Introduction

This paper is part of an ongoing research effort (Briscoe, 1997, 1998, 1999, 2000a, b) to develop a formal model of language acquisition, demonstrate that an innate language acquisition device (LAD) could have coevolved with human (proto)language(s) given plausible assumptions, and explore the consequences of the resulting model of both language and the language faculty for theories of language change. The paper builds on the earlier work by examining the model's ability to account for the process of creolization (Bickerton, 1981; 1984; 1988; Roberts, 1998) within a selectionist theory of language change.

Sections 9.1 and 9.2 describe the theoretical background to this research. Section 9.2 presents a detailed model of the LAD utilizing generalized categorial grammars embedded in a default inheritance network integrated with a Bayesian statistical account of parameter setting. Section 9.3 reports experiments with this model demonstrating feasible and effective acquisition of target grammars for a non-trivial fragment of universal grammar (UG). Section 9.4 describes the simulation of an evolving population of language learners and users. Section 9.5 reports experiments with the simulation model which demonstrate linguistic selection for grammatical variants on the basis of frequency and learnability. Section 9.6 reports further experiments demonstrating evolution of the LAD by genetic assimilation of aspects of the linguistic environment of adaptation. Section 9.7 describes the experiments modeling the demographic and linguistic context of creolization. Section 9.8 summarizes the main findings and outlines areas of further work.