| Despite the widespread adoption of task-based language teaching in English as a foreign/second language (EF/SL), task difficulty has continued to be a controversial issue in task-based assessment. This research explores how feasible it is to implement a task-based approach to testing within the existing Chinese National English Curriculum (CNEC). The central problem that the study has addressed is how task difficulty can be established through the use of a theoretically-motivated analytic scheme, grounded in contemporary task research. The scheme needs to be able to take a test-task as input, and facilitate rating of the test task on a number of dimensions.;Initially, the scheme proposed by Norris Brown, Hudson, and Yoshioka (1998) was explored for validity and practicality in the Chinese context. Since this scheme did not work very effectively, a new analytic scheme for curriculum-based test-tasks was developed through a series of developmental cycles. At the end of this process, an Input-Processing-Output (IPO) task difficulty matrix was established which provided reliable estimates of rated test-task difficulty at both global and analytic levels. The IPO task difficulty matrix designed and used in the series of studies appears to have adequately distinguished between three levels of difficulty of a CNEC theme.;The scheme provides a practical tool, but also a theoretical perspective on how tasks themselves can be analyzed effectively. The main developmental cycles required experienced and trained ELT professionals to use the analytic rating scheme. A later phase employed less experienced and trained teachers and explored how they could be trained to use the rating scheme. This phase established that the scheme can be used in "non-expert" English language teaching (ELT) situations, provided that there is adequate training. Validity was explored through the use of a range of discourse analysis measures of the writing performance of a number of school-based EFL learners in China.;The research establishes (1) that the difficulty of the tasks can be used to sequence effectively both for testing and for teaching; (2) the methods by which the principles of task selection and task difficulty can be used with a wider range of test generators and test users, and (3) that rated task difficulty connects with objective measures of performance as well as conventional ratings of performance. |
