Content uploaded by Sandra Lyn Martin-Chang
Author content
All content in this area was uploaded by Sandra Lyn Martin-Chang on Sep 14, 2018
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=hssr20
Scientific Studies of Reading
ISSN: 1088-8438 (Print) 1532-799X (Online) Journal homepage: http://www.tandfonline.com/loi/hssr20
Differential Effects of Context and Feedback
on Orthographic Learning: How Good Is Good
Enough?
Sandra Martin-Chang, Gene Ouellette & Linda Bond
To cite this article: Sandra Martin-Chang, Gene Ouellette & Linda Bond (2017) Differential Effects
of Context and Feedback on Orthographic Learning: How Good Is Good Enough?, Scientific
Studies of Reading, 21:1, 17-30, DOI: 10.1080/10888438.2016.1263993
To link to this article: https://doi.org/10.1080/10888438.2016.1263993
Published online: 12 Dec 2016.
Submit your article to this journal
Article views: 258
View Crossmark data
Citing articles: 1 View citing articles
Differential Effects of Context and Feedback on Orthographic
Learning: How Good Is Good Enough?
Sandra Martin-Chang
a
, Gene Ouellette
b
, and Linda Bond
a
a
Concordia University;
b
Mount Allison University
ABSTRACT
In this study, students in Grade 2 read different sets of words under 4 experi-
mental training conditions (context/feedback, isolation/feedback, context/
no-feedback, isolation/no-feedback). Training took place over 10 trials, followed
by a spelling test and a delayed reading posttest. Reading in context boosted
reading accuracy initially; in contrast, the external support garnered from feed-
back resulted in heightened reading accuracy throughout training, as well as
1weeklater.Differentpatternswerenotedinspellingtransfer;first,therewasno
effect of reading feedback on spelling at posttest. Second, the highest spelling
scores were observed when children practiced reading words in isolation versus
in context. In sum, providing feedback and/or context helps children read words
accurately, which in turn seems to create orthographic representations that are
“good enough”to support reading accuracy. However, reading in isolation
seems to produce orthographic representations that are higher in quality, and
therefore better able to support precise spelling.
Becoming literate involves learning to read and spell a multitude of words, yet these two abilities are not
equally challenging (Conrad, 2008;Ouellette,2010). For example, although incomplete—or lower quality
—orthographic representations are often “good enough”for accurate reading, only high-quality
representations allow for accurate spelling (Martin-Chang, Ouellette, & Madden, 2014). According to
the self-teaching hypothesis (Share, 2004), reading independently is ideal for forming high-quality
orthographic representations, a process referred to as orthographic learning. Self-teaching occurs when
children recode graphemes into phonemes as they encounter words that are unfamiliar to them in print
(Share, 1995). If grapheme-by-grapheme decoding is central to orthographic learning, then reading that
relies heavily on these bottom-up processes would be expected to result in higher quality orthographic
representations compared to decoding that is assisted by context or feedback. Alternatively, decoding
that is aided by context or feedback may create additional opportunities to pronounce words correctly.
If binding orthography (print), phonology (pronunciation), and semantics (meaning) is central to
orthographic learning, then the additional opportunities to read the words correctly created by top-
down (context) or external (feedback) support might produce better orthographic learning over time
(Martin-Chang & Levy, 2005). Given the two competing hypotheses, the goal of the current study was to
elucidate how contextual facilitation and corrective feedback affect orthographic learning, as measured in
two ways: first, by children’s ongoing reading performance during training, and again after a 1-week
delay, and second, by their spelling performance at the end of training.
Orthographic development has been defined as a multidimensional construct combining both
word-specific orthography and general orthographic knowledge (Conrad, Harris, & Williams, 2013).
Word-specific orthographic representations are the stored spellings of items on a word-by-word
basis (e.g., remembering which word has the doubled consonant: supper vs. super). In contrast,
CONTACT Sandra Martin-Chang smartinc@education.concordia.ca Department of Education, Concordia University,
1455 de Maisonneuve Boulevard West, Montreal, Quebec H3G 1M8, Canada.
© 2017 Society for the Scientific Study of Reading
SCIENTIFIC STUDIES OF READING
2017, VOL. 21, NO. 1, 17–30
http://dx.doi.org/10.1080/10888438.2016.1263993
general orthographic knowledge is the broader understanding and ability to implement the patterns
of written language (e.g., consonants are doubled after closed syllables [supper], but not after open
syllables [super]). Both types of knowledge drive fluent reading (Castles & Nation, 2008; Shahar-
Yames & Share, 2008). Specifically, as children develop a large bank of word-specific orthographic
representations, the balance of labor gradually shifts from having to ploddingly decode most words
to being able to read the bulk of text “on sight”; Ehri, 2014). At the same time, greater general
orthographic knowledge increases the likelihood that words encountered for the first time in print
will be pronounced accurately (Ehri, 2005; Verhoeven & Perfetti, 2011). In the present study, we
focus on the storage of word-specific orthographic representations, which is also referred to as
orthographic learning.
The role of context on orthographic learning
Share (1995)noted that “self-teaching depends on . . . letter-sound knowledge, some minimal
phonological sensitivity, and the ability to utilize contextual information [meaning] to determine
exact word pronunciations on the basis of partial decodings (p. 160). Therefore, a child who
encounters the word neighbor in isolation, despite understanding the spoken word, might struggle
when reading it in print. He or she may achieve partial decoding (e.g., /n//e//g//b//r/) but fail to
pronounce the word accurately, thereby forfeiting the opportunity to develop a full orthographic
representation for that word (Share, 1995; Wang, Castles, Nickels, & Nation, 2011). However, in
context, readers can draw on the surrounding semantics and syntax to support decoding that
improves word recognition. For example, when given the sentence “My neighbor has the best garden
on our street,”the young reader can often arrive at the correct word—not by guessing, because the
vast majority of text is unpredictable (Gough, Alford, & Holley-Wilcox, 1981), but by using partial
decodings, prior vocabulary knowledge, and the semantic and syntactic cues from the surrounding
context (Tunmer & Chapman, 2002). However, this raises an important theoretical question: If the
amount of independent effort spent decoding (i.e., self-teaching) leads to orthographic learning,
what happens to orthographic learning when decoding is supplemented with support from context?
There are two contrasting views regarding the facilitative factors associated with reading in
context. It is widely agreed that context acts as a scaffold to identify words that are outside of a
child’s zone of proximal development (Archer & Bryant, 2001; Nicholson, 1991). It is less clear
whether reading in context results in word-specific orthographic learning. Some authors argue that it
does not; rather, they posit that children must be primarily focused on bottom-up processes in order
to create lasting word representations in memory. From this vantage point, top-down assistance may
help with short-term word identification, but it is counterproductive in terms of orthographic
learning and is therefore detrimental to long-term reading development (Landi, 2013; Nemko,
1984). This view has been supported by several studies showing that children can read more
accurately in context initially but that those reading improvements do not result in long-term
gains in either reading (Landi, 2013; Nemko, 1984) or spelling (Cunningham, 2006; Nation,
Angell, & Castles, 2007; Ricketts, Bishop, Pimperton, & Nation, 2011).
In contrast, researchers in favor of the long-term benefits of contextual reading point to the
proposal that lexical quality includes phonology, orthography, and semantics and that when words
are read more correctly—as they tend to be in context—it presents opportunities for these three
lexical identities to merge (Martin-Chang & Levy, 2005,2006; Nation & Snowling, 1998; Tunmer &
Chapman, 2002). For example, Ehri (2005,2014) suggested that reading aloud greatly benefits the
retention of semantics and orthography because oral reading promotes the phonological component
necessary for building orthographic representations. Support for this hypothesis comes from training
studies that report greater benefits for reading accuracy (Martin-Chang & Levy, 2005,2006; Martin-
Chang, Levy, & O’Neil, 2007) and orthographic recognition (Cunningham, 2006; Wang et al., 2011)
when difficult and/or irregular words were presented in context.
18 S. MARTIN-CHANG ET AL.
In sum, the results of studies examining the effects of context on orthographic learning have been far
from unequivocal. However, there are four methodological considerations that might help explain the
discrepant findings. First, in the literature just discussed, the type of texts differed considerably across
studies; context ranged from predictable sentences (e.g., Landi, Perfetti, Bolger, Dunlap, & Foorman,
2006) to short paragraphs (e.g., Cunningham, 2006;Wangetal.,2011)tolengthypassages(e.g.,Martin-
Chang et al., 2007). Second, the target words used within the experiments might have also impacted the
findings. Stimuli sets tended to consist of a small number of items, ranging from nonwords (e.g.,
Cunningham, Perry, Stanovich, & Share, 2002) to common words contained in the children’soral
vocabularies (e.g., Martin-Chang et al., 2007), and from being easily decodable (e.g., simple consonant-
vowel-consonant words or consonant-vowel-consonant-e words; Bowey & Muller, 2005) to irregular
words (e.g., Wang et al., 2011). Third, studies differed with respect to the variables of interest, with the
focus being on either word reading accuracy (e.g., Martin-Chang et al., 2007) or spelling production (e.g.,
Cunningham, 2006), making cross-study comparisons challenging. Fourth, several of the studies just
cited (Landi et al., 2006; Martin-Chang & Levy, 2005,2006; Martin-Chang et al., 2007) provided
corrective feedback when the children read inaccurately, making it difficult to compare these studies
to others that have employed true self-teaching paradigms devoid ofany external feedback (e.g., Bowey &
Muller, 2005; Ouellette & Fraser, 2009).
The role of feedback on orthographic learning
In the last decade and a half, much progress has been made in understanding the nature of self-teaching;
however, surprisingly little work has been done comparing the effects of self-teaching (which, as the
name implies, is done without the help of an outside “teacher”)tolearningthattakesplacewith
corrective feedback. As argued by Ehri (2014), seeing a word in writing and hearing its pronunciation
binds orthography to phonology and, over time, unifies these two aspects of word identity. Therefore,
feedback might support orthographic learning by modeling the pronunciation (and granting access to
spoken vocabulary) after initial decoding attempts have failed. If this is the case, then hearing the
pronunciation while still attending to the print might put the child in a better position to phonologically
recode the word without assistance on subsequent trials. Although there has been limited research into
the role of feedback on reading development, there is some evidence that supports this view. For
example, Spaai, Ellerman, and Reitsma (1991) found that students in Grade 1 were able to read more
words correctly after receiving feedback in isolation compared to when feedback was not provided. It has
also been noted that feedback received immediately after a mistake helps in the acquisition and
maintenance of word reading accuracy (Barbetta, Heward, Bradley, & Miller, 1994). Important to
note, in both of the aforementioned studies, the feedback followed a child’s original attempt to read a
word him- or herself. Studies that have examined simple exposures to orthography and pronunciation
pairings, as often provided through software storybook reading, or more recently via text-to-speech
programs, have reported no such beneficial effects (e.g., Staels & Van den Broeck, 2014).
Even when feedback comes after the reading attempts have been made, it has been suggested that
children may reduce the time or effort they spend actively recoding when feedback is available. Indeed,
Landi and colleagues (Landi, 2013; Landi et al., 2006) have predicted that students will be less likely to
spend cognitive resources during initial decoding attempts if they know that the correct pronunciation
will be provided. Phonological recoding has been purported to be a crucial step in self-teaching and
orthographic learning (Share, 2004); if receiving assistance reduces recoding efforts, then the practice
of providing individualized feedback could indeed be detrimental to orthographic learning.
Current investigation
To recapitulate, the body of work exploring self-teaching has employed a variety of methods
resulting in many rich insights. However, the variation within the literature has made it difficult
to compare across studies. With these considerations in mind, the goal of the current investigation
SCIENTIFIC STUDIES OF READING 19
was to examine the interplay between two sources of support—context and feedback—on ortho-
graphic learning as reflected in both reading and spelling accuracy.
Method
Participants
Twenty-five participants in Grade 2 were recruited from an ethnically diverse, suburban elementary
school in eastern Canada. Participants were enrolled in a publically funded bilingual program, which
delivered reading and spelling instruction in French and English on alternating days. All students
spoke English as one of their primary languages. One participant was removed from the sample due
to prolonged absences, and one student was removed due to low scores obtained during screening
tests. The final sample contained 23 participants (nine girls, 14 boys; Mage = 7 years 10 months),
including children from Native (n= 6), African Canadian (n= 2), and Caucasian (n= 15) back-
grounds. Thirty percent of the participants were eligible for the free breakfast/lunch program. In
addition to English, many children also spoke French, although to varying degrees.
All participants were screened with standardized tests of reading, vocabulary, and phonological
awareness, as described next. As indicated in Table 1, standardized scores on these measures
indicated that all children were performing at, or near, age-appropriate levels.
Research design
This experiment manipulated whether reading practice took place in context or in isolation, and
whether feedback was given or withheld. A 2 (context vs. isolation) ×2 (feedback vs. no feedback)
fully crossed within participant design was employed where every participant was trained in all four
conditions. The order of conditions was counterbalanced across participants, and the assignment of
training stimuli (word sets) was counterbalanced across conditions.
The first dependent variable was reading accuracy. It was measured at pretest, during training (over 10
trials), and again after a 7-day delay (delayed posttest). The second dependent variable was spelling accuracy.
It was measured at two time points: at pretest and 1 day after training (posttest). Children were given praise
and encouragement that was not contingent on task requirements during all tasks and conditions.
Materials
Screening measures
To ensure that participants were performing within age-expected levels for literacy and language
development, they were screened for word reading, receptive vocabulary, and phonological awareness.
Accuracy of reading was tested with a subtest of the Wide Range Achievement Test–Third Edition
(Wilkinson, 1993), which involved reading 42 words in isolation. The list begins with single-syllable
high-frequency words and increases in difficulty to low-frequency multisyllabic words. Testing was
discontinued after 10 consecutive errors. The Wide Range Achievement Test–Third Edition has an
internal consistency reliability of α= .89.
Table 1. Standardized scores on all screening measures.
Variable MScore SD Range
PPVT–4 104.54 10.56 86–124
CTOPP 9.04
a
2.44 6–14
WRAT3 99.96 11.78 86–105
Note. PPVT–4 = Peabody Picture Vocabulary Test, Fourth Edition; CTOPP =
Comprehensive Test of Phonological Processing; WRAT3 = Wide Range
Achievement Test–Third Edition
a
Standard score mean of 10.
20 S. MARTIN-CHANG ET AL.
Receptive vocabulary was assessed with the Peabody Picture Vocabulary Test, Fourth Edition
(Dunn & Dunn, 2006). Participants were shown a page containing four different pictures and asked
to identify the picture that matched a word spoken by the examiner. This test has very high internal
consistency (α= .91). The Elision subtest of the Comprehensive Test of Phonological Processing
(Wagner, Torgesen, Rashotte, & Pearson, 2013) was completed to assess phonological awareness. In
this task the researcher asks the participant to repeat spoken words after removing specific segments
(e.g., say “bat”without the /b/). The task included 19 words in total. Three consecutive errors resulted
in termination of the test. The internal consistency reliability for this age group is very good (α= .89).
Training stimuli
Five word sets were adapted from Martin-Chang and Levy (2005), each comprised 25 words (see
Appendix A). Target words were chosen to vary with respect to how challenging they were to read and
write for children in Grade 2. To track learning in both reading and spelling, banks of words were
needed that were neither at ceiling nor at floor prior to training. Table 2 shows that we were able to
achieve this objective; when the reading and spelling tasks were considered separately, it became clear
that there were no apparent differences at pretest between the conditions, Fs(1, 22) < 1.6, ps>.23. Five
local second-grade teachers also vetted the materials and indicated that they felt that the majority of the
words (88.5%) would be contained in the oral vocabularies of children in Grade 2.
Context stories were then created around each word set. They ranged between 130 and 150 words,
but only the 25 items from the corresponding word set were marked as targets. All additional words
were considered to be “fillers”that were needed to create a complete story. The stories were written
to approximate texts that children might encounter in their day-to-day lives. In this respect the
context used in the present study deviated from those conducted previously that used shorter texts
with only one target word repeated several times per story. Here, all 25 target words appeared twice
during each training story/isolated word list, and all stories/lists were read five times.
Procedure
Screening and pretests
All students were individually screened for reading, vocabulary and phonological awareness with
standardized tests administered over two sessions. Spelling was then pretested over a 5-day period,
with one set of 25 words tested each day for 5 days. The words were dictated to small groups. Each
participant sat at his or her own table and was unable to view the spellings of the other children. The
examiner said the word aloud, provided a sentence, and repeated the word (e.g., “Both—I like both
chocolate and vanilla ice cream—both”). Spellings were scored as correct or incorrect. Following a 2-
day delay after the spelling pretests, reading accuracy was assessed for these same word sets. This was
completed according to a similar schedule, with one word list read on each of 5 consecutive days.
Students were tested individually. They were simply asked to read the words presented in list format;
no feedback was given. Reading accuracy was scored as correct or incorrect for each word.
Table 2. Pretest performance on reading and spelling (number correct).
Feedback No Feedback
Pretest Context Isolation Context Isolation
Reading 13 12.74 12.43 12.96
(7.00) (6.67) (7.39) (6.78)
Spelling 4.91 5.00 4.47 5.44
(4.02) (3.77) (3.87) (3.83)
Note. Standard deviations are shown in parentheses. All words were read and spelled in isolation,
without feedback at pretest; the column heads refer to assignment of words in subsequent
experimental conditions.
SCIENTIFIC STUDIES OF READING 21
Training
The training schedule took place over a 2-month period (see Appendix B). Training was conducted
over four separate 5-day blocks; one condition (context/feedback, context/no feedback, isolation/
feedback, and isolation/no feedback) was run during each block. Each set of words was read 10
times, distributed over 4 days, with spelling posttesting on the 5th day. Two trials (one story or one
list) were completed on Days 1, 2, and 3 and four trials (two stories or lists) on Day 4. Once again, the
design was fully counterbalanced with respect to the order and word sets assigned to each condition.
Context conditions: With and without feedback. Students read passages independently in the
context condition. This included the 25 target words as well as the filler words; however, only the
target words were scored for accuracy. In the feedback conditions, children received whole-word
corrective feedback on the target words when they made errors or paused for longer than 2 s.
Feedback was offered only after a child was unable to decode a given word. This was done to ensure
that feedback was not replacing self-teaching attempts (see Staels & Van den Broeck, 2014).
Children in the no-feedback condition did not receive input from the researcher; rather, they
were prompted to “read this as if you were alone.”If the child paused for more than 2 s while trying
to decode a target word, the research encouraged the child to “keep going.”If the child read the word
incorrectly, the researcher recorded it as an error. The children’s reading was audiotaped for scoring
purposes in all conditions.
Isolation conditions: With and without feedback. Words trained in isolation were presented one at
atime on a computer screen in Times New Roman 40 font to mimic flash cards. All words appeared
in the center of the screen for 2 s each and were preceded by a fixation point. Whole-word feedback
was provided to participants following errors or nonresponses in the feedback condition. Once
again, feedback was offered only after a child was unable to decode a given word (see Staels & Van
den Broeck, 2014). Participants in the no-feedback condition were prompted to “read this as if you
were alone.”All sessions were audio-recorded to ensure scoring accuracy. Identical scoring proce-
dures were used in both context and isolation conditions.
Posttesting
On the 5th day of each block, target words were posttested for spelling accuracy. Words were written
individually (not in context) regardless of how they had been trained. The word set not assigned to
any experimental condition (i.e., the list pretested but not trained) was assessed the week following
the completion of all training. For reading accuracy a delayed posttest was administered 7 days after
the final training day with the same materials that were used for training. No feedback was given
during any posttests, regardless of the training condition.
Results
Orthographic learning as reflected in word reading accuracy
Figure 1 displays the mean percentage of words read correctly throughout training. A 2 (context vs.
isolation) ×2 (feedback vs. no-feedback) ×10 (Trial: 1–10), repeated measures analysis of variance
(ANOVA) was conducted to evaluate reading accuracy over the duration of training. The assumption
of sphericity was violated for the main effect of trial as assessed by Mauchly’s test, χ
2
(44) = 255.97,
p<.001; therefore the Greenhouse-Geisser correction was applied where appropriate. All three
main effects were significant: The main effect of context was significant with a medium effect size,
F(1, 22) = 9.76, MSE = 26.74, p= .005, η
p2
= .31; the main effect of feedback was significant with a large
effect size, F(1, 22) = 42.28, MSE = 109.94, p< .001, η
p2
= .66; and the main effect of trial was significant
22 S. MARTIN-CHANG ET AL.
with a large effect size, F(1.47, 32.42) = 49.22, MSE = 42.81, p< .001, η
p2
= .69. In terms of the two-way
interactions, the Context ×Trial interaction was significant, F(9, 198) = 3.29, MSE = 2.26, p< .001,
η
p2
= .13, and the Feedback ×Trial interaction was significant, F(9, 198) = 23.91, MSE = 3.26, p< .001,
η
p2
= .52. However, the Context ×Feedback interaction was not significant, p= .98. The ANOVA also
revealed a statistically significant three-way interaction between context, feedback, and trial, F(3.78,
83.23) = 3.35, MSE = 5.65, p= .015, η
p2
= .13.
To better understand the three-way interaction, the effects of context and trial were considered at both
levels of feedback. Beginning with words read in the no-feedback condition only, a 2 (context vs.
isolation) ×10 (trial) repeated measures ANOVA revealed that the main effect of context was approach-
ing significance, F(1, 22) = 4.10, MSE =32.31,p=.055,η
p2
= .16, and that the main effect of trial was
significant, F(2.15, 47.42) = 19.2, MSE = 2.56, p<.001,η
p2
= .13. It is noteworthy that the effect sizes of
the main effects of context and trial were similar, which indicates a facilitating effect of context and
increased accuracy across trials. The Context ×Trial interaction was not significant (p=.751).
This procedure was then repeated for words read with feedback only. A repeated measures
ANOVA indicated that both the main effects of context, F(1, 22) = 4.83, MSE = 26.59, p= .039,
η
p2
= .18, and trial, F(1.5, 33.48) = 48.49, MSE = 7.71, p< .001, η
p2
= .69, were significant. However
in the feedback condition, the Context ×Trial interaction was also significant, F(35.52, 6.98) = 5.09,
MSE = 6.98, p< .001, η
p2
= .19. To follow up this significant two-way interaction observed during
the feedback condition, we considered the simple main effect of context at the beginning (Trial 1),
middle (Trial 5), and end (Trial 10) of training. This analysis revealed context was significantly
different from isolation only during Trial 1, F(1, 22) = 13.25, p= .001, η
p2
= .38. The simple main
effects of context were not significant at Trial 5 (p= .35) or Trial 10 (p= .74). This pattern of results
may reflect a ceiling effect in the later trials.
As a final step in understanding the three-way interaction, the effects of feedback and trial were
considered at both levels of context. Beginning with words read in isolation, a 2 (feedback vs. no
feedback) ×10 (trial) repeated measures ANOVA revealed that the main effects of feedback,
F(1, 22) = 33.83, MSE = 68.96, p< .001, η
p2
= .61, and trial, F(9, 198) = 44.7, MSE = 5.02,
p< .001, η
p2
= .67, were both significant. The Feedback ×Trial interaction was also significant,
F(9, 198) = 20.76, MSE = 3.16, p< .001, η
p2
= .49. To follow up this significant two-way interaction,
we considered the simple main effect of feedback at the beginning (Trial 1), middle (Trial 5), and
end (Trial 10) of training. This analysis revealed that the differences between the feedback and no-
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
12345678910
Number of words read accurately
Context Feedback
Context No Feedback
Isolation Feedback
Isolation No Feedbac
k
Trials
Figure 1. Mean percentage of words read correctly as a function of training condition over 10 trials.
SCIENTIFIC STUDIES OF READING 23
feedback conditions differed at Trial 5, F(1, 22) = 33.06, p< .001, η
p2
= .60, and Trial 10,
F(1, 22) = 2.38, p< . 001, η
p2
= .59, but not at Trial 1 (p= .23).
This procedure was then repeated for words read in context. A repeated measures ANOVA
indicated that both the main effects of feedback, F(1, 22) = 31.65, MSE = 73.15, p< .001, η
p2
= .59,
and trial, F(9, 198) = 30.12, MSE = 4.2, p< .001, η
p2
= .58, were significant. The Feedback ×Trial
interaction was also significant, F(9, 198) = 8.2, MSE = 2.47, p< .001, η
p2
= .27. To follow up this
significant two-way interaction observed during the context condition, we once again considered the
simple main effects at the beginning (Trial 1), middle (Trial 5) and end (Trial 10) of training. Again,
no differences were found between feedback and no feedback during the first trial of training
(p= .14), but significant differences were found at Trial 5, F(1, 22) = 31.92, p< .001, η
p2
= .59,
and Trial 10, F(1, 22) = 24.31, p< .001, η
p2
= .52.
Finally, the delayed reading posttest data were analyzed in a 2 (context vs. isolation) ×2
(feedback vs. no-feedback) repeated measures ANOVA. This analysis confirmed what is apparent
in Table 3. There was a main facilitating effect of feedback on reading accuracy with a large effect
size, F(1, 22) = 34.23, MSE = 729.14, p< .001, η
p2
= .61; the prior facilitating effect of context was
reduced to a trend, F(1, 22) = 3.18, MSE = 10.45, p= .08, η
p2
= .13; and the Context ×Feedback
interaction was not significant, F(1, 22) = 1.60, MSE = 21.30, p= .22, η
p2
= .07. Pairwise
comparisons confirmed that although performance within the two feedback conditions was
equivalent (p= 1.0), accuracy in both feedback conditions was significantly greater than in the
no-feedback conditions (ps < .001); the two no-feedback conditions (context and isolation) did not
differ from one another (p= .60) at posttest.
Orthographic learning as reflected in spelling accuracy
Figure 2 depicts the spelling performance at pretest and posttest, as a function of training condition (i.e.,
reading practice). A 2 (context vs. isolation) ×2 (feedback vs. no-feedback) ×2 (pretest vs. posttest)
repeated measures ANOVA was conducted. Two of the three main effects were significant with medium
Table 3. Number of words read correctly during delayed posttesting.
Feedback No Feedback
Context Isolation Context Isolation
23.80 22.65 18.98 18.67
(1.83) (2.37) (5.91) (6.37)
Note. Standard deviations are shown in parentheses.
0
2
4
6
8
10
12
Context
Feedback
Context
No Feedback
Isolation
Feedback
Isolation
No Feedback
Control
Pre-test
Post-test
Words spelled correctly
Figure 2. Number of words spelled correctly at pre- and posttesting as a function of reading practice.
24 S. MARTIN-CHANG ET AL.
to large effect sizes, context F(1, 22) = 14.03, MSE =8.714,p=.001,η
p2
= .39, and test time F(1,
22) = 70.87, MSE =8.55,p< .001, η
p2
= .76. These main effects reflect increased spelling accuracy for
words read in isolation relative to the context conditions and improved performance from pretest to
posttest. There was no significant effect of feedback, F(1, 22) = .33, MSE =6.49,p=.57,η
p2
= .01.
However, there was a statistically significant two-way Context ×Time interaction, F(1, 22) = 20.41,
MSE =2.77,p< .001, η
p2
= .48, indicating that the superior performance associated with reading words in
isolation was evident only at posttest (and not at pretest). No other interactions were significant. Post hoc
comparisons with Bonferroni corrections in place confirmed that children spelled more words correctly
after participating in the two isolated-word training conditions compared to the context conditions and
the control condition (i.e., “no reading”practice conditions, all ps < .02). No other pairwise comparisons
were significance (all ps>.23).
Discussion
The aim of the current study was to help clarify how reading with contextual support and/or
corrective feedback affect orthographic learning. Interesting to note, our results paint a very different
picture depending on whether the dependent variable was reading or spelling accuracy. For reading
accuracy, there were initial benefits of context and larger, continued benefits of feedback. Yet for
spelling accuracy, there were no significant effects of feedback, and greater gains were observed after
children practiced reading words in isolation.
Unpacking these complex findings leads to a better understand the circumstances under which
context and feedback are beneficial (or, in contrast, neutral or even harmful) to learning. For example,
reading accuracy was clearly improved by the provision of external feedback, and improved to a lesser
extent by context. These results resemble those noted by Martin-Chang and colleagues, who routinely
provide feedback during training (Martin-Chang & Levy, 2005,2006; Martin-Chang et al., 2007), and
they partially align with the findings of Cunningham (2006) and Wang et al. (2011), who reported
initial accuracy benefits when children read regularly spelled words in coherent passages. The
literature examining feedback is much smaller; however, like Barbetta et al. (1994) and Spaai et al.
(1991), we also found that children learned to read words more efficiently when they were provided
with feedback compared to when they were left to decode on their own. The benefits of feedback on
reading accuracy were equally clear regardless of whether the children read in isolation or context, and
they continued to influence reading 1 week after training. This advocates for providing children with
corrective feedback because it may allow additional opportunities to link words’pronunciations (and
perhaps semantic meanings from preexisting oral vocabulary) to their printed forms.
Together, these findings suggest that learning to read in supportive environments leads to greater
reading success during successive word encounters and substantiates the claim that laying down
word-specific orthographic representations, at least in their most rudimentary forms, is facilitated
initially by reading in context and boosted considerably by corrective feedback. A commonality
between reading in context and receiving feedback is that both conditions increase the probably of
pronouncing/hearing (phonological) the words at the same time as they are being visually analyzed
(orthography). This raises the possibility that binding these lexical identities is the process under-
lying improved reading accuracy (Ehri, 2014; Ouellette, 2010).
In contrast to the results for reading accuracy, where the contribution from feedback was robust and
enduring, the role of reading feedback on spelling transfer was elusive. Whether words were read in context
or isolation, feedback did not appear to effect subsequent spelling transfer. Furthermore, reading in context
was not associated with improved performance when it came to spelling; indeed, the higher levels of
spelling accuracy were witnessed for words that had been practiced in isolation. This raises the question of
why such differing patterns were observed when children were asked to read and spell the same words.
Spelling accuracy is typically regarded as more indicative of the quality of underlying ortho-
graphic representations compared to reading accuracy (Perfetti, 2007). When considering the read-
ing and spelling results together, it appears that the orthographic representations created when
SCIENTIFIC STUDIES OF READING 25
reading in supportive environments (with context and feedback) might not have been well-enough
specified to support accurate spelling production. In other words, they may have been “good
enough”for improving reading accuracy but not for spelling. In contrast, when children read
words in isolation, they seemed to lay down more detailed and precise representations, which
allowed for more accurate spelling performance 1 day following training. The present results thus
suggest that focused attention on bottom-up processes might be more conducive to creating the
types of high-quality representations that can later support accurate spelling.
Implications for practice
Fluent reading and accurate spelling are both extensions of orthographic learning, but of the two, it
has been posited that spelling is more tightly bound to the quality of the stored representation
(Perfetti, 2007). Evidence for this notion comes from the observation that spelling practice shows
high transfer rates to reading accuracy (Conrad, 2008). In contrast, as shown here, success during
reading practice does not guarantee that words will be accurately spelled. These findings support the
Lexical Quality Hypothesis (Perfetti, 2007), which proposes that lexical quality is incremental; as the
orthographic representations evolve from being low to high quality, the literacy outcomes that they
support seem to transition from being the least to most cognitively demanding. In other words, the
improvement in the underlying orthographic representation might be expressed as the transition
from being able to read to being able to spell (for a more general account of how word reading
predicts orthographic learning, see Deacon, Benere, & Castles, 2012).
Classroom implications based on these findings would include providing feedback during reading
practice to boost word reading accuracy, and thus bring about more opportunities to lay down new,
albeit potentially incomplete, orthographic representations. Practices that might be effective in providing
feedback include guided reading (Ford & Opitz, 2008) or working with reading buddies (Fuchs et al.,
2001). In addition to school-based activities, parentsshould be encouraged to listen to their children read
aloud at home and to provide feedback when necessary. The present findings also suggest the validity of
repeatedly reading the same text with regards to increasing reading accuracy (Nation et al., 2007).
As outlined earlier, although contextual and feedback supports may increase opportunities for
orthographic learning, more detailed representations appear to follow isolated word practice. Hence,
once the rudimentary orthographic representations have been established that allow accurate word
reading, further practice with isolated words would be expected to hone the quality of the representa-
tions, enabling both accurate spelling and reading. In the classroom, such practice might take the form of
isolated word reading or spelling practice. Indeed, previous research has demonstrated that spelling
practice with isolated words can lead to superior orthographic learning relative to repeated readings
(Ouellette, 2010;Ouellette&Tims,2014; Shahar-Yames & Share, 2008), and spelling practice with
feedbackimproves both phonological awareness and subsequent word reading performance (Ouellette &
Sénéchal, 2008; Ouellette, Sénéchal, & Haley, 2013). It would be of interest to directly compare reading
and spelling practice using larger, more varied (word or nonword) stimuli sets, over longer training
durations and with multiple outcome measures, as implemented in the present study.
Limitations and future directions
The current study trained children to read a fairly large bank of real words that varied in complexity
and regularity. This allowed us to maintain ecological validity and avoid ceiling effects during the first
several trials of training. Lost in this approach, however, is the role played by word-level lexical
characteristics, such as frequency and consistency. This is clearly a direction for future research. It is
important to note that in the present study, word sets were counterbalanced across training conditions
in a fully within-participant design, thus negating any confounding effects of lexical properties. Still,
this is an interesting avenue for future research to pursue, especially considering previous findings that
the benefits of contextual reading on orthographic development may be limited to irregular words
26 S. MARTIN-CHANG ET AL.
(Wang et al., 2011). In addition, although the use of real words adds ecological validity, studies with
nonwords are able to avoid the issue of preexisting word knowledge. In this respect, the self-teaching
field would benefit from studies using larger, multisyllabic, nonword sets.
It is also worthy to mention that in the present paradigm, feedback was offered only after a child
was unable to decode a given word. This was done to ensure that feedback was not replacing self-
teaching attempts (see Staels & Van den Broeck, 2014); yet the consequence was that feedback was
not provided for all words and varied between participants. It might thus be questioned as to
whether there were enough instances of feedback to bring about change in spelling accuracy. These
concerns are alleviated, however, by the substantial effect size for increased reading accuracy
reported for the feedback conditions, which indicates that feedback was frequent enough to have a
robust effect on student performance in reading accuracy. Therefore it remains surprising that these
effects did not translate into improved spelling at posttest.
Self-teaching is hypothesized to underpin reading fluency. In the current study, fluency was
encouraged by allowing the children only 2 s
1
to read each target word before they were either asked
to “keep going”(no feedback) or given the correct pronunciation (feedback). This time limit might
have been overly restrictive, and future studies may consider lengthening the amount of time that
children have to decode words; this would especially be the case during the initial encounters with
words, when self-teaching is most likely to occur.
Finally, we acknowledge that the limited sample size of the present study may have impacted the
power, even in the fully within-participant design. In particular, some of the more moderate effect
sizes (e.g., accuracy when reading in context without feedback) did not reach statistical significance
in the omnibus testing. It is important for future research to determine if there may still be practical
significance in these statistically nonsignificant findings; one means of addressing this possibility
would be to run similar training studies with larger samples to increase statistical power.
Conclusions
Share (1999,2004) has argued that decoding leads to creating orthographic representations in memory.
The data reported here support and extend this notion. If we had only measured orthographic learning
via spelling, as is often the case, we would have concluded that neither reading in context nor receiving
feedback enhances the creation of orthographic representations in memory. In fact, our data clearly show
that reading words in isolation is preferable if the ability to spell words is the primary goal. However,
when the end goal of instruction is broadened to literacy development, which includes reading accuracy,
the picture becomes more nuanced. Specifically, we are left with two general conclusions: First, pure
decoding in the absence of top-down support from context, and regardless of whether feedback is
provided, is central to creating the type of high-quality orthographic representations that underlie the
ability to do difficult tasks, such as spell words (Share, 2004). Second, as suggested by Ehri (2005,2014;
Rosenthal & Ehri, 2011), saying a word (with the help of context) or hearing a word (via feedback) after
decoding attempts have been made builds orthographic representations that are “good enough”to
support less taxing tasks, such as reading. When conceptualizing lexical quality as a developmental
continuum, it might be said that reading with support can start building rudimentary orthographic
representations but more focus at the word level is required for increasing quality. When it comes to
literacy acquisition, then, the question becomes, How good is good enough?
Funding
We gratefully acknowledge financial support from the Natural Sciences and Engineering Research Council to the first
and second authors.
1
Skilled readers can visually recognize words in less than a second (Martin-Chang et al., 2014)
SCIENTIFIC STUDIES OF READING 27
References
Archer, N., & Bryant, P. (2001). Investigating the role of context in learning to read: A direct test of Goodman’s Model.
British Journal of Psychology,92, 579–591. doi:10.1348/000712601162356
Barbetta, P. M., Heward, W. L., Bradley, D. M., & Miller, A. D. (1994). Effects of immediate and delayed error
correction on the acquisition and maintenance of sight words by students with developmental disabilities. Journal of
Applied Behaviour Analysis,27, 177–178. doi:10.1901/jaba.1994.27-177
Bowey, J., & Muller, D. (2005). Phonological recoding and rapid orthographic learning in third-graders silent reading:
A critical test of the self-teaching hypothesis. Journal of Experimental Child Psychology,92, 203–219. doi:10.1016/j.
jecp2005.06.005
Castles, A., & Nation, K. (2008). Learning to be a good orthographic reader. Journal of Research in Reading,31,1–7.
doi:10.1111/j.1467-9817.2007.00367.x
Conrad, N. (2008). From reading to spelling and spelling to reading: Transfer goes both ways. Journal of Educational
Psychology,100, 869–878. doi:10.1037/a0012544
Conrad, N., Harris, N., & Williams, J. (2013). Individual differences in children’s literacy development: The contribu-
tion of orthographic knowledge. Reading and Writing,26, 1223–1239. doi:10.1007/s11145-012-9415-2
Cunningham, A. E. (2006). Accounting for children’s orthographic learning while reading text: Do children self-teach?
Journal of Experimental Child Psychology,95(1), 56–77. doi:10.1016/j.jecp.2006.03.008
Cunningham, A., Perry, K. E., Stanovich, K. E., & Share, D. L. (2002). Orthographic learning during reading:
Examining the role of self-teaching. Journal of Experimental Child Psychology,82, 185–199. doi:10.1016/S0022-
0965(02)00008-5
Deacon, S. H., Benere, J., & Castles, A. (2012). Chicken or egg? Untangling the relationship between orthographic
processing skill and reading accuracy. Cognition,122, 110–117. doi:10.1016/j.cognition.2011.09.003
Dunn, L., & Dunn, L. (2006). Peabody picture vocabulary test –4. Circle Pines, MN: American Guidance Service.
Ehri, L. C. (2005). Learning to read words: Theory, findings, and issues. Scientific Studies of Reading,9, 167–188.
doi:10.1207/s1532799xssr0902_4
Ehri, L. C. (2014). Orthographic mapping in the acquisition of sight word reading, spelling memory, and vocabulary.
Scientific Studies of Reading,18(1), 5–21. doi:10.1080/10888438.2013.819356
Ford, M. P., & Opitz, M. F. (2008). A national survey of guided reading practices: What we can learn from primary
literacy teachers. Literacy Research and Instruction,47, 309–331. doi:10.1080/19388070802332895
Fuchs, D., Fuchs, L. S., Thompson, A., Svenson, E., Yen, L., Al Otaiba, S., . . . Saenz, L. (2001). Peer-assisted learning
strategies in reading: Extensions for kindergarten, first grade, and high school. Remedial and Special Education,22,
15–22. doi:10.1177/074193250102200103
Gough, P. B., Alford, J. A., Jr., & Holley-Wilcox, P. (1981). Words and contexts. In O. J. L. Tzeng & H. Singer (Eds.),
Perceptions of print (pp. 85–102). Hillsdale, NJ: Erlbaum.
Landi, N. (2013). Learning to read words: Understanding the relationship between reading ability, lexical quality, and
reading context. In M. A. Britt, S. Goldman, & J. F. Rouet (Eds.), Reading—From words to multiple texts (pp.
17–33). New York, NY: Routledge.
Landi, N., Perfetti, C. A., Bolger, D. J., Dunlap, S., & Foorman, B. R. (2006). The role of discourse in developing word
form representations: A paradoxical relation between reading and learning. Journal of Experimental Child
Psychology,94, 114–133. doi:10.1016/j.jecp.2005.12.004
Martin-Chang, S. L., & Levy, B. A. (2005). Fluency transfer: Differential gains in reading speed and accuracy following
isolated word and context training. Reading and Writing,18, 343–376. doi:10.1007/s11145-005-0668-x
Martin-Chang, S. L., & Levy, B. A. (2006). Word reading fluency: A transfer appropriate processing account of fluency
transfer. Reading and Writing,19, 517–542. doi:10.1007/s11145-006-9007-0
Martin-Chang, S. L., Levy, B. A., & O’Neil, S. (2007). Word acquisition, retention, and transfer: Findings from
contextual and isolated word training. Journal of Experimental Child Psychology,96,37–56. doi:10.1016/j.
jecp.2006.08.004
Martin-Chang, S. L., Ouellette, G., & Madden, M. (2014). Does poor spelling equate to slow reading? The relationship
between reading, spelling, and orthographic quality. Reading and Writing,27, 1485–1505. doi:10.1007/s11145-014-
9502-7
Nation, K., Angell, P., & Castles, A. (2007). Orthographic learning via self-teaching in children learning to read
English: Effects of exposure, durability, and context. Journal of Experimental Child Psychology,96, 71–84.
doi:10.1016/j.jecp.2006.06.004
Nation, K., & Snowling, M. (1998). Individual differences in contextual facilitation: Evidence from dyslexia and poor
reading comprehension. Child Development,69, 996–1011. doi:10.1111/j.1467-8624.1998.tb06157.x
Nemko, B. (1984). Context versus isolation: Another look at beginning readers. Reading Research Quarterly,19, 461–467.
doi:10.2307/747917
Nicholson, T. (1991). Do children read words better in context or in lists? A classic study revisited. Journal of
Educational Psychology,83, 444–450. doi:10.1037/0022-0663.83.4.444
28 S. MARTIN-CHANG ET AL.
Ouellette, G. (2010). Orthographic learning in learning to spell: The roles of semantics and type of practice. Journal of
Experimental Child Psychology,107,50–58. doi:10.1016/j.jecp.2010.04.009
Ouellette, G., & Fraser, J. R. (2009). What exactly is a yait anyway: The role of semantics in orthographic learning.
Journal of Experimental Child Psychology,104, 239–251. doi:10.1016/j.jecp.2009.05.001
Ouellette, G., & Sénéchal, M. (2008). Pathways to literacy: A study of invented spelling and its role in learning to read.
Child Development,79, 799–813. doi:10.1111/j.1467-8624.2008.01166.x
Ouellette, G., Sénéchal, M., & Haley, A. (2013). Guiding children’s invented spellings: A gateway into literacy learning.
The Journal of Experimental Education,81, 261–279. doi:10.1080/00220973.2012.699903
Ouellette, G., & Tims, T. (2014). The write way to orthographic learning: Printing versus typing in spelling acquisition.
Frontiers in Psychology,5, 117. doi:10.3389/fpsyg.2014.00117
Perfetti, C. A. (2007). Reading ability: Lexical quality to comprehension. Scientific Studies of Reading,11, 357–383.
doi:10.1080/10888430701530730
Ricketts, J., Bishop, D. V. M., Pimperton, H., & Nation, K. (2011). The role of self-teaching in learning orthographic
and semantic aspects of new words. Scientific Studies of Reading,15,47–70. doi:10.1080/10888438.2011.536129
Rosenthal, J., & Ehri, L. C. (2011). Pronouncing new words aloud during the silent reading of text enhances fifth
graders’memory for vocabulary words and their spellings. Reading and Writing,24, 921–950. doi:10.1007/s11145-
010-9239-x
Shahar-Yames, D., & Share, D. L. (2008). Spelling as self-teaching mechanism in orthographic learning. Journal of
Research in Reading,31,22–39. doi:10.1111/j.1467-9817.2007.00359.x
Share, D. L. (1995). Phonological recoding and self-teaching: Sine qua non of reading acquisition. Cognition,55, 151–218.
doi:10.1016/0010-0277(94)00645-2
Share, D. L. (1999). Phonological recoding and orthographic learning: A direct test of the self-teaching hypothesis.
Journal of Experimental Child Psychology,72,95–129. doi:10.1006/jecp.1998.2481
Share, D. L. (2004). Orthographic learning at a glance: On the time course and development onset of self-teaching.
Journal of Experimental Child Psychology,87,267–298. doi:10.1016/j.jecp.2004.01.001
Spaai, G. W. G., Ellerman, H. H., & Reitsma, P. (1991). Effects of segmentation and whole-word sound feedback on
learning to read single words. The Journal of Educational Research,84, 204–213. doi:10.1080/
00220671.1991.10886017
Staels, E., & Van den Broeck, W. (2014). Orthographic learning and the role of text-to-speech software in Dutch
disabled readers. Journal of Learning Disabilities,48,39–50. doi:10.1177/0022219413487407
Tunmer, W. E., & Chapman, J. W. (2002). The relation of beginning readers’reported word identification strategies to
reading achievement, reading-related skills, and academic self-perceptions. Reading and Writing,15, 341–358.
doi:10.1023/A:1015219229515
Verhoeven, L., & Perfetti, C. (2011). Morphological processing in reading acquisition: A cross-linguistic perspective.
Applied Psycholinguistics, 32, 457–466. doi:10.1017/S0142716411000154
Wagner, R. K., Torgesen, J. K., Rashotte, C. A., & Pearson, N. A. (2013). Comprehensive Test of Phonological Processing
(2nd ed.). Austin, TX: Pro-Ed.
Wang, H., Castles, A., Nickels, L., & Nation, K. (2011). Context effects on orthographic learning of regular and
irregular words. Journal of Experimental Child Psychology,109,39–57. doi:10.1016/j.jecp.2010.11.005
Wilkinson, G. S. (1993). The wide range achievement test: Manual (3rd ed.). Wilmington, DE: Wide Range.
SCIENTIFIC STUDIES OF READING 29
Appendix A
Appendix B
Table A1. Training Words.
List 1 List 2 List 3 List 4 List 5
auntie absolutely announced blanket appeared
bath ancient attempting build area
beamed beneath bellowed bundle creatures
bowl boards camera cancelled crickets
change both captured castle detected
confirmed bridge children darling everything
continued caused climb declared exploring
enormous clatter convinced entrance halted
enough concern decided exclaimed imitated
holler cross entirely front insect
ideal dwelled fool glimpse intelligent
leapt father giggling hideout know
nephew follow instructor jacket noise
niece hazardous nest mittens noticed
nothing hiking picture nanny realized
overjoyed laughing rare ought requested
shower monster regarded school research
soaked ogre returned scurried scientist
soapsuds overheard seize sheltered snickered
splashing relatives shrieking snow student
sweater seemed signal snowflakes tune
swiftly sternly spied window unexpectedly
twins teenagers thrashing winter unusual
water upon towards wrapped while
worried ventured tree wrestled whistling
Table B1. Experimental Training Design.
Monday Tuesday Wednesday Thursday Friday
Trials: 1–2 Trials: 3–4 Trials: 5–6 Trials: 7–10
Spelling posttest:
Word Set 1
Story 1
Weeks 1 & 2: Context feedback Delayed Post test: Story 1
Story 2
Weeks 3 & 4: Context no feedback Delayed Post test: Story 2
List 3
Weeks 5 & 6: Isolation feedback Delayed Post test: List 3
List 4
Weeks 7 & 8: Isolation no feedback Delayed Post test: List 4
Note. The order of the feedback conditions and the training conditions was counterbalanced over all participants. Half of the
children received the no-feedback training condition first, and half of the children received context isolation training first. In
addition, the training words were counterbalanced over all conditions so that each set of words was trained in each of the four
conditions (and used in the control condition).
30 S. MARTIN-CHANG ET AL.
