You are on page 1of 10

c o r t e x 4 6 ( 2 0 1 0 ) 1 2 3 8 e1 2 4 7

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/cortex

Special issue: Research report

Developmental dissociations between lexical reading and


comprehension: Evidence from two cases of hyperlexia

Anne Castles a,*, Alison Crichton b and Margot Prior b


a
Macquarie Centre for Cognitive Science, Macquarie University, Sydney, Australia
b
Department of Psychology, University of Melbourne, Australia

article info abstract

Article history: We report two cases of developmental hyperlexia e JY and AD e who performed at normal
Received 10 October 2008 levels or above in converting print into speech, but who were very impaired in spoken and
Reviewed 15 January 2009 written word comprehension. Our investigations focussed on whether these cases
Revised 27 February 2009 displayed evidence for normal acquisition of lexical reading skills, as indexed by unim-
Accepted 21 August 2009 paired performance for age in reading aloud a set of irregular words, despite poor acqui-
Published online 8 July 2009 sition of semantic knowledge of the same words. In both cases, this dissociation was
evident. The pattern of results was also demonstrated at an item level: the two cases
Keywords: showed no significant differences in reading accuracy for irregular words which they could
Hyperlexia define than for those which they could not. The results provide further evidence for the
Nonsemantic reading existence of a direct-lexical route from orthography to phonology, which is not necessarily
Lexical route mediated by semantic knowledge.
Reading development ª 2010 Elsevier Srl. All rights reserved.

Developmental disorders of reading primarily involve dissociations shown by these individuals has the potential to
impairments in converting printed inputs into their spoken inform both cognitive models of reading and theories of
form, and are typically characterised in terms of impaired reading acquisition, particularly in relation to the role of
functioning of orthographic (lexical) or phonological semantics. Indeed, although the term “hyperlexia” empha-
(nonlexical) reading processes, or both (e.g., Castles and sises the presence of precocious or advanced reading in
Coltheart, 1993; Manis et al., 1996). However, in develop- these cases, it is typically the semantic deficit that is the
mental hyperlexia individuals display a quite different pattern most pervasive symptom e regardless of their overall level of
of abnormal reading behaviour: they perform at normal reading attainment, these cases appear to show evidence of
levels or above in converting print into speech, but are very “reading without meaning”.
impaired in their spoken and written comprehension of In this paper, we draw on data from two cases of devel-
words (Aram, 1997; Aaron, 1989; Healy, 1982; Nation, 1999; opmental hyperlexia to explore dissociations between the
Seymour and Evans, 1992; Silberberg and Silberberg, 1967). acquisition of reading ability and general semantic knowledge
While there is some debate as to whether this syndrome (evident across different modalities), and specifically to
should be classified as a form of developmental dyslexia or examine to what degree knowledge of the meanings of words
not (Grigorenko et al., 2003; Nation, 1999), the pattern of is required for the development of lexical reading processes.

* Corresponding author. Macquarie Centre for Cognitive Science, Macquarie University, Sydney, NSW 2109, Australia.
E-mail address: acastles@maccs.mq.edu.au (A. Castles).
0010-9452/$ e see front matter ª 2010 Elsevier Srl. All rights reserved.
doi:10.1016/j.cortex.2010.06.016
c o r t e x 4 6 ( 2 0 1 0 ) 1 2 3 8 e1 2 4 7 1239

The role of semantic knowledge in reading aloud has been for the correct reading of nonwords. The second mechanism,
the subject of considerable debate, particularly in relation to the O-S-P pathway involves converting orthography to
the degree to which semantics is required for the successful phonology via semantics. Both pathways work in parallel;
reading of irregular words, like yacht or blood. In the prominent however, the O-S-P pathway makes a greater contribution to
Dual Route model of reading (Coltheart, 1984; Morton and performance when processing in the O-P pathway is slowed:
Patterson, 1980; Newcombe and Marshall, 1981) and its for example, when words with irregular or inconsistent
computational implementation, the Dual Route Cascaded orthographic-phonological mappings are being read. The PDP
model (DRC model; Coltheart et al., 1993, 2001), there are two model therefore predicts that the semantic reading pathway is
primary routes from print to speech (see Fig. 1): the nonlexical more important for reading irregular words than for reading
route involves applying a set of letter-sound correspondence regular words or nonwords and, indeed, in the implemented
rules to the printed input to generate a spoken output, and is model of Plaut et al. (1996) normal reading aloud of irregular
required for the successful reading of unfamiliar words or words was dependent on the integrity of this pathway.
nonwords, like glup. The lexical route involves accessing stored Evidence adjudicating between these two representations
representations of the orthographic form of known words, of the role of semantics in reading aloud has come largely
which in turn are linked to their phonological form, thus from cases of acquired semantic impairment, or semantic
allowing correct pronunciation of irregular words that cannot dementia. Semantic dementia is a disorder characterised by
be read correctly via letter-sound translation. Importantly, as a progressive loss of semantic memory in which other aspects
shown in Fig. 1, there are two proposed sub-routes of the of cognitive functioning remain intact (Hodges et al., 1992;
lexical route: one that proceeds via semantics e the lexical- Snowden et al., 1989). If semantic mediation is necessary for
semantic route e and one where spoken word representations the successful reading of irregular words, patients who lose
are accessed directly from their orthographic representations their ability to understand the meaning of irregular words
with no semantic mediation e the direct-lexical route. Thus, should also begin to read them incorrectly, displaying an
although irregular words may be read aloud via semantic acquired surface dyslexia. Consistent with this prediction,
mediation, according to the DRC model it is not necessary for a pattern of surface dyslexia was described in the first
this to be the case for successful lexical reading to take place. comprehensive account of patients with semantic dementia
In contrast, in the major competing model of reading (Warrington, 1975), and has been reported in several subse-
aloud, the Parallel Distributed Processing model (PDP: Plaut quent studies (e.g., Funnell, 1996; Graham et al., 1994;
et al., 1996; Seidenberg and McLelland, 1989), semantic Patterson and Hodges, 1992; Shallice et al., 1983; Woollams
mediation is proposed to be necessary for the successful et al., 2007). However, the pattern of reading impairments
reading of irregular words, at least those of low frequency. The shown by patients with semantic dementia is quite hetero-
PDP model proposes two mechanisms for converting print to geneous (McKay et al., 2007) and several cases of semantic
speech. The first mechanism, the O-P pathway, involves direct dementia have been described in which there was no corre-
computation of phonology from orthography and is required sponding surface dyslexia (Blazely et al., 2005; Cipolotti and
Warrington, 1995; Gerhand, 2001; Lambon Ralph et al., 1995;
Schwartz et al., 1980). As well, there are cases of acquired
phonological dyslexia where the ability to read aloud has been
demonstrated in the absence of nonlexical skills and
comprehension (Funnell, 1983). These latter cases provide
support for the DRC proposal that semantic activation may be
important, but not necessary, for the reading of irregular
words, given the availability of the direct-lexical route as an
alternative to the lexical-semantic route.
The relationship between reading performance and
comprehension in developmental hyperlexia represents
another potential source of input into this debate. Rather than
losing semantic knowledge, individuals with developmental
hyperlexia have never acquired it: typically they are children
or adults with moderate to severe intellectual delay and often
they have a co-morbid diagnosis of autism or Asperger’s
syndrome (Nation, 1999). If such individuals are shown to be
able to read aloud irregular words in the absence of knowledge
of their meanings, this would refute the hypothesis that
semantics necessarily mediates the successful reading of
irregular words, and would provide converging evidence for
the existence of a direct-lexical route for reading aloud.
Further, in contrast to the semantic dementia cases, data from
hyperlexia can potentially inform theories about the role of
Fig. 1 e The DRC model of reading aloud (based on semantic knowledge in the acquisition of reading processes.
Coltheart et al., 2001). The bolded arrow represents Even if it were established through evidence from semantic
the direct-lexical route. dementia that skilled readers’ ability to read irregular words
1240 c o r t e x 4 6 ( 2 0 1 0 ) 1 2 3 8 e1 2 4 7

can remain intact after they have lost semantic knowledge, it training study by Glosser et al. (1997), who describe the ability
does not automatically follow that novice readers’ ability to of a 10-year-old hyperlexic child, LA, to acquire lexical repre-
learn to read irregular words can be intact when they have not sentations without meaning. LA’s spelling and comprehen-
acquired such semantic knowledge. Evidence for this capacity sion of a set of low frequency irregular words was initially
in cases of hyperlexia would therefore uniquely address the tested, with spelling accuracy being found to be highly asso-
question of whether the direct-lexical route can develop ciated with comprehension. However, LA was subsequently
independently of the lexical-semantic route. able to be trained to spell the “semantically unknown” words
The provision of such evidence requires data of a very just as accurately as the “semantically known” words without
particular form: successful reading of a set of irregular words any further semantic information being provided. The authors
in a case of developmental hyperlexia must be demonstrated, concluded that “whole word representations can be acquired,
and an impairment of semantic knowledge for those same items stored and retrieved in the absence of a functional link to
must be shown. Although there have been several studies of semantic memory” (p. 234).
reading and semantic processing in cases of hyperlexia, none In summary, the DRC model of reading predicts that it is
to date has provided such evidence. A small number of studies possible for irregular words to be read aloud correctly in the
has explored irregular word reading in developmental absence of comprehension of those words, while the alter-
hyperlexia, but those studies did not examine comprehension native PDP model predicts that accurate reading of irregular
of the same items (Atkin and Lorch, 2006; Frith and Snowling, words will be unlikely to occur in the absence of compre-
1983; Glosser et al., 1996; Welsh et al., 1987). Conversely, the hension, at least for low frequency words. The dissociation
few studies that have explored reading and comprehension of between reading ability and semantic knowledge evident in
the same words in hyperlexic cases have not differentiated cases of developmental hyperlexia provides an ideal oppor-
the words by regularity (Siegel, 1984; Aram et al., 1984). These tunity to test these competing hypotheses, and also to
studies therefore do not allow for a specific examination of the examine whether normal acquisition of irregular word
functioning of the direct-lexical route in hyperlexia, since, on reading skills can occur in the context of impaired compre-
the DRC model, regular words can be read successfully via the hension. Yet, surprisingly, no study to date has systematically
nonlexical route. explored both reading accuracy and comprehension in
To provide some initial insights on this issue, we con- a single set of words specifically selected to be irregular in
ducted a post hoc examination of the data from those hyper- individuals with hyperlexia. In the present paper, we report on
lexia studies reporting both reading and comprehension data two cases of hyperlexia e JY and AD e where such an inves-
for the same items, to see if any of the words that were read tigation was carried out.
correctly without comprehension were in fact irregular. Siegel
(1984) assessed the pictureeword matching ability of a 6-year-
old hyperlexic girl, AE, using the Peabody Picture Vocabulary 1. Case reports
Test (PPVT) and then assessed her reading aloud of the same
items. AE showed a clear dissociation in performance, scoring 1.1. Case JY
in the 2nd percentile for her age on pictureeword matching
(raw score not provided) but reading aloud 67 of the first 72 JY, a boy, was 10 years 4 months old at the time of initial
items correctly. The semantically unknown items that she contact. He was in Year 4 at a mainstream primary school. JY’s
read correctly were; knee, empty, sail, tool, disappointment, mother reported no difficulties during pregnancy or birth and
awarding and vine. Of these, only awarding would be classified JY reached developmental motor and language milestones at
as irregular.1 Aram et al. (1984) used a similar methodology, appropriate ages. His mother reported that JY had been
comparing reading aloud and semantic knowledge for the interested in reading from an early age. He enjoyed reading
items on the PPVT in the case of MD, a 39-year-old hyperlexic aloud to other children when in kindergarten, and sought out
male. Again, MD showed a dissociation between reading aloud unusual print and fonts. For example, he used all kinds of toys
and comprehension, but the words reported to be read aloud and objects, including his own body, to form letters when he
correctly without comprehension all appeared to be regular: was young.
kangaroo, goggles, peacock, freckle and eagle. However, Aram et JY was assessed by a paediatrician at the age of 4 due to
al. also report MD’s reading aloud and pictureeword matching concerns about his gait and behaviour, particularly his social
performance on a second set of animal, body part and color interaction skills, and was found to score within the clini-
words from Schwartz et al. (1980). Here, he successfully read cally significant range on a screening test for high func-
aloud the words beige, fuchsia, and turquoise without showing tioning autism/Asperger’s syndrome. A cognitive assessment
evidence for comprehension, providing some support for the at 4 years 11 months with the Wechsler Pre-school and
use of the direct-lexical route in hyperlexia. Primary scale of Intelligence (WPPSI; Wechsler, 1989)
The only other evidence for acquisition and use of the revealed an IQ in the low average range (Performance: 87;
direct-lexical route in hyperlexia comes from an experimental Verbal: 83; Full Scale: 83). Although no gross language delay
was reported, qualitative difficulties in his language were
1
noted:
We define “regularity” according to the most frequently
occurring phoneme correspondence for a particular grapheme. In
some instances, context-sensitive rules may apply: for example, “(JY’s) memory for the shape and sound of letters and
the word “knee” would be classified as regular since “k” followed words are also indicative of good rote visual and auditory
by “n” is always silent in the first position of a word. memory skills. JY also gave adequate though rather literal
c o r t e x 4 6 ( 2 0 1 0 ) 1 2 3 8 e1 2 4 7 1241

Table 1 e Performance of JY on standardised tests of reading and comprehension.


Age at test (y;m) Raw score Standard score Age equivalent (y;m) Difference from chron age

PPVT-III 10;4 94 74 7;1 39m


10;11 94 72 7;1 46m

WRMT-R
Word Identification 10;4 87 119 14;0 þ44m
10;11 82 110 12;1 þ14m
Word Attack 10;4 39 114 >18;6 >þ98m

NARA-III
Reading accuracy 10;4 84 109 >13;6 þ33m
10;11 91 109 >13;0 þ25m
Reading rate 10;4 73 100 10;1 3m
10;11 75 95 10;0 11m
Reading comprehension 10;4 9 72 6;11 41m
10;11 15 82 7;5 42m

definitions of familiar words. On the other hand he expe- Reading aloud and comprehension at the text level were
rienced difficulty with understanding abstract verbal assessed using the Neale Analysis of Reading Ability-III
concepts and logical thinking processes”. (NARA-III; Neale, 1999). In this case, the comprehension
measure was reading comprehension rather than auditory
We assessed JY at 10 years 4 months and again at 10 years comprehension. His accuracy in reading aloud text was at
11 months to confirm the assessment of his general intellec- ceiling, and more than two years above age-based expecta-
tual functioning and to determine if he met the criteria for tions. His reading rate was within normal range for his age. In
hyperlexia. The results of a Wechsler Intelligence Scale for contrast, his reading comprehension was at least three years
Children-III (WISC-III; Wechsler, 1991), conducted in the first below age expectations on both testing occasions. Thus, the
test session, again revealed his overall intellectual functioning strong dissociation between reading aloud ability and
to be the low average range (derivation quotient: 84; 16th comprehension displayed by JY extended to the text level, and
percentile, 95% confidence interval: 79e90). His score on the overall, he showed a clear hyperlexic reading pattern.
Vocabulary subtest was extremely low (Scaled Score: 1),
while he displayed a relative strength in Block Design (Scaled
1.2. Case AD
Score: 11).
To examine evidence for hyperlexia, we compared JY’s
AD, a boy, was 8 years 2 months old when initially
comprehension skills with his oral reading skills, at both the
assessed. He was in Year 2 at a public primary school. His
single word and text levels. The results are presented in Table 1.
family migrated to Australia from Singapore when AD was
His auditory comprehension of single words, as measured by
5 years old, but although his parents conversed with each
the PPVT (Dunn and Dunn, 1997) was three years below age
other in Indonesian, they always conversed with AD in
expectations on both testing sessions. In contrast, his reading
English. AD had only ever learned and spoken English. His
aloud of single words, as assessed by the Word Identification
mother reported no complications with AD’s pregnancy or
subtest of the Woodcock Reading Mastery Test e Revised
birth. However, all motor and language milestones were
(WRMT-R; Woodcock, 1987) was more than three years ahead
delayed.
of age expectations on initial testing and, although it moder-
When AD was 2 years old, he was diagnosed by a paedia-
ated somewhat on second testing, remained over one year
trician as having Autism Spectrum Disorder. From 3 years old,
ahead of age expectations.2 His nonword reading ability, as
he attended therapy sessions with an occupational therapist
measured by the Word Attack subtest of the WRMT-R was also
and a speech therapist to address weaknesses in motor skills
extremely advanced for his age (tested at Session 1 only).
and language difficulties (echolalia, pronoun reversal, peculiar
Thus, he showed a clear dissociation between his ability to
word and jargon use). At age 5 years 7 months, a psychological
read aloud and his ability to comprehend single words, and in
assessment resulted in a diagnosis of AD as likely to have mild
fact showed a single word reading ability well in advance of
autism. A cognitive assessment at age 6 years 3 months with
his chronological age.
the WISC-III revealed AD’s overall intellectual functioning to
be in the low average range (Verbal IQ: 75; Performance IQ: 96;
Full Scale IQ: 84).
2
We examined whether there was evidence for a significant We assessed AD with the WISC-III at 8 years 2 months to
change over time in these two scores using the Reliable Change confirm the previous assessment of his overall intellectual
Index (Jacobson and Truax, 1991). Given the limited published
functioning. His performance again fell in the average to low
psychometric data in the WRMT-R manual, reliability statistics
average range, with a derivation quotient of 92 (95% confi-
(testeretest reliability, standard deviation on first testing) were
obtained from the normative data of Cirino et al. (2002). We found dence interval: 87e98). As with JY, his weakest subtest was
no evidence for a significant change in JY’s score from Session Vocabulary (Standard Score: 6), with strengths in Block Design
One (87) to Session Two (82). and Similarities (Standard Scores: 11).
1242 c o r t e x 4 6 ( 2 0 1 0 ) 1 2 3 8 e1 2 4 7

Table 2 e Performance of AD on standardised tests of reading and comprehension.


Age at test (y;m) Raw score Standard score Age equivalent (y;m) Difference from chron age

PPVT-III 8;4 81 78 6;1 27m


9;1 105 87 7;1 24m

WRMT-R
Word Identification 8;4 66 117 9;6 þ14m
9;1 59 96 8;9 4m
Word Attack 8;4 32 106 11;2 þ34m

NARA-III
Reading accuracy 8;4 55 111 9;2 þ10m
9;1 63 104 10;1 þ12m
Reading rate 8;4 49 98 7;10 6m
9;1 59 94 8;8 5m
Reading comprehension 8;4 5 81 6;3 25m
9;1 7 72 6;7 30m

We then examined evidence for a hyperlexic pattern in (mean ¼ 10y, 2m; SD ¼ .26m; range ¼ 10y, 1me10y, 9m), and 14
AD’s reading on two occasions e at 8 years, 4 months and at 9 eight-year-olds (all male) matched in age with AD (mean ¼ 8y,
years 1 month. The results are presented in Table 2. At the 2m; SD ¼ .22m; range ¼ 8y, 0me8y, 6m).
single word level, his auditory comprehension as measured by
the PPVT was at least two years below age expectations on 2.2. Materials and procedure
both testing sessions. In contrast, in Session 1, his single word
reading (WRMT Word Identification) was approximately one A test battery was constructed to assess JY and AD’s semantic
year above age expectations. His performance on single word knowledge and reading ability for a single set of irregular word
reading appeared to drop somewhat in Session 2; however, items. There were 36 irregular words, all nouns and all names
with a standard score of 96, his word reading performance of either animals, body parts or food. They ranged in written
was still well within normal range for his age.3 His nonword frequency from 0 to 482 occurrences per million (mean ¼ 41,
reading (WRMT Word Attack) was more than two years ahead obtained from the CELEX database; Baayen et al., 1995). The
of age expectations. The same basic pattern of dissociation items can be found in Appendix A. Each word was presented
was evident in AD’s text reading and comprehension, as in the context of a single word reading task and in the context
measured by the NARA-III. He performed approximately one of two single word comprehension tasks: definition genera-
year ahead of age expectations in reading aloud accuracy on tion and spoken wordepicture matching.4
both testing sessions and his reading rate was at the level
expected for his age; however, his reading comprehension 2.2.1. Reading aloud task
was more than two years delayed on both testing occasions. The 36 irregular words were presented on cards (lower case
Thus, like JY, AD showed clear evidence overall of a hyperlexic Geneva 14 point font) one at a time, in random order, and the
reading pattern. participants were instructed to attempt to read each word
Having established the presence of hyperlexia in both JY aloud. They were given the opportunity to self-correct if they
and AD, we turned to the key experimental investigations of wished, and their final answer was taken as the response. No
irregular word reading in these cases and, in particular to the feedback on correctness was given. Accuracy out of 36 was
relationship between JY and AD’s ability to read aloud a set of scored.
irregular words and their ability to comprehend those same
words. 2.2.2. Oral definition task
In this comprehension task, the participants were asked to
provide a definition of each of the 36 irregular words. On
each trial, the experimenter said the target word aloud and
2. Experimental investigations
than asked “Can you tell me what (target word) means?”
Participants were prompted with “tell me more about it” or
2.1. Participants
“can you explain what you mean?” if their answer was
unclear or vague. For homophonic words, of which there
Participants in the experimental investigations were the two
were five ( pear, steak, eye, bread and bear), if the partici-
hyperlexic cases described above. Also participating were two
pant’s response reflected the wrong homophone, they
sets of normally-developing control readers, matched in
chronological age with the two hyperlexic cases, who were 4
We also included a “living/non-living” category sorting task.
recruited from two Melbourne primary schools. There were 13 However, the controls all obtained perfect scores on this task.
ten-year-olds (11 males; 2 females), matched in age with JY Also, JY appeared not to understand the task, placing all items in
the “living” category. When asked why, he said “all food is eaten
3
Once again, the Reliable Change Index revealed no evidence by animals, so it’s living”. Given these problems with the task, we
for a significant change in scores over this time period. have not reported the results further.
c o r t e x 4 6 ( 2 0 1 0 ) 1 2 3 8 e1 2 4 7 1243

were asked to think of a different meaning for that word.


Table 4 e Performance of JY and AD, and their age
Two practice items were given, and once again no feed- matched controls, on the irregular word reading aloud
back was provided. and comprehension tasks.
The response of each participant was recorded and later
Task Case Raw Control group
scored by two independent raters. Each definition was scored score
out of a possible maximum of three points. Three points were Mean SD Range
awarded for a full definition; two points for a partial definition; Reading aloud JY 33 32.3 1.7 29e34
one point for appropriate use of the word in a sentence or (max ¼ 36) AD 27 30.1 2.4 27e34
minimal evidence of semantic knowledge, and zero points for
Oral definition JY 65 80.3 6.6 70.0e96.0
no evidence of understanding. The two raters’ scores were (max ¼ 108) AD 29 70.4 10.1 56.0e85.5
averaged for each item to provide a total accuracy score for
Pictureeword match JY 33 35.1 .9 33e36
each participant out of 108. Inter-rater reliability was good
(max ¼ 36) AD 29 34.8 1.1 32e36
(Pearson’s correlation ¼ .72; p < .001; see Cicchetti and
Sparrow, 1981). Examples of definitions receiving particular
ratings are provided in Table 3. It can be seen that JY performed slightly above average for
his age in reading aloud the irregular words, while AD per-
2.2.3. Spoken wordepicture matching task formed somewhat below average. The types of errors made in
Semantic knowledge of the irregular words was further both cases were predominantly regularisations (e.g., “steek”
assessed using a spoken wordepicture matching task, which for steak): all of JY’s errors fell into this category and 6 out of
differed from the oral definitions task in not requiring a verbal the 9 errors made by AD. AD’s remaining errors could be
response from the participants. Each of the 36 words was classified as either visual or phonological output errors (e.g.,
spoken aloud to the participants and they were asked to point “plarm” for palm). Neither case showed a strong bias towards
to its corresponding picture, from a set of three pictures pre- producing lexicalisation responses, with AD producing only
sented. The three pictures included (a) a picture of the target one (dessert read as “desert”, an example of a typical regular-
word, (b) a semantic distractor and (c) a semantically unre- isation response to a potentiophone; Friedmann and Lukov,
lated distractor (see Appendix A). The semantic distractors 2008). JY produced two lexicalisation responses: the regular-
came from the same superordinate category as the target isation of the potentiophone dessert as “desert” and the visual
words (e.g., pear e apple). Position of the target picture (left, error of meringue as “mango”.
centre, right), and the distractors, was varied across the Their scores were compared statistically with those of their
displays. appropriate control group using the modified t-test of
The order of administration of the tasks was: (1) Reading Crawford and Howell (1998). This test directly assesses
aloud, (2) Oral definition, and (3) Spoken wordepicture whether or not an individual case exhibits a statistically
matching. The total testing period lasted approximately 1 h. JY significant deficit on a task. In contrast to the use of z-scores,
and AD were tested at their homes or at the University of the test effectively controls the Type I error rate regardless of
Melbourne. The control participants were tested at their the sample size and is robust to violations of normality
schools. (Crawford and Howell, 1998; Crawford et al., 2004). For neither
JY, t(12) ¼ .40, p ¼ .69, nor AD, t(13) ¼ 1.24, p ¼ .24, was there
evidence for a statistically significant deficit in reading aloud
the irregular words.
3. Results Turning to comprehension of the words, both cases scored
well below average on the oral definition task, with
3.1. Overall performance the performance of both AD, t(13) ¼ 3.96, p ¼ .002, and JY,
t(12) ¼ 2.23, p ¼ .04, being significantly impaired compared to
The overall performances of JY and AD on the irregular word
controls. There was some evidence of ceiling effects in the
reading aloud task and the two comprehension tasks, relative
control group for the spoken wordepicture matching task, but
to their respective age matched controls, are presented in
AD showed a highly significant impairment relative to
Table 4.
controls, t(13) ¼ 5.09, p < .001. JY’s deficit on this task
approached significance, t(12) ¼ 2.13, p ¼ .055.
In summary, the results of JY and AD for reading aloud
Table 3 e Examples of definitions receiving particular and comprehension of the experimental items largely
ratings, for the word meringue (taken from control data mirrored their results for the selection measures: they per-
and AD and JY’s responses).
formed at average levels in reading aloud the words, but
Rating Example tended to perform poorly on tests of comprehension of
3 “A meringue is a dessert. It’s a white, crunchy sort of thing those words. Importantly, these dissociations were demon-
and a lot of bakeries put decorations on them and make, strated specifically for the reading and comprehension of
like, mouse meringues and stuff”. a single set of irregular words. Having established that such
2 “A tasty egg thing, which is the egg white. I can’t really dissociations could be found, we moved to examining in
describe them, but I’ve heard about them”.
more detail the item-specific relationship between the
1 “Like, lemon meringue pie”.
hyperlexic children’s reading aloud and comprehension of
0 “A type of fruit or drink, not sure, just heard the word”.
the irregular words.
1244 c o r t e x 4 6 ( 2 0 1 0 ) 1 2 3 8 e1 2 4 7

Table 5 e Reading aloud accuracy for semantically known Table 6 e Words receiving a oral definition score of 0 that
and unknown words for JY and AD. were correctly read aloud by JY and AD.
Case Known words Unknown words Case Correctly Frequency Definition given (rated 0)
read (per
N Number % N Number % word million)
correctly Correctly correctly Correctly
read read read read JY Shoulder 73 “I don’t know”.
Palm 21 “Don’t know what a palm is
JY 24 22 91.7 12 11 91.7
either”.
AD 4 4 100.0 32 23 71.9
Thigh 14 “I don’t know”.
Salmon 11 “I don’t know”.

AD Shoulder 73 “Shoulder is for something, you


3.2. Item-specific effects: reading accuracy for know, a shoulder can call
semantically known and unknown words something”.
Stomach 43 “Stomach is for breathing
To further explore the degree to which the hyperlexic chil- something. Stomach, oh my
gosh”.
dren’s ability to read aloud irregular words was related to their
Bull 27 “A bull is about something, bulls
semantic knowledge of those words, we simplified their
are, I don’t know”.
scores on the oral definition task from a continuous score Ankle 10 “An ankle is for something, is
(0e3) into semantically “known” and “unknown” words.5 A a hurting, when you hurt”.
word was classified as “known” if the response on the oral Squash 9 “Squash is for cleaning the car”.
definition task was rated 2 or above (averaged across the two Penguin 4 “Penguin is a penguin is a penguin,
raters), and classified as “unknown” if it was rated 0 or 1. The a penguin can do something”.
Wasp 2 “Wasp is for something that you
hyperlexic participants’ reading aloud performance as
put it in the bin”.
a function of this classification is presented in Table 5. It can
Echidna 0 “Echidna. How do you spell that?
be seen that JY’s reading aloud accuracy did not vary at all Echidna is for penguin”.
according to whether or not whether or not he knew the
meaning of word: he scored 91.7% correct for both sets of
items. AD accurately read aloud a higher percentage of words
that were known to him than words that were unknown; of the same set of irregular words in hyperlexia. This disso-
however the difference between these two proportions was ciation was evident between reading aloud and performance
not significant (Fisher’s Exact: p ¼ .55). on an expressive comprehension task e providing oral defi-
It could be argued that even a score of “1” on the oral definition nitions of the words e and also between reading aloud and
task (classified as “unknown” in the analyses above) is evidence of a receptive task e matching the spoken words with their
some, extremely limited, semantic knowledge of a word. There- corresponding pictures (approaching significance in the case
fore, to address the question of whether either of the hyperlexic of JY). As well, both cases not only showed a clear disparity
cases had been able to learn to read aloud any irregular words for between their ability to read the words and their ability to
which they had no semantic knowledge at all, we examined understand their meaning, but actually showed normal
whether there were instances of correct reading aloud of a word irregular word reading accuracy for their age.
which had received an oral definition score of 0 (meaning both These findings complement dissociations reported in cases
raters had scored the definition as showing no evidence of of acquired semantic dementia in which irregular word
semantic knowledge). JY displayed 4 such instances (12% of items reading ability has been maintained despite progressive loss
correctly read) and AD displayed 8 instances (30% of items of knowledge of the meanings of words (Blazely et al., 2005;
correctly read). The items and the definitions given are detailed in Cipolotti and Warrington, 1995; Gerhand, 2001). Together,
Table 6. It can also be seen from Table 6 that several of the these cases provide further evidence for the existence of
correctly read but semantically unknown words were quite low in a direct route from visual word recognition to spoken word
frequency, with 7 of the 12 having CELEX frequencies of less than production which does not proceed via semantics. The
20 occurrences per million. present results from hyperlexia also extend these findings in
suggesting that the lexical-semantic and direct-lexical routes
can not only be independently impaired, but can also be
independently acquired in children learning to read: JY and AD
4. Discussion had apparently never learned the meanings of many of the
words in our test battery, and yet they had been able to learn
The results of JY and AD represent the first clear evidence of to successfully read them aloud, and indeed had been able to
a dissociation between reading accuracy and comprehension do so with a proficiency equivalent to that of normally-
developing readers of the same age.
5
This item-specific analysis could not be carried out for the
The pattern of results we observed was also evident at an
spoken wordepicture matching task, as it had a chance compo-
item level: JY was just as accurate at reading aloud words for
nent of 33%. If a child scored correctly on an item, we could not
know with any certainty whether this was because they knew the which he was not able to provide a satisfactory definition as
item’s meaning or whether it was because they had selected the he was in reading aloud words which he could define. AD
correct picture by chance. appeared somewhat less successful at reading aloud
c o r t e x 4 6 ( 2 0 1 0 ) 1 2 3 8 e1 2 4 7 1245

unknown words, but not significantly so. In both cases, phonological representations, unmediated by semantics, is
a proportion of irregular words were read aloud correctly that characteristic of normal reading acquisition, but that these
(a) had received an oral definitions score of zero, and (b) were direct connections frequently go undetected in normal
low in frequency. This would suggest that, in contrast to readers due to the powerful co-existent presence of the
predictions from the PDP modelling of Plaut et al. (1996), even lexical-semantic route. Adjudicating between these two
low frequency irregular words can, on occasion, be read in the hypotheses will be an important direction for future research.
absence of semantic support. In summary, the present results demonstrate that, in cases
A limitation of our analysis was that it was only possible to of developmental hyperlexia, normal acquisition of irregular
perform the item-specific analyses on an expressive word reading skills can occur in the context of impaired
measure e the oral definitions score e rather than on both comprehension of those same words. In drawing this
a receptive and an expressive measure. As such, it is possible conclusion, we do not wish to suggest that semantic factors
that the ability of the children to demonstrate understanding play no role in reading aloud, or that they are not important
of the meanings of the words was limited by their expressive for the acquisition of lexical representations. Indeed, there is
language difficulties. It is true that both cases showed ample evidence from both skilled readers (Balota et al., 2004;
evidence for expressive language impairments, as indexed by Strain et al., 1995) and developing readers (Ricketts et al.,
the Vocabulary Subtest of the WISC-III. However, while we 2007), that such meaning-based factors do have some influ-
cannot rule out that this may have had some influence on ence on performance. There is also evidence that semantic
performance, we note that (a) both children showed evidence impairments and irregular word reading impairments tend
for receptive language difficulties as well, consistent with often to co-occur, both in acquired semantic dementia cases
a diagnosis of general semantic impairment, (b) both cases (Woollams et al., 2007) and in developmental cases of Poor
obtained a rating of “3” on at least one oral definition item, Comprehenders (Nation and Snowling, 1998; Ricketts et al.,
suggesting that their expressive language difficulties were 2007). Our claim is rather that it is not necessary for trans-
not so severe as to limit their ability to demonstrate compre- lation from print to speech always to proceed via semantics,
hension on our task when it was present. even in those instances where a word has an irregular
JY and AD both showed advanced reading accuracy spelling-sound mapping, and that it is not necessary for
performance on the standardised tests used at selection; children to know the meaning of a word in order for them to
however, their irregular word reading skills as measured in be able to learn to recognise and read it aloud. While the
our experimental investigations were only in the average widely-accepted existence of a lexical-semantic route of some
range for their age. This advanced performance is therefore form accounts for the association between irregular word
likely due to the contribution of other reading sub-processes reading and comprehension, the proposal of a pathway such
to performance on the more general standardised reading as the direct-lexical route of the DRC is required to account for
measures, such as nonlexical reading skills, which were also the growing evidence that these two skills can and do
developed to an advanced level in our two cases. Indeed, both dissociate.
cases scored well ahead of age expectations on measures of
nonword reading. As such, their performance was consistent
with that of a number of hyperlexic cases who have been
reported to be able to read nonwords at normal levels or above Appendix A.
(Aram, 1997; Atkin and Lorch, 2006; Glosser et al., 1996; The 36 experimental items and distractors
Seymour and Evans, 1992). Although not the focus of the
present study, the development of a range of reading sub-
processes in hyperlexia, and the distal factors affecting this Target Semantic distractor Unrelated distractor
development, warrant further investigation. (irregular words)
More research is also required to establish precisely how
1. Ankle Wrist Hammer
children with hyperlexia succeed in acquiring lexical repre- 2. Bear Fox Sport person
sentations for at least some words in the absence of knowl- 3. Biscuit Chocolate Keys
edge of their meanings. One hypothesis is that these children 4. Bread Cake Toy
may successfully acquire orthographic lexical representations 5. Brie Swiss cheese Shorts
and link them directly with pre-existing phonological lexical 6. Broccoli Mushrooms Tongs
7. Bull Goat Gloves
representations through extensive exposure and practice,
8. Calf Dog Cup
brought about by the unusual preoccupation with reading that
9. Dessert Salad Box
is often seen in hyperlexia (Nation, 1999). Such a hypothesis 10. Echidna Platypus Plant
would predict that children with hyperlexia should score 11. Eye Ear Light globe
highly on measures of print exposure (i.e., familiarity with 12. Foot Arm Whistle
printed text; Cunningham and Stanovich, 1991) relative to 13. Forehead Nose Ball
children of the same age. While we are aware of no direct 14. Giraffe Reindeer Eggs
15. Head Chest Football boot
evidence that this is the case, this hypothesis deserves further
16. Heart Head Bell
exploration. However, such a hypothesis assumes some 17. Leopard Horse Iron
“deviance” in the lexical acquisition processes of individuals 18. Meringue Pancake Door handle
with hyperlexia: an alternative proposal is that the estab-
(continued on next page)
lishment of direct connections between orthographic and
1246 c o r t e x 4 6 ( 2 0 1 0 ) 1 2 3 8 e1 2 4 7

Coltheart M, Curtis B, Atkins P, and Haller M. Models of reading


Appendix A (continued)
aloud: Dual-route and parallel-distributed processing
Target Semantic distractor Unrelated distractor approaches. Psychological Review, 100: 589e608, 1993.
(irregular words) Coltheart M, Rastle K, Perry C, Langdon R, and Ziegler J. DRC: Dual
Route Cascaded model of visual word recognition and reading
19. Monkey Koala Rose
aloud. Psychological Review, 108: 204e256, 2001.
20. Palm Toes Telescope
Crawford JR and Howell DC. Comparing an individual’s test score
21. Pear Apple Cards
against norms derived from small samples. The Clinical
22. Penguin Seal Jar
Neuropsychologist, 12: 482e486, 1998.
23. Rhinoceros Elephant Cooker
Crawford JR, Garthwaite PH, Howell DC, and Gray CD. Inferential
24. Salmon Shark Dice
methods for comparing a single case with a control sample:
25. Salt Sauce Garden tool
Modified t-test versus Mycroft et al’s (2002) modified Anova.
26. Shoulder Fist Duck
Cognitive Neuropsychology, 21(7): 750e755, 2004.
27. Spread Croissant Money
Cunningham A and Stanovich KI. Tracking the unique effects
28. Squash Carrot Basket
of print exposure in children: Associations with vocabulary,
29. Steak Chicken Tree
general knowledge and spelling. Journal of Educational
30. Stomach Arm Luggage
Psychology, 8: 264e274, 1991.
31. Thigh Shin Button
Dunn L and Dunn L. Peabody Picture Vocabulary Test. 3rd ed.
32. Tongue Lips Timer
Minnesota: American Guidance Service, 1997.
33. Wallaby Possum Hat
Friedmann N and Lukov L. Developmental surface dyslexias.
34. Wasp Butterfly Phone
Cortex, 44: 1146e1160, 2008.
35. Wolf Tiger Lamp
Frith U and Snowling MJ. Reading for meaning and reading for
36. Worm Grasshopper Pencils
sound in autistic and dyslexic children. British Journal of
Disorders of Communication, 1: 329e342, 1983.
Funnell E. Phonological processes in reading: New evidence
references
from acquired dyslexia. British Journal of Psychology, 74:
159e180, 1983.
Funnell E. Response biases in oral reading: An account of the
Aaron PG. Dyslexia and Hyperlexia: Diagnosis and Management of co-occurrence of surface dyslexia and semantic dementia.
Developmental Reading Disabilities. Dordrecht, the Netherlands: Quarterly Journal of Experimental Psychology: Human Experimental
Kluwer, 1989. Psychology, 49(A): 417e446, 1996.
Aram DM. Hyperlexia: Reading without meaning in young Gerhand S. Routes to reading: A report of a non-semantic reader
children. Topics in Language Disorders: Special Perspectives on with equivalent performance on regular and exception words.
relations between oral language knowledge and reading, Neuropsychologia, 39: 1473e1483, 2001.
17: 1e13, 1997. Glosser G, Grugan P, and Friedman RB. Semantic influences on
Aram D, Rose D, and Horwitz S. Hyperlexia: Developmental phonological and orthographic processing: Evidence from
reading without meaning. In Malatesha R and Whitaker H developmental hyperlexia. Brain and Cognition, 30: 335e337, 1996.
(Eds), Dyslexia: A Global Issue. Lancaster, UK: Martinus Nijhoff Glosser G, Grugan P, and Friedman RB. Semantic memory
Publishers, 1984. impairment does not impact on phonological and
Atkin K and Lorch MP. Hyperlexia in a 4-year-old boy orthographic processing in a case of developmental
with Autistic Spectrum Disorder. Journal of Neurolinguistics, hyperlexia. Brain and Language, 56: 234e247, 1997.
19: 253e269, 2006. Graham KS, Hodges JR, and Patterson KE. The relationship
Baayen RH, Pipenbrock R, and Gulikers L. The CELEX Lexical between comprehension and oral reading in progressive
Database (CD-ROM). Linguistic Data Consortium. Philadelphia: fluent aphasia. Neuropsychologia, 32: 229e316, 1994.
University of Pennsylvania, 1995. Grigorenko EL, Klin A, and Volkmar F. Annotation: Hyperlexia:
Balota DA, Cortese MJ, Sergent-Marshall SD, Spieler DH, and Disability or superability? Journal of Child Psychology and
Yap MJ. Visual word recognition of single syllable words. Psychiatry, 44(8): 1079e1091, 2003.
Journal of Experimental Psychology: General, 133: 283e316, 2004. Healy JM. The enigma of hyperlexia. Reading Research Quarterly,
Blazely AM, Coltheart M, and Casey JC. Semantic impairment 17: 319e338, 1982.
with and without surface dyslexia: Implications for models Hodges JR, Patterson K, Oxbury S, and Funnell E. Semantic
of reading. Cognitive Neuropsychology, 22: 695e717, 2005. dementia e Progressive fluent aphasia with temporal-lobe
Castles A and Coltheart M. Varieties of developmental dyslexia. atrophy. Brain, 115: 1783e1806, 1992.
Cognition, 47: 149e180, 1993. Jacobson NS and Truax P. Clinical significance: A statistical
Cicchetti DV and Sparrow SS. Developing criteria for establishing approach to defining meaningful change in psychotherapy
interrater reliability of specific items: Applications to research. Journal of Consulting and Clinical Psychology,
assessment of adaptive behaviour. American Journal of Mental 59: 12e19, 1991.
Deficiency, 86: 127e137, 1981. Lambon Ralph MA, Ellis AW, and Franklin S. Semantic loss
Cipolotti L and Warrington EK. Semantic memory and reading without surface dyslexia. Neurocase, 1: 363e369, 1995.
abilities: A case report. Journal of the International Manis FR, Seidenberg MS, Doi LM, McBride-Chang C, and
Neuropsychological Society, 1: 104e110, 1995. Peterson A. On the bases of two subtypes of developmental
Cirino PT, Rashid FL, Sevcik RA, Lovett MW, Frijters JC, Wolf M, dyslexia. Cognition, 58: 157e195, 1996.
et al. Psychometric stability of nationally normed and McKay A, Castles A, Davis C, and Savage G. The impact of
experimental decoding and related measures in children with progressive semantic loss on reading aloud. Cognitive
reading disability. Journal of Learning Disabilities, 35: 526e539, Neuropsychology, 24: 162e186, 2007.
2002. Morton J and Patterson K. A new attempt at an interpretation
Coltheart M. Acquired dyslexias and normal reading. In or an attempt at a new interpretation. In Coltheart M,
Malatesha RN and Whitaker HA (Eds), Dyslexia: A Global Issue. Patterson K, and Marshall JC (Eds), Deep Dyslexia. London:
Lancaster, UK: Martinus Nijhoff Publishers, 1984. Routledge and Kegan Paul, 1980.
c o r t e x 4 6 ( 2 0 1 0 ) 1 2 3 8 e1 2 4 7 1247

Nation K. Reading skills in hyperlexia: A developmental Shallice T, Warrington EK, and McCarthy R. Reading without
perspective. Psychological Bulletin, 125(3): 338e355, 1999. semantics. Quarterly Journal of Experimental Psychology: Human
Nation K and Snowling MJ. Semantic processing and the Experimental Psychology, 35A: 111e138, 1983.
development of word-recognition skills: Evidence from Siegel LS. A longitudinal study of a hyperlexic child: Hyperlexia
children with reading comprehension difficulties. Journal as a language disorder. Neuropsychologia, 22: 577e585, 1984.
of Memory and Language, 39: 85e101, 1998. Silberberg NE and Silberberg MC. Hyperlexia: Specific word
Neale M (Ed), Neale Analysis of Reading Ability. 3rd ed. Melbourne: recognition skills in young children. Exceptional Child, 34:
ACER Press, 1999. 41e42, 1967.
Newcombe F and Marshall JC. On psycholinguistic classifications Snowden JS, Goulding PJ, and Neary D. Semantic dementia: A
of the acquired dyslexias. Bulletin of the Orton Society, 31: 29e46, form of circumscribed cerebral atrophy. Behavioural Neurology,
1981. 2: 167e182, 1989.
Patterson K and Hodges JR. Deterioration of word meaning: Strain E, Patterson K, and Seidenberg MS. Semantic effects in
Implications for reading. Neuropsychologia, 30: 1025e1040, 1992. single-word naming. Journal of Experimental Psychology:
Plaut DC, McClelland JL, Seidenberg MS, and Patterson K. Learning, Memory, and Cognition, 21: 1140e1154, 1995.
Understanding normal and impaired word reading: Warrington EK. The selective impairment of semantic
Computational principles in quasi-regular domains. memory. Quarterly Journal of Experimental Psychology,
Psychological Review, 103: 56e115, 1996. 27: 635e657, 1975.
Ricketts J, Nation K, and Bishop DVM. Vocabulary is important Wechsler D. Wechsler Preschool and Primary Scale of Intelligence,
for some, but not all reading skills. Scientific Studies of Reading, Revised. San Antonio: The Psychological Corporation, 1989.
11: 235e257, 2007. Harcourt Brace Janovich, Inc.
Schwartz MF, Saffran EM, and Marin OSM. Fractionating the Wechsler D. Wechsler Intelligence Scale for Children (WISC-III).
reading process in dementia: Evidence for word-specific print- 3rd ed. San Antonio: The Psychological Corporation, 1991.
to-sound associations. In Coltheart M, Patterson K, and Harcourt Brace Janovich, Inc.
Marshall JC (Eds), Deep Dyslexia. London: Routledge and Kegan Welsh MC, Pennington BF, and Rogers S. Word recognition and
Paul, 1980. comprehension skills in hyperlexic children. Brain and
Seidenberg E and McLelland J. A distributed, developmental Language, 32: 76e96, 1987.
model of word recognition and naming. Psychological Review, Woodcock R. Woodcock Reading Mastery Tests e Revised (WRMT-R).
96: 523e568, 1989. Minnesota: American Guidance Services, 1987.
Seymour PH and Evans HM. Beginning reading without Woollams AM, Lambon Ralph MA, Plaut DC, and Patterson K.
semantics: A cognitive study of hyperlexia. Cognitive SD-squared: On the association between semantic dementia
Neuropsychology, 9: 89e122, 1992. and surface dyslexia. Psychological Review, 114: 316e339, 2007.

You might also like