We present a number of high-level computer models simulating the social factors involved in language change. The models all build on the concept of a linguistic community. This community consists of a number of interacting agents that adapt their use of linguistic items to their peers. Linguistic items are treated as cultural traits, pieces of information that are transmitted through the population by social learning. We study the effects of different models for this transmission. The models we discuss are based on Daniel Nettle’s (1999) adaptation of Social Impact Theory. Our models simulate a language learning situation in which the linguistic items acquired by an individual are determined by the community’s linguistic behaviour through a complex function. This function incorporates differentiation within the community in terms of the structure of social status, proximity of agents to each other and the number of individuals carrying different traits. The result is a highlevel description of a heterogeneous linguistic community that allows us to efficiently study broad processes. We extend the basic model by considering different community structures, different learning algorithms and different language structures, incorporating insights from socio-linguistic theories on the transmission of linguistic forms through differentiated language communities. In particular we model William Labov’s (2001) investigations into probability matching in variant use by formalizing the social pressures on language users in a game-theoretic framework.