Proceedings of the Annual Meeting of the Cognitive Science Society

People’s social and linguistic environment changes over the course of their life: infants learn language from a small setof caregivers; children and adolescents practice language skills with their peers; adults speak to other adults and also passon their language to the next generation (Kerswill, 1996, Sankoff 2018). Population models of language change haveexplored network effects but neglected changing networks as a function of agent age. We model a population of Bayesianagents that go through life phases of initial learning, subsequent peer interactions, and transmission to the next generation.We find these age-dependent networks to be more stable than other network architectures. This stability counters previousBayesian modelling results in which languages reliably and rapidly change, converging to the learners prior, suggestingthat languages spoken in populations in which interactions are organised assortatively by age may only weakly reflecthuman priors on language learning.