Pragmatic Models for Generating and Interpreting Instructions and Dialogue

Please bwaldon [at] stanford.edu (email) for the Zoom link.

To generate language, natural language processing systems predict what to say---why not also predict how listeners will respond? We show how language generation and interpretation across varied grounded domains can be improved through pragmatic inference following the Rational Speech Acts framework. We use data from people to train neural speaker and listener models that ground language into a world context, then layer a pragmatic inference procedure on top of these models. This procedure improves models' success when interacting with human partners in grounded instruction generation and interpretation tasks, as well as in a challenging partially-observable spatial reference dialogue game.