Introduction

Human natural language is a complex adaptive system, and the usual account of adaptive complexity in the biological world is natural selection (Pinker and Bloom, 1990). There have been a number of arguments that the human capacity for natural language could not have evolved via natural selection. In this paper I focus on two such arguments. The first is the argument that language could not exist in any intermediate forms, and thus could not be the product of a stepwise selective process. The second is that language is arbitrary, and that selection cannot explain its specific nature (Piatelli-Palmarini, 1989).

In this paper, I argue that even if one accepts the claim that language cannot exist in intermediate forms, it is still possible that it could have evolved via natural selection. I make two crucial assumptions. The first is that humans have some degree of plasticity, and are capable of learning. The model of learning that I invoke is impoverished, and is not meant to reflect actual human capacities. The second assumption is that successful communication confers a reproductive advantage on its possessors. In other words, language is adaptive as long as it is shared. This has become a fairly common assumption in the computational modeling of language evolution; an early example is provided by Hurford (1989). I do not consider arbitrariness to be any argument against adaptation (cf. Ridley, 1990), although I do not dispute the claim that many aspects of language are arbitrary.

Evolution as hillclimbing

Pinker and Bloom (1990) have argued that natural language fits into a continuum of viable communicative systems.