Selective predictive value of rapid automatized naming in poor readers

This study considers the differential predictive value of rapid naming tests for various aspects of later reading, where the differential is between nondisabled and poor readers. Two large-N longitudinal samples of students who have been evaluated from third through eighth grades are studied: (a) a randomly accessed, normally distributed group including students with varying degrees of reading ability (N = 154), and (b) a group of poor readers whose single-word reading in third grade is at or below the population 10th percentile (N = 64). Outcomes in fifth and eighth grade were measured in both groups. Single-word reading in both grades was strongly predicted from third-grade rapid naming only within the poor readers, even when IQ, socioeconomc status, and third-grade single-word reading were statistically controlled. Although rapid naming had predictive value within the large, normally distributed group, its predictive power was entirely absent in the average-reading nondisabled students who were between the 10th and 90th percentiles (n = 122). The fact that rapid naming has predictive power only for poor readers but not for average readers is interpreted as suggesting that impaired readers are qualitatively different from the normal-reading population and are not simply the "tail" of a normal distribution of reading ability. It also seems that it is the automaticity of retrieval, not the knowledge of names itself (as in confrontational naming tasks), that gives the predictive power in rapid naming. These data are considered in light of the one- and two-factor theories of the underlying processes involved in reading disability or dyslexia.