From Learner Corpora to Curriculum Design: An Empirical Approach to Staging the Teaching of Grammatical Concepts

A learner corpus with texts allocated into proficiency levels is a useful resource when designing a curriculum for EFL grammar education, as it can provide insights into which grammatical features are most critical to the learner at each stage of their progress. However, no unproblematic methodology has arisen for using learner corpora to inform curriculum design. Some works have compared the degree of usage of grammatical features by learners with native writers, in an attempt to identify over- and under-use of features by the learner, and thus to take corrective measures. However, differences in usage levels between native and learner populations does not show exactly when in a grmmar curriculum the feature should most critically be taught to learners. Hawkins and Buttery (2010) propose using levels of usage (or negatively, levels of error) at each proficiency level to identify to which level a feature is criterial. Where the level of usage of a feature at one level is significantly different from the level below, it is criterial to that level. Unfortunately, for our data, many features (and error types) differ significantly on level after level, usage at A2 differing from A1, B1 from A2, and so on. So no clear indication is available as to which level the feature most belongs. This paper proposes an alternative approach: instead of attempting to assign features to proficiency levels, we order the features in relation to each other. The learner corpus is used to produce an ordering of grammatical concepts in terms of increasing difficulty for acquisition.