means of a superscript ?, it is possible to write a rule that expresses this situation:
p, t, k become p?, t?, k? when they occur at the beginning of a syllable and are immediately followed by a stressed vowel.
By means of a set of symbols and terms that capture the commonalities among the three sounds, the conditions under which the rule is applied, and the result of its application, a very economical formal rule can be written. (To explain all the complications involved in arriving at such rules would require a course in linguistics. My purpose here is merely to suggest their complexity.)
The Relevance of the Rules to Cognitive Science
Perhaps your head is spinning from this discussion of the formalization of rules of English. Perhaps you are wondering why it was included. There are several reasons. First, to appreciate what was involved in the new linguistics of the 20th century, it is necessary to have at least some understanding of the turn it took in the direction of scientific inquiry and method. Second, establishing a formal means of encoding the rules of language enables important generalizations to be grasped (such as the one that extends the effect of the aspiration rule from one sound to all sounds made in the same manner). If we can capture in this way the rules that characterize languages, we can compare them to see what types of rules characterize human languages in general. From there we can proceed to a greater understanding of what the human brain is equipped with that enables it to "do" language.
Another reason for making the rules explicit is to increase our understanding of the" way children learn their first language. If it is indeed by means of acquiring such rules (however they are represented in the brain), we can more readily understand how it is possible for them to do it in the short time it actually takes.
Still another important reason to formalize the rules of language has to do with the capabilities we are developing, via the computer, to model aspects of human intelligence. Computer programs require very precise and unambiguous instructions. The formalization of the rules of language has enabled computer scientists involved in artificial intelligence to attempt to model human language on the computer.
Identifying and formalizing the rules of language rests on the assumption that they have been internalized, represented in some fashion in the brain — that they are in fact in there, somewhere. Regarding the issue from this perspective requires that we consider how it is possible for an infant to begin to acquire these rules, on the assumption that it is born not knowing them. This assumption is reasonable, because babies will learn to speak the languages spoken in their environment. (If a child is adopted by a family of a different culture from that of the biological parents, where a different language is spoken, the child learns the language of the adoptive culture.) We wonder, of course, how it is possible for the linguistic system of a given language to be absorbed, as it demonstrably is, early in childhood. Is the capacity to "do" language a specialized one, residing in brain functions evolved just for that purpose? Or is it rather, as some have proposed, one among many abilities that arise from certain general cognitive abilities, such as the ability to categorize experiences of our environment (this is a chair; that is a person; this word is a potential sentence subject, but that word can only designate an action)?
Chomsky's hypothesis was that the inborn linguistic capacity of humans is sensitive to just those rules that occur in human languages—and in no other sorts of language or system. In other words, something quite specific to language has evolved in the human brain, something that enables the steady stream of language sounds to be perceived by the infant as distinct in an important way from other sounds and that also enables the infant to begin to break down the flow of this stream into meaningful parts, arranged, as the infant somehow recognizes, in a systematic way. Chomsky likened this capacity to the capacity to walk, in the sense that it is a behavior in which humans naturally engage, without the need for special instruction, as soon as they are developmentally ready—provided the environment permits. That is, walking proceeds if the toddler has undergone maturation sufficient for growth of bones and muscles, and language development occurs if the environment provides exposure to language. Both are part of our biological endowment. It is thus no accident that there are many similarities among languages: The many systems, or grammars, that underlie them are ultimately generated by the human brain.
That there are as many different grammars as there are languages is evident. But linguists cannot discover the grammar of any one of them without careful study of the actual language as it is spoken. At the outset one might propose any number of possible grammars to account for the speech heard in a given language. How many sets of rules might one be able to think up that would generate the things people say? Think for a moment about how many explanations one might propose for how a machine—say, a car—works, in the absence of actually looking inside to see the components and what drives their operation. One might come up with many explanations, ranging from "There are forty little guys in there pedaling for all they're worth" to very complicated schemes for having one element move another, which would then move another. . . something like a Rube Goldberg contraption. Someone might even hit on the correct explanation, and we who actually know how the car works know that there is only one correct explanation.
So it is with language. There are many possible grammars that would serve to produce a given language. But if we adults cannot peer inside the mechanism and learn which is the correct one, how on earth does a small child find it on his or her own? Even if we are willing to assume, with Chomsky, that what the child brings to bear is an innate capacity to project the right grammar on the language spoken in the environment, we might not wish to endow—to overburden—this child with all the grammars to select from that would be necessary for all languages. But if babies do not know in advance which language they will be confronted with at birth, how else could they possibly find the right grammar?
The hypothesis proposed to answer this question is that despite their surface differences all human languages share a fundamental structure, and what is common to them all has come to be known as universal grammar. Innate in all of us, according to this hypothesis, is the ability to apply this universal grammar to whatever languages we are faced with at birth. This explanation is not universally accepted; some have maintained, for example, that the human capacity for language is not richly specified in the brain but is rather a special function of the general cognitive abilities humans possess.
Chomsky's early hypotheses regarding the nature of the human linguistic capacity date from the mid-1950s. During the next 20 years or so, researchers following that line of thinking worked with the kind of explanation discussed earlier in this chapter—that is, that each of us is endowed with an innate capacity to respond to rules of language of the sort I have described. But there are so many languages, and each has its own set of rules. How is it possible for a brain to possess innately a system adequate to the task?
The approach then taken by Chomsky and others was to attempt to "factor out" general principles that hold for all languages, principles that govern application of the rules of languages. Under this new principles and parameters formulation, which crystallized around 1980, it is these that constitute the universal grammar. Variation in languages results from the ways in which these principles apply. There is a finite set of ways in which the principles may apply; these are the parameters (Chomsky, 1995). The parameters have been likened to a set of switches, each having a fixed range of potential settings. The actual language the learner is exposed to provides the data that trigger the setting of the switches. Under the principles and parameters hypothesis, learning the syntax of one's first language is a matter of setting these switches; acquiring language is a process of fixing the parameters in one of the permissible ways. With no requirement that the innate component specify all manner of language-specific rules of the sort described previously, the hypothesized innate machinery of language can be reduced.
Whichever way one looks at it, some of the questions we find ourselves asking regarding the language capacity of humans are:
How is language organized in the brain?
How does it work?
If a person speaks more than one language, how are these languages stored?
Once the rules of a given language have been internalized by the child, how
and where are they represented in the brain?
We cannot take the direct path and look inside for the answers, because even when the brain is exposed to view for medical reasons, as in Penfield's work, one sees neither rules such as VP > V (NP) (PP*) nor principles and parameters lurking there in some recognizable form. But the recently developed sophisticated methods of scanning the brain, as we saw in the preceding chapter, allow us to observe brain activity during the performance of language-related tasks, among others. Nonetheless, we see no rule hopping about the cortex clamoring to be recognized.
There are, however, experiments that do provide us with information. Brain-damaged patients can serve as subjects of such experiments (Damasio & Tranel, 1993). These experiments indicate that certain brain structures link areas of knowledge about traits, sounds, and movements of birds, for example, with the nouns and verbs associated with them. Subjects in the experiments could describe, using verbs, what was taking place in pictures of activities involving birds, but did much worse than non-brain-damaged controls in using the appropriate nouns, such as duck, ostrich, and other bird names. Another conclusion to be drawn, according to the experimenters, is that areas of the brain that handle nouns are not the same as those that handle verbs.
In the next chapter we will turn to the sorts of experiments that have been carried out in recent attempts to understand the acquisition and functioning of language in our species.

Внимание, отключите Adblock

Вы посетили наш сайт со включенным блокировщиком рекламы!
Ссылка для скачивания станет доступной сразу после отключения Adblock!

Скачать