Lingua ex Machina: Reconciling Darwin and Chomsky with the Human Brain (review)

The laudable goal of this book, unfortunately not stated explicitly until page 195, is to ‘bring peace to a conflict that should never have broken out in the first place, and to show, contrary to so much that has been written over the past few decades, that the approaches pioneered by Darwin and Chomsky are fully reconcilable’. This goal, first stated prominently in an important 1990 article by Pinker and Bloom (not cited by Calvin and Bickerton), has been taken up by a growing community of researchers in the intervening decade (see, for instance, many of the papers in Hurford, Studdert-Kennedy, and Knight 1998, as well as Carstairs-McCarthy 1999 and Jackendoff 1999).

The book is written as a dialogue between the two authors, each writing relatively independently; their contributions are even presented in distinct typefaces. Occasionally one author interjects a paragraph or two of comment into the other’s chapter, but for the most part we are dealing with two separate interlaced books. Thus, unlike many collaborations, it is clear here precisely which idea is due to whom.

Derek Bickerton is responsible for what I consider one of the few good ideas in the literature on evolution of language: his proposal (Bickerton 1990) that language evolved in two stages, the second of which is the modern language faculty. The first stage is what he calls ‘protolanguage’, in which communication is accomplished essentially by unstructured strings of words supplemented by pragmatics. What makes B’s proposal of interest is his claim that protolanguage did not go away when modern language evolved. Rather, it still surfaces in situations when full language is either not yet developed (as in early child language) or disrupted (Genie, pidgins, possibly agrammatic aphasics); the communication of the language-trained apes appears to be another manifestation of protolanguage. Thus these situations function as sort of living fossils of earlier stages of human evolution. (A somewhat similar proposal appears in Givón 1979.)

Until the present book, B has argued that modern language emerged from protolanguage in a single catastrophic step (see the title of Bickerton 1998). Here, however, he advocates what I think many will see as a more reasonable gradualist position: Some innovation in the protolinguistic capacity resulted in the possibility of certain elements of syntactic structure, then further innovations, fueled by natural selection, led to the full modern capacity. Like Pinker and Bloom, B recognizes the possible role of the ‘Baldwin effect’ in ramping up evolution of an innate capacity: Put simply, if learning some particular ability promotes fitness, then individuals who can learn it more readily gain a selective advantage. One way to learn something more easily is to have some of it innately prespecified. Thus under the right conditions, some aspects of acquired knowledge can in succeeding generations be reinvented as innate knowledge. But such innate knowledge—knowledge that makes learning easier—is exactly what universal grammar is supposed to be.

To formulate a reasonable gradualist hypothesis, one must (1) suggest a plausible initial innovation that gets the process started and (2) suggest plausible steps from the initial innovation to the modern language capacity. Of course, one must also have a theory of the modern capacity that permits (1) and (2). One of the longstanding difficulties in reconciling Darwin with Chomsky is that the Chomskyan treatment of syntax does not lend itself readily to plausible incremental evolutionary steps. This is why Chomsky has advocated catastrophic emergence of language without the help of natural selection. B observes that the minimalist program offers an improvement in this respect, though not a sufficient one. In an appendix he sketches a modified minimalist theory that he claims is more adequate on both linguistic and evolutionary grounds. Aside from one comment below, I will not attempt to evaluate his proposal; I have argued on similar grounds (1997, 2001) that far more radical surgery is required on the traditional Chomskyan architecture.

But...