Navigating the German school system when being perceived as a student ‘with migration background’: Students’ perspectives on linguistic racism

Standard language ideologies, as a construct characterizing presumably unmarked and stable ways of speaking or writing, pervade all aspects of social life, and schools make no exception. The classroom realities in Berlin, our field of investigation, remain fixed on idealized notions of (White) monolingual standard German. On the basis of eight interviews with multilingual women aged 16-23 who speak German and Turkish, we show how young women who are constructed as having a ‘migration background’ partially align with, but also challenge, teachers’ expectation s regarding their use of German. Specifically, we show that because of their perceived ethnicity, the interviewees are viewed as ‘having an accent’ or ‘using non - standard German’ across contexts. Based on these findings, we argue for a renewed focus on addressivity . Following Flores and Rosa (2015), we propose that in order to value heteroglossic repertoires, a shift from the speakers to the addressees needs to take place.


Introduction
Standard language ideologies (Lippi-Green 1994), as a construct characterizing presumably unmarked and stable ways of speaking or writing, pervade all aspects of social life. The classroom realities in Berlin, our field of investigation, make no exception. On the one hand, the curricular framework for Berlin's schools (Rahmenlehrplan Teil B 2015: 10 1 ) acknowledges heteroglossic repertoires and different communicative choices, including non-standard language use (Wiese et al. 2015). Educational settings, on the other hand, still serve as repositories of standard language ideologies, including both the fact that standard (written) languages are highly valued and the fact that multiethnolects and (some) multilingual practices are not.
In German schools indeed, multilingualism is still treated as not desirable (Wiese et al. 2015) when it comes to so-called 'languages of migration' (Migrantensprachen) (Allemann-Ghionda et al. 2010: 8). Previous studies on how students perceive languages other than German and other multilingual practices have shown that 500 students in tenth and eleventh grade (10-11 years old) not only rarely indicated Russian, Turkish, Polish, and Arabic as likeable, but even frequently explicitly mentioned them as dislikable (Plewnia & Rothe 2011: 229). English, on the other hand, which is associated with elite bilingualism (Fuller 2012), or French, seen as a "world language" (Weltsprache) (Allemann-Ghionda et al. 2010: 8), are associated with advanced foreign language skills (Morek 2018: 258). The distinction between 'languages of migration' and others (e.g., those perceived as elite, advanced) is however a tricky one that depends on the addressees' view. From the perspective of a German speaker who speaks only German at home, is Italian less a language of migration than Turkish? We see already how the construction of whose multilingualism is desirable or not pertains to (internalized) raciolinguistic ideologies "that conflate certain racialized bodies with linguistic deficiency unrelated to any objective linguistic practices" (Flores & Rosa 2015: 151).
At the centre of this paper is the study of the impact of the addressees' language ideologies on multilingual speakers of German and Turkish. On the basis of eight interviews of about one hour with women aged 16-23 who regularly meet in Berlin-Kreuzberg, a district typically perceived as populated by multilingual speakers, we challenge additive approaches to education which emphasize the need to give the students the "codes of power" (Delpit 2006) to empower themselves. The first part of the analysis is devoted to the speakers. We focus on the women's ambivalent positioning toward their linguistic practices, and show how young women who are constructed as having a 'migration background' partially align with, but also challenge expectations regarding their use of German.
In the second part of the analysis, we switch to the addressees. We question the effects of "appropriateness-based additive approaches to language education" 4/39 (Flores & Rosa 2015: 151) on students perceived as having a 'migration background' as they are often subjected to negative evaluations. Specifically, we show that because of their perceived ethnicity, the interviewees are viewed as 'having an accent' or 'using non-standard German' across contexts, especially at school. Our interviewees repeatedly say that depending on social information, including stereotypes, people and especially their teachers report hearing a 'migration background' or supposedly deficient language use. Based on these findings, we argue for a renewed focus on addressivity. Following Flores and Rosa (2015), we propose that in order to value heteroglossic repertoires, a shift from the speakers to the addressees needs to take place. In doing so, we argue that while giving students with a 'migration background' the tools to navigate heteroglossic repertoires is still necessary, these speakers remain subjected to judgements on their ways of speaking, no matter how competent they are in standard German.
Section 1 presents the sociolinguistic reality of Berlin as a diverse city where various repertoires co-exist. In section 2 we outline the political and academic debate on the term 'migration background'. We then describe the interview setting and the participants, Martina Oldani's role as an interviewer as well as the nature and structure of the interviews in section 3. In section 4 we analyze the reported experiences and reflections of the participants with respect to public perception and the role played by school in shaping their linguistic identity with regard to a continuum ranging from collusion to contestation (Martin-Jones & Heller 1996). We show how instances of collusion such as the strong focus on formal correctness at the grammatical level coexist with moments of contestation such as the refusal to be complimented in contrast to an imaginary group of peers 'with no migration background'.
In section 5 we consider how the interviewees' stances offer points of reflection in the educational realm for tackling the biases toward non-standard repertoires and the linguistic insecurity arising therefrom. Going further, we plead for a shift from the speakers to the hearers and educators, showing that speakers already invest their fluid repertoires in empowering creative ways but still face discrimination because of the raciolinguistic language ideologies brought upon them by public and educational discourse (Flores & Rosa 2015). While meaning is co-constructed in interaction, emphasis is too often put on the speaker rather on the addressee. In this "overly simplistic characterization of communication", "the listener is relieved of any responsibility in the communicative act, and the full burden is put directly on the speaker" (Lippi-Green 1994: 184). Although our interviewees are aware of this imbalance, they still carry the "communicative burden" (Chaparro 2014). We show how young speakers who are repeatedly perceived and describe themselves as having a 'migration background' depict their linguistic practices, including 'mistakes' compared to standard German, as an inherent part of their identity. Based on these findings, we discuss implications for teachers. We finally conclude that the only way to critically foster the participants' heteroglossic repertoires is by challenging the addressee's listening skills.

Berlin's diverse sociolinguistic reality, heteroglossic repertoires and standard language ideology
The (socio)linguistic reality of Berlin is characterized by "linguistically diverse innercity neighbourhood[s]" (Wiese et al. 2015: 6). Its neighbourhoods or Kieze have long lost their relative ethnic homogeneity and turned into highly layered and stratified globalized neighbourhoods (Schneider, Heyd & von Mengden 2019). In these dynamic networks, language practices as they are deployed in interaction can be described as repertoires, understood as the dynamic and constantly evolving pool of semiotic resources that individual speakers collect throughout different phases of their lives. Initially coined by Gumperz (1964: 138), linguistic repertoires are defined as the "the arsenal [a speaker can chose] in accordance with the meanings they wish to convey" at the intersection between "grammatical and social restraints". Integrating semiotic resources, communicative repertoires are "the collection of ways individuals use language and literacy and other means of communication (gestures, dress, posture, or accessories) to function effectively in the multiple communities in which they participate" (Rymes 2010: 528). The term repertoire conveys the idea that individuals, and not only groups, or not only as belonging to a group, deploy various resources in different contexts. In her critical reflection on the notion of (ethnolinguistic) repertoire, Benor (2010: 161) suggests integrating both perspectives and keeping the group level in order to (a) take into account the group's history, ideologies and boundaries as well as variation within the group, and (b) to enable the comparison between groups. She also shows that the focus on the group level allows for a better understanding on whether people who use a specific repertoire are then perceived as speakers of a specific variety.
When talking about people with a 'migration background', we do not mean that they consist of a homogeneous group, nor that this feature is the most distinctive one. In the same line, we do not make any claims about whether our interviewees can be described as members of the same (ethnic) group based on their use of specific linguistic and semiotic resources. Indeed, we are not directly interested in the interviewees' linguistic practices. Rather, we look into how the young women we interviewed conceptualize their heteroglossic repertoires, i.e. how they discursively construct their ways of speaking and where they draw boundaries. In that sense, the concept of heteroglossic repertoires is speaker-oriented insofar as it does not take "individual languages or varieties as its starting point, but the experiencing subject with his or her [their] multilayered linguistic repertoire" (Busch 2015: 342).
Against the high variety and flexibility of individual repertoires, standard language ideology is the "bias toward an abstracted, idealized, homogenous spoken language which is imposed and maintained by dominant bloc institutions and which names as its model the written language, but which is drawn primarily from the spoken language of the upper middle class" (Lippi-Green 1994). Importantly, the notion of a standard is a myth disguising the fact that all language productions vary (Lindemann & Moran 2017: 651). Consequently, the language of speakers perceived as non-native, non-upper-middle class and/or non-White speakers is most likely to be noticed and 6/39 described despite the fact that any speaker may use language that deviates from the idealized standard (Lindemann & Moran 2017).
As standard language ideology is not inherently tied with monolingual ideology, we do not equate people with a 'migration background'-a loaded term, as we will see-with multilingual speakers. Rather, we will show how the connection between the two is subtly taken upon by the interviewees, and how 'Germanness' is often associated with monolingualism, while having a 'migration background' means navigating between heteroglossic repertoires, including different languages.

2.
Being constructed as a student with 'migration background': statistical reality or problematic denomination?
Contrary to its academic recognition in the US context, notably in the research on African American English (AAE) (Holliday 2019: 2), the term race is not widely used in German-speaking or German-based sociolinguistics. The cautious use of the term can be explained by Germany's history and by its relative absence in public discourse apart from right-wing parties-which does not mean that it is not taken up through different categories.
One of the central categories pervading Germany's construction of otherness is the reference to a 'migration background' (Migrationshintergrund), a term that enables people to point to a perceived ethnicity based on the migration trajectory of either the person themself or their parents and grandparents. The term 'with migration background' (mit Migrationshintergrund) is an ideologically loaded construct. Introduced as a political-not a scientific-substitute to 'guest workers' (Gastarbeiter*innen) in 2005, it corresponds to a change in how migration is taken into account in German population statistics. Until 2005, German population statistics used only the criterion of nationality when referring to immigrant populations and their descendants. This meant that people who had acquired German citizenship disappeared from the category of 'foreigners' (Ausländer*innen) who had immigrated, but who, after many years of residence in Germany, ultimately became German (Mannitz & Schneider 2014: 84-85). For the 2005 microcensus, the Federal Statistical Office implemented for the first time the long-standing demand of migration research to record the country of birth of the respondents and that of their parents. 'Migration background' thus refers to people who have actively migrated (first generation) or to children born in the host society with at least one migrated parent (second generation).
Critics against the term 'migration background' have however been expressed, first on the basis that it conflates different populations, especially in terms of nationality, second, based on the idea that its circulation and discourse have become a means of othering. Similar to what has been observed in the Netherlands, having 'a migration background' has become an ethnic categorization representing "social types with specific associations and imagined characteristics, such as a style of dress, sense of humour, or way of speaking" (van de Weerd 2019: 248). The term 'migration background' is controversial because it is also used in exclusionary, derogatory, and 7/39 stigmatizing contexts (Rösch 2017: 48). Scarvaglieri and Zech (2013: 201) argue, on the basis of a corpus study, that the term, 'migration background' is often used to "distinguish between 'real Germans' and people who are officially German, but are not felt to be part of the German community [...] [and who] present challenges for the majority because they are not sufficiently integrated and need social support". People constructed as 'with migration background' are often relegated to be the "others in migration" (Migrationsandere) (Mecheril & Dirim 2010: 17) (for similar comments in the Dutch context, see van de Weerd 2019: 248). In the end, the label 'migration background' has somehow replaced the term 'foreigner' (Ausländer*innen), which is also negatively connotated (Harr, Liedke & Riehl 2018: 5).
In other words, even if the words race (Rasse) is not used in Germany, this does not mean that racial categories are not in place. We can draw a parallel with "colorblind racism" as "the ideological armor for a covert and institutionalized system in the post-civil rights era" which does not resort to ostensibly racist "facile arguments" based on "biological and moral inferiority", but maintains "white privilege without fanfare" by "otheriz[ing] softly" (or apparently softly, we could add) (Bonilla-Silva 2018: 11-12). As Mannitz & Schneider (2014: 84-85) convincingly argue, "pseudoproblem descriptions from the 1990s, such as 'this school has a high proportion of foreigners'-where distinction-conscious parents immediately know what this is supposed to say about school quality-are translated 1:1 into the new terminology ['with a migration background'], which also has the advantage of appearing factually appropriate and politically correct" (our translation from German).
Regarding languaging more specifically, the construct 'migration background' brings together two related aspects: as seen until now, the (presumed) migration history, but also "monoglossic language ideologies" (Flores & Rosa, 2015). In this sense, the category 'with migration background' has become a way for "the white speaking and listening subject" (Flores & Rosa 2015: 151) to position themself as an idealized norm. As Gramling (2016: 26) explains, the first decades of German's mass labor migration are characterized by a relative indifference toward the languages used by migrants. 'Guest workers' (Gastarbeiter*innen) from Turkey in particular were not incentivized to learn German as a second language as they were not meant to stay in the host country. It was only in the early 2000s that "a major shift in the state's concept of immigration ushered forth a progressivist naturalization discourse with linguistically retrogressive consequences" (Gramling 2016: 26). This "apparently post-ethnic logic for naturalization" now means that citizenship is acquired "by territorial birthright", while until Germany's 1999 Citizenship Law and its 2005 Immigration Law, citizenship was strongly tied to "ethnic inheritance alone" (Gramling 2016: 26).
The label 'migration background' is now often used as a synonym for students who learn German as a second language, speak at least one so-called 'language of migration' at home, and are supposedly "deficient" in German (Mannitz & Schneider 2014: 84-85). For this reason, a new terminology has been introduced in the Berlin school landscape: students of non-German linguistic background (Schüler*innen nicht-deutscher Herkunftssprache) or the unfortunate related expression 'students of non-German origin' (Schüler*innen nicht-deutscher Herkunft), thus putting an emphasis on the assumed origin of the family instead of linguistic socialization (Mannitz & Schneider 2014: 84-85). In order to avoid these infelicitous distinctions, Duarte and Gogolin (2013: 8) differentiate between multilinguals in their everyday life and multilinguals at school. Based on this differentiation, students who speak mostly German but also learn, say, English and French at school, are considered 'multilinguals at school', while students such as out interviewees are 'multilinguals in their everyday life'.
To sum up, the term 'with migration background' is a political label aimed at targeting a specific yet vague statistical category of people, which has now turned into an apparently politically correct way of pointing to the 'others' based on either multilingualism, presumed ethnicity, or both. While using the term, we recognize the risk of reproducing negative connotations but decided nonetheless to acknowledge that this is how the interviewees repeatedly refer to themselves and discursively construct their identities. With the use of scare quotes we intend to signalize the problematic nature of the term and to explicitly distance ourselves from its racist connotations.

The interview setting and the participants
Section 3 presents the setting in which the participants were recruited and the interviews took place, the participants' selection, our positionality during field work, and finally, theoretical and methodological accounts on interviewing as a method.

Field site
The interview setting is a women's center in the district of Friedrichshain-Kreuzberg at the center of Berlin, "one of the most multicultural neighborhoods", where individuals from "more than 180 nationalities" cohabit 2 . In 2019, the population of a 'migration background' amounted to 129.888 people, which is almost half of the population of the district (290.386) and 7,7 % of the overall Berlin population 3 . In Friedrichshain-Kreuzberg, the population of Turkish descent amounts to 20,7% in 2019. The interviewees are mostly students of Martina who participated in a program offered to young women aimed at facilitating their educational path. The program is funded by state funds such as the Senate Department for Health, Care and Equality, Department for Women and Gender Equality (Senatsverwaltung für Gesundheit, Pflege und 2 https://www.berlin.de/special/immobilien-und-wohnen/stadtteile/kreuzberg/, accessed on 27.12.2021. 3 Amt für Statistik Berlin-Brandenburg, cited on https://www.berlin.de/bafriedrichshain-kreuzberg/ueber-den-bezirk/zahlen-und-fakten/, accessed on 27.12.2021. Gleichstellung, Abteilung Frauen und Gleichstellung), as well as private funds such as the Pfefferwerk foundation, whose aim is to "promote diversity and equal opportunities" (Unser Anliegen ist es, Vielfalt und Chancengleichheit zu fördern 4 ), and the Berliner Sparkasse foundation, a local bank.
The program is addressed to "young people of color, young Black people, and young people with migratory background or seeking asylum" (Jugendliche of Color, Schwarze Jugendliche und Jugendliche mit Flucht-oder Migrationsgeschichte). It decidedly targets inequalities in the educational system, and aims at supporting young women in their educational path by offering 1 to 1 lesson with teacher trainees free of charge. The lessons can be in different main subjects, typically German, English, math, and biology. Moreover, the women's center provides a space where students can meet, study, get advice or simply hang out together. Apart from the target groups mentioned above, there are no selection criteria, so that every woman who enrolls in the first school week will get a spot depending on availability.
The projects has existed for several decades and is well established in the neighborhood, which explains why many young women, including our interviewees, had heard about the project from sisters and cousins. The local implementation, together with the fact that often several family members have benefited from the program, creates a familiar context. Some of the teacher trainees have been students in the projects themselves (such as one of the interviewees, Ruja).

Participants' selection
The eight interviewees, aged 16 to 23, were born and raised in Berlin and speak German fluently. They describe themselves as speaking more than one language, including German and Turkish as languages used every day (both L1), and English, French and Spanish, rarely used outside school settings. The specific combination of languages was not a criterion when selecting the girls for the interviews.
Five out of eight interviewees had been coached in the aforementioned program for at least one year at the time when the interviews took place. Aylin, Fiona, Edna and Kevser 5 had been supported while preparing for their Abitur (final school degree which grants access to higher education) or during their apprenticeship (Edna). Esra and Cynthia frequent the program but have not been Martina's students. Finally, Ruja had been a student and took part in the program some years ago, while today she is herself a teacher trainee and colleague in the program.
The program is aimed at young women only, which enables us to explore the intersection between ethnicity and gender. As Martina's 1 familiarity and long-term commitment in the program (two years) are crucial to make the interview setting a safe space, we did not try to recruit any male participant outside of the program. Further, in their reflection on the use of 'correct' German against standard language ideologies 10/39 in educational settings, the interviewees do not directly address gender as a possible variable to take into consideration. For this reason, while we recognize that the allfemale composition of our interviews could be complemented by other studies, we believe that the claims made in this paper are not necessarily to be read under a gender lens.

Author's positionality
Martina, a White woman in her early thirties who grow up bilingual speaking Italian and German but identifies as nonnative speaker of German, is currently working in secondary education in Berlin. At the time of the interviews, she was involved in the program as a teacher trainee. She worked as a freelance teacher 3-5 times a week in the afternoon, offering German and English lessons to young women. The courses took place individually once a week, in some rare cases twice a week.
Due to the relaxed atmosphere at the women's center, the regular encounters as part of the program, and the emphasis on non-hierarchical relationships within the program, Martina most likely does not represent the formal educational system in the eyes of the interviewees. Moreover, in order to mark the difference with the private lessons, the interviews took place in a side room next to the common rooms where the lessons take place. This does not rule out, however, that Martina also represents an element of school authority and still is regarded as a teacher while also having a different ethnic background than the interviewees.
When asked "What kind of German are you speaking with me?" (Was für ein Deutsch sprichst du mit mir?), Aylin provides a nuanced description of the repertoires mobilized with Martina: Example 1: "I'm not scared to make mistakes" (Aylin_2019.08.22_32:44:-33:33) 6 1 AYLIN ich mein ich kenn dich jetzt auch schon länger 2 und viel besser 3 und ich kann dir auch vertrauen in der hinsicht 4 und weil wir uns auch besser kennen 5 fühl ich mich auch so gelassen 6 also ich seh dich also natürlich seh ich dich irgendwo 7 als eine lehrerin 8 also du hilfst mir auch mit lernen und alles drum und dran 9 aber du bist schon mehr so (-) also meine familie 10 dann kommen meine freunde und dann kommst wirklich du 11 dann bist du in dieser ebene und da bin ich auch so gelassen 12 also ich habe keine angst so fehler zu machen beim reden 6 Throughout the paper the excerpts from the transcripts will be indicated as follows: (first name of the interviewee (modified)_date of the interview (year.month.day)_minutes of the audio file. See 3.4. for the transcription conventions.

11/39
13 ich würde es dir auch nicht übelnehmen wenn du mich verbessert 14 ich freu mich 1 AYLIN I mean I've been knowing you for some time now 2 and much better 3 and can trust you in this respect 4 and because we know each other better 5 I also feel relaxed 6 well I see you well of course I see you somehow 7 as a teacher 8 well you help me also with studying and all that 9 but you are really more (-) well my family 10 then there are my friends and you come next actually 11 you are on this level and I am relaxed with you 12 well I'm not scared to make mistakes while talking 13 I wouldn't be offended if you'd correct me 14 I'm happy An intricate aspect of qualitative interviews involves the relationship between interviewer and interviewees. The social desirability effect has been described as "normconsistent behaviour that is deviant from spontaneous language production" (Krug & Sell 2013: 75). Thus, different, possibly changing assumptions could inform interviewees' alignment. On the one hand they could assume that as a teacher (l. 7) assigned with a certain authority on language use, as she is able to "help" (l. 8) and "correct" them (l. 13), Martina may embody prescriptive norms on language use. On the other hand, the interviewees might as well expect Martina, as a nonnative and multilingual German speaker who they have been known for a long time (l. 2, 5) and "trust" (l. 4) and situate at an informal and relatively intimate level just after family and friends (l. 9-11) to be open to different German varieties, and more specifically to a more "relaxed" register (l. 5, 11).

Interviewing as method
The interviews took place in German and were audio recorded. In order to make room for the interviewees' knowledge, experiences and unelicited accounts, the interviews were conducted as semi-structured interviews revolving around a loose set of topics: (1) biographical information, (2) linguistic repertoires (3) outside perception and (4) own positioning. Although we acknowledge the participants' fundamental heterogeneity, relying on a set of prompts in a similar order enabled us to achieve consistency across the interviews and to identify recurring patterns (see 3.5). In exploring language ideologies, we consider how macro frames are (re)constructed discursively in interaction. While our focus on the interactional level is explained by the nature of our data, we do not mean that macro and micro levels can be analytically 12/39 separated from each other or even co-exist as separate entities, "as if social life simply unfolded in more or less intimate, proximate, local, grounded or contained situations" (Carr & Lempert 2016: 8). Rather, the distinction enables us to explore how language ideologies are produced and reproduced locally by social actors during the interviews as communicative events, following "the premise that attitudes are not static, i.e. they are not fixed in the minds of individuals and easily retrieved" (Liebscher & Dailey-O'Cain 2009: 217), but co-created in the interaction while being informed by "systems of power" (Cavanaugh 2020: 55).

Qualitative data analysis
The qualitative data analysis unfolded in two main steps. First, Martina transcribed the data loosely. After the first round of content-wise transcription, she classified extracts according to broad categories such as family, friends, school, job interviews, medical appointments, etc. She then visualized a core set of topics that she systematized as clusters, which was followed by detailed transcripts of selected extracts and their translations into English. The detailed transcription follows the principles of interactional sociolinguistics (Schiffrin 1994;Gumperz 2003;Jaspers 2012). The transcripts account for utterances and not sentences, indicates silences and hesitations, and renders variations in pronunciation (Selting et al. 2009: 359). Colons indicate a lengthening 7 , emphasis is marked by capital letters 8 , pauses are indicated as (.) is they are up to 0.2 second long and as (-) if they are 0.2 to 0.5 seconds long, otherwise with numbers (2.0), paraverbal phenomena are indicated between double brackets ((laughs)), and omissions in the transcript as ((…)). Martina finally drafted the structure and analysis of the paper (which was originally her MA's thesis).
In a second step, Naomi worked with the material again. Crucially, in this phase we worked independently which allowed for our subjective interpretation not to be influenced by our co-author from the onset. We found the process of collaborative, yet independent coding, especially illuminating, as Martina conducted the interviews and knew the participants closely, while Naomi relied exclusively on the audio files. In this second phase, Naomi systematically documented the transcriptions and the codifications, which she rendered in a table. The first columns contain the date of the interview, its duration, the interviewee's name, their pseudonym, and then the beginning and end of the extract considered (e.g. 32:44:-33:33 in (1), see footnote 6). The next columns are devoted to the loose transcript, the detailed transcript, the translation into English, and finally the topics. This visual representation enabled us to see all the interviews at once (instead of having them in different files) and to follow our work's progression easily: we could see at a glance whether we had transcribed only approximately, or if we already had achieved the second round of detailed 13/39 transcription. Moreover, it allowed for a smooth workflow by giving us an overview of all data before the second round of analysis. Crucially, we consider that this double step is an inherent part of the analysis, as transcription is interpretation (Bucholtz 2000).
After the phase of "data reduction", the critical step was to transform "an inchoate corpus to a systematically organized set from which a subset can document representative trends" (Smagorinsky 2008: 397). We indicated next to the extracts, which we divided as small as possible (one or two minutes), which theme was predominant to us, and came to the following categories, divided into subcategories. We presented categories and subcategories in two separate columns in the table in order to allow for a quick retrieval of broad themes. In the subcategories, we used the participants' wording (for instance, 'good/bad German') and tried to group similar instances together (for instance, we classified occurrences of 'correct/incorrect German' as 'good/bad German' although it is not strictly the same and we scrutinized them differently in the analysis). By merging our clusters, we came to the following representation:

Broad topics
How the topics are handled language(s) used German, English, Turkish, good/bad German, Kiezdeutsch, Denglish language use multilingualism, foreign languages, code-switching, colloquial language, 'high German' (Hochdeutsch), swear words, native language judgements on languages from the in-group easy/difficult, positive/negative, same/different, (un)important, fluent, funny, understandable, weird, average, exaggerated, antisocial, (un)known, male, cool judgements on languages from the out-group age, environment, personal situation, nationality Figure 1: Representation of the themes and subthemes we identified in our data While some topics were expected due to the nature of our questions (for instance, we explicitly asked for judgements on language decay or on multiethnolects), others surprised us and enabled us to show the central-and, until now, relatively unexplored-role that gender plays in the construction of heteroglossic repertoires in Berlin (Truan & Oldani 2021). For the present article, we analyze the clusters around the use of varieties of German in particular. We put them in relation with experiences in educational settings, including the participants' exchanges with teachers and fellow students. We also consider similar institutional settings such as the university or doctor's appointments that may be exhibit comparable power relations.

Examining Agency in Language Practice: Collusion and Contestation
In the first part of the analysis, we focus on the speakers. In navigating between the standard language ideologies represented in the educational system, the interviewees' stances oscillate between moments of collusion and contestation (Martin-Jones & Heller 1996: 4) in relation to the unmarked and dominant monolingual norm of 'Germans', understood as 'Germans with no migration background'.
On the one hand, people may engage in practices of collusion by maintaining and possibly reinforcing standard language ideologies, even outside the classroom. On the other hand, people may resist and discursively challenge the legitimacy of such hegemonic views and show how contestation can take place. While it is useful to examine discursive moments of collusion and contestation separately in order to better understand the mechanisms and arguments employed therein, both analytical categories are best understood as part of a continuum. It is thus likely that individuals do not always and consistently assume an affirmative or dissenting position.
We show that the nonadherence to the two polar opposite choices of collusion and contestation frames the experiences of students 'with migration background' in interaction. Although all born and raised in Germany, the interviewees aim for 'sounding (more) German' by improving their grammar and 'losing their accent' that indexes ethnicity to others. At the same time, they refuse to align with constructs of 'Germanness' by embracing a repertoire that is not theirs. The feeling of being inbetween contestation and collusion thus mirrors the feeling of being a gendered, racialized, and educated subject who is conforming to German educational norms.
15/39 4.1 Aspiring to 'Germanness': "I would really like to speak a little more like a German" In this section, we show that a standard language ideology becomes visible primarily as participants talked about grammar and phonology. The participants do not share a consistent notion of what grammar is, but all consider that (German) grammar must be acquired. These various views on grammar demonstrate the subjectivity of heteroglossic repertoires, which is why the willingness to improve one's grammar is closely linked to 'sounding more German' rather than aligning with standard varieties per se: Example 2: "I would really like to speak a little more like a German" (Aylin_2019.08.22_28:36-29:52) 1 AYLIN grammatik ist und war nie meins 2 DAS würd ich verbessern wollen weil (2.0) 3 ich find's irgendwo schon ehm richtig dass man 4 fehlerfrei deutsch sprechen SOllte ((…)) 5 also ich spreche meinerseits also meiner meinung nach 6 spreche ich so deutsch dass man mich noch versteht 7 aber ich würde so gerne so (2.0) 8 bisschen mehr wie so ne DEUtsche sprechen 9 so mein akzent vor allem so 1 grammar has never been my thing 2 THAT I would like to improve because (2.0) 3 I DO think it's right that one 4 should be able to speak german flawlessly ((…)) 5 I mean I think I for my part well in my opinion 6 I speak german in a way that people are just able to understand me 7 but I would really like to speak a (2.0) 8 little more like a GERman 9 well mostly my accent As she affiliates with school norms (collusion), Aylin, who was approaching her final exams and about to successfully conclude her schooling attaining the German Abitur at the time of the interview, explicitly states that grammar has never been her strength, going as far as to describe her German as barely understandable (l. 6, more on the section on linguistic insecurity). As she describes her language use, Aylin presents her German as insufficient on two main grounds: first, as implicitly deficient, as it needs to be "improved" (l. 2); second, as implicitly different, as it is not the German Germans speak. Importantly, Aylin does not only wish she would to speak German without making mistakes (l. 4) because it would be inherently "right" (l. 3), but also-and this is crucial for our interviewees-because she would like to "sound more German" (l. 16/39 8), and, going further, to be perceived as "German": "I would really like to speak a little more like a German" (l. 7-8). This example is a first instance of how nationality ('being German') is tied up with language use ('speaking like a German'). We suggest that 'Germanness' is not, however, a category based only on nationality, as all interviewees, born in Berlin after 2000, hold the German citizenship and have been socialized in the German capital all their life. Rather, speaking "more like a German" (l. 8) is an ethnic categorization through which multilingual German-Turkish Aylin implies that she does not perceive herself as belonging to the unmarked monoglossic majority. Doing 'being a German' "today is to demonstrably speak like one-by displaying discursive and communicative competence in German" (Gramling 2016: 27) by drawing on a monoglossic norm. It remains open whether Aylin considers 'sounding German' as an extra skill to add to her multifaceted heteroglossic repertoire, or as one to replace-or erase-her current way of speaking by aligning with linguistic behavior favoring homogeneity. In any case, Aylin takes upon dominant discourses in the German public sphere equating 'Germanness' with 'German only' and thus with monoculturalism/monolingualism and Whiteness-although the term remains controversial in Germany (see Section 2). For this reason, our interviewees never selflabel as 'Germans' (see van de Weerd 2020: 358 for similar observations in Dutch classrooms).
Similar to Aylin, Ruja underlines the importance of speaking without "grammar mistakes" (l.5) although she believes that they only occur when she finds herself in a context in which she needs to speak "pure German" (l.1-2) or "only German" (l.3): Example 3: "then I start to stammer or I make grammar mistakes" (Ruja_2020.01.08_03:54-04:16) 1 RUJA fällt mir das voll schwer auf so nur REInes DEUtsch zu reden ((…)) 2 wenn ich versuche nur REInes deutsch nur zu reden und also 3 ich muss jetzt drauf achten dass ich NUr DEUtsch rede 4 dann (2.0) dann ich fange an zu stottern 5 oder ich mach grammatikalische fehler 6 obWOHl ich sie eigentlich gar nicht machen würde 1 RUJA it's hard for me to speak only PUre GERman ((…)) 2 when I try to speak only PUre german and well 3 now I have to pay attention that I speak GERman ONLy 4 and (2.0) then I start to stammer 5 or I make grammar mistakes 6 even THOugh I wouldn't make them otherwise Based on the interview, we suggest that "pure German" (reines Deutsch) stands in contrast to codeswitching between German and Turkish, as Ruja explains later that she does not speak "pure Turkish" (kein reines Türkisch), but a mix (Mixmax). Speaking pure German would thus imply speaking only German (nur Deutsch, l.3), and thus complying to a monolingual language ideology. A second possible interpretation, less likely in this specific extract but which occurs earlier in the interview, is the idea that "pure German" points to a formal or academic register, as Ruja otherwise states that she speaks "pure German" only at the university. It thus seems that the notion of "pure German" is located at the cross-roads between resisting multilingual practices (which would be "impure") and aligning with highly valued repertoires. Although Ruja implicitly recognizes heteroglossic repertoires as an essential part of her identity, making it difficult to "speak only pure German" (l. 1), the feeling of not being (socially) allowed to use her full repertoire on every occasion causing discomfort. Ruja's awareness about the social desirability and appropriateness of standard German does not help her to produce it, on the contrary she perceives herself as "stammering" and making "grammar mistakes" (l. 4-5).
Because they recognize how the pressure to produce "pure German" is a way to regulate their language use, the interviewees challenge the validity of the norms at a later point in the interview, thus showing exemplarily how moments of collusion at the beginning of the interview are followed by instances of contestation as the interview unfolds: Example 4: "it doesn't matter grammatically so much" (Ruja_2020.01.08_11:05-12:05) 1 RUJA ehm schlechtes deutsch ist so (2.0) 2 mit den ganzen artikelfehlern 3 aber mittlerweile ist es also ich denk je älter ich geworden bin 4 merk ich jetzt 5 DIE sprache ist gar nicht so::: WIchtig jetzt 6 man also ich merke das wichtigste ist 7 das man sich ausdrücken kann ((…)) 8 das ist schon gut genug vor allem für jemand 9 der nicht mit deutschen eltern aufgewachsen ist 10 sondern mit anderen sprachen ist es schon ziemlich gutes deutsch ((…)) 11 weil deutsche lernen ja NUR deutsch zuhause vielleicht aber (.) 12 wir mit migrationshintergrund 13 müssen nebenbei noch eine weitere sprache lernen 14 ja und das fällt einem schon dann schwer und 15 für uns ist es dann schon GUt 1 RUJA ehm bad german is when (2.0) 2 with all those wrong articles 3 but by now it's well I think the older I get 4 I realize that 18/39 5 language isn't tha::t IMPortant now 6 one well I realize that the most important thing is 7 to be able to express oneself ((…)) 8 that is good enough especially for someone 9 who didn't grow up with german parents 10 but with other languages it is quite good german ((…)) 11 because germans probably ONLY learn german at home (.) 12 we with migration background 13 have to learn another language on the side 14 yes well and that is hard 15 for us that is GOOD Ruja, who describes herself and the in-group as "we with migration background" (l. 12), and, per extension, as multilinguals who grew up "with other languages" (l. 10) "on the side" (l. 13), clearly refuses to uphold the idea that using the "wrong articles" automatically means "bad German" (l. 1) 9 . Instead she offers a functional perspective on what 'good German' is, namely the level of German needed "to express oneself " (l. 7). This position is in line with many other statements from the other interviewees, who stress that "good German is that with which you can communicate, so everyone understands what you mean by it" 10 . Adopting this position requires considering internalized standard ideologies from the distance, and it is only as she gets older (l. 3) and has already left the school system that 23-year-old Ruja makes such statements. Ruja's position applies not only to people 'with migration background' (l. 12) or multilingual speakers, but she further specifies that it is especially true for speakers "who didn't grow up with German parents" (l. 9). This critical stance on standard language ideologies, a typical instance of contestation, strongly resonates with what Kevser replied to the question what qualifies as "good German": Example 5: "if one speaks 5-6 languages" (Kevser_2019.08.21_28:47-29:03) 1 KEVSER aber wenn man so 5-6 sprachen kann 2 und man einfach durcheinander kommt 3 oder wenn man sich wenigstens bemü::ht 4 dann ist es auf jeden fall was anderes 5 also dieses sich bemühen ist für mich das wichtigste 6 wenn man sich bemüht zu sprechen dann find ich es 7 dann ist es automatisch schon gutes deutsch 1 KEVSER but if you speak like 5-6 languages 9 The articles refer to the fact that nouns encode grammatical gender in German. 10 Original: "Gutes Deutsch ist das, mit dem man sich verständigen kann, also jeder versteht, was man damit meint" (Cynthia_ 2020.01.30_16:20-16:30).

19/39
2 and simply get confused 3 or if you're at least try::ing 4 then it's definitely something else 5 it's the trying hard that counts for me 6 if you're trying hard to speak then I think it's 7 then it's automatically good german In referring to "Germans" (l.11 in (4)) as a category they disaffiliate from, the interviewees construct Germans as monolingual speakers. They construct themselves (Germans as well but also people 'with migration background') as multilingual ("speak[ing] like 5-6 languages", l. 1), demonstrated by the use of the third-person pronoun man (translated as a generic you): the pronoun encompasses the speaker without being exclusive. The interviewees' multilingual identity, in turn, is accounted for as a possible explanation for "get[ting] confused" (l.2). While the categories multilingualism-as the practice consisting in speaking several languages-and 'with migration background'-which for, the interviewees, refers to their migration trajectory and the fact that their parents or grandparents were not born in Germany -only partly overlap, they are often considered together in the ways the interviewees talk about their lived experience. The importance of multilingual practices enables Kevser to engage in partial contestation: Without fully rejecting the standard language ideologies she has been exposed to, Kevser underscores that standard German is only part of many students' language repertoires and that in order to assess whether someone is a 'good German' speaker, the whole repertoire of a speaker needs to be taken into account-in her case, the fact that she speaks several languages. At this point of the interview, Kevser does not mention heteroglossic repertoires (as in (6) below) but rather multilingual practices. Conversely, disregarding or even devaluating repertoires outside a monolingual 'standard' language means ignoring the complexity of a speaker's linguistic reality.
All participants expressed great interest in improving their German with respect to grammar. The collected data does not, however, reveal a clear picture of what the participants mean when they refer to grammar. Repeated references are made to articles and the grammatical gender of nouns (see e.g. (3) l. 5, (4) l. 2; (10), l. 14) as well as syntax (use of "full sentences": (9) l. 4) and diversified lexical proficiency (for instance the use of technical terms in extracts not cited here). Despite recognizing the social desirability of complying with "pure German" (3), the interviewees also articulate their doubts as to whether this norm should equally apply to multilingual speakers. On several occasions, the interviewees mention multilingualism as a source of confusion and mistakes. The value of formal correctness per se and in particular of grammar is eventually called into question when compared to the significance of making oneself understood. 20/39

(Dis)affiliating from 'Germanness' while performing 'the good migrant student'
Speaking 'correct' German is tightly connected to the idea that speaking and sounding 'like a German' is constructed as desirable and socially valued. Nonetheless, some participants were not willing to forge a divide between an idealized category of monolingual Germans who (should) speak impeccable German and Germans 'with migration background' who do not. The next example, an instance of contestation, exhibits Fiona's resistance to be classified in an imaginary homogenous group of people 'with migration background': Example 6: "Wow, you speak so well compared to the others" (Fiona_2019.11.15_10:30-11:27) 1 FIONA dann sagen die mir irgendwie dass ich richtig gut sprechen kann 2 im gegensatz zu den anderen (2.0) 3 was mich irgendwie wundert weil ich mein (2.0) 4 inwieFErn anders? inwieFErn gut? ((…)) 5 ich weiß nicht ob ich das als kompliment sehen kann aber ehm ((…)) 6 also ich find's irgendwie komisch dass man dann einfach 7 als TÜrkin oder allgemein (3.0) 8 jemand der ehm mit migrationshintergrund ist 9 und dann diese DEUtsche sprache richtig beherrscht 10 dass man dann irgendwie heraussticht 11 dass man irgendwie gesagt bekommt 12 boah du redest voll gut im gegensatz zu den anderen 1 FIONA then they somehow tell me that I speak quite well 2 compared to the others 3 and I somehow wonder cause I mean (2.0) 4 HOw different? HOw better? ((…)) 5 I don't know if I can take it as a compliment but ehm ((…)) 6 I mean to me it's weird that one cannot 7 simply as a turk or in general (3.0) 8 as someone ehm with migration background 9 and really masters this GERman language 10 that you somehow stick out 11 that you're somehow told 12 wow you speak so well compared to the others In rendering judgments on her way of speaking in reported speech (l. 12), Fiona illustrates how others usually express both surprise and admiration ("wow") when hearing her, and how they compare her to "others"-meaning others 'with migration background' (compare with line 7 in example 10). With her wording "how different? how better?" (l. 4), Fiona unmistakeably refuses to be complimented for her good German, thus clearly showing that apparently positive stances on language performance can become objects of contestation. In doing so, the participants refuse to engage in "the idea of 'articulate as an exceptionalizing discourse'" (Alim & Smitherman 2012: 30) and call it out as an instance of "subtextual racism, one that speakers may not even realize that they hold and perpetuate" (Alim & Smitherman 2012: 39). Performing the "good migrant student" as argued by Bartlett (2007: 217) in her study about transmigrant student's bilingual literacies and educational trajectories means to negotiate one's identity in social interactions across classroom contexts as a successful 'migrant student'. This is why Fiona challenges the stereotypical notion behind the compliment by asking why someone 'with migration background' should not speak flawless German (l. 6-10). A speaker can influence how much effort they place in linguistic accuracy but cannot prevent their interlocutor from labelling these skills as the exception that proves the rule. The social construction of the presumed 'rule' according to which people 'with migration background' tend to speak poor German is here constructed as problematic, if not entirely inaccurate.

I'm not on such a level
Similarly confronted with the idealized image of "WOW-german" (l. 6), Aylin's imaginary performance of a "good migrant student" consists of impressing interlocutors (possibly teachers) with striking German skills, especially if being hypothetically asked "wow where have you learned your German?" (l. 7), which implies 22/39 that she is expected not to have learnt it at home. In this case, however, being different (because exceptional, "wow") from the rest of the students 'with migration background' would be welcome, thus presenting a contrast between collusion and contestation. Expected by their teachers not to produce 'correct' German, students 'with migration background' are nonetheless being strongly encouraged to fulfil standard language ideologies if they want to reach academic success. This double bind explains, in turn, why all interviewees alternate between both poles of collusion and contestation.

Linguistic insecurity: "well till today I'm afraid to raise my hand"
In this last subsection, we address stances relating to linguistic insecurity as "a measure of the subject's recognition of an exterior standard of correctness" (Labov 2006: 277) pertaining to "an aspect of social mobility rather than social insecurity". This is why Romaine (2000: 75) summarizes linguistic insecurity as "[t]he behavior of the lower middle class", "governed by their recognition of an exterior standard of correctness and their insecurity about their own speech".
While linguistic insecurity was first described as a class factor pertaining to the speaker's perception of correctness, other definitions have revolved around the idea of inferiority, i.e. the "speakers' feeling that the variety they use is somehow inferior, ugly or bad" (Meyerhoff 2006: 292). The shift to the idea of inferiority allows for a focus on the speaker's positioning, thus making whether they comply or not with social and linguistic norms less directly relevant. Whereas the importance of social class in triggering feelings of linguistic insecurity should not be undermined and probably plays a role in the interviewees' positioning, the participants do not make their social background predominant when accounting for their experiences. In this regard, we thus use the term linguistic insecurity to refer to the speakers' unease toward their own way of speaking, whether their language use could be described as standard or not.
Despite the fact that all interviewees had obtained their intermediate degrees (MSA, mittlerer Schulabschluss) and were preparing for their Abitur or had already obtained it, they all repeatedly mention being constantly corrected, especially by their teachers, which fuels feelings of insecurity: Example 8: "well till today I'm afraid to raise my hand" (Kevser_2019.08.21_9:43-10:57) 1 KEVSER davor wusst ich n davor dacht ich dass ich nie probleme 2 mit der deutschen sprache hatte 3 also ich hab mich jetzt zwar nicht SO:: sicher gefühlt 4 aber trotzdem sicherer als JEtzt 5 und dann kamen diese bemerkungen in der elften klasse 6 üben sie die groß und kleinschreibung 7 üben sie die kommasetzung so 8 und danach hat es angefangen dass ich so ANgst hatte 23/39 9 auch beim sprechen 10 also ich hatte eine zeit lang angst zu sprechen 11 und was falsch zu sagen also bis heute 12 habe ich immer noch ANgst mich zu melden 13 und ehm was zu sagen ((…)) 14 so dass ich während des sprechens ein artikel falsch sage oder so 15 dann hab ich schon gemerkt dass ICh nicht RIchtig spreche 16 also dass ich die deutsche sprache nicht richtig beherrsche 17 mit (.) immer noch mit 18 jahren also ich bin bald 19 1 KEVSER before that I di' before I never thought 2 that I had problems with the german language 3 well I didn't feel THA::T confident 4 but still more confident than NOw 5 and then this comments started in the 11. grade 6 practice spelling 7 practice punctuation and things like that 8 after that it started that I was AFRaid 9 also when speaking 10 well for a while I was afraid to speak 11 and say something wrong well till today 12 I'm still AFRaid to raise my hand 13 and ehm to say something ((…)) 14 that while speaking I use the wrong article or something like that 15 then I did realize that I don't speak COrrectly 16 well that I don't master the german language entirely 17 at (.) still at 18 years well I'll soon be 19 Kevser portrays herself as in constant fear of making mistakes and therefore avoiding participating in class. Importantly, she attributes her linguistic insecurity to her German skills, stressing that she does not experience such feelings in natural sciences classes. Considering that she started her accounts on her German skills on a rather positive note, it seems legitimate to argue that her reflections on school experiences persuaded her to conclude otherwise, as the narrative is marked by the shift this comment has meant in Kevser's linguistic biography in the form of a before/after (before, l. 1; and then, l. 5; after that, l. 8; till today, l. 12; still, l. 13). As these young women pursue their educational path, they become more aware of the standards for speaking associated with the spaces they inhabit, which, in turn, often makes them more self-critical. These apparently incidental anecdotes are pervasive in the interviewees' identity construction: This excerpt presents a moment of collusion to a norm presented here by the German teacher who suggests that Kevser might not be "German"-meaning not 'only German' or not 'monolingual German'-by asking since when she has been living in Germany. In admitting that she would even soon turn 19, the interviewee shows to have internalized such norm to some extent, and that not complying with this norm is considered a personal failure. It is only at the end of the interview that Kevser explicitly engages in contestation when stressing the double standards she has been exposed to by her teacher who failed her exam in German by giving her the worst possible score: "I don't know if it was because I spoke or wrote wrongly or if it was because I'm a foreigner". Kevser suggests that her being a 'foreigner' (Ausländerin) may explain why she has been treated differently by her German teacher, ultimately leading her to leave the school. While Berlin-born Kevser is not, in fact, 'a foreigner', the discourses of othering are so prevalent that they are taken up (for similar observations in Dutch schools, see van de Weerd 2019: 251; van de Weerd 2020: 358), even though the interviewees also deconstruct the racism at play in the everyday encounters they experience in school contexts. This is why, similarly to the repeated instances of 'migration background', the interviewees draw on the negatively connoted categorization 'foreigner' (Ausländer*innen) (Harr, Liedke & Riehl 2018: 5) to rationalize the difference of treatment they are facing.
Once at the university, Ruja still experiences linguistic insecurity, but also challenges its legitimacy: Example 10: "wow then she does know something" (Ruja_2020.01.08_32:11-33:37) 1 RUJA ja Also in der UNi bin ziemlich schüchtern also 2 vermeide versuch also ich versuch einfach nicht viel zu reden 3 weil ich denk einfach die sind vi::el besser als ich in der sprache 4 und die werden mich nicht verstehen ((…)) 5 in den gruppenarbeiten da teil ich SCHOn meine meinung mit 6 und da sind die meisten immer so SCHOn erstaunt 7 so boah also die (3.0) die kann doch was ((lacht)) 8 ja also das hab ich in der uni auf jeden fall gelernt 9 also dass was ich meinte auch mit dem alter was kommt 10 dass die sprache eigentlich dann doch nicht (.) 25/39 so:: viel ausmacht ehm (2.0) 11 dass die sprache nicht so wichtig ist wie der inhalt ((…)) 12 genau also ich hab jetzt gemerkt dass manche die me::lden sich 13 sprechen richtig me::ga gutes deutsch 14 aber die kommen gar nicht zum punkt la::bern la::bern 1 RUJA yes Well at UNi I am also quite shy 2 I avoid try well I try not to speak so much 3 cause I simply think that the other are so:: much better in the language 4 and they won't understand me ((…)) 5 in group works I DO share my opinion 6 and most people react QUIte surprised 7 like wow then she (3.0) does know something ((laughs)) 8 yes I definitely learned that at the university 9 also what I meant that it also comes as I'm getting older 10 that language isn't (.) tha::t important ehm (2.0) 11 that language isn't as important as content ((…)) 12 exactly I realized now that some people ra::ise their hands 13 and speak really su::per good german 14 but then they never get to their point pra::ttle pra::ttle Ruja deconstructs the assumption according to which speaking standard German (here probably academic German) automatically allows for conclusions as to the validity of the content of the utterances. Stressing the importance of content over the form then becomes a moment of resistance (contestation). Ruja's fellow students are surprised to discover that she has much to say, possibly more than what they may "prattle" (l. 14), which highlights the role of the addressee in the construction of 'non-proficiency' based on Ruja's perceived ethnicity: Because they expect Turkish-German women not to be articulate, monolingual Germans express astonishment when they realize that the participants are, in fact, fully able to express an opinion, which reminds us of perceptions of Barack Obama as "articulate while Black" (Alim & Smitherman 2012).
On the one hand, the participants acknowledge the importance of formal standard German and express the wish to improve their skills. On the other hand, there are numerous instances where they contrast feedback and expectations with their everyday multilingual experiences that reflect their proficient use of several repertoires, among them standard German. Despite the interviewees' partial resistance to the expectations and norms they are faced with, linguistic insecurity, leading to nonengaging in active participation in class, raises the following question: how can educational approaches challenge the current situation in Berlin schools? 26/39

How to promote heteroglossic repertoires? A discussion on additive approaches
Section 5 relates the stances and language ideologies from the interview data to the challenges raised by an analytic focus on the speakers. We first show that the participants have a high level of register awareness and discursively reconstruct when 'standard German' becomes advantageous for them (e.g., during medical appointments). These positionings could be interpreted as an argument in favor of additive approaches aiming at helping students navigate between language varieties by adding them instead of cancelling them. One answer to subtractive models that solely aim at increasing students' standard language competences disregarding or even replacing other linguistic resources (Creese & Martin 2003) are additive approaches (Delpit 2006;Cummins 2017) which foster register awareness as a way to develop an understanding of when which heteroglossic repertoires are the most appropriate. While employing different means, additive approaches all refuse the linguistic assumption that students should lose their heteroglossic repertoires to make space for proficiency in the standard language.
In what follows, however, we argue that additive approaches, which implicitly assume that some language repertoires belong to the private sphere and should not be used in public, thus being relegated to repertoires with overt prestige, have concrete and damaging effects on how legitimate speakers feel about their ways of speaking. Specifically, we show that because of their perceived ethnicity, the interviewees are consistently construed as having an accent or even being unintelligible. We argue that we need to refocus on the addressee instead of letting the "communicative burden" (Chaparro 2014) fall on the speaker only, as Lippi-Green already argued based on examples from the workplace, schools, and the media (1994: 184). Our data indeed show that no matter how competently the interviewees navigate their heteroglossic repertoires, they remain subjected to racism.

Language as a weapon: 'Speaking standard standard good German'
Adopting additive approaches in Berlin schools to foster heteroglossic repertoires can be beneficial to students. For one, participants repeatedly signal their wish to improve their standard German skills in addition to already known repertoires, especially in highly institutionalized environments: Example 11: "so that I can really knock him out with my German" (Ruja_2020.01.08_12:10-12:52) 1 RUJA ja also ich würd schon SEHr gern richtig hoch hoch 2 GUtes deutsch sprechen damit ich auch 3 jetzt so wenn ich zum beispiel beim arzt bin oder so 4 damit ich sie auch mal mit MEINEM deutsch 5 die richtig (.) fertig machen kann wenn was passiert 6 MARTINA dich so ein bisschen verteidigen kannst?
27/39 7 RUJA genau dass ich mit meinem (.) DEUtsch die einfach zeigen kann 8 so ehm (2.0) ich hab was in der birne so 9 die schätzen ja einem immer ein nach ehm nach der sprache 10 nach dem DEUtsch ein wie man GUT oder schlau ist 11 und sprache sagt ja nichts über die intelligenz oder so aus like ehm (2.0) that I'm bright like that 9 so they always judge someone by ehm the language 10 like by the GERman how GOOD or smart one is 11 and language actually says nothing about intelligence or things like that Ruja first states that she needs standard German skills (beyond the ones she already has) to be able to defend herself (l. 3-5). Second, while resenting people's assumptions about her intelligence based on her language use (l. 10-11), she is also well aware on how 'speaking standard standard' (l. 1), or, as she rephrases it, 'good German' (l. 2), may help her achieve her goals, thus showing how, for our interviewees, contesting the norm does not mean refusing to comply. In her will to 'knock him [the doctor] out' (l. 4-5), Ruja suggests that she experiences interactions with doctors in conflictive terms. Her wish to be able to defend herself verbally and to sustain her intellectual skills (l. 7-8) further resonates with Delpit's additive approach that goes beyond the mere promotion of the ability to navigate between repertoires when appropriate, as she demands that pupils are taught linguistic "codes of power" as well as their arbitrariness and "the power relations they represent" (2006: 45). The asymmetry of the encounter explains why for Ruja, it probably appears reasonable or more efficient to "tak[e] on the communicative burden" (Chaparro 2014: 44) instead of questioning the person of authority's right to "judge" her (l.9). The interviewees often feel responsible for "mak[ing] themselves understood" (Chaparro 2014: 54), although they also state their readiness to understand anyone, no matter how they speak: Example 12: "I don't really pay attention to how the person speaks" (Kevser_2020.01.08_08:19-08:59) 1 KEVSER die sind für mich so alle gleich mit sprache 2 da gibt es für mich keinen unterschied 28/39 3 mir ist es wichtig 4 dass ich mich mit ihnen gut unterhalten kann 5 ich achte nicht so drauf wie die person redet 6 sondern ob ich mit der sprache klarkomme 1 KEVSER they are all the same for me with the language 2 there is no difference to me 3 what matters to me 4 is that I can chat well with them 5 I don't really pay attention to how the person speaks 6 but rather if I'm getting along with the language Even if they embody the ideal addressee insofar as they focus on the message rather than the form, the interviewees often do not experience a reciprocal language awareness. This is why the shift from 'standard standard' (l. 1) and 'good German' (l. 2) to 'MY German' (l. 4, l. 7) in (10), is symbolic of the importance of making standard repertoires their own. Although she refuses to associate the command of a language with the person's intelligence (l. 8-11), Kevser indicates that the ability to speak 'standard standard' positively impacts her identity-but once it has become 'hers': Example 13: "I mean I still like this language" (Kevser_2019.08.21_32:09-33:02) 1 KEVSER aber was ich jetzt NICHt verändern will einfach ehm (2.0) ((…)) 2 auch einfach diese LUst an der sprache 3 also ich hab ja trotz (.) ich mag ja die sprache trotzdem 4 und ich hab ja trotzdem so ne lust beim sprechen aber 5 genau DAs will ich behalten 6 auch wenn ich weiß ich mach fehler 7 das mach ja jeder heutzutage od macht halt jeder 8 also einfach diese LUst würd' ich nicht verändern wollen an meiner sprache 1 KEVSER but what I WOULDN't want to change is simply ehm (2.0) ((…)) 2 this PLEasure with the language 3 I mean I still (.) like this language anyway 4 and I still enjoy speaking but 5 that's exactly WHAt I want to keep 6 even if I know that I make mistakes 7 everyone does that nowadays or everyone does that 8 well simply I wouldn't want to change this PLEasure with my language By stating "I still like this language" (l. 3), Kevser seems to defend her right to keep her way of speaking alongside with an improved formal standard German. She also 29/39 specifies that making mistakes and hence not complying with standard language is something "everyone does nowadays" (l. 7).
So far the data supports additive approaches to language teaching in school to foster heteroglossic repertoires. Following this line of thought, addressing and teaching more repertoires as well as the ability to switch between them in Berlin's classrooms could be a promising way to go, and even more so as such an approach is quite likely to earn the acceptance of students.

Reevaluating the role of the addressee: "I simply don't believe that you can hear that"
There are, however, some drawbacks to "appropriateness-based additive approaches to language education" (Flores & Rosa 2015: 151). The critical point is the intricate connection between additive approaches and the idea of appropriateness of heteroglossic repertoires for specific spaces and occasions. This is exemplified by several interviewees who describe their way of speaking as inappropriate not only "for strangers outside" (draußen eben mit fremden personen), as Aylin frames it, but in public spaces altogether (Truan & Oldani 2021: 11). The view shared by many interviewees that their German is inappropriate for most public spaces including school, doctor's practice, but also public transportation, may result in public spaces being perceived as "sites of unbelonging" (Muller, Schmidt & Weber 2019: 177).
Let us unpack the possible contradiction and its undesired effects: On the one hand, following additive approaches, students are told that 'their way of speaking' is legitimate and valuable. On the other hand, the acceptance of heteroglossic repertoires remains confined to private, informal interactions (Flores & Rosa 2015: 159;Wei 2018: 159) 11 . Although informal interactions can also be visible (for instance, talking with a friend on the street), our interview data has shown that as young women, our interviewees actively pursue a politics of nonvisibility in public spaces (Truan & Oldani 2021: 11). For our interviewees, making heteroglossic repertoires legitimate only in interactions with friends thus indirectly means accepting them as far as they remain invisible.
A second ground on which additive approaches have been criticized is that they tend put the emphasis on what kind of language the speakers are able to produce and not what the addressee is able to understand (Flores & Rosa 2015: 160). This second aspect echoes Esra's experience in the waiting room of an interview for an internship where she is told that one hears her 'migration background': Esra receives feedback about her German sounding "migrant" by "Germans" (l. 2)here we see, once again, how 'Germanness' is implicitly constructed as monolingual and monocultural. While Esra finds the remark on her "language" (l. 8) "unusual" (l. 13) and "strange" (l. 16), she also describes the way of speaking of others as an "accent" (l. 4), thus reminding us that "accent is how the other speaks" (Lippi-Green 1994: 166). Without fully accepting as legitimate and truthful the assessment made upon her, Esra provides elements of explanation on why listeners may have "heard" a 'migration background': they were unacquainted with her way of speaking. Esra states that nothing special is to be heard in her speaking (l. 15), while she assumes that for her interlocutors, no one else spoke like her "in their view" (l. 19). Even if the expression in deren Augen, literally translated as 'in their eyes', is here used metaphorically ('in their view'), the reference to what is audible and what is observable suggests that the perception of an accent and the categorization of her as 'with migration background', and, going further, 'non German', are based on her appearance. In this respect, Esra seems to resist the widespread belief that Germans 'with no migration background' have the legitimacy to evaluate accents and supports previous findings that showed how people report hearing a 'foreign' accent where there might be none (König 2019: 123)-a phenomenon Lindemann (2002) has summed up as "listening with an attitude".
At first sight this episode seems to have little to say about how to achieve educational equity and the role of additive approaches specifically. Flores and Rosa (2015: 149), however, make an important point when saying that language learners "can be understood to inhabit a shared racial positioning that frames their linguistic practices as deficient regardless of how closely they follow supposed rules of appropriateness". Here, we see that in a formal situation (a job interview) that has been consistently described by the interviewees as one where they speak standard German, their compliance with the "supposed rules of appropriateness" does not prevent them from being othered. Hence, this example shows how appropriateness-based models of language education tend to overlook the ways in which "particular people's linguistic practices can be stigmatized regardless of the extent to which they approximate or correspond to standard forms" (Flores & Rosa 2015: 152).
Our data suggests that taking into consideration the role of the addressee is at least as important as teaching different repertoires and language awareness to speakers. Consequently, the focus of language teaching should not only be on what repertoires speakers can produce, adding to it a critical reflection of the power relations behind prestige and stigmatization, but also on the reception end of it. At the same time the interviewees emphasize the lack of focus on the addressees and their listening skills.

32/39
The fact that depending on social information, including stereotypes, people report hearing a 'migration background' or deficient language use reveals that teaching solely the speakers (language learners) sis insufficient.

Reaffirming one's identity: "I wouldn't want to adapt"
Finally, one last recurring motif deserves attention, namely the tension between adaptation to the norm versus linguistic identity and authenticity, which we will delve into on the basis of one representative example (also see (13)). Reflecting on her German skills compared to some of her peers, Kevser signalizes limits to the extent to which she would like to improve her German in school: Example 15: "I wouldn't want to adapt" (Kevser_2019.08.21_31:00-32:58) 1 KEVSER ich hab ja auch wie gesagt zwei sehr gute freundinnen in der schule 2 die sind auch deutsche also komplett ((…)) 3 so wie die schreiben und so so würde ich es niemals können 4 und so wie DIE ehm ich weiß nicht 5 ich würd mich jetzt auch nicht a:npassen wollen 6 die person ist so wie sie ist und das ist gut so 7 wenn sie's beherrscht ist gut so 8 aber ich glaube es würde nicht zu mir ste [stehen] 9 also es würde nicht GUT an mir aussehen 10 wenn ich die sprache komp [komplett] so 100% beherrschen würde glaube ich ((…)) 11 das bin ich ((lacht)) 12 und wenn ich statt der kuchen das kuchen sage 13 dann bin ich es halt auch 14 ich würd mich da nicht so anpassen wollen 15 aber jeder da wie er ist 1 KEVSER as I said I also have two very good friends in school 2 they are also german totally so ((…)) 3 I would never be able to write like they do 4 and like THEm ehm I don't know 5 but I wouldn't want to a::dapt either 6 a person is the way they are and that's good 7 so if they master it it's also good 8 but I think it wouldn't fi 9 well it wouldn't look GOOD on me 10 if I'd master the language comp [completely] like 100% I think 11 that's me ((laughs)) and if I say the cake [NEUTRAL] instead of the cake [MASCULINE] 13 then it's also me 14 I wouldn't want to adapt 15 but anyone the way they are The interviewee expresses the wish to be able to gain more self-confidence in speaking standard German while remaining true to herself and refusing to "adapt" (l. 5) compared to the "pure Germans" (see (3)) who are here described as "German totally so" (l. 2). She describes it as if mastering German "100%" would not "look good on her" (l. 8-9) insofar as using 'incorrect' German articles is a distinctive feature of her repertoire (l. 12-13). Both the non-readiness to adapt, which can be indirectly read as a challenge to appropriateness-based approaches, together with the pleasure expressed when speaking German (also see (14)), show how deep her identity is intertwined with her individual (non-standard) language use. The fact that the participant associates "master[ing] the language 100%" (l. 10) with adapting to external expectations suggests that she perceives 'correcting' her language use as act of undesired assimilation. Her explicit reiteration "that's me" followed by "that's also me" (l. 11-13) is unmistakeable in this respect and reminds us of Ruja's willingness to improve her German (only) if it becomes "hers" in (13).
After having been constructed as 'others' during their whole educational path, one of the ways to challenge both the expectations posed upon 'their German' as well as their perceived 'non Germanness' is to reinvest their use of German only-and not multilingual and multiethnolectal practices-as the locus where their hybrid identities can unfold altogether. By that, we mean that refusing to adapt and perform an idealized, standardized, and possibly non-racialized German is an act of power for our interviewees, one through which not using 'pure German' is the only way to feel like themselves.

Discussion: Implications for teachers
Based on our participants' experiences, we turn to implications for teachers. The analysis unfolded in two steps: we first focused on the interviewees as central agents of their own practices, and identified instances of collusion and contestation in how they describe their relationship to 'Germanness'. Specifically, we showed that the interviewees' desire to use standardized repertoires in German may be seen as an empowerment tool, which may, at first sight, be viewed as an argument in favor of additive approaches (Delpit 2006;Cummins 2017). From the perspective of additive approaches indeed, being able to critically use varieties of standard German in situations in which it may be beneficial to the speakers (job interview, school, doctor's appointment) may be regarded as a win. Sometimes, the interviewees express difficulty in navigating heteroglossic repertoires freely, as Ruja who says "start[s] to stammer or make[s] grammar mistakes" when she needs to stick to "pure German" (3). Altogether, our data supports previous studies that showed that multilingual speakers, and especially racialized speakers of multiethnolects, are fully able to accommodate their addressee(s) and to adjust their way of speaking accordingly (Bunk & Pohle 2019), as Aylin explains in (7): "when I speak to teachers, I cannot speak as I want".
Complying with the unmarked White way of speaking is not enough, however, to become a full member of the mainstream society. Promoting the speaker's competence only fails to recognize that the judgments on the interviewees' ways of speaking also is the result of prejudices based on their perceived 'migration background'. Refusing to perform the "good migrant student" (Bartlett 2007) becomes an instrument of power that does not, however, prevent the participants from experiencing linguistic insecurity (see (8)). The interviewees often experience racism, for instance by their teachers asking how long they have been living in the country (see (9)), by other applicants at job interviews (see (14)), and by medical professionals who do not take them seriously (see (11)). In the most extreme cases, it may lead them to switch schools, as in Kevser's case. In other words, no matter how eager they are to improve their skills in standard German and how competent they are nonetheless, the participants sometimes have no other choice than literally leaving the site of harm. This is why in the second part of the analysis, we pleaded for a switch to the addressees, showing that the interviewees remain racialized despite their ability to navigate heteroglossic repertoires freely. Although these women 'with migration background' constantly adapt to their interlocutors, both when speaking and listening, they are not faced with the same readiness to share the "communicative burden" (Chaparro 2014). By recentring the students' experiences, our findings thus show that "appropriateness-based additive approaches to language education" are not sufficient to dismantle "raciolinguistic languages ideologies" in German schools (Flores & Rosa 2015: 151).
First initiatives in Berlin have shown that offering teachers "a set of classroom materials to be implemented into their own teaching" (Wiese et al. 2015: 17) in order to foster linguistic diversity opens up promising avenues. In the program crated and evaluated by Wiese et al. (2015: 5), teachers, including experienced ones (rather than prospective teachers) are at the center. The program showed positive effects on the teachers' attitudes towards linguistic diversity (Wiese et al. 2015: 12), yet it did not aim primarily at implementing classroom materials, as those were seen as an addition once the teachers' awareness towards linguistic diversity had been raised.
In highlighting the students' experiences, this paper invites for further research on pedagogical practices able to teach awareness of linguistic diversity (which is partly already the case), but also crucially, foster competence the teachers and students in both speaking and hearing critically different repertoires. The (in)ability to listen without projecting what one may hold true about the speaker's identity is problematic not only among teachers, but also in other groups. Even once enrolled at the university, our interviewees still are met with astonishment: "wow then she does know something" 35/39 (see (10)), which means that some fellow students had not experienced linguistic and ethnic diversity before or were still entangled in their own bias. It is only by fostering everyone's listening skills equally, and probably by beginning at school, that we can empower all students by recognizing power dynamics in language.

Conclusion
In this paper we qualitatively analyzed how multilingual speakers of German and Turkish with a 'migration background' construct their relationship to heteroglossic repertoires. We localized potential sites of harm and/or resilience, thus inviting for a focus on both exploring various aspects of one's identity through the use of heteroglossic repertoires, but also on fostering an open attitude toward unknown or unfamiliar repertoires.
Despite their high degree of language awareness, gendered, racialized, highly educated students have to assess the risks and benefits in complying to standard language ideologies while surviving and thriving in sometimes hostile interactional contexts. Even if they unequivocally condemn teachers and doctors complimenting their German and/or pretending not to understand them, it remains extremely hard, for Turkish-German women, to call out figures of authority. Fostering the use of heteroglossic repertoires thus means departing from correcting students with a 'migration background' to treating them as legitimate speakers by encouraging fluid repertoires and creativity as identity building.
Yet-and this is our central point-focusing on the speakers only is not enough. While giving students 'with migration background' the tools to navigate heteroglossic repertoires is still necessary, a shift from the speakers to the addressees needs to take place, as these speakers remain subjected to judgements on their ways of speaking, no matter how competent they are in standard German. The listening audiences have a widely unacknowledged responsibility to the multilinguals' sense of self. For this reason, we strongly believe in recognizing the addressee's central role in legitimizing heteroglossic repertoires, especially when they are in a position of power. Going beyond additive approaches, we propose that restricting nonstandard repertoires to certain contexts implicitly invites the participants to pursue a politics of nonvisibility in public spaces (Truan & Oldani 2021: 11). It is only by tackling the addressee's listening skills that racialized women will navigate their repertoires really freely-without facing damaging social and professional consequences.