Learning and Playing in Wubble World

Abstract

Children do not learn the meanings of words from parsing and understanding gigabytes of text; instead meanings are learned from competent speakers who relate language to what's happening in the child's environment. We present a word learning algorithm that operates in a video game environment where the players fill the role of the competent speakers and train softbots to learn language as a child would. We provide empirical evidence that the word learning algorithm successfully learns the meanings for some words in this environment and the children enjoy playing the game.