This paper presents a teaching sequence designed to enable trainee translators to identify text type / genre. The author’s experience of teaching suggests that students find such identification difficult, and that this undermines their general understanding of what constitutes fidelity to the text. The approach taken is based in the notions of discourse and genre. The paper demonstrates how teaching is sequenced to build student understanding of these concepts, and how to apply them to text identification. Although the approach is useful for many types of text, analysis focuses at this stage on texts from medical discourse. Scrutiny of genres in this discourse helps illustrate how, even within what is often conceived as 'objective' scientific discourse, translation fidelity can still be conditioned by cultural difference. This of course, raises questions as to precisely what is meant by 'parallel' texts.