A Developmental Robot Model of Early Language Acquisition
Résumé
Recent NLP techniques have enabled a considerable advance in the generation and understanding of natural language. But given the way these neural NLP systems learn, and the astronomical amounts of data required, they cannot provide answers about how human infants learn and acquire language, as they do not follow the same language development trajectory. We propose a robot cognitive model of early human language acquisition inspired by the way human babies learn language. The robot relies on social interaction, making it an active learner, with a caregiver to acquire motivation-grounded language. The robot's modular architecture enables it to be situated in the same conditions as a human child acquiring language. The aim of this model is to provide a tool for testing hypotheses related to questions about the process of language development in humans.
Origine | Fichiers produits par l'(les) auteur(s) |
---|