Heterogeneity of deficits in developmental dyslexia and implications for methodology

RAND! C. MARTIN ) 0 1 0 Preparation of this manuscript was supported by NIH-NIDCD Grant 00218 to Rice University. The author would like to thank Mary Lesch and Henry L. Roediger III for their comments on an earlier version. Psychology, Rice University , P.O. Box 1892, Houston, TX 77251 ( 1 Rice University , Houston, Texas Most of the studies overviewed by Fanner and Klein (1995) in their review of temporal processing disorders in developmental dyslexia have taken a group study approach in which a group of dyslexic readers is compared with a group of normal readers on some task thought to be relevant to the reading disorder. Because of the acknowledged heterogeneity of deficits among developmental dyslexics, this group study methodology is inappropriate and is likely to lead to findings in one lab that cannot be replicated in another. The single case study methodology, which has been used successfully in the study of adult neuropsychological impairments, should be adopted in the study of developmental impairments as well. In the case study approach, each individual is studied thoroughly with tasks designed to tap the various components of the cognitive domain under study in order to determine which components are spared and which impaired. Data are not averaged, but reported separately for each case. Some recent findings from case studies on developmental dyslexia are reviewed and suggestions are made as to how the case study approach could be used in analyzing whether a temporal processing disorder, or any other hypothesized factor, is causal to the reading disorder. - In many respects Farmer and Klein (1995) have done an admirable job of summarizing and evaluating a large body of evidence on whether groups of dyslexic readers differ from groups of normal readers on various aspects of temporal processing tasks in the auditory and visual modalities. Their review establishes that at least some subset of dyslexic readers perform poorly on such tasks. They have presented data and arguments that provide a plausible case as to how a temporal processing deficit might be related to a deficit in processing speech. That is, given the rapidity of change in the speech signal over time, a temporal processing disorder might lead to im paired development of phonological representations. Thus, they have provided a means of linking a temporal processing disorder in the auditory domain to the phono logical deficits that have been noted in many dyslexic readers. As has been argued by several researchers, a phonological deficit could prevent the development of letter-sound conversion skills and consequently impede the development of reading (e.g., Bradley & Bryant, 1983; Stanovich, 1988). (Farmer and Klein's efforts at linking a temporal processing disorder in the visual do main to possible causes of dyslexia seem less convincing; see the comment by Rayner, Pollatsek, & Bilsky, 1995, for discussion.) The overview that Farmer and Klein have provided will most likely serve the important role of stimulating research on whether the phonological deficits noted in many dyslexic readers might be caused by a temporal processing disorder. In my objection to Farmer and Klein's paper, I am con cerned not so much with the specifics oftheir arguments, as with the general approach to the study of develop mental dyslexia exemplified in the studies contained in their review. Thus, my criticisms are not directed so much at the analyses and interpretations offered by Farmer and Klein, but rather at the previous studies which form the basis of their interpretations. Although Farmer and Klein ackowledge that there may be several different sources of developmental dyslexia, in most of the studies that they review, the researchers have treated developmental dyslexic subjects as a homogeneous group and seem to have sought a single underlying cause of the deficit. Indeed, the history of the field of developmental dyslexia has been a sequence of different unitary expla nations (e.g., unusual distribution of capacities across the two hemispheres, visual perceptual deficits, impairments in the control of eye movements, phonological deficit, temporal processing deficit), with the proponents of one unitary hypothesis seeking to discredit the rest (see Ellis, 1985, for a review). As Ellis notes, a disease model seems to be at work here, in which dyslexia is treated as a disease such as tuberculosis, which one either has or does not, and it is the researchers' task to discover the one cause of this disease. However, dyslexia is more analogous to a symptom, such as a cough, than to a dis ease. Just as a cough can result from many different dis ease processes, dyslexia could result from different un derlying sources. Given the complexity of reading single words and text and the variety of perceptual and cogni tive processes involved (i.e., visual, orthographic, pho nological, attentional, semantic, syntactic, memorial), there would seem to be no a priori reason why only one of these should fail to develop normally nor, alterna tively, why there should be only one underlying compo nent that fails to develop normally and causes the down fall of all. A corollary of the single-deficit approach is the use of group study methodology, employed by most of the stud ies reported in the Farmer and Klein paper. That is, a group of dyslexic individuals (defined as having poor reading abilities relative to other cognitive skills) is com pared with a control group on some measure thought to be related to the reading disorder. This type of group study methodology was common in studies ofadult neuro psychological disorders in the past, but it has come under increasing attack in recent years (Caramazza, 1984; Cara mazza & McCloskey, 1988; Ellis, 1987). The basic prob lem with the group study approach is that it is theoretically unjustifiable to average the performance of individuals who have different functional deficits. In the worst case, the average performance may characterize none of the individual subjects' performance. In the best case, the av erage performance may reflect a majority of cases, but hide the existence of divergences from the average in a significant number of cases. Unfortunately, when a sta tistically significant difference is obtained between the disordered groups' performance and that of the control subjects, the inference often seems to be drawn that the difference characterizes most, ifnot all, of the individu als in the disordered group. In fact, the distributions might be nearly entirely overlapping but with small differences between the group means that are significant with large enough sample sizes. 1 For example, as is indicated in Farmer and Klein's paper, in one of the earlier studies on a possible temporal processing disorder in disabled read ers, Tallal (1980) reported that 12 of 20 disabled readers scored within normal range on the temporal processing task even though a significant group differen (...truncated)