Working Memory Capacity and Reading Skills: A Study Comparing Dutch-English Bilingual and Dutch Monolingual Children
Monica Santoro s1229702
m.l.santoro@fsw.leidenuniv.nl
Universiteit Leiden. Instituut Universiteit Leiden. Instituut Pedagogische Wetenschappen. Onderwijsstudies
Josefine Karlsson MSc. PhD Student Brain and Education Lab. Universiteit Leiden. Instituut Universiteit Leiden. Instituut Pedagogische Wetenschappen. Onderwijsstudies
a.k.j.karlsson@fsw.leidenuniv.nl
Dr. Linda van Leijenhorst
Assistant Professor, Department of Education and Child Studies. Universiteit Leiden. Instituut Pedagogische Wetenschappen. Onderwijsstudies
Foreword
After years, too many years, away from academic work, this Master’s Thesis has been a long and arduous undertaking on my part. I want to thank Ms’ Josefine Karlsson on whose daily supervision I relied and from whom much knowledge I received. My great appreciation goes to Dr. Linda Leijenhorst for her pertinent comments on my results and her general overview. I would like to acknowledge the invaluable help I received from the Scriptie Atelier at FSW.
The participation of the Rijnlands Lyceum Oegstgeest has been crucial for this study. My thankful recognition goes to the Head of the International Department, Mr Mathijs
Hekkelman and to the School Administrator, Mrs Cheryl Embleton, for their enormous support and arrangement of logistics during the testing of children. This thesis could not have been possible without the collaboration and enthusiasm of the children here studied. I sincerely thank them and their parents who allowed their participation.
Finally, and most important, I want to say thanks to my family, Roberto, Lucia and Elías-Mateo, for all the support and encouragement they so patiently gave me.
Contents
Introduction and theoretical framework………...5
Research questions………..10
Methods...…...………...…………11
Participants………11
Instruments: Computer tasks……….11
Instruments: Pen and paper tasks………...………14
Procedures……….15
Results……….17
Data inspection………..17
First research question………...17
Second research question………..19
Third research question……….27
Discussion………..…….32
Abstract
We investigated possible differences between Dutch monolingual and Dutch-English bilingual children ages 11 to 13, in terms of working memory, reading comprehension and the influence of the former on the latter. We addressed three possible manifestations of these differences. First, we studied whether bilinguals and monolinguals perform differently on working memory tests. Our study, however, did not find any significant differences between the two groups. Second, we studied differences in performance of monolingual and bilingual children in understanding while reading sentences. We compared accuracy and reaction times on a reading comprehension task including sentences with temporal connectives. No differences between groups were detected at the level of accuracy. In reaction times, however, the analyses revealed significant interactions between groups and position of the temporal connective in the sentence. These interactions occurred for different connectives in monolingual and bilingual children, suggesting that the groups have different approaches to reading comprehension. Third, we explored a possible correlation between working memory and reading comprehension for the whole sample, first merged and then split by groups of monolinguals and bilinguals. The results showed medium to strong negative correlations between working memory tasks and reading comprehension tasks for the whole sample as well as for each of the groups. These correlations, however, are manifested differently in monolingual and bilingual children, a fact that provides another indication of different comprehension mechanisms for each of the groups. Our pilot study indicates, therefore, that bilingualism is associated to differences in reading comprehension mechanisms and in how these mechanisms correlate with working memory capacities. Further elucidation of these differences, by more extensive investigations, could be useful for the design of adapted educational approaches.
Introduction
Reading comprehension (RC) and cognitive capacities like working memory (WM) are essential tools needed throughout life. These are unique human abilities related to everything we do, from mathematical understanding, to our reasoning ability. We continuously make inferences, retrieve information and thus assemble the whole body of knowledge that makes our daily live function (Dehaene, S. 2009; Van den Broek & White 2012). Therefore, perhaps the most important objective of primary education programs is the early and healthy development of these skills to ensure successful secondary and academic studies and, eventually, the consolidation of adults able to perform within the increasing complexity of modern life. Educational goals have to be accomplished both in an inclusive and efficient manner; a demand that leads to the design of programs adapted to different categories of children. This fact, in turn, requires a careful characterization of the particularities and attributes of each of different categories, for instance, children with learning disadvantages, the gifted or the multilingual. This study focuses on the bilinguals.
In these times of globalization and universalization of education, educators are increasingly confronted with a rapidly increasing population of multilingual children. In fact, most children in the world grow in a multilingual environment, both in developed countries, where migrating parents and international trips are common, and in developing countries, where tribal dialects coexist with official languages. The education of these children confronts particular challenges, but can also make use of particular advantages. Studies geared to disentangling both of these aspects, thus, should be considered a priority to build efficient education policies for today needs. This study aims to detect differences in reading comprehension mechanisms and working memory attributes between bilingual and
Bilingualism is commonly defined in the literature as high level of proficiency in two languages. Monolinguals might have basic notions of other languages but their linguistic proficiency and competence concentrate only in one language. While in former times the coexistence of more than one language was considered to be a disadvantage for children in the initial years of education, more recent research shows the opposite. Indeed, executive control and working memory have been shown to develop earlier in bilingual children than in comparable monolinguals (Adi-Japha, Berberich-Artzi & Libnawi, 2010; Bialystok, 2010; Carlson and Meltzoff, 2008; Yang, Yang & Lust, 2011; Morales, Calvo & Bialystok, 2013) In fact, even bilingual adults continue to outperform monolingual adults on tasks that require the use of executive control (Bialystok, Craik, Klein &Viswanathan, 2004; Treccani, Argyri, Sorace, & Della Sala, 2009). It is known that some aspects of the learning environment may modify the development of executive control, and bilingualism has been shown to be one such aspect (Bialystok, Craik, Green, & Gollan 2009). Executive control is not one single capacity but in fact a set of mental processes, including: Task shifting: the ability to maintain focus on a task and to a new one when our goals or external circumstances change; updating: constructing meaning according to the current context and integrating incoming information; inhibition: selectively attend to linguistic structures in the target language while ignoring interference from the other language that may have some variance in common (Ridderinkhof, van den Wildenberg, Segalowitz & Carter, 2004; Prior & MacWhinney, 2010).
Studies aiming to determine why bilingualism modifies executive control have used different tasks, but all involved some component process considered to be part of cognitive control functions. Sufficient evidence was found to support the existence of a curvilinear relationship between bilingualism and critical thinking disposition (Albert, Albert, Radsma, 2002) and research has also shown that bilingual children have an advantage in at least two processes: inhibition and creative thinking. Inhibitory processes are considered as important
components of intelligence (Dempster, 1991) and they also influence our ability to function in everyday life (Garavan, Ross & Stein, 1999). The brains of bilingual children are trained very early on how to inhibit irrelevant information coming from one language while attending to both languages. There is overwhelming evidence that both languages are always active to some degree, even in contexts that clearly support only one of the languages (Francis, 1999; Kroll & de Groot, 1997; Thierry &Wu, 2007). Manifestations of creativity and originality in bilingual and monolingual children were studied (Carringer, 1974, 2012) and significant differences were found in creativity test performance in high school students with varying degrees of bilingual proficiency. Through his studies Carringer concluded that bilingualism promotes creative thinking abilities and may serve to free the mind from the “tyranny of words”, thus, allowing bilinguals to focus their attention on ideas and not words. In similar studies, Landry (1973, 1974) concluded that cognitive flexibility, produced by learning a second language, was conducive to both divergent thinking and originality. More recent studies found considerably higher independence and cognitive flexibility of both verbal and spatial ability in bilinguals (Konaka, 1997; Benet-Martínez, Lee & Leu, 2006).
Results from experimental research suggest that WM is involved in a broad range of cognitive abilities (Baddeley, 1986; Gilhooly, Logie & Wynn, 1993), and particularly, it seems to be a direct predictor of reading comprehension. Researchers studying reading ability, vocabulary and verbal skills in children found that WM capacity and the components of comprehension -that is, inference making, comprehension monitoring, and story structure knowledge- predicted the differences in children’s reading comprehension levels (Cain, Oakhill, & Bryant, 2004). In addition, these authors found a strong direct relation between verbally mediated WM and reading comprehension thus, confirming a relationship among working memory, reading comprehension, and higher level language skills (Daneman, & Carpenter, 1983). Moreover, individual differences in reading comprehension may reflect
differences in WM capacity, specifically in the trade-off between its processing and storage functions. A poor reader's processes may be so inefficient, that they lessen the amount of additional information that can be maintained in WM. Thus, reading comprehension and executive functions are both intertwined in WM and they reflect the processes of WM capacity.
Reading comprehension (RC) involves the formation of a meaning-based representation of texts, often called mental model or situation model (Van den Broek & Kremer, 2000). We achieve this complex task of understanding using cognitive skills (Cain&Nash, 2011) and through different mind processes like decoding, inference making, integrating new and old information and monitoring understanding newly acquired
knowledge. In turn, working memory allows us to recall and manipulate information to create our own concepts and ideas. WM does not only refer to passive storage, but as a more active part in human processing of data, serving both for executing processes and for storing the products of these processes (Baddeley & Hitch, 1974). Its central role in reading
comprehension involves the saving of pragmatic, semantic and syntactic information from previous texts and its use in disambiguating and integrating the subsequent text.
In this research, sentences with temporal connectives are used, in which children have to decide on chronology of events to evaluate reading comprehension performance and verbal and numerical working memory tasks were chosen to investigate the ability to
manipulate and store information, i.e., working memory capacity.
To conclude this study, the correlation between working memory and reading comprehension, (heavily documented in the literature) was verified for the whole sample and separately by groups, monolingual and bilingual children.
Studies have found that WM appears to be equally correlated to both reading comprehension and reading speed, while reasoning had a stronger correlation with reading comprehension than with reading speed (Seigneuric, Ehrlich, Oakhill and Yuill, 2000).
We chose to study Dutch-English bilinguals because these two languages allow a similar acceptance of the use of two simple past tenses in the same sentence without creating some kind of dissonance in the reader as it would, for instance, in French-English bilingual readers. Children, in both monolingual and bilingual groups, were selected with advice from their teachers to ensure they all had good reading skills and no diagnosed developmental disorders. School principals provided information about children’s SES to ensure we had quite a homogeneous sample, with most children coming from medial class families.
Research Questions
Possible similarities and differences between monolingual and bilingual children were investigated regarding the three aspects described above: Working memory capacity, reading comprehension and the influence of the former on the latter. Specifically, the following three questions are here answered:
First question related to working memory as part of executive control functions: Do monolinguals and bilinguals perform equally well on WM capacity tests? Two competing hypotheses are contemplated: First, bilinguals have higher WM. According to several studies (Bialystok et al., 2008, 2010, 2011, 2012) bilinguals constantly juggling with two languages are forced to train their cognitive capacities and they become more efficient at using those cognitive skills. They are naturally trained to inhibit irrelevant information. Second
alternative hypothesis: bilinguals have lower working memory capacity because the constant juggling of two languages puts a higher load on their cognitive capacities.
Second question in reference to reading comprehension: Are there differences between monolingual and bilingual children in understanding sentences with temporal connectives? The alternative hypotheses are: First, bilingual children outperform
monolinguals at global comprehension because bilinguals constantly juggling two languages are probably more efficient at using strategies to understand semantics. Second, bilinguals are slower than monolinguals in understanding sentences with temporal connectives because suppressing one language at all times makes comprehension slower.
Third question: Is there a correlation between working memory and reading
comprehension? If so, is this correlation different in mono and bilingual children? Given that cognitive capacities, particularly WM, are generally related to a faster understanding of semantics (McVay & Kane 2012) we expect to find a correlation between WM and RC.
Methods
Participants
Students from the Dutch monolingual group (N = 17) were 11-13 years old (M=12.77, N=17, SD=.27), 11 females and 6 males were recruited from Dutch schools in the Hague area. Such data had previously been gathered by Leiden University bachelor students, for an initial research on which this study is based.
The Dutch–English bilingual group (N = 17) children were 11-13 years old
(M=13.15, N=17, SD = .69), 11 females and 6 males. They were recruited from international schools in the Leiden-The Hague area, most of them coming from international medial class families living at the time in the Netherlands. Data on this group were gathered using the same protocol as for the monolingual group. All of the bilinguals are fluent speakers and readers of English and Dutch and most of them speak Dutch and some other language at home. The bilingual children have attended international schools in English since kinder garden. Some of them speak only Dutch and English at home and school but were also exposed to more languages as part of their living experiences abroad.
Instruments: Computer Tasks
Event structure cues task (ES-Cue task). This is a computer task that measures reading
comprehension of the child through sentences with temporal connectives called Event Structure Cues task (ES-Cue-task) in which the child has to decide on chronology of events. In this task the student is presented first with a sentence containing two events: The mouse ate cheese before it played tennis. Then the child pushes a button and three pictures appear in the screen. Two of the pictures have representations of two different actions and in the third picture the same two actions are represented together as happening at the same time. Children are told this is a game called the “what happened first?” game. As soon as they read and
understood the meaning of the sentence the child has to press a key, marked with a sticker on the keyboard, with his index finger. Each sentence however, is displayed for a maximum of 10 seconds; if the child does not press the key within that time, the trial moves on to the next question. An asterisk (*) appears before going to the pictures screen. The children are told that the task is timed and to read each sentence at their usual pace, not faster or slower, making sure they understand the meaning. It is not possible to go back to read the sentence once more to check the meaning. The researcher gives children the opportunity to practice with some rounds of the game before starting the real session.
In each sentence, the ES-Cues, before, after and while, could appear either in initial position in the sentence, or in medial position, that is, a word situated between the two actions. Thus, we deal with six conditions in this task: 1) before-initial, 2) before-medial, 3) after-initial, 4) after-medial, 5) while-initial, and 6) while-medial. While refers to actions taking place at the same time, therefore, conditions 5 and 6 are control conditions.
The data collected with this task are: the reading time taken to read the sentence (in milliseconds per syllable), the reaction time for answering the question (in milliseconds) and the number of correct answers given (in percentages).
Mental counters task, as part of Mariette Huizinga’s battery of tests is a timed computerized
task designed to measure working memory. The child needs to keep scores as in sport matches. In the middle of the computer screen appear two lines side by side. Below or above this line a box will appear to indicate a minus point or a plus point respectively. The
participants are asked to keep track of the score (adding and subtracting) that changes each time a box appears. For instance: The starting position is 0-0. A box appears above the left line the score is now 1-0. Another box will appear above the left line. The score is now: 2-0. A box will appear above the right line. The score is now: 2-1. When there appears a block
under the line the child needs to subtract 1. If a box appears under the left line, the score is now 1-1. The child was instructed to push the button when the score is more than a certain number given at the beginning of each trial. For example, if the command is: "press the button if the score on any line is more than 2”, the child needs to press the button when the score of any line reaches 3. The first block in the game starts with two lines, in the second block there are three lines on which to keep scores making the task more difficult. The rest remains the same. The data we collected with this task was the number of correct answers given in percentages.
Local-global task. This timed computer task measures cognitive flexibility. In this game
children need to match the geometric figures they see on a computer screen, which are drawn at both ends of a line, with the geometric figure that will randomly appear in the medial of the screen. One rule requires the child to attend to local stimuli; another rule requires the child to attend to global stimuli. The child is instructed which button key in the keyboard belonged to which cue for each rule set. When the rules change, children need to switch their tactics. They use their right and left index fingers to designate right or left choice. The task runs on its own but in between blocks the child can take a small pause and continue by pressing the space bar.
The procedure for the local trials is as follows: When the figures at the end of the line are small (left and right) the child needs to look closely at the small figures within the large figure. When a large figure made of many small squares shows in the medial, the child indicates the small square at the end of the line with his left/right index finger. When a large figure made of many small rectangles shows in the medial, the child indicates the small rectangle at the end of the line with his index finger.
For the global trials the procedure is the following: When the figures at the end of the line are large (left and right) the child needs to look closely at the large figures in the
medial. When a large figure is a rectangle, the child indicates the large rectangle at the end of the line with his left/right index finger. When a large figure is a square, the child indicates the large square at the end of the line with his left/right index finger.
For combined local-global trials the child proceeds as follows: The figures at the end of the line change every fourth time, so the figures are sometimes small, left and right, and then the child needs to go to the small figurines within the medial figure. And sometimes the figures are large, left and right, and then the child needs to go to the large figure in the medial. A trial in which the rule switches is called a switch trial. Trials in which the rule repeated the previous trial are called non-switch trials. The data we collected with this task are: reaction time in milliseconds and the number of correct answers given (in percentages).
Instruments: Pen and Paper tasks
Sentence span measurement task. This is a non-timed complex working memory task
designed to measure the processing and storage functions. The task is based on Swanson’s battery of tests (1992, 1993), which, were designed according to Baddeley’s definition (1986) that states that the tests ‘‘require simultaneous processing and storage of information’’ to ‘‘measure various contents’’. In this test children listened to a series of sentences and were asked to recall the final word of each sentence. Before repeating the words to be remembered, however, they were asked to answer a comprehension question about one of the previous sentences.
There are 5 levels with two sets each. We started testing from level 2, with two sentences in each set, and stopped if children made an error in each set within one level. The levels get increasingly more difficult as one sentence per level is added, finishing at level 5 thus with 5 sentences and one question. Scores were calculated taking the percentage of correct words recalled if the question was correctly answered. Level 1 of the task starts with a
series of practice trials containing two sentences and one question. Example: 1. The baby’s toy fell under the bed. (The word to remember is ‘bed’), 2.They ran to the back of the house. (The word to remember is “house”), QUESTION: What fell under the bed? (The answer is “toy”)
In order to obtain a reliable predictive value of these tasks for reading comprehension abilities, the results of both, sentence processing and recall components, were taken to
investigate whether the whole of WM is a predictor of reading abilities.
This study concentrates especially on tasks measuring working memory (SSM and MC) and cognitive flexibility (LG), with switching and non-switching conditions. The ES-Cue is a task expected to measure reading comprehension. Data on the three remaining tasks,
the Progressive Matrices Test (Raven, 1995), the Picture Comprehension task and the Curriculum-based measurement maze (CBM-maze) will not be discussed further in this paper
Procedures
The children were tested twice. For the first session, participants were tested
individually in a quiet room at their school premises, after school hours. They performed five tasks, three on a computer and two with pen and paper. In a second session, two weeks later, two more pen and paper tasks were given to the whole group gathered in a room at their school, after school hours. In both sessions, the tasks were administered in a fixed order. The individual session. Testing began with the ES-Cue task for each condition. On the
computer keyboard, the keys with letters A, S, and D, were covered by stickers. Each trial began with a sentence visible for 10ms and following this, an asterisk appeared in the screen and after 50ms the three pictures appeared. This task measures reading comprehension. The percentage of correct answers, when answers were given, was calculated and the reaction time in milliseconds measured.
The next task was the Swanson task. The SSM file for the auditory stimulus was played through loudspeakers. Once the children had listened to the sentences and the
question, the task was paused to give them time to answer the question and to tell all ending words they could remember. The answers were written down on paper by the researcher. This task measures verbal working memory.
For the Local Global task (LG), there were total 144 trials; in the local modality and in the global modality, presented in randomized order. These trials were further designated as switch (switching between two sets of rules) and non-switch trials. This task measures
cognitive flexibility. The session continued with the Picture Comprehension task to check the stimuli in the first task and finished with the MC task.
In the MC task there were 10 trials in the two-line block and 10 trials in the three-line condition, presented in randomized order. Those trials were further designated as switch non-switch trials. Practice trials with feedback preceded each task. This task measures working memory and cognitive capacity.
Children were offered short breaks in between tasks. None in the bilingual group tookbreaks; the monolingual children were offered some lemonade at a break in the medial of the test session. The session lasted approximately 50 min.
The group session. About two weeks after the individual sessions, the whole group of
children was given two paper tasks. The session started with the Raven task, which they completed in less than 30 minutes; continued with a short practice of the CBM maze, which lasted 30 seconds. The session finished with two maze tasks with two different texts: The Big Five with 54 choices and The Human Body with 55 lasting two minutes each. The whole session lasted for about 40 minutes. At the end of the testing sessions, children were given a small gift for their participation.
Results
Data Inspection
Univariate and multivariate inspections were conducted. Inspections of all continuous variables, their box-plots and shape of histograms showed that the distributions of the
different scores were reasonably normal. Observation of box-plots indicated no outliers. Testing assumptions were checked before each statistical analysis and no violation to those assumptions were found. Levine’s test of covariance matrices of the dependent variables showed no significant differences in variances. Scatter plots showed no obvious evidence of non-linearity. Visual exploration of the data revealed no outliers or missing values.
First research question
A one-way between-groups multivariate analysis of variance (MANOVA) was performed to investigate our first research question searching for possible differences in working memory capacity between bilingual and monolingual children.
Our dependent continuous variables were five: 1) Mental Counters, Average Reaction Time; 2) Mental Counters, Average Percentage Correct; 3) Local Global, Difference Between Switch minus Non-Switch Average Percentage Correct; 4) Local Global, Difference Score Percentage: Switch minus Non-Switch Reading Time; 5) SSM percentage of number of words answered correctly, when questions were answered.
Our independent categorical variable was the group of monolingual and bilingual children.
Table 1 shows the number of children participating, as well as mean and standard deviation values for each of our dependent variables by group of monolinguals and bilinguals.
Table1. Descriptive Statistics of Working Memory Tasks by Groups Group Mean S.D. N MentalCounters Average ReactionTime Monolinguals 640.60 63.74 17 Bilinguals 663.47 72.17 17 Total 652.03 68.04 34 MentalCcounters AverageCorrect (%) Monolinguals 84.26 7.05 17 Bilinguals 82.94 8.67 17 Total 83.60 7.81 34 LocalGlobal Switch(-) NonSwitchReactionTime Monolinguals 122.50 63.91 17 Bilinguals 153.30 89.21 17 Total 137.90 77.99 34 LocalGlobal Switch(-) Non-Switch (%) Monolinguals -1.56 3.65 17 Bilinguals -.80 4.64 17 Total -1.18 4.13 34
SSM (Swanson test) Words remembered (%)
Monolinguals 43.48 20.17 17
Bilinguals 48.31 17.17 17
A Box's test confirmed that the observed covariance matrices of the dependent variables are equal across both groups of bilinguals and monolinguals (F= 1.02, p = .43). The Levene test confirmed that error variances of the dependent variables are equal for both groups (t = -4.22, p = .08)
Results from the multivariate analysis of variance revealed no significant difference in WM capacity between bilinguals and monolinguals p = .64, thus failing to confirm any of the two competing hypotheses about a difference in WM of bilingual and monolingual children participating in this research.
Second research question
To answer the second research question on reading comprehension, children were given the ES-Cue task, which requires the understanding of chronology of events. Two one-way repeated-measure ANOVAs were performed to compare the scores of accuracy and reaction times in the ES-Cue task. Our dependent variables accuracy and reaction times have six conditions: Time connector while, in position initial and medial, time connector before, in position initial and medial and time connector after, in position initial and medial. Our independent categorical variable was the group of monolingual and bilingual children. Table 2 shows mean and standard deviation of scores for the six resulting conditions within
Descriptive Statistics of Reading Comprehension Task by Groups. ES-Cue tasks, all conditions
Conditions Group N Accuracy Reaction Time
M SD M SD
While Initial Monolinguals 17 99.57 1.73 1047.03 372.80
Bilinguals 17 98.73 3.77 1007.37 379.23
Total 34 99.15 2.92 1027.20 370.83
While medial Monolinguals 17 99.15 2.37 975.70 400.35
Bilinguals 17 99.34 2.69 891.49 377.47
Total 34 99.25 2.50 933.59 385.51
Before Initial Monolinguals 17 95.43 7.56 1629.24 475.22
Bilinguals 17 97.79 4.33 1677.93 466.61
Total 34 96.61 6.19 1653.59 464.40
Before medial Monolinguals 17 92.42 13.45 1650.83 412.13
Bilinguals 17 97.31 4.66 1492.93 357.80
Total 34 94.87 10.21 1571.88 388.39
After Initial Monolinguals 17 82.87 22.70 1982.88 700.82
Bilinguals 17 87.62 7.28 1623.44 429.98
Total 34 85.25 16.78 1803.16 600.88
After medial Monolinguals 17 85.08 28.40 1592.81 416.57
Bilinguals 17 91.34 13.64 1522.84 532.06
The accuracy scores of both groups on the ES-Cue task were analysed performing the first one-way repeated-measure ANOVA. The test yielded the following results: Both groups performed equally well. Neither main effect of group p = .32, nor of position were found p = .61. However, there was a main effect of the temporal connective used F (2, 31) = 8.20, p = < .00, partial ɳ² = .35which was further analysed. There were no interaction effects of temporal connective by group, p = .22, nor of temporal connective by position, p = .17 in this accuracy test.
The main effect of the temporal connective was next analysed with a follow up ANOVA (Wilks' lambda = .66, F (2, 31) = 7.98, p = < .00, partial ɳ² = .34). All children performed more accurately on sentences with connective while (M = 99.20, SE = .34, p = .01) than on sentences with connective before (M = 95.74, SE = 1.26, p = < .00). Secondly,
children performed more accurately on sentences with connective while than sentences with connective after (M = 86.87 SE = 3.25), p = <.00). Finally, all children performed more accurately in sentences with connective before than with connective after, (M = 86.87, SE = 3.25, p = < .00).See figure 1.
Figure 1. Accuracy 50 60 70 80 90 100
While Before After While Before After Monolingual Bilingual Initial Medial A C U R A C Y (%)
Next, the reaction time scores of both groups on the ES-Cue-task were analysed performing the second one-way repeated-measure ANOVA. The test yielded the following results: There was no significant main effect of the between-subject variable, p = .38. We did find main effects of the within-subjects variables: position F (1, 32) = 28.2, p = < .00, partial ɳ² = .47 and, temporal connective used F (2, 31) = 47.2, p =< .00, partial ɳ² = .75.
Regardless of group or temporal connective in the sentence, all children had faster reaction times on sentences with position medial than on sentences with position initial, p = < .00, M (medial) = 1348, SE (medial) = 57, M (initial) = 1505, SE (initial) = 65.
As to the main effect of temporal connectives, regardless of group or position of the temporal connective in the sentence, children in the sample had faster reaction times on sentences with temporal connective while than on sentences with temporal connective before, p = < .00, M (while) = 976, SE (while) = 69, M (before) = 1617, SE (before) = 70. Likewise, sentences
with connective while led to shorter reaction times than sentences with temporal connectives after, p = < .00, M (after) = 1687, SE (after) = 83. The difference between reaction times for
sentences with temporal connectives before and sentences with after is not significant, p = .28.
There was no two-way interaction effect between any pair of variables, (temporal connective-group, p = .49; position-group, p = .74; temporal connective-position, p = .15).
There was, however, a significant three-way interaction among temporal connectives by position by group, Wilks´Lambda = .83, F (2, 31) = 3.25, p = .05, partial ɳ² = .17 which justifiedfurther analyses. Thus, follow up mixed between-within analyses of variance were performed separately per condition, while, before and after, to investigate this three-way interaction in detail.
These analyses showed that for sentences with while as an ES-Cue there was a main effect of position on performance, F (1, 32) = 18.25, p = <.00, partial ɳ² = .37. Children were slower when temporal connective while was in initial position, M (While Initial) = 1035, SE
(While Initial) = 64, than when in medial position, M (WhileMedial) = 916, SE (While Medial) = 63.
However, there was no main effect of group, p =.58, nor interaction effect of group by position , p =.09.
For sentences with after as an ES-Cue there was no main effect of group on reaction time, p =.24, but we did find a main effect of position on performance F (1, 32) = 12, p = <.00, partial ɳ² = .27. Children were slower when temporal connective after was in initial position, M (After Initial) = 1820, SE (After Initial) = 100, than when in medial position, M (After
Medial) = 1555, SE (After Medial) = 82. On the other hand, no interaction effect of position by
group on performance was found , p =.11.
For sentences with before as an ES-Cue there was no main effect of group on reaction time p =.75, but we did find a main effect of position on performance F (1, 32) = 4.04, p =.05, partial ɳ² = .12. Children were slower when temporal connective before was in initial position, M (Before Initial) = 1660, SE (Before Initial) = 80 than when in medial position, M
(Before Medial) = 1575, SE (Before Medial) = 66. On the other hand, an interaction effect of position
by group was found F (1, 32) = 6.37, p = .001, partial ɳ² = .17. This prompted us to perform additional one-way repeated measures ANOVAs to find out how this interaction occured. We separated bilinguals from monolinguals.
The results show that the interaction is manifested in each of the groups, monolingual and bilingual, albeit in a different manner. The difference on reaction time by position is significant for bilingual children F (1, 16) = 8.4, p =.010, partial ɳ² = .34, but not significant
for monolingual children, p =.69. Bilingual children performed faster when the connective before was in initial position, M (Before Initial) = 1690, SD (Before Initial) = 458, than when in
medial, M (Before Medial) = 1500, SD (Before Medial) = 356. In order to illustrate these results, a bar
graph was constructed (Figure 2).
Figure 2 also shows a difference between groups with temporal connective after. Ths difference was tested and the results showed that there was an interaction effect of position for sentences with connective after only for monolinguals F (1,16) = .12, p =.003, partial ɳ² = .43, but not for bilinguals p = .20. Monolingual children were significantly slower when temporal connective after was in initial position, M (After Initial) = 1983, SD (After Initial) =
Figure 2. Reaction time (for questions answered) 0 500 1000 1500 2000 2500
While Before After While Before After Monolingual Bilingual Initial Medial R E A C T T I C M T E I O N (millisecs)
Third research question
In order to address our third and final research question on the possible correlations between working memory capacity and reading comprehension we first reassembled the two groups into one (N=34). The goal was to verify that our join sample exhibited correlations between WM and RC similar to previous findings in the literature. According to McVay & Kane (2012), cognitive capacities, particularly working memory, are generally related to a faster understanding of semantics. We checked whether the resulting correlation between WM and RC was present in our sample. Therefore, a correlation analysis between all working memory variables and the reaction times for all reading comprehension conditions was conducted. The Pearson product-moment correlation coefficients are summarized in Table 3.
Table 3. Correlations between scores of measures of WM and RC (whole sample)
While I (RT) While M (RT) Before I (RT) Before M (RT) After I (RT) After M (RT)
Mental Counters Average Correct Pearson Corr. -.11 -.10 -.27 -.14 -.34* -.17
Sig.(2-tailed) N .51 34 .57 34 .120 34 .40 34 .04 34 .31 34 Mental Counters Average RT Pearson Corr.
Sig.(2-tailed) N
.05 .08 .23 .28 -.05 .06
.77 .65 .18 .10 .76 .70
34 34 34 34 34 34
Diff.Switch(-) NonSwitch RT Pearson Corr. -.13 -.18 -.12 -.12 -.16 -.33
Sig.(2-tailed) .46 .29 .46 .49 .36 .05
N 34 34 34 34 34 34
Diff.Switch(-) NonSwitch (%) Pearson Corr. -.05 .04 -.11 .02 -.01 -.00
Sig.(2-tailed) .77 .80 .515 .89 .95 .98
N 34 34 34 34 34 34
SwansonTask (SSM%) Pearson Corr. -.10 -.16 -.40* -.54** -.40* -.36*
Sig.(2-tailed) .57 .35 .01 .00 .01 .03
N 34 34 34 34 34 34
*. Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed).
Table 3 shows significant negative correlations between the percentage of words remembered in the Swanson task and the Mental Counter task (both indicators of WM) and reaction time, (the time children take to understand chronological events), as tested by the ES-Cue task. This means that children who score lower in the working memory tasks
(remembering fewer words in the Swanson task or giving less correct answers in the Mental Counters) also have slower reaction time in the ES-Cue reading comprehension task.
We used the criteria according to Cohen (1988) and the findings are: 1) A strong negative correlation between the Swanson task and connective before medial, r = -.54, p < .01, showing that the percentage of words remembered in the Swanson task explains 30% of the variance in reaction time with connective before in medial position. 2) a medium negative correlation between the Swanson task and before initial, r = -.40, p < .05; showing that the percentage of words remembered in the Swanson task explains 16% of the variance in reaction time with connective before in initial position, 3) a medium negative correlation between the Swanson task and after initial, r = -.40, p < .05, showing that the percentage of words remembered in the Swanson task explains 16% of the variance in reaction time with connective before in initial position, 4) a medium negative correlation between the Swanson task and after medial, r = -.36, p < .05, showing that the percentage of words remembered in the Swanson task explains nearly 13% of the variance in reaction time with connective after in medial position. In addition, there is a medium negative correlation between the Mental Counter scores and reaction times when the connective after in initial position, r = -.34, p < .05, showing that the percentage of correct answers given in the Mental Counters task explains 12% of the variance in reaction time with connective after in initial position.
These analyses effectively confirm the existence of a significant relationship between working memory capacity and reading comprehension in our joint sample. This confirmation led to an analysis of WM and RC correlations discriminated by group. It was expected that
this analysis reflect the previously reported differences in reaction times for bilingual and monolingual children. The results, summarized in Table 4, do reveal suggestive differences between the groups. While significant correlations between the Swanson and the Mental Counters scores (both testing WM) with the six temporal connective conditions while, before and after in initial and medial positions of the ES-Cue task, are observed for each of the groups, they are not the same in each group.
For bilinguals, the negative correlations between the percentage of words remembered in the Swanson task (indicator of WM) and the temporal connectives are, medium negative correlation when before initial, r = -.49, p < .05 and strong negative correlation when before medial r = -.63, p < .01, showing that the percentage of words remembered in the Swanson task explains over 24% of the variance in reaction time with connective before initial, and nearly 40% of the variance in reaction time with connective before medial.
For monolinguals, the significant negative correlations appear between the Swanson scores and temporal connectives after initial r = -.52, p < .01 and after medial r = -.49, p < .01, showing that the percentage of words remembered in the Swanson task explains 27% of the variance in reaction time with connective after initial, and 24% of the variance in reaction time with connective after medial.
Lastly, Table 4 also shows a strong negative correlation between the reaction times in three of the conditions of the ES-Cue reading task and the Mental Counter scores (testing WM) only in bilingual children: While Medial, r = -.54, p < .01; Before Initial r = -.66, p < .05; After Initial r = -.51, p < .01 showing that the percentage of correct answers given by bilinguals in the Mental Counters explains nearly 30% of the variance in reaction time with connective while in medial position, 43.5% with connective before in initial position and 26% with connective after in initial position. No similar correlations were found in monolinguals.
While Initial While Medial Before Initial Before Medial After Initial After Medial
Monolinguals MentalCounters Pearson Correlation -.09 -.04 .11 .24 -.08 .06
N 17 AverageRT Sig. (2-tailed) .74 .88 .68 .36 .75 .80
MentalCounters Pearson Correlation .29 .31 .19 .12 -.35 15
Avrge Correct Sig. (2-tailed) .26 .23 .47 .64 .17 .57
DifferenceSwitch(-) Pearson Correlation -.24 -.37 -.27 -.11 -.24 -.40
NonSwitch RT Sig. (2-tailed) .35 .15 .29 .66 .36 .11
DifferenceSwitch(-) Pearson Correlation .23 .39 .23 .23 -.03 .25
NonSwitch(%) Sig. (2-tailed) .38 .13 .37 .37 .91 .33
SSM Words (%) Pearson Correlation .06 -.09 -.36 -.47 -.52* -.49*
Sig. (2-tailed) .82 .74 .15 .05 .03 .05
Bilinguals MentalCounters Pearson Correlation .18 .24 .34 .38 .04 .08
N 17 AverageRT Sig. (2-tailed) .50 .36 .19 .13 .87 .75
MentalCounters Pearson Correlation -.45 -.54* -.66** -.46 -.51* -.41
Avrge Correct Sig. (2-tailed) .07 .03 .00 .06 .04 .11
DifferenceSwitch(-) Pearson Correlation -.05 .01 -.06 -.06 .02 -.28
NonSwitch RT Sig. (2-tailed) .86 .96 .82 .81 .95 .28
DifferenceSwitch(-) Pearson Correlation -.27 -.25 -.41 -.14 .07 -.15
NonSwitch(%) Sig. (2-tailed) .30 .34 .10 .60 .80 .56
SSM Words (%) Pearson Correlation -.28 -.23 -.49* -.63** -.14 -.25
Sig. (2-tailed) .28 .38 .05 .01 .59 .33
Discussion
This study investigates possible differences between Dutch monolingual and Dutch-English bilingual children ages 11 to 13, in terms of working memory and reading comprehension and the influence of the former on the latter. Three hypotheses are presented: First, bilinguals outperform monolinguals on working memory tests. Results here presented, however, show no differences between the two groups. Second, we studied the possible differences between monolingual and bilingual children in reading comprehension. We compared accuracy and reaction times on a reading comprehension task including sentences with temporal connectives. No differences between groups were detected at the level of accuracy. In reaction times, however, multivariate analyses of variance revealed significant interactions between groups and position of the temporal connective in the sentence. These interactions occur for different connectives in monolingual and bilingual children, suggesting that the groups have different approaches to reading comprehension. Third, we explored a possible correlation between working memory and reading comprehension for the whole sample, first merged and then split by groups of monolinguals and bilinguals. The results showed medium to strong negative correlations between working memory tasks and reading comprehension tasks for the whole sample as well as for each of the groups, albeit in different manner.
The lack of significant differences in performances on WM capacity test indicate neither an advantage of the inhibitory training of bilingual children nor a disadvantage of the extra cognitive load to which bilinguals subject their working memory. Thus, our results failed to confirm any of the two competing hypotheses preliminarily proposed in this research. The juggling between languages does not significantly affect in any positive or negative sense the efficiency of working memory in the bilingual children here studied. These findings are in line with previous studies evidencing that bilingual children’s academic
performance is, at least, just as good as that of monolinguals and that bilingualism is not a cognitive disadvantage (Cummings 1996).
Findings on the second research question about the possible differences between monolingual and bilingual children in understanding sentences with temporal connectives indicate that reading comprehension processes are not identical in bilingual and monolingual children. Differences were detected in reaction times (milliseconds to give an answer in a reading comprehension task) for temporal connectives before and after. Bilinguals´ responses were significantly faster when before was in the medial position. This difference in reaction times is absent in monolinguals. In turn, monolinguals exhibit a significantly shorter reaction time when the temporal connective after is placed in medial position than when placed initially. Bilinguals do not show an analogous difference. While these results still require wider statistical confirmation, they suggest that monolinguals are faster when the right answer comes in the last clause of the sentence. For bilinguals, in contrast, speed seems to increase for sentences written in chronological order. We speculate that these differences are not inconsistent with the hypothesis that bilingual children base their comprehension on grasping the overall sense of a phrase (De Souza-Fleith, Renzulli & Westberg 2010), and thus they are helped by sentences respecting a chronological order, while monolingual children focus rather on individual parts of the sentence, hence perform better when they do not have to go too far to find the right answer.
The third research question in this study refers to the possibility of a correlation between working memory and reading comprehension for both groups separately and merged into one. As expected, a significant correlation between working memory and reading
comprehension was indeed found for the merged sample (table 3). This result confirms that our sample conforms to previous samples in the literature (Cain, Oakhill, & Bryant, 2004). The analysis of correlations between WM and RC discriminated by group, on the other hand,
revealed interesting differences. While significant negative correlations between the
Swanson scores (testing WM) and temporal connective conditions were observed for each of the groups, for bilinguals they appeared for connective before in initial and medial position, while for monolinguals they appeared for connective after in initial and medial position. Suggestively, these are the two types of conditions for which an interaction between position and group was also detected (see Figure 2). In addition, Table 4 shows a strong negative correlation between reaction times in reading tasks and Mental Counter scores (testing WM) only in bilingual children. Similar correlations were not observed for monolingual children. This lack of similar results might suggest that the working memory capacities associated to these Mental Counters tasks tend to be more used by bilingual children to interpret written texts. We conclude that the attributes associated to bilingualism, such as inhibition training, creativity and versatility, lead to significant effects on how working memory capacities correlate with performances in reading comprehension.
An important limitation of the present study is the relatively small sample of children tested. This was due to the difficulty to generate sizeable data of Dutch-English bilingual children. The organization of the testing procedures run against a number of issues: limited number of international schools around the Leiden-The Hague willing to participate in our research; school authorities too busy to organize after-hour sessions; schedule conflicts that lead to a long sequence of postponed sessions, and lack of manpower to face the time consuming individual testing procedures that are part of the study.
The present results, however limited, are nevertheless encouraging and justify the planning of follow up systematic research on the issues of working memory and reading comprehension of bilinguals, discussed here. This research could be refined by including factors such as the number of languages mastered by each child (many children do speak three or more languages nowadays) and with which degree of simultaneity these languages
were acquired. Simultaneous acquisition has more chance of leading children to novel approaches based on shared syntactic structures, a fact that could have important consequences on their reading comprehension ability.
Studies along the lines proposed here are important to design teaching methods and techniques that take into account the particular attributes associated to bilingualism, many of them clearly advantageous, for instance the enhancement of executive control, inhibitory control and cognitive flexibility (Bialystok & Viswanathan 2009). Adapted pedagogical instruction could contribute to develop bilinguals´ full potential for learning. Bilingualism, and multilingualism in children is, in fact, a widespread phenomenon both in developed and developing countries. In the former, due to the growing internationalization of human activities and in the latter due to the traditional coexistence of local dialects with official languages. Many bilingual education programs have been created, but the education of these children still confronts challenges. Chiefly among them is the priority to exploit the
advantages and particularities of bilingual children, so to allow them to capitalize fully on their attributes and allow them to accede to a further level of performance. We do hope that our pilot effort be followed by a more ambitious program in this direction.
References
Adi-Japha, E., Berberich-Artzi, J., & Libnawi, A. (2010). Cognitive flexibility in drawings of bilingual children. Child Development 81(5):1356-66.
Albert, R. T., Albert, R. E., Radsma, J. (2002)
.
Relationships among bilingualism, critical thinking ability, and critical thinking disposition. Journal of Professional Nursing. (8), 4: 220–229.Baddeley, A. D. (1986). Working-memory. Oxford: Clarendon Press.
Baddeley, A. (2012). Working memory: Theories, models, and controversies. Annual Review of Psychology. 63, 1-29
Baddeley, A.D., Hitch, G.J., (1974). In: Bower, G. (Ed.). Working memory. Recent Advances in Learning and Motivation. Academic Press.
Benet-Martínez, V., Lee, F., Leu, J. (2006). Biculturalism and Cognitive Complexity. Expertise in Cultural Representations. Journal of Cross-Cultural Psychology, 4: 386-407
Bialystok, E., Barac, R., Blaye, A. & Poulin-Dubois, D. (2010). Word mapping and executive functioning in young monolingual and bilingual children. Published in final edited form as: Journal Cognitive Development 1; 11(4): 485–508.
Bialystok, E., (2010). Global–local and trail-making tasks by monolingual and bilingual children: beyond inhibition Developmental Psychology 46, (1), 93–105.
Bialystok, E., (2011). Reshaping the mind: the benefits of bilingualism Canadian Journal of Experimental Psychology. 65, (4), 229–235
Bialystok, E., Craik Fergus, I.M. & Luk, G. (2012). Bilingualism: consequences for mind and brain. Trends in Cognitive Sciences. 16, (4).
Bialystok, E. & Viswanathan, M. (2009). Components of executive control with advantages for bilingual children in two cultures. Cognition (112) 3: 494–500.
Cain, K., Oakhill, J., & Bryant. P. (2004). Children’s reading comprehension ability: Concurrent prediction by working memory, verbal ability, and component skills, Journal of Educational Psychology 96, (1), 31– 42.
Cain, K., Nash, H. M. (2011). The influence of connectives on young readers’ processing and comprehension of text Journal of Educational Psychology. 103, (2), 429–441.
Carlson, S. M. & Meltzoff, A. N. (2008). Bilingual experience and executive functioning in young children. Developmental Science. 11, 282–298.
Carringer, D. C., (1974). Creative Thinking Abilities of Mexican Youth. The Relationship of Bilingualism. Journal of Cross-Cultural Psychology, 4: 492-504.
Cummins, J. (1996). The Influence of Bilingualism on Cognitive Growth: A Synthesis of Research Findings and Explanatory Hypotheses. Working Papers on Bilingualism, No. 9. Bilingual Education Project, The Ontario Institute for Studies in Education.
Cummins, J. (1999). Linguistic Interdependence and the Educational Development of Bilingual Children. Review Of Educational Research (49 ) 2: 222-251.
Daneman, M., & Carpenter. P, (1983). Individual differences in integrating information between and within sentences. Journal of Experimental Psychology: Learning, Memory, & Cognition. 9, 561-584.
Dempster, F.N. (1991). Inhibitory processes: A neglected dimension of intelligence. Intelligence 15: 157–173.
De Souza-Fleith, D., Renzulli, J. & Westberg, K. (2010). Effects of a creativity-training program on divergent thinking abilities and self-concept in monolingual and bilingual classrooms. Creativity Research Journal. 14, 3-4, 373-386.
Francis, W.S. (1999). Cognitive integration of language and memory in bilinguals: Semantic representation. Psychological Bulletin. 125(2), 193-222
Garavan, H., Ross, T.J., Stein, E.A. (1999). Right hemispheric dominance of inhibitory control: An event-related functional MRI study. Proceedings of the National Academy of Sciences 96: 8301–8306.
Gilhooly, K., Logie, R. & Wynn, V. (1993). Counting on working memory in arithmetic problem solving Memory & Cognition 22 (4) 395-410.
Huizinga, M., Dolan, C.V. & Van der Molen, M. W. (2006). Age-related change in executive function: Developmental trends and a latent variable analysis. Neuropsychologia, 44, 2017-2036.
Kendeou, P. & Papadopoulos, P., in Espin, C.A., McMaster, K.L., Rose, S. & Wayman, M. Editors (2012). A measure of success: the influence of curriculum-based measurement on education. Ch. 29. Minneapolis: University of Minnesota Press.
Konaka, K. (1997). The relationship between degree of bilingualism and gender to divergent thinking ability among native Japanese-speaking children in the New York
Kroll, J. & de Groot, A. (1997). Lexical and conceptual memory in the bilingual: Mapping form to meaning in two languages Tutorials in bilingualism: Psycholinguistic perspectives, Volume, 169-199.
Landry, R.G. (1973). The relationship of second language learning and verbal creativity. The Modern Language Journal. 57, 110-113.
Landry, R.G. (1974). A comparison of second language learners and monolinguals on divergent thinking tasks at the elementary school level. The Modern Language Journal. 58, 10-15.
Luk, G., Anderson, J.A.E., Craik, F. I.M., Grady, C. & Bialystok, E (2010). Distinct neural correlates for two types of inhibition in bilinguals: Response inhibition versus
interference suppression Brain and Cognition 74, 347–357.
McVay, J.C.,& Kane, M.J. (2012). Why does working memory capacity predict variation in reading comprehension? On the influence of mind wandering and executive attention. Journal of Experimental Psychology 302-320.
Morales, J., Calvo, A. & Bialystok, E (2013). Working memory development in monolingual and bilingual children Journal of Experimental Child Psychology 114, 187–202. Morales, J., Gómez-Ariza, C.J. & Bajo, M.T. (2013). Dual mechanisms of cognitive control in
bilinguals and monolinguals. Journal of Cognitive Psychology. 25:5, 531-546. Prior, A., & MacWhinney, B. (2010). A bilingual advantage in task switching. Bilingualism
Language and Cognition. 13, 253–262.
Pyykkönen, P., & Järvikivi, J. (2012). Children and situation models of multiple events. Developmental Psychology, 48, 2, 521-529.
Raven, J.C. (1995). Colored Progressive Matrices. Oxford: Oxford Psychological Press. Ridderinkhof, K.R., van den Wildenberg, W.P.M., Segalowitz, S.J. & Carter, C.S. (2004).
Neurocognitive mechanisms of cognitive control: The role of prefrontal cortex in action selection, response inhibition, performance monitoring, and reward-based learning. Brain and Cognition 56: 129–140.
Seigneuric, A., Ehrlich, M., Oakhill, J. V., & Yuill, N. M. (2000). Working memory resources and children's reading comprehension. Reading and Writing: An Interdisciplinary Journal. 13, 81-103.
Swanson, H. L. (1992). The generality and modifiability of working memory. Journal of Educational Psychology. 84, 473–488.
Swanson, H. L. (1993). Individual differences in working memory: A model testing and subgroup analysis. Intelligence. 17, 285–332.
Thierry & Wu, (2007). Brain potentials reveal unconscious translation during foreign-language comprehension. Edited by Dale Purves. Duke University Medical Center Durham. NC.
Treccani, P., Argyri, C., Sorace, G. & Della Sala, T. (2009). Internal And External Interfaces In Bilingual Language Development: Beyond Structural Overlap. International Journal of Bilingualism 13:195-210.
Van den Broek, P.W. & Kremer, K.E., (2000). The mind in action: What it means to
comprehend during reading, in Book: Reading for Meaning: Fostering Comprehension in the medial Grades. Teachers College Press edited by Barbara M. Taylor, Michael F. Graves, Paulus W. van den Broek.
Van den Broek, P. & White, M.J., (2012). Cognitive processes in reading and the
measurement of comprehension, Ch. 25 in book: A measure of success: The influence of curriculum-based measurement on education. Espin, C.A., McMaster, K.L.S. & Wayman, M. (Editors).
Yang, S., Yang, H., & Lust, B., (2011). Early childhood bilingualism leads to advances in executive attention: Dissociating culture and language Bilingualism: Language and Cognition 14, (03), 412-42.
