Reading on the computer with orthographic and speech feedback

Abstract

In this paper we present an overview of a computer program directed toward the remediation of children's deficits in word recognition and phonological decoding. In the present studies, 138 children read stories on the computer, in their school, for a half hour per day during a semester. Children were trained to request synthetic-speech feedback (DECtalk) for difficult words by targeting the words with a mouse. Different groups received whole-word feedback, wherein targeted words were highlighted and spoken as a unit, or segmented feedback, wherein segments of words (onsets, rimes, or syllables) were sequentially highlighted and spoken by the computer, requiring the child to pay attention to and blend the segments. Both whole-word and segmented feedback resulted in almost twice the gains in standardized word recognition scores compared to control groups that spent an equal time in their normal remedial reading program. Most important, the computer-trained groups improved their phonological decoding of nonwords at about four times the rate of the control group. However, there was a significant interaction between level of deficit severity and optimal feedback condition. The most severely disabled readers showed the largest phonological decoding gains from syllable feedback, while the largest gains for the less severely disabled readers were from onset-rime feedback. The disabled readers' level of phonological awareness at pre-test was the strongest predictor for gains in word recognition and phonological decoding. Implications of the results for future training programs are discussed.