This question on the nature of what passes for MT instruction has, to the best of our knowledge, hardly been asked, whether in relation to the experiments or with regard to other situations in which it is claimed that the MT is being used . It is all too easy to assume that because there exists some policy requiring the use of the MT at a certain level, or that because glossaries of technical terms have been compiled, teaching therefore takes place in the given language, as opposed to an admixture of that language and some other for instance. Are results of success or failure based on code-mixing, for instance, truly reflective of the advantages or otherwise of MT instruction? The foregoing suggests that research is MTE needs to go in new directions. Among others, research in MTE needs to go beyond pronouncements, policy or otherwise, in order to investigate what actually happens to or with MTs in classroom contexts. Research also needs to yield data that allow for comparison of pronouncements (in respect of policy, materials, manpower development, etc) and the reality of the classroom setting.