According to M.A.K. Halliday the language can be usefully considered a particular form of behaviour, which is identified in "meaning", in "making choices of meaning". Thus the language is not an "abstract" and "unconditioned" knowledge, but a real behaviour, socially and contextually conditioned above all by the social functions that it must accomplish in the various social contexts.
Nevertheless the author doesn’t consider conditioning and determinations in a mechanistic vision. The speaker chooses what he/she wants to say or communicate but he/she should carefully choose the favourite option of meaning in that particular social context: the social context “advises” of what can be said, what is better to say. Besides the social context often determines the number of usable options of meaning and the features of those options, beyond the speaker’s awareness. They are universes of meaning that Halliday calls "semantic potential". This theoretical basis allowed B. Bernstein and his team to specifically and empirically investigate a "classical" theory of anthropology and sociology, before him only stated and generically supported: the learning and the use of the language are socially determined. In the author’s opinion this theoretical basis allows to outline a different and "more productive" epistemological approach to the "language", and new solutions to the practical problems regarding the language learning.