Modelowanie konkurencji między językami

Wojciech Borkowski , Andrzej Nowak , Peter W. Culicover

Abstract

MODELLING LANGUAGE COMPETITION – Summary Competition between languages has recently been in the centre of attention of computational and analytical research which has mainly concentrated on developing models that can reproduce empirically observed distribution of language sizes, which is very close to log-normal. Shultze and Staufer (2006) presented the best-known attempts of sociophisicist. While some of them are very interesting, they tend to ignore recent findings in the field of linguistics and social theory. Simulation model we present here is based on well-known and commonly accepted theory of social influence (Nowak at all, 1990). We also drew on model described by Culicover, Nowak and Borkowski (1999) that was originally designed to explain language change but seems very suitable for modeling language competition. Social influence can be responsible for many important linguistic phenomena, such as the different ways of creating verb clusters in west Germanic dialects or the similiarity of grammar features of disparate languages. In this paper we demonstrate that, if we introduce the big spatial differences in the strenght of different factors (social agents- local communities) which most probably have Pareto-like distribution and when we model a vast area (square lattice of 500 by 500 agents – local communities) the model can describe qualitative reproduction of empirically known distribution of languages.