Video Recordings

Languages persist through a cycle of learning and use - we learn the language of our community through immersion in that language, and in using that language to meet our communicative goals we generate more linguistic data which others learn from in turn. In previous work we have used agent-based computational models and experimental methods from psycholinguistics to simulate this process and show how biases in learning and use can explain some of the fundamental structural features shared by all languages. For example, the fact that languages exploit regular compositional rules for generating meaningful expressions allows languages to be relatively learnable but also exceptionally powerful tools for communication, and we can show that this structure arises naturally as languages adapt to the constraints from learning and use inherent in their transmission. In this talk I’ll review these older findings, then talk about more recent work using the same approach to identify the mechanisms responsible for some other cross-linguistically frequent configurations, in particular teasing apart the contribution of distinct pressures from learning and use in creating frequency-irregularity correlations in inflectional systems and recursive numeral systems.