Prosodic variation across sampling tasks in normal and dysarthric speakers.

Speech assessments are commonly based on structured elicitation tasks. Despite the value of these tasks, the extent to which their results are a valid reflection of natural speech performance is being increasingly questioned. This is particularly warranted in the light of research findings indicating significant differences in normal speech behaviour across sampling tasks. There is, however, a paucity of research into how disordered speakers' performance varies across elicitation tasks. This study investigated ten prosodic parameters in structured and unstructured speech tasks (reading and conversation) in 12 dysarthric and 12 control subjects. The results suggest that the nature of sampling task affected dysarthric speakers differently to the control group. The implications of these findings for the assessment of disordered speakers are addressed.