This study is concerned with the introduction of synthesized speech into the human - computer interface. A simulated hotel booking system is described which asks for information via either the screen or a DECtalk speech synthesizer. The degree of redundancy within the computer's response set is varied. The perceived 'friendliness' of the system increased with the use of synthesized speech incorporating redundancy. In addition, this type of output influenced the quality of the users' responses, such that they 'modelled' the computer's speech.