Think before you talk

Measuring a user's level of cognitive load while they are interacting with the system could offer another dimension to the development of adaptable user interfaces. High levels of cognitive load affect performance and efficiency. However, current methods of measuring cognitive load are physically intrusive and interrupt the task flow. Certain speech features have been shown to change under high levels of load and are good candidates for cognitive load indices for usability evaluation and automatic adaptation of an interface or work environment. A speech-based dual-task user study is presented in which we explore the behaviour of speech pause features in natural speech. The experiment yielded new results confirming that speech pauses are useful indicators of high load versus low load speech. We report an increase in the percentage of time spent pausing from low load to high load tasks. We interpret these results within the framework of Baddeley's modal model of working memory and detail how such a measure could be utilized in the cognitive load measurement.