You are on page 1of 9

Learning Disabilities Research & Practice, 16(3), 152–160

Copyright 
C 2001, The Division for Learning Disabilities of the Council for Exceptional Children

The Double-Deficit Theory of Reading Disability Does Not Fit All


Peggy T. Ackerman, Carol A. Holloway, and Patricia L. Youngdahl
University of Arkansas for Medical Sciences
Roscoe A. Dykman
Arkansas Children’s Hospital

The double-deficit theory of reading disability (Wolf & Bowers, 1999) was examined in a
sample of 56 reading-disabled and 45 normal-reading elementary school children (aged 8 to
11). As hypothesized, the two groups differed markedly on all phonological analysis tasks
and on rapid continuous naming of digits and letters (the double deficits), but they differed
as well on orthographic tasks, attention ratings, arithmetic achievement, and all WISC-III
factors except perceptual organization. Within the reading-disabled (RD) sample, children in
the double-deficit subgroup were no more impaired in reading and spelling than those with a
single deficit in phonological analysis, and those with a single deficit in rapid naming were
no more impaired than those with neither deficit. Multiple regression analyses suggest that a
multiple causality theory of RD is more plausible than a double-deficit theory.

Numerous reports in the past decade document phonologi- child must have auditory phonological sensitivity as well as
cal impairments in children who experience difficulty learn- other skills. One of these skills is rapid naming.
ing to read and spell (see reviews by Goswami & Bryant, Our interest in studying continuous naming speed deficits
1990; Pennington, 1991; Wagner & Torgesen, 1987). Many in reading-disabled children stemmed from the pioneering
researchers consider phonological impairment to be the work of Denckla and Rudel (1976) and from the impressive
major deficit, and a few theorize it may be the only proximal body of research by Wolf, Bowers, and colleagues (Bowers,
cause of reading failure (Gough & Walsh, 1991; Stanovich, Sunseth, & Golden, 1999; Bowers & Wolf, 1993; Wolf, 1991;
1992; Tunmer & Hoover, 1992). Gough and Walsh (1991) Wolf & Bowers, 1999; Wolf, 1999; Wolf & Obregon, 1992;
state that children have to learn the cipher (i.e., that sounds 1995) and by Spring and his associates (Spring & Capps,
map onto letters) in order to become proficient readers. They 1974; Spring & Davis, 1988; Spring & Perry, 1983). Bowers
concede that children do learn to read perhaps hundreds of and Wolf (1993) were the first to propose the double-deficit
sight words well before they master the cipher, but argue that theory of reading disability. They report that children with
it is hard to acquire an adequate reading vocabulary without both phonological and rapid naming deficits are the poorest
the cipher. They further posit that word-specific knowledge readers, but those with either deficit are less proficient than
(needed to pronounce nonphonetic exception words) can be children who are fast continuous namers and phonologically
acquired only with the aid of the cipher. competent.
Perhaps the most stringent measure of “the cipher” is abil- Although some investigators (e.g., Wagner & Torgesen,
ity to pronounce nonsense words (Goswami, 1993). Indeed, in 1987) have viewed continuous naming speed as a measure
a large sample of elementary school children with reading or of phonological competence, we have found only a modest
attention disorders, we (Ackerman & Dykman, 1993) found a correlation between naming speed and phonological sensitiv-
correlation of 0.85 (p < 0.001) between number of real words ity to rhyme and alliteration (Ackerman & Dykman, 1993).
and number of nonsense words successfully read. Both real Baddeley (1986) had proposed that a slow articulation rate
and nonsense word-reading levels were best predicted by the could be the overarching determinant of both slow nam-
same set of underlying variables: age and verbal IQ (forced ing and impaired phonological sensitivity, but our findings
first), plus sensitivity to rhyme and alliteration as measured (Ackerman & Dykman, 1993) did not support this theory.
by Bradley’s (1984) auditory oddity task, continuous naming Rather, our data support Bowers and Wolf (1993) in show-
speed (Denckla & Rudel, 1976; Wolf, 1991), and an auditory ing that both naming speed and phonological skill account for
echoic memory measure (Cohen & Netley, 1981). Thus, our unique variance in word-list reading level. However, our find-
data support the Gough and Walsh position in that children ings suggest there may be more than two contributory factors.
who can read short nonsense words can read real words profi- This is also the position of Badian (1997), whose studies
ciently; that is, they “have the cipher,” to use their term. But, implicate visual matching of alphanumeric stimuli (i.e.,
our data further suggest that in order to master the cipher, the orthographic processing). Further, the studies of Meyer,
Wood, Hart, and Felton (1998) suggest that poor rapid
Requests for reprints should be sent to Peggy T. Ackerman, C.A.R.E.
naming alone is not sufficient to cause chronic poor reading.
Unit, Department of Pediatrics, Arkansas Children’s Hospital, 800 Marshall Yet if investigators choose to study rapid naming or coding
Street, Little Rock, AR 72202. or any of several other measures of processing speed, they
LEARNING DISABILITIES RESEARCH 153

will almost without exception find a significant difference consent form approved by our Internal Review Board and the
between groups of normal and disabled readers (Ackerman children freely gave their assent. Parents were compensated
& Dykman, 1996; Dykman & Ackerman, 1991; Fawcett & $50 and children $10. Meal tickets were provided for lunch
Nicholson, 1994; Tallal, Miller, & Fitch, 1993; Wolff, Ovrut, at the ACH cafeteria.
& Drake, 1990). Prior to acceptance into the project, all children were
The present study was undertaken to look at the relative given the Wechsler Intelligence Scale for Children-III
contributions of phonological analysis and rapid naming (WISC-III) and the basic reading, spelling, and numerical
to literacy acquisition in normally intelligent elementary operations subtests of the Wechsler Individual Achievement
school children. It was assumed, however, that these two Test (WIAT). Those accepted into either group achieved
abilities would not be sufficient to explain all instances of either a Full Scale or Verbal IQ of 85 or higher. The children
reading failure, and, given our past research history, we designated as normal readers (NR) had standard scores of
particularly wished to evaluate the effects of attentional 90 and higher on the WIAT reading and spelling subtests.
problems, arithmetic acquisition, and cognitive abilities. Those designated as RD had standard scores of 86 or lower
The cognitive abilities of greatest interest to us are those on one or both subtests. The Wide Range Achievement Test
tapped by the once clinically popular ACID pattern subtests Revision-3 (WRAT-3) was given to obtain confirmatory data.
of the Wechsler Intelligence Scale for Children (Arithmetic,
Coding, Information, Digit Span). These subtests challenge
short-term, long-term, and working memory, as well as
speed of mental processing, and are frequently depressed in Behavioral Data
learning-disabled children (Ackerman, Dykman, & Peters,
1977; Dykman & Ackerman, 1991; Swartz, 1974). Subjects were not excluded for a known or suspected diag-
We hypothesized that our samples of normal and disabled nosis of Attention Deficit Hyperactivity Disorder (ADHD).
readers would be robustly separated on measures of phono- However, any child taking stimulant medication was required
logical analysis, naming speed, attention, verbal skills, short- to omit the medication on the day of testing. The accompany-
term and working memory, and arithmetic achievement. We ing parent was asked to complete the Child Behavior Check-
also hypothesized that children with multiple deficits would list (Achenbach, 1991) and an ADHD questionnaire adapted
be the most retarded readers. As a corollary we assumed from DSM-IV. The questionnaire listed 18 symptoms, with
that children will vary in degree of impairment on under- severity of each to be rated on a 4-point scale (0 = not a
lying abilities and that multiple regression analyses would problem, 1 = just a little, 2 = pretty much, 3 = very much).
therefore explain literacy skill differences better than catego- The first 9 items assessed problems with attention, the next 5
rization. assessed overactivity, and the last 4 impulsivity.
Finally, we wished to see whether there are orthographic
skills that are separable from phonological skills and thus
might identify children who are poor readers and spellers Phonological Awareness Battery
despite phonological analysis strengths.
1. An abbreviated (13 item) form of the Test of Audi-
tory Analysis Skills (TAAS) (Rosner & Simon, 1971),
recently used by other investigators (Badian, 1996;
METHOD 1997; Fawcett & Nicolson, 1995). Each word is pro-
nounced by the examiner who then instructs the child
Participants to repeat the word, then to repeat it again but to omit
a specified sound. On the first of two demonstration
Two groups of elementary school children (aged 8 to 11) trials, the examiner says “cowboy” and then asks the
served as participants. The reading-disabled (RD) group child to repeat the word and then repeat the word with-
(N = 56) included 23 Caucasian boys, 13 African-American out “boy.” Then follows “steamboat.” Test words are
boys, 17 Caucasian girls, and 3 African-American girls. The arranged in order of difficulty, and testing is discon-
normal-reading (NR) comparison group (N = 45) included tinued after 4 consecutive errors. The examiner pro-
25 Caucasian boys, 2 African-American boys, 13 Caucasian nounces the specific sound to be omitted and not the
girls, and 5 African-American girls. letter name(s). For example, item 4 is “coat” with the
The reading-disabled children were referred to the project “kuh” sound omitted. Badian (1997) reported a test–
from several sources: The Child Psychiatry and the Dennis retest reliability coefficient of 0.84 for this task.
Developmental Clinics at Arkansas Children’s Hospital 2. Bradley’s (1984) Auditory Sound Categorization Test.
(ACH), local child psychologists, and local schools. Letters The Bradley Test consists of 24 series of 4 words
describing the project were mailed to all these sources and each, wherein 1 word does not sound like the other
referrals were invited. The control children were recruited 3. Sixteen of the trials involve rhyme judgment and 8
via advertisements placed on bulletin boards at ACH. require detection of alliteration. Normal readers have
All children considered for selection had to be in good near perfect scores on this task by age 8 or 9 (Bradley,
health with no limiting physical disabilities, and they had 1984).
to have had normal schooling opportunities. Additionally, 3. Pig-Latin Test (Olson, Wise, Conners, & Rack, 1989).
all had to speak English as a first language. Parents signed a The subject is asked to segment the initial phoneme
154 ACKERMAN ET AL.: DOUBLE-DEFICIT THEORY

from spoken words, place the phoneme at the end Rapid Automatized Naming (RAN)
of the word, add the sound “ay,” and pronounce the
result. Thus, pig becomes ig-pay; dog becomes og- Three types of stimuli were presented to the children for
day; bird becomes ird-bay; etc. Following 5 training rapid naming: digits only, letters only, and alternating digits
trials, wherein the examiner demonstrated what was and letters. The children saw 50 stimuli (5 columns and 10
to be done to each word, 20 such 1-syllable words rows) on each 5- by 7-inch card. They were asked to read
were presented. The examiner provided the correct the stimuli aloud (row after row) as rapidly as possible.
answer if the child erred. Testing was discontinued if The children were given the numbers card first, the letters
the child erred 5 times in a row. second, and the alternating list last. The examiner used a
4. Rhyme Fluency Test. Fluency was assessed with a stopwatch to time each trial.
rhyme-generation task (Olson et al., 1989). Four stim-
ulus words (eel, ate, cat, kite) were used and the chil-
dren were asked to name as many words as they could RESULTS
that rhymed with each (1 minute per trial). The exam-
iner illustrated by giving rhymes to and. The reading-disabled (RD) and normal-reading (NR) groups
were well matched for age. As is usually the case in referred
RD samples, male children outnumbered females almost 2
Phonological Decoding to 1. The ratio of Caucasian to African-American children
within the RD group was 2.5 to 1, which is reflective of the
The 1- and 2-syllable pseudoword lists of the Decoding Skills regional population ratio for school-age children. The gender
Test (Richardson & DiBenedetto, 1985) were given to each and race composition of the NR group is similar to that of
child. The child sees a series of cards with 5 pseudowords on the RD group except that African-American boys comprise
each card and is asked to pronounce each item. There are 30 a higher percentage of the RD than NR group (23% versus
1-syllable and 30 2-syllable items. Testing is discontinued on 4%).
each list when the child makes 5 consecutive errors. Preliminary analyses (ANOVAs) of selection variables
evaluated possible gender and race differences and inter-
actions within the group. No significant interactions were
Orthographic Skills found. African-American (AA) children scored significantly
lower on WISC-III Verbal, Performance, and Full Scale IQs
Four tests assessed the children’s recognition of written or- (F1, 93 = 16.46, 6.98, and 16.76, respectively, p < 0.01 for
thographic patterns. each). Mean differences for the three IQs were 12.6, 8.3,
Part 1 of a written test of rhyme recognition utilized or- and 11.4 points. Boys had significantly higher Performance
thographically similar (regular) word-word or word-nonword IQs than girls (F1, 93 = 4.12, p = 0.045): 102.2 (12.0) versus
pairs (for example cat, hat; fat, zat), 20 of each type. Foils 98.7 (12.7). Reading and spelling standard scores from the
were orthographically similar but nonrhyming words (e.g., WIAT and WRAT-3 were somewhat lower for AA children
cat, car) and nonwords (fat, fav), 20 of each type. The but differences were not significant. There were no gender
completely randomized list was presented to the child and he differences on the reading and spelling tests.
or she was asked to highlight the pairs that rhymed. In Part IQ and achievement test group differences (with gender
II of the rhyme-recognition test, the word-word pairs were and race ignored) are given in Table 1. Since the groups were
orthographically dissimilar; 20 rhymed (e.g., blue, new) and selected to differ in reading and spelling, the large mean dif-
20 did not (e.g., rows, hole). The children were given as ferences are as expected. Given a substantial correlation in the
much time as they needed to complete these tests. WIAT and WRAT-3 standardization samples between arith-
Another task assessed spelling recognition. Ten graded metic and reading/spelling scores, it is predictable that the
lists of 10 stimulus trials were used. The lists sampled from NR and RD groups would also differ in arithmetic scores.
preprimer, primer, and 1st- to 7th-grade master lists. The Three of the four WISC-III factor scores and the ACID pat-
child was asked to highlight the real word in each trial. One tern (Arithmetic, Coding, Information, Digit Span; Swartz,
foil was a phonologically legitimate spelling of the real word 1974) separated the groups.
(a pseudohomonym) and the other foil was not. For example, Table 2 presents behavioral rating scores for the two
a trial from the 1st-grade list was room, rume, ruom; a trial groups. Given the high level of comorbidity of ADHD and
from the 4th-grade list was blays, blaze, blais. Testing was learning disabilities (Dykman & Ackerman, 1991), it was to
discontinued at the level where the child got fewer than 5 be expected that the RD group would have more adverse rat-
correct answers. ings on both the Child Behavior Checklist and the ADHD
A second spelling test assessed the child’s ability to spell rating scale. In mixed model ANOVAs of these behavioral
irregular words relative to regular words at his or her spelling ratings, race and gender did not emerge as significant fac-
level ability. The word lists of the Boder-Jarrico Test (1982) tors.
were used here. If a child’s spelling grade level on the As predicted, all the oral tests in the phonological skills
WIAT was grade 2, for example, we asked the child to battery significantly separated the two groups (see Table 3).
write the spellings of 10 words from the grade 2 reading Also, as expected, the Decoding Skills Test, requiring cor-
list; 5 were phonetically regular words and 5 were irregular rect phonological pronunciation of pseudowords, provided
words. the most robust group differences.
LEARNING DISABILITIES RESEARCH 155
TABLE 1 TABLE 3
Demographic and Selection Measures: Group Means and Standard Phonological Skills: Group Means and Standard Deviations
Deviations (SD)
Reading Disabled Normal Reading
Reading Disabled Normal Reading
Mean SD Mean SD t(99df)
(N = 56 ) (N = 45 )
Test of Auditory 3.93 1.92 1.96 1.73 5.37∗∗∗
Measure Mean SD Mean SD t(99df)
Analysis (errors)
u Bradley Sound 5.50 4.22 1.69 1.98 5.99∗∗∗
Age (months) 116.6 8.3 115.8 8.1 0.48
WISC-III Categorization
Verbal IQ 94.3 10.3 106.9 12.3 5.58∗∗∗ (errors)
u Pig Latin (correct) 10.86 7.51 16.73 5.69 4.47∗∗∗
Performance IQ 97.6 11.0 105.0 12.8 3.15∗∗
Full Scale IQ 95.4 9.3 106.3 11.9 5.18∗∗∗ Rhyme Generation 15.88 7.32 22.60 8.74 4.21∗∗∗
(correct)
Verbal 95.3 10.7 106.8 12.5 4.97∗∗∗
Decoding Skills I 8.33 5.83 21.89 5.67 11.76∗∗∗
Comprehension
Perceptual 99.4 11.9 103.9 14.2 1.73 (correct)
Organization
u Decoding Skills II 3.66 3.77 16.53 7.26 10.78∗∗∗
(correct)
u Freedom from 90.4 9.9 103.5 12.4 5.75∗∗∗
Distractibility Note. u = unequal variances, ∗∗∗ p < 0.001.
Perceptual Speed 98.6 12.0 108.2 11.8 3.98∗∗∗
ACID pattern 33.1 6.2 43.5 6.8 8.06∗∗∗
WIAT
TABLE 4
u Reading ss 79.7 7.0 103.6 12.4 11.51∗∗∗
Orthographic Skills: Group Means and Standard Deviations (SD)
u Spelling ss 79.5 7.3 101.8 14.1 10.20∗∗∗
Arithmetic ss 92.4 10.4 102.2 11.2 4.56∗∗∗
WRAT Reading Disabled Normal Reading
u Reading ss 76.9 9.6 101.3 12.8 10.63∗∗∗ Mean SD Mean SD t(99df)
u Spelling ss 78.7 7.2 100.3 14.5 9.13∗∗∗
Arithmetic ss 90.2 11.8 102.5 10.2 5.53∗∗∗
u Rhymes O+ (errors) 7.30 9.44 4.27 5.86 1.98∗
Rhymes O− (errors) 12.29 4.00 4.47 3.91 9.86∗∗∗
Note. SS = standard u = unequal ∗ p < 0.05,
scores, variances, u Spelling Detection 44.20 22.71 78.53 14.86 9.14∗∗∗
∗∗ p < 0.01, ∗∗∗ p < 0.001.
(correct)
Spelling, Regular 3.14 1.15 3.80 1.08 2.93∗∗
(correct)
The tests of orthographic skills likewise significantly sep- Spelling, Irregular 1.41 1.44 2.67 1.22 4.63∗∗∗
arated the groups (see Table 4), as did the Rapid Automatized (correct)
Naming tasks (Table 5). Note. O+ = orthographically similar, O− = orthographically dissimi-
lar, u = unequal variances, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001.

RD Subgroup Analyses
formance on the RAN tests to define a naming speed deficit.
A primary reason for this study was to compare the Error scores of 4 or higher on the TAAS, 5 or higher on the
reading spelling performance of RD children with single- Bradley, and 8 or more on the 20 Pig Latin items are con-
versus double-deficit profiles. To that end, we created four sidered suspect for children in the age range here studied,
subgroups using performance on the oral TAAS, Bradley, judging from prior studies comparing normal and disabled
and Pig Latin tests to define a phonological deficit and per- readers (Badian, 1997; Bradley, 1984; Olson et al., 1989).
Thus, we elected to classify a subject as phonologically im-
paired if his or her error scores on two of the three tests
TABLE 2 reached these cut-scores. Rhyme fluency was not used in this
Behavioral Data: Group Means and Standard Deviations classification algorithm since it does not as directly assess the
ability to break words into component sounds.
Reading Disabled Normal Reading
Child Behavior
Checklist Mean SD Mean SD t
TABLE 5
Total ss 59.8 11.5 50.8 11.0 4.03∗∗∗ Rapid Automatized Naming
Internalizing ss 57.7 12.1 52.9 11.6 2.01∗
u Externalizing ss 56.3 13.1 48.6 10.3 3.31∗∗∗ Reading Disabled Normal Reading
ADHD Mean SD Mean SD t(99df)
Attention sum 15.8 6.3 8.5 7.4 5.25∗∗∗
Hyperactivity & u Digits(sec) 35.6 10.4 24.8 5.9 6.57∗∗∗
Impulsivity sum 11.7 8.2 6.0 5.8 4.11∗∗∗ Letters (sec) 40.5 10.5 29.5 7.5 5.98∗∗∗
u Alternating (sec) 50.4 15.8 35.7 9.3 5.85∗∗∗
Note. ss = standard scores, u = unequal variances, ADHD = Attention
Deficit Hyperactivity Disorder, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. Note. u = unequal variances, sec = seconds, ∗∗∗ p < 0.001.
156 ACKERMAN ET AL.: DOUBLE-DEFICIT THEORY

TABLE 6
Phonological and Rapid Naming Values for RD Subgroups

Neither Slow Phono Both


N = 18 N = 13 N =9 N = 16 F3,52 p Pairwise

TAAS errors 2.7 2.8 5.7 5.3 18.44 <0.000 1 = 2<3 = 4


Bradley errors 2.6 3.3 9.7 8.2 17.82 <0.000 1 = 2<3 = 4
Pig Latin errors 3.9 7.0 13.3 14.4 10.09 <0.000 1 = 2<3 = 4
RAN time sum 102.9 141.8 101.0 155.2 17.23 <0.000 1 = 3<2 = 4
Rhyme sum 19.4 19.0 11.6 11.8 6.42 0.001 1 = 2>3 = 4
Note. RD = reading disabled, TAAS = Test of Auditory Analysis Skills, RAN = Rapid Automatized Naming.

The child was classified as a slow namer if the sum of his The subgroups did not differ on other WISC-III factors,
or her times for the three RAN tasks exceeded 116 seconds or on CBCL or ADHD ratings. Nor did they differ on the
(see Badian, 1997). These cut-score decisions are somewhat orthographic tasks listed in Table 4.
arbitrary but gave the sharpest discrimination of the RD and
NR groups and provided four RD subgroups with reasonable
numbers for analyses: 18 had neither deficit, 16 had both, 13 Regression Analyses
were slow only, and 9 were phonologically impaired only.
The four subgroups did not differ in age but the The above subgroup analyses did not reveal the hypothesized
racial makeup varied. The “neither deficit” subgroup had incremental deficiencies in reading and spelling scores at-
no AA children, and these children were overrepresented tributable to dual underlying deficits in phonological analysis
in the “phonological only” and “both” subgroups (chi- and naming speed. The “slow only” subgroup was no more
square = 12.01, p < 0.01 for 3 df). impaired than the “neither” subgroup, and the dual-deficit
Table 6 shows how widely the subgroups were separated (“both”) subgroup was no more deficient than the “phono-
by the cut-score algorithm. Of interest, 3 children in the NR logical only” subgroup. Still, in the sample as a whole both
group would be classified as “slow only” by this algorithm, phonological and naming deficits as well as other factors sep-
4 as “phonological only,” and 1 would be in the “both” sub- arated the NR and RD groups. Thus, we opted to use stepwise
group. regression analyses to better understand the data.
Reading and spelling standard scores for the four RD sub- In order to limit the number of independent variables
groups appear in Table 7. In pairwise contrasts, the “slow entering into these analyses, we computed a single factor
only” and “neither” subgroups did not differ on any of the score for the phonological variables, which were all highly
literacy measures even though the “neither” subgroup had intercorrelated, and we computed a single RAN measure
higher WISC-III Verbal Comprehension and ACID scores (the sum of seconds on the three subtests). In addition to
than the “slow only” subgroup. The “neither” subgroup had the WISC-III Verbal Comprehension factor scores, we in-
higher scores on all measures than the “phonological only” vestigated the ACID factor (arithmetic, coding, information,
and “both” subgroups. The “slow only” subgroup performed and digit span) because of its discriminating power in earlier
better than the “phonological only” and “both” subgroups on studies of learning-disabled children (Ackerman et al., 1977).
the reading but not spelling tests. The greatest separation of Other independent variables included the ADD index and
the subgroups was on the Decoding Skills Test where the WRAT-3 arithmetic standard scores. Race and gender were
“phonological only” and “both” subgroups were far worse included in initial exploratory analyses but were not found
than the other two. to explain any significant variance. The same was the case

TABLE 7
Differences Between RD Subgroups

Neither Slow Phono Both


N = 18 N = 13 N =9 N = 16 F3,52 p Pairwise

WISC Verbal Factor 101.9 93.6 87.9 93.6 4.81 0.005 1>2=3=4
ACID Pattern 36.1 31.1 32.4 31.7 2.39 0.080 1 > 2,4
WIAT Read ss 82.9 81.5 76.8 76.2 3.84 0.015 1=2>3=4
WIAT Spell ss 83.3 80.4 75.8 76.7 3.76 0.016 1 > 3,4; 1 = 2
WRAT Read ss 81.8 80.6 72.8 70.6 6.52 0.001 1=2>3=4
WRAT Spell ss 81.4 80.2 75.2 76.3 2.53 0.067 1 > 3,4; 1 = 2
Literacy Sum 329.4 322.7 300.6 299.9 5.81 0.002 1=2>3=4
Decoding Skills 18.5 16.4 3.8 5.8 19.96 <0.000 1=2>3=4
Note. WISC = Wechsler Intelligence Test for Children, ACID = WISC Arithmetic + Coding + Information + Digit Span, WIAT = Wechsler Individual
Achievement Test, WRAT = Wide Range Achievement Test, ss = Standard Scores.
LEARNING DISABILITIES RESEARCH 157
TABLE 8
Correlation Matrix and Factor Loadings

1 2 3 4 5 6 7 8 Loadings on Factor I

1. Literacy Sum — 0.923


2. Decoding Skills Sum 0.886 — 0.864
3. ADD Index −0.576 −0.490 — −0.649
4. ACID factor 0.755 0.650 −0.538 — 0.892
5. WISC Verbal Factor 0.605 0.567 −0.306 0.756 — 0.742
6. Phonological Factor −0.703 −0.777 0.406 −0.606 −0.574 — −0.801
7. RAN Sum −0.594 −0.528 0.433 −0.560 −0.345 0.519 — 0.686
8. WRAT Arithmetic 0.671 0.505 −0.459 0.760 0.517 −0.451 −0.429 — 0.758
Note. ADD = Attention Deficit Disorder, ACID = WISC Arithmetic + Coding + Information + Digit Span, WISC = Wechsler Intelligence Scale for Chil-
dren, RAN = Rapid Automatized Naming, WRAT = Wide Range Achievement Test. All correlation coefficients are significant at p < 0.01 or higher.

for hyperactivity ratings and internalizing and externalizing p < 0.01). An obvious initial question is how well the
scores from the CBCL. For dependent measures we used the independent variables predict the DST scores. Table 9 sum-
sum of correct responses on the Decoding Skills Tests and marizes these analyses. Only the phonological factor scores
the sum of standard scores on the WIAT and WRAT-3 read- and the ACID factor scores were accepted to achieve a robust
ing and spelling tests (labeled literacy sum). Intercorrelations R of 0.81 (p < 0.000). If the ACID factor is not included in
of these dependent and independent variables are given in the regression for DST, the ADD and Verbal Comprehension
Table 8. Principal components factor analysis of these vari- factors are accepted along with the phonological factor to
ables yielded only one factor with an eigenvalue > 1.0. Load- yield a virtually identical R (0.810). If the ADD index is not
ings are shown in the last column. included, the WRAT-3 arithmetic score is chosen in Step 2
The first set of regression analyses used the entire sample. to yield an R of 0.792. The RAN variable enters only if the
Decoding Skills Test (DST) scores and reading/spelling ACID, ADD, and arithmetic measures are excluded. Then,
scores are generally robustly correlated (Ackerman & it is taken third after the phonological factor and Verbal
Dykman, 1993), which was the case here (R = 0.886, Comprehension to yield an R of 0.799.
Next the literacy sum scores were predicted, excluding
TABLE 9 DST scores. In addition to ACID and phonological factor
Regression Analyses: Entire Sample scores, ADD and WRAT arithmetic scores were accepted to
yield an R of 0.844 (p < 0.000). If ACID is not allowed to en-
Dependent R R2 Final Beta Weight ter, the solution is as shown for Literacy Sum II. When DST is
allowed to enter in the prediction of literacy sum (see example
I. Decoding Skills Sum III), the phonological factor drops out but WRAT arithmetic
Step 1. Phonological factor 0.777 0.603 −0.605 and ADD remain along with RAN to yield an R of 0.931
Step 2. ACID factor 0.809 0.654 0.284 (p < 0.000). If only the phonological and RAN variables
II. Decoding Skills Sum
are allowed to enter, R is only 0.753, with the phonological
Step 1. Phonological factor 0.777 0.603 −0.606
Step 2. ADD index 0.800 0.639 −0.195
factor explaining 49% of the variance and RAN adding 7%.
Step 3. Verbal comprehension 0.810 0.656 0.160 A second set of multiple regression analyses was limited
I. Literacy Sum to the RD group. At issue here is the degree of reading/
Step 1. ACID factor 0.755 0.571 0.271
Step 2. Phonological factor 0.816 0.666 −0.368 TABLE 10
Step 3. ADD index 0.832 0.693 −0.181 Regression Analyses: RD Group
Step 4. WRAT arithmetic 0.844 0.713 0.216
II. Literacy Sum Dependent R R2 Final Beta Weight
Step 1. Phonological factor 0.703 0.494 −0.317
Step 2. WRAT arithmetic 0.807 0.652 0.285 I. Decoding Skills
Step 3. ADD index 0.830 0.689 −0.195 Step 1. Phonological Factor 0.773 0.597 −0.773
Step 4. RAN sum 0.841 0.707 −0.169 I. Literacy Sum
Step 5. Verbal comprehension 0.850 0.722 0.157 Step 1. Phonological Factor 0.615 0.378 −0.535
III. Literacy Sum Step 2. WRAT Arithmetic 0.742 0.551 0.424
Step 1. Decoding skills 0.886 0.785 0.662 II. Literacy Sum
Step 2. WRAT arithmetic 0.923 0.852 0.283 Step 1. Phonological Factor 0.615 0.378 −0.287
Step 3. RAN sum 0.928 0.861 −0.096 Step 2. WRAT Arithmetic 0.742 0.551 0.458
Step 4. ADD index 0.931 0.867 −0.094 Step 3. Decoding Skills 0.768 0.590 0.313
Note. ADD = Attention Deficit Disorder, ACID = WISC Arithmetic + Note. ADD = Attention Deficit Disorder, ACID = WISC Arithmetic +
Coding + Information + Digit Span, WISC = Wechsler Intelligence Scale Coding + Information + Digit Span, WISC = Wechsler Intelligence Scale
for Children, RAN = Rapid Automatized Naming, WRAT = Wide Range for Children, RAN = Rapid Automatized Naming, WRAT = Wide Range
Achievement Test. All correlation coefficients are significant at p < 0.01 or Achievement Test. All correlation coefficients are significant at p < 0.01 or
higher. higher.
158 ACKERMAN ET AL.: DOUBLE-DEFICIT THEORY

spelling impairment, which ranges from mild to severe. The Verbal IQ scores and reading and spelling standard scores
literacy sum scores for the RD children ranged from 231 (averaged for the WIAT and WRAT-3). The faster namers
to 357 (SD = 27.0), and the Decoding Skills (DST) sum had higher Verbal IQs (96.3) but no larger gap between IQs
ranged from 0 to 32 (SD = 8.7). Table 10 summarizes these and reading and spelling (16 and 15, respectively) than the
analyses. The DST sum is predicted only by the phonological slower namers (IQ = 92.7, gap = 16 for both reading and
factor. The literacy sum is predicted by the phonological spelling).
factor and the WRAT arithmetic scores (R = 0.742). If The current RD sample includes more girls and more
the literacy sum independent variables list includes DST African-American children than our past studies, but we
scores, the phonological factor scores are taken as step 1 found no gender or race differences in mean RAN scores
because they were more strongly correlated with literacy and no interactions with group. Age was not significantly
sum than the DST (–0.615 versus 0.553). However, the correlated with RAN scores. The RD sample also includes a
beta weight for the DST was somewhat higher in the final large number of children with attention problems, but slow
solution. and fast namers (median split) did not differ on the ADD
The orthographic tasks formed a single factor that was index.
so robustly correlated with reading and spelling standard RAN has been hypothesized to relate to orthographic pro-
scores and the phonological factor and DST scores as to cast cessing, but within our RD sample the slower and faster
doubt on its use as an independent variable. Rather, this factor namers (median split) did not differ significantly on any of
appears to tap phonological as much as orthographic ability our orthographic tasks. The slow namers had virtually iden-
(see Velutino, Scanlon, & Chen, 1995, on this issue). tical reading and spelling standard scores (77 and 78) as did
the faster namers (80 for both). Thus, the slow namers were
not more impaired in spelling than reading.
Follow-Up Analyses of RAN

Because the subgroup and regression analyses did not show DISCUSSION
as adverse effect of slow naming as expected, we performed
several post hoc exploratory analyses with RAN scores to try Insofar as differentiating normal and disabled readers, this
to discover why our findings are not strongly supportive of study successfully replicated other studies from our labora-
the double-deficit theory. tory and from other RD investigators. As expected, the RD
Even in our first major study to combine RAN and phono- and NR groups clearly differed on oral word analysis skills
logical variables, we had to include Verbal IQ and a test of and on rapid naming speed. They also differed on parental
short-term memory in order to achieve a robust multiple R ratings of attention and behavioral problems and on three of
in the prediction of reading level (Ackerman & Dykman, the four WISC III factor scores as well as the once clinically
1993). The current model (Table 9) incorporating ACID pat- popular ACID factor (Swartz, 1974; Ackerman et al., 1977;
tern scores, standardized arithmetic scores, and ADD ratings Frederickson, 1999). Although selected for reading/spelling
provides an even better fit. Examination of Table 8 offers at deficits, the RD group was also impaired in arithmetic. The
least a partial explanation. Note that RAN scores are sig- comorbidity of attention disorder and RD has often been re-
nificantly related to ACID scores (–0.56) and phonological ported (Dykman & Ackerman, 1991; Willcutt & Pennington,
scores (0.52), which have the highest correlations with lit- 2000) as has the comorbidity of reading/spelling and arith-
eracy sum. Of the four WISC subtests making up the ACID metic disabilities (Ackerman & Dykman, 1995). While read-
pattern, RAN times are most strongly correlated with Coding ing disabilities and attention disorders occur at all levels of
(–0.50, p < 0.001) but each of the other associations are also intelligence, it is nonetheless the case that referred and/or
significant (Arithmetic, –0.41; Information, –0.40, and Digit recruited samples of RD children generally have somewhat
Span –0.34, all p = 0.001). When RAN sum was included lower IQs and higher levels of attention problems than stan-
with all the WISC subtests in a principal components fac- dardization samples. Likewise, samples of ADHD children
tor analysis, its loading was –0.71 on a presumed perceptual generally have somewhat lower IQs and achievement scores
speed factor with Coding (0.78) and Symbol Search (0.70). than standardization samples.
Information, Arithmetic, and Digit Span had high loadings A primary reason for the current study was to assess pos-
with the other Verbal subtests. Thus, ACID scores appear to sible additive effects of impaired oral word analysis (phono-
offer a good estimate of academic aptitude because of the logical) skill and slow continuous naming on degree of read-
inclusion of measures of long-term, short-term, and working ing/spelling impairment. This question was addressed first
memory, as well as speed of processing. RAN is not as inclu- by subgrouping and then by stepwise multiple regression
sive a measure as ACID, but does appear to measure speed analysis. The subgrouping maneuver revealed, as expected,
of mental processing just as well as Coding (see Ackerman that those RD children with phonological deficits read and
& Dykman, 1996). spelled worse than those without serious phonological defic-
Earlier we had found that RAN scores were not as charac- its. However, there was no evidence of an additive effect of
teristic of “garden variety” (non-IQ discrepant) poor read- slow naming. Rather, the RD children with double defic-
ers as of discrepant poor readers (Ackerman & Dykman, its read and spelled no worse than those with a phono-
1993; Ackerman, Weir, Metzler, & Dykman, 1996). With the logical deficit only, and the children with a single deficit
present RD children, we created two RAN groups (median in naming performed no worse than those with neither
split at 116 seconds), and looked at the differences between deficit.
LEARNING DISABILITIES RESEARCH 159

The various multiple regression analyses likewise failed strength in memory allowed them to acquire academic skills
to show much of an adverse impact from slow naming when at a normal rate. As if to defy those who wish to categorize
other contributory variables were considered. RAN scores children, the NR child with the naming/phonological deficit
were robustly correlated with literacy sum (reading plus also had high ADD scores yet was above average in reading
spelling from the WIAT and WRAT-3) but explained only and arithmetic although low average in spelling. Her phono-
7% of additional variance once the phonological factor en- logical deficit is questionable, however, since she decoded 31
tered. The ACID factor and WRAT-3 arithmetic scores were pseudowords on the DST. Indeed, only 1 NR child who met
much stronger predictors when taken with the phonological the cut-score algorithm for a phonological deficit had a very
scores. Within the RD group, once the phonological factor low DST score, his being 12.
was accepted, the RAN sum did not enter in the prediction Denckla and Cutting (1999) have recently reviewed the
of literacy sum. history of testing for rapid automatized naming. They con-
Our failure to find strong support for the phonologi- clude that rapid naming is a complex ability, and that the RAN
cal/naming speed double-deficit theory of reading disabil- tasks tap at minimum both visual-verbal ability (language do-
ity may, of course, simply be attributable to the vagaries of main) and processing speed (executive domain). Our factor
sampling. But, it could also be that the lesson of the blind analysis (Table 8) suggests there is an overall “g” factor for
men and the elephant is being ignored by some investiga- literacy achievement and that rapid naming loads robustly on
tors (but see Wolf & Bowers, 1999, pp. 422–423). Let us as- the factor, although not as strongly as phonological analysis.
sume that rapid naming of alphanumeric stimuli is an index Our various statistical maneuvers also show that there are yet
of speed of mental processing and resistance to mental fa- other underlying abilities that must be measured to account
tigue during continuous processing, both necessary to good for all instances of impaired literacy acquisition.
working memory. Evidence of this comes from significant Unfortunately, we did not obtain a measure of reading
correlations of RAN sum with three WISC subtests: Cod- speed for connected text. In the age group studied here,
ing (–0.498, p < 0.001), Arithmetic (–0.409, p < 0.001), and where level of word recognition skill varied from early 1st
Digit Span (−0.335, p < 0.001). RAN times are also corre- to 8th grade, it was not possible to find an appropriate
lated with Information (–0.403, p < 0.001) and Vocabulary passage for oral reading on which all subjects could be
(–.324, p = 0.001), suggesting association with the “g” fac- compared. In an earlier study of adolescent RD subjects
tor. Additionally, RAN times are correlated with ADD in- (Ackerman & Dykman, 1996), we were able to compare
dex scores (0.433, p < 0.001), and with Arithmetic achieve- the reading speed of slower and faster continuous namers
ment scores (–0.420, p < 0.001) as well as with Reading (median split). These subjects all read aloud a 3rd-grade-
and Spelling achievement scores (–0.602 and –0.553, respec- level paragraph. The slower namers were the slower readers
tively, p < 0.001 for each). and vice versa (Chi Square = 10.03, p < 0.01). Further, those
Our regression analyses suggest there are multiple deficits adolescents who were slow namers only were as prone to
underlying literacy acquisition and that rapid naming is asso- be slow readers as slow namers with phonological deficits,
ciated with several of these. However, other measures, such and those with phonological deficits only were prone to be
as the ACID factor, the ADD index, and Arithmetic achieve- faster readers of the 3rd-grade text than those who were
ment, provided stronger associations with literacy acquisition slow namers only. Thus, isolated slow naming may eventu-
than RAN times. Also RAN times and the phonological factor ate in reading impairment, even though the student has “the
were not independent (R = 0.52). cipher.” Wolf and Bowers (1999) concur on this point. As
Two puzzling questions arise from, but are partially an- we have noted before (Ackerman & Dykman, 1996), slow
swered by, our regression analyses. First, why is it that there readers are not as apt to read for pleasure and likely will not
are children who have normal oral phonological analysis acquire the spelling proficiency that accrues from repeated
skill and are adequate namers (the neither subgroup) yet are exposure to print. Additionally, very slow readers tax working
poor readers? What weaknesses do they exhibit? Secondly, memory and are therefore not as good comprehenders of text
why are there children who exhibit deficits common to poor (Baddeley, 1986).
readers yet who read at age-expected levels? How do they Several practical implications arise from the findings
compensate? of the present study. Foremost is the advisability of eval-
To address the first question, in the “neither” RD sub- uating RD children’s abilities in memory, attention, and
group (N = 18), all except 6 had either low ACID scores speed of processing as well as their phonological awareness
(<36) or high ADD scores (≥15) or both. Adverse scores on skills. As Lovett and her colleagues (Lovett, Lacerenza, &
these two measures clearly contribute to literacy deficits (see Borden, 2000) have amply demonstrated, the remedial pro-
Tables 8 and 9). Also, the “neither” group, despite having oral cess should include instruction in strategies for recognizing
phonological analysis skills in the range of normal readers, and remembering recurring orthographic patterns as well as
did not perform as well as the NR group on the Decoding direct instruction in phonological analysis. Indeed, strategic
Skills Test (mean total decoded of 18.5 versus 38.4). Most instruction may be just what is needed for RD children who
had not mastered “the cipher.” do seem to have phonological awareness, yet not the know-
As for the second question, in the normal-reader group, how to utilize this oral analysis ability in decoding written
there were 3 who were slow namers only (>116 seconds), 4 words. The issue of whether automaticity and fluency can be
with phonological deficits only, and 1 with both deficits. None abetted in slow namers who are also slow readers has only
of these NR children with naming or phonological deficits had begun to be addressed (see Wolf, Miller, & Donnelly, 2000).
ACID scores <36 or arithmetic standard scores <85. Perhaps Early results suggest that Wolf ’s so-called RAVE-O program
160 ACKERMAN ET AL.: DOUBLE-DEFICIT THEORY

(Retrieval, Automaticity, Vocabulary, Elaboration, Orthogra- Goswami, U., & Bryant, P. (1990). Phonological skills and learning to read.
phy) can produce positive results. East Sussex, UK: Erlbaum.
Gough, P. B., & Walsh, M. A. (1991). Chinese, phonicians, and the
orthographic cipher of English. In S. A. Brady & D. P. Shankweiler
(Eds.), Phonological processes in literacy (pp. 199–210). Hillsdale,
NOTE NJ: Erlbaum.
Lovett, M. W., Lacerenza, L., & Borden, S. L. (2000). Putting struggling
This research was supported by Grant 5R01 HD34182 from readers on the PHAST track: A program to integrate phonological and
the National Institute of Child Health and Human Develop- strategy-based remedial reading instruction and maximize outcomes.
Journal of Learning Disabilities, 33, 458–476.
ment and by Arkansas Children’s Hospital Research Insti- Meyer, M. S., Wood, F. B., Hart, L. A., & Felton, R. H. (1998). Longitudinal
tute. We are grateful to coworkers Nancy B. Stewart, Shane course of rapid naming in disabled and nondisabled readers. Annals of
Eilts, and Dannette Rook for their help in data collection Dyslexia, 48, 91–114.
and manuscript preparation. We are also grateful to Jeanette Olson, R., Wise, B., Conners, F., & Rack, J. (1989). Deficits in disabled
readers’ phonological and orthographic coding: Etiology and remedia-
McGrew, Dr. Larry Clarke, Dr. Glen Lowitz, and local ele- tion. In C. Von Euler, I. Lundberg, & G. Lennerstand (Eds.), Brain and
mentary school principals for help in recruiting reading dis- Reading (pp. 233–242). New York: Stockton Press.
abled participants. Pennington, B. (1991). Diagnosing learning disorders: A neuropsychologi-
cal framework. New York: Guilford Press.
Richardson, E., & DiBenedetto, B. (1985). The decoding skills test. Parkton,
MD: York Press.
REFERENCES Rosner, J., & Simon, D. P. (1971). The auditory analyses test: an initial report.
Journal of Learning Disabilities, 4, 384–392.
Achenbach, T. M. (1991). Manual for the child behavior checklist/4-18 and Spring, C., & Capps, C. (1974). Encoding speed, rehearsal, and probed recall
1991 profile. Burlington, VT: University of Vermont, Department of of dyslexic boys. Journal of Educational Psychology, 66, 780–786.
Psychiatry. Spring, C., & Davis, J. M. (1988). Relations of digit naming speed with three
Ackerman, P. T., & Dykman, R. A. (1993). Phonological processes, im- components of reading. Applied Psycholinguistics, 9, 315–334.
mediate memory, and confrontational naming in dyslexia. Journal of Spring, C., & Perry, L. (1983). Naming speed and serial recall in poor and
Learning Disabilities, 26, 597–609. adequate readers. Contemporary Educational Psychology, 8, 141–145.
Ackerman, P. T., & Dykman, R. A. (1995). Reading disabled adolescents Stanovich, K. E. (1992). Speculations on the causes and consequences of
with and without comorbid arithmetic disability. Developmental Neu- individual differences in early reading acquisition. In P. B. Gough,
ropsychology, 11, 351–371. L. C. Ehri, & R. Treiman (Eds.), Reading acquisition (pp. 307–342).
Ackerman, P. T., & Dykman, R. A. (1996). The speed factor and learning Hillsdale, NJ: Lawrence Erlbaum.
disabilities: The toll of slowness in adolescents. Dyslexia, 2, 1–21. Swartz, G. A. (1974). The language learning system. New York, NY: Simon
Ackerman, P. T., Dykman, R. A., & Peters, J. E. (1977). Learning disabled & Schuster.
boys as adolescents: Cognitive factors and achievement. Journal of the Tallal, P., Miller, S., & Fitch, R. H. (1993). Neurobiological basis of speech:
American Academy of Child Psychiatry, 16, 296–313. A case for the pre-eminence of temporal processing. Annals of the New
Ackerman, P. T., Weir, N. L., Metzler, D., & Dykman, R. A. (1996). A study York Academy of Sciences, 682, 27–47.
of adolescent poor readers. Learning Disabilities Research & Practice, Tunmer, W. E., & Hoover, W. A. (1992). Cognitive and linguistic factors
11, 68–77. in learning to read. In P. B. Gough, L. C. Ehri, & R. Treiman (Eds.),
Baddeley, A. (1986). Working memory. Oxford, England: Clarendon Press. Reading acquisition (pp. 175–214). Hillsdale, NJ: Erlbaum.
Badian, N. A. (1997). Dyslexia and the double deficit hypothesis. Annals of Velutino, F., Scanlon, D., & Chen, R. (1995). The increasingly inextricable
Dyslexia, 47, 69–87. relationship between orthographic and phonological coding in learning
Badian, N. A. (1996). Dyslexia: A validation of the concept at two age levels. to read: Some reservations about current methods of operationalizing
Journal of Learning Disabilities, 29, 102–112. orthographic coding. In V. W. Berninger (Ed.), The varieties of ortho-
Boder, E., & Jarrico, S. (1982). The Boder test of reading-spelling patterns. graphic knowledge. II: Relationships to phonology, reading, and writing
New York: Grune & Stratton. (pp. 47–111). Dordrecht, the Netherlands: Kluwer Academic.
Bowers, P., & Wolf, M. (1993). Theoretical links among naming speed, Wagner, R. K., & Torgesen, J. K. (1987). The nature of phonological process-
precise timing mechanisms, and orthographic skill in dyslexia. Reading ing and its causal role in the acquisition of reading skills. Psychological
and Writing, 5, 69–86. Bulletin, 101, 192–212.
Bowers, P. G., Sunseth, K., & Golden, J. (1999). The route between rapid Willcut, E. G., & Pennington, B. F. (2000). Comorbidity of reading disability
naming and reading progress. Scientific Studies of Reading, 3, 31–53. and attention-deficit/hyperactivity disorder: Differences by gender and
Bradley, L. (1984). Assessing reading difficulties: A diagnostic and remedial subtype. Journal of Learning Disabilities, 33, 179–191.
approach. London: Macmillan Education Ltd. Wolf, M. (1991). Naming speed and reading: The contribution of the cogni-
Cohen, R. L., & Netley, C. (1981). Short-term memory deficits in reading tive neurosciences. Reading Research Quarterly, 26, 123–141.
disabled children, in the absence of opportunity for rehearsal strategies. Wolf, M. (1999). What time may tell: Towards a new conceptualization of
Intelligence, 5, 69–76. developmental dyslexia. Annals of Dyslexia, 49, 3–28.
Denckla, M. B., & Cutting, L. E. (1999). History and significance of rapid Wolf, M., & Bowers, P. G. (1999). The double-deficit hypothesis for
automatized naming. Annals of Dyslexia, 49, 29–42. the developmental dyslexias. Journal of Educational Psychology, 91,
Denckla, M. B., & Rudel, R. (1976). Rapid “automatized” naming (R.A.N.): 415–438.
Dyslexia differentiated from other learning disabilities. Neuropsycholo- Wolf, M., Miller, L., & Donnelly, K. (2000). Retrieval, automaticity, vocab-
gia, 14, 471–479. ulary, elaboration, orthography (RAVE-O): A comprehensive, fluency-
Dykman, R. A., & Ackerman, P. T. (1991). Attention deficit disorder and based reading intervention program. Journal of Learning Disabilities,
specific reading disability: Separate but often overlapping disorders. 33, 375–386.
Journal of Learning Disabilities, 24, 96–103. Wolf, M., & Obregon, M. (1992). Early naming deficits, developmental
Fawcett, A. J., & Nicolson, R. I. (1994). Naming speed in children with dyslexia, and a specific-deficit hypothesis. Brain and Language, 42,
dyslexia. Journal of Learning Disabilities, 27, 641–646. 219–247.
Fawcett, A. J., & Nicolson, R. I. (1995). Persistent deficits in motor skill of Wolf, M., & Obregon, M. (1995). Naming-speed deficits and the “double
children with dyslexia. Journal of Motor Behavior, 27, 235–240. deficit” hypothesis: Implications for diagnosis and practice in reading
Frederickson, N. (1999). The ACID test: Or is it? Educational Psychology disabilities. Unpublished manuscript.
in Practice, 15, 2–8. Wolff, P., Michel, G., Ovrut, M., & Drake, C. (1990). Rate and timing preci-
Goswami, U. (1993). Phonological skills and learning to read. Annals of New sion of motor coordination in developmental dyslexia. Developmental
York Academy of Sciences, 682, 296–311. Psychology, 26, 349–359.

You might also like