Bản ghi email: Cross-Situational Learning with Bayesian Generative Models for Multimodal Category and Word Learning in Robots