Computational and Robotic Models of Early Language Development

摘要

This chapter focuses on the models of early language development, ranging from vocal development to the onset of speech as a communication medium and to early word learning. It analyzes models in terms of several of the general causal mechanisms: cross-situational statistical learning, embodiment, social interaction, self-organization of brain-body-environment couplings, intrinsic motivation, and the links between learning and evolution. Indeed, many robotic models are entirely implemented in a virtual world with a simulator of the robot body and its physical/ social interaction with the environment. Robotic models have been developed to study the roles and structures of the constraints. Developmental robotics models have further studied various links between sensorimotor learning and social language learning. Modeling the development and learning of language has inspired researchers in computer science, psychology, and robotics to adopt diverse approaches to the many challenges involved.