In November 2023, I started working on the FWO fundamental research project "It's all frequency? Testing usage-based theories of language change using agent-based models", supervised by prof. dr. Dirk Speelman, prof. dr. Freek Van de Velde and dr. Dirk Pijpops. The project focusses on how language change is shaped by how often we use certain words, so-called 'frequency effects'. I use computer simulations for this research, which is relatively unique within linguistics.
Computer simulations?
Computer simulations are becoming more and more important in the field of linguistics, as linguists have realised they can use computer simulations as a theoretical sandbox for linguistic theories. Using computer simulations, researchers can program thousands of virtual speakers – also called "agents" – and have them communicate with each other. Through these interactions, agents can make decisions that affect the course of the simulation, allowing researchers to study how different factors might influence the dynamics of language change over time. As such, simulations are a useful method to explore and test hypotheses about language change, and can help us better understand the processes that drive this complex phenomenon.
About the project
In my PhD project, I focus on how frequency effects shape language change. While usage-based theory of language has been conducive to some highly impactful explanatory mechanisms of language change, such as the reducing effect, syntactic priming effect and analogy, critics suggest that these mechanisms are often deployed on an 'ad-hoc' basis. My PhD project emanates from the idea that concrete implementations of such theories could take a more central place in usage-based linguistics, and serve as a means to test and refine usage-based models of the theories mentioned above. The overarching goal of my project is to use agent-based computer simulations to implement the usage-based frequency mechanisms mentioned above in a computational simulation, such that more easily falsifiable hypotheses can be inferred. By testing what minimal conditions are necessary for the three general effects mentioned above to occur in the simulation models, I will be able to fill in their blank spots and assess their potential for language change on the diachronic level. Through these investigations, I will also demonstrate how useful computer simulations can be in this context and exemplify how they can be used to solidify (or undermine) theoretical claims.
Past experience in computer simulations
In the past, I used agent-based simulations to model the divergence of the Standard Dutch pronuncations in the Netherlands and Belgium. This study is published in the Computational Linguistics in the Netherlands Journal.