Language and Mind and Noam Chomsky: Part 2

Chomsky’s Universal Generative Grammar

As noted in the previous chapter, Professor Chomsky’s thoughts about human language capabilities were constructed in opposition to two alternative, major points of view that he encountered early in his academic career, namely,

· Claude Shannon’s Information Theory and its application to problems in computational linguistics such as converting oral speech into written language, and

· the behaviorism approach adopted by the psychologist B. F. Skinner in his book entitled Verbal Behavior that sought to explain language acquisition in humans as the repeated application of a simple stimulus/ response/reward (S/R) mechanism.

As Chomsky explains: (Language and Mind, p 4)

To introduce a technical term, we must isolate and study the system of linguistic competence that underlies behavior… And this system of linguistic competence is qualitatively different from anything that can be described in terms of the taxonomic methods of structural linguistics, the concepts of S-R [Stimulus:Response] psychology, or the notions developed within the mathematical theory of communication or the [computational] theory of simple automata.

The sheer breadth of Chomsky’s argument encompassing these established orthodoxies was audacious. Structural linguistics was the dominant school of thought at the university Chomsky attended. Behaviorism had a strong hold on the study of Psychology in many colleges and universities across the country. And Information Theory had recently emerged, alongside cybernetics, as an important new way of looking at living beings and their ability both to interact with their environment and regulate their behavior and communicate with other members of their species. Chomsky’s view was that the creative aspect of language usage, the generative grammar that people use to construct (and interpret) sentences, contradicts both the stochastic model that Information Theory was based on and the reductionist Behaviorist explanation.

Chomsky didn’t simply argue that neither Information Theory nor behaviorism provided an adequate description of the flexible human language facility. He formulated an alternative theory that, in his estimation, fit the observable facts of language acquisition and language creativity better. In this section, first I will try to explain the basis of Chomsky’s objections to the prevailing models of human language he encountered. I will then proceed to consider in greater detail the alternative model he proposed in their place.

Before attempting to describe Chomsky’s position in more depth, however, it will help to understand Chomsky’s assessment of the role that grammar plays in the assembly of words into well-formed (i.e., consistent with the rules of grammar) speech acts. Contemplating what some of the grammatical rules for constructing sentences might be, Chomsky asks us to consider the following sentence “frames,” which, for future reference, I will label English Language Example 1:

a. John _____ Bill that he should leave.
b. John _____ Bill to leave.
c. John _____ to leave.
d. It is ______ that Bill will leave.

and the following past tense form of the verbs persuaded, expected and happened, all similar, and all very well-known and familiar to native speakers of American English. Observe that we can legally plug some of the verbs into some of the sentences, but none of the verbs fit into all of the sentences:

  Verb       Legal sentences     Illegal sentences
persuaded       a, b               c, d
expected       b, c, d              a
happened         c                 a, b, d

From rich examples like this, first published as an Appendix to Eric Lenneberg’s Biological Foundations of Language in 1967, Chomsky asserts that a full, operational set of grammatical rules for American English needs to be able to account for regularities of this type. (Chomsky, 2006) Clearly, the kind of “grammar” that Chomsky is talking about here goes well beyond the scope of the diagramming of sentence structure that I was taught in English class back in Middle School, or the prescriptive rules of English grammar that govern Subject-Verb agreement, for example. Although Chomsky never says so explicitly, the variability of usage illustrated here seems to have a semantic dimension, an assertion that Chomsky conspicuously avoids, but some of Chomsky’s better-known students (like Ray Jackendoff) have made explicit.

In addition, notice that when “persuaded” is used in sentence

a. John persuaded Bill that he should leave.
b. John persuaded Bill to leave.

the two sentences are roughly equivalent in meaning — if there is a significance semantic difference, it’s not immediately apparent to me. This equivalency on the surface suggested to Chomsky that both of these sentences were derived from the same root. Then a transformational rule was applied that directed replacing the phrase that he should leave with a simpler variant, namely, the infinitive to leave. In much of Chomsky’s early work on syntax, working primarily from examples of proper American English sentences like these, he tries to identify the deep structure of various root sentences and then enumerate the kinds of transformational rules that can be applied to them that generate their visible, surface structure.

No doubt, transformational rules like that do exist in written English. For example, we know there is a rule in English that is used to transform an original sentence into one that uses passive voice. Both variants again mean roughly the same thing, but their usage scenarios vary a bit. (For a punchier narrative when constructing the career and skills resume that we use in a job search, we are encouraged to use active voice.) Another example of a transformational rule is the inversion of the usual subject-predication order in English whenever you want to ask the musical question “Who do you love” The Chomskyan (Chomsky, 2006) position is that the root sentence “you love who<?>” is initially formulated and then a transformational rule is applied that reverses the subject and verb predicate. (Note that the case for some sort of transformational rule is bolstered by the fact that in speech the sentence “You love who?” (emphasizing word “who” to mark the sentence as a question) can be, for example, an appropriate response by a distraught Prince Hamlet to learning that his mother is sleeping with the murderer of her ex, the young man’s father. (Yikes! That Oedipal conflict might make for a good tale.)

These are some of the intricate rules of American English syntax that mainly play a role in written composition, where we learn the prescription against passive voice, for example, handed down through history by teachers of Rhetoric. I am not going to concern myself any more with them here, other than to note Chomsky’s early emphasis on deep structure and transformational rules, which then sparked additional academic research in the years that followed his development of a new paradigm for the structure of human language.

I want to focus instead on the unusual notion of a generative grammar that Chomsky identified that then led him formulate a mathematical framework that could model the capacity of humans to generate an endless variety of meaningful sentences in their native language, what Chomsky describes as the creative aspect of human language production. Formally, a language Grammar is represented by its Vocabulary, a set of terminal strings, a set of Production rules, and a set of Starter symbols, which serves as a model for computing the set of valid sentences that can be generated:

G=(V,T,P,S)

In discovering a set of production rules, Chomsky proposes to model the “creative” aspect of human language usage. We will look at how production rules are applied to a dictionary of words to generate meaningful sentences.

Note that instead of “sentences,” I will refer here instead to speech acts, to use the technical term popularized by the philosopher John Searle, one of the most astute critics of Chomsky’s formulation. Searle’s term explicitly includes oral communication, which is the primary language skill that human children acquire, before they are induced to learn to Read and Write. (Note: Searle, in turn, credits the English philosopher J. L. Austin, particularly in a lecture that was later published as “How to Do Things with Words” in 1955, for originally expounding on the notion of speech acts.)

An important observation is that the same set of grammatical rules is employed by the humans that listen to (or read) these sentences. The humans who are on the receiving end of a communicative act have the ability to recognize sentences as well-formed speech acts, which serves as a prerequisite to any effort to construe their semantic content. The fact that you are a native speaker of some human language implies that you are a Native Listener (a useful term I just made up) as well.

With reference to the creative aspect of language generation, Chomsky points out that the grammatical rules that native speakers have internalized allow the speaker to formulate sentences that they have never been spoken before. Indeed, they may generate sentences that have never been even uttered (or written) before by any human being. Here is where Chomsky claims the “blank slate” approach of children learning their native language solely through experience fails to measure up to this creative aspect of the flexible language usage skills that children readily exhibit. In effect, Chomsky has us metaphorically wandering the infinite shelves of Jorge Borges’ Library of Babel, endowed with a capability to understand sentences we have never heard before and constructing sentences that likely never existed before.

He also wants to separate this innate ability to marshal the production rules of grammatical language usage to generate novel sentences from the semantic knowledge of what words mean and which words can be put together in well- formed sentences. To make that assertion, Chomsky offers the following example sentence:

 Colourless green ideas sleep furiously.

which he very cogently argues is a nonsensical English sentence that, nevertheless, we can recognize as being properly formed, i.e., grammatically correct. From this example, he argues that it is important to consider the syntax of a language, which studies the application of grammatical rules to constructing speech acts, in isolation from semantic considerations. It is a point well-taken, but also inconsistent with the English Language Example 1 from Chomsky that I showed you earlier. (Reading Chomsky’s dense, technical arguments can be fun because he adheres to Emerson’s position that “a foolish consistency is the hobgoblin of little minds.”)

In summary, when Chomsky refers to meaning, namely, the semantic content of a message, his emphasis is on the generative process that the speaker (or writer) relies on to attempt to communicate a message. The same process is used in reverse by the recipient, working backwards from the message to the underlying interpretation and meaning. Within a language community, it should be clear that the speaker’s knowledge of grammar is used in the message generation phase. The message recipient necessarily relies on that same, shared understanding of grammatical rules to interpret the speech act that she receives and cares to understand the meaning of.