Growing a vocabulary

Abstract

This paper describes how simple Boolean Network models can be adapted to explore the way a vocabulary network might grow. The models described have two main parameters — a New Word parameter, which determines the rate at which new words are added to the lexicon, and an independent New Link parameter, which determines the rate at which links form between words. Delaying the application of the New Link parameter by a small amount allows a core of words to establish itself. With these very simple assumptions, a network structure with some of the features of real lexicons can grow itself. The paper describes how the parameter values interact, and how the size of the initial core affects the way the lexicon grows. The paper also speculates about how a more realistic model of lexical growth might be constructed by making the basic processes in the models slightly more complex.