Abstract: Advances in artificial intelligence and digital humanities have offered opportunities for analyzing text at unprecedented scale. I present a computational approach to investigate sociomoral meanings via longitudinal text. The approach is based on the premise that language use can inform people’s sociomoral perception, and we build this methodology on recent development of word embeddings from artificial intelligence. I illustrate the approach in two case studies. In the first study, I show how word embeddings can inform the emergence of gendered language in caretakers and children during early child development. In the second study, I show how similar methods combined with domain knowledge can inform people’s changing moral sentiment toward concepts such as slavery and democracy over centuries. I close by discussing future challenges and opportunities in the intersection of artificial intelligence, language, and society.
Yang Xu: Assistant Professor, Department of Computer Science, Cognitive Science Program, University of Toronto. He obtained his BA/MEng from the University of Cambridge and a PhD in Machine Learning from Carnegie Mellon University. He took a postdoctoral position in the Department of Linguistics at UC Berkeley. His current research intersects language, computation, and cognition, with an emphasis on characterizing the time-varying properties of the lexicon and how natural language informs sociomoral changes.
Great Explorations, is a series of academic discussions we hope will inspire our communities. Our special thanks to our engaged, critical thinkers who attend our speaker series. The dialogues that came out of these talks, both in-person and online via Zoom, are integral to our campus community and Scarborough communities at large. We welcome your input for future topics, as well as a review of previous recorded sessions available on YouTube.