If, as I have suggested, the concept of ‘racism’ has been poorly defined and its implications vaguely understood, how then is it best to approach the study of ideas that affect or result from the relationship between the races (or, more specifically in this context, between people allocated to different colour categories)? I claim first, that the study of discourse is the best way of treating ideas about race, second, that ‘racial discourse’ is a narrower category than ‘discourse dealing with race’ and third, that discourse is most usefully studied in social context.

First, it is best to accept for methodological reasons that only ideas as expressed in language must constitute the subject matter of the study. This is not to assert that ideas cannot exist independently of a public language, nor is it to deny that communication with others is possible without use of the spoken or written word – natural and conventional signs are frequently available. For the sake of simplicity, however, I intend to confine my attention to ideas as expressed in language, and, moreover, to publicly expressed language capable of being used in communication between two or more persons: what is referred to here as ‘discourse’.

Although it is useful to recognise that there are likely to be correlations between linguistic expressions, affective states of mind, and social behaviour, and these are of utmost significance in the field of race relations, discourse – unlike prejudice – is not defined in terms of propensities to feel or act in particular ways.