The aim of this paper is to offer further empirical evidence in support of Bernstein's concept of code from the social semiotic perspective of systemic functional linguistics. When the concept of code was first introduced in the 1960s, it was severely criticised by many famous socio-linguists. It will be argued that these socio-linguists failed to understand the significance of language as a meaning potential, and so they also failed to appreciate the inherent relation between meaning and mental development (Halliday, 1975; Vygotsky, 1978). The concept of code is important in any viable theory of mental development which does not hide behind a mythical homogeneity. Human beings develop different forms of consciousness, and code theory offers a detailed and coherent account of both the social origin and the nature of this variation. In fact, the panoramic scope of Bernstein's sociological theory clarifies the relations of social positioning, coding orientation, communication, and consciousness to learning in official pedagogic sites. In presenting some of the results of my investigation in variant forms of communication between mothers and their young children, I will show that: (a) the variation is systematic, (b) the nature of the variation needs Bernstein's code theory for its interpretation, and (c) the fact of variation can only be accounted for by its relation to social positioning.