Received wisdom in Primary science classroom teaching advocates the use of open questions for elicitation. The socio-cultural perspective on discourse, however, leads one to question the usual classification of questioning by its grammatical formality and leads instead to an analysis of the discourse in the socio-cultural context which includes the background of the children. In this paper we develop this view, motivated by a desire to explain anomalies to the empirical trend found in particular schools/classes in deprived areas: closed questions are preferred as part of the restricted code and yet open questions are unexpectedly found to be successful when anaphoric cohesion is evident. We suggest that "open" and "closed" are not generally "sufficient" concepts, and after Bernstein, Halliday and Hasan we offer "anaphoric and exophoric cohesion", in "restricted and elaborated" codes as appropriate analytic tools in a socio-cultural semiotic perspective on discourse.