deficit, nor may they all have a similar causal relation to reading and spelling problems. One spoken language deficit could be necessary and sufficient to cause the disorder (a prerequisite) and thus be somewhat independent of the other deficits. Put another way, one deficit could be part of the final common pathway leading to deficits in reading and spelling, whereas other deficits could be variable by-products of the different etiologies leading to that final common pathway. Hence, these other deficits would be correlates of the primary deficit and could contribute to the development of the disorder. The various etiologies that lead to dyslexia are unlikely to affect the development of phonological processes in exactly the same way. Thus etiologic heterogeneity could produce a cluster of symptoms with a family resemblance, a "fuzzy set," without there being a unitary underlying deficit that explained the symptom cluster. Moreover, these correlated deficits need not be strongly correlated with the primary deficit in a psychometric or functional way. They are correlates merely because they are part of the "excess baggage" carried by converging etiologies. Stanovich ( 1988) has proposed a related conception in which all the reading-related cognitive skills are continuously arranged in a multidimensional space; at the center of this space is the core deficit that is the cause of dyslexia.
This discussion leads into the answer to our second question, which was concerned with whether the phonological skills in spoken language are a unitary domain. The factor analytic evidence argues against a unitary model and instead supports a multiple factors model. This evidence is complementary with the other evidence indicating a differential relationship between specific phonological processing skills and reading. Moreover, evidence from other domains of neuropsychology should make us cautious about assuming that a single processor underlies a seemingly unitary functional domain (Churchland, 1985). We now know that the brain has multiple visual and memory systems, and that an important dimension for distinguishing these systems concerns their relationship to conscious awareness. Phoneme awareness is usually defined as a metalinguistic skill involving conscious awareness of phonemic segments. Perhaps its place in the domain of phonological processing skills is special because of its relationship to consciousness, just as episodic memory is distinct among memory skills.
This research was supported by the following grants to the first author: A NIMH RSDA (MH00419) and project grants from NIMH (MH38820), NICHD (HD19423), the March of Dimes (12-135), and the Orton Dyslexia Society. Marshall Haith's work was supported by a NIMH RSA (MH00367).