The impact of population dynamics on language evolution: a computational study

Historical context. The past 13 years have seen an explosion in interest in the origins and evolution of human language, sparked by the early works of Hurford (1989) and Pinker & Bloom (1990). Recent publications in major interdisciplinary journals (Nowak et al. 2000; Nowak et al. 2001) confirm that research into the origins and subsequent evolution of language is of scientific interest to a wide array of researchers, including linguists, cognitive scientists, biologists, computer scientists and primatologists. The Language Evolution and Computation Research Unit (LEC) has been at the forefront of research into this new topic. Typically, the evolution of language is seen as primarily a biological problem (e.g. Pinker & Bloom (1990), Nowak et al. (2000) and Nowak et al. (2001)), with the origins and structure of language explicable in terms of the biological evolution of an innate Language Acquisition Device (Chomsky 1980). The LEC’s major contribution to the field has been to draw attention to the possible role of social and cultural processes in explaining the evolution of language. LEC researchers have developed the Iterated Learning Model (ILM, see Kirby (2001), Brighton & Kirby (2001) and Kirby et al. (2002)), which allows the effects of repeated cultural transmission of linguistic structures to be investigated. Briefly, in the ILM the linguistic competence of one generation of individuals (their I-language) determines their observable linguistic behaviour (their E-language), which is observed by a new generation of individuals who induce their own I-language based on the observed E-language. Computational implementations of the Iterated Learning Model show that characteristic features of language, such as compositionality (Kirby 2000) and a regular-irregular distinction (Kirby 2001) can emerge purely as a result of the iteration of this expression-induction cycle.