This paper critically examines one particular issue against the background of changes in South Africa's higher education system consequent upon the advent of a non-racial democracy Á the possibility of implementing multilingual instructional polices that include indigenous African languages in its universities. Currently, a great deal of applied linguistic work is being carried out on the creation of word lists and dictionaries, via translation and term creation. This paper concurs that this is a necessary step in language adaptation for the task envisaged. It uses Saussurian semantics to show that translation and/or creation of terms is not a relatively transparent activity. An examination of Saussure's notion of 'semantic value' leads to a post-structuralist concern with use and function. This orientation to language is considered further in light of the discourse-oriented work of M.A.K. Halliday and J.R. Martin on the language of science in Writing Science, Literacy and Discursive Power, published by Falmer Press in 1993. Such discourse unfolds within a community of practice. The practices required for a translated science involve semiotic inculcation, and can be compared to the efforts of 19th-century Christian missionaries who worked on expanding Xhosa for religious purposes. These efforts show the complexities of developing Xhosa as a possible language of science, which any language planner will have to face.