Jump directly to the Content
Jump directly to the content
Article

John H. McWhorter


It Takes Three to Tango

Neither syntax nor semantics maps the full richness of everyday speech.

Ray Jackendoff's Foundations of Language is a response to what the author sees as a crisis. In the 1960s, Noam Chomsky electrified theoretical linguistics with his hypothesis that humans possess an innate capacity for language. The heart of his conception was a linguistic "deep structure," invariant across the species, with individual languages varying according to how they translate this foundation into "surface structure," plugging in their own words and rules of grammar.

Showcase example of "deep" versus "surface": we are more likely to render Who do you want to see? as Who do you wanna see? than we are to render Who do you want to win the game? as Who do you wanna win the game? Now why is that so? Chomsky's theory proposed that if we look at the deep structure of the latter case, who immediately precedes to win—the verb that who is, if you think about it, the actual subject of in the sentence. (And if you are already having trouble sorting all this out, that's merely evidence of the efficiency of our innate linguistic capacity, which allows us to make such complex distinctions effortlessly, so long as we don't have to explain what we're doing.)

Thus what begins as You want who to win the game? becomes Who do you want to win the game? when who "moves" to the front of the sentence in surface structure. And the reason we are loath to say Who do you wanna win the game? is that when who moves, it leaves behind a "footprint" that blocks want and to from coming together as a contraction: Who do you want __ to win the game?

As originally proposed, Chomsky's deep structure was taken as synonymous with "meaning" itself. Psychologists and philosophers were fascinated with the possibility that linguistics had identified universal structures of meaning underlying the bewildering variousness of the world's languages. Indeed, the notion of deep structure, like relativity or the Uncertainty Principle in physics, quickly became unmoored from its context to be bandied about with abandon in pop intellectual circles. But to nonlinguists, whether scholars or civilians, "traffic rule" issues such as the wanna quirk were of little interest, while syntacticians were much more interested precisely in such details than in exploring the nature of meaning in a larger sense. The trend continues, and the promise Chomsky's hypothesis once held outside linguistics has long been dismissed as a mirage.

Meanwhile, many linguists have themselves rejected Chomsky's view and focused instead on meaning, or semantics. To them, the operations that fascinate the syntactician appear a needlessly elaborated, epistemologically suspect distraction, isolating linguistics from other disciplines. The semanticist notes that in the sentences The ball rolled down the hill and Beth rolled the ball down the hill, rolled is identical in terms of grammar. But the semantics are different: in the second sentence, rolled refers to someone having caused the ball to roll, rather than it just rolling itself. That is, there is an underlying element of causation that expresses itself throughout any language's grammar, and semanticists point to other such elements as well. There is more for the philosopher or psychologist to grab onto in this approach; it seems to address what we think of as "language" more directly than fretting over why we don't say Who do you wanna win the game?

The unseemly warring between the two sides that raged throughout the 1970s has today devolved into a sullen stalemate. The syntacticians hope that the traffic rules alone will explain most of the quirks within languages and the variations between them. Paying minimal attention to the messiness of real-world meanings, they tend to dismiss those who do as "unscientific"—even when overall their work addresses but a sliver of what most of us would call "language" at all. Meanwhile, semanticists pride themselves on forging a more "realistic" or perhaps even humanistic model. But at the end of the day, few of them attempt to demonstrate how their strategies explain the relevant facts better than syntactic ones, such that their work often seems tangential to what most would consider "linguistics" at all.

Jackendoff bemoans this scarcity of cross-fertilization and fears that the syntacticians in particular—victorious in America due largely to Chomsky's charisma and influence—have grown too introverted to communicate effectively with other thinkers. In this book he fashions a compromise, informed by decades of celebrated work in both syntax and semantics. The book is a magnum opus seeking a model of language compatible with how humans process and produce it online.

Jackendoff criticizes the "syntactocentrism" of Chomskyan work, where the traffic rules are the driving force in how we speak, with other aspects, such as meaning and how we translate thoughts into actual sounds (phonology), as "garnish." But Jackendoff also notes that in practice, syntax and semantics alike are too limited in what they map for either one to be seen as central. In the sentence The chair has a stain on it, the syntax involves a subject (The chair), a verb (has), an object (a stain) and then what is called a prepositional phrase, on it. But this last bit doesn't readily offer anything to the semanticist: what the sentence "means" is that there is a stain on the chair. The on it adds no additional element of "meaning." Chalk up one for the syntacticians.

But then, when the waitress says The ham sandwich over in the corner wants more coffee, her referring to a person by means of what he or she ordered requires a feint of abstraction that has nothing to do with the fact that The ham sandwich is the subject of the sentence. If linguistics seeks to identify how we translate the world around us into speech, then a theory that dismisses things like this as "beside the point" will deserve the charge of navel-gazing.

Jackendoff argues that language is based on three modules of equal importance. The first step in saying something is not the syntactic but the conceptual module, the home of meanings: an utterance begins with the thoughts that it consists of. Syntax comes second: here thoughts are arranged into sequences of words. Jackendoff's conception of syntax is a novel one. Linguists traditionally distinguish between "rules" that we apply regularly (such as "add-ed to a verb to make it past) and things that must be "stored individually" because they are unique, such as individual words or irregular verb forms like bought or went, that we cannot create by applying the regular rules.

Jackendoff discards this division and treats as "stored" everything from words, to idioms like kick the bucket (whose meanings have so little to do with their words that they are essentially "words" in themselves), to "constructions" using words in unusual ways (He belched / cried / spent up a storm), to regular rules like the -ed one.

Indeed, Jackendoff pointedly warns against our temptation to treat such idioms and constructions as mere "static" when any language in fact has thousands of them—they are central to speaking. For him, they constitute a continuum between the extreme poles occupied by words on one end and rules on the other. All of these things are stored in the memory and called upon to create utterances when combined, in an overriding process called "UNIFY." Thus a word (walk) is combined with a rule here ("add -ed for past"), or plugged into a construction like the up a storm one there, and what is universal and innate is the broad default tendencies in how such elements are combined.

Finally, the phonological module produces the sequences of sounds that we actually pronounce. Just as semantics and syntax work independently too often for either to be treated as central, our sound system has a way of working "against" the other two. For example, when we say The man's coming, the 's, a truncation of the verb form is, hangs on the noun man instead. If syntax were "the real deal," then the 's would hew to its fellow verb coming. It is the exigencies of English's sound system that prevent us from keeping nouns and verbs in their separate corners and saying The man scoming.

In light of burgeoning examinations of how language, if innate, could have evolved via natural selection, Jackendoff also argues that his conception is more compatible with evolutionary theory. Namely, the modules he proposes appear to recapitulate an evolutionary sequence for language that we can deduce from various types of evidence.

Aphasics, for example, lose grammatical capabilities but retain full memory of individual words, suggesting that language begins in the conceptual module, with syntax coming afterward. Primates can be taught a rudimentary sign language, where basic concepts of meaning can be strung together—which suggests not only that meanings come first but also that what distinguishes us from primates is the other two modules: apes can think the thoughts but have no way of translating them into speech.

Even particularities within modules appear confirmed. Jackendoff, like many, divides syntax between the "phrasal" (word order) and the "morphosyntactic" (endings like -ed or Latin case endings). He hypothesizes that the phrasal comes first in producing an utterance, and indeed, aphasics often lose the ability to handle endings while still controlling word order.

Despite Jackendoff's clear, engaging prose, Foundations of Language is written largely for academic linguists, psychologists, and philosophers. The author's accommodations to the reader are generally just those necessary for the latter two groups, and overall one needs a basic familiarity with modern linguistics to follow the text. The arc of argument also flags a bit in the final third on semantics, where he is concerned more with indicating what schools of thought he finds cogent than fashioning his own reinterpretations. In such a wide-ranging and richly considered work, however, this is inevitable. Jackendoff compensates with cogent suggestions for further research throughout the text, with these especially frequent in this last section.

It is a truism in linguistics that books are used only for reference while articles truly drive debate. That is unfortunate for this book, because Jackendoff's teachings are some of the sharpest, most comprehensive ones on linguistics currently available. He is a boon to the field, and I wish all linguists would sit at his feet by reading this book.

John H. McWhorter is associate professor of linguistics at the University of California at Berkeley. He is the author most recently of The Power of Babel: A Natural History of Language (Times Books).

Most ReadMost Shared