Assessing aspects of children's written grammar: automating the process

Abstract

Teachers are expected to make regular assessments of pupils' writing, providing feedback about their strengths and weaknesses. If the computer could assist teachers in this task by effectively analysing the grammar element in children's writing, then the teacher's work load would be significantly reduced. A computer generated ‘text analysis’ of grammatical features would have the benefit of providing diagnostic information systematically and efficiently. Although text analysis is not new, it is generally restricted to applications aimed at adults. The development of a quantitative text analysis program, specifically geared to analysing children's writing was therefore considered profitable. This required an understanding of both the development of children's writing abilities and the criteria of the National Curriculum levels of attainment for the Standard Assessment Tasks (SATs). Ten grammar features were selected for analysis. Stories reflecting the spread of attainment associated with the end of Key Stage 2 were analysed. The computer successfully distinguished positive and negative grammar features within these texts. Nevertheless, an assessment of qualitative genre details made by teachers will still be required, in order to provide a balanced assessment which includes all elements that contribute to effective writing.