In this chapter we turn our attention to what seem like more down to earth, practical matters. Issues like how we train raters and interlocutors, or how a room is laid out for a speaking test. But the practical matters are also extremely important for theoretical reasons. For example, the interlocutor may be exhorted to be ‘friendly’ (Underhill, 1987: 42-3) in a face-to-face speaking test, but this alone is really to miss the point. As we saw in Figure 4.4 the interlocutor (who may also be the rater) plays an important role in the testing process. The interlocutor interacts with the test taker. It is quite possible that the score is in some way affected by the interlocutor and the nature of the interaction. The question that we have to ask is whether this is inevitable in any act of communication in a speaking test. Or should the score be treated as ‘independent’ of the interlocutor and interaction?