ABSTRACT
When people argue for the need for descriptivism, they typically appeal to the need to provide a semantics that accounts for the cognitive role of these terms: the role they play in reasoning and action. Semantic theories are empirical theories: they have to mesh with evidence about how terms are used. This chapter argues that these considerations suggest the need for a more thoroughgoing descriptivism than what has been proposed by the neo-descriptivist theories—a version of the traditional cluster theory. It briefly addresses the question of which expressions are natural kind terms (NKTs). The driving idea is easily stated: the NKTs are just those that pick out, or purport to pick out, categories united independently of our parochial and often ignorant ways of sorting things together on the basis of perceived, often superficial, similarities; in short, independently of us.
