Where Does Language Come From?

Rarely does the question of the origins of human language provoke serious attention from linguists. It is not that the subject is a trivial one, or that it lacks a history of speculative debate. But discussions of how language originated tend to pass over the very features of human language that make it human. They deal neither with its formal lexico-grammatical structure nor with why languages make particular semantic distinctions in particular syntactic ways. How did it come about, for example, that inflection, word order, and morphology do so much work in so many languages; or that virtually all languages make such neat work of such knotty matters as noting causativeness and transitivity or distinguishing what linguists call “unmarked,” or ordinary, expressions from marked, extraordinary expressions?

If language seems like a miraculous, proliferating tree, most guesses about its roots have been notoriously unproductive and singularly dull. Did language derive from gestures (to pick one of the more recent dreary proposals)? Well, why not? But how did the gestures take on the complexity of a linguistic system? The same can be said about the claims that language emerged from “grunts and groans” and “yo-heave-hos” and other such sounds. Small wonder the Société de Linguistique of Paris banned the topic of origins from its meetings in the late nineteenth century. Besides, questions about the inherent structure and systematic qualities of particular languages—what Saussure in the opening years of this century called the “synchronic” issue—elbowed aside nineteenth-century concern with “diachronic,” or historical, matters, among which were those relating to origins.

But for all the banning of the subject and the boredom with it, speculation about the roots of language has a way of resisting extinction. In recent decades, diachronic linguistics has been given little attention but interest in “roots” has been kept alive by the enormous growth of work on the child’s acquisition of language deriving from Chomsky’s bold hypothesis that language is innate. Perhaps the Cartesian miracle that was alleged by Chomsky to produce language in the child could produce it in the species as well—perhaps a universal, deep-structure, formal grammar has been built into the mind, through which human beings could first invent language and then recognize the local rules for speaking their mother tongue properly, once such a specific local language had come into being. But nobody was prepared to assert that some languages were, so to speak, living, primitive fossils. Efforts like Joseph H. Greenberg’s to catalogue the universals of all known living languages helped little with the issue of origins, though they went some way toward sparing us from our own linguistic parochialism. If you subscribe to the idea (still almost universal) that all languages are equally advanced, then how could an account of what they have in common tell you how it all started?

And then of course there are chimpanzees. But what did research on them tell us, except that these amiable pongids could, given human tutelage, pick up some of the logic of linguistic distinctions? They do not do so in the wild on their own—even if one captive in Oklahoma spontaneously used sign language to call a Brazil nut a “rock berry.”

There might, therefore, seem no reason for excitement when a little-known linguist comes along with a book on the roots of language. So much for prediction based on past scholarly boredom. Even if Derek Bickerton’s book were wrong in its major details (which is probably not the case), it would still be an original and important contribution to this vexed subject. It is the only recent book we know that addresses itself to the possible origins of real language and offers hypotheses about why, in view of the vast number of possible forms language might have taken, it took some of the forms that it did.

The form of Bickerton’s argument and its use of evidence are ingenious, bold, and in the end bothersome—but no part of his case can be ignored. There are occasions, Bickerton argues, when human language is reinvented, when linguistic Big Bangs take place. These are the occasions—often surprisingly sharp in onset and short in duration—when creole languages emerge out of pidgin languages (of which more in a moment). But there are also “little bangs,” occasions when invention occurs in the child’s acquisition of language, as when novel syntactical forms not found in his mother tongue (and therefore not “imitatable”) emerge in his speech, later to disappear when his speech becomes more standard. Nobody but Bickerton has ever thought to compare systematically these inventive “errors” in the early speech of children with the inventions of creole speech.

A word first about “creolization,” an extraordinary linguistic phenomenon that until now has hardly entered the contemporary discussion of the “deep structure” of language. In the linguistic sense “creole” does not refer to specific cultures such as those of Louisiana or Haiti, but to all languages that arise from pidgins. Pidgins are almost structureless, make-do, rather unstable “languages” that evolve to handle limited social exchange between speakers of different standard languages who find themselves living in contact with each other. This happened when slaves in antebellum America who spoke different languages in Africa had to work and live together, or more recently when sugar-cane farm laborers were thrown together in Hawaii. Many of these people spoke only Japanese, Tagolog, Chinese, Portugese, or Hawaiian, in a society whose “superstrate”—i.e., official and dominant—language was English. In some generation subsequent to the contacts that produced the pidgin, the children of pidgin speakers began to speak a general-purpose language, a creole, which henceforth became their mother tongue and the mother tongue of their children. The creole that emerged has a stable structure and syntactical rules that often differ both from those of all the so-called substrate languages of the initial pidgin speakers and from the structure of the superstrate language. The emergent creole has a structure and rules which (unlike the limited original pidgin) serve the full range of linguistic functions that any mature language will be found to serve.

That transition is a remarkable one. But from the opportunistic point of view of the linguist in pursuit of origins, what is unusual is that unlike ordinary languages whose history is necessarily lost in the unrecorded oral past, both the predecessor pidgin and the newly emergent creole are often recorded and available. Many pidgins were first produced in response to modern colonialism and, after a few generations, became genuine creoles. Some of these creoles were so recently formed that they are virtually unchanged today. So it is possible to find fairly well-documented, widely scattered creoles that arose from pidgins whose speakers came from different dominant or superstrate languages. A researcher in a Land Rover can find an informant whose speech embodies new creole rules, and ask him about a list of examples and counterexamples. On the way back to town, he can ask a pidgin speaker of the preceding generation who does not speak that creole how he would say the same thing. What do the creoles have in common, and how do they differ from the pidgins that preceded them?

Take as an example the word order for deploying Subject, Verb, and Object in a sentence in Hawaiian Creole English. Hawaiian Creole permits orders of SVO that are meaningless in Hawaiian Pidgin English (and in most or all of the substrate languages). But more important, as Bickerton shows, while the variable SVO orders in pidgin are simply permissive and allow for the variability of inexpert speakers (whose speech often reveals their language origin), the rules that create variants of SVO order in Hawaiian Creole English are used to fulfill sophisticated “fronting” functions as when the Object, Verb, and Subject occur in this order:

difren bilifs dei got, sam gaiz

“Different beliefs they got, some guys.”

“Some guys have different beliefs.”

That sentence would not be understood by a pidgin speaker. Where did creole speakers pick up this rule of word order? How did Hawaiian Creoles come to put the object at the front of the sentence contrary to the rules—or the absence of rules—in the substrate languages? Is there some linguistic or cognitive push toward such “fronting,” or toward creating a perspective in which the object takes precedence, that impels a speaker to invent a rule for doing it even if such a rule doesn’t exist in the immediate linguistic environment? And suppose you found phenomena comparable to this one in, say, Guyanese Creole, Seychelles Creole, Papiamentu, Sranan, and Mauritian Creole?

Bickerton’s book is so full of examples of common rules that come from so many different creoles with such diverse substrate and superstrate languages that his findings cannot be ignored. The common rules could not have been “learned” from the linguistic environment alone because they do not exist there. They had to be invented, to spring afresh from the mind of man.

And now suppose it appears to be the case that “errors” in word order among children acquiring different standard languages may also be interpreted as attempts to carry out that same function of fronting to which we first alluded. The errors, so-called, would not have to follow specific rules for the movement of the Subject, Verb, and Object (as in the earlier example cited for creoles). All that one needs to show is either that children make a swift and errorless “search” for forms in the language that will perform the same functions as the creole forms or that they incorrectly invent such forms if they do not exist. What is crucial is that the young language “learner” seems to be driven to make certain distinctions in the language he uses, and if he fails to find one that is accessible in his standard language, he will create one on his own. The child in this view is also a language creator, inventing forms different from the ones found in his environment. He must also be getting the structure “from his mind.” The structures of creole and of child language, Bickerton argues, are astonishingly similar, so both must come from a universal “bioprogram” for language that shapes the human mind.

Bickerton’s book is dense with closely reasoned linguistic examples of the kinds of distinctions he thinks are part of the “underlying push” given by the bioprogram of language. Space, alas, does not permit us to sample much of his detailed evidence, and the details will not be easily understood by those not used to linguistic analysis. (The book, for all that, can still be read with enjoyment by the interested nonlinguist, thanks to its broad perspective and the lively style of its general argument.) Rather than our trying to evaluate the details (a task better left to the scholarly controversy that Bickerton’s thesis will foment), we would do better to consider his claims about the nature of the “push” (or pushes) to language and find out whether his posited language bioprogram serves him as well as he supposes it does. His arguments are highly technical, but their implications for a theory of mind and language are enormous.

As we have noted, Bickerton conceives the underlying bioprogram as generating linguistic inventions to match primitive distinctions that are present innately in the nature of human cognition. Learning a language “consists of adapting this program, revising it, adjusting it to fit the realities of the cultural language [the learner] happens to encounter. Without such a program, the simplest of cultural languages would presumably be quite unlearnable.” The structure of language (as with Chomsky) is in the structure of mind. In some sense, then, language is “known” ab initio by any member of the human race. But Bickerton develops this view with a new twist. He argues a more differentiated case: the evidence both from creoles and from language acquisition by children compels us to accept at least four primary mental distinctions as in some way “understood” prior to language. If they were not, then we could never grasp them linguistically.

The four distinctions are basic to all human communication. The first is between the specific and the nonspecific (SNSD), the distinction usually and minimally handled by a system of articles, illustrated by such expressions as (specific) “A dog just bit me,” and (nonspecific) “Mary can’t stand to have a dog in the room.” All creoles use or invent SNSD by distinguishing an indefinite from an absent article (called a “zero-marked” article by linguists); and children learn pure SNSD distinctions in their native language without making errors.

The second is a distinction between state and process (SPD), also “innate.” This is attested to by the speed, errorlessness, and the virtual absence of confusion in the mastery by English-speaking children of the principle that –ing forms apply only to verbs denoting a process and not to verbs referring to a state—a child may say “I eating” but not “I being here.” There are, as Bickerton shows, parallels in creoles.

The third distinction is between “punctual” and “nonpunctual” (PNPD). It is a distinction that marks the difference between verbs for single, punctate events (to hit) and others that are steadily repeated or habitual (to live). The past tense, when first grasped, is reserved for punctuals only, both in language acquisition (French and Italian being the plainest examples) and in creoles.

The last of the Big Four innate distinctions is CNCD, the causative-non-causative distinction. Where the distinction is clearly made in a standard language (as it is in creoles), it is learned without error—whether in a language like New Guinean Kaluli, where the causal role of agent is marked by an inflection on the subject of the sentence, or in Turkish, where an affix is added to a verb to indicate its causal status.

To put it simply, Bickerton argues that the bioprogram predisposes human beings to experience their lives, and then to find primitive linguistic expression for definiteness versus indefiniteness, process versus state, boundedness versus unboundedness in events, and causativeness versus other states of action and being. Those familiar with linguistics will recognize this thesis immediately as elaborating a basic “Tense-Modality-Aspect system” that organizes experience according to what might be called “arguments of action,” and then goads the aspiring speaker to find or invent linguistic forms for expressing basic elements of these arguments. But one does not have to be a linguist to appreciate the boldness of Bickerton’s argument which, on the basis of a breathtaking and well-considered sweep through the literature on language acquisition and on creoles, sets forth a plan of the kind of Mind that must have started the phylogenetic Original Language Game.

The plan Bickerton presents is not only a bold but also a flexible one, for it does not woodenly specify innate linguistic forms but only specifies what kinds of forms will best fit the distinctions that our language-searching minds have been looking for. Indeed, Bickerton argues that our so-called cultural languages often become less well fitted to the essential demands of our alleged bioprogram (as in decreolization, where a superstrate language begins to take over) and become both more difficult to learn and reflective of more complex experience. Bickerton in fact goes so far as to claim that some languages, notably the “purer” original creoles, are less evolved than others. Those purer languages are akin to child language (including its invented “errors”). In spite of Bickerton’s brave effort to avoid giving the impression that original creoles are childish or otherwise linguistically inferior to standard languages, his claims will surely be taken as a challenge to the relativism that has long been orthodox among linguists.

How do his arguments sound to a pair of psychologists, one of whom knows something about language acquisition (J.B.), the other something about Creole languages (C.F.)? Bickerton’s chief problem is the assumption that twentieth-century creoles are created in a linguistic vacuum and by processes comparable to those that originally produced language. But is Bickerton looking at raw creation when he examines creoles? Do the original creole speakers he studied live in a world containing only pidgin? Obviously not. The first generation of creole speakers is born into a world already shaped by language, in spite of their pidgin-speaking parents. They often have grandparents who speak a standard language, parents who understand that language. They are part of a local culture created by speakers of standard substrate languages living in a social structure created by speakers of a standard superstrate language. They come into a world already constituted by language.

Bickerton’s argument about the child’s acquisition of standard languages is marred by the very same assumption about a cultural vacuum. We should note that the author’s wrath about current theories of language acquisition falls on what he calls the “Bruner-Snow fallacy,” which is the view that even if children have innate linguistic structures, they also receive a simplified, well-formed, and repetitive experience—or “input”—of language from parents, adults, and even other children that helps “teach” them language. “This is quite simply untrue,” we are told. Perhaps so, but perhaps we had better have a look at Bickerton’s bioprogram to see whether, indeed, it causes language to spring largely unassisted from the mind.

Consider his own account. Fascinated by the “novelty” of creole he remarks: “All I want to claim is that if we persist in believing that the child must have input in order to learn, we shall continue to misunderstand completely the way in which he does learn a developed, natural language” (p. 139). But later he says: “the world had to be recreated in the image of language before anybody could communicate about anything at all” (p. 218). Or in his happy phrasing on the same page, it is not the case “that when we were ready to talk, all the things in the universe stood there waiting—rock and river, dodo and elephant, storm and sunrise, thirst and evil, love and dishonor—all waiting patiently for their labels.”

Yes, the one thing that we know with some clarity from the last decade of intensive research on language acquisition is that the child requires some knowledge of the world and how it is arranged socially, and something as well about the human intentions that will be encoded in speech acts, before he is able to use his “innate knowledge” of worthwhile distinctions to search for, or invent, linguistic forms that correspond to those distinctions. It is impossible to doubt that there is some kind of bioprogram for language that steers the searching and the invention. But that program cannot begin to work until the child is inducted into a social world where language has already made a deep impression in shaping and even in constituting the reality to which speech will refer.

Whether you are learning Mauritian Creole or Mandarin Chinese, you are introduced into a world shaped by language before you ever start on grammatically well-formed speech. The history of the human species is not carried in the mind, let alone in the genome, alone, but in the culture that carries highly systematized information from generation to generation. The idea that the person intending to speak creole or the child of a family speaking a standard language is starting with no input is bizarre. Nor is such an exaggerated claim necessary for Bickerton’s challenging and intriguing general argument. As he puts it himself, whatever the nature of the bioprogram, it necessarily interacts with the cultural reality and the linguistic forms that are available to the child.

But Bickerton’s effort to show how a bioprogram is at work is a superb piece of scholarship. His book will echo for many years to come—whether it gets all the particulars right or not.