Researchers in many countries have observed that middle class children as a group are more successful in the educational system than working class children. More of the former, for example, reach college. Professor Basil Bernstein of the University of London has argued that there is a link between social class and educational failure and that this link is language. In a series of papers published from 1958 to 1973 Bernstein has developed a theory of the 'structure and process of cultural transmission,' or socialization, part of which may be summarized as follows. One characteristic of many (but not all) working class families is that the status of different members of the family is clearly defined the authority of the father, for instance, derives from the fact that he is the father. This type of family Bernstein calls positional, and he contrasts it with the person-centered family type, more common (although not omnipresent) in the middle classes. In the latter, status, authority and interpersonal relationships are 'negotiated' according to the unique characteristics of each family member. This negotiation, or lack of it, is reflected linguistically. The following conversation might take place in a positional family: Child: Can I have an ice cream? Mother: No. Child: Why not? Mother: Because I said so. In order to justify her refusal of the child's request, the mother resorts to her authority as mother. The equivalent exchange in a 'person-centered' family might go like this. Child: Can I have an ice cream? Mother: No. Child: Why not? Mother: Because if you have an ice cream now, you won't want your lunch later on. This time an attempt is made to justify the decision in logical terms. In both cases a 'reason' is given for denying the ice cream, but the 'rational' nature of the explanation given by the second mother leads her to the explicit expression of a statement of condition—'if you have an ice cream now'—and result—'you won't want your lunch later on.' Now Bernstein is not saying that middle class parents are more rational or articulate or intelligent than working class parents. He notes, however, that if this sort of difference distinguishes a large proportion of the conversations these two children hear in their childhood, then it is reasonable to expect the middle class child to enter school, at age five or six, with the ability to understand and produce a more varied linguistic repertoire, a more 'elaborated code' than his working class school friend. The latter may be just as intelligent, but he will probably possess a more 'restricted' linguistic code. Bernstein also recognizes that not all middle class parents' interactions with their children will be like the (imaginary) example quoted, nor all working class parents' conversations with their children like the second (imaginary) example. He argues that the middle class child, however, is more likely to reach school age with mastery of both codes, restricted and elaborated. Many (but not all) working class children, on the other hand, will possess the 'restricted code' only. This may be just as rich and powerful linguistically, just as complex, just as adequate as a means of expression, but it is not the language of the (often middle class) teacher, of books, of schools, or, more generally, of educational success. There are several problems with Bernstein's theory, even in its complete form. In common with other critics, Labov has noted the vagueness of the notion of 'code' and, with another population, shown how differences in the speech elicited from working class and middle class subjects are sometimes the product of the elicitation procedures themselves. Rosen has attacked what he sees as the confused political definitions of several of Bernstein's central concepts, including social class, and the lack of linguistic data with w