Segmentation in models of reading

A componential approach to understanding reading and its difficulties in preadolescent readers

The present study is predicated on the logic of interrelated functional information processing components as an approach to understanding reading and its difficulties in preadolescent readers. The structural equation modelling (and its variants) involved these three latent components: (a) orthographic/phonological component, (b) morphological component, and (c) sentence and paragraph comprehension component. These components were subserved by a total of ten measurable tasks, all administered on-line via the microcomputer under laboratory conditions with reaction time measures as indices of mental representation of word knowledge and sentence/paragraph comprehension. The latent dependent component of reading performance was subserved by standardized vocabulary and reading comprehension tests. The total sample consisted of 298 children in grades, 4, 5, and 6. Maximum likelihood analyses using LISREL show that the data in general do not disconfirm the proposed model for grade 4 readers. The three-component model, with some variables set free, provides a reasonable fit for the grade 5 data but less claim could be made about the goodness of fit for grade 6. The results show the mutually reinforcing and mutually facilitating effects of multilevels and multicomponents of reading. Word structure and word knowledge are particularly predictive of reading. The present Phase 1 work would be validated in a follow-up of another cohort of readers and would also lead to the systematic training of some of the components with poor readers.