Content uploaded by David L. Share
Author content
All content in this area was uploaded by David L. Share on May 17, 2023
Content may be subject to copyright.
Subtypes of reading disability in a shallow orthography:
a double dissociation between accuracy-disabled
and rate-disabled readers of Hebrew
Michal Shany &David L. Share
Received: 12 January 2010 /Accepted: 4 October 2010 / Published online: 25 November 2010
#The International Dyslexia Association 2010
Abstract Whereas most English language sub-typing schemes for dyslexia (e.g., Castles &
Coltheart, 1993) have focused on reading accuracy for words varying in regularity, such an
approach may have limited utility for reading disability sub-typing beyond English in which
fluency rather than accuracy is the key discriminator of developmental and individual
differences in reading ability. The present study investigated the viability of an accuracy/
fluency-based typology in a regular orthography, pointed Hebrew. We sought evidence of true
or “hard”accuracy/rate subtypes in the strict (double dissociation) sense of selective impairment
on only one dimension in the presence of normal levels of performance on the other dimension.
In a nationally representative sample of fourth graders, we were able to identify a specific
accuracy-disabled sub-group as well as an equally specific rate-disabled subgroup. Validating
this subdivision, we show that the nature of reading performance in these subgroups and their
converging cognitive/linguistic profiles are unique and distinctive on variables other than the
measures used to define them. While the rate-specific disability appeared to reflect a general
deficit in speed of processing affecting reading rate, and rapid automatized naming of print-
related material, the accuracy-only disability subgroup displayed selective deficits in
phonological awareness and morphological knowledge. Biosocial, demographic, and instruc-
tional factors, furthermore, did not explain the sub-group differences. It appears that both these
subtypes are equally prevalent each counting close to 10% of the population.
Keywords Accuracy .Double-dissociation .Dyslexia .Hebrew .Rate .Subtypes
The field of reading disabilities has achieved genuine progress in the past few decades (see, e.g.,
Beaton, 2004; Vellutino, Scanlon, & Snowling, 2004; Stanovich, 2000; National Reading
Panel, 2000; Rayner, Foorman, Perfetti, Pesetsky, & Seidenberg, 2001; Snowling & Hulme,
2005). Today, there is a wide consensus that the cognitive basis of many specific (and non-
Ann. of Dyslexia (2011) 61:64–84
DOI 10.1007/s11881-010-0047-4
M. Shany (*):D. L. Share (*)
The Department of Learning Disabilities, Edmond J. Safra Brain Research Center in Learning
Disabilities, The University of Haifa, Mt. Carmel, Haifa 31905, Israel
e-mail: shany.michal7@gmail.com
e-mail: dshare@construct.haifa.ac.il
specific) reading difficulties is a phonological deficit. Although phonological deficits among
disabled readers are most salient in tasks measuring phonological awareness, phonological
deficits extend well beyond these reading-related tasks into speech perception, phonological
learning, and memory (see Shankweiler & Fowler, 2004 for a review of the full picture).
The phonological deficit hypothesis (see Shankweiler & Liberman, 1989), however,
does not presume to explain all the variance in reading disabilities (Share & Stanovich,
1995). It is now clear that at least some aspects of phonological processing (such as
phonological awareness) may be relatively unimpaired (at least as regards accuracy)in
some disabled readers (Morris et al., 1998; Wolf & Bowers, 1999). These considerations
highlight the possibility of multiple sources of reading disability. The notion of
heterogeneity among dyslexics has naturally spurred attempts to demarcate subtypes of
reading disability (see, e.g., Boder, 1971; Lyon, Stewart, & Freedman, 1982; Lyon &
Watson, 1981; Mattis, French, & Rapin, 1975; Morris et al. 1998). Indeed, the typologies of
reading disabilities have a long but inglorious history (Fletcher et al., 1997; Morris et al.,
1998; Stanovich & Siegel, 1994; Lovett 1984a,1984b; Stanovich, 1991). The main
problem is that most typologies have been grounded more in clinical intuitions and/or
untested theoretical preconceptions than in a well-developed and rigorously tested theory of
reading. A notable exception to this irresolution is the subtyping approach inspired by the
highly influential Coltheart/Baron dual-route model of word reading (Coltheart et al., 1993,
2001) which has attracted considerable interest (as well as controversy) among reading
disability researchers (see, e.g., Bailey, Manis, Pedersen, & Seidenberg, 2004; Castles &
Coltheart, 1993; Harm & Seidenberg, 1999; Jackson & Coltheart, 2001; Manis et al., 1996;
Milne, Nicolson, & Corballis, 2003; Stanovich et al. 1997a,1997b).
The distinction between surface and phonological subtypes of reading disability (both
developmental and acquired dyslexia) is founded on the separability (specifically, double
dissociation) of non-lexical and lexical procedures for reading words aloud. The central axiom
of the Coltheart/Baron dual-route model is that no single procedure yields correct
pronunciations of both nonwords (e.g., slint) and exception words (e.g., pint). Nonwords can
only be correctly pronounced via grapheme–phoneme correspondence rules, the “non-lexical”
route; exception words require an additional procedure, the “lexical”route, because they
cannot be pronounced by the rules. Applying this framework to reading disability subtypes, it
has been claimed that surface dyslexics have a selective impairment in the lexical route,
whereas phonological dyslexics are selectively impaired in the non-lexical route. (Of course,
dyslexics with impairments in both routes will constitute the most severely disabled readers
and, if the lexical and non-lexical routes are correlated, will be the most prevalent.) A good
deal of controversy, however, has surrounded the question of the prevalence and validity of
the two key “single-deficit”subtypes, especially the issue of whether these groups should be
defined relative to chronological-age matched control groups as opposed to reading-age
control groups (see Jackson & Coltheart, 2001; Manis et al., 1996; Stanovich et al. 1997a,
1997b). A broader, more cross-linguistic concern is the fact that the surface/phonological
typology focuses almost exclusively on the accuracy of oral reading, in particular, the
pronunciation accuracy of non-words and exception words (see Share, 2008,foradetailed
discussion). Unlike English, however, most orthographies have highly regular grapheme–
phoneme correspondences (see Daniels & Bright, 1996; Joshi & Aaron, 2006; Seymour, Aro,
& Erskine, 2003), with relatively few “exception”words; hence, the applicability of the dual-
route framework beyond English has come into question (e.g., Hutzler et al., 2004; Share,
2008; Ziegler & Goswami, 2005). By implication, this problem extends to the surface/
phonological typology derived from the dual-route model (see, e.g., Genard, Mousty, Content,
Alegria, Leybaert, & Morais, 1998;Ho,Chan,Chung,Lee,&Tsang,2007).
Subtypes of reading disability in a shallow orthography 65
Accuracy of oral reading, however, is largely a non-issue for the majority of the world’s
(alphabetic) orthographies, in which exceptions are rare and, consequently, performance
levels approach ceiling by the end of grade 1 (Seymour et al., 2003). When accuracy
rapidly asymptotes, speed and fluency become the discriminating measures of develop-
mental and individual differences (Breznitz, 1997; Cossu 1999a,1999b; de Jong & van der
Leij, 2003; Leppanen, Niemi, Aunola, & Nurmi, 2006; Lyytinen, Aro, & Holopainen, 2004;
Nikolopoulos, Goulandris, Hulme, & Snowling, 2006; Wimmer, 1993; Yap & Van der Leij,
1993; Zoccolotti et al., 1999). In German (Wimmer, 1993), Dutch (Yap & van der Leij,
1993), Norwegian (Lundberg & Hoien, 1990), Italian (Zoccolotti et al., 1999), Greek
(Porpodas, 2006), Finish (Lyytinen et al., 2004), Hungarian (Csepe, 2006), and Hebrew
(Breznitz, 1997), most dyslexics attain high levels of reading accuracy but remain slow.
This implies that an exclusive focus on reading accuracy may have limited utility for
reading disability subtyping beyond English: accuracy/fluency dissociations (rather than
nonword/exception dissociations) may be a more profitable avenue for a universal, i.e.,
cross-linguistic subtyping scheme.
One of the few researchers to explore a subtyping scheme based on accuracy/rate
criteria, was Lovett (1984a,1984b,1987) who classified a clinical sample (ages 8–13) into
accuracy-disabled and rate-disabled subgroups. To be subtyped as “accuracy-disabled,”a
child had to score at least 1.5 years below grade level expectations on multiple measures of
word recognition accuracy, but no rate criterion was designated for this group. To be
subtyped as “rate-disabled,”a child had to score “close to, at, or above grade level”on
accuracy and at least 1.5 years below grade level on reading speed. The accuracy disabled,
however, were also impaired in reading rate, and, in fact, were slower readers than the rate-
disabled group. Thus, the accuracy-disabled group was, overall, more severely impaired
readers, or “doubly-disabled,”that is rate- and accuracy-disabled, whereas the rate-disabled
were only “singly-disabled.”Conceptualizing reading development as a developmental
progression from first achieving error-free (i.e., accurate) performance to later-developing
speeded performance, Lovett suggested that the two subgroups represent different points on
a developmental continuum of reading acquisition. Consistent with this observation, the
accuracy (plus rate) subgroup were more impaired on a range of reading and spelling tasks
and also displayed a broad array of oral language deficits. An alternative interpretation of
these data is that speed and accuracy are two correlated measures tapping a single common
underlying construct; the rate-disabled subgroup is simply less severely disabled on this
common dimension than the accuracy-and-rate disabled subgroup. By this account, a uni-
dimensional taxonomy can still be maintained merely by distinguishing different levels of
severity on this continuum. Lovett did not explore the possibility of true double dissociation
by attempting to identify an accuracy-only disabled group.
Leinonen et al. (2001), however, identified a rate-only disabled subgroup in addition to
an accuracy-only disabled subgroup in their subtyping study of a sample of Finnish adult
dyslexics. A group of 84 dyslexics (average age 30) with one or more close relatives with
dyslexia were compared to a normative sample of 100 adults. Dyslexics were selected from
a larger sample of adults, defined as one standard deviation or more below the norm on
either accuracy or speed of oral text reading “or alternatively in spelling accuracy.”Errors
and reading times for oral text reading were transformed to zscores within the dyslexic
sample then the data cluster analyzed. Beyond the majority of dyslexics who were either
mildly impaired (n=44) or severely impaired (n=5) on both rate and accuracy, two
additional subgroups emerged with selective impairments. A “hasty”subtype (n= 22) were
found to be relatively fast but inaccurate, and a “hesitant”subgroup (n= 13) committed
relatively few errors but were slow. Although these data suggest true double dissociation,
66 M. Shany, D.L. Share
the accuracy of the hesitant subgroup as well as the speed of the hasty subgroup were
significantly below normative levels. Thus, these subtypes can only be considered “soft”
not “hard”subgroups, in the sense of intact performance on one dimension co-occurring
with impaired performance on the other (Stanovich & Siegel, 1994). Both subgroups
displayed impairment of differing degrees on rate and accuracy. A further complication is
that the hasty subgroup was almost 10 points lower on IQ, and potentially confounding age
and gender differences were not reported.
Our study sought evidence of “hard”accuracy/rate subtypes in the strict (double
dissociation) sense of selective impairment on only one dimension in the presence of
normal levels of performance on the other dimension. We also sought to establish the
relevant prevalence of these subtypes in the general population by employing an unselected,
nationally representative sample of fourth graders.
Method
The parent sample
The sample in the present study consisted of a subset of a two-stage stratified nationally
representative sample of 1,478 children in grades 2, 4, and 6. In the first stage, 52 schools
were randomly selected out of 1,164 regular elementary schools in the country and a
further 20 (out of 147) special education schools. From each school, one class was
selected at each of three grade levels: 2, 4, and 6. Thus, the original (parent) sample
included an over-representation of the special education population (80% from regular
schools and 20% from special education schools). In the second stage, children were
randomly selected within each class after written parental consent was obtained (the rate
of non-compliance was about 1%). The final sample consisted of 461 grade 2 children,
512 grade 4 children (388 from regular education, 124 from special education,
respectively), and 511 grade 6 children.
The sample in the present study
The sample in the study reported here included only grade 4 children. At this point, children
have moved beyond the initial stage of code instruction and demonstrate mastery of both
versions of Hebrew orthography: pointed and unpointed (Share & Levin, 1999). (Hebrew is
a dual-version orthography, with unpointed script providing only partial vowel information
by means of a set of four vowel letters, and a pointed or fully vowelled script using vowel
diacritics to convey complete vowel information (Ravid, 2005; Share & Levin, 1999;
Shimron, 1993). Thus, pointed Hebrew can be considered a transparent (shallow)
orthography whereas unpointed Hebrew is considerably deeper. Children initially learn to
read using the shallow pointed system but later (around grade 3) switch to the deeper
unpointed version of the script which is the norm for skilled readers.)
Since the original sample included an over-representation of special education children,
the same sampling procedure was applied to the special education sector in order to reflect
the true proportion (2.5%) of children in special education schools. Of the 124 children
from special education, 10 classes were first randomly selected, then one child per class.
Finally, only children who had complete scores for both accuracy and speed of word
reading (the criterion measure for designating accuracy-disabled and rate-disabled
subgroups) were included in the final sample which totaled 386 children.
Subtypes of reading disability in a shallow orthography 67
Criteria for selection of reading disability subgroups
The selection of reading disability subgroups was based on the accuracy (percentage of words
correctly pronounced) and rate (in words per minute) for reading isolated pointed (i.e., fully
pointed) words. These measures are standard for identifying disabled readers in shallow
orthographies (e.g., Breznitz, 1997;Cossu1999a,1999b; de Jong & van der Leij, 1999;
Leppanen et al. 2006; Lyytinen et al. 2004; Nikolopoulos et al. 2006;Wimmer,1993;
Zoccolotti et al., 1999).
As is well-established in the case of pointed Hebrew orthography (Share & Levin, 1999;
Shatil & Share, 2003), the mean level of accuracy was high (M=86.0%, SD=12.69) with
children reading an average of 48.9 words per minute (SD=18.32). These figures signify
high levels of mastery of the alphabetic code of Hebrew among the sample as a whole. It is
important to note that the measure of reading rate was uncorrected for accuracy, that is, it
represented pure reading rate unadjusted for accuracy, rather than a fluency composite (such
as words correctly read in one minute) that combined accuracy and rate.
As anticipated, the simple bivariate correlation between accuracy and rate was only 0.50;
the first-order polynomial (quadratic) was r=0.55, suggesting only partial overlap between
children’s oral reading accuracy and rate. It seems reasonable, therefore, to expect a certain
degree of dissociation between these two dimensions.
In order to identify reading-disabled subgroups, we followed conventional practice in the
reading disability literature (e.g., Siegel, 1999), using a 25th percentile low achievement cut-off
on each measure. This corresponded to accuracy below 80.1% and a reading rate less than 35.5
words per minute. Normal achievement was designated at or above the 35th percentile,
corresponding to an accuracy score above 84.2% and a reading rate higher than 41.53 words
per minute. Thus, children scoring below the 25th percentile mark on accuracy but above the
35th percentile on rate were designated accuracy-disabled; children scoring below the 25th
percentile on reading rate but above the 35th percentile on accuracy were labeled rate-disabled,
and children scoring below the 25th percentile on both rate and accuracy were labeled
accuracy-plus-rate-disabled (or doubly disabled). The rate-disabled group numbered 38 children
(9.8% of the total sample), the accuracy-disabled subgroup included 36 children (9.3% of the
sample), and the accuracy-plus-rate-disabled subgroup included 51 children (13.2% of the
sample). Thus, approximately one third of our reading disabled children were classified as rate-
disabled, one third as accuracy-disabled, and slightly more than one third as doubly-disabled.
As indicated in Table 1, the rate-disabled group exhibited a selective rate deficit. Their
accuracy level (89.7%) was close to the national average (86%) but their reading rate was
only around half the national average (28 words per minute) falling at the 14th percentile. The
accuracy-disabled subgroup displayed the reverse pattern; their accuracy was very poor
(73%) corresponding to only the 15th percentile, whereas reading rate was normal (53.5
words per minute, percentile ranking 61). The group with combined rate and accuracy deficits
was significantly below average on both measures. Furthermore, their accuracy and rate
scores were poorer than the corresponding means for the accuracy-disabled and rate-disabled
subgroups, although the difference attained significance only in the case of accuracy. It
appears that this doubly-disabled subgroup constituted the most severely disabled readers.
In order to ensure that these outcomes were not somehow an artefact of this one criterion, we
also re-defined our subgroups using a more stringent 16th percentile cutoff. The results painted
essentially the same picture. Although there were slightly fewer accuracy-only disabled readers
compared to rate-only disabled readers (16 versus 23; the doubly-disabled sub-group were again
the most numerous, n=28), the same picture of double dissociation was obtained: The accuracy-
disabled subgroup displayed normal reading rate levels (above the national mean) but impaired
68 M. Shany, D.L. Share
Table 1 Descriptive statistics for reading deficit subgroups: means (in number and % of children) and SD (in parentheses)
Variable Group
Rate deficit
(n= 38)
Accuracy deficit
(n= 36)
Rate & accuracy
deficits (n= 51)
National norms
(n= 396)
Statistical analyses
Performance according to criteria for subgroups selection
Isolated pointed words: % Accuracy (38 items) 89.7 (5.24)
a
†73.5 (7.99)
b
†64.6 (14.99)
c
†86.0 (12.7) F
(2,120)
=58.0 p < .001
Isolated pointed words: Rate 28.0 (5.95)
a
†53.5 (12.07
b
24.0 (7.70)
a
†48.9 (18.3) F
(2,120)
=131.4 p < .001
Gender:
Boys 17 (51.5%) 18 (54.5%) 25 (52.19%) 173 (49.1%) χ
2
(2)
=0.7, n.s
Girls 16 (48.5%) 15 (45.5%) 23 (47.9%) 179 (50.9%)
Country of birth:
Born in Israel 33 (86.8%) 32 (88.9%) 42 (84.08%) 350 (91.1%) χ
2
(2)
=.437, n.s
Immigrants (more than 4 years in Israel) 5 (13.2%) 4(11.1%) 8 (16.0%) 34 (8.9%)
Socio-economic index of schools (rating) 1153.5 1148.0 1151.8 1145.3 F
(2,120)
=.349, n.s
Districts in Israel:
North 10 (26.3%) 6 (16.7%) 19 (37.3%) 102 (26.4%) χ
2
(6)
=9.72, n.s
Central (Tel-Aviv) 9 (23.77%) 4 (11.1%) 4 (7.8%) 50 (13.0%)
Central (Jerusalem) 14 (36.8%) 21 (58.3%) 23 (45.1%) 196 (50.8%)
South 5 (13.2%) 5 (13.9%) 5 (9.80%) 38 (9.8%)
Community/population:
Urban 26 (76.5%) 26 (78.8%) 30 (66.7%) 267 (69.2%) χ
2
(2)
=.169, n.s
Non-urban 8 (23.5%) 7 (21.2%) 15 (33.3%) 119 (30.8%)
Schools:
Regular 37 (97.4%) 35 (97.2%) 45 (88.2%) 378 (97.9%) No statistics available
Special Education 1 (2.6%) 1 (2.8%) 6 (11.8%) 8 (2.1%)
Subtypes of reading disability in a shallow orthography 69
Variable Group
Rate deficit
(n= 38)
Accuracy deficit
(n= 36)
Rate & accuracy
deficits (n= 51)
National norms
(n= 396)
Statistical analyses
Religiosity
Non-religious 28 (73.7%) 30 (83.3%) 42 (82.4%) 300 (77.7%) χ
2
(2)
=1.37, n.s
Religious 10 (26.3%) 6 (16.7%) 9 (17.6%) 86 (22.3%)
Method of reading instruction
Systematic alphabetic code 19 (55.9%) 15(44.1%) 30 (66.7%) 195 (51.5%) χ
(2)
2
=3.53, n.s
Holistic/global 15 (45.5%) 18(54.5%) 15 (33.3%) 149 (39.3%)
Mixture of alphabetic and holistic 1 3 4 35 (9.2%)
Note: Means with the †symbol indicate significant differences from national norms at .01
Means with different superscripts indicate significant differences at .01
70 M. Shany, D.L. Share
accuracy (8th percentile rank) and the rate-disabled subgroup the reverse pattern (above average
accuracy, 9th percentile reading rate). This pattern, moreover, was replicated in the validation
decoding measures; consonant–vowel decoding and pseudoword decoding.
Reading and verbal/linguistic measures
All reading, verbal memory, and linguistic measures were taken from the only individually
administered test battery with national norms available in Hebrew, “Alef Ad Taf”(“AtoZ;”
Shany, Lachman, Shalem, Bahat, & Zieger, 2006). For all reading tests, accuracy in percent
and rate in (uncorrected) words per minute were recorded.
Reading measures
Decoding
Consonants and vowels This subtest focuses on children’s ability to read 64 CV combinations
of consonants and vowels which constitute the basic building blocks of printed words. The
subtest includes all 22 regular (i.e., non-final) and five word final letters of Hebrew. Each
consonantal letter appears four times, each with a different vowel diacritic. Each vowel also
appears four times. Cronbach’s alpha for accuracy (based on the entire sample of 512 fourth
graders) is 0.91.
Pseudowords This sub-test consists of 33 nonwords; 24 represent permissible morpho-
phonological structures in the Hebrew language (e.g., “nirpag”) and nine items consist of phonol-
ogical strings that do not exist in the native language (e.g., “tutsted”). Alpha-Cronbach is 0.90.
Word reading
Pointed isolated words This test involves reading single pointed (i.e., fully voweled) words
in isolation. All 38 words are nouns representing different levels of frequency [(for
example, the high frequency word, דג , pronounced /dag/ (‘fish’) versus the low frequency
word, עפעפיים ,/afapayim/ (‘eyelids’)], lengths, and morphological structures. Words were
chosen based on frequency judgments made by 34 teachers and 5 linguists. Percentage
accuracy and words per minute are calculated. Alpha-Cronbach (grade 4) is 0.85.
Speeded reading of pointed isolated words taken from basal text This subtest contains 50
words taken from a 100-word (narrative) text in a fourth-grade basal reader. Subjects are
required to read the words as quickly and accurately as possible. Reliability (based on the
correlations with the test of isolated pointed words) is 0.82 for accuracy and 0.74 for rate.
Text reading
Pointed texts Two texts were included. The first, a speeded test which consists of a
paragraph taken from a story in a fourth-grade basal reader. Subjects are required to read the
text aloud as quickly as possible. The second text is a narrative story for the same grade
level and containing 196 words. Children are asked to read this text at their own convenient
Subtypes of reading disability in a shallow orthography 71
pace. Reliability estimates are based on computing the cross-correlations between accuracy
and rate for both texts. Both values were 0.87.
Unpointed text A 100-word paragraph of pointed text was used for evaluating the reading
of unpointed (partly voweled) text. The same instructions and scoring were used. Reliability
is based on the correlations with accuracy and rate on a parallel text (unpointed narrative
story for the same grade level). The reliability for accuracy was 0.69 and 0.79 for rate.
Verbal/linguistic measures
Verbal memory
Verbal fluency Children are asked to retrieve as many words as they can, beginning with a
specific letter. Three letters were included, b, g, and d, and 30 s were allocated for each
letter. The total number of words retrieved is calculated. Reliability was derived from the
correlations between the letters band g(0.52), band d(0.42), and gand d(0.47).
Story memory (factual details) This test requires participants to listen to a (110 words) story
about a balloon competition. The child then hears a series of 10 questions and has to answer
them by retrieving factual details from the story. Percent of correct responses is calculated.
No reliability is available for this test but the correlation with working memory was 0.76.
Working memory This subtest (Shany et al., 2006) was adapted from Daneman and
Carpenter (1980). In this test, the participant hears sentences, is asked to complete the
missing word in each, and then to repeat the missing words in the correct sequence. For
example: The hen lays ___ (eggs); An airplane flies in the ___(sky).“Eggs, sky.”The
number of items in each set increases from two to six. The child receives 1 point for each
correctly repeated word and an extra point for repeating all the words in each set in the
correct sequence. The Alpha-Cronbach for this test is 0.81.
Speed of processing (rapid automatized naming)
Digit naming speed (rapid automatized naming digits) The subtest consists of five digits: 1,
5, 9, 3, and 7, each repeated randomly 10 times. The 50 printed digits are presented to the
child, who has to read them aloud as fast as possible. The number of digits per minute is
calculated. Reliability, based on the correlation with the rapid automatized naming (RAN)
letters test, is r=0.69.
Letter naming (RAN letters) This subtest consists of five (non-final) Hebrew letters: ס(s), א,
(a), ד(d), ג(g), and ל(l), each repeated randomly 10 times. The 50 printed letters are
presented to the child, who has to read them aloud, as fast as possible. Number of letters per
minute is calculated.
Object naming speed (RAN objects) This subtest contains pictures of five common objects:
flower,cat,book,watch, and flag, each repeated randomly 10 times. Reliability is based on
the correlation with a parallel test (rapid naming of alternating stimuli; RAS) and r=0.59.
72 M. Shany, D.L. Share
Alternating stimulus naming speed This additional 50-item naming subtest consists of
pictures, numbers, and letters, each repeated randomly. Reliability is based on the
correlation with rapid naming of objects (RAN objects) and is r= 0.59.
Morphological knowledge
Morphological knowledge (inflections and derivations) Morphological knowledge (inflec-
tions and derivations). This subtest was developed by Prof. Yosef Shimron and Dr.
Ver e d Va a k ni n f o r t he “Alef Ad Taf”test (Shany et al., 2006). Children were orally
presented with past tense verb forms containing three root (consonantal) letters. (In
Hebrew, these three letters form the word’s root or base morpheme.) After a brief pause, a
sentence is then presented in which the last word is missing. There are two subtasks
required here. First, nine items have to be completed with a word that derives from the
initially presented verb. For example, when presented with the root סכנ ,skn,andthe
sentence “Snakes are…,”participants are required to complete the sentence with the
correct derivation, in this case מסוכנים ,mesukanim (‘dangerous’). Then, nine items have to
be completed by choosing the correct word out of four alternatives. Percentage accuracy
is recorded. Cronbach’s alpha is 0.72.
Syntax
Grammatical knowledge This subtest examines grammatical knowledge via understanding
of pronominal reference. The test includes a short story with 20 anaphoric references. The
story is read to the child who is requested to clarify the references. For example: in the
sentences “The sailor found a small cat. He took him to his cabin.”the word he refers to the
sailor and the word him, to the cat. Percent of correct responses is recorded. Cronbach’s
alpha is 0.82.
Phonological awareness
Syllable deletion The subtest focuses on children’s ability to segment spoken real words
into syllables [e.g., “Say /geshem/ (rain).”“Now say it without the /ge/.”Response: /shem/
(name)]. Target syllables appear in initial, medial, and final positions. All of the products of
this syllable-level word manipulation are real words in Hebrew. The subtest includes 14
items. Percent of correct responses is recorded. Alpha-Cronbach is 0.87.
Phoneme deletion The subtest focuses on children’s ability to segment 16 spoken real
words into phonemes [e.g., “Say xatul (cat).”“Now say it without the /x/.”Response: /atul/
]. None of the outcomes of this phoneme-level word manipulation is a real word in Hebrew.
Percent of correct responses is recorded. Alpha-Cronbach is 0.87.
Synthesizing phonemes into words The subtest focuses on children’s ability to synthesize
spoken phonemes into real words [e.g., k-a-f-t-o-r / ‘kaftor’(button)]. There are eight
familiar nouns ranging from three phonemes to eight phonemes. Alpha-Cronbach is
0.72.
Subtypes of reading disability in a shallow orthography 73
Analyzing words into phonemes This subtest focuses on a child’s ability to analyze spoken
words into phonemes [e.g., ‘deshe’/ d-e-sh-e (grass)]. There are eight familiar nouns
ranging from three phonemes to nine phonemes. Alpha-Cronbach is 0.87.
Procedure
Data collection took place towards the last third of the school year, between April and June.
The tasks were administered individually by trained graduate students of education. All
tasks were administered in a single 2-h session. The cognitive/linguistic measures were
administered first, followed by the reading tasks.
Results
Background characteristics of the reading-disabled subgroups
The first set of analyses were conducted to check for potential differences between the
subgroups on a variety of biosocial, demographic, and instructional variables (see Table 1).
Two sets of analyses were undertaken. The first set compared each of the three disabled
subgroups with the national norms, using the carat symbol to signify a significant
difference. Because of the large number of group comparisons, αwas set at 0.01 instead of
0.05. The second set of analyses compared the three subgroups with one another using
either a one-way analysis of variance with post hoc Bonferonni follow-up comparisons or
Chi-square tests. Distinct superscripts were used to indicate significant differences.
There were similar proportions of boys and girls in all groups (χ
2
<1.0, n.s.). For all three
subgroups, a similar proportion (averaging 86%) of the children was born in Israel (χ
2
<1.0,
n.s.) or were immigrants who had attended school in Israel since first grade. The Ministry of
Education’s index of school-level socio-economic status, which reflects the degree of
affluence (or poverty) of the school’s catchment area (based on the number of school years
completed by parents, income level, housing density, computer ownership, etc), did not differ
between the groups (F<1.0, n.s.). Approximately half of the children in each group came
from the greater Jerusalem area and no significant differences were found either for the
geographical distribution, or for urban versus non-urban communities. The proportion of
children in special education settings was similar for both the selectively disabled subgroups
(close to the national average) but, unsurprisingly, there were a larger number of special
education children in the doubly-disabled group. Across subgroups, similar proportions of
children were religious and non-religious (χ
2
<1.0, n.s.).
As regards reading instructional methods,
1
teacher-reported initial (grade 1) reading
instruction materials were classified into code-emphasis approaches, whole-word approaches,
or mixed code and whole word approaches. No significant difference in the distribution of
these methods was found across the subgroups (χ
2
<1.0, n.s.). Half of the methods were
classified as code emphasis, close to 40% whole word approaches, and the remainder, mixed.
Summarizing, the background data reported in Table 1suggest that the distinct reading
and cognitive–linguistic profiles for the three subgroups reported below are not confounded
by biosocial, demographic, or instructional factors.
1
Data was collected in 2003 before the recommendations of the “steering committee for reform of reading
instruction in literacy education”in Israel were implemented.
74 M. Shany, D.L. Share
Validation of the reading deficit subgroups
The validity of the present subdivision into reading disability subtypes depends on showing that
the nature of the reading performance of these subgroups and their verbal/linguistic profiles are
unique and distinctive on variables other than the measures used to define them. For this
purpose, we again conducted two varieties of analyses. First, we compared each sub-group’s
mean performance to national norms using single sample ttests with alpha set at 0.01. Second,
we examined differences among the three dyslexic subgroups by performing one-way
analysis of variance while controlling alpha at the level of theoretically coherent sets (or
families) of variables (i.e., decoding and word reading, text reading, verbal memory, RAN,
morphological/grammatical knowledge, and phonological awareness.) For each of the
variables in these six sets, we divided the alpha level for the overall analysis of variance
(ANOVA) by the number of variables in the set. Thus, for the two-variable set of
morphological/grammatical knowledge, alpha was set at 0.025 (0.05/2), for the three-variable
sets (decoding and word reading, text reading, phonological awareness, and verbal memory)
at 0.0167 (0.05/3) and the four-variable (RAN) set at 0.0125 (0.05/4). Significant overall tests
were then followed up with post hoc comparisons using the Bonferroni correction.
Performance of the subgroups on reading measures other than those used for group
selection
Table 2presents the descriptive statistics for the reading measures for the three subgroups
as well as the national norms. Data are presented first for accuracy then for rate. Our main
aim here was to determine whether the deficits of the two single-deficit groups (accuracy
disabled and rate disabled) remain selective and stable across measures other than those
used to define the groups.
Decoding and reading isolated words: accuracy
The subgroups were initially defined on a measure of isolated word reading. Here, we
examine performance on three independent measures of isolated word and pseudoword
reading: consonant–vowel (CV) sub-syllables, pseudowords, and pointed words.
Both the accuracy-disabled and accuracy-plus-rate disabled groups (but not the rate-
disabled group) performed significantly below the national norms (as indicated by the carat
symbol) on all three measures in this set. In addition, all three omnibus tests indicated
significant differences between the sub-groups. On both the CV and pseudoword reading
tests, post hoc comparisons revealed that both the accuracy-disabled and rate-plus-
accuracy-disabled subgroups were significantly less accurate than the rate-disabled group
and did not differ from each other. On the test of reading isolated pointed words (taken from
text), the accuracy-disabled and accuracy-plus-rate disabled groups performed significantly
below the national norm whereas the rate-disabled subgroup matched (indeed, slightly
exceeded) this norm. The doubly-disabled group, but not the accuracy-only subgroup, was
inferior to the rate-disabled group. Consistent with the original word decoding selection
criteria (see Table 1), the accuracy-plus-rate-disabled subgroup scored even more poorly
than the accuracy-disabled subgroup, pointing to more severe group-wise decoding deficits.
Turning to the set of text-reading measures, both the accuracy-disabled and accuracy-plus-
rate disabled groups scored below the national norm on all three measures. Confirming the
selective nature of the rate-disabled group’s deficit, their accuracy was consistently close to the
national norm, even on unpointed text. The accuracy-disabled group scored significantly below
Subtypes of reading disability in a shallow orthography 75
Table 2 Descriptive statistics across reading measures according to deficit subgroups: Comparison to national norms via one-sample t-test and comparisons between subgroups
via independent samples t-test
Variable Group
National norms
(n= 386)
Rate-disabled
(n= 38)
Accuracy-disabled
(n= 36)
Rate-plus-accuracy
disabled (n= 51)
M (SD) M (SD) M (SD) M (SD)
% Accuracy
Decoding
Consonants and vowels –63 items 72.6 (15.9) 73.6 (13.9)
a
60.2 (14.9)
b
†58.8 (14.3)
b
†
Pseudowords - 33 items 61.3 (22.7) 61.2 (19.2)
a
41.2 (15.8)
b
†36.5 (19.7)
b
†
Word reading
Pointed isolated words from text- 50 words 92.7 (10.5) 93.8 (3.9)
a
87.9 (7.5)
b
†76.7 (19.5)
c
†
Text reading
Pointed text for Grade 4 - 196 words 95.2 (5.5) 94.6 (3.3)
a
92.7 (3.2)
b
†87.4 (10.2)
c
†
Pointed text for grade 4- 100 words (an accelerated test) 94.2 (7.3) 94.6 (2.9)
a
90.4 (5.7)
b
†83.1(13.2)
c
†
Unpointed text for grade 4 - 100 words 90.5 (9.9) 89.4 (6.1)
a
85.7 (6.6)
a
†75.3 (16.1)
c
†
Rate (words per minute - wpm)
Decoding
Consonants and vowels - 63 items 49.5 (16.9) 43.2 (16.6)
a
†50.8 (21.4)
b
40.0 (16.4)
a
†
Pseudo-words - 33 items 20.8 (7.7) 14.4 (3.8)
a
†24.2 (8.7)
b
14.6 (4.7)
a
†
Word reading
Pointed isolated words from text- 50 words 72.3 (25.4) 54.7 (19.9)
a
†73.3 (922.8)
b
38.9 (20.7)
c
†
Text reading
Pointed text for grade 4 - 196 words 92.4 (31.9) 75.8 (42.5)
a
†94.4 (36.1)
b
60.7 (31.7)
c
†
Pointed text for grade 4- 100 words (an accelerated test) 83.8 (28.7) 61.5 (19.9)
a
†80.5 (18.1)
b
45.8 (19.2)
c
†
Unpointed text for grade 4 - 100 words 68.4 (24.4) 46.3 (15.4)
a
†69.9 (17.7)
b
38.3 (17.5)
a
†
76 M. Shany, D.L. Share
the rate-disabled group on the two pointed measures whereas the rate-plus-accuracy-disabled
group once again evinced a more severe accuracy deficit than the accuracy-only subgroup.
Reading rate
The results for reading rate paint a similar picture to the accuracy data. The mean reading
rate of the rate-disabled subgroup was significantly below national norms on all three
isolated word reading measures, and significantly below the accuracy-disabled subgroup’s
performance on two of the three measures. (The overall test for consonants and vowels did
not attain significance.) The accuracy-only subgroup scored fractionally above the national
norm on each of these three word reading measures. The rate-plus-accuracy-disabled
subgroup displayed a reading rate deficit comparable to the rate-disabled subgroup on the
pseudoword measure, but, paralleling the accuracy data, showed an even more severe
deficit on the isolated word reading measure.
The same pattern of findings emerged on the text reading measures. The rate-disabled and rate-
plus-accuracy-disabled groups were significantly below national norms on all three measures
whereas the accuracy-disabled group displayed normal reading rates. All three overall tests were
significant; follow-up post hoc comparisons revealed that the rate-only disabled group were
significantly slower than the accuracy-disabled subgroup on two of these measures (the 100-word
pointed text and the unpointed text), with the doubly-disabled subgroup slower still.
To summarize, the aim of the analyses presented in Table 2was to validate the identity of
the subgroups. Although the subgroups were selected on the basis of only two measures—
accuracy and rate for the same set of isolated pointed words, the data clearly showed that the
selectivity of the respective deficit in each single-deficit subgroup was robust across a range
of reading measures. On almost all measures, the rate-disabled readers were selectively
slower despite normal accuracy, and their accuracy extended from pointed to unpointed
words. The accuracy-disabled subgroup showed the opposite profile; inaccurate reading but
with normal reading rates. The doubly-disabled subgroup tended to be the most severely
disabled readers with the lowest accuracy and reading rates, particularly for word reading.
Performance of the reading-disabled subgroups on verbal /linguistic measures
Table 3presents the descriptive statistics for the cognitive/linguistic measures for the three
subgroups as well as the national norms.
Verbal memory
The accuracy-disabled subgroup performed significantly below the norm on story memory
and verbal fluency tests but not on working memory. Thus, there is some evidence to
support the idea that this group has a non-rate-related, general verbal memory deficit. There
is no reading or rate of processing component in these memory tests. Accordingly, on all
three measures, the rate-disabled subgroup performed normally. The accuracy-plus-rate-
disabled subgroup performed significantly below the norms on all memory tests. None of
the three overall ANOVA tests reached significance.
Speed of processing
Both groups with rate deficits were significantly slower than the norm on rapid naming of
letters, numbers, and objects, while the accuracy-only group performed normally. The
Subtypes of reading disability in a shallow orthography 77
Table 3 Descriptive statistics across cognitive/linguistic measures according to deficit subgroups: Comparison to national norms via one-sample t-test and comparisons between
subgroups via independent samples t-test
Variable Group
National norms
(n= 386)
Rate-disabled
(n= 38)
Accuracy-disabled
(n= 36)
Rate & Accuracy
disabled (n= 51)
M (SD) M (SD) M (SD) M (SD)
Verbal Memory
Verbal fluency (# words retrieved for three letters) 17.2 (5.8) 17.0 (6.4)
a
14.7 (4.8)
a
†14.6 (5.5)
a
†
Story memory - factual details (% accuracy) 77.9 (22.9) 75.5 (20.8)
a
64.2 (27.1)
a
†64.9 (26.6)
a
†
Working memory (number of words and correct sequences retrieved) 14.6 (8.5) 12.8 (7.3)
a
12.3 (6.6)
a
12.3 (6.1)
a
†
Speed of processing (Rapid Automatized Naming - RAN)
RAN - alphanumeric - numbers & letters (IPM) 110.9 (18.8) 103.9 (16.9)
ab
†111.4 (15.5)
a
97.8 (23.1)
b
†
RAN-objects - 50 items (IPM) 70.2 (15.0) 64.4 (10.2)
a
†68.0 (15.7)
a
65.0 (18.6)
a
†
RAS-alternate stimulus - 50 items (IPM) 82.1 (16.5) 72.8 (12.5)
ab
†81.3 (18.3)
a
70.8 (16.3)
b
†
Morphological knowledge -18 items (% accuracy) 87.3 (12.9) 86.5 (12.3)
a
79.2 (13.6)
b
†73.3 (16.3)
b
†
Grammatical knowledge - 20 items (% accuracy) 82.4 (15.4) 81.6 (13.2)
a
77.2 (17.9)
a
69.3 (22.8)
b
†
Phonological Awareness
Syllable deletion - 14 items (% accuracy) 87.9 (16.7) 89.8 (11.9)
a
78.2 (20.8)
b
†71.5 (26.2)
b
†
Phoneme deletion - 16 items (% accuracy) 71.9 (25.7) 72.5 (18.6)
a
54.4 (25.1)
b
†45.5 (29.2)
b
†
Synthesis of phonemes to words - 8 words (% accuracy) 75.4 (25.4) 70.0 (28.7)
a
59.9 (28.8)
a
†56.7 (30.1)
a
†
Analyses of words into phonemes - 8 words (% accuracy) 45.8 (34.2) 44.6 (32.2)
a
30.9 (36.4)
a
†27.2 (29.0)
b
†
IPM- Items per minute
78 M. Shany, D.L. Share
overall ANOVA attained significance in both the RAN letters and RAS tests; follow-up
comparisons showed that, in both cases, the rate-plus-accuracy subgroup were inferior to
the accuracy-disabled subgroup, with the rate-only subgroup lying in between.
Morphological knowledge
Children with accuracy deficits performed significantly below the norm, and also below the
rate-disabled group, who performed normally. The pair-wise difference between the
accuracy-disabled and rate-disabled subgroups approached significance (p=0.09)
Grammatical knowledge
The analyses indicate that only the doubly-disabled group performed significantly below
the norm and below the two singly-disabled groups. The accuracy-disabled group
performed lower than the rate-disabled group, but this difference was not significant.
Phonological awareness
Both groups with accuracy deficits had a similar impairment in phonological awareness in
comparison to the norm. The doubly-disabled subgroup performed below the norm on all
measures and the accuracy-disabled subgroup performed below the norm on measures of
deletion, and synthesis and analysis of phonemes. The rate-disabled subgroup performed
normally across a range of phonological awareness tests. Significant overall tests for each
of the four measures followed-up by Bonferonni post hoc comparisons revealed that the
rate-disabled readers were superior to both accuracy-disabled subgroups on phoneme
deletion, with the difference on syllable deletion approaching significance (p=0.06).
Overall, examination of cognitive/linguistic abilities in our groups supports the idea that
the accuracy-specific deficit is reflected in a deficient knowledge base in verbal memory,
morphological knowledge, and phonological awareness, but not RAN rate-of-processing.
Dyslexics with selective rate deficits tended to have problems of speed of processing but
not in the fidelity of their verbal/linguistic knowledge base in the two domains that have
been shown to be the most critical in Hebrew reading—phonology and morphology.
Discussion
The current study provides converging evidence in support of a subtyping scheme based on
selective impairments in two aspects of reading performance: rate and accuracy. The pattern
of performance revealed by our nationally representative sample of Israeli fourth-graders
points to a double dissociation resulting in subgroups delineated by intact performance in
one dimension co-occurring with a reliable deficit in the other dimension. Our data are
especially compelling regarding intact performance on the unimpaired dimension because
we compared sub-group performance to norms obtained from the same (co-eval) nationally
representative cohort. These results are further reinforced by the fact that biosocial,
demographic, and instructional factors did not explain the reported group differences in rate
and accuracy, and by the finding that the subgroups differed not only on a validation set of
reading measures not used to define the subgroups, but also on a number of cognitive/
linguistic variables that converge with selective deficits in speed-of-processing on the one
hand, and deficient linguistic knowledge on the other.
Subtypes of reading disability in a shallow orthography 79
Our results, therefore, strengthen and extend previous studies (Lovett 1984a,1984b,
1987; Leinonen et al., 2001) proposing a distinction between rate and accuracy in reading
disability subtyping. While previous studies have identified rate-disabled and accuracy-
disabled subgroups among reading-disabled children (Lovett, 1984a,1984b,1987) and
adults (Leinonen et al., 2001), in none of these cases have the specific deficits in one
dimension been reported alongside intact performance in the other. Lovett (1984a,1984b,
1987) identified a rate-only disabled sub-group with intact levels accuracy, but her
accuracy-disabled group also showed deficits in reading rate. The “hasty”and “hesitant”
subgroups of Leinonen et al. (2001) each displayed impairment of differing degrees on both
rate and accuracy.
Employing standardized measures for identifying disabled readers, the present study
revealed three distinct subgroups: a rate-disabled group performing, on average, at the 15th
percentile with respect to reading rate while performing well the within normal range on
accuracy; an accuracy-disabled group with mean performance at the 15th percentile level
on accuracy yet maintaining normal reading rate; and a doubly-disabled group, performing
very poorly on both rate and accuracy measures. Thus, the single-deficit subgroups
represent the first report of truly “hard cases”of selective reading breakdown in terms of a
dissociation between reading rate and accuracy, as opposed to merely “soft cases”of
relative strengths and weaknesses (see Stanovich & Siegel, 1994). Furthermore, based on
our nationally representative sample, each of these groups appears to be equally prevalent
in the population, with rates of 9.8%, 9.3%, and 13.2%, respectively, or around one
exemplar of each of the three subtypes in every ten readers.
The present findings support the idea that reading rate-disabled and accuracy-disabled
subgroups represent true reading disability subtypes, presumably with different, though
possibly related, etiologies and underlying mechanisms. Further support for this
interpretation is provided by the finding that the two single-deficit groups differed not
only with respect to the accuracy and rate measures by which they were defined, but also
with respect to their broader cognitive–linguistic profiles. While the rate disability appeared
to reflect a general deficit in speed of processing, affecting reading rate and rapid
automatized naming of print-related material, the accuracy disability went hand-in-hand
with linguistic deficits affecting phonemic awareness and morphological knowledge. Thus,
the dissociation between the groups appears to extend beyond domain-specific reading
deficits to non-reading cognitive/linguistic processes.
Although the current investigation successfully demarcated two single-deficit reading
disability subgroups, these data only allow conclusions to be drawn with regard to children
who are beyond the initial stage of code instruction and demonstrate mastery of a shallow
orthography (Share & Levin, 1999). Additional studies in other shallow as well as less
transparent orthographies will be required to fully characterize the subgroups in terms of
development and perhaps shed light on contributing etiological factors, employing
longitudinal designs as well as examining different age groups cross-sectionally.
Differential response to individualized instruction and intervention will be an important
ingredient in this work too.
We conclude with a consideration of the possibility (voiced by one of our anonymous
reviewers) that the profiles of our two single-deficit subgroups may not represent true
disability in the sense of deficits in basic or underlying processes but rather idiosyncratic
reading strategies or styles arising from one or more non-cognitive variables such as
personality characteristics (e.g., reflective/impulsive cognitive style), instructional methods/
environmental factors or even transient situational/motivational factors. By this account, it
might be questioned whether our subgroups could be truly classified as dyslexic subtypes in
80 M. Shany, D.L. Share
the sense that a different underlying or basic cognitive deficit explains different and unique
patterns or profiles or (disordered) reading behavior.
As Castles and Coltheart (1993) have pointed out, the common denominator among
reading strategies/reading styles accounts is that these behaviors are separate from basic
underlying cognitive processes of the type appearing in most causal accounts of dyslexia.
Our data clearly refute this position because we found basic and distinct non-reading
deficits consistent with double-dissociation at the level of reading processes. The accuracy-
disabled group were deficient on RAN objects, not just on the RAN alphanumeric (digits
and letters) subtests that some regard as just expressions (or even consequences) of reading
experience. In addition, the impairments observed in the accuracy-only subgroup alone on
morphological inflections and derivations, as well as pronominal reference reflect
knowledge of common spoken-language forms not literate, written-language forms such
as straightforward verb inflections (e.g., run runs, running). The phonological awareness
results provide further evidence for basic underlying deficits. Although phoneme awareness
has been construed by some, (including the second author, see Share, 1995), as a reading-
specific capability, the syllable deletion task which is known to be literacy-independent (see
Morais, Alegria, & Content, 1987; Share, 1995), revealed the accuracy-disabled subgroup
to be significantly below norms, whereas the rate-disabled actually scored slightly above
the national average. There is general agreement that access to syllables (but not phonemes)
is not primarily dependent on the experience of learning to read an alphabetic orthography
(Morais et al. 1987; Share, 1995). So here again, we have another independent stratum of
evidence, one that neatly dovetails with the nature of reading speed/accuracy profiles.
Specifically, our data indicate basic cognitive deficits that match the rate-deficient profile of
the rate-disabled group who show proficiency in their knowledge base on a host of
variables including phonological awareness and morphological knowledge. It is worth
emphasizing that the lack of deficits on the accuracy/knowledge measures constitutes yet
another strength of the double-dissociation pattern. By the same token, the accuracy-
disabled subgroup exhibited knowledge deficiencies that dovetail with their accuracy-
impaired profile and, furthermore, on the two variables (morphology and phonological
awareness) that have been shown to be the strongest correlates of reading difficulties in
Hebrew (Share & Levin, 1999; Share, 2003) and also its Semitic cousin—Arabic (see Abu-
Rabia, Share, & Mansour, 2003). Thus, the double dissociation observed on the initial word
reading measure used to define our sub-groups, and which was corroborated on additional
reading measures not used to define the sub-groups, was further confirmed by the
converging data from our cognitive/linguistic measures. In this sense then, our data provide
a double validation of sub-group validity.
Further counterevidence against the strategies/styles account comes from the fact that we
did not find reliable differences in instructional methods that are often cited as the source of
different reading styles or strategies. Both types of readers were found in phonic-emphasis
and whole-word/meaning emphasis classrooms with similar frequency. A host of other
variables were also dismissed as potential non-cognitive sources of reading styles/strategies
including gender, socio-economic status, migrant status, religiosity and more.
We must nonetheless concede that the case for double-dissociation is not entirely “open-
and-shut”. Most, but not all the variables in a given set displayed the desired pattern. To be
sure, in comparison with national norms, six out of six reading accuracy scores among the
accuracy-disabled were significantly below average compared to six out of six that were
comparable to norms in the rate-disabled subgroup. The same uniformity was also found for
reading rate. However, the head-to-head comparisons between accuracy-only and rate-only
sub-groups were not uniformly significant; for both rate and accuracy, only four of the six
Subtypes of reading disability in a shallow orthography 81
in each set were significant, thus the weight of evidence could be characterized as positive
but not overwhelming. Some of our measures had less than ideal reliability and this can
obscure group differences. There was also substantial variability within sub-groups
indicating considerable overlap in performance levels. And most reading accuracy measures
were close to ceiling levels (as is the reality of a consistent orthography) which necessarily
attenuates group discrimination. In this context, it is worth commenting that this problem
provides a telling illustration of the importance of using fluency in the assessment of
developmental and individual differences: Accuracy levels in a regular orthography, even
among our accuracy-disabled dyslexics were extremely high, over 90% for word reading in
grade-appropriate connected text. Even when we defined our dyslexic subgroups according
to a more stringent 16th percentile criterion (i.e., one or more standard deviations below the
national mean) the accuracy-disabled subgroup achieved 91% accuracy when required to
read aloud a (self-paced) 196-word grade-level pointed text (the rate-disabled dyslexic
subgroup matched the 95% national norm). These data emphasize the importance of
fluency as a benchmark measure of reading performance—no less important than accuracy.
Our subtypes do not represent idiosyncratic reading styles or strategies but a fundamental
problem in achieving the word reading speed and automaticity necessary for reading
comprehension (Perfetti, 1985).
A growing number of countries (English-speaking included) are now adopting dyslexia
definitions that place reading speed on an equal footing with reading accuracy (British
Psychological Society, 1999, Lyon, Shaywitz, & Shaywitz, 2003, Tunmer & Greaney,
2010). The approach adopted in the present study offers a useful way of operationalizing
this new generation of dyslexia definitions and, furthermore, proposes a subtyping scheme
with the potential for wide cross-linguistic application across both consistent and
inconsistent orthographies.
References
Abu-Rabia, S., Share, D. L., & Mansour, M. S. (2003). Word recognition and basic cognitive abilities among
reading-disabled and normal readers in Arabic. Reading and Writing, 16, 423–442.
Bailey, C. E., Manis, F. R., Pedersen, W. C., & Seidenberg, M. S. (2004). Variation among developmental dyslexics:
Evidence from a printed-word-learning task. Journal of Experimental Child Psychology, 87, 125–154.
Beaton, A. A. (2004). Dyslexia, reading and the brain: A sourcebook of psychological and biological
research. Hove and New-York: Psychology Press.
Boder, E. (1971). Developmental dyslexia: A diagnostic screening procedure basedon three characteristic
patterns of reading and spelling. Journal of Learning Disabilities, 4, 297–342.
Breznitz, Z. (1997). The effect of accelerated reading rate on memory for text among dyslexic readers.
Journal of Educational Psychology, 89, 287–299.
British Psychological Society. (1999). Dyslexia, literacy and psychological assessment: Report of a working
party of the division of educational and child psychology of the British Psychological Society. Leicester,
England: Author.
Castles, A., & Coltheart, M. (1993). Varieties of developmental dyslexia. Cognition, 47, 149–180.
Coltheart, M., Rastle, K., Perry, C., Langdon, R., & Ziegler, J. (2001). A dual route cascaded model of visual
word recognition and reading aloud. Psychological Review, 108, 204–256.
Cossu, G. (1999a). Biological constraints on literacy acquisition. Reading and Writing, 11, 213–237.
Cossu, G. (1999b). The acquisition of Italian orthography. Learning to read and write: A cross-linguistic
perspective. In learning to read and write: A cross-linguistic perspective. Cambridge studies in cognitive
and perceptual development (pp. 10–33). New York, USA: Cambridge University Press.
Csepe, V. (2006). Literacy acquisition and dyslexia in Hungarian. In R. M. Joshi & P. G. Aaron (Eds.),
Handbook of orthography and literacy (pp. 231–247). Mahwah, NJ: Erlbaum.
Daneman, M., & Carpenter, P. A. (1980). Individual-differences in working memory and reading. Journal of
Verbal Learning and Verbal Behavior, 19, 450–466.
82 M. Shany, D.L. Share
Daniels, P. T., & Bright, W. (1996). The world’s writing systems. New York: Oxford University Press.
De Jong, P. F., & van der Leij, A. (1999). Specific contributions of phonological abilities to early reading acquisition:
Results from a Dutch latent variable longitudinal study. Journal of Educational Psychology, 91, 450–476.
De Jong, P. F., & van der Leij, A. (2003). Developmental changes in the manifestation of phonological deficit in
dyslexic children learning to read a regular orthography. Journal of Educational Psychology, 95,22–40.
Fletcher, J. M., Morris, R., Lyon, G. R., Stuebing, K. K., Shaywitz, S. E., Shankweiler, D. P., et al. (1997).
Subtypes of dyslexia: An old problem revisited. In B. A. Blachman (Ed.), Foundations of reading:
Acquisition and dyslexia (pp. 95–114). Mahwah, NJ: Erlbaum.
Genard, N., Mousty, P., Content, A., Alegria, J., Leybaert, J., & Morais, J. (1998). Methods to establish
subtypes of developmental dyslexia. In P. Reitsma & L. Verhoeven (Eds.), Problems and interventions in
literacy development (pp. 163–176). Dordrecht: Kluwer.
Harm, M. W., & Seidenberg, M. S. (1999). Phonology, reading acquisition, and dyslexia: Insights from
connectionist models. Psychological Review, 106, 491–528.
Ho, C. S. H., Chan, D. W., Chung, K. H., Lee, S.-H., & Tsang, S. M. (2007). In search of subtypes of
Chinese developmental dyslexia. Journal of Experimental Child Psychology, 97,61–83.
Hutzler, F., Ziegler, J. C., Perry, C., Wimmer, H., & Zorzi, M. (2004). Do current connectionist learning
models account for reading development in different languages? Cognition, 91, 273–296.
Jackson, N. E., & Coltheart, M. (2001). Routes to reading success and failure: Toward an integrated
cognitive psychology of atypical reading. Hove, UK: Psychology Press.
Joshi, R. M., & Aaron, P. G. (2006). Handbook of orthography and literacy. Mahwah, NJ: Erlbaum.
Leinonen, S., Muller, K., Leppanen, P. H. T., Aro, M., Ahonen, T., & Lyytinen, H. (2001). Heterogeneity in
adult dyslexic readers: Relating processing skills to the speed and accuracy of oral text reading. Reading
and Writing., 14, 265–296.
Leppanen, U., Niemi, P., Aunola, K., & Nurmi, J. E. (2006). Development of reading and spelling Finnish
from preschool to grade 1 and grade 2. Scientific Studies of Reading, 10,3–30.
Lovett, M. W. (1984a). The search for subtypes of specific reading disability: Reflections from the cognitive
perspective. Annals of Dyslexia, 34, 155–178.
Lovett, M. W. (1984b). A developmental perspective on reading dysfunction: accuracy and rate criteria in the
subtyping of dyslexic children. Brain & Language, 22,67–91.
Lovett, M. W. (1987). A developmental approach to reading disability: Accuracy and speed criteria of normal
and deficient reading skill. Child Development, 58, 234–260.
Lundberg, I., & Hoien, T. (1990). Patterns of information processing skills and word recognition strategies in
developmental dyslexia. Scandinavian Journal of Educational Research, 34, 231–240.
Lyon, G. R., Shaywitz, S. E., & Shaywitz, B. A. (2003). A definition of dyslexia. Annals of Dyslexia, 53,1–14.
Lyon, R., Stewart, N., & Freedman, D. (1982). Neuropsychological characteristics of empirically derived
subgroups of learning disabled readers. Journal of Clinical Neuropsychology, 4, 343–365.
Lyon, R., & Watson, B. (1981). Emprically derived subgroups of learning disabled readers: Diagnostic
characteristics. Journal of Learning Disabilities, 14, 256–261.
Lyytinen, H., Aro, M., & Holopainen, L. (2004). Dyslexia in highly rthographically regular Finnish. In I.
Smythe, J. Everatt, & R. Slater (Eds.), International book of dyslexia: A cross-language comparison and
practice guide (pp. 81–91). Chichester, England: Wiley.
Manis, F. R., Seidenberg, M. S., Doi, L. M., & McBride-Chang, C. (1996). On the bases of two subtypes of
development dyslexia. Cognition, 58, 157–195.
Mattis, S., French, J. F., & Rapin, I. (1975). Dyslexia in children and young adults: Three independent
syndromes. Developmental Medicine and Child Neurology, 17, 150–163.
Milne, R. D., Nicolson, T., & Corballis, M. C. (2003). Lexcial access and phonological decoding in adult
dyslexic subtypes. Neuropsychology, 17, 362–368.
Morais, J., Alegria, J., & Content, A. (1987). The relationships between segmental analysis and alphabetic
literacy: An interactive view. Cahiers de Psychologie Cognitive, 7, 415–443.
Morris, R. D., Stuebing, K. K., Fletcher, J. M., Shaywitz, S. E., Lyon, G. R., Shankweiler, D. P., et al. (1998).
Subtypes of reading disability: Variability around a phonological core. Journal of Educational
Psychology, 90, 347–373.
National Reading Panel. (2000). Teaching children to read: An evidence-based assessment of the scientific
research literature on reading and its implications for reading instruction. Washington, DC: National
Institute for Child Health and Human Development.
Nikolopoulos, D., Goulandris, N., Hulme, C., & Snowling, M. J. (2006). The cognitive bases of learning to
read and spell in Greek: Evidence from a longitudinal study. Journal of Experimental Child Psychology,
94,1–17.
Perfetti, C. A. (1985). Reading ability. New York: Oxford University Press.
Subtypes of reading disability in a shallow orthography 83
Porpodas, C. (2006). Literacy acquisition in Greek: Research review of the role of phonological and
cognitive factors. In R. M. Joshi & P. G. Aaron (Eds.), Handbook of orthography and literacy (pp. 189–
199). Mahwah, NJ: Erlbaum.
Ravid, D. (2005). Hebrew orthography and literacy. In R. M. Joshi & P. G. Aaron (Eds.), Handbook of
orthography and literacy (pp. 339–364). Mahwah, NJ: Erlbaum.
Rayner, K., Foorman, B. R., Perfetti, C. A., Pesetsky, D., & Seidenberg, M. S. (2001). How psychological
science informs the teaching of reading. Psychological Science in the Public Interest, 2,31–73.
Seymour, P. H. K., Aro, M., & Erskine, J. M. (2003). Foundation literacy acquisition in European
orthographies. British Journal of Psychology, 94, 143–174.
Shankweiler, D., & Fowler, A. E. (2004). Questions people ask about the role of phonological processes in
learning to read. Reading and Writing, 17, 483–515.
Shankweiler, D., & Liberman, I. Y. (1989). Phonology and reading disability: Solving the reading puzzle.
Ann Arbor: University of Michigan Press.
Shany, M., Lachman, D., Shalem, Z., Bahat, A., & Zieger, T. (2006). “Aleph-Taph”—A test for the diagnosis
of reading and writing disabilities, based on national Israeli norms. Tel Aviv: Yesod Publishing.
Share, D. L. (1995). Phonological recoding and self-teaching: Sine qua non of reading acquisition.
Cognition, 55, 151–218.
Share, D. L. (2003). Dyslexia in Hebrew. In N. Goulandris (Ed.), Dyslexia in different languages: Cross-
linguistic comparisons (pp. 208–234). London: Whurr.
Share, D. (2008). On the Anglocentricities of current reading research and practice: The perils of over
reliance on an “outlier”orthography. Psychological Bulletin, 134, 584–615.
Share, D. L., & Levin, I. (1999). Learning to read and write in Hebrew. In M. Harris & G. Hatano (Eds.),
Learning to read and write (pp. 89–111). Cambridge: Cambridge University Press.
Share, D. L., & Stanovich, K. E. (1995). Cognitive processes in early reading development: accommodating
individual differences into a model of acquisition. Issues in Education, 1,1–57.
Shatil, E., & Share, D. L. (2003). Cognitive antecedents of early reading ability: A test of the modularity
hypothesis. Journal of Experimental Child Psychology, 86,1–31.
Shimron, J. (1993). The role of vowels in reading: A review of studies of English and Hebrew. Psychological
Bulletin, 114,52–67.
Siegel, L. S. (1999). Issues in the definition and diagnosis of learning disabilities: A perspective on
Guckenberger vs. Boston University. Journal of Learning Disabilities, 32, 304–319.
Snowling, M. J., & Hulme, C. (2005). The science of reading: A handbook. Oxford, UK: Blackwell.
Stanovich, K. E. (1991). Discrepancy definitions of reading disability: Has intelligence led us astray?
Reading Research Quarterly, 26,7–29.
Stanovich, K. E. (2000). Progress in understanding reading: Scientific foundations and new frontiers. New
York: Guilford Press.
Stanovich, K. E., & Siegel, L. S. (1994). Phenotypic performance profile of children with reading disabilities:
A regression-based test of the phonological-core variable-difference model. Journal of Educational
Psychology, 86,24–53.
Stanovich, K. E., Siegel, L. S., & Gottardo, A. (1997a). Converging evidence for phonological and surface
subtypes of reading disability. Journal of Educational Psychology, 89,114–127.
Stanovich, K. E., Siegel, L. S., Gottardo, A., Chiappe, P., & Sidhu, R. (1997b). Subtypes of developmental dyslexia:
Differences in phonological and orthographic coding (pp 114–139).InB.Blachman(Ed.),Foundations of
reading acquisition and dyslexia: Implications for early intervention. Mahwah, NJ: Erlbaum.
Tunmer,W., & Greaney, G. (2010). Defining dyslexia. Journal of Learning Disabilities, 43, 229–243.
Vellutino, F. R., Fletcher, J. M., Snowling, M. J., & Scanlon, D. M. (2004). Specific reading disability
(dyslexia): What have we learned from the past four decades? Journal of Child Psychology and
Psychiatry, 45,2–40.
Wimmer, H. (1993). Characteristics of developmental dyslexia in a regular writing system. Applied
Psycholinguistics, 14,1–33.
Wolf, M., & Bowers, P. G. (1999). The double-deficit hypothesis for the developmental dyslexia. Journal of
Educational Psychology, 91, 415–438.
Yap, R., & van der Leij, A. (1993). Word processing in dyslexics: An automatic decoding deficit? Reading
and Writing, 5, 261–279.
Ziegler, J. C., & Goswami, U. (2005). Reading acquisition, developmental dyslexia and skilled reading
across languages: A psycholinguistic grain size theory. Psychological Bulletin, 131,3–29.
Zoccolotti, P., De Luca, M., Di Pace, E., Judica, A., Orlandi, M., & Spinelli, D. (1999). Markers of
developmental surface dyslexia in a language (Italian) with high grapheme-phoneme correspondence.
Applied Psycholinguistics, 20, 191–216.
84 M. Shany, D.L. Share
