Proceedings of the Annual Meeting of the Cognitive Science Society

Words show complex dynamics of meaning change. In somecases, a word may acquire novel senses. In other cases, ex-isting senses of a word may become obsolete. The rates atwhich words gain and lose senses may vary, but it is an openquestion which factors might account for this variation. Build-ing on work in computational linguistics and cognitive science,we develop a computational approach that explores this ques-tion by leveraging word sense records from a large histori-cal database of English. Our results suggest that polysemouswords tend to gain and lose senses more than words with fewersenses, and that these effects are robust when word frequencyand length are both controlled for. These results are consis-tent with recent findings on the mechanisms of emergent wordmeanings and they further suggest stability in the temporal dy-namics of word meanings.