Fast and slow namers: benefits of segmentation and whole word training.

Poor readers in Grade 2 (mean age 7 years 7 months) were categorized into fast and slow namer groups based on their performance on a Rapid Automatized Naming (RAN) task. The fast and slow groups were then trained to read words using 3 different training regimes: one that taught onset/rime segmentation, one that taught phonemic segmentation, and one that used whole word repetition. The main results were that the slow namers acquired the words more slowly across experiences than the fast namers, irrespective of training condition, but they were particularly disadvantaged when trained with word-level units. Unlike beginning nonreaders, poor Grade 2 readers showed poorer retention following onset/rime training compared with phoneme or word level training, even when final level of learning was controlled. Further, they showed the best generalization to reading new words and nonwords following phoneme training and the worst following whole word training, even when final level of acquisition was controlled. The data are related to the P. G. Bowers and M. Wolf (1993, Reading and Writing, 5, 69-85) double-deficit hypothesis and to the specific deficits associated with early reading failure.