ABSTRACT
This chapter reflects on two reprinted articles, published in 2001 and 2009. It portrays the first as programmatic and offering a general model of performance on tasks, and the second as more informed by a range of results which flesh out the original article without changing its basic arguments. The remainder of the chapter updates what task-based research can contribute to testing and assessment. Major theories of language testing are discussed in light of the task literature. Then the central construct of task difficulty is discussed and it is argued that difficulty for Conceptualiser operations has to be considered separately from difficulty for Formulator operations (in Levelt’s terms). This leads into a revision of the model of assessment provided in the 2001 reprinted article, and also an extended discussion of the construct of ability for use. It is argued that such an ability is central to task-based assessment, and mediates between underlying competences and actual performance.
