概況

This study uses a complexity–accuracy–lexis–fluency (CALF) framework to explore performance on 2 speaking tasks: a narrative picture-based task and an interactive decision-making task. A preliminary aim is to compare performance on the 2 tasks, using a wide range of CALF measures to explore where scores are similar and where they are different. However, the research design is within-subjects, and so it is also possible to look at performance by participants across the 2 tasks to explore whether particular CALF measures show more consistency than others, and whether any such consistency is related to hypothesised stages in the process of speaking. The main focus of the study is to use the technique of cluster analysis to examine whether there are performance profiles across CALF measures that go beyond consistency with the individual performance variables. The data suggest that there is something of a tension between a discourse-oriented style—emphasising speed, fluency, and subordination—and a clause-oriented style—emphasising longer clauses, dysfluency, and slow performance. In addition, it is reported that there is some limited style consistency across the 2 tasks, with just under half of the 47 participants using the same style in each task.