USING COMPUTER-BASED ERROR ANALYSIS TO IDENTIFY CHILDREN WITH LEARNING DIFFICULTIES

This paper describes the development of a computer-based error analysis procedure for identification of children with learning difficulties. The procedure is based on the use of profiles established through aggregative case survey analysis of seventeen small-scale studies conducted between 2001 and 2005, as well as two large-scale studies conducted over the same period which compared the types of spelling errors made by learners in different grades at primary school. These initial studies produced evidence of distinctive patterns of error made by both mainstream and remedial learners.

Our subsequent studies have focused on the use of the tests and diagnostic procedures we have developed in classrooms and clinics. Based on statistical comparison of both frequencies and proportions of errors made by mainstream and remedial children, as well as regression and discriminant function analyses to establish the types of errors made by children, we have developed procedures for computer-based error analysis which can be applied in classrooms by teachers working with large groups. We have used error profiles for identification of children requiring special treatment programmes, based on the preponderance of particular types of spelling errors. The results of our studies suggest that patterns of written spelling error made by learners may be relevant indicators of learning difficulties in school, for the reason that the types of written spelling errors made by remedial learners are persistent as they progress through school.

The type of information yielded by our procedures can also be used for planning remedial instruction which targets the specific types of errors made by individual children in the classroom, as well as for clinical use in planning individual instructional programmes. Computerised scoring and profile analysis carries with it the possibility that the procedures can be used on a large-scale. Our research indicates that the tests we have developed are both content and construct valid when measured against other commonly available scholastic tests of reading, writing and spelling, and in addition yield detailed diagnostic information on the types of spelling errors made by children. The tests we have developed are also reliable, and have good potential as group tests for use as screening instruments to identify children with dyslexia, tapping error patterns which are predictors of learning difficulties.