PRIOR KNOWLEDGE OF ENGLISH IN DUTCH PRIMARY SCHOOL PUPILS BEFORE RECEIVING L2 INSTRUCTION

(1)

PRIOR KNOWLEDGE OF ENGLISH

IN DUTCH PRIMARY SCHOOL

PUPILS BEFORE RECEIVING L2

INSTRUCTION

LUCINDA CECILIA ZONNEVELD

S2376040

MA in Applied Linguistics

Faculty of Liberal Arts

University of Groningen

Supervisors:

Dr. M.C.J. Keijzer (primary supervisor)

Prof. dr. W.M. Lowie (second reader)

19

th

of June 2017

Word Count: 16.263

(2)

page

Acknowledgements ... 4

0. Abstract ... 5

1. Introduction ... 6

1.1. Social and theoretical relevance ... 7

2. Theoretical background ... 8

2.1. Early start of a second language ... 8

2.1.1. The essence of English ... 9

2.2. Previous studies on pupils’ English L2 knowledge ... 9

2.3. Incidental versus intentional learning ... 12

2.3.1. Incidental learning ... 12

2.3.2. Intentional learning ... 13

2.3.3. Studies comparing incidental and intentional learning ... 13

2.4. English in Dutch primary schools ... 14

2.4.1. The start of English in Dutch primary schools ... 15

2.4.2. Different ways of teaching English ... 16

2.5. English in Dutch media ... 16

(3)

3.4.2. Listening test ... 26 3.4.3. Reading test ... 27 3.4.4. Speaking test ... 27 3.4.5. Questionnaires ... 27 3.4.6. Statistical analyses ... 28 4. Results ... 29

4.1. Results per language test ... 29

4.2. Pupils’ background information ... 30

4.3. Relationships between the four language tests ... 31

4.4. Relationships between language skills and other factors ... 34

4.4.1. Vocabulary test ... 34

4.4.2. Listening test ... 34

4.4.3. Reading test ... 35

4.4.4. Speaking test ... 35

4.5. Differences between pupils with different language backgrounds ... 37

5. Discussion ... 39

5.1. Language tests ... 39

5.2. Pupils’ background information ... 39

5.3. Outliers ... 40

5.4. Modulating factors in predicting foreign language success ... 41

5.4.1. Media ... 41

5.4.2. Parental background ... 44

5.4.3. Attitude towards English ... 45

5.5. Teaching implications ... 45

6. Conclusion ... 47

(4)

Acknowledgements

Firstly, I would like to thank Merel Keijzer, my supervisor, for her useful tips, expertise, guidance and enthusiasm about my thesis topic. Her help made it easier to finish this long intensive project. Secondly, I would like to express my gratitude to all the members of the applied linguistics department in Groningen. Through their interesting courses, I have learned a lot about language systems, statistics, teaching and presenting my work, which was helpful in writing this thesis.

Thirdly, I am grateful that my old primary school, the Sint Jozef, gave me the opportunity and time to administer the language tests in the two year 6 classes. Therefore, I want to thank, in particular, the teachers of these two classes and their pupils whom I all know by name. Without their participation, none of this would have happened.

(5)

0. Abstract

Only a few studies have investigated the influence of media exposure on English L2 knowledge of Dutch children who started formal L2 instruction (Naber & Lowie, 2012; Verspoor, De Bot & Van der Heiden, 2007). These previous studies tested the media influence on one or two language skills (i.e. listening or speaking), but not on all four, and only examined the effects of a few factors on L2 proficiency. However, the present study investigated all language skills and included, besides various media forms, more variables on the children’s language backgrounds and contact with English to see how these can influence English L2 proficiency. In this study, 60 Dutch primary school children that had not yet started formal L2 English instruction participated. These pupils completed a vocabulary, listening, reading/writing and speaking test. With the help of a student and parental questionnaire, information on the pupils’ language background, media use, contact with English and the background of their parents was obtained, which was related to the language test results. The results of the language tests showed substantial between-subject variation among the pupils and strong relationships between the four language tests. Nonetheless, most children found the productive language tests the hardest, whereas the pupils had less difficulty with the vocabulary test. The language test scores were positively influenced by several media forms: using YouTube and social media, watching English broadcasts, listening to English music and conversing with English-speakers. Of all these predictors, speaking frequently in this L2 seems to be most beneficial since there were strong significant relationships between speaking English outside of class and all language test scores. Additionally, children who were already multilingual in two or more language systems other than English also showed higher scores on the speaking test. In conclusion, it can be stated that being frequently exposed to English through various (media) forms lead to better L2 proficiency scores, even in the absence of formal instruction at primary school.

(6)

1. Introduction

Over the past few years, the role of the English language and culture in Dutch society has increased enormously. Since English is the most widely spoken language in the world, the European Union considers the language a lingua franca, which has resulted in EU countries obligatorily teaching a second language, in particular, English in primary schools (Adam & Gonzalez, 2010). This situation is no different in the Netherlands, where already in 1986 English was introduced to the primary school curricula (Oostdam, 2010; Oostdam & Toorenburg, 2002). Moreover, the Netherlands decided to provide subtitles to English television programs and movies instead of dubbing them. With the introduction of the subtitles together with the expansion of English broadcasts and viewing behaviours of Dutch natives, a situation started wherein English is often heard in Dutch households (Webb & Rodgers, 2009). All members of these households are thus exposed to English; children too receive this kind of input. In consequence, children are exposed to various English media forms from a very young age onwards, even before they receive formal instruction in this L2, which for most Dutch primary schools starts when the pupils are around the age of 11 (Thijs, Trimbos, Tuin, Bodde & De Graaff, 2011). It is interesting to know what effects such extensive media exposure can have on L2 English knowledge, because through this shift in the media it is most likely that children can already produce some English words and sentences before they starting formal instruction. A previous study by Eyckmans and De Wilde (2016), who investigated the effects of English media exposure on 12-and-13-year old Flemish pupils revealed that, in particular, gaming had a positive influence on L2 knowledge. Different language backgrounds of the parents and gender did not show any effects on the acquisition of English in their work.

In this current study, children’s prior knowledge of English is investigated by a listening, reading/writing, speaking and vocabulary test. The children’s scores on these measures are compared to the pupils’ and parents’ answers about their language background, media consumption and other contact with the English language, collected through questionnaires. The following research question forms the basis of this study:

(7)

1.1. Social and theoretical relevance

The results from this study can be useful for various parties. Firstly, information on this topic can help policy makers readjust the age at which children should receive their first formal English L2 instruction in school. In the current situation, the vast majority of primary school pupils start with their first instruction in year 7, when the children are ten or eleven years old. With the findings of this study, policy makers can decide if this age needs to be pushed upwards or backwards. Secondly, teachers, in particular those instructing English, can be informed about the prior knowledge of their students. When it is clear what children already know about the L2 before they start formal instruction, teachers can easily tune in on their students’ current level of knowledge. Subsequently, the teachers do not have to start with the basics of the English language.

(8)

2. Theoretical background

2.1. The early start of a second language

There is a general expectation that if pupils start learning a second language at an early age they will reach a high level of proficiency. At a European conference in Lisbon that took place in 2010, EU members agreed to stimulate bilingual education for young learners. Afterwards, countries like Spain and France mandatorily started teaching second languages in primary schools, in line with early L2 teaching as a worldwide phenomenon (Adam & Gonzalez, 2010). Besides easier communication within the European community and learning linguistic knowledge and skills, learning a foreign language is also an introduction to a new culture. In this way, pupils learn more about themselves and their place in society, which most likely influences the child’s self-image (Abdallah-Pretceille, 1991, p.96).

However, there are also a number of studies that have questioned whether earlier is always better (Corda, Phielix & Krijnen, 2002; Goorhuis-Brouwer & De Bot, 2005; Naber & Lowie, 2012). They have investigated the effect that learning an L2 early in life has on the development of the mother tongue. They expected that a considerable amount of input in a second language would hinder the development of an L1. Nevertheless, the results proved that learning an L2, in all cases English, was not detrimental for mother tongue performance (Dutch in all cases). A study by Broekhof (2007) even showed that the possession of more language systems if anything else benefits the mother tongue development, as children and adults who mastered two languages possess more general knowledge about languages and know how languages operate in the brain. However, the Onderwijsraad (in English: Educational Council) (2008) has found that the possession of more language systems increases the chance of interferences: words and elements of one language are adapted to another. According to Poulisse and Bongaerts (1994), these language switches and interferences are related to the learner’s level of proficiency, with interferences being especially prevalent at early stages of development.

In spite of the provided advantages and importance of English as a subject in primary schools, pupils are generally not tested on English in their final primary school exam. Such standardized English tests do exist: CITO1 developed these for year 7 and 8 of Dutch primary schools and some schools have now tentatively started using them as a measure of the pupils’ progress in this L2 (Carpay, 1990a; CITO Me2! Engels, 2010). However, these CITO tests are only developed to determine the English L2 receptive language skills, listening and reading, and

(9)

vocabulary knowledge on the A1 level of the Common European Framework of Reference (CITO Me2! Engels, 2010).

2.1.1. The essence of English

In the current socio-economical situation of the Netherlands, speaking English is essential. A mutual language is needed to trade with other countries, to study abroad or to be understood in an English-speaking country in general. These are only some examples in which speaking English, in addition to a mother language, can be helpful. In addition, the Netherlands is a small country with a language that is only spoken by 23 million people compared to a group of 328 citizens whose mother tongue is English. Adam and Gonzalez (2010) explain that in the European Union the English language has attained the position of lingua franca, which is seen as indispensable for communication with other parties in the Union. Due to the increased importance of the language, the Dutch Educational Council recommends English as a required course in primary schools, especially seeing that most other European countries introduce English at an even earlier age, i.e. at age 7 (Oostdam, 2010; Van Ooijen, 2002). Additionally, the Netherlands is taking a special position in the acquisition of English, because unlike many other countries, English (media) input is so extensive.

2.2. Previous studies on pupils’ English L2 knowledge

(10)

(11)

the L2 learning process in a more conscious way. Moreover, through the evolvement of the L1 developed cognitive skills, the acquisition of a second language is supported (Naber & Lowie, 2012, p.19). Difficulties are especially expected for children with language developmental disorders, such as dyslexia, but the Onderwijsraad nevertheless does advise the same starting age, as they are not likely to show difficulty acquiring receptive L2 skills (Onderwijsraad, 2008, p.43). Several other studies investigated the English L2 knowledge of high schools pupils in the Netherlands, acquired through media exposure, school and other resources (Berns, De Bot & Hasebrink, 2007; Bonnet, 2004; Verspoor, De Bot & Van der Heiden, 2007). In the study by Berns, De Bot and Hasebrink (2007), participants indicated in percentages how much English they had acquired through three different forms: school, media and other forms with the latter classified as personal networks Compared to Germany, France and Spain, Dutch high school pupils indicated to learn less English in school, 53% to 58%, whereas the other countries reported to acquire more than 60% of English L2 knowledge through formal school instruction (Berns, De Bot & Hasebrink, 2007; Bonnet, 2004). Despite the fact that Dutch high school pupils indicated to have less interaction with the English language in school, the differences with other countries with respect to the amount of English learned through school was only up to 5%. Furthermore, Bonnet (2004) and Berns, De Bot and Hasebrink (2007) revealed that the most important factors of media exposure were television, popular music and movies. Computers and written media, such as books and newspapers, were seen as less essential in learning English as a second language. The latter study by Berns, De Bot and Hasebrink (2007) found a weak but significant correlation between L2 proficiency and contact with English. The term ‘contact’ included contact with friends and family members, contact through electronic media like computers, television and music, contact through literature and contact on holidays. The study by Bonnet (2004) showed similar findings.

(12)

more English media does not necessarily make a difference in listening comprehension. A possible explanation for better listening scores for pupils in religious education is most likely due to the way of teaching: in the religious class there was more interaction with the English language between the pupils and teacher.

2.3. Incidental versus intentional learning

The participants in the study by Verspoor, De Bot and Van der Heiden (2007) received English input incidentally, through media, and intentionally, through L2 instruction in school. These two broad perspectives, incidental and intentional learning, are generally used for second language learning and teaching. Intentional learning includes the memorization of mainly vocabulary items and explicitly being offered grammatical rules, whereas incidental learning encompasses the uptake of different words and structures by engaging in various communicative activities, especially activities that involve reading and listening, causing the learner to focus on the meaning instead of the form of the language (Hulstijn, 2003, p.349). In the field of applied linguistics, these two terms are also called explicit and implicit learning.

2.3.1. Incidental learning

In addition to Hulstijn’s definition (2003), Schmidt (1994a, p.16) defines incidental learning as “the learning of one thing (e.g. grammar) when the learner’s primary objective is to do something else (e.g. communicate)”. Gass (1999) recommended extending the meaning of incidental learning to include the learning of grammatical structures without exposure to examples on how these structures work. This recommendation notwithstanding, incidental learning is mostly taken to relate to the acquisition of lexical items, acquired through reading and other forms of lexical input without explicitly offering word lists to memorize (Hulstijn, 2005; Hulstijn, 2013; Vidal, 2011). Such other forms of input can include playing games in an L2 or reading English subtitles of movies. Even though the term incidental learning states that vocabulary learning is an unconscious process, learners can still be made aware of the strategies that they apply to acquire words (Ellis, 1994).

(13)

translated to another context, they are better memorized.

2.3.2. Intentional learning

Intentional learning involves the memorization of successive facts, which asks high demands of the working memory (Ellis, 2009, p.3). Since intentional learning is a conscious process, learners are aware of their learning and can explain others what they have learned (Ellis, 2009). In this learning process, learners are (made) to take responsibility for their own learning and regulate their learning strategies as well as their learning processes (Hung, 2014). This method is still often used in secondary schools and at the end of primary school for learning English, although in primary schools a more naturalistic approach is often taken. Within intentional learning, learners can set their own learning goals that could be activated by internal or external reasons, by which intrinsic motives can persist throughout an individual’s lifespan and external reasons mostly last only for a short period of time. Examples of intrinsic intentional learning are personal interests or needs and an extrinsic motive can be a secondary school teacher who highlights the importance of L2 knowledge. In this latter setting, pupils pay selective attention to English and use consciously different strategies to uncover the meaning of linguistic elements (Ellis, 1994).

2.3.3. Studies comparing incidental and intentional learning

(14)

dictionary, do not always result in the contextually appropriate meaning. Therefore, she advised to combine the tactics of these two types of learning to prevent extraordinary definitions of words. Another study by Reber, Walkenfield and Hernstadt (1991) investigated to what extent individual differences play a role in implicit and explicit learning and did so by comparing participants’ performances on different tasks. The implicit task included a standard artificial grammar-learning task whereas the explicit one was a completion problem-solving task. A strong correlation was found between the performances on the explicit task and the participant’s IQ although this relationship was not discovered for the performances on the implicit task. A possible explanation for this finding can be found in the evolutionary model, which argues that unconscious, implicit induction systems are much older than conscious explicit learning processes and that his antiquity possesses specific patterns of function that distinguish implicit and explicit processes from one another (cf. Reber, Walkenfield & Hernstadt, 1991).

A study by Trahey (1996) investigated the long-term effects of the memorization of adverb placement rules by young French ESL learners. A year earlier, the participants had received - over a period of two weeks - linguistic data on English adverb placement after which they had to form sentences. It turned out that the learners were not able to produce English grammatical sentences, but these sentences were grammatically correct in French. The follow-up test showed that after one year, the participants still made mistakes in the placement of these adverbs. This means that implicit instruction did not result in the controlling of these structures. Another study by Rousse-Malpat and Verspoor (2012) investigated the effectiveness of explicit versus implicit methods in French speaking. After two years of instruction, free speech data of high school children who received either implicit or explicit instruction were scored on grammatical accuracy and general proficiency. The study revealed that after two years the children who received implicit instruction outperformed the pupils with explicit instruction on both proficiency and accuracy. This means that the ‘unconscious’ process of implicit instruction was found to be more beneficial on the long-term than rule learning and memorization in explicit instruction.

2.4. English in Dutch primary schools

(15)

Dutch primary school pupils still start with formal English instruction in year 7, when children are 10 or 11 years old.

2.4.1. The start of English in Dutch primary schools

(16)

2.4.2 Different ways of teaching English

Due to the internalisation and the European Union, most European countries including the Netherlands, provide high school pupils with the opportunity to learn more than two foreign languages, whereas in primary schools only one L2 is obligatory (Oostdam, 2010, p.9). Over the last years, there has been a steep rise of Dutch primary schools that teach second languages at an earlier stage. These so-called VVTO (in Dutch: Vroeg Vreemde Taal Onderwijs, Early Foreign Language Education) schools can offer English, German, French or Spanish from year 1 onwards, but are often only specialized in one of these languages. In other words, they start their foreign language curriculum when children enter school, which is at age 4 in the Netherlands. The number of VVTO schools in the Netherlands has thus risen enormously: in 2007, there were only 127 schools offering this type of education whereas five years later the number of schools had increased to 658. The vast majority of these VVTO schools offer English as a second language (Inspectie van het Onderwijs, 2013).

2.5. English in Dutch media

With the rise of internationalisation and globalisation, television programs, movies and series from all around the world are accessible in almost every country. The Netherlands and Belgium decided to provide subtitles to English broadcasts instead of dubbing them (Eyckmans & De Wilde, 2016). The introduction of this decision, the expansion of English spoken television programs and the viewing behaviours of Dutch speakers together caused a situation wherein English is frequently heard in Dutch and Flemish living rooms (Webb & Rodgers, 2009). It has even been claimed that English forms part of Dutch and Flemish households’ wallpaper (Eyckmans & De Wilde, 2016, p.1). Through this repeated exposure to the language, one can argue that the English language is part of the Dutch culture. Furthermore, this exposure to the English language also holds for the youngest Dutch citizens: children. This implies that Dutch native children come into contact with English long before they receive formal instruction in primary school on this L2.

(17)

provided better results on the English to Dutch translation test, but not on the Dutch to English translation test, which may indicate that children have difficulties retrieving English words from their brain and the production of these words. It should also be pointed out that translation is a highly specific skill that may not be the best test to use with Kuppens’ objective in mind. However, Sylvén and Sundqvist (2012) found similar results in their study about gaming and English L2 learning. 11-to-12-year-old children participated in a reading, listening and vocabulary test to measure their level of proficiency. Moreover, they filled in a questionnaire and kept a diary in which they documented the amount of time they were in contact with the English language outside of school. The evidence revealed that the frequency of gaming correlated with the proficiency level of the second language. In other words, more gaming helped to gain a higher proficiency level in English. An earlier study by the same author also investigated a possible relationship between playing video games and L2 proficiency, but in this case older learners who were between 15 and 16 years old were tested (Sundqvist, 2009). Indeed, a positive correlation between the two variables was found, especially with regard to the L2 vocabulary knowledge wherein the boys outperformed the girls. This outcome can be explained by the frequency of gaming and the various types of games that the boys played. Several other studies too reported positive benefits of gaming on L2 knowledge, in particular in relation to vocabulary improvement (Marsh & Tainio, 2009; Miller & Hegelheimer, 2006; Ranalli, 2008).

2.6. Hypotheses

On the basis of what is currently available, the present study investigates the English prior knowledge of Dutch primary school children before they start formal English instruction. The work reported here is essentially a replication of the study by Eyckmans and De Wilde (2016). As such, the same hypotheses and predictions are used.

Eyckmans and De Wilde (2016) first of all found no significant differences between the proficiency scores of boys and girls. Other studies showed either an advantage for boys or girls (Kuppens, 2010; Sundqvist, 2009). Despite this inconclusive evidence on better performances for one of the gender, the study by Eyckmans and De Wilde (2016) is seen as the guideline for the formulation of the hypothesis on gender since their study is most similar to the one presented here. This leads to the following hypothesis:

H1: There are no differences in English L2 proficiency scores between boys and girls

(18)

pupils with parents who had different language backgrounds. However, the results showed no significant differences between pupils with native and non-native Flemish-speaking parents. Consequently, with this information the next hypothesis is formed:

H2: There are no differences in English L2 proficiency scores between pupils with different language backgrounds

Numerous studies demonstrated the advantages of extensive media exposure, most notably gaming, on L2 learning (Berns, De Bot & Hasebrink, 2007; Bonnet, 2004; Eyckmans & De Wilde, 2016; Kuppens, 2010; Naber & Lowie, 2012; Marsh & Tainio, 2009; Miller & Hegelheimer, 2006; Ranalli, 2008; Sundqvist, 2009; Sylvén & Sundqvist, 2012; Verspoor, De Bot & Van der Heiden, 2007). The majority of these studies indicated that playing games is most beneficial for the acquisition of English L2 vocabulary (Marsh & Tainio, 2009; Miller & Hegelheimer, 2006; Ranalli, 2008; Sundqvist, 2009; Sylvén & Sundqvist, 2012). The other studies reported the positive influences of English television, music, movies and computer hours on the L2 proficiency level (Berns, De Bot & Hasebrink, 2007; Bonnet, 2004; Eyckmans & De Wilde, 2016; Naber & Lowie, 2012; Verspoor, De Bot & Van der Heiden, 2007). Through these exclusively positive findings, it is expected that media exposure positively correlate with the level of proficiency in the second language. Therefore, the following hypothesis is formulated:

H3: Extensive media exposure, most notably gaming, results in better L2 proficiency scores

Heesters et al. (2008) investigated how Dutch primary school children view English as a school subject. Their study discovered that pupils found the subject important, but not fun or easy. Eyckmans and De Wilde (2016) reported a contrasting finding in which pupils were primarily positive towards the English language. Since the latter study has many similarities to the study described in this paper, their outcome is the basis for formulating the fourth hypothesis, which is as follows:

(19)

3. Method

In this chapter, the method underlying the present study is explained in detail. In particular, the operationalization of the variables in the research question, the participants, the tests and the procedure are discussed.

3.1. Operationalization

Before the research question whether Dutch primary school children have prior knowledge of English before they start with formal instruction in the language can be answered, it is important to operationalize some basic constructs underlying this question. More specifically, prior knowledge is here operationalized by measures of the pupils’ receptive and productive knowledge of English. Receptive knowledge of a language includes the ability to listen and read, whereas productive knowledge consists of the other two language skills: writing and speaking (Saville-Troike, 2006, p.136). According to Saville-Troike (2006), listening and speaking activities merit the highest priority for interpersonal contexts to generate communication.

In this study, receptive knowledge was measured through a listening test with 25 questions and to measure productive knowledge a ten-minute recorded conversation consisted of four parts was completed. Reading and writing skills were investigated through a combined test with 50 questions. Lastly, vocabulary knowledge was determined through a 120-item multiple-choice test (Peabody Picture Vocabulary Test, see paragraph 3.3.1) wherein a word was uttered after which the pupils had to choose the corresponding picture on a slideshow.

The other term in the research question that requires further explanation is the term formal instruction. Formal instruction means that the school included the subject English in the curriculum, obliging teachers to provide English lessons a set number of hours per week. The term does not imply that pupils cannot participate in English classes outside of school: formal instruction only concerns obligatory lessons of English within school time.

3.2. Participants

The study was performed with the help of 60 participants between the age of 8 and 11 years old. These children were pupils in year 6 of the Dutch primary school St. Jozef, who did not yet start formal English instruction. In this school, English lessons were provided to children in year 7, which made this group of participants suitable for the study.

(20)

all children participated in the study, pupils with learning difficulties were also included. The mean age in class 6a was 9.47 years with a range between 9 and 11 years. In the other class, class 6b, the mean age was slightly lower, namely 9.27 years with a minimum of 8 and a maximum of 10 years.

In both classes, the quantity of boys was higher, respectively 16 boys in class 6a and 18 in class 6b. This means that in total 26 girls participated in the study: 14 girls in class 6a and 12 in class 6b (see Appendix K).

The St. Jozefschool is located at the border of the cities Amstelveen and Amsterdam. Through an internal investigation, the school determined 65% of its pupils to come from Amstelveen and 35% from Amsterdam (St. Jozefschool, 2015). This area is known for its highly-educated inhabitants, where in families both parents work on a fulltime basis. Therefore, some of these two-earner households have a babysitter or au pair that takes care of the children after school. Most of these au pairs are non-natives of the Dutch language and speak English to the children. Furthermore, the number of citizens with an international background increased over the past few years (Gemeente Amstelveen, 2013; Gemeente Amsterdam, 2014). Information on the pupils’ parents confirmed both statements since the majority of the parents were found to be highly-educated, respectively 93.3% of both the mothers and fathers were postgraduates or had a higher education (Dutch: HBO) degree. Interestingly, approximately one-seventh of them spoke another language than Dutch as their L1, respectively 10 mothers and 6 fathers. More specifically, there were six English-speaking mothers, four mothers who spoke another language than Dutch or English and the six fathers were equally divided between the English-speaking group and the group that had a different L1 than English or Dutch, leaving 49 mothers and 51 fathers with Dutch as their mother tongue. Since participant number 60 did not provide information on his parents and two pupils’ fathers passed away, information on the language background of 116 parents was available. In short, it can be stated that the population was right skewed in socio-economic status.

(21)

3.3. Materials 3.3.1. Vocabulary test

To test the vocabulary knowledge of the pupils of year 6, a Peabody Picture Vocabulary Test – fourth edition (PPVT-4) was used (Dunn & Dunn, 2007). A PPVT is a multiple-choice vocabulary test that is designed to assess one-word receptive knowledge in children and adults (Dunn & Dunn, 2007). A PPVT is divided in different sets consisting of twelve items and is linked to the vocabulary knowledge that native speakers of specific ages possess. When a test taker makes more than three errors in a set, the examiner stops the test. For example, a participant can reach set 2 of the PPVT, which is identical to the vocabulary knowledge of 5-year old English native. However, due to lack of time, the PPVT was not examined at an individual level, but through a plenary session in both classes and by means of a PowerPoint version. Prior to the PPVT, it was determined to stop at set 12, which corresponds with the vocabulary knowledge of an 11-to-12-year old native speaker. This was realized since Dale, Van der Es and Tanner (2010) state that learners have to be tested one level above their current level, which is also known in the linguistic field as i + 1. This choice implied that the participants had to answer 120 questions on an answer sheet. This answer sheet can be consulted in Appendix A. Image 1 shows the used example of the PPVT where the word ‘boy’ was expressed by the examiner, which corresponds with number 3 out of the four drawings.

(22)

1

2

3

4

Image 1: Example item of the Peabody Picture Vocabulary Test

3.3.2. Listening test

The listening test was a standardized Cambridge test developed for L2 English learners who are in the last year of primary school or the first year of secondary school. The test consisted of 25 questions spread out over five parts. Through an audio recording, the pupils had to fill in missing words, colour elements in a drawing, answer multiple-choice questions, draw lines between names and characters, and connect images with numbers. All audio fragments were played two times, so that the children had the possibility to check or fill in missing answers. In total, this part of the language test lasted 25 minutes.

(23)

Image 2: A multiple-choice question in part 4 of the standardized Cambridge listening test

3.3.3. Reading/writing test

The reading/writing test was, again, a standardized Cambridge test developed for 12-year-old children including 50 questions divided over 7 parts. The sets increased in difficulty. Every part started with a Dutch and English explanation about what was expected followed by one or two examples. The participants had to find the correct word that fit a definition, evaluate statements about a drawing, choose the right answer out of a list of options, fill in the correct word(s) in a sentence by reading a story and had to insert missing words from their own vocabulary knowledge into an email. The reading/writing test had to be finished in 40 minutes since this amount of time was a recommendation by the Cambridge test developers. Henceforth, the reading/writing test will be called the reading test, but also comprises a writing component.

Image 3 presents an example of two sentences where missing words had to be filled in. In most exercises, the participants chose the correct answer out of a wordlist whereas in this task an answer had to be retrieved from the participants’ own English vocabulary knowledge. Image 3 is an example of the exercise in part 7; the complete exercise and other parts of the reading test can be found in Appendix C.

(24)

3.3.4. Speaking test

Like the listening and reading test, the speaking test was also a standardized Cambridge test developed for older children at primary school. The test lasted approximately 10 minutes per participant, but the exact time varied since the participants’ experience with the English language differed. The speaking test, which was audio recorded, contained four parts of which three parts were dialogues with the examiner. In these three parts, the pupils had to find and explain differences, ask and answer questions about two characters, and had to respond to questions about their families, hobbies, favourite food and holiday. In part 3 of the speaking test, participants were asked to complete a story about Tom and his dad who were buying a new TV. This exercise together with part 1 and 2 of the test were supported with illustrations to visualize the continuation of the story, the characters and the differences.

The examiner first explained in Dutch the four speaking exercises. The examiner continuously emphasised that the participant did not have to worry about committing errors. Next, the examiner asked if the participant understood what was expected from him or her. When the participant did not confirm, the examiner explained another time the purpose of the exercise. Moreover, the examiner underlined that if the pupil did not know the answer or did not understand what was asked, the pupil had to shake his head or express that he did not know the answer or understand the question. However, the examiner attempted to encourage the pupils to express only some English words or explain in Dutch what was meant. Through the composed protocol, all participants received the same instructions and questions.

Image 4 shows part 3 of the standardized Cambridge speaking test, where the individual had to complete the story of Tom and his dad who were buying a new television. With the help of five pictures, the participant was asked to explain how the story developed. The entire Cambridge speaking test can be found in Appendix D.

(25)

Image 4: The storytelling exercise in part 3 of the standardized Cambridge speaking test

3.3.5. Questionnaires

To gain insights into the language background and media use of the participants, the pupils but also their parents filled in a questionnaire. Both questionnaires displayed two matrixes with questions about the amount of time a child spent on a certain form of media in English and the home language, which was mostly Dutch. The forms of media use were watching television, listening to music, reading literature, playing games and surfing on social media and YouTube. These questions were answered through a Likert-scale. Other questions tapped speaking in English, the mother tongue of the family and background information of the parents. These questions were open or had to be answered with a yes or no. Since the participants’ questionnaire was developed for older children, the researcher read a Dutch protocol aloud to guide the pupils through the questions. This protocol can be found in Appendix G, whereas the questionnaires for the parents and children can be consulted in Appendices H and I.

3.4. Procedures

(26)

study. At the end of the letter, the researcher provided her email address and phone number to parents with questions or objections. This letter can be consulted in Appendix J.

Per test, it was determined how every exercise was scored, so what was seen as a correct and incorrect answer, and how many points every question received. In the following paragraphs, the used procedures per test are described in detail. Moreover, it is explained which statistical analyses were used to provide answers to the four hypotheses.

3.4.1. Vocabulary test

The first collective test in the testing week was the vocabulary test. On Wednesday morning, the children in one of the year 6 classes completed the test. After the morning break, the other class performed the same test. This order was persistent throughout the testing week: first class 6a was tested followed by class 6b. The vocabulary test was a 120-items PPVT, which was administered through a plenary session. Every correct answer was awarded a point whereas questions, where no answers or incorrect responses were counted as wrong (0), resulting in a maximum score of 120 points.

The listening test was the second plenary assignment that was carried out on Thursday in the testing week. On Thursday morning, class 6a completed the test whereas after the lunch break class 6b did this test. The standardized Cambridge listening test included five parts with a maximum score of 5 each, meaning that the maximum score for the test, as a whole, was 25 points. The scoring of part 1, 3 and 4 of the listening test was straightforward since questions in these exercises were closed: only one letter or name was correct. The scoring of the other two parts took place on the basis of an answer sheet (see Appendix F).

(27)

3.4.3. Reading test

The last plenary language test was completed on Friday morning and Friday mid-afternoon: the standardized Cambridge reading test. The test consisted of seven parts with in total 50 questions, so 50 points was the maximum score. Exercise 1 and 6 were worth ten points, exercise 2 and 5 seven points, exercise 3 and 7 five points and six points was the value of exercise 4. The researcher corrected the test with an answer sheet provided by the Cambridge institute. This answer sheet can be consulted in Appendix F. There was only one adjustment to the answer sheet: the answer to question 1 of part 5 was ‘lake’ whereas some children answered ‘big, blue lake’ as explained in the text. Both options were seen as a correct answer to the question.

For the speaking test, conversations between the examiner and pupils (always one-on-one) were recorded outside the classroom in the hallway. Since this interview was in front of two classrooms and the sport hall of the school, there was a great deal of background noise, which hindered the clarity of the conversations on the audio files. On Tuesday of the testing week, three children from both classes were interviewed whereas on Wednesday twelve pupils from only class 6b were questioned. Thursday morning four participants from class 6b were recorded and the rest of the week the rest of class 6a conversed with the examiner. This means that three-quarter of the pupils participated in the individual speaking test: every pupil from class 6a and the half of the children from class 6b. The one-on-one speaking test session took anywhere between 7 and 15 minutes.

The speaking skills of the pupils were measured through a rubric, which was developed by Eyckmans and De Wilde (2016). This rubric was divided in four elements: grammar, pronunciation, vocabulary and interactive communication. Every aspect received a grade from 0 to 5, where 5 was seen as a perfect performance and 0 a score below the A2 level of the Common European Framework of Reference. There was space left to provide comments on every item and to mention what the pupil did right and wrong in the interview. The grades of the four elements together represented the overall score for a child on speaking which had a maximum of 20 points. The rubric used to determine the speaking abilities of the participants can be consulted in Appendix E.

3.4.5. Questionnaires

(28)

time spent on a media form received a 0 whereas the highest amount received a 5. Most open questions served as background information to be consulted when significant outcomes occurred. 3.4.6. Statistical analyses

(29)

4. Results

4.1 Results per language test

Table 1 shows the results of the four language tests: the vocabulary, listening, reading and speaking test.

Table 1: Frequencies of the four language tests

Test N Mean Std. deviation Minimum Maximum

Vocabulary (max=120) 59 77.15 13.16 51 115

Listening (max=25) 51 13.20 5.27 5 24

Reading (max= 50) 53 18.32 10.58 2 44

Speaking (max=20) 45 8.67 5.61 1 20

As can be seen in Table 1, there is a large range of 64 and a standard deviation of 13.16 for vocabulary, which are both - compared to the other language tests - high.

In the listening test, 51 pupils participated. This number was lower because on the Wednesday that this test was administered, a number of gifted children were absent as they took part in special educational activities. As can be seen in Table 1, the mean score on the listening test appears to lie around chance level, with scores not yet reaching half of the maximum score. But here too the range of scores was substantial.

53 pupils completed the reading test, since seven children of class 6b had to follow music lessons at that time. The mean of 18.32 was less than half of the 50 points that could be earned for this test. The high standard deviation for the reading test indicates a large variation between the pupils’ performances, which is in line with the findings on the listening test.

Since the language tests had to be finished in one week and time was sparse through compulsory lessons, three-fourth of the participants sat with an examiner individually to be tested on their spoken English proficiency; all pupils of class 6a and 15 pupils of class 6b joined. The range for the overall speaking part is almost the same as the maximum score for this test, which is 20, and the average is a bit below half of the maximum score of 20.

(30)

4.2. Pupils’ background information

In the questionnaire, the pupils indicated - among other things - their gender, age, and class (6a or 6b). These variables can play a modulating role in language proficiency scores and so between-subject comparisons (independent samples t-tests) were carried out on these variables to see if there were no a priori differences between the pupils. First, to find out if gender played a role in language proficiency, the scores on the four language skills were compared between boys and girls. Table 2 shows the division between girls and boys per test, their mean scores and standard deviations. Despite the fact that there were differences in mean and standard deviations (with boys mostly outperforming girls except on the reading test), these differences between boys and girls at no point reached significance levels on any of the language tests. This indicates that none of the genders outperformed the other.

Table 2: Test scores between boys and girls

Test Gender N Mean Std. deviation Min Max

Vocabulary (max = 120) Girl 26 74.65 11.97 51 104

Boy 33 79.12 13.89 52 115

Listening (max= 25) Girl 22 12.77 5.80 5 24

Boy 29 13.52 4.92 7 24

Reading (max= 50) Girl 22 19.82 9.65 5 43

Boy 31 17.26 11.23 2 44

Speaking (max= 20) Girl 21 8.38 5.21 2 18

Boy 24 8.92 6.04 1 20

Since all participants at the time of testing were in their 6th_{year of primary school, there was not} much variation in the age between the pupils; 62% of the children were 9 years old and 35% was 10. The other 4% can be divided over one girl with the age of 11 and one boy who was eight years old. Through the homogeneity of the participants’ age, it is not surprising that the test scores show no significant relationships on this demographic variable, assessed by means of several correlation analyses between the test scores and the variable age.

(31)

Table 3: Test scores for class 6a and 6b

Test Class N Mean Std. deviation Min Max

Vocabulary (max = 120) 6a 30 77.07 11.84 52 104 6b 29 77.24 14.61 51 115 Listening (max = 25) 6a 25 13.52 4.55 5 24 6b 26 12.88 5.96 5 24 Reading (max = 50) 6a 30 20.33 9.56 6 43 6b 23 15.70 11.48 2 44 Speaking (max = 20) 6a 30 9.33 5.54 2 20 6b 15 7.33 5.72 1 20

Other interesting findings that resulted from the pupils’ questionnaires provided more insight on the pupils’ use of English. More than half of the children, thirty-two, indicated that they used the English language often, but only one-quarter mentioned that they were actively looking for situations in which they could use the L2. In addition, around three-quarters of the children reported to come into regular contact with English. On the other hand, 61% of the parents indicated on the parental questionnaires that their children did not have much contact with the English language. This discrepancy is probably caused by the English influence in games and videos on the Internet were the parents are not aware of, since parents do not always know what their children do online. Moreover, two parents reported that their child followed English lessons, for example through an online program, and all fifteen parents who filled in the second page of the questionnaire indicated only to use Dutch at home. Lastly, almost all child participants answered positively to the question whether they liked English as a language; only two pupils showed a negative attitude towards this L2.

4.3. Relationships between the four language tests

(32)

Table 4: Outcomes correlation analyses between the four language tests

Vocab Listen Read SpVoc SpGram SpPron SpInt SpOver

Vocab r 1 .77 .66 .74 .70 .68 .77 .80 Sig. .000 .000 .000 .000 .000 .000 .000 .000 Listen r .77 1 .57 .75 .68 .76 .76 .82 Sig. .000 .000 .000 .000 .000 .000 .000 .000 Read r .66 .57 1 .62 .71 .66 .70 .75 Sig. .000 .000 .000 .000 .000 .000 .000 .000 SpVoc r .74 .75 .62 1 .84 .72 .78 .94 Sig. .000 .000 .000 .000 .000 .000 .000 .000 SpGram r .70 .68 .71 .84 1 .66 .74 .91 Sig. .000 .000 .000 .000 .000 .000 .000 .000 SpPron r .68 .76 .66 .72 .66 1 .61 .82 Sig. .000 .000 .000 .000 .000 .000 .000 .000 SpInt r .77 .76 .70 .78 .74 .61 1 .89 Sig. .000 .000 .000 .000 .000 .000 .000 .000 SpOver r .80 .82 .75 .94 .91 .82 .89 1 Sig. .000 .000 .000 .000 .000 .000 .000 .000

The numbers in Table 4 are highly significant which implies strong relationships between all tests and even between certain parts of tests. In other words, a pupil whose English L2 proficiency level is high performed equally well on every other language test. The same is true for pupils with a lower English L2 proficiency level.

(33)

girl with a Dutch-speaking father and a mother from Portugal with whom she regularly talked in Portuguese. She stated that she cannot speak English, does not often have contact with the L2 and is not looking for situations to use the language. However, she browses the Internet especially YouTube and Snapchat.

Image 5: Stem-and-leaf-plots of the four language tests including outliers

It is also interesting to see who scored markedly lower on one of the language tests. These children can possibly change their media habits to become more proficient in English. The pupils with low overall scores were girls as well as boys who indicated to have almost no contact with the English language through media or with speakers of the L2. Remarkably, they all reported to frequently listen to English music. On the other hand, their media exposure in the home language was quite substantial, 30 minutes up to one hour per media form.

(34)

the lowest score on the speaking test. Participant 44, with the lowest reading score, already handed in his reading test after five minutes, but performed well on the other language tests and had more regular contact with English. Overall, all these pupils did not look for situations to use English or speak the language with others.

Among the 60 participants, there was one boy with dyslexia. By viewing his results, it became apparent that he performed above average on every language test except for the reading test. This indicates that children with learning disabilities do not per se have to have difficulties with every aspect in L2 learning.

4.4. Relationships between language skills and other factors

By performing a correlation analysis between the language tests and the answers to the questionnaires, which were about media use, language background and contact with the English language, relationships between the language skills and other factors can be uncovered. In the following sections, the most interesting findings per language test are reported, which were all derived from a two-tailed bivariate correlation analysis, using the Pearson r statistic. The entire correlational matrices are provided in Appendix L.

4.4.1. Vocabulary test

The correlation analysis showed, besides relationships between the language tests, several positive and negative correlations between the vocabulary test and other factors. A negative influence on the English L2 vocabulary knowledge was found to come out of the amount of time children watch Dutch television, since there was a strong negative correlation between the vocabulary scores and watching Dutch television with subtitles, r = -.36; p < 0.01, and the variable watching television in the home language, r = -.28; p < 0.05, which was mainly Dutch. In other words, those children who reported watching many Dutch television programs overall scored less well on the English proficiency measures across the different domains. On the other hand, there was a strong positive relationship between speaking English and vocabulary knowledge, r = .48; p < 0.01. This variable in relation to speaking was gathered through a question in the pupils’ questionnaire about how often a day they spoke English.

(35)

language showed a positive correlation with the listening component of the English language test, r = .34; p < 0.05 and, similar to what was attested in the vocabulary test, a strong positive relationship was found between listening and speaking in English, r = .53; p < 0.01. Again, this variable was an indication by the pupils about how often they spoke the language. Remarkable was a negative correlation between listening and how often a pupil used the English language, r = -.37; p < 0.05. This indicates that – on the basis of these results - a frequent user of English as an L2 performed more poorly on a listening test than a peer who did not indicate to use English frequently outside of school.

4.4.3. Reading test

The correlation analysis showed links between the reading test scores and two other measured features, namely speaking in English and exploring YouTube or other social media in the home language. The former variable was retrieved from the pupils’ questionnaire, whereas the latter was based on the information gathered from the parental questionnaire. Both factors showed a strong positive relationship to reading, respectively r = .60; p < 0.01 for speaking English and r = .64; p < 0.01 for using social media in the home language.

(36)

Table 5: Correlations between speaking parts and factors

Speaking part Factor r p

Vocabulary Speaking English .42 < .01

Grammar Watching English TV without subtitles .30 < .05

Listening to English music .30 < .05

Speaking English .49 < .01

Playing games in the home language .31 < .05

Language mother .33 < .05

Language father .30 < .05

Language brothers and sisters .33 < .05

Listening to English music (PQ) .73 < .01

Education mother (PQ) -.10 < .05

Pronunciation Watching English TV without subtitles .32 < .05 Watching English TV with English subtitles .32 < .05

Reading literature in the home language .30 < .05 Using YouTube in the home language .34 < .05

Interactive communication

Using YouTube in the home language .34 < .05

Watching English TV without subtitles .66 < .01 Reading literature in the home language (PQ) .61 < .05

Overall Watching English TV without subtitles .31 < .05

Using YouTube in the home language .33 < .05

Language brothers and sisters .32 < .05

In Table 5 it can be seen that factors that frequently return as influential aspects on one of the speaking parts and the overall speaking proficiency are the languages of the parents and brothers and sisters, wherein the language of the mother appears to be the most important since in four of the five cases there was a significant relationship between the mother’s language and a speaking element. Watching English television without subtitles also positively influenced an individual’s performance on speaking with the exception of vocabulary. In addition, the overall speaking score revealed a positive relationship between listening to English music r = .59; p < 0.05 and using YouTube in the home language, which was mainly Dutch, r = .33; p < 0.05. There was, furthermore, a strong positive correlation between how often someone spoke English and the overall performance on English speaking, r = .48; p < 0.01.

(37)

proved beneficial to read literature in the home language since this shows positive correlations between one another, respectively r = .30; p < 0.05 for pronunciation and r = .61; p < 0.05 for interactive communication. On the other hand, a negative relationship was found between the education of the mother and the grammatical aspect in speaking, r = -.10; p < 0.05. This means that a child whose mother was lowly-educated tended to score higher on English L2 speaking. However, this relationship was only based on one mother who was lowly-educated, so this outcome is most likely a confound.

4.5. Differences between pupils with different language backgrounds

To uncover if there were differences in test scores between pupils with different language backgrounds, two One-Way ANOVAs were carried out: one for each parent’s language background. By inserting the data in SPSS, languages were divided in three groups: Dutch, English and others. Before performing the One-Way ANOVAs, it was checked if the Levene’s test for equality was not significant for any language test. The Levene’s test yielded several significant outcomes and a non-parametric Welch’s F test was performed since this test derived to be robust despite the fact that homogeneity of variance has been violated. This test did not show any significant effects, which indicated that no differences in test scores between the pupils’ mothers with different language backgrounds were found.

(38)

Table 6: Descriptives for the language of the father

Test Language N Mean Std. deviation Min Max

Vocabulary (max= 120) Dutch 51 76.43 12.89 51 115

English 3 78.33 13.32 67 93

Other 3 83.67 15.57 69 100

Listening (max = 25) Dutch 43 13.02 12.89 5 24

English 3 16.33 5.51 11 22

Other 2 14.00 5.66 10 18

Reading (max = 50) Dutch 45 17.16 10.06 2 44

English 3 21.67 13.58 9 36

Other 2 28.50 14.85 18 39

Speaking (max = 20) Dutch 39 7.92 5.20 1 20

English 2 6.50 2.83 5 8

(39)

5. Discussion 5.1. Language tests

A correlation analysis showed that the four language tests were positively related to one another, indicating that a pupil’s level of English L2 proficiency stays consistent throughout the tests that each tap different skills (listening, reading, writing, speaking). In other words, if a child performed well on reading, he or she also did well on the three other language tests. Having said this, the results also clearly revealed a substantial range in scores on all the four subtests, which implied a wide variety of English L2 proficiency levels in the two year 6 classes. This outcome is consistent with the findings from the study by Eyckmans and De Wilde (2016), which is considering the differences in media exposure, language backgrounds and contact with the English language among the 60 pupils, not surprising. Moreover, the vast majority of the pupils scored above chance level on the vocabulary test; the mean was even measured at 77.15. This positive finding is probably due to the fact that some simple words and English words that sound similar to words in the Dutch language were asked. Moreover, with a little bit of exposure to the English language through media or other forms, Dutch pupils already acquired a substantial number of words in this L2, which can be caused by the English influence in Dutch television. On the other hand, the pupils scored below chance level on the speaking and reading/writing test, which were tests to measure the productive knowledge. It is understandable that pupils received low scores on these language tests because Dutch 9-and-10-year old children are generally more trained on their listening skills through English music and television and not on their productive knowledge skills. These skills are learned during formal instruction, which did not yet start for these children. Actually, it is astonishing that many children already know a large number of words and can express some sentences in the English language without having any instruction yet. The scores on the reading/writing test are not surprising, since the low grades fit the expectation of uninstructed children who normally do not know much about English yet.

5.2. Pupils’ background information

(40)

English L2 vocabulary knowledge. However, in the current study the boys overall did slightly better than the girls though at no point did this difference become significant. Therefore, hypothesis 1, there are no differences between proficiency scores of boys and girls, can be accepted. This finding on gender that was revealed in this study is consistent with the one found in the study by Eyckmans and De Wilde (2016) who also did not find a difference in performance as a function of gender. A possible cause for no significant differences on gender is probably due to the fact that the boys in the current study were found to have more parents with another L1 than Dutch compared to the girls and two other boys had lived abroad where they acquired English. On the other hand, it is often claimed that girls, in general, are better in learning languages (cf. Heinzmann, 2009), which can be an explanation for no differences in gender between the language test scores. The two boys who acquired English abroad and the pupils with parents who spoke another language than Dutch were almost equally divided among class 6a and 6b, which was probably the reason why no differences between these two groups were found. The variable age did not show significant relationships either, which was most likely due to the homogeneity of the group participants.

5.3. Outliers

The characteristics of the children who performed extremely well were further investigated to see which variables positively influence the level of English L2 proficiency. Two of these children received a large amount of English input through their English-speaking mother and babysitter or through the country where one of the children had lived for some years, which in both cases can be described as incidental learning (Gass, 1999; Hulstijn, 2005; Hulstijn, 2013; Vidal, 2011). Through their long-time exposure and input to the English language, it is not remarkable that these two children scored extremely high. The third child was unique since she had already developed two language systems: Dutch and Portuguese. A reason why she scored extremely high can be due to the fact that she already possessed two language systems by which the acquisition of a third language developed in a short timespan since L3 learners have been claimed to possess more general knowledge about languages and know how languages operate in the brain (Broekhof, 2007). Nonetheless, the Onderwijsraad (2008) stated that the possession of several language systems could cause an increase of interferences, which was not true for this case. The acquisition process of this girl was likely sped up through exposure to the English language on YouTube and other social media without any negative consequences.

(41)

contact with English through media or with L2 speakers. The absence of English input through speaking and media probably caused the low scores of these children. This means that various forms of English input are key in the acquisition of this second language.

5.4. Modulating factors in predicting foreign language success 5.4.1. Media

(42)

which the test was administered, especially for class 6b: The reading test was administered on Friday afternoon when the sun was shining outside and the teacher promised the children to play outside when the test was over. Therefore, this could have caused children to not fully focus on the test anymore. Moreover, the participants had to focus for a long time because the reading test consisted of seven parts, altogether taking about 40 minutes to complete. Older children, for whom this test was developed, are more used to be concentrated for a longer period of time, but this could not be expected from 9-and-10-year old children. In consequence, a large group of participants could not finish the reading test in the set time of 40 minutes. Moreover, the pupils were also not satisfied with the speed of the PPVT presentation, which was, according to them, too fast. Since these language tests were developed for 12-year old non-native children, it is even more ingenious that these pupils are already able to answer some of the questions, because the tests are highly above their current level of English. Moreover, these pupils did not receive any formal L2 instruction yet. Therefore, it is phenomenal that these 60 pupils can already score some points on the tests by which a few even outperform the level for 12-year old children.

Despite the fact that many studies (Berns, De Bot & Hasebrink, 2007; Bonnet, 2004;

Eyckmans & De Wilde, 2016; Kuppens, 2010; Marsh & Tainio, 2009; Miller & Hegelheimer, 2006; Naber & Lowie, 2012; Ranalli, 2008; Sundqvist, 2009) discovered the advantages of different media forms like television, gaming, movies and popular music on vocabulary knowledge, none of these forms showed a positive influence on the vocabulary test scores in this study. This outcome is probably due to the fact that children who scored high on this test, which was a large number of pupils, were all exposed to English media a different amount of time. In other words, there were children who did not receive much English input and children who did, whereas both scored high on the vocabulary test. This can be caused by the large influence that English has on the Dutch everyday language through which acquisition of L2 vocabulary takes place more easily. On the other hand, only negative effects were found on the media forms of watching Dutch television with subtitles and watching television in the home language on the vocabulary scores. Since the home language was predominantly Dutch, this finding seems logical because watching television in one language does not improve the abilities in another language.