At this week’s MCQLL meeting, Yang Xu, ​from the Department of Computer Science and Cognitive Science Program at University of Toronto, will give a talk titled Word meaning extension in humans and machines. An abstract of the talk follows.

When: Tuesday, October 18, 15h00–16h00 (Montréal time, UTC-4)

Where: MCQLL meetings this semester are in hybrid format. We will meet in-person in room 117 of the McGill Linguistics Department, 1085 Dr-Penfield. If you’d like to attend virtually, Zoom meetings will be held here.

All are welcome to attend.


Humans often make creative use of words to express novel meanings, a phenomenon known as word meaning extension (WME). WME is one of the most commonly attested forms of lexical creativity, but machines are relatively impoverished in interpreting novel word meanings. I describe research aimed at closing this machine-human gap from the dual perspectives of human cognition and natural language processing. I first present work that formalizes the theory of chaining as probabilistic models to account for the cognitive mechanisms of WME through time. I then show that similar models with enriched knowledge can be applied to simulating children’s lexical innovation and improving the automated processing of creative lexical use such as slang. I close by discussing open issues toward a generative approach to modelling the lexicon.