Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/232559309

Diagnostic Assessment of Learning Disabilities: From Assessment of


Performance To Assessment of Competence

Article in European Journal of Psychological Assessment · January 1997


DOI: 10.1027/1015-5759.13.1.10

CITATIONS READS

16 6,066

1 author:

Jacques Grégoire
Université Catholique de Louvain - UCLouvain
126 PUBLICATIONS 3,528 CITATIONS

SEE PROFILE

All content following this page was uploaded by Jacques Grégoire on 08 March 2017.

The user has requested enhancement of the downloaded file.


European Journal of Psychological Assessment, Vol. 13, Issue 1, pp. 10–20 © 1997 Hogrefe & Huber Publishers

Jacques Grégoire Diagnostic Assessment of Learning Disabilities

Diagnostic Assessment of Learning Disabilities:


From Assessment of Performance To Assessment
of Competence
Jacques Grégoire
Université Catholique de Louvain, Belgium

Models underlying diagnostic assessment of learning disabilities are seldom explained and discussed. More
often too, those assessments are offered without empirical support. The goal of the current paper is to reflect
on the diagnostic assessment of learning disabilities. A distinction can be made between assessments which
focus on performance and those which focus on competence. Competence-based diagnostic assessments
provide more useful information to practitioners to understand learning disabilities and to organize special
education. Historically, the Piagetian theory opened the way to assessments of competence. But the assess-
ments based on Piagetian theory suffer from methodological shortcomings. In this paper, significant changes
for diagnostic assessments which are due to the recent development of cognitive psychology are emphasized.

Keywords: Diagnostic assessment, learning disabilities, cognitive psychology

The assessment of school learning remains strongly From Performance to Competence


influenced by the behaviorist framework. Teachers’
and school psychologists’ attention is often focus on
the directly and objectively observable reality: the The Concepts of Performance and
student’s performance. This viewpoint can be justi- Competence
fied when the goal of the assessment is the selection,
What is a diagnostic assessment of learning disabili-
the ranking or the certification. In this case, quanti-
ties? This assessment is not obvious and many as-
fying a performance and comparing it to a cut score
sumptions, often implicit, are made.
give enough information to come to a fair decision.
(1) The most important assumption relates to the
But today, greater attention is paid to formative
evaluation and diagnostic assessment and leads to a reality to be assessed. Since the work of Chomsky
(1965), the distinction between performance and
shift in the practitioners’ needs. Diagnostic assess-
competence is widely accepted. Performance refers
ment is not a trivial process. In this paper, the main
assumptions of this kind of assessment are dis- to the observable behaviors produced by a person.
Competence is a set of organized abilities assumed
cussed. The distinction between an assessment
to underlie performance. Competence is not directly
which is focused on performance and an assessment
observable and can only be inferred from perform-
which is focused on competence is especially em-
ance.
phasized. The evolution of the diagnostic assess-
ment focusing on competence is reviewed from the The distinction between performance and com-
Piagetian framework to the cognitivist framework. petence can be used to classify diagnostic assess-
The relevance, but also the shortcomings, of these ments. Some assessments focus on the task and on
frameworks are discussed. The most promising di- performance; they are basically atheoretical. On the
rections for the diagnostic assessment are then indi- other hand, some assessments are rooted in a theo-
cated and the questions which need to be answered retical model of the reality underlying performance.
in the future are emphasized. Several constraints of the diagnostic tools come
from this initial choice. In the case of an assessment
which is focused on competence, the relationship
between the performance and the task used to re-
veal it needs to be validated. We need to be sure that
the selected tasks measure abilities related to the
competence to be appraised. In the case of an as-
Diagnostic Assessment of Learning Disabilities 11

sessment which is focused on performance, no infer- consequently, the score of a complex task is usually
ences to constructs are made and hence the validity equal to the sum of the scores of the basic tasks.
requirements are less important. Still, evidence is In the next section, the first assumption will be
needed that the performance is being recorded. more deeply discussed. Through some examples, the
Also validity evidence is needed to generalize from need to focus the diagnostic assessment on the per-
performance on particular tasks to large sets of son’s competencies will be emphasized. The choice
tasks. Therefore, the choice of tasks for performance of the conceptual framework to assess competen-
assessment can be very important. cies will also be discussed.
(2) A second assumption underlying the diagnos-
tic assessment of learning relates to the importance
given to the context. Some authors (e. g., Resnick &
Resnick, 1990) emphasize that the abilities take dif-
Restrictions of Diagnostic Assess-
ferent shapes according to the context in which they ments Which Focus on Performance
are being used. For these authors, a valid assessment
can only be contextualized. To be correctly assessed, Problems of Score Interpretation
a person has to be confronted with a significant and Most tests designed to diagnose learning disorders
motivating task. The emphasis placed on context is focus on performance. One of the reasons comes
certainly relevant, but it complicates assessment too. from the criteria generally used to define learning
The same context may not be equally stimulating disabilities. Following the DSM-IV (American Psy-
and facilitating for all persons. This raises some chiatric Association, 1994), “learning disorder” is
questions as if equity and valid generalizations can present when the ability measured by a stand-
be made. Furthermore, it is incorrect to assert that ardized test is substantially below the expected one,
some tests are contextualized and others are decon- given the age, the intelligence and the education of
textualized. Sternberg (1990, p. 221) observes that, the person. Table 1 shows mathematics disorders as
even in standardized tests, “the items are contextu- an illustration. The criteria are similar for disorders
alized with respect to the environment in which they of reading and written expression.
are supposed to measure performance-namely, the Tests focused on performance do not provide any
environment of the school.” Consequently, the de- basis for understanding the underlying cause of the
bate is not between contextualized assessment and observed phenomena. As Snow and Lohman (1989,
decontextualized assessment (all assessment is con- p. 268) rightly point out, “Item writers are more
textualized) but rather concerns the extent to which likely to be content specialists working from test
generalizations can be made upon series of contex- specifications that bear no relation to the specifica-
tualized assessments. Messick (1994, p. 18) empha- tions of relevant psychological theory.” Conse-
sizes that, in order to avoid the proliferation of quently, items are usually a sample of relevant con-
qualitatively different abilities according to the con- tent domain knowledge, but not of learning. This
text of use, inferences have to be drawn taking into kind of test does not lead to any explanation of the
account the consistency of performance across con- performance.
texts. This consistency is then considered a reliable
indicator of competence.
(3) Finally, a third assumption of the diagnostic
assessment is related to the legitimacy of breaking
complex competencies into basic abilities. Resnick
and Resnick (1990) defend an assessment which re- Table 1. Diagnostic criteria for mathematics disorders fol-
spects the complexity of the competence. They ar- lowing the Diagnostic and Statistical Manual of Mental
gue that an evaluation which focuses on basic abili- Disorders (DSM-IV).
ties does not take into account the interactions A. Mathematical ability, as measured by individually
among abilities. It is true that a competence is more administered standardized tests, is substantially below
than a sum of abilities. The interactions are impor- that expected given the person’s chronological age,
measured intelligence, and age-appropriate education.
tant, too. However, it is difficult to imagine how a B. The disturbance in criterion A significantly interferes
diagnostic test could avoid analyzing complex com- with academic achievement or activities of daily living
petence into some basic abilities. Furthermore, we that require mathematical ability.
can have doubts as to the possibility to directly ap- C. If a sensory deficit is present, the difficulties in
mathematical ability are in excess of those usually
praise the complexity of a competence. Indeed, the associated with it.
scoring of the results is almost always analytical and,
12 Jacques Grégoire

Table 2. Diagnostic assessment of mathematical knowledge with KEY-MATH-R.


Area Subtests Domains (e.g.,)
Basic concepts Numeration
Rational numbers
Geometry

Operations Addition Models and basic facts


Algorithms: whole numbers
Adding rational numbers
Subtraction
Multiplication
Division
Mental computation

Applications Measurement
Time and money
Estimation
Interpreting data
Problem solving

As an illustration (see Table 2), consider the Key mains behaviorist. No hypothesis is proposed about
Math Revised (Connolly, 1988). This test is explicitly the procedures used by the student to solve a pro-
presented as “a diagnostic inventory of essential blem. And no explanation is proposed to understand
mathematics.” Three main area of knowledge are failures. Thus, when a person does not correctly an-
assessed: (1) basic concepts, (2) operations and (3) swer the question “2/5 + 1/5,” we can say only that
applications. Each area is divided into subtests (e. g., the student is unable to add fractions with a common
“addition,” “subtraction” . . . ) and domains (e. g., denominator. Or, if a picture with three children and
“models and basic facts,” “algorithms: whole num- 18 squares of chocolate is shown to a student and he
bers,” “adding rational numbers” . . . ). The items cannot answer the question “How many squares of
have been carefully selected using the Rasch model chocolate will each person receive if the chocolate
(see, for example, Hambleton, Swaminathan & Ro- squares are divided equally?,” we can deduce that
gers, 1991). This means that the author assumes that, this student has difficulty mastering the operation of
for each subtest, there exists a unidimensional latent division. But, why is this student unable to succeed?
continuum on which the items can be ordered and What is the relationship between this failure and the
linked to a level of ability. But is it defensible to con- abilities of the student? How can this student be
sider some domains of the mathematical knowl- helped? . . . We have no answers to these questions!
edge, e. g., data interpretation or problem solving, as Through this example, it can be seen that an as-
unidimensional? From raw scores, derived scores sessment focused only on performance appears to
can easily be computed: standard scores, scaled be insufficient for real diagnostic assessment. At
scores, grade or age equivalents . . . allowing to com- best, it allows a first appraisal of the problem a per-
pare the person’s scores to the scores of a repre- son is having. But it does not provide a deep under-
sentative sample of the population. standing of the difficulties encountered by a student
Though designed with great care, Key Math raises and, therefore, does not provide a sound foundation
some questions. The division into area and subtests for identifying learning disabilities and how they
is not supported by any theoretical model of mathe- might be corrected.
matics learning. Indeed, we recognize the traditional
divisions used in the mathematics curriculum in pri-
mary school. It is a division into subjects and not an Problem of Assessing the Difference
analysis supported by a model of mathematical Between Observed Score and
learning. The interpretation of the results cannot Expected Score
only be made from the pattern of standard scores,
but also from the item level for which the author pro- Another shortcoming of tests focus on school per-
poses a description in terms of behavioral objectives formance comes from norms which are often ex-
(e. g., “the student can add fractions with common pressed in terms of grade equivalent scores. Grade
denominator”). This viewpoint on the results re- equivalent scores can be more or less accurate. This
Diagnostic Assessment of Learning Disabilities 13

way to express norms is popular because it is easy Diagnostic Processes which Focus
to understand by teachers and by psychologists and on Competence
because it takes into account the developmental na-
ture of the measured constructs. Unfortunately,
The Piagetian Model
grade equivalent scores raise many problems and
can be the cause of serious errors of interpretation. Having emphasized the shortcomings of assessment
The first criticism comes from the meaning of models which focus on performance, the advantages
school grades defined on the basis of the average per- as well as the problems of models which focus on
formance of a sample. Pretending that given speed competence will be considered next.
of reading corresponds to the average performance Historically, the Piagetian theory is the first ref-
of second grade pupils, is a deceitful expression. It erence to an assessment focus which is on compe-
legitimizes the idea of a “normal” level of perform- tence (Inhelder, 1943). The Piagetian model was
ance at the second grade, independently of the con- very appealing to practitioners during the sixties
text of learning. We are then led to think that a pupil and the seventies. But today, it is less popular. How-
having a score below this level of performance does ever, it is still used for the diagnosis of mathematical
not reach the end level of the second grade, as if the learning disorders. Therefore, it deserves some at-
mean score of the sample of second grade pupils tention. It allows to go further than the sole report
were an absolute level. But this is incorrect! The pu- of success or failure and enables to catch the per-
pils assessed can be at a normal level of performance son’s way of thinking. In this sense, the Piagetian
according to a teacher’s requirements. In other model provides an understanding of the observed
words, when we express norms in terms of school phenomena and also quite a number of ways for
grades, we risk intermingling interindividual vari- special education. Unfortunately, the Piagetian
ability with interclass variability. Many researchers model is not without shortcomings and therefore it
have shown that this latest variability is very impor- appears to have limited utility in the assessment of
tant. For example, from the scores of a sample of learning disabilities.
1500 5th grade Belgian pupils, Grisay (1989, p. 3) ob- Criticisms can be illustrated with the assessment
served a significant variability of the average per- of the learning of the concept of number. Let’s re-
formances in French from one class to another. An member that, for the Piagetian epistemology, “the
analysis of the factors of variability showed that person is active, author too and builder of the ob-
“nearly 45% of the differences between pupils ob- ject” (Bideaud, 1991, p. 19). This basic assumption
served on the standardized test are related to the deeply determines the concept of number proposed
school class,” whereas the social class account for by Piaget. According to him, number is not an ex-
only 5% of the differences in performance. ternal reality passively absorbed by the child. On the
The assessment of the discrepancy between the contrary, number is built by the child thanks to his
observed and the expected score is even more diffi- logical abilities. It is self-evident that number is not
cult when the seriousness of the discrepancy needs a property of objects in the same way as color or
to be interpreted. Starting from which value is a dis- hardness. Number is a relation between objects that
crepancy or a pathological feature and therefore a the person builds by reflective abstraction. Its real
learning disability? This question raises a more gen- root is inside the person.
eral problem: the legitimacy of cut scores, that is, a Piaget proposed a psychogenetic explanation of
score which provides a dividing line between one number where logic plays the role of a normative sys-
state which one considers to be normal, and another tem progressively built by the child. From this view-
state, which one considers to be pathological. In point, “the study of the development of number will
most cases, such scores are relative and reflect, at be the one of the operations ‘on number’ and of the
best, a consensus. Obviously, the problem of cut seriation and the classification subsystems which
scores does not arise for the most serious cases found it” (Bideaud, 1991, p. 20). In his book written
where the pathology is self-evident. But, when the with Szeminska, The Child’s Conception of Number
results are around the cut score, the question be- (1941), Piaget specifies the logical abilities that chil-
comes much more difficult to answer and other in- dren progressively acquire and coordinate to master
formation than test scores has to be considered. To the concept of number. These abilities develop al-
base important decisions on the results of single most in parallel and their integration begins before
tests with arbitrary cut points is to participate in a their complete mastery. In this theoretical frame-
flawed process which can only undermine the valid- work, the following operations have to be assessed
ity of diagnostic assessments. when the learning disorders of number is diagnosed:
14 Jacques Grégoire

1. The operation of “seriation,” which consists in or- The student who computes a difference between
dering objects according to their differences. The two positive integers should be aware that the result
child has to distinguish objects considering only must necessarily be included in the highest term of
one or several variables (height, weight . . . ). The the operation. In the previous example, the result
person creates an asymmetrical relationship be- must be logically included in “506.” Consequently,
tween objects and disregards equivalencies. At when students give the answer “544,” they not only
the numerical level, seriation is illustrated by the make procedural errors but also logical ones. With-
acquisition of ordered series of integers: 5 is big- out any computation, students should recognize that
ger than 4 which is bigger than 3. their answer cannot be correct.
2. The operation of classification. It consists in ar- This short example demonstrates the relevance
ranging objects in a set, disregarding their differ- of the Piagetian model for the diagnosis of mathe-
ences and paying attention only to their shared matical learning disorders. But, unfortunately, intro-
qualities. The relationship so created is that of ducing the Piagetian theory in the area of assess-
equivalence. At the numerical level, classification ment is not as easy as it first appears. Many meth-
is the basis of the cardinal side of number. Cardi- odological problems arise:
nal number represents a set of elements consid- 1. Although the Piagetian assessment is strong from
ered as equivalent, in this case as units. a theoretical viewpoint, it is weak from a psy-
3. The conservation of number is intrinsically related chometric viewpoint. Standardization and norms
to the evolution and to the integration of logical are generally inadequate. The assessment is al-
abilities described previously. Indeed, so that the ways a comparison. If they do not follow some
children may be conserving, “empirical knowl- methodological criteria, these comparisons are
edge needs to be interpreted and corrected by not valid nor reliable.
reason” (Kamii, 1990, p. 38). The child who has 2. An other methodological problem concerns the
constructed the logico-mathematical structure of characterization of intermediate behaviors. From
number can tackle problems of conservation a diagnostic viewpoint, it is often insufficient to
from a logical point of view free from appear- classify a behavior as at the operational level or
ances. In order to be free of the sensitive appear- not. A more precise apprehension of the level of
ance, the thought must be reversible: it has to be the cognitive development is often useful. The in-
able to perform the same operation in the two di- termediate levels described by Piaget himself can
rections, while being conscious that it is the same be interesting starting points, but certainly are in-
action. The child so understands that the modifi- sufficient for diagnostic assessment
cation of the spatial arrangement of a set of ele- 3. The main methodological question is the validity
ments does not alter its cardinal. He can mentally of the tasks used for assessment. Are Piagetian
come back to the initial situation and free himself tasks measuring what we think they measure, and
from surface changes. If in a set nothing is added only that? When the assessment is focus on com-
nor subtracted, it keeps the same cardinal despite petence empirical evidence is needed that the se-
the appearance. lected tasks correctly reveal the competence
which is intended to be assess. Numerous studies
Even created 50 years ago, the Piagetian model of about Piagetian tasks has shown that this require-
number still is a reference for diagnostic assessment. ment is very difficult to meet. For example, Fayol
Some students’ errors can only be understood if re- (1990), through an important analysis of research
lated to the mastery of number. Consider the exam- on assessment of conservation of number, notices
ple of a subtraction of which the result is wrong: that the results show sometimes a very important
impact of variables that Piaget considered as sec-
506 ondary. This is the case for the way in which the
– 42 questions are expressed and the spatial arrange-
544 ment of the sets of elements.

If the error is analyzed in terms of procedure, the Consequently, is it possible to conduct a valid assess-
hypothesis can be forwarded that, thinking it was ment of the logical abilities of students? Perhaps it
impossible to compute “0–4”, the student has in- is impossible to assess pure reasoning. This one is
verted the terms and computed “4–0”. The person always expressed in a particular context that influ-
did not follow the correct procedure, that is to bor- ences positively or negatively the student’s per-
row in the hundreds column for the tens column. formance. As it is impossible to remove the influ-
Diagnostic Assessment of Learning Disabilities 15

ence of context, the only solution is to follow Mes- tion of different kinds of knowledge: mathematical,
sick’s proposal (1994, p. 19) of a “cross-contextual linguistic, factual . . . They also pay more attention
assessment.” Only the consistency of results across to the representations and to the kind of knowledge
different contexts can be used as a valid indicator of used by the person. Particularly, the concepts of de-
the student’s level of competence. clarative and procedural knowledge and the rela-
The problem of context is related to a fundamen- tionship between the two (Anderson, 1983) have
tal shortcoming of the Piagetian model. Recall that been widely applied to understand the school learn-
Piaget was not a psychologist; he was a philosopher ing (e. g., Hiebert, 1986; Gagné, Yekovich, & Yek-
and a logician. His viewpoint on the cognitive devel- owich, 1993). This goes together with great attention
opment is normative. Across all his work, his main paid to the level of automaticity of procedures,
goal was to define the logical norms useful for the which is appraised via the speed of process and/or
building and the evolution of knowledge. But, Piaget the resistance to interferences from other tasks per-
did not pay attention to the social, cultural and spe- formed simultaneously. For example, Fischer (1996)
cifically psychological conditions influencing the de- has developed a computer program to assess the
velopment of knowledge. If, the pragmatic aspect of knowledge of elementary numerical computations
knowledge may be unimportant for the logician, it is (addition, subtraction, division and multiplication of
the opposite for the psychologist and the teacher. integers). Across a school year, he observed a dra-
Their goal is not to construct a general model of the matic increase in the speed of 5th grade students’
thinker but to teach and to help persons in particular answers. The speed of answering indicated a shift in
contexts. This difference of viewpoint on the building the procedure. At the beginning of the school year,
of thought is the main reason for relative failure of students often needed to compute before giving an-
the application of Piagetian tasks for assessment. swers. At the end of the school year, they needed
On the other hand, the solution may not be to only to retrieve the correct answer from their mem-
discard Piagetian thought, but to revise it if possible ory. This shift in the nature of the procedure leads
to enhance the validity of the assessments. The real to a great reduction in cognitive involvement. The
person and his context have to be taken into account students can then use their working memory for
in order to develop what Bideaud and Houdé (1991, other problems, such as the control of goals. This ex-
p. 90) call: “a constructivism-in-context” linking the ample shows how the constrains of the cognitive sys-
pragmatic and the normative aspects of the thought. tem, e. g., the limited capacity of working memory,
The recent developments of cognitive psychology are taken into consideration in a cognitive-based di-
gives us a new conceptual framework allowing this agnostic assessment.
enlargement of the classical Piagetian theory. Within the general framework of cognitive psy-
chology, numerous attempts have been made to de-
sign useful tasks for diagnostic assessment. For ex-
ample, Royer, Cisero, and Carlo (1993) reviewed
Cognitivist Models more than forty tasks designed to assess cognitive
skills. Their conclusion from this review is quite dis-
Contribution of Cognitivist Models to appointing:
Learning Assessment In all the research that [they] read, there was not
As Richelle emphasizes (1993, p. 9 3), “There is not a single report of a reliability index for an assess-
one cognitivism, there are variations of cognitiv- ment procedure, and indexes of validity were
ism.” Under the same term, we use to gather some- available only as inferences of the form, “if the
times very different models although linked by a measures were not valid, the experiment would
common will to transcend the behaviorism and to not have come out as it did.” (Royer et al., 1993,
study the information processing carried out be- p. 235)
tween the stimulus and the response.
In comparison with the Piagetian theory, cogni- Researchers in cognitive psychology show an im-
tivist models are more concerned with the under- portant lack of concern about the psychometric
standing of the real person in context. For example, properties of the measures they use. More deeply,
when they study problem solving, researchers want these researchers often have no general model for
to understand all the complexity of the information assessment. The tasks they design are not related to
processing without limitation to the sole logical a measurement model. Consequently, practitioners
components. They take into account the co-ordina- have no conceptual framework for organizing a di-
16 Jacques Grégoire

agnostic assessment with these tasks nor to interpret ing, were in fact measuring simultaneous processing.
the results. The need for such a framework for test This is the case for tasks like Magic Window and
development has recently been emphasized by Photo Series. Consequently, in the definitive version
Nichols (1994). of the test, only three tasks are measuring sequential
Two ways that have been used to design tests for processing whereas seven tasks are measuring si-
diagnostic learning assessment will be considered multaneous processing. Furthermore, all three tasks
next: are memory tasks (memory of hand movements,
1. Some researchers have tried to measure very gen- number recall, memory of word order). So we have
eral processes, useful in a wide range of situations. to question the real nature of the total score on the
This viewpoint will be illustrated with the Kauf- Sequential Processing Scale. There is considerable
man Assessment Battery for Children (K-ABC). risk when oversimplifying the cognitive reality.
2. Other researchers have focused their attention on Second, factor analysis results show that the fac-
domain-specific procedures, useful only in partic- tors loading on the tasks are not always the expected
ular situations. This viewpoint will be illustrated ones. For example, for children at age 4, the Triangles
with a test for diagnostic assessment of reading Subtest loaded equally on the sequential factor and
disabilities, based on the model of Patterson and the simultaneous factor. Furthermore, factor load-
Morton (de Partz, 1994). ings can change across age groups. For example, at
age 8, the Triangles Subtest does not measure the
sequential factor. Similar observations can be made
Assessment of General Processes about other subtests. The Hand Movements Subtest
better reflects the sequential factor for 4-year-old
The K-ABC was created by Kaufman and Kaufman children and the simultaneous factor for 8-year-old
(1983). The authors chose to develop a test which children.
would reveal a defect in some general information These observations lead the practitioner to ques-
processing. This defect could be the reason for some tion the way of interpreting failure or success from
learning disabilities. Consequently, the diagnosis us- scores on these subtests. Is it possible to make a
ing the K-ABC provides proposals for special edu- valid assessment of procedures applying to a wide
cation focus on the training of procedures. range of situations? This problem of validity seems
This test was designed on the basis of Das’s mod- difficult to solve. Previously, the problem of validity
el of information processing, which is itself built on was raised for tests assessing the g factor and for
Luria’s work. Following this model, there are two tasks assessing the stages of development in Piaget’s
general processing modalities: the simultaneous model. These tests measure a “domain-specific” ver-
processing and the sequential one. The first one re- sion instead of a “domain-general” version of the
quires the processing of many stimuli at once. The competence of interest. The only solution to this
second one requires the processing of stimuli fol- problem may be to assess the intended competence
lowing a serial order. Picture recognition is an illus- in a large range of situations. But such an assessment
tration of a simultaneous processing. Listening to is very time consuming.
sentences is an illustration of a sequential process-
ing. But these two illustrations are not really good
examples of the use of the two processing modali- Componential Models and
ties. Indeed, when persons are confronted with a Assessment
complex task, such as the ones proposed, they do
not use only one processing modality. Usually, the Many cognitivist models propose to represent cogni-
two processing modalities are mixed and each one tive functioning in the form of an organization of
plays a part at different stages of the task. components, each one responsible for a specific pro-
The K-ABC is an interesting illustration of the cessing of the information. The componential mod-
difficulty of assessing only one modality of process- els are an important conceptual framework for diag-
ing. The shortcomings of the test are the general nostic assessment. There are a guide for the diagnos-
shortcomings of this kind of assessment which is to tic assessment and a reference for the interpretation
appraise general-domain procedures. In the case of of observed performance.
the K-ABC, we observe first that: The test does not The componential models currently used refer
equally measure the simultaneous and the sequen- only to domain-specific procedures. Consequently,
tial modalities. The authors noticed that some tasks, each local model is only useful to appraise a compe-
originally designed to measure sequential process- tence related to very specific problems. This assess-
Diagnostic Assessment of Learning Disabilities 17

e. g., “bact,” “erol”) and cannot read a correct but


irregular word (e. g., “blood”), we can deduce that
the problem is inside the addressing procedure. But,
we cannot already determine if the problem is at the
level of the orthographic lexicon or at the level of
the phonologic lexicon. Other tasks are needed to
answer this question. For example, if the pupil is able
to correctly pronounce the words corresponding to
pictures, we can think that the phonologic lexicon is
not involved. So we have to make a conditional
analysis of the results. This analysis relies on logical
reasoning conducted on the basis of the theoretical
model. However, in the present situation, such rea-
soning does not take into account the psychometri-
cal properties of the tasks. The error of measure-
ment of each of the tasks is often neglected and the
conclusions are drawn assuming that all the tasks
are reliable. But it is often far from being true!
Moreover, local models used for assessment are
Figure 1. Componential model for the assessment of rea-
“minitheories” (Richelle, 1993, p. 221) generally not
ding of words (de Partz, 1994). linked together. We cannot dismiss the hypothesis
that some problems are related to the global cogni-
tive functioning. These problems can, for example,
ment is illustrated with a model developed to ap- concern attention or memory functioning. But com-
praise acquired reading disabilities (de Partz, 1994). ponential models have difficulties taking into ac-
This model can be seen as an architecture organ- count more general problems. Consequently, there
izing the successive processes involved in oral read- is a risk of being simplistic when using only a local
ing. The figure shows the two procedures that a per- model, and believing it can take into account all the
son can use to read a word. Following the first pro- facets of the competence we want to appraise.
cedure, called addressing procedure, the word is first
linked with the visual representation of the word en-
coded in long-term memory and next with the sound The Future of Diagnostic Assessment
representation of the same word, which is then pro-
nounced. This procedure is the one we use to read After the previous analysis of current viewpoints on
irregular words like “few” or “bomb,” of which the diagnostic assessment of learning, some future pros-
pronunciation can only be deduced from the recog- pects can be drawn. The recent development of cog-
nition of the whole word. The second procedure to nitivist models does not imply that the Piagetian
read a word, called assembling procedure, begins theory is out of date. Each theoretical model pro-
with the analysis of the word into letters (graphems). vides specific concepts to describe the characteris-
Each graphem is then matched with a basic sound tics of the cognitive functioning underlying an ob-
(phonem). Finally, all the graphems are assembled served performance. The Piagetian theory provides
and the word is pronounced. This procedure is used a particularly strong conceptual framework to
when a reader meets a new word which is not en- understand the construction of the logical structure
coded in his/her long-term memory in the ortho- of the thought. Consequently, it will remain espe-
graphic lexicon. Consequently, such a word cannot cially helpful to understand the development of
be read following the first procedure described. some concepts, like the concept of number. In this
Using such a model, the assessment consists in case, it focuses our attention on logical operations
discovering the defective component(s). For this that constitute the roots of the concept of number.
purpose, tasks are needed to precisely appraise each It also puts the stress on the progressive co-ordina-
of the components. We also need to interpret each tion of the operations.
result in relation with the results in the other tasks. But the assessment of the abilities related to
For example, if a pupil can read regular pseudo- school learning cannot be restricted to logical abili-
words (i. e., words that do not belong to the English ties. Cognitivists have emphasized that we need to
language lexicon, but that are regularly constricted, take into account other features of the thought, es-
18 Jacques Grégoire

pecially the representation of knowledge in long- 1993, p. 7): (1) each production rule embedded in a
term memory and the constrains related to the production system is a modular piece of knowledge
working memory. Most of the assessment tools de- representing a well-defined step of cognition; (2) a
veloped in cognitive psychology are measuring pro- production system is achieved by stringing together
cedural knowledge, that is “knowing how.” The pro- production rules “by appropriate setting of goals
cedural knowledge can be domain-general or do- and other writing to working memory, and by read-
main-specific. The assessment tools focused on ing from working memory”; (3) the condition-ac-
more specific situations seem the more promising. tions rules are asymmetric; (4) the production rules
This is the case of the test we described, which mea- are abstract and can apply in multiple situations.
sures the procedures involved in the reading of The ACT theory, very shortly described here, can
words. The results it provides are very specific since be a powerful framework to develop tests for diag-
they concern only the reading situation; but they are nostic assessment of learning problems. The Ad-
certainly more valid and, consequently, more inter- vance Computer Tutoring Project at Carnegie Mel-
esting for practitioners. This assessment related with lon University has developed successful intelligent
a componential model of the procedural knowledge tutorial systems based on the ACT theory. The ap-
we want to appraise is the best illustration of the proach taken by these tutors has lots in common
power of an assessment which focus on competence. with the mastery learning approach (Anderson,
In order to create other tests based on a compo- 1995) where the diagnostic assessment plays an im-
nential model, a general framework is needed. The portant role (Huberman, 1988).
ACT theory developed by Anderson (1983, 1993) Referring to the ACT theory, a componential
seems the most suited for such a function. The ACT analysis of the complex skills is needed. Complex
(Adaptive Control of the Thought) theory is a the- skills have to be decomposed into a large number of
ory of the cognitive architecture. The basic proposal production rules. The relevance of such an analysis
of this theory is that a cognitive skill is composed of is related to the second assumption of the diagnostic
production rules (Anderson, 1993, p. 4). A produc- assessment discussed in the first section. Some au-
tion rule is a condition-action pair (IF-THEN). If thors, such as Shepard (1991), see this componential
the conditions exist, then the actions are executed. analysis as an error. This is related to the failure of
The production rules are organized around a set of the task analysis used in the past in mastery learning.
goals, creating a production system (see Table 3 for But this analysis was made in a behaviorist frame-
an example). A production system is a precise and work. As Anderson (1995, p. 396) emphasized there
complete theory of a particular task. The main fea- is no evidence to generalize the perceived failures of
tures of the production systems are (Anderson, the behaviorist program to any program that at-

Table 3. Production rules for addition1 (Anderson, 1993, p. 3).


NEXT- COLUMN
IF the goal is to solve an addition problem and c1 is the rightmost column without an answer digit
THEN set a subgoal to write out an answer in c1

PROCESS-COLUMN
IF the goal is to write out an answer in c1 and d1 and d2 are the digits in that column and
d3 is the sum of d1 and d2
THEN set a subgoal to write out d3 in c1

WRITE-ANSWER-CARRY
IF the goal is to write out d1 in c1 and there is an unprocessed carry in c1 and d2 is the
number after d1
THEN change the goal to write out d2 and mark the carry as processed

WRITE-ANSWER-LESS-THAN-TEN
IF the goal is to write out d1 in c1 and there is no unprocessed carry in c1 and d1 is less than 10
THEN write out d1 and the goal is satisfied

WRITE-ANSWER-GREATER-THAN-NINE
IF the goal is to write out d1 in c1 and there is no unprocessed carry in c1 and d1 is 10 or
greater and d2 is the ones digit of d1
THEN write out d2 and note a carry in the next column and the goal is satisfied
1c
1, d1, d2 and d3 denote variables that can take on different values for different instances of each production
Diagnostic Assessment of Learning Disabilities 19

tempts to analyze a skill into components. Given the sures the procedures involved in the reading of
correct units, this analysis can be successful. words. This kind of test, related with a componential
Even if the ACT theory proves to be a powerful model of the information processing, proves to be a
framework for diagnostic assessment, some general very efficient tool for diagnostic assessment. To de-
problems will unfortunately remain: velop such componential diagnostic test, a theory of
1. Is it possible to use only local models of the cog- the cognitive architecture is needed. We have em-
nitive functioning to understand all the learning phasized that Anderson’s ACT theory is particularly
disorders? We have to recognize that some defec- suited for this function. However, even if the ACT
tive procedures are very general. Consequently, theory proves to be a powerful framework for diag-
using only local models, we tend to neglect the nostic assessment, some questions remain for the fu-
role of such procedures in learning disabilities. ture.
2. The local models and the related tasks have usu-
ally been well validated by researchers. But the Author’s Address:
reliability of the results gathered with such tasks Professor Jacques Grégoire
is generally unknown. More collaboration be- Université Catholique de Louvain
tween cognitivists and psychometricians seems to Faculte de Psychologie et des
be necessary. Sciences de l’Education
3. Finally, using local models, it is much more com- Place du Cardinal Mercier, 10
plicated and time consuming to conduct an as- 1348 Louvain-la-Neuve
sessment which focus on competence than to Belgium
make a simple assessment of performance. That
is why, we think that normative tests which focus
on performance will still have a place in diagnos- References
tic assessment. Indeed, they allow a fast screening
of the student’s knowledge. If some problems are American Psychiatric Association. (1994). Diagnostic and
observed, a deeper investigation can then be con- statistical manual of mental disorders: DSM-IV. Wash-
ducted in a more restricted domain of knowledge ington, DC: American Psychiatric Association.
using an assessment focused on competence. Anderson, J. (1983). The architecture of cognition. Cam-
bridge, MA: Harvard University Press.
Anderson, J. (1993). Rules of the mind. Hillsdale, NJ: Erl-
baum.
Conclusion Anderson, J. (1995). Learning and memory. New York:
John Wiley & Sons.
In this paper, we have stressed that the assessments Bideaud, J. (1991). Les chemins du nombre. Confronta-
which focus on competence, according to Piagetian tions et perspectives. In J. Bideaud, C. Meljac, & J.-P.
theory or cognitivist theories, give more interesting Fischer (Eds.), Les chemins du nombre (pp. 435–450).
Lille: PUL.
information for special education than assessments
Bideaud, J., & Houde, O. (1991). Cognition et develop-
which focus on performance. Indeed, the purpose of ment. Berne: Peter Lang.
diagnosis is not only to quantify successes and fail- Chomsky, N. (1965). Aspects of the theory of syntax. Cam-
ures but it is also, and, above all, to understand the bridge, MA: MIT Press.
meaning of the observed performance. For such an Connolly, A. J. (1988). Key math revised: a diagnostic inven-
understanding, we need to refer to models of learn- tory of essential mathematics. Circle Pines, MI: American
ing and cognitive functioning. These models allow Guidance Service.
us to interpret the observed scores. de Partz, M.-P. (1994). L’approche clinique cognitive des
dyslexies et des dysgraphies acquises. Unpublished doc-
Some assessment tools are focused on proce-
toral thesis, Université Catholique de Louvain, Bel-
dures which are useful in a wide range of situations. gium.
This is the case of the K-ABC, which measures two Fayol, M. (1990). L’enfant et le nombre. Neuchâtel: De-
general procedures that process a great variety of lachaux & Niestlé.
information. Unfortunately, because the tasks in- Fischer, J.-P. (1996). La mesure des temps de réponse aux
cluded in the tests measure these procedures in calculs numériques élémentaires en situation scolaire.
some specific situations, it is difficult to draw con- In J. Grégoire (Ed.), Evaluer les apprentissages. Les ap-
ports de la psychologie cognitive (pp. 72–88). Bruxelles:
clusions about a person’s ability to correctly use the De Boeck.
procedures in general. The assessment tools focused Gagné, E. D., Yekovich, C. W., & Yekovich, F. R. (1993).
on more specific procedures seem more promising. The cognitive psychology of school learning. New York:
This is the case of the test we described, which mea- Harper Collins.
20 Jacques Grégoire

Grégoire, J. (1992). Evaluer les troubles cognitifs au moyen quences in the validation of performance assessments.
des épreuves piagétiennes? Analyse de quelques Educational Researcher, 23(2), 13–23.
problèmes méthodologiques. Archives de Psychologie, Nichols, P. D. (1994). A framework for developing cogni-
60, 177–204. tively diagnostic assessments. Review of Educational
Grisay, A. (1989). La recherche APER fait chuter les taux Research, 64, 575–603.
de retard scolaire. Revue de la Direction Générale de Piaget, J., & Szeminska, A. (1941). La genèse du nombre
l’Organisation des Etudes, 11, 3–19. chez l’enfant (7ème éd., 1991). Neuchâtel: Delachaux et
Hambleton, R. K., Swaminathan, H., & Rogers, H. J. Niestlé.
(1991). Fundamentals of item response theory. Newbury Resnick, L. B., & Resnick, D. P. (1990). Assessing the
Park, CA: Sage. thinking curriculum: New tools for educational reform.
Hiebert, J. (Ed.). (1986). Conceptual and procedural knowl- In R. B. Gifford & M. C. O’Connor (Eds.), Changing
edge: the case of mathematics. Hillsdale, NJ: Erlbaum. assessments: alternative views of aptitude, achievement
and instruction (pp. 37–75). Boston: Kluwer.
Huberman, M. (1988). La pédagogie de maîtrise: idées-
Richelle, M. (1993). Quoi de neuf sur l’esprit? Paris: PUF.
force, analyses, bilans. In M. Huberman (Ed.), Assurer
la réussite des apprentissages scolaires (pp. 12–44). Royer, J. M., Cisero, C. A., & Carlo, M. S. (1993). Tech-
Neuchâtel: Delachaux et Niestlé. niques and procedures for assessing cognitive skills. Re-
view of Educational Research, 63, 201–243.
Inhelder, B. (1943). Le diagnostic du raisonnement chez les
débiles mentaux. Neuchâtel: Delachaux et Niestlé, Shepard, L. A. (1991). Psychometricians’ beliefs about
(2ème éd., 1969). learning. Educational Researcher, 20, 2–16.
Snow, R. E., & Lohman, D. (1989). Implications of cogni-
Kamii, C. (1990). Les enfants réinventent l’arithmétique.
tive psychology for educational measurement. In R. L.
Berne: Peter Lang.
Linn (Ed.), Educational measurement (pp. 261–331).
Kaufman, A. S., & Kaufman, N. (1983). Kaufman assess- New York: Macmillan.
ment battery for children. Circle Pine, MN: American Sternberg, R. J. (1990). T & T is an explosive combination:
Guidance Service. Technology and testing. Educational Psychologist, 25,
Messick, S. (1994). The interplay of evidence and conse- 201–222.

View publication stats

You might also like