Handbook of Learning Disabilities

128 CAUSES AND BEHAVIORAL MANIFESTATIONS

files as well as the underlying brain mecha- typically in subcortical structures, and have
nisms of each. In addition, we discuss the been reported in 40 to 93% of children with
implications of these findings with respect NF-1 (Steen et al., 2001). While UBOs are
to idiopathic LD (i.e., that which as yet has commonly seen in individuals with NF-1,
no established genetic etiology). their biological and clinical significance is
still not fully understood. Macrocephaly, or
Neurofibromatosis Type 1 enlargement of the head, has long been ob-
served in about 50% of individuals with
Genetic and Physical Aspects of NF-1 NF-1; MRI studies are consistent with the
interpretation that this is due to enlarged
NF-1 is one of the most common single-gene brains (megalencephaly; Cutting, Koth, Bur-
disorders, with an incidence of 2 to 3 cases nette, et al., 2000; Moore et al., 2000; Said
per 10,000 in the population (Friedman, et al., 1996; Steen et al., 2001). Recently,
1999) and equal prevalence rates across sex magnetic resonance spectroscopy imaging
and race. Approximately 50% of all cases of (MRSI) has revealed metabolic abnormali-
NF-1 are familial, inherited in an autosomal ties in NF-1, with elevated N-acetylaspar-
dominant manner, with the remaining cases tate/Choline ratios in the thalamus (Wang et
being spontaneous mutations (Crowe, al., 2000).
Schull, & Neel, 1956). The locus of NF-1 ab-
normality has been found to be a rather large Our Approach to Studying NF-1
region on the long arm of chromosome 17 at
17q11.2 (Barker et al., 1987; Goldgar, Green There are several types of methodologies to
Parry & Mulvihill, 1989; Gutman & use when studying genetic disorders; one
Collins, 1993). Although there is a DNA test approach is to study the impact of the ge-
for NF-1, it is currently diagnosed based on netic disorder on cognition by comparing
physical symptoms from NIH consensus cri- performance of those affected to norms
teria (see Table 8.1). In addition, MRI find- based on the general population (e.g., Dilts
ings have become more and more integral to et al., 1996; Eliason, 1986; Moore, Slopis,
the confirmation of NF-1. Schomer, Jackson, & Levy, 1996; North et
al., 1994). Another approach, one we also
NF-1 affects different aspects of the cuta- have taken, is to use a sibling-matched-pair
neous, skeletal, and central nervous sys- (one NF-1-affected, one unaffected) design;
tems. Common manifestations of NF-1 this approach is exemplified in several of
include Lisch nodules, cutaneous and plexi- our published studies (e.g., Cutting, Huang,
form neurofibromas, axilary café au lait Zeger, Koth, & Denckla, 2002; Hofman,
spots, nerve tumors, and optic gliomas Harris, Bryan, & Denckla, 1994; Mazzocco
(Stumpf, Alksne, & Annegers, 1988). In ad- et al., 1995). A sibling-matched-pair design,
dition, T2 weighted hyperintensities, other- unlike that which involves an unrelated in-
wise referred to as unidentified bright ob- dependent control group from the general
jects (UBOs), are seen on MRI scans, population, takes into account familial and
environmental factors (Mackintosh, 1998).
TABLE 8.1. NIH Consensus Criteria for NF-1 Statistical “purists” have balked at the issue
Diagnosis of nonindependence of the groups being
compared. Both methodologies have advan-
ț Six or more café au lait macules tages; however, within the context of trying
to understand the direct gene-to-brain-to-
ț Two or more neurofibromas or one plexiform behavior influences, the sibling-pair ap-
neurofibroma proach allows for a more focused under-
standing of the influence of the NF-1 gene
ț Freckling in the axilla or inguinal region on brain–behavior relationships.

ț An optic glioma Cognitive Profile

ț A distinct osseous lesion Though there is virtually universal agree-
ment that mental retardation is rare with
ț Two or more Lisch nodules

ț A first-degree relative who meets the above
criteria for NF-1

Note. Two or more must be present for diagnosis.

Attention: Relationships between ADHD and LD 129

NF-1, the IQs of those affected are lower followed up the Hofman and colleagues
than the family of origin (indexed by sib- study with a larger sample size (20 sibling
lings) would predict and are shifted into the pairs) and an expanded battery of visuospa-
lower portion of the normal range for the tial, language, reading, and attention abili-
general population. In addition, LD is re- ties. Children with NF-1 were found to per-
ported in approximately 25 to 61% of chil- form lower than expected (by reference to
dren with NF-1 (North et al., 1997; Riccar- sibling) on a variety of language tests
di, 1981; Stine & Adams, 1989), which is (Boston Naming Test, Phoneme Segmenta-
much higher than the estimates of 5 to tion, Phonological Memory, Token Test,
17.5% in the general population (Shaywitz Letter–Word Identification, Passage Com-
and Shaywitz, 1999). Originally, it was prehension, and the Test of Written Lan-
thought that individuals with NF-1 had guage). Normal performance was observed
“nonverbal” LD (NVLD) because early on Rapid Automatized Naming, Word Flu-
studies documented significant impairments ency, and Grammaticality Judgment. Poor-
on Performance IQ and on a test called er-than-expected performance (by reference
Judgment of Line Orientation, both consid- to sibling) was observed on only a few tests
ered to be tests of visuospatial ability (Ben- of visuospatial ability (Judgment of Line
ton, Hamsher, Varney, & Spreen, 1983, Orientation and Block Design), as well as
Weschler, 1974). Therefore, we embarked two tests of controlled processing or execu-
on studying NF-1 in the expectation of ob- tive function (a continuous performance
taining a clearer understanding of NVLD. test—number of omissions and the Wiscon-
However, since this time, a variety of stud- sin Card Sorting Test—categories). Discrep-
ies, including many from our laboratory, ancy-based reading disability was again
have documented that NF-1 is strongly as- confirmed to be more prevalent in children
sociated with deficits in the verbal domain with NF-1 as compared to their siblings; in
(e.g., Cutting et al., 2000; Hofman, Harris, addition, as is quite typically viewed as
Bryan, & Denckla, 1994; Mazzocco et al., characteristic of reading disability, children
1995; North et al., 1997); in fact, deficits in with NF-1 showed deficits on phonological
the verbal domain appear to be more wide- measures. Mazzocco and colleagues con-
spread and academically debilitating than cluded from these findings that while chil-
those in the “nonverbal” domain. An addi- dren with NF-1 do have visuospatial
tional surprise in our research was that deficits, deficits in the verbal domain were
ADHD is associated with NF-1 to a much far more numerous and academically rele-
higher degree than expected from general vant, thus indicating that NF-1 is not a
familial prevalence of ADHD (Koth, Cut- model for NVLD (and thereby resulting in
ting, & Denckla, 2000). In particular, five the article’s title proclaiming NF-1 the “not-
studies, which we review herein, highlight so-nonverbal learning disability”). In terms
different aspects of our findings about the of understanding reading disability, Maz-
cognitive profile of NF-1 and what they re- zocco and colleagues commented that “chil-
veal about idiopathic LD (Cutting, Koth, & dren with NF-1 illustrate the conceptual dif-
Denckla, 2000; Cutting et al., 2002; Hof- ficulty underlying discrepancy-based
man et al., 1994; Koth et al., 2000; Mazzoc- reading disability, wherein language disor-
co et al., 1995). der influences both measures between
which a discrepancy is calculated” (p. 519).
In an initial study of 12 sibling pairs,
Hofman and colleagues (1994) found that In an effort to understand similarities and
in addition to poor performance on tests of differences between NF-1 and idiopathic
visuospatial ability (Judgment of Line Ori- reading disabilities, Cutting, Koth, and
entation; Block Design), children with NF-1 Denckla (2000) compared children with
also had significantly lower-than-expected NF-1 to children with reading disabilities
scores on measures of reading and writing from the general population. Another goal
ability; furthermore, even after controlling of the study was to compare a discrepancy-
for IQ, a disproportionate representation of based reading disabilities group with our
reading disabilities was characteristic of NF-1 group to determine the impact of lan-
children with NF-1 as compared to their guage deficits on ability to meet discrepan-
siblings. Mazzocco and colleagues (1995) cy-based definitions of reading disabilities.

130 CAUSES AND BEHAVIORAL MANIFESTATIONS

As expected, children with NF-1 had significant familial effects on cognition,
deficits similar to those of children with again confirming the need to consider inves-
reading disability; both groups had difficul- tigating the NF-1-associated deficits within
ty with reading and reading-related tests the familial context afforded by the sibling-
(Rapid Automatized Naming; Denckla & pair design.
Rudel, 1976) and phoneme segmentation
measures. Interestingly, the NF-1 group did Because ADHD symptomology has been
not perform poorly on the Rapid Automa- reported in children with NF-1, Koth and
tized Naming measure but did on both colleagues (2000) compared the prevalence
phonological measures, whereas the reading of ADHD in children with NF-1 as com-
disability group performed poorly only on pared to their unaffected siblings and par-
Rapid Automatized Naming, thus potential- ents. The goal of the study was to determine
ly providing some support for the dissocia- whether ADHD could be included in the
tion between phonological and Rapid Au- phenotype of NF-1, or whether it was an
tomatized Naming measures (Wolf & unrelated disorder within families. Frequen-
Bowers, 1999). Other differences between cy of ADHD among children with NF-1,
the groups were that children with NF-1, their siblings, and their parents was com-
unlike children with reading disability, pared. Results indicated that a higher per-
showed deficits in visuospatial areas as well centage of children with NF-1 (42%) had
as a broader language deficit. Overall, this ADHD than either siblings (13%) or par-
study showed that many children with NF-1 ents (5%), suggesting that ADHD is in part
were not able to meet the discrepancy defin- associated with the NF-1 cognitive pheno-
ition of reading disabilities because of their type. The origins of this association between
more “global” verbal impairments (i.e., ADHD and NF-1 are not entirely clear at
lower Verbal IQ) but nonetheless showed this time. One of the features of NF-1 is
the hallmark deficits (phonological) associ- UBOs, which are often seen the same brain
ated with reading disabilities. structures implicated in ADHD (basal gan-
glia, thalamus, cerebellum, and brainstem).
One aspect of our study of NF-1 was to It may be that UBOs disrupt certain critical
examine growth of certain cognitive func- frontally related circuits, similar to those
tions. Based on the Hofman and colleagues thought to be disrupted in idiopathic
(1994) and Mazzocco and colleagues ADHD, thus giving rise to ADHD in chil-
(1995) studies, a pattern of “spared” and dren with NF-1. Future studies examining
“impaired” tests (as compared to sibling’s the relationship between presence of ADHD
performance) emerged; those tests that were and presence and location of UBOs will be
impaired were Vocabulary, Block Design, able to elucidate this issue.
and Judgment of Line Orientation, whereas
those tests that were “spared” were Picture In summary, findings regarding the cogni-
Completion, Picture Arrangement, and tive phenotype of NF-1 indicate that children
Rapid Automatized Naming (letters and with NF-1 appear to have both visuospatial
numbers). Ten sibling pairs were followed deficits and reading disabilities, the latter in
longitudinally and compared on the the familiar context of many oral language
“spared” and “impaired” these tests (Cut- deficiencies, a likely contribution to the low-
ting et al., 2002). Results showed that chil- ering of their Verbal IQ (thus making fulfill-
dren with NF-1 did not “catch up” to their ing the criteria for a discrepancy-based LD
siblings on “impaired” measures; however, more difficult). Findings from the Koth and
on the “spared” measures they continued to colleagues (2000) study also indicate that
perform similarly to their siblings. On aver- NF-1 appears to be associated with ADHD.
age, across the six cognitive measures, there Thus, NF-1, while initially appearing to be a
was no significant difference between the model of NVLD because of selective visu-
groups in terms of growth rates. Interesting- ospatial deficits, has proven to be a model
ly, variation among families for level of per- for what is typically seen in LD: difficulty
formance was larger than variation among with reading and language, with comorbid
siblings (with and without NF-1) within a ADHD. In addition, ADHD/executive func-
family on Vocabulary and Rapid Automa- tion may influence performance on certain
tized Naming, thus providing evidence of tests; for example, it may be that the origin
of poor performance on the Judgment of

Attention: Relationships between ADHD and LD 131

Line Orientation stems from ADHD/execu- Ramsby, 1998; Itoh et al., 1994). To ad-
tive function test-taking demands. Though dress this issue further, we recently studied
the ADHD-related executive function UBOs longitudinally in 12 children with
deficits have not yet been fully explored in NF-1 (Kraut et al., 2002); we examined a
NF-1, further study, particularly in relation number of regions occupied by UBOs, num-
to lesions in the basal ganglia and cerebel- ber of UBOs per brain region, and UBO vol-
lum, may reveal critical circuits involved in ume per brain region. Findings indicated
ADHD in general and in its related executive that the number of regions occupied by
function deficits. A challenge, not unique to UBOs, as well as UBO number and/or vol-
the NF-1 studies, is to tease apart language ume for all brain regions, diminished be-
contributions from the truly “central execu- tween the ages of 7 and 12; however, there
tive” issues when deficits on tests such as was an increase during adolescence. The re-
card sorting or word efficiency are employed lationship of UBOs to changes in cognitive
in an “executive” battery. functioning over time was not examined in
this study but will be a part of future studies
Neuroimaging Findings in order to elucidate further the impact of
UBOs and disruption of particular neuronal
Neuroimaging findings with NF-1 have al- pathways on cognition, in particular with
lowed for the study of areas of the brain regard to the influence of disruption of
that may affect specific aspects of cognition; neural circuits during critical time periods.
in particular, examination of megalen-
cephaly and UBOs in relation to cognition In an effort to better characterize UBOs,
allows for exploration of disruption in spe- we used MRSI to examine the biochemical
cific brain regions and/or neural circuits composition of the brain of males with NF-
that may contribute to selected cognitive 1 (Wang et al., 2000). Findings revealed that
deficits. To date, we have used two different there was elevated Choline and normal N-
imaging modalities, anatomical magnetic acetylaspartate (NAA) in subjects who were
resonance imaging (aMRI) and MRSI in our less than 10 years old, but in subjects older
study of NF-1. than 10 years, there was reduced NAA and
normal Choline. These changes (consistent
It has been debated whether or not UBOs in terms of NF-1-associated decreased
are related to cognitive impairment in NF-1; NAA/Choline ratios) were found in UBOs
some studies have found a relationship in the basal ganglia and, surprisingly, most
while others have not (see Ozonoff, 1999). prominently (and independently of UBOs)
To address the ongoing controversy, Denck- in the thalamus. It was concluded from this
la and colleagues (1996) examined the num- study that the metabolic abnormities found
ber of locations of UBOs and volume of using MRSI might indicate more wide-
UBOs in relation to the relative “lowering” spread white matter abnormalies; specifical-
of IQ in children with NF-1 (as compared to ly, elevated Choline in younger subjects may
their siblings). Findings showed that the reflect increased myelin turnover, which
number of locations (basal ganglia, cerebel- might result in axonal injury (reflected by
lum, brainstem, and other subcortical struc- reduced NAA) in older subjects. This hy-
tures) occupied by UBOs accounted for pothesis has also been put forth by other in-
42% of the variance in the “lowering” of vestigators from other sources of informa-
IQ in children with NF-1. It was concluded tion (e.g., neuropathological evidence;
that “IQ [as a global measure of cognition] DiPaolo et al., 1995). Therefore, UBOs seen
. . . might reasonably be more adversely af- on aMRI scans may be indicators of larger,
fected by multiple interruptions in CNS more global developmental white matter
[central nervous system] connections than abnormalities in NF-1.
by volume replacement in one or more spe-
cific locations” (p. 101). Other neuroimaging findings from our
laboratory have focused on volumetric
In addition to the controversy as to analyses. Cutting, Koth, Burnette, and col-
whether UBOs are related to the cognitive leagues. (2000) found that in a sample of 19
deficits in NF-1, there has also been some males with NF-1, approximately 50% were
suggestion from cross-sectional studies that megalencephalic. Megalencephaly was not
UBOs decrease over time (DiMario & significantly associated with familial or spo-

132 CAUSES AND BEHAVIORAL MANIFESTATIONS

radic origin of NF-1 or presence or absence and circuits, but not genes) may in turn be
of UBOs. However, megalencephaly was as- applicable to understanding brain-based
sociated with verbal impairment (specifical- origins of idiopathic ADHD and LD, as the
ly, lower Vocabulary subtest scores). Fur- systems and circuits involved, regardless of
ther study of megalencephaly examining the reason such are abnormal, may be criti-
volume of lobar subdivisions and gray and cally important.
white matter in relation to UBOs has re-
vealed a more complex picture of megalen- Further Study of NF-1
cephaly in NF-1 (Cutting et al., in press), in
particular with regard to presence of We are currently undertaking study of NF-1
ADHD. Cutting and colleagues (in press) using multiple neuroimaging modalities
found a strong relationship between the (aMRI, MRSI, and functional MRI); this
presence of comorbid ADHD in males with study involves sibling pairs as well as a con-
NF-1 and megalencephaly. In this study, the trol group and a reading disability group
brain volumes of 18 males with NF-1 were from the general population. One goal of the
compared to those of 18 age-matched con- study is to understand how the chemical ab-
trols. Seven of the 18 males with NF-1 were normalities that exist in the brains of chil-
diagnosed with ADHD. As compared to dren with NF-1 are related to the reading
controls, males with NF-1 without ADHD and language deficits associated with NF-1.
were megalencephalic, whereas males with Another goal is to determine how differences
NF-1 with ADHD were not megalencephal- in brain activation when reading are linked
ic. However, all males with NF-1, regardless to the cognitive and academic impairments
of ADHD status, showed increased volume associated with NF-1, and how these may be
of white matter in the frontal lobes; the NF- different/similar to those of children with id-
1 without ADHD group showed increased iopathic reading disabilities. Based on previ-
volume of white matter in the parietal lobes. ous research findings, we are hypothesizing
Consistent with reports of decreased frontal that chemical markers of neuronal abnor-
lobe volumes in idiopathic ADHD, presence malities will exist in the thalamus (a “relay”
of ADHD in NF-1 was associated with a de- station in the brain) and correlate with read-
crease in the volume of gray matter in the ing, language, and articulation deficits in
frontal lobe, namely, right prefrontal. NF-1, as defined by the “lowering” of the
Marked parietal white matter enlargement cognitive score of each child with NF-1 rela-
was seen if UBOs were present in the basal tive to that of his or her unaffected sibling.
ganglia in NF-1 males who did not have We also hypothesize that children with NF-1
ADHD. Findings from this study indicate will exhibit a pattern of activation in the lan-
that ADHD is an important comorbid diag- guage centers of the brain when performing
nosis to consider and appears to be associat- reading-like tasks similarly to children with
ed with a different neuroanatomical profile, reading disabilities, but anomalously in com-
specifically reduction in brain volume (as is parison to normal readers. The goal of this
also observed in idiopathic ADHD). In ad- study is to basic neurobiological factors and
dition, strong evidence for the association of their affect on cognition, particularly read-
white matter abnormities with NF-1, re- ing and language (although ADHD will also
gardless of comorbid ADHD, was found. be a factor) in NF-1 and reading disabilities,
thus furthering our understanding of gene-
In summary, neuroimaging findings with to-brain-to-behavior relationships as related
NF-1 indicate a complex picture of a variety to reading and language disorders, as well as
of anomalies: UBOs, megalencephaly, and ADHD.
metabolic evidence of neuronal and myelin-
ic abnormalities. How these abnormalities Tourette Syndrome
in NF-1 are related to each other, as well as
to cognition and presence of ADHD, is still Tourette syndrome (TS) has been a focus of
under investigation. Understanding these re- the LDRC over the last 12 years. TS is a
lationships may further reveal our under- neuropsychiatric disorder with a prevalence
standing of neural circuitry and brain re- rate of approximately 1 per 1,000 males
gions that affect cognition in NF-1. This
knowledge (at least at the level of systems

Attention: Relationships between ADHD and LD 133

and 1 per 10,000 females and is character- (Harris et al., 1995; Mahone, Koth, Cutting,
ized by a variety of waxing and waning and Singer, & Denckla, 2001; Mahone et al., in
changing motor and vocal tics, and has an press; Schuerholz et al., 1997; Schuerholz,
onset usually prior to 15 years of age (Leck- Baumgardner, Singer, Reiss, & Denckla,
man, King, & Cohen, 1999). Although the 1996). For example, Harris and colleagues
precise genetic mechanism for TS is not yet (1995) found that impairments in planning,
known, and is looking more polygenic than cognitive flexibility, response inhibition, and
single-gene in mechanism, it has a strong ge- self-monitoring were observed in the ADHD
netic/familial component (Leckman & Co- and TS+ADHD groups as reflected by poor
hen, 1999). Individuals with TS have a high performance on the Rey Osterreith Complex
rate of comorbidity with ADHD and obses- Figure, the Wisconsin Card Sorting Task
sive–compulsive disorder (Golden, 1984; (WCST; Categories Achieved and Set
Singer, Schuerholz, & Denckla, 1995). It is Breaks), and the Test of Variables of Atten-
estimated that approximately 50 to 60% of tion (TOVA); however, the only impairments
children with TS also have ADHD; more- observed for the TS-only group were slow
over, it has been reported that approximate- and variable reaction time on the TOVA,
ly one-third of children with TS also have suggesting subcortical or basal ganglia in-
some type of LD (e.g., Burd, Kauffman, & volvement for this group. Other studies of
Kerbeshian, 1992; Golden, 1984). The fo- groups of TS-only, TS+ADHD, and ADHD
cus of study of TS in the LDRC was to un- have further clarified differences between
derstand the neuropsychological and neu- these groups (Schuerholz et al., 1996).
roanatomical similarities and differences Schuerholz and colleagues (1996) found that
between children with TS, TS plus ADHD, while 23% of the TS sample had an LD, this
and ADHD. In particular, the goal was to was because LD was present only in children
discern the role of ADHD, as mediated by who had TS and ADHD. Other findings
its cognitive correlate, executive dysfunc- from this study mirrored those of Harris and
tion, in producing LD; on the brain sys- colleagues, with slow and variable reaction
tems/circuits level, we focused on differ- time observed in all groups; an unexpected
ences in prefrontal–subcortical systems that finding in this study was poor performance
might differentiate these three groups. Our on Letter–Word Fluency in the TS-only
main findings were that LD and most as- group. It was suggested from this finding
pects of executive dysfunction are not par- that TS might be associated with a slowing in
ticularly characteristic of pure TS (when mental search (“bradyphrenia”) resulting in
free of comorbid ADHD), with the sole ex- poor linguistic productivity, different from
ception of cognitive slowing. LD (written motor slowing. This hypothesis was further
expression in particular) and an array of ex- clarified in a study of neuromotor function-
ecutive dysfunctions are more widespread in ing in children with TS-only, TS+ADHD,
children with TS when ADHD is also pre- and ADHD-only (Schuerholz et al.). On
sent. Therefore, our most pertinent findings timed motor movements, children with
with regard to the TS project are actually in ADHD (with or without TS) were found to
regard to what we have found about ADHD be slow relative to their age peers, while TS-
and its impact on manifestations of LD. only was associated with relatively unim-
Consequently, we provide some discussion paired performance. Therefore, it was sug-
of our findings with regard to TS; however, gested that while both TS and ADHD are
most of our discussion focuses on ADHD. associated with slowing on choice reaction
time tasks, this slowing is caused by different
Cognitive Profile of TS and ADHD deficits. Children with TS show cognitive
slowing, or “bradyphrenia,” whereas chil-
Studies of several cohorts of children with TS dren with ADHD show motoric slowing
from the LDRC have revealed that children (“bradykinesia”).
with TS-only have relatively few impair-
ments in executive functioning and do not Mahone and colleagues (2001) recently
have significant LD; instead, significant im- examined two aspects of executive function,
pairments in executive functioning and LD organization and response inhibition
are present when there is comorbid ADHD (thought to reflect dorsolateral and or-
bitofrontal circuitry), in a second cohort of

134 CAUSES AND BEHAVIORAL MANIFESTATIONS

children with TS-only, ADHD-only, and Cutting, Koth, Mahone, and Denckla (in
controls. It was hypothesized that children press), in an effort to further clarify how
with TS-only would show deficits in organi- children with ADHD (who presumably
zation, whereas children with ADHD would have impairments in executive function)
show deficits in both organization and re- may show difficulty in the process of learn-
sponse inhibition. In contrast to our previ- ing, recently examined the mechanisms un-
ous studies, Mahone and colleagues exam- derlying verbal learning in children with
ined not only total outcome scores but also and without ADHD. Children with ADHD
process variables, or how the groups com- (none of whom had RD) were compared on
pleted the task. In addition to overall perfor- both process and product scores from the
mance, process variables examined included CVLT-C. Findings indicated that while chil-
semantic clustering on a list-learning task dren with ADHD initially learned the same
(California Verbal Learning Test for Chil- number of words as controls, they were
dren; CVLT-C), clustering on semantic and weak in recalling the words after delays,
letter–word fluency, intrusions on the CVLT- suggesting that children with ADHD are
C, and errors on semantic and letter–word less efficient learners. Sex-related findings
fluency. Findings were somewhat inconsis- revealed that regardless of ADHD diagno-
tent with those of our previous studies: Dif- sis, boys and girls performed differently.
ferences were found only in the number of Boys used semantic clustering less frequent-
intrusions on the CVLT-C, which were ab- ly and recalled fewer words from the middle
normally elevated in both the TS and ADHD region of the list than girls; girls also outper-
groups. No differences were observed be- formed boys in terms of overall perfor-
tween groups for total score on Letter Word mance, despite lower verbal IQs. These
Fluency or process variables from Letter findings showed that children with ADHD
Word Fluency. It was hypothesized that dif- exhibit unexpected weaknesses in the
ferences between these results and the process of learning.
Schuerholz and colleagues (1996) study may
have been because of differences between In summary, neuropsychological findings
samples; the more recent cohort had higher from the TS project have yielded an under-
IQs and had been screened for a third co- standing that most of the school-related dif-
morbidity, obsessive–compulsive disorder ficulties reported in children with TS are
(i.e., more stringently excluded). Mahone most often associated with presence of
and colleagues speculated that “overgrow- ADHD-related deficits in executive func-
ing,” or compensation for subcortical defi- tion, although there is an issue with slow-
ciencies when afforded the maturation of the ness (i.e., “processing speed”) in this group,
frontal cortex, might allow for normal per- as well as with ADHD. (We have now seen
formance on executive measures. this slowing phenomenon in our clinical ex-
perience with children referred for possible
Other investigations stemming from the LD who also have TS.) Findings from vari-
TS project have focused only on ADHD. For ous studies reveal that ADHD-related exec-
example, Reader, Harris, Schuerholz, and utive function deficits negatively influence
Denckla (1994) investigated executive func- not only in purely “academic” subjects but
tion in 48 children with ADHD (no TS). Im- also the process of learning, in that children
pairments were found on the WCST (num- with ADHD do not apply effective learning
ber of categories achieved and set breaks) as strategies to material. More discussion of
well as the TOVA (errors of omission, slow the implications of these findings is in the
and variable reaction time), but intact per- section “Future Directions,” but nonethe-
formance on Word Fluency as well as the less these findings do imply the need for fur-
Rey Complex Figure. The effect of comorbid ther clarification of the role of ADHD and
reading disability (ADHD+RD) on executive executive function in learning difficulties.
function was also examined; findings indi-
cated that there were no differences between Neuroimaging Findings in TS and ADHD
the ADHD and ADHD+RD groups on any
of the executive function measures with the Several volumetric neuroimaging studies
exception of less variability on the TOVA for that have been conducted through the
the ADHD+RD group. LDRC lend support to the hypothesis that

Attention: Relationships between ADHD and LD 135

neurobiological mechanisms in TS and groups, neuroanatomical findings suggest
ADHD involve frontal/subcortical circuits. that each disorder affects the nervous sys-
In ADHD, however, abnormalities have tem somewhat differently (albeit in the same
been found in cortical (frontal), subcortical “neighborhood”). Our group’s more recent
(basal ganglia), and cerebellar structures. It work (Frederickson et al., 2002) elucidates
may be that abnormalities in one or anoth- the source of the fibers (white matter) that
er, or even two or three of these structures, in pure TS enlarge the rostral corpus callo-
give rise to different types of severities of sum. Interestingly, it should be noted that
impairment in ADHD. Following we pre- the abnormalities found in TS and ADHD
sent a brief description of neuroanatomical in the corpus callosum are confined to boys,
findings from the LDRC. at least on the level of volumetric anatomy,
according to another study from our group
In a study of the basal ganglia in children (Mostofsky et al., 1999).
with TS, Singer and colleagues (1993)
found in a group of predominantly male Examination of gray and white matter
subjects that TS was associated with re- cerebral volumes in TS and ADHD has re-
versed asymmetry in the putamen and vealed that whereas cortical abnormalities
lenticular region. While control subjects are present in both disorders, the cortical ab-
showed left-greater-than-right asymmetry in normality found in TS is much more subtle
the putamen and lenticular regions, TS sub- than that found in ADHD (Frederickson et
jects showed significantly less left predomi- al., 2002; Mostofsky et al., 2002). Frederick-
nance in these areas. Comorbid ADHD was son and colleagues (2002) found right
associated with significantly smaller left frontal lobe abnormalities in boys with TS,
globus pallidus volumes, and a separate specifically an enlarged percentage of frontal
subsequent follow-up study documented the lobe white matter as compared to overall
same finding in boys with ADHD alone frontal lobe tissue. Because of the known
(Aylward et al., 1996). These studies sup- unidirectionality of the white matter in-
port the hypothesis of subcortical abnor- volved (frontal lobe to basal ganglia), these
malities in both TS and ADHD, but consid- findings, combined with those of the Singer
erably more subtle ones in “pure” TS cases and colleagues (1993) of loss of normal
(loss of normal L > R asymmetry). asymmetry in the basal ganglia, led Freder-
ickson and colleagues to speculate that
Baumgardner and colleagues (1996) ex- frontal lobe abnormalities may be primary
amined corpus callosum morphology in and underlie the basal ganglia anomalies in
children (predominantly males) with TS- TS. Mostofsky and colleagues (2002) recent-
only, ADHD, and TS+ADHD. Results re- ly examined gray and white matter frontal
vealed dissociation between the anatomies lobe and sublobar volumes of boys with
of TS and ADHD. TS was associated with ADHD. Results indicated, as many previous
larger-than-normal volumes in four of five studies have also found, that boys with
areas of the corpus callosum (splenium, ADHD had smaller total cerebral volumes;
isthmus/posterior body, midbody, and ros- this reduction was primarily due to smaller
tral body); conversely, ADHD was associat- frontal lobe volumes. Within the frontal
ed with smaller rostral body volumes. Para- lobe, both gray and white matter volumes
doxically, the opposite direction of the two were reduced, suggesting that ADHD is asso-
volumetric abnormalities led to apparently ciated with decreases in both the cell bodies
“normal” rostral body volumes of the co- and axons of the frontal lobe. Sublobar vol-
morbid TS+ADHD group, as though two umetric findings indicated reduction in pre-
pathologies, equal but opposite. These find- frontal, premotor, and deep white matter
ings support the view that there are abnor- volumes. These findings suggest that ADHD
malities in frontal/subcortical circuitry in is associated not only with abnormalities in
both TS and ADHD, in that the anterior prefrontal cortex (which in this study includ-
part of the corpus callosum provides inter- ed dorsolateral and orbitofrontal regions)
hemispheric connections for the frontal cor- but also with premotor cortex (which in this
tex; specifically, the rostral body has axons study included supplementary motor associ-
that link premotor and supplementary mo- ation areas, Broca’s area, and the frontal eye
tor areas. Consistent with neuropsychologi- fields). Abnormalities observed in cognitive
cal findings contrasting TS and ADHD

136 CAUSES AND BEHAVIORAL MANIFESTATIONS

(“executive functions”), social (disihibited of which and how many (one-, two-, or
or impulsive behavior), motor, and oculomo- three-level) brain anomalies are present.
tor tasks in ADHD, many from our labora-
tory (e.g., Barker et al., 2001; Denckla & Future Directions
Rudel, 1978; Mostofsky, Lasker, Singer,
Denckla, & Zee, 2001; Reader et al., 1994; Though the findings of the LDRC at the
Shue & Douglas, 1992), lend support to Kennedy Krieger Institute over the past 12
Mostofsky and colleagues’ findings of ab- years have resulted in further understanding
normalities in prefrontal and premotor cor- of the relationships between LD and ADHD,
tex and suggest that rather than a single cir- with particular regard to the mediating influ-
cuit being impaired in ADHD, a number of ence of executive function overlapping these
parallel fronto–striatal circuits, perhaps be- two, a lot remains to be understood. Al-
cause of a common developmental abnor- though much of the research on LD has ac-
mality, are impaired. cumulated a body of knowledge about basic
reading disabilities, the findings from the
Cerebellar anomalies in ADHD suggest LDRC at the Kennedy Krieger Institute illus-
that this disorder also encompasses abnor- trate the complexities associated with eluci-
malities in fronto–cerebellar circuits (Ber- dating deficits in higher-order functions. For
quin et al., 1998; Mostofsky, Reisss, Lock- example, the study of basic reading disabili-
hart, & Denckla, 1998). Mostofsky and ty has resulted in knowledge about how to
colleagues (1998) found decreased size of identify and remediate children with phono-
the posterior vermis, specifically the inferior logically based basic reading disability; how-
posterior lobe (lobules VIII-X), in boys with ever, much remains to be understood about
ADHD. Because there are connections be- the complex interrelationships between ex-
tween the cerebellum and prefrontal cortex, ecutive function, language, and academic
Mostofsky and colleagues speculated that skills other than basic reading, such as read-
these cerebellar abnormalities might con- ing comprehension, mathematics reasoning,
tribute to the deficits observed in executive and written expression. All these achieve-
function thought to arise from prefrontal ments require the ability to plan, organize,
cortex. and self-monitor, several key components of
executive function. In addition, deficits in
In summary, neuroimaging findings re- the more “basic” level of executive function,
garding TS and ADHD reveal anomalies in response inhibition and working memory
frontal–subcortical structures. Both disor- (associated with ADHD), in relation to all
ders show abnormal frontal lobe volumes, types of LD, including basic reading disabili-
with additional abnormalities in subcortical ties, are still not well understood. Systematic
structures. Nonetheless, the volumetric study of the interrelationships between the
anomalies associated with each disorder are development of executive function, lan-
distinct in nature: TS appears to be associat- guage, and skills in other academic areas be-
ed with subtle cortical abnormalities. In sides basic reading is critical in terms of un-
contrast, there appear to be substantial find- derstanding how LD may manifest itself
ings with regard to ADHD in both wide- differently at each age and stage of develop-
spread frontal–cortical and subcortical ment; for example, maturation of the frontal
structures; the fact that reduction in frontal, cortex may play a critical role in developing
basal ganglia, and cerebellar volumes have those skills for reading comprehension (be-
all been found reflect the multifaceted na- yond what is accounted for by basic reading)
ture of this disorder and suggest that a vari- that require working memory and higher-
ety of levels of anomalies in multiple path- order thinking.
ways can result in ADHD.
Acknowledgments
It remains to be seen what would emerge
from studying a large group of children This work was supported by the following National
with ADHD with aMRI, to discover who Institutes of Health grants: P50 NS 35359 (Learn-
has one, two, or three levels of brain anom-
alies (including combinations of pairs) and
what motor and cognitive profiles would
correlate with each possible anatomic pro-
file. Prognosis might well differ as functions

Attention: Relationships between ADHD and LD 137

ing Disabilities Research Center), ND 07414 (Post- process theory of development of skilled writing
doctoral Fellowship), and HD 24061 (Mental Re- (pp. 57–81). Greenwich, CT: JAI Press.
tardation and Developmental Disabilities Research Berquin, P. C., Giedd, J. N., Jacobsen, L. K.,
Center), as well as a grant from the Department of Hamburger, S. D., Krain, A. L., Rapoport, J. L.,
Defense (DAMD 17-00-1-0548). & Castellanos, F. X. (1998). Cerebellum in
attention-deficit hyperactivity disorder: A mor-
References phometric MRI study. Neurology, 50, 1087–
1093.
Adams, M. J. (1990). Beginning to read: Thinking Burd, L., Kauffman, D. W., & Kerbeshian, J. (1992).
and learning about print. Cambridge, MA: MIT Tourette syndrome and learning disabilities. Jour-
Press. nal of Learning Disabilities, 25, 598–604.
Crowe, F. W., Schull, W. J., & Neel. J. V. (1956). A
Aylward, E. H., Reiss, A. L., Reader, M. J., Singer, clinical, pathological, and genetic study of multi-
H. S., Brown, J. E., & Denckla, M. B. (1996). ple neurofibromatosis. Springfield, IL: Charles C
Basal ganglia volumes in children with attention Thomas.
deficit hyperactivity disorder. Journal of Child Cutting, L. E., Cooper, K. L., Koth, C. W., Mostof-
Neurology, 11, 112–115. sky, S. H., Kates, W. R., Denckla, M. B., & Kauf-
mann, W. E. (in press). Megalencephaly in NF1:
Barker, C. A., Garvey, M. A., Bartko, J. J., Denckla, Predominantly white matter contribution and
M. B., Wasserman, E. M., Castellanos, F. X., & mitigation by ADHD. Neurology.
Ziemann, U. (2001). The ipsilateral silent period Cutting, L. E., Huang, G., Zeger, S., Koth, C. W., &
(iSP) in children with attention deficit hyperactiv- Denckla, M. B. (2002). Specific cognitive func-
ity disorder (ADHD). Psychological Bulletin, tions remain “spared” and “impaired” over time
121, 65–94. in children with neurofibromatosis type–1:
Growth curve analyses of neuropsychological
Barker, D., Wright, E., Nguyen, K., Cannon, L., profiles. Journal of the International Neuropsy-
Fain, P., Goldgar, D., et al. (1987). Gene for von chological Society, 8, 838–846.
Recklinghausen neurofibromatosis is in the peri- Cutting, L. E., Koth, C. W., Burnette, C. P., Abrams,
centrometic region of chromosome 17. Science, M. T., Kaufmann, W. E., & Denckla, M. B.
236, 1100–1102. (2000). The relationship of cognitive functioning,
whole brain volumes, and T–2 weighted hyperin-
Barkley, R. A. (1997a). ADHD and the nature of tensities in neurofibromatosis type 1. Journal of
self-control. New York: Guilford Press. Child Neurology, 15, 157–160.
Cutting, L. E., Koth, C. W., & Denckla, M. B.
Barkley, R. A. (1997b). Behavioral inhibition, sus- (2000). How children with neurofibromatosis
tained attention, and executive functions: Con- type 1 differ from “typical” learning disabled
structing a unifying theory of ADHD. Psycholog- clinic attenders: Nonverbal learning disabilities
ical Bulletin, 121, 65–94. revisited. Developmental Neuropsychology, 17,
29–47.
Baumgardner, T. L., Singer, H. S., Denckla, M. B., Cutting, L. E., Koth, C. W., Mahone, E. M., &
Rubin, M. A., Abrams, M. T., Colli, M. J., & Denckla, M. B. (in press). Evidence for Unexpect-
Reiss, A. L. (1996). Corpus callosum morpholo- ed Weaknesses in Learning in Children with at-
gy in children with Tourette syndrome and atten- tention deficit hyperactivity disorder without
tion deficit hyperactivity disorder. Neurology, 47, reading disabilities. Journal of Learning Disabili-
477–482. ties.
Davis, C. J., Knopik, V. S., Olson, R. K.,
Benton, A. L., Hamsher, K. D., Varney, N. R., & Wadsworth, S. J., & DeFries, J. C. (2001). Genet-
Spreen, O. (1983). Judgment of Line Orientation. ic and environmental influences on rapid naming
In A. L. Benton et al. (Eds.), Contributions to and reading ability: A twin study. Annals of
neuropsychological assessment: A clinical manu- Dyslexia, 51, 231–247.
al. New York: Oxford University Press. DeFries, J. C., & Alarcón, M. (1996). Genetics of
specific reading disability. Mental Retardation
Berninger, V. W., Abbott, R. D., Abbott, S. P., Gra- and Developmental Disabilities Research Re-
ham, S., & Richards, T. (2002). Writing and views, 2, 39–47.
reading: Connections between language by hand DeFries, J. C., Filipek, P. A., Fulker, D. W., Olson,
and language by eye. Journal of Learning Dis- R. K., Pennington, B. F., Smith, S. D., & Wise, B.
abilities, 35, 39–56. W. (1997). Colorado Learning Disabilities Re-
search Center. Learning Disabilities: A Multidis-
Berninger, V. W., & Hart, T. (1992). A developmen- ciplinary Journal, 8, 7–19.
tal neuropsychological perspective for reading Denckla, M. B., Hofman, K., Mazzocco, M. M.,
and writing acquisition. Educational Psycholo- Melham, E., Reiss, A. L., Bryan, R. N., Harris, E.
gist, 27, 415–434. L., Lee, J., Cox, C. S., & Schuerholz, L. J. (1996).
Relationship between T2-weighted hyperintensi-
Berninger, V. W., & Rutberg, J. (1992). Relation- ties (UBOs) and lower IQs in children with neu-
ship of finger function to beginning writing: Ap-
plication to diagnosis of writing disabilities. De-
velopmental Medicine and Child Neurology, 34,
155–172.

Berninger, V. W., & Swanson, H. L. (1994). Modi-
fying Hayes & Flower’s model of skilled writing
to explain beginning and developing writing. In
E. Butterfield (Ed.), Children’s writing: Toward a

138 CAUSES AND BEHAVIORAL MANIFESTATIONS

rofibromatosis–1. American Journal of Medical the International Neuropsychological Society, 1,
Genetics, 67, 98–102. 511–516.
Denckla, M. B., & Rudel, R. G. (1976). Rapid Au- Hofman, K. J., Harris, E. L., Bryan, N., & Denckla,
tomatized Naming Test (R. A. N.): Dyslexia dif- M. B. (1994). Neurofibromatosis type 1: The
ferentiated from other learning disabilities. Neu- cognitive phenotype. Journal of Pediatrics, 124,
ropsychologia, 14, 471–479. S1-S8.
Denckla M. B., & Rudel R. G. (1978). Anomalies Hooper, S. R., Swartz, C. W., Wakely, M. B., de
of motor development in hyperactive boys with- Kruif, R. E. L., & Montgomery, J. W. (2002). Ex-
out learning disabilities. Annals of Neurology, 3, ecutive functions in elementary school children
231–233. with and without problems in written expression.
Dilts, C. V., Carey, J. C., Kircher, J. C., Hoffman, R. Journal of Learning Disabilities, 35, 57–68.
O., Creel, D., Ward, K., Clark, E., & Leonard, C. Itoh, T., Magnaldi, S., White, R. M., Denckla, M.
O. (1996). Children and adolescents with neu- B., Hofman, K. J., & Naidu, S., & Bryan, R.
rofibromatosis 1: A behavioral phenotype. De- (1994). Neurofibromatosis type 1: The evolution
velopmental and Behavioral Pediatrics, 17, of deep gray and white matter MRI abnormali-
229–239. ties. American Journal of Neurology, 15, 1–7.
DiMario, F. J., Jr., & Ramsby, G. (1998). Magnetic Koth, C. W., Cutting, L. E., & Denckla, M. B.
resonance imaging lesion analysis in neurofibro- (2000). The association of neurofibromatosis
matosis type 1. Archives of Neurology, 55, type 1 and attention deficit hyperactivity disor-
500–505. der. Child Neuropsychology, 6, 185–194.
DiPaolo, D. P., Zimmerman, R. A., Rorke, L. B., Kraut, M. A., Gerring, J. P., Cooper, K. L, Thomp-
Zackai, E. H., Bilaniuk, L. T., & Yachnis, T. A. son, R. E., Denckla, M. B., & Kaufmann, W. E.
(1995). Neurofibromatosis type 1: Pathologic (2002). Longitudinal evolution of T2-weighted
substrate of high-signal intensity foci in the brain. hyperintensities in children with neurofibromato-
Radiology, 195, 721–724. sis Type 1. Manuscript submitted for publication.
Eliason, M. J. (1986). Neurofibromatosis: Implica- Leckman, J. F., & Cohen, D. J. (1999). Evolving
tions for learning and behavior. Journal of Devel- models of pathogenesis. In J. F. Leckman & D. J.
opmental Pediatrics, 7, 175–179. Cohen (Eds.), Tourette’s syndrome: Tics, obses-
Fredericksen, K. A., Cutting, L. E., Kates, W. R., sions, compulsions (pp. 155–176). New York:
Mostofsky, S. H., Singer, H. S., Cooper, K. L., Wiley.
Lanham, D. C., Denckla, M. B., & Kaufmann, Leckman, J. F., King, R. A., & Cohen, D. J. (1999).
W. E. (2002). Disproportionate increases of white Tic and tic disorders. In J. F. Leckman & D. J. Co-
matter in right frontal lobe in Tourette syndrome. hen (Eds.), Tourette’s syndrome: Tics, obsessions,
Neurology, 58, 85–89. compulsions (pp. 23–42). New York: Wiley.
Friedman, J. M. (1999). Epidemiology of neurofi- Lyon, G. R. (1995). Toward a definition of dyslexia.
bromatosis type 1. American Journal of Medical Annals of Dyslexia, 45, 3–27.
Genetics, 89, 1–6. Mackintosh, N. J. (1998). IQ and human intelli-
Geary, D. C. (1990). A componential analysis of an gence. Oxford: Oxford University Press.
early learning deficit in mathematics. Journal of Mahone, E. M., Cirino, P. T., Cutting, L. E., Cer-
Experimental Child Psychology, 49, 363–383. rone, P. M., Hagelthorn, K. M., Hiemenz, J. R.,
Geary, D. C. (1992). Counting knowledge and skill Singer, H. S., & Denckla. M. B. (in press). Validi-
in cognitive addition: A comparison of normal ty of the Behavior Rating Inventory of Executive
and mathematically disabled children. Journal of Function in children with ADHD and/or Tourette
Experimental Child Psychology, 54, 372–391. Syndrome. Archives of Clinical Neuropsychology
Geary, D. C. (1993). Mathematical disabilities: Mahone, E. M., Koth, C. W., Cutting, L. E., Singer,
Cognitive, neuropsychological, and genetic com- H. S., & Denckla, M. B. (2001). Executive func-
ponents. Psychological Bulletin, 114, 345–362. tion in fluency and recall measures among chil-
Golden, G. S. (1984). Gilles de la Tourette’s syn- dren with Tourette syndrome or ADHD. Journal
drome following methylphenidate administra- of the International Neuropsychological Society,
tion. Developmental Medicine and Child Neurol- 7, 102–111.
ogy, 16, 76–78. Mazzocco, M. M. M. (2001). Math learning dis-
Goldgar, D. E., Green, P., Parry, D. M., & Mulvhill, ability and math ld subtypes: Evidence from stud-
J. J. (1989). Multipoint linkage analysis in neu- ies of Turner syndrome, fragile x syndrome, and
rofibromatosis type 1: An international collabo- neurofibromatosis type 1. Journal of Learning
ration. American Journal of Human Genetics, Disabilities, 34, 520–533.
44, 6–12. Mazzocco, M. M. M., Turner, J. E., Denckla, M. B.,
Gutman, D. H., & Collins, F. S. (1993). Neurofi- Hofman, K. J., Scanlon, D. C., & Vellutino, F. R.
bromatosis type 1: Beyond positional cloning. (1995). Language and reading deficits associated
Archives of Neurology, 50, 1185–1193. with NF1: evidence for not-so-nonverbal learning
Harris, E. L., Schuerholz, L. J., Singer, H. S., Read- disability. Developmental Neuroscience, 11,
er, M. J., Brown, J. E., Cox, C., Mohr, J., Chase, 503–522.
G. A., & Denckla, M. B. (1995). Executive func- Moore, B. D., Slopis, J. M., Jackson, E. F., De Win-
tion in children with Tourette syndrome and/or ter, A. E., & Leeds, N. E. (2000). Brain volume in
attention deficit hyperactivity disorder. Journal of children with neurofibromatosis type 1: Relation

Attention: Relationships between ADHD and LD 139

to neuropsychological status. Neurology, 54, Singer, H. S., & Denckla, M. B. (1997). Neuro-
914–920. motor functioning in children with Tourette syn-
Moore, B. D., Slopis, J. M., Schomer, D., Jackson, drome with and without attention deficit hyper-
E. F., & Levy, B. M. (1996). Neuropsychological activity disorder. Neurology, 12, 438–442.
significance of areas of high signal intensity on Shaywitz, S. E., & Shaywitz, B. A. (1999). Dyslexia:
brain MRIs of children with neurofibromatosis. From epidemiology to neurobiology. In D. D.
Neurology, 46, 1660–1668. Duane (Ed.), Reading and attention disorders:
Mostofsky, S. H., Cooper, K. L., Kates, W. R., Neurobiological correlates (pp. 113–128). Timo-
Denckla, M. B., & Kaufmann, W. E. (in press). nium, MD: York Press.
Smaller prefrontal and premotor volumes in boys Shue, K. L., & Douglas, V. I. (1992). Attention
with attention deficit/hyperactivity disorder. Bio- deficit hyperactivity disorder and the frontal lobe
logical Psychiatry, 52, 785–794. syndrome. Brain and Cognition, 20, 104–124.
Mostofsky, S. H., Lasker, A. G., Singer, H. S., Singer, H. S., Reiss, A. L., Brown, J. E., Aylward, E.
Denckla, M. B., & Zee, D. S. (2001). Oculomo- H., Shih, B., Chee, E., Harris, E. L., Reader, M.
tor abnormalities in boys with Tourette syndrome J., Chase, G. A., & Bryan, R. N. (1993). Volu-
with and without ADHD. Journal of the Ameri- metric MRI changes in basal ganglia of children
can Academy of Child and Adolescent Psychiatry, with Tourette’s syndrome. Neurology, 43,
40, 1464–1472. 950–956.
Mostofsky, S. H., Reiss, A. L., Lockhart, P., & Singer, H. S., Schuerholz, L. J., & Denckla, M. B.
Denckla, M. B. (1998). Evaluation of cerebellar (1995). Learning difficulties in children with
size in attention deficit hyperactivity disorder. Tourette’s syndrome. Journal of Child Neurology,
Journal of Child Neurology, 13, 434–439. 10, S58–S61.
North, K. N., Joy, P., Yuille, D., Cocks, N., Mobbs, Smith, S. D, Kelley, P. M., Askew, J. W., Hoover, D.
E., Hutchins, P., McHugh, K., & de Silva, M. M., Deffenbacher, K. E., Gayan, J., Brower, A.
(1994). Specific learning disability in children M., & Olson, R. K. (2001). Reading disability
with neurofibromatosis type 1: Significance of and chromosome 6p21. 3: Evolution of MOG as
MRI abnormalities. Neurology, 44, 878–883. a candidate gene. Journal of Learning Disabili-
North, K. N., Riccardi, V., Samango-Sprouse, C., ties, 34, 512–519.
Ferner, R., Moore B., Legius E., Ratner, N., & Steen, R. G., Taylor, J. S., Langston, J. W., Glass, J.
Denckla, M. B. (1997). Cognitive function and O., Brewer, V. R., Reddick, W. E., Mages, R., &
academic performance in neurofibromatosis 1: Pivnick, E. K. (2001). Prospective evaluation of
consensus statement from the NF1 cognitive dis- the brain in asymptomatic children with neurofi-
orders task force. Neurology, 48, 1121–1127. bromatosis type 1: Relationship of macrocephaly
Ozonoff, S. (1999). Cognitive impairment in neu- to T1 relaxation changes and structural brain ab-
rofibromatosis type 1. American Journal of Med- normalities. American Journal of Neuroradiolo-
ical Genetics, 89, 45–52. gy, 22, 810–817.
Reader, M. J., Harris, E. L., Schuerholz, L. J., & Stine, S. B., & Adams, W. V. (1989). Learning prob-
Denckla, M. B. (1994). Attention deficit hyperac- lems in neurofibromatosis patients. Clinical Or-
tivity disorder and executive dysfunction. Devel- thopaedics and Related Research, 245, 43–48.
opmental Neuropsychology, 10, 493–512. Stumpf, D. A., Alksne, J. F., & Annegers, J. F.
Riccardi, V. M. (1981). Von Recklinghausen neu- (1988). Neurofibromatosis. Archives of Neurolo-
rofibromatosis. New England Journal of Medi- gy, 45, 575–578.
cine, 305, 1617–1627. Wang, P. Y., Kaufmann, W. E., Koth, C. W., Denck-
Said, S. M., Yeh, T. L., Greenwood, R. S., Whitt, J. la, M. B., & Barker, P. B. (2000). Thalamic in-
K., Tupler, L. A., & Krishman, K. R. (1996). volvement in neurofibromatosis type 1: Evalua-
MRI morphometric analysis and neuropsycho- tion with proton MR spectroscopic imaging.
logical function in patients with neurofibromato- Annals of Neurology, 47, 477–487.
sis. Neuroreport, 7, 1941–1944. Weschler, D. (1974). Weschler Intelligence Scale for
Schuerholz, L. J., Baumgardner, T. L., Singer, H. S., Children—Revised. New York: Psychological
Reiss, A. L., & Denckla, M. B. (1996). Neu- Corporation.
ropsychological status of children with Tourette’s Wolf, M., & Bowers, P. G. (1999). The double-
syndrome with and without attention deficit hy- deficit hypothesis for the developmental dyslexias.
peractivity disorder. Neurology, 46, 958–965. Journal of Educational Psychology, 91, 415–438.
Schuerholz, L. J., Cutting, L. E., Mazzocco, M. M.,

9

RAN’s Contribution to
Understanding Reading Disabilities

Patricia Greig Bowers
Galit Ishaik

Much research evidence has accumulated derlying slow naming has not been resolved,
demonstrating that phonological processing, the results of the many studies in this area
especially sensitivity to the individual have enriched our knowledge about reading
phonemes in oral language, plays an impor- acquisition and reading fluency. The per-
tant role in learning to read not only English spective taken in this chapter is that slow
but other languages as well. A major issue naming speed marks a second core deficit
for the field investigating cognitive bases for associated with reading disabilities whose
reading disabilities is that of the sufficiency effects are reflected in a particular profile of
of this factor in explaining reading difficul- reading skills.
ties. Are other cognitive differences (e.g.,
working memory and naming speed) which The possibility that naming simple visual
characterize reading disabled versus normal- stimuli and reading tap similar processes
ly achieving readers just correlates of the was first suggested by Geschwind (1965a,
phonological problems or consequences of 1965b). However, Denckla (1972) and
poor reading? Or, are they somewhat inde- Denckla and Rudel (1974, 1976) provided
pendent correlates or causes of reading diffi- evidence that it was the speed rather than
culties? This chapter addresses one variable the accuracy of naming such stimuli (letters,
for which this question has been debated, digits, color patches, and pictures of simple
the rapid naming of highly familiar visual objects) that was related to reading skill.
symbols. Is slow naming a marker for un- They reported that naming speed for stimu-
derlying problems associated with reading lus arrays (five highly familiar items repeat-
acquisition not explained by phonological ed 10 times in semirandom order) distin-
difficulties (Bowers & Wolf, 1993; Wolf & guished children with reading disabilities
Bowers, 1999)? Or, is slow naming speed a from children with other learning disabili-
type of phonological problem partially dis- ties as well as from normally achieving chil-
tinct from phoneme awareness but still tap- dren. They called the test they devised
ping a similar underlying deficit (e.g., Wagn- Rapid Automatized Naming (RAN). (Fig-
er, Torgesen, & Rashotte, 1994)? Although ure 9.1 displays the RAN Digits format.)
the debate about the nature of the deficit un- Number and letter arrays are usually the
more sensitive discriminators of reading

140

RAN’s Contribution to Understanding Reading Disabilities 141

26 49 726479 dressed were whether speed of processing
97 26 472946 deficits were limited to the language do-
74 62 946297 main and whether the relationship between
46 27 924976 naming speed and reading was found in
62 79 476249 languages other than English. Naming
speed’s role in predicting response to reme-
FIGURE 9.1. RAN Digits. Adapted from diation attempts and the type of remedia-
Denckla and Rudel (1974). tion particularly relevant to children with
naming-speed deficits have been investigat-
skill (e.g., Wolf, Bally, & Morris, 1986), ed more recently. A theoretical basis for the
but in samples containing young children empirical relationships between naming
or severely dyslexic participants, time to speed and reading has been much more dif-
name color and object arrays also distin- ficult to establish, but some progress along
guish groups well (e.g., Meyer, Wood, Hart, these lines is described later in the chapter.
& Felton, 1998).
Our lab’s contribution to this literature
The RAN format has been adopted by has been to help establish the parameters of
many researchers. However, other formats the RAN–reading relationship and to ex-
have also been used. For example, as early plore theoretical issues concerning its basis.
as 1974, Spring and Capps reported that the Our empirical work has been conducted us-
speed of naming 50 single digits on one line ing children in grades 2 through 5 in nor-
was associated with reading disability. Re- mal, publicly funded classrooms in several
cently, a rapid naming test with a slightly small cities in Ontario. Sampling strategies
different format has been included in the have varied. Some studies report results
Comprehensive Test of Phonological Pro- from the whole range of abilities found in
cessing (Wagner, Torgesen, & Rashotte, such classrooms, whereas other studies
1999), with published normative informa- screen subjects from those classrooms to fit
tion. Denckla and Wolf (in press) are pub- subtypes based on rapid naming and phone-
lishing RAN stimuli closer to the original mic awareness (PA) measures. PA is typical-
set and format with normative data. The ly assessed by a phoneme deletion measure,
terms “RAN,” “rapid naming,” and “nam- the Auditory Analysis Test (Rosner & Si-
ing speed” are often used interchangeably to mon, 1971). Another strategy is selecting
indicate serial list measures, with perfor- children considered reading disabled,
mance reported either by the time to name chronological age controls, and sometimes
whole lists or by items per second, a metric reading age controls. Children are called
with better psychometric properties. Wolf reading disabled or poor readers if they
(1986) has developed a “Rapid Alternating score at or below the 25th percentile on
Stimuli” (RAS) serial list, which alternates standardized tests of word recognition. Be-
numbers and letters or numbers, letters, and cause speed of naming simple items increas-
colors. RAS discriminates dyslexics from es with age for all children, with gains di-
controls well but is not further reviewed minishing by grade 5 for normal readers
here. (Flowers, Meyer, Lovato, Felton, & Wood,
2001), we have preferred to study children
During the 1980s, Blachman (1984), in a small age range to minimize the impact
Mann (1984), Wolf (1982), Bowers, Steffy, of age-related variance. In our samples, the
and Swanson (1986), and Wagner and reliability of rapid naming tests is impres-
Torgesen (1987) independently pursued the sive: test–retest reliabilities are above .90
role naming speed played in the emerging and stability over 1- and 2-year periods
picture of the correlates and precursors of above .85. Our work over the years reflects
reading disability or dyslexia. Research in the three broad themes outlined below,
the 1990s investigated a variety of topics, which are then reviewed in greater depth.
such as rapid naming’s relationships to var-
ious reading subskills and its degree of in- Our initial focus was to investigate the in-
dependence from other cognitive processes dependence of rapid naming’s contribution
related to dyslexia. Other questions ad- to reading from that of phonological aware-
ness, memory span, and verbal ability. The
early work revealed that controlling for

142 CAUSES AND BEHAVIORAL MANIFESTATIONS

memory span or verbal ability in samples Focus 1: Independent Contributions of
from either clinic (Bowers, Steffy, & Tate, RAN to Reading
1988) or classroom (e.g., Bowers & Swan-
son, 1991) did not appreciably affect the Should rapid naming be considered one of
moderately strong relationships found be- several phonological skills as suggested by
tween rapid naming and reading. Determin- Wagner and colleagues (1994)? Or, are the
ing the extent of the independent versus phonological aspects of rapid naming only
overlapping contributions of phoneme part of its complex nature, with individual
awareness and naming speed to a variety of differences in the ability to rapidly integrate
different skills in reading was a much more several processes the more distinctive at-
complex task. tribute of the test (Wolf, 1991; Wolf, Bow-
ers, & Biddle, 2000)? Many studies have
A second focus of research involved a used factor-analytic techniques to determine
fruitful and continuing collaboration with the factor structure of various reading-relat-
Maryanne Wolf of Tufts University. During ed tasks (e.g, DeJong & van der Leij, 1999,
a sabbatical at Tufts in 1990–1991, Bowers in Dutch; Wagner et al., 1994, in English-
and Wolf joined forces to try to understand speaking samples; Wagner & Torgesen,
the implications of work on rapid naming 1987). Measures of these variables form
done in our respective labs (and those of three factors reflecting phonemic awareness,
others) for theories of dyslexia, especially phonological memory, and RAN, with the
delayed growth of orthographic skill. While first two factors being more strongly corre-
our initial conceptualization was published lated with each other than with rapid nam-
in Bowers and Wolf (1993), the University ing. Recent work in our lab (Ishaik, Bowers,
of Waterloo lab has continued to explore & Steffy, 2001) suggests that measures of
the association between orthographic pro- verbal working memory (involving both
cessing and rapid naming (e.g., Bowers, storage and manipulation of verbal materi-
Golden, Kennedy, & Young, 1994; Bowers, al) overlap considerably with phoneme dele-
Sunseth, & Golden, 1999). In addition, tion and sound categorization, common
both Wolf and Bowers have pursued the im- measures of phoneme awareness, but are
plications of the separable contributions to distinct from rapid naming when predicting
reading of phonological awareness and reading accuracy.
naming speed by positing subtypes of read-
ers with no, only one, or “double” deficits An early concern in our lab was whether
in the two cognitive skills. Wolf has ex- phonemic awareness and rapid naming
plored effects of interventions for severely were related to different types of reading
dyslexic children, typically those with dou- skill. Average and poor readers were select-
ble deficits. She and her colleagues (Wolf, ed in grade 2 and followed until grade 4,
Miller, & Donnelly, 2000) developed a re- providing evidence about the concurrent
mediation program, “Retrieval, Automatici- and predictive relationships between these
ty, Vocabulary Elaboration, Orthography variables and a variety of reading skills
(RAVE-O),” as a supplement to phonologi- (Bowers, 1995; Bowers & Swanson, 1991).
cal training; it targets remediation of the Controlling for oral vocabulary skill, both
fluency and orthographic skill deficits asso- phonemic awareness and rapid naming typi-
ciated with slow RAN. cally contributed shared and unique vari-
ance to word recognition, with phonemic
A third focus of our work has been to un- awareness playing a larger role. Rapid nam-
derstand the “why” of the association be- ing’s unique contribution to nonword de-
tween rapid naming and reading (e.g., Bow- coding was small relative to the strong con-
ers, 2001; Bowers & Wolf, 1993; Wolf & tribution of phonemic awareness. However,
Bowers, 1999). Specifically, what types of RAN’s strong, unique contribution to the
processes relevant to reading are being latency of correct identification of regular
tapped by this simple test? We explored this and exception words, whether of high or
issue in the early years by comparing differ- moderate frequency, as well as to reading
ent formats for rapid naming and later by comprehension, contrasted with the in-
devising measures to test theoretical links significant unique contribution of phonemic
between RAN and specific aspects of read- awareness to these measures. Naming
ing skill.

RAN’s Contribution to Understanding Reading Disabilities 143

speed’s contribution to reading comprehen- struct. That the pattern of relationship
sion was fully explained through its associa- strength differs for PA and RAN is evident
tion with latency of word recognition. Levy from Table 9.1. Other researchers do not al-
(2001) reviewed work highlighting the cen- ways find such strong relationships (e.g.,
tral importance of word recognition speed Torgesen et al., 1997), but the patterns are
to reading fluency and comprehension and replicated. Reflecting a profile opposite to
confirmed the special relationship of RAN that of phonemic awareness, RAN is more
to such speed. related to recognition of exception words
and to knowledge of orthographic patterns
The pattern of differential relationships than to phonological decoding.
of phonemic awareness and naming speed
to types of reading skill found in the early Levy (2001) underlines an important dis-
study in our lab has been replicated in sever- tinction between factors associated with
al other studies (e.g., Carver, 1997; Corn- learning to read unfamiliar words and fac-
wall, 1992; Manis, Doi, & Bhadha, 2000; tors associated with speeding the processing
Manis, Seidenberg, & Doi, 1999; Torge- of print by automatizing access to represen-
sen, Wagner, Rashotte, Burgess, & Hecht, tations of words already somewhat familiar
1997). Table 9.1 (excerpted from Manis et to the child. RAN is related to both factors
al., 2000) highlights this finding. It is based but perhaps in different ways. To study each
on 85 children tested at the end of grade 2 of these factors, different designs are re-
who were representative of the full range of quired. Study of automatizing access to
reading abilities in classrooms of two public print, as indexed by text reading speed, pre-
elementary schools. Only children with lim- supposes a high level of reading accuracy. If
ited English were excluded from the study. text is too difficult, poor readers’ decoding
For the Commonality Analyses presented deficits will affect speed and obscure the
here, Wechsler Intelligence Scale for Chil- view of other factors also affecting fluency.
dren—III (WISC-III; Wechsler, 1991) Vo- Therefore, careful choice of text is impera-
cabulary was entered at Step 1 to control tive to a study of determinants of fluent
for general verbal ability. Then a measure of reading (Young & Bowers, 1995). Noting
RAN and of phoneme awareness was en- the strong relationship between rapid nam-
tered at Steps 2 or 3 to provide estimates of ing and latency of correct word recognition,
the common and unique variance con- researchers in our lab and others have
tributed to various reading measures. Both looked at predictors of reading fluency be-
RAN Digits and Letters were administered fore and after practice with text. The chil-
and PA was measured by both Sound Dele- dren in the longitudinal study reported ear-
tion and Sound Blending. In general, lier (i.e., Bowers & Swanson, 1991) took
stronger relations were evident for RAN part each year in a study of repeated reading
Letters and for Sound Deletion, and these of text chosen to be at a level of difficulty
are presented here. However, the pattern appropriate to each child’s reading skill
was similar for both measures of each con- (Bowers, 1993). Not surprisingly, their text

TABLE 9.1. Hierarchical Regression Analyses Predicting Reading Subskills: Unique and Common
Variance for RAN-Letters (RAN) and Sound Deletion (PA)

Letter

Word Word Orthographic String Exception

Variable Identification Nonwords Attack Comprehension Choice Choice Words

Vocabulary 13.6*** 8.1** 5.7* 23.4*** 1.6 0.5 11.8**
Common 16.8 12.9 12.4 9.7 10.9 9.5 16.0

RAN–PA 17.1*** 7.1** 5.5** 9.3*** 12.8*** 11.0** 21.7***
RAN unique 13.9*** 18.7*** 27.7*** 12.3*** 9.6***
PA unique 7.6** 4.2*

Note. From Manis, Doi, and Bhadha (2000, Table 3, p. 329) Copyright 2000 by Pro-Ed, Inc. Adapted with
permission.
*p < .05; **p < .01; ***p < .001.

144 CAUSES AND BEHAVIORAL MANIFESTATIONS

reading speed on the first reading of the text ly, Meyer and colleagues (1998) report that
was associated with their digit naming variance in rapid naming in grade 3 poor
speed. However, the fluency of those chil- readers was related predictively to grade 8
dren with better naming speed increased reading skill, whereas in grade 3 average
more after repeated reading of text than did readers, it was unrelated to grade 8 skill.
the fluency of slower naming children, even Davis, Knopik, Olson, Wadsworth, and De-
after controlling for initial fluency. A subse- Fries (2001) found that the correlation be-
quent study of practice with words and tween RAN and reading in a group of
nonwords reported in Bowers and Kennedy twins, at least one of whom had a reading
(1993) and a study of text reading by Young disability (“low range group”), was higher
(1997) with grade 5 children with reading than in the group of control twins, neither
disabilities produced a similar pattern of re- of whom had a reading disability (“normal
sults. That is, rapid naming was associated range group”). Not all studies have found
not only with initial fluency but also with stronger correlations between RAN and
gains in fluency after practice, suggesting reading in samples composed of just poor
greater automatizing skill of those with readers (e.g., Torgesen et al., 1997). Never-
faster naming speed. Not all studies have theless, an implication of findings may be
confirmed the details of these results. For that RAN contributes variance to reading
example, Levy, Abello, and Lysynchuk accuracy measures in the lower ranges of
(1997) found word and text practice as suc- skill, while only its relationship to reading
cessful in increasing word recognition speed fluency may persist at higher levels.
for slow RAN poor readers as it was for
faster RAN poor readers. RAN deficits are heritable to some extent
(e.g., Compton, Davis, DeFries, Gayan, &
Although the connection between reading Olson, 2001; Davis et al., 2001). Compton
speed and RAN is especially strong, word and colleagues (2001) report separable heri-
recognition accuracy is also associated with tabilities for phonemic awareness and rapid
RAN, as evident in Table 9.1. Measured in naming in their sample of monozygotic and
kindergarten or grade 1, RAN predicts dizygotic twins. Furthermore, “subjects
reading accuracy in grades 1 to 3 (Wagner with deficits in alphanumeric RAN skill
et al., 1997; Wolf et al., 1986) and even in tend to have deficits in word reading skills
grade 5 (Kirby, Parrila, & Pfeiffer, 2001). that are influenced, in part, by a common
Differences in naming speed between chil- set of genes” (Compton et al., 2001, p.
dren with reading disabilities and normally 285). Grigorenko and colleagues (2001)
achieving children are found at many ages have suggested locations on chromosome 1
(e.g., Lovett, 1987). Even adult dyslexics and 6 for reading impairments associated
continue to be characterized by slow nam- with rapid naming, a linkage especially
ing speed (e.g., Felton, Naylor, & Wood, strong for those with deficits in both rapid
1990). Success in learning to read English as naming and phonemic awareness.
a second language is predicted by phonolog-
ical awareness and naming speed (Geva, Research investigating the role of RAN in
Yaghoub-Zadeh, & Schuster, 2000), just as reading achievement in languages other
is learning to read English when it is one’s than English has been especially informa-
native language. tive. Compared to its role in English-speak-
ing samples, RAN plays a relatively larger
Variability on RAN within poor reader role in prediction of reading in languages
groups is associated with reading perfor- such as German (Wimmer, 1993) and Dutch
mance as well. McBride-Chang and Manis (Van den Bos, 1998). In those languages,
(1996) found that both rapid naming and phonological demands are more easily met
phonemic awareness contributed indepen- than in English due to the higher regularity
dently to variability in reading accuracy of symbol/sound correspondence. Accuracy
within a poor reader group, but variance in of word recognition of German-speaking
oral vocabulary did not. On the other hand, dyslexics is quite high by grade 2 (as is
within an average-and-above reader group, phonemic awareness). Nevertheless, slow
rapid naming contributed no independent reading and poor spelling are persistent bot-
variance to reading, but both phonemic tlenecks to performance (Wimmer,
awareness and oral vocabulary did. Similar- Mayringer, & Landerl, 2000) and, consis-

RAN’s Contribution to Understanding Reading Disabilities 145

tent with English-language research, are as- before conclusions can be drawn with con-
sociated with slow RAN. As well, Chinese fidence.
dyslexics are characterized by RAN deficits
even more strongly than by phonological In summary, slow naming speed forms a
awareness deficits (Ho, 2001), presumably factor separate from factors for phonemic
because Chinese orthography does not re- awareness or working memory. It is highly
quire as much reliance on the phoneme level related to fluent reading and has a profile of
of analysis of words as does English. Phono- relationships to measures of reading subskill
logical memory (i.e., immediate repetition accuracy different from that of phonemic
of digits and of syllables presented auditori- awareness. For example, unlike phoneme
ally) discriminated Chinese dyslexics from deletion, RAN is related more strongly to
both chronological-age and reading-age orthographic skill than to phonemic decod-
controls, and rapid naming, whether on ing. RAN is more highly related to reading
continuous lists or discrete item presenta- in samples of readers of low skill, such as
tions, discriminated Chinese dyslexics from beginning readers or children with RD than
age controls (Ho & Lai, 2000). in higher-skilled samples. RAN deficits are
heritable somewhat separately from phone-
The degree to which RAN deficits are mic awareness deficits. RAN’s role in read-
specific to reading difficulties as distinct ing proficiency varies according to the
from other learning and attention problems characteristics of the orthography of the
has been another area of concern. Denckla language. For example, in languages in
and Rudel (1976) reported such specificity which decoding is “easy” due to the high
in comparison to other learning disorders. regularity of symbol–sound correspon-
More recently, researchers have distin- dence, RAN plays a larger role in distin-
guished samples of children diagnosed guishing dyslexics from normally achieving
as attention-deficit/hyperactivity disorder readers than it does in English. RAN al-
(ADHD) with and without reading disabili- phanumeric deficits do not characterize
ty (RD) and those with RD alone or no di- ADHD children without a reading disorder
agnosis; they found that RAN was associat- but may be associated with arithmetic com-
ed with RD, not ADHD (Felton, Wood, putation difficulties.
Brown, & Campbell, 1987; Nigg, Hinshaw,
Carte, & Treuting, 1998). Several re- Focus 2: The Double-Deficit Hypothesis
searchers have reported an interesting dis-
tinction between the more automatized The fact of independent contributions to
RAN digits and letters and the less automa- reading skill by phonemic awareness and
tized RAN colors and objects. It appears rapid naming, and the differential profile of
ADHD and control children do not differ contributions to reading subskills, led to the
on RAN Digits or Letters but do differ on hypothesis that children with deficits in
RAN objects or colors (Carte, Nigg, & both skills would be the poorest readers
Hinshaw, 1996; Semrud-Clikeman, Guy, & (Bowers & Wolf, 1993; Wolf & Bowers,
Griffin, 2000; Tannock, Martinussen, & 1999). This pattern of data was demonstrat-
Frijters, 2000). The specificity of the rela- ed by reanalysis of several data sets in the
tionship between naming speed for the Wolf and the Bowers labs, as well as labs of
more automatized symbols and reading dis- Manis and of Lovett (Lovett, Steinbach, &
abilities compared to the more general rela- Frijters, 2000; Manis et al., 2000; Wolf,
tionship of naming speed for stimuli requir- Bowers, & Biddle, 2000). In several types of
ing more controlled processing is a samples, it was possible to select children
fascinating pattern worthy of further study. without deficits in either skill, with only a
Although the literature is sparse, RAN deficit in one but not the other skill, and
deficits may be associated with arithmetic those with double deficits. The no-deficit
computation difficulties as well as reading and double-deficit groups were the best and
disabilities (Hecht, Torgesen, Wagner, & worst readers across many measures, with
Rashotte, 2001), perhaps associated with single-deficit children having variable pro-
the commonly found covariance between files, sometimes similar to each other on
disabilities in math and reading. Greater re- reading tasks and sometimes different.
search focus on math disability is needed

146 CAUSES AND BEHAVIORAL MANIFESTATIONS

Many studies have adopted a subtyping somewhat varied details of the findings. Re-
strategy but have varied the selection crite- sults from sampling full classrooms are dis-
ria when studying children with these deficit cussed first. Subtypes based on RAN and
patterns. Work in our lab has typically PA screening of children from full grade 3
screened full class samples to select children classes (Bowers et al., 1999; Sunseth &
with strengths and/or weaknesses in RAN Bowers, 2002), have found single-naming-
and phoneme deletion skill (Baker, 2002; speed-deficit children (above 50th percentile
Bowers et al., 1999; Sunseth & Bowers, for their grade in phonemic awareness and
2002), a strategy also used by Manis and below 30th percentile in rapid naming)
colleagues (2000). Reading scores are free scoring in the average range on word identi-
to vary with such a strategy. Other studies fication tasks but being slow readers on
have first selected poor readers (e.g., Lovett easy text and poor spellers. They had espe-
et al., 2001; Meyer et al., 1998; Morris et cially poor performance on a spelling recog-
al., 1998) and then defined subgroups by nition test in which the correct spelling of a
RAN and PA scores among other variables. word needed to be chosen from several
In another variation, children who are poor foils. Single-phonological-deficit children
readers (e.g., Levy, Bourassa, & Horn, (above the 50th percentile in rapid naming
1999) have been subdivided into those with and below the 30th percentile on phonemic
relatively slow or fast RAN. The children in awareness) were low average in word recog-
both groups have poor phonological skills, nition but read the easy text reasonably
and therefore represent a single-phonologi- quickly; they had similarly poor spelling
cal-deficit group and a double-deficit group. skill. Both groups scored below the level of
no-deficit children on all spelling tasks.
Varied selection methods have led to

TABLE 9.2. Performance of Subtypes of Grade 3 Children

Double Phonological Naming speed Double
asset deficit
deficit deficit (n = 16)
(n = 17)
Measures (n = 17) (n = 18) 11.3 (2.7)b
1.4 (.2)b
Defining variables 23.2 (3.2)a 11.4 (2.4)b 22.4(2.6)a
AAT 2.1 (.3)a 2.1 (.2)a 1.5 (.1)b
RAN:D items/sec

Reading Accuracy Standard Scores 116.9 (16.0)a 92.2 (9.1)bc 100.3 (16.3)c 83.8 (6.9)b
(WJ-R tests; Woodcock & Johnson, 120.1 (22.2)a 88.4 (7.8)b 100.2 (11.9)c 82.3 (6.5)b
1989)

Word Identification
Word Attack

Reading speed and accuracy on 15.3 (4.6)a 21.3 (4.8)b 31.5 (13.8)c 37.1 (11.1)c
GORT III “easy” passage 0.1 (.3)a .4 (.8)ab 1.3 (2.7)ab 2.1 (1.8)b
(Wiederholt & Bryant, 1992)

Seconds
Errors

Spelling Standard Scores

Dictation (Test of Written Spelling III;

Larsen & Hammill, 1994)

Predictable words 103.5 (10.5)a 89.4 (10.2)b 91.7 (6.0)b 84.9 (6.6)b
88.2 (9.0)b 88.2 (10.7)b 80.9 (6.8)b
Unpredictable 97.5 (14.5)a 85.8 (7.0)b 81.4 (8.5)bc 78.9 (5.5)c

Recognition (Peabody Individual 96.5 (13.7)a

Achievement Test—Revised;

Markwardt Jr., 1989)

Note. Values sharing superscripts do not differ from each other. Data from Sunseth & Bowers (2002).

RAN’s Contribution to Understanding Reading Disabilities 147

However, whether they differed significantly and slow RAN poor readers, their compara-
from double-deficit children was less pre- bly low scores on PA tasks suggest they are
dictable. Table 9.2 highlights some of the re- equally well described as single phonologi-
sults reported by Sunseth and Bowers cal deficit (PD) and DD children, respective-
(2002). ly. Groups differed initially on relative read-
ing skill, with the slow RAN children worse
Sunseth and Bowers (2002) found that readers than the faster RAN children. Also
the single-deficit groups did not differ in the in the Levy laboratory, samples of grade 2
percent of children categorized as poor poor readers selected by Conrad (2002)
readers, despite the better word identifica- found that the slow and fast poor reader
tion scores of single-naming-speed-deficit groups had similarly poor word identifica-
children on average. (Note however the tion and phonemic awareness scores. How-
large standard deviation on Word Identifi- ever, the slow RAN group had poorer scores
cation of these children.) Approximately on a careful revision for this age group of
30% of either single-deficit group could be the Olson, Kliegl, Davidson, and Foltz
considered poor readers using the 25th per- (1985) Orthographic Choice test. Choice of
centile cutoff definition on a Word Identifi- the correct spelling of a word from its ho-
cation test, while over 90% of double- mophone foil (e.g., truk vs. truck) on this
deficit (DD) children were so categorized. task requires word-specific orthographic
Similarly, Bowers and Newby-Clark (2002) knowledge, with decoding skill unhelpful.
report single-deficit groups having 20%
poor readers and a DD group with 81% Differential response of the deficit groups
poor readers. to instruction has also been studied. Levy
and colleagues (1999) investigated effects of
Subtyping full class samples based on training slow and fast RAN (i.e., DD and
kindergarten phonological processing and PD) poor readers. Twenty sessions of word
RAN scores has revealed later reading practice were given, varying the unit in the
achievement differences between groups. word made salient to participants. That is,
Kirby and colleagues (2001) found that the words were segmented in both visual and
grade 5 reading achievement of the DD aural presentations based on phonemes or
group lagged behind the no-deficit group by onset/rime, or were presented as whole
almost 2 years. Interestingly, they also found words. For example, in the onset/rime con-
that unlike the subtyping based on grade 3 dition, “band” would be pronounced in a
scores, the single-RAN-deficit group scored segmented fashion by the examiner as b/and
almost as poorly as did the DD group, while as well as shown visually, with “b” colored
the single-PA-deficit group fared much bet- differently from “and.” Controlling for
ter. That many kindergarten children’s PA their initially less skilled reading, DD chil-
deficits can be remedied by phonics training, dren made less progress over time than did
with rapid naming deficits being less affected the children with faster RAN under all con-
by interventions, may account for the differ- ditions of training. Although segmented
ence between results of Kirby and colleagues practice was associated with better progress
and Sunseth and Bowers. for all readers, it made a greater difference
for the slow RAN (i.e., DD) poor readers,
Another method for categorizing sub- for whom whole word practice was particu-
types relevant to the DD hypothesis selects larly disadvantageous. Levy (2001) pursued
poor readers first and then categorizes this issue by providing whole word practice
them. Most children in such samples (in to comparable DD children in which the
English-speaking countries) do have a phonological information was always at the
phonological deficit, but there is more vari- unsegmented whole word level, but the vi-
ability in rapid naming. Morris and col- sual information either did or did not make
leagues (1998) found that a subtype defined the rime unit salient. Under whole word
by cluster analysis that had both PA and conditions in which the rime unit was em-
RAN deficits was one of the most severely phasized visually by training the words in
impaired reader groups. Using a simpler blocks with the same rime, rather than in
strategy, Levy and colleagues (1999) and randomly mixed order, the DD children
Levy (2001) divided grade 2 poor readers learned about as well as DD children had in
into groups above and below the median of
a poor reader sample on RAN. Called fast

148 CAUSES AND BEHAVIORAL MANIFESTATIONS

the previous study with words segmented to form orthographic codes for commonly
both aurally and visually. These children seen letter strings. Many theories of skilled
seem to need extra orthographic support to reading posit interacting phonological and
notice the visual similarities between words. orthographic “routes” to word recognition.
Perhaps a neat correspondence could be
Lovett and colleagues (2000) reported found between a child having double
that in a sample of severely dyslexic chil- deficits and impairment in both orthograph-
dren, both single- and double-deficit chil- ic and phonological routes. Demonstrating
dren made sizable gains after instruction. clearly such a correspondence has proved
However, the DD children did not transfer difficult.
their knowledge to uninstructed words as
well as did single-deficit children. Although Several strategies have been employed in
they used multiple regression procedures an attempt to unravel the mystery of
rather than a subtyping strategy, Torgesen RAN–reading relationships. These include
and colleagues (1999) also found effects of varying the format of the test to explore ef-
slow RAN on later reading. In their sample fects on reading and devising tests tapping
of kindergarten children with weak letter hypothesized mediating links between RAN
naming accuracy and phonemic awareness and reading. Although no definitive solu-
skill, slower RAN was associated with less tion has been found, researchers in our lab
response to remediation over a 2½-year pe- and in many others have learned much
riod. about the parameters an eventual explana-
tion will need to accommodate.
In summary, a subtyping scheme based on
relative strengths and weaknesses in phone- Initially, Lynn Swanson asked whether
mic awareness and rapid naming has proved the format of the rapid naming test (discrete
robust and useful in distinguishing charac- item vs. serial list) mattered to the associa-
teristics of reading in children in full class tion with reading (Bowers, 1995; Bowers &
samples and in samples of children with Swanson, 1991; Swanson, 1989). If only the
RD. Children with double deficits typically serial list format is predictive of reading, the
read more poorly than did children in the more limited identification and name re-
other subgroups. Furthermore, within poor trieval required by report of a single isolated
reader groups, children with double deficits number or letter may not be focal to the re-
respond less well to remediation efforts than lationship. Instead, other processes involved
children with single phonological deficits, in managing a list of items, including atten-
even after controlling for any initial differ- tional ones, may be more important. Bow-
ences in reading skill. ers and Swanson (1991) reported that grade
2 average and poor readers differed on both
Focus 3: Why is Rapid Naming Related to discrete trial naming latencies and serial list
Reading?: Theoretical Explorations items per second. This conclusion was con-
firmed in the grade 4 data of these children
Bowers and Wolf (1993) and Bowers and (Bowers, 1995) and replicated recently in
colleagues (1994) were impressed by evi- Chinese children (Ho & Lai, 2000). Bowers
dence for RAN–orthographic skill relation- (1995) reported that the two methods of
ships. Orthographic skill (knowledge and measuring rapid naming correlated highly.
use of the specific letter patterns found in However, the serial list had the stronger re-
words) is known to be associated with lation with reading skill, and the contribu-
phonological skill and print exposure (Cun- tion to reading of discrete trial latencies was
ningham & Stanovich, 1990). However, entirely accounted for by the serial list mea-
RAN contributes additional variance to or- sure. Something “extra” was contributed by
thographic skill (e.g., Bowers et al., 1999; the need to name one item and then name
Conrad, 2002; Manis & Freedman, 2001; the next on the list in quick succession.
Manis et al., 2000; Sunseth & Bowers, However, results suggested that preprocess-
2002). Bowers and Wolf hypothesized that ing of the next item on the list did not ac-
the rapid naming deficit was associated with count for this “extra” variance. Following
reading skill through the processes underly- Swanson (1989), children were presented
ing rapid naming affecting a child’s ability with a computer task in which five numbers
(or letters) were displayed, with an arrow

RAN’s Contribution to Understanding Reading Disabilities 149

pointing to the one item they were to name, This result highlights the crucial role in the
always the item in the second position. Con- RAN–reading relationship played by differ-
ditions varied the relevance of the items to ential access time to symbol names, consis-
the right. In the relevant condition block of tent with the previously reported correla-
trials, the item to the immediate right of the tion with reading of pause duration on
target would be the next target to be named; serial lists, and latency of response on dis-
in the irrelevant condition block, it would crete trials.
not be. If faster serial list performance re-
flected some preprocessing of the items to How general are the processes that under-
the right of the target, differences between lie RAN performances? The answer to this
these conditions, especially for faster RAN question is unclear. Carver (1997) and Kail,
children, would be expected. Instead, no Hall, and Caskey (1999) argue that the as-
differences were found. In summary, al- sociation reflects general cognitive process-
though serial list presentation is not neces- ing speed. Kail and colleagues found that
sary to naming speed’s association with rapid naming was uniquely predicted by
reading skill, it does provide additional general speed of processing measures (i.e.,
reading-related variance. However, the ex- Visual Matching and Cross-Out tasks from
tra ingredient may not be the preprocessing the Woodcock–Johnson Tests of Cognitive
of subsequent items on the list. Ability). Controlling for age, RAN and
print exposure contributed unique variance
There have been other attempts to under- to reading recognition, but processing speed
stand processes associated with RAN by an- no longer did. They interpret this pattern of
alyzing serial list performance. Obregon findings to mean that RAN’s relationship to
and Wolf (1995) analyzed the responses of reading overlapped with the slightly smaller
children as they named items on the RAN, variance contributed to reading by more
timing various aspects of the response. Slow general processing speed. They argue that
and fast RAN children differed only in the “naming and reading are linked because
length of their pauses between naming skilled performance in both naming and
items, not in the articulation time for items reading depends, in part, on the rapid exe-
or in time managing the start of a new row cution of the underlying processes” (p.
of items. Similarly, Neuhaus, Foorman, 312). In a sample of normally achieving
Francis, and Carlson (2001) reported that children in grades 1 to 3, Cutting and
in first- and second-grade students, RAN Denckla (2001) report that scores on these
pause durations for numbers, letters, and same processing speed measures are indi-
objects were differentially related to read- rectly related to reading, through variance
ing, while articulation duration was rarely shared not only with RAN but also with
related to reading. The RAN letters pause phonemic awareness, memory span, and or-
time was the most robust predictor of sever- thographic knowledge. In our lab, Baker
al reading measures and predicted reading (2002) found that grade 2 DD readers dif-
even after controlling for pause time for ob- fered from other subgroups on Cross Out
jects. and Number Comparison processing speed
tasks similar to those used by Kail and col-
Scarborough and Domgaard (1998) test- leagues.
ed several hypotheses about the source of
the variance in RAN related to reading by This “domain general” view of process-
devising many different serial list tasks, sys- ing speed’s association with reading is con-
tematically altering just one variable (e.g., trasted with the “domain specific” view es-
the number of different items in a list). In- poused by Wimmer and Mayringer (2001),
terestingly, most alterations did not affect who used a different set of visual processing
the RAN–reading relationship. The one tasks. They found that latency of response
condition crucial to the task was actually for visual discrimination tasks not involving
naming letters rather than giving a yes/no familiar letters or numbers did not distin-
decision about whether the symbol (printed guish German children with rate or accura-
in different fonts) had a particular name. cy and rate reading problems from normal
The decision task reduces the demand to lo- reading controls, despite RAN deficits dis-
cate a new name because one name is al- criminating both groups quite well. Con-
ways kept in mind as decisions are made. flicting results provide no basis presently for

150 CAUSES AND BEHAVIORAL MANIFESTATIONS

strong conclusions about whether RAN rep- resulting variability in orthographic sensi-
resents domain general versus domain- tivity. Cutting and Denckla (2001) provide
specific processing speed associated with some support for this position, as the shared
reading. variance of processing speed and RAN was
related to orthographic skill in their normal-
A large literature about perceptual pro- ly achieving young reader sample. However,
cessing speed of dyslexics does not focus on unlike our hypothesis, RAN was also direct-
RAN performance but seems relevant to ly related to reading apart from the shared
constructs that may be tapped by RAN. variance with processing speed and ortho-
Farmer and Klein (1995) and Wolf, Bowers, graphic skill.
and Biddle (2000) review evidence concern-
ing visual and auditory reaction time and Manis and colleagues (1999) have sug-
other basic perceptual process findings asso- gested a different basis for the RAN–ortho-
ciated with dyslexia. Their reviews suggest graphic skill correlation, arguing that both
that differences between dyslexics and con- RAN and orthographic skill reflect the abil-
trols appear when stimuli are presented at ity to learn arbitrary associations. Still oth-
faster speeds and in series. Breznitz (2001) ers consider it more likely that rapid naming
has provided evidence that dyslexic children and orthographic skill are separate deficits
and adults have slower event-related poten- (e.g., Badian, 1997; Berninger, Abbott,
tials (ERP) to visual and auditory stimuli. Billingsley, & Nagy, 2001). Further empiri-
Her hypothesis that it is the greater asyn- cal work exploring these theoretical issues
chrony of the responses that undermines the has led to a revision of the original hypothe-
amalgamation of phonological and ortho- sis which places less stress on the mediating
graphic knowledge for dyslexics will be dis- role of orthographic skill while continuing
cussed more fully later. Keen and Lovegrove to highlight the impact of letter string pro-
(2000) report that dyslexics have a “slug- cessing deficits (Bowers, 2001).
gish” visual processing system. Fawcett and
Nicolson (2001) have reviewed their find- To explore the hypothesized link between
ings suggesting that dyslexics show poorer letter string processing efficiency, reading
automatization of many skills, both linguis- skill, and RAN, Bowers (1996) devised the
tic and nonlinguistic. Nicolson and Fawcett Quick Spell Test (QST). It had three sub-
(2001) provide a framework in which cere- tests, four letter simple words (e.g., went),
bellar problems underlie both articulatory pseudowords (e.g., meft), and all-consonant
and automatizing deficits relevant to litera- illegal nonwords (e.g., dlhw) which were
cy. Stein (2001) also cites dyslexia-related presented to each child in mixed order on a
difficulties in cerebellar functioning that can computer screen for 250 ms, with the child’s
be indexed by motion detection tests. task simply to name the letters seen. There
were 10 letter strings in each subtest and
Bowers and Wolf’s (1993) argument for a number of strings correctly reported was
special relationship between orthographic scored. Bowers and colleagues (1999) found
skill and RAN was not centered on how that naming speed was a strong correlate of
general processing speed factors might be QST performance in grade 2 and grade 3
reflected in RAN but, rather, on how factors children. Although the association of RAN
associated with slow RAN would affect with accuracy of processing briefly present-
reading acquisition. Thus this position is ed letter strings was confirmed, the pattern
unaffected by the outcome of the domain- of results did not clearly implicate ortho-
general/domain-specific debate. Our hy- graphic skill as the route through which
pothesis asserted that processes reflected in RAN was associated with reading. When
RAN underlie letter recognition speed in comparing single- and double-deficit chil-
text. If letter identification proceeds too dren on these subtests, the most consistent
slowly, letter representations in words discriminator of groups was the illegal non-
would not be activated in sufficiently close words. The double-deficit children were
temporal proximity to induce sensitivity to particularly poor at reporting letters in these
commonly occurring orthographic patterns. strings; single-deficit children were interme-
In essence, Bowers and colleagues (1994) diate in their performance and no-deficit
predicted that the variance in RAN associat- children were reasonably accurate. All
ed with reading would be mediated through groups exhibited a “word superiority ef-

RAN’s Contribution to Understanding Reading Disabilities 151

fect” such that the performance was best for Both good and poor grade 4 readers made
words, next for pseudowords, and worse more correct identifications of letters in let-
for nonwords. If the original hypothesis was ter strings with high bigram frequency (e.g.,
correct, RAN deficit and DD children blbs) compared to the original low bigram
would have smaller effects of orthographic frequency letter strings (e.g., dlhw). In this
structure than other children. sample, only the original nonword strings
significantly differentiated good and poor
The data reported in Table 9.3 are from a readers, with differences between groups
subsequent study of grade 3 children (Sun- narrowed by the sensitivity to common pat-
seth & Bowers, 2002). (Some data from this terns even in nonwords.
study were reported in Table 9.2.) As indi-
cated earlier, children were divided into Summarizing the results of several studies
groups based on their strengths and weak- using the QST, Bowers (2001) reported that
nesses in phonemic awareness and naming poor reader and double-deficit groups dif-
speed, with strengths defined as above the fered most from other groups on the letter
50th percentile and weaknesses as below the strings with the least orthographic struc-
30th percentile. Again, QST nonword letter ture, and each group’s accuracy benefited
strings discriminated double-deficit from similarly from each additional increase in
single-deficit children. Although all deficit orthographic structure. Van der leij and Van
groups performed more poorly than the Daal (1999) present further evidence that
double-asset group on strings with greater dyslexics are particularly slow at processing
orthographic structure, they did not differ nonwords with low frequency clusters and
significantly from one another on these benefit from presentation of nonwords with
strings. In that study, Sunseth and Bowers higher frequency clusters. Much earlier,
also administered an embedded word test Horn and Manis (1985) showed that
devised by Hultquist (1997) as a measure of dyslexics used orthographic structure in vi-
orthographic skill. The same double-deficit sual search and lexical decision tasks as well
children showed more errors detecting as did normally reading controls. These re-
words embedded in strings of consonant let- sults were in the context of dyslexics’ lesser
ters (e.g., pjgirlwjwz) than single-deficit overall accuracy in visual search and slower
children, even when controlling for their latency in lexical decision tasks.
poorer performance reading similar isolated
words (e.g., rock). Thus DD children were This pattern of data suggested a more
more affected by the surrounding consonant complex route for RAN effects than pro-
strings. posed by the original hypothesis. Bowers
(2001) interpreted the findings to mean that
Rueffer (2000) revised the QST by adding naming speed may affect sight word skill
a list of nonwords with high bigram fre- mainly through its association with a base-
quency to assess just how sensitive children line for speed of visual letter string identifi-
were to the presence of orthographic pat- cation, upon which orthographic knowl-
terns even in illegal all-consonant strings. edge adds perceptual facilitation effects. In

TABLE 9.3. Performance on QST and Embedded Words by Subtypes of Grade 2 Children

Double Phonological Naming speed Double

asset deficit deficit deficit

Quick Spell Test: Number correct/10 9.7 (.8)a 8.1 (1.8)b 7.3 (1.1)b 6.4 (2.1)b
9.2 (1.0)a 6.2 (1.6)b 5.8 (1.3)b 4.7 (2.2)b
Word 8.8 (1.5)a 4.8 (1.7)b 4.7 (1.2)b 2.9 (1.4)c
Pseudoword
Nonword

Hultquist Embedded Word Test: % correct 95.3 (5.1)a 69.8 (13.1)b 78.5 (14.9)b 56.7 (13.1)c
99.0 (2.3)a 84.6 (11.3)b 89.2 (13.1)ab 75.3 (14.3)b
Embedded
Nonembedded

Note. Values sharing superscripts do not differ from each other. Data from Sunseth & Bowers (2002).

152 CAUSES AND BEHAVIORAL MANIFESTATIONS

our studies, this baseline is independently letter string followed by a probe) but also at
affected by PA as well as naming speed. longer ones, up to 2½ seconds. The baseline
(One might suppose that the memory com- difficulties in processing are not overcome
ponent of PA might be associated with this by just more time to inspect targets.
baseline because the strings are not pro-
nounceable.) Orthographic knowledge once We may need to interpret our findings
attained helps speed the perceptual process- about RAN-related difficulties in processing
es involved in letter recognition in strings strings of letters with low or high ortho-
with high orthographic structure, but base- graphic structure within a broader frame-
line effects persist. Acquiring orthographic work of processing speed effects on the
knowledge may be impeded by the slow amalgamation of phonological and ortho-
baseline but also by other factors. Certainly graphic codes. Breznitz (2001) focuses at-
orthographic skill requires much practice tention on the degree of asynchrony in time
with common sublexical patterns, especially between the auditory (phonological) and vi-
for double-deficit poor readers. Levy (2001) sual (orthographic) processing of print. She
showed how special efforts to make sublexi- has found that normally reading children
cal patterns visually salient improved the and adults have a natural asynchrony based
word recognition of even these poor read- on the different speed of visual and auditory
ers. information processing, as indexed by ERP
responses to appropriate stimuli. However,
The focus on RAN effects on baseline let- the “gap” is not particularly large, and pre-
ter string recognition with additive effects of sumably within a space of time that can be
orthographic knowledge coming from sev- resolved, such that connections between the
eral sources may be more consistent with two systems can be forged. However,
studies suggesting three deficits: phonologi- dyslexics have ERP latencies (both P200
cal awareness, naming speed, and ortho- and P300) that are longer to both auditory
graphic awareness (e.g., Badian, 1997). But and visual stimuli, especially the auditory
it is also unsurprising that naming speed is ones. This finding can be interpreted to
particularly associated with “sight” word mean that dyslexics take longer to per-
(orthographic) codes, as individual differ- ceive/classify and integrate into working
ences in processing strings of unrelated let- memory various types of simple stimuli,
ters forms the baseline on which ortho- even nonlinguistic material. The most sensi-
graphic knowledge speeds recognition of tive association between reading skill and
real words. speed of processing indices for children was
between reading and a “gap” score devised
Recent work by Conrad (2002) is consis- by subtracting the visual from the auditory
tent with this newer interpretation. She did ERP latencies for graphemes and phonemes.
find that double-deficit poor readers (com- Her data suggest that the link in time be-
pared to those poor readers with faster tween the visual and verbal systems rather
RAN performance, i.e., single-phonologi- than the processing time for one of them is
cal-deficit poor readers) had significantly central. Retrieving verbal labels must be in-
poorer performance on several tests of or- tegrated with visual pattern recognition. Is
thographic accuracy, replicating the it a sluggish visual system or an inefficient
RAN–orthographic skill association. How- verbal system that acts as bottleneck to the
ever, her studies also suggest the separate ef- integration? Or is the verbal–visual connec-
fects of RAN and orthographic skill on tion system itself not performing at an opti-
reading. Using a probe task at letter, letter mal rate? The efficiency of the integration
cluster, and word levels, she concluded that of two systems remains the core of what is
the DD children “have difficulty processing being measured. Work not only by Breznitz
individual letters in a string, whether or not but also by Wimmer and Mayringer (2001),
the string is orthographically regular.” Yet Berninger and colleagues (2001), and Levy
she also replicated our finding that even DD (2001) emphasizes the importance of bring-
children make use of orthographic structure ing these processes into synchrony, and the
to aid word processing. Most intriguing is penalties incurred when slow processing of
the finding that the letter processing deficit one or more elements impedes their integra-
of DD children occurs not only at relatively tion. To rapidly name items on a RAN task
brief presentation rates (one second for the

RAN’s Contribution to Understanding Reading Disabilities 153

requires the visual identification and phono- sociation of RAN with orthographic skill
logical naming systems to be suitably syn- has led to the hypothesis that RAN reflects
chronized. a baseline speed of identifying letter strings
on which orthographic knowledge builds
This perspective might resonate with the perceptual facilitation effects. Thus, we hy-
second hypothesis described by Wolf and pothesize that RAN reflects the efficient in-
Bowers (1999), which emphasizes the dif- tegration of verbal and visual information
ferent processing stages in word recognition at a fairly basic level, which in turn may be
in which speed requirements are crucial. Vi- related to the degree of asynchrony of pro-
sual naming speed as indexed by RAN can cessing speed for visual and auditory infor-
be considered to reflect the rapid integration mation.
of lexical access and retrieval processes with
lower-level visual, auditory, and motoric An Alternative View
(articulatory) processes.
The hypothesis just described is consistent
We believe that this unique combination of (a) with much data but still awaits confirma-
actual subprocesses used in reading and (b) tion. An alternative view about slow RAN
similar efficiency or processing speed require- and related deficits in reading is that they
ments needed in subprocess integration has reflect underspecified or immature phono-
made naming speed tasks one of the two best logical representations whose effects are
predictors of reading achievement (along with seen in slow naming and poor verbal short-
phonemic awareness tasks) across all lan- term memory, as well as the more obvious
guages studied to date. At the same time, the deficit in PA (Pennington, Cardoso-Martins,
multicomponential nature of naming speed Green, & Lefly, 2001). In this view, speed of
suggests that naming speed deficits could re- processing does not play a direct role in
sult from multiple, underlying sources. (Wolf reading skill but, rather, is another way in
& Bowers, 1999, p. 430) which poor phonological processing is re-
vealed. Perhaps this interpretation coincides
Berninger and Abbott (1994) also empha- with that of Wagner and colleagues (1994),
size the multiple connections between as- who describe rapid naming as one of the
pects of visible language needed for accurate phonological processing abilities. “Time
and fluent reading. Berninger and col- will tell” which view is more accurate.
leagues (2001) suggest that “the time score
for RAN reflects both the efficiency (speed) Implications for Remediation
and automaticity (direct access) of integrat-
ing the orthographic and phonological lay- Because even children with RAN deficits
ers . . .” (p. 402). Moreover, they agree that can speed perceptual processes relevant to
different individuals may be slow on RAN reading using their knowledge of ortho-
for different reasons. graphic constraints, a focus on ways to
boost that knowledge may be key to reme-
In summary, the effort to understand the diation efforts. Although the baseline for
basis for the RAN–reading relationship has perceptual identification of letter strings
led to a rather complex set of findings. Our may change only through maturational
current interpretation of those findings sug- processes, compensation for deficits in these
gest that RAN may reflect the relative ease lower-level processes is possible through the
of amalgamation of visual–orthographic effects of print exposure and decoding skill
and name retrieval processes. The degree to on orthographic knowledge. Practice with
which asynchrony of the two processes im- commonly occurring letter patterns may in-
pedes their amalgamation is reflected in deed need to be extraordinarily intense to
both RAN and reading. The search for clues overcome baseline differences associated
in the format of the RAN for the nature of with RAN. The RAVE-O (Wolf, Miller, &
the association has highlighted the pause Donnelly, 2000) program of remediation fo-
times between naming items as the aspect of cuses on increasing the fluency of the sever-
the serial list related to reading. That laten-
cy to name an isolated symbol is also associ-
ated with reading suggests that greater ac-
cess time to the symbol’s name is reflected in
the pauses. Attempts to understand the as-

154 CAUSES AND BEHAVIORAL MANIFESTATIONS

al (orthographic, phonological, and seman- Bowers, P. G. (1995). Tracing symbol naming
tic) components of reading skill and may speed’s unique contributions to reading disabili-
provide this training. By directly teaching ties over time. Reading and Writing: An Interdis-
orthographic patterns and gradually build- ciplinary Journal, 7, 189–216.
ing up the speed of access to them through
practice with timed games, RAVE-O (in Bowers, P. (1996, April). The effects of single and
conjunction with systematic phonics) has double deficits in phonemic awareness and nam-
boosted the reading performance of severely ing speed on new tests of orthographic knowl-
dyslexic children (Lovett, 2001). Levy edge. Paper presented at the annual meeting of
(2001) has shown that double-deficit chil- the Society for the Scientific Study of Reading,
dren need extra support to learn sublexical New York.
orthographic patterns. Once having learned
those patterns, they benefit from the in- Bowers, P. G. (2001). Exploration of the basis for
creased processing efficiency derived from rapid naming’s relationship to reading. In M.
orthographic knowledge. Wolf (Ed.), Dyslexia, fluency and the brain (pp.
41–63). Timonium, MD: York Press.
Interventions targeted to the particular
deficit profile of dyslexic children are rec- Bowers, P. G., Golden, J. O., Kennedy, A., &
ommended, with remediation focused on Young, A. (1994). Limits upon orthographic
fluency and/or accuracy of decoding as knowledge due to processes indexed by naming
needed. No one method is apt to address the speed. In V. W. Berninger (Ed.), The varieties of
difficulties of the variety of children with orthographic knowledge: Theoretical and devel-
RD in our schools. Careful integration of opmental issues (pp. 173–218). Dordrecht, The
remediation efforts, informed by the idea Netherlands: Kluwer Academic.
that dyslexics, especially those with double
deficits in PA and RAN, require much Bowers, P. G., & Kennedy, A. (1993). Effects of
greater support to develop and use ortho- naming speed differences on fluency of reading
graphic knowledge, may lead to more suc- after practice. Annals of the New York Academy
cessful remediation programs. of Sciences, 682, 318–320.

References Bowers, P. G., & Newby-Clark, E. (2002). The role
of naming speed within a model of reading acqui-
Badian, N. (1997). Dyslexia and the double-deficit sition. Reading and Writing: An International
hypothesis. Annals of Dyslexia, 47, 69–87. Journal, 15, 109–126.

Baker, K. (2002). Visual processes and the double- Bowers, P., Steffy, R., & Swanson, L. (1986). Nam-
deficit hypothesis for reading disabilities. Unpub- ing speed, memory and visual processing in read-
lished doctoral dissertation, University of Water- ing disability. Canadian Journal of Behavioral
loo, Waterloo, Ontario, Canada. Science, 18, 209–223.

Berninger, V., & Abbott, R. (1994). Multiple ortho- Bowers, P. G., Steffy, R., & Tate, E. (1988). Com-
graphic and phonological codes in literacy acqui- parison of the effects of IQ control methods on
sition: An evolving research program. In V. memory and naming speed predictors of reading
Berninger (Ed.), The varieties of orthographic disability. Reading Research Quarterly, 23,
knowledge I: theoretical and developmental is- 304–319.
sues (pp. 277–317). Dordrecht, The Netherlands:
Kluwer Academic. Bowers, P.G., Sunseth, K., & Golden, J. (1999). The
route between rapid naming and reading
Berninger, V., Abbott, R. D., Billingsley, F., & Nagy, progress. Scientific Studies of Reading, 3, 31–53.
W. (2001). Processes underlying timing and flu-
ency: Efficiency, automaticity, coordination and Bowers, P. G., & Swanson, L. B. (1991). Naming
morphological awareness. In M. Wolf (Ed.), speed deficits in reading disability: Multiple mea-
Dyslexia, fluency and the brain (pp 383–414). Ti- sures of a singular process. Journal of Experi-
monium, MD: York Press. mental Child Psychology, 51, 195–219.

Blachman, B. A. (1984). Relationship of rapid nam- Bowers, P. G., & Wolf, M. (1993). Theoretical links
ing ability and language analysis skills to kinder- between naming speed, precise timing mecha-
garten and first-grade reading achievement. Jour- nisms and orthographic skill in dyslexia. Reading
nal of Educational Psychology, 76, 610–622. and Writing: An Interdisciplinary Journal, 5,
69–85.
Bowers, P. (1993). Text reading and rereading: Pre-
dictors of fluency beyond word recognition. Jour- Breznitz, Z. (2001). The determinants of reading
nal of Reading Behavior, 25, 133–153. fluency: A comparison of dyslexic and average
readers. In M.Wolf (Ed.), Time, fluency and de-
velopmental dyslexia (pp. 245–276). Timonium,
MD: York Press.

Carte, E. T., Nigg, J. T., & Hinshaw, S. P. (1996).
Neuropsychological functioning, motor speed,
and language processing in boys with and with-
out ADHD. Journal of Abnormal Child Psychol-
ogy, 24, 481–498.

Carver, R. P. (1997). Reading for one second, one
minute, or one year from the perspective of read-
ing theory. Scientific Studies of Reading, 1, 3–43.

Compton, D. L., Davis, C. J., DeFries, J. C., Gayan,

RAN’s Contribution to Understanding Reading Disabilities 155

J., & Olson, R. K. (2001). Genetic and environ- status predict the course of reading development?
mental influences on reading and RAN: An Annals of Dyslexia, 51, 49–71.
overview of results from the Colorado Twin Geschwind, N. (1965a). Disconnection syndrome in
Study. In M. Wolf (Ed.), Time, fluency and devel- animals and man (Part I). Brain, 88, 237–294.
opmental dyslexia (pp. 277–303). Timonium, Geschwind, N. (1965b). Disconnection syndrome
MD: York Press. in animals and man (Part II). Brain, 88, 585–644.
Conrad, N. J. (2002). Letter processing in children Geva, E., Vaghoub-Zadeh, Z., & Schuster, B.
with naming speed deficits. Unpublished doctoral (2000). Understanding individual differences in
dissertation, McMaster University, Hamilton, word recognition skills of ESL children. Annals
Ontario, Canada. of Dyslexia, 50, 123–154.
Cornwall, A. (1992). The relationship of phonolog- Grigorenko, E. L., Wood, F. B., Meyer, M. S., Pauls,
ical awareness, rapid naming, and verbal memory J. E. D., Hart, L. A., & Pauls, D. L. (2001). Link-
to severe reading and spelling disability. Journal age studies suggest a possible locus for develop-
of Learning Disabilities, 25, 532–538. mental dyslexia on chromosome 1p. American
Cunningham, A. E., & Stanovich, K. E. (1990). As- Journal of Medical Genetics (Neuropsychiatric
sessing print exposure and orthographic process- Genetics), 105, 120–129.
ing skill in children: A quick measure of reading Hecht, S. A., Torgesen, J. K., Wagner, R. K., &
experience. Journal of Educational Psychology, Rashotte, C. A. (2001). The relations between
82, 733–740. phonological processing abilities and emerging
Cutting, L. E., & Denckla, M. B. (2001). The rela- individual differences in mathematical computa-
tionship of rapid serial naming and word reading tion skills: A longitudinal study from second to
in normally developing readers: An exploratory fifth grades. Journal of Experimental Child Psy-
model. Reading and Writing: An Interdiscipli- chology, 79, 192–227.
nary Journal, 14, 673–705. Ho, C. S-H. (2001, June). The cognitive profile and
Davis, C. J., Knopik, V. S., Olson, R. K., multiple-deficit hypothesis in Chinese develop-
Wadsworth, S. J., & DeFries, J. C. (2001). Genet- mental dyslexia. Paper presented to annual meet-
ic and environmental influences on rapid naming ings of the Society for the Scientific Study of
and reading ability: A twin study. Annals of Reading, Boulder, CO.
Dyslexia, 51, 231–248. Ho, C. S-H., & Lai, D. N.-C. (2000). Naming speed
DeJong, P. F., & Van der Leij, A. (1999). Specific deficits and phonological memory deficits in Chi-
contributions of phonological abilities to early nese developmental dyslexia. Learning and Indi-
reading acquisition: Results from a Dutch latent vidual Differences, 11, 173–186.
variable longitudinal study. Journal of Educa- Horn, C. C., & Manis, F. R. (1985). Normal and
tional Psychology, 91, 450–476. disabled readers’ use of orthographic structure in
Denckla, M. B. (1972). Color-naming defects in processing print. Journal of Reading Behavior,
dyslexic boys. Cortex, 8, 164–176. 17, 143–161.
Denckla, M. B., & Rudel, R. G. (1974). Rapid “au- Hultquist, A. M. (1997). Orthographic processing
tomatized” naming of pictured objects, colors, abilities of adolescents with dyslexia. Annals of
letters and numbers by normal children. Cortex, Dyslexia, 47, 89–109.
10, 186–202. Ishaik, G., Bowers, P., & Steffy, R. (2001, June).
Denckla, M. B., & Rudel, R. G. (1976). Rapid au- Phonological awareness tasks dissected. Poster
tomatized naming (RAN): Dyslexia differentiated presented at the annual meeting of the Society for
from other learning disabilities. Neuropsycholo- the Scientific Study of Reading, Boulder, CO.
gia, 14, 471–479. Kail, R., Hall, L. K., & Caskey, B. J. (1999). Pro-
Denckla, M. B., & Wolf, M. (in press). Rapid auto- cessing speed, exposure to print, and naming
matic naming (RAN) and Rapid alternating stim- speed. Applied Psycholinguistics, 20, 303–314.
uli naming (RAS). Austin, TX: Pro-Ed. Keen, A. G., & Lovegrove, W. J. (2000). Transient
Farmer, M. E., & Klein, R. M. (1995). The evi- deficit hypothesis and dyslexia: examination of
dence for a temporal processing deficit linked to whole-parts relationship, retinal sensitivity, and
dyslexia: A review. Psychonomic Society, 2, spatial and temporal frequencies. Vision Re-
460–493. search, 40, 705–715.
Fawcett, A. J., & Nicolson, R. I. (2001). Speed and Kirby, J. R., Parrila, R. K., & Pfeiffer, S. L. (2001,
temporal processing in dyslexia. In M. Wolf June). Naming speed and phonological aware-
(Ed.), Time, fluency and developmental dyslexia ness as predictors of reading development.
(pp. 23–40). Timonium, MD: York Press. Paper presented at the annual meeting of the So-
Felton, R. H., Naylor, C. E., & Wood, F. B. (1990). ciety for the Scientific Study of Reading, Boulder,
Neuropsychological profile of adult dyslexics. CO.
Brain and Language, 39, 485–497. Larsen, S. C., & Hammill, D. D. (1994). Test of
Felton, R. H., Wood, F. B., Brown, I. S., & Camp- Written Spelling: Third edition. Austin, TX: Pro-
bell, S. K. (1987). Separate verbal memory and Ed.
naming deficits in attention deficit disorder. Jour- Levy, B. A. (2001). Moving the bottom: Improving
nal of Learning Disabilities, 22, 3–13. reading fluency. In M. Wolf (Ed.), Time, fluency
Flowers, L., Meyer, M., Lovato, J., Felton, R., & and developmental dyslexia (357–379). Timoni-
Wood, F. (2001). Does third grade discrepancy um, MD: York Press.

156 CAUSES AND BEHAVIORAL MANIFESTATIONS

Levy, B. A., Abello, B., & Lysynchuk, L. (1997). Be- Dyslexia, fluency and the brain (pp.159–188).
ginning word recognition: Benefits of training by Timonium, MD: York Press.
segmentation and whole word methods. Scientif- Nigg, J. T., Hinshaw, S. P., Carte, E. T., & Treuting,
ic Studies of Reading, 3, 129–157. J. J. (1998). Journal of Abnormal Psychology,
107, 468–480.
Levy, B. A., Bourassa, D. C., & Horn, C. (1999). Obregon, M., & Wolf, M. (1995, April). A fine-
Fast and slow namers: Benefits of segmentation grained analysis of serial naming duration pat-
and whole word training. Journal of Experimen- terns in developmental dyslexia. Poster presented
tal Child Psychology, 73, 115–138. at the annual meeting of the Society for Research
in Child Development, Indianapolis, IN.
Lovett, M. W. (1987). A developmental approach to Olson, R. K., Kliegl, R., Davidson, B. J., & Foltz,
reading disability: Accuracy and rate criteria in G. (1985). Individual and developmental differ-
the subtyping of dyslexic children. Brain and ences in reading disability. In G. E. MacKinnon
Language, 22, 67–91. & T. G. Waller (Eds.), Reading research: Ad-
vances in theory and practice (Vol. 4, pp. 1–64).
Lovett, M. (2001, November). Reading disabilities Orlando, FL: Academic Press.
can be remediated: Lessons from research at the Pennington, B. F., Cardoso-Martins, C., Green, P.
Hospital for Sick Children. Workshop presenta- A., & Lefly, D. L. (2001). Comparing the phono-
tion at the Research Into Practice Conference of logical and double deficit hypotheses for develop-
the Learning Disabilities Association of Ontario, mental dyslexia. Reading and Writing: An Inter-
Toronto, Canada. disciplinary Journal, 14, 707–755.
Rosner, J., & Simon, D. P. (1971). The Auditory
Lovett, M. W., Steinbach, K. A., & Frijters, J. C. Analysis Test: An initial report. Journal of Learn-
(2000). Remediatiing the core deficits of develop- ing Disabilities, 4(7), 384–392.
mental reading disability: A double-deficit per- Rueffer, K. A. (2000). An examination of the fac-
spective. Journal of Learning Disabilities, 33, tors underlying the development of skilled read-
334–358. ing. Unpublished master’s thesis, University of
Waterloo, Waterloo, Ontario, Canada.
Manis, F. R., Doi, L. M., & Bhadha (2000). Nam- Scarborough, H. S., & Domgaard, R. M. (1998,
ing speed, phonological awareness and ortho- April). An exploration of the relationship be-
graphic knowledge in second graders. Journal of tween reading and rapid serial naming. Paper
Learning Disabilities, 33, 325–333. presented at the annual meeting of the Society for
the Scientific Study of Reading, San Diego, CA.
Manis, F. R., & Freedman, L. (2001). The relation- Semrud-Clikeman, M., Guy, K., & Griffin, J. D.
ship of naming speed to multiple reading mea- (2000). Rapid naming deficits in children and
sures in disabled and normal readers. In M. Wolf adolescents with reading disabilities and atten-
(Ed.), Dyslexia, fluency and the brain (pp. tion deficit hyperactivity disorder. Brain and Lan-
65–92). Timonium, MD: York Press. guage, 74, 70–83.
Spring, C., & Capps, C. (1974). Encoding speed, re-
Manis, F. R., Seidenberg, M. S., & Doi, L. M. hearsal, and probed recall of dyslexic boys. Jour-
(1999). See Dick RAN: Rapid naming and the nal of Educational Psychology, 66, 780–786.
longitudinal prediction of reading subskills in Stein, J. (2001). The neurobiology of reading diffi-
first and second graders. Scientific Studies of culties. In M. Wolf (Ed.), Dyslexia, fluency and
Reading, 3(2), 129–157. the brain (pp. 3 –22).Timonium, MD: York Press,
Sunseth, K., & Bowers, P. G. (2002). Rapid naming
Mann, V. (1984). Review: Reading skill and lan- and phonemic awareness: Contributions to read-
guage skill. Developmental Review, 4, 1–15. ing, spelling, and orthographic knowledge. Scien-
tific Studies of Reading, 6, 401–429.
Markwardt Jr., F. C. (1989). The Peabody Individ- Swanson, L. B. (1989). Analyzing naming speed-
ual Achievement Test—Revised (PIAT-R). Circle reading relationships in children. Unpublished
Pines, MN: American Guidance Service. doctoral dissertation, University of Waterloo.
Tannock, R., Martinussen, R., & Frijters, J. (2000).
McBride-Chang, C., & Manis, F. R. (1996). Struc- Naming speed performance and stimulant effects
tural invariance in the associations of naming indicate effortful, semantic processing deficits in
speed, phonological awareness, and verbal rea- attention-deficit/hyperactivity disorder. Journal
soning in good and poor readers: A test of the of Abnormal Child Psychology, 28, 237–252.
double deficit hypothesis. Reading and Writing: Torgesen, J. K., Wagner, R. K., Rashotte, C. A.,
An Interdisciplinary Journal, 8, 323–339. Burgess, S., & Hecht, S. (1997). Contributions of
phonological awareness and rapid automatic
Meyer, M. S., Wood, F. B., Hart, L. A., & Felton, R. naming ability to the growth of word-reading
H. (1998). The selective predictive values in rapid skills in second to fifth-grade children. Scientific
automatized naming within poor readers. Journal Studies of Reading, 1(2), 161–185.
of Learning Disabilities, 31, 106–117. Torgesen, J. K., Wagner, R. K., Rashotte, C. A.,
Rose, E., Lindamood, P., Conway, T., & Garvan,
Morris, R., Stuebing, K., Fletcher, J., Shaywitz, S.,
Lyon, R., Shankweiler, D., Kata, L., Francis, D.,
& Shaywitz, B. (1998). Subtypes of reading dis-
ability: A phonological core. Journal of Educa-
tional Psychology, 90, 1–27.

Neuhaus, G., Foorman, B. R., Francis, D. J., &
Carlson, C. D. (2001). Measures of information
processing in Rapid Automatized Naming (RAN)
and their relation to reading. Journal of Experi-
mental Child Psychology, 78, 359–373.

Nicolson, R. I. & Fawcett, A. (2001). Dyslexia,
learning and the cerebellum. In M. Wolf (Ed.),

RAN’s Contribution to Understanding Reading Disabilities 157

C. (1999). Preventing reading failure in young caused by slow visual processes? In M. Wolf
children with phonological processing disabili- (Ed.), Dyslexia, fluency and the brain (pp.
ties: Group and individual responses to instruc- 93–102). Timonium, MD: York Press.
tion. Journal of Educational Psychology, 91, Wimmer, H., Mayringer, H., & Landerl, K. (2000).
579–593. The double-deficit hypothesis and difficulties in
Van den Bos, K. P. (1998). IQ, phonological aware- learning to read a regular orthography. Journal of
ness, and continuous naming speed related to Educational Psychology, 92, 668–680.
Dutch children’s poor decoding performance Wolf , M. (1982). The word-retrieval process and
on two word identification tests. Dyslexia, 4, reading in children and aphasics. In K. Nelson
73–89. (Ed.), Children’s language (pp. 437–493). Hills-
Van der Leij, A., & Van Daal, V. H. P. (1999). Au- dale, NJ: Erlbaum.
tomatization aspects of dyslexia: speed limita- Wolf, M. (1986). Rapid alternating stimulus (RAS)
tions in word identification, sensitivity to increas- naming: A longitudinal study in average and im-
ing task demands, and orthographic paired readers. Brain and Language, 27, 360–
compensation. Journal of Learning Disabilities, 379.
32, 417–428. Wolf, M. (1991). Naming speed and reading: The
Wagner, R. K., & Torgesen, J. K. (1987). The nature contribution of the cognitive neurosciences.
of phonological processing and its causal role in Reading Research Quarterly, 26, 123–141.
the acquisition of reading skills. Psychological Wolf, M., Bally, H., & Morris, R. (1986). Auto-
Bulletin, 101, 192–212. maticity, retrieval processes, and reading: A lon-
Wagner, R. K., Torgesen, J. K., & Rashotte, C. A. gitudinal study in average and impaired readers.
(1994). Development of reading-related phono- Child Development, 57, 988–1000.
logical processing abilities: New evidence of bidi- Wolf, M., & Bowers, P. G. (1999). The double-
rectional causality from a latent variable longitu- deficit hypothesis for the developmental dyslexi-
dinal study. Developmental Psychology, 30, as. Journal of Educational Psychology, 91,
73–87. 415–438.
Wagner, R. K., Torgesen, J. K., & Rashotte, C. A. Wolf, M., Bowers, P. G., & Biddle, K. (2000). Nam-
(1999). Comprehensive Test of Phonological Pro- ing-speed processes, timing, and reading. A con-
cessing. Austin, TX: Pro-Ed. ceptual review. Journal of Learning Disabilities,
Wagner, R. K., Torgesen, J. K., Rashotte, C. A., 33, 387–407.
Hecht, S. A., Barker, T. A., Burgess, S. R., Don- Wolf, M., Miller, L., & Donnelly, K. (2000). Re-
ahue, J., & Garon, T. (1997). Changing causal re- trieval, Automaticity, Vocabulary Elaboration,
lations between phonological processing abilities Orthography (RAVE-O): A comprehensive, flu-
and word-level reading as children develop from ency-based reading intervention program. Jour-
beginning to fluent readers: A five-year longitudi- nal of Learning Disabilities, 33, 375–386.
nal study. Developmental Psychology, 33, Woodcock, R. W., & Johnson, M. B. (1989). Wood-
468–479. cock–Johnson Psycho-Educational Battery—
Wechsler, D. (1991). Wechsler Intelligence Scale for Revised. Allen, TX: DLM Teaching Resources.
Children—Third edition. San Antonio, TX: The Young, A. (1997, March). Relationship of phono-
Psychological Corporation. logical analysis and naming speed to training ef-
Wiederholt, J. L., & Bryant, B. R. (1992). Gray Oral fects among dyslexic readers. Paper presented at
Reading Tests, third edition. Austin, TX: Pro-Ed. the annual meeting of the Society for the Scientif-
Wimmer, H. (1993). Characteristics of developmen- ic Study of Reading, Chicago.
tal dyslexia in a regular writing system. Applied Young, A., & Bowers, P. G. (1995). Individual dif-
Psycholinguistics, 14, 1–34. ference and text difficulty determinants of read-
Wimmer, H., & Mayringer, H. (2001). Is the read- ing fluency and expressiveness. Journal of Exper-
ing–rate problem of German dyslexic children imental Child Psychology, 60, 428–454.

10

Basic Cognitive Processes
and Reading Disabilities

Linda S. Siegel

This chapter reviews the literature on the measure. This inconsistency constitutes a
normal course of the development of read- fundamental problem with the definition of
ing and also examines what happens when this critical variable. The lack of integration
reading skills fail to develop adequately in in this field is a result of the lack of clarity
children with reading disabilities. in regard to the basic operational defini-
tions. Siegel and Heaven (1986) reviewed
The chapter discusses the development of these definitional issues, but one of the most
reading by analyzing it in terms of a theoret- significant issues is the difference between
ical approach that is focused on the basic reading comprehension and word recogni-
cognitive processes. First, however, it tion. Tests of reading comprehension typi-
considers some important conceptual and cally involve the reading of text and multi-
methodological issues in this field. ple-choice questions about the text; tests of
word reading involve the reading of single
Controversies and Methodological Issues words. Reading comprehension tests are
timed; word reading tests are not. Although
A great deal of inconsistency and controver- reading comprehension may appear to be
sy exist in the research on reading and read- the fundamental aspect of reading and is
ing disabilities. Therefore, any discussion of clearly the ultimate goal of reading, the
reading and reading disabilities must start measurement of reading comprehension is a
with a clarification of some basic definition- methodologically complex issue full of pit-
al issues and assumptions. The confusion in falls. The issues in the measurement of read-
the field results from lack of clear, theoreti- ing comprehension were examined in detail
cally motivated, and consistent operational by Siegel and Heaven, Siegel and Ryan
definitions of two major constructs, reading (1989b), and Tal and Siegel (1996); but the
and reading disability. Although the ques- fundamental problem is that measures of
tion of what reading means may sound triv- reading comprehension are confounded by
ial, hundreds of tests are called reading tests, a number of other processes, such as back-
and reading is defined in a different way in ground knowledge, vocabulary, and reading
each one and hence each yields a different speed, and available tests of reading com-

158

Basic Cognitive Processes and Reading Disabilities 159

prehension usually involve not making an there will be an uninterrupted continuum
inference from the text material but merely from painfully thin to inordinately fat. It is
finding a verbatim answer in the text. In entirely arbitrary where we draw the line
contrast, tests of word recognition measure between ‘normal’ and ‘obese,’ but that does
more basic processes and responses are not not prevent obesity being a real and worry-
confounded with differences in reading ing condition, nor does it prevent research
speed, background knowledge, and test- into the causes and cures of obesity being
taking strategies. In addition, the use of both valuable and necessary” (p. 172). Ellis
reading comprehension scores as the inde- also wrote, “Therefore, to ask how preva-
pendent variable or the basis of the defini- lent dyslexia is in the general population
tion of reading disability can yield different will be as meaningful, and as meaningless,
results from the use of word recognition as asking how prevalent obesity is. The an-
scores (e.g., Siegel & Ryan, 1989a, 1989b; swer will depend entirely upon where the
Stanovich, Nathan, & Zolman, 1988). line is drawn” (p. 172). No virus, or specific
brain lesion, or biochemical disturbance,
Also, from a theoretical perspective, word has been shown to be the cause of dyslexia,
recognition is fundamental to comprehen- so it is not an illness in the traditional med-
sion (e.g., Gough & Tunmer, 1986; ical sense. Because a reading disability is an
Stanovich, 1982a, 1982b). The ability to educational problem and not a medical one,
read isolated words is highly correlated with and because it cannot be treated by any of
text comprehension (e.g., Shankweiler & the traditional medical means, professionals
Liberman, 1972). The problems of the be- are often reluctant to use the term “dyslex-
ginning reader or the disabled reader are ia.” However, it is a real condition that de-
clearly at the level of the word. Problems at serves study and treatment.
the word level interfere with the reading of
connected text (Shankweiler & Liberman, Reading problems are best conceptualized
1972). Because word decoding is critical to as a continuum with varying degrees of
comprehension and is the basic process in severity. Clearly, a problem at any level de-
reading, the discussion in this chapter con- serves attention and treatment, but the di-
centrates on the development of word viding line between a reading problem and
recognition. no problem is arbitrary. Fear and disdain of
the term “dyslexia” is common in North
Definitional Issues: A Digression America but seems less common in other
parts of the world. I can offer no empirical
Continuum versus Dichotomy evidence to support these speculations, but I
suspect that the sociopolitical context has
Another critical issue involves what consti- influenced the terminology. The egalitarian
tutes the appropriate definition of a reading philosophy and the cultural ethos of North
disability. Throughout this chapter, I use America may lead to the perception that a
the term “reading disability” instead of label, such as dyslexia, applied to a child
“dyslexia.” The terms “reading disability” may reduce access to educational opportu-
and “dyslexia” are actually synonymous, nities. Therefore, for these considerations,
but certain considerations have led to the and for those who find the term “dyslexia”
widespread avoidance of the term “dyslex- offensive, I generally use the term “reading
ia” in many parts of the world, particularly disability,” although, as far as I am con-
by, although not limited to, the educational cerned, their meaning is identical.
community. I do not understand why the
term “dyslexia” is often viewed as if it were Subtypes
a four-letter word not to be uttered in polite
company. However, I will speculate briefly. One of the issues that has been raised in the
Dyslexia is often taken to imply an illness, study of reading disability is whether or not
such as measles, when, in fact, in the words individuals with reading disabilities can be
of Ellis (1985), it is more similar to a prob- separated into subtypes. However, no reli-
lem such as obesity. As Ellis has written, able evidence supports the concept of sub-
“For people of any given age and height types and no clear subtypes have been delin-
eated (see Siegel & Heaven, 1986; Siegel,

160 CAUSES AND BEHAVIORAL MANIFESTATIONS

Levey, & Ferris, 1985; Siegel & Metsala, Often, the individual with reading dis-
1992, for a review of studies and method- abilities is defined as a person whose read-
ological issues). On the contrary, children ing score is significantly lower than would
with a reading disability show a remarkable be predicted from his or her IQ. (Individu-
homogeneity in the profiles of their cogni- als who fit this definition have traditionally
tive abilities (e.g., Siegel & Ryan, 1989b), been labeled “dyslexic.”) If an individual
and, when heterogeneity is found, it seems has a lower reading score but it is not sig-
to result from the particular definition used nificantly lower than would be predicted by
in the study. his or her IQ, the individual is not defined
as dyslexic. This definition is referred to as
Evidence indicates that the definition of the discrepancy definition. However, a
reading disability used in a study can influ- number of investigators have provided evi-
ence the conclusions made about the hetero- dence that a discrepancy between IQ and
geneity of the population. For example, reading is not necessary for an individual to
Siegel and Ryan (1989b) have shown that if have reading disabilities. For example, I
reading disability is defined as a deficit in have compared (Siegel, 1992) dyslexics, de-
word reading skills, all the children with fined as children whose reading scores were
reading problems have deficits in phonolog- low (standard scores < 90) and significantly
ical processing, working memory and short- (1 standard deviation) below their IQ
term memory, and syntactic awareness. The scores, and poor readers, whose reading
pattern is similar if a deficit in pseudo-word scores were low (standard scores < 90) but
reading skills is used as the basis for defin- not below the level predicted from their IQ.
ing reading disability. However, if reading These two groups did not differ on any
disability is defined on the basis of a deficit reading, spelling, or phonological process-
in reading comprehension, the group that ing tasks and on most language and memo-
emerges is heterogeneous and does not ry tasks, in spite of the fact that the mean
show deficits in phonological processing IQ score of the dyslexics was 25 points
and syntactic skills but does show deficits in higher than that of the poor readers. Both
working memory and short-term memory. these groups had scores on the reading,
Thus, if and when subtypes appear within spelling, phonological processing, language,
the population with reading disabilities, and memory tasks that were significantly
they may be artifacts of the definition used. below normal readers. The critical variable
was the presence or absence of a reading
IQ and Reading disability.

When issues related to reading disabilities Indeed, if the relative contributions of IQ
are examined, the question is always raised and pseudo-word reading are compared, IQ
as to the role of IQ and whether any differ- contributes little independent variance be-
ences in cognitive processes between individ- yond that contributed by pseudo-word
uals with reading disabilities and normal reading to the prediction of word reading
readers are merely a result of differences in and reading comprehension scores (Siegel,
IQ. However, no reliable evidence indicates 1993). Most of the variance is contributed
that IQ level plays a causative role in the de- by phonological processing as measured by
velopment of reading skills. On the contrary, pseudo-word reading. In summary, intelli-
evidence from a number of sources indicates gence as measured by IQ scores seems irrel-
that reading is not strongly related to intelli- evant to the definition and analysis of read-
gence as measured by IQ tests. Children with ing disability.
reading disability at all IQ levels show equal
difficulty with phonological processing tasks Definitions
such as pseudo-word reading, recognizing
the visual form of a pseudo-word, and pseu- Throughout this chapter children who have
do-word spelling (Siegel, 1988). Therefore, low scores on reading tests are called poor
the presence of a reading disability, not a readers, whether or not their reading scores
particular IQ, determines the pattern of cog- are significantly lower than shat would be
nitive strengths and weaknesses in regard to predicted by their IQ scores. Typically, a
language, memory, and phonological skills. reading score at or below the 20th or 25th

Basic Cognitive Processes and Reading Disabilities 161

percentile is used. Good or average readers languages, specifically, Stevenson, Stigler,
are defined as having scores on reading tests Lucker, Hsu, and Kitamura (1982) for Chi-
at or above the 30th, 35th or 40th per- nese and Japanese and Lindgren, De Renzi,
centile (depending on the study). For the and Richman (1985) for Italian. However,
aforementioned reasons, word reading tests, in both of these studies, deficit in reading
as opposed to reading comprehension tasks, comprehension was used as the measure of
yield the clearest definition of normal and a reading problem, and as discussed previ-
atypical reading. Comparisons between dis- ously, this definition does not address the
abled and normal readers are typically cognitive deficits that underlie severe read-
based on chronological age, and most of the ing problems, specifically phonological pro-
studies reviewed in this chapter use chrono- cessing.
logical age to make these comparisons.
However, another type of design is possible. Liberman, Liberman, Mattingly, and
This design involves what is called a read- Shankweiler (1980) outlined the complexi-
ing-level match. ties of studying the relationship between the
acquisition of reading skills and different
An alternative to studying both the devel- orthographies:
opment of reading skills and the differences
and similarities between disabled and nor- Orthographies vary considerably in the de-
mal readers is to match disabled and normal mands they make on the beginning reader.
readers on reading age, also called reading This variation has two essentially independent
level (e.g., Backman, Mamen, & Ferguson, aspects: first, the depth of the orthography, its
1984). This type of design is used in an at- relative remoteness from the phonetic repre-
tempt to identify differences between read- sentation; and second, the particular linguistic
ing disabled and normal readers that are unit—morpheme, syllable, or phoneme—that
merely consequences of differential experi- is overtly represented. A deep orthography,
ence with print. The theory underlying this like that of English, demands greater phono-
type of comparison is that children who are logical development on the reader’s part than
poor readers probably read less and there- a shallow orthography, like that of Viet-
fore do not have the same exposure to print. namese. Logographies (such as the Chinese
If so, a chronological age match confounds writing system), syllabifies (such as old Persian
differences that reflect experience with print cuneiform), and alphabetic systems (such as
and differences that reflect factors that English) demand successively increasing de-
cause reading disability. grees of linguistic awareness. (p. 146)

Basic Cognitive Processes in Reading Clearly, the consideration of other lan-
guages is important and I include evidence
Theoretical Approach from other languages when it is available,
though such evidence is meager.
I have postulated five processes that are
possibly significant in the development of Phonological processing involves a variety
reading skills in the English language of skills, but in the context of the develop-
(Siegel, 1993). The processes involve ment of reading skills, the most significant is
phonology, syntax, working memory, se- the association of sounds with letters (i.e.,
mantics, and orthography. This chapter re- the understanding of grapheme–phoneme
views the role of all these processes in the conversion rules and the exceptions to these
development of reading skills. Unfortunate- rules). This skill is the basis of decoding
ly, most of the information that is available print, and although other routes can be used
about the development of reading is based to obtain meaning from print, the phonolog-
on studies conducted with English, a lan- ical route is clearly an important one and
guage that has the highest degree of irregu- critical in the early development of reading
larity of the correspondence between letters, skills (e.g., Jorm, 1979; Stanovich, 1988a,
more properly graphemes, and phonemes, 1988b).
the sounds represented by letters and letter
combinations. Some studies have addressed Syntactic awareness, also called grammat-
the prevalence of reading problems in other ical sensitivity, refers to the ability to under-
stand the syntax of the language. This abili-
ty appears to be critical for fluent and
efficient reading of text, and it requires
making predictions about the words that

162 CAUSES AND BEHAVIORAL MANIFESTATIONS

come next in the sequence. Syntactic factors the connections that it has with other words
may influence the difficulty of reading single in a semantic network.
words, such as function words, preposi-
tions, and auxiliary verbs, which are diffi- Orthographic processing refers to the un-
cult to integrate in a semantic network. Ehri derstanding of the writing conventions of
and Wilce (1980) have shown that begin- the language in question and knowledge of
ning readers acquire information about the the correct and incorrect spellings of words.
syntactic properties of function words when All alphabetic systems include legal and ille-
they have been trained to read these words gal and more and less probable sequences of
in the context of a sentence. Therefore, the letters, and a fluent reader uses knowledge
ability to process syntax may be an impor- of these sequences to some extent. Position-
tant aspect of word learning. al constraints and probabilities that letters
will occur in certain positions are additional
Working memory refers to the retention of aspects of orthographic knowledge used by
information in short-term storage while pro- the skilled reader.
cessing incoming information and retrieving
information from long-term storage. Work- The following sections provide details of
ing memory is relevant to reading because the growth of these skills in children who
the reader must decode and/or recognize are normal readers and also in children with
words while remembering what has been reading disabilities.
read and retrieving information such as
grapheme–phoneme conversion rules. Work- Phonological Processing
ing memory may also be critical to the read-
ing of individual words, particularly in the Current theories of the development of
beginning of the acquisition of word reading reading skills in English stress that phono-
skills because the grapheme–phoneme con- logical processing is the most significant un-
version rules for each segment of the word derlying cognitive process. Stanovich
must be held in memory while the remaining (1988a, 1988b) outlined arguments for this
segments of the word are processed. Longer position. Phonological processing involves a
words, in terms of the number of syllables, variety of functions, but in the context of
place increasing demands on working mem- the development of reading skills, the most
ory. In addition, the complexity of a particu- significant is the association of sounds with
lar rule will influence the difficulty of word letters or combinations of letters. This func-
recognition because the number of possible tion is referred to as the understanding of
alternative grapheme–phoneme pronuncia- grapheme–phoneme conversion rules, and
tions may have an influence on ease or diffi- because of the irregular nature of the corre-
culty of reading a particular word. Given spondences in English, learning these rules
more alternative pronunciations, reading is a complex process. The child who is
will be slower and less accurate until the in- learning to read must map oral language
dividual items are mastered. More rules onto written language by decomposing the
might be searched and applied to the word word into phonemes and associating each
being read. For example, “c” and “g” have letter (or combination of letters) with these
multiple pronunciations at the beginning of phonemes.
English words, and, therefore, words or
pseudo-words starting with these letters may DUAL-ROUTE THEORIES
be more difficult than words or pseudo-
words beginning with other letters, especial- The development of phonological processing
ly for beginning readers. and the development of reading can be un-
derstood in the context of “dual-route” the-
Semantic processing refers to the under- ories of reading. These theories have a vari-
standing of meaning. Theoretically, word ety of manifestations, but their basic premise
meanings are coded in semantic networks is that two possible routes are involved in
and are retrieved through these networks. In gaining access to the meaning of print (e.g.,
the context of reading, semantic processing Coltheart, 1978; Forster & Chambers,
is relevant to the retrieval of words. For ex- 1973; Meyer, Schvanevelt, & Ruddy, 1974).
ample, the ease of retrieving the meaning of One of these routes involves direct lexical ac-
a word may depend, at least partially, on cess—that is, visually reading a word with-

Basic Cognitive Processes and Reading Disabilities 163

out any intermediate phonological process- MEASUREMENT OF PHONOLOGICAL
ing. The orthographic configuration of a
word is directly mapped onto an internal vi- PROCESSING SKILLS
sual store in lexical memory. The other
route, the phonological route, involves the The task of the beginning reader is to ex-
use of grapheme–phoneme conversion rules tract these grapheme–phoneme conversion
to gain lexical access to a print stimulus. rules. The alternative is simply to memorize
Grapheme–phoneme conversion rules are each word as a visual configuration and to
used to translate a graphemic code into a associate a meaning with it. This kind of
phonemic one. This route is referred to as learning may occur, but it is inefficient and
nonlexical because the application of the makes tremendous demands on visual mem-
rules does not rely on word-specific pronun- ory. In English, no one-to-one correspon-
ciations. Instead, grapheme–phoneme con- dence exists between a letter (or letters) and
version rules are presumed to be stored ex- a sound. The same letter represents different
plicitly and used to determine a word’s sounds and the same sound may be repre-
pronunciation. According to this model, sented by different letters.
pseudo-words can be read only by means of
a nonlexical route, as, by definition, a pseu- In an alphabetic language such as English,
do-word cannot have a lexical representa- the best measure of phonological processing
tion. skills is the reading of pseudo-words, that
is, pronounceable combinations of letters
Dual-route theories have been chal- that can be read by the application of
lenged. For example, the reading of non- grapheme–phoneme conversion rules, but
words is influenced by their similarity to that are, of course, not real words in Eng-
real words, and regular words that have ir- lish. Examples include pseudo-words, such
regular orthographic neighbors are read as “shum,” “laip,” and “cigbet.” Pseudo-
more slowly than regular consistent ones, words can be read by anyone who is famil-
indicating reciprocal influences of these two iar with the grapheme–phoneme conversion
routes. If pseudo-words were read only by rules of English even though they are not
grapheme–phoneme conversion rules, then real words and have not been encountered
the reading of pseudo-words should not be in print or in spoken language before.
influenced by their similarity to real words,
and regular words should not be influenced The development of the ability to read
by the characteristics of their orthographic pseudo-words has been studied extensively
neighbors. Furthermore, multiple-level (e.g., Calfee, Lindamood, & Lindamood,
models (e.g., Brown, 1987) and connection- 1973; Hogaboam & Perfetti, 1978; Siegel
ist models (e.g., Seidenberg & McClelland, & Ryan, 1988; Venezky & Johnson, 1973).
1989) that have been proposed involve a Ample evidence indicates that children with
variety of postulated units and processes dyslexia have a great deal of difficulty read-
but not two distinct routes. (For an extend- ing pseudo-words. Studies such as those of
ed discussion of these issues, see Besner, Bruck (1988), Ehri and Wilce (1983),
Twilley, McCann, & Seergobin, 1990; Snowling (1980), Siegel and Ryan (1988),
Glushko, 1979; Humphreys & Evett, 1985; and Waters, Bruck, and Seidenberg (1985)
Metsala & Siegel, 1992). However, in spite have shown that disabled readers have more
of a certain ambiguity about the validity of difficulty reading pseudo-words than do
dual-route theories, conceptualizations of normal readers matched on either chrono-
reading in terms of dual-route theory repre- logical age or reading level. For example,
sent one way of examining the development Siegel and Ryan studied the development of
of reading skills and the performance of the ability to read pseudo-words in normal
children with a reading disability. I will dis- and disabled readers ages 7 to 14 years old.
cuss tasks used to measure both these kinds By the age of 9, the normal readers were
of processing, the direct lexical access and quite proficient and performed at almost a
the use of grapheme–phoneme conversion perfect level for even the most difficult pseu-
rules and the performance of reading dis- do-words, with, in some cases, as many as
abled and normal readers on these types of three syllables. Similarly, Backman, Bruck,
tasks. Hebert, and Seidenberg (1984) showed that
10-year-olds perform as well as adults on
tasks involving the reading of pseudo-

164 CAUSES AND BEHAVIORAL MANIFESTATIONS

words. However, Siegel and Ryan found read and write (Vandervelden & Siegel,
that the performance of the children with 1995). Several general accounts of the
reading disabilities was quite different. process by which the child learns to read
These children appear to acquire these read- have been proposed. Ehri and Wilce (1983)
ing skills late in development and even chil- postulated three phases in this process. In
dren with reading disabilities at the age of phase 1, unfamiliar words become familiar
14 were performing no better than normal and the child pays attention to the compo-
readers at the age of 7. nent letters of a word. In phase 2, words
come to be recognized as wholes with delib-
To control, at least partially, for experi- erate processing of grapheme–phoneme cor-
ence with print, Siegel and Ryan (1988) respondences, and the meanings of words
used a comparison of disabled and normal are accessed automatically. In phase 3, the
readers matched on reading grade level. speed of processing increases significantly.
Even when the disabled readers and the nor- However, less skilled readers do not show
mal readers were matched on reading level this automaticity or the growth of speed in
(hence the disabled readers were consider- identifying words and nonwords.
ably older than the normal readers), the per-
formance of those with reading disabilities Harris and Coltheart (1986) proposed
on a task involving the reading of pseudo- four phases in learning to read. Initially, chil-
words was significantly poorer than that of dren learn to read a small set of words
the normal readers. through the direct access or visual route; that
is, they recognize words without sounding
Thus, difficulties with phonological pro- them out. Then children learn a small set of
cessing seem to be the fundamental problem words on which they have been instructed.
of children with reading disability, and this Then, around 5 or 6 years of age children
problem continues to adulthood. Many rely on partial cues and relate printed words
adults with a reading disability become rea- to items stored in memory. Phonological
sonably fluent readers but still have difficul- recoding occurs at the next stage and
ty reading pseudo-words or read them slow- grapheme–phoneme conversion rules are
ly (e.g., Barwick & Siegel, 1996; Bruck, used extensively. But grapheme–phoneme
1990; Shafrir & Siegel, 1994). conversion rules are inadequate for many
languages in which the correspondence be-
For children learning to read English, the tween letters and phonemes is not perfect;
learning of grapheme–phoneme conversion hence, an orthographic stage, with no
rules is a result of systematic instruction, phonological recoding of words, is the final
and the extraction of the rules is a result of stage.
repeated encounters with print. No evidence
is available as to how much of the develop- Gough and Juel (1991) also proposed a
ment of decoding skills is a result of specific series of stages by which the child learns to
instruction in the grapheme–phoneme con- read. In the first stages, the child learns to
version rules and how much is a result of pair sounds with a printed word in an asso-
experience with print. In any case, the un- ciative process. According to Gough and
derstanding of the grapheme–phoneme con- Juel, the child selects one cue from the
version rules develops rapidly in the first printed word and the response is associated
years of experience with print under normal with that one cue. To illustrate this process,
conditions. Gough reported an unpublished study in
which children 4–5 years old were asked to
DEVELOPMENTAL STAGES OF learn four words on cards. One of the cards
had a thumbprint in the lower left corner.
PHONOLOGICAL PROCESSING The children learned the word on the card
with the thumbprint much faster than those
No conclusive evidence exists as to the on the other three but often could not iden-
process by which these skills develop. Before tify the word unless the thumbprint was on
the child learns to apply phonological skills the card, and would, in the presence of the
to print, the child must develop phonological thumbprint, incorrectly label a word with
awareness skills. Phoneme awareness refers the word that had been on the card with
to the ability to segment spoken vowels into the thumbprint. Thus, the children ap-
component parts called phonemes. This abil-
ity develops reciprocally with learning to

Basic Cognitive Processes and Reading Disabilities 165

peared to be learning the word–sound asso- Sprenger-Charolles, 1991). These studies are
ciation based on the overall visual stimulus discussed in detail later in this chapter. In
without attention to individual letters. That contrast, the child using the cipher will make
is, they were learning a sound–picture asso- errors indicating a misapplication of rules.
ciation and incorrectly using part of the vi-
sual stimulus, in this case an irrelevant ele- ACQUISITION OF GRAPHEME–PHONEME
ment. In terms of the dual-route theory,
these children were apparently using the di- CONVERSION RULES
rect access or visual route but doing so in-
efficiently. Although we have evidence about the inade-
quate phonological skills of children with
Gough (in Gough & Juel, 1991) provided reading disabilities, little is known about the
an additional demonstration of this use of precise manner in which the complex
partial cues. He taught children 4–5 years grapheme–phoneme conversion rules of the
old to read four words and then determined English language are acquired. The studies
whether they could recognize a word when reported previously have involved global
half of it was hidden. Some of the children measures of pseudo-word reading. This type
could recognize the word if the first part of measure is an important first step, but in
was hidden (“du” in duck) but not if the order to understand the process of reading, a
second part was hidden, and some could more detailed analysis is needed. Venezky
recognize the word when the second part and Johnson (1973) said, “A single ‘word at-
was hidden but not the first. They appeared tack’ score has little diagnostic value, espe-
to be using only partial visual cues. cially for those children who fall in the mid-
dle ranges between mastery and complete
According to Gough and Juel (1991), in failure” (pp. 109–110). The ascertainment
the next stage the child must map spoken of the order and nature of the acquisition of
language onto printed words using a these rules is an important step in the under-
process called cryptanalysis, that is, learning standing and treatment of reading skills. A
the correspondences of sounds and letters number of investigators have begun to work
(the orthographic cipher). Gough and Juel on the problem of specifying the order of ac-
distinguished between this cipher and what quisition of these grapheme–phoneme con-
is called phonics. They characterized the version rules with the expectation that the
rules of English phonics as explicit and the rules are acquired in a relatively fixed and
cipher as a larger set of regularities that may predictable order in a manner similar to the
be learned as rules or that may be represent- way syntactic structures develop in oral lan-
ed by analogies. They asserted that the use guage (e.g., Guthrie & Seifert, 1983; Siegel
of phonics rules is a slow and laborious & Faux, 1989; Snowling, 1980). To study
process of associating each sound with a let- these issues, we showed disabled and normal
ter, holding the sound in memory, and readers words and pseudo-words that in-
blending all the individual sounds to make a volved a variety of grapheme–phoneme con-
word. version rules, such as consonant blends, r-in-
fluenced vowels, and inconsistent vowels
Gough and Juel (1991) noted that the test (Siegel & Faux, 1989). We found that com-
of mastering the cipher is the reading of plexity, as measured by the number of sylla-
pseudo-words. They obtained a correlation bles in a pseudo-word, was a significant de-
of .55 between the reading of real words and terminant of the difficulty of reading the
pseudo-words. Siegel and Ryan (1988) ob- pseudo-word. Pseudo-words with two or
tained a correlation of .86 for English and, more syllables were quite difficult for older
for Portuguese, Da Fontoura and Siegel disabled readers (11–13 years) even though
(1995) obtained a correlation of .63. Chil- normal readers had become quite proficient
dren who are “using the cipher,” in their ter- by the age of 9 to 10. Even simple vowels and
minology, will make more reading errors consonant blends were not mastered by the
that are nonwords than children who are not oldest children with reading disabilities in
using it; that is, the child not using it will be the study (ages 11–14) when they were re-
more likely to guess another word. A num- quired to read pseudo-words such as “mog,”
ber of studies have shown that children who “lun,” and “spad,” although most of the 7-
cannot read well make just these sorts of er-
rors (e.g., Johnston, 1982; Siegel, 1985;

166 CAUSES AND BEHAVIORAL MANIFESTATIONS

and 8-year-old normal readers had no diffi- pseudo-words with consonant blends at
culty with these features in words or pseudo- grade levels 2, 3, and 6. In some cases, the
words. reading disabled and normal readers did not
differ; however, these cases often resulted
In most cases, even when the disabled from floor or ceiling effects.
readers appeared to demonstrate mastery of
grapheme–phoneme conversion rules when English orthography is characterized by
they read a word, they were unable to read unpredictable correspondences between
a pseudo-word with the same rule. The graphemes and phonemes. That is, when
reading disabled experienced unusual diffi- reading a given grapheme, one often cannot
culty when reading pseudo-words. Even predict its pronunciation. Some words are
when they could read words with particular regular (e.g., “paid,” “gave,” and “heat”)
grapheme–phoneme correspondences in and can be read using the rules of pronunci-
consonant-vowel/consonant words, such as ation of their component graphemes. Other
“ran,” “wet,” and “sit,” they could not words are irregular or exceptions, and they
read pseudo-words such as “han,” “fet,” violate grapheme–phoneme conversion
and “rit,” and although they could read rules and have no rhymes with similar
words involving consonant blends, such as spelling patterns (e.g., “said,” “have,” and
“hunt,” “spot,” and “help,” they could not “great”). Words in another category also
read pseudo-words of a similar structure, have irregular grapheme–phoneme corre-
such as “lunt,” “grot,” and “melp.” spondences but also have unusual spellings
that do not occur in many other words,
This superiority of words over pseudo- such as “aisle,” “ache,” and “tongue.” Wa-
words suggests that the children with read- ters, Seidenberg, and Bruck (1984) found
ing disabilities were using some sort of di- that younger normal and poor readers were
rect lexical access which, of course, they sensitive to the effects of irregular spelling
could use when they read words but which and irregular grapheme–phoneme corre-
was not possible in the reading of pseudo- spondence and took longer to read words
words. This direct lexical access probably with these characteristics. The children also
involves processing each word as a picture showed the effects of frequency, in that the
(visual representation) rather than a series regular exception differences were greater
of letters with sounds. This visual represen- with low-frequency words, such as “pint”
tation is retrieved from long-term memory. and “wool.” Because children with reading
disabilities have poor phonological skills,
One relatively simple rule of English, with they are more likely to rely on context when
few exceptions, is that a final e in a one- reading (e.g., Bruck, 1988).
syllable word makes the vowel long. This
rule was not mastered by the oldest children Other studies have shown that poor read-
with reading disabilities in this study. That ers have difficulty with exception words
is, the older disabled readers could correctly (Manis & Morrison, 1985; Seidenberg,
read the words that reflected the rule (e.g., Bruck, Fornarolo, & Backman, 1985). How-
“like,” “cute,” and “nose”) but not the ever, still others have not revealed any differ-
comparable pseudo-words (e.g., “rike,” ence between regular and irregular words for
“fute,” and “mose”). This difficulty is quite disabled readers (Frith & Snowling, 1983;
surprising because this rule is repeatedly Seymour & Porpodos, 1980; Siegel & Ryan,
stressed in reading instruction and is nor- 1988). If regular words with regular pronun-
mally mastered early in the development of ciations are not read more easily than irregu-
reading skills. In many instances, the scores lar words, grapheme–phoneme conversion
of the children with reading disabilities were rules are apparently not being used. In addi-
significantly lower than those of normal tion, disabled readers are much less likely
readers who were matched on reading grade than normal readers to regularize the vowels
level. For example, the disabled readers had in irregular words (Seidenberg et al., 1985).
significantly lower scores than did the nor-
mal readers of the same reading age on the One set of hypotheses that has been ad-
following tasks: reading one-syllable pseu- vanced is that the development of reading
do-words at grade-level 3; two-syllable skills is accompanied by increasing reliance
pseudo-words at grade-level 4–5; multisylla- on the visual/orthographic route. At the ear-
ble pseudo-words at grade-level 6; and ly stages of acquisition, readers rely heavily

Basic Cognitive Processes and Reading Disabilities 167

on phonological information, but good Although most errors on the exception
readers learn to recognize high-frequency words involved regularizations (e.g. “come”
words automatically. Words are largely rec- pronounced as “coam”) rather than errors
ognized by direct access through the visual that were not (“come” pronounced as
route. Doctor and Coltheart (1980) found “came”), younger children made fewer regu-
that good readers relied more on phonologi- larizations than did older children and high
cal mediation when judging the meaningful- school students. However, fewer errors in-
ness of sentences. They used four types of volved giving regular inconsistent words an
meaningless: sentences that sounded cor- irregular pronunciation (e.g., “bone” read as
rect, but in print had an incorrect real word “bun” like “done”). Poor readers were not
(e.g., “I have know time”); meaningless sen- as skilled at using grapheme–phoneme con-
tences with a pseudo-word (“I have bloo version rules and had more difficulty with
time”); meaningless sentences containing orthographic patterns that had multiple pro-
real words (“I have blue time”); meaningful nunciations. Poor readers also had more dif-
sentences with a pseudohomophone (e.g., “I ficulty than normal readers with the excep-
have noe time”). The children were required tion, inconsistent, and ambiguous words and
to read these sentences and were asked tended to make fewer regularization errors.
whether the sentences made sense. Sen- Poor readers also had more difficulty with
tences that sounded correct when phonolog- pseudo-words. Under normal circumstances,
ically recoded (e.g., “I have know time” and as children get older they become more
“I have noe time”) produced more incorrect skilled at reading the irregular and unpre-
responses than did sentences that were dictable aspects of English orthography.
meaningless when phonologically recoded Poor readers, however, continue to have dif-
(e.g., “I have blue time” and “I have bloo ficulty with the orthographic features that
time”). However, the difference decreased are not predictable but do well with high-
with age, and the investigators concluded frequency regular words. This pattern of
that young readers rely on phonological en- findings is consistent with the findings by
coding and older readers rely on visual en- Doctor and Coltheart (1980) about a shift
coding through the direct route. from phonological recoding to direct visual
access.
Backman and colleagues (1984) found
that beginning readers appear to be using Seidenberg and colleagues (1985) also
the visual route for high-frequency words found that poor and disabled readers took
but they are also learning more about longer and were less accurate in reading
grapheme–phoneme conversion rules. words with homographic patterns (e.g.,
Young readers and poor readers had diffi- “one,” as in “done” and “gone”) than nor-
culty reading homographic patterns, that is mal readers. Exception words were the
orthographic patterns with multiple pro- hardest for good readers, but they read reg-
nunciations such as “ose” in “hose,” ular inconsistent, ambiguous, and regular
“lose,” and “dose.” Backman and col- words equally well. This pattern suggests
leagues showed good and poor readers reg- that they were significantly influenced by
ular words (e.g., “hope”), exception words grapheme–phoneme conversion rules be-
(“said”), regular inconsistent words, that is, cause exception words, by definition, vio-
words with regular pronunciations but with late these rules and these words were the
irregular orthographically similar neighbors most difficult to read. Poor and disabled
(e.g., “paid” and “said”), ambiguous words readers made more errors on exception, reg-
(e.g., “clown” because “own” can be pro- ular inconsistent, and ambiguous than on
nounced as in “down” or “blown”), and regular words. Manis and colleagues (1987)
pseudo-words constructed to test these or- found that children with reading disabilities
thographic features. Young normal readers had more difficulty than normal readers in a
read the regular words that were of high fre- task that required learning to associate sym-
quency quite well but made more errors on bols with words or symbols with other sym-
exception, regular inconsistent, and am- bols, particularly when the rule was incon-
biguous words. Older good readers per- sistent. This type of rule learning is
formed at a level comparable to high school analogous to the grapheme–phoneme con-
comparison subjects. version rules of English. However, the dis-

168 CAUSES AND BEHAVIORAL MANIFESTATIONS

abled and normal readers did not differ in More vowel spellings correspond to a
learning the association when no rule was particular vowel phoneme than consonant
applicable. Therefore, children with reading spellings to a particular consonantal
disabilities do not appear to have a deficit in phoneme. Consequently, misreadings of
visual memory that does not involve linguis- vowels occur more frequently than misread-
tic stimuli. ings of consonants (Fowler, Shankweiler, &
Liberman, 1979; Weber, 1970). Unlike con-
Relatively few detailed studies of the ac- sonants, which are more likely to be mis-
quisition of specific grapheme–phoneme read in the final than initial position, the
conversion rules have been conducted. position of a vowel has no effect on the
Venezky and Johnson (1973) studied the ac- probability that it will be misread. Unlike
quisition of reading the letter “c,” pro- consonant errors, vowel errors are unrelat-
nounced as “k” or “s,” and the letter “a,” ed to their target sound, that is, they are
pronounced short (ae) or long (e) using random in regard to phonetic features. Ac-
pseudo-words such as “cipe,” “acim,” and cording to Fowler, Liberman, and
“bice.” They found that for normal readers, Shankweiler (1977), vowels are less clearly
the rules for the long and short “a” ap- defined and are more subject to individual
peared early in reading acquisition, but the and dialect variation. Vowels are the foun-
rule for the “c” pronounced as “s” ap- dation of the syllable and code the prosodic
peared much later. The initial “c” as “s” features, and consonants carry the informa-
was learned more slowly than the pronunci- tion.
ation of “c” in the medial position. Venezky
and Johnson speculated that the child may English vowels have the property that
not be exposed to as many words with “ce,” their pronunciation can change depending
“ci,” and “cy” and the teaching may not on the context. An example is the rule that
emphasize the multiple pronunciations of an “e” at the end of a word usually makes
“c.” Although Venezky and Johnson did not the vowel long. The reading of vowels is
specifically test poor readers, they noted “context free” if this rule is ignored and the
that the scores on their reading task were vowel is pronounced with the short vowel
correlated with reading comprehension sound (e.g., “cape” read as “cap”), and the
scores. reading is context dependent if the rule is
followed (Fowler et al., 1979). Fowler and
VOWELS colleagues (1979) administered pseudo-
words to young normal readers and found
English vowels tend to have more complex that most of the responses to vowels were
and irregular pronunciations than English not random but were either context depen-
consonants. The grapheme–phoneme corre- dent or context free, that is, the children
spondences of English vowels are unpre- were using the possible sounds for that
dictable. At this time, the understanding of vowel. Context-dependent responses in-
the relationship between the nature of Eng- creased with increasing age, indicating an
lish vowel orthography and the develop- awareness of the context in which the possi-
ment of reading skills and problems cannot ble spellings of phonemes occur. Even the
be determined because, as Shankweiler and youngest readers, who had received only 1
Liberman (1972) have noted: year of reading instruction, could apply
their knowledge of orthographic regularities
This generalization applies to English. We do to pseudo-words.
not know how widely it may apply to other
languages. We would greatly welcome the ap- As noted earlier, disabled readers are less
pearance of cross–language studies of reading likely to regularize the vowels in irregular
acquisition, which could be of much value in words. Bryson and Werker (1989) adminis-
clarifying the relations between reading and tered a pseudo-word reading task to dis-
linguistic structure. That differences among abled readers to determine whether they
languages in orthography are related to the in- would be more likely to read vowels as con-
cidence of reading failure is often taken for text dependent. As normal readers gained
granted, but we are aware of no data that di- reading skills, they made more context-
rectly bear on this question. (p. 310) dependent responses. Some of the children
with reading disabilities (those with signifi-

Basic Cognitive Processes and Reading Disabilities 169

cantly higher performance than verbal IQ short vowel sounds. What they called spe-
scores) made more context-free responses cial rule word production, with such vowel
than age- and reading-level matched con- sounds as in “food,” “join,” and “bulk,”
trols. Some of the children with reading dis- were learned even later. Typically, the poor
abilities did not make context-free errors. readers’ mastery of these complex rules was
However, it should be noted that these chil- slower and less adequate than that of the
dren were defined on the basis of below- good readers.’
grade-level scores on a reading comprehen-
sion and/or text reading test. As noted The increased likelihood of vowel errors
earlier, children with low scores on these does not appear to be a result of inadequate
types of reading tests may not have poor perception of sounds or difficulties with
word recognition or decoding skills; there- speaking. When children were asked to re-
fore, these children may not have been read- peat the words that they had been asked to
ing disabled in the sense used in this chapter. read, Shankweiler and Liberman (1972)
found that fewer errors occurred on vowels
Bryson and Werker (1989) noted that than consonants and that the errors were
poor readers and younger normal readers, evenly distributed between the initial and fi-
when attempting to read double vowels, ei- nal positions.
ther sounded out the first letter and ignored
the second or sounded out each individual In languages other than English, vowels
letter. Often, the poor readers sounded out have more regular patterns with fewer rep-
the final silent “e,” therefore adding a resentations of each vowel sound. One such
phoneme. They appeared to be reading let- language is Hebrew, in which the ortho-
ter by letter. graphy is transparent, that is, the
grapheme–phoneme conversion rules are
Seidenberg and colleagues (1985) found predictable. Children learning to read both
that both poor readers and clinically diag- English and Hebrew can be tested to com-
nosed, probably dyslexic readers made pare these two very different orthographies.
more vowel than consonant errors. Most of In a comparison of English-speaking chil-
these errors involved the incorrect lengthen- dren learning to read Hebrew as a second
ing or shortening of the vowel. The more se- language, we (Geva & Siegel, 2000) found
verely disabled readers produced errors that that the incidence of errors in reading vow-
involved substitution of a totally different els was significantly higher in English than
vowel (e.g., “lake” for “like”); poor readers in Hebrew. Other children who had reading
produced mispronunciations of the target disabilities (in both languages) made many
vowel on the exception words; good readers vowel errors in English but few in Hebrew.
tended to regularize them (“come” pro- Younger children with reading disabilities
nounced to rhyme with “home”). The read- made vowel errors in both languages. How-
ing disabled and poor readers were less like- ever, other types of errors were more com-
ly to make these kinds of errors. Poor and mon in Hebrew. Hebrew has many visually
disabled readers were less likely to regular- similar letters and more errors were made
ize a pseudo-word that could be pro- involving visually confusable letters in He-
nounced like a regular or an exception word brew than in English. In addition, because
(e.g., “naid” that could be pronounced to Hebrew has a transparent orthography, one
rhyme with “said” or “paid”). Using pseu- can decode it syllable by syllable and pro-
do-words, Smiley, Pasquale, and Chandler nounce it properly and read the word with-
(1976) also found that poor readers made out the proper stress. Failure to read the
more errors on vowels, especially long vow- word with the stress on the correct syllable
els, than did good readers. Shankweiler and was more common in Hebrew than in Eng-
Liberman (1972) conducted detailed analy- lish. In English, a syllable-by-syllable decod-
ses of the errors that were actually made in ing would usually result in vowel errors
misreading vowels. Vowels that have many (e.g., pronouncing the vowel as a short
orthographic representations—such as /u/, vowel when the word ends in “e” and per-
which is represented by u, o, oo, ou, oe, ew, haps even pronouncing the final silent “e”).
and ie—were the most difficult to read. Order errors, in which a consonant was
omitted or the order of the consonants was
Guthrie and Seifert (1977) found that confused, were more common in English
long vowel sounds were learned later than

170 CAUSES AND BEHAVIORAL MANIFESTATIONS

than Hebrew, possibly because Hebrew for only a small portion of the errors made
words can be decoded in a linear manner in reading words in the Shankweiler and
from right to left and the linear strategy Liberman (1972) study, even though they
does not always work successfully in Eng- used lists designed to elicit these errors. Fur-
lish. thermore, sequence reversals such as “saw”
read as “was” were uncorrelated with letter
CONSONANTS reversals such as “b” read as “d.” However,
consonant errors were more common than
Consonants in English are more regular vowel errors.
than vowels in that particular consonantal
phonemes are represented in fewer ways. Werker, Bryson, and Wassenberg (1989)
Consequently, consonants are less likely to examined the reading of consonants and
be misread. Shankweiler and Liberman found that both disabled and normal read-
(1972) and Fowler and colleagues (1977) ers made more phonetic feature substitution
found that consonants in the initial position errors than orientation reversal substitu-
were more likely to be read correctly than tions. Also, children with a reading disabili-
consonants in the final position. (In the ty made more consonant addition errors.
Shankweiler and Liberman study, the posi- Most errors were not reversal errors. Al-
tions of the vowels and the particular con- though some reversals are found in young
sonants used were not counterbalanced; but children regardless of reading ability (Tay-
this methodological problem was corrected lor, Satz, & Friel, 1979; Vellutino, Steger, &
in the Fowler and colleagues study.) The Kandel, 1972), these reversal errors may be
reason for this positional effect is not clear. linguistic rather than perceptual because re-
It could result from guessing a word on the versals of orientation (“b” read as “d”) are
basis of the initial letter rather than trying not correlated with reversal of sequencing
to apply grapheme–phoneme conversion (“was”–“saw”). Reversals occur with
rules to the word because of poor reading words but not with single letters presented
ability and underdeveloped phonological tachistoscopically, and consonants are con-
skills. Fowler and colleagues noted that the fused when they differ by a single phonetic
initial segment is easiest to isolate and un- feature regardless of visual similarity. Sei-
like the final one does not require analysis denberg and colleagues (1985) found that
of the syllable. Therefore, children with in- disabled readers make more substitution er-
adequate phonological skills might be ex- rors (“belt” for “best”) and insertion errors
pected to be able to process the first conso- (“grave” for “gave”) than slow readers,
nant but not the later ones. who make more errors than normal readers.

Consonant errors were closely related to Werker and colleagues (1989) noted that
their target sound but vowel errors were Seidenberg and colleagues (1985) confound-
not. For example, “b” and “p” were more ed phonetic feature and orientation reversal
likely to be substituted for each other than substitutions by calling them both reversals
“b” and “s.” Consonants with more com- (“deed” for “beed”) and inversions (“deed”
plex orthographies (i.e., the ones that can be for “deep”). Werker and colleagues studied
represented by more than one letter), were orientation reversal errors in which one let-
more difficult, but this effect cannot explain ter was read as another differing in left/right
the initial–final consonant difference. or up/down orientation, such as “b” for
“d,” and phonetic feature errors in which
The error patterns were not the same for one letter was misread as another differing
vowels and consonants (vowel errors were in a single phonetic feature such as voicing
independent of position, consonant errors “b” versus “p” and place of articulation
were not; vowel errors were not closely re- (“b” and “d” are both voiced but “b” is bi-
lated to the target, consonant errors were). labial and “d” is alveolor). They found that
The errors evidently do not reflect visual normal and disabled readers were equally
difficulties because visual difficulties should likely to make orientation reversal errors.
not work differently with vowels and conso- All groups made more phonetic feature than
nants. In addition, visual difficulties do not orientation reversal errors. Therefore, errors
appear to be characteristic of beginning were the result of phonetic and not visual
readers. Word and letter reversals accounted similarities. The order of types of errors was

Basic Cognitive Processes and Reading Disabilities 171

as follows: phonetic > addition > omission > these words correctly. Even when matched
sequencing. The children with reading dis- with normal readers of the same reading lev-
abilities made more errors than normal el, the disabled readers made significantly
readers that involved adding a consonant. more errors than did the normal readers.
The normal readers made more phonetic Compared to the normal readers, the
feature substitutions than any other type of younger children with a reading disability
error. Disabled readers seemed to be reading were significantly less likely to use a rule-
letter by letter. The most common type of based strategy and more likely to use an
addition errors involved homorganic errors, analogy strategy. This pattern suggests a
that is, closing a syllable with the consonant greater reliance on the visual route.
sound already existing (e.g., “ap” to
“pap”). Reading disabled, not normal read- OTHER PHONOLOGICAL SKILLS
ers, made these errors. Intrasyllable addi-
tions, reading “ope” as “olpe,” were less Pseudo-word reading is not the only task
common but did occur especially among the that distinguishes poor from normal read-
disabled readers and typically involved the ers. Another task is the spelling of pseudo-
addition of the liquids, “r” and “l.” Werker words. Obviously, pseudo-words can be
and colleagues speculated that errors result spelled only by using phoneme–grapheme
from knowledge of individual letters but conversion strategies as no lexical entry ex-
that the disabled readers have trouble ists. Disabled readers had significantly low-
knowing and retrieving the rules when they er scores on a task that involved the spelling
must combine letters. In addition, they may of pseudo-words, even when the disabled
rely on articulatory information when readers were at the same reading level as
sounding out words so that they retrieve the younger normal readers (Siegel & Ryan,
pronunciation of letters that are close in 1988).
place of articulation to the target letter.
One type of evidence of phonological
Smiley and colleagues (1976) found that processing skills is the use of phonological
disabled readers made more errors on the recoding in short-term memory such that
variable consonants (e.g., “c” and “g”). The rhyming (confusable) stimuli are more diffi-
reading disabled group had particular diffi- cult to remember than nonrhyming stimuli.
culty with the “s” pronunciation of “c,” the A number of studies have shown that
“j” pronunciation of “g,” the initial “ch” younger poor readers are less disrupted by
sound, and two-syllable words ending in rhyming stimuli (e.g., Byrne & Shea, 1979;
“y.” The good readers made more plausible Mann, Liberman, & Shankweiler, 1980;
(similar to the correct answer) errors than Shankweiler, Liberman, Mark, Fowler, &
did poor readers. Fischer, 1979; Siegel & Linder, 1983). How-
ever, Johnston (1982) and Siegel and Linder
ANALOGY VERSUS RULES (1983) found that older dyslexic children do
show phonetic confusability, although their
Other kinds of tasks have been used to mea- short-term memory for letters was signifi-
sure the development of the understanding cantly poorer than that of age-matched con-
of grapheme–phoneme conversion rules. The trols. This latter finding is not surprising as
reading of pseudo-words that can be read by phonological recoding skills are likely to be
analogy or by grapheme–phoneme rules, involved in any verbal memory task and the
such as “puscle,” “fody,” and “risten,” has dyslexics’ poor verbal memory may be a
been studied (Manis, Szeszulski, Howell, & function of inadequate phonological abili-
Horn, 1986). For example, “puscle” can be ties.
pronounced as if it rhymed with “muscle” or
with the “cl” pronounced, and “fody” can Performance on a variety of phonological
be pronounced like “body” or with a long tasks distinguishes disabled from normal
“o.” Children with a reading disability had a readers. Children with reading disabilities
great deal of difficulty with these pseudo- were slower than normal readers in decid-
words. The children with reading disabilities ing whether two aurally presented words
were significantly less able than normal read- rhymed, presumably because of inadequate
ers of the same chronological age to read use of phonological recoding in memory
(Rack, 1985). Phonemic awareness, the

172 CAUSES AND BEHAVIORAL MANIFESTATIONS

ability to recognize the basic phonemic seg- essential. Even in Chinese (Cantonese), chil-
ments of the language, is obviously an im- dren with reading problems have difficulty
portant component of phonological pro- with tone and rhyme discrimination and
cessing. Difficulties with phonemic have significantly lower scores than do nor-
awareness predict subsequent reading prob- mal readers on tasks measuring these
lems (e.g., Bradley & Bryant, 1983; Mann, phonological skills (So & Siegel, 1997).
1984; Wallach & Wallach, 1976). Poor
readers also have deficits in phonological SYNTACTIC AWARENESS
production tasks, for example, naming ob-
jects represented by multisyllable words and Syntactic awareness is the ability to under-
repeating multisyllabic words and difficult stand the basic grammatical structure of the
phrases with alliteration. Pratt and Brady language in question. Siegel and Ryan
(1988) found differences between good and (1988) have investigated the development of
poor readers on the ability to segment these skills in disabled and normal readers
words into phonemes and delete sounds using an Oral Cloze task, a Sentence Cor-
from words. Good readers were more accu- rection task, and the Grammatical Closure
rate in judging the length of a word or pseu- subtest of the Illinois Test of Psycholinguis-
do-word. Good readers were more disrupt- tic Abilities. In the Oral Cloze task, a sen-
ed than poor readers by misspellings in text tence is read aloud to the child and the child
that were phonologically inappropriate is required to fill in the missing word. Ex-
(“robln” for “robin”), indicating that the amples include the following: “Jane _____
good readers were using phonological cues her sister ran up the hill”; “Betty _____ a
(Snowling & Frith, 1981). hole with her shovel”; “The girl_____ is tall
plays basketball.” In the sentence correction
Children with a reading disability also task, a sentence that is syntactically incor-
have difficulty recognizing the visual code rect is read aloud to the child, who is then
of sounds (Siegel & Ryan, 1988). In the required to correct the sentence. Examples
Gates McKillop test, children hear pseudo- include the following: “Animal are kept in
words such as “wiskate” and are asked to zoos”; “Can you read them book?”; and “It
select the correct version of the word from was very cold outside tomorrow.” In the
among four printed choices: “iskate,” Illinois Test of Psycholinguistic Abilities
“wiskay,” “wiskate,” and “whestit.” Poor Grammatic Closure subtest, the child is re-
readers had significantly lower scores than quired to supply the missing word in a sen-
normal readers on this task. Although this tence that is read aloud while the examiner
task involves skills that are relevant to points to pictures illustrating the sentence.
spelling, aspects of it are relevant to phono- For example, “Here the thief is stealing the
logical processing, including the segmenta- jewels. Here the jewels have been ______.”
tion involved in analyzing the pseudo-word In this example, the child must understand
and in decoding the alternatives. the irregular past tense of the verb “to
steal” in order to supply the correct word.
THE DEVELOPMENT OF PHONOLOGICAL SKILLS When the disabled and the normal readers
were compared on these three tasks, the
IN OTHER LANGUAGES children with a reading disability performed
at a level that was significantly lower than
We have been discussing only English up to the normal readers. More difficult tasks
this point. Children who have difficulty might have yielded differences between the
learning to read Portuguese have difficulty older dyslexics and the normal readers but
reading pseudo-words (Da Fontoura & the differences were certainly significant in
Siegel, 1995) and children learning Hebrew the elementary school years. Brittain (1970)
as a second language also have difficulty found that performance on a test of the pro-
with pseudo-words (Geva & Siegel, 2000). duction of morphology (similar to the ITPA
English is an alphabetic language with a sig- Grammatic Closure) was related to reading
nificant amount of irregularity; Chinese is a ability in grade 1 and 2 children.
morphemic orthography in which the char-
acters have meaning and in which phono- Other evidence suggests that children
logical information about pronunciation is with reading problems have difficulty with
sometimes coded in a character but is not

Basic Cognitive Processes and Reading Disabilities 173

syntactic awareness. Guthrie (1973) found lower scores on Portuguese oral cloze than
that disabled readers performed at a lower did children who were good readers of Por-
level than both chronological-age- and read- tuguese. Testing native speakers of Hebrew,
ing-level-matched normal readers on a read- Bentin, Deutsch, and Liberman (1990)
ing cloze task that measured syntax compre- found that disabled readers in Hebrew were
hension, even though the disabled readers less accurate at judging whether the syntax
had an adequate sight reading vocabulary of a sentence was correct and correcting a
to perform this task. Although reading dis- sentence with incorrect syntax. In addition,
abled children were not studied, Goldman good readers were more influenced by con-
(1976) found that the understanding of text in identifying unclear words and made
complex syntax (e.g., sentences such as more errors than disabled readers that in-
“John tells Bill to bake the cake” and “John volved substituting a syntactically correct
promises Bill to bake the cake”) was related word but one that was not the word they
to performance on a reading comprehension had heard.
test. Cromer and Wiener (1966) found that
poor readers made more errors than normal Working Memory
readers that indicated a lack of awareness of
syntax on text reading tasks. Glass and Per- Working memory is the ability to retain in-
na (1986) found that performance on an formation in short-term memory while pro-
oral-language sentence comprehension test cessing incoming information. In reading,
was poorer for children with a reading dis- working memory means the decoding or
ability than for normal readers. Willows recognizing of words or phrases while re-
and Ryan (1981) found that less skilled membering what has been read. Siegel and
readers were not as accurate as normal Ryan (1989a) studied working memory in
readers at substituting a missing word in a normal and disabled readers and dyslexics,
reading cloze procedure. Although difficul- using a task based on one developed by
ties in the processing of syntax may be an Daneman and Carpenter (1980). In the
artifact of working-memory problems, this modified version of this task, the child is
possibility is relatively unlikely as we have read aloud two, three, four, or five sentences
found that children with reading disabili- and is asked to fill in a missing word at the
ties, except at the ages of 7 to 8, are as like- end of each sentence. The child is then re-
ly to show correct verbatim recall of sen- quired to remember the missing words. Ex-
tences of varying length and grammatical amples include the following: “In the sum-
complexity (Siegel & Ryan, 1988). Byrne mer it is very _____. People go to see
(1981) has also shown that poor readers monkeys in a _____. With dinner we some-
had more difficulty than good readers only times eat bread and _____.” The child was
with certain types of syntactic structures; then required to repeat the three words that
the complexity of sentence structure, not the he or she selected in the order of presenta-
length of the sentence, was a determinant of tion of the sentences. The disabled readers
performance. performed significantly more poorly than
did the normal readers on this task, indicat-
Some evidence from other languages indi- ing significant difficulties with working
cates that children with reading difficulties memory in the disabled readers. Similar dif-
experience syntactic difficulties. Children ficulties with working memory have been
with reading problems in Chinese (Can- noted in Chinese (So & Siegel, 1997), He-
tonese) demonstrated poorer performance brew (Geva & Siegel, 2000), and Por-
in an oral cloze test involving syntactic tuguese (Da Fontoura & Siegel, 1995).
awareness of Chinese (So & Siegel, 1997).
Similar results were found for Canadian Semantic Processing
children who spoke Portuguese as a first
language, received instruction in reading in The three basic cognitive processes de-
English, and attended a Portuguese Heritage scribed previously are important for the de-
Language Program in Portuguese (Da Fon- velopment of reading skill and are signifi-
toura & Siegel, 1995). The children who cantly disrupted in disabled readers. Two
had low scores on Portuguese word and other processes, semantic and orthographic,
pseudo-word reading tests had significantly

174 CAUSES AND BEHAVIORAL MANIFESTATIONS

are also involved in reading, but children read as “mommy.” Temple, among others,
with reading disabilities do not seem to ex- argued that these errors may have been due
perience the same degree of difficulties with to chance. This explanation seems unlikely
these processes as with the preceding three. for several reasons. Normal readers do not
make these errors. The substitutions all
READING ERRORS make sense in terms of having similar
meaning and no pairings are random.
Two types of analyses indicate that the se- Given the total speaking vocabulary of
mantic processing skills of poor readers are 10,000–20,000 words of children this age,
relatively intact. One type is analysis of er- these particular errors seem unlikely to oc-
rors made in word-reading tasks and the cur by chance.
other is analysis of sentence processing. The
analysis of errors made in reading single In the one report of semantic errors in
words can reveal important information single-word reading among French-speak-
about the reading process. A number of ing children, Sprenger-Charolles (1991) ad-
studies indicate that some children with se- ministered a task in which children were re-
vere reading problems make semantic errors quired to read words or pseudo-words that
in the reading of single words. An impor- were attached to pictures. Some pictures
tant point is that these errors are made in were correctly named; others were given a
reading single words with no context cues. name related to the correct name but not
Johnston (1982) reported the case of an 18- synonymous (e.g., “limace,” slug, was writ-
year-old girl who made semantic errors such ten under a picture of a snail); and others
as “down” read as “up,” “chair” read as were given pseudo-word names that dif-
“table,” and “office” read as “occupation,” fered in a single letter from the real name
and who could not read any pseudo-words. (e.g., “falise” instead of “valise” or “pan-
I have shown that a small group of children talin” instead of “pantalon”). The children
with reading disabilities make semantic sub- were required to say whether or not the cor-
stitutions while reading single isolated rect name was attached to the picture. Se-
words (Siegel, 1985). All these children had mantic errors (e.g., “locobotive” read as
very poor, or nonexistent, phonological pro- “train,” “binyclette,” a nonword similar to
cessing skills and were unable to read a sin- the real word “bicyclette,” read as “velo”
gle pseudo-word. These types of semantic [bike]) were quite common for a group of
errors indicate that phonological processing poor readers, average age 10, but virtually
is not used at all because none of the sounds never occurred in the group of good read-
implicit in the stimulus word is produced in ers.
the response. In addition, the printed equiv-
alent of the response is not visually similar Normal readers at the earliest stages of
to the target word. However, this type of er- reading may sometimes appear to make
ror indicates that some semantic processing these semantic errors. Seymour and Elder
is occurring and that although the word is (1986) studied 4½–5½-year-old children
not being read correctly, some semantic in- who had received reading instruction that
formation is being processed. This type of emphasized a sight vocabulary and that did
error is made only in the early stages of not involve systematic instruction in
reading acquisition. Normal readers do not grapheme–phoneme conversion rules. When
appear to make this type of error. The types reading single words, these children made
of errors that normal readers typically make semantic errors such as “boat” read as
involve the substitution of a visually and/or “yacht,” “milk” read as “tea,” “little” read
phonologically similar word (e.g., “look” as as “wee.” Thus, semantic coding of words
“book,” “chicken” as “children,” and appears to be the first aspect of words to be
“away” as “way”). acquired, and semantic coding will be used
if the child lacks an understanding of
Temple (1988) reported the case of a 9- spelling–sound correspondences either be-
year-old poor reader who could not read cause these correspondences have not been
pseudo-words and who made some seman- taught or because they have not been ac-
tic substitutions when reading single words, quired because of cognitive factors, as in
such as “eye” read as “blue” and “mother” reading disability. These types of errors in-
dicate that grapheme–phoneme conversion

Basic Cognitive Processes and Reading Disabilities 175

rules are not being used at all and that the were less so. These data indicate that se-
phonological processing is virtually nonex- mantic processing is primary for reading
istent. and at the earliest stages, or with disabled
readers, semantic processing is operating
Other evidence exists of the accuracy of even when other processes are much less ef-
semantic processing in disabled readers. ficient.
Frost (1998) found that dyslexics could re-
spond as quickly and as accurately as nor- Waller (1976) studied good and poor
mal readers when required to make deci- readers and found that poor readers were as
sions about whether two words belonged to likely as good readers to remember many of
the same semantic category but were signifi- the semantic aspects of what they had read
cantly slower on a phonological task that but were less likely to remember whether a
involved making a decision about whether lexical item was singular or plural and
two orthographically dissimilar words whether a past or present tense was used.
rhymed. This pattern of errors indicates relatively in-
tact semantic processing but difficulties
SENTENCE PROCESSING with the syntactic processing.

Skills involved in processing the semantic Some evidence indicates that children
aspects of sentences appear to be adequate with reading disabilities may even be superi-
in children with a reading disability. In the or to normal readers in their use of semantic
sentence correction task described earlier, context. Frith and Snowling (1983) admin-
some of the sentences were syntactically istered a task in which reading disabled and
correct but meaningless. Examples include normal readers, matched on reading level,
the following: “There are flowers flying in were required to read sentences with homo-
the garden”; “In the summer, it snows”; and graphs (with the correct pronunciation)
“The moon is very big and bright in the such as “He had a pink bow” and “He
morning.” The reading disabled did not made a deep bow.” The performance of the
have any difficulty correcting these sen- children with reading disabilities was supe-
tences and performed at a level similar to rior to that of the normal readers, indicating
that of the normal readers. This finding that the disabled readers were better able to
contrasts with their performance on sen- make use of semantic/syntactic cues.
tences where the correction of syntax was
required. Therefore, the children with read- Orthographic Processing
ing disabilities have a deficit in the process-
ing of syntactic information, but this deficit Orthographic processing involves the
does not extend to processing of semantic awareness of the structure of the words in a
information. Lovett (1979) found that read- language. For example, in English one does
ing competence in young readers was not not find “v” at the end of a word or any
related to the ability to remember the se- words that start with “dl” or have “zxg” in
mantic aspects of what had been read. them. Olson, Kliegl, Davidson, and Foltz
Lovett required children to read short pas- (1985) have developed two tasks that pro-
sages and then to recognize whether a sen- vide a direct contrast of the visual (ortho-
tence had been in the passage when the sen- graphic) and phonological processing
tence was identical or differed slightly in routes. In the Visual task, the child is shown
semantic, syntactic, or lexical context. The a real word and a pseudo-word (e.g.,
children at all reading levels were easily able “rain”–“rane,” and “boal”–“bowl”) and
to recognize changes in the semantic con- has to select the correct spelling. In the
tent, were less able to recognize syntactic Phonological task, the child has to specify
changes, and had much more difficulty in which of two pseudo-words, presented visu-
recognizing lexical changes (e.g., “picked ally, sounds like a real word (e.g.,
up” changed to “lifted up”). Even when the “kake”–“dake” and “joap”–“joak”). Each
children were required to read material be- of these tasks is designed so that only one
tween reading the sentence and remember- process can operate. That is, in the Visual
ing it, semantic information remained avail- task both choices sound exactly the same,
able, but syntactical and lexical information so that visual memory for the orthography
of a word must be used; phonological

176 CAUSES AND BEHAVIORAL MANIFESTATIONS

processes are not helpful in this case be- level than the normal readers. Therefore, in
cause sounding out the words would pro- comparison to the data on phonological
duce the identical response to each word. processing, the orthographic processing of
For the Phonological task, recall of the visu- the reading disabled is quite good. These
al pattern would not be useful because nei- data indicate that orthographic processing
ther alternative is a correct orthographic is not as impaired in dyslexics as is phono-
pattern in the English language. However, logical processing. These data indicate that
one of the alternatives, when sounded out, semantic and orthographic processing occur
does produce an English word, although it in reading but that the use of these processes
is obviously not the correct orthographic can disrupt normal reading and cause er-
form. rors.

These tasks were administered to disabled The preceding discussion has been based
and normal readers, ages 7 to 16 years. Not on what might be called orthographic aware-
surprisingly, the disabled readers performed ness skills. Some evidence suggests that dis-
more poorly on the Phonological task than abled readers are more sensitive to the visual
age-matched and reading-level-matched aspects of printed stimuli than better readers.
normal readers and did not catch up to the For example, Steinhauser and Guthrie
normal readers until the age of 13. They (1974) found that poor readers were faster
also performed more poorly on the Visual than good readers of the same reading level
task than age-matched normal readers until on a task that involved circling individual
age 13. However, the disabled readers per- letters in a text. However, poor readers were
formed at a significantly higher level on the worse than good readers when required to
Visual task than did the reading-level- circle phonemes. A visual matching proce-
matched normal readers at reading level 2. dure can be used to circle individual letters,
This finding indicates good visual memory but phonemes probably require some
skills in the disabled readers relative to their phonological coding. These data suggest that
level of word reading. It indicates that the individuals with reading disabilities are pay-
reading disabled were paying attention to ing attention to the visual aspects of printed
the visual aspects of a word rather than the stimuli, but because of differences in phono-
phonological aspects. logical skills, they have more difficulty with
these aspects of print. Snowling (1980) also
Another aspect of the awareness of ortho- found that children with a reading disability
graphic structures is the ability to recognize were more accurate than normal readers of
legal and illegal orthographic combinations the same reading level on a task involving se-
of English letters. Siegel, Share, and Geva lecting the visual form of an aurally present-
(1995) developed a task to assess this abili- ed pseudo-word. This superiority of the
ty. Children were shown 17 pairs of pro- group with reading disabilities occurred only
nounceable pseudo-words, one containing a at the lowest reading level studied (age 7).
bigram that never occurs in an English word However, the children with reading disabili-
in a particular position and the other con- ties performed significantly more poorly
taining a bigram that occurs in English. Ex- than reading-level-matched normal readers
amples are “filv”–“filk,” “moke”–“moje,” on a task involving recognition of the audi-
“vism”–“visn,” and “powl”–“lowp.” This tory form of a visually presented pseudo-
task was administered to disabled and nor- word. Clearly, this latter task involves
mal readers, ages 7 to 16 years. phonological processing skills and the Audi-
tory to Visual task relies on visual skills that
The performance of the poor and normal are operating normally, or perhaps in a supe-
readers did not differ except at the youngest rior manner. The children with reading dis-
ages. At 7–8, the children with reading dis- abilities did not differ from normal readers
abilities made significantly more errors than in the Auditory–Auditory task in which they
normal readers of the same chronological had to judge whether two aurally presented
age, but an important point is that the chil- pseudo-words were the same or different, so
dren with reading disabilities did not per- the difficulties of poor readers were not due
form more poorly than the age-matched to problems in auditory discrimination.
normal readers at ages 9 to 16. However,
when matched on reading level, the disabled The reading disabled did not show an im-
readers performed at a significantly higher

Basic Cognitive Processes and Reading Disabilities 177

provement with age on the Visual–Visual However, the reading disabled were signifi-
task, but the normal readers did, suggesting cantly more accurate in the complex aspects
that the disabled readers did not use a of English orthography than normal readers
phonemic code in the visual matching task of the same spelling level.
and that the normal readers were probably
converting the visual stimuli to a phonemic We (Lennox & Siegel, 1993) found that
code. The normal readers performed at the the spelling errors of children who were
same level on the Visual–Visual, Audito- poor spellers were more similar visual
ry–Visual, and Visual–Auditory tasks. How- matches to the correct word than were
ever, the children with reading disabilities those of good spellers of the same spelling
performed significantly better on the Visu- age. However, the misspellings of poor
al–Visual task than on the two crossed- spellers were less phonologically accurate
modality tasks, suggesting again that the vi- than those of good spellers of the same
sual stimuli (pseudo-words) were not spelling age. These findings indicate that the
phonologically recoded. All the studies im- poor spellers were more likely to use visual
ply that the direct or visual access route is memory than phonological strategies in
relatively intact in the reading disabled, but spelling.
that the phonological route is impaired.
These results suggest that individuals
Evidence from adults with reading dis- with a reading disability may be able to
ability indicates that phonemic coding does compensate for their difficulties in phono-
not occur, at least not to the same extent as logical processing. Rack (1985) found that
in normal readers. We (Shafrir & Siegel, children with reading disabilities make use
1991) found that adults with reading dis- of an orthographic code in memory. Read-
abilities reported using a visual scanning ing disabled and normal readers, ages 8 to
strategy, rather than phonological recoding, 14, were presented four lists of words to
in reading tasks that involved matching learn. The words in a list were orthographi-
words and pseudo-words. The adults with cally similar and rhyming (e.g.,
reading disabilities who did use a phonolog- “farm”–“harm”), orthographically similar
ical recoding strategy in the word task and not rhyming (e.g., “farm”–“calm”), or-
showed significantly longer latencies than thographically dissimilar and rhyming (e.g.,
those who used a phonological recoding “farm”–“warm”), and orthographically
strategy, suggesting that the visual strategy dissimilar and not rhyming (e.g.,
may be more efficient for disabled readers. “farm”–“pond”). Whether the presentation
was visual or auditory, orthographic simi-
Evidence from spelling tasks indicates larity improved the performance of reading
that adults with reading disabilities have an disabled more than normal readers, indicat-
adequate knowledge of English orthogra- ing that the disabled readers were more sen-
phy and, in some cases, a greater degree of sitive to orthographic effects. Phonetic simi-
knowledge than do normal readers. Pen- larity did not predict recall for the disabled
nington and colleagues (1986) scored the readers but it did for the normal readers.
spelling errors of adults with reading dis- Children with reading disabilities remem-
abilities and normal reading adults accord- bered more orthographically similar targets
ing to a simple system in which any ortho- than did the normal readers and fewer
graphically illegal sequence occurred (e.g., rhyming targets, indicating that they were
“ngz” in “angziaty” for “anxiety”) and a making more use of an orthographic rather
complex system in which errors indicating a than a phonetic code. Normal readers of the
lack of knowledge of more subtle aspects of same reading age did not show this effect.
orthography were scored, for example, Children with reading disabilities took
knowing that vowel clusters can be repre- longer to say yes for rhyming pairs that
sented by one vowel (“iou” in “precious” is were orthographically dissimilar (“farm”–
the sound of /u/) or knowing that “phys” “calm”) than for those that were ortho-
occurs in many words (e.g., “physics” and graphically similar (“head”–“lead”). Read-
“physician”) and represents the same sound ing-level-matched normal readers did not
in all of them. The reading disabled and show this effect.
normal readers did not differ in the preser-
vation of simple orthographic features. However, Katz (1977) found that poor
readers were not as accurate as good read-