How relevant is connectionist modelling of reading to educational practice? Some implications of Margaret Snowling’s article

Abstract

Why is it that we can read ‘gave’ and ‘have’ in quick succession without even noticing that the words have identical spelling patterns but different pronunciations? Why is it that we read ‘eat a mint’ and ‘drink a pint’ without any confusion about the pronunciations of mint/pint? Plaut, McClelland, Seidenberg & Patterson’s (1996) connectionist modelling of ‘parallel distributed processing networks’ provides an explanation which Margaret Snowling considers the most plausible metaphor in accounting for the cognitive research findings that she reviews. In commenting on Snowling’s article, I shall focus mainly on the practical implications of Plaut et al.’s model for the teaching of reading and for our understanding children with reading difficulties.