A study of prosodic alignment in interlingual map-task dialogues

This paper reports results from a study of how speakers adjust their speaking style in relation to errors from Automatic Speech Recognition (ASR), while performing an Interlingual map task. The dialogues we analysed were collected using a prototype speech-to-speech translation system which adds 3 elements to the communication which we think of as “filters”: Automatic Speech Recognition (ASR), Machine Translation (MT) and Text To Speech (TTS). Our belief is that these filters affect the speakers’ performance in terms of cognitive load, resulting in adaptation of their communicative behaviour. The study shows that the participants do adjust their speaking style and speaking rate as a way of adapting to the errors made by the system. Specifically, the results show that (a) system errors influence speaking rate, and (b) the perceived level of cooperation by the interlocutors increases as system error increases.