528539

research-article2014
JLDXXX10.1177/0022219414528539Journal of Learning DisabilitiesCarretti et al.

Article
Journal of Learning Disabilities

Oral and Written Expression in Children

2016, Vol. 49(1) 65­–76
© Hammill Institute on Disabilities 2014
Reprints and permissions:
With Reading Comprehension Difficulties sagepub.com/journalsPermissions.nav
DOI: 10.1177/0022219414528539
journaloflearningdisabilities.sagepub.com

Barbara Carretti, PhD1, Eleonora Motta, BA1,

and Anna Maria Re, PhD2

Abstract
Several studies have highlighted that children with reading comprehension difficulties also have problems in tasks that
involve telling a story, in writing or verbally. The main differences identified regard poor comprehenders’ lower level of
coherence in their productions by comparison with good comprehenders. Only one study has compared poor and good
comprehenders’ performance in both modalities (oral and written), however, to see whether these modalities differently
influence poor comprehenders’ performance. We qualitatively and quantitatively compared the performance of good and
poor comprehenders in oral and written narrative tasks with the aim of shedding light on this issue. Regression analyses
were also used to explore the role of working memory and vocabulary in explaining individual differences. Our results
showed that the two groups produced narratives of comparable length, with similar percentages of spelling mistakes,
whereas they differed in terms of the quality of their narratives, regardless of the modality. These differences were
qualified by analyzing the children’s use of connective devices, and poor comprehenders were found to use a higher
proportion of additive devices than good comprehenders. Regression analyses showed that working memory (particularly
the intrusion errors measure) explained a modest part of the qualitative differences in narrative production. Implications
for our theoretical understanding of poor comprehenders’ profiles and education are discussed.

Keywords
poor comprehenders, oral expression, written expression, reading comprehension

To understand a written text completely, readers need to Flower, 1980) have shown the importance of context and
build an integrated and coherent mental representation of long-term memory, and of several cognitive processes,
its meaning (Gernsbacher, 1997). The quality of this men- including planning, transcription, and revision. Writing an
tal representation, or “mental model” (Johnson-Laird, essay involves not only producing ideas, but also organiz-
1983) or “situation model” (Kintsch, 1998), depends on ing them consistently with the task’s objectives. Ideas have
several cognitive and metacognitive factors. From the cog- to be generated, assessed, and compared with the writer’s
nitive viewpoint, reading comprehension requires the acti- aims and the requirements of the task. Objectives may be
vation of an adequate (lexical and semantic) background general and poorly defined at the outset (Bereiter &
knowledge that enables readers to identify relevant infor- Scardamalia, 1987), becoming better defined and more spe-
mation within the text and infer information that is not cific during the transcription and revision of the essay. In
explicitly stated. Working memory plays a crucial part in these latter two important phases of expressive writing,
the construction of a mental model because it enables irrel- other processes relating to orthographic competence and
evant information to be excluded and facilitates connec- monitoring become crucial too (Swanson & Berninger,
tions between items of information in the text (van den 1996). Clearly, all or some of these numerous cognitive pro-
Broek, 2010). From the metacognitive standpoint, readers cesses (planning, producing, organizing and transcribing
need to understand the goal of reading and recognize the ideas, and revising the work as a whole) may be sensitive to
characteristics of different text genres to choose appropri-
ate reading strategies. While reading, they should also 1
Department of General Psychology, University of Padova, Italy
monitor their level of understanding from time to time. 2
Department of Developmental Psychology and Socialization, University
Such control processes ensure the creation of a coherent of Padova, Italy
representation of the text’s meaning.
Corresponding Author:
Like reading, writing—in terms of expressing one’s own Anna Maria Re, Department of Developmental Psychology and
ideas—is a complex activity because many cognitive abili- Socialization, Università degli studi di Padova, Via Venezia 8, Padova,
ties need to be used in the process. Classical studies on 35131, Italy.
expressive writing (Burnett & Kastman, 1997; Hayes & Email: annamaria.re@unipd.it
66 Journal of Learning Disabilities 49(1)

a limited working memory capacity (Berninger & Swanson, verbal topic prompt; in the other, they started from a
1994; McCutchen, 1996, for review; Swanson & Berninger, sequence of six pictures. The results showed that, when nar-
1996). rating from topic prompts, the less-skilled comprehenders
The above considerations led us to hypothesize a close were more likely than either the age-matched skilled com-
link between the ability to express ideas in writing and the prehenders or the younger group matched for comprehen-
ability to understand what others have written, given that sion age to produce stories in which the main events were
both abilities are related to an individual’s vocabulary causally unrelated. The less-skilled comprehenders’ perfor-
knowledge, working memory capacity, and metacognitive mance improved, however, when they narrated from the
control. Consistently with this hypothesis, several studies sequence of pictures, and they were more likely to produce
have demonstrated a two-way relationship between under- stories containing a causally integrated sequence of events
standing texts and writing narratives (see, e.g., Berninger & in this latter condition. Cain and Oakhill’s study thus sug-
Abbott, 2010), but reading comprehension level predicted gested a strong relationship between story comprehension
the quality of written expression better than verbal language skills and quality of oral story production.
skills (listening comprehension or oral expression), thus In a subsequent study, Cain (2003) further analyzed this
suggesting a distinction between the two. With few excep- issue by comparing the sequence of pictures with two kinds
tions, this pattern was found for children from third to sev- of verbal topic prompt, that is, a very short title (as in the
enth grade. The importance of reading comprehension for previous study), and a longer, more informative one, to
developing good written composition skills was also guide the children’s narrative production. In fact, the differ-
reported in a longitudinal study by Abbott, Berninger, and ences identified in Cain and Oakhill’s (1996) earlier study
Fayol (2010), who showed that text comprehension ability might have stemmed from the fact that the pictures were
had significant longitudinal paths to text composition in more informative and provided a skeleton for the narra-
children from second to sixth grade. tive’s construction. Cain’s (2003) later results again demon-
Judging from these data, the complexity of expressive strated that poor comprehenders produced narratives with
writing presumably poses a number of difficulties for any less well-integrated event structures than skilled compre-
child, but especially for children who have problems with henders or younger children matched for comprehension
the underlying processes (Berninger & Rutberg, 1992; age, particularly when short topic titles were used as verbal
Hooper, 2002), such as comprehension difficulties. The story prompts, whereas their performance improved when
most typical problems seen in poor comprehenders concern they were given more informative story prompts. In other
the ability to draw inferences, identify-relevant information words, stories generated from more informative titles or
in a text, connect items of information together, recognize sequences of pictures had more coherent event structures
text structure, and monitor their level of reading compre- than those generated from short topic title prompts. The
hension (e.g., Cain & Oakhill, 2007). Work on poor com- more informative prompts also had a stronger positive
prehenders has generally shown that they perform well in influence on the coherence of the less-skilled compre-
decoding, but have difficulty with semantic aspects of (oral henders’ stories than on that of the other groups, such that
and written) language when it comes, for example, to under- group differences apparent in the short topic title condition
standing discourse structure (Bishop & Snowling, 2004). were not seen when either more informative titles or picture
Poor comprehenders might therefore have much the same sequences were used as prompts.
difficulties in expressing their own ideas verbally or in writ- Cragg and Nation (2006) analyzed the case of written
ing because they find it hard to correlate events and repre- narrative. Groups of poor and good comprehenders were
sent their connections in a mental model, a process that is asked to write an extended narrative about an event depicted
essential to both understanding and producing a story (Arfé in a sequence of fifteen pictures. After a delay, participants
& Boscolo, 2006; van den Broek, 1997; van den Broek, were subsequently asked to recall the story orally. The
Linzie, Fletcher, & Marsolek, 2000). results of this study showed that poor comprehenders had
The written and oral narrative skills of poor compre- age-appropriate spelling skills, and their written narratives
henders have been the object of only a few studies. A first did not differ from those produced by control children in
study conducted by Cain and Oakhill (1996) explored the terms of length or syntactic complexity. But their narratives
ability to develop oral narratives from verbal or visual captured less of the story content, they reported fewer of the
prompts. Two narrative tasks were presented to 7- and main ideas, and their story structure was less sophisticated,
8-year-old children grouped by reading comprehension suggesting that poor comprehenders were unable to capture
level (less-skilled comprehenders and skilled compre- the causal meanings in the story. Their oral story recall mir-
henders), and compared with children matched for reading rored their written output.
comprehension age (i.e., the less-skilled comprehenders In a more recent study, Carretti, Re, and Arfé (2013) fur-
were paired with chronologically younger children). In one ther analyzed the issue of writing skills in poor compre-
task, the children were asked to tell a story starting from a henders by asking groups of poor and good comprehenders
Carretti et al. 67

to produce two different kinds of written text (descriptive inferences can be drawn on the role of expression modality.
vs. narrative) using two different kinds of prompt (pictorial Different hypotheses can be advanced on how the written
vs. verbal) in the narrative text condition. Only five pictures and oral modalities might compare: On one hand, it may be
were used for the pictorial prompt to make the task less that good and poor comprehenders do not differ as a func-
demanding in terms of the amount of information to pro- tion of expression modality because the cognitive processes
cess, and very informative verbal prompts were used to involved in oral and written production are substantially the
avoid differences between the two modalities deriving from same, although the two skills have been demonstrated to be
differences in the amount of information provided. Results separate (Berninger & Abbott, 2010); on the other hand,
showed that poor comprehenders’ performance was mini- writing involves not only cognitive skills but also metacog-
mally influenced by the modality of the prompt: They gen- nitive processes (e.g., Cornoldi, Del Prete, Gallani, Sella, &
erally performed less well than good comprehenders with Re, 2010), so any differences emerging between good and
both verbal and pictorial prompts. Their performance was poor comprehenders could be due to poor comprehenders’
affected by the text genre, however; that is, they performed well-documented more limited metacognitive aptitude
as well as the good comprehenders in the descriptive text (e.g., Cataldo & Cornoldi, 1998). In other words, the writ-
condition, but not in the narrative text condition, especially ing process could be more difficult for poor comprehenders
in terms of coherence and structure. A story’s structure has because it requires metacognitive knowledge and several
to do with the teller’s ability to organize a text according to metacognitive processes, such as the ability to reflect on
the temporal and causal relationships in the narrative. Low narrative aims and structure and to revise the text. These
story structure scores thus reflect narratives that simply skills are not usually well mastered by poor comprehenders
describe the content of each picture (in the pictorial prompt in the context of reading, and this would lead to a lower
condition) or list the events (in the verbal prompt condi- performance on their written expression too by comparison
tion), instead of reporting causal connections between with good comprehenders. Some of these metacognitive
events. The poor comprehenders in this study obtained low competencies (e.g., revision) are not involved in oral
story structure scores both when they had to organize a nar- expression, however, so differences between poor and good
rative on the basis of a set of pictures and when they had to comprehenders could be less strong.
write about a familiar event in the verbal prompt condition. The first goal of the present study was therefore to ana-
The authors concluded that the poor comprehenders’ worse lyze the effect of using a written versus an oral modality to
performance in the narrative text condition depended on the produce narratives, considering both quantitative measures
characteristics of the text genre, where coherence and cau- (length, spelling mistakes) and qualitative aspects (e.g.,
sality were important elements. adherence to the requirements of the task, richness in gen-
Summarizing the above-mentioned studies, children eral, and in terms of vocabulary, syntax, and text structure).
with poor comprehension skills have written and oral In particular, examining the quality of the text structure
expression difficulties, but their performance is partially enabled us to assess the children’s mastery of the elements
influenced by the prompt and particularly by the text genre: typical of a narrative, that is, their inclusion of a beginning,
They find it harder to produce narrative than descriptive which introduces the events and the character; a middle,
texts, probably because the quality of a narrative text relies explaining the initiating events that motivate the protago-
on its coherence and cohesion (e.g., Cain, 2003), aspects to nist’s internal reactions; and an end, reporting the conse-
which poor comprehenders are usually less sensitive quences of the events and the conclusion. This enabled us to
(Perfetti, Landi, & Oakhill, 2005). assess the coherence of the narrative produced by two
Whether or not the oral and written expression modali- groups of good and poor comprehenders. Text coherence
ties differently affect a poor comprehender’s performance was qualified by analyzing the texts’ cohesion in terms of
has never been assessed directly. Most studies considered the type of connectives used. Cohesion is an important
written and oral expression separately, showing that poor aspect of narrative because it maintains local and global
comprehenders fared worse than good comprehenders in links between events, and one way of measuring this char-
both conditions. An exception was the already cited study acteristic in a text is to focus on the use of connectives, seen
by Cragg and Nation (2006), who asked poor and good as a system for connecting semantic content across proposi-
comprehenders to orally recall the main ideas of their writ- tions (Shapiro & Hudson, 1991).
ten narratives. But in this story recall situation (based on an Following the approach taken by Shapiro and Hudson
initially written production) any information not included in (1991) and Cain (2003), we distinguished between four
the written narratives would be unlikely to find a place in a classes of conjunctions: additive, temporal, adversative,
subsequent oral production, so the oral output was some- and causal. Additive connectives (e.g., “and”) and continu-
what biased by the previously administered written task. ative connectives (e.g., “now”) were taken to mean an inde-
Another problem concerns the fact that tasks used to pendent relationship between two clauses; temporal
elicit narratives have also varied considerably, so no connectives (including “then,” “later,” “first”) were
68 Journal of Learning Disabilities 49(1)

Table 1.  Description of the Sample by Group.

Good Comprehenders Poor Comprehenders

  M SD M SD
Age 9.67 0.49 9.50 0.52
Reading comprehension 12.67 0.65 5.25 1.60
Lexical decision task 35.58 6.92 32.92 5.13
PMA Spatial Abilities subscale 15.92 4.23 13.67 3.06
Nonword reading (times) 30.92 14.35 35.25 10.02
Nonword reading (errors) 1.08 0.90 1.50 1.31
Writing speed (number of graphemes) 36.58 6.93 41.50 8.77
Nonword writing (errors) 1.50 1.38 1.50 1.31
WM updating correct recall 22.67 2.81 18.67 5.05
Proportion of intrusion errors 0.36 0.14 0.64 0.36
PMA Vocabulary 23.08 7.57 15.75 8.79

Note. PMA = Primary Mental Abilities; WM = working memory.

considered as denoting a temporal relationship between Vocabulary knowledge has also emerged as an important
clauses; adversative (e.g., “but”) and causal connectives factor in a poor comprehenders’ profile (e.g., Hulme &
(e.g., “because”) were deemed to indicate a dependent rela- Snowling, 2009) and is associated with writing/oral skills
tionship between clauses. (e.g., Babayiğit & Stainthorp, 2011; Berninger, 1999).
Our second aim was to explore the role of certain cogni-
tive measures associated with expressive writing, assessing
their contribution to explaining the performance of good Method
and poor comprehenders. In fact, all previous studies on
narrative production have focused on describing the charac-
Participants
teristics of poor comprehenders’ expressive skill, whereas The study involved 12 poor comprehenders (6 males and 6
the source of their difficulty was never taken into consider- females), aged 8 to 10 years, and 12 good comprehenders (6
ation. Our objective was to start analyzing which cognitive males and 6 females), matched on school grade and type of
processes are involved in poor comprehenders’ narrative school. The children were selected from an original sample
difficulties. From a theoretical standpoint, this would shed of 290 children attending the fourth and fifth grades. All
light on the association between oral and written language children came from families with Italian as their first lan-
skills in studies on individual differences and lead to practi- guage. The two groups were selected on the basis of the
cal suggestions on how to support the narrative skills of general criteria proposed by Cornoldi, De Beni, and
poor comprehenders. We chose to consider working mem- Pazzaglia (1996) and were matched for estimated IQ,
ory, in terms of maintenance (correct recall) and inhibition administering the Spatial Relations subscale of the Primary
(intrusion errors), and vocabulary knowledge for their cen- Mental Abilities (PMA) test (Thurstone & Thurstone, 1963)
tral role in reading comprehension (e.g., Cain, Bryant, & and for word decoding, administering the lexical decision
Oakhill, 2004; Oakhill & Cain, 2012) and narrative expres- task (Caldarola, Perini, & Cornoldi, 2012). The Spatial
sion (e.g., Vanderberg & Swanson, 2007; Yeung, Ho, Chan, Relations subscale of the PMA involves finding one of four
& Chung, 2013). possible figures that can be combined with a given model to
Verbal working memory might explain some of the dif- produce a square within six minutes. The task consisted of
ference between poor and good comprehenders’ expressive 25 items. The lexical decision task (Caldarola et al., 2012)
writing/oral skills, being a weakness of poor comprehenders involves silently reading a list of words and nonwords and
(see the meta-analysis by, e.g., Carretti, Borella, Cornoldi, identifying the nonwords as rapidly as possible.
& De Beni, 2009) and involved in the writing process (e.g., The two groups differed in a standardized reading com-
McCutchen, 1996) and in oral production (Duinmeijer, de prehension test appropriate for their age, the MT test
Jong, & Scheper, 2012). In particular, both the maintenance (Cornoldi & Colpo, 2011), with poor comprehenders
and executive components of verbal working memory may obtaining scores below the 25th percentile, whereas good
be crucial because they enable relevant information to be comprehenders obtained scores above the 75th percentile
kept active and to be integrated in a consistent mental (see Table 1).
model. This clearly emerged from classical studies on read- The above-mentioned tasks were administered collec-
ing comprehension (e.g., Daneman & Carpenter, 1980), tively in a session lasting about 1 hour. Then selected par-
so it probably applies to written/oral production too. ticipants met the experimenter individually and were
Carretti et al. 69

administered several tasks to assess their basic reading and Syntactic structure: sentence construction (use of direct
writing skills. In particular, they completed a nonword read- or indirect discourse) and sentence coordination/subor-
ing task and a nonword writing task taken from an Italian dination, that is, the use of connectives, the appropriate
battery for assessing dyslexia and dysorthographia (Sartori, use of verb tenses, and the proper agreement between the
Job, & Tressoldi, 2007). A writing speed test was also gender and number of nouns, verbs, and adjectives
administered to measure automatization of basic writing
skills, in which participants were asked to write the syllable The qualitative parameters would provide information
“le” continuously for one minute (Tressoldi, Cornoldi, & on the two typical level of analysis used to assess narratives,
Re, 2012). that is, the macrostructure and the microstructure. In par-
Participants obtained comparable performance on all the ticular, the global impression and story structure parameters
reading decoding and writing tasks (see Table 1), as well as provide information on the global coherence of the text,
in the Spatial Relations subscale, whereas they clearly dif- whereas the lexicon and syntactic structure elucidate the
fered in the reading comprehension task, F(1, 22) = 220.58, microstructure of the written and oral texts.
p < .001, η2 = .90. Based on the standardized procedure in the original man-
ual (Tressoldi et al., 2012), each parameter was assessed on
a 5-point scale (1 = considerably below grade, 2 = some-
Materials
what below grade, 3 = grade-appropriate, 4 = somewhat
Oral and written expression.  The tasks consisted of cartoon above grade, and 5 = considerably above grade). Two of
strips adapted from an Italian battery for assessing writing the authors (blinded to participants’ grouping according to
skills (Tressoldi et al., 2012). There were two strips con- their comprehension skills) rated the written and oral narra-
taining five cartoons each, one strip telling the story of a tives. The interrater agreement was almost perfect (accord-
child falling off a tree, the other describing an old lady ing to the guidelines of Landis & Koch, 1977), so the
being robbed by a thief in the street; the two strips were analyses were run on the first author’s assessment. Cohen’s
designed to have the same characteristics (in terms of famil- kappa coefficients are shown in the appendix, along with
iarity, number of characters, number of inferences required). separate descriptive statistics for each rater.
Each participant was asked to give a written or verbal
account of the story represented in the cartoon strip to Text cohesion.  To gain a better understanding of the differ-
enable a friend to understand what happened to the main ences between our poor and good comprehenders, their per-
character (the child or the old lady). The oral narratives formance on the oral and written production tasks was also
were audiotaped and then transcribed. assessed in terms of text cohesion. The proportion of con-
Performance on the two tasks was measured both quan- nectives used in the narrative was computed, distinguishing
titatively and qualitatively. First, the number of words pro- between additive, temporal, adversative and causal words
duced and the number of spelling mistakes (as a percentage (Cain, 2003; Cain, Patson, & Andrews, 2005; Shapiro &
of the total number of words) were recorded and considered Hudson, 1991), and calculating their proportions in all the
as quantitative measures (the latter only applied to the writ- propositions in the narrative, where a proposition consisted
ten expression task). of a subject and a predicate, as suggested by Cain (2003).
For the qualitative measures, two blinded raters were
asked to examine the texts considering five qualitative Understanding of cartoon strips.  After completing each writ-
parameters and using established assessment measures (Re, ing task, participants were asked two questions to ascertain
Cazzaniga, Pedron, & Cornoldi, 2009; Tressoldi et al., their real understanding of the cartoon strips; one of the
2012), that is, questions focused on factual information represented in the
cartoon series, the other required an inference about the
Adherence: participants’ ability to meet the task’s event. Correct answers were awarded 1 point each.
requirements; a higher score was therefore assigned to
written texts addressed to a fictional friend and describ- Working memory updating task.  This task was adapted from
ing what happened or what the participant had seen the relevance-based updating task proposed by Palladino,
General impression: the coherence and richness of the Cornoldi, De Beni, and Pazzaglia (2001; see also Belacchi,
ideas presented Carretti, & Cornoldi, 2010). It consisted of six lists, each
Text structure: the organization of the text was judged to comprising eight highly familiar and concrete object words
be satisfactory when it was arranged into three parts, (Barca, Burani, & Arduino, 2002) easily comparable for
with a beginning (introducing the event), a middle size. The task involved participants listening to each list of
(explaining what happened), and an end (reporting the object words and then selecting the three smallest objects in
outcome) each list. The number of correctly recalled words and the
Lexicon: the appropriateness and variety of the words percentage of intrusion errors (calculated by dividing the
used number of intrusions by the number of correctly recalled
70 Journal of Learning Disabilities 49(1)

Table 2.  Descriptive Statistics for Oral and Written Expression Tasks by Group.

Good Comprehenders Poor Comprehenders

  Oral Written Oral Written

  M SD M SD M SD M SD
Number of words 63.25 20.18 68.67 22.22 58.33 14.24 61.75 31.45
% of spelling mistakes 1.18 1.11 0.68 1.05
Adherence 3.17 0.58 3.25 0.45 3.00 0.43 2.83 0.39
General impression 3.67 0.98 4.00 1.04 2.33 0.49 2.67 0.88
Text structure 3.58 0.90 3.92 1.08 2.58 1.08 2.58 0.67
Lexicon 3.67 1.15 3.82 0.94 2.58 0.52 2.42 0.52
Syntactic structure 3.67 1.23 4.00 1.04 1.92 0.52 2.25 0.97

words and multiplying by 100) were the dependent no differences emerged in these parameters, the effect of
variables. group, F(1, 22) = 2.71, p = .11, η2 = .11, and modality, F(1,
22) = 3.19, p = .09, η2 = .12, and the interaction between
Vocabulary.  This task, taken from the PMA test (Thurstone them, F < 1, were not significant.
& Thurstone, 1963), consisted of 30 words and partici- For general impression, the results showed a main effect
pants were asked to identify synonyms of each word from of group, F(1, 22) = 29.26, p < .001, η2 = .57, with the good
among 4 options. The task was timed and participants had comprehenders’ oral and written production being judged
5 minutes to complete as many items as they could. The more consistent and richer than that of the poor compre-
total number of correct answers was the dependent henders. The effect of modality, F(1, 22) = 4.63, p = .043, η2
variable. = .17, was significant too, the written texts being judged
better than the oral narratives. The Group × Modality inter-
Procedure action, F(1, 22) = 1.16, p = .294, η2p = .05, was not
significant.
The writing tasks and the working memory task were
For text structure, our results showed a main effect of
administered individually in a session lasting about 45 min-
group, F(1, 20) = 12.68, p < .01, η2 = .37, that is, the
utes. The version of the cartoon strip and the oral or written
good comprehenders’ narrative followed the typical
expression modality used for the task were counterbalanced
structure of a story, whereas this was not the case for the
across participants, to avoid effects relating to the material
poor comprehenders. The effect of modality, F < 1, and
being used.
the Group × Modality interaction, F < 1, were not
significant.
Results For lexicon, again there was only a main effect of group,
Quantitative parameters. The differences between the F(1, 22) = 16.17, p < .01, η2 = .42, the good comprehenders’
two groups in terms of the number of words produced were use of lexicon being found more appropriate than that of the
analyzed with a mixed-design, repeated-measures ANOVA, poor comprehenders. Neither the effect of modality, F < 1,
with Group as the between-subjects factor and Modality nor the Group × Modality interaction, F(1, 22) = 1.57, p =
(oral vs. written) as the within-subject factor. No differ- .22, η2p = .06, was significant.
ences emerged between the two groups, F < 1, neither the For syntactic structure, the analysis yielded only a main
effect of Modality nor the Group × Modality interaction effect of group, F(1, 22) = 25.13, p < .001, η2 = .53, the
proving significant (for both F < 1; see Table 2). good comprehenders’ syntax being considered more appro-
No differences emerged between the two groups for the priate than the poor comprehenders’ (see Table 2). The
percentage of spelling mistakes in the written expression effect of modality, F(1, 22) = 3.03, p = .095, η2 = .12, and
task either, F(1, 22) = 1.31, p = .262, η2 = .06. the Group × Modality interaction, F < 1, were not
significant.
Qualitative parameters.  The descriptive statistics for the
qualitative parameters are shown in Table 2. We first consid- Text cohesion.  Text cohesion was assessed by analyzing
ered the participants’ adherence to the task’s requirements: the proportions of connectives used in the oral and written
Carretti et al. 71

Figure 1.  Proportion of connectives in the clauses produced by good and poor comprehenders.

narratives (see Figure 1). A 2 (groups: good vs. poor) × 2 Working memory updating. Univariate ANOVA on
(modalities: written vs. oral) repeated-measures analysis of the number of correctly recalled words and the propor-
variance was run for each type of connective word (addi- tion of intrusion errors showed a significant difference
tive, temporal, adversative, and causal). in both these dependent variables, F(1, 22) = 5.75, p
For the additive connectives, our results showed a main < .05, η 2 = .21 and F(1, 22) = 6.16, p < .05, η 2 = .22,
effect of modality, F(1, 22) = 9.45, p < .01, η2 = .32, with respectively, with poor comprehenders recalling fewer
oral productions containing a higher proportion of additive words and making more intrusion errors than good
connectives than written texts, and also of group, F(1, 22) = comprehenders.
5.22, p < .01, η2 = .19, with poor comprehenders using a
higher proportion of additive connectives than good com- Vocabulary.  Univariate ANOVA on the number of cor-
prehenders. The Group × Modality interaction was not sig- rect answers showed a significant difference between the
nificant, F < 1. groups, F(1, 22) = 4.79, p < .05, η2 = .18, again with good
The analysis on temporal connectives yielded no sig- comprehenders outperforming poor comprehenders.
nificant effects: group, F < 1; modality, F < 1; Group ×
Modality, F < 1. The same applied to the adversative con- Correlations. Since modality did not affect the two
nectives: group, F(1, 22) = 1.32, p = .26, η2 = .05; modal- groups’ performance differently, the qualitative assessments
ity, F < 1; Group × Modality, F(1, 22) = 2.71, p = .11, were averaged and correlated with the cognitive measures,
η2 = .11. so we considered narrative production, regardless of the
For the causal connectives, the Group × Modality inter- modality. Although the indexes were generally different
action was significant, F(1, 22) = 4.94, p < .05, η2 = .18. from zero, only the proportion of intrusion errors was asso-
Post hoc comparisons with Tukey’s test showed that the ciated with the global quality of the text and the richness of
two groups differed marginally (p = .062) in the oral lexicon (see Table 3). These correlations can be considered
modality, with good comprehenders using a higher pro- large according to Cohen’s (1988) guidelines.
portion of causal connectives than poor comprehenders.
Good comprehenders also tended to use more causal con- Regression analyses. Hierarchical regression analyses
nectives in the oral condition than in their written texts were run to estimate the percentage of variance of the quali-
(p = .011). No differences emerged from the other com- tative measures (criterion variables) explained by group,
parisons. There were no significant effects of either group working memory (WM; correct recall and intrusion errors),
or modality. and vocabulary. The WM measures (correct recall and intru-
sion errors) were entered in Step 1, the vocabulary measure
Understanding of cartoons.  The two groups’ performance in Step 2, and the group variable in Step 3 (converted into
was compared using a univariate ANOVA and no differ- a dummy variable, 1 for good comprehenders, 0 for poor
ences emerged. comprehenders).
72 Journal of Learning Disabilities 49(1)

Table 3.  Correlations Between Qualitative Indexes and The effects of the written and oral modalities had not
Cognitive Measures (Vocabulary and Working Memory [Recall previously been examined simultaneously, however; the
and Intrusions]). few studies examining these issues generally focused on
Vocabulary WM Recall WM Intrusions one modality or the other. An exception was the study by
Cragg and Nation (2006), who asked poor and good com-
General impression .300 .324 –.485* prehenders to orally recall the main ideas of their written
Text structure .211 .283 –.223 narratives, but (as mentioned in the introduction) basing the
Lexicon .329 .286 –.415*
story recall situation on an initially written production
Syntactic structure .278 .338 –.368
meant that the oral output could have been somewhat biased
Note. WM = working memory. by the previously administered written task.
*p < .05. To further examine the narrative abilities of poor com-
prehenders, the first goal of the present study was to iden-
tify any influence of the written or oral modality of the
In the first step, WM measures (correct recall and intru- narrative task on individual differences in reading compre-
sion errors) contributed to the explained variance for the hension. We concentrated on narrative tasks because they
general impression measure (R2 = .20, p < .05) with the pro- had proved harder for poor comprehenders to handle than
portion of intrusion errors as a unique predictor (β = –.73, p other kinds of production (Carretti et al., 2013). Two car-
< .05); no contribution was made by vocabulary. But when toon strips showing familiar characters were presented and
the group dummy variable was entered, the contribution of participants were asked to tell the story in writing or ver-
the WM measures was no longer significant, so the group bally. To examine the role of certain cognitive processes in
dummy variable (β = .60, p < .001) was the only salient accounting for individual differences in narrative produc-
predictor. The model explained 45% of the variance. tion, we also administered a verbal WM task and a vocabu-
As for text structure, only the group dummy variable lary knowledge test.
contributed to explaining the variance (R2 = .26, β = .68, Our findings indicate that using the oral or written
p < .01). modality made no difference to the performance of poor
The quality of the groups’ lexicon was marginally pre- comprehenders, whose output was worse than that of
dicted by WM (R2 = .14, p = .07) and group (R2 = .22, p < good comprehenders, in terms of richness and lexical and
.05); the group dummy variable (β = .58, p < .05) was the syntactic complexity whichever modality they used, con-
only salient predictor, however. firming the results of previous studies (e.g., Carretti et
When it came to syntactic structure, the group dummy al., 2013). On the other hand, the two groups did not dif-
variable (β = .73, p < .001) was the only salient predictor. fer in terms of the length of their narratives, the percent-
The model explained 45% of the variance. age of spelling mistakes they made, or their adherence to
the task’s requirements, meaning that poor compre-
henders’ difficulties are not due to the cognitive effort
Discussion and Conclusions involved in completing the task or understanding what it
Several reports in the literature have suggested that compre- requires. In other words, the lower quality of poor com-
hension skills are associated with the ability to tell a story prehenders’ written and oral narratives cannot be attrib-
verbally or in writing because much the same cognitive pro- uted to differences in their basic writing skills (they did
cesses are involved (Berninger et al., 2006). In a 1-year lon- just as well as the good comprehenders in tests measur-
gitudinal study, for instance, Babayiğit and Stainthorp ing their writing speed and spelling).
(2011) explored the association between reading and writ- Our results give the impression that part of the difference
ing skills in two cohorts of primary school children. Their lies in that narratives produced by poor comprehenders are
results showed that qualitative measures of narrative text less cohesive (as our analysis on the connectives they used
writing, assessing the clarity and richness of the text, were would suggest). Judging from our findings, poor compre-
associated with reading comprehension skills at different henders tend to use more additive connectives than good
time points. These results led authors to suggest the need for comprehenders, and fewer causal connectives—particu-
a more comprehensive approach to the study of literacy larly in the oral modality—and this would explain the gen-
skills (e.g., Abbott et al., 2010; Juel, 1988; Shanahan, 1984; erally lower quality of the poor comprehenders’ output:
Shanahan & Lomax, 1986). rather than a story (that also involves reporting causal con-
This relationship also emerged when individual differ- nections between events), their narratives seemed more a
ences in reading comprehension were considered, poor list of events with a description of each picture. On the
comprehenders revealing difficulties in both oral (e.g., strength of these results, poor comprehenders’ lower-qual-
Cain, 2003; Cain & Oakhill, 1996) and written forms of ity narrative productions may derive from a weaker ability
expression (e.g., Cragg & Nation, 2006), particularly when to represent the story’s meaning. In other words, our poor
they had to produce a narrative (Carretti et al., 2013). comprehenders were less able to fully understand the events
Carretti et al. 73

depicted in the cartoons and this affected their narratives. important mechanism in explaining WM performance and
This would be consistent with the idea that comprehension individual differences, particularly in relation to reading
skills transcend expression modality, as suggested by other comprehension (e.g., Cain, 2006; Carretti, Cornoldi, De
reports of less-skilled comprehenders having more trouble Beni, & Romanò, 2005; Pimperton & Nation, 2010), so fail-
understanding stories presented in auditory or pictorial for- ure to control information may partly explain differences in
mats (e.g., Gernsbacher, Varner, & Faust, 1990). In the case written/oral expression. The variance explained by the dif-
of the present study, however, the poor and good compre- ference between the good and poor reading comprehension
henders’ comparable performance on answering the com- groups nonetheless remained the stronger predictor of writ-
prehension questions would challenge this explanation. ing quality, suggesting that other variables not considered in
Some caution is warranted, however, because only two the present study, such as metacognitive components, may
questions were asked in the present study, one requiring the have a part to play.
recall of factual information and the other an inference. To write a narrative adequately, writers should be able to
Further studies would be needed to completely disentangle assess the importance of the events involved and the rela-
this issue. tionship between them, and to organize this information by
As mentioned previously, our analysis on connectives providing causal and temporal links. Differences in the
suggested that a poor narrative production is associated organization of a narrative’s structure could depend partly
with, or derives from a poor use of connectives, and this on an individual’s prior knowledge, but the events narrated
would fit in with the report from Cain et al. (2005) of poor in the cartoon used in our study are very common in real
comprehenders having more difficulty understanding and life, so their role in explaining any differences should be
using connectives. In the Cain et al. study, poor and good very small. A possible source of individual differences
comprehenders were shown texts in which some conjunc- could relate to people’s metacognitive knowledge about
tions had been omitted and asked to choose the appropri- story structure. Several studies have suggested that writers
ate word from among three options. The poor are supported by their familiarity with the story structure of
comprehenders’ performance was generally worse in narrative (e.g., Scardamalia & Bereiter, 1986), so a possible
terms of the number of correct answers, and they com- reason for the difference between poor and good compre-
pleted the cloze inappropriately, irrespective of the type of henders might relate to the former’s inadequate knowledge
conjunction. So there may be a link between their poor of text structure, which is a common feature of poor com-
understanding and their poor use of conjunctions in oral or prehenders (e.g., Cain & Oakhill, 1996; Yuill & Oakhill,
written forms of expression, as already suggested by Cain 1991).
(2003), for instance. An idea-generating phase is followed by writers tran-
It therefore seems that poor comprehension skills affect scribing and then revising their texts. Other processes relat-
people’s ability to tell a story in written and oral form, and ing to orthographic competence and monitoring become
their difficulty stems partly from the fact that their narra- crucial in the latter two, fundamental phases of expressive
tives lack those elements that give a story its structural writing (Swanson & Berninger, 1996). Since orthographic
complexity. It is important to add that a note of caution is skills are not usually impaired in poor comprehenders (e.g.,
warranted in interpreting our results, however, because Cragg & Nation, 1996), the monitoring phase might be crit-
narrative production was measured using only two tasks ical, given poor comprehenders’ tendency not to adopt
(one for the written and one for the oral modality). metacognitive strategies to monitor their level of perfor-
Obtaining more indicators of the same skills would enable mance (e.g., Garner, 1981). This has been observed particu-
us to be more confident about the general applicability of larly in the context of reading, but an aptitude for
our findings, so future research should include more tasks metacognition usually has a more general effect, so such a
to obtain more robust measures and improve their tendency might plausibly be expected to emerge in other
reliability. learning contexts too. Future research should therefore pay
Concerning the processes involved in narrative produc- attention to the role of metacognitive knowledge and moni-
tion, our correlation analyses showed that WM measures toring in accounting for individual differences between
are associated with the quality measures we considered, as poor and good comprehenders.
in previous reports (e.g., Swanson & Berninger, 1996; From an educational standpoint, it would be important
Vanderberg & Swanson, 2007). In particular, the intrusion to know what stage of the writing process poses problems
errors measure, which refers to the ability to inhibit no lon- for poor comprehenders to design remediation programs.
ger relevant information, contributed to the variance Some studies have demonstrated that expressive writing
explained in the general impression of the narratives’ performance can be enhanced by providing facilitators
quality and, marginally, in the vocabulary quality designed to improve an individual’s text organization (e.g.,
parameters. Inhibition has often been considered an Re, Caeran, & Cornoldi, 2008). For instance, the method
74 Journal of Learning Disabilities 49(1)

used in the study by Re and colleagues (2008) was based on In conclusion, our study indicates that reading compre-
breaking down the writing process into a series of steps and hension and narrative expression are related, and individual
providing boxes in which children were asked to write their differences in comprehension skills are uninfluenced by the
ideas. When children with attention-deficit/hyperactivity use of an oral or written modality (poor comprehenders’
disorder were instructed to use this procedural aid when performance is poor in both cases). Texts produced by poor
planning how to organize their texts, the quality of their out- comprehenders are characterized by a greater use of addi-
put improved. It would be interesting to assess the utility of tive conjunctions, and this probably explains the lower
this procedure in the case of poor comprehenders too. quality of their output.

Appendix
Descriptive Statistics Separately for the Two Independent Raters and Their Level of Agreement (Cohen’s
Kappa Coefficient)

Rater 1 Rater 2

  M SD M SD Cohen’s κ
 Adherence Written 3.12 0.45 3.12 0.45 1.00
  Oral 2.96 0.55 3.00 0.51 .90
  General impression Written 3.33 1.17 3.33 1.17 .97
  Oral 3.08 0.97 3.00 1.02 .87
  Text structure Written 3.25 1.11 3.29 1.12 .83
  Oral 3.13 1.08 3.08 1.10 .89
 Lexicon Written 3.13 1.03 3.17 1.01 .93
  Oral 3.17 1.01 3.13 1.03 .93
  Syntactic structure Written 3.12 1.33 3.08 1.21 .84
  Oral 2.83 1.27 2.79 1.28 .82

Note. All coefficients are significant at p < .001.

Declaration of Conflicting Interests New insights from a transparent orthography. Journal of

Educational Psychology, 103, 169–189.
The author(s) declared no potential conflicts of interest with
Barca, L., Burani, C., & Arduino, L. S. (2002). Word naming
respect to the research, authorship, and/or publication of this
times and psycholinguistic norms for Italian nouns. Behavior
article.
Research Methods, Instruments, & Computers, 34, 424–434.
Belacchi, C., Carretti, B., & Cornoldi, C. (2010). The role of work-
Funding ing memory and updating in Coloured Raven Matrices perfor-
The author(s) received no financial support for the research, mance in typically developing children. European Journal of
authorship, and/or publication of this article. Cognitive Psychology, 22, 1010–1020.
Bereiter, C., & Scardamalia, M. (1987). The psychology of written
composition. Hillsdale, NJ: Lawrence Erlbaum.
References Berninger, V. (1999). Coordinating transcription and text genera-
Abbott, R. D., Berninger, V. W., & Fayol, M. (2010). Longitudinal tion in working memory during composing: Automatic and
relationships of levels of language in writing and between constructive processes. Learning Disability Quarterly, 22,
writing and reading in Grades 1 to 7. Journal of Educational 99–112.
Psychology, 102, 281–298. Berninger, V. W., & Abbott, R. D. (2010). Listening comprehen-
Arfé, B., & Boscolo, P. (2006). Causal coherence in deaf and sion, oral expression, reading comprehension, and written
hearing students’ written narratives. Discourse Processes, 42, expression: Related yet unique language systems in grades 1,
271–300. 3, 5, and 7. Journal of Educational Psychology, 102, 635-651.
Babayiğit, S., & Stainthorp, R. (2011). Modeling the relation- Berninger, V. W., Abbott, R. D., Jones, J., Wolf, B. J., Gould, L.,
ships between cognitive-linguistic skills and literacy skills: Anderson-Youngstrom, M., . . . Apel, K. (2006). Early devel-
Carretti et al. 75

opment of language by hand: Composing, reading, listening, Cohen, J. (1988). Statistical power analysis for the behavioral sci-
and speaking connections; three letter-writing modes; and ences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.
fast mapping in spelling. Developmental Neuropsychology, Cornoldi, C., & Colpo, G. (2011). Prove di lettura MT-2 per la
29, 61–92. scuola primaria [MT-2 tasks for the assessment of reading in
Berninger, V. W., & Rutberg, J. (1992). Relationship of finger primary school]. Florence, Italy: Giunti OS.
function to beginning writing: Application to diagnosis of Cornoldi, C., De Beni, R., & Pazzaglia, F. (1996). Reading
writing disabilities. Developmental Medicine and Child comprehension profiles. In C. Cornoldi & J. Oakhill (Eds.),
Neurology, 34, 155–172. Reading comprehension difficulties: Processes and interven-
Berninger, V. W., & Swanson, H. L. (1994). Modifying Hayes and tion (pp. 113–136). Mahwah, NJ: Lawrence Erlbaum.
Flower’s model of skilled writing to explain beginning, devel- Cornoldi, C., Del Prete, F., Gallani, A., Sella, F., & Re, A. M.
oping writing. In E. C. Butterfield (Ed.), Children’s writing: (2010). Components affecting expressive writing in typical
Toward a process theory of the development of skilled writing and disabled children. Advanced in Learning and Behavioral
(pp. 57–81). Greenwich, CT: JAI. Disabilities, 23, 269–286.
Bishop, D. V. M., & Snowling, M. J. (2004). Developmental dys- Cragg, L., & Nation, K. (2006). Exploring written narrative in
lexia and specific language impairment: Same or different? children with poor reading comprehension. Educational
Psychological Bulletin, 130, 858–886. Psychology, 26, 55–72.
Burnett, R., & Kastman, L. (1997). Teaching composition: Current Daneman, M., Carpenter, P. A. (1980). Individual differences in
theories and practice. In G. Phye (Ed.), Handbook of aca- working memory and reading. Journal of Verbal Learning
demic learning: Construction of knowledge (pp. 268–305). and Verbal Behavior, 19, 450-466.
San Diego, CA: Academic Press. Duinmeijer, I., de Jong, J., & Scheper, A. (2012). Narrative abili-
Cain, K. (2003). Text comprehension and its relation to coherence ties, memory and attention in children with a specific lan-
and cohesion in children’s fictional narratives. British Journal guage impairment. International Journal of Language and
of Developmental Psychology, 21, 335–351. Communication Disorders, 47, 542–555.
Cain, K. (2006). Individual differences in children’s memory and Garner, R. (1981). Monitoring of passage inconsistency among
reading comprehension: An investigation of semantic and poor comprehenders: A preliminary test of the “piecemeal
inhibitory deficits. Memory, 14, 553–569. processing” explanation. Journal of Educational Research,
Cain, K., Bryant, P. E., & Oakhill, J. (2004). Children’s read- 74, 159–162.
ing comprehension ability: Concurrent prediction by work- Gernsbacher, M. A. (1997). Two decades of structure building.
ing memory, verbal ability, and component skills. Journal of Discourse Processes, 23, 265–304.
Gernsbacher, M. A., Varner, K. R., & Faust, M. E. (1990).
Educational Psychology, 96, 31–42.
Investigating differences in general comprehension skill.
Cain, K., & Oakhill, J. (1996). The nature of the relationship
Journal of Experimental Psychology: Learning, Memory and
between comprehension skill and the ability to tell a story.
Cognition, 16, 430–445.
British Journal of Developmental Psychology, 14, 187–201.
Hayes, J. R., & Flower, L. S. (1980). Identifying the organiza-
Cain, K., & Oakhill, J. (Eds.). (2007). Children’s comprehension
tion of writing processes. In L. W. Gregg & E. R. Steinberg
problems in oral and written language: A cognitive perspec-
(Eds.), Cognitive processes in writing (pp. 3–30). Hillsdale,
tive. New York, NY: Guildford.
NJ: Lawrence Erlbaum.
Cain, K., Patson, N., & Andrews, N. (2005). Age- and ability-
Hooper, S. R. (2002). The language of written language: An intro-
related differences in young readers’ use of conjunctions.
duction to the special issue. Journal of Learning Disabilities,
Journal of Child Language, 32, 877–892.
35, 2–6.
Caldarola, N., Perini, N., & Cornoldi, C. (2012). DLC: Una prova Hulme, C., & Snowling, M. J. (2009). Developmental disorders
di decisione lessicale per la valutazione collettiva delle abil- of language, learning and cognition. Oxford, UK: Wiley-
ità di lettura [DLC: A lexical decision task for the collective Blackwell.
assessment of reading]. Dislessia, 9, 93–108. Johnson-Laird, P. N. (1983). Mental models. Cambridge, UK:
Carretti, B., Borella, E., Cornoldi, C., & De Beni, R. (2009). Cambridge University Press.
Role of working memory in explaining poor comprehenders’ Juel, C. (1988). Learning to read and write: A longitudinal study
performance: A meta-analysis. Learning and Individual of 54 children from first through fourth grades. Journal of
Differences, 19, 246–251. Educational Psychology, 80, 437–447.
Carretti, B., Cornoldi, C., De Beni, R., & Romanò, M. (2005). Kintsch, W. (1998). Comprehension: A paradigm for cognition.
Updating in working memory: A comparison between poor New York, NY: Cambridge University Press.
and good comprehenders. Journal of Experimental Child Landis, J. R., & Koch, G. G. (1977). The measurement of observer
Psychology, 91, 45–66. agreement for categorical data. Biometrics, 33, 159–174.
Carretti, B., Re, A. M., & Arfé, B. (2013). Reading comprehension McCutchen, D. (1996). A capacity theory of writing: Working
and expressive writing: A comparison between good and poor memory in composition. Educational Psychology Review, 8,
comprehenders. Journal of Learning Disabilities, 46, 87–96. 299–325.
Cataldo, M. G., & Cornoldi, C. (1998). Self-monitoring in poor Oakhill, J., & Cain, K. (2012). The precursors of reading ability in
and good reading comprehenders and their use of strategy. young readers: Evidence from a four-year longitudinal study.
British Journal of Developmental Psychology, 16, 155–165. Scientific Studies of Reading, 16, 91–121.
76 Journal of Learning Disabilities 49(1)

Palladino, P., Cornoldi, C., De Beni, R., & Pazzaglia, F. (2001). Shapiro, L. R., & Hudson, J. A. (1991). Tell me a make-believe
Working memory and updating processes in reading compre- story: Coherence and cohesion in young children’s picture-
hension. Memory and Cognition, 29, 344–354. elicited narratives. Developmental Psychology, 27, 960–974.
Perfetti, C. A., Landi, N., & Oakhill, J. (2005). The acquisition of Swanson, L. H., & Berninger, V. W. (1996). Individual differences
reading comprehension skill. In M. J. Snowling & C. Hulme in children’s working memory and writing skill. Journal of
(Eds.), The science of reading: A handbook (pp. 227–247). Experimental Child Psychology, 63, 358–385.
Oxford, UK: Blackwell. Thurstone, T. G., & Thurstone, L. L. (1963). Primary Mental
Pimperton, H., & Nation, K. (2010). Suppressing irrelevant infor- Abilities. Chicago, IL: Science Research Associates.
mation from working memory: Evidence for domain-specific Tressoldi, P. E., Cornoldi, C., & Re, A. M. (2012). Batteria per
deficits in poor comprehenders. Journal of Memory and la valutazione della scrittura e della competenza ortografica
Language, 62, 380–391. nella scuola dell’obbligo [BVSCO battery for the assessment
Re, A. M., Caeran, M., & Cornoldi, C. (2008). Improving expres- of writing skills of children from 7 to 13 years old]. Florence,
sive writing skills of children rated for ADHD symptoms. Italy: Giunti OS.
Journal of Learning Disabilities, 41, 535–544. van den Broek, P. (1997). Discovering the cement of the universe:
Re, A. M., Cazzaniga, S., Pedron, M., & Cornoldi, C. (2009). The development of event comprehension from childhood to
Io Scrivo: Valutazione e Potenziamento delle Abilità di adulthood. In P. van den Broek, P. Bauer, & T. Bourg (Eds.),
Espressione Scritta [I can write: Assessment and empower- Developmental spans in event comprehension and represen-
ment of expressive writing skills]. Florence, Italy: Giunti OS. tation (pp. 321–342). Mahwah, NJ: Lawrence Erlbaum.
Sartori, G., Job, R., & Tressoldi, P. E. (2007). DDE-2. Batteria per van den Broek, P., Linzie, B., Fletcher, C., & Marsolek, C. (2000).
la valutazione della dislessia e della disortografia evolutiva The role of causal discourse structure in narrative writing.
[Battery for the assessment of developmental dyslexia and Memory & Cognition, 28, 711–721.
dysorthographia]. Florence, Italy: Giunti OS. Vanderberg, R., & Swanson, H. L. (2007). Which components
Scardamalia, M., & Bereiter, C. (1986). Research on written com- of working memory are important in the writing process?
position. In M. Wittroch (Ed.), Handbook of research on Reading and Writing, 20, 721–752.
teaching (3rd ed.). New York: Macmillan Education Ltd. Yeung, P., Ho, C. S., Chan, D. W., & Chung, K. K. (2013).
Shanahan, T. (1984). Nature of reading-writing relations: An Contribution of oral language skills, linguistic skills, and tran-
exploratory multivariate analysis. Journal of Educational scription skills to Chinese written composition among fourth-
Psychology, 76, 466–477. grade students. Discourse Processes, 50, 498–529.
Shanahan, T., & Lomax, D. (1986). An analysis and comparison Yuill, N., & Oakhill, J. (1991). Children’s problems in text com-
of theoretical models of the reading-writing relationship. prehension: An experimental investigation. Cambridge, UK:
Journal of Educational Psychology, 78, 116–123. Cambridge University Press.