Joseph K. Torgesen, Florida State University
Learning Disabilities Summit: Building a Foundation for the Future White Papers
This paper is available in alternative formats: | Download Word | Download pdf |
Pages: | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 |
In order to justify a recommendation that children be identified as learning disabled by showing they have the kind of intrinsic processing weaknesses that are a central part of the definition, we must have reasonable evidence that the type of learning disability specified in the definition does, in fact, exist. In order to validate the theoretical elements in the definition of learning disabilities from a scientific perspective, all that is required is to show that children with neurologically based, intrinsic learning disabilities are a reality. Even one case of a child with this type of disorder can serve as an "existence proof" for the definition and concept.
However, validation of the definition from the perspective of learning disabilities as a field in special education (which can be considered a social-political-educational movement) is much more difficult. This type of validation requires nothing less than evidence that a significant portion of children currently being served in learning disabilities programs fit the essential elements of the definition. It is on this point that the theoretical assumptions of the definition are most frequently attacked. For example, Jim Ysseldyke and his colleagues have reported on a program of research showing that school-identified learning disabled children cannot be differentiated from other kinds of poor learners on the basis of their patterns of intellectual abilities (Ysseldyke, 2001). In his book The Learning Mystique, Gerald Coles (1987) also mounted an extensive attack on the idea that most school-identified learning disabled children have neurological problems as the basis of their learning difficulties. In fact, he is right in showing that the evidence for this idea is exceedingly weak.
In contrast, the evidence in support of the idea that constitutionally based, intrinsic processing weaknesses can produce important patterns of learning disability in specific children is very strong. I will now briefly review two coherent lines of research that provide support for the concept of learning disabilities as it is presently defined.
This theory, which is perhaps the most completely developed and widely supported current conceptualization of a specific learning disability type (Torgesen, 1999), starts with the observation that children identified as severely reading disabled most frequently experience extreme difficulties in acquiring word level reading skills. Even more specifically, the outcome to be explained by this theory is these children's inordinate difficulties mastering the alphabetic principle in learning to read (Rack, Snowling, & Olson, 1992; Stanovich & Siegel, 1994). These children have extreme difficulties learning to use grapheme-phoneme correspondences to decipher words they have not seen before in print.
The psychological explanation for this overt learning difficulty is that these children have difficulties processing the phonological features of words (Liberman, Shankweiler, & Liberman, 1989). These phonological processing difficulties manifest themselves on a variety of non-reading measures including tests of phonological awareness, rapid automatic naming, verbal short-term memory, and speech perception (Manis, McBride, Seidenberg, Doi, & Custodio, 1993; Stanovich & Siegel, 1994; Torgesen, 1999). Investigation of the relationships between these variables and reading growth has been the focus of intense study over the past two decades, and there is now a substantial body of both longitudinal-correlational (Wagner, Torgesen, & Rashotte, 1994; Wagner et. al, 1997) and experimental (Bradley & Bryant, 1985; Hatcher, Hulme, & Ellis, 1994; Lundberg, Frost, & Peterson, 1988) evidence indicating that differences among children on these language skills are causally related to variability in the rate at which children acquire early word reading abilities.
At the next level of explanation, the neurobiological locus of the specific processing weakness, there is consistent evidence indicating that poor readers exhibit disruption primarily but not exclusively in the left hemisphere serving language. Thus, neurobiologic investigations using postmortem brain specimens (Galaburda, Menard, & Rosen, 1994), brain morphometry (Filipek, 1996), and diffusion tensor magnetic resonance imaging (Klingberg et al., 2000) suggest that there are subtle structural differences in several regions of the brain between children who are learning to read normally and children with reading disabilities. There is also emerging evidence from a number of laboratories using functional brain imaging that indicates an atypical pattern of brain organization in children with reading disabilities. These studies show reductions in brain activity while performing reading tasks usually, but not always, in the left hemisphere (Shaywitz et al., 2000). In a recent summary of the evidence concerning the neurobiological substrate for specific reading disabilities, Zeffiro and Eden (2000) conclude that, "the combined evidence demonstrating macroscopic, morphologic, microscopic neuronal, and microstructural white matter abnormalities in dyslexia is consistent with a localization of the principal pathophysiological process to perisylvian structures predominantly in the left hemisphere" (p. 23). However these authors also hint at the possible need to enlarge our conceptualization of the biological differences between dyslexic and typical children by pointing out that there is emerging evidence for brain abnormalities in these children extending beyond the classically defined language areas.
At the level of etiology of the neurobiological and processing differences that cause difficulties acquiring accurate and fluent word reading skills, there is strong and consistent evidence that these kinds of information processing weaknesses are significantly heritable (Olson, 1999). That is, reading disabilities run in families, and a child with a parent who has a reading disability is approximately 8 times more likely to have a reading difficulty than a child of unaffected parents.
Although the theory of phonologically based reading disabilities is widely accepted at present, there are several interesting problems remaining that are relevant to the issues considered in this paper. The most fundamental question involves the nature of the specific processing limitation that interferes with performance on both reading and non-reading measures of phonological skill. For example, oral language measures of phonological awareness are strongly predictive of difficulties acquiring alphabetic reading skills, but phonological awareness is defined as a kind of knowledge and understanding about words and phonemes, not as a basic psychological processing capability (Wagner & Torgesen, 1987). In other words, deficits in phonological awareness are an outcome of processing weakness and are not a direct measure of an intrinsic processing disability. It is true that children must engage in phonological processing (the processing of phonological information) in order to succeed on measures of phonological awareness, but the processing required on phonological awareness tasks is also supported by knowledge about words and phonemes that is acquired through experience and instruction. The causal relationship between phonological awareness and reading is actually reciprocal (Ehri, 1989; Morais, Alegria, & Content, 1987; Perfetti, Beck, Bell, & Hughes, 1987); differences in initial levels of phonological awareness cause different success in learning to read, and different responses to early reading instruction cause further differences in growth of phonological awareness.
Currently the two leading hypotheses concerning the information processing weaknesses that causes performance difficulties on both measures of phonological awareness and alphabetic reading skill are a speech-specific perceptual processing problem (Studdert-Kennedy & Mody, 1995) and a more general problem processing rapidly changing or rapidly successive acoustic stimuli (Tallal, 1980). It is interesting that measures of neither of these information processing skills are used as widely as measures of phonological awareness to predict the emergence of reading difficulties in young children or to verify the diagnosis of specific reading disability in older children.
Another point of controversy within the theory of phonologically based reading disabilities at present concerns the question of whether rapid automatic naming tasks are primarily measures of phonological processing skill or whether they measure a different kind of processing capability that influences aspects of reading growth other than the initial attainment of accuracy in using alphabetic reading strategies. For example, Wolf and Bowers (Bowers, Golden, Kennedy, & Young, 1994; Wolf & Bowers, 1999; Wolf, 1991) and their colleagues have argued against viewing rapid automatic naming tasks as primarily phonological in nature, and instead they emphasize the visual and speed components of these tasks. They propose that rapid naming tasks assess the operation of a "precise timing mechanism" that is important in the formation of the visually based representations of words that allow them to be recognized as whole units in text. If Wolf and Bowers are correct, this would mean that that an additional (other than phonological), as yet unspecified processing weakness causes reading failure in some children.
A final issue that is important in the present context is that individual differences in phonological awareness, which is the primary measure of phonological processing capability used in research and diagnosis of reading disabilities, are only roughly 50% heritable. The other half of the variability in phonological awareness is produced by environmental factors, such as the language environment in the home and factors related to socioeconomic status (Hecht, Burgess, Torgesen, Wagner, & Rashotte, 2000). Further, we know that phonological awareness and reading have a reciprocal causal relationship (Wagner et al., 1997). Thus, in current practice, we have strong evidence that one of the most commonly used measures of children's intrinsic processing weaknesses in the phonological area is influenced both by constitutionally based differences in processing capability and by environmental/instructional factors at home and school.
Children with nonverbal learning disabilities (NLD) were originally identified by their particularly poor performance on mechanical arithmetic tasks (Rourke & Finlayson, 1978; Rourke, Young, & Flewelling, 1971). Over the past 30 years, Rourke and his colleagues have expanded their description of these children's academic difficulties to include problems with graphomotor skills (early problems with printing and cursive writing), difficulties in reading comprehension, mathematical reasoning, and tasks in science that involve complex concept formation. In Rourke's work, it is the pattern of strengths and weaknesses in academic skills, rather than their absolute levels, that is the most defining feature of the syndrome. Thus, children with NLD show striking weaknesses in math computation skills relative to their word recognition and spelling skills. Their deficits in reading comprehension are also relative to their much stronger word-level reading skills. These children show persistent difficulties in academic subjects that require problem solving and complex concept formation relative to their strengths on tasks that require simple rote learning. Children with NLD have also been shown to have quite severe social/behavioral problems.
Rourke's theory does not identify intrinsic cognitive deficits within an information processing model of mechanical arithmetic or other academic outcomes. Rather, he describes these children's intrinsic processing weaknesses in terms of a pattern of neuropsychological assets and deficits. The theory indicates how a core of primary neuropsychological difficulties involving tactile perception, visual-spatial-organizational skills, and complex psychomotor functions lead to a variety of difficulties with academic and social/behavioral outcomes.
Rourke's theoretical description of children with NLD also includes explicit discussion of areas of normal cognitive development. Early in development, these areas of strength include auditory perception, simple motor behaviors, and rote memory ability. Later, these intact areas of functioning produce normal levels of skill in phonological processing, receptive language, verbal knowledge and associations, and verbal output. As with outcomes at the academic and social skill level, it is the pattern of performance (strength versus weaknesses) that is most important in identifying children for the diagnosis of NLD. Thus, it is performance deficits on visual-spatial-organizational skills relative to performance on measures of vocabulary, or relative deficits on measures of complex motor versus simple motor skills, that are considered the most reliable indicators of the diagnosis.
The major locus of neurological impairment in children with NLD, according to Rourke's theory, is in the right cerebral hemisphere. Specifically, he states that
"the necessary condition for the production of the NLD syndrome is the destruction or dysfunction of white matter that is required for intermodal integration. (For example, a significant reduction of callosal fibers or any other neuropathological state that interferes substantially with 'access' to right hemispheral systems [and thus, to those systems that are necessary for intermodal integration] would be expected to eventuate in the NLD syndrome)"(Rourke, 1988, p. 312).
According to the theory, each individual will manifest specific aspects of the NLD syndrome depending upon both the total amount of white matter that is affected and upon the location and stage of development at which the white matter was damaged.
In terms of etiology, Rourke views the NLD syndrome to be the "final common pathway" for a number of different conditions that produce white matter disease or dysfunction (Rourke, 1995). Examples of such conditions include head injury involving shearing of white matter, hydrocephaly, treatment of acute lymphocytic leukemia with large doses of X-irradiation for a long period of time, congenital absence of the corpus callosum, or significant tissue removal from the right cerebral hemisphere. Other etiologies that might produce the kind of white matter destruction or dysfunction associated with the NLD syndrome include teratogenic effects between conception and birth and extremely low birth weight itself. At present, there is no evidence that NLD is transmitted genetically, except as specific diseases that produce white matter damage may be transmitted genetically (Rourke, 1995).
If it is true that Rourke's neuropsychological assessments are valid measures of intrinsic cognitive processing weaknesses in children with NLD, then his approach to assessment could serve as a prototype for the type of process-oriented direct diagnosis of learning disabilities that is being discussed in this paper. The diagnosis does not depend on the presence of a discrepancy between general IQ and academic achievement, but rather on the identification of a pattern of strengths and weaknesses in neuropsychological functioning that are the core of the learning disability being identified. For Rourke, the diagnosis is made at the neuropsychological level, and the academic and social outcomes are simply the common expression or phenotype of the disorder (Rourke, 1995).
One continuing weakness of the theory of nonverbal learning disabilities, from the present point of view, is that it does not clearly specify how the cognitive, or neuropsychological, limitations of NLD children actually produce the primary academic symptoms such as difficulties with mechanical arithmetic. A useful addition to the theory would be the development of a more complete information processing model of their problems acquiring arithmetic skills. This model would add to the theory in two ways. First, it would help to refine our understanding of NLD children's specific difficulties in acquiring arithmetic skills in a way that might suggest remedial interventions. Second, it might also help to clarify or validate theoretical statements about the underlying cognitive limitations of NLD children. Such a model is important if tight theoretical links are to be established between the academic performance problem and the intrinsic cognitive disabilities of NLD children. A problem that is associated with this latter issue is that the theory has been developed primarily from clinical observation and interpretation of empirical data that is exclusively correlational. Although statements about causality in the theory are embedded within a comprehensive model of neuropsychological development, they have not been subject to rigorous analysis of causal relationships in longitudinal-correlational studies or treatment-intervention studies.
There is very strong evidence that current definitions of learning disabilities are, in fact, a valid description of the learning difficulties of many children. For the theory of phonologically based reading disabilities, the strongest evidence for intrinsic cognitive weaknesses as the cause of the reading disability comes from the consistent evidence that the phonological component of reading skills is strongly heritable (Olson, 1999). For the NLD theory, this evidence comes from neuropsychological studies of brain-behavior relationships in which specific anomalies within the central nervous system have been reliably associated with patterns of performance on neuropsychological measures (Rourke, 1995).
In spite of this evidence for the validity of the intrinsic processing component of the definition, however, serious problems remain in terms of reliable assessment of the critical processing weaknesses that are causally related to the academic outcomes. In the case of phonologically based reading disabilities, the tasks that are most commonly used to predict reading disabilities or to establish the diagnosis in older children are not direct measures of the processing weaknesses that are fundamental to the disorder. In the case of NLD, evidence for causal relationships between the specific neuropsychological problems identified in the syndrome and the academic outcomes associated with the syndrome is still relatively weak.
Funding for NRCLD is provided by the U.S. Department of Education, Office of Special
Education Programs. Award No. H324U010004. Opinions expressed herein are those of the National
Research Center on Learning Disabilities and do not necessarily represent the position of the U.S.
Department of Education.
© 2007 National Research Center on Learning Disabilities
Last Updated: November 19, 2007
Contact: nrcld@ku.edu