On the development of computational models of the English lexicon