Noam Chomsky, around whose work much of the Syntax series revolves, gives listeners a glimpse into the evolution of his own thinking, with an emphasis on areas of linguistics where computational considerations play a major role.
Chomsky briefly outlines the key components of a biologically based linguistics that began to emerge 50 years ago: first, a genetic language endowment (Universal Grammar), which interacts with the external environment, and second, the individual’s development and learning strategies. While UG has been called “controversial,” says Chomsky, the “alternative is magic,” since something has to account for the fact that “my granddaughter picked out part of her environment as language related, and almost reflexively developed a language while her pet kitten, a chimp or songbird, exposed to exactly the same data, didn’t take the first step and couldn’t conceivably take the second.”
Chomsky links a third factor of language involving architecture and the principles underlying data acquisition to natural laws that may apply generally in biology, and not specifically to language. Research suggests that between 50 and 100 thousand years ago, humans made an abrupt evolutionary leap forward in cognitive capacity. Language seems to have emerged at this time. While long-term evolution can lead to great complexity, a sudden leap like this, says Chomsky, tends to yield something “simple, almost perfect -- a perfect solution to design problems imposed by circumstances and conditions prevailing at the time of emergence...” This proposal has been dubbed the Strong Minimalist Theory (SMT), and offers a plausible approach to studying the complexity of language, believes Chomsky. It might prove profitable to “examine the range of phenomena that fall under what’s loosely called language,” and try to “disentangle them so some parts of them conform more or less to SMT.” And here, says Chomsky, issues of computational efficiency play perhaps an overwhelming role.
Chomsky links SMT to transformational grammar, a long-standing component of his linguistic theory. He states that “a simple form of transformational grammar is just the optimal system, and if you don’t have it, you’d have to have an argument as to why you don’t.” Well-designed systems should have simple, sensible properties. He recommends “chipping away at the stipulated properties of Universal Grammar, and technologies proposed to deal with particular problems to see how closely you can show that language does approximate to the perfect design that would be a natural expectation in light of what appears to be evolutionary history.”