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 |
From the material discussed thus far, it is clear that direct diagnosis of learning disabilities by assessment of the intrinsic processing limitations that cause them has a number of important advantages over current discrepancy-based approaches. Conceptually, the process assessment approach is more consistent with definitions that specify deficits in psychological processing capabilities as the proximal cause of poor academic outcomes in children with learning disabilities. As we have just seen, the process approach to diagnosis would also support early identification and intervention as well as targeting of instruction to both the children and in the specific cognitive/neuropsychological areas of greatest need. However, the utility of approaches that emphasize assessment of psychological processing strengths and weaknesses is also critically dependent upon a knowledge base about human learning and cognitive functioning that is not available now, nor is it likely to be available in the immediately foreseeable future. In this section I will briefly describe a range of difficulties that preclude the widespread use of process-oriented approaches to the diagnosis of learning disabilities in present practice.
Current federal regulations specify that children may be identified with learning disabilities that affect learning outcomes in any one of seven areas: 1) oral expression, 2) listening comprehension, 3) written expression, 4) basic reading skill, 5) reading comprehension, 6) mathematics calculation, or 7) mathematics reasoning. A well validated theory of each of these types of learning disorder is required to support the kind of diagnostic approach being evaluated in this paper. To justify diagnosing learning disabilities by assessing the intrinsic psychological processing weaknesses that supposedly underlie them, we must have a well-established understanding of the nature of those processes. Our theoretical understanding of each of these areas of learning disability must start with agreement about how the learning problem, at the outcome level, is to be specified. For example, what exactly is meant by a problem in "written expression"? Can children have more than one kind of problem in this area? If so, what are the several types (and how should each be measured)?
Next, we must be able to identify the specific psychological processing weaknesses that cause the problem with learning outcome. This is extremely difficult to do: It requires several lines of converging evidence to be at least reasonably confident about causality in psychological theory. For example, in the well developed theory of phonologically based reading disabilities, we have evidence from three lines of research that phonological weaknesses are causally related to problems acquiring basic reading skill. In the most convincing research, phonological processing weaknesses have been indexed by performance on measures of phonological awareness. Evidence that individual differences in phonological awareness are causally related to the early growth of alphabetic reading skills comes from: 1) both standard and causal modeling studies of longitudinal-correlational data (Mann, 1993; Stanovich, Cunningham, & Cramer, 1984; Wagner et al., 1994; Wagner et al., 1997); 2) studies showing that older reading disabled children are more impaired in phonological awareness than younger, normal readers matched to them on reading level (Bowey, Cain, & Ryan, 1992); and 3) true experiments that show improved growth in word-level reading skills as a result of prior training in phonological awareness (Cunningham, 1990; Hatcher, Hulme, & Ellis, 1994; Lundberg et al., 1988; Torgesen, Morgan, & Davis, 1992).
When performance on rapid automatic naming tasks is used as a marker for phonological processing difficulties (or some other processing disability), there are two sources of evidence for their causal role in the development of early word reading ability: 1) standard and causal modeling analyses of longitudinal-correlational data (Felton & Brown, 1990; Wagner et al., 1994, Wagner et al., 1997; Wolf & Goodglass, 1986); and 2) differences between younger normal and older reading disabled children matched for reading level (Bowers et al., 1994).
There are beginning attempts to specify the psychological processing problems associated with other forms of academic failure (Berninger, 1994; Berninger & Graham, 1998; Geary, 1993; Geary, Hamson, & Hoard, 2000; Rourke, 1995), but none of these theories is as well developed as the theory of basic reading difficulties caused by phonological processing weaknesses. As an illustration of the difficulties involved in establishing causal relationships between intrinsic processing limitations and learning outcomes, consider the work of Lee Swanson and his colleagues in studying the role of domain-general capacity limitations in working memory as a cause of problems in word reading ability, reading comprehension, and math calculation skills. In a careful and extensive series of studies (Swanson, 1994; Swanson, 1999; Swanson & Alexander, 1997; Swanson & Sachse-Lee, in press), Swanson and his colleagues have shown that children with learning disabilities in either reading or math perform more poorly than typical learners on measures of working memory that require children to both store and process information at the same time. Because of specific correlational patterns in the data, Swanson argues that at least part of the math and reading difficulties of these children is caused by a domain-general limitation in working memory capacity. Apart from the difficulties inherent in arguing the presence of a constitutionally based domain-general processing weakness as the cause of specific learning disabilities, Swanson also must establish that this domain-general capacity limitation is the cause of the learning problems and has not, in fact, been caused by them. Thus far, none of the three categories of causal evidence described earlier consistently supports the hypothesis that constitutionally based, or inherent, domain-general limitations in working memory capacity actually cause specific learning disabilities in reading or math.
It is, in fact, very likely that almost all children with learning disabilities will show performance problems on complex measures of working memory, because these tasks draw so heavily on a variety of knowledge and skills that are acquired during successful learning experiences. Siegler (1998) makes explicit the difficulties involved in interpreting performance problems on these tasks when he describes the various factors that can contribute to differences between older and younger children on many memory tasks:
One explanation is that older children have superior basic processes and capacities. In terms of the computer metaphor, this means that development occurs in the hardware of the system--its absolute information processing capacity or its speed of operation. A second explanation emphasizes strategies. Older children know a greater variety of strategies than young children and use them more often, more efficiently, and more flexibly. A third explanation highlights metacognition--knowledge about one's own cognitive activities. Older children better understand how memory works; they use this knowledge to choose strategies and allocate memory resources more effectively. Finally, older children have greater prior knowledge of the types of content they need to remember or process; this greater content knowledge may be a major source of their superior memory. (p. 178)
It is easy to imagine how the functional capacity of working memory will be affected by the chronic learning failures experienced by children with learning disabilities. Since early failure affects motivation to learn or succeed in school, children become less engaged in putting consistent effort into school learning tasks (Kistner & Torgesen, 1987 ). Not only does this affect acquisition of new knowledge across many domains, but it also undermines growth in the control processes and strategies that help children adapt successfully when asked to perform complex tasks such as those used to measure working memory. In this way, it is plausible that domain-general limits in the functional capacity of working memory would be characteristic of many children with learning disabilities. While it is important to know about these domain-general limitations (because they will affect these children's adaptation to new learning challenges and limit their success on complex tasks), if they arise as a result of chronic failure caused by other domain-specific processing limitations, then they are secondary characteristics (Torgesen, 1993) and not the kind of intrinsic processing limitations specified in the definition.
In sum, to support the widespread application of a diagnostic process that involves the identification of intrinsic processing disabilities, we will need substantial concurrence about what the critical intrinsic processes are that affect every type of learning disability specified by the definition. For almost all of the learning and skill outcomes specified in the definition, it is possible to find an isolated study (or a single investigator) that purports to have discovered a unique processing disability to explain the learning difficulty. However, emerging speculative scientific understanding is not sufficient justification for advocating widespread, everyday measurement of these processes by school psychologists or diagnosticians. For this level of application, we require converging evidence from many different investigators, as well as compelling theoretical descriptions of the mechanisms by which the processing disability acts as a proximal cause of the learning difficulty.
In his discussion of the difficulties involved in diagnosing the presence or absence of specific processing capabilities in children, Flavell (Flavell et al., 1993) described them as "many, varied, and very, very troublesome"(p. 320). In their most general form, these problems arise because of the complex organization and interactions among processes and knowledge in all academic learning and performance outcomes. As Flavell and his colleagues point out,
the mind is a very highly organized device, one whose numerous 'parts' are richly interconnected to one another. It is not a collection or aggregate of unrelated cognitive components, but rather a complexly organized system of interacting components...each process plays a vital role in the operation and development of each other process, affecting it and being affected by it. This idea of mutual, two-way interactions among cognitive processes is an exceedingly important one. (p.3)
Any deficit in academic outcome or performance that fits the definition of a learning disability always involves a complex admixture of a processing weakness (or weaknesses) present at some point in development (perhaps not even concurrently present), an instructional context in which that processing weakness operates, the child's motivational and emotional reaction to the learning difficulties caused by the processing weakness, and the domain-specific knowledge acquired to support performance on the task. As children become older and acquire longer learning histories, measurement ambiguity increases until, when measuring a "psychological processing disability" in a 9-year-old child with suspected learning disabilities, it is extremely difficult to be certain that what we have identified is a constitutionally based, or intrinsic, processing disability.
With this general description of the complexities of cognitive diagnosis as a background, let us now consider three specific problems that make diagnosis of learning disabilities by identifying intrinsic processing weaknesses a daunting prospect. First, psychological processing weaknesses in school-aged children can be identified accurately only by multiple measurements that vary from one another in theoretically meaningful ways. For example, in order to establish that a child has specific difficulties processing rapidly changing or rapidly sequential aspects of the auditory signal (Tallal's temporal processing hypothesis), one would have to present a series of stimuli that required processing across varying temporal durations. Only if the child showed an aberrant effect of rapid, as opposed to slower changes, could one infer that the child was particularly affected at rapid rates of change. Since measurement of many of the basic processing skills underlying poor performance on academic tasks is likely to require very precise delivery of stimuli and/or precise measurement of response times under conditions that eliminate potentially distracting or confounding stimuli, there are likely to be enormous practical difficulties involved in assessing the basic processes and capacities that are alluded to in definitions of learning disabilities.
Another problem with assessing basic processes and capacities is that, as we attempt to assess them outside the context of the task for which they are purportedly required, we run a serious risk of distorting them. As Ericsson (in press) has pointed out, "when investigators design tasks that minimize the relevance of prior knowledge and eliminate redundant stimuli, all these factors combined are likely to induce processes mediating performance that have limited relevance to behavior in everyday life" (p. 12). In other words, humans adapt to the requirements of single, or simple tasks by trying to use the most efficient strategy possible. Strategies that enhance performance on a simple task might actually interfere when the processing skill supposedly measured by that task is embedded in a more complex task environment.
A final difficulty in diagnosing the basic psychological processing weaknesses responsible for difficulties in a particular academic domain is that performance on academic tasks, for which skill is acquired over time, is likely to depend on control processes or knowledge structures that are not required on simpler tasks. These more complex integrative or management processes and knowledge structures will not be assessed when single or elemental processes are measured. The example of long-term working memory is relevant here. When people are first exposed to tasks that are unfamiliar, their performance is tightly constrained by the limited capacity of their working memory. However, these rigid constraints of working memory tend to disappear once individuals have had sufficient skill-building experience with the tasks (Ericsson & Kintsch, 1995). Acquiring almost any academic skill involves acquisition of problem solving routines and knowledge structures that help one to appear more efficient in processing information on that task or in related domains. If relatively small differences in processing capacity or skill give rise to very different learning histories, or if different motivational patterns or learning opportunities produce similar differences in skill acquisition, children will manifest very different information processing skill profiles after several years. The essential point here is that acquisition of academic skills themselves has such an important effect on a child's processing capabilities that it becomes very difficult, indeed, to determine which processing weaknesses are intrinsic and which are acquired.
There are two very difficult problems that severely limit the viability of approaches to the diagnosis of learning disabilities that depend upon identification of intrinsic or constitutionally based psychological processing weaknesses. The first problem is that we do not have a complete understanding of the psychological processing capabilities that are required to attain good learning outcomes in all the areas specified in the definition and regulations. Although individual psychologists, in school or private practice, often speculate about the specific processing weaknesses that underlie a child's academic performance problems, these speculations are most often not supported by reliable scientific evidence. They are a kind of "psychometric phrenology" that has limited diagnostic reliability or instructional usefulness.
The second problem involves technical issues that interfere with the valid assessment of basic psychological processing weaknesses within the complexly organized cognitive systems of children who have substantial learning histories. It is most difficult to know for certain whether performance problems on psychological tests reflect intrinsic processing limitations or whether performance is limited by deficits in acquired knowledge structures and acquired automatic processing routines.
Overall, the foundation for reliable and valid assessment of the intrinsic psychological processing weaknesses of children with learning disabilities is not strong enough to recommend it for widespread application in schools. The premature use of process-oriented approaches to diagnosis and treatment has lead the learning disabilities field down many blind alleys (Hallahan & Cruickshank, 1973; Torgesen, 1979) in the course of its history. Although there is now good evidence that current definitions of learning disabilities are valid for many children, we are still not ready to directly apply the concept of intrinsic processing weaknesses in the routine diagnosis of learning disabilities in school. We will now consider an alternative that, although it does not involve assessment of intrinsic processing weaknesses, is still consistent with the definition and may enable critical early interventions to be more widely applied for children with learning disabilities.
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