AUTOMATED PREDICTION OF THE WRITTEN ERRORS OF TERTIARY LEVEL ESL AND EFL LEARNERS

Abstract

This paper summarizes a two-year project at the Chinese University of Hong
Kong (CUHK) from 2006 to 2008. IELTS Online Writing Assistant (IOWA) is
a computer-based teaching system designed for students in support of their
preparation for the writing module of the International English Language
Testing System (IELTS). Situated in Error Analysis (EA) (Corder, 1981;
Mitchell, 2004; Lightbown, 1998) and Computer-Assisted Instruction (CAI),
the system is designed to address both students' local (i.e., word and sentence)
and global (i.e., discourse) errors. It attempts to predict which types of error
tertiary level ESL and EFL learners in Hong Kong are most likely to make by
assessing their ability to find errors in a prepared script. There are two major
research questions: How effective is the system in predicting students'
predispositions to committing writing errors and to creating discourse
problems? How can teachers better use such a diagnostic tool to complement
their teaching practice to address different issues of error categories? The
study concludes that: (1) its predictive performance varies greatly depending
on the types of error; (2) testing alone—with limited feedback—is effective at
reducing the incidence of certain types of error, especially low-frequency and
structural errors, in students’ productive writing; (3) certain types of error,
which are most difficult for IOWA to predict, could be better instructed by
teachers in an ordinary classroom context. Overall, the study has laid
important foundations for enhancing our students’ IELTS writing test
preparation.

Keywords: Computer-Assisted Instruction (CAI), International English Language Testing System ( IELTS), Error Analysis (EA)