Вы находитесь на странице: 1из 49

This article was downloaded by: [Texas A & M International University]

On: 24 August 2015, At: 02:56


Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: 5 Howick Place, London, SW1P 1WG

Neuropsychological
Rehabilitation: An International
Journal
Publication details, including instructions for authors
and subscription information:
http://www.tandfonline.com/loi/pnrh20

Thinking outside the boxes:


Using current reading models to
assess and treat developmental
surface dyslexia
a a
Caroline Law & Linda Cupples
a
Macquarie Centre for Language Sciences, Macquarie
University, Sydney, NSW, Australia
Published online: 18 Aug 2015.
Click for updates

To cite this article: Caroline Law & Linda Cupples (2015): Thinking outside the
boxes: Using current reading models to assess and treat developmental surface
dyslexia, Neuropsychological Rehabilitation: An International Journal, DOI:
10.1080/09602011.2015.1064453

To link to this article: http://dx.doi.org/10.1080/09602011.2015.1064453

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the
information (the “Content”) contained in the publications on our platform.
However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or
suitability for any purpose of the Content. Any opinions and views expressed
in this publication are the opinions and views of the authors, and are not the
views of or endorsed by Taylor & Francis. The accuracy of the Content should
not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions,
claims, proceedings, demands, costs, expenses, damages, and other liabilities
whatsoever or howsoever caused arising directly or indirectly in connection
with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes.
Any substantial or systematic reproduction, redistribution, reselling, loan, sub-
licensing, systematic supply, or distribution in any form to anyone is expressly
forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Downloaded by [Texas A & M International University] at 02:56 24 August 2015
Neuropsychological Rehabilitation, 2015
http://dx.doi.org/10.1080/09602011.2015.1064453

Thinking outside the boxes: Using current reading


models to assess and treat developmental surface
dyslexia
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

Caroline Law and Linda Cupples


Macquarie Centre for Language Sciences, Macquarie University, Sydney,
NSW, Australia

(Received 31 May 2014; accepted 16 June 2015)

Improving the reading performance of children with developmental surface


dyslexia has proved challenging, with limited generalisation of reading skills
typically reported after intervention. The aim of this study was to provide tai-
lored, theoretically motivated intervention to two children with developmental
surface dyslexia. Our objectives were to improve their reading performance,
and to evaluate the utility of current reading models in therapeutic practice.
Detailed reading and cognitive profiles for two male children with develop-
mental surface dyslexia were compared to the results obtained by age-
matched control groups. The specific area of single-word reading difficulty
for each child was identified within the dual route model (DRM) of reading,
following which a theoretically motivated intervention programme was
devised. Both children showed significant improvements in single-word
reading ability after training, with generalisation effects observed for untrained
words. However, the assessment and intervention results also differed for each
child, reinforcing the view that the causes and consequences of developmental
dyslexia, even within subtypes, are not homogeneous. Overall, the results of the
interventions corresponded more closely with the DRM than other current
reading models, in that real word reading improved in the absence of enhanced
nonword reading for both children.

Keywords: Surface dyslexia; Dual route model of reading; Intervention; Devel-


opmental; Reading.

Correspondence should be addressed to Linda Cupples, Macquarie Centre for Language


Sciences, Macquarie University, Sydney NSW 2109, Australia. E-mail: linda.cupples@mq.
edu.au
No potential conflict of interest was reported by the authors.

# 2015 Taylor & Francis


2 LAW AND CUPPLES

INTRODUCTION
There has been a recent increase in the number of intervention studies for
children with developmental dyslexia that encompass a theoretically motiv-
ated, targeted approach to treatment devised on the basis of individual assess-
ment results. The dual route model (DRM) of reading has been utilised in
many of these studies and the interventions have shown some success in
improving the reading (and spelling) performance of individual children
(e.g., Broom & Doctor, 1995a, 1995b; Brunsdon, Coltheart, & Nickels,
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

2005; 2006; Brunsdon, Hannon, Coltheart, & Nickels, 2002; Brunsdon,


Hannon, Nickels, & Coltheart, 2002).

Dual route model (DRM) of reading


According to the DRM (Coltheart, Curtis, Atkins, & Haller, 1993; Coltheart,
Rastle, Perry, Langdon, & Ziegler, 2001; Morton & Patterson, 1987) skilled
readers make use of two qualitatively different mechanisms when converting
print to speech: lexical and non-lexical. The lexical mechanism involves rec-
ognition of the whole word and retrieval of its pronunciation from memory,
whereas the non-lexical mechanism involves assigning the word’s pronuncia-
tion through a process of decoding or application of letter-sound rules. The
DRM is illustrated in Figure 1.
As the figure shows, both reading mechanisms begin with the identification
of visual features and then letters, which are encoded in their correct serial
positions within the word. The final stage, also shared by both mechanisms,
is the phoneme unit level, which contains individual speech sounds within
a word and encodes them in their correct positions for articulation.
Intervening processing stages differ between the lexical and non-lexical
mechanisms. Within the lexical mechanism, information about letter identity
passes through an orthographic lexicon (which contains orthographic rep-
resentations of known words) to a phonological lexicon (which contains the
corresponding phonological representations of words), either directly or via
the semantic system. By contrast, the non-lexical mechanism contains a
letter-sound or grapheme-to-phoneme conversion module, which assigns a
word’s pronunciation using the phonemes most commonly associated with
the constituent graphemes contained in the word. In this sense, the non-
lexical mechanism is rule-governed and “hard-wired”; that is, pronunciations
are determined by the rules that are programmed into the system, with no
flexibility to use less common grapheme-to-phoneme correspondences. The
non-lexical mechanism is generally considered to operate more slowly than
the lexical mechanism because of the need to assemble a word’s pronuncia-
tion via serial application of grapheme-to-phoneme correspondences. By con-
trast, letters in words are processed in parallel by the lexical mechanism (see
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 3
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

Figure 1. The Dual Route Model of Reading. The series of processes constituting the lexical route are
connected with dashed lines (— ..—), while the non-lexical route components are connected with
dotted lines ( . . . ). Inputs to and outputs from the system are shown in solid lines. Taken from
Castles, Bates, Coltheart, Luciano, and Martin (2006, pp. 92–103). Published by Wiley
Publications. Reprinted with permission. #United Kingdom Literacy Association 2006. Published
by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street,
Malden, MA 02148, USA.

Grainger & Ziegler, 2011 for more information about orthographic processing
in dual route models).
According to the DRM, both the lexical and non-lexical mechanisms will
process all word types when reading. However, only the lexical mechanism
will provide a correct pronunciation for irregular words, such as yacht,
which do not conform to spelling-to-sound rules, and only the non-lexical
mechanism will provide a correct pronunciation for pronounceable nonwords,
such as fot, which do not have a matching lexical entry. In line with this div-
ision of labour, the DRM predicts three main types of reading difficulties that
4 LAW AND CUPPLES

have been confirmed in developmental research conducted over the past 20


years (Castles, Bates, & Coltheart, 2006; Castles & Coltheart, 1993; Manis,
Seidenberg, Doi, McBride-Chang, & Peterson, 1996). Children with develop-
mental surface dyslexia, the focus of the current study, experience difficulties
with the lexical mechanism (resulting in poorer reading of irregular words
than nonwords). By contrast, children with developmental phonological dys-
lexia experience difficulties with the non-lexical mechanism (resulting in
poorer reading of nonwords than irregular words). Difficulties with both
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

mechanisms result in developmental mixed dyslexia, reflected in poor


reading of both irregular words and nonwords.
Despite these confirmatory findings, the DRM is not without its critics, par-
ticularly in its application to developmental reading disorder, on grounds that
it represents a skilled reading system and cannot be used to describe the
dynamic processes by which a child learns to read (Snowling, Bryant, &
Hulme, 1996). Castles et al. (2006) refuted this claim, however, providing
evidence from typically developing readers and children with dyslexia indi-
cating that the DRM architecture stayed the same during the development
of reading, with only quantitative changes occurring in the system as chil-
dren’s reading ability improved. Thus, Castles et al. claimed that use of the
DRM was entirely appropriate for developing readers as well as for skilled
readers.
Another potential criticism of the DRM relates to implementation of the
non-lexical reading mechanism as a set of explicit rules for mapping gra-
phemes onto phonemes. Based on investigations of German nonword
reading, Perry, Ziegler, Braun, & Zorzi (2010) concluded that a connectionist
dual process model (CDP+) provided a better account of human reading
behaviour.

Connectionist dual process model (CDP1)


The CDP+ (Perry, Ziegler, & Zorzi, 2007; Zorzi, 2010) has a similar archi-
tecture to the DRM, containing a lexical and a non-lexical reading mechanism
or process. It differs however in regard to implementation of the non-lexical
process. Instead of incorporating explicit rules for converting graphemes to
phonemes, the CDP+ contains a connectionist network, which is trained
through exposure to a lexical database to learn the statistical patterns under-
lying spelling–sound relationships. This network approach provides for
greater flexibility and variability in the mapping of sounds to letters, including
the influence of surrounding graphemic context, such as the word body (i.e.,
the orthographic equivalent of the phonological rime).
As the CDP+ has a dual-process architecture similar to that of the DRM, it
predicts the same types of selective developmental reading disorders; in par-
ticular, surface dyslexia, resulting from a weakness in the lexical reading
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 5

process, and phonological dyslexia, resulting from a weakness in the non-


lexical process. In confirmation, Zorzi, Houghton, and Butterworth (1998)
biased their network model to use a lexical reading strategy (as if there
were damage to the non-lexical route) and found evidence of a reduced
ability to read nonwords, as in phonological dyslexia.
In sum, both the DRM and CDP+ provide a principled account of the
existence of selective reading disorders, such as surface and phonological
dyslexia, through the inclusion of two separate reading mechanisms or pro-
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

cesses. An alternative interpretation was proposed by Harm and Seidenberg


(1999) building on Seidenberg and McClelland’s (1989) connectionist
model of reading, in which a single mechanism exists to read all types of
words, both regular and irregular, and nonwords.

Harm and Seidenberg’s connectionist model


The architecture of Harm and Seidenberg’s (1999) model is illustrated in
Figure 2. According to this model the process of learning to read relies
on development of a single procedure for mapping orthography onto pho-
nology, which uses rule-like behaviour extracted from a set of training
words. Harm and Seidenberg argued that different types of damage to a
computational network such as this one, either before or during the

Figure 2. The connectionist model of reading from Harm and Seidenberg (1999). From Harm and
Seidenberg (1999, 491– 528). Published by American Psychological Association. Reprinted with
permission.
6 LAW AND CUPPLES

course of learning to read, could result in different types of reading diffi-


culties. In line with this view, they simulated phonological dyslexia by
impairing the computational network’s phonological representations
before it learnt to read. Mild damage produced a phonological dyslexic
pattern (impaired nonword reading in the presence of unimpaired reading
of irregular words); whereas more severe damage produced impairments
in both nonword reading and irregular word reading. Surface dyslexia
was also simulated, but not in its pure form (i.e., impaired reading of irre-
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

gular words in the presence of unimpaired nonword reading). Rather, Harm


and Seidenberg adopted the view that surface dyslexia reflects a general
reading delay rather than a specific impairment in reading irregular
words (but see Peterson, Pennington, & Olson, 2013, for an opposing
view). In line with their perspective, Harm and Seidenberg’s simulations
focused on slowing the network’s acquisition process by reducing the
amount of training, slowing the rate of learning, degrading orthographic
input, or reducing the number of hidden units that mediate between ortho-
graphy and phonology in order to impair the network’s capacity to learn
higher-order dependencies (i.e., those spanning multiple letters). In all of
these cases, the computational network showed a pattern of reading diffi-
culty characterised by impairments in both irregular word reading and
nonword reading, although irregular word reading was more impaired.

Theoretically motivated intervention for developmental surface


dyslexia
Theoretical models of reading can be evaluated in the extent to which
they accord with observed patterns of human behaviour, including devel-
opmental dyslexia. They can also be evaluated in terms of their utility
for guiding clinical assessment and intervention for children who present
with developmental reading difficulties, and for interpreting post-interven-
tion results (i.e., whether intervention effects are consistent with the model
or indicate a need for revision). Published clinical research using reading
models as a basis for intervention is still relatively rare. Some previous
intervention studies (e.g., Brunsdon, Hannen, Coltheart et al., 2002) have
shown that the DRM can be useful in guiding clinical treatment decisions,
although clinical judgement is also required in order to interpret a child’s
reading profile, make intervention decisions, and interpret changes in
reading performance as a result of intervention. To date, other reading
models (such as the CDP+ and other connectionist models) have been
used primarily to simulate predicted performance rather than to guide clini-
cal intervention practices (although see Harm, McCandliss, & Seidenberg,
2003; Welbourne & Lambon Ralph, 2005, for attempts to simulate rehabi-
litation effects).
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 7

An advantage of using the DRM to devise interventions for developmental


surface dyslexia in particular lies in the utility of the model for identifying
specific intervention targets within the lexical reading route (see Figure 1).
By contrast, the view that surface dyslexia reflects a general reading delay
or lack of reading experience (as per Harm & Seidenberg’s, 1999, connection-
ist model) provides little specific guidance as to how an effective intervention
programme might be structured or implemented. Results from two previous
intervention studies that used the DRM as their theoretical basis provide
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

clear illustrations of the value of this approach. Both Broom and Doctor
(1995a) and Brunsdon, Hannan, Coltheart et al. (2002) provided intervention
aimed at enhancing visual orthographic processing within the lexical reading
mechanism of the DRM.
Broom and Doctor (1995a) conducted a treatment study with an 11-year-
old male participant, D.F., who performed significantly below the level of a
group of 25 control participants in reading irregular words. D.F. also
made regularisation errors when reading these words (e.g., “tomb” read as
/t mb/), consistent with a weakness in the lexical reading mechanism and
a consequent over-reliance on the non-lexical mechanism (developmental
surface dyslexia). Using the DRM as a framework, Broom and Doctor
hypothesised that the breakdown in D.F.’s reading system was within the
lexical route, and more specifically the orthographic lexicon, as he had
more difficulty reading aloud low frequency irregular words than high fre-
quency irregular words, and he made homophone confusion errors (e.g.,
defining “pale” as if it were “pail” and “stare” as if it were “stair”)
suggesting that the breakdown was prior to semantics. The researchers
designed an intervention plan specifically to enable D.F. to develop and
utilise an orthographic reading strategy. Intervention consisted of treating
irregular words (that were matched on frequency with a list of untreated
irregular words) by asking the participant to write the words, discuss their
meanings, and name them. Words were practised in clinic sessions and as
homework. Post-intervention, D.F.’s oral reading of the trained irregular
words showed significant improvement. By contrast, there was no improve-
ment in performance on the matched, untreated irregular words, indicating
that the intervention did not generalise to untrained words. In their
summary, Broom and Doctor suggested that future interventions for
surface dyslexia should be aimed at establishing an orthographic lexicon
by training letter groupings to encourage generalisation, rather than adopting
a whole-word approach. They also suggested that words should be relevant
to daily reading activities for participants.
Different results were reported by Brunsdon et al. (2002) in their interven-
tion study with T.J., a 10-year-old boy who presented with mixed dyslexia;
that is, poor reading of both irregular words and nonwords associated with
damage to both lexical and non-lexical mechanisms of the DRM. Brunsdon
8 LAW AND CUPPLES

et al. hypothesised that the impairment in T.J.’s lexical reading mechanism


was located in the visual word recognition system (orthographic lexicon),
because assessments suggested that his phonological output lexicon and
semantic system were intact. They focused their intervention on increasing
the efficiency with which T.J. recognised high frequency regular and irregular
words using flash cards (some cards contained mnemonic cues for the target
words whereas others did not). T.J. was trained on a set number of target
words each week, which were practised at home in between sessions. After
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

intervention T.J. showed significantly improved performance in his oral


reading of trained words and improved performance in his oral reading of
untrained words (particularly in the second treatment study), but improve-
ments did not generalise to standardised measures of reading. Brunsdon
et al. reported that the intervention was successful in improving T.J.’s
visual word recognition using the lexical mechanism, although they noted
that there also appeared to be some transfer of learning to the non-lexical
mechanism, because T.J. showed improved performance in some aspects of
phonological processing post-intervention; namely, sounding out graphemes
and nonword repetition. This observed improvement in non-lexical
processing could be seen to contradict the independence of lexical from
non-lexical reading mechanisms as specified in the DRM. However, the
researchers argued against this interpretation on the grounds that “training
focused on visual word recognition does not entirely preclude some learning
of basic sublexical skills, especially when training in the lexical processing of
regular words is included” (p. 414).

The current study


Previous research demonstrates the potential value of using a theoretical
reading model, such as the DRM, to guide intervention for developmental
surface dyslexia. Nevertheless, important questions remain as to how inter-
vention programmes should be designed in order to optimise training
effects for individual children and to ensure generalisation of learning to
untreated words. These issues were addressed in the present study. The
main aim was to devise and evaluate methods for enhancing visual ortho-
graphic processing within the lexical reading route that would maximise gen-
eralisation of learning. To this end, intervention targeted letter groupings
rather than whole words, as recommended by Broom and Doctor (1995a).
Two boys, who presented with developmental surface dyslexia, participated
in two single case studies, each consisting of assessment and two stages of
intervention for their reading based primarily on the DRM. Both interventions
targeted visual word recognition processes: one through enhancing the speed
and accuracy with which participants identified briefly presented orthographic
sequences (Intervention I), and the second through provision of a strategy for
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 9

identifying commonly occurring letter patterns within words (Intervention II).


The DRM was selected as the primary theoretical framework because of its
use in previous research, and its utility for identifying specific intervention
targets within the lexical reading mechanism for children with surface dys-
lexia. Our overarching hypotheses based on previous research were as
follows.
1. The DRM would provide a useful framework for confirming the diag-
nosis of developmental surface dyslexia in two male participants and
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

for identifying specific areas of weakness within each participant’s


lexical reading route. This hypothesis was evaluated in Phase 1 of
the study.
2. The DRM would provide a useful framework for devising theoretically
motivated targeted treatment programmes to address specific areas of
weakness identified within each participant’s lexical reading route.
The success of the approach would be reflected in improved real
word reading for both boys post-intervention, consistent with enhanced
development of the lexical reading mechanism. This hypothesis was
evaluated in Phases 2 and 3 of the study.

In each phase of the current study, evaluation of these formal hypotheses


was supplemented by a more exploratory consideration of the extent to which
the results were consistent with expectations based on the DRM and/or other
current models of reading; in particular, the CDP+ and Harm and Seiden-
berg’s (1999) connectionist model (both described earlier). This secondary
focus of the research applied to all three phases.

GENERAL METHODOLOGY
This research encompassed three distinct phases with the same two partici-
pants taking part in each phase. Although the methodological design of the
research was two single case studies, the results of the interventions were
compared between participants to examine possible differences in outcomes,
especially as they related to the individual children’s processing strengths and
weaknesses. The first phase involved a comprehensive assessment of each
participant’s reading skills as identified within the DRM. Various reading,
language, and cognitive assessments were also administered during this
phase, to provide a comprehensive and detailed participant description. Inter-
vention was conducted during the second and third phases with selected pre-
and post-intervention assessments administered at the beginning and end of
each phase, respectively. In what follows, the method, results and a brief dis-
cussion of each phase are described.
10 LAW AND CUPPLES

Participants

H.F.
H.F.’s data were collected over a 312 year period when he was aged between
7;6 (years; months) and 11;1. He was one of twin boys (his sibling having no
reading difficulties). His motor and speech and language skills were age
appropriate and there were no reported medical or emotional difficulties
prior to the study. Australian English was the only language spoken at
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

home. Comprehensive language testing was not conducted as part of this


study because our focus was on the component processes engaged during
single word reading as represented in the DRM. It is nevertheless worth
noting that previous testing, when H.F. was 5 years old, indicated that his
verbal and non-verbal abilities were both in the high average range (on the
Wechsler Pre-school and Primary Scale of Intelligence–Revised; Wechsler,
1989). In addition, his school reported no concerns regarding H.F.’s oral
language. Of more direct relevance to the integrity of components of the
DRM, testing conducted as part of this study showed no evidence of articula-
tion difficulties or impairments in expressive or receptive vocabulary (single
word semantics, see Phase 1) or nonword repetition (see Phase 3). Indeed,
H.F.’s vocabulary knowledge placed him in the top 25% of children the
same age (see Table 3).
H.F.’s assessment and intervention sessions were conducted in a quiet
room at his home. His scores on non-standardised assessments were com-
pared to two separate age-matched control groups at different stages of the
study. A younger control group (N ¼ 14, mean age ¼ 9;2, SD ¼ 10.03
months) was used to evaluate H.F.’s initial assessment results and his
response to the first phase of intervention, which was conducted when he
was between 9 and 10 years of age. An older control group (N ¼ 15, mean
age ¼ 10;9, SD ¼ 5.4 months) was used to evaluate his response to the
second phase of intervention, which was conducted between the ages of
10;6 and 11;1. The control participants were recruited from two schools in
NSW. They were identified by their teachers as children with no learning,
language or cognitive difficulties and their parents gave written permission
for their participation in the study. Their results are presented alongside
H.F.’s in Tables 3, 4, and 6.

R.C.
R.C.’s data were collected over a 3 year period when he was aged between
9;0 and 12;0. He was the younger of two children. His speech and language
skills were age appropriate and there were no other reported medical or
emotional difficulties prior to the study. Australian English was the only
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 11

language spoken at home. Prior to taking part in the study, R.C. had under-
gone a comprehensive speech and language assessment which was adminis-
tered by a qualified speech-language pathologist. His oral language skills
were reported to be in the average to above average range. R.C.’s general
language skills were not further assessed as part of this study due to our par-
ticular focus on components of the DRM. Throughout testing, no articulation
difficulties were noted, and no impairments were evident in R.C.’s receptive
vocabulary (single word semantics, assessed in study Phase 1) or nonword
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

repetition (assessed in study Phase 3).


R.C.’s assessment and intervention sessions were conducted at school or
home. As for H.F., his scores on non-standardised assessments were com-
pared to two separate age-matched control groups at different stages of the
study. A younger control group (N ¼ 15, mean age ¼ 10;7, SD ¼ 5.4
months) was used to evaluate R.C.’s initial assessment results and his
response to the first phase of intervention, conducted when he was between
10 and 11 years of age. An older control group (N ¼ 17, mean age ¼ 11;2,
SD ¼ 3.9 months) was used to evaluate his response to the second phase of
intervention, which was conducted between the ages of 11;0 and 12;0. As
for H.F., the control participants were recruited from two schools in NSW.
They were identified by their teachers as children with no learning, language
or cognitive difficulties and their parents gave written permission for their
participation in the study. Their results are presented alongside R.C.’s in
Tables 3, 5 and 6.

PHASE 1: ASSESSMENT
Assessment materials were selected to investigate a range of reading and
related skills, as detailed in Table 1. Where appropriate, the same assessments
were administered to the younger control groups to obtain comparison scores.

Reading tests

Oral reading
Word identification and Word attack subtests from the Woodcock Reading
Mastery Tests –Revised (WRMT-R; Woodcock, 1987) were administered to
examine oral reading of real words and nonwords, respectively. These subt-
ests were administered according to the test manual. Oral reading of
regular words (which conform to spelling-to-sound rules, e.g., check), irregu-
lar words (which do not conform to spelling-to-sound rules, e.g., have), and
nonwords, was subsequently investigated explicitly using the Coltheart and
Leahy (1996) word list. Words and nonwords were printed on flashcards
12 LAW AND CUPPLES

TABLE 1
Phase one assessment materials according to target variable and relevant DRM
components

Variable DRM component Assessment materials

Oral reading Lexical and non-lexical WRMT–R word identification and word
mechanisms attack
Coltheart and Leahy (1996) regular,
irregular, and nonwords
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

Homophone selection Orthographic lexicon Manis et al. (1996) orthographic choice


Castles and Coltheart (1996) homophone
selection (2 tasks)
Visual-orthographic Letter identification/ Discrimination/recall of symbols, symbol
processing orthographic lexicon sequences, and nonwords
Seymour and Evans (1993) identity
matching task
Sound–symbol Orthographic lexicon, PhAB Rapid naming
associations phonological lexicon Learning associations between visual
symbols and names
Vocabulary Semantics Peabody Picture Vocabulary
Test– 3rd Edn
Test of Word Finding (H.F. only)
Phonological N/Aa McBride-Chang (1995) phoneme
awareness segmentation
Auditory perception N/Aa PALPA subtest 1, auditory analysis of
nonwords

WRMT-R ¼ Woodcock Reading Mastery Tests– Revised; PhAB ¼ Phonological Awareness


Battery; PALPA ¼ Psychological Assessment of Language Processing in Aphasia.
a
Phonological awareness and auditory perceptual ability are not part of the DRM but are widely
acknowledged as relevant pre-reading skills.

and shown individually to each participant. Their responses were audio-


recorded and transcribed.

Homophone selection
Three tasks were used to assess recognition of homophones: orthographic
choice (Manis et al., 1996); and two homophone selection tasks, one using
pairs of regular words, and the other using irregular word—nonword pairs
(Castles & Coltheart, 1996). In orthographic choice, participants were
asked to select the real word from a choice of two on each trial, one of
which was a homophonic nonword (e.g., sponge—spunge). In the two homo-
phone tasks, word pairs were shown on individual cards and a definition of
one of the words was read aloud. Participants were instructed to select the
written word that went with the spoken definition. There was no time limit
for these tasks and the word sets in the homophone tasks were controlled
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 13

for target word location (left or right side). Results for orthographic choice
were compared to norms cited in Manis et al., whereas results for the
Castles and Coltheart homophone tasks were compared to scores from the
respective age-matched control groups.

Visual orthographic processing

Discrimination and recall of single symbols and nonwords


Downloaded by [Texas A & M International University] at 02:56 24 August 2015

Participants’ ability to discriminate single symbols and individual non-


words was assessed first. A single symbol or nonword (“the target”) was dis-
played on a flashcard and then removed from view. A second flashcard was
presented immediately after the first. It contained two symbols or nonwords:
one was the target and the other a distracter (distracter nonwords were homo-
phones, e.g., zoal—zole). Participants were asked to identify the symbol or
nonword they had seen previously. Stimuli comprised a total of 25 symbols
and 20 nonword homophone pairs. Each symbol and nonword was used
once as target and once as distracter. Target location on the flashcards (left
or right) was balanced.
Ability to recall the symbols and nonwords used for discrimination was
assessed next. A sub-set of 10 symbols used in the discrimination task was
intermixed with 10 new symbols and shown to participants on individual
flashcards. They were asked to report whether or not they had seen the
symbols in the previous task. An analogous task was designed and
implemented using a sub-set of 10 previously seen nonwords and 10 new
nonwords.

Sequential symbol recall—immediate and delayed


Ability to recall symbol sequences was assessed using a selection of 32
symbols, which were randomly assigned to four sets of two symbols each,
four sets of three symbols each, and three sets of four symbols each. All of
the symbol sequences were printed on flashcards and presented to participants
for five seconds each. Immediately after presentation, participants were asked
to re-create the symbol sequence from a set of flashcards that contained indi-
vidual printed symbols, both seen (correct) and unseen (distracters). The task
was repeated one week later when a one-minute delay was introduced
between seeing the symbol sequence and beginning to recreate it with the
individual symbol cards.

Visual orthographic processing efficiency


An orthographic encoding task based on the identity matching task
described by Seymour and Evans (1993) was used to investigate participants’
14 LAW AND CUPPLES

accuracy and speed of orthographic processing. A total of 80 pairs of non-


words was displayed in random order on a computer screen for exposure dur-
ations of 650 milliseconds and 340 milliseconds. Nonword pairs were either
(a) the same, (b) different by one letter, or (c) different by two letters or more.
Participants were asked to identify whether the nonwords in each pair were
the same or different. Answers and response times were recorded using
DMDX software (Forster & Forster, 2003).
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

Sound–symbol associations

Rapid naming
The naming speed test from the Phonological Awareness Battery (PhAB;
Frederickson, Frith, & Reason, 1997) was administered to investigate rapid
automatised naming ability (RAN) for digits and pictures. Scores were
devised using the norms in the administration manual.

Learning associations between visual symbols and their names


Participants’ ability to learn associations between sounds and symbols was
assessed in three phases: (1) a learning phase in which each of nine symbols
was presented three times with its corresponding name; (2) a recall phase in
which participants were asked to recall as many of the symbol names as they
could; and (3) an association phase in which participants were shown the
symbols individually and asked to recall their names.

Vocabulary
The Peabody Picture Vocabulary Test–3rd Edition (PPVT-III; Dunn & Dunn,
1997) was administered to assess receptive vocabulary. On each trial, partici-
pants were asked to select one of four pictures to match a spoken word, with
correct performance taken to reflect intact semantic processing at the single
word level as per the lexical mechanism of the DRM (see Figure 1). The
Test of Word Finding (TWF; German, 1986/1989) was also administered to
H.F., because he had not completed previous standardised testing for
naming which is also part of the lexical mechanism of the DRM. The TWF
was not administered to R.C., because his naming skills had been evaluated
previously as part of a comprehensive spoken language assessment where
his performance for naming was found to be in the typical range for his age.

Phonological awareness and auditory perception


Phonological awareness (PA) was assessed using the Phoneme Segmentation
Test described in McBride-Chang (1995). Participants’ scores were compared
to those reported for McBride-Chang’s control participants (mean age ¼ 9;3,
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 15

SD ¼ 6.2 months). An auditory perception task was also administered. For


this purpose, we used subtest 1, auditory analysis of nonwords, from the
Psychological Assessment of Language Processing in Aphasia (PALPA;
Kay, Lesser, & Coltheart, 1992). Participants’ results were compared to the
published norms for that subtest.

RESULTS
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

H.F.
Table 2 shows the scores attained by H.F. on the WRMT-R and the Coltheart
and Leahy word list at the initial assessment. Table 3 shows his scores on the
additional language, cognitive, and silent reading measures, along with corre-
sponding control data where applicable.

Oral reading
On the WRMT-R, H.F.’s oral reading of real words (word identification)
was below the level expected for a child of his age (13th percentile). By con-
trast, his oral reading of nonwords (word attack) was at the 35th percentile,
well within the average range (which extends from the 20th to the 80th per-
centile). He made regularisation errors (e.g., reading “find” as /f nd/) and
visual errors (e.g., reading “hurry” as “hungry” and “stove” as “story”).
This pattern of superior performance on nonwords as compared to real
words was confirmed by H.F.’s reading of the Coltheart and Leahy (1996)
word list, where he performed within the average range on nonwords and
regular real words, both of which can be read aloud using spelling-sound
rules, but well below average on irregular words (see Table 2). H.F.’s
pattern of reading difficulty can be characterised as a pure form of develop-
mental surface dyslexia; that is, a selective difficulty in reading irregular
words.

Vocabulary, cognition, and silent reading


As the data in Table 3 show, H.F’s receptive and expressive vocabulary,
phonological awareness, auditory perception (PALPA 1), and ability to
associate names with symbols were comparable to or better than age-
matched controls. By contrast, he showed some deficits in silent reading
(homophone selection), visual-orthographic processing, and rapid naming.

Homophone selection. On the Orthographic Choice Task, H.F. identified


27/52 (52%) words correctly, a score that falls more than 2 standard devi-
ations below the mean score reported by Manis et al. (1996) for a group of
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

16
LAW AND CUPPLES
TABLE 2
Results on standardised oral reading assessments for H.F. and R.C. at initial assessment, pre-intervention II, and post-intervention II

H.F. R.C.

Number correct (percentile) Number correct (percentile)

Assessment Initial Pre-int II Post-int II Post–Pre (sig) Initial Pre-int II Post-int II Post– Pre (sig)

WRMT-R
Word identification (106) 33 (13) 56 (9) 75 (52) +19 (p , .001) 25 (1) 50 (2) 58 (5) +8 (p ¼ .04)
Word attack (45) 22 (35) 29 (39) 36 (70) +7 (n.s.) 22 (29) 27 (29) 29 (34) +2 (n.s.)
Coltheart and Leahy a
Irregular (30) 8 (8) 18 (6) 20 (9) +2 (n.s.) 9 (2) 9 (, 1) 18 (3) +9 (p ¼ .004)
Regular (30) 22 (24) 28 (20) 29 (43) +1 (n.s.) 21 (6) 25 (5) 29 (49) +4 (n.s.)
Nonwords (30) 16 (24) 25 (38) 28 (72) +3 (n.s.) 16 (14) 23 (15) 24 (38) +1 (n.s.)
a
Comparison scores taken from Edwards and Hogben (1999).
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 17

non-reading disabled children aged 8.5 years (SD ¼ 0.64) (see Table 3).
Additionally, H.F. scored 14/30 on the regular word homophone assessment,
a score that does not differ significantly from chance (Binomial, p ¼ .86), and
8/30 on the irregular word—nonword pairs, choosing the incorrect nonword
21 times, and giving no response to the homophone pair “shove – shuv”.
H.F.’s homophone scores were well below the range of the control group
(see Table 3).
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

Visual orthographic processing. In regard to visual orthographic proces-


sing, H.F’s scores were within the range observed for his age-matched control
group with one exception. Delayed recall of nonwords was significantly
poorer than that of the control group, indicating some difficulty in retaining
visual-orthographic information, a deficit that might affect the long-term
storage of words’ orthographic representations. H.F. also showed a different
pattern of results from the control group on the identity matching task.
Control participants extracted information at the faster presentation rate
(340 ms) as accurately as they did at the slower rate (of 650 ms), thus result-
ing in a non-significant effect of exposure duration on response accuracy, t(1,
13) ¼ .893, p ¼ .388. In fact, the control group’s mean score for accuracy
increased by 1.3 items on average at the faster presentation rate, and none
of the individual control participant’s scores decreased in accuracy by more
than 6 (out of 80) items at the faster presentation rate compared to the
slower rate. Notably, this pattern of results was obtained despite the control
participants also responding significantly more quickly to items presented
at the faster rate, t(1, 13) ¼ 5.119, p , .001. By contrast H.F.’s score
decreased by 9 out of 80 items, McNemar x 2(1, N ¼ 80) ¼ 7.00, p ¼ .008.
This decrease in performance suggests that he was not able to extract infor-
mation about the nonword pairs as efficiently as control participants,
especially at the shorter exposure duration. In terms of response times, H.F.
also showed a pattern that was different from the control participants, with
mean response times that fell above the control group range at both presen-
tation rates (see Table 3).

Rapid automised naming. H.F. scored slightly below the level expected
for his age on digit naming but in the typical range for picture naming; indi-
cating a selective weakness in RAN.

R.C.
Table 2 shows the scores attained by R.C. on the WRMT-R and the Coltheart
and Leahy word list at the initial assessment. Table 3 shows his scores on the
additional language, cognitive, and silent reading measures, along with corre-
sponding control data where applicable.
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

18
TABLE 3
Initial assessment results on language, cognitive, and silent reading measures for H.F., R.C., and their respective control groups

LAW AND CUPPLES


H.F. R.C.

Control group Control group

Assessment Score Mean (SD) Range Score Mean (SD) Range

Vocabulary
PPVT-III raw score (percentile) 130 (75) – – 138 (82) – –
TWF raw score (percentile) 73 (77) – – N/A – –
Homophone tests
Orthographic choice (% correct) 52% 74.6 (10.3)a – 42.3% 74.6 (10.3)a –
Regular-Regular word (no. correct out of 30) 14 24.3 (3.5) 21–30 19 28.6 (1.9) 23–30
Irregular-nonword (no. correct out of 30) 8 25.8 (4.0) 17–30 7 29.7 (0.9) 27–30
Visual orthographic processing (no. correct)
Symbol discrimination (25) 25 24.9 (0.3) 24–25 25 25 (0) 25–25
Symbol recal – delayed (20) 18 17.8 (1.2) 16–20 19 18.6 (1.7) 14–20
Sequential symbol recall – immediate (11) 11 11 (0) 11–11 11 11 (0) 11–11
Sequential symbol recall – delayed (11) 10 10.3 (1.4) 6– 11 9 10.9 (0.4) 10–11
Nonword discrimination (40) 38 37.1 (1.6) 35–39 38 39.1 (1.5) 36–40
Nonword recall – delayed (20) 8 15.28 (1.8) 13–18 14 15.6 (2.0) 13–20
Identity matching (650 ms)
Number correct (80) 77 72.8 (4.6) 65–78 69 73.2 (5.3) 65–79
Reaction time (ms) 1750 1365 (189) 1101–1690 1723 1414 (150) 1178–1690
Identity matching (340 ms)
Number correct (80) 68 74.1 (4.2) 65–79 61 74.7 (4.9) 65–80
Reaction time (ms) 1452 1098 (202) 870– 1450 1367 1190 (229) 908 –1685
RAN (percentiles)
Pictures 42 2
Digits 14 8
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA


Sound symbol learning
Symbol learning 9 8.2 (0.97) 6 –9 9 8.53 (0.83) 6 –9
Sound recall 4 2.71 (1.26) 1 –6 5 3.67 (1.05) 2 –6
Sound-symbol 2 1.43 (1.1) 0 –3 1 2.93 (1.67) 0 –7
PA
Phoneme segmentn (no. correct out of 24) 19 13.9 (5.3)b 12 13.9 (5.3)b
PALPA 1
Same (36) 36 35.7 (0.56)c 36 35.7 (0.56)c
Different (36) 33 35.09 (2.34)c 34 35.09 (2.34)c

Dashes indicate that data were not collected for that cell. PALPA ¼ Psychological Assessment of Language Processing in Aphasia; RAN ¼ rapid automa-
tised naming.
a
Taken from control group scores in Manis et al. (1996); bNorms taken from McBride-Chang (1995); cNorms taken from the PALPA assessment by Kay et al.
(1992).

19
20 LAW AND CUPPLES

Oral reading
On the WRMT-R, R.C.’s oral reading of real words (word identification)
was at the 1st percentile, a level well below that expected for a child of his
age, whereas his reading of nonwords (word attack) was at the 29th percentile,
within the average range. R.C. made regularisation errors (e.g., reading “island”
as / zl n/ and “tomb” as /t mb/) and decoding errors (e.g., reading “cough” as
/k l h / and “peng” as /p1nd/). This pattern of poorer performance on real
words than nonwords was confirmed by R.C.’s reading of the Coltheart and
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

Leahy (1996) word list, where he performed less accurately in reading irregular
words than nonwords, even though his scores on both item types were below
age expectations (see Table 2). R.C.’s pattern of reading difficulty can be
characterised as developmental surface dyslexia to the extent that his reading
of irregular words was poorer than his reading of nonwords. His reading diffi-
culty was not as pure (selective) as that of H.F., however.

Vocabulary, cognition, and silent reading


The data in Table 3 show that R.C. resembled H.F. in achieving scores that
were comparable to an age-matched control group on assessments of recep-
tive vocabulary, PA, auditory perception (PALPA 1), and ability to associate
names with symbols. By contrast, he showed some deficits in silent reading
(homophone selection), visual-orthographic processing, and rapid naming.

Homophone selection. On the Orthographic Choice Task, R.C. identified


22/52 (42.3%) words correctly, a score that falls more than 2 standard devi-
ations below the mean score reported by Manis et al. (1996) for a group of
non-reading disabled children aged 8.5 years (SD ¼ 0.64). In addition, he
scored 19/30 on the regular word homophone assessment, a score that does
not differ significantly from chance (Binomial, p ¼ .20), and 7/30 on the irre-
gular word—nonword pairs. These homophone scores were well below the
range of his age-matched control group (see Table 3).

Visual orthographic processing. As for H.F., R.C’s performance in visual


orthographic processing was generally within the range observed for his age-
matched control group. There was, however, one apparent weakness, which
related to the retention of sequential symbol information after a delay. R.C.
performed just outside the control group range on this aspect, which might
affect the long-term storage of orthographic representations. Results from
the Identity Matching task also revealed a different pattern for R.C. compared
to his age-matched control group in terms of accuracy. As for H.F.’s control
group, participants extracted information at the faster presentation rate
(340 ms) as accurately as they did at the slower rate (of 650 ms), thus
resulting in a non-significant effect of exposure duration on response
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 21

accuracy, t(1, 14) ¼ 1.314, p ¼ .210. The control group’s mean score for
accuracy increased by one item on average at the faster presentation rate,
and none of the individual control participant’s scores decreased in accuracy
by more than 5 (out of 80) items at the faster presentation rate compared to the
slower rate. This pattern of results was once again obtained despite the control
participants responding significantly more quickly to items presented at the
faster rate, t(1, 14) ¼ 4.66, p , .001. By contrast, R.C. showed a significant
decrease in score of 8 out of 80 items, McNemar x 2(1, N ¼ 80) ¼ 6.125, p ¼
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

.013. This decrease suggests that R.C. was not able to extract information
about the nonword pairs as efficiently as control participants at the shorter
exposure duration. This pattern was obtained despite R.C.’s mean response
time being within the range observed for the matched control group at the
faster presentation rate and just above the control range at the slower presen-
tation rate (see Table 3).

Rapid automatised naming. Results from the rapid naming subtests show
that R.C. exhibited difficulty with naming speed compared to his age-matched
peers, with scores falling below the 10th percentile on both digit and picture
naming.

DISCUSSION
The oral reading results obtained by both H.F. and R.C. are consistent with a
diagnosis of developmental surface dyslexia; that is, relatively poorer per-
formance in reading irregular words than nonwords. However, in the case
of R.C., who was slightly older at the time of testing, there were also some
milder difficulties apparent in the reading of regular and nonwords on the
Coltheart and Leahy (1996) word list. According to the DRM, a difficulty
in reading irregular words reflects a weakness in the lexical mechanism,
two important components of which are the orthographic lexicon (for long-
term storage of words’ orthographic representations) and semantics.
Neither H.F. nor R.C. showed evidence of semantic difficulties at the
single word level, with receptive vocabulary (PPVT-III) scores at the 75th
and 82nd percentiles, respectively. Three homophone tasks were adminis-
tered to assess the integrity of the orthographic lexicon (see Coltheart, Mas-
terson, Byng, Prior, & Riddoch, 1983, for the use of homophones in the
assessment of surface dyslexia). Both H.F. and R.C. performed well below
the level of age-matched control participants on all three tasks, consistent
with an impairment or weakness at this level.
Further investigation of the participants’ visual-orthographic processing
was conducted using a range of tasks requiring discrimination or delayed
recall of symbols, symbol sequences, and nonwords. Although most of
22 LAW AND CUPPLES

these tasks did not discriminate clearly between the participants and their
respective age-matched control groups, H.F. was less able to recall
nonword letter patterns after a delay and R.C. showed a mild difficulty
with sequential symbol recall after a delay. More importantly, however,
both boys achieved significantly and markedly poorer than expected perform-
ance on a test of identity matching, the results of which showed that neither
H.F. nor R.C. was able to extract orthographic information about pairs of non-
words as efficiently as members of their respective control groups at the
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

shorter exposure duration of 340 ms. Finally, RAN proved difficult for both
boys, but especially for R.C., whose scores on picture and digit naming fell
at the 2nd and 8th percentiles, respectively. By contrast, H.F. showed evi-
dence of a milder difficulty with digit naming only.
In sum, the weaknesses observed in H.F.’s and R.C.’s oral reading, homo-
phone selection, orthographic processing efficiency, and RAN, are all consist-
ent with a diagnosis of developmental surface dyslexia, which, according to
DRM, reflects under-development of the lexical reading mechanism (see
Figure 1). On the other hand, neither boy performed poorly on assessments
of PA (phoneme segmentation), auditory perception (PALPA 1), or sound
symbol learning, all skills that could have assisted in the development of
an adequate non-lexical reading mechanism. This pattern of results provides
support for our first hypothesis in that the DRM provided a useful framework
for: (1) confirming the diagnosis of developmental surface dyslexia in both
male participants; and (2) identifying specific areas of weakness within
each participant’s lexical reading route. In particular the results show evi-
dence of weaknesses at the orthographic lexicon, letter identification, and
the phonological lexicon.
In devising an intervention programme for H.F. and R.C. the question
arises as to whether, and if so how, their observed weaknesses in orthographic
processing efficiency, and to a lesser extent RAN, might have contributed to
their reading difficulties. According to the DRM, the lexical reading mechan-
ism and hence the ability to read irregular words correctly, relies on parallel
processing of letters and recognition of words and letter patterns as familiar.
In this context, the pattern of results obtained on our identity matching task is
relevant, in that both H.F. and R.C. were less efficient than age-matched con-
trols in extracting letter pattern information from orthographic stimuli. It is
possible that this pattern of results reflected their use of a serial, letter-by-
letter reading strategy rather than parallel mapping (Grainger & Ziegler,
2011), which in theory could make their ability to extract specific letter pos-
ition information slower and less accurate than controls. This difficulty, along
with cognitive difficulties related to RAN, which may hinder the formation of
links between orthographic and phonological representations (Manis, Seiden-
berg, & Doi, 1999; Wolf & Bowers, 1999), may have resulted in reduced
development of the orthographic lexicon.
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 23

As regards alternative theoretical interpretations, both the DRM and the


CDP+ attribute developmental surface dyslexia to a selective weakness in
the lexical reading mechanism. By contrast, Harm and Seidenberg (1999)
adopted the view that developmental surface dyslexia is a general reading
delay, which may be caused by a lack of reading experience or slow rate of
learning, access to degraded orthographic input, or a difficulty linking ortho-
graphy with phonology. It is a limitation of this research that we cannot defi-
nitively rule out any of these potential factors, but especially the possibility of
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

degraded orthographic input (given both boys’ results on the identity match-
ing task), or a difficulty in linking orthography to phonology (in light of RAN
results). Nevertheless, H.F.’s reading profile suggests a pure form of develop-
mental surface dyslexia (problems reading irregular words but not nonwords
in the presence of typical phonological skills), a profile that Harm and Seiden-
berg did not simulate.

PHASE 2: INTERVENTION I
In line with our assessment results showing evidence of a weakness at the
level of the orthographic lexicon in both participants (according to the
DRM), reading intervention focused on development of this component. In
doing so, we were conscious that previous interventions for surface dyslexia
aimed at establishing word-specific entries within the orthographic lexicon
have involved mnemonic cues with little success in generalising to untreated
items (with the exception of Brunsdon, Hannan, Nickels et al., 2002). In this
study therefore, unlike previous research, treatment was devised to remediate
the efficiency of visual-orthographic processing directly by targeting inter-
vention at the level of letter identification and the orthographic lexicon and
to observe effects on reading and related skills.
At the time of writing, there was only one published therapy technique
designed to address poor development of the orthographic lexicon due to pro-
blems with visual-orthographic processing efficiency. Judica, De Luca, Spi-
nelli, and Zoccolotti (2002) reported successful training for a group of 18
Italian children with developmental surface dyslexia. The study showed that
reading performance on words and nonwords improved and that fixation dur-
ations reduced to the normal range when children were trained to read aloud
and spell briefly presented single words. The authors concluded that training
had taught the children to extract letter information with increased speed and
accuracy. In Judica et al.’s intervention, items were real words, ranging from
high to low frequency, which were displayed for durations of 60 to 150 ms.
Post-intervention results showed evidence of generalisation, in that children’s
reading accuracy and speed improved for nonwords as well as real words, as
did their ability to make lexical decisions. The study by Judica et al. was
24 LAW AND CUPPLES

used as a basis for the first phase of intervention described here, although there
were several important differences between this research and theirs.
Participants in the study by Judica et al. (2002) were presented with real
words during intervention, which they were asked to read aloud and spell.
In this way, all components of the lexical reading mechanism of the DRM
were directly targeted for treatment. By contrast, in this research, participants
were not asked to read or spell real words during treatment. Rather, nonwords
were employed as training stimuli and presented to participants in the exper-
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

imental format of the same–different (identity) matching task that was used
during the assessment phase to identify differences in visual-orthographic
processing efficiency. In short, participants were asked to indicate by pressing
a button on a computer whether the letters and their positions in pairs of
briefly presented nonwords were the same or different. Nonwords were
selected for the intervention to ensure: (1) that participants had to focus on
constituent letters with no lexical information available to assist them, and
(2) that intervention was not targeting development of specific entries
within the orthographic lexicon. Using this procedure, we aimed to ensure
that intervention targeted the early stages of orthographic processing, in par-
ticular, the transfer of information from letter identification to the ortho-
graphic lexicon, but no other component of the lexical reading mechanism
(e.g., semantics or the phonological lexicon).
Based on findings reported by Judica et al. (2002), we hypothesised that
training participants to perform a same–different task on orthographic
sequences presented at short duration would encourage them to extract infor-
mation about letter identity and pass it on to the orthographic lexicon with
increased speed and accuracy, possibly through use of a parallel processing
strategy (a strategy that is reportedly required for development of the
lexical reading mechanism) (Seymour & Evans, 1993). In line with this
view, we predicted improvements in both regular and irregular word
reading from pre- to post-intervention.

Materials and procedure

Pre- and post-intervention assessment


To evaluate the effectiveness and specificity of training, assessments of
oral reading, silent reading (lexical decision), phonological processing, and
RAN were conducted pre- and post-intervention.

Real word reading. Participants’ oral reading progress was assessed


using the 300 most frequent words in English (Fry, 1980), which were
shown individually on a computer screen. H.F. and R.C. were asked to read
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 25

each word aloud. The errors made on this list by each individual participant
were incorporated into a lexical decision (silent reading) task.

Lexical decision. Words read aloud incorrectly from the list of 300 most
frequent words were matched individually with orthographically legal non-
words that differed by one letter from the real word targets (Lexical
decision—Frequent words). This procedure resulted in a list of 29 word—
nonword pairs for H.F. and 55 pairs for R.C. Items in the list for each partici-
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

pant were then randomly ordered and shown individually to the participants
using DMDX software. Participants were asked to indicate using left and
right shift keys on the computer keyboard, whether each item was a real
word or a nonword.
In addition, a homophone lexical decision task was administered using
DMDX. Irregular words and visual foils (e.g., worry—wurry) taken from
the Castles and Coltheart (1996) homophone test were presented one at a
time on a computer screen. H.F. and R.C. were asked to look at each item
and decide whether or not it was a real word. The computer shift keys were
used to respond.

Phonological processing. Assessments of phonological processing


ability were included both pre- and post-intervention to evaluate the speci-
ficity of the remediation programme, which we predicted would not affect
phonological ability. These tests were therefore used as a measure to identify
general improvement in reading-related skills that were unrelated to the inter-
vention. Three tasks were used for this purpose: McBride-Chang’s (1995)
Phoneme Deletion Test, Perin’s (1983) Spoonerism Test, and McBride-
Chang’s (1995) Position Analysis Test. For phoneme deletion, participants
were presented with individual spoken nonwords and asked to repeat each
one aloud and then to say the nonword again without a particular sound;
for example, to say “melvz” without the “v”. In the Spoonerism task, partici-
pants were asked to exchange the first letters of two spoken words; for
example, to exchange the first two letters of “Bob Marley” to become
“Mob Barley.” For the Position Analysis task, participants were presented
with individual spoken nonwords and asked to repeat each one, and then to
identify which sound came before or after another sound contained in the
word; for example, to identify which sound comes before the “r” in
“kroost” (correct response ¼ /k/).

Rapid naming. To measure any change in rapid naming ability, the PhAB
naming speed subtest was re-administered pre- and post-intervention as per
the assessment instructions.
26 LAW AND CUPPLES

Intervention

Tasks. For the intervention, pairs of nonwords were shown to participants


on a computer screen. The nonwords differed by one letter at the beginning,
middle or end. Each list of nonword pairs was displayed at different presen-
tation speeds over the course of the intervention, beginning at a slower speed
and gradually increasing to a faster speed.
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

Materials. A total of 108 pairs of four-letter, single-syllable nonwords


was devised (e.g., kier—kief). These nonword pairs were used to create
three intervention lists: In List 1, the nonwords in each pair differed by a
letter at either the beginning (zain—tain) or the end (kier—kief); in List 2
the nonwords in each pair differed by a letter at either the beginning or the
middle (kier—kiar); and in List 3, the nonwords in each pair differed by a
letter at either the middle or the end.

Procedure. Intervention sessions took place twice weekly in a quiet room


at the participants’ homes. Each intervention session lasted approximately 45
minutes. A single nonword list was presented on a computer screen four times
during each intervention session using the DMDX software. H.F. and R.C.
used the right and left shift keys to record where they thought the letters dif-
fered in each nonword pair (e.g., left shift key ¼ nonwords differed at the
beginning, right shift key ¼ nonwords differed at the end). At the end of
each presentation, they were provided with feedback on the number of
correct responses and were encouraged to beat that score on the next presen-
tation. Both participants were given instructions before each presentation and
were told where to expect the letter differences (e.g., at either the end or the
beginning of the word). Four practice items were administered before each
presentation with verbal feedback. In each session, the same list was given
at the same speed for all four trials. Each intervention list was used for
four intervention sessions. On the first of the intervention sessions for a
given list, each nonword pair was displayed on the computer screen for
650 ms, for the second intervention session the nonword pairs were displayed
for 507 ms, on the third session for that list the nonword pairs were displayed
for 340 ms, and on the final administration the nonword pairs were displayed
for 170 ms. Intervention List 1 was used for the first four intervention ses-
sions, intervention List 2 for the next four sessions, and intervention List 3
for the final four sessions. Thus, there were 12 intervention sessions in
total, spanning six weeks. The DMDX software collected response times
and errors made by the two participants. The average reaction time was cal-
culated and recorded for all of the items judged correctly.
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 27

Reliability
Inter-rater reliability data were collected for 15% of the pre- and post-inter-
vention oral reading data. Participants’ responses were scored by an indepen-
dent speech pathologist who was unaware of their pre- versus post-
intervention status. Classification reliability (as correct or incorrect) ranged
from 91–100% with a mean of 96.5%.
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

RESULTS

H.F.
Results for H.F. following intervention I are summarised in Table 4. He made
statistically significant gains in oral reading of Fry’s (1980) list of 300 most
frequent words, with his score now falling in the range of the age-matched
control group. When the words on the list were analysed as regular and irre-
gular according to the DRM (Rastle & Coltheart, 1999), H.F. showed

TABLE 4
Intervention I results for H.F.

H.F.’s score Significance a H.F.’s control group

Assessment Pre- Post- p-value Mean (SD) Range

Fry’s 300-word list 271/300 293/300 , .001 293 (5.94) 279 –300
Lexical decision– Homophones
No. correct 33/60 46/60 .001 46 (5.03) 35–52
Reaction time (ms) 2533.14 1312.04 , .001 1473 (482) 1006–2564
Lexical decision– Frequent words
No. correct 47/58 45/58 0.804 – –
Reaction time (ms) 2229.03 1460.79 0.019 – –
Identity matching (650 msec)
No. correct 77/80 72/80 0.063 72.77 (4.6) 65–78
Reaction time (ms) 1749.92 1202.16 ,.001 1365 (189) 1100–1689
Identity matching (340 msec)
No. correct 68/80 67/80 1.000 74.08 (4.19) 65–79
Reaction time (ms) 1451.84 1127.76 0.011 1098 (202) 869– 1450
RAN (percentile)
Pictures 60 55 – – –
Digits 22 32 – – –
Position analysis 22/24 21/24 1.000 – –
Spoonerisms 15/18 15/18 1.000 – –
Phoneme deletion 17/24 22/24 0.125 – –

Dashes indicate that data were not collected for that cell. RAN ¼ rapid automatised naming.
a
Changes in response times were analysed using repeated measures t-tests. Item analyses of correct
responses were conducted using McNemar x 2 test.
28 LAW AND CUPPLES

improved ability to read both word types. His score on irregular words
improved from 80/90 pre-intervention to 88/90 post-intervention, McNemar
x 2(1, N ¼ 90) ¼ 4.90 p ¼ .027; and his score on regular words improved
from 191/210 to 205/210, McNemar x 2(1, N ¼ 210) ¼ 7.68 p ¼ .006. H.F.
also improved significantly in making lexical decisions about printed homo-
phones, achieving levels of accuracy and response speed similar to the control
participants after intervention (see Table 4). While no changes were observed
in H.F.’s accuracy on the frequent word–lexical decision task from pre- to
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

post-intervention, there was a significant improvement in response speed on


this task, thus suggesting an increase in H.F.’s orthographic processing effi-
ciency (i.e., improved response speed without a corresponding decrease in
accuracy).
Turning attention to the identity matching task, which reflects most closely
the skills targeted during intervention, H.F.’s response speed showed a signifi-
cant improvement from pre- to post-intervention to reach a level similar to
that of the control group. Moreover, this improvement in speed of responding
was achieved without a significant reduction in accuracy, particularly at the
shorter exposure duration of 340 ms (see Table 4).
Finally, and as expected, there were no significant changes observed in the
control measures of RAN or phonological processing (including position
analysis, spoonerisms and phoneme deletion), indicating that the intervention
was selective in addressing visual-orthographic processing.

R.C.
R.C. made statistically significant and substantial gains in oral reading of the
300 most frequent words from pre- to post-intervention I, but unlike H.F., his
score remained below the level of the matched control group (see Table 5).
Further analysis showed that R.C. resembled H.F. in showing significantly
improved performance on regular words, from 173/210 pre-intervention to
197/210 post-intervention, McNemar x 2(1, N ¼ 210) ¼ 14.69 p , .001.
On the other hand, although his reading of irregular words improved (from
72/90 to 79/90), the change was not significant, McNemar x 2(1, N ¼ 90) ¼
1.57 p ¼ .211. R.C. also showed no significant improvement in response
time or accuracy on the two lexical decision tasks.
Like H.F., R.C. showed improved performance on the identity matching
task from pre- to post-intervention. In the case of R.C., however, response
speed changed little over the course of the intervention, whereas accuracy
at the faster presentation rate improved substantially (by 8 items out of 80)
to within the control group range (see Table 5).
Finally, and as expected, R.C.’s scores for the control measures of RAN
and phonological processing showed no significant improvements from
pre- to post-intervention, indicating the specificity of the intervention effects.
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 29

TABLE 5
Intervention I results for R.C.

R.C.’s score Significance a R.C.’s control group

Assessment Pre- Post- p-value Mean (SD) Range

Fry’s 300-word list 245/300 276/300 , .001 299.2 (1.32) 297 –300
Lexical decision—Homophone
No. correct 34/60 28/60 .310 53.63 (4.99) 43 – 60
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

Reaction time (ms) 929.85 ms 721.67 ms .121 1400 (339) 1034–2158


Lexical decision—Frequent words
No. correct 70/110 66/110 .134 – –
Reaction time (ms) 1259.3 ms 1364.84 ms .253 – –
Identity matching (650 msecs)
No. correct 69/80 72/80 .250 73.2 (5.29) 65–79
Reaction time (ms) 1723.42 1424.97 .152 1414 (150) 1177–1689
Identity matching (340 msecs)
No. correct 61/80 69/80 .008 74.67 (4.94) 65–80
Reaction time (ms) 1366.82 1215.01 .951 1190 (226) 908 –1685
RAN (percentile)
Pictures 14 11 – – –
Digits 22 9 – – –
Position analysis 20/24 20/24 1.000 – –
Spoonerism 13/18 18/18 .063 – –
Phoneme deletion 18/24 20/24 .727 – –

Dashes indicate that data were not collected for that cell. RAN ¼ rapid automatised naming.
a
Changes in response times were analysed using repeated measures t-tests. Item analyses of correct
responses were conducted using McNemar x 2 test.

DISCUSSION
The aim of Intervention I for H.F. and R.C. was to increase the efficiency of
visual-orthographic processing. Activities were devised on the assumption
that weakness in visual-orthographic processing prevented typical develop-
ment of the orthographic lexicon for both boys. In terms of the DRM, the
intervention targeted areas of specific weakness, namely letter identification
and the efficient transfer of information to the orthographic lexicon. The pro-
posal based on previous research was that promotion of a more holistic strat-
egy in the process of letter identification would facilitate development of the
lexical route.
The intervention was effective in that both boys showed significant
improvement in the primary intervention task—identity matching—from
pre- to post-intervention. H.F.’s response speed improved to within the
control group range at both exposure durations with no reduction in accuracy,
and R.C.’s accuracy at the shorter exposure duration improved to within the
control group range with no significant reduction in speed. As predicted if the
30 LAW AND CUPPLES

intervention were successful, both boys’ oral reading of commonly occurring


untrained real words (Fry’s 300-word list) also improved significantly from
pre- to post-intervention, lending support to the view that increased efficiency
in the processing of visual-orthographic information would lead to improved
accuracy in real word reading. In a general sense, these findings provide
support for our second research hypothesis in that a targeted treatment pro-
gramme devised using the DRM as a framework resulted in improved real
word reading for both participants, consistent with enhanced development
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

of the lexical reading mechanism.


Despite the similarly positive intervention outcomes for both participants,
they also differed in respect to their post-intervention performance on irregu-
lar word reading and lexical decision. Whereas H.F. showed significantly
improved performance on irregular word reading and both lexical decision
tasks after intervention, R.C. did not. In fact, R.C.’s performance on the
lexical decision tasks remained essentially unchanged from pre- to post-inter-
vention, with his score for homophone lexical decision being at chance level
at both assessments. It is possible that R.C.’s failure to respond accurately on
some or all of these measures might have reflected a strategic emphasis on
speed of response rather than accuracy. Some evidence for this suggestion
comes from the Lexical Decision—Homophone task, in which R.C.’s
response times were considerably shorter than those of all control group
participants.
H.F.’s and R.C.’s results are essentially in line with those reported by
Judica et al. (2002) in that the ability to read real words improved significantly
after intervention. In contrast, however, only H.F. showed some improvement
in homophone lexical decision and neither boy improved in his ability to
make accurate lexical decisions about frequently occurring words (although
H.F. was faster to respond post-intervention). Some of these differences
between our results and those reported by Judica et al. are presumably due
to methodological differences between the two studies (e.g., Judica et al.
used real words in their intervention whereas we used nonwords) and
perhaps also differences between Italian and English orthographies, which
necessitate the use of different criteria for identifying a child with surface dys-
lexia in the two languages.
Judica et al. (2002) concluded that their intervention had “increased the
speed and accuracy of stimulus processing” (p. 195), because participants
were able to view the word for only brief periods and therefore became
more efficient at identifying the letters in the word during this brief exposure.
This account would also explain the results of our study, in that the partici-
pants may not have had time to complete serial processing on the presented
orthographic sequences (their original strategy), and so may have adopted a
more parallel (holistic) processing approach, which has been argued to
favour development of the orthographic lexicon (Seymour & Evans, 1993)
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 31

and a more mature orthographic system (Grainger & Ziegler, 2011). Post-
intervention, both H.F. and R.C. scored within the range of the control
group on the identity matching task in terms of response speed and accuracy
at both exposure durations. Furthermore, the large differences in accuracy
between the longer and shorter exposure durations that were observed pre-
intervention for both boys had reduced at the post-intervention assessment
(from 9/80 to 5/80 for H.F., and from 8/80 to 3/80 for R.C.). Moreover,
both boys showed a significant improvement in oral reading of the 300
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

most frequent words of English, suggesting an increase in the operational effi-


ciency of their lexical reading mechanism. Unfortunately, a nonword reading
measure was not included in the pre- and post-assessments for this phase of
intervention, making a direct comparison between lexical and non-lexical
mechanisms impossible. It is important to note, however, that neither boy
showed a significant change in any of the phonological processing measures
included in the pre- and post-assessments, all of which would be expected to
correlate with nonword decoding ability.
With respect to alternative theoretical perspectives, it would seem that any
of the three models considered here could account for increased levels of oral
reading as a consequence of an improvement in the quality and efficiency of
early orthographic processing. What is not so clear is how the different
models (including the DRM) would account for the pattern of selective
improvement seen in R.C., whose reading of regular words, but not irregular
words, improved post-intervention. One possibility is that R.C.’s additional
difficulties with phonological output (reflected in poor RAN scores) may
have interfered with efficient development of the orthographic lexicon and
its links to the phonological output lexicon (in dual route terms). In this
case irregular words would be more affected than regular words, whose
reading could be supported by the non-lexical route.

PHASE 3: INTERVENTION II
The first phase of intervention was successful in leading to a significant
improvement in both participants’ oral reading of highly frequent real
words. Nevertheless, R.C. still scored below the level of his age-matched
peer group in oral reading of highly frequent words (Fry’s 300 word list)
and on a homophone lexical decision task incorporating irregular words
and visual foils. Furthermore, during oral reading assessments it was clear
that both boys often did not recognise letter combinations that commonly
occur in English words. Consequently, when they came across familiar
English letter patterns or chunks in oral reading (e.g., “ow” or “ough”) they
continued to decode them incorrectly, often letter-by-letter, rather than learn-
ing the phoneme sequences associated with these higher-order orthographic
32 LAW AND CUPPLES

units. Based on this clinical observation and the suggestion by Broom and
Doctor (1995a) that future interventions for developmental surface dyslexia
should target letter groupings rather than whole words in order to encourage
generalisation, this skill was identified as the focus for our second interven-
tion programme.
The plan for this intervention was to utilise the participants’ relative
strengths in decoding to assist with irregular word reading and potentially
to encourage further development of the orthographic lexicon. The logic
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

underlying our approach was as follows. Irregular words can contain letter
patterns that are essentially uncommon grapheme-to-phoneme correspon-
dences. Thus the letter combination -ear is most often pronounced as in the
words dear and near, making these words regular. By contrast, the word
pear is irregular because it contains a less common grapheme-to-phoneme
mapping for the same letter sequence. If multiple grapheme-to-phoneme cor-
respondences could be learnt, however, (i.e., that -ear can be read aloud as in
“dear” or “pear”) then some words could potentially become “less” irregular
(see Share, 1995, for further discussion on word irregularity), thus enabling
them to be “decoded”. Once decoded, these irregular words (and perhaps
other words containing the same letter sequences) could become more fam-
iliar, thus assisting development of the orthographic lexicon. In line with
this reasoning, H.F. and R.C. were taught a selection of letter patterns that
are frequent in English but that have multiple correspondences, in an
attempt to encourage them to learn letter patterns and their corresponding
phonemic representations.
From a theoretical perspective, this intervention approach appears to be
more in line with the CDP+ and connectionist models of reading than
with the DRM. Thus, the CDP+ model (which encompasses the DRM’s
dual-pathway architecture) and Harm and Seidenberg’s connectionist
model both map graphemes onto phonemes in a probabilistic and context-
dependent manner according to their statistical distribution in the lexicon.
By contrast, according to the DRM, the non-lexical reading mechanism
uses a rule-governed process for mapping graphemes onto phonemes,
which is considered incapable of reading irregular words (Jackson &
Coltheart, 2001), although it is able to apply multi-rules where a number
of letters represent a single phoneme (e.g., -igh). It is important to remember,
however, that the aim of this intervention was to provide participants with a
conscious strategy for recognising words containing particular letter-sound
patterns, not to alter the operation of the non-lexical reading mechanism
per se, which is, after all, not under conscious control. Nevertheless, the
different reading models do appear to make different predictions about inter-
vention outcomes.
According to the DRM, the taught multi-letter strategies would not be
encompassed within the non-lexical reading mechanism, but could encourage
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 33

development and/or reorganisation of the orthographic lexicon, thus resulting


in improved reading of real words but not nonwords. On the other hand, both
the CDP+ model and Harm and Seidenberg’s connectionist model would
appear to be entirely consistent with an improvement in performance on all
word types, including nonwords, because both models assign nonword and
real word pronunciations in accordance with the statistical distribution of
letter-sound associations in the lexicon. In summary, the aim of the interven-
tion in this phase was to teach H.F. and R.C. to recognise frequently occurring
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

English letter patterns that have multiple phonemic correspondences and


encourage them to use this knowledge when reading single words aloud.

Materials and procedure

Intervention
Common English letter patterns (phonograms) with two or more pronun-
ciations were selected from Fry, Polk, and Fountoukidis (1984); for
example, ow as in low versus how. Words whose pronunciations could not
be achieved through simple letter-by-letter decoding were also included in
the set and considered “irregular” for this task even though they could be con-
sidered to follow the spelling rules of English (e.g., eight).
A list of 259 words containing selected letter patterns was shown to H.F.
and R.C. one by one on a computer screen. They were asked to say each
one aloud. Their responses were used to create an individualised probe
word list for each of them. Any letter pattern that was read aloud correctly
more than 80% of the time was discarded. Any letter patterns that were not
read aloud correctly 80% of the time were used to construct the list.
Because each boy made different errors on the initial list, their probe lists dif-
fered too. H.F.’s probe word list comprised 80 words containing six different
letter patterns, whereas R.C.’s probe word list comprised 129 words contain-
ing 13 different letter patterns. All of the words on the probe word lists were
randomised and printed onto flashcards (Ariel 36 point). The full probe list
was administered to each boy on two occasions, three weeks apart, in order
to obtain a baseline. The first baseline score was used as the pre-intervention
score for analysis purposes. H.F. made 27 errors at baseline 1 and 22 errors at
baseline 2. This difference between the two testing sessions was not signifi-
cant, McNemar x 2(1, N ¼ 80), p . .10. R.C. made 58 errors at baseline 1
and 60 errors at baseline 2, also a non-significant difference between the
two testing sessions, McNemar x 2(1, N ¼ 129), p . .10. The letter patterns
selected for training H.F. were: -ough (including -ought), -eigh (including
-eight), -ow (including -own), -otch, -ew, -our. For R.C. the selected letter pat-
terns were: ough (including -ought), -eigh (including -eight), -ow (including
-own), -otch, -ew, -our, -ost, -ould, -aught, -itch, -ind, -ight, -are. The words
34 LAW AND CUPPLES

on each probe word list were divided into two sets. One set contained words
that would be trained and the other set were untrained words that were shown
to the participants only when reading aloud their respective probe word lists.
The trained and untrained words used with each participant were compared on
two measures of phonotactic probability (reflecting the frequency of occur-
rence of their individual phonological segments or biphones) and a measure
of neighbourhood density (the number of words that differ from a target
word by a single sound) using the online calculator of Storkel and Hoover
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

(2010). Trained words did not differ significantly from untrained words in
the average frequency of their positional segments or biphones. The mean fre-
quency of occurrence of individual phonological segments in trained and
untrained words was .0469 versus .0435, respectively, for H.F., t(78) ¼
1.011, p ¼ .315; and .0494 versus .0483 for R.C., t , 1; whereas the mean
biphone frequency for trained and untrained words was .0034 versus .0035
for H.F., t , 1; and .0043 versus .0045 for R.C., t , 1. On the other hand,
trained words for H.F. came from more densely populated lexical neighbour-
hoods than untrained words, with an average of 12.1 versus 8.1 neighbours,
respectively, t(78) ¼ 2.40, p ¼ .019. The corresponding difference was
smaller and non-significant for R.C., with an average of 10.67 versus 8.67
neighbours, t (127) ¼ 1.62, p ¼ .107. Each letter pattern and its associated
training words were printed on flashcards individually. The training cards
used in the intervention sessions had the letter pattern underlined (e.g.,
sleigh); however, the training words in the probe list did not have the letter
patterns underlined (so they looked the same as the untrained words on the
list). The inclusion of untrained words was important in order to examine
whether any generalisation would take place from the training words contain-
ing a particular letter pattern to untrained words containing the same letter
pattern. The probe words were re-administered twice post-intervention,
four weeks apart.
Training took place in a quiet room at each participant’s home. Each train-
ing session lasted approximately 45 minutes. H.F. was trained on the words
twice a week for four weeks, whereas R.C. was trained twice a week for
six weeks. At the beginning of each session, the probe word list was read
aloud by the participant. No feedback or assistance was provided at these
times. Next, the letter pattern taught in the previous session and its words
(from the training set only) were revised. Each letter pattern was revised
only once during treatment.
The letter pattern to be trained during a session was introduced in isolation
(e.g., -ough). The different pronunciations for the letter pattern were dis-
cussed and the participant was asked to remember some words that contained
the letter pattern (e.g., crow and cow for the letter pattern -ow). He was asked
to visualise a scene where the two things were pictured and to try to remember
this scene for that letter pattern. He was then shown the training words
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 35

associated with the letter pattern and was encouraged to say the letter pattern
sounds aloud until he came across a word that he knew. The participant was
praised when he read the word aloud correctly.

Pre- and post-intervention assessment

Oral reading. The participants’ ability to read single words aloud was
assessed pre- and post-intervention using the Coltheart and Leahy (1996)
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

word list, the WRMT-R word attack and word identification subtests, and
the 400 most frequent English words (Sakiey & Fry, 1979). H.F. and R.C.
were shown the Coltheart and Leahy words on flashcards (Ariel 36 point)
in a pseudo-random order and the Sakiey and Fry 400 most frequent
English words were shown individually on a computer screen (Ariel 48
point). Participants were asked to read each word aloud, with no time limit
imposed. The WRMT-R word attack and word identification subtests were
administered according to instructions in the test manual.

Silent reading. Silent reading was assessed pre- and post-intervention


using lexical decision and homophone selection. The homophone selection
task was devised using 39 items from the intervention set. These items
were divided into two sets, 20 items were paired with other real words to
form a set of real word homophone pairs (e.g., slay—sleigh) and 19 items
were matched with nonword homophones (e.g., height—hite). Twenty-two
of the words were trained in the intervention and 17 were generalisation
words. One homophone pair contained two trained words. The word pairs
were printed on flashcards, Ariel font 36. A description for one of the
words was spoken aloud to each participant and he was asked to point to
the word that went with the description (e.g., for the homophone pair “pore
– pour”, the target description was “To tip liquid from one container to
another”).
For lexical decision, the Manis et al. (1996) orthographic choice task com-
prising 52 word–nonword pairs was re-administered (from the initial
assessment).
Written word to picture matching: Words on the probe word list whose
meanings could be represented by a picture using the Boardmakerw software
programme were printed individually on the right hand side of an A4 page.
Three pictures were printed on the left hand side of the same page. The pic-
tures consisted of the target and two distracters. All pictures used were repre-
sentative of probe words. Each picture was used three times in the assessment,
once as target and twice as distracter. The list contained 117 words for R.C.
(12 words could not be represented by pictures) and 73 for H.F. (7 words
could not be represented by pictures). Each participant was shown the
36 LAW AND CUPPLES

words one at a time and asked to say each one aloud. He was then asked to
select the picture that went with the word.

Real word and nonword repetition. Real and nonword repetition was
included as a non-reading control for this second phase of intervention. It
consisted of 200 items: 100 real words and 100 nonwords matched for
number of syllables. The real words were divided into high frequency
words (above 200 per million occurrences according to Kucera & Francis,
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

1967) and low frequency words (below 20 per million occurrences). Words
and nonwords contained between one and five syllables, with 20 real
words (10 high frequency and 10 low frequency) and 20 nonwords for
each syllable length.
Each item was spoken aloud by the examiner and the participant was asked
to repeat it immediately. Items were presented in the same pseudo-random
order to each participant ensuring even distribution of the various item
types. This test was also administered to the respective control groups.

RESULTS

H.F.
H.F.’s results are shown in Tables 2 and 6. He showed significant improve-
ment in his oral reading of real words on the word identification subtest of
the WRMT-R, see Table 2, McNemar x 2(1, N ¼ 106) ¼ 17.05, p , .001;
Sakiey and Fry’s list of 400 most frequent English words, McNemar x 2(1,
N ¼ 400) ¼ 9.48, p ¼ .002; and the oral reading probe, McNemar x 2(1, N
¼ 80) ¼ 25.04, p , .001, see Table 6. By contrast, there was no significant
change in his oral reading accuracy for regular, irregular, or nonwords on
the Coltheart and Leahy (1996) word list, and no significant improvement
in his oral reading of nonwords on the WRMT-R word attack sub-test (see
Table 2). In regard to silent reading, H.F. was significantly more accurate
in selecting printed homophones to match spoken descriptions after interven-
tion, McNemar x 2(1, N ¼ 39) ¼ 9.60, p ¼ .002, but there was no significant
improvement in word–picture matching for probe words, lexical decision on
the Manis et al. (1996) orthographic choice test, or the ability to accurately
repeat words and nonwords (see Table 6).

Trained versus untrained probe word scores


Generalisation of learning to untaught words was investigated by compar-
ing H.F.’s oral and silent reading performance on trained and untrained words
pre- versus post-intervention. He showed significantly improved performance
from pre-intervention to post-intervention in his oral reading of both training
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA


TABLE 6
Intervention II results for H.F., R.C., and their respective control groups on non-standardised measures

H.F. R.C.

H.F.’s control R.C.’s control


Score group (N ¼ 15) Score group (N ¼ 17)

Assessment Pre- Post- Diff. Mean (SD) Range Pre- Post- Diff. Mean (SD) Range

Frequent words reading (400) 378 395 +17∗∗ 398.73 (2.15) 398 –400 378 368 210 399.5 (0.81) 398–400
Orthographic choice (52) 35 33 22 – – 38 36 22 – –
Probes
Oral reading 53/80 80/80 +27∗∗∗ 76.07 (1.43) 77–80 71/129 117/129 +46∗∗∗ 125.9 (2.15) 121–128
Word– picture matching 67/73 72/73 +5 70.9 (2.15) 65–73 97/117 117/117 +20∗∗∗ 115.5 (1.32) 114–117
Homophone (39) 23/39 36/39 +13∗∗ – – 23/39 33/39 +10∗ – –
Repetition (200) 197 197 0 197.2 (2.24) 196 –200 197 197 0 198 (1.9) 194–200

Item analyses of correct responses were conducted using McNemar x 2. Dashes indicate that data were not collected for that cell.

p , .05; ∗∗ p , .01; ∗∗∗ p , .001.

37
38 LAW AND CUPPLES

words, from 28/44 pre-intervention to 44/44 post-intervention; McNemar


x 2(1, N ¼ 44) ¼ 14.06, p , .001, and untrained generalisation words, from
25/36 pre-intervention to 36/36 post-intervention; McNemar x 2(1, N ¼ 36)
¼ 9.09, p ¼ .003. A similar percentage increase and perfect performance
was observed on both word types post-intervention, despite the significant
difference in neighbourhood density, which favoured trained words. Results
for the probe homophone task reveal a different pattern, however, in that
H.F. made significantly fewer errors on trained probe items after intervention
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

than before, from 12/22 pre-intervention to 21/22 post-intervention,


McNemar x 2(1, N ¼ 22) ¼ 7.11, p ¼ .008; whereas his performance on
untrained items showed no such improvement, from 11/17 pre-intervention
to 15/17 post-intervention, McNemar x 2(1, N ¼ 17) ¼ 1.50, p ¼ .221.
On the word–picture matching task, the improvement in performance
from pre- to post-intervention was not statistically significant for trained or
untrained words. This lack of a significant change was presumably due to a
ceiling effect as H.F. made only six errors on this test at the pre-intervention
assessment.

R.C.
R.C’s results are shown in Tables 2 and 6. Like H.F., he showed significant
improvement in his oral reading of real words on the word identification
subtest of the WRMT-R from pre- to post-intervention (see Table 2),
McNemar x 2(1, N ¼ 106) ¼ 4.08, p ¼ .043; and a significant improvement
in reading the probe word list (see Table 6), McNemar x 2(1, N ¼ 129) ¼
44.02, p , .001. He also showed no improvement in his oral reading of non-
words on either the WRMT-R word attack subtest or the Coltheart and Leahy
(1996) word list (see Table 2). However, R.C. also differed from H.F. in
showing a significant improvement on the Coltheart and Leahy irregular
words, McNemar x 2(1, N ¼ 30) ¼ 7.11, p ¼ .008; but no improvement in
his reading of Sakiey and Fry’s list of 400 most frequent English words.
Despite these observed improvements in R.C.’s oral reading of real words,
his performance remained below the 10th percentile on two of the five stan-
dardised measures (see Table 2). In regard to silent reading, R.C. showed evi-
dence of improved performance from pre- to post-intervention in selecting
printed homophones to match spoken descriptions, McNemar x 2(1, N ¼
39) ¼ 4.50, p ¼ .034; and in word–picture matching for probe words,
McNemar x 2(1, N ¼ 117) ¼ 18.05, p , .001. As was the case for H.F., he
showed no significant improvement in lexical decision on the Manis et al.
(1996) orthographic choice test, or the ability to accurately repeat
words and nonwords (which were close to ceiling prior to intervention)
(see Table 6).
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 39

Trained versus untrained probe word scores


Generalisation of learning to untaught words was investigated by compar-
ing R.C.’s performance on trained and untrained words pre- versus post-inter-
vention. Like H.F., he showed significantly improved performance in his oral
reading of training words, from 43/72 pre-intervention to 65/72 post-interven-
tion, McNemar x 2(1, N ¼ 72) ¼ 20.05 p , .001; and untrained generalis-
ation words that contained the same letter patterns, which improved from
28/57 pre-intervention to 52/57 post-intervention, McNemar x 2(1, N ¼ 57)
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

¼ 22.04, p , .001. In addition, his performance on the probe homophone


selection task improved significantly for trained words, from 12/22 pre-inter-
vention to 21/22 post-intervention, McNemar x 2(1, N ¼ 22) ¼ 7.11, p ¼
.008; but not untrained words, from 11/17 pre-intervention to 12/17 post-
intervention, McNemar x 2(1, N ¼ 17) ¼ 0.00, p ¼ 1.0. Finally, R.C. differed
from H.F. in showing significantly improved word–picture matching
performance for both trained items and untrained items from pre- to
post-intervention, with 50/62 versus 62/62 for trained items, McNemar
x 2(1, N ¼ 62) ¼ 10.08 p ¼ .001; and 47/55 vs. 55/55 for untrained items,
McNemar x 2(1, N ¼ 55) ¼ 6.13 p ¼ .013.

DISCUSSION
The aim of this second phase of intervention was to train pronunciation of
irregular English letter patterns and encourage the participants to recognise
these patterns as an alternative strategy to either whole word recognition or
letter-by-letter decoding. We predicted that this intervention approach
would be effective in increasing the boys’ familiarity with the taught letter
patterns and words containing them, thereby enabling them to establish,
restructure and/or reorganise entries within the orthographic lexicon. In line
with this prediction, both participants improved in their oral reading of
trained and untrained probe words and in their homophone selection of
trained, but not untrained, probe words after intervention. In addition, both
boys showed improved oral reading on the standardised WRMT-R word
identification assessment after intervention, confirming that generalisation
to a totally different word set had occurred. This overall pattern of results pro-
vides further evidence in favour of our second research hypothesis, in
showing that another targeted treatment programme that was devised
within the framework of the DRM resulted in improved real word reading
for both participants, consistent with enhanced development of the lexical
reading mechanism.
Once again, despite the overall similarity between H.F. and R.C. in their
response to intervention, the individual patterns of results were somewhat
different for each boy. R.C. showed improved comprehension of trained
40 LAW AND CUPPLES

and untrained written probe words on the word–picture matching task,


whereas H.F. did not (presumably due to a ceiling effect). Similarly,
whereas R.C. improved significantly in his oral reading of irregular words
on the Coltheart and Leahy (1996) word list, H.F. showed small and non-sig-
nificant improvements on all word types. By contrast, H.F.’s ability to read
aloud words from the 400 most frequent words list improved significantly,
but R.C.’s performance on this task did not reveal any gain in skill. This
finding is surprising in the context of R.C.’s overall pattern of reading
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

improvement. It is possible, however, that the 400-word reading test (the


longest single reading assessment used here) simply placed too great a
strain on R.C.’s ability to maintain attention to the task, given that reading
was challenging for him. Further evidence that extended reading tasks
caused difficulty for R.C. comes from the findings obtained on the 300-
word reading test used for Intervention I. Although R.C. showed significant
improvement on that task from pre- to post-intervention, his post-intervention
score remained well below the level of the control group range (see Table 5).
Finally, neither boy showed changes on the Manis et al. (1996) orthographic
choice task, or the controlled frequency repetition task. The specificity of the
intervention results, along with the non-significant effect of intervention on
nonword decoding (WRMT-R word attack and Coltheart and Leahy non-
words), confirms that the boys were not experiencing an overall, generalised
improvement in their reading performance over the course of the second
phase of intervention, but rather were responding to the highly individualised
and selectively targeted intervention programme.
In sum, the general pattern of results reported here indicates that training
poor readers to select consciously between alternative phonemic mappings
for particular graphemic units until a meaningful word is identified can
help to compensate for a poorly functioning lexical reading mechanism.
For the participants described here, this intervention approach enabled
better reading of trained words and generalisation to untrained words,
which might have occurred by analogy to the training words (see Glushko,
1979). To our knowledge, this intervention approach has not been attempted
in previous research, thus making this demonstration the first of its kind.
Although we have interpreted the results of this second intervention phase
in terms of the DRM, it might be argued that the observed generalisation to
untrained words suggests involvement of the non-lexical reading mechanism.
On this alternative view, the improvements documented here could be due to
compensation by the non-lexical reading mechanism for a weakness in the
lexical mechanism, rather than to direct remediation of the lexical mechan-
ism. As noted earlier, this interpretation would be more consistent with
models such as CDP+ or the connectionist model of Harm and Seidenberg
(1999), in which phonology is assigned probabilistically, rather than with
the DRM. It would seem to be a less plausible interpretation in the present
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 41

context, however, given that neither participant improved significantly in the


ability to read nonwords aloud. Nevertheless, further research is required
before firm conclusions can be drawn.

GENERAL DISCUSSION
In this research, we hypothesised that the DRM would provide a useful frame-
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

work for (1) identifying the specific oral reading difficulties experienced by
two young boys and (2) designing a theoretically motivated, targeted treat-
ment programme. These hypotheses were confirmed. Both participants
demonstrated a particular difficulty in irregular word reading compared to
nonword reading; a pattern recognised as developmental surface dyslexia.
More specifically, both participants had difficulty with the input stage of
the lexical reading mechanism, in particular, at the orthographic lexicon.
Further examination of the boys’ associated language and cognitive abilities
revealed deficits in RAN and visual-orthographic processing, which might
have prevented typical development of the DRM’s lexical reading mechan-
ism, and led both boys to compensate for their difficulties by adopting a
simple (mainly single letter to single sound) decoding approach for all
word types. Based on this detailed assessment, two interventions were
devised to target different aspects of orthographic processing. The first inter-
vention targeted visual-orthographic processing by increasing the efficiency
with which pronounceable letter strings (nonwords) were identified;
whereas the second intervention provided training in the identification and
decoding of common letter patterns in irregular words.
Overall, the interventions resulted in improved performance in the oral
reading of real words but not nonwords for both boys, thus confirming our
second research hypothesis that post-intervention improvements in reading
would be evident for both boys and consistent with expectations based on
the DRM. More remarkably, the observed improvements were seen on stan-
dardised tests, an effect that has rarely been reported in intervention studies
for surface dyslexia. One of the participants, H.F., even obtained oral
reading scores for single words that were in an age-appropriate range at the
end of the study. Despite the general similarity in how the two participants
responded to the intervention, there were also differences between them.
H.F. showed significantly improved homophone recognition following Inter-
vention I, an indication that the intervention had improved his ability to use
the lexical mechanism to access semantics. By contrast, R.C. showed
improvement in this regard only after Intervention II. These differing
results confirm that although both participants presented with impairment at
the orthographic lexicon, the intervention outcomes were slightly different
in each case. This individual variation within the same reading disability
42 LAW AND CUPPLES

subtype reinforces the view that children with surface dyslexia do not form a
homogeneous group and emphasises the need for specific assessment and
individualised treatment for each child presenting with developmental
surface dyslexia (or any other type of reading disability).
The positive outcomes obtained in this study confirm and extend results
reported previously by Broom and Doctor (1995a) and Brunsdon et al.
(2002). Whereas both of these earlier studies reported better reading of
words learned during intervention, only Brunsdon et al. found evidence of
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

limited generalisation to untaught words following intervention. By contrast,


our results show clear evidence of generalisation to untaught words following
both intervention phases, and most notably to a standardised assessment of
single word oral reading (the WRMT-R). We attribute these stronger gener-
alisation effects at least partly to our intervention focus on sub-lexical graphe-
mic units rather than whole words, as suggested by Broom and Doctor.
Another distinguishing feature of this research that may have contributed to
the difference between studies, however, was the focus on improving the effi-
ciency of lexical processing strategies rather than word-specific knowledge,
especially in regard to letter identification, which was targeted in Intervention
I. Further research is required before any firm conclusions can be drawn
regarding the relative effectiveness of these (and other) intervention
approaches.
Results from single case studies such as H.F. and R.C. are invaluable in
establishing the types of interventions that can be effective for children
with particular areas of weakness in their single word oral reading skills.
An additional but related question, however, concerns the role of unimpaired
processes in determining intervention success for a given child. In the current
study, both boys responded positively to Intervention II, which built on their
strengths in decoding to encourage the establishment of entries within the
orthographic lexicon. Whether similar improvements would be seen in chil-
dren with inferior decoding skills (i.e., a mixed form of developmental dys-
lexia) is an open question that can only be answered by future research. In
attaining a clearer picture of how any child’s overall profile of reading and
related strengths and weaknesses can influence the effectiveness of different
intervention strategies, our collective research findings will make a major
contribution to future clinical practice.
Although the primary aim of this research was to use the DRM to devise an
effective intervention programme for two children with developmental
surface dyslexia, a supplementary focus was to explore the extent to which
post-intervention results were consistent with expectations based on other
current models of reading; in particular, the CDP+ and Harm and Seiden-
berg’s (1999) connectionist model. Although all three reading models
could account for aspects of the intervention effects, we have suggested
that the DRM is most consistent with the selective improvements in real
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 43

word reading (as compared to nonword reading) observed following Interven-


tion II. It would be premature, however, to discount any of the theoretical
models at this stage. Instead, clinical researchers should make use of the
range of viable theoretical perspectives in their efforts to ensure accurate
assessment of and effective intervention for children’s reading difficulties.

Limitations
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

Despite the overall positive outcomes achieved in this research, two difficul-
ties were identified in using the DRM for intervention purposes. First, as
noted earlier, the model does not explicitly incorporate ideas for intervention
so a thorough knowledge of the DRM and some additional level of clinical
expertise are required in order to devise appropriate interventions. As more
studies are conducted using this approach, the results of devised intervention
tasks will hopefully give some insight into the working of the particular sub-
components of the model that are targeted for intervention. This knowledge
will result in more established therapy techniques for each component.
Second, due to the relative originality of this research in using the DRM
for assessment and intervention of developmental difficulties it was discov-
ered that there is little in the way of standardised data directly related to
the DRM, in order to describe which components should have been acquired
(and at what level) by a typically developing child at various ages. Although
these problems were encountered while conducting this research, they were
by no means insurmountable. Importantly, the approach is in the relatively
early developmental stages, and further research should begin to fill in
some of the missing pieces.
Two methodological notes are also in order. First, because our focus was
on single word reading as represented in the DRM, a comprehensive formal
assessment of wider language and articulation skills was not conducted as part
of this study. Furthermore, an assessment of expressive vocabulary was admi-
nistered to only one of the two participants to avoid re-assessment of skills
measured previously in the other boy. Although these strategies had the
benefit of reducing time spent in test administration for this study, they
also resulted in a less complete data set than would otherwise have been
the case. Second, although this research incorporated age-matched control
groups of boys with no reading difficulties for comparison with the exper-
imental participants, it did not include a group of untreated control partici-
pants matched for type of reading disability. Accordingly, it is possible that
the improvements observed during intervention phases of the current study
resulted not from the intervention per se, but from natural maturation and/
or activities taking place outside the research clinic. In the current study,
two precautions were taken to minimise the likelihood of such confounding.
First, each phase of intervention was limited to just 6 weeks in duration to
44 LAW AND CUPPLES

reduce the potential impact of natural maturation or external influences.


Second, as is generally the case in single case intervention research of the
type reported here, participants served as their own controls through the
inclusion of assessment tasks that were not expected to show an effect of
intervention. Since the pattern of results confirmed this expectation, it is
likely that improvements in treated skills resulted from the intervention
itself rather than reflecting influences external to the study or a generalised
positive influence of participation in the research. However, a stronger
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

form of control would be provided through the added inclusion of an


untreated control group in future research.
In conclusion, the DRM was used in this research to identify areas of weak-
ness in the reading skills of two boys, H.F. and R.C., and to design an effec-
tive targeted, theoretically motivated, and individualised intervention
programme. It is hoped that the present findings will stimulate further
research of this nature, which will ultimately lead to the identification of
more ways to help struggling readers, and in doing so add valuable infor-
mation to the debate about how children learn to read.

ORCID
Caroline Law http://orcid.org/0000-0002-0795-6502
Linda Cupples http://orcid.org/0000-0003-3659-1642

REFERENCES
Boardmaker (Version 5) [Computer software]. Solana Beach, CA: Mayer – Johnson.
Broom, Y., & Doctor, E. (1995a). Developmental surface dyslexia: A case study of the efficacy
of a remediation programme. Cognitive Neuropsychology, 12, 69 – 110. doi:10.1080/
02643299508251992
Broom, Y., & Doctor, E. (1995b). Developmental phonological dyslexia: A case study of the
efficacy of a remediation programme. Cognitive Neuropsychology, 12, 725 – 766. doi:10.
1080/02643299508251400
Brunsdon, R., Coltheart, M., & Nickels, L. (2005). Treatment of irregular word spelling in
developmental surface dysgraphia. Cognitive Neuropsychology, 22(2), 213 – 251. doi:10.
1080/02643290442000077
Brunsdon, R., Coltheart, M., & Nickels, L. (2006). Severe developmental letter-processing
impairment: A treatment case study. Cognitive Neuropsychology, 23(6), 795 – 821. doi:10.
1080/02643290500310863
Brunsdon, R., Hannan, T., Coltheart, M., & Nickels, L (2002). Successful treatment of
sub-lexical reading deficits in a child with dyslexia of the mixed type. Neuropsychological
rehabilitation, 12(3), 199 – 229. doi:10.1080/09602010244000048
Brunsdon, R., Hannan, T. Nickels, L., & Coltheart, M. (2002). Treatment of lexical processing
in mixed dyslexia: A case study. Neuropsychological Rehabilitation, 12(5), 385 – 418.
doi:10.1080/09602010244000174
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 45

Castles, A., Bates, T., & Coltheart, M. (2006). John Marshall and the developmental dyslexias.
Aphasiology, 20(9/10/11), 871– 892. doi:10.1080/02687030600738952
Castles, A., Bates, T., Coltheart, M., Luciano, M., & Martin, N. (2006). Cognitive modelling
and the behaviour genetics of reading. Journal of Research in Reading, 29(1), 92 – 103.
Castles, A., & Coltheart, M. (1993). Varieties of developmental dyslexia. Cognition, 47,
149 – 180. doi:10.1016/0010-0277(93)90003-E
Castles, A., & Coltheart, M. (1996). Cognitive correlates of developmental surface dyslexia: A
single case study. Cognitive Neuropsychology, 13(1), 25 – 50. doi:10.1080/
026432996382051
Coltheart, M., Curtis, B., Atkins, P., & Haller, M. (1993). Models of reading aloud: Dual-route
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

and parallel-distributing-processing approaches. Psychological Review, 100, 589 – 608.


doi:10.1037/0033-295X.100.4.589
Coltheart, M., & Leahy, J. (1996). Assessment of lexical and non-lexical reading abilities in
children – Some normative data. Australian Journal of Psychology, 35, 469 – 496. doi:10.
1080/00049539608259520
Coltheart, M., Masterson, J., Byng, S., Prior, M., & Riddoch, J (1983). Surface dyslexia. The
Quarterly Journal of Experimental Psychology, 35(3), 469 – 495. doi:10.1080/
14640748308402483
Coltheart, M., Rastle, K., Perry, C., Langdon, R., & Ziegler, J. (2001). DRC: A dual route
cascaded model of visual word recognition and reading aloud. Psychological Review,
108, 204 – 256. doi:10.1037/0033-295X.108.1.204
Dunn, L., & Dunn, L. (1997). The Peabody picture vocabulary test. 3rd ed. (PPVT-III). Circle
Pines, MN: AGS Publishing.
Edwards, V., & Hogben, J. (1999). New norms for comparing children’s lexical and non-lexical
reading: A further look at sub-typing dyslexia. Australian Journal of Psychology, 51, 37 –
49. doi 10.1080/00049539908255333
Forster, K. I., & Forster, J. C. (2003). DMDX: A Windows display program with millisecond
accuracy. Behaviour Research Methods, Instruments & Computers, 35, 116 – 124. doi:10.
3758/BF03195503
Frederickson, N., Frith, N., & Reason, R. (1997). Phonological awareness battery. Windsor,
UK: Nelson Publishing Co Ltd.
Fry, E. (1980). The new instant word list. Reading Teacher, 34(3), 284 – 289.
Fry, E., Polk, J., & Fountoukidis, D. (1984). The reading teacher’s book of lists. Englewood
Cliffs, NJ: Prentice Hall.
German, D. (1986/1989). Test of word-finding. Dallas, TX: PRO-ED.
Glushko, R. J. (1979). The organisation and activation of orthographic knowledge in reading
aloud. Journal of Experimental Psychology: Human Perception and Performance, 5(4),
674 – 691.
Grainger, J., & Ziegler, J. (2011). A dual route approach to orthographic processing. Frontiers
in Psychology, 2, 1 – 13. doi:10.3389/fpsyg.2011.00054
Harm, M. W., McCandliss, B. D., & Seidenberg, M. S. (2003). Modelling the successes and
failures of interventions for disabled readers. Scientific Studies of Reading, 7, 155 – 182.
doi:10.1207/S1532799XSSR0702_3
Harm, M. W., & Seidenberg, M. S. (1999). Phonology, reading acquisition, and dyslexia:
Insights from connectionist models. Psychological Review, 106, 491– 528. doi:10.1037/
0033-295X.106.3.491
Jackson, N. E., & Coltheart, M. (2001). Routes to reading success and failure: Towards an inte-
grated cognitive psychology of atypical reading. N.Y.: Taylor and Francis.
Judica, A., De Luca, M., Spinelli, D., & Zoccolotti, P. (2002). Training of developmental surface
dyslexia improves reading performance and shortens eye fixation duration in reading. Neurop-
sychological Rehabilitation, 12(3), 177–197. doi:10.1080/09602010244000002
46 LAW AND CUPPLES

Kay, J., Lesser, R., & Coltheart, M. (1992). Psycholinguistics assessments of language proces-
sing in aphasia (PALPA). Hove, U.K.: Lawrence Erlbaum Associates Ltd.
Kucera, H., & Francis, W. N. (1967). Computational analysis of present-day American English.
Providence, RI: Brown University Press.
Manis, F. R., Seidenberg, M. S., & Doi, L. M. (1999). See Dick RAN: Rapid naming and
the longitudinal prediction of reading sub skills in first and second graders. Scientific
Studies of Reading, 3, 129 – 157. Retrieved from http://dx.doi.org/10.1207/
s1532799xssr0302_3
Manis, F. R., Seidenberg, M. S., Doi, L. M., McBride-Chang, C., & Peterson, A. (1996). On the
basis of two subtypes of developmental dyslexia. Cognition, 58, 157 – 195. doi:10.1016/
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

0010-0277(95)00679-6
McBride-Chang, C. (1995). What is phonological awareness? Journal of Educational Psychol-
ogy, 87(2), 179 – 192. Retrieved from http://dx.doi.org/10.1037/0022-0663.87.2.179
Morton, J., & Patterson, K. E. (1987). A new attempt at an interpretation, or, an attempt at a new
interpretation. In M. Coltheart, K. E. Patterson, & J. C. Marshall (Eds.), Deep dyslexia
(pp. 91 – 118). London: Routledge & Kegan Paul.
Perin, D. (1983). Phonemic segmentation and spelling. British Journal of Psychology, 74(1),
129 – 144. Retrieved from http://dx.doi.org/10.1111/j.2044-8295.1983.tb01849.x
Perry, C., Ziegler, J. C., Braun, M., & Zorzi, M. (2010). Rules versus statistics in reading aloud:
new evidence on an old debate. European Journal of Cognitive Psychology, 22, 798– 812.
doi:10.1080/09541440902978365
Perry, C., Ziegler, J. C., & Zorzi, M. (2007). Nested incremental modeling in the development
of computational theories: The CDP+ model of reading aloud. Psychological Review, 114,
273 – 315. doi:10.1037/0033-295X.114.2.273
Peterson, R. L., Pennington, B. F., & Olson, R. K. (2013). Subtypes of developmental dyslexia:
Testing the predictions of the dual-route and connectionist frameworks. Cognition, 126,
20 – 38. doi:10.1016/j.cognition.2012.08.007
Rastle, K., & Coltheart, M. (1999). Serial and strategic effects in reading aloud. Journal of
Experimental Psychology: Human Perception and Performance, 25, 482 – 503. doi:10.
1037/0096-1523.25.2.482
Sakiey, E., & Fry, E. (1979). 3000 instant words. Highland Park, NJ: DREIER Education
Systems.
Seidenberg, M. S., & McClelland, J. L. (1989). A distributed, developmental model or word
recognition and naming. Psychological Review, 96, 523 – 568. Retrieved from http://dx.
doi.org/10.1037/0033-295X.96.4.523
Seymour, P. H. K., & Evans, H. (1993). The visual orthographic processor and developmental
dyslexia. In D. Willows & R. Kruk (Eds.). Visual processing in reading and reading disabil-
ities (pp. 347 – 376). Hillsdale, England: Lawrence Erlbaum.
Share, D. L. (1995). Phonological recoding and self-teaching: The sine qua non of reading rec-
ognition. Cognition, 55, 151 – 218. Retrieved from http://dx.doi.org/10.1016/0010-
0277(94)00645-2
Snowling, M. J., Bryant, P. E., & Hulme, C. (1996). Theoretical and methodological pitfalls in
making comparisons between developmental and acquired dyslexia: Some comments on
A. Castles and M. Coltheart 1993. Reading and Writing, 8, 443– 451. doi:10.1007/
BF00404005
Storkel, H. L., & Hoover, J. R. (2010). An on-line calculator to compute phonotactic probability
and neighborhood density based on child corpora of spoken American English. Behavior
Research Methods, 42, 497– 506. doi:10.3758/BRM.42.2.497
Wechsler, D. (1989). Preschool and primary scale of intelligence-revised. San Antonio, TX:
Psychological Corporation.
ASSESSING AND TREATING DEVELOPMENTAL SURFACE DYSLEXIA 47

Welbourne, S., & Lambon Ralph, M. (2005). Using computational, parallel distributed
processing networks to model rehabilitation in patients with acquired dyslexia: An initial
investigation. Aphasiology, 19(9), 789– 806. doi:10.1080/02687030500268811
Wolf, M., & Bowers, P. G. (1999). The double deficit hypothesis for the developmental dyslex-
ias. Journal of Educational Psychology, 91, 415– 438. doi:10.1037/0022-0663.91.3.415
Woodcock, R. (1987). The woodcock reading mastery test – Revised. Circle Pines, MN: AGS.
Zorzi, M. (2010). The connectionist dual process (CDP) approach to modelling reading aloud.
European Journal of Cognitive Psychology, 22, 836–860. doi:10.1080/09541440903435621
Zorzi, M., Houghton, G., & Butterworth, B. (1998). The development of spelling-sound
relationships in a model of phonological reading. Language and Cognitive Processes,
Downloaded by [Texas A & M International University] at 02:56 24 August 2015

13(2 – 3), 337 – 371. doi:10.1080/016909698386555

Вам также может понравиться