What Employers Mean When They Talk About Language Proficiency: a Civic Integration Policy Meets the Importance of Informal Skills

In 2013, a Norwegian language test was made mandatory for immigrants participating in the Norwegian Introduction Programme (NIP). The rationale behind the law was that documented test scores would allow employers to better evaluate these jobseekers’ language skills, thereby helping reduce the employment gap between the immigrant and majority population. However, although previous research has found that mandated language classes have only a modest, if any, effect on labour force participation, a Norwegian study has identified a positive correlation between high scores on the Norwegian test and the probability of being employed. Whether this correlation is a result of the test diploma is, however, uncertain. In this paper, we investigate how employers in private, mostly service-oriented industries make use of the information they get through this documentation. Based on a unique linkage of two types of data sources—a survey experiment and interviews with employers—we find that the Norwegian test does not signal good Norwegian skills more clearly than a candidate’s self-assessment of their language level. We argue that this is related to a mismatch between the policy instrument’s focus on enhancing immigrants’ formal skills, on the one hand, and employers’ valuation of informal skills, on the other. We discuss these findings in light of notions about cultural cloning and homosocial reproduction. Rather than drawing on theories of discrimination to understand employers’ evaluations of immigrant job applicants, as is often done, we treat the test as an example of a larger societal change, the “civic turn”.


Introduction
The road to employment is often difficult for immigrants. One reason for this is a widespread and well-documented practice of ethnic discrimination in hiring (Di Stasio & Larsen 2020, Quillian & Midtbøen 2021. Employers often justify lower call back rates for applicants with an immigrant background by citing their uncertainty about applicants' language skills (Midtbøen, 2014;Oreopoulos, 2011). When the Norwegian government passed a bill in 2013 that required all participants in the Norwegian Introduction Programme (NIP) to take a standardised language test, the rationale was that formal documentation of language skills would encourage fairer hiring processes by providing relevant and reliable information to employers (Prop. 79 L (2010(Prop. 79 L ( -2011. In this paper, we investigate what employers really mean when they talk about the importance of language skills and how their perspectives relate to the Norwegian government's ideas. More concretely, based on two types of primary data sources and methodological approaches:-(1) a survey experiment and (2) in-depth interviews with employers-we ask: How do employers make use of the information they get through a documented language test score? And what information about immigrant jobseekers' language skills do employers consider relevant?
While many studies show that language competency is important for job market inclusion (Thijssen et al., 2021;Chiswick & Miller, 2015;;Oreopoulos, 2011), few researchers have unpacked the role language plays in hiring processes (Goodman & Wright, 2015). While language can refer to a formal competency that can be tested and measured, it can also refer to familiarity with the norms of how to use the language properly. Previous studies have confirmed that employers often privilege candidates who are familiar and reject those who are perceived as deviant (Bjørnset et al. 2021a), a pattern some have tried to explain using notions like homosocial reproduction (Kanter 1977) or cultural cloning (Essed & Goldberg 2002). Viewed through this lens, employers might be more interested in jobseekers' informal language skills than their test scores. Thus, in this paper, we also investigate how employers in the private, and mostly service-oriented, sector make use of Norwegian test scores and in what way this points to discrepancies with the Norwegian government's rationale and presumption about the relevance and utility of this test as a policy instrument.
As discrimination varies depending on both sector and industry, it is likely that employers will value information about Norwegian-test scores differently depending on where they work. As far as we know, only one previous correspondence test study has been conducted in Norway with both ethnicity and sector varying. This study found that the highest discrimination rates occurred in the private sector, making the instrument in question potentially more pertinent here. In serviceoriented industries where customer contact is part of the daily routine, employers are likely more concerned with a candidate's language competencies than in other industries (Midtbøen & Rogstad 2012a).
In this paper, we make two contributions, one theoretical and the other empirical. First, rather than drawing on theories of discrimination to understand employer evaluations of immigrant job applicants, as is often done, we investigate what documenting Norwegian skills means by treating the test as an example of a larger societal change-the so-called civic turn. In doing so, we move beyond the micro-situation of the hiring process, which enables us to analyse the implications of this policy instrument in a larger societal and political context. Second, in a field dominated by debates over policy content and concepts, we draw on two different sources of data to study what employers mean when they talk about immigrant jobseekers' language skills and how they understand the relevance of a particular example of a civic turn policy instrument. As immigrant employment is one of the major integration challenges in Europe, this is of critical importance for further policy development, as well as for better understanding the process of integration.

Background: the "Civic Turn" in Integration Policies
Some 50 years after the start of the great post-World War II migration, it is widely believed that state policies set up to accommodate newcomers are insufficient, regardless of the national model (Joppke, 2007). Over time, there has been a weakening of national distinctiveness and a convergence of integration policies, creating a "civic turn" (Joppke, 2007;Borevi et al., 2017). In this article, we employ the term to denote a set of formal policy instruments. These are typically integration contracts, classes, tests, and diplomas thought to increase immigrants' skills and formal qualifications (Borevi et al., 2017). The idea is that imposing more obligations and requirements will enable newcomers to be self-sufficient by formalising their competencies in a way that will help them accommodate and integrate into the host society (Goodman, 2010;Joppke, 2007). Since the 1990s, most European countries, including Norway, have introduced measures to formalise these types of requirements. Unlike most of the literature on the civic turn in integration policies, which focuses on the naturalisation trajectory and early integration, we examine a labour market policy instrument. Labour market inclusion is of particular concern when it comes to integration, and this policy area is also characterised by a shift to an emphasis on duties over rights, which can be thought of as part of the broader civic turn in integration policy (Borevi et al., 2017).

Language Skills and Discrimination
We know that many immigrants experience steep downward mobility upon their arrival in the destination country (Feliciano, 2020;Ichou, 2014). A lack of credentials recognised in the host country labour markets, as well as insufficient language skills and pertinent knowledge, all make inclusion in the mainstream economy difficult. Newer research demonstrates that integration measures may explain differences in integration outcomes (Hernes et al., 2020). However, the literature addressing the civic turn has primarily been interested in debating whether Western countries are increasingly converging (e.g. Joppke, 2007) or national models remain resilient (e.g. Goodman, 2014). Our objective is not to contribute to this debate. Rather, our aim is to investigate how employers make use of the information they get through the policy in question.
The rationale behind the test rests on the idea that discrimination is a way of managing the imperfect information characterizing hiring decisions, known as statistical discrimination (Fibbi, Midtbøen & Simon 2021). Lack of knowledge about language proficiency will make risk-aversive employers hire a candidate that he knows masters the native language, namely, non-immigrant jobseekers. The policy's explicit intention is thus to make hiring decision based on merit rather than more biased intuitions and subjective valuations (Bachman and Purpura 2008, Innst. 370 L (2010. However, we argue that the consequences of the policy can be different than intended if employers use language as a proxy for sameness.

Language Competency and Labour Market Participation Among Immigrants: Previous Knowledge
Integration into the workforce can be understood as a process whereby a jobseeker's competencies are matched with an employer's needs. However, the applicant needs not only to have the competence required for the position but also to be able to demonstrate that competence in a way that meets the employer's expectations. Language proficiency is an example of a competency that employers perceive as important for employment (Thijssen et al., 2021;Djuve & Kavli, 2019;Chiswick & Miller, 2015;Yao & van Ours, 2015). As stated in the introduction, many employers justify lower callback rates of applicants with an immigrant background by citing uncertainty about whether they have good enough language skills to do the job (Midtbøen, 2014;Oreopoulos, 2011). This is the essence of statistical discrimination: drawing conclusions about a person's less visible characteristics based on more or less valid knowledge about a group with which they are associated (Arrow, 1973;Birkelund et al., 2020;Fryer & Jackson, 2008;Mullainathan, 2002;Thijssen, Coenders & Lancee, 2021;Kaas & Manger, 2012). In this sense, uncertain language skills can be used as a proxy for lower productivity and therefore explain lower callback rates for job seekers with an immigrant background (Ahmad, 2020;Baert, 2018;Midtbøen & Rogstad, 2012b;Heath & Cheung, 2006).
Several studies have explored different mechanisms that can explain the effect of mandated language classes on labour force participation, finding that these policies have only a modest effect if any (Ersanilli & Koopmans, 2011;Lochmann et al., 2019). However, previous research from Norway shows that the higher an applicant scores on the Norwegian test, the more likely they are to be employed (Djuve, Kavli & Sterri, 2017). Whether this is due to the direct effect of the test diploma in the hiring process is, however, uncertain. A challenge of studying the relation between language skills and employment is that linguistic competence is seen as a formal skill that can be measured and documented on the one hand and as an informal, cultural, and communicative competence on the other hand. When employers cite uncertainty about candidates' language skills to explain lower callback rates of immigrant jobseekers, it is not clear exactly what skills they are referring to. What we do know is that employers often use subjective and non-transparent criteria such as "personal suitability" when hiring (Bjørnset et al. 2021a) and that they often value jobseekers' ability to signal appropriate cultural norms more than their ability to signal their formal skills (Bye et al., 2014, Rogstad & Sterri, 2018. The positive evaluation of a jobseeker's ability to signal "appropriate cultural norms" can be understood as part of what social psychologists' term "homosocial reproduction", whereby employers give advantages to individuals similar to themselves (Kanter, 1977). Thus, hiring is not only a process of matching a candidate's skills with employer's needs as stated above, but also a process of cultural matching between candidates and employers (Rivera 2012). The formation of this "in group favouritism" often includes some sort of "out-group rejection" (Fibbi, Midtbøen & Simon 2021). One practice that might ensure such in-group preference, thus reproducing ethnic inequality in the labour market, is when employees' referrals or employers' own social networks are used in the recruitment process (Stainback, 2008). Because information tends to flow through ethnically similar others, employers might draw from an ethnically homogenous applicant pool. Further, members of dominant groups often give an advantage to individuals who share their ascriptive characteristics. The ability to present oneself in a way that "matches" an employer's normative expectations, often described as being a good "cultural fit", seems therefore to be important for immigrant jobseekers (Bye et al., 2014). However, not all segments of the labour market are dominated by native majority employees. Some industries and occupations are characterised by a clustering of specific ethnic groups, creating "immigrant niches" (Waldinger 1994). In these segments of the labour market, ethnicity can function as a resource rather than an obstacle (Friberg & Midtbøen, 2019). Our findings are, however, not drawn from these types of industries.
This means that although mastery of Norwegian is increasingly understood as a key competency for employees, it is not necessarily enough to get a person hired. The Norwegian authorities' increased use of language testing to grant or restrict political rights, and the justification that it will benefit immigrants in their job search makes it particularly important to investigate whether the test does indeed improve immigrants' chances in the labour market. Based on previous knowledge, it is not clear how employers assess the Norwegian test as a signal of language competency and whether it meets the stated need for information. In this paper, we attempt to fill this knowledge gap.

The Norwegian Test as a Mandatory Requirement
Even though the integration of immigrants into the labour market has long been a common political goal in Norway, there is still a more than 10% difference in employment between the immigrant population and the majority population (65 vs 78%) (SSB, 2021). A key feature of the political solution to this integration challenge is the Norwegian Introduction Programme (NIP) for newly arrived immigrants, which launched in 2003 and consists of two (sometimes three) years of mainly language and on-the-job training (Djuve & Kavli, 2019Tronstad, 1 3 2019). In line with the "civic turn" in Europe, Norwegian integration policy has changed in that the authorities have become more concerned with testing and measuring immigrants' Norwegian skills, which has been linked to various rights (Ministry of Children, Equality and Inclusion 2013). When Norwegian language testing was made mandatory for participants in NIP in 2013, the idea was that that "employers should become familiar with -and able to rely on -these tests when making [hiring] decisions" (Prop. 79 L (2010-2011):11). The authorities thus seem to think that the test signals the formal language skills employers are interested in and that not having this documentation may hurt jobseekers' chances in the labour market (Prop. 79 L (2010-2011): 10). As of 2017, a language test is also required to obtain citizenship (Statsborgerloven § 8). The expanded use of the Norwegian test as a tool for measuring formal language skills in various arenas shows that the Norwegian authorities view this instrument as objective and measurable.
The test measures candidates' Norwegian skills based on the Common European Framework for Languages, with levels A1 and A2 (basic level), B1 and B2 (independent level) and C1 (advanced level), through four separate subtests in listening comprehension, reading comprehension, written presentation, and oral communication (Kompetanse Norge, 2018Norge, , 2015. The scale is standardised, meaning that the levels of the test are the same regardless of where and when the test is taken. Employers should therefore be able to evaluate an immigrant's formal Norwegian skills based on the test result and then have a better basis for a fair employment. In an interview with Aftenposten in conjunction with the government's proposal for a new integration law and stricter language requirements, Jan Tore Sanner, then Minister of Education and Integration, said, "The Norwegian language is the key to participating in small and large communities in society and to gaining access to the workforce. You have to learn the language to get a job" ("Regjeringing skaper norskkravet" [Government tightening Norwegian requirement, Jan Tore Sanner, Minister of Education and Integration in Norway 2019, to Aftenposten], 2019). Formal mastery of Norwegian is thus increasingly understood as one of the most important factors for successful integration and a key competency for inclusion in the labour market.

Our Study: Data and Methods 1
To better understand whether language test scores provide employers in private sector with useful information about applicants' Norwegian skills, we have used two different data sources: (1) a survey experiment and (2) qualitative in-depth interviews. The findings are complementary each other, thus increasing the validity of the results. Triangulation of this kind tests the consistency of the findings from the survey experiment and the interviews separately (Bryman, 2012). While the individual data sources have limitations, the unique linkage of both quantitative and qualitative strategies is a viable option to obtain findings that complement each other.

3
What Employers Mean When They Talk About Language Proficiency:…

(1) Survey Experiment
To investigate what information the Norwegian test gives employers about Norwegian language skills, Norstat AS conducted a survey with an experimental design on our behalf in September 2019 among 600 employers in Norway, mainly in the private sector. This method involves randomly dividing respondents into two groups and asking both groups the same question, keeping all conditions equal except one, in this case the way Norwegian skills are presented (Auspurg et al., 2017;McDonald, 2019). In this case, the respondents were asked to decide whether an applicant's Norwegian skills were good enough to be employed, given the information they had received. All employers were informed that the applicant was a woman who came from, and was educated in, Eritrea and who had lived and worked in Norway for the past seven years. In addition, about half of the respondents were informed that the applicant had indicated mastery of Norwegian in writing and orally, while the other half were informed that the applicant had documented her Norwegian skills through the Norwegian test level B2. The vignette question is presented in its entirety in Fig. 1. A vignette question like ours gives respondents more information than a regular question and ensures that their answers are based on the same scenarios, thus improving internal validity (Nock & Guterbock, 2010). The internal validity is also strengthened by the fact that the respondents are randomly distributed between the two groups. This way, there can be no systematic differences between the respondents, and we can assume that any differences in the responses of the two groups are due to the way Norwegian skills are presented. This design allows us to control for potential biases (see Appendix for an overview of the randomisation between the two groups) (Bryman, 2012;Hellevik, 2011). As such, the design is considered to provide high internal validity (Bryman, 2012).
The survey experiment was conducted as part of Norstat's quarterly survey of Norwegian companies (the NaeringsBuss survey). The Norstat survey contains multiple questions on a variety of subjects, and the respondents did not know that they were involved in an experiment or what the aim of our study was. Norstat uses a stratified random sample (quota sample) of Norwegian companies from the Brønnøysund Register Centre (Enhetsregisteret). The general manager (CEO or equivalent) was asked to respond to the survey.
Quota sampling based on the number of employees was used to select the participants in the survey leading to the over-representation of large companies in the sample (see dropout analysis in figure in appendix). In addition, the study population was drawn from industries with industry codes A through N in Statistics Norway's standard for industry grouping. This includes mostly private companies. Companies in industry and wholesale are over-represented. The excluded industry codes O through S include public administration, education, health and social services, and cultural activities. This is a weakness of the data because we lack information on certain industries in which immigrants are strongly represented. For example, many immigrants work in the health sector. It is reasonable to assume that businesses that have more employees with an immigrant background are also more likely to know about or make use of language tests. This is probably the case within the public sector where some jobs require passing scores on the Norwegian test to be employed and/ or to obtain a permanent position Ødegård & Andersen, 2020). At the same time, important industries with a high proportion of immigrant workers are included, such as hospitality and catering, transportation, and construction. Norstat contacted potential respondents by telephone, calling as many companies as necessary to obtain a sample of 600. Most who did not respond to the survey did not pick up the phone when called (42%), did not have the time to take the survey (21%), or did not want to participate or were not interested in the topic (13%). Seven per cent of the people called responded to the survey (Valen, 2020). When the response rate is low, there is a danger that a specific group will be accidentally omitted from the analysis.
These limitations make our sample biased, which means that the findings cannot be generalised beyond the industries represented in this sample. Although there is little potential for generalisation, the findings in this article should still be relevant. Because discrimination varies depending on both sector and industry (Midtbøen & Rogstad, 2012a), the only way to investigate the effects of language testing policy is to provide sector-and industry-specific findings, as we do in this article.
Another limitation of the study is that it does not include the perspectives of immigrant jobseekers themselves, presenting only employers' perspectives. Further, the category of "immigrants" refers to a rather heterogeneous group, which we take to include people "who have immigrated themselves, and who were born abroad with foreign-born parents and four foreign-born grandparents" (Dzamarija, 2019). A limitation in this regard is that in our study, "immigrant" is further narrowed down to a very concrete example, a woman from Eritrea who has lived and worked in Norway for the past seven years. Therefore, we cannot generalise our findings beyond this narrow definition of an immigrant.

(2) Qualitative In-Depth Interviews
The interviewees in this study were recruited in connection with a project we did on behalf of the Ministry of Labour and Social Affairs, in which a small field experiment was conducted with two fictitious Eritrean applicants (one listed a Norwegian test level B2 on her resumé while the other stated she had "good Norwegian language skills" in her application) (see Bjørnset et al., 2021b). A challenge with this methodological approach is that it violates the ethical standard that participation in research should, as a rule, be based on information and consent. To counter these issues, we approached all the employers who had contacted one or both fictitious Eritrean applicants. As soon as an applicant received a call or e-mail, we registered the callback, withdrew the application, and sent a letter to the employer with information about the study, encouraging them to contact us to be interviewed. This means that all interviewees based their participation on informed consent. We sent ten letters and were able to recruit eight employers for interviews.
While a common critique of qualitative interviews is that self-reports and behaviour are often conflated (Jerolmack & Khan, 2014), in this study, all interviewees were at the end of a real recruitment procedure, and both researchers and participants knew how they had recently behaved. Thus, the information obtained in the interviews was not based on what the interviewees told us that they would do in a hypothetical situation but on what they had actually done. However, people often present themselves in ideal ways. Although the interview was based on something that had actually happened, the reasons the interviewees gave for their behaviour must be understood partly as a form of self-presentation (Gubrium & Holstein 2009).
The interviews were semi-structured in the sense that we, the researchers, always came prepared with an interview guide with three sets of topics we wanted to talk to the interviewees about (Brinkmann 2014). First, we asked about the recruitment process in which they had recently hired a candidate for a real position. In this part of the interview, we wanted to know why they chose to call one or the other or both candidates back, whether they had noticed the difference between them, who they ended up hiring, and so on. Second, we asked about their company's routines, their experiences with hiring candidates with immigrant backgrounds, and how they assess language competency in the hiring process. Finally, we asked them about their knowledge of, experiences with, and views on the Norwegian test and language as a competency more generally. The interviews were conducted over the telephone and were audiotaped and transcribed.
Although we acknowledge the constructed nature of interviews (e.g. Gubrium & Holstein 2009), our analysis draws on a content-based approach (Braun & Clarke, 2006). This means that we coded the transcribed interviews into different categories, including broad categories, such as "language", "communication," and "culture", and more concrete categories, such as "knowledge about the levels", "knowledge about the test", "composition of the firm", "assessing language skills," and "employers' evaluations of good language skills". Even though we wanted to understand what happened in the recruitment process, we were even more interested in understanding the importance of language in interviewees' hiring decisions. The small number of interviewees makes it difficult for us to provide solid conclusions about employers' assessments. Still, although the findings in this article are obviously not statistically generalisable, we argue that they provide information of value beyond the interview context. The interviewees were sampled by virtue of their position in the labour market as employers with recent experience recruiting for their business. By focusing on these experiences rather than on their individual characteristics, others in similar situations might find their experiences relatable.
Seven of the interviewed employers worked in private companies, all of which were accounting firms. The companies varied in size, the smallest having 20 employees and the largest over 2000 employees. The interviewees are not directly comparable to the survey respondents because they represented different businesses. The interviewees were exclusively from accountancy firms, which was not the case for the survey respondents. However, both groups of participants worked in the private sector, with most of them working in firms where customer contact was part of the daily routine. Thus, combining our different data sources enabled us to provide more sophisticated and complex analysis than would have been possible with a single-method approach. The survey experiment alone cannot explain why and under what conditions employers acted as they did (Midtbøen & Rogstad, 2012a). Interviewing a selection of the employers that participated in a field experiment allowed us to investigate their behaviour in the context in which it occurred.

Survey Experiment: the Norwegian Test of No Importance
The first finding, presented in Fig. 1, is that more than 70% of the employers in the survey experiment stated that the candidate (to a large or fairly large extent) had good enough Norwegian skills to be employed by them, regardless of how they signalled their Norwegian skills. However, the vignette contained other information that indirectly signalled Norwegian skills, such as length of residence and work experience from Norway. We cannot say whether the employers focused on these signals to establish that the candidate had good enough Norwegian skills to be employed, or whether that determination was based on the Norwegian test score or the subjective assessment.
This finding suggests that the relative difference between the two forms of assessment is more important than the absolute difference. Based on the stated policy intentions, we would expect that documented language skill would make employers more likely to consider the candidate's language skills to be good enough compared to other indicators. Our findings from the vignette question showed the opposite. When the candidate offered a subjective assessment of their own Norwegian skills, 77% of the employers believed that the candidate's language skills (to a large or fairly large degree) were good enough to be employed. In comparison, when the candidate referred to passing the Norwegian test with a level of B2, 71% of employers believed (to a large or fairly large degree) that their language skills were sufficient. The difference in how employers assessed the information from the two instruments is significant at the 5% level. The most important takeaway from the survey experiment is, in other words, that employers are not more likely to hire a candidate with a documented Norwegian test score than a candidate with a subjective assessment of Norwegian skills. This indicates that the test does not give a clearer signal of good Norwegian skills than a candidate's subjective self-assessment of their language level. This finding is supported by the limited field experiment referred to in the method section (Bjørnset et al., 2021b).
The distribution of answers in Fig. 1 may be affected by the fact that the respondents could have perceived it as not politically correct to suggest that they distrusted a candidate's positive self-assessment of his or her Norwegian skills. We also know that the sample is biased. It is reasonable to assume that the answers would have been different if, for example, respondents from the health sector, where the language test is mandatory, had been included. In line with previous studies, we found that larger companies were more likely to consider the candidate who had passed the Norwegian test to have good enough Norwegian skills to be employed (not shown in figure) (Valen 2020;Heath & Cheung, 2006). Larger companies are believed to have more formalised recruitment processes, and there are more large companies in the sample than in the population. Therefore, these findings are not representative of companies outside the study sample. Nevertheless, the answers from the survey indicate that a documented language test does not give employers information to assess immigrant jobseekers' language competency to a greater degree than a subjective assessment.

Interviews with Employers: Language as a Proxy for Relational and Cultural Norms
When making the Norwegian test a mandatory part of NIP, the Norwegian authorities argued that documenting a score on a Norwegian test in job applications would help convey relevant information about immigrants' language skills to employers, which would in turn reduce discrimination in hiring practices (Innst. 370 L (2010(Innst. 370 L ( -2011. This expectation seems to rest on assumptions akin to those of the theory of statistical discrimination, which holds that employers may discriminate even though they are both rational and non-prejudiced. By this account, discriminatory treatment in hiring situations is a result of how employers manage the lack of information about candidates' productivity-related characteristics, of which language competency is one. In other words, employers draw conclusions about an individual's productivity based on statistical information about a group (Midtbøen, 2014).
If this is the case, more information about a candidate's language skills would reduce employer uncertainty about the candidate's ability to contribute to the productivity of the workplace. Our results, including the qualitative interviews, indicate that the Norwegian test does not seem to be considered a tool that provides relevant information about candidates' language skills. The interviewees stated that they did not look for language test results when evaluating these applications. In fact, none of the interviewed employers saw this documentation as a relevant 1 3 indicator of language skills among candidates with an immigrant background. One reason seems to be that they lacked information about the test (e.g. Birkeland, Tkachenko, & Ulven, 2019;Haugsvaer, 2018). This is illustrated in following interview exchange: When talking about the test, employers demonstrated their ignorance about the test, for example, not knowing the levels of mastery indicated by the test, mixing up "high" and "low" levels, or confusing the test with other language tests available. Another interviewee stated "B1, B2, C1, the Bergenstest, I don't know what they mean, but it doesn't really matter". The implication is that the indicator becomes irrelevant because it is fundamentally unclear. In fact, none of our employer interviewees saw this documentation as a relevant indicator of language skills among candidates with an immigrant background.

Language as an Indicator of social and relational skills
Although lack of information seems to be part of the reason employers considered the test irrelevant, our analysis indicates that the type of skills the employers considered "language skills" was also important. When the interviewees expressed the importance of "good language skills", they were referring to a candidate's relational and communicative skills. As one employer stated: Good communication is so much more than just "language skills". It's about how you enter into dialogue with other people. How you involve others, how you make contact with them, how service-oriented you are, and so on.
Broadly put, our employer interviewees used "language" to mean norms about appropriate behaviour in the workplace. When talking about the importance of "language", one employer told us, "Well, the candidate must master the language that the customer prefers to use". Here, mastery of language refers to the candidate's ability to be adaptive. Viewing "language competency" and "adaptiveness" as similar traits was common. Examples of other "adaptive communicative skills" our employer interviewees looked for are knowing how quickly to answer an e-mail, how often and how loudly to talk in meetings, how to approach authorities, and whether or not to use exclamation marks when writing to a customer. Another interviewee told us, "the people working in my firm must be able to treat our customers with respect and to communicate with them professionally. This is difficult, including for Norwegians".
For this interviewee, language competency is not necessarily something that native speakers have, and immigrants lack. Rather, it refers to a set of informal social and relational skills, such as an individual's ability to treat someone with respect and to be professional. It is important to remember, however, that the findings in this study are based on data from industries where customer contact is highly important. It might be that these employers are more concerned with a candidate's ability to use their communicative skills to adapt and adjust than employers in other industries.
However, the interviewees also legitimised their concern about the relational and social dimensions of candidates' "language" skills by referring to the importance of the social dynamic between co-workers and the work environment in general. One said: Well, I think it is an advantage to be Norwegian when there are mostly Norwegians working here. You know, there are so many references, things a foreigner wouldn't understand. Like TV shows we all watched as kids, or like who the Norwegian prime ministers were in the 80s, and stuff like that. It is important to have this knowledge to be included.
The knowledge this employer refers to indicates whether a candidate "fits in" or is "suitable" for the position (Horverak, Bye, Sandal, & Pallesen, 2013). The employer interviewed explicitly states that he understands cultural similarity between the candidate and the workers in the firm as important for inclusion. His preference of someone similar implies an exclusion of someone different. This practice will, in the long run, create a situation of "cultural cloning" understood as a systematic reproduction of sameness (Essed & Goldberg 2002). While concerns about language as a "relational skill" might be especially pertinent in the service sector, the importance of hiring candidates with informal skills that allow them to "fit in" is found across sectors and industries (Bjørnset et al. 2021a). The employers in this study also argued that informal, soft skills are the most important thing to consider when hiring a candidate as the formal skills required for a position can often be learned. One employer said: The job in itself is not really hard, it is easy to learn. What we look for is someone curious, good at problem solving, you need to have the guts to approach people, be service minded, make people cooperate.
This employer is clearly looking for a candidate with a certain personality. Their reasoning is that because the required formal skills can be easily learned, it is not crucial to detect them during the recruitment process. However, the other vaguer and more informal skills the employer refers to, such as curiosity, the courage to approach people, service mindedness, and the ability to cooperate, are not framed as something a candidate can easily learn. This tendency to view informal skills as less learnable, and therefore more important, when hiring has been found in previous research as well (Bjørnset et al., 2021a). All of this suggests that although employers often talk about "language" as a skill they care about in the hiring process, it does not necessarily mean that they are concerned about "language proficiency" in terms of the candidate's vocabulary. Rather, "language" might serve as a proxy for other vaguer qualities, such as the ability to adapt, relational sensitivity, and adherence to cultural norms. If this is the case, the value of the "objective" information provided by the Norwegian test becomes irrelevant.

Language Proficiency and Cultural Adaptability as Reciprocal Skills
The ability to present oneself in a way that conforms to an employer's normative expectation is, in social psychology, often described as "cultural fit" (Bye, Horverak, Sandal, Sam, & Van de Vijver, 2014), "cultural cloning" (Essed & Goldberg 2002), or "homosocial reproduction" (Kanter 1977. Language proficiency in an interview setting is often taken as a sign of a high level of cultural fit and as a reassuring sign of similarity. Thus, language and cultural norms are intertwined and difficult to separate. Their relationship might even be reciprocal in the sense that proficiency in the host language might indicate cultural adaptability and similarity while signs of cultural adaptability might indicate host language proficiency. Thus, when language skills are invisible, employers might look for signs of adaptability to provide more information about a candidate's language skills.
In our interviews, we found that employers considered signs of social integration suggestive of language skills. Being part of a network with native Norwegian speakers was more indicative of a person's language proficiency than attending a Norwegian language course. One employer, whose wife had migrated from a Southeast Asian country a few years previously, made the following observation: Well, I meet people at my wife's Norwegian class. They have passed several tests, but the way I see it, they do not speak Norwegian any better than my wife, even though she would definitely fail the tests they have passed. So in my opinion, and this is just my personal opinion, experience with speaking Norwegian daily is considerably more important than documenting a passed test.
For this employer, an indication of membership in a network with native Norwegians would be a more reliable and relevant source of information about a candidate's language skills than a test score. Another employer addressed his concerns about a candidate's social life: Maybe he has been around his own "countrymen" a lot and not learned proper Norwegian. We are a bit concerned about that. How quick and motivated will such a candidate be?
These employers seem to evaluate candidates based on gut feeling and emotions (e.g. Bjørnset et al., 2021a;Hedenus & Backman, 2020). If an employer gets the feeling that a jobseeker is not "socially integrated" (which in this case means being included in networks similar to the employer), they may extrapolate to make a judgment about the candidate's language proficiency. At the same time, language is seen to indicate similarity and understood to be intertwined with other informal skills. While informal skills are hard for employers to "observe", they are almost impossible to document in an application or through a test score. This does not mean that having a minimum of formal language skills is irrelevant for being hired. In a sense, it might be necessary but is hardly ever sufficient.

3
What Employers Mean When They Talk About Language Proficiency:…

Conclusion
In the literature, civic integration policies are often introduced as measures that will serve to accommodate immigrants (Borevi et al., 2017). Regarding the Norwegian test specifically, it is believed that making it a requirement encourages fairer hiring processes by providing relevant and reliable information about immigrants' language proficiency. However, while previous research has found that mandated language classes have only a modest effect, if any, on labour force participation (Ersanilli & Koopmans, 2011;Lochmann et al., 2019), research from Norway shows that the higher an applicant scores on the Norwegian test, the greater the probability of their being employed (Djuve, Kavli & Sterri, 2017). Whether this is due to the direct effect of the test diploma in the hiring process is, however, uncertain. In this paper, our aim was to investigate how employers in private, mostly service-oriented industries make use of this specific type of civic integration policy instrument. Our analysis and findings are based on two primary data sources and methodological approaches-a survey experiment and in-depth interviews with employers. This unique combination of quantitative and qualitative strategies is a viable option to obtain findings that complement each other We have shown that documentation of language skills does not necessarily give employers the information they are looking for. The employers in this study work in industries where customer contact is very important, and we show how they are more concerned with a candidate's familiarity with communicative norms than with formal, measurable language skills.
This study does not include the perspectives of immigrant jobseekers themselves, but previous studies indicate that many immigrants take the Norwegian test in the hopes that it will increase their chances of being hired (Djuve et al., 2017). Some report being told that passing the Norwegian test all but guarantees employment. For the industries represented in this study, we identified a mismatch between the importance placed on documenting language skills through the Norwegian test and successful labour market integration.
One reason why this specific policy instrument does not work as intended in these industries might be that there is a mismatch between its focus on enhancing immigrants' formal skills, on the one hand, and employers' valuation of informal social and cultural skills as crucial for their employability, on the other. Rather than being concerned about a candidate's "language proficiency" in terms of their vocabulary, employers often used "language" as a proxy for other qualities, such as the ability to adapt, relational sensitivity, and adherence to cultural norms. Our findings show that a test score did not signal the competency the employers in this study were most apprehensive about. Instead, information about a candidate's social integration was considered a more relevant as it was an indication of similarity, and thus a high level of "cultural fit". The "fit" between candidates and employers that are culturally similar is often referred to as "cultural cloning", or "homosocial reproduction". While this type of cultural matching is not necessarily conscious, it still leads to an exclusion of "out -groups", a reproduction of sameness. These forms of "systemic" discrimination are harder to prohibit by legislation, as the phenomenon are deeply entrenched in our everyday practices, our social life and existing organizational cultures (Essed & Goldberg 2002, Fibbi et al. 2021.
As our analysis focuses on employers in the private sector, we are unable to generalise our findings beyond the industries represented in our sample, but the findings in this article are still relevant and contribute to the overall discourse about language proficiency among immigrant jobseekers. However, as discrimination varies depending on both sector and industry, the only way to produce knowledge about how employers evaluate language tests is to provide sector-and industry-specific findings. Based on our findings in this paper, we recommend that future research should investigate the measures employed in civic integration policies and, more specifically, address the question of whether an increased use of language tests has other societal consequences beyond hiring decisions. As the term "immigrants" refers to a very heterogeneous group, future research should also investigate whether the use of language tests differs between different groups of immigrants. Because language is a major factor in defining identity and a key part of successful social and cultural integration, adjusting requirements may change the larger context of cohesion and societal integration. Further, we recommend expanding the research design to include other sectors and industries than the ones applied here.

Interview guide-employers: language
Anonymity, voluntary participation, and right to decline 1. Do you consent to the material from this interview being used for research purposes? 2. Both the company and you as an individual will be guaranteed complete anonymity.

3
What Employers Mean When They Talk About Language Proficiency:… (Choose whether you want to know how your company "fares") 21. The letter you received from us states that we can inform you of the outcome of the specific test that deals with position X. Are you interested in what the outcome was? 22. What do you think about this outcome? 23. Why do you think this was the result?

About language
24. How important is proficiency in Norwegian to work in your company, and particularly in relation to position X?
a. Is it most important for job performance or to fit into the working environment?
25. What are your experiences with the language proficiency of people with an ethnic minority background? 26. When you receive an application from someone with a foreign-sounding name, does it tell you anything about the person's language proficiency? 27. Are there other indicators that can be used? a. Length of residence b. Where they studied c. Work experience 28. What about language tests? a. Are you aware of the various language tests? b. Do you emphasise the need for formal language qualifications? 29. Have you clarified the level of proficiency in Norwegian that is needed to perform specific jobs at your company? a) How do you determine this? b) Is the language requirement set by you or the company? c) Is this done prior to publishing a job announcement? d) Does the level of language proficiency required vary depending on the type of position?
30. Do skills in the Norwegian language tell you more than a person's ability to do a job? 31. Can Norwegian skills be developed at work? Do you facilitate this? Uncertainty 32. What is the first thing you look at when you receive a job application? 33. In positions like this, is it about finding "the one" for the job, or is primarily about not recruiting the wrong person?