Questions in everyday discourse consist of a situated exchange in which the questioner and answerer are in a roughly symmetrical relationship in which each is entitled to request information from the other. Questioners typically do not have the information that they are requesting. The answerer is not obliged to answer, but there is a normal Gricean expectation that the answer will provide the information requested. Courtroom questioning differs markedly, in that lawyers usually have a particular version of events in mind that they are attempting to confirm with the witness. Usually witnesses are compelled to answer, and do not have the right to ask questions. Therefore courtroom questions differ from everyday questions in both their social and their information characteristics. These differences mean that courtroom questions are different from everyday questions along a range of linguistic parameters. At the overall narrative or spoken text level, the lawyer is constructing a version of events element by element – neither he nor the witness normally provides a full narrative during the interaction. At the exchange level, normally only the lawyer asks questions, and only the witness answers questions – an asymmetrical pattern – and evaluative lawyer third parts are common. At the level of question structure, coercive grammatical forms are strongly over-represented when compared to everyday conversation.