Adequate speech recordings for intonation research should meet the following conditions: (1) Naturalness, (2) Control of semantic, linguistic and pragmatic content, (3) Standardization of the recording situation. A method is described here using simulated dialogues comprising two alternating roles: Role one serves to set the desired situational framework for subsequent responses and is prerecorded on tape. Role two consists of fixed responses presented on a monitor screen at the appropriate moment in the dialogue. A subject is requested to play the 'life' dialogue partner listening to role one and using the presented responses. His/her spoken utterances are recorded on tape for subsequent analysis. Some possible applications are discussed.