Abstract:

This paper focuses on the use of Information and Communication Technology (ICT) to identify error patterns for screening and diagnostic purposes. The information can then be used to design effective remedial interventions. Our presentation focuses on research conducted in South Africa based on procedures developed over a number of years in mainstream and clinic schools. The research involves use of an instrument called the Phonic Inventories which yields information on the types of spelling errors made by children.

Studies thus far have been based on statistical comparison of both frequencies and proportions of errors made by mainstream and remedial learners, as well as regression and discriminant function analyses to establish whether the types of errors made by learners can be used for predictive and classification purposes. Similar methodologies are being used in current studies where spelling error patterns made by children in mainstream classrooms are being used for diagnostic and screening purposes, as well as for planning instruction.

The Inventories test the ability to use phonic skills in writing words and consists of three levels. The test is scored to yield thirteen error types, which are further categorised in terms of the position of the error in the word. Frequencies of error can then be combined to form profiles of the types of errors made by individual children. Proportions of error types can also be derived.

In working with the Inventories in schools and clinics, each level has been found to correlate highly with tests of reading and spelling and to yield reliability coefficients in the region of 0.90. The profiles of errors have been found by teachers and remedial therapists to be helpful for purposes of focusing instruction. Certain types and profiles of errors have been found to be both prevalent and persistent in children diagnosed as having learning disorders. The Inventories can thus be characterised as reliable and content valid as well as construct valid, and to have high potential both in terms of discriminant and ecological validity.

Test data is analysed and profiled via a computer-based analysis and reporting system. Teachers are presently still required to manually score test results and enter this data for analysis. However, once data has been entered users can quickly discern whether a child has potential learning disabilities and devise appropriate remedial strategies. Results can also be profiled against groupings selected by the teacher or compared against previous test scores to clarify where remediation has or has not been effective.

Our research has identified that the process of manually scoring test data is a factor that needs to be controlled as scored results differ depending on the evaluator’s knowledge of phonics and English language orthography. An additional function, currently being debugged, allows teachers to enter raw data which are automatically scored for analysis. Raw research data can be fed into the system via a web interface or scanned into its data base via an Optical Character Recognition system. Actuarial statistical analysis of manually scored data against automatically scored data will confirm the reliability and validity of the automated scoring process.

The time consuming and onerous task of scoring tests manually is therefore reduced and time required to initiate effective remedial action greatly improved.

Keywords:

phonic inventories, screening, diagnosis, remedial intervention, testing, phonic.