``The effect of fitness in the emergence of creole'' Makoto Nakamura, Takashi Hashimoto, Satoshi Tojo and Kenny Smith Japan Advanced Institute of Science and Technology and University of Edinburgh mnakamur@jaist.ac.jp We report results relating to the role of fitness in the emergence of creoles. The framework of population dynamics has become involved in language evolution in recent years, the goal being to establish a mathematical and evolutionary theory of language. The language dynamics equation proposed by Nowak et al. [1] is representative of this effort. We have revised their model to be more realistic, in order to study the emergence of a creole in the context of population dynamics. Our prime revision is that the transition rate between languages is sensitive to the distribution of languages in the population at each generation. In addition, we introduce an exposure probability term, which determines the degree of influence from other languages during acquisition. Using this approach, we have previously shown the process of the emergence of a creole [2] and the conditions on similarities among languages required for a creole to emerge and be dominant [3]. In these models, and those of Nowak et al., it is assumed that language speakers bear offspring in proportion to their success at communicating (their fitness). However, in the real world case creoles do not emerge because creole speakers have more offspring than speakers of other pre-existing languages. Rather, the influence of infants' learning biases during the repeated cultural transmission of the creole forces it to take on certain structures. Therefore, we need to understand what role fitness plays in the emergence of creoles in these models. We will then be able to compare the behaviour of the modified language dynamics equations with and (more realistically) without the fitness term. The major difference is that the range of exposure probabilities at which a creole emerges is remarkably larger in the model without fitness than the one with fitness. When fitness does not play a role, the creole exists over the whole range of exposure probabilities. In contrast, in the model where fitness does play a role, there is a critical value for the exposure probability. Above this critical value, the creole is maintained in the population. Below it, the creole is eliminated. We conclude from these numerical analyses that fitness in fact acts to suppress the emergence of a creole when children are not significantly affected by other languages during the process of language acquisition. [1]Nowak, M. A., Komarova, N. L., & Niyogi, P. (2001). Evolution of universal grammar, Science, Vol 291, 114--118. [2]Nakamura, M., Hashimoto, T., & Tojo, S. (2003). The Language Dynamics Equations of Population-Based Transition -- a Scenario for Creolization, Proceedings of the International Conference on Artificial Intelligence (IC-AI'03), CSREA Press, 689--695. [3]Nakamura, M., Hashimoto, T., & Tojo, S. (2003). Creole Viewed from Population Dynamics, Proceedings of the Workshop/Course on Language Evolution and Computation in ESSLLI, 95--104.