How can artificial intelligence inform us about language acquisition in children?

This event has ended.

While decades of research in developmental psychology have identified key mechanisms behind children's language learning, debate continues over the relative importance of statistical learning, cross-referential learning, social learning, and nativist theories.

In this talk, I argue that one way to make progress on this question is to use machine learning to simulate and analyze language acquisition in children. I will first present a series of modeling studies that aim at identifying potential learning mechanisms and types of input data needed for computer models to mimic infant early sound and word learning (simulation). I will then go on to present our efforts in building automatic speech-processing algorithms that can be used to measure the influence of the environment on language acquisition (automatic analysis).