In his last writings, Kuhn discussed not only strict taxonomic kinds but also ‘kinds of materials’ and ‘singletons’. This suggests that, for Kuhn, scientific lexicons are richer and more complex than simple conceptual taxonomies. At the same time, the distinctions among various components of lexicons may obscure the primary objective of their study: to elucidate the belief system of a scientific community through linguistic analysis. This original purpose can be restored by setting aside concerns related to meaning acquisition and by viewing a scientific lexicon as a representation of the inferential patterns employed by members of a scientific/linguistic community. As I will show, Kuhn appears to take this approach in some of his mature writings. Finally, given how his philosophy connects scientific language to the activities of scientific communities, I will dispute claims that Kuhn underwent a ‘linguistic turn’.