Speaking task design and complexity and fluency features of adolescent English learners

This paper reports the results of a study designed to explore the relationships between speaking test task variables and linguistic features of spoken responses on a speaking assessment for Grade 7 multilingual English learners (age 12–13) in U.S. public schools. Speaking task responses from 30 high-proficiency test takers were transcribed and coded for 14 complexity measures and 9 fluency measures. Subsequent analyses of simple main effects indicate that many of the complexity and fluency measures significantly differ by task variables. We observed more meaningful differences in the complexity measures, but fluency features also varied in the key area of pause time. The findings reveal that tasks designed to elicit extended spoken responses did indeed prompt spoken language with more complex features than tasks designed to elicit shorter, simpler responses across academic disciplines. Results of the effects of task variables on fluency measures indicate that Grade 7 students pause more in response to more complex tasks. The results provide some preliminary validity evidence for the assessment by indicating expected performance differences based on target task difficulty level and have implications for the design and scoring of speaking assessment tasks for young learners.