In present-day English and most European languages, the normative language is based to a large extent upon the usage of the speakers of the language who have the highest social prestige. As a result, the model of prescriptivism which has been constructed by linguists working on these languages assumes that prescriptivism necessarily reinforces existing social inequalities (e. g. Aitchison 1981;Milroy and Milroy 1985;Crowley 1989;Cleary and Lund 1993; Battistella 1999, inter alia). There are languages, however, where this is not the case, in which the 'correctness' of a form is determined by factors independent of the social status of the people who use it, and the situation in such languages is consistently misunderstood and misrepresented by linguists who apply mainstream sociolinguistic models to them (see, e. g., Ibrahim's 1987 criticism of Ferguson's 1959 interpretation of Arabic diglossia). The present paper, using data from a variety of languages, presents a descriptive framework for the categorization of languages according to a number of different parameters of 'correctness'; within this framework, languages such as English represent only one manner in which the parameters of 'correctness' may be aligned.