Academic writing centres in multilingual settings: Intermediary agents of higher education language policy?

Academic writing centres in non-Anglophone countries navigate language policies that encourage the development of local languages and English within multilingual university settings. These policies entail diverging language ideologies that connect language use to diversiﬁcation and standardisation aims. Although writing centres are stakeholders in academic language development and therefore participants in the policy process, their role and agency in this process has been neglected. The study explores how writing centre staff perceive their work in relation to university language policies and aims. Data were collected via a nation-wide survey of centres in Sweden and seven focus groups. The results demonstrate that the staff claim agency as language brokers. Institutional ideologies of monolingualism and assessment regimes that value the text product over the learning process provide limitations. The re-ported practices illuminate the special status of Swedish for English text production, and opportunities for grassroots agency as intermediaries between students, lecturers and policies.


Introduction
With the massification and diversification of higher education, European non-Anglophone universities have introduced services that support students' academic writing development.In Sweden, these writing centres cater for two academic languages, Swedish and English, which corresponds to language policies at Nordic universities that aim to develop the national language(s) and English as the lingua franca of science ( Hultgren, Gregersen & Thøgersen, 2014 ).Besides the two dominant languages, a plethora of other languages is used at Swedish universities due to the linguistic diversity of the staff and students (e.g.Kaufhold & Wennerberg, 2020 ).
This multilingual reality has been acknowledged by the Nordic Council of Ministers ( Gregersen et al., 2018 ) in terms of engaging with multilingual speakers nationally and reaching out internationally.Their report thus echoes aims of widening access to higher education nationally and universities' internationalisation.It recommends academic language development with a focus on the national language(s) and English in multilingual contexts where the lecturers and students do not share the same first/stronger language (L1).At the same time, such language policies index conflicting orientations towards language, for instance perceiving multilingualism as a valuable resource or as a problem (cf.Källkvist & Hult, 2020 ).
Writing centres centrally deal with the policy recommendation of language development in the context of university aims for diversification ( Lennartson-Hokkanen, 2016 ) and internationalisation, and are likely to encounter contrasting language orientations.In their pedagogic work, writing centre staff, i.e. learning developers (LDs), are entwined in language policy processes and might function as "intermediary agents" ( Hornberger & Johnson, 2007 ) between policies and language users.Despite their growing number and increasing professionalisation in non-Anglophone countries such as Sweden ( Bjernhager & Grönvall Fransson, 2018 ), their role and agency in this policy process has been neglected.
The aim of the article is therefore to explore LDs' perceptions of their pedagogic agency in relation to university language policies and aims within the multilingual setting of their institutions.More specifically, the study examines three research questions: 1 In what ways are university language policies and aims meaningful for the centres' work?
2 How do the LDs orient towards the multilingualism they encounter at their higher education institutions?3 How do the LDs evaluate their pedagogic agency in the context of their institutions?
The article contributes to Clarence's (2019) call for contextualising writing centre work "within the current dominant discourses shaping higher education" (p.118).The context of Northern European higher education provides an instructive case due to current professionalisation processes within writing centres as well as the strong influence of academic English and the legal protection of the local language(s).The purpose is to identify dilemmas and opportunities for pedagogic development in this multilingual context.

Academic writing centres in Sweden
In common with academic writing centres in other non-Anglophone countries (e.g.Brinkschulte, Grieshammer & Stoian, 2018 ;Okuda, 2019 ), the Swedish centres support literacies in both the local language and English.The centres were established in line with widening access effort s ( Swedish Council for Higher Education, 2016 ) since 1998 ( Bjernhager & Grönvall Fransson, 2018 ).Their initial remit was to support students' development of academic Swedish and to improve student completion rates ( Lennartson-Hokkanen, 2016 ).With the increase in Englishmedium instruction programmes and the aim for international publishing in English by research students, support for academic English writing has been added to the remit.The centres thus encounter effects of university policies relating to widening access and participation, internationalisation in relation to Englishmedium instruction, and parallel language use.
Currently there are 29 centres that cater for most of the Swedish higher education institutions (50 universities, university colleges and other higher education providers, according to the Swedish Higher Education Authority, 2020 ).They provide one-toone writing consultations and larger workshops, and contribute to academic programmes at other academic departments.They combine US writing centre pedagogies with local educational traditions in a setting with two dominant academic languages, Swedish and English ( Gustafsson & Ganobcsik-Williams, 2016 ).Writing centre staff, i.e. learning developers (LDs), have various professional backgrounds, often as teachers in secondary education or other academic language backgrounds, and some hold a PhD ( Bjernhager & Grönvall Fransson, 2018 ).In contrast to other countries, Swedish centres usually do not employ peer-tutors.
The writing centres' fundamental pedagogic aim is to support students to become independent academic writers through collaborative discussions and guidance on writing processes ( Gustafsson & Ganobcsik-Williams, 2016 ) without assessment.Treating writing as a process rather than solely focusing on the product is highly valued ( Clarence, 2019 ;Lennartson-Hokkanen, 2016 ;Okuda, 2019 ).Woodward-Kron (2007) for instance demonstrates the multiple functions a consultation can accomplish, which go beyond surfacelevel editing of texts and instead focus on meaning-making.Other researchers have however questioned this process ideal and contrast it with students' expectations that favour a more directive, result-focused teaching approach ( Aldohon, 2020 ;Clark, 2001 ).Thus, what academic writing development involves and how language is to be regulated is debated.
Debates about the role of writing centres in students' language development are imbued with beliefs about how language affects knowledge production and how language should be managed.With their contribution to language development, LDs are inevitably "participants in the policymaking process" ( Tollefson, 2011 , p. 372).Their agency is, however, rarely considered, despite their potential role as intermediaries between institutional policies and students' writing development.This article contributes to filling this gap by adopting a language ideological perspective and considering the LDs' agency in the language management process.

Language ideologies and literacy mediators
The article takes a language ideology perspective on academic writing and draws on insights from language policy research.A language ideology perspective considers writing to be embedded in ideological practices ( Street, 2001 ), i.e. it is situated within institutional power relations with "assumptions about the nature and value of academic writing conventions for participation in knowledge making" ( Lillis, 2019 ).Drawing on linguistic anthropology, we can define language ideologies as beliefs about language that are used to interpret and justify relations between language and society ( Kroskrity, 20 0 0 ;Silverstein, 1979 ).Ideologies are not fixed entities but interactional resources and speakers can draw on multiple ideologies ( Woolard & Schieffelin, 1994 ).Language ideologies are also central to language policy research.Following Spolsky (2012) , language policy has three components: the management of languages often within a formal entity, such as a university; the language practices of the people connected to this entity; and language ideologies.Language ideologies influence how people use language(s) and how they manage language use, not least in writing centre education.
There have been a range of studies on how language use is institutionally managed in various areas of higher education.In the area of classroom teaching, Madiba (2013) shows how teachers and students at a university in South Africa participate in opening up spaces for complementary language use in the course design, enabled by national language policy frameworks.However, Antia and Dyers (2016) are more cautious and point to the limitations of multilingual education in a context where languages other than English carry less status in South African education despite national policy.Similarly, in Northern European higher education, policy frameworks seem to open up spaces for multilingualism.For instance, the Nordic Council of Ministers suggest that Nordic universities "have a democratic duty to maintain and develop scientific dialogue, both in international research circles and with broad groups of citizens" and that this "requires multifaceted linguistic competencies" ( Gregersen et al., 2018 , p. 8).These discourses around democratisation and diversification relate to both widening access to higher education for students from underrepresented backgrounds, and international students ( Cots, Llurda & Garrett, 2014 ;Soler & Gallego-Balsà, 2019 ).Yet, such diversification perspectives often meet standardisation goals ( Fabricius, Mortensen & Haberland, 2017 ) and institutional language preferences that gravitate towards standard varieties of the national language(s) and/or English ( Kuteeva et al., 2020 ).Thus, contrasting ideologies have to be negotiated.
This article examines beliefs about language(s) and literacies connected to institutional discourses of democratisation, diversification and standardisation.The focus is on the LDs' agency to mediate and regulate language use within their institutional context.Following Ahearn (2001) , agency is defined as a "socio-culturally mediated capacity to act" (p.112) where agents do not act in isolation from other social actors ( Liddicoat & Taylor-Leech, 2020 ).
To analyse agency in literacy mediation, textual and ideological dimensions need to be investigated.On the level of interventions in the text, the study adopts the term literacy brokers as used by Lillis and Curry (2006) to describe the influence of language professionals on professional academic writers' work.In this study, the term captures the LDs' work of regulating language at the text level.Paxton (2014) adds an ideological dimension when she characterizes LDs' literacy mediation between students and their academic supervisors as an act of balancing between the resources the For analytic purposes, these two positions can be connected to the roles of advocates, who wish to change language practices ( Spolsky, 2018 ), and sponsors of literacy , who enable and regulate literacy "and gain advantage by it in some way" ( Brandt, 1998 , p.166).The latter economic metaphor indicates a continuation of institutional ways of knowledge production into which the learners are enlisted.

Materials and methods
To gain insights into LDs' views on their pedagogic agency in relation to university language policies and aims, the study adopted a two-stage qualitative research approach.It identified trends at the national level via a questionnaire and explored these trends in focus groups with seven local centres.The study was designed in collaboration with the Academic Writing Service at Stockholm University to ensure it incorporated practitioners' concerns.The contact person (the second author) was involved in the data collection and was not a study participant.

Data collection
Table 1 provides an overview of the collected data.First, we distributed an online questionnaire to the LDs across Sweden via the national centres' network (with ca.126 members).A reminder was sent to each centre directly.We received 43 responses from LDs at centres which are either affiliated with the university library ( N = 19), constitute an independent unit or are part of student services ( N = 15), or are part of a language or education department ( N = 8).One respondent did not reveal the organisational affiliation.The questionnaire (Appendix A) surveyed perceptions of university policies and aims regarding language use, internationalisation and widening participation, and the LDs' own language use with colleagues and students.To elicit reported practices, we developed four consultation scenarios (two for Swedishlanguage and two for English-language writing support) and asked how the LDs would respond to the multilingual students in these scenarios.
Second, we selected seven centres for focus groups (FGs) to explore the topics in the questionnaire further (Appendix B) and examine how the various perspectives and beliefs about language use are negotiated amongst the centre staff ( Litosseliti, 2003 ).Instead of providing consultation scenarios, we elicited narratives around multilingualism in the centres' work.The selected centres serve medium to large-size Swedish universities or university colleges.To capture some of the contextual variability that we noticed in the questionnaire, we selected centres that are located in different geographical areas in Sweden and have different organisational affiliations (i.e.library, independent unit, academic department).For one centre, we had to conduct an interview because only one member of the team was available.The interview lacks the group discussion but includes contextualised and shifting perspectives on language use and was therefore included in the thematic analysis (see 4.2).
We audio-recorded three focus groups face to face at the premises of the selected centres, and three plus the interview via online videoconferencing (due to travel restrictions at the onset of the Covid-19 pandemic).Depending on the size of the centres and the availability of staff, the number of participants in the focus groups ranged from two to six participants.For one centre we had 14 participants (Group 6).To enable an online discussion in this large group, we divided the group and conducted two focus groups in parallel.A total of 34 LDs participated.The difference in size and modes most likely affected the dynamics of the group discussions ( Myers, 1998 ), but the topics engaged participants in lively discussions irrespective of the mode.The focus group interviews lasted between 34 and 67 min and participants spoke Swedish and/or English.The recordings were transcribed verbatim and anonymised.
In line with ethical research practice, all participants were informed about the aim of the study after which they consented to participate.The FG participants were reminded about their right to withdraw any information they had provided during the discussion.This meant that we removed some expressions of opinions from the transcript before the analysis.The FG participants also received a first draft of this article to comment or ask for changes in the interpretation of the data.

Data analysis
To study the perceptions and reported practices collected in the questionnaire, we employed a thematic analysis.This analysis included the listing of reported languages and the "descriptive coding" ( Miles, Huberman & Saldaña, 2020 , p. 65) of recurring themes in the open-question answers on perceptions of university policies and consultation practices.These codes were applied in the analysis of the FGs.The FG discussions on the role of multilingualism in the LDs' work yielded additional codes (see Appendix C for list of codes).The thematic codes provided an overview of perspectives that were raised ( what ).
Since orientations towards language are interactional resources and speakers can draw on conflicting views to position themselves (cf.Woolard & Schieffelin, 1994 ), it was also important to analyse how the FGs negotiated the different perspectives that were raised.To analyse these negotiations, joint narratives of centre work were identified in the transcripts, i.e. instances where participants narrated and evaluated one event or several related events ( Georgakopoulou, 2015 ).In these narratives, the participants' stance taking ( Du Bois, 2007 ) was analysed by identifying how they evaluated their agency and how they positioned themselves in the narratives and towards other stakeholders ( Baynham, 2011 ) with focus on mediating roles ( Section 3 ).For the analysis we used the original transcripts in Swedish and/or English to ensure that nuances would not get lost in translation.

Results and discussion
In answer to research question 1, Section 5.1 introduces perceptions of university policies and aims derived from the ques- tionnaire and the FG discussions.To answer the second research question, Section 5.2 analyses in addition how the participants negotiated diverging perspectives on multilingual literacy development based on two longer excerpts from the FGs.Section 5.3 discusses the findings from the previous sections in terms of how the LDs evaluate their pedagogic agency and position themselves within their institutions.All excerpts derive from the FGs and are identified with the number of the group (e.g.FG2) and the number of the participant after the hash sign (e.g.#3).Appendix D lists the transcription conventions.For the excerpts that were originally in Swedish, translations by the first author are used and the original quotes are provided in Appendix E.
The translations are as literal as possible but some of the special expressions had to be substituted by near-equivalents for comprehensibility.

In what ways are university language policies and aims meaningful for the centres' work?
The questionnaire responses suggest that most of the LDs are aware of an institutional language policy ( Table 2 ).Of the 29 respondents who confirmed that their institution has a language policy, about two thirds agreed that the policy has an influence on their work.These respondents connected the impact to their use of Swedish and English with colleagues, students and on their website.This result reflects perceptions of parallel language use that focus on the two dominant academic languages used in Swedish higher education ( Karlsson, 2017 ).Some respondents also suggested that the centre exists as a consequence of the policy.Those who stated that the policy does not influence their work expressed doubts about the practical use of the policy.
The FGs further reveal that the perception of the language policy depends on the situatedness of the centre in the institution.For instance, a participant in Group 7 claimed (1): "I have never thought about this and nobody has said anything about a language policy here" (FG7#5); while another centre has been (2) "specifically recognised" ( lyfts fram ) by the vice chancellor (FG2#4).It was often the senior staff who expressed a greater awareness of the policy, and in several FGs it was mentioned that the centres have been involved in ongoing discussions about their institutions' language policies.Frequently this was combined with an acknowledgement that these activities have little impact on the daily work of the centres, which the following excerpt exemplifies: (3) I was actually partly involved when we updated the language policy […] But it does n't come in at all, there isn't anything that is connected to how we should guide students through their education language-wise.(FG6a#3) In terms of the institutions' aims of internationalisation and widening participation, 37 out of the 43 questionnaire respondents agreed that these aims influence their daily work.Internationalisation was mainly connected to the use of academic English by students and staff.The FGs added an intercultural aspect as expressed by FG4#1 (4): "it is about translating […] other academic cultures to our academic culture".
In the questionnaire, some considered the aim of widening participation to be one of the reasons for establishing writing centres in Sweden, in line with previous findings ( Lennartson-Hokkanen, 2016 ).In terms of academic writing, the centres support the use of academic Swedish and the socialisation into academic ways of working.However in the FGs the meaning of the term widening participation was also problematised and several participants considered it as a form of categorising students: (5) I want to give possibilities for all […] irrespective of like the language background […] one needs to learn the academic way […] I'm a bit against this with dividing into different categories.(FG7#3) Excerpt 5 emphasizes not only the rejection of the term but also the pedagogic ethos of inclusiveness.Relatedly, several LDs underlined that the students come with diverse individual needs and cannot be categorised.
The FG discussions led to the formulation of dilemmas for the centres that relate to diverging beliefs about the nature of academic literacy and conflicts between university aims.Excerpt (6) suggests that there is a mismatch between students' needs and university practices which is caused by a simplified view of what it takes to gain competence in an academic language: (6) The risk with this is that people just dump everything on us.There is a big sort of problem that students don't speak the language, Swedish probably, as we are in Sweden, or English.And then, just go to the unit of academic language and they will sort everything out for you.I think that sort of high up in the hierarchy it is taken for granted that people just speak these languages.Just write in English is easy, kind of thing.And then they come to us.(FG6a#6) The mismatch between projected task and reported reality means that the LDs encounter unrealistic expectations.
Excerpt 7 formulates the dilemma caused by contrasting aims and provisions most concisely: (7) There are competing perceptions.We have these policies about widening participation, internationalisation and [...] multilingualism-as-a-resource but there are also other discourses, monolingual norm [...] that turns into a collision where both teachers and students get caught in the middle, also the centre, which should then maybe in some way try to solve this, which isn't so easy.(FG1#2) Students and teachers are described in excerpt (7) as being caught between these policies while the centre is seen as an additional actor who has the difficult task of providing solutions.The centre is positioned as an intermediary between teachers, students and the different policies.
Overall, the LDs agree that the educational policies of widening participation and internationalisation and to a lesser extent the university language policies influence their work, especially at an organisational level.Language ideologies related to these aims can lead to dilemmas for the centres.Importantly, the widening partic- ipation label, which is supposed to signal inclusion and diversification ( Swedish Council for Higher Education, 2016 ), is to an extent rejected because it categorizes students into problem groups and leads to exclusion.

How do the LDs orient towards the multilingualism they encounter at their higher education institutions?
The questionnaire indicates that most LDs provide consultations for either Swedish or English writing, suggesting a separation of the two languages.Eleven respondents worked with both languages in parallel.The questionnaire asked about whether the LDs use other languages than the target language of the text in the consultations, even if these included only occasional phrases ( Table 3 ).Of the 35 who answered the questions regarding their consultation on Swedish texts, 25 reported that they use other languages, which turned out to be mainly English.The languages listed in Table 3 in brackets were only mentioned once or twice.Most respondents explained that they used another language for comparison between languages.
Of the 19 who answered the questions for consultations on English texts, 12 suggested that they use other languages, usually Swedish.The explanation was that these languages are used for comparison or that Swedish is the student's preferred language.The findings indicate that English and Swedish have a special status as multilingual resources.
In response to the consultation scenarios with multilingual students, few respondents suggested that they would draw on the students' first or stronger languages during the consultation.Of the 30 informants who responded to the Swedish target language scenarios, seven wrote they would advise to make notes in or compare with the stronger language before formulating the text in the target language.In the 15 responses to the English target language scenarios, there was only one respondent who suggested they would draw on multilingual resources in the same way as they described for the Swedish scenarios.Despite the probing of multilingualism in various questions on language use with colleagues and students preceding the scenarios, multilingual practices did not seem to be of primary concern for consultation strategies.
The FG discussions provided more space to negotiate views on the use of other languages than the target language of the text.Two central topics occurred across the FGs in narratives around supporting multilingual students at the centres: using multilingualism as a resource (see excerpt 7); and the special role of Swedish when talking about English texts.To gain insights not only into the variety of views but also how these are negotiated amongst LDs, two longer excerpts will be discussed.These excerpts were chosen because they concisely touch on most views around the two topics that were raised across the FGs.

Using multilingualism as a resource
The theme of using multilingualism as a resource was discussed as an ideal in the writing process and in opposition to the institutional norm of monolingualism.Excerpt (8) follows the description of the dilemma in excerpt (7).

Excerpt (8)
In lines 1-5, FG1#2 evaluates the use of multilingualism as a resource to be problematic but possible.Such possibilities are presented in two short narratives as examples that include "performed direct speech" ( Baynham, 2011 , p. 64).The pedagogic rea-sons for the engagement presented in the first example (lines 5-8) are to motivate and to connect to existing knowledge.The student's prior knowledge is valued as a relevant resource ( Lillis & Scott, 2007 ).The second narrative as example (lines 9-16) focuses on comparing linguistic structures of an L1 and the target language.Again, FG1#2 underlines that the student has a valuable resource and, importantly, that it does not matter whether the LD and the student share the language in these types of situations.In terms of mediation, FG1#2 positions herself as an advocate valuing the use of non-dominant language resources in the writing process.
FG1#4 adds her own experience (lines 17-24) but focuses on motivating and acknowledging the resources, especially in terms of quantity.In line 26, this acknowledgement turns from the valuable resource to the student's additional effort s as a second language writer.The contrastive "but" introduces the issue of mistakes and the perspective shifts from the writing process to the product invoking the concept of a standard language.
FG1#2 returns to the comment about the effort and explains how she uses her own experience of writing in an L2 to empathize with the student (lines 30-34).FG1#4 joins in with this pedagogic evaluation through backchanneling and commenting on normalising the effort (lines 33-38).Several participants add further examples, which are omitted in the excerpt.FG1#4 adds the final point in this sequence and returns to the importance of motivating students.Again, she shifts perspectives from process to product signalled by "even though".In line 42, assignments are introduced as primary writing purpose and an institutional monolingual ideology is endorsed.While the aspect of motivation is formulated as a weak obligation using the modal verb 'should' ( skulle ), the need to comply with institutional requirements is formulated as a fact.FG1#4 concludes by aligning with the product-focus, which FG1#2 sanctions as a given for the LDs.This is followed by general agreement in the group.
Throughout this sequence, the interlocutors position themselves as language brokers with reference to both the writing process and the textual product.The positions of advocate and sponsor are negotiated in that the conversation moves from realising an ideal of valuing students' diverse resources to promoting the continuation of the dominant monolingual norm in higher education.It is important to underline that the participants agree with both discourses and narrate how they try to do justice to these opposing views at the same time.Following the ideal can only be accomplished in "small ways" that might be pedagogically sound but seem to be counterintuitive in the face of dominant institutional norms, students' expectations (who want to pass assignments), and departmental expectations of "sorting students out" (cf.excerpt 6).

Using Swedish as a multilingual resource
As indicated in the questionnaire, the situation discussed in excerpt (8) changes if Swedish is the additional resource for talking about English texts.In the following excerpt, the LDs debate the use of Swedish as a natural choice, if shared by teacher and student, and the use of English as a pedagogic decision.Excerpt (9) follows the question as to whether other languages are used with students.

Excerpt (9)
FG6a#6 commences in Swedish with a short generic narrative explaining that most of the students who write English texts also speak Swedish and that it is therefore natural to use Swedish (line 2).The possibility that this spoken Swedish can be influenced by the English language of the text is characterised as Swenglish and evaluated as "interesting".The laughing voice further indicates that this observation is in some way remarkable, which may point to a perceived deviation from a standard language norm (cf.Kuteeva, 2020 ).In contrast to excerpt (8), no distraction from the text product is suspected.FG6a#6 concludes her turn by clarifying that English is the natural choice if the students do not speak Swedish.
FG6a#4 switches to English and adds to the question of language choice.In the hypothetical narrative (lines 10-16), she expresses a preference for using the target language but highlights the students' preference as a guiding principle.The language choice becomes a feature of a student-centred pedagogy.
The importance of sharing language skills at the level of "maximum competence" ( Blommaert & Backus, 2011 , p. 15) is made explicit by FG6a#2.Reporting that he is lacking this level of competence, he characterizes it as a "luxury".This positive evaluation contrasts with that of other multilingual language resources, which are relevant in small ways (excerpt 8).Yet FG6a#2 concludes that having to rely on English is rarely a problem.FG6a#1 adds another view in favour of using English and moves the discussion on from supporting students' reflections on academic writing to issues of language teaching and proficiency.
The first speaker of this sequence concludes by acknowledging and disaligning herself from the perspective raised by FG6a#1.Instead she returns to her initial point that speaking Swedish with a Swedish speaker is a natural choice.She now speaks English and aligns in this sense with the preceding speakers enacting the point that language choice is context-dependent.In addition, she problematizes the LDs' pedagogic role(s), an issue which will be discussed in Section 5.3 .
All participants but FG6a#1 position themselves and the students primarily as users of Swedish and English.All of them are language brokers engaging with the students' texts.Identifying their institutional position as advocates or sponsors is difficult.The pedagogic position of FG6a#1 points to a sponsorship role with a focus on the language of the textual product.The complementary language use presented by others indicates, on the one hand, advocacy for change because the LDs do not follow the parallel language use as envisaged by the majority of Swedish university language policy documents ( Karlsson, 2017 ).On the other hand, the use of the two dominant institutional languages is in line with practices that have been observed at Swedish universities for some time ( Söderlundh, 2012 ) and have been reflected in the policies of a few Swedish universities ( Karlsson & Karlsson, 2020 ).In this sense, the use of Swedish might be read as a form of sponsorship of established language policy practices.
While there were speakers of other languages among the FG participants, none of these spoke about using their other languages even if shared with the students.In fact, one participant explained that she avoids using this third language that she shares with many students and instead speaks English because this is expected by the international students.Unless Swedish is a shared resource, English monolingualism rules, a phenomenon which Fabricius et al. (2017) term the paradox of internationalisation.

How do the LDs evaluate their pedagogic agency in the context of their institutions?
In the narratives of multilingual practices, the LDs position themselves as language practitioners, professionals, and institutional service providers.In these different positions, there is more or less room for agency as intermediaries of language policy.
Within the practitioner position, the LDs take on the role of literacy sponsors orienting towards institutional conventions.In a few reported practices, they position themselves as advocates of complementary language use or translanguaging, in the sense of combining linguistic resources associated with different language codes for meaning making ( Li, 2018 ).Comparing linguistic features across languages ( Table 3 ) and speaking Swedish about English texts (excerpt 9) is common.The LDs also position themselves as different types of language users: often as monolinguals, even if they are able to draw on some knowledge of other languages; or as multilinguals, if they use Swedish to talk about English texts.As practitioners, the LDs construct their agency largely within institutional assessment regimes and a maximum-competence view of multilingualism.
As we have seen in the discussion so far, the LDs' professional positioning is complex.Similar to Bjernhager & Grönvall Fransson (2018) findings, the LDs in our sample have a variety of language-related professional backgrounds.Depending on the centre, the LDs might have additional affiliations at the university.For instance, FG1#4 stressed (10) that the centre work "is not the only or single function" they have.Despite their various career backgrounds, the LDs generally identify themselves in the FGs as pedagogic agents (11): "Well, basically we are also pedagogues" (FG4#2).But how exactly this fits with their institutional roles seems to be ambiguous, as evident in FG6a#1's reflection (12): "are we teachers or are we supervisors or are we something special that's not either and a little bit of both".In excerpt ( 9), FG6a#6 justifies her use of Swedish by positioning herself as an adviser rather than a language teacher.She thus claims agency in her lan-guage choice and connects it to the ethos of the non-directive and student-centred pedagogy ascribed to writing centres ( Gustafsson & Ganobcsik-Williams, 2016 ).
The positioning as institutional service providers depends on the centres' status within their universities, for instance whether the centre is a part of the library service or explicitly recognised centrally (excerpt 2).Similar to writing centre staff in countries with more established traditions (e.g.Paxton, 2014 in South Africa;Swales et al., 2001 in the USA), the LDs commented critically that their academic support and development work is often positioned as an add-on to subject curricula or as a remedial service (excerpt 6).However, irrespective of their organisational affiliation, the LDs report that their collaboration with departments and subject teachers strengthens their position (13): "lecturers are more and more interested in what we do and in what way, that means that lecturers also know that they need more tools to reach students" (FG4#1).In other FGs, such collaborations are characterised as "professional development" for lecturers who learn how to talk about language and writing strategies (FG1#2).The LDs are thus not only literacy brokers for students but also for lecturers.Providing specific support for writing in the disciplines is a development we see across European higher education ( Gustafsson & Ganobcsik-Williams, 2016 ).In such cross-departmental collaborations, the LDs create spaces for institutional agency and influence departmental practices ( Clarence, 2012 ;Eklund Heinonen, Lennartson-Hokkanen & Nord, 2018 ).
As mentioned in Section 5.1 , many of the centres have been involved in language policy discussions at the management level.But at this level the LDs often perceive themselves to be in a less powerful position as intermediaries or they present a hypothetical stance, e.g. ( 14): "we should be more involved in the big decision that deal with what affects us" (FG7#3).This situation might be shaped by a central conceptualisation of the centres' task to develop autonomous skills (cf.excerpt 6), and the centres' lack of a research remit ( Lennartson-Hokkanen, 2016 ).The potential for agency at the institutional level is thus limited.

Conclusion
The article discussed LDs' perceptions of their pedagogic agency in relation to language policies within the multilingual setting of Swedish universities.One of the centres' core tasks is to support students in developing their academic Swedish and/or English in line with language policies as formulated by the Nordic Council of Ministers ( Gregersen et al., 2018 ).In addition, the LDs generally agree that their work is in line with widening participation and internationalisation aims as they support students' diverse individual needs.Thus language policies, widening participation and internationalisation aims constitute the remit of the writing centres' work.To some extent, these policies provide a space for agency in that they inform the LDs' status as language brokers for students, and within their cross-departmental collaborations for lecturers.In this role, the LDs engage in talk around the students' texts and make not only language visible ( Lillis, 2006 ) but also the writing process.Depending on where in the institution the centre is situated, the policies can be used to emphasize the relevance of the centre for the university.At the same time, the LDs discuss the problematic aspects of these policies, e.g. that language policy documents often provide little practical guidance for the centres' work or that widening participation categories might reinforce marginalisation rather than promote inclusion.
Language ideologies that are associated with these policies and aims are often contradictory.The study demonstrates how the LDs navigate and appropriate these ideologies when they reflect on their work and evaluate their capacity to act.On the one hand and in line with the pedagogical ethos of writing centre work, they express a desire to include and value diverse language resources and academic experiences and thus echo discourses of democratisation and diversification (cf.Soler & Gallego-Balsà, 2019 ).This aim is predominantly presented as an ideal that can only be realised in small ways.An exception is the use of Swedish when talking about English texts, as has been observed in English-medium-instruction contexts ( Söderlundh, 2012 ).On the other hand, the LDs report acting as sponsors of literacy promoting the institutionally endorsed languages Swedish and English.
In addition, the study reveals the intricate connection between the dominant institutional "parallel monolingualism" ( Heller, 1999 ) and the prevailing essayist literacy perspective that values "product over process" ( Clarence, 2019 , p. 126).While the LDs stress the importance of supporting the writing process and the benefits of drawing on multilingual resources, the focus often shifts to the result and the use of the target language.These contrasting perspectives echo the notion of the "writing centre paradox" ( Carter, 2009 ), which suggests that LDs represent the students as well as the higher education system.In the multilingual context of the study, this means that the LDs continuously negotiate their aims of valuing the students' multilingual resources and adhering to institutional monolingual assessment regimes.
The LDs' reported practices suggest that this paradox both enables and limits their capacity to act as intermediaries between institutional policies and students' writing development.Of course, the reports might differ from actual practices as noticed by Aldohon (2020) .Nevertheless, the FGs facilitated a critical discussion of various practices and their feasibility (see excerpts 8,9).Further research can complement the findings by observing consultations and workshops.Yet, there is no reason to doubt that LDs advocate the use of other language resources in small ways and thus engage in "language policy from below" ( Mortensen, 2014 ) within the constraints of their position as institutional service providers.
Professionalisation effort s in the f orm of the current national network activities as well as advancing multilingual pedagogies for writing centres can further strengthen the LDs' agency.In terms of pedagogy, more sustained support, such as regular writing group sessions, can focus on writing processes and develop multilingual writers (e.g.Brinkschulte et al., 2018 ) who can activate their resources in strategic ways ( Flowerdew, 2015 ).With a process focus, the question of whether the languages are shared between students and LDs might be less relevant.In addition, the fact that Swedish seems to be a legitimate complementary language for consultations on English texts can be a starting point for reflecting on the use of other languages.In this way, writing centres would not only follow the Nordic Council of Ministers in their call for incorporating more languages in parallel ( Gregersen et al., 2018 ), but use them complementary in the writing process.The question on language ideologies also concerns the linguistic repertoire of the LDs and raises the question in what way experiences of academic writing in different languages, including non-dominant languages such as Arabic, can be valuable resources.This might have implications for staffing where multilingual competence might become a merit (cf.Carroll, 2017 ).
To realize the potential for advancing pedagogies for academic writing in multilingual settings, the professional identity of LDs needs to be strengthened.This can be achieved through reflections on the LDs' own stances on multilingualism, which was a side effect of the FGs noted by some participants and is in line with more extensive continuous professional development initiatives (e.g.Weaver, 2019 ).A part of this work could be to discuss recent policy documents as argumentative support for pedagogies that develop "multifaceted linguistic competencies" ( Gregersen et al., 2018 , p. 8).The LDs' work on reports (e.g.Nodén, 2019 ) or studies (e.g. Eklund Heinonen et al., 2018 ) as well as national network meetings and participation in relevant conferences can strengthen the academic position of the centres.These activities should be centrally supported by making research resources available ( Lennartson-Hokkanen, 2016 ).More work needs to be done to achieve long-term collaborations with academic departments and to enable process-oriented support over product-oriented assimilation.

Declaration of Competing Interest
None.

Table 1
Overview of data collection.

Table 2
University language policy impact.

Table 3
Reported use of other languages ( L ) than target language (TL) in consultations.