Human Language Acquisition by Computers

In this paper, we present a new system for learning structured knowledge based on multimodal language acquisition in human-computer interaction. The structured knowledge is represented in the computer system using natural language. New language knowledge is acquired in two ways: (1) by acquiring new linguistic units (words or phrases) and their corresponding semantics, and (2) by acquiring new language rules (sentences) and their corresponding computer actions. The system represents an adaptive dialogue interface capable of interpreting the user’s commands and inputs and of learning new linguistic terms and rules. This system easily acquires new lexical knowledge using different input modalities, such as, speech, typing, touch, video and pointing. Key-Words: Language acquisition, adaptive dialogue systems, computer language learning, structured knowledge.