• Keine Ergebnisse gefunden

59 they could be balanced bilinguals, but they could also be, and this is the more frequent type, highly unbalanced bilinguals.

On a number of occasions, we have already stated that these bilingual speakers, i.e.

immigrants and their children and grandchildren, are very frequent in current societies, due to global developments in our modern, western societies (see Chapters 1 and 2). Therefore, bilingual heritage speakers that grow up in monolingual countries are increasingly taking foreign language classes together with monolingual foreign language learners (Montrul 2016:

3). As the discussion in Chapter 3.1.2 demonstrated, it seems to make a difference whether a language learner has previous knowledge of one or of two languages. This has already been shown to be the case in many studies that compared L2 and L3 learners. However, this typology of heritage speakers should have also established that these bilingual learners show qualitative differences from balanced bilinguals and also from monolingual speakers that have acquired a foreign language in a classroom situation.

To sum up, the previous discussion confirms that a heritage speaker can indeed be seen as one type of a bilingual speaker. In addition, it seems as if it is actually a rather frequent phenomenon for children to grow up as bilingual heritage speakers, hence with a heritage language and with a majority language, i.e. the language of the environment. Surprisingly little research has so far investigated multilingual development in child bilingual speakers with a heritage language living in an area where another language is the majority language. Therefore, we will focus on exactly this group of language learners. The so-called control groups will be monolingual speakers (i.e. monolingual German, Russian, Turkish, and Vietnamese speakers) that start learning English as a foreign language in school. Hence, English is neither their heritage language nor the language of the environment; they grow up with one language, the language of the country they are living in, and study a foreign language in school.

Before we can continue to analyze these groups of learners of English, some further concepts need to be introduced and clarified. The subsequent chapter addresses a topic that is said to be a property of bilingual or multilingual speakers; hence, of the types of learners that are in focus of the current study.

60 […] metalinguistic awareness is a major subject of TLA [third language acquisition]

research. Several leading publications have identified metalinguistic awareness as a key component of multilingual competence and as a factor that sets multilingual learners apart from monolingual learns, providing the former with a strategic advantage for further language learning.

In the following, we want to assess what metalinguistic awareness is, if knowing more languages equals higher metalinguistic awareness, and how it is associated with advantages in non-native language acquisition (see Chapter 3.7). First, we provide two definitions of metalinguistic awareness.

(1) Metalinguistic awareness is the ability to think flexibly and abstractly about language;

it refers to an awareness of the formal linguistic features of language and the ability to reflect upon. Metalinguistic awareness allows the individual to step back from the comprehension or production of an utterance in order to consider the linguistic form and structure underlying the meaning of the utterance. (Malakoff 1992: 518)

(2) Thus metalinguistic awareness refers to the ability to focus attention on language as an object in itself or to think abstractly about language and, consequently, to play with or manipulate language. A multilingual certainly makes more use of this ability than a monolingual. One might even state that linguistic objectivation is the multilingual’s most characteristic cognitive ability. (Jessner 2006: 42)

Both definitions are clearly comparable and show some overlap. Metalinguistic awareness is defined as the structural knowledge about language that goes beyond the understanding of the meaning of language use. Furthermore, this knowledge about language(s) is neither limited to a specific language, nor is it limited to just one language. Jessner (2006) goes one step further and agrees with Bono (2011) in that heightened metalinguistic awareness is associated with bi- or multilingual speakers in comparison to monolingual speakers. Bi- and multilingual speakers have an increased structural knowledge, because they have theoretical insights into more than just one language, and this increased structural knowledge may be advantageous in further language acquisition (Jessner 2006: 42).

All this is part of the so-called M-factor (Jessner 2006, 2008). M-factor, or multilingualism factor, represents the features that evolve in speakers that have access to more than one language (Jessner 2008: 275) and this ultimately leads to “an enhanced level of metalinguistic awareness and metacognitive strategies” (Jessner 2006: 35). This means that metalinguistic awareness is part of the cover term M-factor that refers to language aptitude and language skills that can be found in multilingual learners, hence, it is a characteristic trait of third language learners (Jessner 2006: 56; Jessner 2008: 275). Such additional qualities develop, because of the increased contact of different languages within one speaker (Jessner 2008: 275).

This discussion about metalinguistic awareness is relevant for the current study, because it seems that there is considerable interaction between metalinguistic awareness and cross-linguistic influence. The following paragraphs analyze why heightened metacross-linguistic

61 awareness seems to be a feature of people that have knowledge of more than one language, and whether there are limitations and restrictions to this rule.

Third language acquisition was stated to be more complex than second language acquisition, because cross-linguistic influence is not limited to occur from the L1 to the L2 or the other way around, but three languages can possibly influence each other (Jessner 2008: 271;

see Figure 2; Chapter 3.1). Hence, this enlarged linguistic system is what divides bilinguals or multilinguals and monolinguals, and it is claimed to cause the development of skills and competences which further influence foreign language acquisition (Jessner 2008: 275).

Apart from cross-linguistic influence from all previously acquired languages, another component, namely the aforementioned development of skills and competences on a cognitive level, needs to be mentioned. There seems to be something that is responsible for causing the language acquisition process to be qualitatively different in second than in third language acquisition which cannot be explained alone with the fact that more languages are available (Jessner 1999: 203). Jessner argues this to be metalinguistic awareness; once again it is “the ability to focus on linguistic form and meaning […] to categorize words into parts of speech;

[to] switch focus between form, function, and meaning; and [to] explain why a word has a particular function” (2008: 275, 277).

We find a similar concept, which is labeled “multi-competence”, in Cook (2016c) and Franceschini (2016). Multi-competence can be defined as “the overall system of a mind or a community that uses more than one language” (Cook 2016c: 2). This implies that multilingual speakers are not simply speakers that know several languages, but that the availability of more than one language adds something else. Franceschini (2016: 105) defines this as the “third quality”. More explicitly, she states that it is “a quality that represents more than the sum of its parts” (Franceschini 2016: 105). She explains that the experience a language learner gains throughout his or her live, all the linguistic competences, also the awareness of these competences, and the general awareness of the linguistic systems, adds to the individual linguistic repertoire (Franceschini 2016: 105). Such multilingual, or multi-competent individuals become “flexible speaker[s]” (Franceschini 2016: 106), because they are able to use their language repertoire freely, and they can communicate flexibly in any of their known languages. This may be a conscious process or even an unconscious potential (Franceschini 2016: 107). In this sense, Cook (2016c) and Franceschini (2016) somehow add to Jessner’s (1999, 2008) metalinguistic awareness, because in this definition of multi-competence, we find both theoretical knowledge, i.e. language awareness, and also practical usage based properties.

62 Yet, this is not unique to bilinguals or multilinguals. Franceschini (2016: 109) remarks that the concept of multi-competence is not limited to bi- or multilingual speakers but that monolingual speakers can also be multi-competent users. Multi-competence is not an exclusive feature of speakers with knowledge of more than one language. Monolinguals can also have

“communicative expertise”; they can be flexible speakers of one language due to a flexible use of that one language in different social and communicative contexts (Franceschini 2016: 109).

Hence, the concept of multi-competence applies either to different languages or to different varieties of one language. Similarly, Jessner (2008: 277) reports that monolinguals also have metalinguistic knowledge, though only from one language; hence, the metalinguistic awareness of bilinguals is said to be higher than of monolinguals (Jessner 2008: 277). Bilinguals or multilinguals are capable of reflecting to a different extent on their language usage than monolinguals: their multiple languages allow them, for instance, to compare these distinct systems (Jessner 1999: 203). This might lead to detecting helpful similarities or contrasts between languages.

Taking up Jessner’s (2008) argument, Cenoz (2013: 75) also states that previous learning experience and the knowledge of two languages, hence two different linguistic systems, are the reasons why bilinguals enlarge their level of metalinguistic awareness.

Language learning involves certain techniques; the more languages you learn, the more learning strategies you experience (Cenoz 2013: 76). This adds to the theoretical knowledge about learning languages in general. In addition, the learners gain theoretical knowledge about the languages may be able to reflect upon them on an abstract level. As a consequence, so Jessner (1999: 203), such higher developed metalinguistic awareness enhances the development of further learning strategies (1999: 203) which could convert into an advantage for bilingual learners. Keeping this in mind, it does not come as a surprise that Jessner equates higher metalinguistic awareness with a higher success rate in (foreign) language acquisition (Jessner 2008: 277).

Bono (2011: 49) also argues in her study that multilingual learners have a high level of metalinguistic awareness; yet, she calls it “linguistic awareness”. According to her explanations, metalinguistic knowledge develops in the process of foreign language acquisition, because this is when you reflect on the structure and features of a language (Bono 2011: 49).

Bono (2011), however, does not refer to a bilingual advantage, but instead, she claims that previously acquired foreign languages are responsible for a heightened metalinguistic awareness, which in turn helps them to rely on formerly acquired languages. She continues by

63 claiming that learners should explicitly be made aware of similarities and differences between their languages (Bono 2011: 49).

The study by Bono (2011) has two crucial implications: first, knowing two languages may not automatically lead to heightened metalinguistic awareness; it may not be a subconscious characteristic or something that automatically develops once you know more than one language. It may rather be that explicit explanations or explicit theoretical training is necessary. This is linked to the second implication, namely that foreign language learners, i.e.

those who receive formal training instead of bilinguals who require both of their languages naturally, show heightened levels of metalinguistic awareness. Similarly, Cook also argues that

“raising awareness of language in general helps second language learning” (2016a: 51). In the context of the current discussion, this could be easily extended to further language acquisition.

Cook calls the result of this awareness raising “language awareness” and he projects general educational advantages, i.e. the formerly mentioned advantages in language acquisition (2016a:

51). Both Cook (2016a) and Bono (2011) refer to language acquisition that happens in a tutored setting.

Another study that also does not support the claim that bi- or multilinguals have higher metalinguistic awareness than monolinguals is Spellerberg (2016). The explanation for this finding, however, is somewhat different. Spellerberg (2016) also focused on metalinguistic awareness, and she wanted to find out if and how this affects the academic achievement of monolinguals, bilinguals, and multilinguals, and also how an additional factor, the socio-economic status, affects metalinguistic awareness. Spellerberg’s (2016) study included 219 high school students in Denmark at the age of 14 to 16. She separated the students into three groups based on self-reports: 106 monolingual Danish students, 26 bi- and multilingual students that spoke Danish at home, and 87 bi- and monolingual students that did not speak Danish at home (Spellerberg 2016: 26). All students study English in school as a foreign language and the majority reported to use English exclusively in the school context and only a small number of the participants confirmed to use English actively in their daily lives outside of school (Spellerberg 2016: 26). She used a comprehensive Metalinguistic Awareness Test, based on Pinto et al. (1999), to measure metalinguistic awareness (Spellerberg 2016: 25) and she also took the results of the school leaving exams (Spellerberg 2016: 29). The latter scores are used as a measure for academic achievement. The correlation of the scores based on the metalinguistic awareness test and the school leaving exams showed that metalinguistic awareness correlates positively with the exam scores (Spellerberg 2016: 31). In general, these results show that the higher the metalinguistic awareness, the better the academic achievement.

64 For the current study, especially the exam scores of English, i.e. the foreign language of the participants, are of interest. Spellerberg (2016: 35) reports that metalinguistic awareness also correlates positively with the exam results in English; hence, for foreign language learning, it seems to be advantageous to have higher metalinguistic awareness levels. However, this was not the strongest correlation; in fact, it was the weakest, compared to all other exam scores (Spellerberg 2016: 35). Nevertheless, it supports the idea that metalinguistic awareness helps with further language acquisition.

However, and this is a striking finding, the bi- or multilingual participants did not outperform the monolingual participants in terms of metalinguistic awareness (Spellerberg 2016: 34). The opposite was the case: the monolinguals outperformed those students that know more than one language (Spellerberg 2016: 36). Between the bi- or multilingual students, the mean score for metalinguistic awareness did not differ statistically significantly. Overall, however, the positive correlation of metalinguistic awareness and academic achievement was visible for all participants, irrespective of the number of known languages. This finding runs counter to previous research such as Cenoz (2013), who would have expected bi- and multilinguals to outperform monolingual students, because higher metalinguistic awareness is associated with bilinguals and also with a facilitating effect on further language acquisition.

In addition, and this may have a huge impact on the overall exam results, too, Spellerberg (2016) found that socio-economic status had an effect on the metalinguistic awareness scores. Lower status resulted in lower metalinguistic awareness scores (Spellerberg 2016: 36). Overall, metalinguistic awareness can influence academic achievement and foreign language performance. However, bi- or multilingual students do not have a higher level of metalinguistic awareness per se. There are further influencing factors, such as socio-economic status, that affect metalinguistic awareness. This last point is above all interesting, because those bi- and multilingual participants that did not speak Danish at home had the lowest metalinguistic awareness scores and they also had the lowest socio-economic scores (Spellerberg 2016: 36).

This shows, once again, that acquiring languages is a complex phenomenon that is affected by a number of variables; metalinguistic awareness seems to be one of them; yet, socio-economic status should be included as well, since some of the variation can be explained on the basis of this background variable.

Furthermore, Spellerberg (2016) admits that language proficiency was not controlled for, hence, nothing is known about the participants’ proficiency levels of Danish or of their other languages. This is arguably another influenting factor. Especially against the background of Chapter 3.5 and the characteristics of heritage speakers, it is possible that these students may

65 not be equally proficient in both languages. This is assumed, because the description of the participants in Spellerberg (2016) seems to be comparable to Westergaard et al. (2017), Hopp (2019), and also the participants of E-LiPS (as described in Siemund & Lechner 2015, for instance; see also Chapter 5.3). Spellerberg’s definition of bilingual students in Denmark, i.e.

“children who have a mother tongue other than Danish and who do not learn Danish until they come into contact with the surrounding community or through the teaching in school” (2016:

37), resembles that of heritage speakers given in Chapter 2.5. What is known, however, is that the metalinguistic awareness test was conducted in Danish and all participants attend Danish schools, therefore, they are assumed to have a high command of Danish. In addition, Spellerberg refers to Danish as the “majority language” (2016: 20). This again corresponds to the Norwegian or German contexts referred to earlier. Therefore, it is quite likely, that the other language (or the other two languages) may be their weaker language(s), the typical situation found in heritage speakers (see Montrul 2016). This could explain their lower metalinguistic awareness skills. Lower language skills because of an unbalanced bilingual status may result in lower metalinguistic awareness.

This last claim finds some support in a study investigating how the degree of bilingualism, i.e. balanced bilinguals versus unbalanced bilinguals, affects metalinguistic awareness (Cohen 2013). Cohen (2013) analyzes young, primary school French-English bilinguals attending an international school in France, and separates them into two groups, balanced bilinguals, on the one hand, and unbalanced bilinguals (here called dominant10 bilinguals), on the other hand. In order to assess and compare “control of linguistic processing”

of these two groups of bilingual children, they performed a number of metalinguistic tasks, such as a word renaming task and a symbol substitution task in English and in French (Cohen 2013).

Interestingly, only under specific circumstances, do the observed differences reach statistical significance, namely, only if the highest score, i.e. either the score of the English version or the French version, is considered. Then, the balanced bilinguals performed better than the unbalanced bilinguals (Cohen 2013). Furthermore, Cohen (2013) reports that it was not consistent that all unbalanced bilinguals scored higher in their dominant language, some performed better in their weaker language. She explains that unbalanced bilinguals have to pay closer attention to their minority language (in general) and they may therefore more skillful in applying metalinguistic knowledge to this language and not to the dominant language. This is quite intriguing; it indicates that balanced bilinguals may have an advantage over unbalanced

10 For the sake of consistency within the current study, we will use the term unbalanced bilinguals instead of dominant bilingual.

66 bilinguals, meaning that they potentially have higher metalinguistic knowledge than unbalanced bilinguals. However, it also demonstrates that the weaker language needs to be assessed along with the dominant language in order to capture the entire linguistic competence of bilinguals.

These findings are based on just a small sample (n=38), yet it provides further support that balanced bilinguals are different from unbalanced bilinguals in a number of respects, with the degree of metalinguistic awareness being one of them.

To sum up, as this chapter should have shown, and as was addressed on a number of occasions in the previous chapters, language acquisition is a multi-layered and complex process.

Franceschini (2016: 101) summarizes that “[l]anguage acquisition – both untutored and tutored – turned out to be a very intricate matter, closely linked to personal experience, the speaker’s attitude towards societies and the narrow social context.” As a consequence of bilingualism or multiple language acquisition, languages are not simply added to the brain, but they form a complex system within one speaker; Franceschini uses the term multi-competence and explains that this is the “third quality” (2016: 105). A crucial role plays the social situation (for instance socio-economic status) but also the context, in which a language is learned (tutored or untutored, simultaneous or sequential, for example).

Furthermore, we discussed that metalinguistic awareness is not only a property of bi- and multilingual speakers, but also of monolingual speakers (but maybe to a more limited extent). The availability and active access to more than one language, as well as the past experience of foreign language acquisition, may result in increased metalinguistic awareness.

This, however, may be less pronounced in (unbalanced) bilingual heritage speakers. First, they have not necessarily acquired a non-native language other than English, because both of their previously acquired languages may actually have the status of a native language. Second, most heritage speakers may have only limited language skills in their heritage language (see Chapter 3.5) which could negatively affect metalinguistic awareness (see Cohen 2013). Third, and this was particularly prominent in Spellerberg’s (2016) study, heritage speaker may belong to a group with a lower socio-economic background, and since socio-economic status seems to correlate with lower metalinguistic awareness, it may explain this unfavorably situation for bilingual heritage speakers.

In the next chapter, we will now turn to bilingual advantages. This topic is closely related to metalinguistic awareness and we will find a similar argumentation to the one in this chapter.

67 3.7 Bilingual advantages

L2 and L3 acquisition have a lot in common. Many individual and contextual factors such as age, motivation, socio-economic status, among others, influence the acquisition process of languages, no matter if the additional language is the second or the third language. Yet, throughout the previous chapters, we have stressed that there are many differences.

Unquestionably, bilinguals have a broader linguistic repertoire, since they have access to not just one but to two languages. This enlarged linguistic repertoire may be helpful in further language acquisition (Cenoz 2013; Cenoz & Valencia 1994; Jessner 2008; Maluch et al. 2015;

Sanz 2000). However, is it that justified to claim that bilinguals have an advantage over monolinguals simply because their linguistic repertoire is larger? There has been a lot of research on this matter in the recent past; especially noteworthy are Cenoz (2003) and Cenoz (2013) that provide general overviews and include broad discussions. The aim of this chapter is to look a little closer into the popular folk wisdom “the more languages a person knows, the easier it becomes to acquire an additional language” (Cenoz 2013: 74).

First, we need to clarify what we refer to when we talk about advantages. Many studies that analyze bilinguals specifically involve cognitive advantages. In these studies, bilingualism was repeatedly reported to result in better cognitive skills (Aronin & Jessner 2015; Barac &

Bialystok 2011; Barac et al. 2014; Cenoz & Valencia 1994; DESI 2008; Jessner 1999, 2008;

Ringbom 1987). We also saw that bilingualism may increase the level of metalinguistic awareness; however, this was not supported for all types of bilinguals (see Chapter 3.6).

These two advantages, i.e. superior cognitive skills and heightened metalinguistic awareness, however, are not the main focus here. What we rather try to identify are linguistic advantages, as in higher proficiency scores and more target-like foreign language use. Hence, if bilinguals showed an advantage in foreign language acquisition, which is sometimes argued to be the case (see for example Aronin & Jessner 2015; Cenoz 2013; Jessner 2006), this would be the bilingual advantage we are interested in here.

One study that reports such advantages in additional language acquisition of bilingual learners over monolingual learners of a foreign language is a study by Cenoz and Valencia (1994). In their investigation, they included Spanish-Basque bilinguals and Spanish monolinguals in the last year of school (between age 17 and age 19) (Cenoz & Valencia 1994:

199). They measured different English abilities, namely speaking, listening, reading, writing, and vocabulary (Cenoz & Valencia 1994: 200). They clearly demonstrated in their statistical

68 analysis that bilingualism has a positive influence on the performance in English (Cenoz &

Valencia 1994: 204). They argue that the structural knowledge of the two languages Basque and Spanish helped with English – yet, not because of transfer, but because of other side-effects of bilingualism such as increased metalinguistic awareness (see again Chapter 3.6) and because of a higher communicative competence and sensitivity towards different languages (Cenoz &

Valencia 1994: 205). They excluded the possibility of transfer from Basque, because there is no structural similarity between Basque and English that could explain the better performance of the bilinguals (Cenoz & Valencia 1994: 205).

Similarly, Sanz (2000) also examined the role of bilingualism on the acquisition of an additional language, though not in the Basque country but in Catalonia. She looked at Catalan-Spanish bilinguals and Catalan-Spanish monolinguals (Sanz 2000: 23) and interestingly, she also found that the bilinguals performed better in English than the monolinguals (Sanz 2000: 33-34). She claimed that she could confirm an already existing hypothesis, namely that “L1 and L2 literacy has a positive effect on L3 learning” (Sanz 2000: 34). This ultimately leads back to metalinguistic awareness, because due to the literacy development in both languages, metalinguistic awareness is heightened and allows language learners to use their knowledge more efficiently for further language acquisition (Sanz 2000: 36). We will come back to this effect of literacy later again.

Further support comes from Agustín-Llach (2017). She also investigated Spanish-Basque bilingual participants like Cenoz and Valencia (1994) and compared their performance in English with monolingual Spanish participants (Agustín-Llach 2017: 5). Again, she could identify a bilingual advantage, because the bilinguals outperformed their monolingual peers in English (Agustín-Llach 2017: 10).

These three studies clearly demonstrate that bilingual speakers are better in learning English than monolingual speakers. Yet, this seems a bit too optimistic and simplistic, especially considering what the aforementioned discussions (especially Chapter 3.1 and 3.6) have shown. Indeed, contrary to the assumption that there exists an overall bilingual advantage, there are several large-scale attainment tests of high school children that report that bilingual or multilingual children, mostly immigrant children, score poorly on such tests, in comparison to their monolingual peers (see for example OECD 2010; Stanat et al. 2010, 2016). What is reported in these studies is that children who speak another language than the majority language of the country at home with their family are outperformed by monolingual students in terms of school performance (Lechner & Siemund 2014a: 320). This is interesting, because one would

69 expect exactly this group, the bilingual students, to have an advantage, at least in the scores for foreign languages.

A closer look, however, reveals that the underperformance can most likely not be explained with the fact that these children have a migration background or are bilingual, or that they use another language than German at home (Lechner & Siemund 2014a: 336, 340-341).

This claim is based on a small study (that will be discussed in Chapter 5.1 in more detail) that could neither replicate the bilingual advantage found in the DESI study (2008) nor could it support results from PISA (2009) (Klieme et al. 2010; OECD 2010) that attributed bilingual immigrant students a disadvantage in school (Lechner & Siemund 2014a: 336, 341). What they could show, however, was that the socio-economic status of the families had a significant impact. A low socio-economic status correlated with poorer results in the English tests and vice versa (Lechner & Siemund 2014a: 339). Hence, they did not find evidence for bi- or multilingualism being responsible for comparatively poorer performance in English. However, at the same time, they did also not find evidence for an increased metalinguistic awareness that is advantageous for multilinguals. In sum, this shows that bilingualism may not necessarily be advantageous or detrimental, when further background variables are controlled. It could not have any effect at all.

The same can be found in Hopp (2019). He analyzed Turkish-German bilingual primary school children and compared their performance in English with monolingual German school children (see Chapter 3.1.2). Yet, he could not find a difference between these two groups;

hence, no bilingual advantage or disadvantage was reported.

Slightly different to these two studies are the findings in Şahingöz (2014). She analyzed Russian-German and Turkish-German heritage speakers, age 16, who grow up in Germany and study English as a foreign language at school (Şahingöz 2014: 90). The area of investigation is English word order in both written and spoken production data. Due to some cross-linguistic influence from the heritage languages in the English production data (Şahingöz 2014: 234), which had a negative effect on the target-like English word order, she identified small disadvantages of the bilingual participants compared to their monolingual peers (Şahingöz 2014: 237). This result is the opposite of what was found for instance in Cenoz and Valencia (1994). Şahingöz (2014) shows that bilingualism can even be disadvantageous.

Ghezlou et al. (2018) make similar observations, because they can also not find bilingual advantages in their study, but they found a lower performance of the bilinguals when compared to the performance of the monolingual learners of English. The bilingual participants they investigated had acquired Azeri as their L1 and Persian as their L2 (Ghezlou et al. 2018: 177).

70 They acquired the L1 naturally (as their native language) and they started to learn the L2 Persian at the age of seven as part of their educational training. The monolingual participants are speakers of Persian and they started to receive formal training of Persian also at the age of seven. Persian was the language of instruction at school for both groups. At the time of the study, both groups were enrolled at a university and learned English as an L3 or L2 respectively (Ghezlou et al. 2018: 178). The bilinguals are located at an Azeri speaking city and the monolinguals at a Persian speaking city. Azeri and English are not typologically similar, yet they share pre-nominal adjective placement (Ghezlou et al. 2018: 178). Nevertheless, the authors could not find “superiority of bilinguals over monolinguals” (Ghezlou et al. 2018: 179), because they only find non-facilitative influence from Persian and no facilitative influence from Azeri in the bilingual data. Overall, the bilinguals were outperformed by the monolingual participants (Ghezlou et al. 2018: 181). Hence, they conclude that in their language acquisition scenario, i.e. in subtractive bilingualism, where the L2 becomes the more dominant language, bilinguality does not provide the bilinguals with an advantage (Ghezlou et al. 2018: 180-181).

Furthermore, Siemund and Lechner (2015) made an interesting observation when they investigated two different age cohorts, younger learners at the age of 12, and older learners at the age of 16. Indeed, they found a bilingual advantage, yet, this was only visible in the younger learners and not in the older learners (Siemund & Lechner 2015: 157-158). They explained that the initial bilingual advantage disappears at later stages of the language acquisition in school (Siemund & Lechner 2015: 158). The older cohorts performed similarly; no group showed to be superior to the other.

These nuanced results find support in Maluch et al. (2016). In their study based on a large-scale longitudinal project carried out in Germany, they identified clear bilingual advantages in foreign language acquisition for students in school year six (Maluch et al. 2016:

116). They compared the performance of monolingual German and different groups of bilingual students in an English Cloze-test11 (Maluch et al. 2016: 113-114). The home languages spoken by the bilingual students were largely Turkish, Arabic, Chinese and numerous other languages.

The bilinguals were not seen as one homogeneous group, but they were subdivided further, according to a number of background variables, such as language dominance and age of onset of learning German (Maluch et al. 2016: 113). As has just been stated, there was a clear advantage for the bilinguals, compared to their monolingual peers in the early phase of the

11 A Cloze test usually consists of four texts that include word gaps which need to be completed by the students.

This test instrument is used to assess reading comprehension, spelling, grammar, and vocabulary knowledge in the respective language and is widely used in language studies (see also Lehmann & Lenkeit 2008).

71 study, hence, when the participants were in school year six. However, in school year eight, these findings could not be replicated but represent a different pattern: overall, there is no bilingual advantage visible in the data, except for the bilinguals that reported to mostly use German in their homes (Maluch et al. 2016: 116). The authors explain these varying patters with, on the one hand, “cognitive and linguistic advantages” of bilinguals, but also, on the other hand, a major impact of the monolingual language practices found in the English classroom (Maluch et al. 2016: 116). Since the potential of the bilingual students is not supported in the classroom settings, their advantages vanish throughout secondary-schooling.

Further, less straightforward findings are presented in Maluch et al. (2015) and Maluch

& Kempert (2017). The study by Maluch et al. (2015) includes an impressive number of participants. They analyzed almost 3,000 students between the age of 10 and 15; the participants are divided into monolingual German students and five bilingual groups, depending on their home languages Arabic, Chinese, Polish, Turkish, and other (Maluch et al. 2015: 79). They also relied on a Cloze test and controlled for a number of background variables such as cognitive capacity, social and family background, gender, and age (Maluch et al. 2015: 79). Their assessment demonstrates a general trend: bilingual children showed a higher foreign language attainment than their monolingual peers (Maluch et al. 2015: 82). Yet, they admit that they found considerable variation between the five different bilingual groups and that additional variables, especially proficiency of the language of instruction, had a strong impact on the results (Maluch et al. 2015: 82-83). The latter point is a remarkable result: they noticed that only those students who had high language skills in German had an advantage over the monolinguals in the English assessment, and that those students that had weak proficiency scores in German performed lower in English than their monolingual peers (Maluch et al. 2015:

82-83). In sum, however, they argue for a bilingual advantage of immigrants and support the view that immigrant bilingualism, even if the minority language is not part of the formal education, can be seen as a resource (Maluch et al. 2015: 83).

In a similar vein, Maluch and Kempert (2017) investigate in a follow-up study how further factors such as manner of language acquisition and language use of bilingual students affect the acquisition of English as a foreign language. As a major advancement compared to the studies that were discussed above, Maluch and Kempert (2017: 6) subdivide the bilingual participants according to manner of heritage language acquisition (informal acquisition at home versus additional formal instruction), age of onset of learning the minority and majority language (simultaneous bilinguals versus sequential bilinguals), and use of languages (frequently switching between languages versus occasionally switching versus never switching)

72 on a self-report basis. Interestingly, they do not reveal an overall bilingual advantage when acquiring a foreign language, but only certain groups show an advantage over the monolingual participants (Maluch & Kempert 2017: 8). Those bilinguals that had additional formal education in their heritage language scored higher than the monolinguals; yet, for those who only learned the minority language informally at home, bilingualism was not shown to be an advantage for English, because they had lower scores than the monolinguals (Maluch &

Kempert 2017: 8). Furthermore, the sequential bilinguals did not differ from the monolinguals and also those students that reported to switch only infrequently or never between their two languages showed no differences when compared to the results of the monolingual participants (Maluch & Kempert 2017: 8). What this clearly shows is that we need to differentiate, when conducting language acquisition studies.

Looking at these contradictory findings, i.e. bilingual advantages on the one hand (Aronin & Jessner 2015; Cenoz 2013; Cenoz & Valencia 1994; Agustín-Llach 2017; Jessner 2006; Sanz 2000), and no bilingual advantages on the other hand (Ghezlou et al. 2018; Hopp 2019; Lechner & Siemund 2014a; Şahingöz 2014; Siemund & Lechner 2015), plus the mixed findings presented in Maluch & Kempert (2017) and Maluch et al. (2015), we may ask why we find such remarkable differences. First, we may want to ask what all the studies that found bilingual advantages have in common, before we can decide if these results are generally true for all bilingual learners.

Noticeably, the three studies mentioned first, all used participants that come from official bilingual regions where both languages, i.e. Spanish and either Catalan or Basque, have a high standing (Cenoz & Valencia 1994: 197-198; Sanz 2000: 26). The authors acknowledge that the special social situation found in such bilingual communities clearly adds to the results (Cenoz & Valencia 1994: 204-205; Sanz 2000: 38). Both languages have a high social value in these areas and children receive formal education in both languages. However, the authors refrain to extend their findings to all bilingual situations. Sanz, for example, is careful with generalizations (2000: 38) and Agustín-Llach (2017) also states that these are mere tendencies and that no general “bilingual superiority” could be attested in her study (Agustín-Llach 2017:

9). This clearly shows that they limit their findings to their specific bilingual situations.

Furthermore, later, Cenoz (2013: 77) states that one cannot easily generalize because

“language acquisition is a complex phenomenon that is also influenced by many other factors.”

Hence, Cenoz mentions, and here we come back to what we have already seen in the previous section, that not only the number of languages has an influence, but that further variables, such as the community and especially socio-economic status, also play important roles (Cenoz 2013: