Lexical Semantic Networks and Language Change

Abstract. Word lists from many languages of the world are available in a digital form and one should be able to learn about language relationships by analyzing them. This, however, requires an understanding of measures of relatedness between words. Since words are mappings between sounds and meanings, this requires understanding of both domains. Over the last few years, we made much progress on the phonetic side, and we have developed good algorithms for comparing the sounds of words. A similar development of a linguistically salient definition of semantic distance is the next step. We have recently developed the concept of a lexical semantic network and studied its universality, but have yet to develop a state-process model of linguistic change based on it, or, indeed to show that cognates from actual linguistic families show semantic shifts expected from our semantic network. This working group will bring together experts with interest in these questions, and make possible the next major step in this direction.