It can be argued that just as there are different kinds of literacy, there are different kinds of illiteracy. A proximate illiterate, i.e. an illiterate who has easy access to a literate person, is clearly better off than someone without such access. The existing literature that takes account of these differences (1) defines an illiterate person to be a proximate illiterate if he or she lives in a household with at least one literate person and (2) derives new measures of literacy which typically exceed the standard literacy rate. The latter risks generating policy complacency. The aim of this paper is to suggest a measure of literacy that is not limited by (1) and (2). The measure is axiomatically characterized and its use is illustrated with a numerical exercise for the provinces of South Africa.