The Birth of Language

To the Editors:

In his review of Robert Berwick and Noam Chomsky, Why Only Us: Language and Evolution [NYR, August 18], Ian Tattersall concludes that “no other scenario currently available from linguistics fits the archaeological facts better than the essentials of Berwick and Chomsky’s vision.” This is a remarkable claim for a “vision” blind to the social context, whatever it was, in which language must have evolved and offering no account whatever of the origin of words. Words are the first step into language. It’s through words learned in conversation that children enter language. Without words, no syntax, without syntax, no language, as Berwick and Chomsky themselves fully acknowledge.

Since, for Berwick and Chomsky, language is primarily an instrument of thought, only secondarily of communication, they postulate “computational atoms, word-like but not words” as concepts, the units of thought or meaning that Merge combines but that no one ever speaks. A recurrent refrain of the book is the origin and nature of these uniquely human concepts and the words that symbolize them, representing our whole mental world, past, present, and future, real and imagined, concrete and abstract. In the face of this puzzle, Berwick and Chomsky throw up their hands and declare it a mystery, a view that Tattersall evidently shares.

Oddly, Berwick and Chomsky mention Derek Bickerton as one of the few authors to have recognized the “mystery,” but do not address his ingenious answer. Bickerton in More Than Nature Needs (2014), perhaps the most original, erudite, and deeply thought book yet published on the evolution of language, argues that the first step toward human concepts was taken by Homo erectus’s use of gestures or vocal sounds for displacement, that is for reference to objects and events not immediately present to the senses.

From clear evidence of “confrontational scavenging” by Homo erectus, Bickerton conjures the scene of scouts recruiting absent members of their group to ward off dangerous rival scavengers and to help in butchering a newly found dead megafauna. Over many hundreds of thousands of years of sustained confrontational scavenging, words emerged from the slowly differentiating vocal apparatus, each step changing the conditions of selection on the path toward the “language niche.” Words, initially used for immediate reference, or only in displacement while scavenging, freed themselves little by little from their original contexts to form a general-purpose lexicon. Thus grew the syntax-free, pragmatically ordered protolanguage (peremptorily dismissed by Berwick and Chomsky) that Merge (or some parallel alternative) would transform into the hierarchically structured form of a natural language. On this view, in full accord with modern studies of neuroplasticity, behavior built language in the brain; the brain did not build language in behavior.

Speculative though Bickerton’s scenario is, it is no more speculative than Berwick and Chomsky’s and has the critical added virtues not only of explaining the origin of human concepts and the words that symbolize them, but also of setting the birth of language in an evolutionarily plausible social context. In our view, no other scenario better fits the facts than Bickerton’s.

Michael Studdert-Kennedy
President and Director of Research Emeritus
Haskins Laboratories
New Haven, Connecticut

Herbert Terrace
Professor of Psychology and Psychiatry
Columbia University
New York City

Ian Tattersall replies:

Studdert-Kennedy and Terrace’s response to my review epitomizes the way in which discussion of language origins (and consequently of attitudes toward Chomsky) has become polarized. The problem was, and continues to be, twofold. Language is the unique property of the sole surviving hominid species Homo sapiens, so there is no comparative basis on which to approach it; and languages themselves change so fast that the original properties of the underlying system are lost in the mists of time, depriving researchers of a historical record. Everything we want to know about when and how language originated is thus a matter of indirect inference from proxy information.

Under such circumstances, a minimalist (if you’ll excuse the expression) approach to any aspect of language origins is much to be preferred; and it is odd to find Studdert-Kennedy and Terrace objecting to Berwick and Chomsky’s focus on the cognitive algorithm underlying language to the exclusion of “social context” and “the origin of words.” For consideration of such things would only have served to muddy already murky waters.

Derek Bickerton’s extravagant but evidently much-admired 2014 scenario is a case in point, piling one unsubstantiable assertion upon another until all we are left with is a well-crafted but entirely untestable story. If Studdert-Kennedy and Terrace’s appeal to authority really had to involve citing Bickerton, they might have done better to quote his 1995 remark that “true language, via the emergence of syntax, was a catastrophic event, occurring within the first few generations of Homo sapiens.”

For in that case, Bickerton’s linguistic speculation hewed much more closely to what we can infer from the actual material record of early human behavior. As I hope to have reasonably clearly shown in my review, a hugely consequential qualitative behavioral inflection demonstrably occurred in our lineage, not only abruptly and recently, but within the tenure of our species Homo sapiens. This behavioral shift was most plausibly associated with the sudden and spontaneous invention of language by members of a species that already possessed the necessary neural wiring; and no scenario of the mental algorithm underlying this event better fits the facts than Berwick and Chomsky’s.