Identifying drivers of language change using neural agent-based models.

In this project, I will determine social, cognitive and language-specific factors that drive language change. Specifically, I will investigate the influence of community structure, language contact, and learnability on morphological simplification. I will do this by simulating interactions between speakers using agent-based computer models, evaluated on data from the Alor-Pantar islands in Indonesia, and other data sets.

Agent-based models enable research on language change on long time scales, for languages for which only descriptive data is available, from a bottom-up perspective. To study cognitive factors for language change, I will use deep neural networks from artificial intelligence, as models of language comprehension and production. To examine social factors, I will evaluate different representations of the agent environment, community structures and language contact situations. I will extend phylogenetic models from historical linguistics, to not only reconstruct language ancestry, but also give insight in the processes that brought about language change, by combining coalescent simulations from bioinformatics and agent-based models. By performing this project, I will contribute to the knowledge about general drivers of language change and to the agent-based methods that could be used to study language change.