Handbook of psychology volume 7 educational psychology

466

Learning Disabilities

differentiate subgroups of disabled readers, and (d) the use of

a large number of tests and elaborate multivariate statistical

procedures to search for more homogeneous subgroups of

disabled readers.

We examine attempts at subtyping according to these clas-

sification systems. It should be noted that there is a funda-

mental difference between the first type of subtyping scheme

and the other three types. In the first type, patterns of achieve-

ment in other areas besides reading (e.g., spelling and arith-

metic) are examined. Typically these studies—unlike the

other three types—do not attempt to divide disabled readers

into reading subtypes; rather, they try to differentiate disabled

readers from other groups of LD children.

Classification Based on Reading Patterns.

Some clas-

sification schemes based on qualitative differences in reading

performance have been attempted. For example, Lovett

(1984) proposed that there are two subtypes of disabled read-

ers: rate disabled readers who read slowly but have adequate

decoding skills and accuracy disabled readers who have

below-average decoding skills. Lovett found a number of

differences between the accuracy and the rate disabled read-

ers. However, the conclusions reached by Lovett may be a

function of her definition. It seems to be quite possible that

her rate disabled readers were not really reading disabled. For

example, they had to be 1.5 years below grade level on four

out of five of the rate measures. Grade level is an imprecise

measure and represents an ordinal rather than an interval

scale. Subtracting grade levels is not an appropriate way of

defining individual differences. In addition, the same level of

grade retardation may mean a different discrepancy at vari-

ous levels in the developmental continuum. For example, a

younger child with even a slight grade-level retardation

would score at a low percentile or standard score level,

whereas an older child might not necessarily have a low per-

centile score even at a 1.5- or 2-year grade-level retardation

(see Siegel & Heaven, 1986, for detailed evidence on this

issue). As many of the children in Lovett’s study were at an

older age level, in which a 1.5 grade level retardation does

not necessarily indicate a serious impairment, the use of this

criterion may mean that many of these children were not se-

riously reading disabled. The rate disabled children were not

obviously impaired in spelling, language, comprehension,

regular and exception word reading, or phonics, whereas

the accuracy disabled children were. Although these rate

disabled children may have been slower than average in read-

ing speed (even this is unknown because of the grade-level

criterion), it is not clear that they were different from normal

readers. More important is that it is not clear that such chil-

dren should be called disabled or dyslexic. They clearly have

many good reading skills—for example, reading nonwords,

reading exception words, and understanding syntax. They are

probably below average in reading speed but not really dis-

abled. Superficially, although it appeared that Lovett found

subtypes, closer examination reveals that not all these chil-

dren are truly reading disabled. 

Vernon (1977) proposed a system for defining subgroups

of poor readers based on variations in the level of the compo-

nent skills involved in reading, such as letter discrimination,

letter-sound correspondence, or slow reading. Unfortunately,

such a system is not very useful without precise operational

definitions or a methodology for separating these subgroups.

Vernon’s analysis of reading appears to represent sequential

development of skills related to reading rather than skills that

are absent or present simultaneously. Another attempt at

defining groups of reading-disabled children was made by

Boder (1968, 1971, 1973), who developed a screening proce-

dure for diagnosing reading disabilities on the basis of three

so-called reading (but actually spelling) patterns. Boder’s

screening procedure consists of a two-part test, flashed and

nontimed presentation of words, and an individual spelling

test based upon the child’s reading performance. Spelling is

assessed by asking the children to spell words from their

sight vocabulary (at or below reading level) and then an equal

number from their unknown vocabulary at grade level. The

rationale for this method is that including words from the

children’s sight vocabulary allows for assessment of a child’s

ability to revisualize words and inclusion of unknown words

taps the ability to spell phonetically (Boder, 1971). On the

basis of this screening procedure, Boder outlined a catego-

rization scheme to classify children with dyslexia based on

three patterns. The first subgroup is called dysphonetic be-

cause these children are supposedly characterized by an

inability to develop phonetic skills, have difficulty in sound-

symbol integration, and read in whole-word gestalts—that is,

they appear to see reading as a pattern recognition task. The

second subgroup is called dyseidetic; children in this group

are characterized by the opposite problem. They experience

difficulty in forming so-called whole-word visual gestalts

and must sound out every word as if they were encountering

it for the first time. The third subgroup is called alexic and are

characterized by the problems of both the other groups. It is

important to note that the categorization of a child is based on

spelling—not reading—patterns.

Camp and Dolcourt (1977) developed and tested two

parallel standardized reading and spelling forms to increase

the utility of Boder’s concept utilization of subtypes. The

word lists were revised to contain half phonetic (e.g., cave)

and half dysphonetic (e.g., calf) words. In this new proce-

dure, the examiner selected the list of spelling words that

Subtypes

467

corresponded to the grade level the child could read 50% cor-

rectly. Camp and Dolcourt determined that dyseidetic indi-

viduals were diagnosed primarily by spelling performance,

and because of Boder’s definition (less than or equal to 50%

of sight vocabulary correctly spelled and misspellings of

known irregular words), several children would be classified

as dyseidetic even though they were reading above grade

level. Children who have good reading but poor spelling

skills are common (e.g., Lennox & Sigel), but they are not

really dyslexic.

A number of studies have been carried out using Boder’s

classification system, but many of them suffer from method-

ological and definitional problems that make their compari-

son extremely difficult and interpretation of their findings

rather tenuous. For example, electrophysiological evidence

for subgroups of developmental dyslexia has been reported

by Fried, Tanguay, Boder, Doubleday, and Greensite (1981).

Using Boder’s diagnostic approach Fried et al. applied event

related potential (ERP) techniques to the study of word and

musical chord auditory information processing in the left and

right hemispheres of their dyslexic participants and com-

pared their performance with that of a normal control group.

They found that latency differences between ERPs evoked by

word and musical chord stimuli were greater for the left

hemisphere of the normal children, as expected from studies

conducted in adults (Brown, Marsh, & Smith, 1973). The dy-

seidetic children, all of who could phonemically decode and

encode reading material well, also exhibited a normal pattern

of greater waveform differences in the left as opposed to the

right hemisphere, although the magnitude of these differ-

ences differed from that in the controls (a finding that the in-

vestigators attributed to differences in attentional factors

between the groups). In contrast, the dysphonetic group did

not show the greater word-musical chord ERP waveform dif-

ferences in the left hemisphere. Fried et al. interpreted these

data as suggesting that the left hemisphere may not have a

fully developed capacity to process auditory information in

the normal manner. An unexpected finding was that the alexic

participants produced results similar to those of the normal

readers, despite the fact that they were postulated to have

problems with both the right and left hemispheres. It is possi-

ble that the waveform differences observed in this group

were a matter of chance. In any case, the differential perfor-

mance of the dysphonetic and dyseidetic readers may be a

function of severity in that the dyseidetic individuals may not

have been really reading disabled.

A study by Telzrow, Century, Whitaker, Redmond, and

Zimmerman (1983) investigated the demographic and neu-

ropsychological characteristics of children in the various

reading categories defined by performance on the Boder Test

and found some differences between them. Unfortunately,

the authors of this study fluctuated between calling their

participants learning disabled and reading disabled to the ex-

tent that one does not really know much about the population

under discussion. One is not told how the original diagnosis

of developmental dyslexia was made, what measures were

used, and what criteria for inclusion were adopted. Hence, it

is possible that many were not really reading disabled.

A study by Nockleby and Galbraith (1984) compared the

performance of dysphonetic individuals, those with non-

specific reading disabilities, and normal readers on eight

dependent-variable tasks—four of which were described as

requiring analytic-sequential processing and four that re-

quired processing in a simultaneous gestalt fashion. The ana-

lytic sequential tasks included Auditory Sequential Memory

(from the Illinois Test of Psycholinguistic Ability, or ITPA),

Visual Sequential Memory (ITPA), the Lindamood Auditory

Conceptualization Test, and Sound Blending (ITPA). Simul-

taneous gestalt tasks included a facial memory task using

40 photographs of men and women that had to be identified

later, a tactile-visual recognition test in which children had to

recognize shapes placed in their hand by viewing a response

card of those shapes, the Benton Visual Retention Test, and

the Raven Coloured Progressive Matrices. It was hypothe-

sized that a subgroup of reading-disabled children catego-

rized by the Boder test as dysphonetic would perform poorly

on tasks requiring analytic-sequential processing and nor-

mally on tasks requiring simultaneous gestalt processing. A

subgroup of disabled readers categorized as having nonspe-

cific reading retardation were predicted to perform as well as

a comparison group of normal readers on all perceptual and

memory tasks. The dysphonetic and nonspecific groups did

not perform significantly differently from the controls on any

of the simultaneous gestalt processing measures. Dyspho-

netic individuals performed significantly below the control

group on the visual and auditory memory test. Both dyspho-

netic and nonspecific groups performed significantly below

the control group on the Lindamood test. Sound blending was

the only analytic-sequential task that did not discriminate

among the reader groups. The authors concluded that these

results support the hypothesis that dysphonetic dyslexic chil-

dren are deficient in one information-processing strategy

(analytic-sequential) and normal in the other process (si-

multaneous gestalt) and that children identified by Boder as

having nonspecific reading retardation may have essen-

tially intact processing for both modes. However, the authors

suggested that the absence of a difference between the non-

specific and the dysphonetic groups on the Lindamood

test appears to be evidence against Boder’s classification

system because the Lindamood has been found to be a valid

468

Learning Disabilities

indicator of problematic phonetic skills in poor readers. That

both groups performed almost identically on this task sug-

gests that both are phonetically disabled. Only one dependent

variable—visual memory—actually distinguished these two

groups. The present results suggest that many children classi-

fied as nonspecific in fact have difficulties processing the

sounds of language.

This study suffers in the way that many of the subtype

studies do: Their definition of a reading disability is ques-

tionable. As outlined by Siegel and Heaven (1986), grade

scores or age levels are not as appropriate as percentiles for

identifying a reading disability. In addition, the Gray Oral

Reading Test (used to define the groups) consists of a series

of graded oral reading passages. Although this study classi-

fied reading-disabled subjects into Boder’s three subtypes,

the dyseidetic group was ignored after the classification. It

would be extremely valuable to know how the dyseidetic

group performed on the Lindamood test, for example. If their

performance was equally as poor as that of the other readers,

considerable doubt would be cast on the dyseidetic category

as a useful subtype. It is also important to know whether the

dyseidetic group would in fact exhibit performance inferior

to that of the other groups on the simultaneous gestalt pro-

cessing tasks.

Malatesha and Dougan (1982) found that when Boder’s

scheme was used, dysphonetic and normal readers had differ-

ent patterns of scores on a dichotic listening test. Again, this

apparent heterogeneity appears to be a function of the defini-

tion used. First, 1-year grade-level retardation in reading was

used as the criterion for reading disability. As noted earlier,

grade-level retardation is not a valid measure because grade

levels are ordinal rather than interval levels of measurement.

Furthermore, it does not represent the true degree of difficulty

at all developmental levels. Second, a reading comprehen-

sion test was used (Gates MacGinitic). Variables such as

reading speed, vocabulary, prior knowledge, attention, and

memory contribute to scores on this measure, and a child may

achieve a low score for any or all of these reasons.

Evidence for the homogeneity of the reading-disabled

population and a failure to validate the subtypes outlined by

Boder comes from work by Van den Bos (1984). He pre-

dicted that the children Boder classified as dyseidetic should

not have a deficit in processing auditorily presented letters

because their problem is theoretically in processing visual

information. In contrast, Van den Bos found that memory for

auditorily presented or visually presented letters did not

differentiate the dyseidetic from the dysphonetic dyslexic

individuals by Boder’s criterion. Conversely, Van den Bos

predicted that according to Boder’s criteria, dyseidetics

should perform more poorly than do dysphonetic dyslexic

individuals on a letter-matching task because they presum-

ably have difficulty with visual information processing. In

fact, there were no significant differences between dyspho-

netic and dyseidetic participants on this task—all the

dyslexic individuals performed significantly more poorly

than did normally functioning individuals on this task. All the

dyslexic participants had a particular problem with a task that

involved determining whether two letters had the same

name—a task that presumably relies on phonological pro-

cessing skills. It should be noted that Van den Bos did not val-

idate Boder’s subtypes. Van den Bos used a criterion of poor

word-recognition skills to define the dyslexic group. There-

fore, he had poor readers in the sense that I believe the term

should be used. A more appropriate definition of dyslexia re-

sulted in homogeneity within the group.

In another reading performance-oriented system, Larsen

and Parlenvi (1984) suggested that there may be significant

individual differences and perhaps subtypes in groups of poor

readers. In their study of reading in Sweden, they noted that

some second-grade poor readers were not as disrupted as good

readers in accuracy or rate of reading of inverted words,

whereas others were more disrupted. However, these authors

were not specific about the type of reading test used to define

their groups; hence, they may have had a heterogeneous group

of poor readers who represented a continuum of severity.

Those readers whose performance was less disrupted or even

possibly improved by the inverted stimuli may have been

the very poor readers (dyslexic in the sense I support) who

were reading through the visual route. The others may not

have been really reading disabled in the traditional sense. As

can be seen from the descriptions of the previously discussed

studies, classification based on reading performance has not

clearly differentiated valid subtypes of disabled readers.

Classification Based on Neuropsychological Models

A study by Kinsbourne and Warrington (1963b) represented

one of the initial attempts to explore the possibility of distinct

subgroups of dyslexic children. They divided children re-

ferred because of reading backwardness into two groups on

the basis of WISC Verbal-Performance IQ discrepancies.

Group 1 consisted of six children who had at least a 20-point

Verbal-Performance (V-P) discrepancy in favor of perfor-

mance. They were termed the language-retarded group.

Group 2 consisted of seven children with a 20-point V-P

discrepancy in favor of Verbal. These children were termed

the Gerstmann group (Kinsbourne & Wanington, 1963a).

Group 1 exhibited delays in speech acquisition, verbal com-

prehension, and verbal expression. Their mean reading and

spelling ages were almost the same and were significantly

Subtypes

469

lower than their arithmetic ages (WISC subtest). Group 2

exhibited finger agnosia as characterized by poor perfor-

mance on tests of finger order and differentiation. They

showed significant retardation in right-left orientation and in

arithmetic difficulty. The mean spelling age was 1 year lower

than the mean reading age. The mean arithmetic age was

nearly the same as the reading age.

Although age levels rather than percentile scores were used

in this study and therefore the exact severity of the academic

skills deficit remains unknown, the findings are interesting

because of the patterns they present. The language-disabled

group with poorer reading and spelling performance; the

Gerstmann group with poorer spelling, arithmetic, and hand-

writing performance; and a possible third group with both

patterns of deficit have been identified by other investigators,

such as the groups of Rourke et al., Siegel et al., and Fletcher

et al. as elaborated earlier.

Johnson and Myklebust (1962, 1967) suggested that indi-

viduals can have auditory dyslexia and visual dyslexia. Their

classification was based on clinical descriptions rather than

empirical evidence. Persons with visual dyslexia appear to

experience deficits in visual perception and in memory; con-

sequently, they have visual discrimination problems that re-

sult in confusion of letters and words that look the same.

They can make discriminations but only very slowly. Indi-

viduals with auditory dyslexia experience difficulty in re-

membering auditory stimuli and in stringing these stimuli

into sequences. They experience problems in distinguishing

similarities and differences in sounds, perceiving a sound

within a word, synthesizing sounds into words, and dividing

words into syllables. It is important to note that there is no

reliable empirical evidence for these groups as distinct

subtypes.

A similar attempt by Mattis, French, and Rapin (1975) and

Mattis (1978) identified three subgroups of children with

reading disorders: (a) language, (b) articulation and grapho-

motor, and (c) visual perception. The language-disordered

group had difficulty discriminating similar sounds, repeating

sentences, and following complex verbal commands (Token

Test), but they had relatively intact visuospatial skills. The

articulation and graphomotor group had adequate language

skills, but people in this group had difficulties with articula-

tion of speech sounds and copying shapes. A third group, the

visuospatial-disordered group, had difficulty remembering

visual stimuli, some problems with spatial concepts (Raven’s

Coloured Progressive Matrices), and relatively intact verbal

skills. It is quite possible that the latter two groups were older

and did not have reading problems that were as serious as

those of the first group. The definition of reading disability

used by Mattis et al. contributed to this apparent heterogeneity;

this is not to say that their classification scheme is incorrect.

Rather, the heterogeneity in their sample may be a function of

the definition of the reading-disabled group. Mattis et al. used

children who were classified as retarded by two or more

grades according to the WRAT. The problem with a grade-

level retardation definition of reading disability has already

been described. Although this scheme may not be a problem

with younger children, it is with older children because a

two-grade discrepancy between actual reading grade level

and expected grade level may actually mean that the child has

a relatively high score and therefore cannot be considered

reading disabled. Information about the ages and reading per-

centile scores of the three subgroups was not presented, so we

do not know if this is the case. The group called visual-

perceptual disorder by Mattis et al. (1975) may have been

older and may not have had as serious a reading problem

(or even a reading problem at all). In addition, many of the

skills measured by Mattis et al. (1975) seem to be deficient

in this population; furthermore, without a normal comparison

group, we do not know whether these results represent rela-

tive strengths and weaknesses or absolute low levels of per-

formance by the groups.

Classification Based on Statistical Techniques

An example of a statistical approach to the definition of sub-

types is that of Doehring, Trites, Patel, and Friedorowicz

(1981), who investigated the interaction of subtypes of read-

ing disabilities with language and neuropsychological

deficits. A variety of tests of reading and reading-related skills

were administered, and the results for the total sample and for

a variety of subsamples were analyzed by factor analysis to

ascertain the stability of the types of reading disabilities

that were identified. Three factors emerged from the classifi-

cation: Factor 1/Type 0, difficulty with oral reading; Factor 2/

Type A, auditory-visual association difficulty (difficulty in

the types of silent reading skills that require the association of

the spoken equivalents of printed letters, syllables, and

words); and Factor 3/Type S, a sequencing deficit (difficulty

with syllables and words as compared to letters and numbers).

After analyzing the reading test results, the authors at-

tempted to explore the possibility that different types of lan-

guage deficits might be involved in different types of reading

disabilities. Twenty-two language measures were used to

measure both lower level and higher level language skills. In

general, it was found that individuals with reading problems

were about as impaired in language skills as they were in

reading skills. The pattern of language deficits suggested that

the greatest difficulty was at relatively low levels of language

skills (that is, phonemic segmentation and blending, serial