Multilingual practices and normative work in the context of a civic integration project in Sweden

ABSTRACT This article explores multilingual practices and normative work in intercultural encounters in a language café set up as part of an integration project promoted by The Equmenia Church in Sweden. The method used in the study combines an ethnographic approach with two research methodologies used for studying interaction: conversation analysis and membership categorisation analysis. For this article, I have selected two examples extracted from video-recordings of interactions when volunteers and migrants at the café are engaged in normative work concerning sensitive, culture-related topics. The article zooms in on the mediating work of a multilingual facilitator, who is assigned the role of assisting as an interpreter in the integration project. It was found that the facilitator does something more than simply translate. Specifically, she also addresses moral issues through (i) explaining and clarifying; (ii) adding information; (iii) expanding on the agenda of the primary party; and (iv) expressing stances and taking sides. Based on these findings, I discuss the implications of this study, highlighting the role of lay multilingual speakers’ mediating work for intercultural communication and for migrants’ early process of integration.


Introduction
Due to increasing mobility, societies are becoming ethnically and culturally diversified at an unprecedented pace. However, migrant-receiving countries often struggle with segregation; that is, with the physical and social divides between people of what is perceived as distinct ethnic categories (Finney 2009). Integration, on the other hand, is about fostering intergroup contact and mutual exchange, which bring knowledge of the other and hence contribute to the reduction of prejudiced views of the other (Allport 1979). Migrant-receiving countries across the globe face many challenges in achieving this ideal of integration.
Prejudiced views of the other are linked to the idea of a national identity (e.g. Anderson 2016;Billig 1995;Wodak et al. 2009), which is reproduced through ideological habits of practice and belief. These less visible, taken-for-granted forms of 'banal nationalism' (Billig 1995), are deeply rooted in everyday life. Central to the formation of national identity, is the contradistinction from outgroups. Through stereotyping categorisations we construe distinct notions of 'them' and 'us', thereby assuming ourselves as the standard or the unmarked normality against which 'their' deviations appear notable.
This article explores non-professional forms of interpreting and normative work in the context of a civic integration project promoted by a regional missionary evangelistic organisation in Sweden. The concept 'normative work' is here used in relation to the theoretical framework of national identity. It refers to the ideological dynamics of everyday reasoning and categorisation of self and others in intercultural communication. In line with the idea of national identity, the article demonstrates how stereotyping categorisations of 'them', which are at odds with norms concerning local moral standards, are invoked in the table talk at a language café set up as part of the integration project. For this article, I have selected sequences of multilingual interactions between Arabic-speaking migrants/café visitors (with a newly arrived Iraqi woman as the focal participant) and Swedish-speaking volunteers. The article focusses on how a lay multilingual speaker (without training in interpreting), who is bilingual in Swedish and Arabic and is assigned the role to assist as an interpreter in the integration project, relays parts of the ongoing conversation for the co-participants in the other language, and thereby facilitates their understanding and participation in the interaction. The bilingual speaker's means of representing prior talk is here referred to as multilingual/ mediating practices. The bilingual speaker is referred to as a facilitator. The aim of the article is to identify the facilitator's multilingual practices and explore the role they have for the management of categorisations that reflect conventional expectations of normative behaviours. I argue that the facilitator's mediating work allows the migrant to challenge negative beliefs about her culture and to position her identity as a good example.
Since the first seminal studies (e.g. Berk-Seligson 1990: Wadensjö 1998, the research community has predominantly focussed on interpreting and translation as professional activities performed in institutional settings. These early studies have established the theoretical grounds for later research in the area. Specifically, a central theoretical point concerns the interpreter's participation and involvement in ongoing talk. In particular, the research trend about dialogue interpreting has modified the view of the interpreter as a translator, i.e. who provides plain renditions of others' turns of talk. Wadensjö's (1998) distinction between relaying and coordinating talk has contributed to a changed view of the interpreter's roles and responsibilities, of what it means to translate in interaction, and of the activities it involves. Typically, studies have focused on the function of relaying talk; that is on the role of the interpreter as a translator. On the other hand, the function of coordinating talk considers the interpreter as an active participant in a triadic interaction, and not simply as a translator who transcodes into another language. In accordance with this argument, it has been shown that mediators do not engage only in turn-by-turn translation, but also in other activities that are relevant to meeting interactional goals (Metzger 1999;Roy 2000;Baraldi 2012). In other words, this research has shown that mediators do more than providing plain renditions of participants' turns. They do so through explicating, modifying and expanding, thereby making sense of what has been framed as critical/delicate issues in talk (Mason 2006;Baraldi and Gavioli 2012;Gavioli 2015). In conclusion, this dynamic perspective points to the fact that interpreting, as any type of human interaction, is about the situated construction of meaning (cf. Baker 2006).
Interpreting and translating were originally regarded as innate skills developed in all bilinguals (Harris 1977). Although mundane, non-professional forms of interpreting undoubtedly have existed long before the professionalisation of this activity, it is not until recently that we have seen a growing interest for nonprofessional activities of translation and interpreting (for early studies see Harris 1977;Müller 1989). Lay multilingual speakers and non-professionals (i.e., people with no special training in interpreting) take on the task of interpreting in a variety of situations and conditions. In institutional settings, such as healthcare, nonprofessionals (e.g. staff members and relatives) appointed to perform interpreting activities may fill a gap that arises out of a lack of professional interpreters, or a shortage of economic resources, a situation that has been rather well explored in previous studies (e.g. Tse 1996; see Flores 2005 for a review). Most of these studies point to the poor quality and negative effects of using non-professional interpreters, such as misunderstandings and distortions of information. However, in recent approaches to the study of face-to-face mediated interaction, nonprofessional interpreting and lay translation are treated as sociocultural and interactional phenomena in their own right, and not simply as alternatives to professional interpreting. This branch of research focusses on lay multilingual speakers who engage in occasional translating as ad hoc mediators, in mundane, everyday settings or at workplaces and in service-providing contexts, sometimes referred to as brokering in the literature (e.g. Tse 1996;Del Torto 2008;Skårup 2004;Bolden 2012;Jansson and Wadensjö 2016;Harjunpää 2017).
As a growing body of research indicates (see Pérez-Gonzalez and Susam-Saraeva 2012 for a review), non-professional forms of translation and interpreting are being conducted by an increasingly heterogeneous range of agents in society, such as museum communities, religious congregations, commerce and marketing, and digital marketplaces. This broadening of the field of linguistic mediation reflects the emergence of new forms of engagement in new sites of cross-cultural contact in public life, resulting from migration and resettlement of populations. The crucial role played by translators and interpreters in legal, health and human services settings in migrants' early process of integration and in intercultural communication has been highlighted in previous research (Cronin 2006;Pérez-Gonzalez and Susam-Saraeva 2012, p. 153;Inghilleri 2017). In global societies, where institutions and civic organisations have limited funding for professional interpreting services, it is becoming increasingly frequent to rely on in-house bilingual employees, community members or volunteering colleagues who act as interpreters (e.g. Hokkanen 2012; Pérez-Gonzalez and Susam-Saraeva 2012; Traverso 2012; Roels et al. 2015;Jansson and Wadensjö 2016). The current study advances this line of research by casting light on the practices of non-professional interpreting in a public site of intercultural communication in civic society. Using the methodology of conversation analysis (CA) combined with the ethnomethodological approach of membership categorisation analysis (MCA), the study investigates how translation is intertwined with issues of face, category and identity.
The article is organised as follows. First, I briefly describe the setting and discuss the data and methodology applied in the study. Then, I analyse two examples in which the participants are engaged in normative work. In the first example, the normative perspective concerns beliefs about the poor learning outcomes of migrants with long residence in Sweden. In the second example, beliefs about morally questionable ways of getting married in the migrant's culture are invoked. In the analysis of both examples, I demonstrate how the migrant is made accountable for telling her story related to these topics and how her story is portrayed in the interpreter/facilitator-mediated interaction with the volunteers. It was found that the facilitator does something more than simply translate. In fact, she also addresses moral issues through: (i) explaining and clarifying; (ii) adding information; (iii) expanding on the agenda of the primary party; (iv) and expressing stances and taking sides. Finally, I discuss the implications of the findings from this study, highlighting the role of non-professional interpreters for intercultural communication and for migrants' early process of integration.

Setting and participants
The present study draws on fieldnotes and interviews, adopting an ethnographic approach, and on video recordings (60 hours) of table talk in a language café organised by an evangelistic church in a suburban multilingual area in Sweden. Audio-recorded interviews with the café organiser, the volunteers, three multilingual employees and the pastor in the church were carried out by the author. Video data were collected by the author and two collaborators. The fieldwork was carried out from August 2018 until June 2019 and comprises 28 visits at the café (23 by the author and 5 by the collaborators). All data were collected following the ethical guidelines established by the Swedish Research Council and with the approval of the Regional Committee for Research Ethics (Dnr 2017/2283-31/5).
From its start in 2010, the café was running one day a week. Since January 2018, when the organisers received funding from an integration project, the café runs two days a week, for a total of five hours; attendance is on a drop-in basis. The participants of the café are the volunteers, the pastor in the church and the migrants. The volunteers are retired men and women, who are ethnic Swedes. The migrants have moved to Sweden for various reasons (e.g., to seek asylum or to reunite with their family, etc.), mostly from Syria, Afghanistan, Iran, Iraq and Eritrea.
The café is a subproject of a larger integration project, called Tillsammans för asylsökandes egenmakt och deltagande ('Together for asylum seekers' empowerment and participation') promoted by a regional missionary evangelistic organisation in Sweden and funded by government grants from the County Administrative Board (project period 2018-2019). The organisers of the café use the short name Asylprojektet ('the Asylum project'). The grants cover salary costs for three multilingual facilitators who were employed to work ten hours a week in the café project. The overarching goal of the project is to 'strengthen asylum seekers' empowerment and participation through bringing knowledge, enabling mutual exchange of experiences, offering community and networking' (quote from the application for funding; the translation is mine).
The café can be described as a hybrid setting, where the participants engage in activities with a pedagogical agenda as well as in unplanned activities such as informal chatting with no predetermined topics. In interviews, the café-organiser and the volunteers describe the café as an open meeting place for socialising, where newly arrived migrants can experience a sense of belonging and community regardless of their language skills. At the same time they state that the ultimate goal is that the migrants learn Swedish.
The multilingual employees are speakers of the café visitors' languages (Arabic, Farsi, Amhari and Tigrinia). As employees in the café project, they are officially assigned the role to assist as interpreters. Note that professional interpreting was never used in this setting. Another assigned role is to help the migrants with guidance in their language about everyday matters related to their settlement. All this is expressed in documents (in the application for funding and in the café organiser's reports to the funder), as well as in interviews with the café organiser and the pastor, who emphasise the importance of helping with translation and interpreting for facilitating participation. In small group sessions during conversation-for-training and language work with vocabulary and grammar the facilitators' interpreter help was often enlisted by the volunteers. They also assisted as interpreters during fronted information sessions in plenum held by the café organiser or by an invited guest (e.g. about Swedish traditions and Swedish civil service).
The current study focusses on the multilingual practices of the Arabic-speaking facilitator (whom I name Fatima) in two examples drawn from situations when she is enlisted to help with translation in conversations between volunteers and visitors. These two examples were chosen of two reasons: (1) they represent a recurrent situation in my data: when the facilitators were involved in face-toface interaction in small group conversations and (2) they illustrate recurrent multilingual practices adopted by the facilitators. Specifically, they provide examples of an intercultural mediator performing cultural brokering (e.g. Tse 1996;Angelelli 2004;Del Torto 2008). They illustrate how the facilitator by virtue of her multicultural competence acts as an intermediate between the café visitors and the volunteers. Both examples involve a newly arrived migrant from Iraq in her twenties (Rim) with five months of residency in Sweden, and a Kurdish woman from Iraq in her fifties (Noor) with thirteen years of residency in Sweden, both regular visitors at the café. The volunteers are three members of the church, two men (Bert and Arne) and a woman (Ada). All three are attending the café almost every session. Bert and Ada have been engaged in the café since its start, whereas Arne has participated as a volunteer for a little more than a year.
Fatima, a refugee-background Syrian woman with six years of residence in Sweden, has no special training in interpreting. In Syria, she studied law. She is a member of the church and was herself a regular visitor at the café as a newly arrived migrant. In interviews, she reports that she most of all likes to assist as an interpreter in small group conversations: 'there are many Arabic-speaking guests in the café and it is difficult for those who don't speak English to talk with a Swede' (quote from interview; my translation from Swedish). She reports that helping with interpreting makes it easier for those people to have contact with Swedes.

Conversation analysis and membership categorisation analysis combined with ethnography
The method used in this study combines an ethnographic approach with two research methodologies used for studying interaction: CA (Sidnell and Stivers 2013) and MCA (cf. Sacks 1992; see Antaki and Widdicombe 1998). Long-term ethnographic fieldwork provided knowledge about routine activities and especially about the practices of the multilingual employees. I spent time in the café setting, observing the table talk, chatting with the participants without interfering in their conversations and avoiding assessing their practices. This work eventually granted me the opportunity to video record instances of occasional interpreting in small group conversations at the tables.
The recordings have been analysed from an emic (participant-relevant perspective) with the methodological tools afforded by CA and MCA. Specifically, MCA focusses on the locally invoked common sense knowledge that the larger society has about a category (e.g. an ethnic group or a population), to which people orient as they accomplish ordinary activities. Categories as defined by Sacks (1992) are social types, bound up with specific actions (category-bound activities) or characteristics/features (category predicate) that both produce and reproduce conventional expectations of normative behaviours within a group or a social identity. This reflexivity between foreground and background (i.e., the phenomenon for which local activities become meaningful in relation to the environment where they occur) is based on the concept of indexicality, which has its roots in semiotics (Silverstein 2003). The definition given here is specific to the use of the term in MCA (Lepper 2000). The principle of indexicality in MCA focuses on explicit mentioning of categories as well as members' presuppositions of their own and others' positioning. These presuppositions are demonstrated through evaluative orientations (e.g. assessments) and linguistic structures (e.g., question design). To claim the relevance of a category, then, requires the analyst to make inferences (Sacks 1992;Lepper 2000), which means that the analyst necessarily has to draw on extra-contextual resources.
In this article, the CA attention on the micro-details of interaction is combined with the ethnomethodological approach of MCA to analyse: (i) how the participants engage in normative work and, in doing so, mobilise categorisations; and (ii) how this work is managed through the facilitator's mediating, multilingual practices.

Transcriptions
Data excerpts have been transcribed following CA conventions for multimodal transcription (Mondada 2014; see the Appendix); when relevant for the analysis, I have added drawings (based on screenshots from the video) to illustrate the participants' embodied conduct. Conversational video-recorded data in Arabic has been transcribed and then translated into Swedish by two proficient speakers of this language. Translations from Swedish into English were made by the researcher and proofread by a Swedish-speaking, English native speaker. Each Swedish utterance is given an English translation in italics beneath it. Translations of the Swedish spoken language into English have been made, with the attempt to make it comprehensible, albeit not always altogether idiomatic. When needed, two translation glosses are provided under each line in Arabic. The first gloss follows the word order of the transcribed talk word-by-word. The second gloss is an idiomatic English translation. Arabic script was used for transcribing Arabic speech. The Arabic script was then romanised, following the original pronunciation as closely as possible.

Managing categorical generalisations of the learning outcomes of migrants
In the following example (Excerpts 1a, 1b, 1 c and 1d), categorical generalisations of the poor learning outcomes of migrants with long residence in Sweden are managed through the facilitator's mediating work. The analysis shows how the facilitator teams up with the migrant as a co-explainer, which helps portraying the migrant and her husband as exemplary language learners. The volunteer Ada is chatting with Rim and Noor. Fatima is sitting at the same table, assisting with translating when needed (seating arrangement is shown in Figure 1). Prior to the fragment represented in Excerpt 1a, the participants have been talking about how to practice Swedish. Noor has mentioned that she reads Swedish fairy tales aloud for her grandchildren.

Establishing a Swedish-only norm as the best practice
In Excerpt 1a, a Swedish-only norm is collaboratively established. The participants seem to believe that maximum practice and maximum exposure are the best ways to learn an additional language. At least, this is the argument that the volunteer brings up in this excerpt. With a polar yes/no question, Ada asks Rim whether she speaks Swedish at home (line 5). The question is produced while pointing at Rim. The format of the question makes relevant a confirming answer (see Boyd and Heritage 2006 on polar questions), which Rim produces (line 6). Noor too responds with an affirmative reply (line 7), thereby aligning with the emerging norm concerning the use of Swedish in the home environment. Excerpt 1a. 'Do you speak Swedish at home?' In line 10, there is a shift in participation framework, with Ada categorising the action of talking Arabic at home as a bad habit. This generalisation is accomplished through a shift in reference from the Swedish personal pronoun ni ('you', plural) to the indexical non-specific term en del dom ('some people they'). The categorical description of 'some people's' deviant behaviour is used to invoke membership to a particular non-accepted category; that is, the category of 'those migrants who talk a minority language at home'. Such description reveals the expected normative behaviour of speaking Swedish at home. This category-bound best practice activity, tied to Rim, is assessed with a praise (line 8), whereas the deviant behaviour of using a minority language is assessed with the negative descriptor dumt ('silly', line 20). The inappropriateness of this behaviour is further emphasised through Ada's use of reported speech in line 12. The voice of those people who speak Arabic at home is animated with a smiley voice quality. This overlay of her own voice enables Ada to express her stance towards this category-bound activity. Noor, for her part, rejects being included in this membership category, positioning herself instead as a speaker of Swedish (lines 14 and 17).
In line 13, overlapping with Ada's reported speech, Rim produces an in-breath, followed by hesitation tokens delivered with a creaky voice as she simultaneously smiles and glances at Fatima (Figure 2). This embodied action marks the onset of a word search (see Excerpt 1b).

Speaking on behalf of the migrant
In Excerpt 1b, Rim accounts for how she practices Swedish at home, thus distancing herself and her habits from 'some people's' bad behaviours as described before. She thereby aligns with the volunteer's normative framework. In this process of accounting, Fatima teams up with Rim as a coexplainer through speaking on behalf of her.
In line 21, Rim turns to Fatima, and invites her to help with a word in Swedish, which Fatima provides, arg ('angry'). Rim repeats the sought-for-word with rising, try-marked (cf. Lerner 1996) intonation, while maintaining mutual gaze with Fatima; the combination of the prosody of Rim's turn and her embodied behaviour makes relevant confirmation from Fatima (line 25). Having resolved the word search, Rim redirects her gaze to Ada (line 26) and accounts for how she practices Swedish at home. With this explanatory account, she portrays herself as the 'ideal' language learner who communicates with her partner in the language of the receiving country, rather than using Arabic, which they both have in common. Throughout Rim's telling in Swedish, Fatima displays her engagement through smiling and gazing at Rim.
((Ten lines of transcribed talk omitted; Ada states that training Swedish at home is the best way to learn Swedish; Noor rephrases Rim's telling in Arabic)) In line 34, Rim tries to say that her husband gets mad when she does not understand much (lite (.) inte förstår, 'a little not understand'). As she formulates her turn, she manifests a problem with the delivery of the Swedish word arg ('angry'). Specifically, after the first turn constructional unit, Rim produces an in-breath and various hesitation tokens, thereby indicating her engagement in a word search. Eventually, Rim produces a try-marked candidate outcome for the search, 'erj?', which is close to the target word arg ('angry') in Swedish. As she does so, she turns to Fatima. Thus, both the prosody of the candidate word and Rim's embodied behaviour invite Fatima's assistance. With a terminal overlap, Fatima produces a completion of Rim's turn-in-progress in line 35: han blir arg ('he gets angry'). In line 37, she adds å nervös ('an' edgy'), which nuances the description of the domestic scene of practicing Swedish (the verb root assab-in Arabic in inflected forms can be used to mean both 'angry' and 'edgy'). Rim without delay displays her affiliation with Fatima's added nuance by producing the stretched acknowledgement token ja::::: ('yeah') and various laughter tokens (line 38). With her anticipatory completion (Lerner 1996) of Rim's turn in line 34, Fatima not only provides a solution to the word search, but also assists Rim in explaining the details of her story to Ada. She thereby talks on behalf of Rim, aligning with the activity of producing accounts accompanied by category-bound descriptors. Fatima's gazing at Ada (line 35) further demonstrates her involvement in the interaction as a conversational partner.
Subsequently, Ada enquires whether Rim gets mad when her husband gets angry at her (line 39). In response to that, Rim laughs (line 40), while Noor takes Ada's turn as a display of misunderstanding, as if Ada had applied the descriptor arg ('angry') to the wrong person referent (i.e., Rim; see line 41). Fatima again steps in as a spokesperson for Rim (line 42) and re-establishes intersubjectivity among the participants with the turn: han blir arg ('he gets angry') with emphasis on the indexical reference han ('he') that ties the categorical activity to the husband. Ada confirms her understanding with a repetition of Fatima's turn, which attests that intersubjectivity is restored.

Challenging negative characterisations
In Excerpt 1 c and 1d, it is Rim's husband who is in focus. Fatima's mediating practices enable Rim to challenge negative characterisations associated with the category 'migrants with long residency in Sweden' and portray her husband as an educated person.
In line 64, when shared understanding about the length of residency of Rim's husband (twelve years) has been established, Ada shifts reference from a personal 'he' to a generic 'they', which involves a change in participation framework. Here, she asserts knowledge of Swedish as normal standard for people with twelve years in Sweden. She then moves into a categorical description of the poor learning outcomes of 'some' (en del) migrants (line 67), who only speak their language despite their long residency in Sweden (lines 66-68). Throughout the delivery of her turn and even during the subsequent pause (line 69), Ada keeps looking at Rim, whose face expression (see her eyebrows drawn together (see line 67)) seems to display confusion. Indeed, in line 70, Rim turns to Fatima and initiates repair (Sacks, Schegloff, and Jefferson 1974) with an open-class repair initiator in Arabic followed by a claim of non-understanding (sheno maftehamet, 'what I didn't understand'). Fatima immediately completes the repair by translating Ada's trouble-source turn into Arabic (lines 72-78). Note that Fatima, unlike the primary party, uses third-person reference in her rendition (see lines 75-76), which ties the category-bound activity 'speaking Swedish' to Rim's husband and not to people with twelve years of residency in general. Furthermore, she refers to 'many people' (keteer 'alam, see line 77) as lacking proficiency in Swedish (line 78) versus 'some people' in the original utterance, thereby broadening the scope of this category membership. In partial overlap with the end of Fatima's rendition, Rim, maintaining mutual gaze with Fatima, responds with two negation particles in Arabic (line 79), thus rejecting the category 'poor skills in Swedish' as applicable to her husband. Then, with her gaze redirected at Ada, Rim provides an assessment of her husband's skills in Swedish: jättebra (0.2.) talar svenska ('very good speaks Swedish', line 81). This praiseworthy description of her husband thus challenges Ada's membership categorisation.
In Excerpt 1d, Rim provides further descriptions of her husband's skills.
The excerpt starts with Ada's moral upshot (line 83): if her husband is good at Swedish, he should help Rim when she does not understand. She goes on to produce a piece of general advice: the best way to learn Swedish is to speak it (line 88). Rim does not respond to this; instead, she refocuses the talk on her husband's skills. Specifically, she turns to Fatima and explicitly asks her to explain to Ada that her husband is a pharmacist (line 89) and that he has studied in Sweden (line 91). Rim's request is formulated as golelha ('tell her'). Gavioli (2015) points to the dual function of such turn structures: (i) marking issues dealt with in the current speaker's turn as critical or delicate; and (ii) projecting an involvement of the language mediator to do something more than simply translate. The criticality of the issue dealt with here is observable in the affective stance embodied by Rim's smiley face and by the laughter tokens produced in the request turn (line 89). With the golelha-structure, Rim marks that the issue dealt with here is in need of attention and explanation. It is critical in the sense that it deals with the social identity of Rim's husband as a well-established citizen in the Swedish society, a membership categorisation that challenges the negative description provided by Ada (see Excerpt 1 c). Fatima's response to Rim's request displays her alignment with the criticality highlighted by Rim. Fatima's rendition (lines 93 and 96), in fact, is not just a translation of Rim's words, since she explains that Rim's husband is educated in Sweden with the descriptor utbildad här i Sverige ('educated here in Sweden', line 96), which is an expansion of Rim's words. Specifically, Fatima selfrepairs the descriptor studerade ('studied', line 96) with the category predicate utbildad ('educated'), which associates more clearly to the professional identity of Rim's husband as a pharmacist.
In all instances of mediating talk described in this example (Excerpts 1a, 1b, 1 c and 1d), the facilitator's help is enlisted and sanctioned by the migrant. However, the facilitator does more than translate talk. She speaks on behalf of the migrant to explain details of critical importance for Rim's identity as a language learner and for her husband's identity as an educated person. She does so by teaming up with Rim as a co-explainer: by adding nuances (e.g. the descriptor nervös, 'edgy' in Excerpt 1b); by correcting misunderstandings when intersubjectivity is at risk; and by making expansions of the original utterance (e.g. the descriptor utbildad här i Sverige, 'educated here in Sweden' in Excerpt 1d). All these practices of doing more than translating, together with the Arabic language translation in Excerpt 1 c, help the migrant to reject categorical generalisations of bad behaviour (Excerpt 1a, 1 c) and portray herself and her husband as good examples.

Managing beliefs about ways of getting married
In the following example (Excerpts 2a, 2b and 2 c), beliefs about morally questionable ways of getting married in the migrant's culture are invoked. Apart from the facilitator Fatima, the example involves the two Iraqi café visitors, Rim and Noor, and two volunteers, Bert and Arne (see Figure 3 for the seating arrangement).

Enlisting interpreter help
Excerpt 2a starts when the participants are four and a half minutes into the conversation. Prior to this excerpt, Rim was asked why she married her husband and whether somebody decided that. Whereas Rim got married for love, an Afghani asylum seeker who lives at Arne's place was forced by her father to marry a much older man when she was thirteen years old. In the excerpt below, Bert and Arne refer to this woman (lines 1-2).
Excerpt 2a. How did you end up getting married? In line 4, there is a quick shift from talking about the Afghani woman to talking about Rim's marriage, which possibly causes a problem in understanding (line 8). In response to that, Bert and Rim summon Fatima, who is engaged in a conversation at another table. When Fatima arrives at their table, Bert announces the problem, framing it as a sensitive topic by explicitly acknowledging its delicacy (nu (0.4) nu har vi delikat problem, 'now (0.4) now we have a delicate problem', line 33) and by addressing its complicated nature (hur va're (hhegentligenhh, 'how was it actually', line 40). In lines 43-55, Arne reports what they have been talking about.

Expressing stance and siding with the migrant
Excerpt 2b immediately follows upon Excerpt 2a. Translation in this excerpt is preceded by the facilitator's orientation to the problematic nature of their prior topical talk. Excerpt 2b. 'Family secrets have become public'. In line 64, Fatima's use of the resumption marker okay orients to the closing of the previous sequence and to the beginning of the new activity; that is, providing a translation of Arne's talk. While maintaining mutual gaze with Rim throughout her turn (onset in line 64, continued in lines 65 and 67), Fatima rephrases Arne's talk, (okay ʿam yesa'lke huwa eyqol enuu enti shlon (.) shlon (0.7) etzawajty etʿarrafty ʿala zuiaik, 'okey he's asking you he's saying how did you how did you get married met your husband'). Rim starts a response in lines 66 and 68, but abandons her turns as Fatima produces further talk. Specifically, in line 69, Fatima adds a polar question (habaytu baʿath?, 'were you in love?') that makes relevant a yes or no response. The answer is provided by Rim with no delays and is formulated in Swedish: as she gazes at Arne and Bert, she says j¿a ('yes', line 70). In line 72, she reiterates and reinforces her response by saying: ja absolut ('yes absolutely'). Rim's use of the intensifier absolut does more than confirming (note also Bert's verbal and embodied uptake of this formulation in line 74, Figure 5) and challenges the most fundamental presuppositions of a polar question, that both yes and no are possible answers (cf. Stivers 2011). The facilitator's mediating work here is more than simple rendition: she takes up Arne's concern about Rim's marriage and adapts its telling to the recipient. The question design enables Rim to contest the presupposition that she might belong to that category of Iraqi women who would be willing to marry a man for reasons other than love. By asserting that such a possibility is unthinkable, she positions herself as a good example, taking 'the moral high ground' (cf. Stivers 2011, 86).

Dealing with moral issues
The discussion in Excerpt 2 c is a direct continuation of Excerpt 2b. At this point, when the original questioner has received an answer, the facilitator re-engages in mediating work. Due to limited space, the final part of this sequence is not shown here (29 lines of transcription omitted).
Excerpt 2 c. 'How did you meet him?' ((The discussion about the kinship between Rim and her husband continues; 29 lines of transcription omitted.)) In this excerpt, the facilitator's mediating work deals with moral issues concerning blood relationship between spouses. The excerpt shows how the facilitator accounts for the kinship between Rim and her husband, thereby speaking on behalf of Rim, to reassure the volunteers that any possibility of blood relationship between them is excluded.
In line 79, Fatima asks Rim in Arabic how she met her husband. This question was never mentioned in Arne's original utterance. Once Rim has provided a response to Fatima (her husband is the brother of her uncle's wife), Fatima's retelling of her answer to the volunteers develops into a lengthy explanation in Swedish of the kinship between Rim and her husband. Specifically, Fatima reports that Rim and her husband know each other (line 89), but are not cousins (line 92). This additional piece of information, that Rim and her husband are not related by blood, was not provided in the primary party's turn (see lines 81-82 and 84) and prefaces the information that Rim's husband is a relative of hers. In other words, the addition of this detail (that Rim and her husband are not cousins) seems to be doing prefatory work, aimed at preventing strong reactions from the Swedish volunteers in whose cultural background marriage among relatives is not common, and not accepted. This piece of information is met with positive assessments produced by Arne (lines 93 and 95), who thus orients to a normative perspective according to which marriage among people who are related by blood should not be permitted. Furthermore, the assistance provided by the volunteers during the course of Fatima's retelling adds to this normative work. Specifically, in line 96 Fatima produces an incomplete turn: ja men finns e::m ('yes but [there] is e::m'), which Arne completes with lite släkt där ('some kinship there', line 97), thereby suggesting that Rim and her husband are related to some extent, which Fatima confirms in line 98. Later on, Bert provides assistance by teaming up with Fatima as a co-explainer, suggesting second cousins and other kinship terms as applicable to Rim and her husband (lines 100, 105, 107).
Finally, in lines 109-110, Fatima gazes at Arne and produces a turn characterised by speech perturbations, where she retells what Rim has explained to her in Arabic, that her husband is the brother of her uncle's wife (see line 81). Although Arne acknowledges Fatima's explanation as reassuring in the continuation of this sequence, Fatima resumes explanatory talk and provides additional clarifications (not shown here; 27 lines omitted), seemingly to reassure the volunteers about the nature of Rim's marriage.
Overall, this example (Excerpt 2a, 2b, 2c) shows how the facilitator sides with the migrant and expresses stances in dealing with a topic that is framed as sensitive. In doing so, she does more than translating: she also addresses moral issues through explaining critical issues about kinship and blood relationship. She does so through (i) expanding on the question agenda of the primary party (Arne), and (ii) adding information.

Concluding discussion
In this article, the mediating work of a facilitator, a lay multilingual speaker, in conversations between ethnic Swedes and Arabic-speaking migrants at a language café set up as part of a civic integration project has been examined. Two examples have been analysed in which the participants engage in normative work concerning moral issues and, as part of this work, mobilise categorisations of themselves and others. The ethnomethodological approach of MCA (Sacks 1992;Antaki and Widdicombe 1998) has been my focus when approaching the normative work in the interaction.
The examples illustrate how lay people's ideologies and beliefs about minority ethnic groups are locally invoked in intercultural encounters. In this way, the volunteers' normative work exemplifies how national identities (e.g. Anderson 2016;Billig 1995;Wodak et al. 2009) are reproduced through discursive practices in everyday life. The analyses attest how a normative framework is invoked in the conversations, as demonstrated through shifts in reference and participation framework (i.e. from talking about a specific person to talking about people from the migrants' country in general). In both examples, the volunteers' evaluative orientations, as displayed through question design and assessments, demonstrate presuppositions concerning their own identity and the positioning of the migrant (Rim) regarding moral issues.
A crucial finding in this study concerns the contribution of the multilingual facilitator. The data examined here show examples of mediated interaction which allows the migrant to let her voice be heard and to participate in the normative discourse, which is critical for the positioning of the migrant's identity. The analyses show how the facilitator is invited to deal with issues marked as delicate in the interaction, by the migrant with the golelha ('tell her') structure (see Excerpt 1d; cf. Gavioli 2015) and by the volunteers' request for assistance with a sensitive question (see Excerpt 2a). The facilitator, in dealing with these issues is shown to do more than translate. There is ample proof of her involvement in the interaction. As demonstrated in the analyses, her contributions are not limited to plain renditions of turns. In fact, she also 'coordinates' (Wadensjö 1998) the interaction (see also Baraldi and Gavioli 2012) by taking positions and highlighting co-actors' perspectives. This confirms the dialogical view of the mediator as an active participant in a triadic interaction (Wadensjö 1998). The examples show how the facilitator, by virtue of her multicultural and multilingual competence, makes sense of what is said in the interactions on both sides. She does so through adding information and explaining moral issues that are raised as delicate by the participants. The facilitator thereby makes reference to possible contextual assumptions (cf. Mason 2006). An illustrative example of this is provided in the second extract, when the facilitator engages in a lengthy explanation about blood relationship between Rim and her husband (Excerpt 2 c), possibly drawing on her background knowledge about the setting. This confirms Baker's (2006) argument that mediated interaction is about situated construction of meaning. All in all, the practices of voicing and speaking on behalf of the translatee help the migrant to tell her story and negotiate her cultural identity that challenges negative characterisations of practices in her culture. In this respect, the analyses bring to the fore the mediating work of an intercultural mediator, who in fact performs cultural brokering (e.g. Angelelli 2004;Del Torto 2008).
Overall, these findings highlight the crucial role played by non-professionals' mediating work in the migrant context (cf. Cronin 2006;Roels et al. 2015;Inghilleri 2017). A scenario with no interpreter present would leave the migrant voiceless. With a professional interpreter (which was never used in this setting), the dynamics would probably be different. The lack of the element of brokering would probably require more assertiveness on the part of the migrant to respond to the volunteers' views. This raises issues of access and inequality (Roels et al. 2015), and migrants' agency to negotiate and express their cultural identity, as not all migrants can count on the support and rapport of an intercultural mediator as an empowering process.

Disclosure statement
No potential conflict of interest was reported by the author.

Funding
This work was supported by Vetenskapsrådet [The Swedish Research Council].

Notes on contributor
Gunilla Jansson is Professor of Swedish Language at the Department of Swedish Language and Multilingualism at Stockholm University, Sweden. Her research interests include multilingual communication in working life and multilingualism and ethnicity in dementia and elderly care. She works within the theoretical and methodological frameworks of applied conversation analysis and ethnography. Recent projects include 'The language café as a social venue and a site for language training' (The Swedish Research Council 2017-03628). 2014-2017).