REVISITING CLASSIFICATION AND
IDENTIFICATION


Barbara K. Keogh 

BARBARA K. KEOGH, Ph.D., is professor emerita. University of California Los Angeles. 

In his invitation to contribute to this special issue, 
LDQ Editor David Scanlon suggested I discuss issues of 
classification and identification. I am pleased to do that 
as I think some of the continuing controversies about 
learning disabilities (LD) are related to those topics. 
More specifically, I suggest that many problems have to 
do with confusion between the two. Despite years of 
effort and an extraordinary increase in the number of 
individuals considered as having LD, we continue to 
grapple with vagaries and inconsistencies in classification, 
definition, and identification. We continue to face 
critical challenges about "what is LD? "who is LD and 
who isn't' LD?" and "how do we know?" 

Our problems are in part related to a lack of clear 
boundaries between LD and other conditions. They also 
relate to definitions that serve political, legislative, 
advocacy, or intervention needs as well as research or 
"scientific" purposes. To muddle the problem further, 
classification problems are compounded by limited and 
often inadequate or inappropriate operational methods 
of identification. 

Problems and confusions in classification and identification 
are not new, and they are not limited to the 
field of LD. Over 40 years ago, Zigler and Phillips (1961) 
published a seminal article titled "Psychiatric Diagnosis: 
A Critique," in which they examined problems in classification 
and identification of psychiatric conditions. 
For those of us in the LD field the article is as relevant 
today as it was in psychiatry then. Zigler and Phillips 
argued that "Reduced to essentials, diagnostic classification 
involves the establishment of categories to which 
phenomena can be ordered" (p. 608). 

Such efforts result in a map or a taxonomy that systematically 
describes a field. In this sense, classification 
is fundamental in organizing any field, and LD is no 
exception. In contrast to classification, identification 
refers to the assignment of individuals or exemplars to 
classification categories and, thus, relates to opera


tional decisions, methods, and measures. Who is included 
and who is not included as LD is a function 
in part of what measures, procedures, and criteria are 
used. Which psychometric tests? Where are the cutoff 
points? How large must a discrepancy be? As fisherman 
have long known, the size of the net affects the size 
of the fish that are netted. 

Zigler and Phillips argued that three criteria were 
essential in determining the adequacy and utility of 
a classification system: homogeneity, reliability, and 
validity. In their analysis, homogeneity refers to the 
similarity (or lack of) of those included in a category. 
Reliability refers to the degree of agreement when placing 
individuals in categories. Finally, validity refers to 
how well category membership informs prognosis or 
treatment. 

Applying these criteria to classification in LD, consider 
first the question of homogeneity. There are different 
definitions of LD and there have been changes 
over the years (see Kavale & Forness, 2000, for discussion). 
The LD classification is defined as much by what 
it isn't as by what it is, and it is accompanied by a long 
list of exclusions, for example, learning problems due 
to generalized cognitive limitations, to social/cultural 
conditions, or to instructional inadequacies. The result 
is a broad band of learning problems gathered under 
the LD rubric, thus posing a challenge to homogeneity. 

In Zigler and Phillips' discussion, reliability raises the 
question of consistency of identification. Specifically, 
is there agreement about who should be included or 
who should be excluded from the category? In LD, 
decision rules about inclusion and exclusion differ 
widely, and identification may be influenced by a 
number of extraneous conditions. For example, where 
you live and go to school and your cultural and familial 
status, in addition to the nature and degree of your 
learning problem, may determine whether or not you 
will be identified as LD. School districts using the 

Learning Disability Quarterly 100 


aptitude-achievement discrepancy model of LD differ 
in what tests are used and in how large a discrepancy 
must be for students to be eligible for services. 
Response-to-intervention (RTI) provides a model of 
identification that differs from the traditional discrepancy 
model, but ultimately confronts the same question: 
After sorting out instructional effects, who should 
be considered LD and who should not? On what basis 
do we make that decision? These are important questions 
as countless individual students are affected, 
some become eligible for needed services while others 
are denied the same services. Because the "rules" that 
define eligibility and the operational procedures used 
to identify who is and who is not LD vary, the reliability 
of identification is questionable, thus threatening 
the meaningfulness of the classification. 

Finally, validity in classification has to do with the 
kind of information classification provides, the nature 
of the correlates of the classification. Applied to LD, 
what does class membership tell us except that an individual 
has a learning problem? Even when individuals 
are reliably identified as LD, as in psychiatric diagnosis, 
Zigler and Phillips noted that "class membership conveys 
... little to the solution of the pressing problems of 
etiology, treatment procedures, prognosis, etc." (p. 612). 
Said directly, classification as LD does not necessarily 
direct treatment responses and may not lead to appropriate 
decisions about what to do. There is a striking 
range of intervention programs for individuals with 
LD, some emphasizing educational techniques, others 
aimed at changing underlying processing problems, still 
others treating LD within a psychodynamic framework. 
Unfortunately, some interventions are based primarily 
on strong beliefs rather than on evidence of effectiveness. 
Sometimes, too, intervention decisions are a 
function of advocacy and availability rather than of 
individual characteristics and needs. 

THE NEED FOR CLASSIFICATION 

Up to this point the discussion has painted a rather 
dismal picture of the state of the LD field, and that is 
somewhat justified. Despite decades of effort, we still 
struggle with questions and problems of classification 
and identification, leading some to ask if there really is 
a condition called LD. My answer is that LD is real and 
that it describes problems that are distinct from other 
conditions subsumed under the broad category of problems 
in learning and achievement; that is, generalized 
and pervasive cognitive deficits or instructional inadequacies. 
Because LD is only one possible reason for 
learning problems, any overarching classification of 
learning problems must allow for other categories. That 
does not minimize the importance of a classification 
system within LD itself; that is, a system of classification 

based on what we know about LD in terms of etiology, 
expression, and prognosis across the developmental 
spectrum. 

It is time to have serious discussions about what LD is, 
rather than what it isn't. As a start, I suggest we separate 
the condition from how it is operationalized. That is, 
that we separate classification and identification, and 
that we focus on the conceptual organization of what 
we think constitutes LD rather than building a classification 
system based on already identified individuals 
and measures. 

When focusing on LD as a classification issue, we 
know a lot about various aspects of LD. There are both 
behavioral and academic expressions. LD may be 
expressed in different academic or content areas: reading, 
arithmetic, spelling, or any of the areas specified in 
the definition proposed by the U.S. Department of 
Education (1997). Indeed, it is the specificity of the 
problem that in part defines LD, and that is a point to 
be emphasized. The last 10 years of research in reading 
has resulted in important steps in understanding reading 
acquisition and reading problems. For example, we 
have solid evidence about the importance of phonological 
processing and its correlates in early reading, and 
we have evidence of effective intervention techniques 
directed at that particular problem. But reading disability 
is only one aspect of learning disabilities. Thus, 
many questions are still unanswered. 

Linking intrinsic processing deficits to specific problems 
in academic achievement and behavior provides 
a promising direction for developing a classification 
system that is clear and conceptually defensible. We 
can expect that there will be different views about the 
nature and specificity of the processing problems. See, 
for example, Torgesen's and Swanson's discussions 
of phonological processing and working memory and 
the responses to them in Identification of Learning 
Disabilities, Research to Practice (2002) edited by Bradley, 
Danielson, and Hallahan. While different, these approaches 
present reasonable arguments that specific LD 
can be differentiated from other problem conditions 
both conceptually and empirically. 

It is encouraging that consensus is beginning to 
emerge on several points: LD is real and is intrinsic to 
the individual; the expression or evidence of an LD is 
related to specific processing deficits; and specific processing 
deficits are different from general cognitive 
delays or deficits. However, we continue to face the 
difficult task of operationalizing the constructs due to 
issues of measurement, which in turn lead to issues of 
reliability in the identification of individuals. A further 
unanswered and controversial issue has to do with intelligence 
and its role in classification and identification 
of LD. Note that I refer to intelligence, not to IQ. The 

Volume 28, Spring 2005 101 


limitations of IQ, especially in the controversial aptitude-
achievement discrepancy identification model are 
well known. Certainly there are problems in how intelligence 
is operationalized, measured, and applied in the 
identification of individuals. There are also continuing 
questions about the role of intelligence in the classification 
in LD. 

Finally, what is required in a classification system in 
LD? At the least it must differentiate LD from other 
problem conditions. It must be comprehensive, covering 
the many expressions of LD. It must take into 
account the hierarchical organization of classes and 
subclasses. It must be equitable, ensuring equal opportunity 
for individuals to be included or excluded. It 
should inform instruction and treatment/intervention. 
Hopefully, it would also inform prevention. And, it 
must satisfy the requirements of homogeneity, reliability, 
and validity. 

The development of a classification system and the 
specification of identification procedures is not a simple 
or easy task. It requires careful and critical thinking and 
discussion, and a great deal of empirical work. Classifications 
and methods of identification will likely change 
over time as we learn more. Yet, they remain important 

to the survival of LD as a field. It is time we confront 
them with reason and with evidence. To return to Zigler 
and Phillips, "Systems of classification must be treated 
as tools for further discovery, not as bases for polemic 
disputation" (p. 616). Good advice. 

REFERENCES 

Bradley, R., Danielson, L., & Hallahan, D. P. (Eds.). (2002). 
Identification of learning disabilities: Research to practice. Mahwah, 
NJ: Lawrence Erlbaum. 

Kavale, K. A., & Forness, S. R. (2000). What definitions of learning 
disability say and don't say. Journal of Learning Disabilities, 33, 
239-256. 

Swanson, H. L. (2002). Learning disabilities is a specific processing 
deficit, but it is much more than phonological processing. In R. 
Bradley, L. Danielson, & D. P. Hallahan (Eds.), Identification of 
learning disabilities: Research to practice (pp. 643-651). Mahwah, 
NJ: Lawrence Erlbaum. 

Torgesen, J. K. (2002). Empirical and theoretical support for direct 
diagnosis of learning disabilities by assessment of intrinsic processing 
weaknesses. In R. Bradley, L. Danielson, & D. P. 
Hallahan (Eds.), Identification of learning disabilities: Research to 
practice (pp. 565-613). Mahwah, NJ: Lawrence Erlbaum. 

U.S. Department of Education. (1977). Nineteenth annual report to 
Congress on the implementation of the Individuals with Disabilities 
Education Act. Washington, DC: Author. 
Zigler, E., & Phillips, L. (1961). Psychiatric diagnosis: A critique. 
Journal of Abnormal and Social Psychology, 63, 607-618. 

Learning Disability Quarterly 102