WORKING MEMORY, READING COMPREHENSION

Running head: WORKING MEMORY, READING COMPREHENSION

Working memory contributions to reading comprehension components in middle childhood children

Elisavet Chrysochoou, Zoe Bablekou, & Nikolaos Tsigilis

Aristotle University of Thessaloniki, Greece

Addresses: Dr Elisavet Chrysochoou Department of Early Childhood Education, Aristotle University of Thessaloniki, Thessaloniki 54124, Greece Tel.: +30 6937 382929/ + 30 2311 247994 e-mail: [email protected]

Associate Professor Dr Zoe Bablekou Department of Early Childhood Education, Aristotle University of Thessaloniki, Thessaloniki 54124, Greece Tel. & Fax: +30 2310 995033 e-mail: [email protected]

1

Dr Nikolaos Tsigilis Department of Physical Education and Sport Science, Aristotle University of Thessaloniki, 60, I. Polemi Str., Thessaloniki 54248, Greece Tel.: +30 2310 320858 email: [email protected]

Correspondence: Dr Elisavet Chrysochoou, 133 Tsimiski Str., Thessaloniki 54621, Greece Tel.: +30 6937 382929/ + 30 2311 247994 e-mail: [email protected]

2

Abstract The paper examines working memory contributions to reading comprehension sub-skills in Greek children (mean age 9:1 years). The phonological loop of the Baddeley and Hitch working memory model was assessed with three recall tasks (words, nonwords and digits) and a word list matching task. The central executive (CE) was assessed with three tasks (listening, counting, and backward digit recall). Participants were also given a receptive vocabulary task, a reading fluency task, and written stories accompanied by different comprehension questions. Canonical correlation analyses showed that the comprehension variables were related to the CE, rather than the phonological loop measures. CE functions were more strongly associated with elaborative inference generation (requiring demanding off-line processing) and comprehension control (involving metacognitive monitoring). Smaller, yet significant associations were observed between the CE and the necessary inference and literal comprehension measures, whereas a moderate relationship was found in the case of the simile comprehension variable. Among the CE variables, listening recall demonstrated the highest loading on the canonical function, followed by moderate, yet significant counting and backward digit recall loadings. Vocabulary was found to fully mediate several associations between working memory and comprehension measures; however, the relation between listening recall and elaborative inferences was partly mediated. Reading fluency and, on several occasions, Greek vocabulary knowledge did not mediate the relations between CE measures and comprehension skills assessed. This study demonstrates the usefulness of CE measures for identifying young children’s possible difficulties in carrying out specific reading comprehension processes.

Key-words: reading comprehension skills, central executive, phonological loop, working memory, children

3

ACKNOWLEDGEMENTS * This research was partly supported by a post-doctoral grant of the Research Committee of the Aristotle University of Thessaloniki, Greece. * We would like to thank Professor David P. MacKinnon for making valuable suggestions regarding the mediation analyses conducted.

4

Working memory contributions to reading comprehension components in middle childhood children

The relationship between reading comprehension and working memory in childhood has been studied extensively (Cain, Oakhill, & Bryant, 2004; Dufva, Niemi, & Voeten, 2001; Leather & Henry, 1994; Seigneuric & Ehrlich, 2005; Seigneuric, Ehrlich, Oakhill, & Yuill, 2000; Swanson & Berninger, 1995; Yuill, Oakhill, & Parkin, 1989). However, most work in this field investigates English- or French-speaking children and has assessed comprehension as a single component rather than assessing specific reading comprehension subcomponents, offering relatively limited data on their relationship with children’s working memory capacity. Besides, for many years working memory skills have been assessed with several simple span (e.g. word or digit span) and complex span tasks (e.g. listening or counting span) that were not included in broader, standardized working memory test batteries (e.g. Dufva et al, 2001; Leather & Henry, 1994; Seigneuric et al., 2000). The Working Memory Test Battery for Children (Pickering & Gathercole, 2001) has enabled a theoretically more stable and broad-ranging assessment of working memory capacities, in the frame of the Baddeley and Hitch (1974; see Baddeley, 2007) multi-component working memory model. This test battery was used in our study with Greek-speaking children, along with a story comprehension task that tapped five comprehension-related skills. Literal questions tapped children’s recall of text information. Inferential questions were of two types: necessary inferences that measured ability to incorporate general knowledge with text information, in order to fill gaps of meaning necessary for extracting the text ideas, and elaborative inferences that simply enriched text representation. Simile comprehension questions assessed the ability to comprehend similes in the text. Finally, comprehension control questions provided children with information that was inconsistent with the gist, thus evaluating text representation adequacy and meta-cognitive control processes. In the frame of the Baddeley and Hitch model, the present study can help improve understanding of the involvement of the phonological loop and central executive (CE) components in each off-line text comprehension

5

process. Young children, who have not mastered their linguistic skills yet, may not manage to efficiently process linguistic constructs within stories (on-line) and may particularly need to depend on off-line processes, carried out in relation to questions following the stories, on the basis of their representations in long-term memory (see Gathercole & Baddeley, 1993). As far as story comprehension is concerned, the sophisticated functions of the CE in the multi-component working memory model (Baddeley, & Hitch, 1974; see Baddeley, 2007) are expected to support the activation of story information, syntactic and semantic processing in relation to comprehension questions, and temporary maintenance of the processing products before responses are provided (see Baddeley, 1996, 2007; Gathercole & Baddeley, 1993). The phonological loop would play a less significant role: it could maintain the phonological records of semantically complex questions and facilitate their off-line processing (see Baddeley, 1997; Martin, 1990; Gathercole & Baddeley, 1993; Waters, Caplan, & Hildebrandt, 1987). In line with this assumption, Leather and Henry (1994) found that two measures of CE function (listening and counting span) made a unique contribution to 7.5 year olds’ overall performance in a reading comprehension task, after phonological loop contributions had been taken into account. Moreover, Swanson and Howell (2001) reported that the phonological and executive systems made independent contributions to the prediction of age-related changes in reading comprehension and word recognition, which were stronger in the case of the CE (see also Daneman & Merikle, 1996; Georgiou, Das, & Hayward, 2008; Gottardo, Stanovich, & Siegel, 1996). Similarly, in a longitudinal study with 11-17 year-olds, Swanson and Jerman (2007) showed that working memory (controlled attention), rather than short-term memory (phonological loop) was associated with growth in reading comprehension and reading fluency. Deriving from the above, it should come as no surprise that the CE, rather than the phonological loop functions have also been found to relate to comprehension difficulties. Less-skilled comprehenders performed significantly worse than good comprehenders on word- and number-based CE assessments (Cain, 2006b, Experiment 2; Oakhill, Hartt, &

6

Samols, 2005), but they did not differ on verbal and numerical measures of short-term memory (Cain, 2006b, Experiment 1). Despite the evidence reported with regard to children’s overall performance on reading comprehension tasks, research relating working memory to specific comprehension skills is scarce (see Cain, 2006a). There is some evidence to suggest that neither memory for the text itself (Cain & Oakhill, 1999; Oakhill, 1984) nor knowledge deficits (Cain, Oakhill, Barnes, & Bryant, 2001) can fully account for children’s inference generation difficulties. Besides this, it has also been demonstrated that children’s comprehension monitoring ability is impaired when inconsistent sentences are apart in a text (setting higher memory demands), rather than adjacent (see summary of findings in Cain, 2006a). To the best of our knowledge, however, there is only one study that has investigated the relations between specific comprehensionrelated skills and working memory capacity in typically developing children, by carrying out independent working memory assessments (rather than manipulating working memory demands within texts). In doing so, Cain and colleagues (2004) found that performance on literal questions about short stories was not related to the CE measures taken at either 8.5 or 10.5 years of age. In contrast, these researchers found that a sentence span (a CE measure) was significantly correlated at both age levels to comprehension processes that require deeper, constructive processing, such as inference generation and comprehension monitoring. The present study aims at providing better understanding of the relations between component skills of reading comprehension and different verbal working memory measures, by taking into account the possible mediating roles of vocabulary and word reading skills; their contributions to children’s overall performance in reading comprehension tasks are wellestablished (e.g. Seigneuric & Ehrlich, 2005; Seigneuric et al., 2000; Swanson & Berninger, 1995). Besides vocabulary and word decoding, comprehension depends heavily on higherlevel processes, such as integration of text information with prior knowledge in the context of inference generation and simile comprehension, as well as on metacognitive control processes involved in comprehension monitoring (see Cain, 2006a; Cain et al., 2004; Oakhill, Cain, & Bryant, 2003; Yuill & Oakhill, 1991). Storage and processing coordination, strategy selection

7

and operation, activation and manipulation of long-term memory information in relation to these processes would tax attentional CE resources (see Baddeley, 1996, 2007; Gathercole & Baddeley, 1993). Thus, the reading comprehension and CE measures should be significantly associated, independent of any vocabulary and word reading mediations. There is some evidence to support the above hypothesis. Cain and colleagues (2004) used a comprehension task that set significant memory demands. Children were discouraged from returning to the texts, in order to answer orally presented questions. Researchers found that a composite CE measure made a significant contribution to overall reading comprehension performance at three assessment points (7.5, 8.5 and 10.5 years), after the effects of vocabulary knowledge, verbal IQ, and word reading ability had been controlled for. In a study with French children, Seigneuric and colleagues (Seigneuric & Ehrlich, 2005; Seigneuric et al., 2000) used a reading comprehension task that assessed different comprehension components, including sentence processing, syntactic processing, fact retrieval, pronominal reference and inference generation in texts. Memory demands were kept to a minimum (children provided written answers to fill-in blank questions and were allowed to look back at the texts); yet, these researchers found that CE measures made a significant, unique contribution to 9 year olds’ (Seigneuric & Ehrlich, 2005) and 10 year olds’ (Seigneuric et al., 2000) overall reading comprehension performance when contrasted with vocabulary and decoding skills (see also Swanson & Berninger, 1995; and Yuill et al., 1989 for similar findings). Nevertheless, there is no evidence yet on the mediating role of vocabulary and word reading fluency in the relationship between verbal working memory capacity and different reading comprehension skills. Clearly, further work is needed to examine the relations between the phonological and executive components of working memory and the comprehension of similes, the generation of different types of inferences (e.g. necessary as opposed to elaborative ones), the ability to monitor one’s comprehension. The above comprehension elements are crucial for efficient comprehension and their evaluation may uncover difficulties affecting educational attainment (see Block, Rodgers, & Johnson, 2004;

8

Cain, 2006a; Cain & Oakhill, 2006; McGee & Johnson, 2003). Going beyond prior work, we provide a more detailed understanding of these relationships in a study with Greek-speaking children who are assessed with various phonological loop and CE tasks and five types of comprehension questions (i.e. literal, necessary, and elaborative inferences, simile comprehension, and comprehension control) about written stories. In terms of age, we were interested in 8-10 year old children, because of the significant changes occurring during this developmental phase in both working memory (see Cowan, 1997; Gathercole, 1998, 1999; Gathercole & Baddeley, 1993; Pickering & Gathercole, 2001) and comprehension skills (see Barnes, Dennis, & Haefele-Kalvaitis, 1996; Curtis, 1980; Paris, Lindauer, & Cox, 1977; Swanson & Howell, 2001). These were also the youngest children who could cope with the demands of inference generation, simile comprehension and comprehension control within a reading comprehension context. We stated one general and one more specific research question: 1) Is there a relationship between reading comprehension skills and the two working memory components in middle childhood? We expected that the CE measures will be better correlated with the reading comprehension scores, as compared to the phonological loop measures (Hypothesis 1). 2) Are the observed relations mediated by vocabulary and reading fluency? We expected the CE measures to be significantly related to the reading comprehension skills, irrespective of any vocabulary and reading fluency contributions (Hypothesis 2).

Method Participants Ninety-two children (Mean age = 9:1, Range = 8-10:1) participated, coming from different schools of the city of Thessaloniki (about a million inhabitants), Greece. Forty-three children attended third-grade (Mean age = 8:7) and forty-nine were fourth-grade (N = 49, Mean age = 9:6). They were randomly selected, according to their odd or even ID number in their class list. Participants did not have a language or learning difficulties diagnosis. In

9

selecting the population, classroom teachers’ evaluations were also taken into account; according to those, participants did not manifest any such problems at school. Design Participants were individually tested in the school setting. Three testing sessions were carried out, one for the working memory, one for the vocabulary, and one for the reading comprehension and reading fluency measures. Each session lasted approximately 25-30 minutes and took place on a different day of the week. Testing sessions (working memory, vocabulary, reading comprehension/ reading fluency) were given in a fixed random order, following every child’s ID number in the experiment participant list. Reading comprehension and reading fluency measures (carried out in the same session) were counterbalanced. Different measures within the working memory session were given in a fixed order, according to the test manual instructions. Τasks were presented in a fashion that avoided overtaxing the same working memory component during successive subtests (see Pickering & Gathercole, 2001). Specifically, the digit recall and the word list matching (phonological loop measures) were administered first, followed by the listening recall (CE measure); then the word list recall (phonological loop measure), the counting recall (CE measure), the nonword list recall (phonological loop measure), and finally, the backward digit recall (CE measure) were given. With regard to the reading comprehension session, stories and questions were presented in a fixed random order, again following each child’s ID number in the list. Tasks and procedure Reading comprehension was assessed using a battery of five short stories (one practice and four assessment ones) and comprehension questions. Stories and some of the questions were drawn from Oakhill’s (1984) and Cain and Oakhill’s (1999) studies. Extra questions were added and the stories were appropriately adapted for use with Greek children (Chrysochoou, 2006). Every story was followed by ten questions, two questions assessing each of five comprehension elements (see Appendix). Participants read each story once and the researcher asked the questions, while children did not have access to the stories. Stories and questions were presented in a fixed random order. The child’s number of correct

10

responses to the questions tapping every single comprehension skill in the four stories constituted his/ her test score for every skill (Max: 8). As mentioned in the introduction, literal comprehension questions tapped children’s ability to recall information explicitly stated in the story. Children were also required to appropriately integrate text information with general knowledge so as to make sense of information implied, rather than explicitly stated in the text. Two types of inferential questions were included: those requiring inferences necessary for extracting the basic meaning of the text and those tapping the generation of elaborative inferences, which simply enrich text representation. There also were questions referring to comprehension of similes in the text. The fifth category, comprehension control questions, contrasted the basic meaning of a story and assessed the accuracy of story gist (see Appendix). We used the tasks included in the Working Memory Test Battery for Children (Pickering & Gathercole, 2001) to assess the verbal components of the Baddeley and Hitch working memory model. Language-based tasks were translated from English to Greek and were appropriately adapted. The phonological loop was assessed with four tasks. Three of the tasks tapped serial recall of spoken verbal stimuli that varied in terms of familiarity (digits, words or nonwords). Finally, in a word list matching task, children had to decide if word order was the same in two sequences of words presented to them orally. Words (nouns, adjectives, and verbs) were translated from English to Greek; as in the English battery, they provided minimal opportunities of being semantically grouped in the recall lists (see Pickering & Gathercole, 2001). Instead of the 1-syllable words used in the English battery, tasks consisted of 2-syllable words; 1-syllable verbs are scarce in Greek and 1-syllable adjectives and nouns simply do not exist (Porpodas, 2002). The nonword list recall task consisted of 2-syllable nonwords, created using the same pool of sounds as in the words used for the word list recall subtest, following the English task. Three tasks, placing demands for simultaneous verbal processing and storage were used to tap the CE of the model (a listening recall, a backward digit recall, and a counting recall task). In a listening recall task, children listened to a set of short sentences. They judged the

11

veracity of every sentence (e.g. scissors cut paper is true, balls are square is false) and they then had to recall the final word of each sentence in the order it was presented (e.g. paper, square). Sentences were translated from English to Greek and the spoken duration of each sentence was kept up to 2 seconds, the same as in the English version of the task. Let us note that, besides concurrent storage and processing, the listening recall task taps an executive function that is important in comprehension: the activation, maintenance and manipulation of information in long-term memory (see Baddeley, 1996). In the backward digit recall task, children listened to a list of digits and had to recall it in reverse order. Finally, in the counting recall task, they overtly counted a set of dot arrays and were required to serially recall the tallies. Children were presented with blocks of trials in every task. Each block consisted of six trials of the same difficulty level. For example, in the first block of the word list recall task, the child was presented with one word, in the second block, with lists of two words, etc. In the first block of the counting recall (CE) task, the child was presented with one set of dots arrays, in the second block, with two sets of dot arrays, and so on. List length increased gradually, along with task demands (digit recall: 1-9, word list matching: 2-9, word list recall: 1-7, nonword list recall: 1-6, listening recall: 1-6, counting recall: 1-7, backward digit recall task: 2-7). Testing ceased when children failed any three trials in a particular block. Therefore, a child’s score on the test was the sum of his/ her correct responses. Moreover, children’s receptive vocabulary was assessed with the Peabody Picture Vocabulary Test (Dunn, 1965), which has been translated and used in several Greek studies with young children (e.g., Chrysochoou & Bablekou, 2010; Natsopoulos, Stavroussi, & Alevriadou, 1998). Participants were presented with one word and four pictures at a time and they were asked to find the appropriate word-picture matching. Testing ceased after six errors in eight consecutive trials. Performance on the test was the sum of correct responses. Finally, in a reading fluency task (Kotoulas, 2007) participants were asked to read as accurately and as quickly as possible each of 100 stimuli that appeared on a computer screen. The test consisted of 80 words (nouns, verbs, adjectives, adverbs, pronouns, prepositions,

12

conjunctions, clauses), mixed with 20 nonwords (constructed by replacing several syllables or changing syllable order in existing words) assessing decoding processes that could not have been supported by the child’s lexicon. Words were drawn from the reading textbooks used in all six elementary school grades, so that children would be familiar with some, but not all words in the test (see Kotoulas, 2007). Stimuli were representative of all Greek language phonemes and graded on the basis of their syllable number (ranging from 1-6). Because of the test grading nature, presentation order remained fixed. Let us note that, contrary to evidence from irregularly spelled languages like English (see Alcock, Nokes, Ngowi, Musabi, Mbise, Mandali, Bundy, & Baddeley, 2000; Frith, 1985; Marsh, Friedman, Desberg, & Saterdahl, 1981; Thorstad, 1991), the logographic strategy does not play a role in beginning to read in Greek (see Porpodas, 2001, 2002). Phonologically based reading becomes fluent in Greek-speaking children by 7 years already (Bablekou, Pita, & Kiosseoglou, 2001; Porpodas, 2001). This could be attributed to the extreme regularity between graphemes and phonemes in Greek (see Harris & Giannouli, 1999; Holton, Mackridge, & Philippaki-Warburton, 1997). There is evidence to suggest that in other regularly spelled languages (e.g. Italian and Spanish) children also use a systematic phonological strategy and they learn to read more quickly and accurately than Englishspeaking children (Thorstad, 1991; see also Alcock et al., 2000). As word reading in Greek becomes accurate early on, a reading measure that also takes reading speed into account (reading speed improves significantly during elementary school years), was considered to be a more sensitive index of children’s reading ability (Porpodas, 2002). Reading fluency thus equaled the total number of words participants read accurately in the first minute of the reading task (Kotoulas, 2007; see also Lesaux, Rupp, & Siegel, 2007; Lionetti & Cole, 2004; vanAuken, Chafouleas, Bradley, & Martens, 2002).

Results Descriptive statistics and preliminary analyses

13

Mean scores, standard deviations and score ranges for the overall group and for the two age levels (8.5 and 9.5) separately were calculated in every single task and they are presented in Table 1. Scores fell within ranges that avoided attenuation resulting from either floor or ceiling effects. In general, the mean working memory scores were comparable to those obtained in relevant studies with young children (see Alloway, Gathercole, Willis, & Adams, 2004; Gathercole, Pickering, Ambridge, & Wearing, 2004; Gathercole, Pickering, Knight, & Stegmann, 2004).

Table 1 about here

Multivariate analyses of variance showed that participants’ performance on the reading comprehension tasks (Wilks’ λ= .08, F (5, 86) = 1.58, p = .18), the phonological loop (Wilks’ λ= .05, F (4, 87) = 1.06, p = .38) and the CE tasks (Wilks’ λ= .92, F (3, 88) = 2.48, p = .07) was not differentiated across grades. Thus, further analyses were conducted on the whole sample (N = 92). Correlation coefficients between all measures are given in Table 2. Prior studies have examined the temporal stability of the variables measured using a testretest approach. In particular, good reliability coefficients (.74-.79) have been reported for the vocabulary test when used with 8-10 year-old children (Dunn, 1965). Moreover, Pickering and Gathercole (2001) found test-retest reliability coefficients ranging from .38 to an excellent .82, with a mean of .51, for the seven tests of the working memory battery we employed. With regard to the internal consistency, Cronbach’s alpha coefficients were .79 for the reading comprehension battery, .63 for the phonological loop and .60 for the CE batteries in the present study. These findings are comparable to those obtained with Greek-speaking elementary school children (Chrysochoou & Bablekou, 2010; see also Chrysochoou, 2006 for reliability and validity analyses). Moreover, reliability of the Greek reading fluency task was examined in a study of 280 elementary school children and the reliability coefficients (Cronbach’s alpha) reported were excellent for the words (0.90 for the third- and 0.91 for the

14

fourth-grade children) and adequate for the nonwords used in the task (0.61 for the third- and 0.69 for the fourth-grade) (see Kotoulas, 2007).

Table 2 about here

Multivariate relationships In relation to the first research question and hypothesis, the multivariate relationship between the reading comprehension measures and the measures of every working memory component was examined using canonical correlation analysis. This is a powerful multivariate statistical technique, recommended for examining the association between two sets of variables (see Hair, Anderson, Tatham, & Black, 1995). It can offer protection against type I error (Sherry & Henson, 2005). In addition, simultaneous examination of the complex pattern of relationships among several variables offers a better approximation of reality. Correlation analysis also places fewer assumptions on the type of data, the information derived is of higher quality and the results easier interpretable. In fact, many frequently used parametric statistical techniques (e.g., t-tests, analysis of variance, MANOVA, multiple regression analysis) can be regarded as special cases of canonical analysis (see Tabachnick & Fidell, 2007; Thompson, 2000). Overall, two canonical correlation analyses were conducted to examine the multivariate relationship between (a) the reading comprehension and phonological loop variables, and (b) the reading comprehension and CE variables. Let us note that Tabachnick and Fidell (2007) propose that 10 participants per variable are adequate for reliable findings to be obtained. In the present study, the participant/ variable ratio was 10.2 (92/9) for the first canonical correlation analysis and 11.5 (92/8) for the second one. Additionally, the canonical correlation results should be interpreted in light of the magnitude of the canonical correlation coefficient derived (Stevens, 2002). The first canonical analysis had a non significant value (Rc = .26, p = .87), suggesting a weak association which is of little practical importance (Stevens, 2002).

15

However, a second analysis, conducted on the comprehension and CE variables, revealed a significant canonical correlation coefficient of Rc = .49 (Wilks’ λ = .65, p = .001), verifying our first hypothesis. Α moderate canonical correlation (.50) based on 100 participants can be detected about 67% of the time (see Stevens, 2002); thus, our analysis has good chances to detect a meaningful effect. The relative contribution of a variable to the canonical associations was based on canonical loadings. These loadings represent the simple linear correlation between a variable and its respective canonical variate and they are interpreted in a fashion similar to factor analysis loadings. Therefore, a loading is expected to range from 0 to 1 and is interpreted in the same manner as a correlation coefficient. According to Stevens (2002), interpretation of a loading should be a function of sample size. Based on the guidelines provided (Table 11.1, p. 384), a loading was considered significant if it exceeded the value of .512. Loadings of the two sets of variables on the canonical function are shown in Table 3. They were all above the cut-off point, with most of them yielding a high value. Highest loadings were demonstrated by the listening recall measure of CE function, the elaborative inference, and the comprehension control measures. Good loadings were demonstrated by the necessary inference and literal comprehension measures. Significant, but moderate, were the loadings of the counting and backward digit recall measures (CE), as well as of simile comprehension measures. The strength of the relationship between the two sets of variables was examined using the average squared canonical correlation (Cramer & Nicewander, 1979). Since only one canonical function was significant, the average squared canonical correlation equals the squared of the obtained canonical correlation, which indicates that the model explained about 24% of the variance shared between the variable sets.

Table 3 about here

16

Mediation analyses We then examined whether the observed associations between the working memory and the reading comprehension measures were mediated by vocabulary and reading fluency. According to our second hypothesis, the CE measures will make independent contributions to the prediction of the reading comprehension skills after controlling for vocabulary and reading fluency. It should be pointed out that the mediation hypothesis would only be valid if the following conditions are satisfied (see Baron & Kenny, 1986): (i) the initial variable is related to the outcome (path c), (ii) the initial variable is related to the mediator (path a), (iii) the mediator is related to the outcome variable, controlling for the effects of the initial variable (path b), and (iv) for full mediation to be established, the initial variable does not have an effect on the outcome, after controlling for the mediator; otherwise, mediation is only partial (path c΄). However, the first condition for establishing a mediation effect has been seriously questioned (MacKinnon & Fairchild, 2009; MacKinnon, Fairchild, & Matthew, 2007) and was thus not taken into account. Structural equation modeling procedures were employed to conduct a mediation analysis, using AMOS ver. 5.0 (Arbuckle, 1997). The advantage of this procedure over a series of regression analyses is that all equations are simultaneously conducted, thus reducing type I error. Preliminary multivariate analysis showed that there were significant differences between the two age groups on vocabulary and reading fluency (Wilks’ λ= .681, F (2, 889) = 20.81, p < .001). Thus age was included in the mediation model as a predictor of all variables to adjust for its effect (MacKinnon, 2008). Prior to conducting the analyses, the required assumptions were examined. Results are only provided for those variables that satisfied the above mentioned requirements (Table 4). Specifically, vocabulary fully mediated most associations in the analyses conducted. Our second hypothesis was only verified in the case of the elaborative inference and listening recall measures: they also shared a unique association, after controlling for vocabulary knowledge. Finally, mediation analyses were not conducted in relation to reading fluency, since the criteria set for paths a and b were not met in any case.

17

Let us note that the participant/variable ratio in the mediation analyses was satisfactory (92/4 = 23), guaranteeing relatively stable results (see Tabachnick & Fidell, 2007). Moreover, employment of bootstraping procedures, provided by AMOS, contributes to our confidence of the derived results (see Arbuckle, 1997). These procedures were used to calculate indirect effects (on 2000 bootstrap samples), along with their significance levels and a 90% confidence interval (see Table 4).

Discussion The present study adds to our knowledge on the relationship between children’s working memory capacities and reading comprehension in several ways. Going beyond prior work (which has focused on a single comprehension skill) we demonstrated significant relations between several comprehension skills and CE (rather than phonological loop) resources, verifying our first hypothesis. It should be noted that relevant evidence has been scarce so far (Cain, 2006a) and has concerned CE measures only, whereas important skills (Cain et al, 2004) such as the comprehension of similes, necessary and elaborative inference generation, and off-line comprehension control processes had not been investigated in relation to working memory capacity. We found that the CE resources (with listening recall being the most sensitive measure) were more closely related to elaborative inference generation and comprehension control; they both require demanding off-line processing, with comprehension control also involving meta-cognitive monitoring processes. The present study was also a first attempt to investigate the mediating role of vocabulary and reading fluency in the relations between specific comprehension skills and concurrent working memory measures. Vocabulary fully mediated several relations; however, in line with our second hypothesis, listening recall (CE) was related to elaborative inference generation, irrespective of any vocabulary contributions. Moreover, reading fluency in Greek, and in many cases vocabulary, did not mediate the relations between the CE and comprehension measures. In the paragraphs that follow we discuss evidence more thoroughly, indicating contributions to literature and educational practice.

18

Among the inferential skills assessed, elaborative inference generation was more closely related to CE functions than necessary inference skills. Both inferential questions tap recall of on-line and off-line processing products in relation to their comprehension. Off-line processing, however, might be especially called for in the case of the elaborative inferences, thus requiring extra CE support: it has been suggested that they are not necessarily drawn online, since they simply enrich text representation rather than affect gist comprehension (see Corbett & Dosher, 1978; Garnham, 1982; Yuill & Oakhill, 1991). In line, vocabulary mediated the relation between the listening recall and necessary inference measures fully, whereas the relation between the elaborative inference and listening recall measures only partly. The veracity of this assumption should be explored further, along with the role of working memory capacity in other types of inferential skills (e.g. inferences relying on the integration of information stated in the text, rather than involving general knowledge, too). The comprehension control measure was also very strongly related to CE function in the canonical correlation analysis, demonstrating a loading equal to that of the elaborative inference variable. This is not surprising considering that the comprehension control questions also involve significant off-line processing of story information, combined with taxing meta-cognitive control procedures. Dependence of comprehension control on attentional CE resources was observed in a study of young children’s oral comprehension skills, too (Chrysochoou & Bablekou, 2010; see also Chrysochoou, 2006). The identification of factors that contribute to each specific comprehension skill in oral and reading settings, as well as the relationships between oral and reading comprehension skills in childhood is of vital importance and should be explored further (see also Nation, Cocksey, Taylor, & Bishop, 2010). Oral comprehension, for example, has been found to significantly mediate the association

between

preschooler’s

working

memory

function

and

their

reading

comprehension performance in second grade (see Dufva et al., 2001). The loading of the literal comprehension variable on the canonical function was significant, yet weaker than those demonstrated by the comprehension control and the inferential variables. Indeed, literal comprehension is considered to require more superficial

19

maintenance and manipulation of text information, rather than the kind of deep, constructive processing involved in drawing inferences or controlling one’s comprehension (Yuill & Oakhill, 1991; see also Cain, 2006a). Not surprisingly, children attained the highest mean scores in the literal and the simile questions, whereas simile comprehension shared the weakest (yet significant) relation with the CE in the canonical analysis conducted. It would be interesting to examine if this could be attributed to children’s greater reliance on general knowledge (namely on long-term memory) when interpreting similes, thus requiring less conscious, effortful processing and less taxing problem-solving executive operations (see Baddeley, 2007; Gathercole & Baddeley, 1993). Whether a different form of simile questions would differentiate findings (e.g. asking participants “How did Alexis manage to reach the biscuit tin” instead of repeating the simile in the questions we used “Like an acrobat, Alexis reached the biscuit tin; what does it mean?”, see Appendix) also remains to be seen. A first attempt was made to assess how two key reading comprehension predictors (vocabulary and reading fluency) (see Cain et al., 2004; Seigneuric & Ehrlich, 2005; Seigneuric et al., 2000) mediate the relations between working memory and specific (rather than general) comprehension measures. We found that vocabulary fully mediated most associations in the analyses conducted, after having controlled for age. Among the phonological loop measures only word list matching was entered in the analyses, nevertheless it was not associated with any comprehension variable after controlling for vocabulary. In some cases, this specific working memory measure has been differentiated from the three serial recall tasks of phonological loop capacity (digit, word, and non word recall) in studies aiming at the development of the Working Memory Test Battery for Children; its validity at different age levels should be examined further (see Gathercole and Pickering, 2000; see also Pickering & Gathercole, 2001). The listening recall measure of CE demonstrated the highest loading in the canonical correlation analyses, probably because it shares most with language comprehension: they both involve sentence comprehension, strategy selection and operation, long-term memory search and retrieval, and verbal recall (see Baddeley, Logie, Nimmo-Smith, & Brereton, 1985;

20

Gottardo et al., 1996). Yet, vocabulary fully mediated most associations between the specific CE measure and comprehension skills in the mediation analyses. There was one exception: even after this stringent verbal ability control, listening recall continued to share a unique association with the elaborative inference measure, in harmony with our second hypothesis. It is also noted that vocabulary was not a mediator, neither between listening recall and comprehension control nor between the other two CE measures (counting and backward digit recall) and all five comprehension skills. Our findings fit well with evidence from 9.5 yearolds’ oral comprehension skills (Chrysochoou & Bablekou, 2010). It could thus be concluded that verbal skills underlie (see also Nation, Adams, Bowyer-Crane, & Snowling, 1999), but cannot fully account for the observed associations between verbal working memory and comprehension

variables.

Investigations

of

CE

contributions

to

overall

reading

comprehension performance in English-speaking (e.g. Cain et al., 2004) and French-speaking children (Seigneuric et al, 2000) have arrived at similar conclusions. A final line of similar data derives from Greek-speaking children’s oral comprehension (Chrysochoou & Bablekou, 2010). Interestingly, we found no relations between reading fluency in Greek and the working memory or reading comprehension variables in the analyses examining satisfaction of the mediation criteria. Reading fluency was not a mediator in our case, in contrast to evidence with English-speaking children (see Baddeley, 2007; Cain, 2006a; Dehn, 2008; Swanson & Howell, 2001). However, there is evidence to suggest that children learning to read in Greek and in languages with regular grapheme-phoneme correspondence rely on a phonological strategy (Bablekou et al., 2001; Thorstad, 1991) and they learn to read more quickly and accurately than children who speak English (Porpodas, 2001, 2002; also see Alcock et al., 2000; Thorstad, 1991). By the age of 10, Greek children approach reading orthographically (Bablekou et al., 2001). It could thus be proposed that, with age and reading experience, phonological loop contributions to phonological recoding gradually subside (see Gathercole & Baddeley, 1993); this happens earlier with Greek readers, along with the indirect effect of the phonological loop (via word reading) on reading comprehension performance. Besides,

21

the phonological loop was not related to the higher-level processes in the present comprehension task (e.g. inference generation and comprehension control), despite its reported contributions to the temporary maintenance of semantically complex material that necessitates off-line processing (see Baddeley, 1997; Gathercole & Baddeley, 1993). As shown in the canonical correlation analyses, such higher-level processing demanded CE involvement; however, the involvement was not mediated by reading fluency in Greek. Lack of relevant studies limits discussion, and therefore relevant knowledge would benefit notably from direct comparisons of the mediating role of reading fluency (and vocabulary) in the relations between verbal working memory components and reading comprehension skills, in different age groups and linguistic contexts. Summing up, the present study is indicative of the relevance of the multi-component working memory model to the understanding of specific text comprehension processes. We demonstrated the important role the executive system (as opposed to phonological loop) resources can play when young children are faced with higher-order reading comprehension skills; such skills require concurrent storage and off-line processing of text information, regulation of information flow from long-term to working memory (e.g. in drawing inferences), and metacognitive control processes (see Baddeley, 1996, 2007, and Gathercole & Baddeley, 1993 for a discussion of CE functions). The present evidence also denotes the mediating role of vocabulary, as opposed to the non-mediating role of reading fluency in a linguistic context (Greek) that is different from those most research comes from (e.g. the English and the French). Nevertheless, it seems that the period between 8 and 10 years is also a sensitive one for the relationships between the CE and reading comprehension in Greek. Future research with larger populations coming from different linguistic contexts could provide a more profound understanding of the working memory functions influencing every specific comprehension skill, at different age levels. So far, the limited number of longitudinal and intervention studies inhibit the formulation of causally linked and well-integrated text comprehension models (see Cain, 2006a). This is of vital importance for creating effective learning environments and for facilitating reading development (see Block et al., 2004); also

22

for identifying comprehension difficulties and their causes early on, thus preventing poor educational attainment (see Cain, 2006a; Cain & Oakhill, 1999, 2006). Future research should explore the role of certain executive functions, such as switching retrieval plans (Baddeley, 1996, 2007) during reading comprehension. It should also shed more light into possible effects of the executive system on reading comprehension difficulties. Relevant investigations are still limited (see Sesma, Mahone, Levine, Eason, & Cutting, 2009) and do not involve specific comprehension components. A challenging research field emerges here, which could enlighten us further on whether it is more appropriate to view the CE as a unitary attention controller serving multiple functions, or as an executive committee of interacting, still relatively autonomous control processes (see Bablekou, 2009; Baddeley, 1996).

23

References Alcock, K. J., Nokes, K., Ngowi, F., Musabi, C., Mbise, A., Mandali, R., Bundy, D., Baddeley, A. (2000). The development of reading tests for use in a regularly spelled language. Applied Psycholinguistics, 21, 525-555. Arbuckle, J. L. (1997). AMOS users’ guide: Version 3.6. Chicago: SPSS Alloway, T. P., Gathercole, S. E., Willis, C., & Adams, A. M. (2004). A structural analysis of working memory and related cognitive skills in young children. Journal of Experimental Child Psychology, 87, 85-106. Bablekou, Z. (2009). Nous: Cognitive models of working memory. In C. Mourlas, N. Tsianos, & P. Germanakos (Eds.), Cognitive and emotional processes in web-based education: Integrating human factors and personalization (pp.86-109). Hershey, PA: Information Science Reference. Bablekou, Z., Pita, R., & Kiosseoglou, G. (2001). A study of silent reading development in children aged 7-10 years. In V. Deligianni-Kouimtzi and A. Psalti (Eds.), Scientific Annals of the School of Psychology, Aristotle University of Thessaloniki [in Greek], Vol. 4 (pp. 181- 205). Thessaloniki, Aristotle University of Thessaloniki Editions. Baddeley, Α. D. (1996). Exploring the central executive. The Quarterly Journal of Experimental Psychology, 49 A (1), 5-28. Baddeley, Α. D. (1997). Human memory: Theory and Practice (Revised Edition). East Sussex: Psychology Press. Baddeley, A. (2007). Working memory, thought, and action. New York: Oxford Psychology Series. Baddeley, A. D., & Hitch, G. (1974). Working memory. In G. A. Bower (Ed.), Recent advances in learning and motivation, Vol. 8. New York: Academic Press. Baddeley, A. D., Logie, R., Nimmo-Smith, I., & Brereton, N. (1985). Components of fluid reading. Journal of Memory and Language, 24, 119-131.

24

Barnes, M.A., Dennis, M., & Haefele-Kalvaitis, J. (1996). The effects of knowledge availability and knowledge accessibility on coherence and elaborative inferencing in children from six to fifteen years of age. Journal of experimental Child Psychology, 61, 216-241. Baron, R. M. & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173-1182. Block, C. C., Rodgers, L. L., & Johnson, R. B. (2004). Comprehension process instruction: Creating reading success in grades K-3. N.Y. : The Guilford Press. Bowyer-Crane, C., Snowling, M. J., Duff, F. J., Fieldsend, E., Carroll, J. M., Miles, J., Gotz, K., & Hulme, C. (2008). Improving early language and literacy skills: Differential effects of an oral language versus a phonology with reading intervention. Journal of Child Psychology and Psychiatry, 49, 422-432. Cain, K. (2006a). Children’s reading comprehension: The role of working memory in normal and impaired development. In S. J. Pickering (Ed.), Working memory and Education (pp. 61-91). London: Academic Press. Cain, K. (2006b). Individual differences in children’s memory and reading comprehension: An investigation of semantic and inhibitory deficits. Memory, 14(5), 553-569. Cain, K., & Oakhill, J. V. (1999). Inference making ability and its relation to comprehension failure in young children. Reading and Writing: An Interdisciplinary Journal, 11, 489-503. Cain, K. & Oakhill, J. V. (2006). Profiles of children with specific reading comprehension difficulties. British Journal of Educational Psychology, 76, 683-696. Cain, K., Oakhill, J. V., Barnes, M. A., & Bryant, P. E. (2001). Comprehension skill, inference-making ability and their relation to knowledge. Memory and Cognition, 29, 850859. Cain, K., Oakhill, J. V., & Bryant, P. E. (2004). Children’s reading comprehension ability: Concurrent prediction by working memory, verbal ability, and component skills. Journal of Educational Psychology, 96 (1), 31-42.

25

Chrysochoou, Ε.

(2006).

Working

memory

contributions

to

children’s

listening

comprehension in early and middle childhood years. Unpublished Ph.D. Thesis, Department of Early Childhood Education, Aristotle University of Thessaloniki [in Greek]. Chrysochoou, Ε. & Bablekou, Z. (2010). Phonological loop and central executive contributions to oral comprehension skills of 5.5 to 9.5 years old children. Applied Cognitive Psychology, DΟΙ: 10.1002/acp. 1723. Corbett, A. T. & Dosher, B. A. (1978). Instrument inferences in sentence encoding. Journal of Verbal Learing and Verbal Behavior, 17, 479-491. Cowan, N. (1997). The development of working memory. In N. Cowan (Ed.). The development of memory in childhood (pp. 163-199). London: Psychology Press. Cramer, E. M. & Nicewander, W. A. (1979). Some symmetric, invariant measures of multivariate association. Psychometrica, 44, 43-54. Curtis, M. E. (1980). Development of components of reading skill. Journal of Educational Psychology, 72(5), 656-669. Daneman, M. & Merikle, P. M. (1996). Working memory and language comprehension: a meta-analysis. Psychonomic Bulletin & Review, 3, 422-33. Dehn, M. J. (2008).Working memory and academic learning: Assessment and intervention. New Jersey: Wiley Dufva, M., Niemi, P., & Voeten, M. J. M. (2001). The role of phonological memory, word recognition, and comprehension skills in reading development: From preschool to grade 2. Reading and Writing, 14 (1-2), 91-117. Dunn, L. M. (1965). Peabody Picture Vocabulary Test. Minnesota: AGS. Frith, U. (1985). Beneath the surface of developmental dyslexia. In K. Patterson, M. Coltheart and J. Marshall (Eds.), Surface Dyslexia (pp. 301-330). London: Lawrence Erlbaum. Garnham, A. (1982). Testing psychological theοries about inference making. Memory and Cognition, 10, 341-349. Gathercole, S.E. (1998). The development of memory. Journal of Child Psychology and Psychiatry, 39(1), 3-27.

26

Gathercole, S.E. (1999). Cognitive approaches to the development of short-term memory. Trends in Cognitive Sciences, 3(11), 410-419. Gathercole, S. E., & Baddeley, A. D. (1993). Working memory and language. UK: Psychology Press Ltd. Gathercole, S. E., & Pickering, S. J. (2000). Assessment of working memory in six- and seven-year old children. Journal of Educational Psychology, 92 (2), 377-390. Gathercole, S. E., Pickering, S. J., Ambridge, B., & Wearing, H. (2004). The structure of working memory form 4 to 15 years of age. Developmental Psychology, 40 (2), 177-190. Gathercole, S. E., Pickering, S. J., Knight, C., & Stegmann, Z. (2004). Working memory skills and educational attainment: evidence from national curriculum assessments at 7 and 14 years of age. Applied Cognitive Psychology, 18, 1-16. Georgiou, G. K., Das, J. P., & Hayward, D. V. (2008). Comparing the contribution of two test of working memory to reading in relation to phonological awareness and rapid naming speed. Journal of Research in Reading, 31(3), 302-318. Gottardo, A., Stanovich, K. E., & Siegel, L. S. (1996). The relationships between phonological sensitivity, syntactic processing, and verbal working memory in the reading performance of third-grade children. Journal of Experimental Child Psychology, 63, 563582. Hair, J. F., Jr., Anderson, R. E., Tatham, R. L., & Black, W. C. (1995). Multivariate data analysis with readings (4th ed). Upper Saddle River, NJ: Prentice-Hall. Harris, M. & Giannouli, V. (1999). Learning to read and spell in Greek: The importance of letter knowledge and morphological awareness. In M. Harris & G. Hatano (Eds.), Learning to read and write: A cross-linguistic perspective (pp. 51-70). New York: Cambridge. Holton, D., Mackridge, P., & Philippaki-Warburton, I. (1997). Greek: A Comprehensive grammar of the Modern Greek Language. London: Routledge Kotoulas, V. (2007). Phonological awareness evaluation in the primary school years and an investigation of its relationship to written language difficulties. Unpublished Ph.D. Thesis, Department of Special Education, University of Thessaly, Greece [in Greek].

27

Leather, C. V., & Henry, L. A. (1994). Working memory span and phonological awareness tasks as predictors of early reading ability. Journal of Experimental Child Psychology, 58, 88-111. Lesaux, N. K., Rupp, A. A., & Siegel, L. S. (2007). Growth in reading skills of children from diverse linguistic backgrounds: Findings from a 5-year longitudinal study. Journal of Educational Psychology, 99, 821-834. Lionetti, T. & Cole, C. (2004). A comparison of the effects of two rates of listening while reading on oral reading fluency and reading comprehension. Education and Treatment of Children, 27(2), 114-129. MacKinnon, D. P. (2008). Introduction to statistical mediation analysis. Mahwah, NJ: Erlbaum. MacKinnon, D. P. & Fairchild, A. J. (2009). Current directions in mediation analysis. Current Directions in Psychological Science, 18, 16- 20. MacKinnon, D. P. & Fairchild, A. J., & Matthew, S. F. (2007). Mediation analysis. Annual Review of Psychology, 58, 593-614. Marsh, G., Friedman, M. P., Desberg, P. & Saterdahl, K. (1981). Comparison of reading and spelling strategies in normal and reading disabled children. In M.P. Friedman, J.P. Pas and N. O’Connor (Eds.), Intelligence and learning (pp. 363-367). N.Y.: Plenum. Martin, R. C. (1990). Neuropsychological evidence on the role of short-term memory in sentence processing. In G. Vallar & T. Shallice (Eds.), Neuropsychological impairments of short-term memory (pp. 390-427). Cambridge: Cambridge University Press. McGee, A. & Johnson,H. (2003). The effects of inference training on skilled and less-skilled comprehenders. Educational Psychology: An International Journal of Experimental Education, 23, 49-59. Nation, K., Adams, J. W., Bowyer-Crane, C. A., & Snowling, M. J. (1999). Working memory deficits in poor comprehenders reflect underlying language impairments. Journal of Experimental Child Psychology, 73, 139-158.

28

Nation, K., Cocksey, J., Taylor, J. S. H., Bishop, D. V. M. (2010). A longitudinal investigation of early reading and language skills in children with poor reading comprehension. Journal of Child Psychology and Psychiatry, 51, 1031-1039. Natsopoulos, D., Stavroussi, P., & Alevriadou, A. (1998). On the concept of comparison in mentally retarded and nonretarded children. Journal of Psycholinguistic Research, 27(3), 321-337. Oakhill, J. V. (1984). Ιnferential and memory skills in children’s comprehension of stories. British Journal of Educational Psychology, 54, 31-39. Oakhill, J. V., Cain, K., & Bryant, P.E. (2003). The dissociation of single-word reading and text comprehension: Evidence from component skills. Language and Cognitive Processes, 18, 443-448. Oakhill, J. V., Hartt, J., & Samols, D. (2005). Levels of comprehension monitoring and working memory in good and poor comprehenders. Reading and Writing: An Interdisciplinary Journal, 18(7-9), 657-686. Paris, S.G., Lindauer, B.K., & Cox, G. L. (1977). The development of inferential comprehension. Child Development, 48, 1728-1733. Pickering, S., & Gathercole, S. Ε. (2001). Working Memory Test Battery for Children. London: The Psychological Corporation. Porpodas, C. (2001). Cognitive processes in first grade reading and spelling of Greek. Psychology (The Journal of the Hellenic Psychological Society)- A special issue devoted to Research on Reading, Spelling and Dyslexia in Europe, 8 (3), 384-400. Porpodas, C. (2002). Reading. Patra: Author’s Edition [in Greek]. Seigneuric, A. & Ehrlich, M. F. (2005). Contribution of working memory capacity to children’s reading comprehension: A longitudinal investigation. Reading and Writing: An Interdisciplinary Journal, 18(7-9), 617-656. Seigneuric, A., Ehrlich, M. F., Oakhill, J. V., & Yuill, N. (2000). Working memory resources and children’s reading comprehension. Reading and Writing: An Interdisciplinary Journal, 13, 81-103.

29

Sesma, H. W., Mahone, E. M., Levine, T., Eason, S. H., & Cutting, L. E. (2009). The contribution of executive skills to reading comprehension. Child Neuropsychology, 15, 232246. Sherry, A. & Henson, R. K. (2005). Conducting and interpreting canonical correlation analysis in personality research: A user-friendly primer. Journal of Personality Assessment, 84, 37-48. Stevens, J. (2002). Applied multivariate statistics for the social sciences (4th ed). Mahwah, NJ: Lawrence Erlbaum Associates. Swanson, H. L. & Berninger, V. (1995). The role of working memory in skilled and less skilled readers’comprehension. Intelligence, 21, 83-108. Swanson, H. L. & Howell, M. (2001). Working memory, short-term memory, and speech rate as predictors of children’s reading performance at different ages. Journal of Educational Psychology, 93(4), 720-734. Swanson, H. L. & Jerman, O. (2007). The influence of working memory on reading growth in subgroups of children with reading disabilities. Journal of Experimental Child Psychology, 96(4), 249-283. Tabachnick, Β. & Fidell, L. (2007). Using multivariate statistics (5th ed). Boston, MA: Pearson Education. Thompson, B. (2000). Canonical correlation analysis. In G. G. Laurence & P. R. Yarnold (Eds). Reading and understanding more multivariate statistics (pp. 285-316). Washington, DC: American Psychological Association. Thorstad, G. (1991). The effect of orthography on the acquisition of literacy skills. British Journal of Psychology, 82, 527-537. vanAuken, T., Chafouleas, S., Bradley, T., & Martens, B. (2002). Using brief experimental analysis to select oral reading interventions: An investigation of treatment utility. Journal of Behavioral Education, 11(3), 163-179.

30

Waters, G. S., Caplan, D., & Hildebrandt, N. (1987). Working memory and written sentence comprehension. In M. Coltheart (Ed.), Attention and performance XII (pp. 531-555). Hove, UK: Lawrence Erlbaum Associates Ltd. Yuill, N., & Oakhill, J. V. (1991). Children’s problems in text comprehension: An experimental investigation. Cambridge: Cambridge University Press. Yuill, Ν., Oakhill, J. V., & Parkin, A. (1989). Working memory, comprehension ability and the resolution of text anomaly. British Journal of Psychology, 80, 351-361.

31

APPENDIX Sample of stories and reading comprehension questions

Based on Oakhill’s (1984) “Tim and the Biscuit Tin”: Alexis waited until he was alone in the house. The only sound he could hear was his father’s axe in the shed. Alexis checked every room. His mother had gone shopping. He pushed a chair in front of the sink. Like an acrobat, Alexis reached the biscuit tin. The tin was behind the sugar. Alexis stretched and managed to lift the lid up. As he reached inside the tin, the door opened. There stood his sister, smiling at him. However, Alexis’s face looked like a lemon (=turned very pale; Greek expression used in cases of sudden fear). Literal questions What happened as Alexis reached inside the tin? Where was the biscuit tin? Necessary inference questions Why did Alexis want to be alone? What did he want the chair for? Elaborative inference questions Which room was Alexis in? What was Alexis’s father doing? Simile comprehension questions “Like an acrobat, Alexis reached the biscuit tin”; what does it mean? “Alexis’s face looked like a lemon”; what does it mean? Comprehension control questions Why didn’t Alexis ask his father to give him the biscuits? Why was Alexis happy to see his sister?

32

Table1. Means, standard deviations and score ranges for 8.5 and 9.5 year-olds and for all

participants

8.5 yrs

9.5 yrs

All participants

Μ

SD

Range

Μ

SD

Range

Μ

SD

Range

Digit recall

26.56

2.99

21-34

27.02

3.83

13-35

26.80

3.45

13-35

Word list matching

25.02

8.12

9-45

24.71

6.24

18-51

24.86

7.14

9-51

Word list recall

21.44

3.50

15-33

21.47

2.84

17-30

21.46

3.15

15-33

Non word list recall

12.09

1.91

7-19

12.80

1.83

9-19

12.47

1.89

7-19

Listening recall

11.05

2.98

6-17

12.61

2.71

7-22

11.88

2.94

6-22

Counting recall

19.51

4.70

12-34

20.69

4.87

12-31

20.14

4.80

12-34

Backward digit recall

12.02

4.45

6-25

13.27

3.01

6-20

12.68

3.78

6-25

Literal comprehension

5.23

1.72

0-8

5.94

1.35

3-8

5.61

1.56

0-8

Necessary inference

4.14

2.03

0-8

5.14

1.81

1-8

4.67

1.97

0-8

Elaborative inference

4.21

1.47

1-7

4.86

1.23

1-6

4.55

1.38

1-7

Simile comprehension

6.26

1.24

4-8

6.63

1.19

4-8

6.46

1.22

4-8

Comprehension control

2.47

2.23

0-7

3.22

2.37

0-7

2.87

2.33

0-7

Vocabulary

95.30

20.65

55-138

116.29

17.33

74-140

106.48

21.59

55-140

Reading fluency

39.40

5.10

30-52

45.41

7.73

24-69

42.60

7.26

24-69

Note: Maximum score is 8 for every comprehension measure taken. For the vocabulary and the working memory tasks there is no fixed maximum score, since the procedure is terminated on the basis of specific criteria. Reading fluency equals the total number of words participants read accurately in the first minute of the reading task.

33

Table 2. Pearson correlation coefficients among the phonological loop, central executive, reading comprehension, vocabulary, and reading fluency measures

1. Digit recall

1

2

3

4

5

6

7

8

9

10

11

12

13

14

-

.37**

.47**

.22*

.13

.32**

.30**

.06

.08

.10

.03

.20

.21*

.10

-

.55**

.32**

.37**

.40**

.28**

.19

.06

.11

.04

.14

.23*

.10

-

.28**

.29**

.30**

.20

.18

.16

.14

.13

.12

.21*

.05

-

.21*

.29**

.19

.13

.11

.21*

.09

.13

.12

.25*

-

.24*

.37**

.30**

.31**

.38**

.34**

.35**

.39**

.35**

-

.43**

.24*

.27**

.18

.03

.22

.24*

.22*

-

.24*

.22*

.23*

-.03

.27**

.16

.34**

-

.55**

.51**

.34**

.34**

.35**

.15

-

.54**

.57**

.54**

.52**

.26*

-

.45**

.39**

.38**

.17

-

.37**

.51**

.20

-

.30**

.16

-

.30**

2. Word list matching 3. Word list recall 4. Nonword list recall 5. Listening recall 6. Counting recall 7. Backward digit recall 8. Literal comprehension 9. Necessary inference 10. Elaborative inference 11. Simile comprehension 12. Comprehension control 13. Vocabulary 14. Reading fluency

-

Note: *p < .05. **p < .01 ***p < .001

34

Table 3. Canonical loadings of the central executive and reading comprehension variables

Function 1 Reading comprehension Literal comprehension

.72

Necessary inference

.74


.80


.52


.80

Central Executive Listening recall

.91

Counting recall

.57

Backward digit recall

.65

35

Table 4. Vocabulary mediation effects on the relation between working memory and reading comprehension

Paths Initial variable

Outcome variable

a

b

c΄

Indirect effects

90% CI

Word list matching


.19*

.26*

.12ns

.05*

.01-.12

Word list matching

Necessary inference

.19*

.47***

-.06

.09*

.03-.17

Word list matching


.19*

.30**

.02ns

.06*

.02-.12

Word list matching


.19*

.51***

-.08ns

.10*

.03-.18

Word list matching


.19*

.26*

.07ns

.05*

.01-.11

Listening recall


.29**

.22*

.18ns

.07*

.01-.16

Listening recall

Necessary inference

.29**

.42***

.12ns

.12***

.05-.21

Listening recall


.29**

.22*

.26*

.07*

.01-.15

Listening recall


.29**

.44***

.16ns

.13***

.06-.23

Note: Path a: the relation between the initial variable and the mediator, Path b: the relation between the mediator and the outcome variable, controlling for the effects of the initial variable, Path c΄: partial effect between the initial and the outcome variables, controlling for the effects of the mediator. CI: Confidence interval. * p