1 Executive functions in children: Examining the bilingual advantage in Frisian-Dutch bilinguals,
Dutch monolinguals, Dutch EFL learners, and Frisian-Dutch bilingual EFL learners
Name: Jan-Willem Wijers Student number: 1506439 Program: MA Applied Linguistics Supervisors: Dr. A van Hout and N. Houtzager
Word count: 13,342
2 Summary
Two well-established methods, the Flanker and the Wisconsin card sorting test, were used to measure executive functions while controlling for age, intelligence and socio-economic status.
This study examined four groups of 9- and 10-year-old Dutch monolinguals, Frisian-Dutch bilinguals, Dutch foreign language learners, and Frisian-Dutch foreign language learners and the relationship between bilingualism and executive function performance. In addition to the contrast between monolingual and bilingual children, two groups of foreign language learning children were included to provide more insight in the amount of language input needed to investigate whether a ‘bilingual advantage’ could be observed. The results of both the Flanker and Wisconsin Card sorting task gave no indication that bilingual children or foreign language learning children benefit in terms of executive function performance. It did become clear, however, that there were two influences on executive functions - age and intelligence - affect one of the executive functions tasks, namely the Flanker task.
3 Table of contents:
1. Introduction 2. Background
2.1 Executive functions 2.2 Important terminology
2.3 History of bilingualism and executive functions 2.4 Bilingualism and executive functions in children 2.5 Sequential bilingualism
2.6 Current language climate in Europe 2.7 This study
3 Methods
3.1 Participants
3.2 Materials and procedure 4. Results
4.1 Raven Standard Progressive Matrices 4.2 Flanker
4.3 Wisconsin card sorting test 4.4 Results overview
5. Discussion and conclusion
4 1. Introduction
The notion that bilinguals are somehow different from monolinguals has been around for quite some time. Obviously the fact that bilinguals speak more than one language fluently sets them apart from monolinguals, but there is also evidence that speaking more than one language may yield additional benefits as well. For example, being bilingual may influence certain cognitive processes, such as executive functions. In the past it was widely assumed that being bilingual - or multilingual for that matter - was detrimental to cognitive development (Adesope, Lavin, Thomson, & Ungerleider, 2010). Years of research, however, have suggested that being multilingual can, in fact, have a positive influence on cognitive development. Studies have not always found positive effects exclusively; it is frequently observed that bilinguals usually have a smaller receptive vocabulary and lower verbal fluency (Kalashinkova & Mattlock, 2012). Many researchers, however, find positive differences in an area of the cognitive system called executive functions (Poarch & van Hell, 2012; Esposito et al 2013; Carlson & Meltzoff, 2008;
Morales, Calvo & Bialystok (2013); Bialystok & Barac, 2012; Bialystok, 2011). This positive effect in executive function performance related to language is also known as the ‘bilingual
advantage’. Much is still unknown about this bilingual advantage. There is by no means any consensus as to whether and how being multilingual influences cognitive functions: some studies find no differences in executive performance between monolinguals and bilinguals (Gathercole et al., 2014; Morton & Harper, 2007). Moreover, most language-related executive functions studies focus on bilinguals and research on young language learners and young bilingual foreign language learners, however, is limited. There is also much debate regarding the mechanisms that may govern differences in cognitive development and just what executive functions entail, anyway. In order to contribute to the discussion about bilingualism and its relation to cognitive development, a study was conducted to examine a group of young Frisian- Dutch bilinguals, Dutch EFL (English as a Foreign Language) learners, and Frisian-Dutch bilingual EFL learners. Its goal was to investigate the effects of ‘lingualism’ on executive functions in young learners. The data used in this study is part of a larger study on executive functions and
5 bilingualism1. This data is contrasted with the performance of monolingual Dutch learners of the same age, who acted as a control group.
2. Background
The following section first explains what executive functions are, defines their sub-functions and briefly discusses their position in the field of cognitive psychology. Next, a brief history of language related executive functions studies is given. Before moving on to recent studies on bilingualism and its relation to executive functions, the terminology of bilingual research is defined. Lastly, the position of this study in relation to existing studies is explained.
2.1 Executive functions
Executive functions are a set of cognitive skills that are vital to goal-oriented problem solving.
This set of cognitive skills includes working memory and inhibitory control, and it plays a role in the ability to switch between tasks. Working memory refers to the process of temporarily retaining information about stimuli that came in through the senses or was retrieved from long term memory. In addition to keeping information ‘in mind’, working memory also allows for manipulation of this information. The effectiveness of the working memory can be trained (Buchsbaum, & D’Espesito, 2013). Inhibitory control refers to the ability to reject misleading clues (Carlson, Zelazo & Faja, 2013). Task switching is the ability to switch between tasks, skills or cognitive sets rapidly. Children are generally more effective task switchers than adults (Nobre & Kastner, 2014). In addition to age, important influences to executive functions include socio-economic status, culture, language, care-giving and sleep (Carlson, Zelazo & Faja, 2013). Executive functions continue to improve rapidly until adolescence sets in, its greatest growth occurring between the age of six and eight. Figure 1 shows the growth expressed as
1 Stevenson, C., Saarloos, A.M. Wijers, J.W., de Bot, K. (2014). A bilingual advantage for young beginning EFL learners? (provisional name). Unpublished.
6 aggregate scores on the Flanker and Wisconsin card sorting task, which are also the two tasks administered in this study (Carlson, Zelazo & Faja, 2013).
Figure 1. Aggregate scores on Flanker and Wisconsin card sorting tasks related to age.
Two important measuring tools of executive functions, the Flanker task and the dimensional card sorting tasks, show that improvement is correlated with age. As mentioned earlier, language is not the only factor influencing executive functioning; other factors also contribute.
Calvo and Bialystok (2014) found significant effects from socioeconomic status (SES) and
bilingualism on executive function performance, even though intelligence scores were the same across groups. Children from middle class environments performed better on executive
functions tests than working class children, and bilinguals outperformed monolinguals on executive functions tasks. The influence of SES on executive functions, which before the discovery of executive functions was called intelligence, was also one of the most important conclusions of the classic study by Peal and Lambert. They found that the influence of SES on intelligence was one of the factors that had been frequently overlooked by earlier studies.
Intelligence, in turn, was usually not associated with executive functions, but such a relationship has previously been established. One study compared fluid and crystallized intelligence to three measures of executive functions, namely shifting, updating working memory, and inhibition control. They found no correlation between intelligence and switching
7 and inhibition control, but they did find a strong correlation between intelligence and updating working memory (Friedman et al., 2006). This would suggest that not all aspects of executive functions are equally influenced by intelligence, although there is a relation between the two.
Although executive functions have been frequently researched and studies go back to the 1930’s, there is still no consensus on how to accurately describe them. The main debate in executive function studies centers on whether executive function is one unitary construct or if it consists of several separate components that are interrelated (Carlson, Zelazo & Faja, 2013).
Miyake et al. (2000) proposed a model including three separate sub-functions in executive functions, namely inhibition control, shifting and updating. The latter refers to the updating of working memory. Although this is a widely supported view, it is still subject to heavy debate because the three identified variables intercorrelate.
There are many different methods to test executive functions. The Dimensional Change Card Sort is a test that can test switching capabilities and can be used with young children (Zelazo, 2006). Simon-tasks come in a variety of forms, but the common denominator in these tasks is that the participant is presented with a stimulus that is either on the same side as the target location (congruent trial) or on the opposite side (incongruent trial). It measures inhibitory control. If for example a participant is asked to push the left button when he or she sees a cow and the right button when he or she sees a chicken, then reaction times will be quicker when a chicken is shown on the right side of a screen than on the left side (Gioia, Isquith, Guy, &
Kenworthy, 2000). The Flanker task operates on basically the same principle: the participant is presented with arrows that are bordered by arrows that either point in the same or in the other direction of the target arrow (Jordan & Morton, 2012). The Attentional Network Task is
basically a Flanker task, with the addition of extra lines to create distraction (Gooding, Braun &
Studer, 2006). The Wisconsin Card Sorting test measures a wider spectrum, but mainly taps into working memory. The test will be described in detail in the ‘materials’ section (Chelune &
Baer,1986). Lastly, the Stroop task is a classic test in which words describing colors are presented in different colors. It is a test of conflict management. Participants are presented with a word describing a color and are asked to say that word. When the color of the word
8 matches the word itself (congruent), then reaction times will be lower than when they do not (incongruent trials). This test knows many variations, both verbal and non-verbal (Carlson, Zelazo & Faja, 2013).
2.2 Important terminology
Before discussing the research in the field, it is important to explain the several types of
language combinations that are discussed in this study. Since the terms “bilingual”, “sequential bilingual”, “monolingual” and “second language learners” are in risk of overlapping somewhat, clear definitions are given below.
For monolingual speakers, the only language they use on a regular basis is their mother tongue.
These speakers receive no formal education in another language, but usually get exposed to other languages through modern mass media. Bilingual speakers grow up hearing and speaking two languages from birth. Usually, their parents each speak a different language or they live in an area that has two official languages that are spoken to the same degree. Another common situation is when one language is spoken at home and the other one in the community. When these individuals use both languages to the same extent they are called balanced bilinguals. In the case of sequential bilinguals, the speaker came into contact with a second language after learning the first. These individuals can become balanced bilinguals when they have had a lot of exposure to the second language. With sequential bilinguals, the large exposure to the second language means that the speaker has become very proficient in the second language, even though it was not learned from very early on. An example good example is when immigrant children are immersed in the language of the language of the country they moved to. In terms of second language exposure, sequential bilinguals are placed in between bilinguals from birth and second language learners because they have seen more exposure to a second language than second language learners, but less than bilinguals from birth. Second or third language learners receive formal education in a language other than their mother tongue. This can range from a couple of hours a week up to so called immersion programs which are taught fully in the target language.
9 2.3 History of bilingualism and executive functions
Although it was long believed that bilingualism had a detrimental effect on cognitive development, especially intelligence (Peal & Lambert, 1962), the general stance towards
bilingualism is currently not so negative (Adesope, Lavin, Thomson, and Ungerleider, 2010). The starting point of this change in perspective was the work by Peal and Lambert (1962), who critically reviewed earlier research on the relation between bilingualism and intelligence and performed their own study of this relation at six Montreal schools. They concluded that much of previous research fell short in terms of controlling for many independent variables
influencing intelligence. Some of these variables are, for example, socio-economic status (SES) and age. It turned out that these variables were influencing intelligence scores to such a degree that not controlling for them made it seem that bilingualism is detrimental to intelligence. In their own study they did not find negative effects related to bilingualism; in fact, they found significant positive effects of bilingualism on intelligence. Peal and Lambert looked at
intelligence, because the concept of executive functions did not exist at the time.
The study of Peal and Lambert has since been replicated and expanded to investigate a wider segment of the cognitive system. Currently, it is an accepted view in the field of linguistics that bilingualism improves a number of cognitive skills due to the extensive use of and switching between two different language systems (Bialystok, Craik & Luk, 2012). Bilingualism and its relation to cognitive functions has been the focus of a great deal of attention and many
researchers have conducted studies in this specific field. Although many find bilingualism to be related to better executive functioning, there are also studies contradicting that assertion.
These studies will be reviewed and contrasted in the following sections in an attempt to gain understanding about the current state of research.
10 2.4 Bilingualism and executive functions in children
A meta-analysis of 63 studies (Adesope, Lavin, Thomson, & Ungerleider, 2010), spanning across many language conditions and several measures of cognitive functions, revealed that
bilingualism is reliably associated with improved cognitive functions, although the effect sizes range from small to large. Adesope et al. point out that a lot is still unknown about the relation between executive functions and bilingualism and that further research is needed.
If being bilingual has an effect on executive functioning, the question is how. One possible explanation is that in the bilingual brain both languages are active at the same time, which means that one of those languages needs to be suppressed (Traxler 2012). This suppression is necessary to ensure that the other language does not interfere with the language spoken at any given moment. Consequently, there should be a system that controls the selection of active languages to make sure the correct one is used (Poarch and van Hell, 2012). Since bilingual speakers have much experience suppressing one language in favor of another, this training helps to suppress conflicting input and to select the corresponding set of rules for the desired language.
Bialystok, Craik and Luk (2012) discuss how bilingualism affects the brain. They explain the concept of functional neuroplasticity: experiences change the way the brain is organized.
Modified brain structures have been found in video game players, taxi drivers and skilled musicians. In these individuals, areas of the brain that govern the skills they excel in are found to be enlarged compared to corresponding areas in the brains of individuals who do not share their particular talents (Bialystok, Craik & Luk , 2012). The constant training that bilinguals experience while keeping two languages apart may similarly affect brain structure. Evidence for the influence of bilingualism on the structure of the brain comes, for example, from a study by Jasinka and Petitto (2013), who found differences between monolinguals and bilinguals using near-infrared spectroscopy. Interestingly, they not only found differences in brain structure between monolinguals and bilinguals, but they also discovered that the age of exposure to a second language caused a different organization in the brain. It turns out that individuals who were exposed to a second language later in life recruited more of the areas of the brain that
11 govern general cognitive and language specific processes than those who had been exposed to a second language much earlier on. Bilinguals who learned a second language later on showed greater use of the prefrontal cortex, which is known to govern abstract thinking and has a clear relation with executive functions.
Although many studies assume that extensive training is necessary to bring about differences in executive functions and brain structure, there is also evidence that not a great deal of exposure to two languages is needed. Kovacs and Mehler (2009) found a bilingual advantage for children as young as 7 months. In addition, Poulin-Dubois, Blaye, Coutya and Bialystok (2011) tested 24- month old children to investigate if a bilingual advantage could be found at a very young age.
They only found significant results for an adapted Stroop task for children, but not for other executive function tasks like working memory and task switching. In addition, in an overview from 2001, Bialystok asserted that attention control is the first of a specific set of cognitive functions is to be positively influenced in young bilinguals. Attentional control is the ability to select a relevant clue despite conflicting input. This would indicate that even at a young age increased performance on conflict tasks is found for bilingual children, although not all aspects of executive functioning are equally affected at this young age. If bilingual effects can be found at such a young age, they may not result just from extensive training. It is possible that there are other factors not previously considered that affect executive performance in young children.
The majority of research has focused on bilinguals from birth. A notable exception is the work by Poarch and Van Hell (2012) who looked not only at second language learning children, but who also compared bilinguals from birth and bilingual third language learners in two separate experiments. The first experiment looked at German monolingual, German-English bilingual, German-English bilingual foreign language learners and monolingual foreign language learners with a mean age of 3.9 years. These children were tested using an Attentional Networks Task and the Simon task as experimental stimuli. The results revealed that bilingual and bilingual foreign language learners showed better performance on the Simon task compared to the monolingual speakers, indicating that bilinguals and bilingual foreign language learners have
12 better inhibitory control than monolinguals. The study did not find significant differences
between the monolingual foreign language learners and the other test groups. They concluded that multilingual children have greater cognitive control than monolinguals. Cognitive control is the ability to adapt to changing situations and to process information. This is comparable to the task switching capability discussed earlier. The second experiment used the same participants, but employed other tests. This experiment did not find any differences in terms of reaction times, which indicates that there are no differences in monitoring capabilities. Upon further analysis significant results for inhibitory control were found: The bilinguals and bilingual foreign language learners performed better than the monolingual second language learners, indicating that bilinguals and bilingual foreign language learners were less distracted by interference.
Further evidence for the improved ability to ignore interference comes from a study by
Esposito, Baker-Ward and Mueller (2013) who found that preschoolers who were bilingual from birth performed better on tasks that required suppression of interfering signals than
monolinguals.
2.5 Sequential bilingualism
Although many studies have looked into the possible beneficial effects of bilingualism on executive functioning, not many studies have looked into the effects of sequential bilingualism.
There is evidence indicating that children learning an L2 are fundamentally different in their language development compared to bilinguals from birth, because these speakers already have a language system in place when they begin the process of acquiring a second one
(Kalashnikova & Mattlock, 2012). In addition, the difference in state of cognitive maturity when a learner first comes into contact with a language is also assumed to play a role in how a second language is attained (Paradis, 2008). This might account for differences in acquisition patterns between sequential and bilinguals from birth. If language development in sequential bilinguals is different from language development in those who acquire two languages simultaneously, then it is possible that its effect on executive functions also differs. However, other theories propose that these differences are mainly caused by variations in exposure and that there are
13 no fundamental differences in the way sequential bilinguals acquire a language compared to bilinguals from birth (Gathercole, 2007).
A study by Kalashnikova and Mattock (2012) looked into English monolingual children and English-Welsh sequential bilinguals of the same age (M 4.6 years). These sequential bilinguals started learning in Welsh from age 3, but given that Welsh is an official language, they had previously heard the language spoken in the community and in the media. Welsh was not spoken at home by the participants. Kalashnikova and Mattock found significant differences for attentional control, using a dimensional card sorting test. No significant differences were found to exist in other areas, like meta-linguistic awareness and meta-presentational understanding.
They concluded that in young sequential bilinguals attentional control is the first part of the cognitive system to be influenced by second language exposure. Attentional control is one of the elements of executive functions that other studies have are frequently found to be improved in bilinguals. For example, studies by Bialystok (1988) and Bialystok and Majumder (1998) also found that sequential bilinguals and bilinguals from birth performed better than monolinguals on attentional control tasks. Only on analytical tasks did bilinguals from birth perform better than sequential bilinguals.
Another study exploring executive functions and sequential bilingualism through immersion programs was performed by Bialystok and Barac (2012). It found that performance on meta- linguistic tasks was related to level of proficiency, and executive control tasks performance to amount of time spent in an immersion program. To test executive control, Flanker and switch tasks were performed by the participants. This suggests that although it was previously
assumed that only full bilinguals would benefit in terms of increased executive functions, it may also be possible that, through practice, sequential bilinguals can improve their performance.
Not all studies find effects in executive functions in young bilinguals or sequential bilinguals, however. Carlson and Meltzoff (2008) could not find any significant effects for sequential bilinguals. Their study compared monolinguals, balanced bilinguals and children in a second language immersion program. Balanced bilinguals from birth performed better than the other groups in attention control tasks, which confirms earlier notions that bilinguals from birth may
14 benefit more in certain executive functions than sequential bilinguals or second language learners.
Finally, a large scale study on English-Welsh bilinguals (n=650), early sequential bilinguals (n=557) and English monolinguals (n=354) was performed by Gathercole et al. (2014). Using a card sorting task, Simon task and metalinguistic judgment tasks, the researchers did not find overwhelming evidence for a bilingual advantage. The bilinguals only outperformed the
monolinguals on the Simon task, and then only under one specific condition and in one specific age group. There were no differences in Stroop task performance. In their discussion
Gathercole et al. mention that other studies which found no bilingual advantage looked at fully bilingual language communities instead of sequential bilinguals or advanced language learners (Dunabeita et al., 2014). In addition, a study by Morton and Harper on French-English bilinguals and English monolinguals did also not find a bilingual advantage on a group of children that were well matched in terms of socio-economic status (Morton & Harper, 2007). One
explanation for these results, then, is that bilingualism might be less modular than previously assumed. If the two languages in a bilingual brain are not distinctly separate, but rather are part of a network, then it stands to reason that the links within a language are stronger than
between languages. In other words, full bilinguals do not have two separate languages, which would require switching between sets, but rather their two languages are part of one unified system. This is roughly comparable to the ability of individuals to tap into different registers in their native language. From this, it follows that fully fluent bilinguals may not actually switch between languages at all. Rather, their proficiency in both languages enables them to use the right language in the right context. Under this assumption fully bilingual speakers do not need to exert as much effort to keep languages apart as less fluent second language speakers do.
2.6 Current language climate in Europe
Language policy makers believe that early language learning has distinct advantages in addition to the acquisition of an extra language. Nikolov and Djigunovic (2011) explain how the
Commission of the European Community defines these advantages in a broad sense. According
15 to the Commission, the advantages of early language learning are, among others, better first language (L1) skills and better attitudes towards other cultures. From the perspective of the Commission, it does not matter which second language is learned. In fact, aiming to teach only a single European second language would undermine the European linguistic diversity. Still, English seems to be the most frequently taught second language in Europe.
In Europe, foreign languages are usually taught through dedicated courses which cover general topics like counting or basic conversational skills, or by using them to teach other courses. The latter approach includes Content and Language Integrated Learning (CLIL) programs. Generally, the time devoted to foreign language teaching takes between one hour in language awareness programs and up to several hours for the CLIL programs of the curriculum. Typically, language awareness programs are limited in terms of the time committed to language teaching, and they are often taught by teachers with no specific language teaching credentials whose proficiency falls short of native-speaker level. Most foreign language teaching programs, however, do not aim to create proficient second language speakers. Rather, their goal is to raise more
awareness about other languages and cultures. Politicians have long been concerned that teaching foreign languages may have a negative impact on L1 proficiency, but most studies that have looked into this matter have found no such negative influences (Nikolov and Djigunovic 2011).
Although it has long been common to teach at least one foreign language at high schools in the Netherlands, during the last few years early second language teaching has gained more sway.
Now, in the current primary school system it is mandatory to teach English at ages 10 to 12.
Primary schools in the bilingual province of Friesland are also obliged to teach Frisian (“Vakken en kerndoelen basisonderwijs”, 2014). There are also 1100 schools in the Netherlands that teach English from the age of 4 (“Basisonderwijs”, 2014). Although foreign language teaching in primary schools is normally limited to a maximum of 3 hours and 45 minutes per week (2 hours of which are mandatory by law), a pilot will be started in September 2014 where 30 to 50 percent of all teaching is done in a foreign language. A total of 20 schools are participating in this pilot, which will run for 5 years (“Eerste basisscholen van start met tweetalig onderwijs”,
16 2014). The shift from no foreign language teaching at primary schools to a mandatory 2 hours a week in the last two years of primary schools, up to pilots that promote extensive language teaching, are a clear sign that both government and the general public assign great importance to early foreign language teaching. Another testament to growing interest in early foreign language teaching is the Foreign Languages In Primary schools Project (FLIPP). This project studied the effects of early foreign language teaching at 14 primary schools which teach foreign languages to young children. Two monolingual schools acted as a control group. The main findings of the project include that the proficiency of the teacher is a good predictor for
eventual proficiency of the students; the first language does not suffer from the extra input of a foreign language; and increased time devoted to teaching English significantly improves
proficiency (Unsworth, de Bot, Persson & Prins, 2012).
2.7 This study
Early second language programs gain more and more popularity in the Netherlands and that there is evidence pointing towards the possible cognitive benefits of learning more than one language. Since the debate on bilingual advantage is still ongoing and it remains unclear how much exposure to a second or third language is needed to observe differences in executive functioning, the present study investigates if (1) a bilingual advantage can be found for Dutch- Frisian bilingual children and (2) if the advantage can be found for second language learners, both monolingual and bilingual ones. The hypothesis under investigation is that bilinguals will show a bilingual advantage by outperforming monolinguals at ages 9 and 10. In turn, it is hypothesized that foreign language learning children do not show an advantage over children who do not learn a foreign language in the same age range as the bilingual and monolingual groups. This result is expected based on the previous research that found bilingual advantages for bilinguals and very proficient foreign language learners. Several young learners in four different language groups were tested using a number of executive function tasks. These groups consisted of monolinguals, bilinguals, monolingual second language learners, and
17 bilingual language learners. This study tries to answer several questions regarding bilingualism and executive functions, namely:
1) Do bilinguals from birth show a bilingual advantage when compared to monolinguals? Based on previous studies it is likely that bilinguals from birth will exhibit improved executive
functions. Other studies, however, have found indications that bilinguals from birth are so skilled in managing two languages that there is no need for an interference suppression mechanism to keep the languages apart. If that is the case, then no bilingual advantage should be found for this group.
2) If bilinguals from birth have an advantage in executive function performance due to the management of two language systems, does the added influence of learning a third language at school give them a greater advantage over their bilingual peers? It is assumed that the
influence of the third language at school will not be great enough to show differences in performance. This may be caused by the general language goals that are set forth by the European Union, as their focus lies primarily on awareness and not proficiency. Since language proficiency is expected to be rather low, the influence of an extra language on the cognitive system is also expected to be low (Nikolov and Mihaljevic, 2011).
3) In addition, do early foreign language learners show a bilingual advantage in executive functions when compared to monolinguals? Again, there is a good chance that the maximum input of foreign language teaching (between 2 to 3.75 hours a week) in these individuals is simply not great enough to have an effect on executive functions. The prediction is that this will likely be the case. Studies have shown that very young bilinguals can already show differences in executive function performance, which may mean that a limited total exposure to a language can have profound effects. It is important to remember, however, that those young bilinguals are exposed, in roughly equal amounts, to two language systems in a short period of time. The foreign language learners in this study are only exposed to a maximum of two hours a week in the form of formal teaching, which may not be enough to really force them to manage two languages.
18 4) Lastly, if young foreign language learners are shown to have an advantage in executive function performance, then do bilingual foreign language learners differ from monolingual foreign language learners in the extent to which they experience this advantage? Bilingual foreign language learners already have two language systems in place and receive additional third language input. This is an interesting comparison, because there are several factors at play. The expectation is that bilingual foreign language learners will outperform monolingual foreign language learners, because they have had more experience managing the two
languages that are already in place. Since the influence of a limited amount of foreign language learning is likely to be small, any effects can likely be traced back to the comparison between monolinguals and bilinguals from birth.
3 Methods
The following section gives a description of the participants. This is followed by a brief description of the testing materials and their corresponding procedures. Lastly, it covers the design of the experiment.
3.1 Participants
For this study a total of 184 participants from four different primary schools were tested, aged between 8.3 and 11.2 years. At every school two groups of children – grades 5 and 6 - were tested,. To control for socio-economic status, the so-called ‘leerling-gewicht’ (student-weight) measure was used. This is a value assigned to every Dutch school by the Dutch government to give a rough indication of the average educational level of the students’ parents. All schools included in this study have a ‘leerling-gewicht’ of 0, which equates to the highest tier of socio- economic status (“Leerlingen”, 2013).
Two of the schools are from the province of Friesland, which is in the north of the Netherlands.
It is a bilingual province with two official languages. The two schools in Friesland were selected
19 to include bilingual speakers of Frisian and Dutch. The majority of the subjects speaks Frisian at home and also speaks fluent Dutch. All selected participants from the Frisian schools use Frisian as a home language and grew up hearing both Dutch and Frisian. One of the schools in
Friesland teaches mainly in Dutch and also teaches in and about the Frisian language from age 4 onwards for a couple of hours a week. According to the school, up to a maximum of 40 percent of the total teaching is done in Frisian. The children, as well as the school itself, can therefore be considered to be fully bilingual. The other school in Friesland also teaches in Dutch and Frisian in approximately the same degree, with the exception that this school also teaches English as a foreign language from a young age, namely grade 1. Like the children from the first school these children are fully bilingual, but in addition they are also foreign language learners.
These children are taught English by a native speaker between 1.5 and 2 hours a week. This equates to a maximum of 300 to 400 hours of English teaching, depending on the grade the students are in, from age 4 up to the moment of testing. This rough calculation does not take sick-days or holidays into account.
The two remaining schools are located in the adjacent province of Groningen, where Dutch is the only official language. These schools were chosen because they have a majority of
monolingual speakers. One of the two schools is strictly monolingual and this school takes pride in the fact that they do not provide early second language education. The other is a school well known for its foreign language education and the children receive a minimum of 1.5 to a maximum of 2 hours a week of English by a native speaker.
Children who spoke any languages other than the ones spoken and/or taught at school (for example, children of immigrants) were left out of the analysis, as were children who did not (yet) speak one or more of the languages spoken and/or taught at school. Lastly, children with known learning disorders or concentration issues were also removed from the data set. Table 1 shows an overview of the test groups.
20 Table 1
Descriptive statistics participants
Language group Age group in years
Mean age in months (SD)
Range in years N
Monolinguals 9 108 (4.0) 1.3 28
10 121 (4.0) 1.3 20
Bilinguals 9 110 (5.2) 1.5 18
10 120 (3.4) 11 15
Monolingual foreign language learners
9 109 (3.5) 1.1 23
10 121 (4.6) 1.3 20
Bilingual foreign language learners
9 107 (5.7) 1.8 18
10 120 (6.1) 1.8 17
Table 1
3.2 Materials and procedure
A brief description of the testing procedure is given below:
After acquiring parental consent and briefing the teachers, the participants were tested on 3 separate days in testing sessions lasting a maximum of 45 minutes each. During these sessions the participants performed a variety of executive functions, intelligence and language tasks. In addition, they also filled out a questionnaire about their language situation and socio-economic status. All tasks were performed individually on computers provided by the schools or laptops brought by the researchers. The only exception was the Raven Standard Progressive Matrices test, which was provided in paper format. All tasks were carried out either in class with the whole group or in smaller groups in a separate, quiet room. In case of any disturbances or abnormalities, the head researcher made notes for later reference.
The following tasks were used for this study.
21 Raven Standard Progressive Matrices
For this study the 20-minute paper and pencil version was used to estimate fluid reasoning.
Fluid reasoning is an intelligence measure used to assess the ability to solve problems that do not require previous knowledge. The 20-minute version is regarded to be a good predictor for performance on the Raven Advanced Progressive Matrices Test (Hamel & Schmittmann, 2006).
The short version was chosen because the participants were subjected to an already time- consuming battery of tests. The participants were required to finish as many puzzles in 20 minutes as possible. These puzzles all consisted of an array of figures with one item missing.
The participants were asked to choose which of six provided items fit the array best. An example of one such puzzle can be seen in Figure 1. The total number of correct items in 20 minutes was used to estimate fluid reasoning. Before starting the task, the researcher
performed two trials with the whole class in order to ensure that the children understood the task. These trials were deducted from the total score.
Figure 2.Example item of the Raven Standard Progressive Matrices test.
Flanker Task
22 To investigate inhibitory control, a computerized version of the Flanker task was used (Fan, McCandliss, Sommer, Raz, & Posner, 2002). In the practice trials, participants were asked to indicate if an arrow on the screen pointed to the left or the right by pressing a corresponding button on the keyboard. After the practice trials, the participants were presented with five arrows in a row and asked to press whichever button corresponded to the arrow in the center.
The other four arrows either pointed in the same direction of the target arrow or in the opposite direction, the former being a congruent and the latter an incongruent trial. Figure 2 shows a congruent trial. In this case all arrows point in the same direction and there is no conflicting information. Figure 3 is an example of an incongruent trial. In this case, the target arrow points to the left, while the other arrows point to the right. The arrows that point to the right are used to create a distraction. At the beginning of this test the participants were asked to press the correct button as fast as possible, but also to focus on accuracy. If the participants did not press a button within 1800 ms, the trial was considered to be wrong.
Figure 3. Congruent trial in the Flanker task.
23 Figure 4. Incongruent trial in the Flanker task.
Wisconsin Card Sorting Task
The Wisconsin Card Sorting Task (WCST), is used to measure the ability of participants to adapt to changing rules after continuously applying a correct rule. Participants are asked to match a presented card with one of four other cards. The cards can be matched terms of color, shape, number or a combination of the three. In the case of shape, the participant could choose between stars, crosses, circles or triangles. When the sorting condition was color, the
participant could choose between red, green, yellow and blue. Lastly, when the condition was to sort on numbers, the number of figures on the cards ranged from one to four. Figure 4 shows a correct trial when the matching rule is shape. Figure 5 shows what happens when the participant selected a card based on color when another rule should have been applied.
Whenever a wrong choice is made, only one option remains, since the system will never use the same rule twice in a row. If the participant remembered the last rule and knows which rule was wrong on the previous trial, logic dictates that only one rule is left. A smiley icon indicates if the correct rule is applied, as can be seen in figure 4. A crying icon indicates a wrong choice, like in
24 Figure 5. The rule will stay the same until the program switches the rule after 10 correct trials or 5 incorrect trials within a set. The program switches the rule without warning. In this study a computerized version was used, which was based on the original version with cards. The scoring and administration is the same in both the paper-based and the computerized version.
Figure 5. Correct WISC trial.
25 Figure 6. Incorrect WISC trial.
4. Results
The design of the study consisted of a four-level between-subjects condition with test groups based on ‘lingualism’; monolinguals, bilinguals, monolingual foreign language learners, and bilingual foreign language learners. What follows is a description of three analyses which were done to compare intelligence and executive function performance across the groups. The first analysis looked at intelligence as measured by the Raven Standard Progressive Matrices. The values found were used to determine if there were any mean differences in intelligence
26 between the groups and if intelligence interacted with executive function scores. The next two analyses were based on the output from the Flanker and Wisconsin Card sorting tasks. For every aspect of these tests, correlations were performed to see if age (expressed in months) and intelligence were related to each other and with the test-variables. If the scores interacted, then the variable was used as a covariate in an ANCOVA. When there was no interaction, one- way ANOVAs were performed to see if there were performance differences between the groups.
4.1 Raven Standard Progressive Matrices
Table 2 gives a representation of the mean, raw scores on the Raven Standard Progressive Matrices test.
Table 2
Mean raw Raven scores
Lingualism Raven Score (SD)
Monolinguals 36.4 (5.32)
Bilinguals 32.8 (7.49)
Monolingual foreign language learners 36.3 (5.71) Bilingual foreign language learners 33.5 (5.72) Table 2
A Pearson correlation revealed that there was a weak but significant positive correlation
between age in months and Raven Standard Progressive Matrices score, r(157) =0.2, p<0.05. To cancel out this age effect, age was used as a covariate while comparing the intelligence scores between groups using an ANCOVA. The ANCOVA for Lingualism on Raven score, controlling for age in months, found statistically significant main effects, F(3, 154)= 3.886, p=0.010, interaction F(3,154)=, 6.592, p=0.016.
The results for the pair-wise comparison are shown in Table 3.
27 Table 3
ANCOVA-Significant Pairwise comparisons between groups on Raven
Group comparison (Raven scores with adjusted mean) p monolingual fll* (36.2) bilinguals (32.7) 0.013
monolinguals (36.5) bilinguals (32.7) 0.006
monolinguals (36.5) bilingual fll* (33.6) 0.027 Table 3 – This Table only includes significant differences between the groups.
*Foreign language learners
The pair-wise comparison reveals that the monolinguals performed better on the Raven Standard Progressive Matrices test than the bilinguals and bilingual foreign language learners.
The monolingual foreign language learners perform better than the bilingual foreign language learners.
Summarizing, there were significant differences in intelligence scores between the two
monolingual groups on the one hand and the two bilingual groups on the other. Since there are significant differences in intelligence between the language groups, and since intelligence will likely affect performance on executive functions tasks, Raven scores are used as a covariate for the Flanker and WISC analyses below.
4.2 Flanker
The following section discusses the results of the Flanker task and a number of analyses. The Flanker task yields four basic scores; congruent reaction time, congruent accuracy, incongruent reaction time, and incongruent accuracy. Reaction times are calculated only for correct items and are presented in milliseconds. Accuracy is calculated by taking the percentage of correct items in relation to the total trials. A fifth variable is derived from the basic scores by taking the reaction times of congruent minus incongruent trials to give a Flanker effect score. Pearson correlations were performed to test the influence of age and intelligence on the Flanker score.
28 When age and intelligence interacted with Flanker scores, these factors were used as covariates in the analysis.
Flanker scores and interactions
The results for reaction times and accuracy on the congruent Flanker trials are included in Table 4.
Table 4
Congruent Flanker trials
Lingualism Reaction time (SD) Accuracy (SD)
Monolinguals 717.5 (150.3) 0.93 (0.14)
Bilinguals 769.2 (152.2) 0.91 (0.13)
Monolingual foreign language learners 728.1 (159.6) 0.95 (0.11) Bilingual foreign language learners 755.4 (171.5) 0.91 (0.16) Table 4
Since intelligence may have an influence on the performance on executive functions tasks, correlation analyses between Raven scores and Flanker scores were performed. Correlating Raven scores with reaction times on the congruent Flanker task yielded weak, but significant negative correlation, r=0.194, p=0.014. There was no significant correlation between Raven scores and the accuracy in congruent trials, r=0.152, p>0.05. In addition, a Pearson correlation showed that there is a strong negative correlation between age in months and reaction time scores on congruent trials, r=-0.345, p<0.05, but there was no correlation between age in months and accuracy on congruent trials, r=-0.003, p=p>0.05.
The results for reaction times and accuracy on the congruent Flanker trials are included in Table 5.
29 Table 5
Incongruent Flanker trials
Lingualism Reaction time (SD) Accuracy (SD)
Monolinguals 733 (159.5) 0.88 (0.19)
Bilinguals 780.1 (161.5) 0.85 (0.2)
Monolingual foreign language learners 743.1 (156.6) 0.79 (0.24) Bilingual foreign language learners 764.5 (163.5) 0.84 (0.2) Table 5
A Pearson correlation between Raven scores and reaction times on incongruent Flanker trials gave a weak but significant negative correlation, r=-0.018, p=0.025. This indicates an interaction between intelligence and reaction time in both types of trials. More intelligent participants are quicker in both types of trials than participants with lower intelligence. There was also a weak but significant positive correlation between Raven scores and accuracy in the incongruent trials, r=0.17, p=0.036. This indicates that higher intelligence results in a higher accuracy in
incongruent trials. Reaction times in incongruent trials also showed a significant strong
correlation with age in months, r=-0.36, p=<0.00. Older participants are markedly quicker than younger ones in Flanker tests. In contrast, there was no significant correlation between age in months and accuracy scores incongruent trials.
Lastly, the Flanker effect was calculated. This is the difference between the reaction times of the incongruent trials minus the reaction times of the congruent trials. It needs to be noted that this Flanker effect score was calculated for each individual separately before calculating the means and is not the difference between the mean scores. The Flanker effect score is an indication of performance difference between congruent trials and incongruent trials. The Flanker effect scores can be seen in table 6.
30 Table 6
Flanker Effect Lingualism
Flanker effect (SD)
Monolinguals 15.53 (47.04)
Bilinguals 10.9 (46.98)
Monolingual foreign language learners 15 (27.43) Bilingual foreign language learners 9.04 (39.15) Table 6
There was no interaction between the Flanker effect score and age in months, r=-.115, p>0.05 and Flanker effect scores and Raven scores, r=0.041, p>0.05.
Flanker scores between subjects
Since intelligence and age in months proved to be an influence on congruent reaction time, an ANCOVA was performed with Raven score and age in months as covariates. The ANCOVA for Lingualism on congruent reaction times, controlling for age in months and Raven scores, found no statistically significant main effects, F(5, 158)= 0.637, p>0.05, interactions F(5, 158)= 18.172, p<0 and F(5, 158)=1.744 , p>0.05.
Since there was no interaction between accuracy in congruent trials and age in months and Raven scores, one-way ANOVAs were performed to investigate differences in accuracy between the groups. A one-way ANOVA did not reveal significant differences between the groups, F (3,158) =0.747, p>0.05.
Since there was also an interaction between Raven scores and reaction times of incongruent trials, Raven score was used a covariate in an ANCOVA. In addition, age in months was also of influence on the scores, so age in months was also included in the analysis. The ANCOVA for Lingualism on incongruent reaction times, controlling for age in months and Raven scores,
31 found no statistically significant main effects, F(5, 158)= 0.511, p>0.05, interactions F(5, 158)=
20.656, p<0 and F(5, 158)=1.236 , p>0.05.
There was an interaction between Raven scores and accuracy on the incongruent trials. Since intelligence influenced accuracy scores on incongruent trials, an ANCOVA with Raven scores as a covariate was performed. The ANCOVA for Lingualism on incongruent reaction times,
controlling for age in months and Raven scores, found no statistically significant main effects, F(5, 158)= 0.511, p>0.05, interactions F(5, 158)= 20.656, p<0 and F(5, 158)=1.236 , p>0.05.
Since age and intelligence did not influence Flanker effect scores, a one-way ANOVA was performed. A one-way ANOVA did not reveal significant differences between the groups, F (3,158) =0.235, p>0.05.
Overall, none of the analyses found significant differences between the groups on any of the scores of the Flanker task.
4.3 Wisconsin card sorting test
The following section discusses the scores on the Wisconsin Card Sorting Test. It covers the total achieved categories per set, total accuracy, a measure of distraction, and perseverance scores. As with the Flanker task, correlations were performed to investigate interaction between variables.
The basic measure of performance in the Wisconsin Card Sorting Test is the total number of categories achieved (that is, the number of times a participant understood a particular
matching condition). The maximum amount of categories that can be achieved is 6. A category is considered to be achieved when at least 5 correct responses per matching condition were completed. The scores from the Wisconsin Card Sorting Test on total categories can be seen in table 7. At first glance the scores are very similar across groups.
32 Table 7
Total achieved categories WISC
Lingualism categories (SD)
Monolinguals 4.71 (1.53)
Bilinguals 4.33 (1.45)
Monolingual foreign language learners 4.45 (1.6) Bilingual foreign language learners 4.57(1.5) Table 7
A Pearson correlation revealed that there was no correlation between age and total categories achieved, r=-0.07, p>0.05. A Pearson correlation between Raven score and total categories showed no significant correlation, r=0.136, p>0.05. This means that there is no discernible influence of age and intelligence on the mean total categories achieved. A one-way ANOVA did not reveal significant differences between the groups, F (3,157) =0.446, p>0.05.
Another measure of Wisconsin card sorting test performance is total accuracy. The accuracy scores can be found in table 8.
Table 8
Accuracy WISC
Lingualism Accuracy (SD)
Monolinguals 0.71 (0.13)
Bilinguals 0.65 (0.16)
Monolingual foreign language learners 0.67 (0.15) Bilingual foreign language learners 0.67 (0.13) Table 8
A Pearson correlation between age in months and total accuracy revealed no significant correlation. In addition, there was no correlation between Raven score and total accuracy on the Wisconsin card sorting task. Since there were no significant influences of age and
33 intelligence, a one-way ANOVA was performed to uncover differences in performance on accuracy. A one-way ANOVA did not reveal any significant differences between the groups.
The distraction value measures incorrect answers in a continuous series of the same rule or the inability to switch to a correct rule. A higher value indicates worse performance. A Pearson correlation did not reveal a relation between the distraction measure and Raven score. Age in months and the distraction measure of the Wisconsin card sorting test did also not correlate significantly. Table 9 shows descriptive statistics for distraction in the Wisconsin card sorting test.
Table 9
Distraction WISC
Lingualism Distraction (SD)
Monolinguals 0.16 (0.10)
Bilinguals 0.20 (0.09)
Monolingual foreign language learners 0.19 (0.12) Bilingual foreign language learners 0.20 (0.08) Table 9
Since there was no influence of age and intelligence on the distraction measure a one-way ANOVA was performed. The one-way ANOVA did not find significant differences between the groups. This means that all groups performed equally on this measure.
Lastly, the perseverance score is a measure of the ability to switch between rules. The more a participant keeps on using a rule that is not relevant anymore, the higher the score. A higher score means worse performance in terms of executive functioning. There was no correlation between age in months and Raven, and perseverance scores on Wisconsin trials. Table 10 gives the descriptive statistics for perseverance.
34 Table 10
Perseverance WISC
Lingualism Perseverance (SD)
Monolinguals 0.013 (0.018)
Bilinguals 0.014 (0.018)
Monolingual foreign language learners 0.014 (0.017) Bilingual foreign language learners 0.013 (0.018) Table 10
In order to assess possible performance differences for the perseverance measure in the Wisconsin Card Sorting Test a one-way ANOVA was performed. These results can be found in Table 11.
Table 11
ANOVAs WISC
Test F(df) p
Total categories 0.446 (3,157) >0.05
Accuracy 1.614 (3,157) >0.05
Distraction 1.259 (3,157) >0.05
Perseverance 0.050 (3,157) >0.05
Table 11
The results from this test yielded no significant differences between the groups.
4.4 results overview
One of the background measures, intelligence measured through the Raven Standard Progressive Matrices, was strongly influenced by age. When controlling for age, significant
35 differences in intelligence were found between the monolinguals and bilinguals, monolinguals and bilingual foreign language learners, and monolingual foreign language learners and
bilingual foreign language learners. Since there were significant differences in intelligence between the groups, extra care was taken to account for the influence of age on the executive functions tests.
In a number of scores of the Flanker task there was an influence of age or intelligence. When controlling for these influences, there were no observable performance differences between the different language conditions. Age and intelligence did not seem to have a large influence on the Wisconsin card sorting task and did not influence any of the subscores obtained on this test. Again, like with the Flanker test, the Wisconsin card sorting test did not find mean
differences between the groups.
5. Discussion and conclusion
The first research question in this study set out to answer was whether bilingual children perform better than monolingual children on executive function tasks and have what is known as a ‘bilingual advantage’. Since the majority of studies that looked at bilinguals from birth found this bilingual advantage (see the meta-study by Adesope, Lavin, Thomson, & Ungerleider (2010)), significant advantages for bilinguals were expected to be found in this study as well.
Analyses of the results from the Flanker task and the Wisconsin Card Sorting Task, however, did not reveal differences in performance. It must be concluded that bilingual children do not have an advantage over monolingual children - at least not at this age, on not on these specific executive functions. There are a number of reasons why this might be the case.
The majority of studies have looked at bilinguals from birth and early sequential bilinguals. The literature suggests that advantages of being bilingual are correlated with the extent that two languages are used (Gathercole et al., 2007). The more balanced the use of both languages, the more often the speaker must switch. In this study, therefore, it was expected that the two foreign language learner groups would not perform differently from their non-foreign language
36 learner counterparts due to a limited amount of language input in the foreign language. It was surprising, however, to find no differences in performance between monolinguals and
bilinguals either. Since these bilinguals have been hearing and speaking two languages from birth, one would expect differences in executive function performance, due to the constant managing of two languages.
However, when reviewing the extant literature it becomes clear that some, but not all, not all researchers find these differences. For example, Kalashnikova and Mattock (2012) found significant differences between early sequential bilinguals and monolinguals, whereas Gathercole et al. (2007) did not. Both were studying the same language groups. In addition, most studies have looked at students in immersion programs (for example Bialystok and Barac, 2012; Carlson and Melzoff, 2008; Esposito et al., 2013; Poarch and van Hell, 2012) or immigrant children (Morales, Calvo and Bialystok, 2013; Bialystok, 2011; Calvo and Bialystok, 2014). Only Poulin-Dubois (2011) and Jasinka and Petitto (2013) looked at actual bilinguals from birth and found significant differences.
Gathercole et al. (2007) suggested that bilinguals from birth are less likely to show advantages in executive function and that only advanced foreign language learners benefit in executive functioning. There is a possibility that advanced foreign language learners are able to tap into their switching capabilities to a greater extent than early language learners with only a limited exposure to a foreign language. Full bilinguals, on the other hand, will be so proficient in using two languages in different contexts, that using two languages might have no bearing on their switching capabilities. If this is true, then the only groups of multilinguals that may benefit in terms of executive functioning are young advanced language learners. Advanced foreign
language learners, however, were not included in this study, and this may be why no significant effects were found.
Several studies have demonstrated the importance of controlling for external influences in executive functions research. In language related executive functions studies, the focus is usually on age-groups and socio-economic class through either questionnaires or geographic
37 location. The first reason this study did not find ‘lingualism’ effects where other studies did could be in the way such external influences were, or were not, controlled for.
Many of the studies that were discussed earlier did not control for intelligence (Calvo &
Bialystok, 2014 ; Bialystok, 2011; Morales, Calvo & Bialystok, 2013; Jasinka & Petitto, 2013;
Poulin-Dubois et al., 2011), This study did, and found that intelligence does play an important role in executive function performance: intelligence was shown to affect participants’ reaction time scores, though not their accuracy, in the Flanker task. It is possible that no significant results were found because intelligence was rigorously controlled for – the significant results found in previous studies may have been due to the effect that intelligence has on executive functions.
This study demonstrated that age has an influence on executive function performance as well - even when the age range under investigation spans no more than 1.8 years. Carlson, Zelazo and Faja (2013) showed in the aggregate scores for the Flanker and the Wisconsin Card Sorting Test that children exhibit a strong increase in executive function performance between 4 and 12 years of age – an age range frequently studied. Given the steep increase in executive function performance during this time, it is likely that even within a group of children not more than one year apart in age, a few months can have profound effects on executive function performance.
Many other researchers in the field acknowledge the influence of age on executive function performance and therefore control for age (Jasinka & Petitto, 2013; Poarch & van Hell, 2012;
Esposito et al., 2013; Carlson & Meltzoff, 2008; Morales, Calvo & Bialystok, 2013; Bialystok &
Barac, 2012; Bialystok, 2011). It is interesting to observe, however, that although all of the studies discussed herein select participants based on age, age is then always used to group participants in an age category. The ranges in age per group varied between roughly 36 months (Jasinka & Petitto, 2013; Poarch & van Hell, 2012), to 24 months, (Esposito et al., 2013; Carlson
& Meltzoff, 2013) and 12 months (Morales, Calvo & Bialystok, 2013; Bialystok & Barac, 2012;
Bialystok, 2011). The smallest range that could be found was 5 months (Kalashnikova &
Mattlock, 2012). Grouping of children into 12- or 24- month cohorts seems to be common practice. None of these studies have included age as a covariate or in a regression analysis to
38 consider its effect within the group. Since age turned out to be of influence on several subtests in this study and previous studies have shown dramatic increases in executive performance growth at a young age, it could therefore be possible that the influence of age may have influenced the results of the aforementioned studies. In this study, age was used as a within- cohort variable, giving a more precise representation of age-related effects on executive function performance
It is interesting to note that intelligence and age both influenced performance on the Flanker task, but did not appear to influence performance on the Wisconsin Card Sorting Test. This difference may be due to the fact that the tests each tap into different aspects of executive functions. The Flanker task is a fairly specific test designed to focus on the inhibition control aspect of executive functions (Carlson, Zelazo & Faja, 2013), whereas the Wisconsin card sorting test is a compound test which looks at several aspects of executive functions, like switching and working memory (Carlson, Zelazo & Faja, 2013). The Wisconsin Card Sorting Test requires participants to use a larger set of skills, thus possibly cancelling out the effects of age and intelligence.
Another factor that possibly influenced the results obtained by this study is the manner in which socio-economic status was controlled for. Previous research has demonstrated that the influence of socio-economic status on executive functions is substantial (Calvo & Bialystok, 2014). The measures this study used to control for socio-economic status may therefore not have been precise enough: geographic proximity and ‘leerlinggewicht’ (student weight) were used to select participating schools, and it was assumed that the combination of these two global measures would lead to a reliable control. However, it may be that SES requires more stringent control. If so, then the small differences in the participants’ socio-economic status that were present in this study may have obscured the effect of their ‘lingualism’. Many other studies on executive functions controlled for SES with the use of a questionnaire, which was frequently developed by the researchers themselves. The absence of a standardized
questionnaire makes cross-study comparisons difficult. Nevertheless, a questionnaire was also developed for and administered during this study, but due to time constraints its results could
39 not be incorporated in the analysis. Future analyses on this data will take this questionnaire into account.
A second reason that no ‘lingualism’ effects were found may have to do with the amount of relatedness between the spoken languages that were under investigation. Studies that did find significant results often investigated combinations of English and Spanish (Esposito et al., 2013;
Carlson & Melzoff, 2008), English and French (Poulin-Dubois et al, 2011; Jasinka & Petitto, 2013;
Bialystok & Barac, 2012), English and Welsh (Kalashnikova & Mattlock, 2012; ), English and Hebrew (Bialystok & Barac, 2012), English and German (Poarch & van Hell, 2012) or a
combination of English and other languages (Morales, Calvo & Bialystok, 2013; Bailystok, 2011;
Calvo & Bialystok, 2014). If executive function performance is improved by having to keep two language systems separate, it is possible that the relatedness of the languages in question influences the need to keep those languages apart. Languages that are very similar are therefore more likely to become part of one system, whereas two languages that are very different may be more separated. This study looked at different combinations between Dutch, Frisian and English. These three languages are all Germanic in origin and have many shared vocabulary and syntax features. Since these languages are hard to keep apart they may possibly become part of the same system. Consequently, bilingual speakers do not switch between two sets of languages, but rather keep them in one system. The question, then, is this: all else being equal, does the relatedness of the languages a speaker commands affect his or her executive function performance? It appears this has not been looked at before, and future research is needed to help answer this question.
A third factor that may have influenced this study’s results is the fact that testing was carried out at the schools. These made for inherently unpredictable environments. Many other, similar studies tested the children individually in a quiet room (Poarch & van Hell, 2012; Esposito et al, 2013; Carlson & Meltzoff, 2008; Morales, Calvo & Bialystok, 2013; Bialystok & Barac, 2012;
Bialystok, 2011) or laboratory setting (Jasinka & Petitto, 2013). The study by Bialystok and Barac (2012) does not specify how the tests were carried out. For this study, the tests were done at school to allow for more efficient testing and to observe the children in a natural environment.
