Handbook of psychology volume 7 educational psychology

474

Learning Disabilities

deficits that differentiate them from children with at least

reading deficits as defined by depressed scores on the Reading

subtest on the WRAT. The cognitive and neuropsycholog-

ical profiles of children identified as specific arithmetic-

disabled are also different from those of normally achieving

children.

Evidence (Fletcher, 1985b; Siegel & Linder, 1984; Siegel

& Ryan, 1984, 1988, 1989a, 1989b) suggests that those chil-

dren meeting the criteria of the specific arithmetic-disabled

subtype have deficits in short-term and working memory that

are dependent on the type of stimulus and the aspect of mem-

ory assessed. Specifically, Siegel and Linder (1984), in a

study of the role of phonemic coding in short-term memory,

compared three groups of children—one with reading dis-

abilities (as defined by scores on the WRAT Reading subtest

of equal to or below the 25th percentile and no cutoff on the

other two WRAT subtests), a second with arithmetic disabili-

ties (as defined by scores on the WRAT Arithmetic subtest of

equal to or above the 30th percentile), and a normally achiev-

ing group (as defined by scores of greater than or equal to the

30th percentile on all three WRAT subtests). The children,

aged 7–13, were administered a series of tasks that involved

the visual or auditory presentation of rhyming and non-

rhyming letters and either an oral or written response. Pat-

terns and levels of performance were compared statistically

across three age groups (i.e., 7–8, 9–11, 12–13) and between

each subtype and normally achieving children. Due to statis-

tical problems, noncomparable age distributions, and small

sample sizes, it was not possible to compare across subtypes.

Results indicated that both older disabled groups—like their

normal counterparts—had significantly poorer recall of

rhyming as opposed to nonrhyming letters (except for the

oldest—12–13 years—arithmetic-disabled group, in which

the authors suggest that the children may be functioning at the

upper limit of their visual short-term memory). For stimuli

presented visually, the overall performance levels of both LD

groups were significantly lower than those of the normally

achieving group. For the auditory stimuli, only the reading-

disabled group differed significantly from the normally

achieving peers.

Fletcher (1985b) found differences in memory tasks

between LD groups as defined by WRAT scores. He

compared four groups of LD children (a reading-spelling-

disabled group, a reading-spelling-arithmetic-disabled group,

a spelling-arithmetic-disabled group, and an arithmetic-

disabled group) and a normally achieving group of children

on storage and retrieval aspects of memory for verbal and

nonverbal stimuli. He found that relative to the normally

achieving controls, both the arithmetic and the arithmetic-

spelling-disabled subgroups had significantly lower storage

and retrieval scores on the nonverbal task but did not differ

from each other; the reading-spelling subgroup differed only

on retrieval scores on the verbal task; and the reading-

spelling-arithmetic subgroup differed on the retrieval scores

on the verbal task and storage and retrieval scores on the non-

verbal task. As with Siegel and Linder (1984), the differences

between subgroups depended on the type of stimulus (verbal

vs. nonverbal) and the aspect of memory (storage or re-

trieval) being assessed.

Siegel and Ryan (1988) also compared reading-disabled

(as defined by WRAT subtest scores), specific arithmetic-

disabled (as defined by WRAT subtest scores), and normally

achieving children on a variety of skills involving grammati-

cal sensitivity, phonology, and short-term memory. In

general, it was found that older specific arithmetic-disabled

children performed in a manner similar to that of the nor-

mally achieving group in grammatical sensitivity and pho-

nological tasks. Some exceptions were found in that the

arithmetic-disabled children in the 7–10 age group performed

more poorly on a sentence repetition task; this difficulty was

attributed to the short-term memory component of the task.

Additionally, this age group performed more poorly than did

normally achieving children on the nonword spelling sec-

tions (a writing task) of the phonics task. However, in tasks

that measure short-term memory (phonological coding), the

specific arithmetic-disabled group performed in a manner

similar to that of the reading-disabled group and significantly

more poorly than did the normally achieving group. The au-

thors conclude that although both of the two disabled groups

(compared with normally achieving children) have deficits in

short-term memory, only the reading-disabled group had

deficits in tasks said to represent a language disorder.

Siegel and Ryan (1989a) examined the same groups, using

two working memory tasks—one involving sentences and the

other involving counting. Again, the disabled groups differed

from each other on the types of memory deficits observed. The

reading-disabled group differed from the normally achieving

children on both tasks, whereas the arithmetic-disabled chil-

dren differed from their normally achieving peers only on the

counting task. It appears from the research (Fletcher, 1985b;

Siegel & Linder, 1984; Siegel & Ryan, 1988, 1989a, 1989b)

that although both subtypes of LD children have deficits in

short-term and working memory, problems in children with

reading deficits are more generalized and involve both

verbal and nonverbal aspects of memory, whereas those in

children with arithmetic deficits and normal or above-normal

reading are more limited to visual, nonverbal, and numerical

material.

Evidence from a number of sources (Fletcher, 1985b;

Rourke & Finlayson, 1978; Share, Moffitt, & Silva, 1988;

Assessment of Learning Disabilities 

475

Siegel & Feldman, 1983; Spellacy & Peter, 1978; Webster,

1979) indicates that specific arithmetic-disabled children

(as defined by deficient scores on the WRAT Arithmetic

subtest and the age-appropriate scores on the WRAT Reading

and Spelling subtests—Group 3) have age-appropriate

auditory-perceptual and verbal abilities but are deficient

on measures of visual-perception and visuospatial abilities.

However, reading-disabled children (as defined by being

relatively proficient at arithmetic as compared with their

WRAT Reading and Spelling subtest scores—Group 2) have

age-appropriate visual-perception and visuospatial abilities

but are deficient on measures of auditory-perceptual and

verbal abilities (Rourke & Finlayson, 1978). Also, Group 3

(arithmetic-disabled) children exhibit difficulty in tasks such

as the Halstead Category Test, which require higher order

visuospatial analysis and visual-perceptual organization

(Strang & Rourke, 1983b). They also appear to exhibit

deficits in measures of psychomotor abilities and on tests such

as the Tactile Performance Tests (Reitan & Davison, 1974),

the Grooved Pegboard Test, and the Maze Test designed to

identify tactile-perceptual impairment (Rourke & Strang,

1978; Siegel & Feldman, 1983; Spellacy & Peter, 1978). On

the other hand, Rourke and Strang (1978) and Strang and

Rourke (1983) found that Group 2 children (relatively profi-

cient in arithmetic, compared with their reading and spelling)

are proficient at these tasks.

In addition, Rourke and Strang (1978) claim that the arith-

metic subgroup (Group 3) exhibited normal right-hand per-

formance but impaired left-hand performance—the exact

opposite of the Group 2 children, who had impaired right-

hand performance but normal left-hand performance.

Strang and Rourke (1983a, 1983b) suggest that the arithmetic-

disabled subgroup has deficiencies in nonverbal concept-

formation

compared

with

other

disabled

subgroups.

Specifically, when the types of errors made of the Arithmetic

subtest of the WRAT were analyzed, it was found that the spe-

cific arithmetic subtype tended to make a larger number of er-

rors, make a greater variety of errors, and attempted to answer

questions without an apparent understanding of the strategies

needed to solve the problems (Strang & Rourke, 1985a,

1985b). This error pattern was not found in children with read-

ing disabilities (Group 2).

As with the research with memory deficits cited earlier,

Rourke and Finlayson (1978), Rourke and Strang (1978), and

Strang and Rourke (1983a, 1983b) suggest that the character-

istics described are different from those of other learning-

disabled students (who showed deficits on all the WRAT

subtests, Group 1, or just on the reading and spelling subtests

compared with the arithmetic subtest, Group 2). This finding

has led Rourke et al. (Rourke, 1982, 1983, 1985, 1987; Rourke

& Finlayson, 1978; Rourke & Fisk, 1988) to hypothesize that

those children with arithmetic deficits belong to the larger

nonverbal LD group with right-hemisphere processing prob-

lems, whereas those children with deficits in reading as well

as arithmetic belong to the larger linguistic learning-disabled

group with left-hemisphere processing problems. Clearly,

however, children who only have severe deficits in arithmetic

can be differentiated from children with reading difficulties

and from normally achieving children on cognitive and neuro-

psychological profiles.

In light of the previously described controversy and

research findings, the use of specificity assumptions in the

definition of learning disabilities is questionable; this is

true regardless of whether one refers to domain specificity

(the limitation of the disability to one or two cognitive areas)

or population specificity (failure to use subtypes).

ASSESSMENT OF LEARNING DISABILITIES 

Determining who is learning disabled requires careful and sys-

tematic assessment. The three following questions address the

assessment of learning disabilities: (a) How should achieve-

ment be measured; (b) which tests should be used; and (c) what

cutoff scores should be used to identify a learning disability?

The Measurement of Achievement

The arguments about the definition of learning disability

center on the determination of whether an individual meets

specific criteria for the diagnosis of a disability. First of all,

to measure whether there are significant difficulties, one

must use a systematic assessment of these academic areas;

standardized (norm-referenced) tests appear to be the best

way to do this. Why use norm referenced tests? The answer

is simple: Those making the assessment want to compare an

individual with others of the same age and know whether

that person has a significant problem. A standardized test is

the best way to accomplish this task. Nonstandardized as-

sessments can be used, but they do not provide normative

information that can be used for the purposes of comparison.

With a nonstandardized or informal assessment, it is impos-

sible to know whether an individual has made the number

and type of errors that are typical of his or her age group and

therefore are normal and expected, or whether the errors

are atypical and unexpected and indicative of a problem.

Nonstandardized tests may play an important role (to be

discussed later), but the core assessment should use stan-

dardized tests. However, nonstandardized assessments do

have a role in the evaluation of writing; this role is discussed

later in the chapter.

476

Learning Disabilities

To assess learning disabilities, there are several types of

tests that should be used. Specifically, an assessment of an

individual for the possibility of a reading disability should in-

clude a measure of word-recognition skills. Word recognition

is one of the critical building blocks in gaining meaning from

print, and it is important to know whether these basic skills in

this area are significantly below average (Stanovich, 1982).

An assessment should include a reading test that involves the

reading of what are called pseudowords—pronounceable

combinations of English letters that can be sounded out with

the basic rules of grapheme-phoneme correspondences. This

type of test assesses the awareness of the phonological as-

pects of a language that is the key to decoding words in an

alphabetic language such as English. Difficulties with these

phonological skills are the basis of a reading disability (e.g.,

Bruck, 1990; Felton et al., 1990; Shafrir & Siegel, 1994a;

Siegel & Ryan, 1988). A test of text reading—specifically, a

reading comprehension test—should be included. Obviously,

the measurement of text reading skills is particularly impor-

tant to measure what individuals remember and understand

from what they have read.

There should be a test of spelling involving the dictation

of words; this parallels the type of spelling required in writ-

ing in the academic setting.

There should be a test of computational arithmetic skills to

determine what the individual understands about the funda-

mental arithmetic operations. An assessment of mathematical

problem-solving skills should be included.

There should also be an assessment of writing skills; this

type of assessment is quite difficult for a variety of reasons.

The time involved to allow someone to write may be extensive

because one must allow time for planning as well as for the

actual writing. Also, many individuals have learned to use a

computer and prefer to write using a computer. Therefore, a

proper assessment of writing might use a computer, which

may not be feasible in most assessment contexts. It may be

acceptable to ask the individual to bring in a sample of his or

her writing, but some type of brief assessment in the context of

the assessment is useful. The scoring of these written products

is subjective and there does not appear to be agreement on

what constitutes a widely accepted scoring system. However,

Berninger (1994) has proposed a system that appears to have

potential to consistently evaluate writing. She suggests six di-

mensions to evaluate writing: (a) handwriting quality (legibil-

ity), (b) handwriting fluency (number of words copied within

time limits), (c) spelling single words from dictation (on stan-

dardized lists of increasing difficulty), (d) spelling in compo-

sition (percentage of correctly spelled words), (e) composition

fluency (number of words produced within time limits), and

(f) composition quality (content and organization of para-

graph construction).

Identification of whether there is a learning disability

should use a simple system. Brief tests of word recognition,

decoding (pseudoword reading), reading comprehension,

spelling, writing, and computational arithmetic and mathe-

matical problem solving will detect most (if not all) of the

learning disabilities. A low score on any of these tests is a

danger signal. More detailed testing can then be conducted,

but any testing should be related to remediation and not used

without consideration of what new and useful information is

provided by the test and whether it is really necessary—a

point that is discussed in detail later in this chapter.

Types of Tests

There is considerable confusion in the field about how to

measure achievement. Attempts at defining and studying

learning disabilities suffer from a common fallacy of assum-

ing that all tests that have the same label (e.g., intelligence

test, reading test) measure the same skill (Siegel & Heaven,

1986; Siegel et al., 1985). When one considers the area of

achievement tests, the labeling fallacy becomes even clearer.

There appears to be almost an infinite variety of ways to mea-

sure reading, spelling, and arithmetic. The choice of which

tests to use can determine whether a disability will be found.

Consider the case of measuring reading achievement.

There are four types of material that are typically found in

reading tests: (a) pseudowords, (b) single words, (c) sen-

tences, and (d) paragraphs. For the reading of pseudowords,

the individual is asked to read a set of pronounceable combi-

nations of letters that test the understanding of the relation-

ship between letters or groups of letters and their sounds.

This type of test is simple and direct and measures a funda-

mental skill.

Tests of the reading of single, real words typically require

that the individual read a set of words aloud. These words

may vary on several dimensions, but usually these dimen-

sions are not systematically assessed. For example, words

may be regular—that is, they follow the letter-sound corre-

spondence rules of English (e.g., fat, block)—or they may be

exceptions to these rules—that is, they involve irregularity or

unpredictable letter-sound correspondence (e.g., have, said,

island, sword). A person may have difficulty with the irregu-

lar but not the regular ones. The words may be more familiar

and in the person’s vocabulary, such as cat, book, and red, or

they may be less familiar, such as predatory, terpsichorean,

and oligarchy. Obviously, what is familiar and what is

unknown depends on the age, vocabulary, and experience

of the individual. Some words may be read correctly because

they have been encountered before, whereas others may

be read incorrectly (but almost correctly; e.g., intrigue read

in-tri-gue instead of in-treeg) because they are not part of the

Assessment of Learning Disabilities 

477

individual’s vocabulary and have not been encountered

before. The confounding of familiarity with other dimensions

of a word makes the construction of a word-reading list a

difficult task. In the case of each particular word, one simply

does not know when a person reads a word correctly whether

he or she has merely memorized it. Note that pseudowords do

not present this difficulty.

The reading of both words and pseudowords assesses the

basic problem in a reading disability—that is, difficulty with

phonological processing; both tasks are relatively straightfor-

ward. However, the measurement of text processing becomes

more complex. Text processing is typically measurable by

tests that involve the reading of sentences and paragraphs. In

both cases, there are often clues about the word from the sur-

rounding context. Typically, nouns follow articles, verbs fol-

low subjects, adjectives precede nouns, and so on. When an

individual reads a word in context correctly, we do not know

whether he or she has read the word or made a good guess

from the context. Note that this problem is not an issue with

the reading of single words or pseudowords.

The reading of sentences or paragraphs may occur silently

or aloud. If the reading is silent, there is no way of assessing

what the person is actually reading, although this type of

reading may be more similar to what occurs in many reading

situations. Questions about what has been read are the princi-

pal means to assess comprehension. In most cases, memory is

a very important aspect of the testing of sentence and para-

graph comprehension. Often, the material is removed so that

it is not available when the questions are asked. The person

may have read quite well but may forget the answer to the

question. Even when the material is available, the individ-

ual’s performance is timed, or a fixed time is allowed to

complete the test. At least some of the variance between indi-

viduals may be caused by variations in reading rate or speed

of task completion—not a differential understanding of the

material. There is, however, a significant difference between

a slow reader and one who may not even be able to decode

the words in the first place. Some students are able to decode

the words and answer the questions on a reading comprehen-

sion test but need more time. Some have a problem with

decoding the words. An assessment should be able to differ-

entiate these two difficulties.

Reading tests vary in the output or type of response that is

required. Some require an oral output that may involve some

degree of facility with expressive language, whereas others

require a written output—for example, answering multiple-

choice questions. Still others involve having the person select

a synonym for a word; in reality, this test is a measure of

vocabulary. An individual may select the incorrect word not

because of poor reading skill, but because he or she is not

sure of the correct synonym.

The actual comprehension questions themselves may vary

in several dimensions. They may involve inferences, mem-

ory for details, or the general point of the passage. It is very

likely that a large part of reading comprehension ability con-

sists of memory skills (e.g., Tal & Siegel, 1996). The individ-

ual must decode words and obtain meaning from them, but

also he or she must retain the information in working mem-

ory and be able to answer questions about the content of the

reading passage. It seems apparent, however, that memory is

still a significant factor in tasks in which recall of exact word-

ing or details is not essential. In these cases, the meaning

must be retained and then operated on in some manner to

produce an expected answer.

The individual’s familiarity with the material in the text

can determine how the person will score on a reading com-

prehension test (e.g., Drum, Calfee, & Cook, 1981; Marr &

Gormley, 1982; Schneider et al., 1989). For example,

Schneider et al. found that background knowledge about soc-

cer influenced comprehension of and memory for a story

dealing with soccer, but there were no significant differences

between children with high and low verbal aptitude skills.

Therefore, background knowledge was a critical factor in text

comprehension, but verbal ability was not.

Time to read can also be an integral part of the reading

score. A number of factors can contribute to differences in the

time taken to read a passage. For example, a person who

recalled information about the story may have a faster time

than does someone who could not recall the target informa-

tion but who could remember its spatial location and look

back quickly, who may in turn have a faster time than does a

person who could not remember anything about the target

information and had to search throughout the passage.

Daneman (1984) has reported that much of the variance in

reading comprehension scores disappears if individuals are

allowed to look back at the passage that has just been read.

Another difficulty with reading comprehension tests is

that frequently the questions can be answered with a reason-

able amount of accuracy without reading or comprehending

the passage (e.g., Tal & Siegel, 1996). Such questions as

Where was the cow kept? can be answered by good guessing;

cows are not likely to be kept in cars, closets, or bathtubs.

Obviously, the problem with having so much variability in

the measurement of reading comprehension is that many

different skills are assessed. Theoretically, there are many

types of possible reading difficulties if this kind of measure

of reading is used because the person could have a problem in

any one or more of these components. Clearly, some of these

combinations are more likely than others are, but the point is

that it is unclear which dimensions are creating the problem

when the individual achieves a low score on one of these

sentence or paragraph reading tests. An individual may