Reading comprehension among typically developing Swedish-speaking 10-12-year-olds: examining subgroups differentiated in terms of language and decoding skills.

Abstract

Based on data from 156 typically developing 10-12-year-olds from Sweden, reading comprehension skills were studied in three subgroups: those classified with specific poor word decoding skills (n = 10), those with specific poor language comprehension (n = 12), and those with mixed difficulties in word decoding and language comprehension (n = 11). The mixed poor group achieved significantly lower scores than both specific groups in reading comprehension, and was the only group displaying poor reading comprehension test results relative to the performance of the full sample. Results are indicative of the necessity of a combined effect of poor word decoding and language in reading comprehension difficulties for this group. Implications and limitations are discussed.

Full text

Page 1

Correspondence: Jakob Å sberg, University of Gothenburg, Department of Psychology, Box 500, G ö teborg, 405 30, Sweden. E-mail: psyjaas@psy.gu.se
(Received 13 July 2009 ; accepted 4 May 2010 )
ORIGINAL ARTICLE
Reading comprehension among typically developing Swedish-speaking
10 – 12-year-olds: Examining subgroups differentiated in terms of
language and decoding skills
JAKOB Å SBERG 1 , MARIKA CARLSSON 2 , ANN-MARIE ODERSTAM 2 &
CARMELA MINISCALCO 2
1 University of Gothenburg, Department of Psychology, Box 500, G ö teborg, 405 30, Sweden, and 2 University of Gothenburg,
Institute of Neuroscience and Physiology, Division of Speech and Language Pathology, G ö teborg, Sweden
Abstract
Based on data from 156 typically developing 10 – 12-year-olds from Sweden, reading comprehension skills were studied in
three subgroups: those classifi ed with specifi c poor word decoding skills ( n ? 10), those with specifi c poor language com-
prehension ( n ? 12), and those with mixed diffi culties in word decoding and language comprehension ( n ? 11). The mixed
poor group achieved signifi cantly lower scores than both specifi c groups in reading comprehension, and was the only group
displaying poor reading comprehension test results relative to the performance of the full sample. Results are indicative of
the necessity of a combined effect of poor word decoding and language in reading comprehension diffi culties for this group.
Implications and limitations are discussed.
Key words: Language comprehension , listening , reading comprehension , school age , simple view of reading , word decoding
Introduction
Reading comprehension (RC) is the main goal of
reading and of great importance for general school
achievement. Prior research has established that RC
sometimes becomes a more evident challenge for
some children during the middle-school grades (i.e.
grades 4 – 6, or approximately 10 – 12 years of age)
when instruction shifts from ‘ learning to read ’ to
‘ reading to learn ’ (cf. (1)).
The most infl uential model of individual differ-
ences in RC is the ‘ simple view of reading ’ (2,3).
Gough and Tunmer (2) predicted that RC can break
down for three reasons according to the simple view:
because of specifi c diffi culties in either 1) word
decoding (the term ‘ word decoding ’ is used synony-
mously with ‘ word recognition ’ in the present text),
or 2) language/linguistic comprehension skills, or 3)
due to a combined weakness of both sets of skills.
While much research supports the ‘ simple view of
reading ’ at a general level (4,5), data appear not
wholly consistent with the predictions made by
Gough and Tunmer, at least not for readers after the
primary school years. For example, if they were pro-
vided with suffi cient time to decode the text, adults
with near-average oral vocabulary comprehension
but below average word decoding skills have been
shown to reach normal levels of RC (6). Thus, at
least adult word decoding can be slow and laborious
and still ‘ good enough ’ to allow the successful extrac-
tion of meaning from written text. This fi nding is also
compatible with interactive-compensatory models of
reading (cf. (7,8)), in which semantic and word
decoding-related processes interact, and where weak-
nesses in one area (e.g. phonology) sometimes can
be compensated for if other processing resources
(e.g. comprehension) function normally. However,
this may not apply equally to children. Bishop et al.
recently demonstrated that a group of 9 – 10-year-
olds with dyslexia, who presented with severe word
decoding problems but adequate oral language com-
prehension, achieved lower scores than typically
developing readers in RC (9). This result indicates
that specifi c word decoding diffi culties are enough to
hamper RC among children.
Logopedics Phoniatrics Vocology, 2010; Early Online, 1–5
ISSN 1401-5439 print/ISSN 1651-2022 online © 2010 Informa UK, Ltd.
DOI: 10.3109/14015439.2010.491518
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Goteborgs University on 08/22/10
For personal use only.

Page 2

2 J. Å sberg et al.
Research on children with oral language diffi cul-
ties has shown that literacy diffi culties are very com-
mon in this group (cf. (8)). While these children
often appear to have mixed diffi culties in both word
decoding and RC, a subgroup of children — so-called
‘ poor comprehenders ’ — has also been identifi ed.
These children present with poor RC and poor oral
language comprehension in spite of intact word
decoding skills (e.g. (9)). In most previous research,
participating children with poor comprehension have
been recruited based on low screening results in RC
and not on low language comprehension results. It
is therefore not well known if children with poor oral
language comprehension typically experience poor
RC even if their word decoding skills are found to be
adequate. In addition, there is a need to examine these
questions in non-English speaking samples as there
are some reasons to assume that the profi le of normal
word decoding but diffi culties in oral comprehension
might be more common among children reading more
shallow orthographies (such as Swedish) as compared
to deeper ones (such as English). This hypothesis fol-
lows from research of English-speaking children (10),
where poor oral vocabulary has been found to be asso-
ciated with impaired reading of irregularly spelled
words but not directly with non-word reading. Hence,
meaning-based processing appears to be an important
component of ‘ sight ’ word recognition. As there is a
greater proportion of irregularly spelled words in writ-
ten English than in almost all other languages, this
would seem to mean that there could be a greater
chance for an overlap between oral comprehension
diffi culties and word decoding diffi culties among Eng-
lish-speaking children and greater chances for the dis-
sociation of these sets of skills among for example
Swedish-speaking children. Theories of literacy should
be tested across different languages before generaliza-
tions are made (11).
The present study was set up to examine if spe-
cifi c poor performances in word decoding or oral
language comprehension skills among Swedish
10 – 12-year-olds are independently suffi cient to pro-
duce low RC, or if RC diffi culties follow only when
poor word decoding and poor oral comprehension
skills co-occur.
Method
Participants and instruments
Assessment was done by two master students in
logopedics (the second and third author) as part of
the collection of norm data for the Swedish version
of Test for reception of grammar — version 2 (TROG-2)
(12). Ten- to 12-year-olds were recruited from
grades 4 to 6 in schools in counties at the outskirts
of Gothenburg. The selected counties were equiva-
lent to the average in Sweden in terms of mean
income level for the year of assessment (2006).
Information about the study was sent out to a total
of 213 families. Written informed consent from par-
ents was necessary for participation in the study. In
addition, to ensure a representative sample for the
non-disabled population, on-going special needs
education or speech and language therapy due to
language problems were exclusion criteria. Further-
more, all participating children needed to have
Swedish as their fi rst language or judged by their
parents to have skills in Swedish that were equally
good or better than a second language. This resulted
in a total of 156 children meeting the inclusion cri-
teria for the study (85 girls, 71 boys, mean chrono-
logical age 11.06 years). None of the children were
placed in classes for children with intellectual dis-
abilities or autism.
Participants were assessed in groups for word
decoding using the Word Chains Test (13) and RC
using the SL40 test (14). The Word Chains Test is
widely used in Sweden and has proven to be a fast
and effi cient indicator of word decoding/recognition,
showing high correlations with other measures of the
construct (13). Jacobson reports a test-retest reliabil-
ity of 0.85 for fi fth-graders (approximately 11 years
old). An English version of this test is available (15).
The task in the Word Chains Test is to mark with a
pencil where divisions should be made in a chain of
words without inter-word blank spaces (e.g. ‘ car-
housetree ’ ). All words in the test were expected to be
in the participants ’ spoken language vocabulary and
of high frequency in the Swedish language. The num-
ber of correctly solved chains in two minutes was
registered. In the current sample, the mean raw score
on the Word Chains Test was 18.7, with a standard
deviation (SD) of 6.1.
The SL40 is a test of RC that is commonly used
in clinics, schools, and research in Sweden and
Denmark (e.g. (16,17)). The task in SL40 is to
identify the one picture out of a choice of four that
matches a target sentence or short story for mean-
ing. Especially the later items involve more com-
plex grammatical constructions. The test consists
of 40 items, and the number of correctly solved
items in 15 minutes was registered. The mean raw
score on the SL40 reading comprehension task was
37.0 (SD 3.6).
All participants were also individually assessed in
a quiet room at their schools for oral language com-
prehension using TROG-2. The task in TROG-2 is to
identify the one picture out of a choice of four that matc-
hes a target sentence for meaning. The items are of
an increasing diffi culty. The test has been widely used
internationally in the fi eld of language and literacy
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Goteborgs University on 08/22/10
For personal use only.

Page 3

Word decoding and language comprehension
3
skills and diffi culties in children (e.g. (9)). Children
were given all items in the test booklet, and the num-
ber of correctly solved blocks, out of a maximum of
20, was registered. The mean raw score on the
TROG-2 test was 16.6 (SD 2.0).
Study groups and statistical analyses
Test raw scores were age-corrected by entering test
results as dependent variables and age as indepen-
dent variable in three separate regression analyses
according to the procedure described by White et al.
(18). The unstandardized residuals were used as the
corrected scores. These corrected scores were con-
verted into z -scores to allow identifi cation of children
with poor performance. A z -score of zero represents
the group mean. A z -score of -1 represents a perfor-
mance one standard deviation (SD) below the mean.
Descriptives and correlations between test results are
reported in Table I. Non-parametric Spearman cor-
relations were used to test relations among measures
due to a generally high performance on the compre-
hension measures, and particularly so for RC, with
a tendency towards ceiling levels. These distributions
thus violated the assumptions regarding normality.
Positive and signifi cant correlations were obtained
among the measures.
To form study groups the following criteria
were used:
1. Specifi c poor word decoding group : More than 1
SD below the mean in word decoding, but no
more than 0.25 SD below the mean in oral lan-
guage comprehension.
2. Specifi c poor language comprehension group : More
than 1 SD below the mean in oral language com-
prehension, but no more than 0.25 SD below
the mean in word decoding.
3. Mixed poor group : To get a large enough sample
in this group, cut-off was set to more than 0.67
SD below the mean in word decoding, and more
than 0.67 SD below the mean in oral language
comprehension. This cut-off represents the 25th
percentile in the normal distribution. (It is
also of importance in the design adopted that
the mixed poor group is not more impaired in
either word decoding or oral language compre-
hension in relation to the matched specifi c
group.)
A total of 33 children out of 156 typically developing
children met criteria for one of these groups: 10 (6.4%)
for the Specifi c poor word decoding group , 12 (7.7%)
for the Specifi c poor language comprehension group , and
11 (7.1%) for the Mixed poor group.
Group differences were evaluated with analysis of
variance (ANOVA) with Tukey post hoc . A two-tailed
signifi cance level of P ? 0.05 was adopted. Group-wise
comparisons were also evaluated with non-parametric
Mann-Whitney U tests due to the non-normality of
the data set. As the pattern of results was identical, only
P values from parametric post-hoc tests are reported
here.
Background data on the study groups and results
from overall group comparisons as per ANOVA are
presented in Table II. Children with Specifi c poor word
decoding did not differ from the Mixed poor group in
word decoding ( P ? 0.63) but achieved as expected
higher scores in oral language comprehension ( P ?
0.001). Children with Specifi c poor oral comprehension
did not differ from the Mixed poor group in oral lan-
guage comprehension ( P ? 0.95), but achieved as
expected higher scores in word decoding ( P ? 0.001).
The two groups with specifi c diffi culties differed, in
Table I. Test descriptives and Spearman correlations in the whole
sample ( n ? 156) of 10 – 12-year-old children.
z -score
min-max1.2.3.
1. Language
comprehension
2. Reading
comprehension
3. Word decoding
−4.71 – 1.97
0.421 b 0.221 a
−5.14 – 1.11
0.295 b
−3.05 – 3.55
a P ? 0.01.
b P ? 0.001.
Correlations are tested two-tailed.
Table II. Background data from three groups of children.
Group
Measure Poor word decoding
Poor language
comprehension Mixed poor group F value/ P (df ? 2, 30)
n
Chronological age
Word decoding
Language comprehension
10 1211
11.89 (0.72) a
−1.50 (0.58) a
0.34 (0.25) a
11.58 (1.00) a
0.38 (0.43) b
−1.60 (0.52) b
11.22 (0.84) a
−1.29 (0.56) a
−1.69 (1.15) b
1.6/ P ? 0.22
44.3/ P ? 0.001
24.5/ P ? 0.001
Test results are expressed in age-corrected z -scores (standard deviations in brackets) based on performance of the normative sample
( n ? 156).
a,b Means in the same row that do not share subscripts differ at P ? 0.05 according to Tukey post hoc .
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Goteborgs University on 08/22/10
For personal use only.

Page 4

4 J. Å sberg et al.
different directions, in word decoding ( P ? 0.001)
and oral language comprehension ( P ? 0.001). This
design allows us to elucidate necessary and suffi cient
conditions for RC diffi culties.
Results
Both groups with specifi c diffi culties had RC results
close to the normative z -score of zero ( Specifi c poor
word decoding : M ? 0.03, SD ? 0.89; Specifi c poor
language comprehension : M ? 0.11, SD ? 0.54). In
contrast, the Mixed poor group scored more than one
standard deviation below the normative mean in RC
( M ? -1.11, SD ? 1.72). The result from an ANOVA
indicated an overall group difference in RC, F (2, 30)
? 3.9, P ? 0.031. Post-hoc tests revealed that the
mixed poor group scored signifi cantly below the two
specifi c groups (both P ? 0.05), who in turn did not
differ statistically from each other ( P ? 0.99).
Discussion
By necessity there must be a limit where skills in
either oral language comprehension or word decod-
ing are so low that they independently will constrain
RC, e.g. for children with hyperlexia (cf. (19)) or
severe dyslexia (cf. (20)). However, this research on
subgroups of mainstream Swedish 10 – 12-year-olds
showed that a specifi c weakness in either oral lan-
guage comprehension or word decoding seemed not
suffi cient to cause low RC results. Only if children
had low test scores in both sets of skills did poor RC
follow, suggesting the presence and necessity of a
combined effect in our study samples. Importantly,
this pattern of results could not be explained by dif-
ferences in the degree of severity of word decoding/
oral language comprehension diffi culties in the spe-
cifi c groups versus the mixed poor group as these
background variables were tightly controlled for.
The present study strongly suggests that we must
consider and support children ’ s literacy development
also in the context of their oral language comprehen-
sion development (8). The results also have conse-
quences for assessment. In Sweden, the Word Chains
Test and SL40 are often used in schools to identify
children at risk for literacy diffi culties. Presumably,
much might be won by including a test of oral lan-
guage comprehension in such screening.
Findings also indicated that 10 – 12-year-old
children are able to compensate for word decoding
diffi culties during RC if their oral language com-
prehension skills are adequate. One possible mech-
anism allowing such compensation was studied by
Nation and Snowling (21), namely the facilitative
usage of semantic context during word recognition.
One would expect RC to be more successful if
poor decoders are able to benefi t by top-down sup-
port from semantic context in the process of iden-
tifying individual words. This might have been the
case for our Specifi c poor word decoding group . Like-
wise, our results are consistent with an idea that
this compensation mechanism was not available to
the Mixed poor group .
The fi nding of profi cient RC performance in the
Specifi c poor word decoding group was in contrast to
the results reported in some previous research (9).
Potentially, differences between studies could be due
to the age of the participants or to the severity of
their decoding problems, but it also appears possible
that the choice of test to assess RC could play a role.
For example, several English studies on RC have
used the Neale Analysis of Reading Ability instru-
ment (22). In this instrument, the child is asked com-
prehension questions only on text passages on which
he or she has performed few decoding error s. As
children with word decoding diffi culties often make
more such errors they have not always been given the
chance to answer all comprehension questions (see
(21), for a discussion of this issue). In the present
study separate tests of word decoding and RC were
used, which appears to be a less biased procedure.
However, a limitation also in our study must be men-
tioned in this context. The performance in RC
approached ceiling levels in the larger sample. This
may in turn have infl ated the performance of our
study groups. On the other hand, it would appear to
affect all three groups equally.
How can one understand the result that children
with specifi c oral language comprehension diffi cul-
ties managed to extract linguistic meaning when pre-
sented with written materials? This result seems
unintuitive considering that text is a written form of
oral language. One possible explanation is that the
written texts were not as linguistically challenging as
the oral language test constructions. This seems likely
on inspection of test items, and a more challenging
RC test may have been diffi cult for the Specifi c poor
language comprehension group . But as an explanation
of our fi ndings, this would not seem to fully account
for the poorer performance of the Mixed poor group .
Another possibility is that written materials allow for
a closer scrutiny by its ‘ lasting features ’ as compared
to speech. Indeed, a large body of research has dem-
onstrated that performance improves for children
with language impairments when the presentation
rate of the linguistic material is lowered (23). Poten-
tially, the Specifi c poor language comprehension group
drew upon such a ‘ scrutiny ’ -strategy during RC,
which thereby allowed them to extract the texts ’
meaning. In contrast, such a strategy was likely not
within reach to the same extent for children with the
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Goteborgs University on 08/22/10
For personal use only.

Page 5

Word decoding and language comprehension
5
combination of poor language comprehension and
poor word decoding skills, as evident in the weaker
RC performance seen in the Mixed poor group .
All in all, the present study shows that the combi-
nation of low performance in word decoding and oral
language comprehension is particularly problematic
for RC performance. Our interpretation is that this
might be due to a blocking of compensation mecha-
nisms or strategies that are usually available to typi-
cally developing children of this age during the process
of meaning-making from written text. Obviously, there
could also be other differences between the specifi c
groups and the mixed group that have not been con-
sidered in the present study. Hence, further studies are
needed that look more into if and how multiple risk
factors, inside and outside the child, interact to cause
RC diffi culties. In addition, we proposed a language-
bound perspective in the introduction. Specifi cally, we
put forward the idea that there may be a greater chance
for an overlap between oral language comprehension
diffi culties and word decoding diffi culties among Eng-
lish-speaking children, and greater chances for the dis-
sociation of these sets of skills among children reading
more shallow orthographies, such as Swedish. While
this hypothesis could formally be put to test only in a
comparative cross-linguistic study, it can be noted
here that we had no diffi culties in identifying sub-
groups with specifi c problems in our study sample.
This relative mismatch between language comprehen-
sion and word decoding was also evident in a rather
low (albeit signifi cant) correlation between these two
measures in the total sample.
The current study was a small-scale examination
of subgroups of typically developing 10 – 12-year-olds.
Results in the present study should be considered ten-
tative until they have been replicated in other, and
larger, samples of Swedish-speaking children. Such
replications need to include a broader range of tests of
reading and oral language abilities, which would raise
the reliability of the subgrouping. This is perhaps par-
ticularly important for the outcome measure, RC.
Recent studies have demonstrated that different RC
tests appear to measure partly different skills and that
the ‘ same ’ test can measure different things depending
on age and ability (cf. (4,24)). Preferably, a range of
different RC tests should be used in future research.
Declaration of interest: The authors report no
confl icts of interest. The authors alone are respon-
sible for the content and writing of the paper.
References
Leach JM, Scarborough HS, Rescorla L. Late-emerging
reading disabilities. J Educ Psychol. 2003;95:211–24.
Gough PB, Tunmer WE. Decoding, reading, and reading dis-
ability. Remedial and Special Education. 1986;7:6–10.
1.
2.
Hoover WA, Gough PB. The simple view of reading. Read
Writ. 1990;2:127–60.
Cutting LE, Scarborough HS. Prediction of reading com-
prehension: Relative contributions of word recognition,
language profi ciency, and other cognitive skills can depend
on how comprehension is measured. Sci Stud Read. 2006;
10:277–99.
Catts HW, Adlof SM, Weismer SE. Language defi cits in poor
comprehenders: A case for the simple view of reading. J
Speech Lang Hear Res. 2006;49:278 – 93.
Lesaux NK, Pearson MR, Siegel LS. The effects of timed
and untimed testing conditions on the reading comprehen-
sion performance of adults with reading disabilities. Read
Writ. 2006;19:21 – 48.
Perfetti CA, Roth S. Some of the interactive processes in
reading and their role in reading skill. In: Lesgold AM, Per-
fetti CA, editors. Interactive processes in reading. Hillsdale,
NJ: Lawrence Erlbaum; 1981. p. 269–97.
Bishop DVM, Snowling MJ. Developmental dyslexia and
specifi c language impairment: Same or different. Psychol
Bull. 2004;130:858–86.
Bishop DVM, McDonald D, Bird S, Hayiou-Thomas ME.
Children who read accurately despite language impairment:
Who are they and how do they do it? Child Dev. 2009;80:
593–605.
Ricketts J, Nation K, Bishop DVM. Vocabulary is important
for some, but not all reading skills. Sci Stud Read. 2007;11:
235 – 57.
Harris M, Hatano G, editors. Learning to read and write: A
cross-linguistic perspective. New York: Cambridge University
Press; 1999.
Bishop DVM. Test for Reception of Grammar — Version 2
(TROG-2). San Antonio, TX: Harcourt Assessment; 2003.
Jacobson C. L ä skedjor. Stockholm. Psykologif ö rlaget AB;
2005.
Hellquist B, Magnusson E, Naucl é r K. Ordl ä sningsprov
OS64 och OS120 Meningsl ä sningsprov SL60 och SL40.
Skivarp: Pedagogisk Design; 1997.
Miller-Guron L. Wordchains. Windsor: NFER-Nelson; 1999.
Frost J, Madsbjerg S, Neiders ø e J, Olofsson Å , S ø rensen, P.
Semantic and phonological skills in predicting reading devel-
opment: From 3 – 16 years of age. Dyslexia. 2005;11:79 – 92.
Naucl é r K, Magnusson E. Reading and writing development:
Report from an ongoing longitudinal study of language-
disordered and normal groups from preschool to adoles-
cence. Folia Phoniatr Logop. 1998;50:271–82.
White S, Frith U, Milne E, Rosen S, Swettenham J, Ramus
F. A double dissociation between sensorimotor impairments
and reading disability: A comparison of autistic and dyslexic
children. Cogn Neuropsychol. 2006;23:748–61.
Grigorenko EL, Volkmar F, Klin A. Hyperlexia: Disability or
superability? J Child Psychol Psychiatry. 2003;44:1079–91.
H ø ien T, Lundberg I. Dyslexia: From theory to intervention.
Dordrecht: Kluwer Academic Publishers; 2000.
Nation K, Snowling MJ. Individual differences in contextual
facilitation: Evidence from dyslexia and poor reading com-
prehension. Child Dev. 1998;69:996–1011.
Neale, MD. Neale analysis of reading ability. 2nd ed. Wind-
sor, UK: NFER-Nelson; 1997.
Campbell TF, McNeil MR. Effects of presentation rate and
divided attention on auditory comprehension in children
with an acquired language disorder. J Speech Lang Hear Res.
1985;28:513–20.
Keenan JM, Betjemann RS, Olson RK. Reading comprehen-
sion tests vary in the skills they assess: Differential depend-
ence of decoding and oral comprehension. Sci Stud Read.
2008;12:281–300.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
Logoped Phoniatr Vocol Downloaded from informahealthcare.com by Goteborgs University on 08/22/10
For personal use only.