Sight Word and Phonics Training in Children With Dyslexia

Abstract

The aims of this study were to (a) compare sight word training and phonics training in children with dyslexia, and (b) determine if different orders of sight word and phonics training have different effects on the reading skills of children with dyslexia. One group of children (n = 36) did 8 weeks of phonics training (reading via grapheme-phoneme correspondence rules) and then 8 weeks of sight word training (reading irregular words as a whole), one group did the reverse (n = 36), and one group did phonics and sight word training simultaneously for two 8-week periods (n = 32). We measured the effects of phonics and sight word training on sight word reading (trained irregular word reading accuracy, untrained irregular word reading accuracy), phonics reading (nonword reading accuracy, nonword reading fluency), and general reading (word reading fluency, reading comprehension). Sight word training led to significant gains in sight word reading measures that were larger than gains made from phonics training, phonics training led to statistically significant gains in a phonics reading measure that were larger than gains made from sight word training, and both types of training led to significant gains in general reading that were similar in size. Training phonics before sight words had a slight advantage over the reverse order. We discuss the clinical implications of these findings for improving the treatment and assessment of children with dyslexia.

Full text

http://ldx.sagepub.com/
Journal of Learning Disabilities
http://ldx.sagepub.com/content/early/2013/10/01/0022219413504996
The online version of this article can be found at:
DOI: 10.1177/0022219413504996
published online 1 October 2013J Learn Disabil
Genevieve McArthur, Anne Castles, Saskia Kohnen, Linda Larsen, Kristy Jones, Thushara Anandakumar and Erin Banales
Sight Word and Phonics Training in Children With Dyslexia
Published by:
Hammill Institute on Disabilities
and
http://www.sagepublications.com
can be found at:Journal of Learning DisabilitiesAdditional services and information for
http://ldx.sagepub.com/cgi/alertsEmail Alerts:
http://ldx.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
What is This?
- Oct 1, 2013OnlineFirst Version of Record >>
at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from at Macquarie University Library on October 8, 2013ldx.sagepub.comDownloaded from

Journal of Learning Disabilities
XX(X) 1 –17
© Hammill Institute on Disabilities 2013
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0022219413504996
journaloflearningdisabilities.sagepub.com
Article
Around 5% of children find it unusually difficult to learn to
read even though they have had normal reading instruction,
they have normal intelligence, and they have no known neu-
rological or psychological problems. This condition—which
is often called developmental dyslexia (Hulme & Snowling,
2009)—not only affects children’s academic achievements
but also increases their risk for anxiety, depression, conduct
disorder, and hyperactivity (Carroll, Maughan, Goodman, &
Meltzer, 2005). Thus, we need to discover how to treat poor
reading as effectively as possible.
To date, most treatment trials done with children with
dyslexia have looked at the effects of “phonics” reading
programs. These programs teach children to learn to read
using the grapheme–phoneme correspondence (GPC) rules
(i.e., “letter–sound rules”). The outcomes of these studies
have been systematically reviewed 3 times in the past
decade. The National Reading Panel review considered 38
studies that tested a variety of phonics programs that may
be used in schools (e.g., synthetic, analytic, analogy) with
or without simultaneously training in other skills (e.g., pho-
neme awareness or learning irregular words by sight). In
children with poor reading, phonics training had a moderate
and significant effect on reading accuracy for “nonwords”
(i.e., nonsense words that follow the GPC rules; Cohen’s d =
0.52) and “regular words” (i.e., real words that follow the
GPC rules; Cohen’s d = 0.49), and a small but significant
effect on reading mixed words (i.e., reading both regular
words and “irregular” words, the latter being words that
contain graphemes that “break” the GPC rules, such as
YACHT). A more recent review by Suggate (2010) included
85 studies that tested programs that trained phonics, pho-
neme awareness, and comprehension. Unfortunately, it did
not report on the effect of phonics training in poor readers
per se. However, it did report that the mean effect on reading
across all reading programs was moderate (Cohen’s d =
0.51) for poor readers. Very recently, a Cochrane Review
(McArthur et al., 2012) identified 11 studies that looked at
504996JLD
XXX10.1177/0022219413504996Journal of Learning DisabilitiesMcArthur et al.
research-article2013
1
Macquarie University, Sydney, Australia
Corresponding Author:
Genevieve McArthur, ARC Centre of Excellence in Cognition and Its
Disorders, Department of Cognitive Science, Macquarie University,
Sydney, NSW, 2109 Australia.
E-mail: genevieve.mcarthur@mq.edu.au
Sight Word and Phonics Training in
Children With Dyslexia
Genevieve McArthur, PhD
1
, Anne Castles, PhD
1
,
Saskia Kohnen, PhD
1
, Linda Larsen, BSc Hons
1
,
Kristy Jones, BSc Hons
1
, Thushara Anandakumar, BSc Hons
1
,
and Erin Banales, BSc Hons
1
Abstract
The aims of this study were to (a) compare sight word training and phonics training in children with dyslexia, and (b)
determine if different orders of sight word and phonics training have different effects on the reading skills of children with
dyslexia. One group of children (n = 36) did 8 weeks of phonics training (reading via grapheme–phoneme correspondence
rules) and then 8 weeks of sight word training (reading irregular words as a whole), one group did the reverse (n = 36), and
one group did phonics and sight word training simultaneously for two 8-week periods (n = 32). We measured the effects
of phonics and sight word training on sight word reading (trained irregular word reading accuracy, untrained irregular
word reading accuracy), phonics reading (nonword reading accuracy, nonword reading fluency), and general reading (word
reading fluency, reading comprehension). Sight word training led to significant gains in sight word reading measures that
were larger than gains made from phonics training, phonics training led to statistically significant gains in a phonics reading
measure that were larger than gains made from sight word training, and both types of training led to significant gains in
general reading that were similar in size. Training phonics before sight words had a slight advantage over the reverse order.
We discuss the clinical implications of these findings for improving the treatment and assessment of children with dyslexia.
Keywords
dyslexia, treatment, sight word training, phonics, reading

2 Journal of Learning Disabilities XX(X)
the effect of “pure” phonics training programs in English-
speaking poor readers (i.e., programs that taught reading via
the GPC rules alone with little or no simultaneous training
in any other skills). It reported a large and significant effect
of phonics on nonword reading accuracy (d = 0.76), a mod-
erate and significant effect on word reading accuracy (d =
0.51), and a small to moderate significant effect on GPC
rule knowledge (d = 0.35).
How might phonics training improve word reading in
children? When children first see the word CAT, they have to
(a) identify the letters, (b) translate each grapheme (i.e., a let-
ter or letter cluster) into a speech sound (i.e., a phoneme; e.g.,
“k” “a” “t”), and (c) blend these phonemes together into a
word that is spoken aloud (e.g., “kat”). Boxes 1 (letter identi-
fication), 2 (GPC knowledge), and 6 (phonological output) in
Figure 1 represent these three “components” of phonics read-
ing. Collectively, these three components represent the “pho-
nics” or “nonlexical” reading route in a typical dual route
model of reading (Coltheart, Rastle, Perry, Langdon, &
Ziegler, 2001). Note that this is just one of a number of read-
ing models used in dyslexia research, which all incorporate
some or all of the reading components shown in Figure 1
(e.g., see also Ellis & Young, 1988; Perry, Ziegler, & Zorzi,
2007; Plaut, McClelland, Seidenberg, & Patterson, 1996).
Once a word has been read a number of times via the
phonics route, a memory is formed of the whole word (i.e.,
the combination and order of the letters in the word; e.g.,
SHIP; see Box 3 in Figure 1—the orthographic lexicon
component). This memory activates the meaning of that
word (i.e., a boat; see Box 4—the semantic knowledge
component), the spoken representation of that word (“ship”;
see Box 5—the phonological lexicon component), and the
spoken output of that word (“ship”; see Box 6—the phono-
logical output component which is also apart of the phonics
route). Together, these components form the “sight word”
or “lexical” reading route of the dual route model of reading
in Figure 1. It is important to note that the proposed “knock-
on” effect of the letter–sound route onto the sight word
route—via the development of orthographic representations
that link to semantic representations and phonological rep-
resentations in dual route models—is not unique to dual
route models. Most major models and theories of reading
acquisition—including triangle models and the self-teach-
ing hypothesis—posit that learning words using letter–
sound rules helps develop whole-word written
representations of those words, which are linked to the
meaning and spoken representations of those words (Nation,
2009; Share, 1995). Thus, many (if not most) reading
“Sight word” or “lexical”
reading route
See word
CAT
“Phonics” or “non-
lexical” reading route
1.Letter identification
C A T
3.Orthographic lexicon
CAT
4.Semantic knowledge
furry pet
2.GPC knowledge
“k” “a” “t”
5.Phonological lexicon
‘kat’
6.Phonological output
“kat”
Say word
“kat”
Figure 1. A typical dual route model of reading (Coltheart et al., 2001).

McArthur et al. 3
researchers agree that phonics reading plays an important
role in the development of sight word reading.
Sight word reading is particularly important for reading
English because one third of written words in English con-
tain letters that do not follow the letter–sound rules (i.e.,
they are “irregular”; Coltheart et al., 2001). For example,
ACH in YACHT sounds like “o” and not “a” “ch.” Most
irregular words can be partially read with the GPC rules
since some of the letters are regular (e.g., Y and T in YACHT
follow the letter–sound rules “y” and “t”). However, to be
learned accurately, irregular words like YACHT must be
memorized as a whole word (i.e., read via the sight word
reading route).
Given the importance of sight word reading in English,
and given that many children with dyslexia have problems
with sight word reading as well as phonics reading
(Brunsdon, Hannan, Nickels, & Coltheart, 2002; Castles
& Coltheart, 1993; Goulandris & Snowling, 1991), it is
surprising that no one has tested the specific effect of sight
word training in a group of children with dyslexia. Thus,
the first aim of this study was to compare the effect of
sight word training to phonics training in children with
dyslexia. We measured gains in the sight word route using
trained and untrained irregular words (i.e., words that
comprise letters that do not follow the common letter–
sound rules, such as YACHT), which can be correctly read
by the sight word route but not the phonics route. We mea-
sured gains in the phonics route using “nonwords” (i.e.,
nonsense words that comprise letters that all follow the
common GPC rules, such as GRENTY). Since nonwords
do not have representations in the sight word route (i.e., in
the orthographic lexicon), they can be read correctly only
via the phonics route. We also included two general mea-
sures of reading that included both irregular and regular
words (i.e., a word reading fluency test and a reading com-
prehension test).
We made four predictions about the outcomes relating to
our first aim. First, regarding trained irregular words, from
the few case studies that have trained irregular word reading
in children with dyslexia (e.g., Broom & Doctor, 1995;
Brunsdon et al., 2002; Rowse & Wilshire, 2007), we pre-
dicted that sight word training would lead to a statistically
significant gain in reading accuracy for trained irregular
words, and that this gain would be larger than that made
from phonics training. Second, regarding untrained irregular
words, we predicted from the self-teaching hypothesis that
phonics training would lead to gains in reading untrained
words (Share, 1995; Wang, Castles, Nickels, & Nation,
2011). It was not possible to predict the effect of sight word
training on untrained irregular words since the aforemen-
tioned case studies produced mixed findings on this issue,
and because there is no well-specified theory of how gener-
alization may occur as a result of sight word training. Third,
regarding nonword reading accuracy and fluency, we
predicted that phonics training would lead to statistically
significant gains in measures of the phonics route, and that
these gains would be larger than those made from sight word
training. Fourth, regarding word reading fluency and read-
ing comprehension, we predicted that sight word training
and phonics training would have similar-sized effects on
both of these measures since the irregular words in each test
should benefit from sight word training and the regular
words should benefit from phonics training.
The second aim of the current study was to determine if
the order of phonics and sight word training is important for
treating children with dyslexia. There is a widely held, yet
not uncontested (Connor et al., 2009; Wyse & Goswami,
2008), belief that phonics reading is the foundation of read-
ing and so should precede other types of instruction.
According to this view, teaching reading via the GPC rules
(i.e., phonics reading) should empower children to fully
decipher or “decode” regular words or names that they have
never encountered before (e.g., GOBLET, PROFESSOR
MCGONAGALL), and partially decode irregular words
that they have never encountered before (e.g., FRIEND).
Repeatedly decoding a word—either fully or partially—
will create full or partial representations of the whole word
(see the orthographic lexicon in Figure 1, as well as Rose,
2006; Share, 1995). This will increase the ability to recog-
nize that word by sight. According to this view, training
phonics reading before sight word reading should lead to
greater gains in sight word reading than vice versa. We
tested this prediction by comparing groups of children with
dyslexia who did (a) phonics training followed by sight
word training (phonics + sight word group) and (b) sight
word training followed by phonics training (sight word +
phonics group). We also included a third group who did
phonics training and sight word training simultaneously
(mixed + mixed group) because, despite the widespread
belief that phonics should be trained before sight word read-
ing, many teachers, clinicians, and reading programs train
phonics reading and sight word reading simultaneously.
Thus, we wished to compare the efficacy of mixed + mixed
training to phonics + sight word training and sight word +
phonics training.
In sum, this study had two aims, which were to (a) com-
pare sight word training and phonics training in children
with dyslexia and (b) determine if different orders of sight
word training and phonics training have different effects on
the reading skills of children with dyslexia. We predicted
that (a) sight word training would lead to statistically sig-
nificant gains in sight word reading measures (trained and
untrained irregular words) that would be larger than gains
made from phonics training, (b) phonics training would
lead to statistically significant gains in phonics reading
measures (nonword reading accuracy and fluency) that
would be larger than gains made from sight word training,
and (c) phonics training and sight word training would have

4 Journal of Learning Disabilities XX(X)
similar-sized significant effects on measures of reading that
taxed both phonics and sight word reading (word reading
fluency and reading comprehension).
Method
The Macquarie University Human Ethics committee
approved the methods outlined below. All children and their
parents gave informed written or verbal consent to partici-
pate in the research. Below we outline the method accord-
ing to CONSORT 2010 guidelines (Shulz, Altman, &
Moher, 2010).
Trial Design
Figure 2 illustrates the order of testing sessions and training
phases in this study. At Test 1, children completed the
screening and outcome measures (see Table 1). After 8
weeks of no training, they returned to do the outcome mea-
sures (Test 2). This “no-training” period indexed the degree
to which children made gains on the outcome measures due
to factors beyond the training involved in this study, such as
test–retest effects, familiarity with the testing environment,
regression to the mean, or growth in reading due to normal
schooling (see the discussion for why we controlled for
gains during a no-training period within groups rather than
between groups). The phonics + sight word group then did
8 weeks of phonics training (and then Test 3) followed by 8
weeks of sight word training (and then Test 4). The sight
words + phonics group did the same except the order of
training was reversed. The mixed + mixed group did pho-
nics and sight word training on alternate days for 8 weeks
(and then Test 3) and then the same again for another
8 weeks (and then Test 4).
Participants
Children were recruited from schools, from clinics, and via
newspaper advertisements. They were included in the study
if they (a) were aged from 7 to 12, (b) scored below the
average range for their age (i.e., had a z score lower than –1)
on the Castles and Coltheart 2 (CC2) irregular-word read-
ing test or nonword reading test (Castles et al., 2009), (c)
had no history of neurological or sensory impairment as
indicated on a background questionnaire, and (d) used
English as their primary language at school and at home
(see the screening measures in Table 1). It is noteworthy
that although all children were tested for their nonverbal
intelligence, children with nonverbal IQ scores below the
average range were not excluded from the study since intel-
ligence does not appear to predict reading ability or response
to treatment (Gresham & Vellutino, 2010; Siegel, 1989).
The children were given four screening tests. We tested
nonword reading and irregular-word reading accuracy with
the CC2 reading test (Castles & Coltheart, 1993; Castles
et al., 2009). The test comprises 40 nonwords (e.g.,
GRENTY), 40 irregular words (e.g., YACHT), and 40 regu-
lar words (e.g., MARSH). The type of item was presented
in a random fixed order (e.g., nonword, irregular word,
regular word, regular word, nonword, irregular word) with
items within each type presented in order of difficulty. Each
item was printed on a separate card. Children were asked to
read the item on each card until they made five errors in a
row for any one type of item (e.g., irregular words). At this
Phonics+sight word (N = 36) Sight word+phonics (N=36) Mixed+mixed (N = 32)
Test 1
2-3 hours
Screening measures
Outcome measures
Screening measures
Outcome measures
Screening measures
Outcome measures
No training
8 weeks
No training No training No training
Test 2
2-3 hours
Outcome measures Outcome measures Outcome measures
Train 1
8 weeks
Phonics Sight words Mixed
Test 3
2-3 hours
Outcome measures Outcome measures Outcome measures
Train 2
8 weeks
Sight words Phonics Mixed
Test 4
2-3 hours
Outcome measures Outcome measures Outcome measures
Figure 2. Testing and training phases for the three training groups.

McArthur et al. 5
Table 1. Means and Standard Deviations for the Screening and Outcome Measures.
Phonics + Sight Word Sight Word + Phonics Mixed + Mixed
M SD M SD M SD ANOVA
Screening
Age (years) 9.42 1.71 9.19 1.64 9.21 1.67 ns
Nonverbal IQ (s) 97.50 14.16 95.56 17.12 101.12 14.25 ns
CC2 irregular words (z) –1.31 0.67 –1.38 0.64 –1.26 0.58 ns
CC2 nonwords (z) –1.50 0.57 –1.27 0.54 –1.32 0.53 ns
CC2 regular words (z) –1.41 0.57 –1.29 0.59 –1.30 0.59 ns
Training time
Sight word training (sessions/40) 31.94 11.96 34.67 7.70 33.83 7.88 ns
Phonics training (sessions/40) 36.10 5.78 33.71 10.10 33.83 7.88 ns
Phonics training (mins/800–1,000) 821.16 330.03 679.89 360.96 738.96 265.64 ns
Test 1
Trained irregular accuracy (r) 11.17 7.77 10.97 8.09 11.16 8.07 ns
Untrained irregular accuracy (r) (s) 8.58 6.39 8.50 6.37 9.03 6.28 ns
Nonword reading accuracy (r) 7.06 6.35 9.08 6.49 8.31 7.29 ns
Nonword reading fluency (r) 11.67 9.24 13.08 6.95 14.66 11.19 ns
Nonword reading fluency (s) 81.11 8.76 85.36 7.91 86.09 8.55
Word reading fluency (r) 42.47 16.88 40.83 16.84 41.78 16.74 ns
Word reading fluency (s) 87.14 8.53 86.94 8.27 87.66 8.91
Reading comprehension (r) 14.25 8.37 13.75 8.20 14.59 7.89 ns
Test 2
Trained irregular accuracy (r) 12.22 8.00 12.28 7.99 12.44 8.22 na
Untrained irregular accuracy (r) (s) 9.67 6.46 9.81 6.79 10.44 6.02 na
Nonword reading accuracy (r) 8.33 5.50 9.25 6.47 9.19 6.83 na
Nonword reading fluency (r) 13.69 7.63 14.86 7.03 15.06 9.98 na
Nonword reading fluency (s) 83.86 8.86 86.56 7.04 86.72 6.98
Word reading fluency (r) 46.44 16.26 44.08 18.12 42.91 16.37 na
Word reading fluency (s) 89.36 10.35 90.44 17.50 88.25 7.82
Reading comprehension (r) 16.08 7.33 15.64 7.50 15.78 7.28 na
Test 3
Trained irregular accuracy (r) 13.83 8.15 16.22 8.21 16.19 8.20 na
Untrained irregular accuracy (r) (s) 10.67 6.67 11.03 6.54 11.78 6.89 na
Nonword reading accuracy (r) 9.81 6.50 10.50 5.97 10.75 7.44 na
Nonword reading fluency (r) 15.39 9.14 16.17 7.52 16.59 10.90 na
Nonword reading fluency (s) 84.75 9.17 87.78 7.56 86.69 7.74
Word reading fluency (r) 49.17 17.30 45.25 16.21 46.53 16.25 na
Word reading fluency (s) 87.56 19.03 89.11 9.01 89.66 10.39
Reading comprehension (r) 17.78 7.59 17.31 6.86 18.53 6.25 na
Test 4
Trained irregular accuracy (r) 16.31 8.56 16.11 8.64 17.44 7.96 na
Untrained irregular accuracy (r) (s) 12.28 7.31 10.89 6.78 12.44 6.93 na
Nonword reading accuracy (r) 10.06 6.70 12.72 6.76 10.81 7.42 na
Nonword reading fluency (r) 15.83 9.67 16.11 8.29 16.75 11.73 na
Nonword reading fluency (s) 83.94 10.20 85.89 8.27 85.66 9.37
Word reading fluency (r) 49.81 16.82 50.36 18.09 47.84 15.73 na
Word reading fluency (s) 88.50 11.57 93.19 9.07 89.03 9.55
Reading comprehension (r) 19.03 6.82 17.97 6.63 18.47 6.78 na
CC2 irregular words (z) –1.21 0.78 –1.19 0.73 –1.20 0.82 na
CC2 nonwords (z) –1.31 0.60 –1.06 0.71 –1.19 0.53 na
Note. CC2 = Castles and Coltheart 2 (Castles et al., 2009) reading tests; m = minutes; na = not applicable since group comparisons based on difference
scores (see Figure 4); ns = nonsignificant effect; r = raw score; s = standard score; z = z score.

6 Journal of Learning Disabilities XX(X)
point, the presentation of that type of item was stopped.
Presentation of the other types of items continued until the
child made five errors for each of the other types of items or
they reached the end of the test. A child was given 10 s to
read each word before being prompted to try the next word.
Scores were z scores that had a mean of 0 and SD of 1.
We measured nonverbal IQ with the Kaufman Brief
Intelligence Test–2 Matrices subtest (Kaufman & Kaufman,
2004). In each trial, children saw an incomplete picture
matrix and had to select the missing portion from six pos-
sible options. Scores were standardized with a mean of 100
and an SD of 15.
We used a questionnaire to determine if children had any
known problems with their hearing, vision, neurology, or
psychology, which could account for their reading problem.
We also used this questionnaire to establish if the children
used English as their primary language at both school and
home.
Interventions
Sight word training. Children were asked to do five sight-
word training sessions per week for 8 weeks. Parents
recorded the number of sessions that they completed by
placing a sticker on a calendar for each training session
completed. Each sticker was a gold medallion marked with
a $ sign. This indicated to the child that they had earned $1
for that session.
Each training session, which was designed to take 30 min,
used one of 30 lists of 24 irregular words that increased in
difficulty both between and within lists. A word was consid-
ered irregular if at least one grapheme in the word did not
correspond to its most frequent pronunciation. We did not
train any irregular words that were in the untrained irregular
word measure, the word reading fluency measure, or the
reading comprehension measure (see the Outcomes sec-
tion). Irregular words were used in training because these
words can be read correctly only via the sight word reading
route, and so any improvements in reading after training
irregular words should reflect changes in the sight word
reading route.
In the first 5 to 10 min of the first sight word training
session, a parent tested their child’s ability to read the easi-
est list of 24 irregular words (List 1) by presenting each
word on a flash card. Words that the child read correctly
were placed in a “correct pile.” Words that they read incor-
rectly were corrected and were then placed in an “incorrect
pile.” In the next 15 to 20 min of the session, the child
played a computerized game called Dingo Bingo by
MacroWorks®, which was adjusted by Shane Davis (the
CEO) to present the same list of irregular words that were
used for the flash cards. In this game, a child is presented 9
to 24 words with each written in one section of a grid (i.e.,
as in a bingo game). In each trial, the program “says” one of
the words aloud, and the child is required to click on the
correct word. They receive points for each correctly
“clicked” (i.e., identified) word. The parent noted any
words that the child failed to read in Dingo Bingo, and
added them to the incorrect pile. In the final 5 to 10 min of
the session, the parent re-presented the child with the flash-
cards in the incorrect pile. If a child got less than 2 incorrect
(i.e., 0 or 1 mistake), then they moved onto the next list the
next day (List 2). Otherwise they redid the same list in the
next training session.
It is noteworthy that the sight word training focused on
reading accuracy rather than fluency. It is also noteworthy
that training was done at home with the support of both a
parent and computer. We took this approach for clinical rea-
sons. An overarching goal of our research program is to
increase the accessibility of reading training to children
with dyslexia so we can help as many children as we can
regardless of location or economic situation. To achieve this
goal, we need to find ways to deliver training that is afford-
able and convenient as well as theoretically rigorous and
scientifically validated. The delivery of training via com-
puter and a parent is convenient because it can be done at
home, avoiding the stress and effort involved in driving
through heavy traffic to attend the university five times per
week for city children. It also allowed us to include children
in outer suburbs into the study since they only needed to
drive to the university four times for test sessions. It is also
inexpensive since the training does not depend on the skills
of a trained reading specialist.
Phonics training. Children and parents were instructed to do
the phonics training at home for 30 min per day, 5 days per
week, for 8 weeks. All training was done on a computer for
the same clinical reasons outlined above. As for the sight
word training, children used tokens on a calendar to indi-
cate how many 30-min sessions they completed. They
received $1 per session. In addition, the training software
recorded the number of minutes the child spent at each
level of the training. This time did not include time spent
moving between levels or for the instructions provided by
the software. Thus, we expected the software to record 20
to 25 min per session, which was a total of 800 to 1,000
min overall.
Like the sight word training, phonics training focused on
accuracy rather than fluency. The phonics training was a
modified version of the Lexia® Strategies for Older
Students, which uses a wide variety of games and exercises
to teach the pairing of written stimuli (i.e., letters, letter
clusters, syllables, morphemes, whole words, phrases, and
sentences) to the spoken versions of those stimuli. For
example, in one exercise children are asked to pair together
syllables to create a complete word. In another exercise,
they are asked to find words within a grid of letters. And in
another exercise, they are asked to type a letter to complete

McArthur et al. 7
a word. The modifications—provided by the managing
director of Lexia® Learning Australia, John Dyson—
involved removing all exercises that included irregular
words. We also removed exercises that presented sentences
and paragraphs of text. Thus, the program focused on train-
ing GPCs either alone, within parts of words (i.e., sylla-
bles), or within regular words.
Mixed training. The mixed training was the same as the pho-
nics and sight word training except that each type of train-
ing was done on alternate days. So, on the first day the child
did phonics training, the second day sight word training, the
third day phonics training, and so on.
Outcomes
Since both sight word training and phonics training
focused on improving word reading accuracy rather than
fluency, the primary outcomes of this study were trained
and untrained irregular word reading accuracy (see the
Sight Word Reading section below) and nonword reading
accuracy (see the Phonics section below). We also mea-
sured word and nonword reading fluency and reading
comprehension, which were secondary outcomes of the
study.
Although a number of our outcome measures had stan-
dard scores, we indexed training gains using raw scores
since these avoid problems arising when children shift from
the upper end of one age band in the pretraining session
(which inflates their standardized score) to the lower end of
the next age band in the posttraining session (which deflates
their standardized score). Subtracting an inflated standard
score from a deflated standard score creates the impression
that training has impaired children’s ability even if the train-
ing had a positive effect.
Sight word reading. We measured the accuracy of the sight
word route by asking children to read aloud 60 irregular
words that were presented on flashcards. Half of the words
were included in the sight word training program (“trained
irregular words”) and half were not (“untrained irregular
words”). Untrained irregular words were matched to the
trained irregular words in terms of their written frequency,
length in letters, and relative irregularity (i.e., the propor-
tion of irregular GPCs in a word relative to the total number
of GPCs in that word). Scores were total correct trained
irregular words (out of 30) and total correct untrained irreg-
ular words (out of 30).
We measured accuracy of responses to trained irregular
words to get a direct measure of the effect of the sight word
training, which trained irregular words. We measured
responses to untrained irregular words to assess whether
phonics or sight word training led to improvements beyond
trained irregular words. Specifically, phonics might improve
untrained irregular words by improving children’s ability to
use their GPC knowledge (see Component 2 in Figure 1) to
sound out part of an irregular word (e.g., “y” “t” in YACHT)
and then “guess” the word using their knowledge of spoken
words that might match (i.e., see the phonological lexicon
in Figure 1; Share, 1995). Sight word training might
improve untrained irregular words by improving a child’s
access to partially learned written words in the orthographic
lexicon (see component 3 in Figure 1; Brunsdon, Coltheart,
& Nickels, 2005).
Phonics reading. We tested the effect of the training pro-
grams on two measures of the phonics reading route. Both
tests used nonwords as stimuli since nonwords can only be
read correctly by the phonics reading route. We tested non-
word reading accuracy using 20 untrained nonwords printed
on flashcards. A child was asked to read each nonword
aloud. All items were monosyllabic, comprised three or
four letters (e.g., urk, clon), and translated to two, three or
four phonemes. Half the items contained digraphs (e.g., th,
ai, oo), and half single-letter correspondences (e.g., t, p, e).
Scores were total correct out of 20.
We indexed nonword reading fluency using the Test of
Word Reading Efficiency (TOWRE) Nonword subtest
(Torgesen, Wagner, & Rashotte, 1999). This comprised 63
increasingly difficult nonwords that can be read correctly
using the GPC rules. The children were asked to read as
many nonwords as they could in 45 s. Scores were the total
responses correct out of 63.
Both reading routes. We tested the output of both reading
routes using two general measures of reading that included
both irregular and regular words. The TOWRE Sight Word
subtest (Torgesen et al., 1999), which measured word read-
ing fluency, comprised 104 irregular (sight word route) and
regular words (sight word and phonics route) that increased
in difficulty. Children were asked to read as many words as
possible in 45 s. Scores were the total responses correct out
of 104.
We tested reading comprehension using the Test of
Everyday Reading Comprehension (McArthur et al., 2013).
This included 13 “everyday” reading stimuli, such as a text
message, a medicine label, or a cafeteria menu. For each
stimulus, children were asked two literal (i.e., not inferen-
tial) questions based on the information in the text. Scores
were the total responses correct out of 26.
Sample Size
A flow diagram of the number of participants in each stage
of the study is shown in Figure 3. At the end of the study,
there were 36 children in the phonics + sight word group, 36
children in the sight word + phonics training group, and 32
children in the mixed + mixed training group.

8 Journal of Learning Disabilities XX(X)
Sequence Generation
For funding-related reasons, we had to complete all data
collection for this study in 2 years. This meant that we had
to recruit children for the study from Month 1 to Month 18,
since the last child recruited into the study would take 6
months to complete his or her training. For reasons beyond
our control, there was a 3-month delay in the development
of the sight word training. The only solution to this unex-
pected problem was to divide the 18-month recruitment
period into three periods of time. We then allocated children
recruited in Months 1 to 6 to the phonics + sight word train-
ing group, children recruited in Months 7 to 12 to the sight
word + phonics group, and children recruited in Months 13
to 18 to the mixed + mixed group.
There is good evidence that this quasi-randomized allo-
cation procedure did not bias the outcomes of this study.
First, the groups were very well matched prior to training
(see Table 1). Second, for all bar one outcome, groups made
similar gains after 16 weeks of training, indicating that allo-
cation did not produce any group that was unusually respon-
sive or unresponsive to treatment. Third, for the exceptional
outcome, the group difference was in the predicted direc-
tion, indicating that superior group performance was a
result of a genuine experimental effect rather than a group
allocation effect. Fourth, this study was designed so that
there could be no possible bias between allocation to inter-
vention and control groups since each individual partici-
pated in both control and intervention periods, and any
gains in the control period were controlled for in the inter-
vention period statistically (i.e., we used a double-baseline
design that gauged the effect of no training in each and
every participant before they did training).
Allocation Concealment
Each recruitment period had a fixed start date and an end
date. Children were allocated to their group according to
Figure 3. A flow diagram of the participant numbers at each stage of the study.

McArthur et al. 9
when they were recruited for the study. Since children could
be allocated to only one group, it is highly unlikely that lack
of allocation concealment introduced bias into the study.
Implementation
The first author defined the three recruitment periods in this
study (i.e., the start and finish dates). The last four authors
enrolled participants into the study. These authors arranged
the appropriate training, which depended on the fixed
recruitment period. All training was conducted by computer
(phonics training) or by a combination of computer and a
parent (sight word training). All instructions to parents were
provided via documents (i.e., there was no one-on-one
training of parents). If parents were unclear about any
aspect of the training, they contacted one of the last four
authors for support. These authors all joined the study at
different points in time, and so varied in their support expe-
rience. Thus, parents in later stages of the study (i.e., sight
word + phonics and mixed + mixed) did not necessarily
receive support from more experienced study authors than
parents in the phonics + sight word group.
Blinding
Unlike drug trials, cognitive treatment trials find it difficult
to guarantee double blinding because the type of training
cannot be completely concealed from a volunteer. However,
neither parents nor children were told their group alloca-
tion, and it is highly unlikely that they had the expertise to
ascertain the type of training that they were receiving (i.e.,
they were blind to group allocation). Furthermore, all chil-
dren received exactly the same type of training in this study.
The only difference was the order in which they did the
training. This would further obscure group allocation to
children and their parents.
Regarding personnel, it is unethical and impractical to
hire and fire trained reading experts in long-term continu-
ous studies to ensure that different testers are used before
and after training. Thus, we employed four casual testers to
help two principal testers. With careful planning, we
ensured that no tester assessed the same child twice, and no
tester was aware of the child’s group allocation (i.e., the
tester was blind to group allocation).
Results
Participant Flow
A flow diagram of the number of participants in each stage
of the study is shown in Figure 3. We screened 193 children.
43 children were excluded because they did not fit our cri-
teria, and 9 withdrew before the first test session. Of the 43
children allocated to the phonics + sight word group, 2
dropped out of the no-training period (4.6%), 2 out of the
phonics training (4.6%), and 3 out of the sight word training
period (6.9%; final n = 36). Of the 44 children who started
in the sight word + phonics group, 2 dropped out in the no-
training period (4.5%), 2 out of the sight word training
(4.5%), and 4 out of the phonics training (9.1%; final n =
36). Of the 54 who started in the mixed + mixed group, 8
dropped out of the no-training period (14%), 8 out of the
first mixed training phase (4%), and 6 out of the second
mixed training phase (11%). All participants withdrew from
the training on their own accord for a variety of reasons
relating to a participant’s personal or family circumstances.
We did not exclude any participants that completed the
study from the analysis. Thus, the drop out rate in this study
was low, and the reason for drop out appeared to be
random.
Baseline Data
Before analyzing training effects on outcome measures, it is
important to establish that training groups do not differ on
screening or outcome measures prior to training. This
ensures that any improvements in outcome measures after
training are not due to regression to the mean effects (i.e.,
according to statistical probability, any person or group
with an extreme score at one point in time is likely to have
a more moderate score at a second point in time even if
there is no actual change in their ability). We used a between
groups ANOVA to compare the screening and outcome
measure scores of the three training groups prior to training
(i.e., at Test 1). The relevant data are shown in Table 1. Prior
to training, there was no significant difference between the
three groups on any screening or outcome measure.
Training Fidelity
Before analyzing training effects on outcome measures, it is
also important to establish that training groups did not differ
in the amount of training that they did with the sight word
and phonics programs. As mentioned above, for the phonics
training, the computer recorded the number of minutes
spent at each level. For both the phonics training and the
sight word training, parents indicated how many sessions
were done using $1 sticker tokens on a calendar. Note that
we could not record the exact number of minutes done for
the sight word training because (a) it was partly conducted
with a parent and (b) the computerized part of each session
was not programmed to record number of minutes Dingo
Bingo played. However, each sight word training session
was carefully designed to take 30 min to match each pho-
nics training session. Thus, each token roughly represented
a 30-min session.
It is noteworthy that although children were asked to do
five 30-min sessions for 8 weeks for each training program

10 Journal of Learning Disabilities XX(X)
(40 sessions in total), based on previous training studies, we
expected that this request would prompt children to manage
4 sessions per week (32 sessions in total) given unexpected
illnesses, holidays, busy schedules, and the occasional “bad
day” (McArthur, Ellis, Atkinson, & Coltheart, 2008). Our
expectation proved correct since children in each group
reported 31 to 36 training sessions for each program. The
amount of time (phonics training) and number of sessions
(sight word training) completed by each group are illus-
trated Table 1. There was no significant difference between
groups in the amount of training done for each type of
training.
Numbers Analyzed
The analyses included 36 children in the phonics + sight
word group, 36 children in sight word + phonics group, and
32 children in the mixed + mixed group. We analyzed the
data of participants in the groups to which they were origi-
nally allocated. We conducted an available case analysis on
the data (i.e., based on participants with complete data)
rather than an intention-to-treat analysis (ITT; i.e., based on
all allocated participants including those with missing data)
for three reasons. First, as mentioned above, missing data
were minimal and apparently random. Second, ITT pro-
vides not a true test of a treatment (i.e., as it was designed to
be done) but rather the execution of the treatment in a par-
ticular study (Peduzzi, Henderson, Hartigan, & Lavori,
2002). Third, ITT requires missing data to be imputed, but
there is no agreement about how this is best done. For
example, the most common method in randomized con-
trolled trials—last observation carried forward—has been
criticized for numerous reasons (Lavori, 1992; Peduzzi
et al., 2002).
Outcomes
Figure 4 shows each group’s mean and 95% confidence
intervals (CIs) for gains in raw scores (i.e., difference
scores) for each outcome measure (i.e., trained irregular
word accuracy, untrained irregular word accuracy, non-
word reading accuracy, nonword reading fluency, word
reading fluency, and reading comprehension) from Test 1
(pretraining) to Test 2 (after 8 weeks of no training; T1T2)
after 8 weeks of training (T1T3), and after 16 weeks of
training (T1T4; see Table 1 for mean raw scores from
which these difference scores were calculated). A positive
difference score means that a child’s score was better in the
latter test. The first three means and 95% CIs (i.e., vertical
lines) in each graph belong to the phonics + sight word
group. The second three vertical lines relate to the sight
word + phonics group. The last three lines relate to the
mixed + mixed group. It is noteworthy that in the 1970s, it
was claimed that difference scores had inferior reliability.
However, work in the 1980s and 1990s established that this
is not the case (Bonate, 2000). Difference scores have
superior reliability when degree of improvement differs
between trainees (Rogosa & Willett, 1983) and are more
appropriate than posttreatment means for quasi-randomized
trials (Cribbie & Jamieson, 2004).
The first mean and 95% CI for each group, labeled
T1T2, represents the shift in raw scores from Test 1 to
Test 2. These two tests flanked an 8-week period of no
training, and so any positive gain reflects an improvement
due to factors beyond the training effects in this study. Any
T1T2 line that does not fall across the 0 line (i.e., the point
of no difference between raw scores at each test time) is
statistically significant (p ≤ .05). These are marked with an
asterisk.
The second line for each group, labeled T1T3, reflects
the difference in raw scores on the test between Test 1
(before training) and Test 3 (after the first 8 weeks of train-
ing). If this line does not cross the 0 line (i.e., no difference
in raw scores between Test 1 and 3) then this means there
was a statistically significant increase in raw scores between
tests (marked with an asterisk). For this significant increase
to be considered a training effect, it had to be significantly
larger than gains made in the no-training period (i.e., T1T2;
also marked with an asterisk). Any T1T3 line that repre-
sents a significant training effect is marked with two
asterisks.
The third line for each group, labeled T1T4, is the differ-
ence in raw scores between Test 1 (before training) and Test
4 (after 16 weeks of training). If this line does not fall over
the zero line (marked with an asterisk), and if the raw differ-
ence score is significantly larger than the gains made over
the no-training period (T1T2; marked with an asterisk) then
there is a effect of 16 weeks of training (i.e., marked with
two asterisks). For T1T2, T1T4, and T1T3, we have pro-
vided the Cohen’s d effect sizes calculated from the differ-
ence scores (i.e., mean group difference score/SD group
difference score) for each effect underneath the asterisk
marks (Howell, 2010). We used Cohen’s d effect sizes since
these are most commonly used in controlled trials and allow
for direct comparisons between studies that use different
outcome measures.
As well as testing if each type of training had an effect,
we wanted to determine if there was a reliable difference
between the size of effect for the different types of training
done over 8 weeks (i.e., phonics versus sight words versus
mixed). For each outcome, we used a between-group
ANCOVA (controlling for each group’s gains over the no-
training period on that outcome measure) to compare T1T3
gains for the phonics + sight word, sight word + phonics,
and mixed + mixed groups. We also wanted to determine if
the different orders of training had different effects on each
outcome. We tested this in two steps: We used (a) a
between-groups ANCOVA (controlling for T1T2

McArthur et al. 11
Figure 4. Means and 95% confidence intervals (CIs) for groups’ gains in raw scores (i.e., difference scores) for each outcome
measure (i.e., trained irregular word accuracy, untrained irregular word accuracy, nonword reading accuracy, nonword reading
fluency, word reading fluency, and reading comprehension) from Test 1 (pretraining) to Test 2 (after 8 weeks of no training; T1T2)
and after 8 weeks of training (T1T3) and after 16 weeks of training (T1T40).
Note: A positive difference score means that a child’s score was better in the latter test. The first three means and 95% CIs (“lines”) in each graph
represent data for the phonics + sight word group. The second three lines relate to the sight word + phonics group. The last three lines relate to the
mixed + mixed group. Any line that does not fall across the 0 line (i.e., the point of no difference between raw scores at each test time) represents
a statistically significant gain (p ≤ .05), and is marked with an asterisk. Any such gain that is significantly larger than the T1T2 gain in the same group
represents a significant training effect. Such gains are marked with two asterisks.

12 Journal of Learning Disabilities XX(X)
no-training period gains) to compare T1T4 gains for each
group and (b) a repeated measures group (phonics + sight
word, sight word + phonics, mixed + mixed) by gain
(T1T3, T1T4) ANCOVA (controlling for T1T2 gains over
the no-training period). Below, we interpret the result for
each outcome measure in turn.
Trained irregular word accuracy. Eight weeks of phonics,
sight word, and mixed training had very large and signifi-
cant training effects on trained irregular word accuracy.
Figure 4 reveals that the two groups that did sight word
training (sight word + phonics group and mixed + mixed
group) made larger gains than the group that did phonics
training. The between-group ANCOVA for the T1T3 data
showed that these group differences were statistically
significant.
Sixteen weeks of phonics and sight word training had a
very large and significant training effect on trained irregular
word accuracy. The between-groups ANCOVA revealed no
significant difference between the groups after 16 weeks of
training. The repeated measures ANCOVA revealed an
effect of gain because the T1T4 gains were larger than T1T3
gains, F(1, 100) = 12.06, p < .01. There was also a signifi-
cant effect of group because the T1T3 and T1T4 gains
(averaged) were larger in groups that did sight word train-
ing before phonics training (i.e., the sight word + phonics
group and the mixed + mixed group), F(2, 100) = 3.28, p =
.04. More interesting, there was a significant group by gain
interaction, F(2, 100) = 8.35, p < .01. This occurred because
the phonics + sight word group made smaller gains in their
first 8 weeks of training (phonics) than the two groups that
did sight word training, but then made much larger gains
than these groups when they did sight word training in the
last 8 weeks. In fact, the phonics + sight word group effec-
tively “caught up” with the other two groups once they
received sight word training. Thus, sight word training
clearly had a larger effect than phonics training on trained
irregular words regardless of order of training.
Untrained irregular word accuracy. Eight weeks of phonics,
sight word training, and mixed training had very large and
significant training effects on untrained irregular word
reading accuracy. Figure 4 shows that the two groups that
did sight word training made slightly larger gains than the
group that did phonics training. However, the between-
group ANCOVA for the T1T3 data revealed that this differ-
ence was not statistically significant.
Sixteen weeks of phonics, sight word, and mixed train-
ing had significant and very large training effects on
untrained irregular word accuracy. The between-groups
ANCOVA showed that the group that did sight word train-
ing before phonics training made smaller gains than the
phonics + sight word group (p < .05) and mixed + mixed
group (p = .07) . The repeated measures ANCOVA revealed
a significant effect of gain because T1T4 gains were larger
than T1T3 gains, F(1, 100) = 5.09, p = .03. There was also
a significant group by gain interaction, F(2, 100) = 6.10, p
< .01: All three groups made similar gains after the first 8
weeks of training regardless of training type, but in the sec-
ond 8 weeks of training, the groups that did sight word
training made much greater gains than the group that did
phonics training. Of interest, the group that got half a “dose”
of sight words in the last 8 weeks of training (i.e., mixed +
mixed group) made smaller gains than the group who got a
full dose (phonics + sight word group). Considered together,
these findings suggest that untrained irregular words
respond similarly to 8 weeks of phonics and sight word
training, but benefit more when phonics precedes sight
words than vice versa.
Nonword reading accuracy. Eight weeks of phonics, sight
word, and mixed training had moderate to large training
effects on nonword reading accuracy. Figure 4 reveals that
the groups that did phonics training in the first 8 weeks
(phonics + sight word group and mixed + mixed group)
made larger gains than the group that did sight word train-
ing. However, the between-groups ANCOVA for the T1T3
data indicated that this difference was not significant.
Sixteen weeks of phonics and sight word training had a
significant and moderate to large training effect in each
training group. The between-groups ANCOVA revealed no
difference in the T1T4 gains of the three groups. The
repeated measures ANCOVA revealed a significant main
effect of gain because T1T4 gains were larger than T1T3
gains, F(1, 100) = 4.85, p = .04. There was also a significant
interaction between group and gain, F(2, 100) = 26.98, p =
.02, because the group that trained with sight words in the
first 8 weeks of training made smaller T1T3 gains than the
two groups that did phonics training. However, when this
group did phonics training in the second 8-week training
period, they made far larger gains than the other two groups
who did sight word training.
Considered together, the outcomes suggest that phonics
training has a larger effect on nonword reading accuracy
than sight word training regardless of training order.
However, this effect was not large enough to reach statisti-
cal significance in samples of around 30 children with
dyslexia.
Nonword reading fluency. Eight weeks of phonics, sight
word, and mixed training had moderate to large training
effects on nonword reading fluency in the phonics + sight
word group and mixed + mixed group but not the sight
word + phonics group (i.e., this effect was large—0.8—but
not significantly larger than the gain made over the no-
training period). Despite the absence of a true treatment
effect in the sight word + phonics group, Figure 4 suggests
that the three groups made similar gains in the first 8 weeks

McArthur et al. 13
of training. This is supported by the between-group
ANCOVA, which revealed no significant difference
between the gains made by the children who did phonics
training, sight word training, or mixed training.
Sixteen weeks of phonics and sight word training had a
significant and moderate to large training effect on nonword
reading fluency in the phonics + sight word group and
mixed + mixed group but not the sight word + phonics
group (i.e., again, the gains were not larger than those made
over the no-training period). The between-groups ANCOVA
revealed no difference between the groups after 16 weeks of
training. The repeated measures ANCOVA revealed no
main effects of group or gain, or an interaction between the
two. This suggests that nonword reading fluency responds
similarly to phonics training, sight word training, and mixed
training regardless of order of training.
Word reading fluency. Eight weeks of phonics training, sight
word training, and mixed training had large and significant
effects on word reading fluency. Figure 4 indicate that the
groups that did phonics training made slightly greater gains
than the group who did sight word training. However, the
between-group ANCOVA for the T1T3 data revealed no
difference between the training groups.
Sixteen weeks of phonics and sight word training had a
large and significant training effect on word reading flu-
ency. The between-groups ANCOVA revealed that there
was no significant difference between the groups that did
different orders of training. The repeated measures
ANCOVA revealed a significant effect of gain because
T1T4 gains were larger than T1T3 gains, F(1, 100) = 6.95,
p = .01. There was also an interaction between gain and
group, F(2, 100) = 3.71, p = .03. This reflected the fact that
the group that did sight word training in the first 8 weeks
made slightly smaller gains than the two groups that did
phonics training, but then made larger gains than these two
groups when they did phonics training in the second 8-week
training period. Considered together, these outcomes sug-
gest that word reading fluency may respond slightly more to
phonics training than sight word training.
Reading comprehension. Eight weeks of phonics training,
sight word training, and mixed training had large and sig-
nificant effects on reading comprehension. Figure 4 sug-
gests that the three training groups made similar gains in the
first 8 weeks of training. This is supported by the between-
group ANCOVA for the T1T3 data, which revealed no dif-
ference between the groups that did different types of
training.
Sixteen weeks of phonics and sight word training had a
large and significant training effect on reading comprehen-
sion. The between-groups ANCOVA revealed no difference
between the groups after 16 weeks of training. The repeated
measures ANCOVA revealed no main effects of group or
gain, or an interaction between the two. This suggests that
reading comprehension responds similarly to phonics train-
ing, sight word training, and mixed training regardless of
order of training.
Discussion
The aims of this study were to (a) compare phonics training
and sight word training in children with dyslexia and
(b) determine if different orders of sight word and phonics
training have different effects in children with dyslexia. We
allocated 104 children with dyslexia to one of three train-
ing groups. One group was given 8 weeks of phonics
training and then 8 weeks of sight word training. The sec-
ond group was given the reverse. The third group was
given two 8-week phases of mixed training. We measured
the effects of training on the phonics reading route, the
sight word reading route, and both routes combined. Below
we interpret the outcomes in relation to the aims and predic-
tions of this study, and consider the implications of the out-
comes for theory and for clinical practice. We conclude
with possible limitations of this study and ideas for future
research.
The Effect of Sight Word and Phonics Training
in Children With Dyslexia
The first aim of this study was to compare sight word train-
ing and phonics training in children with dyslexia. We pre-
dicted that (a) sight word training would lead to statistically
significant gains in sight word reading measures (trained
and untrained irregular words) that would be larger than
gains made from phonics training, (b) phonics training
would lead to statistically significant gains in phonics read-
ing measures (nonword reading accuracy and fluency) that
would be larger than gains made from sight word training,
and (c) phonics training and sight word training would have
significant effects on measures of reading that taxed both
phonics and sight word reading (word reading fluency and
reading comprehension) that would be similar in size.
The outcomes of this study supported the first predic-
tion. Sight word training had a significant effect on trained
and untrained irregular word reading, and in the case of
trained irregular words, this effect was larger than the effect
of phonics training. However, this was not the case for
untrained irregular words, which improved significantly
from both sight word training and phonics training. This
finding is interesting for at least two reasons. First, it indi-
cates that phonics knowledge can help children learn irreg-
ular words. As outlined above, the generalization from
phonics training to untrained irregular words may be driven
by improvements in children’s ability to use GPC knowl-
edge to sound out parts of an irregular word, which allows
them to guess the whole word from the spoken word

14 Journal of Learning Disabilities XX(X)
representations that they have in their phonological lexicon.
The generalization from trained irregular words to untrained
irregular words may result from improvements in children’s
access to partially learned written words in the orthographic
lexicon (Brunsdon et al., 2005).
The outcomes of this study also supported the second
prediction. Phonics training had a significant effect on non-
word reading accuracy and nonword reading fluency, and
in the case of nonword reading accuracy, this effect was
larger—although not significantly so—than the effect of
sight word training (see the Limitations section for why
this effect may have failed to reach statistical significance).
These results suggest that it is important to teach phonics
explicitly to children with dyslexia because these children
appear to learn GPC rules more readily from phonics train-
ing than from exposure to sight words (i.e., sight word
training). This is consistent with a wealth of studies sup-
porting the role of phonics ability as a key foundation for
reading (see Ehri, Nunes, Stahl, & Willows, 2001, for a
review).
The results of this study also supported the third predic-
tion. Both sight word training and phonics training had sig-
nificant effects on word reading fluency and reading
comprehension. The effect of each type of training was
similar in size for reading comprehension. The effect of
phonics training was slightly larger than the effect of sight
word training for word reading fluency, but this was not
statistically significant. Overall, these findings support the
prediction that both sight word training and phonics train-
ing trigger similar-sized gains in these tests since they
include both irregular words (indexing gains in the sight
word reading route) and regular words (indexing gains in
the phonics reading routes).
An unpredicted finding of this study was that sight word
training, even when restricted to irregular words, can pro-
duce some benefits to reading via the phonics reading route.
This is consistent with research by Fletcher-Flinn and col-
leagues suggesting that phonics rules can be deduced
implicitly from exposure to sight words to some extent
(Fletcher-Flinn & Thompson, 2000; Thompson, Fletcher-
Flinn, & Cottrell, 1999).
Does the Order of Sight Word and Phonics
Training Matter?
The second aim of the current study was to determine if the
order of phonics and sight word training is important for
treating children with dyslexia. This aim addressed the
widespread view that teaching children to read via the GPC
rules (i.e., phonics reading) should allow them to develop
whole-word orthographic representations, which should
allow them to fully decode unfamiliar regular words and
partially decode unfamiliar irregular words. Thus, we pre-
dicted that training phonics before sight words (i.e., the
phonics + sight word group) would lead to greater gains in
accuracy of regular and irregular word reading than training
sight words and then phonics (i.e., the sight word + phonics
group). The results showed that training order had a signifi-
cant effect on untrained irregular word accuracy test. This
was in the predicted direction since the groups that did pho-
nics before sight word training made significantly greater
gains than the group who did sight word training and then
phonics training. It is noteworthy that although this group
had poor phonics-related reading for their age (i.e., their
nonword reading was 1.5 SD below the age-mean), they
nevertheless did have some phonics-related knowledge
(i.e., they did not score at 0 on the nonword reading accu-
racy). Thus, although this study may represent typical order
effects in 7- to 12-year-old children with dyslexia, it may
underrepresent the strength of training phonics then sight
words in children who have little or no phonics skills (e.g.,
beginning readers or children with severe phonological dys-
lexia. This idea is supported by the meta-analysis by
Suggate (2010) outlined in the discussion that reported that
phonics training has a larger effect in younger than older
children in general. It is also noteworthy that the superior
effect of training phonics then sight words on untrained
irregular words also provides some support for the idea that
phonics skills help children read unfamiliar words, even
when those words are irregular (Share, 1995).
The widespread idea that phonics training should pre-
cede sight word instruction makes no predictions about
mixed training. We used this group as a “pragmatic” com-
parison group (the mixed + mixed group) since many teach-
ers and clinicians teach phonics and sight word reading at
the same time. For this reason, it would have been concern-
ing to find that the mixed + mixed group made significantly
smaller gains than the phonics + sight word group and the
sight word + phonics group. Fortunately, for all but one out-
come, the mixed + mixed group made similar gains as the
phonics + sight word group and sight word + phonics group.
The exceptional outcome was trained irregular words, for
which the mixed + mixed group (and sight word + phonics
group) made significantly greater gains in their first 8 weeks
of training than the phonics + sight word group. Thus, there
appears no general disadvantage (or advantage) for training
phonics and sight word reading simultaneously in children
with dyslexia.
Clinical Implications
The results of this study—the first to test both sight word
training and phonics in children with dyslexia—provide at
least three insights into improving the treatment of children
with dyslexia. First, this study revealed that training both
sight word reading and phonics reading has significant and
large training effects on the reading skills of children with
dyslexia (mean Cohen’s d = 1.04). It is important to note

McArthur et al. 15
that these large effects resulted from modest, yet consistent,
changes in children’s raw scores. For example, from T1 to
T4, mean scores increased by around 7 for word reading
fluency, and by around 4 on reading comprehension, with
relatively little variance for this heterogeneous population
of children (around 3–6 scores). These findings match those
of the few previous studies that have tested the effect of a
“pure” phonics program in children with dyslexia using
outcome measures with similar scales to ours. Ford (2009),
Hurford et al. (1994), and Lovett, Steinbach, and Frijters
(2000) found respective effect sizes (ESs) of 0.37, 0.46, and
0.71 for nonword reading accuracy (we found 0.5), which
corresponded to score gains of 2.12, 5.22, and 4.77 (we
found 2.75). Similarly, for nonword reading fluency, Ford
(2009) found an ES of 0.38 (we found 0.6), which corre-
sponded to a score gain of 5.89 (we found 5.89). Two of
these three studies (Ford, 2009; Hurford, 1994) presented
their pure phonics training via computer for up to 2 hr per
week (as did we), and two studies trained children on pho-
nics for less than 3 months (as did we; Ford, 2009; Lovett,
2000). Considered together, these studies, albeit limited in
number so far, suggest that relatively pure phonics training
delivered via computer for up to 2 hr per week for less than
3 months has moderate to large effects on various reading
skills, which reflect small yet reliable gains in children with
dyslexia. In terms of clinical practice, these findings sup-
port the use of computer-based reading training for children
with dyslexia. In an ideal world, this training would be used
as “homework” to complement one-on-one sessions with a
therapist. However, in the real world, where one-on-one
sessions are too expensive for many families, these findings
indicate that computer-based reading training, which is typ-
ically much less expensive than therapy sessions, can be
used to promote small but reliable short-term gains in chil-
dren’s reading.
Second, the outcomes of this study support the idea that
many children with dyslexia need more than just phonics
training. They also need to be trained how to read whole
words by sight. Sight word training is particularly important
for irregular words, which this study revealed are most
effectively learned via explicit training of the words them-
selves rather than via phonics or other sight words. It is
noteworthy that LiteracyPlanet (www.literacyplanet.com)
has now integrated the lists of irregular words that we
developed for this study into their bank of exercises (see
More Sight Words under Sight Words module).
Third, contrary to the beliefs of some reading profes-
sionals, training children to read irregular words will not
impair their ability to read via the letter–sound rules.
Children with dyslexia who did sight training (i.e., with
irregular words) in this study did not regress on the tests of
phonics reading (i.e., nonword reading accuracy and the
nonword reading fluency) or tests of both phonics and sight
word reading (i.e., word reading fluency and reading
comprehension). Thus, we found no evidence that training
children with irregular words harms their ability to read
with the letter–sound rules.
Limitations
There are three potential limitations of this study. The first
relates to our use of trainee’s own gains from Test 1 to test
2 to index gains made over the no-training period (i.e.,
within-subjects control). An alternative approach would
have been to test a group of children with dyslexia at Tests
1, 2, 3, and 4 without giving them any training (i.e., between-
subjects control). We decided against this because between-
subjects control data are (a) not as rigorous as within-subjects
control data because they are collected from different chil-
dren who may experience different (e.g., smaller) effects
across a no-training period; (b) ethically questionable since
they necessitate a delay in children’s treatment by 6 months
during a formative period in their reading, schooling, and
self-esteem; and (c) practically problematic because fami-
lies are much less likely to volunteer for a study where there
is a high chance (1 in 3) that their child may be placed in an
untrained control group. It is noteworthy that our use of
within-subjects no-training control data is a conservative
approach since practice effects on tests of language, read-
ing, and general cognition (e.g., attention, memory, reaction
time) tend to asymptote after a second test session (Bartels,
Wegrzyn, Wiedl, Ackermann, & Ehrenreich, 2010; Collie,
Maruff, Darby, & McStephen, 2003; Kohnen, Nickels, &
Coltheart, 2010). However, it is also noteworthy that for
ethical and practical reasons outlined above, this study did
not use an untreated control group, which is the gold-stan-
dard control for treatment trials. Given the promising results
of the current study, it may now be considered more ethical
to test the reliability of the effects in this study in a random-
ized controlled trial that uses an untrained control group.
A second potential limitation of this study is the size of
the treatment groups. It took our team 2 years to identify
141 children with dyslexia who fulfilled the research crite-
ria. A total of 37 children dropped out of the study. The
remaining 104 children were divided into three groups of
36, 36, and 32 children. Although these sample sizes had
power enough to allow the moderate to very large within-
group effects to reach statistical significance, they may not
have been large enough to allow potential training order and
type effects to reach statistical significance. For example, it
can be seen in Figure 4 that the groups that first did phonics
(phonics + sight word group and mixed + mixed group)
made more rapid gains in nonword reading accuracy than
the group who did sight word training (sight word + phonics
group). However, the latter group made impressive gains
once they later did phonics training. This trend makes sense
theoretically, but it was not statistically significant. Thus, a
study with greater statistical power (i.e., more participants)

16 Journal of Learning Disabilities XX(X)
may reveal that some of our theoretically sensible, yet non-
significant, trends are in fact statistically significant.
A third potential limitation of this study was its use of a
quasi-randomized allocation procedure, which resulted
from an unanticipated delay in the development of the sight
word program that was beyond our control. For reasons out-
lined above, there was no evidence that this allocation pro-
cedure biased the outcomes of this study. However, in future
studies, we will use minimization or random allocation to
allocate children to treatment groups.
Summary
In sum, the outcomes of this study suggest that 16 weeks of
phonics and sight word training has large or very large effects
on the phonics and sight word reading of children with dys-
lexia. Furthermore, 8 weeks of phonics, sight word, or mixed
training has moderate to very large effects in these children.
Of particular interest, training phonics reading before sight
word reading appears to have a larger effect on reading
untrained irregular words than the reverse order of training.
These findings represent an advance in treatment of dyslexia
since mean effect in this study was larger (Cohen’s d = 1.04)
than the mean small to moderate effect found by previous
studies. This superior effect size supports the idea that chil-
dren with dyslexia need treatment for sight word reading and
phonics reading, and not just phonics reading alone.
Acknowledgments
We would like to thank all the children and parents who donated
their time and effort to this research. We would like to thank Max
Coltheart for his guidance on many issues related to this article.
We would like to thank Shane Davis (MacroWorks and
LiteracyPlanet) and John Dyson (Lexia) for modifying their train-
ing programs for this trial. And we would like to thank our review-
ers for their valuable contributions to the development of this
article.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
Funding
The author(s) disclosed receipt of the following financial support
for the research, authorship, and/or publication of this article: This
research was funded by NHMRC Project 488518 and ARC
DP0879556.
References
Bartels, C., Wegrzyn, M., Wiedl, A., Ackermann, V., &
Ehrenreich, H. (2010). Practice effects in healthy adults: A
longitudinal study on frequent repetitive cognitive testing.
BMC Neuroscience, 11, 118–130.
Bonate, P. L. (2000) Analysis of pretest-posttest designs. Boca
Raton, FL: Chapman and Hall.
Broom, Y. M., & Doctor, E. A. (1995). Developmental surface
dyslexia: A case study of the efficacy of a remediation pro-
gramme. Cognitive Neuropsychology, 12, 69–110.
Brunsdon, R., Coltheart, M., & Nickels, L. (2005). Treatment of
irregular word spelling in developmental surface dysgraphia.
Cognitive Neuropsychology, 22, 213–251.
Brunsdon, R., Hannan, T., Nickels, L., & Coltheart, M. (2002).
Successful treatment of sublexical reading deficits in a
child with dyslexia of the mixed type. Neuropsychological
Rehabilitation, 12, 199–229.
Carroll, J. M., Maughan, B., Goodman, R., & Meltzer, H. (2005).
Literacy difficulties and psychiatric disorders: Evidence for
comorbidity. Journal of Child Psychology and Psychiatry, 46,
524–532.
Castles, A., & Coltheart, M. (1993). Varieties of developmental
dyslexia. Cognition, 47, 149–180.
Castles, A., Coltheart, M., Larsen, L., Jones, P., Saunders, S.,
& McArthur, G. (2009). Assessing the basic components of
reading: A revision of the Castles and Coltheart test with new
norms. Australian Journal of Learning Difficulties, 14, 67–88.
Collie, A., Maruff, P., Darby, D., & McStephen, M. (2003). The
effects of practice on the cognitive test performance of neu-
rologically normal individuals assessed at brief test–retest
intervals. Journal of the International Neuropsychological
Society, 9, 419–428.
Coltheart, M., Rastle, K., Perry, C., Langdon, R., & Ziegler, J.
(2001). A dual route cascaded model of visual word recogni-
tion and reading aloud. Psychological Review, 108, 204–256.
Connor, C., Piasta, S., Fishman, B., Glasney, S., Schatschneider,
C., Crowe, E., & Morrison, F. J. (2009). Individualizing
student instruction precisely: Effects of child by instruction
interactions on first graders’ literacy development. Child
Development, 80, 77–100.
Cribbie, R. A., & Jamieson, J. (2004). Decreases in post-
test variance and the measurement of change. Methods of
Psychological Research Online, 9, 37–55.
Ehri, L. C., Nunes, S. R., Stahl, S. A., & Willows, D. M. (2001).
Systematic phonics instruction helps students learn to read-
ing: Evidence from the national reading panel’s meta-analysis.
Review of Educational Research, 71, 393–447.
Ellis, A. W., & Young, A. W. (1988). Human cognitive neuropsy-
chology. Hove, UK: Lawrence Erlbaum.
Fletcher-Flinn, C. M., & Thompson, G. B. (2000). Learning to
read with underdeveloped phonemic awareness but lexical-
ized phonological recoding: A case study of a 3-year-old.
Cognition, 74, 177–208.
Ford, C. (2009). The effect of the backward-chaining method of
decoding with computer-assisted instruction on the reading
skills of struggling adolescent readers (Unpublished thesis).
Northern Illinois University, DeKalb.
Goulandris, N. K., & Snowling, M. (1991). Visual memory defi-
cits: A plausible cause of developmental dyslexia? Evidence
from a single case study. Cognitive Neuropsychology, 8,
127–154.
Gresham, F. M., & Vellutino, F. R. (2010). What is the role of
intelligence in the identification of specific learning disabilities?

McArthur et al. 17
Issues and clarifications. Learning Disabilities Research and
Practice, 25, 194–206.
Howell, D. (2010). Statistical methods of psychology. Belmont,
CA: Cengage Wadsworth.
Hulme, C., & Snowling, M. (2009). Developmental disorders
of language, learning and cognition. London, UK: Wiley-
Blackwell.
Hurford, D., Johnston, M., Nepote, P., Hampton, S., Moore, S.,
Neal, J., . . . Huffman, D. (1994). Early identification and
remediation of phonological processing deficits in first-grade
children at risk for reading disabilities. Journal of Learning
Disabilities, 27, 647–659.
Kaufman, A., & Kaufman, N. (2004). Kaufman Brief Intelligence
Test–2. Circle Pines, MN: AGS.
Kohnen, S., Nickels, L., & Coltheart, M. (2010). Skill generalisa-
tion in teaching spelling to children with learning difficulties.
Australian Journal of Learning Difficulties, 15, 115–129.
Lavori, P. (1992). Clinical trials in psychiatry: Should protocol
deviation censor patient data? Neuropsychopharmacology, 6,
39–63.
Lovett, M., Steinbach, K., & Frijters, J. (2000). Remediating the
core deficits of developmental reading disability: A double-
deficit perspective. Journal of Learning Disabilities, 33,
334–358.
McArthur, G., Jones, K., Anandakumar, T., Castles, A., Larsen, L., &
Coltheart, M. (2013). A Test of Everyday Reading Comprehension
(TERC). Australian Journal of Learning Difficulties, 18, 35–85.
McArthur, G. M., Ellis, D., Atkinson, C., & Coltheart, M. (2008).
Auditory processing deficits in children with reading and lan-
guage impairments: Can they (and should they) be treated?
Cognition, 107, 946–977.
McArthur, G. M., Eve, P., Jones, K., Banales, E., Kohnen, S.,
Anandakumar, T., . . . Castles, A. (2012). Phonics training
for English-speaking poor readers. Cochrane Database of
Systematic Reviews, 12, CD009115.
Nation, K. (2009). Form–meaning links in the development of
visual word recognition. Philosophical Transcripts of the
Royal Society of Biological Sciences, 364, 3665–3674.
Peduzzi, P., Henderson, W., Hartigan, P., & Lavori, P. (2002).
Analysis of randomized control trials. Epidemiologic Reviews,
24, 26–38.
Perry, C., Ziegler, J., & Zorzi, M. (2007). Nested incremental mod-
eling in the development of computational theories: The CDP
model of reading aloud. Psychological Review, 114, 273–315.
Plaut, D. C., McClelland, J. L., Seidenberg, M. S., & Patterson,
K. (1996). Understanding normal and impaired word read-
ing: Computational principles in quasi-regular domains.
Psychological Review, 103, 56–115.
Rogosa, D. R., & Willett, J. B. (1983). Demonstrating the reli-
ability of the difference score in the measurement of change.
Journal of Educational Measurement, 20, 335–343.
Rose, J. (2006). Independent review of the teaching of early read-
ing: Final report. Nottingham, UK: DfES.
Rowse, H., & Wilshire, C. E. (2007). Comparison of phonological
and whole word treatments for two contrasting cases of devel-
opmental dyslexia. Cognitive Neuropsychology, 24, 817–842.
Share, D. (1995). Phonological recoding and self-teaching: Sine
qua non of reading acquisition. Cognition, 55, 151–218.
Siegel, L. S. (1989). IQ is irrelevant to the definition of learning
disabilities. Journal of Learning Disabilities, 22, 469–478,
486.
Suggate, S. P. (2010). Why what we teach depends on when:
Grade and reading intervention modality moderate effect size.
Developmental Psychology, 46, 1556–1579.
Thompson, G., Fletcher-Flinn, C., & Cottrell, D. (1999). Learning
correspondences between letters and phonemes without
explicit instruction. Applied Psycholinguistics, 20, 21–50.
Torgesen, J., Wagner, R., & Rashotte, C. (1999). Test of Word
Reading Efficiency. Austin, TX: PRO-ED.
Wang, H.-C., Castles, A., Nickels, L., & Nation, K. (2011).
Context effects on orthographic learning of regular and irreg-
ular words. Journal of Experimental Child Psychology, 109,
39–57.
Wyse, D., & Goswami, U. (2008). Synthetic phonics and the
teaching of reading. British Educational Research Journal,
34, 691–710.