The Causal Role of Phoneme Awareness and Letter-Sound Knowledge in Learning to

Read: Combining Intervention Studies With Mediation Analyses

Author(s): Charles Hulme, Claudine Bowyer-Crane, Julia M. Carroll, Fiona J. Duff and
Margaret J. Snowling
Source: Psychological Science , JUNE 2012, Vol. 23, No. 6 (JUNE 2012), pp. 572-577
Published by: Sage Publications, Inc. on behalf of the Association for Psychological
Science

Stable URL: https://www.jstor.org/stable/41489742

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Sage Publications, Inc. and Association for Psychological Science are collaborating with JSTOR
to digitize, preserve and extend access to Psychological Science

This content downloaded from

156.38.120.41 on Fri, 04 Nov 2022 11:41:22 UTC
All use subject to https://about.jstor.org/terms
 яшшшт aps I ASSOCIATION FOR
яшшшт I ASSOCIATION FOR

Research Report

Psychological Science
23(6) 572-577
The Causal Role of Phoneme Awareness ©TheAuthor(s) 2012
Reprints and permission:

and Letter-Sound Knowledge in Learning sagepub.com/journalsPermissions.nav

DOI: 1 0.1 177/095679761 1435921

to Read: Combining Intervention Studies

http://pss.sagepub.com

USAGE
With Mediation Analyses

Charles Hulme1, Claudine Bowyer-Crane2, Julia M.Carroll3,

Fiona J. Duff4, and Margaret J. Snowling4
'University College London, Sheffield Hallam University, 3University of Warwick, and
4University of York

Abstract

There is good evidence that phoneme awareness and letter-sound knowledge are reliable longitudinal predictors of le
read, though whether they have a causal effect remains uncertain. In this article, we present the results of a mediat
using data from a previous large-scale intervention study. We found that a phonology and reading intervention
letter-sound knowledge and phoneme awareness produced significant improvements in these two skills and in later
reading and spelling skills. Improvements in letter-sound knowledge and phoneme awareness at the end of the interv
mediated the improvements seen in children's word-level literacy skills 5 months after the intervention finished. O
support the conclusion that letter-sound knowledge and phoneme awareness are two causal influences on the deve
of children's early literacy skills.

Keywords
reading, academic achievement, cognitive development, letter-sound knowledge, phoneme awareness

Received 1 0/21/ 1 1; Revision accepted 12/20/1 1

It is important to base interventions for disorders see

on aHulme,
clear Caravolas, Malkova, & Brigstocke, 2005, and
theory of the nature and causes of that disorder (Hulme
Hulme,& Snowling, Caravolas, & Carroll, 2005, for a different
Snowling, 2009). It is also generally accepted that the view). Similarly, it is well established that letter knowledge is
best form
of evidence for addressing causal theories comes from random-
a key independent predictor of the development of word read-
ized experiments that address putative causal processes (Foster,
ing, although this does not necessarily imply that letter knowl-
edge trials
2010). In this article, we show that combining randomized causes improved reading skill; some researchers have
with mediation analyses is a powerful way of advancing theo-
argued that letter knowledge is better considered an indicator
retical understanding of the causes of developmental of disorders.
parental support in literacy or of visual- verbal learning abil-
To illustrate this, we reanalyzed data from a previouslyity (Foulin, 2005).
reported
large-scale randomized trial for children's reading disorders
Castles and Coltheart (2004) argued that the ideal way to
(Bowyer-Crane et al., 2008). test these causal hypotheses would be to train children on a
It is now widely accepted that there is a close relationship
key skill (phoneme awareness or letter knowledge) in isolation
between learning to read and children's phonological skills
and monitor the effect of such training on their later word
(see Bowey, 2005, and Melby-Lervâg, Lyster, & Hulme, reading.
2012, Studies of this type generally show small effects
for reviews). Specifically, it has been argued that children's
(for discussions with respect to phoneme awareness and letter
ability to isolate and manipulate phonemes in spokenknowledge,
words is see Castles & Coltheart, 2004, and Piasta &
one causal influence on the development of word-reading skills
(Muter, Hulme, Snowling, & Stevenson, 2004). Conversely,
Corresponding Author:
other researchers have argued that phoneme-manipulation skills
Charles Hulme, Division of Psychology and Language Sciences, University
may be a consequence rather than a cause of variations in chil-
College London, Chandler House, 2 Wakefield St., London, England
dren's reading skills (Castles & Coltheart, 2004; Morais, Cary,
WCIN2PF

Alegria, & Bertelson, 1979; Ziegler & Goswami, 2005;

E-mail:but
c.hulme@ucl.ac.uk

This content downloaded from

156.38.120.41 on Fri, 04 Nov 2022 11:41:22 UTC
All use subject to https://about.jstor.org/terms
Learning to Read 573

Wagner, addition
2010, to ind
respectively).
that phonemeknowledge
awareness and l
and p
for a true understanding of
standing how letters in printe
Method
in spoken words), and this sug
skills in In the present study,
isolation would we reanalyzed datanot
from Bowyer-Crane
be
ing early reading skills.
et al. (2008). We will describe In
only those aspects of lin
the method
training of letter
and measures that are criticalknowledg
for understanding the analyses
reading tendreported;
to full show
details of the study arelarger
provided in the earlier
jointeffects paper.
of these differen
(Hatcher, Hulme, & Ellis, 199
Abetter approach to examining
in training
Participants
studies is arguably
(Baron & Kenny, 1986;
Participants consisted of 152 children whoMacKi
had an average age
A mediation model
of 5 years 0 months at the examines
beginning of the study. These chil- h
outcome are transmitted
dren were recruited from 19 schools on the basis of their via poor
struct. In theverbal study
ability. From each school, reported
we selected children with the
remedial lowest verbal composite
reading intervention score derived from the Vocabulary t
phoneme and Verbal Reasoning subscales
awareness and of the third edition of the
letter
influences onWechsler
reading Preschool and Primary Scale of Intelligence, United
developm
the theory Kingdom edition
that the (WPPSI-IIIUK; Wechsler, 2003); on average,
improvem
duced by thisthe children selected had scaled scores of 6.42 (Vocabulary) w
intervention
phoneme and 7.62 (Verbal Reasoning).and
awareness The 8 children selected
letter in each
In Bowyer-Crane school were randomly et assigned al.
to the P+R (2008)
or OL programs.
two randomly assigned interv
oral language skills at schoo
received an
Design
intervention prom
letter-sound Children
knowledge,received 20 weeks of intervention delivered in either
along
of simple a group (30 min)with
books or an individual session (20
amin)teachi
on alternat-
reading, or ing days. The P+Rgroup);
P+R, program combined training in letter-sound
the o
program knowledge and oral phoneme oral
targeting awareness with booklangua
reading.
and narrative The OL skills;
program focused on vocabulary the
development, narra-
oral-
end of 20 weeks
tive structure, andof speaking and interventi
listening skills. Both programs
the OL group were delivered
inby the same teaching assistants working in the seg
phoneme
sound children's schools. Children and
knowledge, were assessed four times: before
measur
contrast, thethe OL intervention began
group(Time 1), after 10 weeks of intervention
was ah
ety of oral (Time 2), after 20 weeks of intervention
language measures. (Time 3), and 5
tions had months after the intervention was completed
differential effects (Time 4). o
differences were maintained 5 m
had finished. Following our t
reading made
Tests
by
and procedure
the P+R grou
of the Each child was
training assessed using a broad battery
these of tests of spo-
children
edge and ken
phoneme language, phonological skills, and
awareness, literacy.
examined in the present study
Mediation General cognitive ability. To
models assess general cognitiveresea
allow ability
ing whether at
an Time 1, we administered
intervention the Block Design subtest of the i
tors that WPPSI-IIIUK.
account for the interv
the current study is unique in
learning Phoneme awareness in
disorder . At Time 3,the
phoneme blending, seg-
contex
P+R mentation, and deletion tasks
intervention wasfrom the Test of Phonological
effecti
because of Awareness
its (Hatcher,
impact 2000) were used to assess children's
on abil- phon
knowledge, ity
then to segment and blend
we words. expected
no further direct effects on
variables wasLiteracy
takenmeasures . Four measures of into
literacy were adminis- accou
tive tered. To assess
hypothesis is letter-sound
that knowledge, we asked children to
there w
the say the sounds of 24 letters
intervention would (Times 1 and 2) or 26 lettershave
(Times d

This content downloaded from

156.38.120.41 on Fri, 04 Nov 2022 11:41:22 UTC
All use subject to https://about.jstor.org/terms
 574 Hulme et al.

3 and 4) of
Because causal effects must operatethe forward in time, we a
assessedused measures
at of phonemeall fou
awareness (the phoneme blending,
test (Hatcher et
segmentation, and deletion tasks) and letter-sound knowledge at al
consecutive erro
Time 3 as possible mediators of the intervention effect on word-
(Snowling, Stoth
level literacy skills at Time 4. The analyses were conducted as a
Time 4 series as a
of path models in Mplus (Version meas
6. 1 ; Muthén & Muthén,
presenting
2010) using robust maximum-likelihood
childre estimators to account
"book," "heart";
for the fact that some measures were not normally distributed.
plus "chick," "tr
To allow for the nonindependence of observations arising from
Words the clustering of children within prese
were classrooms, we used robust
(Huber- White) standard errors. The small amount of missing
data was handled by full-information maximum-likelihood esti-
Results
mators (the default in Mplus).
Analyses reported in Bowyer-Crane et al. (2008) showed that
We adopted an iterative approach to developing the final
mediation model (shown in Fig. 1). We first assessed the
the P+R program produced selective improvements on pho-
neme segmentation and blending, letter-sound knowledge,
effects of the intervention on the composite literacy outcome
spelling, prose-reading accuracy, and nonword reading at at Time 4. Because intervention group was dummy-
measure
coded (0 = OL group, 1 = P+R group), model estimates in
Times 3 and 4. Here, we present analyses that focused purely
on literacy scores at Time 4 and the extent to which they can only the dependent variable was standardized expressed
which
be attributed to differences in underlying skills produced
thebymean difference between groups on the outcome measure
the P+R intervention at Time 3. Literacy at Time 4inwas z-score units. In the absence of any covariates, the differ-
assessed with a composite measure of word-level literacy
ence in literacy between the intervention groups was signifi-
cant, d = 0.49, 95% confidence interval = [0.27, 0.71], p <
formed by averaging z scores for early word reading, nonword
reading, and spelling (r = .65 for early word reading and.001.
non-We next added initial levels of reading (scores on the
Early
word reading, r = .73 for early word reading and spelling, and Word Reading test at Time 1) and nonverbal ability
r = .66 for nonword reading and spelling). Descriptive statis-
(scores on the Block Design subtest at Time 1) as covariates
tics for the measures used are shown in Table 1, and correla-
(to control for baseline differences in reading ability and gen-
tions among measures are shown in Table 2. eral cognitive ability on the outcome measures). Finally, we

Table I. Descriptive Statistics for Key Variables Included in the Mediation Model

Phonology-and-reading group Oral-language group

Variable M SD Range M SD Range

Age (months) at Time I 60.60 3.56 55-67 59.84 3.31 54-67
Block Design score at 6.92 2.98 1-15 6.86 3.23 1-16
Time I

Early Word Reading score at 4.88 7.00 0-33 3.04 3.55 0-18
Time I
Phoneme-awareness score at 7.48 4.79 0-17 4.05 3.53 0-12
Time 3

Letter-sound knowledge 23.92 3.14 11-26 22.19 4.82 5-26

score at Time 3

Spelling score at Time 4 2.96 2.42 0-10 1.97 1.40 0-5

Graded Nonword Reading 4.26 5.72 0-20 2.03 4.07 0-20
Test score at Time 4

Early Word Reading score at 27.07 12.16 2-42 22.72 10.52 0-41
Time 4

Literacy composite score 0.23 1 .0 1 - 1 .27-2.35 -0.2 1 0.69 - 1 .30- 1 .77

at Time 4

Note: Block Design (Wechsler, 2003) and Early Word Reading (Hatcher, Hulme, & Ellis, 1994) tests were
administered before the intervention began (Time I). Phoneme awareness (Hatcher, 2000) and letter-sound
knowledge were measured after 20 weeks of intervention (Time 3). The spelling test, the Graded Nonword
Reading Test (Snowling, Stothard, & McLean, 1996), and the Early Word Reading test were administered 5
months after the intervention was completed (Time 4); the literacy composite was based on scores from
those tests.

This content downloaded from

156.38.120.41 on Fri, 04 Nov 2022 11:41:22 UTC
All use subject to https://about.jstor.org/terms
 Learning to Read 575

Table 2. Correlations Among

Intervention group Ear

Variable (dummy-coded) at Time I Time I at Time 3 knowledge at Time 3

Early Word Reading at .18 - - - -

Time I

Block Design at Time I .02 .14 - - -

Phoneme awareness at .38 .36 .31 - -
Time 3

Letter-sound knowledge .2 1 .30 .32 .40 -

at Time 3

Literacy at Time 4 .24 .53 .37 .71 .57

Note: Early Word Reading (Hatcher, Hulme, & Ellis, 1994)

(Time I), phoneme awareness (Hatcher, 2000) and letter-so
was measured 5 months after the intervention was comp

assessed the extent to effect

which on the fit group differences
of the model (Satorra-Bentler scaled x2 differ- in
on Time-4 literacy were mediated
ence = 0.86, n.s.). byof changes
Furthermore, the strength the mediated
paths from intervention
awareness and letter-sound knowledge via phoneme awareness
at Time and letter- 3.
els assessed each of these
sound knowledge
mediators
to literacy did notseparately;
change reliably when the th
ined their joint effects. nonsignificant path was dropped.
The final model showed that the P+R intervention was The effects of the intervention on literacy scores at Time 4
associated with differences in both phoneme awareness
can
and
be completely accounted for by levels of the mediators
letter-sound knowledge at Time 3. These differences, in(letter-sound
turn, knowledge and phoneme awareness) at Time 3.
were associated with reliable influences on literacy scores
Bothatof the compound indirect paths (intervention - > pho-
Time 4. Note that in this model, the direct path from theneme
inter-awareness at Time 3 - > literacy at Time 4 and interven-
vention to literacy at Time 4 was not significant and was tion -> letter knowledge at Time 3 - > literacy at Time 4) were
there-
statistically reliable. In terms of effect sizes, we derived an
fore dropped. All retained paths in this model were statistically
significant; dropping the nonsignificant direct path fromestimate
inter- of the size of the mean difference between the two
vention to literacy at Time 4 had no statistically significant
groups in the literacy outcome that was associated with these

Time 1 Time 3 Time 4

Intervention Literacy ^ n
Group (0,1 ) Outcome

Early Word s' 0.27

Reading ^^05' У
Phoneme 0.70 У
ãõ* I . ?• Aw>reness /
Design | . ' 0.04 0 33 ^ Л
Letter- J
Sound Q
Knowledge

Fig. I. Path model showing the effects of a phonology and reading intervention on children's literacy
scores, as mediated by the intervention's effects on phoneme awareness and letter-sound knowledge.
At Time I, before the intervention, children's single-word reading and general cognitive ability (Block
Design subtest of the third edition of the Wechsler Preschool and Primary Scale of Intelligence;
Wechsler, 2003) were assessed. Phoneme awareness and letter-sound knowledge were tested at Time
3, after 20 weeks of intervention. Literacy was assessed at Time 4, 5 months after the intervention
was completed. The oral-language group was coded as 0 in the model, and the phonology-and-reading
group was coded I. All path weights are standardized on the outcome (Y) variable only.

This content downloaded from

156.38.120.41 on Fri, 04 Nov 2022 11:41:22 UTC
All use subject to https://about.jstor.org/terms
 576 Hulme et al.

indirect Our effects

findings have direct implications for reading instruc-
by
tion and methods of remedial
dependent teaching. The conclusion that
variabl
between phonemic skills and letter-sound knowledge are two causal
groups
z-score influences
units). on the development of reading skills leads to Th rec-
was equal to
ommendations that these 0.4
skills should be directly taught to all s
0.3 1 standard dev
children in the early stages of learning to read. Arguably, such
via phoneme aw
structured teaching may be particularly critical for children,
.027) wassuch as those in attribut
the present study, who are struggling in the
concerns early stages of about
learning to read. th
violations This study focused on the of no
mechanisms of reading develop-
sessed the reliab
ment and reading difficulties. However, the method we used to
strapped standar
test a causal theory is potentially widely applicable in other
results areas.
(p Combining mediation
< analyses .001 with randomized trials a
Finally, note
testing theoretically tha
motivated interventions is a powerful way
phoneme awarene
of advancing the theoretical understanding of the causes of psy-
cated that these
chological disorders. Mediation models allow researchers to go t
tion between th
beyond evaluating whether an intervention is effective to test
specified
theories about thein the
factors responsible for the effects observed. m
lent fit to the da
fit Acknowledgments
index = 1.00,
error of approxim
We thank Arne Lervâg for statistical advice.

Declaration of Conflicting Interests

Discussion
The authors declared that they had no conflicts of interest with
In the study reported here, we tested the theory that respect
the devel-
to their authorship or the publication of this article.
opment of word-level literacy skills is causally influenced by
References
children's early letter knowledge and phoneme awareness. A
phonology and reading intervention (Bowyer-Crane et R.
Baron, al.,
M., & Kenny, D. A. (1986). The moderator-mediator vari-
2008) that directly taught letter-sound knowledge andablepho-distinction in social psychological research: Conceptual,
neme awareness produced significant improvements instrategic
these and statistical considerations. Journal of Personality
two skills and in early word-level literacy skills. Moreand
strik-
Social Psychology , 51, 1173-1182.
ingly, levels of letter-sound knowledge and phoneme aware-
Bowey, J. A. (2005). Predicting individual differences in learning to
read.
ness measured at the end of the intervention fully mediated theIn M. J. Snowling & C. Hulme (Ed.), The science of read-
improvements seen in the children's word-level literacying:
skills
A handbook (pp. 153-172). Oxford, England: Blackwell.
measured 5 months later. Given that children in this study
Bowyer-Crane, C., Snowling, M. J., Duff, F. J., Fieldsend, E.,
were randomly assigned to two interventions, we haveCarroll,
good J. M., Miles, J., & Hulme, С. (2008). Improving early
evidence that the improvements seen in letter-sound knowl-
language and literacy skills: Differential effects of an oral lan-
edge, phoneme awareness, and literacy skills were guage
causalversus a phonology with reading intervention. Journal
effects. Viewed in this light, the results of the mediationof
model
Child Psychology and Psychiatry , 49, 422-432. doi: 10.1 111/
provide strong support for the theory that weaknesses inj.l469-7610.2007.01849.x
letter-
sound knowledge and phoneme awareness are two causes Byrne,ofB. (1998). The foundation of literacy: The child's acquisitions
difficulties in mastering word-level literacy skills (since dif-
of the alphabetic principle. Hove, England: Psychology Press.
ferences at Time 3 in letter-sound knowledge and phoneme
Castles, A., & Coltheart, M. (2004). Is there a causal link from pho-
awareness can be attributed to differences between the two nological awareness to success in learning to read? Cognition,
interventions that the children received). 9 1,11-111.
In alphabetic writing systems, letters (graphemes) are Foster, E. M. (2010). Causal inference and developmental psychol-
used to represent phonemes, and it has been argued that chil- ogy. Developmental Psychology, 46, 1454-1480. doi: 10.1037/
dren need to master the alphabetic principle in order to learn a0020204
to read effectively (Byrne, 1998). To master the alphabetic Foulin, J. N. (2005). Why is letter-name knowledge such a good pre-
principle, children must understand the mappings between dictor of learning to read? Reading and Writing, 18, 129-155.
letters in printed words and the phonemes in spoken words. Hatcher, P. J. (2000). Sound linkage (2nd ed.). London, England:
To do this, they need to possess phonemically structured rep- Whurr.
resentations of speech, as well as letter-sound knowledge.
Hatcher, P. J., Hulme, С., & Ellis, A. W. (1994). Ameliorating early
Our study provides strong support for a theory that early reading failure by integrating the teaching of reading and phono-
reading development depends critically on phonemic skills logical skills: The phonological linkage hypothesis. Child Devel-
and letter-sound knowledge. opment, 65, 41-57.

This content downloaded from

156.38.120.41 on Fri, 04 Nov 2022 11:41:22 UTC
All use subject to https://about.jstor.org/terms
Learning to Read 577

Hulme, С., Developmental

Caravolas, M., Mal
Phoneme 1649.40.5.665
isolation ability is no
sound Muthén, L. K., & Muthén,
knowledge. В. О. (2010). Mplus user s guide (6th ed.). ,
Cognition
Hulme, С., & Los Angeles, CA: Author.
Snowling, M. (200
ders. Piasta,
Oxford, S. В., & Wagner, R. K. (2010). Developing
England: Blackwe early literacy
Hulme, С., skills: A meta-analysis of alphabet
Snowling, M. learning J.,
and instruction.
Ca
Phonological Reading Research Quarterly,
skills are 45, 8-38.(probab
ing to read: APreacher, K. J., & Hayes, A. F. (2008). Asymptotic
comment on and resampling
Cast
ies of Reading,
strategies9, 351-365.
for assessing and comparing indirect effects in multi-
MacKinnon, D.ple mediator
P., models.
Fairchild,
Behavior Research Methods, 40, 879-891.
A.
analysis. Snowling, M. J., Review
Annual Stothard, S. E., & McLean, M. (1996).
of Graded Psy
Non-
Melby-Lervâg, M.,
word Reading Test. Bury Lyster, S.
St. Edmunds, England: Thames Valley
logical skills Test Company. their
and role in
review. Wechsler, D. (2003). Wechsler PreschoolBulletin
Psychological and Primary Scale of Intel-
Morais, J., ligence- Third
Cary, UK EditionAlegria,
L., (WPPSI-Ilf K). San Antonio, TX: J
ness of speechPsychological
as Corp. sequence
a of
nition, 7, Ziegler, J. C., & Goswami, U. (2005). Reading acquisition, de-
323-331.
Muter, V., velopmental dyslexia,
Hulme, С., and skilled reading across languages: A
Snowling
nemes, rimes,psycholinguistic
vocabulary grain size theory. Psychological Bulletin,
and 131,
of early 3-29.
reading development: E

This content downloaded from

156.38.120.41 on Fri, 04 Nov 2022 11:41:22 UTC
All use subject to https://about.jstor.org/terms