Three initial research questions guided the present study:

  1. 1)

    Which forms and frequencies of media-related extramural English contacts do German and Swiss adolescents have?

  2. 2)

    How do socio-economic background and gender influence the pattern and frequency of media-related extramural English contacts?

  3. 3)

    How do media-related extramural English contacts influence students’ English competences?

This chapter summarizes and discusses the central results of the present study and embeds them into the theoretical framework. The discussion for each research question will be presented in a separate subchapter. In addition, further limitations of the present study, recommendations for future research, and implications for the educational setting will be discussed.

7.1 Summary and Discussion of Forms and Frequency of Extramural English Contact

While research has shown a regular and intensive contact with English for the population of several smaller European countries, e.g., Sweden and the Netherlands, empirical data for Germany and Switzerland had been limited until now. The present study was designed to help close the research gap by providing reliable empirical evidence. Given the dominance of the American entertainment industry and the development of entertainment technology in recent decades, it was proposed that adolescents in both countries will have increasingly frequent media-related English contacts outside of school, so-called extramural contacts (H1).

In line with findings from national media studies (MPFS, 2017; Waller et al., 2016), the present study could show that in 2016/2017 almost all students in the sample had access to the internet and owned a smartphone and laptop, which allows easy and regular access to a variety of online content. A majority also had access to a gaming console at home.

Furthermore, the results showed that English-language media content has become a constant presence in the lives of most adolescents in Germany and Switzerland. Even though both countries have a rich pool of German-language media content, adolescents increasingly use new online-based solutions to integrate international media content into their everyday life. When doing so, they tend to favor the original versions over dubbed or translated versions.

Music has shown to be the most frequented form of extramural English contact. In addition, most students surf up to two hours per day on English-language websites and visit social media sites. They also regularly visit video-sharing platforms to watch videos in English. Most students also watch movies and TV series in English regularly. Reading books or newspapers and magazines in English was less popular, yet still, almost 40% in the sample stated that they read English books at least a few times per month. Playing computer games in English is the least frequented activity, yet students who do engage in these computer games spend a considerable time gaming in each session and per week.

While the results support the research hypothesis, there is also a wide variation within each media category, as illustrated by the standard deviations. Thus, while most students have regular contact with English via the media, others only engage in occasional contact or no contact at all. In addition, not all students prefer the same media channels to the same degree.

The data showed small to medium differences between Switzerland and Germany, with Swiss students being more active in listening to audiobooks in English, watching online videos, movies, and TV series in English, and reading books and newspapers in English. By contrast, German students are more active in gaming, and they spend longer hours gaming per sitting and per week. They also surf longer hours per day and read and write longer content online. However, there was no significant difference between students from Germany and Switzerland for the overall amount of extramural contact. It seems that students in both countries might differ in their preference for specific media categories when it comes to English-language content, but not in their overall frequency of media-related extramural English contact. In addition, the effect sizes for the found differences were small.

These results also hold after controlling for socio-economic background factors. Thus, differences in socio-economic background between the subsamples cannot completely explain the differences between the two countries. Therefore, these differences are most likely caused by other factors, such as media traditions or an overall attitude towards English media content. Unfortunately, the present study could not analyze this question in more detail.

Overall, the results found in this study are in line with empirical findings from other European countries (e.g., Kuppens, 2010; Olsson, 2011, 2016; Olsson & Sylvén, 2015; Sundqvist, 2008, 2009a, 2009b, 2011, 2012, 2013; Sundqvist & Sylvén, 2012a, 2012b, 2014; Sylvén, 2006, 2007, 2019; Sylvén & Sundqvist, 2012a, 2012b, 2015, 2017; Toffoli & Sockett, 2010, 2012). The high frequency of extramural English contact in the sample indicates that learners in upper secondary education have reached a level of language proficiency, on which they can comfortably process even more complex authentic language input and are able to interact with the language in a self-regulating way.

However, most students use media channels passively, i.e., they read, listen, and watch English media content. Only a minority of students engaged in active behavior through posting and uploading content themselves. The most common form of active involvement is posting short comments and posts on social media sites or communicating in English. Thus, it seems that learners’ extramural English contacts are characterized mainly by a steady intake of English input rather than a production of output or interaction with others. However, as Toffoli and Sockett (2010) showed, this might also be subject to change over time, as students might grow more comfortable and start to engage in active output production, the longer they engage in extramural English contacts.

When asked about the reasons why they engaged in extramural English contact, students stated that they are fond of English, enjoy the authentic content, and appreciate the original and undubbed versions of movies, TV series, books, and games.

The translation of a book or the dubbing of movies and TV series was also often seen as a distortion of the original work. Statements like these are similar to the sentiment prevalent in Scandinavia and the Benelux states (Berns et al., 2007). The answers also reveal that the dominance of the English language in the entertainment industry and online is an important motivator for students to engage in out-of-school contact. English content is often easier to find, contains more information, and is of a higher quality. For example, similar to results from Toffoli and Sockett (2012), some students in the present study started watching the original version because the translated version was not available yet. Once they had switched, they stayed with the English version and did not want to switch back.

Many students were also aware and convinced of the language-learning opportunities provided by this extramural contact. Some students also stressed the importance of improving and practicing their language skills in a natural setting to prepare them for the future.

However, this awareness does not mean that the possible learning processes are not incidental. Incidental language learning was defined as an unintentional or unplanned process resulting as a by-product of another activity (see section 4.1). It is unstructured learning resulting from daily life activities. Such a by-product can result from implicit learning processes but might also be accompanied by explicit learning processes, during which a person pays at least peripheral attention to certain language forms and engages in hypothesis forming and testing. Therefore, incidental language learning is not marked by any lack of explicit process or conscious knowledge of learning opportunities. Instead, the important distinction is that in the case of incidental learning, the focus of the activity is on understanding and decoding the message from the input and, in the case of interactive activities, on interaction and communication (Elley, 1997; N. C. Ellis, 1994; R. Ellis, 2008; Hulstijn, 2001; Kekra, 2000; Rieder, 2003).

The fast pace of authentic media content, such as movies or games, makes it unlikely that students have time to focus on much more than the content in the moment. Written input and written interaction might provide students with more time to employ certain learning strategies and explicit learning processes. Nevertheless, students’ statements in the survey underlined the voluntary nature of their activities and the strong focus on entertainment rather than active language learning, even though they often were aware of possible learning gains. This is in line with the definition of incidental and informal learning as applied in this study.

When interpreting the results, it should be noted that the generalizability of the findings is limited by the selected sample and the study design.

First, the high level of media-related extramural English contact found in this study is most likely due to the older age group of the sample. The upper secondary students in the sample could have been expected to have reached a competence level in which input from more complex authentic media content can be processed and understood. The higher language competence in the MEWS sample was demonstrated by Keller et al. (2020). The paper shows that the majority of students were found to have already reached the B2 level of the Common European Framework of Reference for Languages (CEFR) in writing at T1, and 15.9% of students reached C1 at T2 (see Keller et al., 2020 for an in-depth analysis of the two writing tasks and the overall writing performance). Köller et al. (2019) also discussed the high language competences at T1 for the German sample in all three language skills. Thus, the older sample probably also allowed the observation of a high level of engagement in demanding authentic material, such as books and movies. As such, the findings of the present study are most likely not representative for all age groups in Germany and Switzerland.

Studies have shown that for beginners, authentic media input is often too overwhelming (d’Ydewalle, 2002; d’Ydewalle & van de Poel, 1999; Huckin & Coady, 1999; Neuman & Koskinen, 1992; Peters, 2018; Rankin et al., 2006; Sylvén & Sundqvist, 2015; Vidal, 2011). It is thus reasonable to assume that a younger sample would have shown much lower levels of media-related extramural English contacts, as not all authentic input might be suitable or comprehensible for them.

Analysis of why students chose to engage in extramural English contact further supports this. Many students have stated that they have no trouble understanding and following everything they read, hear, and watch in English. By contrast, the small number of students who reported that they do not engage in extramural contacts all stated a lack of sufficient language competences as a reason, as they do not understand enough to make it an enjoyable activity.

However, results from Sylvén and Sundqvist (2015) indicate that even children as young as 11 or 12 might already be able to engage in extramural contact with authentic media content. Future research should strive to close this research gap for the German and Swiss populations.

Second, the present sample is also highly selective regarding socio-economic background, as entrance into the Gymnasium is already dependent on family background in both countries. As Section 6.2 has shown, this leads to a smaller variance within the sample regarding technical equipment, an above-average level of educational and cultural capital, and a conducive home environment. Consequently, the results are most likely not representative of all students in Germany and Switzerland. Future research should include a broader range of school tracks and socio-economic backgrounds into their sample to capture their unique media habits (for more details on the possible influence of socio-economic background, see the next subchapter).

Last, it is important to mention that the present study can only capture a fraction of an extremely complex leisure time activity. Given the lack of empirical data for Germany and Switzerland, the study aimed to provide a first overview of media categories and preferred English-language content for both countries. For that reason, the questionnaire included ten media categories designed to capture the most popular media activities: music, radio/ podcasts, books, newspapers/ magazines, movies/ TV series, TV shows, online videos, surfing, and gaming. However, even though the study included follow-up questions for eight of the most relevant categories, the results have probably not captured every nuance of this complex behavior. In addition, students were asked to rate their overall average extramural media exposure. This averaged measure may have been challenging for some students.

Media diaries, as they were used in Sundqvist (2009a) and Olsson and Sylvén (2015), would have been more suitable to gather in-depth information about students’ everyday media activities. This would have also allowed an investigation of the amount of passive language input and (inter)active language production in more detail. Unfortunately, this strategy was not possible for the present study, as students were only tested at two fixed measurement points. As a result, the present study has only scratched the surface, and each category could and has warranted entire studies dedicated to them.

Furthermore, due to logistic limitations, measures for media-related extramural contact could only be implemented at the second measuring point. This makes it impossible to discuss details about students’ development between the two measurement points. Future research should aim to implement longitudinal study designs to track students’ development over time.

7.2 Summary and Discussion of Extramural English Contacts and the Digital Divide

The second research question asked how gender and socio-economic background, as two important social factors, influence the pattern and frequency of media-related extramural English contacts among German-speaking adolescents.

Drawing on findings from the field of gender and women studies and the theoretical framework of West and Zimmerman (1987) and Bourdieu (1987, 2005), gender in the present study was understood as a social, cultural, and discursive category that does not exist a priori and outside of people. Instead, it is rooted in social interaction and is something people do rather than something they are. Through socialization, existing gender structures are inscribed into peoples’ perception, behavior, and evaluation. As a result, forms of doing gender are deeply inscribed in all areas of behavior and everyday life (Bourdieu, 1987, 2005; Drüeke, 2016; Klaus, 2002; Röser & Wischermann, 2004; Schneider, 2002; West & Zimmerman, 1987).

As the media landscape is part of a given society, it does not stand outside of this system. As a result, media production is highly influenced by the existing gender norms. The resulting media content therefore actively participates in producing and reproducing gender definitions (what is considered feminine and masculine) and gender positions (roles, tasks, activities, and opportunities ascribed to gender categories) in an existing power structure (Dorer, 2002; Drüeke, 2016; Klaus, 2002; Peil et al., 2020; Philipp, 2011; Philipp & Garbe, 2007; Röser & Wischermann, 2004; Schneider, 2002). This is evident in the fact that women are often still shown in traditional life contexts and are underrepresented as experts in science and technology in the media (Götz & Prommer, 2020; Peil et al., 2020). Newer media channels did not overcome these stereotypical gender representations (Götz & Prommer, 2020; Prommer et al., 2019). Furthermore, certain media channels and media content are often defined as specific male or female domains, as is especially evident in terms of the internet and the computer as typical male domains (Dorer, 2002; Drüeke, 2016; Klaus, 2002; Peil et al., 2020; Philipp, 2011; Philipp & Garbe, 2007; Röser & Wischermann, 2004; Schneider, 2002).

However, the media does not simply impose gender norms on passive consumers. Instead, people actively use media channels and content to claim and reproduce a specific gender identity. Members of different gender groups claim different media categories and different content as their field of expertise and leisure-time activity and, by doing so, reproduce their membership in said gender group (Drüeke, 2016; Klaus, 2002; Röser & Wischermann, 2004; Straub, 2006). The media is thus not only a gendered field but also a gendering field (Straub, 2006).

Following the theoretical framework, it can be assumed that male and female students in the present study would thus differ in their frequency and preferences for certain media channels and specific media content as they serve as means to reproduce gender (H3.1 & H3.2). Looking at the results, the present study found clear signs for gender-stereotypical behavior among German and Swiss adolescents. Similar to findings from studies investigating German-language media use, female students were shown to be more likely to choose communicative or literacy-based activities and English-language media content. They not only read more English books but also spent more time online on websites providing longer written input, such as fan fiction and social media websites.

However, male students indicated that they read more magazines and newspapers in English. At first, this might seem to contradict the research hypothesis, as studies have repeatedly found female adolescents to be more frequent readers. However, studies have also found that this is not true for all forms of written content. While female adolescents and children can often be shown to have a higher average for reading books and magazines, studies found a higher overall average for male adolescents and children in terms of comics and online newspaper articles (Philipp, 2011; Philipp & Garbe, 2007).

In addition, the present study did not evaluate which newspapers and magazines were read. It is therefore difficult to say what kind of content students choose. Male students might read newspapers and magazines with a sports and technology focus, which would once again be in line with previously identified stereotypical media patterns. These results call for further investigation and underscore the fact that media channels can deliver a wide array of content and topics. Consequently, media habits may not always appear as straightforward as expected. This is also in line with other findings, which show that especially new online media content covers a wide range of topics within one media channel (Philipp, 2011; Philipp & Garbe, 2007). It thus seems not only important to ask which media channels students frequently use but also which content and topics they engage with.

Male students in the present study also indicated that they watch more online videos and were more active gamers. When engaged in gaming, more male students indicated an interest in violent or sports-based gaming categories than female students. In contrast, female students more often named simulation games or quizzes as their preferred gaming genres. Effect sizes for gaming variables were particularly strong.

In line with research hypothesis H3.1, the study found male students to have a higher overall frequency of extramural English contact. Similar to the findings from Sundqvist (2009a), this is mostly due to their high involvement in time-consuming online content and computer games.

With respect to socio-economic background, the present study drew on Bourdieu’s theory of social habitus to explain differences in students’ media preferences. According to the theory, social classes differ in terms of economic, social, and cultural capital, resulting in different life circumstances. These differences lead to differences in behavior and (aesthetic) taste for things like art, music, or movies. This so-called social habitus is passed down through unconscious intergenerational transmission (Bourdieu, 1983, 1987, 2001 [1966])). Consequently, taste for media channels, media content, and the affinity towards technology should be understood as a form of learned social behavior. Parents not only influence their children’s media behavior and attitude towards English media content by imposing rules and restrictions but also do so through their own attitudes in terms of aesthetic taste. Biermann (2009, 2013) used the term media habitus for this part of the social habitus that specifically relates to the taste and affinity for media content and technology.

For media contacts in the native language, empirical research has suggested that adolescents from lower socio-economic backgrounds have a higher probability of owning their own television and gaming consoles, spending more time online and with computer games, and spending less time reading books (MPFS, 2017; Waller et al., 2016). Research also suggests that adolescents from higher socio-economic backgrounds are more likely to have grown up in an English conducive home environment due to the high level of parental education, cultural capital, and foreign language habits (Lindgren & Muñoz, 2013; Rolff et al., 2008). As a result, they are more likely to show a higher overall engagement in regular extramural English activities (H2.1), while they would differ from children from lower socio-economic backgrounds in their preference for certain media categories, genres, and content (H2.2).

While the results were somewhat complex, the overall findings supported these hypotheses. Students from higher socio-economic backgrounds were found to have a significantly higher overall frequency of extramural English contact. The contact was mainly due to contact through books, newspapers and magazines, movies, TV series, TV shows, and online content. In contrast, students from lower socio-economic backgrounds were shown to surf on English-language websites for more extended periods of time but had a lower overall frequency of extramural English contact.

For most categories, the two structural factors (parents’ highest educational level and objectified cultural capital) did not show significant effects. However, objectified cultural capital had a significant effect on literacy-based activities. This finding can be explained by the fact that objectified cultural capital was operationalized by the number of books at home. Unsurprisingly, growing up in a household with many books positively influences students’ likelihood of being avid readers themselves.

The strongest and most stable predictor for extramural English contacts was the use of English within the family. Students from homes in which English is used regularly in daily life reported a higher frequency of extramural English contacts for most media categories and a higher involvement in literacy-based activities.

Surprisingly, however, this factor did not show a significant effect on the frequency of extramural gaming activities. Indeed, the data did not show any effect for socio-economic background factors for gaming in the model controlling both structural and process factors. Hours spent gaming per week was also not affected by any of them. These findings are surprising, given that media studies for media use in the native language have indicated that students from higher economic backgrounds tend to engage less in computer gaming (MPFS, 2017; Waller et al., 2016). The same apparently does not hold for extramural gaming contacts in the present study. This might perhaps be due to the low variation in the data, as not many students in the study engaged in extramural gaming activities overall. It might also be because the questionnaire explicitly included smartphone-based games in the entry question and the follow-up questions about gaming hours. As most students have their own smartphone, engaging in mobile games might not be as determined by social background as gaming on a console.

Apart from the frequency of gaming, the follow-up questions for the other media categories supported the initial findings for each media category. Students from a higher and more conducive socio-economic background are engaged more often in online and offline literacy-based activities and use the internet for entertainment as well as for reading and information.

Looking at preferred gaming genres, the expected effect of socio-economic background on gaming was found, as students from higher and more language conducive socio-economic backgrounds are not as likely to engage in violence-based gaming. Instead, they are more likely to engage in literacy and knowledge-based games and less likely to engage in other gaming genres.

It should be kept in mind that since filters were used for all follow-up questions, the results for these questions can only be interpreted within the specific subsample of gamers, surfers, watchers, and readers. Effects for any given factor thus only indicate the significant or non-significant effect of socio-economic background (or gender) within that subsample.

The effects of socio-economic background factors (as well as the effect of gender) are, of course, not absolute. This is also evident in the present study, as effect sizes usually range from small to medium.

One reason might be the overwhelming dominance of English online and in the music industry, making it difficult to avoid extramural contact in everyday life. This seems to hold especially true for music, as almost all students listen to English-language music daily. It seems that music might transcend class and gender boundaries alike.

Another reason is that the influence of any given social factor on human behavior is complex and has to be understood as part of a complex interaction with other factors. Furthermore, the weight of the influence for each factor might differ from field to field. Bourdieu therefore emphasizes that the relationship between socio-economic background (and by extension gender) and any form of social habitus is statistical, not deterministic (Bourdieu, 1987).

In addition, adolescents also undergo important self-socialization processes − which lead to the development of independent interests—and are influenced by their peer network (Henrichwark, 2009; Kommer, 2008; Senkbeil & Wittwer, 2009; Straub, 2006). Still, social factors, such as family background and gender, most likely influence the direction of these self-socialization processes and peer networks, thus making certain developments more likely than others (Kommer, 2008).

Overall, the results support the theory that the class-specific media habitus, to some extent, influences students. By passing down their media habits, attitudes, and practices towards foreign media production, parents unconsciously influence their children’s tastes and preferences for things such as books, movies, or games. As a result, children from higher socio-economic backgrounds with a more English conducive home environment have a higher chance of extramural English contact. These findings are in line with results from other empirical studies (Henrichwark, 2009; MPFS, 2017; Senkbeil & Wittwer, 2009; Stecher, 2005; Waller et al., 2016).

Despite these interesting findings, it is important to discuss some limitations in terms of research design and sampling. As already mentioned in Section 3.1, the complicated nature of the individual dispositions of a given person is not determined by one single factor, not even one as important as social class, and can therefore not be measured by one (Bourdieu, 1987). Instead, the specific individual dispositions result from more than one factor, each of which influences the specific circumstances of life (social class, age, gender, place of residence). However, according to Bourdieu, even a system of factors would not be differentiated enough, for although all these factors have explanatory power in all fields, the weight of the factors varies from field to field. Thus, even within social classes, class fractions will differ. This is also true for predicting the trajectory and development a person will go through in life (Bourdieu, 1987). In order to understand this complex structure for each field, more complex analyses would be necessary.

In addition, quantitative measures for analyzing the effect of socio-economic background cannot fully reveal the entire complex structure of individual socialization conditions. According to Bourdieu, the individual habitus is produced and reproduced in people’s actions, thoughts, and attitudes. Therefore, traces of it can be found in what people do and say or leave out (Bourdieu, 1987). To fully capture these traces, interpretative research methods are needed. Only then is it possible to show the totality of the complex socialization conditions. Such an analysis is not possible with quantitative data (Stecher, 2005). As a result, the present study was only able to approximate these conditions by applying surface characteristics by which persons can be divided into broad social groups. Consequently, the specific way in which parents consciously or unconsciously influence their children’s extramural contacts or to what extent parents execute control over age-appropriate media content cannot be conclusively determined in this study.

A similar argument can be made for the gender category used in the present study, as the utilized dichotomous gender variable can be criticized for adhering to the traditional binary male/female distinction. As was discussed in Section 3.2.3, understanding gender as an interactive and discursive category would, in theory, require the observation and analysis of gender identities in their performative production, in which the produced gender identity is the result, not the starting point of the analysis (Nentwich & Kelan, 2014). This would also allow for a broader representation of gender fluidity and gender representation outside of the heteronormative binary gender system. Unfortunately, such an in-depth analysis was not possible for the present study due to economic reasons. Instead, gender was categorized by a binary gender variable. Therefore, it is important to emphasize that the found differences should not be understood as natural differences between male and female students. Instead, the results must be seen as an effect of the students' deeper mechanisms of gender identification and reproduction.

Despite these limitations and the somewhat contradicting findings for gaming, it can be concluded that overall, students’ extramural English contacts seem to be influenced by a specific media habitus, i.e., class and gender-specific attitudes and behavioral patterns. Being raised in a household with strong extramural media habits and a higher educational background positively influences students’ frequency of out-of-school contact with the English language. In addition, male students seem to be more prone to frequent extramural contacts, especially through online and gaming content.

The found differences in gender and socio-economic background might lead to differences in incidental language learning outside the classroom, both for female students and students from lower socio-economic backgrounds. This might be less of a problem for female students, who usually outperform male students in test achievement and school grades in English as a foreign language (Hartig & Jude, 2008). However, for students from lower socio-economic backgrounds, the results are problematic, as they already have a lower probability of achieving higher test scores and grades in school (Angelone & Ramseier, 2012; Buchmann et al., 2016; Frank & Sliwka, 2016; Hußmann et al., 2017; Klemm, 2016; Konsortium PISA.ch, 2019; Kuhl et al., 2013; OECD, 2016, 2020; Solga & Dombrowski, 2009). According to the results from this study, they might also benefit less from the incidental language learning outside of school.

7.3 Summary and Discussion of Extramural English Contacts and Language Learning

According to Krashen’s input hypothesis, a high amount of comprehensible language input in a foreign language through natural settings will lead to language learning similar to the way children acquire their native language. Input is comprehensible if it is slightly more complex than a person’s current level of competences (i + 1). Under such circumstances, learners can derive enough unknown words from the surrounding context to understand and interpret the language input. Such input can derive from contact through any form of written or spoken content (e.g., books, movies, conversation) and will automatically lead to language acquisition in the language acquisition device of the brain (LAD), given the affective filter is low enough (Krashen, 1982, 1985, 1989). According to this theory, language input through media content will lead to a natural increase in language competences.

Newer online-based media channels might also provide learners with an opportunity to interact and produce output, as well as receive immediate feedback from other, often more advanced, learners or native speakers. This interaction will not only help to make input more comprehensible but also assist learners in attaining higher levels of language proficiency and being able to self-regulate their use of language as a social tool (Dunn & Lantolf, 1998; Lantolf, 2000, 2005, 2011; Swain, 2005; Vygotsky, 1978). In addition to the input hypothesis, the present study therefore drew on Vygotsky’s sociocultural theory and Swain’s output hypothesis to underline the importance of interaction and output production for incidental language learning.

Drawing on this theoretical framework, it was hypothesized that a higher frequency of extramural English contact will be positively related to students’ language competences (H4). Overall, the analysis in the present chapter supported the research hypothesis. Almost all ten media channels revealed a positive correlation to students reading, writing, and listening skills.

The results have shown the effects of extramural English contacts to be similar for students’ listening, reading, and writing skills. This is interesting for two reasons: first, one might have expected the effect for extramural contact for audio(-visual) media content to be particularly strong on students’ listening skills, while written language input could have been suspected to be linked especially closely to students’ reading comprehension. Second, as writing in a foreign language is a demanding productive activity and composing a high-quality argumentative or synthesis essay demands students to be familiar with rules and conventions of the text genre, writing skills might be highly dependent on intentional learning and formal classroom instruction. It could thus have been expected that this test score might show a weaker relationship with the frequency of extramural English contacts. However, this was not the case.

Nevertheless, the findings are somewhat in line with other empirical findings. Olsson (2011) and Sundqvist (2009a) also found a positive relationship between extramural English contacts with the two productive skills. The authors attributed this to the fact that students with frequent extramural contacts gain a more extensive and diverse language register, which allows them to adapt their language better to different text types (Olsson, 2011).

In addition, for the present study, overall text characteristics, such as vocabulary, grammar, collocations, and prepositions, were important criteria for scoring the essays. This was true for both the integrated and the independent writing task (Keller et al., 2020; Rupp et al., 2019). Many of these criteria are also important for reading and listening skills. This can be understood as an indication that declarative linguistic and metacognitive knowledge of a foreign language provides a critical source from which both receptive and productive skills draw (Schoonen, 2019). This is also in line with findings from Laufer and Nation (1995), which provided evidence that vocabulary size and lexical richness are significant determinants for successful written production and holistic scores in foreign language writing.

If this is indeed the case, extramural English contact should have an especially large effect on those underlying macrofeatures of the writing scores, which best capture these overall language criteria. Unfortunately, proprietary reasons provided a detailed analysis of the macrofeatures for the present study. Future studies will hopefully allow a rescoring of the student texts and thus enable a more detailed evaluation of the relationship between extramural English contact and different text criteria.

Structural equation models further supported the research hypothesis, showing that extramural English contacts positively affect language learning for all three language skills. Again, the path coefficients were similar in size for all three skills. However, the models only explained a small portion of the variance for the three skills. Model Fit indices showed an adequate to a good fit for all three models, yet some uncertainty remains, and the results should be interpreted with caution (Geiser, 2010; Schermelleh-Engel et al., 2003). Further methodological limitations will be discussed below.

The effect of extramural contacts on the three language skills stayed significant, even after controlling for language competence at T1, yet path coefficients decreased considerably in size. This indicates that the effect of extramural contacts on actual language gains between the time points is significant but small. The small effect sizes might seem unfortunate, yet it is not surprising for three reasons: First, students had already reached a high level of language competences at the first measurement point. Language gains between the time points were, therefore, small in general. In addition, the two measurement points did not lie that far apart. As discussed above, incidental language learning is a slow process (R. Ellis, 1999; Letchumanan et al., 2015; Sok, 2014; Webb & Rodgers, 2009), huge language gains could thus not have been expected to take place between the two measurement points.

Second, as students probably choose a diverse set of topics and genres in their extramural contacts, it is possible that the standardized test administered to measure students reading, writing, and listening skills did not capture the unique content students were exposed to through their extramural contacts. Media content usually contains a high number of high-frequency words and phrases. However, some students might engage in extramural contact that provides them with a large amount of non-frequent and peripheral words. For example, students engaged in extramural contact through playing fantasy games might have acquired an extensive set of vocabulary related to the specific genre that would not have factored into their test scores. This might explain both the only medium-sized effect in model 1 and the small effects on language gain between measurement points in model 2.

Third, as this was not an experimental study, the exact level of language complexity provided by students’ extramural English contacts cannot be determined. As already discussed in Section 4.1, lack of comprehensible input might lead to fossilization, which will slow down the learning progress or stop it completely (Krashen, 1985). As students’ extramural contact is guided by their personal interests, they might not automatically progress to more complex and demanding media content as they increase their language competences. Instead, they might engage in the same type of media content for a long period of time, even though they might have already reached the level of difficulty provided by the source. Input that is too easy will then not provide learners with new syntax and only with a limited range of vocabulary (Krashen, 1985). Extramural English input through mainstream media at some point might even fail to provide more advanced students with input complex enough to aid further language learning (although this might be less of a risk with the upper secondary students in this study).

In addition, as the results from M. Peterson (2012) show, online communication often relies heavily on abbreviations and shortcuts, which can be easily memorized and used for quick communication. As a result, language learners might develop strategies for successful online communication with memorized phrases without real language learning taking place (Krashen, 1985).

By contrast, input that is too complex will provide learners with input they cannot decipher (Krashen, 1985). As a result, students would be unable to understand enough of the content to derive unknown words from the surrounding context, and they might not even be able to use other-regulation or object-regulation to work through their zone of proximal development. While intermediate and advanced students might already feel comfortable enough to seek extramural English contact, some learners in the sample might not have yet been able to use the media to its full advantage because they have not yet reached the required level of proficiency.

In conclusion, extramural English contacts hold the risk of unsuitable language input and communication that is either too simple or too complex. For future research, it would be valuable to control for or at least measure the quality and difficulty of language input students receive in their free time. It would also be interesting to see which strategies learners might employ to select comprehensible input and communicative settings and how they progress from one level of difficulty to the next.

With respect to the influence of gender and socio-economic background, the results from the structural equation models confirm findings from Chapter 6. Male students showed a significantly higher probability of engaging in media-related extramural contact. As a result, they might benefit more from incidental language learning. This might also be why this study failed to find a significant effect for gender on the three language skills in M0. This would be in line with Sundqvist (2009a), who concluded that the high test performance from the male students in her sample might have been due to their high level of involvement in extramural contacts.

In addition, the present study was able to demonstrate that the students’ socio-economic background influences the frequency with which they engage in extramural English contact. Here again, structural factors were less relevant than process factors. A positive and conducive home environment significantly affected the frequency of extramural contacts, as did does parents’ perceived importance of English as an investment into their children’s future. Interestingly, parents’ own English competences negatively influenced students’ extramural English contacts. As discussed in Section 6.2.1, this is most likely an effect of educational background in general, as studies have revealed a negative effect for educational background on media activities in the native language. Children from higher educational backgrounds have been shown to spend less time watching TV, surfing the internet, and gaming but more time reading than children from lower educational backgrounds (MPFS, 2017; Waller et al., 2016).

All in all, the results for the socio-economic background indicate that by including English in everyday life activities and placing value on language proficiency, parents directly influence their children in their choices to engage with English media content voluntarily.

Despite these encouraging findings, the results should be interpreted with caution, as the present study design and resulting dataset has some limitations. First, the hierarchical structure of the data posed a challenge for the analysis of the structural equation models. As data was gathered in Germany and Switzerland across different schools, students in the dataset are nested in classes, which are nested in schools, which are nested in countries. In the Swiss dataset, schools were also nested in different cantons. Since entrance into upper secondary education is dependent on socio-economic family background in both countries, both samples are most likely highly selective and not representative for all adolescents in the two countries. In Switzerland, selectivity will most likely be even higher since entrance into upper secondary education is even more restrictive in the Swiss educational system (Keller et al., 2020). In addition, the sampling procedure differed between Germany and Switzerland. Therefore, it is advisable to account for the hierarchical structure of the data in the analysis.

One option would be to compute a multiple group comparison. This would have also made it possible to investigate possible differences in the effects between the two countries. Unfortunately, given the overall significant Chi2 tests of the SEM models in this study, it would not have been possible to determine if any found differences were statistically significant.

Another option would have been to include country as a control variable on the individual level. However, including cluster variables as a control variable on the individual level also introduces uncertainty into the model (Snijders & Bosker, 2012). Instead, it is advised to control for clustering on the appropriate hierarchical level.

MPlus does allow for the computation of the standard errors and chi-square test of model fit while taking into account the clustering structure of the data without modeling specific effects on the higher hierarchical levels (Muthén & Muthén, 1998–2017). Such so-called sandwich estimators can yield robust standard errors, accounting for the multilevel structure of the data, without specifically modelling cross-level effects (Muthén & Muthén, 1998–2017; Snijders & Bosker, 2012). MPlus currently allows for the control of two hierarchical levels, although one has to choose which level the standard errors should be controlled for (Muthén & Muthén, 1998–2017). The literature usually advises controlling for the highest level cluster (Snijders & Bosker, 2012). Unfortunately, small sample sizes within classes and schools, especially in the German data set, combined with missing data, led to some higher-level clusters only having one valid observation on some variables of interest. Consequently, the present data did not allow for simultaneous control of clustering at multiple hierarchical levels in MPlus. Even if this had been an option, the data structure would still have made it necessary to disregard either the class, school, or country level, as only two hierarchical levels can be included simultaneously.

In addition, studies have also shown that controlling for the highest cluster level is not always the best strategy if the number of elements on the highest cluster is small. Under such circumstances, the sandwich estimation might not provide the best model (Snijders & Bosker, 2012). Given the fact that the current study only included two elements at the country level and overall differences between countries were small for media use as well as for socio-economic background factors, the decision was made to control for students nested in classes instead. This also accounts for variance at the class level due to class composition effects, which most likely also influence students peer group connections (see Section 7.4. for a brief discussion about the effect of peer groups on adolescents’ media use). However, the results should be interpreted with some caution, as the complex hierarchical structure could not be fully accounted for.

In light of these challenges, future research should strive to carefully implement sampling and study designs that not only allow for a more thorough statistical modeling of the hierarchical structure but also for a more detailed investigation of cultural and national differences within the German-speaking communities.

Second, the study design also does not allow for a conclusive determination of the direction of the causal relationships between students’ language competences and their media-related extramural English contacts. While it is reasonable to assume that regular and intensive contact with the English language via media content will lead to learning processes, a reverse causal effect is likely also relevant. As a certain threshold needs to be reached to understand authentic media content or to interact with others, it is most likely that students’ language competences influence their ability to access media content or participate in communication with other English speakers. Furthermore, students with a high language competence might also feel more motivated to choose English media content instead of content produced or dubbed in German. It is thus likely that language competence and extramural English contacts have a reciprocal effect on each other. Future research should strive to implement experimental study designs to further investigate the causal relationship between extramural language contact and language learning. This will also increase our understanding of the learning process and the learning gains that can be expected from extramural English contact under different circumstances.

7.4 Further Limitations and Recommendations for Future Research

Overall, the findings in the present study have revealed that extramural English contacts might positively impact learners’ language competences, both for receptive and productive skills. In addition, the results have indicated that the decision to engage in these contacts is in part influenced by a learners’ socialization process into gender norms and a specific form of social media habitus.

While the study has shed important light on out-of-school informal language learning processes, the nature of the study design left some questions unanswered and raised some new ones. The previous subchapters already discussed limitations and possible biases, as they related directly to the results and the discussion of each research question. In this section, some further limitations of the present study and suggestions for future research will be mentioned.

First, the present study could not include other social factors such as the influence of peer networks and self-socialization processes. However, as Senkbeil and Wittwer (2009) and Henrichwark (2009) have pointed out, children and especially adolescents develop interests independently of their parents and undergo important self-socialization processes. Peer networks provide adolescents with meaningful connections and opportunities to talk about shared interests in media topics and genres, as friends often share the same interests. In addition, children and adolescents often have their first contact with the computer and the internet through their peer network (Henrichwark, 2009; Kommer, 2008; Senkbeil & Wittwer, 2009; Straub, 2006).

Furthermore, the internet especially offers young people the possibility to search for information and develop new values, identities, and interests independently of their parents. Media, therefore, also provides adolescents with the opportunity for self-socialization processes (Straub, 2006). These processes might be amplified by the increased autonomy smartphones and computers give adolescents. By using their personal devices, watching a movie might shift from a family activity in the living room to a private activity with minimal parental supervision. Thus, while being rooted in social practice and media habitus, extramural English contact through media channels is most likely also influenced by individual development.

Second, it was also beyond the scope of this study to include psychological factors such as interest or motivation, as they are a complex field of their own. However, research has repeatedly demonstrated the positive moderating role of motivation and interest for second and foreign language learning. The fact that media-related extramural English contacts are engaged in voluntarily most likely provides a strong intrinsic motivation for continued regular language contact outside of the classroom (Senkbeil & Wittwer, 2009). In addition, as interest is defined as the “heightened attention and emotional engagement that emerges when a person has a positive interaction with a content area or a task” (Hidi & Renninger, 2006, p. 112), it is very likely that interest in the media content helps to further increase and foster incidental learning processes.

Still, interest and motivation are part of Krashen’s input hypotheses (1982, 1985, 1989): According to him, input will only lead to acquisition if a person’s affective filter is low enough. The filter will be low if the learner feels motivated, self-confident, self-efficient, and not anxious. A high filter will prevent acquisition, no matter how much input is provided. For students with a low level of anxiety and a high level of motivation, self-efficacy, and a conducive self-concept, extramural English contact will likely result in a higher level of incidental learning. Nevertheless, the exact nature of the relationship in the case of media-related extramural English contacts should be investigated further.

Furthermore, it is very likely that a positive experience and positive feedback in understanding the content of a movie or book, or from a successful interaction with others, will positively influence these psychological factors. This, in turn, can lead to a positive self-concept and self-efficacy as an English speaker and help lower the affective filter. As students grow more accustomed and confident in their role as language users in an international community, their learning revenues might therefore increase. It would also be interesting to see how incomprehensible input, unsuccessful interaction, or negative feedback might influence students’ motivation or self-efficacy.

In addition, a positive experience with English outside of the classroom might also increase interest and strengthen students’ understanding of the personal relevance of English for their personal future. Therefore, extramural English contacts might help turn English from a school subject with general significance into an object of personal significance (Tin, 2013).

Third, future research might also be interested in the interaction between psychological and social factors, such as gender and socio-economic background, and their effect on incidental learning. The present study did not hypothesize that socio-economic background would affect the actual incidental learning process arising from extramural contacts. Drawing on Krashen’s theory, for the socio-economic background to affect the learning process itself, students from different backgrounds would have to systematically differ in their affective filter. However, unlike classroom instructions, extramural English contacts are not mandatory. Students can choose the media category and content they are most interested in. As a result, it can be argued that the affective filter should generally be low for students who decide to engage in extramural contact, thus creating an ideal basis for incidental language learning. Thus, while socio-economic background might influence the probability of engaging in extramural contact, it was not expected to play a moderating role in the learning effect for students who do decide to engage in said contact. The same can be said for gender. However, empirical evidence in this area is still limited, and future research should examine the interactions of these various factors more closely.

Fourth, this study could not include students’ actual communication and interaction strategies with extramural English content. Nevertheless, for a full understanding of the benefits of interactive media use it is important to understand how students interact online and how learners might use other-regulation and object-regulation to work through their zone of proximal development. Some studies have demonstrated that native speakers and more advanced language learners might provide novices with explanations, co-construction, positive and negative feedback (Rankin et al., 2009; Thorne, 2008). However, research in this area is undoubtedly still at the beginning.

In addition, little is known about if and how students employ explicit learning strategies as they engage in extramural contact. Given the fact that students in the present study have indicated that they are aware of the learning potentials deriving from the extramural English contacts, future research should examine which explicit learning strategies students might employ and how these strategies might differ between fast-paced media content such as movies or computer games and the slower activity of online and offline reading.

Fifth, it was also beyond the scope of this study to investigate the effect of extramural English contact on students’ speaking skills. Sundqvist (2009a) was able to show that extramural contact can have a positive influence on learners’ oral skills. This might be especially true for extramural contacts from listening to and watching English content, as it provides direct auditory examples of pronunciation. Anecdotal evidence, for example, links the frequent contact to English music, movies, and TV series with the fact that Dutch and Flemish children are often able to pronounce English words perfectly (Koolstra et al., 2002). It would therefore be interesting to see how regular extramural English contact might lead to different proficiency levels in certain dialects. Due to the dominance of the United States, it could be expected that learners with a high level of extramural English contact might have a special inclination to reproduce the most represented American accents.

Last, the present study did not address the topic of media or ICT literacy, i.e., students ability to access, analyze, and evaluate images and sounds, as well as their ability to communicate with and through these media channels and use the necessary technology appropriately (Henrichwark, 2009). As studies have shown, and as Thorne (2008) points out, the media is a stable part in the lives of most adolescents, and the internet is much more a social fact than a technological one, yet this does not mean that adolescents automatically have a high level of ICT literacy. The present study assumed that students who engage in media-related extramural English contact at least had a high enough media literacy to access the content of interest. However, results from the ICIL study show that not all young people have the same level of ICT literacy. Although the study focused on school and work-related ICT skills, the findings raise doubts about equal access to online and computer-based media content and the competence to navigate and evaluate the found information. The study also revealed that media and ICT literacy strongly depend on social background factors and gender (Fraillon et al., 2014; Lorenz et al., 2014). Similarly, Henrichwark (2009) also provided evidence that not all students have the same ability to understand and evaluate content presented to them via media channels and that students’ media literacy is dependent on socio-economic background.

Results like these are worrisome, especially given that children and adolescents have increasingly autonomous control over their media intake through their own technical devices. Therefore, the competence to understand and evaluate content and information is a crucial skill to navigate the online world and not fall prey to misinformation and radical ideas. Future research should include media literacy and ICT literacy measures to capture their selective effect on extramural English contact and investigate how students process and evaluate the information they receive online.

7.5 Pedagogical Implications

The high frequency of extramural English contacts among adolescents in this study poses new challenges for the educational system. In an ever-changing media landscape, teachers and parents should be aware of the frequency with which adolescents might engage in extramural English contacts and the benefits and risks arising from them. Ignoring these influences means ignoring students' increasingly heterogeneous language backgrounds. Some students might participate in a complex net of extramural activities and are part of a stable online community where they take on the role of a self-sufficient language user. They may have a highly specialized vocabulary, which might not be recognized or valued in the school context. For these students, classroom instruction might even be frustrating or boring.

Other students might be hesitant to engage in any out-of-school contact, relying solely on the input from their teachers. Therefore, these students might be less able to self-regulate their language output or interact with members of the target language community in an informal and natural context. For these students, parents and teachers can serve as important motivators and inspiration.

By the time of this study, results from Grau (2009) suggested few teachers took students’ extramural English contacts seriously. They also seemed to be mostly unreflective about their own role in providing students with appropriate listening and viewing strategies for their out-of-school language contact (Grau, 2009) or encouraging students to find their own voice in the international online community. One reason for this might be missing knowledge and expertise on the teachers’ side (Sundqvist & Sylvén, 2012a). It seems crucial for teachers to be better informed of the newest developments in media and technology.

In addition, to encourage students to seek extramural English contacts, the educational system should also strive for a stronger incorporation of students’ interests and prior knowledge into the classroom. The modern English classroom should strive to build bridges between students’ in- and out-of-school language contacts, without necessarily making them a homework assignment. Doing so can link formal instructions to students' interests and activate prior knowledge. Learners’ extramural English contacts should be seen as an opportunity to increase in-school motivation and interest in English as a subject and as a fruitful addition to the language classroom. Ignoring the interests and language backgrounds students bring to the classroom means to deepen the rift between students’ in-school and out-of-school lives and leaves students to their own devices when it comes to choosing, consuming, evaluating, and reflecting on media content.

Of course, incidental language learning should not be understood as a replacement for thorough and well-structured language instruction within the educational setting. Especially not since incidental language learning has been shown to be a slow and error-prone process and can be especially overwhelming for beginners (e.g., Letchumanan et al., 2015; Sok, 2014; Webb & Rodgers, 2009). It is important to teach learners the basic linguistic principles and lexical system of the target language, as well as to make them aware of (vocabulary) learning tasks and to teach them explicit strategies for doing so. This ensures that learners start their language journey by studying a base vocabulary, grammar, and a robust connection between form and meaning (d'Ydewalle & van de Poel, 1999; Hulstijn, 2001; Neuman & Koskinen, 1992).

Thus, when integrating students' out-of-school interests into the classroom, in-classroom activities should strive to make genre- and context-specific linguistic structures and vocabulary of contemporary media content explicit to the learners. In addition, the classroom should endeavor to equip students with learning strategies for out-of-school language use and promote active language use in and outside of the classroom. In this way, it can be possible to incorporate media content from outside the classroom and allow learners a space to develop their personal interests. This can then help young learners to navigate and guide their own journey on their way to becoming self-sufficient within a media world that is governed by English as its dominant mode of communication.

This will also enable learners to reach a level of proficiency in which a broad range of authentic input will be comprehensible and but help learners to establish a monitor that helps to regulate and check the output before it is uttered (Krashen, 1985). Once these requirements are met, and a base is established, incidental learning deriving from extramural English contacts can be a helpful and interesting additional way of language learning.