Learning to Read With and Without Feedback, In and Out of Context

Abstract

The self-teaching hypothesis posits that enduring orthographic and phonological representations are produced when children independently recode print into speech. However, very little research has examined how children self-teach when initial decoding attempts are weak or ineffective. In this within-participant design, 25 students in Grade 2 learned to read 85 different words in 4 conditions. Words were read in and out of context, with and without feedback. Accuracy rates were recorded throughout 5 training sessions (2 word repetitions per session = 10 repetitions in total). A posttest was administered after a 6-day delay by reinstating the training materials. At the end of training, the highest accuracy scores were observed when children read in context/feedback followed by when they read in isolation/feedback, and then in context/no feedback; the lowest accuracy scores were observed when children read in isolation/no feedback. This pattern remained over the retention period, suggesting that external support from feedback, and top-down support from context, can help create word representations in memory. The results are discussed in relation to the importance of whole-word phonology within self-teaching.

Full text

Learning to Read With and Without Feedback, In and Out of Context
Sandra Martin-Chang
Concordia University
The self-teaching hypothesis posits that enduring orthographic and phonological representations are
produced when children independently recode print into speech. However, very little research has
examined how children self-teach when initial decoding attempts are weak or ineffective. In this
within-participant design, 25 students in Grade 2 learned to read 85 different words in 4 conditions.
Words were read in and out of context, with and without feedback. Accuracy rates were recorded
throughout 5 training sessions (2 word repetitions per session !10 repetitions in total). A posttest was
administered after a 6-day delay by reinstating the training materials. At the end of training, the highest
accuracy scores were observed when children read in context/feedback followed by when they read in
isolation/feedback, and then in context/no feedback; the lowest accuracy scores were observed when
children read in isolation/no feedback. This pattern remained over the retention period, suggesting that
external support from feedback, and top-down support from context, can help create word representations
in memory. The results are discussed in relation to the importance of whole-word phonology within
self-teaching.
Keywords: context, decoding, feedback, isolation, self-teaching, word reading
Share (2004) noted that children “self-teach” the majority of the
words they can read. The ability to self-teach is associated with
two potential factors. The first factor relates to the process of
recoding; namely, focusing on grapheme-to-phoneme correspon-
dences during decoding may help to create well-specified ortho-
graphic representations in memory (Share, 1995). The second
factor relates to the product of recoding; in this case, focusing
simultaneously on whole-word orthography and phonology, may
help amalgamate written words with their spoken pronunciations
(Elbro, de Jong, Houter, & Nielsen, 2012). When decoding is
successful, it is difficult to disentangle the effects of these two
factors: proficient decoding results in the word’s correct pronun-
ciation. However, ineffective decoding creates the opportunity to
examine the second possible factor more closely. For example,
when decoding skills are weak, or the word to be read is difficult,
correct pronunciations are more likely to be activated when chil-
dren read in context. Moreover, when decoding fails to produce the
spoken word altogether, correct pronunciations can be provided
via feedback from a “teacher.” The current experiment explored
the second potential factor involved in self-teaching by examining
children’s growth in reading accuracy as they read in, and out, of
context—with, and without, corrective feedback.
The Self-Teaching Hypothesis
The notion behind self-teaching is that children build up “sight
word” lexicons—words that can be read automatically— by creat-
ing orthographic representations as they decode words. In essence,
“sounding out” new words induces a form of cognitive processing
akin to “focal attention” (Samuels, 1967) because children’s at-
tention is focused on the letters, letter patterns, and letter sequenc-
ing that make up each word’s unique orthographic form (Share,
1999). This process allows children to gradually accumulate gen-
eral orthographic knowledge (knowledge about the language as a
whole), by internalizing orthographic representations of many
specific words (e.g., slowly coming to understanding that the “ss”
spelling pattern is legal at the end, but not beginning of English
words through exposure with words such as less and sell,boss,
sob, etc.). Support for this hypothesis has come from several
empirical investigations showing robust word-specific ortho-
graphic gains after a minimal number of successful reading expe-
riences for both nonwords (Bowey & Muller, 2005;Cunningham,
Perry, Stanovich, & Share, 2002;Ouellette, 2010;Ouellette &
Fraser, 2009;Share, 1999;Wang, Castles, Nickels, & Nation,
2011) and, to a lesser extent, real words (Cunningham, 2006;
Landi, Perfetti, Bolger, Dunlap, & Foorman, 2006).
Ziegler, Perry, and Zorzi (2014) recently simulated the self-
teaching process by building a connectionist model around two
main assumptions: (a) that young children begin to read with a
well-developed spoken vocabulary; and (b) that before reading
begins in earnest, children are explicitly taught phonics. The
Phonological Decoding Self-Teaching Model quickly learned how
to read more than 25,000 words and was able to generalize this
learning to read nonwords. Critical to the self-teaching hypothesis,
it was able to accomplish these tasks without the use of feedback
from a teacher. Interestingly context was also seen to play a key
role in the model. For example, when the correct word was
generated as one of several possible word candidates, it was
This article was published Online First June 2, 2016.
I gratefully acknowledge support from the Social Sciences and Human-
ities Research Council (SO1294) and the children, teachers and principals
who participated in this study. I also thank Gene Ouellette, Kyle Levesque,
and Holly Recchia for their thoughts and comments on an earlier draft of
this article.
Correspondence concerning this article should be addressed to Sandra
Martin-Chang, Department of Education, Concordia University, 1455 de
Maisonneuve Boulevard West, Montreal, Quebec, Canada, H3G 1M8.
E-mail: smartinc@education.concordia.ca
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Journal of Educational Psychology © 2016 American Psychological Association
2017, Vol. 109, No. 2, 233–244 0022-0663/17/$12.00 http://dx.doi.org/10.1037/edu0000131
233

chosen based on the premise that in “real learning situations with
real texts, children have additional information from the story
context, semantics, or syntax to help them choose the correct
target” (p. 6). In short, given alternative pronunciations, Ziegler et
al., suggest that the ability to select the correct pronunciation is
attributed to additional information gained from reading in context.
Self-Teaching in Context
The self-teaching hypothesis is aligned with decades of research
showing the importance of explicit and systematic teaching of
letter-sound correspondences (Snowling & Hulme, 2011). Indeed,
a core tenet of self-teaching is that children need to understand the
alphabetic principle so that decoding can take place. The self-
teaching hypothesis also acknowledges that most word knowledge
is learned implicitly via experience with texts (e.g., Elgort, Perfetti,
Rickles, & Stafura, 2015;Nagy, Anderson, & Herman, 1987).
Words in naturalistic text are difficult to predict (Gough, Alford,
& Holley-Wilcox, 1981); therefore, when children read authentic
texts they seldom rely on guessing from context as their primary
strategy for reading (Nation & Snowling, 1998;Tunmer & Chap-
man, 1995). Still, under certain conditions predicting words from
context becomes more likely. For example, when reading highly
constrained texts (e.g., “Roses are red, violets are _”), individuals
are less likely to fixate on the predictable words. Moreover, when
the predicable words are fixated, the amount of time spent looking
at them tends to be shorter in duration compared with those that are
less predictable (Ehrlich & Rayner, 1981). Under normal circum-
stances, however, rather than simply “guessing” from context,
children use both partial decoding attempts and the structure of the
text to arrive at correct word pronunciations (Nation & Snowling,
1998;Share, 2004;Tunmer & Chapman, 2012). The advantage
gained from this type of crosschecking between semantics and
print has been termed “contextual facilitation.”
The beneficial effects of reading in context have been well
documented (Martin-Chang & Levy, 2006;Martin-Chang, Levy,
& O’Neil, 2007;Roth & Perfetti, 1980;Stanovich, Nathan, West,
& Vala-Rossi, 1985). What remains contentious is whether the
performance gains observed during online contextual reading,
accumulate to result in crystalized learning that is capable of
supporting later reading fluency (for the distinction between online
performance and crystallized learning see Byrne et al., 2013).
Landi and colleagues (2006) were interested in this distinction
between reading performance and generalized learning. They iden-
tified two written word sets that children in Grades 1 and 2
(Experiment 1) were unable to accurately read aloud. Half of the
words were then shown in isolation, and the other half were
presented in predicable sentences. Landi et al. reported that chil-
dren were able to read more words in predictable sentences than in
isolation (13.29 words in context compared with 4.88 words in
isolation). However, both sets of words were read with similar
accuracy when presented in isolation one week later (6.15 words in
context, 5.3 words in isolation). This led Landi et al. to conclude
that children performed better in context initially, but that reading
words in isolation was superior for learning. However, as de-
scribed previously, very highly constrained text, which is generally
not representative of children’s more naturalistic contextual read-
ing, encourages top-down processing and reduces the need for
students to focus on the print. Therefore, Landi et al.’s selection of
text may have inadvertently reduced the opportunities for self-
teaching in context.
Cunningham’s work (Cunningham, 2006) addressed some of
these issues by inviting children in Grade 1 to read four coherent
and four scrambled passages that were longer and less predictable
than the materials used by Landi et al. (2006). Cunningham re-
ported that words were read more accurately when they were
initially presented in a meaningful context (83.6%) compared with
in a scrambled passage (67%). However, 3 days after self-teaching,
the children performed similarly on an orthographic choice task
and a spelling task, regardless of whether the words were first read
in context or isolation. A posttest measure of reading accuracy was
not included—perhaps because there were only eight words to be
learned in total (one word per passage). Therefore, the results of
this study cannot be directly compared with those of Landi et al.,
in terms of word reading accuracy.
Self-teaching in and out of context was also explored in two
experiments by Wang et al. (2011). The authors created a set of
words that were understood verbally by associating nonwords with
meanings. After the pronunciations and definitions of the non-
words were well understood orally, they were presented in writing.
The spellings of the nonwords remained the same over two exper-
iments; however, the pronunciations associated with the nonwords
differed. In Experiment 1 the pronunciations were regular while in
Experiment 2, they were irregular. Wang et al. reported that the
regular words were read more accurately in context during the first
self-teaching trial and that the irregular words were read more
accurately across all four contextual training trials. They con-
cluded that when decoding is difficult, as it is for irregular words,
context helps children read words accurately. However, like Cun-
ningham (2006), Wang et al. used a small learning set (four words
per condition). Therefore, it is unclear whether their findings
would generalize to reading longer passages.
Reviewing the work of Martin-Chang and Levy (2005) can help
resolve some of these questions. They presented average readers in
Grade 2 (Experiment 2) with 85 real words read in a meaningful
story and 85 different words read in isolation. The authors found
that the children read the target words more accurately in context
compared with in isolation during training. They also found that
children read new stories faster and more accurately if the words
in the stories had first been trained in a different context. However,
Martin-Chang and Levy departed from the methodology of other
researchers by electing to give corrective feedback in response to
children’s errors; therefore, their results cannot be directly com-
pared with the self-teaching literature (e.g., Cunningham, 2006;
Landi et al., 2006;Wang et al., 2011).
Self-Teaching and Feedback
A defining characteristic of the self-teaching model is that
feedback from external sources is not required for word acquisi-
tion. However, very little research has compared how self-teaching
in the absence of an expert compares to learning with the assis-
tance of feedback.
When the child receives whole word feedback, the “whole
word” is supplied after an error. This type of feedback is often
negatively contrasted with graphophonemic feedback, which re-
lates individual letters or letter clusters with specific sounds.
Whole word feedback—also termed “terminal feedback”— has
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
234 MARTIN-CHANG

been faulted for ending the children’s decoding attempts, (Evans,
Barraball, & Eberle, 1998). Phonological recoding has been spec-
ulated to be a crucial step in learning how to read words fluently
(Share, 2004); consequently, if receiving assistance discourages
children from actively recoding, or ends the recoding process
prematurely, it might also impair the quality of word representa-
tions in memory. Indeed, Landi and colleagues (Landi, 2013;
Landi et al., 2006) have speculated that students will be less likely
to spend cognitive resources during initial decoding attempts when
they know the correct pronunciation is forthcoming. If this is the
case, feedback would be expected to reduce exhaustive grapheme-
phoneme recoding, to the detriment of self-teaching.
Without question, a teacher or parent who “supplies” a misread
word is doing the important work of recoding the symbols into
speech on the child’s behalf; yet it is possible that hearing the
pronunciation while attending to the print might put the child in a
better position to phonologically recode the word on subsequent
trials. As argued by Ehri (2014), when “readers see a new word
and say or hear its pronunciation, its spelling becomes mapped
onto its pronunciation and meaning” (Ehri, 2014, p. 6). Barbetta,
Heward, Bradley, and Miller (1994) provided evidence to support
this view by teaching five students in Grade 2 to read words in
isolation with immediate or delayed whole-word feedback. The
results showed that immediate feedback was more profitable than
delayed feedback in the acquisition and maintenance of word
reading. The authors speculated that the immediate feedback re-
duced the likelihood that the same mistakes were repeated
throughout training. Additionally, immediate feedback allowed the
pronunciation to be heard soon after the print was seen, which may
have contributed to the amalgamation of the words’ phonological
and orthographic forms.
Current Investigation
The literature discussing self-teaching stresses the importance of
each “successful recoding experience” as if it was one unified
entity (Cunningham, 2006;Share, 1999,2004). However, within
every fruitful decoding attempt there are two factors that could be
contributing to self-teaching: (a) grapheme-phoneme recoding;
and (b) the pairing of whole word orthography and phonology.
This pairing of complete spoken words with their respective letter
strings could be achieved in at least three ways: by pure bottom-up
decoding (the first factor in self-teaching); by decoding that is
supplemented with top-down support from context; or by feedback
that is provided by an external source after a failed decoding
attempt. The question is, does providing support (from context or
feedback) weaken or strengthen long-term word recognition? If
independent grapheme-phoneme recoding is critical to self-
teaching, then situations where there is very little support (e.g.,
reading in isolation without feedback) would be expected to pro-
duce the highest degree of accuracy over training. In contrast, if
pairing whole-word orthography and phonology is central to cre-
ating word representations in memory (Ehri, 2014), then situations
that offer the most support for reading accuracy (e.g., reading in
context with feedback) should result in superior accuracy. The
current investigation tested these hypotheses by having students
read a large set of words, in and out of meaningful text, with and
without feedback.
Method
Design
A within-subjects design was implemented with two experimen-
tal manipulations: the availability of context and the provision of
feedback. Specifically, the first manipulation involved whether the
target words were presented in isolation or in context. The second
manipulation involved whether whole-word feedback was pro-
vided or whether no feedback was given. Taken together, the
training phase of the study consisted of four distinct experimental
reading conditions: context/feedback,isolation/feedback,con-
text/no feedback, and isolation/no feedback. Participants were ex-
posed to four unique sets of target words, which were counterbal-
anced across all four experimental conditions (see Figure 1). The
experiment was conducted in two 18-day blocks. In one block, the
children were given feedback during both the context and isolated
word training conditions. During the other block, the children did
not receive feedback during either context or isolated word train-
ing. Each block contained five training sessions (on Days 1, 3, 7,
9, and 11). A posttest was administered on the last day of each
block (Day 18) to determine if accuracy gains would be main-
tained over a delay.
Participants
Twenty-eight participants were recruited from three Grade 2
classrooms in central Canada. The children’s teachers all reported
Block 1: Feedback*
Monday
Tuesday
Wednesday
Thursday
Friday
Screening
Day 1
Session 1:
List A**
Story B
Day 2
Day 3
Session 2:
List A
Story B
Day 4
Day 7
Session 3:
List A
Story B
Day 8
Day 9
Session 4:
List A
Story B
Day 10
Day 11
Session 5:
List A
Story B
Day 14
Day 15
Day 16
Day 17
Day 18
Retention:
List A
Story B
Block 2: No feedback*
Monday
Tuesday
Wednesday
Thursday
Friday
Screening
Day 1
Session 1:
Story C**
List D
Day 2
Day 3
Session 2:
Story C
List D
Day 4
Day 7
Session 3:
Story C
List D
Day 8
Day 9
Session 4:
Story C
List D
Day 10
Day 11
Session 5:
Story C
List D
Day 14
Day 15
Day 16
Day 17
Day 18:
Retention
Story C
List D
Figure 1. Experimental training design, Story !context, and list !
isolation.
!
The order of the feedback conditions and the training conditions
was counterbalanced over all participants, so that half of the children
received the no-feedback training condition first, and the other half re-
ceived the story condition 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.
!!
Each list and story contained two
repetitions of 85 different training words. Therefore, the children read a
total of 170 unique words (half in a story, half in a list), twice, each day.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
235
FEEDBACK AND CONTEXT

using strong phonics training (e.g., Jolly Phonics) in their regular
classroom instruction. Children with parental consent were
screened with the reading subtest of the Wide Range Achievement
Test—Third Edition (WRAT3; Wilkinson, 1993). In total, three
participants were omitted from the study. The first student had a
low standardized WRAT score ("70), the second was absent for
an extended period of time, and the third was shy and preferred not
to read aloud. The final sample consisted of 25 students (17 girls
and 8 boys). Participants were in Grade 2 and were, on average, 7
and half years old (M!7 years and 7 months, range !6 years and
7 months to 8 years and 3 months). The average standardized
WRAT score was 97 (SD !14, range from 75 to 130). After each
session, the children were thanked with small gifts (e.g., stickers,
pencils). They were also invited to choose an age appropriate book
to take home after each 18-day block.
Materials
To use longer and more complex passages for the children, four
nonoverlapping sets of 85 real words were counterbalanced over
the four training conditions (340 unique words in total/4 condi-
tions !85 words per condition). The words were selected because
they were contained in 7-to-8-year-olds’ spoken vocabularies
(Martin-Chang & Levy, 2005). For the present study, six Grade 2
teachers (including a classroom teacher from the current sample)
also vetted the materials to confirm that they contained words
children could understand orally. The teachers indicated that on
average, only 3.6% of the 340 words might be difficult for children
in Grade 2 to understand; the remaining 96.4% of the words were
deemed to be common in children’s spoken vocabularies.
The materials included words with both regular and irregular
spellings and words with a range of morphological complexities
(see Appendix A). The lists were not equated in terms of spelling
patterns or orthographic complexity; however, the average number
of letters (average !5.7 letters, range 5.5 letters to 5.9 letters) and
morphemes (average !1.44 morphemes, range 1.4 –1.5) per list
were controlled for. The items were not screened to avoid addi-
tional word exposures (see also Cunningham, 2006); therefore, it
was difficult to track how many of the words could already be read
before the first session.
The isolation reading condition was achieved by presenting the
words individually on a computer screen. The target items were
repeated twice within each isolated-word list to produce a total of
170 target words. To create the context condition, four narratives
were created around each of the lists (see Appendix B for an
example). The target words were repeated twice within each train-
ing story to produce a total of 170 target words. The four stories
contained some written words (approximately 30 – 40%) that were
above the children’s reading ability and were all evaluated to be a
4.0 grade level by the Flesh- Kincaid readability test. They were
written to resemble children’s passages and ranged from 675 to
766 words in length. As would be the case within authentic texts,
there was differing amount of contextual support for different
words.
Procedure
Two training conditions were conducted during each 18-day
block (see Figure 1). It was speculated that receiving feedback on
only some trials might create unnecessary confusion for the chil-
dren (e.g., they might wait for feedback on trials where no feed-
back was forthcoming). Therefore, the feedback manipulation was
administered in separate blocks. The order of the blocks was
counterbalanced so that half of the children received feedback in
both conditions (context and isolation) during the first block and
the other half of the children received feedback in both conditions
(context and isolation) during the last block. Within each block,
the order of the conditions was counterbalanced so that context
was presented first for half of the children and presented last for
the rest. It is important to note that, while the context and isolation
manipulation occurred on the same days (either in the feedback or
no feedback blocks), each of the four reading conditions corre-
sponded to different word sets. The delayed posttest occurred on
Day 18 of each block.
Two trained research assistants worked with the children in a
quiet place in their school. The first experimenter conducted the
training phase, while a second experimenter, who was blind to the
feedback condition used during training, conducted the posttest.
Training phase. The students were given praise and encour-
agement throughout the entire experiment (not contingent on cor-
rect reading). However, during the feedback block, the children
were also given whole-word corrective feedback after reading
errors. In the no feedback block, the participants were asked to
read as if they were alone; here, they were not given assistance of
any kind. Within each block (feedback/no feedback), the children
read two material sets; each set included 85 different words. The
children read one story (context) and one list (isolation) during
each training session. The stories and lists contained two repeti-
tions of each word.
During isolated-word training, individual words were presented
on a computer screen for 2 s, followed by a fixation point. The rapid
pace of word presentation kept the task from becoming overly long
and/or monotonous and the frequent encouragement and/or feed-
back from the experimenter kept the task from feeling completely
solitary. To ensure that the lists did not appear too “story like” (i.e.,
not rapid serial visual presentation) the words were shown in a
fixed- randomized order, with the only stipulation being that no
word was presented twice in a row (85 words #2 repetitions !
170 words). If the children were in the feedback block, the exper-
imenter provided whole-word feedback after errors. The children
were not asked to repeat the corrected word. Pauses longer than 2 s
were considered omissions. Inaccurate attempts and omissions
were both marked as errors. If the child was in the no feedback
block, training continued without interruption. The experimenter
discreetly coded the answer as correct or incorrect during the
fixation cross, and prompted the onset of the next word by pressing
a computer key. The sessions were audio recorded so that the
scoring sheets could be verified for accuracy.
During context training, a shared reading paradigm was adopted
where the participants read only the target words (85 #2 repeti-
tions !170 words per session). The shared reading paradigm
equated the task demands of the training conditions (isolation and
context). The children were asked to follow along with the story
and read the words that were bolded and underlined while the
experimenter read the remainder of the story. This style is similar
to one that might be adopted by a parent who pauses to let the
children read some of the words in a “daily reading” story at home.
The experimenter read at a natural pace, pausing at each target
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
236 MARTIN-CHANG

word. If the child was in the feedback condition, the experimenter
would read the target word after the child made an error or failed
to attempt the word after 2 s. The story continued without the child
repeating the corrected word. In the no-feedback condition, the
experimenter would resume with the story after errors. Once again,
the sessions were audiotaped for scoring purposes. The experi-
menter discreetly coded the child’s reading as correct or incorrect
as they were reading and inaccurate attempts and failures to
respond were both marked as errors.
Testing phase. The delayed posttest occurred on the last day
of the 18-day testing block. During the posttest, the children were
asked to read the same materials that they had used during their
training phase (i.e., same contextual story and same isolation word
list) to determine if the reading accuracy gains from the training
sessions had been retained. The only difference between the train-
ing phase and the posttest was that the children were not offered
assistance, regardless of whether they had been in the feedback
condition during training. If the child paused for more than 2 s, the
experimenter reassured the child that it was fine to “skip it” or
“keep going.”
Results
Word Reading in Session 1
As shown in Figure 2, the children were able to read many of the
target words at the onset of training. Accuracy ranged from 61 to
71% during the first session. To determine if there were any initial
differences between the conditions, a 2 (context, isolation) #2
(feedback, no feedback) repeated measures Analysis of variance
(ANOVA) was conducted on the accuracy scores from the first
training session. The significant main effect of context confirmed
that children were able to read more words correctly in context
compared with in isolation at the beginning of training, F(1, 24) !
39.73, MSE !4,121.64, p".001. The difference between word
reading in context versus in isolation corresponded to a very large
effect (r!.79). There was no main effect of feedback, indicating
the accuracy scores in Session 1 were similar regardless of whether
the children were in the feedback or no-feedback condition (p!
.15). The Context #Feedback interaction was not significant (p!
.65). Similar accuracy scores were expected in the two feedback
conditions because the supplementary instruction had not had time
to exert influence on training.
Word Reading in Sessions 1–5
The number of words that could be read inclusively from
Sessions 1–5 was also analyzed. Figure 2 depicts the percentage of
words read correctly during each session. The highest accuracy
scores from Session 2 onward were noted when children read in
context/feedback. The lowest scores were observed when they read
in isolation/no feedback. Figure 2 also shows that when children
read in isolation/feedback, they started training with reduced ac-
curacy, but by Session 3, the children’s scores in the isolation/
feedback condition had surpassed the context/no-feedback condi-
tion.
To evaluate whether participants’ word reading accuracy im-
proved across training sessions, and, if so, whether this change was
modulated by the experimental reading conditions, a 2 (context,
isolation) #2 (feedback, no feedback) #5 (Sessions: 1–5) re-
peated measures ANOVA was conducted. Inspection of Mauchly’s
test indicated that the assumption of sphericity had been violated
for the main effect of Session, $
2
(9) !103.25, p".001, and the
interaction between Feedback and Session, $
2
(9) !65.31, p"
.001. The degrees of freedom associated with these effects were
corrected using Greenhouse-Geisser conservative estimates of
sphericity (ε!.31 for the main effect of Session and ε!.40 for
the Feedback #Session interaction; Field, 2009). Results from the
ANOVA analysis confirmed significant main effects of context
(F(1, 24) !35.51, MSE !21,496.58, p".001, r!.77) and
feedback (F(1, 24) !39.43, MSE !37,243.48, p".001, r!.79),
both of which corresponded to large effect sizes (Cohen, 1988).
However, the Context #Feedback interaction was not significant
(p!.236). Furthermore, the significant main effect of session
showed that children were able to read a greater number of words
as they participated in additional practice sessions, F(1.26,
30.33) !65.24, MSE !31,879.71, p".001. Planned pairwise
comparisons, with Bonferroni adjustments for multiple compari-
sons, confirmed that each session was associated with significantly
greater learning in word reading compared with the performance in
the previous session (all pairwise ps".003). More importantly,
this main effect was qualified by a significant Feedback #Session
interaction, F(1.58, 37.89) !36.34, MSE !5813.50, p".001,
indicating that feedback resulted in accelerated learning over ses-
sions.
Planned comparisons using the “repeated” contrast function was
used to examine the interaction between Feedback and Session
more carefully. Repeated contrasts are especially useful in a
repeated-measures design in which the level of a variable has a
meaningful order (e.g., Session 1, 2, 3, etc.; Field, 2009). In terms
of the current study, the repeated contrasts compared the perfor-
mance of each session to the performance at the previous session
(Session 3 vs. 2), and evaluated the degree of learning in relation
to the effects of feedback. At nearly every successive practice
session, the Feedback condition was associated with significantly
greater learning in word reading accuracy compared with the No
Feedback condition (all ps".004; Session 1 vs. 2, r!.76;
Session 2 vs. 3, r!.65; Session 3 vs. 4, r!.56). Although there
was a significant improvement in reading accuracy between Ses-
sions 4 and 5 overall, the improvement was statistically similar for
Figure 2. Percentage of words read correctly during the five training
sessions and posttest.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
237
FEEDBACK AND CONTEXT

the Feedback and No Feedback condition (Session 4 vs. 5, p!.06.
The Context #Session (p!.50), and the Context #Feedback #
Session interactions (p!.19) were not significant, indicating that
the initial benefit of reading in context was maintained over all the
training sessions.
1
The contextual advantages observed throughout training raised
the question of whether the effects were being driven by a small
number of highly irregular items within the larger word set. To
gain insight into this question, the data were reorganized to be
examined from a within-word perspective. Simply put, this pre-
liminary analysis shed light on how many times the same word
was read more accurately (by different children) when it was
presented in context compared with when it was presented in
isolation. Reorganizing the data in this manner resulted in a less-
than-ideal data set that was very large (85 words #4 lists #2
forms of feedback !680 items) with very few observed values per
item (approximately six participants per condition). Nevertheless,
this exploratory inspection of the data revealed that when read with
feedback, 215/340 words were read more accurately in context,
106/340 words were read more accurately in isolation, and 18/340
words were read equally accurately in context and isolation. A
Wilcoxon signed-ranks test found these differences to be signifi-
cantly different (z!7.324, p".001). An uncannily similar
pattern was observed when the words were read without feedback:
213/340 words were read more accurately in context, 106/240
words were read more accurately in isolation, and 20/340 words
were tied. Once again, a Wilcoxon signed-ranks test found these
differences to be significantly different (z!8.202, p".001).
Although the original study was not designed with these analyses
in mind, the descriptive results suggest that the context effects are
not being driven by a small number of words.
Delayed Posttest
On the 18th day of each training block, the children were once
again presented with the training materials to measure retention of
learning. This follow-up task was administered to determine if the
reading gains made during training would be maintained over a
6-day period. As depicted in Figure 2, the general patterns favoring
context and feedback were also observed during the posttest (con-
text/feedback !93.98%, isolation/feedback !87.93%, context/no
feedback !81.45, isolation/no feedback !73.89). Furthermore,
the accuracy scores improved over the retention period, indicating
that the children treated the delayed posttest like an additional
training session. A 2 (context, isolation) #2 (feedback, no feed-
back) #2 (Session 5, posttest) repeated measures ANOVA con-
firmed that all three main effects were significant and corre-
sponded to large effect sizes (Cohen, 1988): context (F(1, 24) !
24.38, MSE !7,267.92, p".001, r!.71), feedback (F(1, 24) !
48.51, MSE !30,582.28, p".001, r!.82), and time of test (F(1,
24) !7.69, MSE !1,793.13, p!.011, r!.49). The absence of
any significant interactions (all ps%.29) suggested that the effects
of context and feedback were similar at both the end of training
and during the retention task.
Discussion
Successful decoding experiences contain two closely linked
components, each of which could be contributing to the formation
of orthographic representations in memory. The first factor in-
volves the act of decoding, which can only function if the reader
is attending to the orthographic details of the word. The second
factor involves the result of decoding, which happens when the
reader accesses the word’s spoken pronunciation. Thus far, the
lion’s share of research on self-teaching has been dedicated to
exploring the decoding process (the first factor), with fewer re-
sources directed at examining the importance of matching whole-
word phonology and orthography (the second factor). Granted,
when decoding activates the correct spoken word, these two fac-
tors become tightly woven and any additional contribution beyond
decoding is difficult to measure. However, when decoding is
ineffective—as it was for 30 to 40% of the words read in Session
1—it creates an opportunity to experimentally bolster the second
factor and measure resulting changes in word learning. In the
current experiment, self-teaching was tracked while children read
under two conditions—namely, context and feedback— hypothe-
sized to increase the availability of word pronunciations. If the
availability of whole-word phonology contributes to the creation
of word specific representations in memory, then higher learning
rates would be expected in these more supportive conditions.
When contemplating the two feedback conditions it becomes
apparent that reading was more accurate when words were pre-
sented in context compared with in isolation. From the second
session onward, the highest accuracy scores were observed when
children read with the benefit of both context and feedback; this
pattern remained unchanged over a 6-day delay. This may suggest
that generating and/or saying the pronunciation aids children more
than just hearing the correct pronunciation of the word after an
error. Such an interpretation fits nicely with the work of Ehri, who
has found accelerated learning rates when children were asked to
read words out loud compared with silently (see Ehri, 2014 for
review). Alternatively, it could suggest that children are simply
more successful when reading in context, and that this advantage
is maintained even with the additional support of feedback. The
present experiment is not able to tease apart these alternative
hypotheses, however, in either case, the data clearly show that
training is more effective when children have access to the words’
pronunciations, and that effortful grapheme-by-phoneme decoding
is not the only process for creating word representations in mem-
ory.
Context also helped children read more accurately when they
read without feedback. During the first no-feedback session, chil-
dren read 12.36 more words in context than they did in isolation.
This contextual benefit was largely maintained throughout the
duration of training. Indeed, it took children five sessions reading
in isolation without feedback (116.43 words) to achieve the same
degree of accuracy as they had on the first day when reading in
context without feedback (116.64 words). The benefits observed
from context also remained stable over a 6-day delay. Framing
these findings in terms of the two factors hypothesized to impact
self-teaching helps to explain how the top-down support from
context aids in long-term word learning. The contextual facilitation
effect describes how children use partial word decodings and
1
The same pattern of results was found when only words that were read
incorrectly in Session 1 were included in the analysis in a 2 (context,
isolation) #(feedback, no feedback) #4 (Sessions 2–5) ANOVA.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
238 MARTIN-CHANG

contextual constraints to improve children’s immediate word read-
ing performance. The byproduct of increased reading accuracy is
greater access to whole-word phonology and thereby, more oppor-
tunities to amalgamate speech and print, and hence, greater learn-
ing (Ehri, 2014;Share, 2004).
Thus far, it seems that reading with both forms of support
(context and feedback) offers greater benefits than reading without
either of these scaffolds (isolation without feedback). However,
what if only one form of support was available? Do children gain
more from reading in context (without feedback) or by reading
with feedback (out of context)? This is an interesting question
from a theoretical standpoint because of the fact that feedback
pairs a word’s verbal pronunciation with its written form on every
repetition (either the child reads the word correctly, or the pronun-
ciation is provided). In contrast, context allows a greater propor-
tion of words to be read correctly initially, but children will not
generate/hear the correct pronunciations of all of the words on
every trial. Therefore, if seeing and hearing words in close suc-
cession is a driving force behind gains in accuracy across sessions,
then receiving feedback should produce greater benefits than read-
ing in context. This prediction was born out in the current study.
Word reading in the isolation/feedback condition improved at a
faster rate, and ultimately surpassed the context/no feedback read-
ing scores.
The present study elected to use whole word feedback, which is
generally regarded as less helpful than graphophonemic feedback
during parent– child reading activities (Evans et al., 1998). The fact
that marked reading improvements were found after even the least
effective form of guidance was given, makes a convincing case for
the use of feedback. It should be noted, however, that the children
in the current study were in Grade 2 and they already had a basic
foundation of decoding skills. Therefore, these results might not be
replicated with a younger group of children.
Conjectures have been raised that independently generating pho-
nology from print may provide ideal circumstances for long-term
learning. A strong interpretation of this statement suggests that
corrective feedback may, in fact, be disadvantageous to reading
development (Landi, 2013). The data reported here do not support
this hypothesis. Children experienced the most difficulty when
they were reading in isolation without feedback. However, it is
worth highlighting that even under the least supportive conditions,
children were able to learn more than 12 words over and above the
104 words that were read correctly during the first session of
isolated-word reading without feedback. Therefore, the current
study adds to the literature (cf., Share, 1995,1999,2004) by
showing that children are capable of independently using recoding
skills to self-teach with the help of nothing more than pure decod-
ing. However, they learned more than triple as many words at the
end of training when they were given whole-word feedback in
isolation compared with when they were left to read in isolation
alone. These data suggest that hearing the spoken word soon after
an unsuccessful reading attempt may help children form partial
orthographic representations in memory that can be referenced and
refined on future word encounters.
When contemplating the role of feedback in this investigation it
is important to note that children were actively attempting to
decode the words before hearing the word pronunciations. The
same pattern of results would not be expected if the children did
not yet understand the alphabetic principle or if they were simply
“following along” while the teacher read the text. In this sense,
even with feedback, the main components of self-teaching remain
intact. That is, for feedback to result in lasting improvements, it is
hypothesized that the child must still attempt to recode the word in
relation to the print—in this case, the pronunciation provided by
the teacher may simply act as a catalyst to jumpstart future recod-
ing opportunities.
The findings reported here are not in complete agreement with
other studies in the self-teaching literature (e.g., Landi et al., 2006;
Nation, Angell, & Castles, 2007;Ricketts, Bishop, Pimperton, &
Nation, 2011;Wang et al., 2011, Experiment 1); however, there
are several differences between this study and those reported
previously that merit consideration. The first involves whether the
dependent variable is the total number of words that can be read
correctly (online performance) or the number of words learned.
During the present experiment, the same children were substan-
tially more successful when presented with words in context
compared with in isolation. However, because this benefit was
apparent on the first trial, the rate of learning did not differ
between the context and the isolation conditions. Therefore, the
condition that results in superior reading is a matter of opinion. On
the one hand, it could be argued that children learned just as many
words in isolation as they did in context. On the other hand, it
could also be said that presenting words in contexts allows chil-
dren to experience greater success while reading from the first
session onward. Both arguments are equally valid and supported
by the data.
Second, the role of context is most influential when reading is
difficult. Therefore, the nature of the materials is an important
factor with regards to context effects. As noted in the introduction,
several studies in this area have used very small training sets (e.g.,
eight items) and words that can be easily decoded (e.g., CVCe
words such as “yate”). When decoding is relatively simple it is
unlikely that children would need the extra support provided by
context. It is hypothesized that using a much larger word set—with
words that varied greatly in letter length and morphological com-
plexity—made the present study more sensitive to change, and
allowed the advantages of context to be observed.
A final consideration is whether spelling or reading is the
variable of interest. The current experiment focused on self-
teaching in relation to word reading rather than spelling. However,
there could be any number of different mappings between the
underlying orthographic representation and the way a word is
ultimately read. The child could read the word correctly and have
a high quality representation; conversely, the child could read the
word correctly and have a low quality or incomplete orthographic
representation (Martin-Chang, Ouellette, & Madden, 2014).
Therefore, the current findings cannot comment on whether the
benefits associated with contextual reading would generalize to
spelling. In fact, there is reason to believe that they may not. For
example, Kyte and Johnson (2006) found that reading accuracy
during their learning phase was not equally correlated with all of
the tasks that comprised the orthographic learning task. Rather, it
was most closely linked to the reading accuracy posttest measure
and least linked with the orthographic choice task. Given the
wealth of evidence showing that context does not aid (nor hinder)
orthographic learning, it could be possible that the two factors
involved in self-teaching (decoding and whole-word phonology)
have differential effects on reading and spelling development.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
239
FEEDBACK AND CONTEXT

Grapheme-to-phoneme decoding may have a more profound im-
pact on spelling development (suggesting higher quality ortho-
graphic representations in memory), while matching whole-word
orthography to phonology might suffice for reading accuracy but
not precise spelling (suggesting representations that may not be
fully specified). It is also important to consider that a child could
conceivably read a word incorrectly (e.g., read the word “chaos” as
“chase” or “city” as “kitty”) and still have formed a stable ortho-
graphic representation of that word in memory, albeit one that is
associated with an incorrect pronunciation.
Limitations and Future Directions
This study offers a different perspective on self-teaching be-
cause it observed how children learn to read a large bank of real
words over multiple exposures. In many respects the materials
were representative of those children encounter when they read
independently, specifically, they contained many words of varying
difficulty to be learned. However, using real words made it diffi-
cult to determine the proportion of items that were self-taught
during the first session compared with those that were previously
known. Therefore, it would be fruitful to replicate the procedures
discussed here with words that were known verbally, but not in
writing, to determine if the same patterns hold. In addition, the
ability to arrive at a correct pronunciation without a high quality
representation to support reading is partially driven by the prop-
erties of the words themselves. Therefore, examining the interac-
tion between context and word regularity may be insightful. Future
studies should consider using a cross-classified analytic approach
(see Kim, Petscher, Foorman, & Zhou, 2010) to examine the
contribution of both participant characteristics and word features
simultaneously.
A second limitation involves the shared reading paradigm used
during this study. Shared reading, while offering a high degree of
experimental control, does not mimic the independent reading that
generally happens when children self-teach. This manner of pre-
sentation might have felt awkward for good readers, who could
have read the whole text alone. It may also have maximized the
effects of context for poor readers, who might have struggled to
read the surrounding text independently. Therefore, this study
should be replicated when children are reading independently.
Conclusions
Orthographic learning has been attributed with providing read-
ers with fast, accurate, and long lasting access to written words; in
short, self-teaching is the mechanism that has been attributed with
improving reading fluency. The results from the present experi-
ment showed that children learned to read a number of words
without feedback in both context and in isolation, but that im-
provement during training was more pronounced (and equally
lasting) when feedback was provided. The data also showed that
children could read substantially more words accurately when they
were presented with words in context compared with in isolation.
The most advantageous of all the conditions was when children
were given the benefits of both context and feedback.
The conclusions drawn from the present study suggest that
generating (with the help of context) or being given a word’s
pronunciation (via feedback) aids in accurate decoding on future
word encounters. It would seem that children use all available
resources, including feedback and context, to amalgamate the
orthographic, phonological, and semantic properties of words as
they are learning to read.
References
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 Behavior Analysis, 27, 177–178. http://dx.doi.org/10
.1901/jaba.1994.27-177
Bowey, J. A., & 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. http://dx.doi.org/10.1016/j.jecp.2005.06.005
Byrne, B., Wadsworth, S., Boehme, K., Talk, A. C., Coventry, W., Olson,
R., . . . Samuelsson, S. (2013). Multivariate genetic analysis of learning
and early reading development. Scientific Studies of Reading, 17, 224 –
242. http://dx.doi.org/10.1080/10888438.2011.654298
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2
ed.). Hillsdale, NJ: Erlbaum.
Cunningham, A. E. (2006). Accounting for children’s orthographic learn-
ing while reading text: Do children self-teach? Journal of Experimental
Child Psychology, 95, 56 –77. http://dx.doi.org/10.1016/j.jecp.2006.03
.008
Cunningham, A. E., 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.
http://dx.doi.org/10.1016/S0022-0965(02)00008-5
Ehri, L. (2014). Orthographic mapping and the acquisition of sight word
reading, spelling memory and vocabulary learning. Scientific Studies of
Reading, 18, 5–21. http://dx.doi.org/10.1080/10888438.2013.819356
Ehrlich, S., & Rayner, K. (1981). Contextual effects on word perception
and eye movements during reading. Journal of Verbal Learning &
Verbal Behavior, 20, 641– 655. http://dx.doi.org/10.1016/S0022-
5371(81)90220-6
Elbro, C., de Jong, P., Houter, D., & Nielsen, A. (2012). From spelling
pronunciation to lexical access: A second step in word decoding. Sci-
entific Studies of Reading, 16, 341–359. http://dx.doi.org/10.1080/
10888438.2011.568556
Elgort, I., Perfetti, C. A., Rickles, B., & Stafura, J. Z. (2015). Contextual
learning of L2 word meanings: Second language proficiency modulates
behavioral and event-related brain potential (ERP) indicators of learning.
Language, Cognition and Neuroscience, 30, 506 –528.
Evans, M. A., Barraball, L., & Eberle, T. (1998). Parental responses to
miscues during child-to-parent book reading. Journal of Applied Devel-
opmental Psychology, 19, 67– 84. http://dx.doi.org/10.1016/S0193-
3973(99)80028-8
Field, A. P. (2009). Discovering statistics using SPSS (3rd ed.). London,
England: SAGE.
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 Associates.
Kim, Y. S., Petscher, Y., Foorman, B., & Zhou, C. (2010). The contribu-
tions of phonological awareness and letter-name knowledge to letter
sound acquisition—A cross-classified multilevel model approach. Jour-
nal of Educational Psychology, 102, 313–326. http://dx.doi.org/10.1037/
a0018449
Kyte, C. S., & Johnson, C. J. (2006). The role of phonological recoding in
orthographic learning. Journal of Experimental Child Psychology, 93,
166 –185. http://dx.doi.org/10.1016/j.jecp.2005.09.003
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
240 MARTIN-CHANG

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 context in developing word form repre-
sentations: A paradoxical relation between reading and learning. Journal
of Experimental Child Psychology, 94, 114 –133. http://dx.doi.org/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. http://dx.doi.org/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. http://dx.doi.org/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. http://
dx.doi.org/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.
http://dx.doi.org/10.1007/s11145-014-9502-7
Nagy, W., Anderson, R., & Herman, P. (1987). Learning word meanings
from context during normal reading. American Educational Research
Journal, 24, 237–270. http://dx.doi.org/10.3102/00028312024002237
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. http://dx.doi.org/10.1016/j.jecp.2006.06.004
Nation, K., & Snowling, M. J. (1998). Individual differences in contextual
facilitation: Evidence from dyslexia and poor reading comprehension.
Child Development, 69, 996 –1011. http://dx.doi.org/10.1111/j.1467-
8624.1998.tb06157.x
Ouellette, G. (2010). Orthographic learning in learning to spell: The roles
of semantics and type of practice. Journal of Experimental Child Psy-
chology, 107, 50 –58. http://dx.doi.org/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. http://dx.doi.org/10.1016/j.jecp.2009
.05.001
Ricketts, J., Bishop, D., 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. http://dx.doi.org/10
.1080/10888438.2011.536129
Roth, S. F., & Perfetti, C. A. (1980). A framework for reading, language
comprehension, and language disability. Topics in Language Disorders,
1, 15–28. http://dx.doi.org/10.1097/00011363-198012000-00004
Samuels, S. J. (1967). Attentional process in reading: The effect of pictures
on the acquisition of reading responses. Journal of Educational Psychol-
ogy, 58, 337–342. http://dx.doi.org/10.1037/h0020045
Share, D. L. (1995). Phonological recoding and self-teaching: Sine qua non
of reading acquisition. Cognition, 55, 151–218. http://dx.doi.org/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. http://dx.doi.org/10.1006/jecp.1998.2481
Share, D. L. (2004). Orthographic learning at a glance: On the time course
and developmental onset of self-teaching. Journal of Experimental Child
Psychology, 87, 267–298. http://dx.doi.org/10.1016/j.jecp.2004.01.001
Snowling, M. J., & Hulme, C. (2011). Evidence-based interventions for
reading and language difficulties: Creating a virtuous circle. British
Journal of Educational Psychology, 81, 1–23. http://dx.doi.org/10.1111/
j.2044-8279.2010.02014.x
Stanovich, K. E., Nathan, R. G., West, R. F., & Vala-Rossi, M. (1985).
Children’s word recognition in context; spreading activation, expec-
tancy, and modularity. Child Development, 56, 1418 –1428. http://dx.doi
.org/10.2307/1130461
Tunmer, W. E., & Chapman, J. W. (1995). Context use in early reading
development: Premature exclusion of a source of individual differences?
Issues in Education, 1, 97–100.
Tunmer, W. E., & Chapman, J. W. (2012). Does set for variability mediate
the influence of vocabulary knowledge on the development of word
recognition skills. Scientific Studies of Reading, 16, 122–140. http://dx
.doi.org/10.1080/10888438.2010.542527
Wang, H.-C., 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. http://dx.doi.org/10.1016/
j.jecp.2010.11.005
Wilkinson, G. S. (1993). The Wide Range Achievement Test. Wilmington,
DE: Wide Range.
Ziegler, J., Perry, C., & Zorzi, M. (2014). Modeling reading development
through phonological recoding and self-teaching: Implications for dys-
lexia. Philosophical Transactions of the Royal Society, B: Biological
Sciences, 369, 1–9.
(Appendices follow)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
241
FEEDBACK AND CONTEXT

Appendix A
Stimuli Lists and Word Frequency Counts Per Million Words
List 1 Frequency List 2 Frequency List 3 Frequency List 4 Frequency
After 682.59 About 3631.49 Admired 3.67 Away 730.9
Ahead 198.33 Above 48.88 Announced 5.67 Awful 63.41
Animals 39.08 Again 792.71 Arms 59.8 Bald 9.73
Behind 187.86 Always 655.25 Around 736.73 Because 1071.02
Below 28.04 Answered 15.2 Asked 216.25 Belt 24.35
Beneath 11.63 Beating 22.75 Baby 509.37 Boots 19.16
Board 64.16 Before 794.14 Back 2009.16 Boys 224.16
Both 295.33 Began 32.51 Bath 31.12 Branch 10.08
Bounced 2.96 Bird 45.45 Bathroom 61.67 Breeze 8.04
Bridge 45.71 Blackbird .82 Bathtub 6.1 Broke 105
Cars 45.63 Books 67.76 Bedroom 36.71 Called 340.02
Chuckled .08 Bright 44.41 Best 404.37 Camera 57
Close 219.43 Child 157.65 Bowls 2.18 Cannon 8.71
Coming 527.02 Chirping .98 Brand 13.96 Caught 93.94
Could 1629.59 Clever 27.27 Carried 20.12 Check 278.98
Crash 28.65 Confessed 5.96 Change 240.35 Climb 19.75
Cross 55.04 Cried 12.98 Clean 121.24 Couldn’t N/A
Deer 8.71 Demanded 2.76 Closely 9.18 Crack 32.84
Drove 28.86 Different 209.53 Continue 49.55 Crows 2.76
Elizabeth N/A Discovered 28.76 Crowded 8.94 Dead 448.98
Excited 48.61 Each 253.25 Declared 6.53 Decided 88.65
Fallen 16.92 Estimate 4.76 Door 292.06 Diving 6.29
Families 22.33 Ever 709.22 Enough 501.33 Down 1490.3
Fast 137.45 Every 549.16 Even 875.92 Edge 23.51
Father 554.49 Everything 654.88 Everyone 241.65 Feet 120.73
Filled 27.18 Forest 18.88 Family 354.25 Fell 73
Followed 34.1 Found 396 Felt 119.82 Firecrackers .71
Forward 72.33 Freeze 32.16 First 840.57 Five 285.45
Foxes 1.16 Guess 453.98 Good 2610.14 Flapping 1.67
Girls 208.35 Heart 244.18 Grandma N/A Fluttered .06
Glanced .63 Here 4525.25 Hall 51.94 Foot 64.92
Going 2123.29 Home 774.33 Just 4749.14 Getting 484.69
Group 73.76 Huge 48.37 Kept 89.39 Giggling 4.1
Growled .12 Imitate 1.8 Laughed 10.69 Grinned .24
Hands 236.53 Kind 590.69 Lots 60.16 Halfway 13.29
Happy 333.2 Knew 368.96 Louder 10.1 Head 371.51
Hated 28.22 Know 5721.18 Made 561.29 Help 921.12
Held 42.45 Languages 4.1 Making 222.53 High 195
Hello N/A Leaving 141.39 Metal 19.45 Hurry 173.65
Hike 6.53 Listening 62.84 Mind 484.61 Looked 120.9
Inched .02 Many 359.43 Modern 18.24 Make 1387.75
Included 7.49 Maple 3.24 More 1298.59 Mice 6.57
Jumped 21.14 Minute 377.49 Never 1362.55 Mayer N/A
Laughing 52.29 Myself 342.55 Nothing 853.61 Moment 187.04
Little 1446.39 Name 641.86 Ordered 36.96 Much 973.25
Lived 66.04 Nearby 8.33 Papa N/A Mustn’t N/A
Long 675.16 Nimbly .1 Picked 69.29 Nest 11.1
Moved 69.33 Nobody 266.65 Pitcher 3.24 Next 452.75
Named 69.88 Notice 59.25 Plastic 18.76 Only 1083.71
Need 1294.9 Often 57.35 Politely 1.71 Owls 2.12
Noises 7.16 Once 344.88 Pulling 27.14 Park 72.12
Nuts 53.51 Other 735.39 Quickly 56.49 Place 602.67
Onto 36.69 Professor 69.57 Screams 16.9 Quietly 12.33
Others 99.24 Realized 35.96 Seems 167.55 Ready 387.8
Popped 7.92 Right 4008.39 Shall 185.12 Realize 79.06
Possible 114.04 School 333.12 Shiny 7.8 Returned 24.76
Rabbits 6.43 Single 72.08 Should 1061.94 Scare 33.57
Replied 1.16 Size 46.14 Smiled 4.92 Screeching 2.55
(Appendices continue)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
242 MARTIN-CHANG

Appendix A (continued)
List 1 Frequency List 2 Frequency List 3 Frequency List 4 Frequency
Riverbed .43 Slid 1.84 Soapsuds N/A Seen 384.96
Rotten 17.47 Snickered .02 Splashing 1.1 Shots 28.37
Seats 21.76 Some 1727.24 Spouts .06 Sign 133.27
Seemed 54.25 Song 93.69 Started 187.57 Sitting 94.39
Shrill .47 Sound 143.39 Stepped 12.86 Small 124.96
Slowly 25.08 Stopped 75.37 Stick 97.12 Softly 4.73
Something 1500.16 Strange 86.43 Stood 25.78 Sounded 18.86
Sounds 156.27 Study 49.04 Stop 707.27 Sticks 13.61
Stacey’s N/A Suddenly 55.96 Sweater 13.8 Such 291.22
Start 340.1 Things 692.88 Thank 1115.24 Sure 1099.82
Stay 515.65 Those 753.02 Thick 13.98 Take 1891.04
Still 788.73 Thought 808.47 Though 181.94 Teacher 55.73
Story 220.78 Thousands 27.65 Towel 14.16 Teddy N/A
Sudden 33.47 Town 247.92 Trouble 223.55 These 904
Swamps .88 Tracks 16.75 Turned 105.65 Think 2691.39
Tested 10.53 Tried 186.84 Twisted 10.59 Tiptoe .88
Their 655.16 Unknown 15.18 Waiting 211.12 Told 699.59
Today 433.8 Ventured .47 Want 2759.18 Took 342.24
Together 383.39 Walk 215.86 Warm 52.14 Towards 27.43
Trail 19.2 Went 411.51 Warned 15.84 Tree 65
Troll 2.71 Whistle 15.45 Water 225.06 Trying 448.02
Unsure 1.02 Whose 62.49 Whimpered .08 Underfoot N/A
Very 1241.25 Wildly 1.92 Whined .1 Under 261.92
Voice 86.16 Woods 29.06 Will 2123.65 Warning 31.96
Who’s N/A Would 1767.88 Wooden 7.2 Whispered 2.02
Wondered 14.9 Written 44.06 World 455.22 Without 354.65
Wonderful 164 Young 243.18 Years 568.69 Yelled 6.14
Average 231.88 443.38 377.65 291.04
(Appendices continue)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
243
FEEDBACK AND CONTEXT

Appendix B
Example of a Training Story “Mary’s Race Day”
(
!
target words are those bolded and underlined in the story)
The thick door flung open and the children ran outside onto
Barton Street. They were all looking to see the shiny new wagon
with its modern metal frame. Mary’s was the best wagon in
years. The children admired its fancy plastic steering wheel and
its brand new wooden basket. They thought it was the fastest
wagon in the world. The children laughed and yelled louder as
more wagons crowded onto Barton Street.
Mary could hear the children’s screams from her bedroom. She
didn’t want to keep them waiting. She quickly rushed through the
hall into the bathroom for a bath. When Mary finished the bath,
her splashing had left water and soapsuds all over the bathtub.
She picked up a towel and ran it over the spouts and bowls. She
stepped back and admired the shiny bathtub and spouts.
“Finished,” she declared.
She left the bathroom and went into her bedroom. She smiled
while pulling on her warm thick sweater. Today was Mary’s first
wagon race. Nothing could change her wonderful mood. Mary
had never felt more prepared for anything. She stepped back into
the hall. Her towel was full of soapsuds so she carried it to the
plastic bowls. Then she went downstairs. In the kitchen she saw
that her father had picked lots of lemons. He twisted and turned
them until he made ababy pitcher of lemonade. Papa thought the
old fashioned lemonade was better than the new modern stuff
from a can. Mary loved it. She often drank a whole baby pitcher
herself. But today she politely declared that she was full and that
she should stop. Mary was careful to mind her manners and
thank her father. She knew she had the best family. Mary
wouldn’t trade them for the world. Soon everyone was set to go
so they left the house and locked the front door.
On the way to the race, Grandma asked Mary if she should
continue to stretch her arms and run around the block to warm
up her legs. Mary smiled. Her Grandma never ordered to do
things. She just politely suggested them.
“That seems like a good idea,” Mary announced, “I think I
shall try it.”
But when Mary started down the street she wanted to change
her mind. There were lots of people yelling for her and they were
louder than she could have imagined.
“Oh no,” Mary whined. As her family began going towards the
screams and crowded streets Mary’s eyes began to water.
“I’m Scared,” she whined.
“I know this seems difficult, but I have years of experience,”
her Papa warned “You will continue on alone but we shall
follow closely behind.”
Mary felt like pulling on his arms and making him come with
her but she knew that she was old enough to go alone.
Even though she was frightened Mary kept her head up as she
stood by her wagon. Her heart whimpered with fear as they were
ordered to get into their wagons. But she wasn’t waiting long.
Soon the gun fired and the race had started. Mary made aclean
getaway and she was quickly in first place. But then Tom, another
racer, scooted closely past her. Tom also had a brand new metal
wagon. He was making this an exciting race. As the road twisted
down the hill, Mary spotted trouble. There was a large wooden
stick in the middle of the road but there was nothing she could do.
Tom could not be warned.Hejust kept racing towards trouble.
His wagon came to a sudden stop as it hit the stick. Tom fell hard,
splashing into a puddle. Some of the other kids laughed as they
passed him but Mary didn’t want to leave him there. She turned
her wagon around and carried him off the road. She gave him her
sweater and helped him clean his cuts. Tom could not thank her
enough.
“Will you be alright?” Mary asked.
“Yes,” he whimpered, “thanks to you.”
At the end of the day, even though Mary lost the race, her good
deed was announced and everyone stood and clapped for her.
Received August 14, 2015
Revision received April 6, 2016
Accepted April 8, 2016 !
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
244 MARTIN-CHANG