See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/325745588
What is the relationship between reading prosody and reading
comprehension in European Portuguese? Evidence from grades 2 to 5
Article in Journal of Research in Reading · June 2018
DOI: 10.1111/1467-9817.12248
CITATIONS
READS
14
639
4 authors, including:
Luis Querido
Luisa Araujo
Egas Moniz School of Health and Science
Instituto Superior de Educação e Ciências
26 PUBLICATIONS 309 CITATIONS
68 PUBLICATIONS 457 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
THE STRUCTURE OF THE SPEECH LEXICON IN THE ABSENCE OF LITERACY View project
Phonological Awareness View project
All content following this page was uploaded by Luisa Araujo on 14 September 2018.
The user has requested enhancement of the downloaded file.
SEE PROFILE
Journal of Research in Reading, ISSN 0141-0423
Volume 00, Issue 00, 2018, pp 1–28
DOI:10.1111/1467-9817.12248
What is the relationship between reading
prosody and reading comprehension in
European Portuguese? Evidence from
grades 2 to 5
Sandra Fernandes
, Luís Querido
and Arlette Verhaeghe
Faculdade de Psicologia, Universidade de Lisboa, Alameda da Universidade,
Lisbon, Portugal
Luísa Araújo
Instituto Superior de Educação e Ciências, Alameda das Linhas de Torres, Lisbon,
Portugal
The relevance of reading prosody on reading comprehension has been theoretically
proposed and empirically recognised. The present study aimed to investigate longitudinal and concurrent relationships between reading prosody and reading comprehension processes, beyond decoding and word reading efficiency. Two cohorts of
Portuguese children, ranging different grade levels (grades 2 to 3 and 4 to 5), were
tested. Reading prosody was assessed through a rating scale, and text reading comprehension and context fluency effect were used to measure effortful and effortless reading
comprehension processes, respectively. Results showed no bidirectional longitudinal
effects between reading prosody and reading comprehension processes. In contrast,
they revealed a concurrent contribution of reading prosody to the measure that mainly
captures the effortless processes of reading comprehension from grades 2 to 5 but not
for effortful ones. Results also highlighted the key role of decoding and word reading
efficiency in the relationship between reading prosody and reading comprehension.
Keywords: decoding, reading prosody, context fluency effect, reading comprehension
Highlights
What is already known about this topic
• Prosody is an important component of oral reading fluency.
• Decoding is crucial for reading prosody and reading comprehension.
• Reading prosody is associated with reading comprehension.
What this paper adds
• The relation between reading prosody and reading comprehension may depend on grade level and may vary according to orthographic depth.
© 2018 UKLA. Published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4 2DQ,
UK and 350 Main Street, Malden, MA 02148, USA
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
• Prosodic reading has a concurrent predictive role on the effortless processes of
reading comprehension.
• There is no bidirectional links between reading prosody and reading comprehension in European Portuguese until fifth grade.
Implications for theory, policy or practice
• Regarding the relationship between reading prosody and reading comprehension, theory must take into account, beyond the grade level, the consistency
level of the orthographic system.
• Instructional programmes must take into account the important role of
decoding for achieving efficient reading.
• Learning to read in European Portuguese would benefit from reading texts for
practice in reading fluency with prosody.
There is agreement that the definition of reading fluency must include three components:
accuracy, rate and prosody (Hudson, Lane, & Pullen, 2005; Hudson, Pullen, Lane, &
Torgesen, 2009; Kuhn & Stahl, 2003; National Reading Panel, 2000). However, different
theoretical perspectives put different emphases on these three components of reading fluency when considering their relationship with reading comprehension. While some authors
(e.g., Fuchs, Fuchs, & Maxwell, 1988; Torgesen, Rashotte, & Alexander, 2001) emphasise
accuracy and rate in reading fluency, others (e.g., LaBerge & Samuels, 1974; Samuels,
2006) claim that the three components involved in reading fluency – rate, accuracy and
prosody – indicate that decoding and comprehension are occurring simultaneously. Daane,
Campbell, Grigg, Goodman, and Oranje (2005) consider the prosody of oral reading as the
most important part of reading fluency because it reflects comprehension.
Prosody in oral language refers to the ‘rhythmic and tonal aspects of speech: the “music”
of oral language’ (Hudson et al., 2005, p.704), and prosodic reading has been defined as
the appropriate expression or intonation in reading (e.g., Hudson et al., 2005; Kuhn,
Schwanenflugel, & Meisinger, 2010) and, more specifically, as the ability to read text
quickly, accurately and with proper expression (e.g., Kuhn et al., 2010). Thus, prosodic
text reading involves the use of prosodic features (e.g., phrasing, pauses, stress and intonation) of spoken language during reading. According to Kuhn and Stahl (2003), prosodic
features support not only the oral comprehension of speech-language but also reading comprehension. They stated that reading prosody might assist reading comprehension because
it presupposes that the student has segmented text according to major syntactic and semantic elements. In fact, there is evidence that, when students are provided with information
about syntactic and semantic boundaries, their comprehension improves (e.g., O’Shea &
Sindelar, 1983). Given that the development of reading prosody occurs once decoding becomes fluent (LaBerge & Samuels, 1974; Perfetti, 1985), Schwanenflugel, Hamilton,
Kuhn, Wisenbaker, and Stahl (2004) suggested that ‘prosody may serve to mediate between decoding skills and comprehension to enhance comprehension’ (p. 122). This mediation may reflect the importance of syntactic processing for reading comprehension (e.g.,
Kuhn & Stahl, 2003), because prosody can assign to words their syntactic roles within
sentences (e.g., Koriat, Kreiner, & Greenberg, 2002). Another possibility is that prosody
can provide a scaffold for reading comprehension by allowing speech sequences to be held
in working memory and subsequently analysed for further processing (e.g., Frazier,
Carlson, & Clifton, 2006). This scaffolding can help readers to understand sentences and
© 2018 UKLA
READING PROSODY AND READING COMPREHENSION
texts even when accurate and fluent reading is not yet fully established (Benjamin &
Schwanenflugel, 2010). Anyway, it is well established that only when accurate word recognition becomes faster and more efficient, and thus more fluent, the student’s resources can be
directed toward the reading comprehension processes (LaBerge & Samuels, 1974; Perfetti,
1985), and that until there, the focus is mainly on decoding. Furthermore, it is widely accepted that the weight of decoding on reading comprehension decreases during reading development (e.g., Fernandes, Querido, Verhaeghe, Marques, & Araújo, 2017), but as far as
reading prosody is concerned, little is known about its differential weight on reading comprehension along learning to read. However, it has recently been proposed that the relation between reading prosody and reading comprehension is dynamic and depends on reading
level and the characteristics of the language under study (e.g., Veenendaal, Groen, &
Verhoeven, 2016a). Although scarce, there is evidence supporting this statement.
As a matter of fact, several studies using a cross-sectional or concurrent design, either in
English (e.g., Benjamin et al., 2013; Benjamin & Schwanenflugel, 2010; Kuhn & Stahl,
2003; Miller & Schwanenflugel, 2006) or in more transparent languages (in Spanish, Calet,
Gutiérrez-Palma & Defior, 2015; and in Dutch, Veenendaal, Groen, & Verhoeven, 2014,
2015), have consistently established that reading prosody affects reading comprehension
in a positive way. The same type of influence was also obtained in studies using a longitudinal design, in English (Miller & Schwanenflugel, 2008) and in Dutch (Veenendaal,
Groen, & Verhoeven, 2016b).
Firstly, concerning the cross-sectional studies, those conducted by Schwanenflugel and
colleagues (Benjamin et al., 2013; Benjamin & Schwanenflugel, 2010; Miller &
Schwanenflugel, 2006), with English-speaking children, showed that in second and third
grades, prosody accounted for variance in reading comprehension beyond reading skill. With
Spanish children and taking into account their grade level, Calet et al. (2015) also intended to
examine the contribution of prosodic reading to comprehension beyond automaticity in word
reading (nonword reading and reading rate). Their results showed a strong contribution of
automaticity to reading comprehension both in second and fourth grades but more significantly in the former one. However, the opposite pattern was observed for prosodic reading,
which was a better predictor of reading comprehension in fourth grade than in second grade.
In their study carried out with Dutch fourth graders, Veenendaal et al. (2014, 2015) showed
in the same way that reading prosody explained variance in reading comprehension scores,
after controlling for, for example, decoding efficiency and language comprehension.
Regarding longitudinal studies, Miller and Schwanenflugel (2008) examined the contribution of reading prosody in first-grade and second-grade English speaking children, to reading
comprehension outcomes in third grade. They found that decreases in the number of pausal
intrusions between the first and the second grades and the emergence of an early adult-like
intonation contour predicted better reading comprehension in the third grade. Also using a
longitudinal design, Veenendaal et al. (2016b) found a contribution from reading prosody
in fourth-grade and fifth-grade Dutch children to reading comprehension in sixth grade.
Thus, this set of findings from cross-sectional and longitudinal studies indicates that
reading prosody is helpful for reading comprehension either in opaque or transparent orthographies. Importantly, one of these studies also showed that the strength of this effect
may vary depending on the grade level (Calet et al., 2015).
While keeping the assumption that reading prosody promotes and improves reading comprehension, Schwanenflugel et al. (2004) further propose that increased comprehension might
allow the student to read with prosody. The relationship between reading prosody and reading
comprehension could therefore be reciprocal. Regarding this issue of the eventual
© 2018 UKLA
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
bidirectional link between reading prosody and reading comprehension, empirical data seem
not to provide unanimous conclusions. As a matter of fact, in a study with English-speaking
children from grades 2 and 3 and using suprasegmental features (e.g., intonation patterns,
stress placement and rhythm), Schwanenflugel et al. (2004) found that reading prosody predicted (although minimally) reading comprehension but not the reverse. In another study with
English fifth graders, Klauda and Guthrie (2008) used a rating scale measure of reading prosody and found that changes in ratings of syntactic processing (i.e., expressive oral reading)
were reciprocally related to longitudinal changes in reading comprehension (measured early
and late in grade 5) even after controlling for word reading speed and autoregressive effects.
Aiming to test the Klauda and Guthrie’s (2008) hypothesis, Lai, Benjamin,
Schwanenflugel, and Kuhn (2014) carried out a short-term longitudinal study with
English-speaking second graders (over three time points). Using structural equation modelling, they examined the bidirectional effects between reading fluency (a latent variable including reading prosody) and reading comprehension. Their results showed evidence for a
unidirectional effect of the reading fluency measure which includes reading prosody on
reading comprehension but not for a bidirectional one.
More recently, using also a rating scale measure of reading prosody, Veenendaal et al.
(2016a) investigated these relationships between reading prosody and reading comprehension
in a longitudinal study with Dutch children followed from fourth to sixth grades. They took into
account both the autoregressive effect and the ‘decoding efficiency – the fast and accurate retrieval of the phonological code for written words’ (Veenendaal et al., 2016a, p. 3). Their results
suggested a reciprocal relation between text reading prosody and reading comprehension: in
fifth grade, text reading prosody was related to prior decoding and reading comprehension,
whereas in sixth grade, reading comprehension was only related to prior text reading prosody.
In short, the aforementioned studies examining the reciprocal relationship between reading prosody and reading comprehension (in English: Klauda & Guthrie, 2008; Lai et al.,
2014; Schwanenflugel et al., 2004; and in Dutch: Veenendaal et al., 2016a) reached different conclusions probably due to the fact they used different methods (e.g., different control
variables and reading fluency and reading comprehension measures) and also students in
different grades as participants.
Actually, analysing those previous studies (Klauda & Guthrie, 2008; Lai et al., 2014;
Schwanenflugel et al., 2004; Veenendaal et al., 2016a) allowed ascertaining that they did
not always consistently control the effects of decoding skills and word reading efficiency
and/or the autoregressive effect. We therefore considered that the inclusion of such variables as control ones might decrease and even eliminate the effect of reading prosody on
reading comprehension and vice versa. In fact, a significant contribution of decoding fluency (considered either as a latent measure comprising word and pseudoword reading or
as a measure of pseudoword reading) on the relation between reading prosody and reading
comprehension was observed (e.g., Schwanenflugel et al., 2004; Veenendaal et al., 2016a)
when it was included as a control variable. In this way, the contribution of reading prosody
to reading comprehension seems actually mediated by the efficiency in pseudoword and
word reading, which would also allow prosodic reading.
Therefore, this effect of decoding can be a reason why the strength of the relationship
between reading prosody and reading comprehension could change according to the grade
level and also according to orthographic consistency. As it is well known, the consistency
of alphabetic orthographies has a large degree of variation (e.g., Seymour, Aro & Erskine,
2003). Consistent orthographies have a total or a great majority of 1:1 relationships between graphemes and phonemes compared with inconsistent ones. English orthography
© 2018 UKLA
READING PROSODY AND READING COMPREHENSION
has the most inconsistent mapping system, and thus, for English readers, decoding is more
influential to reading comprehension during the early stages of learning to read, than for
more transparent orthographies (refer to Florit & Cain, 2011, for a review). By the same
way, the attainment of efficiency of decoding ability takes considerably longer in English
(refer to Landerl & Wimmer, 2008), and for this reason, the relationship between reading
prosody and reading comprehension would arise later in this opaque orthography than in
transparent ones (e.g., Dutch). In European Portuguese, an orthography with an
intermediate level of consistency (refer to Fernandes, Ventura, Querido, & Morais, 2008;
Sucena, Castro, & Seymour, 2009 for a characterisation of the orthography), the nature,
strength and directionality of the link between reading prosody and reading comprehension
may prove to be different from those found in other orthographies (e.g., English and
Dutch). Besides, to our knowledge, there is no study addressing this relationship from
the beginning (second grade) of reading development to a much later stage of consolidation
of reading ability (fifth grade). In the present study, we thus aimed examining these issues
in European Portuguese, an orthography of intermediate depth, taking into account reading
development.
From the previously mentioned theoretical assumptions of Kuhn and Stahl (2003) and
Schwanenflugel et al. (2004), we hypothesised that prosody assists reading comprehension
and that reading comprehension allows reading prosody, that is, that there is an interactive
relation between reading prosody and reading comprehension processes. In addition, we
expected that this reciprocal relationship varies as a function of the stage reached on learning to read, increasing in step with reading development.
Traditionally, reading comprehension is measured by means of a text reading comprehension (TRC) task (e.g., Calet et al., 2015) that mainly captures effortful processes. Nevertheless, according to Kintsch (1988), a crucial step in reading comprehension is the
identification of meaningful relations, which can be identified either by a quick, effortless
and automatic process that usually proceeds without the reader being aware of it or by a
slow, effortful and strategic process that requires conscious attention by the reader. While
some strategic effortful processes can, with practice, become effortless and automatic, both
of these two types of processes (effortless and effortful) take place during reading
comprehension. Thus, textual reading comprehension involves not only the
unintentional/automatic activation of the meanings and roles of words but also (and perhaps
mainly) the effortful processing of surface and deep meanings, which requires intensive use
of working memory capacities, executive processes, activation of schemes and retrieval of
world knowledge (Kintsch, 1988). Thereby, in the present study, in addition to a traditional
TRC task, we also included a measure designed to reflect and capture in a purest way the
effortless and unintentional processing of meaning and syntactic structure crucial to sentence comprehension and, hence, to text comprehension. This measure, called context fluency effect (CFE), consists of the difference between performance on text reading fluency
and word reading fluency. It reflects the specific facilitation that the processing of textual
properties and the processing of neighbouring words in phrases may have on reading fluency, that is, the part of text fluency that is not determined by the reading of words in random
order but benefits from the effortless processing of the syntactic and semantic cues present
in the sentences of connected text. In fact, it is widely recognised that words in context are
read faster than the same words out of context (e.g., Jenkins, Fuchs, van den Broek, Espin,
& Deno, 2003; Stanovich, 1980). As put forth by Jenkins et al. (2003), ‘context reading fluency depends to a considerable degree on pure (context-free) word recognition skill, but it is
also influenced by processes that originate in context’ (p. 720).
© 2018 UKLA
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
As the reciprocal relationship between reading prosody and reading comprehension can
only be examined using tests administered with significant time lags in between them, longitudinal effects were examined using cross-lag sequential regression analysis (SRA) with
autoregressive controls. This methodology of inclusion of an autoregressive control has
been used in developmental research to establish a bidirectional relationship between
two variables (e.g., Deacon, Benere, & Castles, 2012), allowing for the extraction of a
unique effect of a variable at an earlier time on another at a later time. To our knowledge,
only the studies of Klauda and Guthrie (2008) and of Veenendaal et al. (2016a) used this
autoregressive methodology to study the relationship between reading prosody and reading
comprehension, but they did not incorporate the all set of control variables. Thus, to overcome this limitation, we included, in the present study, controls for multiple variables before the autoregressive control (e.g., age, nonverbal reasoning, word fluency [WF] and
pseudoword fluency [PF]) to reduce the possibility that any effects could be due to a spurious third variable (refer to Kenny, 1975) or due to decoding or word reading efficiency.
Furthermore, we also sought to determine whether individual differences in reading
prosody could account for reading comprehension at the same moment of reading development, using separate SRAs for concurrent relationships (within the same test period at each
grade level). In line with previous research, we expected a predictive effect of reading prosody on reading comprehension (e.g., Klauda & Guthrie, 2008; Miller & Schwanenflugel,
2006, 2008; Schwanenflugel et al., 2004).
Method
Participants
A total of 170 children from two public schools in the Lisbon district participated in the
study. Two cohorts were tested in three testing periods, more precisely throughout the initial 3 months and the final 3 months of the school year in one grade and in the last 3 months
of the subsequent grade: cohort 1 was tested in grades 2 and 3, respectively; cohort 2, in
grades 4 and 5. Depending on the specific test, a minimum of 80 and 75 children participated in the last testing period for cohort 1 and 2, respectively.
Students came from two different classrooms per grade in each school. They were all native speakers of European Portuguese and had an average or above average cognitive functioning – 25th percentile or higher on the Raven’s Coloured Progressive Matrices (Simões,
1994). Children flagged as having learning, emotional or sensory disabilities as referred by
their parents, and teachers were not included in the sample. Given these screening criteria,
15 and 13 children were discarded from participating in cohort 1 and 2, respectively. Data
from a questionnaire answered by the children’s parents (accounting for 74.3% of the participants) indicated that only 6.4% and 4.7% of the fathers and mothers, respectively, did
not have at least high school education, 11.7% (for both) graduated from high school,
38.6% and 48.5% graduated from the university, and, additionally, 9.9% and 7.6% had
some post-graduate education (including 2.9% and 2.3% with a PhD). Only one father
(and no mother) was reported as unemployed. Our sample is thus not representative of
the Portuguese population on the whole but overwhelmingly constituted by children from
middle and upper middle class families.
The distribution of participants tested by grade and gender (F and M) are presented in
Table 1.
© 2018 UKLA
92.55
15.96
0.96
64.74
27.44
0.85
36.12
Ratec
Reading prosodyd
Word fluency
(accuracy)
Ratec
Context fluency effect
Pseudoword
fluency (accuracy)
Ratec
11.34
0.10
18.32
17.27
0.03
4.22
26.63
0.03
0.19
SD
0.67
M
0.20
SD
45.07
0.88
27.41
77.16
0.97
17.13
104.68
0.97
M
End
12.36
0.07
18.24
18.56
0.04
3.54
29.20
0.02
SD
7.78**
13.55**
2.52*
0.28
10.22**
1.01
1.17
7.44**
1.18
t
0.67
34
M
0.22
2.37
SD
Beginning
51.42
0.88
33.12
87.02
0.98
18.62
119.82
0.97
M
End
F = 42; M = 45
N = 87
Grade 4
Note.
a
Raw score.
b
The proportion of correct responses was obtained from the total of questions (24), chance level for complete test was 0.25.
c
Number of correct items per minute.
d
Global rating score, maximum of 24.
*p < .05.
**p < .01.
0.97
M
Text fluency
(accuracy)
4.16
SD
0.52
29
M
Text reading
comprehension
(accuracy)b
Non-verbal reasoning
a
Beginning
F = 37; M = 44
F = 37; M = 46
End
N = 81
N = 83
Beginning
Grade 3
Grade 2
Cohort 1
11.73
0.07
17.78
15.21
0.02
3.54
21.99
0.03
SD
0.86
M
0.13
SD
Beginning
Cohort 2
52.22
0.89
35.82
95.33
0.98
19.01
131.57
0.97
M
End
12.71
0.07
19.14
15.67
0.02
3.17
22.22
0.02
SD
F = 37; M = 39
N = 76
Grade 5
t
1.89a
1.66
1.05
6.89**
0.90
0.57
7.52**
1.72
9.03**
Table 1. Distribution of participants tested by grade and gender (F and M), descriptive statistics (mean [M] and standard deviation [SD]) and time of testing (beginning and/or
end of grade)
READING PROSODY AND READING COMPREHENSION
© 2018 UKLA
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
Measures
Non-verbal ability
The Portuguese adaptation of the Raven’s Coloured Progressive Matrices (Simões, 1994)
was used to measure non-verbal reasoning.
Reading fluency
Text reading fluency. Four different texts were presented in paper sheets. Second graders
received a text composed of 142 words, third and fourth graders a text with 533 words
and fifth graders a text with 589 words. The texts were written in a Times New Roman
16 font type with “normal” spacing between letters and 1.5 spacing between lines. All
of the texts were selected from school manuals not then used as textbooks in the participating schools but appropriate for each grade. The readability level of the texts
was checked with the Flesch Reading Ease Formula adapted for the Spanish language1
(Fernández-Huerta, 1959). Test–retest reliability over a 1-year period was 0.88. We
assessed the three dimensions of reading fluency: accuracy, speed and prosody. Children
were asked to read aloud in an accurate and speeded way, in order to comprehend ‘what
the text said’. Errors and reading times were recorded. Oral reading recordings were obtained using a digital mp3 recorder. The number of correct words per minute (rate) was
used for analysis.
Reading prosody. For the reading prosody measure, we adapted a categorical rating scale
from the four key elements of the Multidimensional Fluency Scale developed by Rasinsky
(2004): (1) expression and volume (making the passage sound like natural language, adequate voice volume, expression and enthusiasm), (2) phrasing (marking phrase boundaries,
clause and sentence units), (3) smoothness (resolves word and structure difficulties easily
and quickly) and (4) pace (reads with appropriate conversational pace). Each element
was classified in six levels (1 = very weak; 2 = weak; 3 = reasonable; 4 = good; 5 = very
good; and 6 = excellent), by four trained judges,2 resulting in a total maximum score of 24,
corresponding to a global reading prosody adequacy level. Each judge listened to the first
minute of text reading of each child, randomly and without knowing the grade of the
reader, and then judged the quality of prosodic reading. An average of the global score
of the four judges was used for reading prosody analysis. There was a high level of
inter-rater reliability (global intraclass coefficient of 0.96).
Word reading fluency. The words were the same as in the text presented for text fluency but
presented in columns in a pseudo-randomised order to prevent the appearance of contextual and/or semantic relationship between the words. The child was asked to read the words
as rapidly and as accurately as possible. The number of words read correctly and the reading times were recorded in a way similar to that used for text fluency. The number of correct words per minute (rate) was used for analysis. Test–retest reliability over a 1-year
period was 0.83.
Pseudoword reading fluency. Pseudowords were constructed from the words of the WF
test, so that both tests had the same number of items. The order of the pseudowords was
randomised. The presentation and procedure were similar to those used for the WF test.
The number of correct pseudowords per minute (rate) was used for analysis. Test–retest reliability over a 1-year period was 0.82.
© 2018 UKLA
READING PROSODY AND READING COMPREHENSION
Reading comprehension
Context fluency effect. This measure of the context effect was based on the difference between performance on tests of text reading fluency and word reading fluency described earlier, that is, the number of words accurately read per minute in a connected text minus the
number of the same words accurately read per minute randomly presented.
Text reading comprehension. In the TRC test, the participants were asked to read six short
texts (58 to 68 words) and answer to four multiple-choice questions for each text. In half of
the multiple-choice questions, participants had to focus on and retrieve explicitly stated
information, that is, locate and understand relevant information or ideas that are explicitly
stated in text. The other half required the participants to make straightforward inferences,
that is, text-based inferences. Each multiple-choice question has only one correct response.
The position of the correct alternative varies randomly within the four possible positions.
The test, which was administered to large groups of participants, has a 7-minute time
limit for all of the grades studied. Test–retest reliability above a 1-and-a-half-year period
was 0.71.
General procedure
The tests were administered by a team of trained graduate students in a quiet setting provided by the school.
All of the fluency measures were individually administered in different sessions in a
fixed order (text, word and pseudoword) and with at least a 1-week interval between the
sessions. These assessments were performed during a several-week period at the end
(April–June) of the academic year. The student’s responses were recorded using a Memup
mp3 recorder during the test session and played back later for analysis purposes. The testers timed the tests using digital countdown stopwatches.
The reading comprehension test was administered by the same trained graduate students
but to large groups of participants.
Results
Descriptive statistics
Descriptive statistics, and t values to compare grade differences within each cohort, are presented in Table 1. Significant differences were found both between grades 2 and 3 and between grades 4 and 5 for TRC and for all reading fluency measures (text, word and
pseudoword) regarding the number of correct items per minute (rate). Concerning reading
fluency accuracy measures, there were no significant differences between grades 2 and 3
and between grades 4 and 5, with the exception of accuracy in pseudoword reading fluency
with higher performance in grade 3 than in grade 2. For the CFE and for reading prosody,
there were no significant differences within each cohort.
Correlational analyses
A summary of the concurrent and longitudinal correlations between the measures is presented in Tables 2 and 3 for cohorts 1 and 2, respectively.
© 2018 UKLA
Reading prosody at
end of grade 2
Pseudoword fluency
(accuracy) at end of grade 2
Pseudoword fluency (rate)
at end of grade 2
Word fluency (accuracy) at end of
grade 2
Word fluency (rate) at end of
grade 2
Context fluency effect at end of
grade 2
Text reading comprehension
(accuracy) at end of grade 2
Reading prosody at end of grade 3
Pseudoword fluency (accuracy) at
end of grade 3
Pseudoword fluency (rate) at end
of grade 3
Word fluency (accuracy) at end of
grade 3
Word fluency (rate) at end of
grade 3
Context fluency effect at end of
grade 3
Text reading comprehension
(accuracy) at beginning of grade 3
2
3
4
5
6
7
8
9
10
11
12
13
14
15
*p < .05.
**p < .01.
Non-verbal reasoning
1
Variable
1
1
1
0.24*
2
1
0.27*
0.10
3
Table 2. Correlation matrix for all measures in cohort 1
1
0.58**
0.52**
0.15
4
1
0.53**
0.62**
0.36**
0.16
5
1
0.51**
0.89**
0.34**
0.44**
0.15
6
1
0.14
0.26*
0.30**
0.25*
0.41**
0.12
7
1
0.40**
0.63**
0.39**
0.64**
0.36**
0.38**
0.17
8
1
0.36**
0.29*
0.32**
0.26*
0.33**
0.12
0.19
0.11
9
1
0.29*
0.36**
0.23*
0.37**
0.56**
0.48**
0.59**
0.34**
0.10
10
1
0.47**
0.40**
0.60**
0.26*
0.84**
0.45**
0.88**
0.46**
0.46**
0.02
11
1
0.31**
0.77**
0.29*
0.36**
0.25*
0.27*
0.57**
0.35**
0.50**
0.26*
0.01
12
1
0.25*
0.85**
0.32**
0.36**
0.53**
0.26*
0.84**
0.42**
0.79**
0.29**
0.39**
0.05
13
1
0.25*
0.38**
0.42**
0.39**
0.33**
0.48**
0.63**
0.42**
0.38**
0.55**
0.43**
0.54**
0.23*
14
1
0.55**
0.65**
0.22*
0.63**
0.37**
0.31*
0.62**
0.37**
0.68**
0.34**
0.67**
0.25*
0.45**
0.29*
15
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
© 2018 UKLA
Context fluency effect at end of
grade 5
14
*p < .05.
**p < .01.
Text reading comprehension
(accuracy) at beginning of grade 5
Word fluency (rate) at end of
grade 5
13
15
Word fluency (accuracy) at end of
grade 5
Text reading comprehension
(accuracy) at beginning of grade 4
8
12
Context fluency effect at end of
grade 4
7
Pseudoword fluency (rate) at end of
grade 5
Word fluency (rate) at end of
grade 4
6
11
Word fluency (accuracy) at end of
grade 4
5
Reading prosody at end of grade 5
Pseudoword fluency (rate) at end of
grade 4
4
Pseudoword fluency (accuracy) at
end of grade 5
Pseudoword fluency (accuracy) at
end of grade 4
3
9
Reading prosody at end of grade 4
2
10
Non-verbal reasoning
1
Variable
1
1
1
0.20
2
1
3
0.04
0.06
Table 3. Correlation matrix for all measures in cohort 2
0.15
1
0.19
0.23*
4
1
0.14
0.53**
0.04
0.00
5
1
.02
0.79**
0.00
0.22*
0.04
6
1
0.11
0.14
0.18
0.03
0.26*
0.04
7
1
0.48**
0.30**
0.09
0.38**
0.22*
0.18
0.12
8
1
0.12
0.03
0.09
0.07
0.01
0.13
0.27*
0.15
9
1
0.06
0.17
0.05
0.02
0.40**
0.18
0.62**
0.15
0.02
10
1
0.21
0.11
0.31**
0.34**
0.59**
0.08
0.77**
0.01
0.28*
0.24*
11
1
0.11
0.24*
0.26*
0.02
0.12
0.22
0.31**
0.03
0.35**
0.10
0.06
12
1
0.10
0.68**
0.12
0.10
0.40**
0.23
0.70**
0.04
0.70**
0.01
0.29*
0.06
13
1
0.18
0.05
0.13
0.10
0.41**
0.27*
0.44**
0.09
0.02
0.15
0.19
0.36**
0.06
14
1
0.29*
0.29*
0.01
0.21
0.03
0.06
0.58**
0.42**
0.22
0.15
0.23*
0.12
0.26*
0.18
15
READING PROSODY AND READING COMPREHENSION
© 2018 UKLA
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
All predictor and dependent variables were significantly positively correlated for cohort
1 (Table 2) with the unexpected exception of reading prosody at grade 2 that did not show a
significant correlation with reading prosody at grade 3. We think that this lack of correlation may be related to the size of our sample. Therefore, we performed a new correlation
analysis using the bootstrapping method (Efron & Tibshirani, 1993), which allows us, with
confidence, to simulate the population distribution of this statistic (i.e., correlation coefficient). The bootstrap correlation coefficient (average of r values across 1,000 samples) between reading prosody at grade 2 and reading prosody at grade 3 was significant
(p < .001). Recent evidence showed that the bootstrap estimates of the correlation coefficient represent valid sampling distribution estimates and provides a viable solution to research with small samples (Sideridis & Simos, 2010). In cohort 1, early word and
pseudoword reading fluency measures (both accuracy and rate) were significantly correlated with later reading prosody (with the exception of pseudoword reading fluency accuracy) and with later reading comprehension processes. Moreover, the significant
correlations also showed that reading prosody at an early grade was associated with later
reading comprehension processes and that the reverse was also true (reading comprehension and CFE at an early grade were associated with later reading prosody).
For cohort 2 (Table 3), with the exception of the significant correlation between early
pseudoword reading fluency (rate) and later TRC, there were no others significant correlations between early word and pseudoword reading fluency measures and later reading prosody or reading comprehension processes. Furthermore, reading prosody at grade 4 was
significantly correlated with reading prosody, CFE, and reading comprehension at grade 5.
Notwithstanding, TRC and context effect at grade 4 were significantly correlated with each
other at grade 5 but were not correlated with reading prosody at grade 5.
These correlations revealed that whereas in an early phase of learning to read (cohort 1),
there was a link between word and pseudoword reading fluency and reading prosody and
reading comprehension processes; in a later phase, no such link was observed. Besides, in
an early phase, a bidirectional association between reading prosody and reading comprehension processes was observed, whereas in a later phase (Cohort 2), only the initial reading
prosody was associated with the further reading comprehension processes but not the reverse.
Sequential regression analyses3
SRAs were used to model relationships between the present variables and to predict the
value of a dependent variable from an (or a set of) independent variable(s) (or predictors).
First, we specified the set of predictor variables that make up the model. Next, we specified
the order of entry of the predicting variables. Then, we tested the relative contribution of
each predictor variable to the dependent one in a sequential regression model.
For all SRAs reported here, age and nonverbal reasoning were entered first to control for
the contributions of age and general nonverbal ability.
Longitudinal bidirectional effects between reading prosody and reading comprehension
processes
We conducted a series of SRAs, which addressed a central aim of the study: to examine the
longitudinal associations between reading prosody and reading comprehension processes
(as measured by the CFE and TRC). The longitudinal bidirectional effects between them,
© 2018 UKLA
READING PROSODY AND READING COMPREHENSION
Figure 1. Schematic general procedure of the sequential regression analyses (SRAs) for the study of the longitudinal effects.
after controlling for autoregressive effects and reading efficiency, were examined by means
of cross-lag SRAs. We evaluated whether the cross-lag contributions of the predictor variable (A at time 1), entered at the third step, to another outcome variable (dependent variable, B, at time 2) survived the additional control of the outcome variable B at time 1,
entered in the second step (refer to Figure 1 for a schematic representation). After the first
cross-lag SRA, an additional SRA was conducted with the same control steps, and predictor and outcome variables switched positions. These two sets of SRAs were conducted to
shed light on the contribution between reading prosody and reading comprehension processes after removing the effects of autoregressor (SRA 1, Figure 1) and of reading efficiency (PF and WF; SRA 2, Figure 1).
For example, when the longitudinal effect of reading prosody at the end of grade 2 on the
CFE at the end of grade 3 was examined, the grade 2 CFE was entered as a control variable
in the second step to examine the contribution of reading prosody to the change in the CFE
from the end of grade 2 to the end of grade 3. Subsequently, the same was done to evaluate
the longitudinal effect of the CFE at the end of grade 2 on reading prosody at the end of
grade 3. This procedure was applied to two pairs of variables in both cohorts: (1) reading
prosody and the CFE, (2) reading prosody and TRC.
We conducted additional SRAs in which we examined whether the significant bidirectional contributions between reading prosody and reading comprehension processes were
still present after the additional control of two skills (PF and WF) entered in the second
and third step, respectively.
The results of the cross-lagged SRAs are presented in Tables 4–6 and 7. This first set of
sequential regressions, in which only the effects of the autoregressor were removed,
showed that there were significant contributions (only in cohort 1) between reading prosody and reading comprehension processes (Tables 4 and 5): reading prosody (refer to
Table 4) at the end of grade 2 accounted for significant variance in TRC at the beginning
of grade 3 (0.04) and in the CFE at the end of grade 3 (0.06).
In the reverse direction (refer to Table 5), TRC at the end of grade 2 accounted for significant variance in reading prosody at the end of grade 3 (0.09). In cohort 2, with the
© 2018 UKLA
RP
0.28
0.50
0.06*
0.36***
0.07
ΔR
RP
TRC
A
NR
Predictor
0.22
0.51
0.14
0.20
β
0.04*
0.34**
0.11*
ΔR
2
Grade 3 TRC
Note. All the beta weights are from the final step of the regression model.
*p < .05.
**p < .01.
***p < .0001.
CFE
3
0.01
A
2
0.11
NR
1
β
Predictor
Step
2
Grade 3 CFE
3
2
1
Step
RP
CFE
A
NR
Predictor
0.08
0.02
0.25
0.40
β
0.05
0.20***
0.01
ΔR
2
Grade 5 CFE
Time of reading comprehension measures assessment
RP
TRC
A
NR
Predictor
0.18
0.51
0.12
0.10
β
0.03
0.29**
0.08
Δ R2
Grade 5 TRC
Table 4. Summary of longitudinal sequential regression analysis predicting grades 3 and 5 reading comprehension measures by reading prosody (RP) after controlling for nonverbal reasoning (NR), age (A), context fluency effect (CFE) or text reading comprehension (TRC)
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
© 2018 UKLA
NR
1
CFE
3
0.22
0.09
0.13
0.07
β
0.04
0.03
0.04
ΔR
TRC
RP
A
NR
Predictor
0.31
0.08
0.14
0.07
β
Note. All the beta weights are from the final step of the regression model.
*p < .05.
RP
2
A
Predictor
Step
2
Grade 3 RP
0.09*
0.03
0.04
ΔR
2
3
2
1
Step
CFE
RP
A
NR
Predictor
Time of reading prosody assessment
0.09
0.30
0.03
0.09
β
0.01
0.07*
0.03
Δ R2
TRC
RP
A
NR
Predictor
Grade 5 RP
0.10
0.29
0.02
0.05
β
0.01
0.09*
0.02
Δ R2
Table 5. Summary of longitudinal sequential regression analysis predicting grades 3 and 5 reading prosody by context fluency effect (CFE) or text reading comprehension
(TRC) after controlling for non-verbal reasoning (NR), age (A) or reading prosody (RP)
READING PROSODY AND READING COMPREHENSION
© 2018 UKLA
NR
1
RP
5
0.15
0.45
0.01
0.30
0.02
0.09
β
0.02
0.20**
0.01
0.25**
0.07
ΔR
RP
TRC
WF
PF
A
NR
Predictor
0.09
0.26
0.26
0.22
0.15
0.20
β
0.01
0.04*
0.03*
0.41**
0.11*
ΔR
2
Grade 3 TRC
Note. All the beta weights are from the final step of the regression model.
*p < .05.
**p < .0001.
WF
CFE
3
4
PF
2
A
Predictor
Step
2
Grade 3 CFE
5
4
3
2
1
Step
RP
CFE
WF
PF
A
NR
Predictor
0.16
0.06
0.06
0.25
0.44
0.15
β
0.05
0.18**
0.00
0.02
0.01
ΔR
2
Grade 5 CFE
Time of reading comprehension measures assessment
RP
TRC
WF
PF
A
NR
Predictor
0.19
0.55
0.16
0.05
0.09
0.12
β
0.03
0.24**
0.01
0.05
0.08
Δ R2
Grade 5 TRC
Table 6. Summary of longitudinal sequential regression analysis predicting grades 3 and 5 reading comprehension measures by reading prosody (RP) after controlling for nonverbal reasoning (NR), age (A), pseudoword fluency (PF), word fluency (WF), context fluency effect (CFE) or text reading comprehension (TRC)
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
© 2018 UKLA
NR
1
WF
RP
CFE
3
4
5
0.21
0.02
0.22
0.08
0.14
0.06
β
0.03
0.00
0.00
0.10**
0.04
ΔR
TRC
RP
WF
PF
A
NR
Predictor
0.22
0.03
0.03
0.17
0.15
0.07
β
Note. All the beta weights are from the final step of the regression model.
*p < .05.
**p < .01.
PF
2
A
Predictor
Step
2
Grade 3 RP
0.03
0.00
0.00
0.10*
0.04
ΔR
2
5
4
3
2
1
Step
CFE
RP
WF
PF
A
NR
Predictor
0.05
0.30
0.15
0.13
0.04
0.06
β
Time of reading prosody assessment
0.00
0.07*
0.01
0.00
0.03
Δ R2
TRC
RP
WF
PF
A
NR
Predictor
Grade 5 RP
0.01
0.13
0.31
0.22
0.25
0.04
β
0.01
0.08*
0.01
0.01
0.02
Δ R2
Table 7. Summary of longitudinal sequential regression analysis predicting grades 3 and 5 reading prosody by context fluency effect (CFE) or text reading comprehension
(TRC) after controlling for non-verbal reasoning (NR), age (A), pseudoword fluency (PF), word fluency (WF) or reading prosody (RP)
READING PROSODY AND READING COMPREHENSION
© 2018 UKLA
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
autoregressive control, there were no significant contributions from reading prosody to any of
the reading comprehension processes nor were there contributions in the reverse direction.
To remove the effect of reading efficiency, we conducted a final set of SRAs, with controls for the variables PF and WF in addition to the autoregressive control (refer to Tables 6
and 7). No significant contributions were observed in either cohorts 1 or 2.
When PF and WF were introduced in the SRAs from reading prosody to reading comprehension measures, there was a significant contribution from PF at grade 2 to each of
reading comprehension measures at grade 3 (0.25 to the CFE and 0.41 to TRC). For students advancing from grades 4 to 5, PF only made a significant (small) contribution to
TRC (0.05) but not to the other measure of reading comprehension (CFE). Furthermore,
WF only made a small contribution to TRC (0.03) in students moving from grades 2 to 3.
Regarding the reverse direction, when we introduced PF and WF in the SRA from reading comprehension measures to reading prosody, only PF (and not WF) made a significant
unique contribution to reading prosody for students moving from grades 2 to 3 (0.10 from
the CFE to reading prosody and from TRC to reading prosody) but not for students moving
from Grades 4 to 5.
Concurrent effects of reading prosody on reading comprehension processes
Concurrent relationships within the same test period at each grade level were also carried out.
This set of regression analyses tested the contribution of reading prosody to reading comprehension processes (as measured by the CFE and/or TRC) in the two cohorts (cohorts 1 and 2).
Thus, we conducted two SRAs (refer to Figure 2 for a schematic representation) that
tested the direct effect of reading prosody as a predictor variable, entered in the second
step, to the outcome variable A (CFE or TRC). Then, we carried out three additional SRAs
to examine whether the contribution of reading prosody, entered in the fourth step, was still
present after the additional control of two skills (PF and WF) entered in the second and
third step, respectively.
The effect of reading prosody (without the controls for PF and WF) on the CFE (refer to
Table 8) was significant in students in all grades tested (0.16, 0.08, 0.06 and 0.16 of explained variance, respectively, for grades 2, 3, 4 and 5). When the additional controls for
PF and WF were introduced (refer to Table 9), significant contributions remained for
grades 2, 4 and 5, (0.07, 0.04 and 0.18, respectively) and attained an almost significant
Figure 2. Schematic general procedure of the sequential regression analyses (SRAs) for the study of the concurrent effects.
© 2018 UKLA
READING PROSODY AND READING COMPREHENSION
level at grade 3 (p = 0.07 and 0.05 of explained variance). The contribution of PF to the
CFE was significant in grades 2 and 3 (0.11 and 0.08 of unique explained variance, respectively) but not in grades 4 and 5. The unique explained variance from WF in the CFE was
significant in grades 2, 4 and 5 (0.05, 0.17 and 0.19, respectively).
Reading prosody (without the controls for PF and WF) accounted for significant variance in TRC (0.12) at the end of grade 24 (refer to Table 8). However, this significant contribution of reading prosody to TRC did not hold with the additional controls for PF and
WF (refer to Table 9). Here, both PF and WF explained significant variance in TRC
(0.34 and 0.07, respectively).
Discussion
In the present study, we aimed at testing the relationships between reading prosody and
reading comprehension processes during learning to read in European Portuguese, an orthography with an intermediate level of opacity. Based on the theoretical proposals that,
on one hand, reading prosody may assist reading comprehension (Kuhn & Stahl, 2003),
and that on the other hand, reading comprehension might allow the student to read with
prosody (Schwanenflugel et al., 2004), we hypothesized that longitudinal effects occur between reading prosody and reading comprehension processes in both directions.
To achieve this purpose, we distinguished two types of reading comprehension processes: effortful and effortless. To evaluate reading comprehension, we used a test that
mainly involves effortful processing: TRC, and to measure effortless processing in reading
comprehension, we assessed the CFE by calculating the difference between text reading
fluency and word reading fluency.
To study these relationships, we took into account a set of control variables (e.g.,
autoregressor, PF and WF). The inclusion of the reading skill as control (PF and WF), in
addition to the autoregressor, is crucial when examining the contribution of reading prosody to reading comprehension along reading development because its respective weight
may vary in the process of reading acquisition. It is widely recognised that during the initial
phase of literacy acquisition, decoding is fundamental to the children’s development of fast
and accurate reading skill and, as such, it may be more helpful than reading prosody for
reading comprehension, whereas reading prosody may be more relevant in an upper phase
when children have already attained automaticity in reading. In fact, Calet et al. (2015) observed a higher contribute of nonword reading and reading rate, in second grade than in
fourth grade. Moreover, ‘the role of prosodic reading in reading comprehension in second
graders was not as relevant as for fourth graders’ (Calet et al., 2015, p.10). In this vein, it is
expectable that the hypothesised relationship between reading prosody and reading comprehension will become stronger throughout the development of reading.
The present results partially confirm our hypothesis that longitudinal effects between
reading prosody and reading comprehension would be bidirectional. Indeed, the SRAs with
only the autoregressive control showed that reading prosody at the end of grade 2 significantly contributed to TRC at the beginning of grade 3 (0.04) and to the CFE at the end of
grade 3 (0.06). In this cohort (cohort 1), we also observed effects in the reverse direction:
TRC at the end of grade 2 significantly influenced reading prosody at the end of grade 3
(0.09). However, for the second cohort (cohort 2), no significant contribution was observed
in either direction.
© 2018 UKLA
RP
0.36
0.06
0.09
β
0.12***
0.03
ΔR
2
RP
A
NR
Predictor
0.42
0.08
0.04
0.16**
0.03
NR
RP
A
0.29
0.08
0.20
2
0.08*
0.07
ΔR
β
Predictor
β
ΔR
Grade 3 CFE
2
Grade 2 CFE
RP
A
NR
Predictor
0.02
0.25
0.17
β
2
0.06*
0.04
ΔR
Grade 4 CFE
Note. CFE = context fluency effect; TRC = text reading comprehension. All the beta weights are from the final step of the regression model.
*p < .05.
**p < .01.
***p < .0001.
2
NR
1
A
Predictor
Step
Grade 2 TRC
RP
A
NR
Predictor
0.01
0.41
0.11
β
0.16***
0.02
Δ R2
Grade 5 CFE
Table 8. Summary of sequential regression analysis predicting grades 2, 3, 4 and 5 reading comprehension measures by reading prosody (RP) after controlling for non-verbal
reasoning (NR) or age (A)
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
© 2018 UKLA
NR
1
WF
RP
3
4
0.10
0.57
0.03
0.06
0.05
β
0.00
0.07**
0.34***
0.03
ΔR
RP
WF
PF
A
NR
Predictor
0.31
0.43
0.56
0.10
0.03
β
2
0.07*
0.05*
0.11***
0.03
ΔR
Grade 2 CFE
RP
WF
PF
A
NR
Predictor
0.26
0.37
0.51
0.07
0.18
β
2
0.05
0.02
0.08*
0.06
ΔR
Grade 3 CFE
RP
WF
PF
A
NR
Predictor
0.22
0.70
0.67
0.11
0.12
β
2
0.04*
0.17***
0.02
0.04
ΔR
Grade 4 CFE
Note. CFE = context fluency effect; TRC = text reading comprehension. All the beta weights are from the final step of the regression model.
*p < .05.
**p < .01.
***p < .0001.
PF
2
A
Predictor
Step
2
Grade 2 TRC
RP
WF
PF
A
NR
Predictor
0.42
0.64
0.63
0.09
0.16
β
0.18***
0.19***
0.02
0.01
Δ R2
Grade 5 CFE
Table 9. Summary of sequential regression analysis predicting Grades 2, 3, 4 and 5 reading comprehension measures by reading prosody (RP) after controlling for non-verbal
reasoning (NR), age (A), pseudoword fluency (PF) or word fluency (WF)
READING PROSODY AND READING COMPREHENSION
© 2018 UKLA
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
These results obtained with the SRAs are to some extent consistent with those obtained
by Schwanenflugel et al. (2004) insofar as these authors also observed reading prosody acting as a predictor of reading comprehension with English-speaking children from grades 2
and 3. However, they did not observe the reverse, that is, reading comprehension acting as
a predictor of reading prosody as we found in the current study.
Thus, our outcomes provide some evidence for long-lasting bidirectional effects between
reading prosody and reading comprehension processes, particularly with regard to those involved in TRC from grades 2 to 3. However, the inclusion of PF and WF in addition to the
autoregressive control in our study was essential to highlight the relationship between reading prosody and reading comprehension, given that most of the previous studies examining
this relationship have not included or used, as control, any measure of decoding – a determinant skill for reading prosody (e.g., Kuhn & Stahl, 2003) and reading comprehension
(e.g., LaBerge & Samuels, 1974) – in their models. Indeed, with these additional control
variables (PF and WF) in the regression model, the contributions we found with the only
autoregressive control did not hold in both directions. In this sense, Lopes, Silva, Moniz,
Spear-Swerling, and Zibulsky (2015) have recently examined the influence of reading
prosody on reading comprehension (but not the reverse influence) among Portuguese second and third graders, showing that the contribution of prosody to reading comprehension
was only marginal when reading rate was accounted for. It is worth to note that no other
control variables (e.g., auto-regressive and pseudoword reading) were included in their regression model. If they had been included, perhaps, no contribution at all of reading prosody to reading comprehension would be observed. In the present study, the inclusion of
these control variables led to apparently inconsistent results with those obtained in previous
studies. Yet, conceivably, had Schwanenflugel et al. (2004), Klauda and Guthrie (2008)
and Lopes et al. (2015) used or included the set of control variables (particularly the
decoding and autoregressive control) as those considered in our study, and their results
would have been similar to those obtained here.
It is important to underline that, in the present study, for students moving from grades 2
to 3 (but not for those moving from grades 4 to 5), decoding skill plays an important role in
the relationship from reading prosody to reading comprehension (0.25 and 0.41, respectively, for CFE and TRC) and also in the reverse direction (0.10 for both, CFE and
TRC). Thus, these results obtained in European Portuguese support the view that decoding
fluency may be the major factor in the longitudinal relation between reading prosody and
reading comprehension processes at least at an early phase of reading development in this
orthography. It is worth noting that this finding is, to some extent, in line with previous research with English-speaking children (Miller & Schwanenflugel, 2006; Schwanenflugel
et al., 2004). In fact, although variables related to pitch (Miller & Schwanenflugel, 2006)
and adult-like intonation (Schwanenflugel et al., 2004) were found to be linked to reading
comprehension, pausing variables of reading prosody were not directly linked to reading
comprehension but were directly related to fluent decoding skills. In a similar way, reading
prosody was, in our study, associated with decoding skills.
It is widely recognised that the attainment of efficiency in decoding skills is dependent
on the degree of orthographic consistency, occurring in earlier grade levels in more consistent orthographies (e.g., Seymour, Aro, & Erskine, 2003). As such, considering the crucial
role of decoding skill in the establishment of the relation between reading prosody and
reading comprehension, it is conceivable that the grade level and also the consistency of
the orthography could impact on results when examining this relationship. Indeed, in a recent longitudinal study with Dutch children from fourth to sixth grade, Veenendaal et al.
© 2018 UKLA
READING PROSODY AND READING COMPREHENSION
(2016a) examined this relationship controlling the autoregressive as well the decoding effects. They observed a reciprocal relationship between reading prosody, measured with a
rating scale, and reading comprehension: decoding and reading comprehension measured
at fourth grade were related to subsequent text reading prosody (sixth grade), and text reading prosody at fourth grade predicted reading comprehension at sixth grade. Thus, compared with those of others longitudinal studies, the results of the present study seem to
be more similar to those found in English than in Dutch, a transparent orthography. In fact,
we did not observe a longitudinal relationship between reading prosody and reading comprehension, in any cohort, when taking into account the entire set of control variables (refer
to Table 6). Perhaps in European Portuguese, an intermediate depth orthography (e.g.,
Sucena et al., 2009), and in English, a direct relationship between reading prosody and
reading comprehension takes place in more advanced grade levels, when reading prosody
and reading comprehension no longer rely too heavily on decoding skills.
Concerning the concurrent effects of reading prosody on reading comprehension processes, the results showed that, after introducing PF and WF as controls in the SRAs, the
contribution of reading prosody to the CFE remained significant (0.07, 0.04 and 0.18 in
grades 2, 4 and 5, respectively) or almost significant (0.05, p = 0.07, in grade 3) but not
for TRC at grade 2. Furthermore, beyond this influence of reading prosody on reading
comprehension, decoding seems to once again play a relevant role. Actually, in grade 2,
PF significantly contributed to reading comprehension (0.11 and 0.34, respectively, for
the CFE and TRC); in grade 3, PF maintained a significant explained variance for the
CFE (0.08); and in grades 4 and 5, PF was predictive of the CFE (unique variance of
0.17 and 0.19, respectively).
Thus, regarding the predictive effect of reading prosody on TRC (measure limited to
grade 2), this did not hold after the control of reading skills (PF and WF), as already observed for the longitudinal contributions, but contrary to what observed for the longitudinal effects, reading prosody predicted CFE through all the examined grades (second to
fifth). This set of results suggests that the distinction between effortful and effortless processes in reading comprehension, assessed, respectively, by the TRC and CFE measures,
may be relevant when considering their relationship with reading prosody. Indeed, our
data revealed that reading prosody concurrently predicted the ability to unintentionally
segment a text according to its major syntactic/semantic elements, that is, effortless processing, from an early phase of reading acquisition until a more advanced one. In contrast, in grade 2, reading prosody was not predictive of reading comprehension
involving intentional and effortful processing. Unfortunately, the missing data collection
of TRC concurrently with the other measures beyond grade 2 do not allow drawing extensive conclusions about this last predictive role of reading prosody at more advanced
grades. It is, however, conceivable that in a later phase of learning to read, TRC involving complex processes would benefit from reading prosody. A recent cross-sectional
study (Calet et al., 2015) with Spanish-speaking children, in addition to showing that automaticity (a measure that includes decoding) contributed strongly for reading comprehension at the second grade, but weakly at the fourth grade, provides evidence of a
reverse pattern for the contribution of prosody to reading comprehension, that is, weak
at the second but stronger at the fourth grade.
The lack of significant concurrent contribution from reading prosody to TRC in our study
may be related to the degree of consistency of European Portuguese, an orthography more
inconsistent than Spanish (Defior, Martos, & Cary, 2002; Seymour et al., 2003), assigning a
more relevant weight to decoding in early grades for European Portuguese. Taking into
© 2018 UKLA
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
account the recent evidence (Fernandes et al., 2017) indicating a decreased role, in later
grades, of decoding on TRC in European Portuguese, it is expectable that reading prosody
will have a contribution to TRC in a more advanced stage of reading acquisition.
The joint analysis of our results and those of the study by Calet et al. (2015) leads us to
suggest that there is no pure link between reading prosody and reading comprehension,
highlighting that decoding skills play a major role in the relationship between prosodic
reading and reading comprehension, at least in an initial phase of learning to read. Thus,
at first sight, there seems to be a limited relationship between prosodic reading and reading
comprehension skills that should be taken into account by acknowledging the key role of
decoding fluency in reading comprehension. Nevertheless, it is important to note that lexical prosody (e.g., stress assignment) considered as an aspect of general reading prosody is
necessary for reading words and pseudowords fluently, having been suggested that it is part
of decoding skills (Schwanenflugel & Benjamin, 2017). Actually, Schwanenflugel and
Benjamin (2017) showed a relation between lexical prosody and reading fluency and reading comprehension. Further studies are necessary to examine this set of issues.
Lastly, regarding the limitations of the present study, it is important to recognise that
some aspects such as the sample dimension and the number of studied relations prevented
us to use a more elegant and analytic method (e.g., path analyses) than SRAs. Future research should consider the use of spectrographic measures allowing for the examination
of a number of characteristics of oral reading related to reading prosody (e.g., pause patterns and pitch changes) or, as suggested by Benjamin et al. (2013), the use of a spectrographically grounded scale. In addition, to better deepen our understanding of the
relationship between reading prosody and reading comprehension, it would be relevant
to develop further longitudinal studies in languages with different levels of opacity and
covering a wider range of grade levels.
In conclusion, our findings bring new evidence into the longitudinal relationships between reading prosody, decoding and reading comprehension in a language with an intermediate degree of orthographic transparency, from early to more advanced levels of
learning to read. Actually, when taking into account reading skills (PF and WF) and the
autoregressive control, there was no evidence of any bidirectional effect between reading
prosody and reading comprehension processes in an intermediate depth orthography – European Portuguese – from grades 2 to 5. Under these conditions, it is important to underline
the key role of decoding in the relation between reading prosody and reading comprehension processes.
As far as concurrent relations are concerned, reading prosody predicted the effortless
processing of reading comprehension, from an early phase of reading acquisition until a
more advanced one, but not the effortful one, raising the relevance of examining the relationship between prosodic reading and reading comprehension taking into account the distinction between effortless and effortful processes of reading comprehension (Kintsch,
1988). A recent training study conducted in Spanish (Calet, Gutiérrez-Palma, & Defior,
2017) revealed that in fourth but not in second grade, reading efficiency benefitted from
both automaticity and prosody training, and sentence comprehension, assessed by the
use of the punctuation marks, improved from automaticity training and even more strongly
from prosody training. Regarding TRC, only prosody training enhanced comprehension
assessed in an intra-training test, in both grades, but interestingly, this effect did not generalise to a post-training test. In our view, these findings of Calet et al. (2017) obtained with
another methodology than ours seem to strengthen the pertinence of taking into consideration the already referred distinction of effortless and effortful processes in reading
© 2018 UKLA
READING PROSODY AND READING COMPREHENSION
comprehension when examining the contribution of reading prosody to reading
comprehension.
Finally, from a teaching/learning perspective, the findings showing an influence from
reading prosody on effortless processes of reading comprehension suggest that children
learning to read in European Portuguese would benefit from guidance in reading and rereading texts with prosody as a way to develop reading comprehension (refer to Calet
et al., 2017 for Spanish).
Acknowledgements
This work was partly supported by FCT – Foundation of Science and Technology under
the PhD research grant SFRH/BD/46432/2008. Data collection was supported and carried
out under the project ‘Developmental benchmarks of reading and writing, in European Portuguese, from first to sixth grade’ (2008–2010), under the auspices of the National Reading
Plan. We thank the students, teachers, administrators and school personnel of the two
intervenient schools. We also thank all the examiners who collected data in the field. We
acknowledge the helpful scientific advice of Professor Catarina Marques concerning the
statistical methodology. We also thank the writer Isabel Alçada for her contribution in
the writing of some texts included in the text reading comprehension test. We would also
like to thank Professor José Morais for his constructive comments on a previous version of
the manuscript. Lastly, we thank the anonymous reviewers for their very helpful comments
and suggestions that greatly contributed to improve the paper.
Notes
1. In the absence of a readability formula for Portuguese (European or Brazilian), we
adopted the Spanish adaptation of Flesch Reading Ease Formula. The European Portuguese language has a greater level of similarity concerning the frequency of mono and
multisyllabic words with Spanish than with English or French.
2. The panel of judges was composed by two teachers from the first cycle of basic education (grades 1 to 4 in Portugal) and two teachers of Portuguese language in the second
cycle of basic education (grades 5 and 6 in Portugal).
3. The number of participants included in each sequential regression analysis is the total of
participants who accomplished all the tasks (listwise option) involved in each regression
analysis. Therefore, the number of participants may vary between analyses and may be
different from the number involved in the correlation analysis.
4. It was possible to consider TRC only in the analyses of the second grade once; in the
other grades, it was tested before reading prosody. Thus, for this reason (temporal
order), we did not carry out regression analyses with this measure beyond second
grade.
References
Benjamin, R.G. & Schwanenflugel, P.J. (2010). Text complexity and oral reading prosody in young readers. Reading Research Quarterly, 45(4), 388–404. https://doi.org/10.1598/RRQ.45.4.2.
© 2018 UKLA
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
Benjamin, R.G., Schwanenflugel, P.J., Meisinger, E.B., Groff, C., Kuhn, M.R. & Steiner, L. (2013). A spectrographically grounded scale for evaluating reading expressiveness. Reading Research Quarterly, 48(2),
105–133. https://doi.org/10.1002/rrq.43.
Calet, N., Defior, S. & Gutiérrez-Palma, N. (2015). A cross-sectional study of fluency and reading comprehension
in Spanish primary school children. Journal of Research in Reading, 38(3), 272–285. https://doi.org/10.1111/
1467-9817.12019.
Calet, N., Gutiérrez-Palma, N. & Defior, S. (2017). Effects of fluency training on reading competence in primary
school children: The role of prosody. Learning and Instruction, 52, 59–68. https://doi.org/10.1016/j.
learninstruc.2017.04.006.
Daane, M. C., Campbell, J. R., Grigg, W. S., Goodman, M. J., & Oranje, A. (2005). Fourth-grade students reading
aloud: NAEP 2002 special study of oral reading (NCES 2006–469). Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for education Statistics. Retrieved from http://nces.ed.
gov/nationsreportcard/pdf/studies/2006469.pdf
Deacon, S.H., Benere, J. & Castles, A. (2012). Chicken or egg? Untangling the relationship between orthographic processing skill and reading accuracy. Cognition, 122(1), 110–117. https://doi.org/10.1016/j.cognition.2011.09.003.
Defior, S., Martos, F. & Cary, L. (2002). Differences in reading acquisition development in two shallow orthographies: Portuguese and Spanish. Applied PsychoLinguistics, 23(1), 135–148. https://doi.org/10.1017/
S0142716402000073.
Efron, B. & Tibshirani, R. (1993). An introduction to the bootstrap. New York: Chapman & Hall.
Fernandes, S., Querido, L., Verhaeghe, A., Marques, C. & Araújo, L. (2017). Reading development in European
Portuguese: Relationships between oral reading fluency, vocabulary and reading comprehension. Reading and
Writing: An Interdisciplinary Journal, 30(9), 1987–2007. https://doi.org/10.1007/s11145-017-9763-z.
Fernandes, S., Ventura, P., Querido, L. & Morais, J. (2008). Reading and spelling acquisition in European Portuguese: A preliminary study. Reading and Writing: An Interdisciplinary Journal, 21(8), 805–821. https://doi.org/
10.1007/s11145-007-9093-7.
Fernández-Huerta, J. (1959). Medidas sencillas de lecturabilidad (Simple readability measures). Consigna, 214,
29–32.
Florit, E. & Cain, K. (2011). The simple view of reading: Is it valid for different types of alphabetic orthographies?
Educational Psychology Review, 23(4), 553–576. https://doi.org/10.1007/s10648-011-9175-6.
Frazier, L., Carlson, K. & Clifton, C. (2006). Prosodic phrasing is central to language comprehension. Trends in
Cognitive Sciences, 10, 244–249. https://doi.org/10.1016/j.tics.2006.04.002.
Fuchs, L.S., Fuchs, D. & Maxwell, L. (1988). The validity of informal reading comprehension measures. Remedial and Special Education, 9(2), 20–29. https://doi.org/10.1177/074193258800900206.
Hudson, R., Lane, H. & Pullen, P. (2005). Reading fluency assessment and instruction: What, why, and how? The
Reading Teacher, 58(8), 702–714. https://doi.org/10.1598/RT.58.8.1.
Hudson, R.F., Pullen, P.C., Lane, H.B. & Torgesen, J.K. (2009). The complex nature of reading fluency: A multidimensional view. Reading & Writing Quarterly, 25(1), 4–32. https://doi.org/10.1080/10573560802491208.
Jenkins, J.R., Fuchs, L.S., van den Broek, P., Espin, C. & Deno, S.L. (2003). Sources of individual differences in
reading comprehension and reading fluency. Journal of Educational Psychology, 95(4), 719–729. https://doi.
org/10.1037/0022-0663.95.4.719.
Kenny, D.A. (1975). A quasi-experimental approach to assessing treatment effects in the nonequivalent control
group design. Psychological Bulletin, 82(3), 345–362. https://doi.org/10.1037/0033-2909.82.3.345.
Kintsch, W. (1988). The use of knowledge in discourse processing: A construction-integration model. Psychological Review, 95(2), 163–182. https://doi.org/10.1037/0033-295X.95.2.163.
Klauda, S.L. & Guthrie, J.T. (2008). Relationships of three components of reading fluency to reading comprehension. Journal of Educational Psychology, 100(2), 310–321. https://doi.org/10.1037/0022-0663.100.2.310.
Koriat, A., Kreiner, H. & Greenberg, S. (2002). The extraction of structure during reading: Evidence from reading
prosody. Memory & Cognition, 30(2), 270–280. https://doi.org/10.3758/bf03195288.
Kuhn, M., Schwanenflugel, P.J. & Meisinger, E.B. (2010). Aligning theory and assessment of reading fluency:
Automaticity, prosody, and the definitions of fluency. Reading Research Quarterly, 45(2), 230–251. https://
doi.org/10.1598/RRQ.45.2.4.
Kuhn, M.R. & Stahl, S.A. (2003). Fluency: A review of developmental and remedial practices. Journal of Educational Psychology, 95(1), 3–21. https://doi.org/10.1037/0022-0663.95.1.3.
LaBerge, D. & Samuels, S.J. (1974). Toward a theory of automatic information processing in reading. Cognitive
Psychology, 6(2), 293–323. https://doi.org/10.1016/0010-0285(74)90015-2.
© 2018 UKLA
READING PROSODY AND READING COMPREHENSION
Lai, S.A., Benjamin, R.G., Schwanenflugel, P.J. & Kuhn, M.R. (2014). The longitudinal relationship between
reading fluency and reading comprehension skills in second-grade children. Reading & Writing Quarterly,
30(2), 116–138. https://doi.org/10.1080/10573569.2013.789785.
Landerl, K. & Wimmer, H. (2008). Development of word reading fluency and spelling in a consistent orthography: An 8-year follow-up. Journal of Educational Psychology, 100(1), 150–161. https://doi.org/10.1037/
0022-0663.100.1.150.
Lopes, J., Silva, M.M., Moniz, A., Spear-Swerling, L. & Zibulsky, J. (2015). Evolução da prosódia e compreensão
da leitura: Um estudo longitudinal do 2.° ano ao final do 3.° ano de escolaridade. Revista de Psicodidáctica,
20(1), 5–23. https://doi.org/10.1387/RevPsicodidact.11196.
Miller, J. & Schwanenflugel, P.J. (2006). Prosody of syntactically complex sentences in the oral reading of young
children. Journal of Educational Psychology, 98(4), 839–843. https://doi.org/10.1037/0022-0663.98.4.839.
Miller, J. & Schwanenflugel, P.J. (2008). A longitudinal study of the development of reading prosody as a dimension of oral reading fluency in early elementary school children. Reading Research Quarterly, 43(4), 336–354.
https://doi.org/10.1598/RRQ.43.4.2.
National Reading Panel (2000). Teaching children to read: An evidence-based assessment of the scientific research literature on reading and its implications for reading instruction. NIH publication no. 00–4769.
Washington, DC: U.S. Government Printing Office.
O’Shea, L.J. & Sindelar, P.T. (1983). The effects of segmenting written discourse on the reading comprehension
of low-and high-performance readers. Reading Research Quarterly, 18(4), 458–465. https://doi.org/10.2307/
747380.
Perfetti, C.A. (1985). Reading ability. New York, NY: Oxford University Press.
Rasinsky, T.V. (2004). Creating fluent readers. Educational Leadership, 61(6), 46–51 Retrieved from http://
educationalleader.com/subtopicintro/read/ASCD/ASCD_364_1.pdf.
Samuels, S.J. (2006). Toward a model of reading fluency. In S.J. Samuels & A.E. Farstrup (Eds.), What research
has to say about fluency instruction, (pp. 24–46). Newark, DE: International Reading Association.
Schwanenflugel, P.J. & Benjamin, R.G. (2017). Lexical prosody as an aspect of oral reading fluency. Reading and
Writing: An Interdisciplinary Journal, 30(1), 143–162. https://doi.org/10.1007/s11145-016-9667-3.
Schwanenflugel, P.J., Hamilton, A.M., Kuhn, M.R., Wisenbaker, J.M. & Stahl, S.A. (2004). Becoming a fluent
reader: Reading skill and prosodic features in the oral reading of young readers. Journal of Educational Psychology, 96(1), 119–129. https://doi.org/10.1037/0022-0663.96.1.119.
Seymour, P.H., Aro, M. & Erskine, J.M. (2003). Foundation literacy acquisition in European orthographies. British Journal of Psychology, 94(2), 143–174. https://doi.org/10.1348/000712603321661859.
Sideridis, G.D. & Simos, P. (2010). What is the actual correlation between expressive and receptive measures of
vocabulary? Approximating the sampling distribution of the correlation coefficient using the bootstrapping
method. International Journal of Educational & Psychological Assessment, 5, 117–133 Retrieved from
https://drive.google.com/file/d/0ByxuG44OvRLPYzFNTmc3aVhmbFk/view.
Simões, M. R. (1994). Investigações no âmbito da aferição nacional do teste das Matrizes Progressivas Coloridas
de Raven (M.P.C.R.). (Investigations in the context of the national assessment of Raven’s Colored Progressive
Matrices). (Unpublished doctoral dissertation). Universidade de Coimbra, Coimbra.
Stanovich, K.E. (1980). Toward an interactive-compensatory model of individual differences in the development
of reading fluency. Reading Research Quarterly, 16(1), 32–71. https://doi.org/10.2307/747348.
Sucena, A., Castro, S.L. & Seymour, P. (2009). Developmental dyslexia in an orthography of intermediate depth:
The case of European Portuguese. Reading and Writing: An Interdisciplinary Journal, 22(7), 791–810. https://
doi.org/10.1007/s11145-008-9156-4.
Torgesen, J.K., Rashotte, C.A. & Alexander, A. (2001). Principles of fluency instruction in reading: Relationships
with established empirical outcomes. In M. Wolf (Ed.), Dyslexia, fluency, and the brain, (pp. 333–356).
Parkton, MD: York Press.
Veenendaal, N.J., Groen, M.A. & Verhoeven, L. (2014). The role of speech prosody and text reading prosody in
children’s reading comprehension. British Journal of Educational Psychology, 84(4), 521–536. https://doi.org/
10.1111/bjep.12036.
Veenendaal, N.J., Groen, M.A. & Verhoeven, L. (2015). What oral text reading fluency can reveal about reading
comprehension. Journal of Research in Reading, 38(3), 213–225. https://doi.org/10.1111/1467-9817.12024.
Veenendaal, N.J., Groen, M.A. & Verhoeven, L. (2016a). Bidirectional relations between text reading prosody
and reading comprehension in the upper primary school grades: A longitudinal perspective. Scientific Studies
of Reading, 20(3), 189–202. https://doi.org/10.1080/10888438.2015.1128939.
Veenendaal, N.J., Groen, M.A. & Verhoeven, L. (2016b). The contribution of segmental and suprasegmental phonology to reading comprehension. Reading Research Quarterly, 51(1), 55–66. https://doi.org/10.1002/rrq.127.
© 2018 UKLA
FERNANDES, QUERIDO, VERHAEGHE & ARAÚJO
Sandra Fernandes, PhD in Cognitive Psychology from the Faculty of Psychology, University of
Lisbon (FPUL), Portugal; Assistant Professor at FPUL; Cognitive Neuropsychologist at Community
Service of FPUL. Research interests on reading and spelling acquisition.
Luís Querido, PhD in Cognitive Psychology from the FPUL, Portugal; Cognitive Neuropsychologist
at Community Service of FPUL, Portugal; Current research focuses on orthographic knowledge and
reading and writing development.
Arlette Verhaeghe, PhD in Psychology; Associate Professor at FFPUL, Portugal; Research interests
on reading and spelling acquisition, and on the impact of literacy on visual object recognition.
Luísa Araújo, PhD in Curriculum and Instruction from the University of Delaware, with a concentration in literacy studies and bilingual education; Full Professor at Instituto Superior de Educação e
Ciências, Lisbon, Portugal; Research interests on literacy and early reading development and second/
foreign language learning.
Received 29 December 2016; revised version received 1 March 2018.
Address for correspondence: Sandra Fernandes, Faculdade de Psicologia, Universidade
de Lisboa, Alameda da Universidade, Lisbon 1649-013, Portugal. Email:
sfernandes@psicologia.ulisboa.pt
© 2018 UKLA
View publication stats