Validation of a large-scale task-based test: Functional progression in dialogic speaking performance

A list of language functions is usually included in task-based speaking test specifications as a useful tool to describe target output language of test-takers, to define TLU domains, and to specify task demands. Such lists are, however, often constructed intuitively and they also tend to focus solely on the types of function to be elicited and ignore the ways in which each function is realised across different levels of proficiency (Green, 2012). The study reported in this chapter is a part of a larger-scale test revision project for Trinity’s Integrated Skills in English (ISE) spoken examinations. Analysing audio-recordings of 32 performances on the ISE spoken examination both quantitatively and qualitatively, the aims of this study are (a) to empirically validate lists of language functions in the test specifications of the operational, large-scale, task-based examinations, (b) to explore the usefulness and potential of function analysis as a test task validation method, and (c) to contribute to a better understanding of varied test-taker language that is used to generate language functions.