Meta-analytic confirmation of the nonword reading deficit in developmental dyslexia

TTn a recent narrative review, Rack, Snowhng, and Aoison (1992) concluded that strong evidence exists for the phonological deficit hypothesis in explamjng severe word reading and spellmg problems that cannot be accounted for by sensory or neurological damage, lack of educational opportumty, or low intelligence. The phonological deficit hypothesis states that in these instances of dyslexia there is a highly specific deficit m the phonological language domain, which ulümately leads to problems in reading and spelling. Dyslexics are supposed to differ from normal readers in those qualitative aspects of reading that emphasize phonological processes. An alternative Interpretation is the developmental lag or delay hypothesis Accordmg to this hypothesis, normal and dyslexic readers differ only in the speed of development, and are equal in terms of qualitative aspects of reading style The developmental lag hypothesis implies that dyslexics will perform poorly on phonological reading tests, but not more so than younger readers at the same reading stage who develop in a normal way. The phonological deficit hypothesis, on the other hand, predicts that dyslexics and (younger) normal readers may have the same word recognition ability but will differ strongly in phonological skills. Rack et al. (1992) describe and analyze a series of studies of nonword reading in dyslexics and reading-level-matched normal readers that may be considered äs crucial tests for the validity of the deficit and the delay hypotheses. In these studies, Snowling's (1980, 1981) paradigm of the nonword reading task has been applied in a variety of ways to assess phonological skill relatively independently of reading abihty. Dyslexics are matched with normal readers in terms of reading level. The readinglevel-matched design is used to compare dyslexics' performance on a nonword task with younger normal readers' performance on the same phonological skill measure. The design controls for differences in reading abiliües that might influence the children's performance on the nonword task. The effectiveness of the design in reaching this goal depends, of course, on the adequacy of the matching procedure. In most studies the matching of dyslexics and normal readers is checked by a word recognition lest that should show only minimal differences between normal and dyslexic readers. Rack et al. (1992) scrutinized all pertinent published studies using the nonword paradigm in the context of the readinglevel-match design. Because the majority of studies showed (a) significant differences in nonword processing between dyslexics and normal readers against the background of (b) equivalence of word recognition abilities in both groups, the authors were convinced that there is "extremely strong evidence for the phonological deficit hypothesis" (p. 49). Furthermore, they analyzed in depth the causes of absence of phonological skill differences in about a third of the studies that seemed to contradict the deficit hypothesis, and pointed to several alternative hypotheses in terms of measures, designs,