Re-examining accuracy measures in second language task-based spoken performance

Abstract

This article reports on a study re-examining the measurement of accuracy in second language task-based performance. Three studies were re-analysed, loosely linked in research designs and experimental variables, but with standardised methods of measuring accuracy. Numerous accuracy measures were used, covering unit of measurement, error gravity, and error linked to length of clause. The contribution of the different measures was compared, largely through effect sizes. It is claimed that some measures are more sensitive than others. Variants of Errors per 100 words were found to be very effective. In contrast the conventional Error Free Clauses measure did not perform consistently well. Clause length accuracy was also distinctive, as was the concept of error gravity. It is proposed that greater understanding of measurement alternatives with accuracy can make a significant contribution to the field, in terms of comparability of studies, as well as the evaluation of theoretical models of second language task-based performance.