Natural kinds

5. The semantics of natural kind terms

Broadly speaking, there are two kinds of theory about this topic. Descriptivist theories say that each natural kind term is associated with a description (or a cluster of descriptions). So, for example, the term ‘squirrel’ might be associated with the description ‘small four-legged grey or red creature with a bushy tail that lives in trees’. This description would describe what Locke calls the nominal essence of the kind squirrel. The theory further says that a natural kind term applies to exactly those things that satisfy the associated description. A competent user of a natural kind term then needs to know which description is associated with the term. Direct reference theories say that a natural kind term may be introduced by identifying certain similar looking samples or exemplars – paradigm cases to which the term clearly applies – and taking the term to apply to exactly those things that belong to the same natural kind as the exemplars. So, for example, ‘squirrel’ might be introduced by identifying certain apparently similar things that have been observed in the wild and then taking ‘squirrel’ to apply to all and only those things that belong to the same kind as the observed things. (See Proper names §§1 and 5;Reference §§3 and 4.)

Descriptivist theories are often contrasted with direct reference theories. Hybrid versions of these theories, however, are available. For example, drawing upon Kripke, one direct reference theory says that a term such as ‘squirrel’ was introduced by speakers who were in a certain causal relation (such as a perceptual relation) to various exemplars, certain paradigmatic squirrels. A sophisticated descriptive theory can exploit that claim. It does this by using the claim in the description that the theory associates with ‘squirrel’. According to the theory, ‘squirrel’ is associated with a description that says (at least) that the term was introduced by speakers who stood in a certain causal relation (such as a perceptual relation) to various exemplars –exemplars that are then themselves described. The tactic is to co-opt what this direct reference theory takes to be the relation between speakers and exemplars, and to build it into the description associated with the term. The result is a theory with elements of both direct reference and descriptivist theories. In the hybrid theory sketched here, we have what is called a ‘causal descriptivist’ theory of the term ‘squirrel’.

Descriptivist theories are often associated with conventionalism about natural kinds. According to the latter view, it is a matter of convention and practical convenience how things and stuffs are classified into kinds. If it is more convenient than not to classify certain things as forming a kind, then it is a matter of convention that those things are so classified. It is not a matter of discovery. Now, if it is taken to be only a matter of convention (and so not something open to investigation and improvement) which description is associated with which natural kind term, descriptivism about natural kind terms leads to conventionalism about natural kinds.

Nevertheless, the descriptivist need not make that initial assumption, and so need not be committed to conventionalism. A descriptivist theory about the semantics of natural kinds is consistent with realism about scientific classification. According to the latter view, it is a matter of discovery, not convention, how things and stuffs are classified into kinds. Science is engaged in a fallible investigation of what kinds there are and of which things and stuffs are in which kinds. The description semantically associated with a natural kind term such as ‘squirrel’ can explicitly allow that much about squirrels is unknown and that scientific investigation can fill out the description.

Lastly, Kripke (1972) and Putnam (1975) presented direct reference theories of natural kind terms alongside their essentialism about natural kind membership: ‘squirrel’ applies to the exemplars and to various other things if and only if the exemplars and the other things have the same essence. These views form an attractive combination, but they are distinguishable and do not stand or fall together.