Get PJ Media on your Apple

Deconstructing Chomsky

If Noam Chomsky’s linguistics are problematic, can his politics be far behind?

by
David Solway

Bio

May 21, 2011 - 12:55 am

Chomsky starts us off with his definition of a language: “Language is a set (finite or infinite) of sentences, each finite in length, and constructed out of a finite set of elements.” The sentence itself comprises a noun phrase followed by a verb phrase. Perhaps he should have stopped there. For he goes on to elaborate a complex “phrase-structure grammar” which he diagrams as a tangled forest filled with inverted trees whose branches consist of various parts of speech fit for a tribe of swinging monkeys. The name of this jungle is “Transformational Grammar.” It purports to map what Chomsky calls the “deep structure,” “underlying strings,” and “recursive properties” common to all the world’s languages and innate to all the world’s speakers. These features are supposed to provide for an economy of grammatical rules that prescribe the convoluted operations by which coherent sentences are constructed. His wielding of Ockham’s Razor, however, seems to produce a lot of unnecessary bleeding and more stubble than is desired. As Judith Greene points out in Psycholinguistics: Chomsky and Psychology, the ramification of syntactic rules in the mind of the individual speaker would “start generating strings at random [and] would obviously be wildly uneconomical.”

Moreover, his theories keep changing — though one would assume reality does not. Aspects of the Theory of Syntax significantly emends the earlier Syntactic Structures. He later went on to revise his ideas even further, jettisoning deep structure and postulating a “universal grammar” deriving from “simple” computational laws — hence The Minimalist Program — which do nothing to pollard his earlier arboreal speculations. Indeed, deep structure keeps its place in the popular domain as Chomsky’s chief contribution to the psychology of language.

When he tells us that the branches of a phrase structure, unlike the links of a word-chain device, act like a kind of blueprint for the finished sentence, we have to take it on his authority. Similarly, what is called a “finite-state grammar,” to quote Steven Pinker in his explication of Chomsky in The Language Instinct, “is just one damn word after another, but with a phrase-structure grammar the connectedness of words in the tree reflects the relatedness of ideas in mentalese.” Really? A formulation of this nature is actually a dictum, and though a hierarchical tree diagram may seem convincing, the map is not the territory. It may not be that way in reality.

What we get in Chomsky is more like a description of what might be or should be the case, but not an explanation of what manifestly is the case. The diagrams function in a certain way; therefore the things they are diagrams of must also behave in the same way. Because the branches of his tree drawings describe arches and parabolas under which the various phrasal segments of a sentence are combined, it must follow that long-distance dependencies are plainly accounted for, that is, that the sentence “remembers” later what came before in the deep structure to produce grammatical agreement among its terms. It’s a nice idea and the term “deep structure” continues to resonate. (Ringing phrases can lead to the belief in entirely phony constructs — think of “Oedipus Complex” and “The dictatorship of the proletariat.”) The trouble is that the term refers exclusively to the idea; neither is confirmable by data gleaned from outside the magic circle. They indicate a hypothesis, not a fact.

In the same way, we are told that parts of speech are not a kind of meaning but, let’s say, they are like Lego pieces that fit one another in prefabricated ways. A given part of speech, for example, a noun, is simply like an item in a menu that must obey certain sequences. Similarly, certain verbs may enter a phrase structure lawfully only by conforming to modular rules or obeying certain formal parameters. But these operational procedures can surely be replaced by the meaning of the word itself (or what this school of thought rather pompously calls the “semantic component”). One can argue that a transitive verb requires an object or an embedded sentence, not because it is following a nexus of prior and invisible rules, but because the sense of the verb in itself requires that you complete the potential or specify the general meaning of the verb in its dictionary acceptation.

In Aspects of a Theory of Syntax, Chomsky tries to get around the dilemma by adding a new constellation of base rules, called a “lexicon,” that would somehow allot meanings to words and sentences. It simply won’t wash — one doesn’t repair a theoretical lesion by simply inventing a prosthesis and tacking it on. Syntax — the order of words — allows us to interpret the overall meaning of a sentence but it cannot output definitions, without which syntax is perfectly helpless and contentless. Analogously, in The Minimalist Program, Chomsky plucks out of thin air a faculty he labels a “parser,” which “assigns a percept to a signal” and which, mirabile dictu, “presumably incorporates the language and much else.” Such maneuvers are plainly illicit and imply, rather, that neither deep structures nor a universal grammar could possibly contain all the information necessary for the semantic interpretation of a sentence.

No less troubling, the notion of Universal Grammar may be correct in the trivial sense that every language has a grammar and is learnable, but whether Chomsky’s scaffolding of rules and “parameters” would apply to all the world’s languages, for instance, Mandarin or Hausa or Barikanchi or Hopi or Nootka (the latter, according to Benjamin Lee Whorf in Language, Thought and Reality, has no parts of speech), is another question entirely. Many scholars are profoundly skeptical whether Chomsky’s enunciations apply across the board even to English and some have furnished strong evidence that they do not. (See, for example, Peter Seuren’s Western Linguistics.)

As Raymond Tallis writes in Not Saussure: “Chomsky’s methodological tactic of treating language as primarily a syntactic structure has led linguistics to an impasse.” The central problem that Chomsky has failed to address is that a “context-based intuition of the speaker’s intentions is necessary not only to determine the meaning but also the grammatical structure of what has been said” (emphasis added). Tallis’ reasoning is at least as persuasive as Chomsky’s — but Tallis has not undertaken to found an academic discipline or create a dendritical pseudo-science.

Tallis’ allusion to “context-based intuition” strikes very much to the heart of the matter, especially when we consider that Chomsky does not adequately distinguish a sentence from an utterance. “The fundamental aim of the linguistic analysis of a language L,” he writes in Syntactic Structures, “is to separate the grammatical sequences which are the sentences of L from the ungrammatical sequences which are not sentences of L and to study the structure of the grammatical sequences.” Syntactic competence, he states, is reflected in “performance”; unfortunately, not all performances would satisfy the criteria he assembles.

Consequently, when he tells us in The Minimalist Program that language is “embedded in performance systems, which access the generative procedure,” he only muddies the waters. For the fact is that people do not speak in sentences and their verbal expressions are often quite perceptibly ungrammatical. Yet the strings of words, gaps, ill-formed sequences, missing suffixes, disheveled syntax, wrong auxiliaries, morphological aberrations, and improper formations in everyday speech are readily comprehensible. Context, and intuition gained from experience, appear to do the work. As former president of the American Psychological Association, the late Charles Osgood, commented in his Lectures on Language Performance, “the situational conditions to which [speakers] are responding are perceptual and cognitive rather than linguistic.” This at least makes sense.

I have merely skimmed the surface but readers who are unfamiliar with Chomsky’s modus operandi can plunge into his scholarly texts for themselves. What they will discover is that Transformational Grammar (no more than Universal Grammar) gives us not an explanation of how language works but a bundle of descriptions of formal and diagrammatic processes accompanied by a glossary of definitions and reams of alphabetic formulae. What is explained is the description, not the thing it is a description of. To use Chomsky’s own terminology, but in a manner he did not intend, his theories enjoy a kind of “descriptive adequacy,” which boomerang back on themselves; their “explanatory adequacy” applies to the description as — at best — a conceivable but not necessarily an actual mechanism for generating an infinite set of grammatical sentences.

In Aspects of the Theory of Syntax, for example, Chomsky lays it down that “a system of rules that in some explicit and well-defined way assigns structural descriptions to sentences” is grounded in “mental processes that are far beyond the level of actual or even potential consciousness.” And here is the predicament. Chomsky’s “system” is, in many respects, pretty well incoherent, but even if it happened to be coherent, there is still no way of determining that it would be valid or that these mental processes demonstrably exist. Put succinctly, Chomskyan psycholinguistics is not a science, but an elaborate Rube Goldberg machine that mobilizes enormous resources to get very little done. Nevertheless, many of us are seduced by an intricately latticed diction and dazzled into submission by indomitable complexity.

Click here to view the 83 legacy comments

Comments are closed.